Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6

Conflicts:

	arch/x86/kernel/io_apic.c

139 files changed, 12319 insertions, 14278 deletions

diff --git a/fs/aio.c b/fs/aio.c
index f658441d566..d6f89d3c15e 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -191,6 +191,20 @@ static int aio_setup_ring(struct kioctx *ctx)
 	kunmap_atomic((void *)((unsigned long)__event & PAGE_MASK), km); \
 } while(0)
 
+static void ctx_rcu_free(struct rcu_head *head)
+{
+	struct kioctx *ctx = container_of(head, struct kioctx, rcu_head);
+	unsigned nr_events = ctx->max_reqs;
+
+	kmem_cache_free(kioctx_cachep, ctx);
+
+	if (nr_events) {
+		spin_lock(&aio_nr_lock);
+		BUG_ON(aio_nr - nr_events > aio_nr);
+		aio_nr -= nr_events;
+		spin_unlock(&aio_nr_lock);
+	}
+}
 
 /* __put_ioctx
  *	Called when the last user of an aio context has gone away,
@@ -198,8 +212,6 @@ static int aio_setup_ring(struct kioctx *ctx)
  */
 static void __put_ioctx(struct kioctx *ctx)
 {
-	unsigned nr_events = ctx->max_reqs;
-
 	BUG_ON(ctx->reqs_active);
 
 	cancel_delayed_work(&ctx->wq);
@@ -208,14 +220,7 @@ static void __put_ioctx(struct kioctx *ctx)
 	mmdrop(ctx->mm);
 	ctx->mm = NULL;
 	pr_debug("__put_ioctx: freeing %p\n", ctx);
-	kmem_cache_free(kioctx_cachep, ctx);
-
-	if (nr_events) {
-		spin_lock(&aio_nr_lock);
-		BUG_ON(aio_nr - nr_events > aio_nr);
-		aio_nr -= nr_events;
-		spin_unlock(&aio_nr_lock);
-	}
+	call_rcu(&ctx->rcu_head, ctx_rcu_free);
 }
 
 #define get_ioctx(kioctx) do {						\
@@ -235,6 +240,7 @@ static struct kioctx *ioctx_alloc(unsigned nr_events)
 {
 	struct mm_struct *mm;
 	struct kioctx *ctx;
+	int did_sync = 0;
 
 	/* Prevent overflows */
 	if ((nr_events > (0x10000000U / sizeof(struct io_event))) ||
@@ -267,21 +273,30 @@ static struct kioctx *ioctx_alloc(unsigned nr_events)
 		goto out_freectx;
 
 	/* limit the number of system wide aios */
-	spin_lock(&aio_nr_lock);
-	if (aio_nr + ctx->max_reqs > aio_max_nr ||
-	    aio_nr + ctx->max_reqs < aio_nr)
-		ctx->max_reqs = 0;
-	else
-		aio_nr += ctx->max_reqs;
-	spin_unlock(&aio_nr_lock);
+	do {
+		spin_lock_bh(&aio_nr_lock);
+		if (aio_nr + nr_events > aio_max_nr ||
+		    aio_nr + nr_events < aio_nr)
+			ctx->max_reqs = 0;
+		else
+			aio_nr += ctx->max_reqs;
+		spin_unlock_bh(&aio_nr_lock);
+		if (ctx->max_reqs || did_sync)
+			break;
+
+		/* wait for rcu callbacks to have completed before giving up */
+		synchronize_rcu();
+		did_sync = 1;
+		ctx->max_reqs = nr_events;
+	} while (1);
+
 	if (ctx->max_reqs == 0)
 		goto out_cleanup;
 
 	/* now link into global list. */
-	write_lock(&mm->ioctx_list_lock);
-	ctx->next = mm->ioctx_list;
-	mm->ioctx_list = ctx;
-	write_unlock(&mm->ioctx_list_lock);
+	spin_lock(&mm->ioctx_lock);
+	hlist_add_head_rcu(&ctx->list, &mm->ioctx_list);
+	spin_unlock(&mm->ioctx_lock);
 
 	dprintk("aio: allocated ioctx %p[%ld]: mm=%p mask=0x%x\n",
 		ctx, ctx->user_id, current->mm, ctx->ring_info.nr);
@@ -375,11 +390,12 @@ ssize_t wait_on_sync_kiocb(struct kiocb *iocb)
  */
 void exit_aio(struct mm_struct *mm)
 {
-	struct kioctx *ctx = mm->ioctx_list;
-	mm->ioctx_list = NULL;
-	while (ctx) {
-		struct kioctx *next = ctx->next;
-		ctx->next = NULL;
+	struct kioctx *ctx;
+
+	while (!hlist_empty(&mm->ioctx_list)) {
+		ctx = hlist_entry(mm->ioctx_list.first, struct kioctx, list);
+		hlist_del_rcu(&ctx->list);
+
 		aio_cancel_all(ctx);
 
 		wait_for_all_aios(ctx);
@@ -394,7 +410,6 @@ void exit_aio(struct mm_struct *mm)
 				atomic_read(&ctx->users), ctx->dead,
 				ctx->reqs_active);
 		put_ioctx(ctx);
-		ctx = next;
 	}
 }
 
@@ -555,19 +570,21 @@ int aio_put_req(struct kiocb *req)
 
 static struct kioctx *lookup_ioctx(unsigned long ctx_id)
 {
-	struct kioctx *ioctx;
-	struct mm_struct *mm;
+	struct mm_struct *mm = current->mm;
+	struct kioctx *ctx = NULL;
+	struct hlist_node *n;
 
-	mm = current->mm;
-	read_lock(&mm->ioctx_list_lock);
-	for (ioctx = mm->ioctx_list; ioctx; ioctx = ioctx->next)
-		if (likely(ioctx->user_id == ctx_id && !ioctx->dead)) {
-			get_ioctx(ioctx);
+	rcu_read_lock();
+
+	hlist_for_each_entry_rcu(ctx, n, &mm->ioctx_list, list) {
+		if (ctx->user_id == ctx_id && !ctx->dead) {
+			get_ioctx(ctx);
 			break;
 		}
-	read_unlock(&mm->ioctx_list_lock);
+	}
 
-	return ioctx;
+	rcu_read_unlock();
+	return ctx;
 }
 
 /*
@@ -1215,19 +1232,14 @@ out:
 static void io_destroy(struct kioctx *ioctx)
 {
 	struct mm_struct *mm = current->mm;
-	struct kioctx **tmp;
 	int was_dead;
 
 	/* delete the entry from the list is someone else hasn't already */
-	write_lock(&mm->ioctx_list_lock);
+	spin_lock(&mm->ioctx_lock);
 	was_dead = ioctx->dead;
 	ioctx->dead = 1;
-	for (tmp = &mm->ioctx_list; *tmp && *tmp != ioctx;
-	     tmp = &(*tmp)->next)
-		;
-	if (*tmp)
-		*tmp = ioctx->next;
-	write_unlock(&mm->ioctx_list_lock);
+	hlist_del_rcu(&ioctx->list);
+	spin_unlock(&mm->ioctx_lock);
 
 	dprintk("aio_release(%p)\n", ioctx);
 	if (likely(!was_dead))
diff --git a/fs/bio-integrity.c b/fs/bio-integrity.c
index 19caf7c962a..77ebc3c263d 100644
--- a/fs/bio-integrity.c
+++ b/fs/bio-integrity.c
@@ -111,7 +111,7 @@ void bio_integrity_free(struct bio *bio, struct bio_set *bs)
 	    && bip->bip_buf != NULL)
 		kfree(bip->bip_buf);
 
-	mempool_free(bip->bip_vec, bs->bvec_pools[bip->bip_pool]);
+	bvec_free_bs(bs, bip->bip_vec, bip->bip_pool);
 	mempool_free(bip, bs->bio_integrity_pool);
 
 	bio->bi_integrity = NULL;
diff --git a/fs/bio.c b/fs/bio.c
index df99c882b80..711cee10360 100644
--- a/fs/bio.c
+++ b/fs/bio.c
@@ -31,7 +31,11 @@
 
 DEFINE_TRACE(block_split);
 
-static struct kmem_cache *bio_slab __read_mostly;
+/*
+ * Test patch to inline a certain number of bi_io_vec's inside the bio
+ * itself, to shrink a bio data allocation from two mempool calls to one
+ */
+#define BIO_INLINE_VECS		4
 
 static mempool_t *bio_split_pool __read_mostly;
 
@@ -40,9 +44,8 @@ static mempool_t *bio_split_pool __read_mostly;
  * break badly! cannot be bigger than what you can fit into an
  * unsigned short
  */
-
 #define BV(x) { .nr_vecs = x, .name = "biovec-"__stringify(x) }
-static struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly = {
+struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly = {
 	BV(1), BV(4), BV(16), BV(64), BV(128), BV(BIO_MAX_PAGES),
 };
 #undef BV
@@ -53,12 +56,121 @@ static struct biovec_slab bvec_slabs[BIOVEC_NR_POOLS] __read_mostly = {
  */
 struct bio_set *fs_bio_set;
 
+/*
+ * Our slab pool management
+ */
+struct bio_slab {
+	struct kmem_cache *slab;
+	unsigned int slab_ref;
+	unsigned int slab_size;
+	char name[8];
+};
+static DEFINE_MUTEX(bio_slab_lock);
+static struct bio_slab *bio_slabs;
+static unsigned int bio_slab_nr, bio_slab_max;
+
+static struct kmem_cache *bio_find_or_create_slab(unsigned int extra_size)
+{
+	unsigned int sz = sizeof(struct bio) + extra_size;
+	struct kmem_cache *slab = NULL;
+	struct bio_slab *bslab;
+	unsigned int i, entry = -1;
+
+	mutex_lock(&bio_slab_lock);
+
+	i = 0;
+	while (i < bio_slab_nr) {
+		struct bio_slab *bslab = &bio_slabs[i];
+
+		if (!bslab->slab && entry == -1)
+			entry = i;
+		else if (bslab->slab_size == sz) {
+			slab = bslab->slab;
+			bslab->slab_ref++;
+			break;
+		}
+		i++;
+	}
+
+	if (slab)
+		goto out_unlock;
+
+	if (bio_slab_nr == bio_slab_max && entry == -1) {
+		bio_slab_max <<= 1;
+		bio_slabs = krealloc(bio_slabs,
+				     bio_slab_max * sizeof(struct bio_slab),
+				     GFP_KERNEL);
+		if (!bio_slabs)
+			goto out_unlock;
+	}
+	if (entry == -1)
+		entry = bio_slab_nr++;
+
+	bslab = &bio_slabs[entry];
+
+	snprintf(bslab->name, sizeof(bslab->name), "bio-%d", entry);
+	slab = kmem_cache_create(bslab->name, sz, 0, SLAB_HWCACHE_ALIGN, NULL);
+	if (!slab)
+		goto out_unlock;
+
+	printk("bio: create slab <%s> at %d\n", bslab->name, entry);
+	bslab->slab = slab;
+	bslab->slab_ref = 1;
+	bslab->slab_size = sz;
+out_unlock:
+	mutex_unlock(&bio_slab_lock);
+	return slab;
+}
+
+static void bio_put_slab(struct bio_set *bs)
+{
+	struct bio_slab *bslab = NULL;
+	unsigned int i;
+
+	mutex_lock(&bio_slab_lock);
+
+	for (i = 0; i < bio_slab_nr; i++) {
+		if (bs->bio_slab == bio_slabs[i].slab) {
+			bslab = &bio_slabs[i];
+			break;
+		}
+	}
+
+	if (WARN(!bslab, KERN_ERR "bio: unable to find slab!\n"))
+		goto out;
+
+	WARN_ON(!bslab->slab_ref);
+
+	if (--bslab->slab_ref)
+		goto out;
+
+	kmem_cache_destroy(bslab->slab);
+	bslab->slab = NULL;
+
+out:
+	mutex_unlock(&bio_slab_lock);
+}
+
 unsigned int bvec_nr_vecs(unsigned short idx)
 {
 	return bvec_slabs[idx].nr_vecs;
 }
 
-struct bio_vec *bvec_alloc_bs(gfp_t gfp_mask, int nr, unsigned long *idx, struct bio_set *bs)
+void bvec_free_bs(struct bio_set *bs, struct bio_vec *bv, unsigned int idx)
+{
+	BIO_BUG_ON(idx >= BIOVEC_NR_POOLS);
+
+	if (idx == BIOVEC_MAX_IDX)
+		mempool_free(bv, bs->bvec_pool);
+	else {
+		struct biovec_slab *bvs = bvec_slabs + idx;
+
+		kmem_cache_free(bvs->slab, bv);
+	}
+}
+
+struct bio_vec *bvec_alloc_bs(gfp_t gfp_mask, int nr, unsigned long *idx,
+			      struct bio_set *bs)
 {
 	struct bio_vec *bvl;
 
@@ -67,60 +179,85 @@ struct bio_vec *bvec_alloc_bs(gfp_t gfp_mask, int nr, unsigned long *idx, struct
 	 * If not, this is a bio_kmalloc() allocation and just do a
 	 * kzalloc() for the exact number of vecs right away.
 	 */
-	if (bs) {
+	if (!bs)
+		bvl = kmalloc(nr * sizeof(struct bio_vec), gfp_mask);
+
+	/*
+	 * see comment near bvec_array define!
+	 */
+	switch (nr) {
+	case 1:
+		*idx = 0;
+		break;
+	case 2 ... 4:
+		*idx = 1;
+		break;
+	case 5 ... 16:
+		*idx = 2;
+		break;
+	case 17 ... 64:
+		*idx = 3;
+		break;
+	case 65 ... 128:
+		*idx = 4;
+		break;
+	case 129 ... BIO_MAX_PAGES:
+		*idx = 5;
+		break;
+	default:
+		return NULL;
+	}
+
+	/*
+	 * idx now points to the pool we want to allocate from. only the
+	 * 1-vec entry pool is mempool backed.
+	 */
+	if (*idx == BIOVEC_MAX_IDX) {
+fallback:
+		bvl = mempool_alloc(bs->bvec_pool, gfp_mask);
+	} else {
+		struct biovec_slab *bvs = bvec_slabs + *idx;
+		gfp_t __gfp_mask = gfp_mask & ~(__GFP_WAIT | __GFP_IO);
+
 		/*
-		 * see comment near bvec_array define!
+		 * Make this allocation restricted and don't dump info on
+		 * allocation failures, since we'll fallback to the mempool
+		 * in case of failure.
 		 */
-		switch (nr) {
-		case 1:
-			*idx = 0;
-			break;
-		case 2 ... 4:
-			*idx = 1;
-			break;
-		case 5 ... 16:
-			*idx = 2;
-			break;
-		case 17 ... 64:
-			*idx = 3;
-			break;
-		case 65 ... 128:
-			*idx = 4;
-			break;
-		case 129 ... BIO_MAX_PAGES:
-			*idx = 5;
-			break;
-		default:
-			return NULL;
-		}
+		__gfp_mask |= __GFP_NOMEMALLOC | __GFP_NORETRY | __GFP_NOWARN;
 
 		/*
-		 * idx now points to the pool we want to allocate from
+		 * Try a slab allocation. If this fails and __GFP_WAIT
+		 * is set, retry with the 1-entry mempool
 		 */
-		bvl = mempool_alloc(bs->bvec_pools[*idx], gfp_mask);
-		if (bvl)
-			memset(bvl, 0,
-				bvec_nr_vecs(*idx) * sizeof(struct bio_vec));
-	} else
-		bvl = kzalloc(nr * sizeof(struct bio_vec), gfp_mask);
+		bvl = kmem_cache_alloc(bvs->slab, __gfp_mask);
+		if (unlikely(!bvl && (gfp_mask & __GFP_WAIT))) {
+			*idx = BIOVEC_MAX_IDX;
+			goto fallback;
+		}
+	}
 
 	return bvl;
 }
 
-void bio_free(struct bio *bio, struct bio_set *bio_set)
+void bio_free(struct bio *bio, struct bio_set *bs)
 {
-	if (bio->bi_io_vec) {
-		const int pool_idx = BIO_POOL_IDX(bio);
+	void *p;
 
-		BIO_BUG_ON(pool_idx >= BIOVEC_NR_POOLS);
-
-		mempool_free(bio->bi_io_vec, bio_set->bvec_pools[pool_idx]);
-	}
+	if (bio_has_allocated_vec(bio))
+		bvec_free_bs(bs, bio->bi_io_vec, BIO_POOL_IDX(bio));
 
 	if (bio_integrity(bio))
-		bio_integrity_free(bio, bio_set);
+		bio_integrity_free(bio, bs);
+
+	/*
+	 * If we have front padding, adjust the bio pointer before freeing
+	 */
+	p = bio;
+	if (bs->front_pad)
+		p -= bs->front_pad;
 
-	mempool_free(bio, bio_set->bio_pool);
+	mempool_free(p, bs->bio_pool);
 }
 
 /*
@@ -133,7 +270,8 @@ static void bio_fs_destructor(struct bio *bio)
 
 static void bio_kmalloc_destructor(struct bio *bio)
 {
-	kfree(bio->bi_io_vec);
+	if (bio_has_allocated_vec(bio))
+		kfree(bio->bi_io_vec);
 	kfree(bio);
 }
 
@@ -157,16 +295,20 @@ void bio_init(struct bio *bio)
  *   for a &struct bio to become free. If a %NULL @bs is passed in, we will
  *   fall back to just using @kmalloc to allocate the required memory.
  *
- *   allocate bio and iovecs from the memory pools specified by the
- *   bio_set structure, or @kmalloc if none given.
+ *   Note that the caller must set ->bi_destructor on succesful return
+ *   of a bio, to do the appropriate freeing of the bio once the reference
+ *   count drops to zero.
  **/
 struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
 {
-	struct bio *bio;
+	struct bio *bio = NULL;
+
+	if (bs) {
+		void *p = mempool_alloc(bs->bio_pool, gfp_mask);
 
-	if (bs)
-		bio = mempool_alloc(bs->bio_pool, gfp_mask);
-	else
+		if (p)
+			bio = p + bs->front_pad;
+	} else
 		bio = kmalloc(sizeof(*bio), gfp_mask);
 
 	if (likely(bio)) {
@@ -176,7 +318,15 @@ struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
 		if (likely(nr_iovecs)) {
 			unsigned long uninitialized_var(idx);
 
-			bvl = bvec_alloc_bs(gfp_mask, nr_iovecs, &idx, bs);
+			if (nr_iovecs <= BIO_INLINE_VECS) {
+				idx = 0;
+				bvl = bio->bi_inline_vecs;
+				nr_iovecs = BIO_INLINE_VECS;
+			} else {
+				bvl = bvec_alloc_bs(gfp_mask, nr_iovecs, &idx,
+							bs);
+				nr_iovecs = bvec_nr_vecs(idx);
+			}
 			if (unlikely(!bvl)) {
 				if (bs)
 					mempool_free(bio, bs->bio_pool);
@@ -186,7 +336,7 @@ struct bio *bio_alloc_bioset(gfp_t gfp_mask, int nr_iovecs, struct bio_set *bs)
 				goto out;
 			}
 			bio->bi_flags |= idx << BIO_POOL_OFFSET;
-			bio->bi_max_vecs = bvec_nr_vecs(idx);
+			bio->bi_max_vecs = nr_iovecs;
 		}
 		bio->bi_io_vec = bvl;
 	}
@@ -1346,30 +1496,18 @@ EXPORT_SYMBOL(bio_sector_offset);
  */
 static int biovec_create_pools(struct bio_set *bs, int pool_entries)
 {
-	int i;
+	struct biovec_slab *bp = bvec_slabs + BIOVEC_MAX_IDX;
 
-	for (i = 0; i < BIOVEC_NR_POOLS; i++) {
-		struct biovec_slab *bp = bvec_slabs + i;
-		mempool_t **bvp = bs->bvec_pools + i;
+	bs->bvec_pool = mempool_create_slab_pool(pool_entries, bp->slab);
+	if (!bs->bvec_pool)
+		return -ENOMEM;
 
-		*bvp = mempool_create_slab_pool(pool_entries, bp->slab);
-		if (!*bvp)
-			return -ENOMEM;
-	}
 	return 0;
 }
 
 static void biovec_free_pools(struct bio_set *bs)
 {
-	int i;
-
-	for (i = 0; i < BIOVEC_NR_POOLS; i++) {
-		mempool_t *bvp = bs->bvec_pools[i];
-
-		if (bvp)
-			mempool_destroy(bvp);
-	}
-
+	mempool_destroy(bs->bvec_pool);
 }
 
 void bioset_free(struct bio_set *bs)
@@ -1379,25 +1517,49 @@ void bioset_free(struct bio_set *bs)
 
 	bioset_integrity_free(bs);
 	biovec_free_pools(bs);
+	bio_put_slab(bs);
 
 	kfree(bs);
 }
 
-struct bio_set *bioset_create(int bio_pool_size, int bvec_pool_size)
+/**
+ * bioset_create  - Create a bio_set
+ * @pool_size:	Number of bio and bio_vecs to cache in the mempool
+ * @front_pad:	Number of bytes to allocate in front of the returned bio
+ *
+ * Description:
+ *    Set up a bio_set to be used with @bio_alloc_bioset. Allows the caller
+ *    to ask for a number of bytes to be allocated in front of the bio.
+ *    Front pad allocation is useful for embedding the bio inside
+ *    another structure, to avoid allocating extra data to go with the bio.
+ *    Note that the bio must be embedded at the END of that structure always,
+ *    or things will break badly.
+ */
+struct bio_set *bioset_create(unsigned int pool_size, unsigned int front_pad)
 {
-	struct bio_set *bs = kzalloc(sizeof(*bs), GFP_KERNEL);
+	unsigned int back_pad = BIO_INLINE_VECS * sizeof(struct bio_vec);
+	struct bio_set *bs;
 
+	bs = kzalloc(sizeof(*bs), GFP_KERNEL);
 	if (!bs)
 		return NULL;
 
-	bs->bio_pool = mempool_create_slab_pool(bio_pool_size, bio_slab);
+	bs->front_pad = front_pad;
+
+	bs->bio_slab = bio_find_or_create_slab(front_pad + back_pad);
+	if (!bs->bio_slab) {
+		kfree(bs);
+		return NULL;
+	}
+
+	bs->bio_pool = mempool_create_slab_pool(pool_size, bs->bio_slab);
 	if (!bs->bio_pool)
 		goto bad;
 
-	if (bioset_integrity_create(bs, bio_pool_size))
+	if (bioset_integrity_create(bs, pool_size))
 		goto bad;
 
-	if (!biovec_create_pools(bs, bvec_pool_size))
+	if (!biovec_create_pools(bs, pool_size))
 		return bs;
 
 bad:
@@ -1421,12 +1583,16 @@ static void __init biovec_init_slabs(void)
 
 static int __init init_bio(void)
 {
-	bio_slab = KMEM_CACHE(bio, SLAB_HWCACHE_ALIGN|SLAB_PANIC);
+	bio_slab_max = 2;
+	bio_slab_nr = 0;
+	bio_slabs = kzalloc(bio_slab_max * sizeof(struct bio_slab), GFP_KERNEL);
+	if (!bio_slabs)
+		panic("bio: can't allocate bios\n");
 
 	bio_integrity_init_slab();
 	biovec_init_slabs();
 
-	fs_bio_set = bioset_create(BIO_POOL_SIZE, 2);
+	fs_bio_set = bioset_create(BIO_POOL_SIZE, 0);
 	if (!fs_bio_set)
 		panic("bio: can't allocate bios\n");
 
diff --git a/fs/buffer.c b/fs/buffer.c
index 10179cfa115..776ae091d3b 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -99,10 +99,18 @@ __clear_page_buffers(struct page *page)
 	page_cache_release(page);
 }
 
+
+static int quiet_error(struct buffer_head *bh)
+{
+	if (!test_bit(BH_Quiet, &bh->b_state) && printk_ratelimit())
+		return 0;
+	return 1;
+}
+
+
 static void buffer_io_error(struct buffer_head *bh)
 {
 	char b[BDEVNAME_SIZE];
-
 	printk(KERN_ERR "Buffer I/O error on device %s, logical block %Lu\n",
 			bdevname(bh->b_bdev, b),
 			(unsigned long long)bh->b_blocknr);
@@ -144,7 +152,7 @@ void end_buffer_write_sync(struct buffer_head *bh, int uptodate)
 	if (uptodate) {
 		set_buffer_uptodate(bh);
 	} else {
-		if (!buffer_eopnotsupp(bh) && printk_ratelimit()) {
+		if (!buffer_eopnotsupp(bh) && !quiet_error(bh)) {
 			buffer_io_error(bh);
 			printk(KERN_WARNING "lost page write due to "
 					"I/O error on %s\n",
@@ -394,7 +402,7 @@ static void end_buffer_async_read(struct buffer_head *bh, int uptodate)
 		set_buffer_uptodate(bh);
 	} else {
 		clear_buffer_uptodate(bh);
-		if (printk_ratelimit())
+		if (!quiet_error(bh))
 			buffer_io_error(bh);
 		SetPageError(page);
 	}
@@ -455,7 +463,7 @@ static void end_buffer_async_write(struct buffer_head *bh, int uptodate)
 	if (uptodate) {
 		set_buffer_uptodate(bh);
 	} else {
-		if (printk_ratelimit()) {
+		if (!quiet_error(bh)) {
 			buffer_io_error(bh);
 			printk(KERN_WARNING "lost page write due to "
 					"I/O error on %s\n",
@@ -2913,6 +2921,9 @@ static void end_bio_bh_io_sync(struct bio *bio, int err)
 		set_bit(BH_Eopnotsupp, &bh->b_state);
 	}
 
+	if (unlikely (test_bit(BIO_QUIET,&bio->bi_flags)))
+		set_bit(BH_Quiet, &bh->b_state);
+
 	bh->b_end_io(bh, test_bit(BIO_UPTODATE, &bio->bi_flags));
 	bio_put(bio);
 }
diff --git a/fs/exec.c b/fs/exec.c
index 1f59ea079cb..02d2e120542 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -773,7 +773,6 @@ static int de_thread(struct task_struct *tsk)
 	struct signal_struct *sig = tsk->signal;
 	struct sighand_struct *oldsighand = tsk->sighand;
 	spinlock_t *lock = &oldsighand->siglock;
-	struct task_struct *leader = NULL;
 	int count;
 
 	if (thread_group_empty(tsk))
@@ -811,7 +810,7 @@ static int de_thread(struct task_struct *tsk)
 	 * and to assume its PID:
 	 */
 	if (!thread_group_leader(tsk)) {
-		leader = tsk->group_leader;
+		struct task_struct *leader = tsk->group_leader;
 
 		sig->notify_count = -1;	/* for exit_notify() */
 		for (;;) {
@@ -863,8 +862,9 @@ static int de_thread(struct task_struct *tsk)
 
 		BUG_ON(leader->exit_state != EXIT_ZOMBIE);
 		leader->exit_state = EXIT_DEAD;
-
 		write_unlock_irq(&tasklist_lock);
+
+		release_task(leader);
 	}
 
 	sig->group_exit_task = NULL;
@@ -873,8 +873,6 @@ static int de_thread(struct task_struct *tsk)
 no_thread_group:
 	exit_itimers(sig);
 	flush_itimer_signals();
-	if (leader)
-		release_task(leader);
 
 	if (atomic_read(&oldsighand->count) != 1) {
 		struct sighand_struct *newsighand;
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index e4a241c65db..04158ad74db 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -1721,7 +1721,7 @@ static loff_t ext4_max_size(int blkbits, int has_huge_files)
 	/* small i_blocks in vfs inode? */
 	if (!has_huge_files || sizeof(blkcnt_t) < sizeof(u64)) {
 		/*
-		 * CONFIG_LSF is not enabled implies the inode
+		 * CONFIG_LBD is not enabled implies the inode
 		 * i_block represent total blocks in 512 bytes
 		 * 32 == size of vfs inode i_blocks * 8
 		 */
@@ -1764,7 +1764,7 @@ static loff_t ext4_max_bitmap_size(int bits, int has_huge_files)
 
 	if (!has_huge_files || sizeof(blkcnt_t) < sizeof(u64)) {
 		/*
-		 * !has_huge_files or CONFIG_LSF is not enabled
+		 * !has_huge_files or CONFIG_LBD is not enabled
 		 * implies the inode i_block represent total blocks in
 		 * 512 bytes 32 == size of vfs inode i_blocks * 8
 		 */
@@ -2021,13 +2021,13 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	if (has_huge_files) {
 		/*
 		 * Large file size enabled file system can only be
-		 * mount if kernel is build with CONFIG_LSF
+		 * mount if kernel is build with CONFIG_LBD
 		 */
 		if (sizeof(root->i_blocks) < sizeof(u64) &&
 				!(sb->s_flags & MS_RDONLY)) {
 			printk(KERN_ERR "EXT4-fs: %s: Filesystem with huge "
 					"files cannot be mounted read-write "
-					"without CONFIG_LSF.\n", sb->s_id);
+					"without CONFIG_LBD.\n", sb->s_id);
 			goto failed_mount;
 		}
 	}
diff --git a/fs/inode.c b/fs/inode.c
index 0487ddba139..098a2443196 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -108,84 +108,100 @@ static void wake_up_inode(struct inode *inode)
 	wake_up_bit(&inode->i_state, __I_LOCK);
 }
 
-static struct inode *alloc_inode(struct super_block *sb)
+/**
+ * inode_init_always - perform inode structure intialisation
+ * @sb		- superblock inode belongs to.
+ * @inode	- inode to initialise
+ *
+ * These are initializations that need to be done on every inode
+ * allocation as the fields are not initialised by slab allocation.
+ */
+struct inode *inode_init_always(struct super_block *sb, struct inode *inode)
 {
 	static const struct address_space_operations empty_aops;
 	static struct inode_operations empty_iops;
 	static const struct file_operations empty_fops;
-	struct inode *inode;
-
-	if (sb->s_op->alloc_inode)
-		inode = sb->s_op->alloc_inode(sb);
-	else
-		inode = (struct inode *) kmem_cache_alloc(inode_cachep, GFP_KERNEL);
 
-	if (inode) {
-		struct address_space * const mapping = &inode->i_data;
-
-		inode->i_sb = sb;
-		inode->i_blkbits = sb->s_blocksize_bits;
-		inode->i_flags = 0;
-		atomic_set(&inode->i_count, 1);
-		inode->i_op = &empty_iops;
-		inode->i_fop = &empty_fops;
-		inode->i_nlink = 1;
-		atomic_set(&inode->i_writecount, 0);
-		inode->i_size = 0;
-		inode->i_blocks = 0;
-		inode->i_bytes = 0;
-		inode->i_generation = 0;
+	struct address_space * const mapping = &inode->i_data;
+
+	inode->i_sb = sb;
+	inode->i_blkbits = sb->s_blocksize_bits;
+	inode->i_flags = 0;
+	atomic_set(&inode->i_count, 1);
+	inode->i_op = &empty_iops;
+	inode->i_fop = &empty_fops;
+	inode->i_nlink = 1;
+	atomic_set(&inode->i_writecount, 0);
+	inode->i_size = 0;
+	inode->i_blocks = 0;
+	inode->i_bytes = 0;
+	inode->i_generation = 0;
 #ifdef CONFIG_QUOTA
-		memset(&inode->i_dquot, 0, sizeof(inode->i_dquot));
+	memset(&inode->i_dquot, 0, sizeof(inode->i_dquot));
 #endif
-		inode->i_pipe = NULL;
-		inode->i_bdev = NULL;
-		inode->i_cdev = NULL;
-		inode->i_rdev = 0;
-		inode->dirtied_when = 0;
-		if (security_inode_alloc(inode)) {
-			if (inode->i_sb->s_op->destroy_inode)
-				inode->i_sb->s_op->destroy_inode(inode);
-			else
-				kmem_cache_free(inode_cachep, (inode));
-			return NULL;
-		}
+	inode->i_pipe = NULL;
+	inode->i_bdev = NULL;
+	inode->i_cdev = NULL;
+	inode->i_rdev = 0;
+	inode->dirtied_when = 0;
+	if (security_inode_alloc(inode)) {
+		if (inode->i_sb->s_op->destroy_inode)
+			inode->i_sb->s_op->destroy_inode(inode);
+		else
+			kmem_cache_free(inode_cachep, (inode));
+		return NULL;
+	}
 
-		spin_lock_init(&inode->i_lock);
-		lockdep_set_class(&inode->i_lock, &sb->s_type->i_lock_key);
+	spin_lock_init(&inode->i_lock);
+	lockdep_set_class(&inode->i_lock, &sb->s_type->i_lock_key);
 
-		mutex_init(&inode->i_mutex);
-		lockdep_set_class(&inode->i_mutex, &sb->s_type->i_mutex_key);
+	mutex_init(&inode->i_mutex);
+	lockdep_set_class(&inode->i_mutex, &sb->s_type->i_mutex_key);
 
-		init_rwsem(&inode->i_alloc_sem);
-		lockdep_set_class(&inode->i_alloc_sem, &sb->s_type->i_alloc_sem_key);
+	init_rwsem(&inode->i_alloc_sem);
+	lockdep_set_class(&inode->i_alloc_sem, &sb->s_type->i_alloc_sem_key);
 
-		mapping->a_ops = &empty_aops;
- 		mapping->host = inode;
-		mapping->flags = 0;
-		mapping_set_gfp_mask(mapping, GFP_HIGHUSER_PAGECACHE);
-		mapping->assoc_mapping = NULL;
-		mapping->backing_dev_info = &default_backing_dev_info;
-		mapping->writeback_index = 0;
+	mapping->a_ops = &empty_aops;
+	mapping->host = inode;
+	mapping->flags = 0;
+	mapping_set_gfp_mask(mapping, GFP_HIGHUSER_PAGECACHE);
+	mapping->assoc_mapping = NULL;
+	mapping->backing_dev_info = &default_backing_dev_info;
+	mapping->writeback_index = 0;
 
-		/*
-		 * If the block_device provides a backing_dev_info for client
-		 * inodes then use that.  Otherwise the inode share the bdev's
-		 * backing_dev_info.
-		 */
-		if (sb->s_bdev) {
-			struct backing_dev_info *bdi;
+	/*
+	 * If the block_device provides a backing_dev_info for client
+	 * inodes then use that.  Otherwise the inode share the bdev's
+	 * backing_dev_info.
+	 */
+	if (sb->s_bdev) {
+		struct backing_dev_info *bdi;
 
-			bdi = sb->s_bdev->bd_inode_backing_dev_info;
-			if (!bdi)
-				bdi = sb->s_bdev->bd_inode->i_mapping->backing_dev_info;
-			mapping->backing_dev_info = bdi;
-		}
-		inode->i_private = NULL;
-		inode->i_mapping = mapping;
+		bdi = sb->s_bdev->bd_inode_backing_dev_info;
+		if (!bdi)
+			bdi = sb->s_bdev->bd_inode->i_mapping->backing_dev_info;
+		mapping->backing_dev_info = bdi;
 	}
+	inode->i_private = NULL;
+	inode->i_mapping = mapping;
+
 	return inode;
 }
+EXPORT_SYMBOL(inode_init_always);
+
+static struct inode *alloc_inode(struct super_block *sb)
+{
+	struct inode *inode;
+
+	if (sb->s_op->alloc_inode)
+		inode = sb->s_op->alloc_inode(sb);
+	else
+		inode = kmem_cache_alloc(inode_cachep, GFP_KERNEL);
+
+	if (inode)
+		return inode_init_always(sb, inode);
+	return NULL;
+}
 
 void destroy_inode(struct inode *inode) 
 {
@@ -196,6 +212,7 @@ void destroy_inode(struct inode *inode)
 	else
 		kmem_cache_free(inode_cachep, (inode));
 }
+EXPORT_SYMBOL(destroy_inode);
 
 
 /*
@@ -534,6 +551,49 @@ repeat:
 	return node ? inode : NULL;
 }
 
+static unsigned long hash(struct super_block *sb, unsigned long hashval)
+{
+	unsigned long tmp;
+
+	tmp = (hashval * (unsigned long)sb) ^ (GOLDEN_RATIO_PRIME + hashval) /
+			L1_CACHE_BYTES;
+	tmp = tmp ^ ((tmp ^ GOLDEN_RATIO_PRIME) >> I_HASHBITS);
+	return tmp & I_HASHMASK;
+}
+
+static inline void
+__inode_add_to_lists(struct super_block *sb, struct hlist_head *head,
+			struct inode *inode)
+{
+	inodes_stat.nr_inodes++;
+	list_add(&inode->i_list, &inode_in_use);
+	list_add(&inode->i_sb_list, &sb->s_inodes);
+	if (head)
+		hlist_add_head(&inode->i_hash, head);
+}
+
+/**
+ * inode_add_to_lists - add a new inode to relevant lists
+ * @sb		- superblock inode belongs to.
+ * @inode	- inode to mark in use
+ *
+ * When an inode is allocated it needs to be accounted for, added to the in use
+ * list, the owning superblock and the inode hash. This needs to be done under
+ * the inode_lock, so export a function to do this rather than the inode lock
+ * itself. We calculate the hash list to add to here so it is all internal
+ * which requires the caller to have already set up the inode number in the
+ * inode to add.
+ */
+void inode_add_to_lists(struct super_block *sb, struct inode *inode)
+{
+	struct hlist_head *head = inode_hashtable + hash(sb, inode->i_ino);
+
+	spin_lock(&inode_lock);
+	__inode_add_to_lists(sb, head, inode);
+	spin_unlock(&inode_lock);
+}
+EXPORT_SYMBOL_GPL(inode_add_to_lists);
+
 /**
  *	new_inode 	- obtain an inode
  *	@sb: superblock
@@ -561,9 +621,7 @@ struct inode *new_inode(struct super_block *sb)
 	inode = alloc_inode(sb);
 	if (inode) {
 		spin_lock(&inode_lock);
-		inodes_stat.nr_inodes++;
-		list_add(&inode->i_list, &inode_in_use);
-		list_add(&inode->i_sb_list, &sb->s_inodes);
+		__inode_add_to_lists(sb, NULL, inode);
 		inode->i_ino = ++last_ino;
 		inode->i_state = 0;
 		spin_unlock(&inode_lock);
@@ -622,10 +680,7 @@ static struct inode * get_new_inode(struct super_block *sb, struct hlist_head *h
 			if (set(inode, data))
 				goto set_failed;
 
-			inodes_stat.nr_inodes++;
-			list_add(&inode->i_list, &inode_in_use);
-			list_add(&inode->i_sb_list, &sb->s_inodes);
-			hlist_add_head(&inode->i_hash, head);
+			__inode_add_to_lists(sb, head, inode);
 			inode->i_state = I_LOCK|I_NEW;
 			spin_unlock(&inode_lock);
 
@@ -671,10 +726,7 @@ static struct inode * get_new_inode_fast(struct super_block *sb, struct hlist_he
 		old = find_inode_fast(sb, head, ino);
 		if (!old) {
 			inode->i_ino = ino;
-			inodes_stat.nr_inodes++;
-			list_add(&inode->i_list, &inode_in_use);
-			list_add(&inode->i_sb_list, &sb->s_inodes);
-			hlist_add_head(&inode->i_hash, head);
+			__inode_add_to_lists(sb, head, inode);
 			inode->i_state = I_LOCK|I_NEW;
 			spin_unlock(&inode_lock);
 
@@ -698,16 +750,6 @@ static struct inode * get_new_inode_fast(struct super_block *sb, struct hlist_he
 	return inode;
 }
 
-static unsigned long hash(struct super_block *sb, unsigned long hashval)
-{
-	unsigned long tmp;
-
-	tmp = (hashval * (unsigned long)sb) ^ (GOLDEN_RATIO_PRIME + hashval) /
-			L1_CACHE_BYTES;
-	tmp = tmp ^ ((tmp ^ GOLDEN_RATIO_PRIME) >> I_HASHBITS);
-	return tmp & I_HASHMASK;
-}
-
 /**
  *	iunique - get a unique inode number
  *	@sb: superblock
@@ -1292,6 +1334,7 @@ int inode_wait(void *word)
 	schedule();
 	return 0;
 }
+EXPORT_SYMBOL(inode_wait);
 
 /*
  * If we try to find an inode in the inode hash while it is being
diff --git a/fs/jfs/inode.c b/fs/jfs/inode.c
index 210339784b5..b00ee9f05a0 100644
--- a/fs/jfs/inode.c
+++ b/fs/jfs/inode.c
@@ -59,8 +59,14 @@ struct inode *jfs_iget(struct super_block *sb, unsigned long ino)
 		if (inode->i_size >= IDATASIZE) {
 			inode->i_op = &page_symlink_inode_operations;
 			inode->i_mapping->a_ops = &jfs_aops;
-		} else
+		} else {
 			inode->i_op = &jfs_symlink_inode_operations;
+			/*
+			 * The inline data should be null-terminated, but
+			 * don't let on-disk corruption crash the kernel
+			 */
+			JFS_IP(inode)->i_inline[inode->i_size] = '\0';
+		}
 	} else {
 		inode->i_op = &jfs_file_inode_operations;
 		init_special_inode(inode, inode->i_mode, inode->i_rdev);
diff --git a/fs/lockd/clntlock.c b/fs/lockd/clntlock.c
index 8307dd64bf4..1f3b0fc0d35 100644
--- a/fs/lockd/clntlock.c
+++ b/fs/lockd/clntlock.c
@@ -14,6 +14,7 @@
 #include <linux/sunrpc/svc.h>
 #include <linux/lockd/lockd.h>
 #include <linux/smp_lock.h>
+#include <linux/kthread.h>
 
 #define NLMDBG_FACILITY		NLMDBG_CLIENT
 
@@ -60,7 +61,7 @@ struct nlm_host *nlmclnt_init(const struct nlmclnt_initdata *nlm_init)
 
 	host = nlmclnt_lookup_host(nlm_init->address, nlm_init->addrlen,
 				   nlm_init->protocol, nlm_version,
-				   nlm_init->hostname);
+				   nlm_init->hostname, nlm_init->noresvport);
 	if (host == NULL) {
 		lockd_down();
 		return ERR_PTR(-ENOLCK);
@@ -191,11 +192,15 @@ __be32 nlmclnt_grant(const struct sockaddr *addr, const struct nlm_lock *lock)
 void
 nlmclnt_recovery(struct nlm_host *host)
 {
+	struct task_struct *task;
+
 	if (!host->h_reclaiming++) {
 		nlm_get_host(host);
-		__module_get(THIS_MODULE);
-		if (kernel_thread(reclaimer, host, CLONE_FS | CLONE_FILES) < 0)
-			module_put(THIS_MODULE);
+		task = kthread_run(reclaimer, host, "%s-reclaim", host->h_name);
+		if (IS_ERR(task))
+			printk(KERN_ERR "lockd: unable to spawn reclaimer "
+				"thread. Locks for %s won't be reclaimed! "
+				"(%ld)\n", host->h_name, PTR_ERR(task));
 	}
 }
 
@@ -207,7 +212,6 @@ reclaimer(void *ptr)
 	struct file_lock *fl, *next;
 	u32 nsmstate;
 
-	daemonize("%s-reclaim", host->h_name);
 	allow_signal(SIGKILL);
 
 	down_write(&host->h_rwsem);
@@ -233,7 +237,12 @@ restart:
 	list_for_each_entry_safe(fl, next, &host->h_reclaim, fl_u.nfs_fl.list) {
 		list_del_init(&fl->fl_u.nfs_fl.list);
 
-		/* Why are we leaking memory here? --okir */
+		/*
+		 * sending this thread a SIGKILL will result in any unreclaimed
+		 * locks being removed from the h_granted list. This means that
+		 * the kernel will not attempt to reclaim them again if a new
+		 * reclaimer thread is spawned for this host.
+		 */
 		if (signalled())
 			continue;
 		if (nlmclnt_reclaim(host, fl) != 0)
@@ -261,5 +270,5 @@ restart:
 	nlm_release_host(host);
 	lockd_down();
 	unlock_kernel();
-	module_put_and_exit(0);
+	return 0;
 }
diff --git a/fs/lockd/host.c b/fs/lockd/host.c
index e05d0441603..abdebf76b82 100644
--- a/fs/lockd/host.c
+++ b/fs/lockd/host.c
@@ -48,6 +48,7 @@ struct nlm_lookup_host_info {
 	const size_t		hostname_len;	/* it's length */
 	const struct sockaddr	*src_sap;	/* our address (optional) */
 	const size_t		src_len;	/* it's length */
+	const int		noresvport;	/* use non-priv port */
 };
 
 /*
@@ -222,6 +223,7 @@ static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni)
 	host->h_nsmstate   = 0;			/* real NSM state */
 	host->h_nsmhandle  = nsm;
 	host->h_server	   = ni->server;
+	host->h_noresvport = ni->noresvport;
 	hlist_add_head(&host->h_hash, chain);
 	INIT_LIST_HEAD(&host->h_lockowners);
 	spin_lock_init(&host->h_lock);
@@ -272,6 +274,7 @@ nlm_destroy_host(struct nlm_host *host)
  * @protocol: transport protocol to use
  * @version: NLM protocol version
  * @hostname: '\0'-terminated hostname of server
+ * @noresvport: 1 if non-privileged port should be used
  *
  * Returns an nlm_host structure that matches the passed-in
  * [server address, transport protocol, NLM version, server hostname].
@@ -281,7 +284,9 @@ nlm_destroy_host(struct nlm_host *host)
 struct nlm_host *nlmclnt_lookup_host(const struct sockaddr *sap,
 				     const size_t salen,
 				     const unsigned short protocol,
-				     const u32 version, const char *hostname)
+				     const u32 version,
+				     const char *hostname,
+				     int noresvport)
 {
 	const struct sockaddr source = {
 		.sa_family	= AF_UNSPEC,
@@ -296,6 +301,7 @@ struct nlm_host *nlmclnt_lookup_host(const struct sockaddr *sap,
 		.hostname_len	= strlen(hostname),
 		.src_sap	= &source,
 		.src_len	= sizeof(source),
+		.noresvport	= noresvport,
 	};
 
 	dprintk("lockd: %s(host='%s', vers=%u, proto=%s)\n", __func__,
@@ -417,6 +423,8 @@ nlm_bind_host(struct nlm_host *host)
 		 */
 		if (!host->h_server)
 			args.flags |= RPC_CLNT_CREATE_HARDRTRY;
+		if (host->h_noresvport)
+			args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
 
 		clnt = rpc_create(&args);
 		if (!IS_ERR(clnt))
diff --git a/fs/lockd/svc.c b/fs/lockd/svc.c
index 56b076736b5..252d80163d0 100644
--- a/fs/lockd/svc.c
+++ b/fs/lockd/svc.c
@@ -45,7 +45,7 @@
 static struct svc_program	nlmsvc_program;
 
 struct nlmsvc_binding *		nlmsvc_ops;
-EXPORT_SYMBOL(nlmsvc_ops);
+EXPORT_SYMBOL_GPL(nlmsvc_ops);
 
 static DEFINE_MUTEX(nlmsvc_mutex);
 static unsigned int		nlmsvc_users;
@@ -300,7 +300,7 @@ out:
 	mutex_unlock(&nlmsvc_mutex);
 	return error;
 }
-EXPORT_SYMBOL(lockd_up);
+EXPORT_SYMBOL_GPL(lockd_up);
 
 /*
  * Decrement the user count and bring down lockd if we're the last.
@@ -329,7 +329,7 @@ lockd_down(void)
 out:
 	mutex_unlock(&nlmsvc_mutex);
 }
-EXPORT_SYMBOL(lockd_down);
+EXPORT_SYMBOL_GPL(lockd_down);
 
 #ifdef CONFIG_SYSCTL
 
diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
index c2e9cfd9e5a..3e634f2a108 100644
--- a/fs/nfs/callback.c
+++ b/fs/nfs/callback.c
@@ -16,6 +16,7 @@
 #include <linux/mutex.h>
 #include <linux/freezer.h>
 #include <linux/kthread.h>
+#include <linux/sunrpc/svcauth_gss.h>
 
 #include <net/inet_sock.h>
 
@@ -182,10 +183,34 @@ void nfs_callback_down(void)
 	mutex_unlock(&nfs_callback_mutex);
 }
 
+static int check_gss_callback_principal(struct nfs_client *clp,
+					struct svc_rqst *rqstp)
+{
+	struct rpc_clnt *r = clp->cl_rpcclient;
+	char *p = svc_gss_principal(rqstp);
+
+	/*
+	 * It might just be a normal user principal, in which case
+	 * userspace won't bother to tell us the name at all.
+	 */
+	if (p == NULL)
+		return SVC_DENIED;
+
+	/* Expect a GSS_C_NT_HOSTBASED_NAME like "nfs@serverhostname" */
+
+	if (memcmp(p, "nfs@", 4) != 0)
+		return SVC_DENIED;
+	p += 4;
+	if (strcmp(p, r->cl_server) != 0)
+		return SVC_DENIED;
+	return SVC_OK;
+}
+
 static int nfs_callback_authenticate(struct svc_rqst *rqstp)
 {
 	struct nfs_client *clp;
 	RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]);
+	int ret = SVC_OK;
 
 	/* Don't talk to strangers */
 	clp = nfs_find_client(svc_addr(rqstp), 4);
@@ -194,21 +219,22 @@ static int nfs_callback_authenticate(struct svc_rqst *rqstp)
 
 	dprintk("%s: %s NFSv4 callback!\n", __func__,
 			svc_print_addr(rqstp, buf, sizeof(buf)));
-	nfs_put_client(clp);
 
 	switch (rqstp->rq_authop->flavour) {
 		case RPC_AUTH_NULL:
 			if (rqstp->rq_proc != CB_NULL)
-				return SVC_DENIED;
+				ret = SVC_DENIED;
 			break;
 		case RPC_AUTH_UNIX:
 			break;
 		case RPC_AUTH_GSS:
-			/* FIXME: RPCSEC_GSS handling? */
+			ret = check_gss_callback_principal(clp, rqstp);
+			break;
 		default:
-			return SVC_DENIED;
+			ret = SVC_DENIED;
 	}
-	return SVC_OK;
+	nfs_put_client(clp);
+	return ret;
 }
 
 /*
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 7547600b617..9b728f3565a 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -143,7 +143,6 @@ static struct nfs_client *nfs_alloc_client(const struct nfs_client_initdata *cl_
 	clp->cl_proto = cl_init->proto;
 
 #ifdef CONFIG_NFS_V4
-	init_rwsem(&clp->cl_sem);
 	INIT_LIST_HEAD(&clp->cl_delegations);
 	spin_lock_init(&clp->cl_lock);
 	INIT_DELAYED_WORK(&clp->cl_renewd, nfs4_renew_state);
@@ -224,31 +223,54 @@ void nfs_put_client(struct nfs_client *clp)
 	}
 }
 
-static int nfs_sockaddr_match_ipaddr4(const struct sockaddr_in *sa1,
-				 const struct sockaddr_in *sa2)
+#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
+static const struct in6_addr *nfs_map_ipv4_addr(const struct sockaddr *sa, struct in6_addr *addr_mapped)
 {
-	return sa1->sin_addr.s_addr == sa2->sin_addr.s_addr;
+	switch (sa->sa_family) {
+		default:
+			return NULL;
+		case AF_INET6:
+			return &((const struct sockaddr_in6 *)sa)->sin6_addr;
+			break;
+		case AF_INET:
+			ipv6_addr_set_v4mapped(((const struct sockaddr_in *)sa)->sin_addr.s_addr,
+					addr_mapped);
+			return addr_mapped;
+	}
 }
 
-static int nfs_sockaddr_match_ipaddr6(const struct sockaddr_in6 *sa1,
-				 const struct sockaddr_in6 *sa2)
+static int nfs_sockaddr_match_ipaddr(const struct sockaddr *sa1,
+		const struct sockaddr *sa2)
+{
+	const struct in6_addr *addr1;
+	const struct in6_addr *addr2;
+	struct in6_addr addr1_mapped;
+	struct in6_addr addr2_mapped;
+
+	addr1 = nfs_map_ipv4_addr(sa1, &addr1_mapped);
+	if (likely(addr1 != NULL)) {
+		addr2 = nfs_map_ipv4_addr(sa2, &addr2_mapped);
+		if (likely(addr2 != NULL))
+			return ipv6_addr_equal(addr1, addr2);
+	}
+	return 0;
+}
+#else
+static int nfs_sockaddr_match_ipaddr4(const struct sockaddr_in *sa1,
+				 const struct sockaddr_in *sa2)
 {
-	return ipv6_addr_equal(&sa1->sin6_addr, &sa2->sin6_addr);
+	return sa1->sin_addr.s_addr == sa2->sin_addr.s_addr;
 }
 
 static int nfs_sockaddr_match_ipaddr(const struct sockaddr *sa1,
 				 const struct sockaddr *sa2)
 {
-	switch (sa1->sa_family) {
-	case AF_INET:
-		return nfs_sockaddr_match_ipaddr4((const struct sockaddr_in *)sa1,
-				(const struct sockaddr_in *)sa2);
-	case AF_INET6:
-		return nfs_sockaddr_match_ipaddr6((const struct sockaddr_in6 *)sa1,
-				(const struct sockaddr_in6 *)sa2);
-	}
-	BUG();
+	if (unlikely(sa1->sa_family != AF_INET || sa2->sa_family != AF_INET))
+		return 0;
+	return nfs_sockaddr_match_ipaddr4((const struct sockaddr_in *)sa1,
+			(const struct sockaddr_in *)sa2);
 }
+#endif
 
 /*
  * Find a client by IP address and protocol version
@@ -270,8 +292,6 @@ struct nfs_client *nfs_find_client(const struct sockaddr *addr, u32 nfsversion)
 		if (clp->rpc_ops->version != nfsversion)
 			continue;
 
-		if (addr->sa_family != clap->sa_family)
-			continue;
 		/* Match only the IP address, not the port number */
 		if (!nfs_sockaddr_match_ipaddr(addr, clap))
 			continue;
@@ -305,8 +325,6 @@ struct nfs_client *nfs_find_client_next(struct nfs_client *clp)
 		if (clp->rpc_ops->version != nfsvers)
 			continue;
 
-		if (sap->sa_family != clap->sa_family)
-			continue;
 		/* Match only the IP address, not the port number */
 		if (!nfs_sockaddr_match_ipaddr(sap, clap))
 			continue;
@@ -470,7 +488,7 @@ static void nfs_init_timeout_values(struct rpc_timeout *to, int proto,
 static int nfs_create_rpc_client(struct nfs_client *clp,
 				 const struct rpc_timeout *timeparms,
 				 rpc_authflavor_t flavor,
-				 int flags)
+				 int discrtry, int noresvport)
 {
 	struct rpc_clnt		*clnt = NULL;
 	struct rpc_create_args args = {
@@ -482,9 +500,13 @@ static int nfs_create_rpc_client(struct nfs_client *clp,
 		.program	= &nfs_program,
 		.version	= clp->rpc_ops->version,
 		.authflavor	= flavor,
-		.flags		= flags,
 	};
 
+	if (discrtry)
+		args.flags |= RPC_CLNT_CREATE_DISCRTRY;
+	if (noresvport)
+		args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
+
 	if (!IS_ERR(clp->cl_rpcclient))
 		return 0;
 
@@ -522,6 +544,8 @@ static int nfs_start_lockd(struct nfs_server *server)
 		.protocol	= server->flags & NFS_MOUNT_TCP ?
 						IPPROTO_TCP : IPPROTO_UDP,
 		.nfs_version	= clp->rpc_ops->version,
+		.noresvport	= server->flags & NFS_MOUNT_NORESVPORT ?
+					1 : 0,
 	};
 
 	if (nlm_init.nfs_version > 3)
@@ -623,7 +647,8 @@ static int nfs_init_client(struct nfs_client *clp,
 	 * Create a client RPC handle for doing FSSTAT with UNIX auth only
 	 * - RFC 2623, sec 2.3.2
 	 */
-	error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_UNIX, 0);
+	error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_UNIX,
+				      0, data->flags & NFS_MOUNT_NORESVPORT);
 	if (error < 0)
 		goto error;
 	nfs_mark_client_ready(clp, NFS_CS_READY);
@@ -965,7 +990,8 @@ error:
 static int nfs4_init_client(struct nfs_client *clp,
 		const struct rpc_timeout *timeparms,
 		const char *ip_addr,
-		rpc_authflavor_t authflavour)
+		rpc_authflavor_t authflavour,
+		int flags)
 {
 	int error;
 
@@ -979,7 +1005,7 @@ static int nfs4_init_client(struct nfs_client *clp,
 	clp->rpc_ops = &nfs_v4_clientops;
 
 	error = nfs_create_rpc_client(clp, timeparms, authflavour,
-					RPC_CLNT_CREATE_DISCRTRY);
+				      1, flags & NFS_MOUNT_NORESVPORT);
 	if (error < 0)
 		goto error;
 	memcpy(clp->cl_ipaddr, ip_addr, sizeof(clp->cl_ipaddr));
@@ -1030,7 +1056,8 @@ static int nfs4_set_client(struct nfs_server *server,
 		error = PTR_ERR(clp);
 		goto error;
 	}
-	error = nfs4_init_client(clp, timeparms, ip_addr, authflavour);
+	error = nfs4_init_client(clp, timeparms, ip_addr, authflavour,
+					server->flags);
 	if (error < 0)
 		goto error_put;
 
@@ -1059,6 +1086,10 @@ static int nfs4_init_server(struct nfs_server *server,
 	nfs_init_timeout_values(&timeparms, data->nfs_server.protocol,
 			data->timeo, data->retrans);
 
+	/* Initialise the client representation from the mount data */
+	server->flags = data->flags;
+	server->caps |= NFS_CAP_ATOMIC_OPEN;
+
 	/* Get a client record */
 	error = nfs4_set_client(server,
 			data->nfs_server.hostname,
@@ -1071,10 +1102,6 @@ static int nfs4_init_server(struct nfs_server *server,
 	if (error < 0)
 		goto error;
 
-	/* Initialise the client representation from the mount data */
-	server->flags = data->flags;
-	server->caps |= NFS_CAP_ATOMIC_OPEN;
-
 	if (data->rsize)
 		server->rsize = nfs_block_size(data->rsize, NULL);
 	if (data->wsize)
@@ -1177,6 +1204,10 @@ struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *data,
 	parent_server = NFS_SB(data->sb);
 	parent_client = parent_server->nfs_client;
 
+	/* Initialise the client representation from the parent server */
+	nfs_server_copy_userdata(server, parent_server);
+	server->caps |= NFS_CAP_ATOMIC_OPEN;
+
 	/* Get a client representation.
 	 * Note: NFSv4 always uses TCP, */
 	error = nfs4_set_client(server, data->hostname,
@@ -1189,10 +1220,6 @@ struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *data,
 	if (error < 0)
 		goto error;
 
-	/* Initialise the client representation from the parent server */
-	nfs_server_copy_userdata(server, parent_server);
-	server->caps |= NFS_CAP_ATOMIC_OPEN;
-
 	error = nfs_init_server_rpcclient(server, parent_server->client->cl_timeout, data->authflavor);
 	if (error < 0)
 		goto error;
diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index cc563cfa694..968225a8801 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -43,6 +43,27 @@ static void nfs_free_delegation(struct nfs_delegation *delegation)
 		put_rpccred(cred);
 }
 
+void nfs_mark_delegation_referenced(struct nfs_delegation *delegation)
+{
+	set_bit(NFS_DELEGATION_REFERENCED, &delegation->flags);
+}
+
+int nfs_have_delegation(struct inode *inode, fmode_t flags)
+{
+	struct nfs_delegation *delegation;
+	int ret = 0;
+
+	flags &= FMODE_READ|FMODE_WRITE;
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation != NULL && (delegation->type & flags) == flags) {
+		nfs_mark_delegation_referenced(delegation);
+		ret = 1;
+	}
+	rcu_read_unlock();
+	return ret;
+}
+
 static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
@@ -119,7 +140,7 @@ void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, st
 	delegation->maxsize = res->maxsize;
 	oldcred = delegation->cred;
 	delegation->cred = get_rpccred(cred);
-	delegation->flags &= ~NFS_DELEGATION_NEED_RECLAIM;
+	clear_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
 	NFS_I(inode)->delegation_state = delegation->type;
 	smp_wmb();
 	put_rpccred(oldcred);
@@ -134,19 +155,35 @@ static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *
 	return res;
 }
 
+static struct inode *nfs_delegation_grab_inode(struct nfs_delegation *delegation)
+{
+	struct inode *inode = NULL;
+
+	spin_lock(&delegation->lock);
+	if (delegation->inode != NULL)
+		inode = igrab(delegation->inode);
+	spin_unlock(&delegation->lock);
+	return inode;
+}
+
 static struct nfs_delegation *nfs_detach_delegation_locked(struct nfs_inode *nfsi, const nfs4_stateid *stateid)
 {
 	struct nfs_delegation *delegation = rcu_dereference(nfsi->delegation);
 
 	if (delegation == NULL)
 		goto nomatch;
+	spin_lock(&delegation->lock);
 	if (stateid != NULL && memcmp(delegation->stateid.data, stateid->data,
 				sizeof(delegation->stateid.data)) != 0)
-		goto nomatch;
+		goto nomatch_unlock;
 	list_del_rcu(&delegation->super_list);
+	delegation->inode = NULL;
 	nfsi->delegation_state = 0;
 	rcu_assign_pointer(nfsi->delegation, NULL);
+	spin_unlock(&delegation->lock);
 	return delegation;
+nomatch_unlock:
+	spin_unlock(&delegation->lock);
 nomatch:
 	return NULL;
 }
@@ -172,6 +209,8 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 	delegation->change_attr = nfsi->change_attr;
 	delegation->cred = get_rpccred(cred);
 	delegation->inode = inode;
+	delegation->flags = 1<<NFS_DELEGATION_REFERENCED;
+	spin_lock_init(&delegation->lock);
 
 	spin_lock(&clp->cl_lock);
 	if (rcu_dereference(nfsi->delegation) != NULL) {
@@ -226,22 +265,47 @@ static void nfs_msync_inode(struct inode *inode)
  */
 static int __nfs_inode_return_delegation(struct inode *inode, struct nfs_delegation *delegation)
 {
-	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	nfs_msync_inode(inode);
-	down_read(&clp->cl_sem);
 	/* Guard against new delegated open calls */
 	down_write(&nfsi->rwsem);
 	nfs_delegation_claim_opens(inode, &delegation->stateid);
 	up_write(&nfsi->rwsem);
-	up_read(&clp->cl_sem);
 	nfs_msync_inode(inode);
 
 	return nfs_do_return_delegation(inode, delegation, 1);
 }
 
 /*
+ * Return all delegations that have been marked for return
+ */
+void nfs_client_return_marked_delegations(struct nfs_client *clp)
+{
+	struct nfs_delegation *delegation;
+	struct inode *inode;
+
+restart:
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
+		if (!test_and_clear_bit(NFS_DELEGATION_RETURN, &delegation->flags))
+			continue;
+		inode = nfs_delegation_grab_inode(delegation);
+		if (inode == NULL)
+			continue;
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
+		spin_unlock(&clp->cl_lock);
+		rcu_read_unlock();
+		if (delegation != NULL)
+			__nfs_inode_return_delegation(inode, delegation);
+		iput(inode);
+		goto restart;
+	}
+	rcu_read_unlock();
+}
+
+/*
  * This function returns the delegation without reclaiming opens
  * or protecting against delegation reclaims.
  * It is therefore really only safe to be called from
@@ -279,83 +343,55 @@ int nfs_inode_return_delegation(struct inode *inode)
 	return err;
 }
 
+static void nfs_mark_return_delegation(struct nfs_client *clp, struct nfs_delegation *delegation)
+{
+	set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+	set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
+}
+
 /*
  * Return all delegations associated to a super block
  */
-void nfs_return_all_delegations(struct super_block *sb)
+void nfs_super_return_all_delegations(struct super_block *sb)
 {
 	struct nfs_client *clp = NFS_SB(sb)->nfs_client;
 	struct nfs_delegation *delegation;
-	struct inode *inode;
 
 	if (clp == NULL)
 		return;
-restart:
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		if (delegation->inode->i_sb != sb)
-			continue;
-		inode = igrab(delegation->inode);
-		if (inode == NULL)
-			continue;
-		spin_lock(&clp->cl_lock);
-		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
-		spin_unlock(&clp->cl_lock);
-		rcu_read_unlock();
-		if (delegation != NULL)
-			__nfs_inode_return_delegation(inode, delegation);
-		iput(inode);
-		goto restart;
+		spin_lock(&delegation->lock);
+		if (delegation->inode != NULL && delegation->inode->i_sb == sb)
+			set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+		spin_unlock(&delegation->lock);
 	}
 	rcu_read_unlock();
+	nfs_client_return_marked_delegations(clp);
 }
 
-static int nfs_do_expire_all_delegations(void *ptr)
+static void nfs_client_mark_return_all_delegations(struct nfs_client *clp)
 {
-	struct nfs_client *clp = ptr;
 	struct nfs_delegation *delegation;
-	struct inode *inode;
 
-	allow_signal(SIGKILL);
-restart:
-	if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) != 0)
-		goto out;
-	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0)
-		goto out;
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		inode = igrab(delegation->inode);
-		if (inode == NULL)
-			continue;
-		spin_lock(&clp->cl_lock);
-		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
-		spin_unlock(&clp->cl_lock);
-		rcu_read_unlock();
-		if (delegation)
-			__nfs_inode_return_delegation(inode, delegation);
-		iput(inode);
-		goto restart;
+		set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+		set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
 	}
 	rcu_read_unlock();
-out:
-	nfs_put_client(clp);
-	module_put_and_exit(0);
+}
+
+static void nfs_delegation_run_state_manager(struct nfs_client *clp)
+{
+	if (test_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state))
+		nfs4_schedule_state_manager(clp);
 }
 
 void nfs_expire_all_delegations(struct nfs_client *clp)
 {
-	struct task_struct *task;
-
-	__module_get(THIS_MODULE);
-	atomic_inc(&clp->cl_count);
-	task = kthread_run(nfs_do_expire_all_delegations, clp,
-				"%s-delegreturn",
-				rpc_peeraddr2str(clp->cl_rpcclient,
-							RPC_DISPLAY_ADDR));
-	if (!IS_ERR(task))
-		return;
-	nfs_put_client(clp);
-	module_put(THIS_MODULE);
+	nfs_client_mark_return_all_delegations(clp);
+	nfs_delegation_run_state_manager(clp);
 }
 
 /*
@@ -363,68 +399,29 @@ void nfs_expire_all_delegations(struct nfs_client *clp)
  */
 void nfs_handle_cb_pathdown(struct nfs_client *clp)
 {
-	struct nfs_delegation *delegation;
-	struct inode *inode;
-
 	if (clp == NULL)
 		return;
-restart:
+	nfs_client_mark_return_all_delegations(clp);
+}
+
+static void nfs_client_mark_return_unreferenced_delegations(struct nfs_client *clp)
+{
+	struct nfs_delegation *delegation;
+
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		inode = igrab(delegation->inode);
-		if (inode == NULL)
+		if (test_and_clear_bit(NFS_DELEGATION_REFERENCED, &delegation->flags))
 			continue;
-		spin_lock(&clp->cl_lock);
-		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
-		spin_unlock(&clp->cl_lock);
-		rcu_read_unlock();
-		if (delegation != NULL)
-			__nfs_inode_return_delegation(inode, delegation);
-		iput(inode);
-		goto restart;
+		set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+		set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
 	}
 	rcu_read_unlock();
 }
 
-struct recall_threadargs {
-	struct inode *inode;
-	struct nfs_client *clp;
-	const nfs4_stateid *stateid;
-
-	struct completion started;
-	int result;
-};
-
-static int recall_thread(void *data)
+void nfs_expire_unreferenced_delegations(struct nfs_client *clp)
 {
-	struct recall_threadargs *args = (struct recall_threadargs *)data;
-	struct inode *inode = igrab(args->inode);
-	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
-	struct nfs_inode *nfsi = NFS_I(inode);
-	struct nfs_delegation *delegation;
-
-	daemonize("nfsv4-delegreturn");
-
-	nfs_msync_inode(inode);
-	down_read(&clp->cl_sem);
-	down_write(&nfsi->rwsem);
-	spin_lock(&clp->cl_lock);
-	delegation = nfs_detach_delegation_locked(nfsi, args->stateid);
-	if (delegation != NULL)
-		args->result = 0;
-	else
-		args->result = -ENOENT;
-	spin_unlock(&clp->cl_lock);
-	complete(&args->started);
-	nfs_delegation_claim_opens(inode, args->stateid);
-	up_write(&nfsi->rwsem);
-	up_read(&clp->cl_sem);
-	nfs_msync_inode(inode);
-
-	if (delegation != NULL)
-		nfs_do_return_delegation(inode, delegation, 1);
-	iput(inode);
-	module_put_and_exit(0);
+	nfs_client_mark_return_unreferenced_delegations(clp);
+	nfs_delegation_run_state_manager(clp);
 }
 
 /*
@@ -432,22 +429,20 @@ static int recall_thread(void *data)
  */
 int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid)
 {
-	struct recall_threadargs data = {
-		.inode = inode,
-		.stateid = stateid,
-	};
-	int status;
+	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
+	struct nfs_delegation *delegation;
 
-	init_completion(&data.started);
-	__module_get(THIS_MODULE);
-	status = kernel_thread(recall_thread, &data, CLONE_KERNEL);
-	if (status < 0)
-		goto out_module_put;
-	wait_for_completion(&data.started);
-	return data.result;
-out_module_put:
-	module_put(THIS_MODULE);
-	return status;
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation == NULL || memcmp(delegation->stateid.data, stateid->data,
+				sizeof(delegation->stateid.data)) != 0) {
+		rcu_read_unlock();
+		return -ENOENT;
+	}
+	nfs_mark_return_delegation(clp, delegation);
+	rcu_read_unlock();
+	nfs_delegation_run_state_manager(clp);
+	return 0;
 }
 
 /*
@@ -459,10 +454,14 @@ struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs
 	struct inode *res = NULL;
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		if (nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
+		spin_lock(&delegation->lock);
+		if (delegation->inode != NULL &&
+		    nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
 			res = igrab(delegation->inode);
-			break;
 		}
+		spin_unlock(&delegation->lock);
+		if (res != NULL)
+			break;
 	}
 	rcu_read_unlock();
 	return res;
@@ -476,7 +475,7 @@ void nfs_delegation_mark_reclaim(struct nfs_client *clp)
 	struct nfs_delegation *delegation;
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list)
-		delegation->flags |= NFS_DELEGATION_NEED_RECLAIM;
+		set_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
 	rcu_read_unlock();
 }
 
@@ -486,17 +485,22 @@ void nfs_delegation_mark_reclaim(struct nfs_client *clp)
 void nfs_delegation_reap_unclaimed(struct nfs_client *clp)
 {
 	struct nfs_delegation *delegation;
+	struct inode *inode;
 restart:
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0)
+		if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0)
+			continue;
+		inode = nfs_delegation_grab_inode(delegation);
+		if (inode == NULL)
 			continue;
 		spin_lock(&clp->cl_lock);
-		delegation = nfs_detach_delegation_locked(NFS_I(delegation->inode), NULL);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
 		spin_unlock(&clp->cl_lock);
 		rcu_read_unlock();
 		if (delegation != NULL)
 			nfs_free_delegation(delegation);
+		iput(inode);
 		goto restart;
 	}
 	rcu_read_unlock();
diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h
index f1c5e2a5d88..09f38379517 100644
--- a/fs/nfs/delegation.h
+++ b/fs/nfs/delegation.h
@@ -17,14 +17,20 @@ struct nfs_delegation {
 	struct rpc_cred *cred;
 	struct inode *inode;
 	nfs4_stateid stateid;
-	int type;
-#define NFS_DELEGATION_NEED_RECLAIM 1
-	long flags;
+	fmode_t type;
 	loff_t maxsize;
 	__u64 change_attr;
+	unsigned long flags;
+	spinlock_t lock;
 	struct rcu_head rcu;
 };
 
+enum {
+	NFS_DELEGATION_NEED_RECLAIM = 0,
+	NFS_DELEGATION_RETURN,
+	NFS_DELEGATION_REFERENCED,
+};
+
 int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
 void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
 int nfs_inode_return_delegation(struct inode *inode);
@@ -32,9 +38,11 @@ int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *s
 void nfs_inode_return_delegation_noreclaim(struct inode *inode);
 
 struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs_fh *fhandle);
-void nfs_return_all_delegations(struct super_block *sb);
+void nfs_super_return_all_delegations(struct super_block *sb);
 void nfs_expire_all_delegations(struct nfs_client *clp);
+void nfs_expire_unreferenced_delegations(struct nfs_client *clp);
 void nfs_handle_cb_pathdown(struct nfs_client *clp);
+void nfs_client_return_marked_delegations(struct nfs_client *clp);
 
 void nfs_delegation_mark_reclaim(struct nfs_client *clp);
 void nfs_delegation_reap_unclaimed(struct nfs_client *clp);
@@ -45,22 +53,11 @@ int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state
 int nfs4_lock_delegation_recall(struct nfs4_state *state, struct file_lock *fl);
 int nfs4_copy_delegation_stateid(nfs4_stateid *dst, struct inode *inode);
 
-static inline int nfs_have_delegation(struct inode *inode, int flags)
-{
-	struct nfs_delegation *delegation;
-	int ret = 0;
-
-	flags &= FMODE_READ|FMODE_WRITE;
-	rcu_read_lock();
-	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation != NULL && (delegation->type & flags) == flags)
-		ret = 1;
-	rcu_read_unlock();
-	return ret;
-}
+void nfs_mark_delegation_referenced(struct nfs_delegation *delegation);
+int nfs_have_delegation(struct inode *inode, fmode_t flags);
 
 #else
-static inline int nfs_have_delegation(struct inode *inode, int flags)
+static inline int nfs_have_delegation(struct inode *inode, fmode_t flags)
 {
 	return 0;
 }
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 3e64b98f3a9..e35c8199f82 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -799,6 +799,9 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 		goto out_bad;
 	}
 
+	if (nfs_have_delegation(inode, FMODE_READ))
+		goto out_set_verifier;
+
 	/* Force a full look up iff the parent directory has changed */
 	if (!nfs_is_exclusive_create(dir, nd) && nfs_check_verifier(dir, dentry)) {
 		if (nfs_lookup_verify_inode(inode, nd))
@@ -817,6 +820,7 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 	if ((error = nfs_refresh_inode(inode, &fattr)) != 0)
 		goto out_bad;
 
+out_set_verifier:
 	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
  out_valid:
 	dput(parent);
@@ -973,7 +977,7 @@ struct dentry_operations nfs4_dentry_operations = {
  * Use intent information to determine whether we need to substitute
  * the NFSv4-style stateful OPEN for the LOOKUP call
  */
-static int is_atomic_open(struct inode *dir, struct nameidata *nd)
+static int is_atomic_open(struct nameidata *nd)
 {
 	if (nd == NULL || nfs_lookup_check_intent(nd, LOOKUP_OPEN) == 0)
 		return 0;
@@ -996,7 +1000,7 @@ static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry
 			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);
 
 	/* Check that we are indeed trying to open this file */
-	if (!is_atomic_open(dir, nd))
+	if (!is_atomic_open(nd))
 		goto no_open;
 
 	if (dentry->d_name.len > NFS_SERVER(dir)->namelen) {
@@ -1047,10 +1051,10 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 	struct inode *dir;
 	int openflags, ret = 0;
 
+	if (!is_atomic_open(nd))
+		goto no_open;
 	parent = dget_parent(dentry);
 	dir = parent->d_inode;
-	if (!is_atomic_open(dir, nd))
-		goto no_open;
 	/* We can't create new files in nfs_open_revalidate(), so we
 	 * optimize away revalidation of negative dentries.
 	 */
@@ -1062,11 +1066,11 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 
 	/* NFS only supports OPEN on regular files */
 	if (!S_ISREG(inode->i_mode))
-		goto no_open;
+		goto no_open_dput;
 	openflags = nd->intent.open.flags;
 	/* We cannot do exclusive creation on a positive dentry */
 	if ((openflags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL))
-		goto no_open;
+		goto no_open_dput;
 	/* We can't create new files, or truncate existing ones here */
 	openflags &= ~(O_CREAT|O_TRUNC);
 
@@ -1081,10 +1085,9 @@ out:
 	if (!ret)
 		d_drop(dentry);
 	return ret;
-no_open:
+no_open_dput:
 	dput(parent);
-	if (inode != NULL && nfs_have_delegation(inode, FMODE_READ))
-		return 1;
+no_open:
 	return nfs_lookup_revalidate(dentry, nd);
 }
 #endif /* CONFIG_NFSV4 */
@@ -1794,7 +1797,8 @@ static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, str
 	cache = nfs_access_search_rbtree(inode, cred);
 	if (cache == NULL)
 		goto out;
-	if (!time_in_range(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
+	if (!nfs_have_delegation(inode, FMODE_READ) &&
+	    !time_in_range_open(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
 		goto out_stale;
 	res->jiffies = cache->jiffies;
 	res->cred = cache->cred;
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index d22eb383e1c..0c381686171 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -592,7 +592,7 @@ static void nfs_file_set_open_context(struct file *filp, struct nfs_open_context
 /*
  * Given an inode, search for an open context with the desired characteristics
  */
-struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, int mode)
+struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, fmode_t mode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_open_context *pos, *ctx = NULL;
@@ -712,14 +712,7 @@ int nfs_attribute_timeout(struct inode *inode)
 
 	if (nfs_have_delegation(inode, FMODE_READ))
 		return 0;
-	/*
-	 * Special case: if the attribute timeout is set to 0, then always
-	 * 		 treat the cache as having expired (unless holding
-	 * 		 a delegation).
-	 */
-	if (nfsi->attrtimeo == 0)
-		return 1;
-	return !time_in_range(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
+	return !time_in_range_open(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
 }
 
 /**
@@ -1182,7 +1175,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 		nfsi->attrtimeo_timestamp = now;
 		nfsi->attr_gencount = nfs_inc_attr_generation_counter();
 	} else {
-		if (!time_in_range(now, nfsi->attrtimeo_timestamp, nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) {
+		if (!time_in_range_open(now, nfsi->attrtimeo_timestamp, nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) {
 			if ((nfsi->attrtimeo <<= 1) > NFS_MAXATTRTIMEO(inode))
 				nfsi->attrtimeo = NFS_MAXATTRTIMEO(inode);
 			nfsi->attrtimeo_timestamp = now;
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index d212ee41caf..340ede8f608 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -63,6 +63,20 @@ struct nfs_parsed_mount_data {
 	struct security_mnt_opts lsm_opts;
 };
 
+/* mount_clnt.c */
+struct nfs_mount_request {
+	struct sockaddr		*sap;
+	size_t			salen;
+	char			*hostname;
+	char			*dirpath;
+	u32			version;
+	unsigned short		protocol;
+	struct nfs_fh		*fh;
+	int			noresvport;
+};
+
+extern int nfs_mount(struct nfs_mount_request *info);
+
 /* client.c */
 extern struct rpc_program nfs_program;
 
diff --git a/fs/nfs/mount_clnt.c b/fs/nfs/mount_clnt.c
index 086a6830d78..ca905a5bb1b 100644
--- a/fs/nfs/mount_clnt.c
+++ b/fs/nfs/mount_clnt.c
@@ -29,47 +29,43 @@ struct mnt_fhstatus {
 
 /**
  * nfs_mount - Obtain an NFS file handle for the given host and path
- * @addr: pointer to server's address
- * @len: size of server's address
- * @hostname: name of server host, or NULL
- * @path: pointer to string containing export path to mount
- * @version: mount version to use for this request
- * @protocol: transport protocol to use for thie request
- * @fh: pointer to location to place returned file handle
+ * @info: pointer to mount request arguments
  *
  * Uses default timeout parameters specified by underlying transport.
  */
-int nfs_mount(struct sockaddr *addr, size_t len, char *hostname, char *path,
-	      int version, int protocol, struct nfs_fh *fh)
+int nfs_mount(struct nfs_mount_request *info)
 {
 	struct mnt_fhstatus	result = {
-		.fh		= fh
+		.fh		= info->fh
 	};
 	struct rpc_message msg	= {
-		.rpc_argp	= path,
+		.rpc_argp	= info->dirpath,
 		.rpc_resp	= &result,
 	};
 	struct rpc_create_args args = {
-		.protocol	= protocol,
-		.address	= addr,
-		.addrsize	= len,
-		.servername	= hostname,
+		.protocol	= info->protocol,
+		.address	= info->sap,
+		.addrsize	= info->salen,
+		.servername	= info->hostname,
 		.program	= &mnt_program,
-		.version	= version,
+		.version	= info->version,
 		.authflavor	= RPC_AUTH_UNIX,
-		.flags		= 0,
 	};
 	struct rpc_clnt		*mnt_clnt;
 	int			status;
 
 	dprintk("NFS: sending MNT request for %s:%s\n",
-		(hostname ? hostname : "server"), path);
+		(info->hostname ? info->hostname : "server"),
+			info->dirpath);
+
+	if (info->noresvport)
+		args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
 
 	mnt_clnt = rpc_create(&args);
 	if (IS_ERR(mnt_clnt))
 		goto out_clnt_err;
 
-	if (version == NFS_MNT3_VERSION)
+	if (info->version == NFS_MNT3_VERSION)
 		msg.rpc_proc = &mnt_clnt->cl_procinfo[MOUNTPROC3_MNT];
 	else
 		msg.rpc_proc = &mnt_clnt->cl_procinfo[MNTPROC_MNT];
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index ea790645fda..4e4d3320437 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -38,8 +38,12 @@ struct idmap;
  ((err) != NFSERR_NOFILEHANDLE))
 
 enum nfs4_client_state {
-	NFS4CLNT_STATE_RECOVER  = 0,
+	NFS4CLNT_MANAGER_RUNNING  = 0,
+	NFS4CLNT_CHECK_LEASE,
 	NFS4CLNT_LEASE_EXPIRED,
+	NFS4CLNT_RECLAIM_REBOOT,
+	NFS4CLNT_RECLAIM_NOGRACE,
+	NFS4CLNT_DELEGRETURN,
 };
 
 /*
@@ -90,12 +94,18 @@ struct nfs4_state_owner {
 
 	spinlock_t	     so_lock;
 	atomic_t	     so_count;
+	unsigned long	     so_flags;
 	struct list_head     so_states;
 	struct list_head     so_delegations;
 	struct nfs_seqid_counter so_seqid;
 	struct rpc_sequence  so_sequence;
 };
 
+enum {
+	NFS_OWNER_RECLAIM_REBOOT,
+	NFS_OWNER_RECLAIM_NOGRACE
+};
+
 /*
  * struct nfs4_state maintains the client-side state for a given
  * (state_owner,inode) tuple (OPEN) or state_owner (LOCK).
@@ -128,6 +138,8 @@ enum {
 	NFS_O_RDONLY_STATE,		/* OPEN stateid has read-only state */
 	NFS_O_WRONLY_STATE,		/* OPEN stateid has write-only state */
 	NFS_O_RDWR_STATE,		/* OPEN stateid has read/write state */
+	NFS_STATE_RECLAIM_REBOOT,	/* OPEN stateid server rebooted */
+	NFS_STATE_RECLAIM_NOGRACE,	/* OPEN stateid needs to recover state */
 };
 
 struct nfs4_state {
@@ -149,7 +161,7 @@ struct nfs4_state {
 	unsigned int n_rdonly;		/* Number of read-only references */
 	unsigned int n_wronly;		/* Number of write-only references */
 	unsigned int n_rdwr;		/* Number of read/write references */
-	int state;			/* State on the server (R,W, or RW) */
+	fmode_t state;			/* State on the server (R,W, or RW) */
 	atomic_t count;
 };
 
@@ -157,9 +169,12 @@ struct nfs4_state {
 struct nfs4_exception {
 	long timeout;
 	int retry;
+	struct nfs4_state *state;
 };
 
 struct nfs4_state_recovery_ops {
+	int owner_flag_bit;
+	int state_flag_bit;
 	int (*recover_open)(struct nfs4_state_owner *, struct nfs4_state *);
 	int (*recover_lock)(struct nfs4_state *, struct file_lock *);
 };
@@ -174,7 +189,6 @@ extern ssize_t nfs4_listxattr(struct dentry *, char *, size_t);
 
 
 /* nfs4proc.c */
-extern int nfs4_map_errors(int err);
 extern int nfs4_proc_setclientid(struct nfs_client *, u32, unsigned short, struct rpc_cred *);
 extern int nfs4_proc_setclientid_confirm(struct nfs_client *, struct rpc_cred *);
 extern int nfs4_proc_async_renew(struct nfs_client *, struct rpc_cred *);
@@ -187,7 +201,7 @@ extern int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
 		struct nfs4_fs_locations *fs_locations, struct page *page);
 
 extern struct nfs4_state_recovery_ops nfs4_reboot_recovery_ops;
-extern struct nfs4_state_recovery_ops nfs4_network_partition_recovery_ops;
+extern struct nfs4_state_recovery_ops nfs4_nograce_recovery_ops;
 
 extern const u32 nfs4_fattr_bitmap[2];
 extern const u32 nfs4_statfs_bitmap[2];
@@ -202,16 +216,18 @@ extern void nfs4_kill_renewd(struct nfs_client *);
 extern void nfs4_renew_state(struct work_struct *);
 
 /* nfs4state.c */
-struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp);
+struct rpc_cred *nfs4_get_renew_cred_locked(struct nfs_client *clp);
 
 extern struct nfs4_state_owner * nfs4_get_state_owner(struct nfs_server *, struct rpc_cred *);
 extern void nfs4_put_state_owner(struct nfs4_state_owner *);
 extern struct nfs4_state * nfs4_get_open_state(struct inode *, struct nfs4_state_owner *);
 extern void nfs4_put_open_state(struct nfs4_state *);
-extern void nfs4_close_state(struct path *, struct nfs4_state *, mode_t);
-extern void nfs4_close_sync(struct path *, struct nfs4_state *, mode_t);
-extern void nfs4_state_set_mode_locked(struct nfs4_state *, mode_t);
+extern void nfs4_close_state(struct path *, struct nfs4_state *, fmode_t);
+extern void nfs4_close_sync(struct path *, struct nfs4_state *, fmode_t);
+extern void nfs4_state_set_mode_locked(struct nfs4_state *, fmode_t);
 extern void nfs4_schedule_state_recovery(struct nfs_client *);
+extern void nfs4_schedule_state_manager(struct nfs_client *);
+extern int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_state *state);
 extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp);
 extern int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl);
 extern void nfs4_copy_stateid(nfs4_stateid *, struct nfs4_state *, fl_owner_t);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 83e700a2b0c..8dde84b988d 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -62,14 +62,12 @@
 struct nfs4_opendata;
 static int _nfs4_proc_open(struct nfs4_opendata *data);
 static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
-static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *);
-static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception);
-static int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs_client *clp);
+static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
 static int _nfs4_proc_lookup(struct inode *dir, const struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
 static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
 
 /* Prevent leaks of NFSv4 errors into userland */
-int nfs4_map_errors(int err)
+static int nfs4_map_errors(int err)
 {
 	if (err < -1000) {
 		dprintk("%s could not handle NFSv4 error %d\n",
@@ -195,6 +193,83 @@ static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dent
 	kunmap_atomic(start, KM_USER0);
 }
 
+static int nfs4_wait_bit_killable(void *word)
+{
+	if (fatal_signal_pending(current))
+		return -ERESTARTSYS;
+	schedule();
+	return 0;
+}
+
+static int nfs4_wait_clnt_recover(struct nfs_client *clp)
+{
+	int res;
+
+	might_sleep();
+
+	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
+			nfs4_wait_bit_killable, TASK_KILLABLE);
+	return res;
+}
+
+static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
+{
+	int res = 0;
+
+	might_sleep();
+
+	if (*timeout <= 0)
+		*timeout = NFS4_POLL_RETRY_MIN;
+	if (*timeout > NFS4_POLL_RETRY_MAX)
+		*timeout = NFS4_POLL_RETRY_MAX;
+	schedule_timeout_killable(*timeout);
+	if (fatal_signal_pending(current))
+		res = -ERESTARTSYS;
+	*timeout <<= 1;
+	return res;
+}
+
+/* This is the error handling routine for processes that are allowed
+ * to sleep.
+ */
+static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
+{
+	struct nfs_client *clp = server->nfs_client;
+	struct nfs4_state *state = exception->state;
+	int ret = errorcode;
+
+	exception->retry = 0;
+	switch(errorcode) {
+		case 0:
+			return 0;
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_BAD_STATEID:
+		case -NFS4ERR_OPENMODE:
+			if (state == NULL)
+				break;
+			nfs4_state_mark_reclaim_nograce(clp, state);
+		case -NFS4ERR_STALE_CLIENTID:
+		case -NFS4ERR_STALE_STATEID:
+		case -NFS4ERR_EXPIRED:
+			nfs4_schedule_state_recovery(clp);
+			ret = nfs4_wait_clnt_recover(clp);
+			if (ret == 0)
+				exception->retry = 1;
+			break;
+		case -NFS4ERR_FILE_OPEN:
+		case -NFS4ERR_GRACE:
+		case -NFS4ERR_DELAY:
+			ret = nfs4_delay(server->client, &exception->timeout);
+			if (ret != 0)
+				break;
+		case -NFS4ERR_OLD_STATEID:
+			exception->retry = 1;
+	}
+	/* We failed to handle the error */
+	return nfs4_map_errors(ret);
+}
+
+
 static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
 {
 	struct nfs_client *clp = server->nfs_client;
@@ -248,7 +323,7 @@ static void nfs4_init_opendata_res(struct nfs4_opendata *p)
 }
 
 static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
-		struct nfs4_state_owner *sp, int flags,
+		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
 		const struct iattr *attrs)
 {
 	struct dentry *parent = dget_parent(path->dentry);
@@ -268,7 +343,8 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
 	p->owner = sp;
 	atomic_inc(&sp->so_count);
 	p->o_arg.fh = NFS_FH(dir);
-	p->o_arg.open_flags = flags,
+	p->o_arg.open_flags = flags;
+	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
 	p->o_arg.clientid = server->nfs_client->cl_clientid;
 	p->o_arg.id = sp->so_owner_id.id;
 	p->o_arg.name = &p->path.dentry->d_name;
@@ -324,10 +400,13 @@ static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
 	return ret;
 }
 
-static int can_open_cached(struct nfs4_state *state, int mode)
+static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
 {
 	int ret = 0;
-	switch (mode & (FMODE_READ|FMODE_WRITE|O_EXCL)) {
+
+	if (open_mode & O_EXCL)
+		goto out;
+	switch (mode & (FMODE_READ|FMODE_WRITE)) {
 		case FMODE_READ:
 			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0;
 			break;
@@ -337,21 +416,23 @@ static int can_open_cached(struct nfs4_state *state, int mode)
 		case FMODE_READ|FMODE_WRITE:
 			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0;
 	}
+out:
 	return ret;
 }
 
-static int can_open_delegated(struct nfs_delegation *delegation, mode_t open_flags)
+static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
 {
-	if ((delegation->type & open_flags) != open_flags)
+	if ((delegation->type & fmode) != fmode)
 		return 0;
-	if (delegation->flags & NFS_DELEGATION_NEED_RECLAIM)
+	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
 		return 0;
+	nfs_mark_delegation_referenced(delegation);
 	return 1;
 }
 
-static void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
+static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
 {
-	switch (open_flags) {
+	switch (fmode) {
 		case FMODE_WRITE:
 			state->n_wronly++;
 			break;
@@ -361,15 +442,15 @@ static void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
 		case FMODE_READ|FMODE_WRITE:
 			state->n_rdwr++;
 	}
-	nfs4_state_set_mode_locked(state, state->state | open_flags);
+	nfs4_state_set_mode_locked(state, state->state | fmode);
 }
 
-static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
+static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
 {
 	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
 		memcpy(state->stateid.data, stateid->data, sizeof(state->stateid.data));
 	memcpy(state->open_stateid.data, stateid->data, sizeof(state->open_stateid.data));
-	switch (open_flags) {
+	switch (fmode) {
 		case FMODE_READ:
 			set_bit(NFS_O_RDONLY_STATE, &state->flags);
 			break;
@@ -381,16 +462,15 @@ static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *
 	}
 }
 
-static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
+static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
 {
 	write_seqlock(&state->seqlock);
-	nfs_set_open_stateid_locked(state, stateid, open_flags);
+	nfs_set_open_stateid_locked(state, stateid, fmode);
 	write_sequnlock(&state->seqlock);
 }
 
-static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *deleg_stateid, int open_flags)
+static void __update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, const nfs4_stateid *deleg_stateid, fmode_t fmode)
 {
-	open_flags &= (FMODE_READ|FMODE_WRITE);
 	/*
 	 * Protect the call to nfs4_state_set_mode_locked and
 	 * serialise the stateid update
@@ -401,20 +481,60 @@ static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_sta
 		set_bit(NFS_DELEGATED_STATE, &state->flags);
 	}
 	if (open_stateid != NULL)
-		nfs_set_open_stateid_locked(state, open_stateid, open_flags);
+		nfs_set_open_stateid_locked(state, open_stateid, fmode);
 	write_sequnlock(&state->seqlock);
 	spin_lock(&state->owner->so_lock);
-	update_open_stateflags(state, open_flags);
+	update_open_stateflags(state, fmode);
 	spin_unlock(&state->owner->so_lock);
 }
 
-static void nfs4_return_incompatible_delegation(struct inode *inode, mode_t open_flags)
+static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
+{
+	struct nfs_inode *nfsi = NFS_I(state->inode);
+	struct nfs_delegation *deleg_cur;
+	int ret = 0;
+
+	fmode &= (FMODE_READ|FMODE_WRITE);
+
+	rcu_read_lock();
+	deleg_cur = rcu_dereference(nfsi->delegation);
+	if (deleg_cur == NULL)
+		goto no_delegation;
+
+	spin_lock(&deleg_cur->lock);
+	if (nfsi->delegation != deleg_cur ||
+	    (deleg_cur->type & fmode) != fmode)
+		goto no_delegation_unlock;
+
+	if (delegation == NULL)
+		delegation = &deleg_cur->stateid;
+	else if (memcmp(deleg_cur->stateid.data, delegation->data, NFS4_STATEID_SIZE) != 0)
+		goto no_delegation_unlock;
+
+	nfs_mark_delegation_referenced(deleg_cur);
+	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
+	ret = 1;
+no_delegation_unlock:
+	spin_unlock(&deleg_cur->lock);
+no_delegation:
+	rcu_read_unlock();
+
+	if (!ret && open_stateid != NULL) {
+		__update_open_stateid(state, open_stateid, NULL, fmode);
+		ret = 1;
+	}
+
+	return ret;
+}
+
+
+static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
 {
 	struct nfs_delegation *delegation;
 
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation == NULL || (delegation->type & open_flags) == open_flags) {
+	if (delegation == NULL || (delegation->type & fmode) == fmode) {
 		rcu_read_unlock();
 		return;
 	}
@@ -427,27 +547,28 @@ static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
 	struct nfs4_state *state = opendata->state;
 	struct nfs_inode *nfsi = NFS_I(state->inode);
 	struct nfs_delegation *delegation;
-	int open_mode = opendata->o_arg.open_flags & (FMODE_READ|FMODE_WRITE|O_EXCL);
+	int open_mode = opendata->o_arg.open_flags & O_EXCL;
+	fmode_t fmode = opendata->o_arg.fmode;
 	nfs4_stateid stateid;
 	int ret = -EAGAIN;
 
-	rcu_read_lock();
-	delegation = rcu_dereference(nfsi->delegation);
 	for (;;) {
-		if (can_open_cached(state, open_mode)) {
+		if (can_open_cached(state, fmode, open_mode)) {
 			spin_lock(&state->owner->so_lock);
-			if (can_open_cached(state, open_mode)) {
-				update_open_stateflags(state, open_mode);
+			if (can_open_cached(state, fmode, open_mode)) {
+				update_open_stateflags(state, fmode);
 				spin_unlock(&state->owner->so_lock);
-				rcu_read_unlock();
 				goto out_return_state;
 			}
 			spin_unlock(&state->owner->so_lock);
 		}
-		if (delegation == NULL)
-			break;
-		if (!can_open_delegated(delegation, open_mode))
+		rcu_read_lock();
+		delegation = rcu_dereference(nfsi->delegation);
+		if (delegation == NULL ||
+		    !can_open_delegated(delegation, fmode)) {
+			rcu_read_unlock();
 			break;
+		}
 		/* Save the delegation */
 		memcpy(stateid.data, delegation->stateid.data, sizeof(stateid.data));
 		rcu_read_unlock();
@@ -455,19 +576,11 @@ static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
 		if (ret != 0)
 			goto out;
 		ret = -EAGAIN;
-		rcu_read_lock();
-		delegation = rcu_dereference(nfsi->delegation);
-		/* If no delegation, try a cached open */
-		if (delegation == NULL)
-			continue;
-		/* Is the delegation still valid? */
-		if (memcmp(stateid.data, delegation->stateid.data, sizeof(stateid.data)) != 0)
-			continue;
-		rcu_read_unlock();
-		update_open_stateid(state, NULL, &stateid, open_mode);
-		goto out_return_state;
+
+		/* Try to update the stateid using the delegation */
+		if (update_open_stateid(state, NULL, &stateid, fmode))
+			goto out_return_state;
 	}
-	rcu_read_unlock();
 out:
 	return ERR_PTR(ret);
 out_return_state:
@@ -480,7 +593,6 @@ static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data
 	struct inode *inode;
 	struct nfs4_state *state = NULL;
 	struct nfs_delegation *delegation;
-	nfs4_stateid *deleg_stateid = NULL;
 	int ret;
 
 	if (!data->rpc_done) {
@@ -507,7 +619,7 @@ static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data
 		if (delegation)
 			delegation_flags = delegation->flags;
 		rcu_read_unlock();
-		if (!(delegation_flags & NFS_DELEGATION_NEED_RECLAIM))
+		if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
 			nfs_inode_set_delegation(state->inode,
 					data->owner->so_cred,
 					&data->o_res);
@@ -516,12 +628,9 @@ static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data
 					data->owner->so_cred,
 					&data->o_res);
 	}
-	rcu_read_lock();
-	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation != NULL)
-		deleg_stateid = &delegation->stateid;
-	update_open_stateid(state, &data->o_res.stateid, deleg_stateid, data->o_arg.open_flags);
-	rcu_read_unlock();
+
+	update_open_stateid(state, &data->o_res.stateid, NULL,
+			data->o_arg.fmode);
 	iput(inode);
 out:
 	return state;
@@ -552,7 +661,7 @@ static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context
 {
 	struct nfs4_opendata *opendata;
 
-	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, NULL);
+	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, 0, NULL);
 	if (opendata == NULL)
 		return ERR_PTR(-ENOMEM);
 	opendata->state = state;
@@ -560,12 +669,13 @@ static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context
 	return opendata;
 }
 
-static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openflags, struct nfs4_state **res)
+static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
 {
 	struct nfs4_state *newstate;
 	int ret;
 
-	opendata->o_arg.open_flags = openflags;
+	opendata->o_arg.open_flags = 0;
+	opendata->o_arg.fmode = fmode;
 	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
 	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
 	nfs4_init_opendata_res(opendata);
@@ -575,7 +685,7 @@ static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openf
 	newstate = nfs4_opendata_to_nfs4_state(opendata);
 	if (IS_ERR(newstate))
 		return PTR_ERR(newstate);
-	nfs4_close_state(&opendata->path, newstate, openflags);
+	nfs4_close_state(&opendata->path, newstate, fmode);
 	*res = newstate;
 	return 0;
 }
@@ -631,7 +741,7 @@ static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state
 {
 	struct nfs_delegation *delegation;
 	struct nfs4_opendata *opendata;
-	int delegation_type = 0;
+	fmode_t delegation_type = 0;
 	int status;
 
 	opendata = nfs4_open_recoverdata_alloc(ctx, state);
@@ -641,7 +751,7 @@ static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state
 	opendata->o_arg.fh = NFS_FH(state->inode);
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
-	if (delegation != NULL && (delegation->flags & NFS_DELEGATION_NEED_RECLAIM) != 0)
+	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
 		delegation_type = delegation->type;
 	rcu_read_unlock();
 	opendata->o_arg.u.delegation_type = delegation_type;
@@ -744,7 +854,7 @@ static void nfs4_open_confirm_release(void *calldata)
 		goto out_free;
 	state = nfs4_opendata_to_nfs4_state(data);
 	if (!IS_ERR(state))
-		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
+		nfs4_close_state(&data->path, state, data->o_arg.fmode);
 out_free:
 	nfs4_opendata_put(data);
 }
@@ -808,12 +918,12 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 	if (data->state != NULL) {
 		struct nfs_delegation *delegation;
 
-		if (can_open_cached(data->state, data->o_arg.open_flags & (FMODE_READ|FMODE_WRITE|O_EXCL)))
+		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
 			goto out_no_action;
 		rcu_read_lock();
 		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
 		if (delegation != NULL &&
-		   (delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0) {
+		    test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0) {
 			rcu_read_unlock();
 			goto out_no_action;
 		}
@@ -877,7 +987,7 @@ static void nfs4_open_release(void *calldata)
 		goto out_free;
 	state = nfs4_opendata_to_nfs4_state(data);
 	if (!IS_ERR(state))
-		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
+		nfs4_close_state(&data->path, state, data->o_arg.fmode);
 out_free:
 	nfs4_opendata_put(data);
 }
@@ -955,10 +1065,11 @@ static int nfs4_recover_expired_lease(struct nfs_server *server)
 	int ret;
 
 	for (;;) {
-		ret = nfs4_wait_clnt_recover(server->client, clp);
+		ret = nfs4_wait_clnt_recover(clp);
 		if (ret != 0)
 			return ret;
-		if (!test_and_clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
+		if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) &&
+		    !test_bit(NFS4CLNT_CHECK_LEASE,&clp->cl_state))
 			break;
 		nfs4_schedule_state_recovery(clp);
 	}
@@ -993,8 +1104,9 @@ static inline int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4
 
 	do {
 		err = _nfs4_open_expired(ctx, state);
-		if (err == -NFS4ERR_DELAY)
-			nfs4_handle_exception(server, err, &exception);
+		if (err != -NFS4ERR_DELAY)
+			break;
+		nfs4_handle_exception(server, err, &exception);
 	} while (exception.retry);
 	return err;
 }
@@ -1031,12 +1143,11 @@ static inline void nfs4_exclusive_attrset(struct nfs4_opendata *opendata, struct
 /*
  * Returns a referenced nfs4_state
  */
-static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
+static int _nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
 {
 	struct nfs4_state_owner  *sp;
 	struct nfs4_state     *state = NULL;
 	struct nfs_server       *server = NFS_SERVER(dir);
-	struct nfs_client *clp = server->nfs_client;
 	struct nfs4_opendata *opendata;
 	int status;
 
@@ -1050,12 +1161,11 @@ static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct
 	if (status != 0)
 		goto err_put_state_owner;
 	if (path->dentry->d_inode != NULL)
-		nfs4_return_incompatible_delegation(path->dentry->d_inode, flags & (FMODE_READ|FMODE_WRITE));
-	down_read(&clp->cl_sem);
+		nfs4_return_incompatible_delegation(path->dentry->d_inode, fmode);
 	status = -ENOMEM;
-	opendata = nfs4_opendata_alloc(path, sp, flags, sattr);
+	opendata = nfs4_opendata_alloc(path, sp, fmode, flags, sattr);
 	if (opendata == NULL)
-		goto err_release_rwsem;
+		goto err_put_state_owner;
 
 	if (path->dentry->d_inode != NULL)
 		opendata->state = nfs4_get_open_state(path->dentry->d_inode, sp);
@@ -1073,13 +1183,10 @@ static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct
 		goto err_opendata_put;
 	nfs4_opendata_put(opendata);
 	nfs4_put_state_owner(sp);
-	up_read(&clp->cl_sem);
 	*res = state;
 	return 0;
 err_opendata_put:
 	nfs4_opendata_put(opendata);
-err_release_rwsem:
-	up_read(&clp->cl_sem);
 err_put_state_owner:
 	nfs4_put_state_owner(sp);
 out_err:
@@ -1088,14 +1195,14 @@ out_err:
 }
 
 
-static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, int flags, struct iattr *sattr, struct rpc_cred *cred)
+static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, int flags, struct iattr *sattr, struct rpc_cred *cred)
 {
 	struct nfs4_exception exception = { };
 	struct nfs4_state *res;
 	int status;
 
 	do {
-		status = _nfs4_do_open(dir, path, flags, sattr, cred, &res);
+		status = _nfs4_do_open(dir, path, fmode, flags, sattr, cred, &res);
 		if (status == 0)
 			break;
 		/* NOTE: BAD_SEQID means the server and client disagree about the
@@ -1230,10 +1337,13 @@ static void nfs4_close_done(struct rpc_task *task, void *data)
 			renew_lease(server, calldata->timestamp);
 			break;
 		case -NFS4ERR_STALE_STATEID:
+		case -NFS4ERR_OLD_STATEID:
+		case -NFS4ERR_BAD_STATEID:
 		case -NFS4ERR_EXPIRED:
-			break;
+			if (calldata->arg.fmode == 0)
+				break;
 		default:
-			if (nfs4_async_handle_error(task, server) == -EAGAIN) {
+			if (nfs4_async_handle_error(task, server, state) == -EAGAIN) {
 				rpc_restart_call(task);
 				return;
 			}
@@ -1272,10 +1382,10 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data)
 	nfs_fattr_init(calldata->res.fattr);
 	if (test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0) {
 		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
-		calldata->arg.open_flags = FMODE_READ;
+		calldata->arg.fmode = FMODE_READ;
 	} else if (test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0) {
 		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
-		calldata->arg.open_flags = FMODE_WRITE;
+		calldata->arg.fmode = FMODE_WRITE;
 	}
 	calldata->timestamp = jiffies;
 	rpc_call_start(task);
@@ -1328,6 +1438,7 @@ int nfs4_do_close(struct path *path, struct nfs4_state *state, int wait)
 	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid);
 	if (calldata->arg.seqid == NULL)
 		goto out_free_calldata;
+	calldata->arg.fmode = 0;
 	calldata->arg.bitmask = server->attr_bitmask;
 	calldata->res.fattr = &calldata->fattr;
 	calldata->res.seqid = calldata->arg.seqid;
@@ -1354,13 +1465,13 @@ out:
 	return status;
 }
 
-static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct nfs4_state *state)
+static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct nfs4_state *state, fmode_t fmode)
 {
 	struct file *filp;
 	int ret;
 
 	/* If the open_intent is for execute, we have an extra check to make */
-	if (nd->intent.open.flags & FMODE_EXEC) {
+	if (fmode & FMODE_EXEC) {
 		ret = nfs_may_open(state->inode,
 				state->owner->so_cred,
 				nd->intent.open.flags);
@@ -1376,7 +1487,7 @@ static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct
 	}
 	ret = PTR_ERR(filp);
 out_close:
-	nfs4_close_sync(path, state, nd->intent.open.flags);
+	nfs4_close_sync(path, state, fmode & (FMODE_READ|FMODE_WRITE));
 	return ret;
 }
 
@@ -1392,6 +1503,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
 	struct dentry *res;
+	fmode_t fmode = nd->intent.open.flags & (FMODE_READ | FMODE_WRITE | FMODE_EXEC);
 
 	if (nd->flags & LOOKUP_CREATE) {
 		attr.ia_mode = nd->intent.open.create_mode;
@@ -1409,7 +1521,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 	parent = dentry->d_parent;
 	/* Protect against concurrent sillydeletes */
 	nfs_block_sillyrename(parent);
-	state = nfs4_do_open(dir, &path, nd->intent.open.flags, &attr, cred);
+	state = nfs4_do_open(dir, &path, fmode, nd->intent.open.flags, &attr, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		if (PTR_ERR(state) == -ENOENT) {
@@ -1424,7 +1536,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 		path.dentry = res;
 	nfs_set_verifier(path.dentry, nfs_save_change_attribute(dir));
 	nfs_unblock_sillyrename(parent);
-	nfs4_intent_set_file(nd, &path, state);
+	nfs4_intent_set_file(nd, &path, state, fmode);
 	return res;
 }
 
@@ -1437,11 +1549,12 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 	};
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
+	fmode_t fmode = openflags & (FMODE_READ | FMODE_WRITE);
 
 	cred = rpc_lookup_cred();
 	if (IS_ERR(cred))
 		return PTR_ERR(cred);
-	state = nfs4_do_open(dir, &path, openflags, NULL, cred);
+	state = nfs4_do_open(dir, &path, fmode, openflags, NULL, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		switch (PTR_ERR(state)) {
@@ -1458,10 +1571,10 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 	}
 	if (state->inode == dentry->d_inode) {
 		nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
-		nfs4_intent_set_file(nd, &path, state);
+		nfs4_intent_set_file(nd, &path, state, fmode);
 		return 1;
 	}
-	nfs4_close_sync(&path, state, openflags);
+	nfs4_close_sync(&path, state, fmode);
 out_drop:
 	d_drop(dentry);
 	return 0;
@@ -1887,6 +2000,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 	};
 	struct nfs4_state *state;
 	struct rpc_cred *cred;
+	fmode_t fmode = flags & (FMODE_READ | FMODE_WRITE);
 	int status = 0;
 
 	cred = rpc_lookup_cred();
@@ -1894,7 +2008,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		status = PTR_ERR(cred);
 		goto out;
 	}
-	state = nfs4_do_open(dir, &path, flags, sattr, cred);
+	state = nfs4_do_open(dir, &path, fmode, flags, sattr, cred);
 	d_drop(dentry);
 	if (IS_ERR(state)) {
 		status = PTR_ERR(state);
@@ -1910,9 +2024,9 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		nfs_post_op_update_inode(state->inode, &fattr);
 	}
 	if (status == 0 && (nd->flags & LOOKUP_OPEN) != 0)
-		status = nfs4_intent_set_file(nd, &path, state);
+		status = nfs4_intent_set_file(nd, &path, state, fmode);
 	else
-		nfs4_close_sync(&path, state, flags);
+		nfs4_close_sync(&path, state, fmode);
 out_putcred:
 	put_rpccred(cred);
 out:
@@ -1974,7 +2088,7 @@ static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
 {
 	struct nfs_removeres *res = task->tk_msg.rpc_resp;
 
-	if (nfs4_async_handle_error(task, res->server) == -EAGAIN)
+	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
 		return 0;
 	update_changeattr(dir, &res->cinfo);
 	nfs_post_op_update_inode(dir, &res->dir_attr);
@@ -2402,7 +2516,7 @@ static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data)
 {
 	struct nfs_server *server = NFS_SERVER(data->inode);
 
-	if (nfs4_async_handle_error(task, server) == -EAGAIN) {
+	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
 		rpc_restart_call(task);
 		return -EAGAIN;
 	}
@@ -2423,7 +2537,7 @@ static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data)
 {
 	struct inode *inode = data->inode;
 	
-	if (nfs4_async_handle_error(task, NFS_SERVER(inode)) == -EAGAIN) {
+	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
 		rpc_restart_call(task);
 		return -EAGAIN;
 	}
@@ -2449,7 +2563,7 @@ static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
 {
 	struct inode *inode = data->inode;
 	
-	if (nfs4_async_handle_error(task, NFS_SERVER(inode)) == -EAGAIN) {
+	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
 		rpc_restart_call(task);
 		return -EAGAIN;
 	}
@@ -2742,19 +2856,25 @@ static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen
 }
 
 static int
-nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server)
+nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
 {
 	struct nfs_client *clp = server->nfs_client;
 
 	if (!clp || task->tk_status >= 0)
 		return 0;
 	switch(task->tk_status) {
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_BAD_STATEID:
+		case -NFS4ERR_OPENMODE:
+			if (state == NULL)
+				break;
+			nfs4_state_mark_reclaim_nograce(clp, state);
 		case -NFS4ERR_STALE_CLIENTID:
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_EXPIRED:
 			rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL);
 			nfs4_schedule_state_recovery(clp);
-			if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) == 0)
+			if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0)
 				rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task);
 			task->tk_status = 0;
 			return -EAGAIN;
@@ -2772,79 +2892,6 @@ nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server)
 	return 0;
 }
 
-static int nfs4_wait_bit_killable(void *word)
-{
-	if (fatal_signal_pending(current))
-		return -ERESTARTSYS;
-	schedule();
-	return 0;
-}
-
-static int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs_client *clp)
-{
-	int res;
-
-	might_sleep();
-
-	rwsem_acquire(&clp->cl_sem.dep_map, 0, 0, _RET_IP_);
-
-	res = wait_on_bit(&clp->cl_state, NFS4CLNT_STATE_RECOVER,
-			nfs4_wait_bit_killable, TASK_KILLABLE);
-
-	rwsem_release(&clp->cl_sem.dep_map, 1, _RET_IP_);
-	return res;
-}
-
-static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
-{
-	int res = 0;
-
-	might_sleep();
-
-	if (*timeout <= 0)
-		*timeout = NFS4_POLL_RETRY_MIN;
-	if (*timeout > NFS4_POLL_RETRY_MAX)
-		*timeout = NFS4_POLL_RETRY_MAX;
-	schedule_timeout_killable(*timeout);
-	if (fatal_signal_pending(current))
-		res = -ERESTARTSYS;
-	*timeout <<= 1;
-	return res;
-}
-
-/* This is the error handling routine for processes that are allowed
- * to sleep.
- */
-static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
-{
-	struct nfs_client *clp = server->nfs_client;
-	int ret = errorcode;
-
-	exception->retry = 0;
-	switch(errorcode) {
-		case 0:
-			return 0;
-		case -NFS4ERR_STALE_CLIENTID:
-		case -NFS4ERR_STALE_STATEID:
-		case -NFS4ERR_EXPIRED:
-			nfs4_schedule_state_recovery(clp);
-			ret = nfs4_wait_clnt_recover(server->client, clp);
-			if (ret == 0)
-				exception->retry = 1;
-			break;
-		case -NFS4ERR_FILE_OPEN:
-		case -NFS4ERR_GRACE:
-		case -NFS4ERR_DELAY:
-			ret = nfs4_delay(server->client, &exception->timeout);
-			if (ret != 0)
-				break;
-		case -NFS4ERR_OLD_STATEID:
-			exception->retry = 1;
-	}
-	/* We failed to handle the error */
-	return nfs4_map_errors(ret);
-}
-
 int nfs4_proc_setclientid(struct nfs_client *clp, u32 program, unsigned short port, struct rpc_cred *cred)
 {
 	nfs4_verifier sc_verifier;
@@ -2916,7 +2963,6 @@ static int _nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cre
 		spin_lock(&clp->cl_lock);
 		clp->cl_lease_time = fsinfo.lease_time * HZ;
 		clp->cl_last_renewal = now;
-		clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
 		spin_unlock(&clp->cl_lock);
 	}
 	return status;
@@ -3074,7 +3120,6 @@ static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock
 	struct nfs4_lock_state *lsp;
 	int status;
 
-	down_read(&clp->cl_sem);
 	arg.lock_owner.clientid = clp->cl_clientid;
 	status = nfs4_set_lock_state(state, request);
 	if (status != 0)
@@ -3091,7 +3136,6 @@ static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock
 	}
 	request->fl_ops->fl_release_private(request);
 out:
-	up_read(&clp->cl_sem);
 	return status;
 }
 
@@ -3181,11 +3225,13 @@ static void nfs4_locku_done(struct rpc_task *task, void *data)
 					sizeof(calldata->lsp->ls_stateid.data));
 			renew_lease(calldata->server, calldata->timestamp);
 			break;
+		case -NFS4ERR_BAD_STATEID:
+		case -NFS4ERR_OLD_STATEID:
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_EXPIRED:
 			break;
 		default:
-			if (nfs4_async_handle_error(task, calldata->server) == -EAGAIN)
+			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
 				rpc_restart_call(task);
 	}
 }
@@ -3248,6 +3294,7 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 
 static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
+	struct nfs_inode *nfsi = NFS_I(state->inode);
 	struct nfs_seqid *seqid;
 	struct nfs4_lock_state *lsp;
 	struct rpc_task *task;
@@ -3257,8 +3304,12 @@ static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *
 	status = nfs4_set_lock_state(state, request);
 	/* Unlock _before_ we do the RPC call */
 	request->fl_flags |= FL_EXISTS;
-	if (do_vfs_lock(request->fl_file, request) == -ENOENT)
+	down_read(&nfsi->rwsem);
+	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
+		up_read(&nfsi->rwsem);
 		goto out;
+	}
+	up_read(&nfsi->rwsem);
 	if (status != 0)
 		goto out;
 	/* Is this a delegated lock? */
@@ -3484,7 +3535,7 @@ static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request
 
 static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
-	struct nfs_client *clp = state->owner->so_client;
+	struct nfs_inode *nfsi = NFS_I(state->inode);
 	unsigned char fl_flags = request->fl_flags;
 	int status;
 
@@ -3496,19 +3547,13 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
 	status = do_vfs_lock(request->fl_file, request);
 	if (status < 0)
 		goto out;
-	down_read(&clp->cl_sem);
+	down_read(&nfsi->rwsem);
 	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-		struct nfs_inode *nfsi = NFS_I(state->inode);
 		/* Yes: cache locks! */
-		down_read(&nfsi->rwsem);
 		/* ...but avoid races with delegation recall... */
-		if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-			request->fl_flags = fl_flags & ~FL_SLEEP;
-			status = do_vfs_lock(request->fl_file, request);
-			up_read(&nfsi->rwsem);
-			goto out_unlock;
-		}
-		up_read(&nfsi->rwsem);
+		request->fl_flags = fl_flags & ~FL_SLEEP;
+		status = do_vfs_lock(request->fl_file, request);
+		goto out_unlock;
 	}
 	status = _nfs4_do_setlk(state, cmd, request, 0);
 	if (status != 0)
@@ -3518,7 +3563,7 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
 	if (do_vfs_lock(request->fl_file, request) < 0)
 		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __func__);
 out_unlock:
-	up_read(&clp->cl_sem);
+	up_read(&nfsi->rwsem);
 out:
 	request->fl_flags = fl_flags;
 	return status;
@@ -3664,11 +3709,15 @@ int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
 }
 
 struct nfs4_state_recovery_ops nfs4_reboot_recovery_ops = {
+	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
+	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
 	.recover_open	= nfs4_open_reclaim,
 	.recover_lock	= nfs4_lock_reclaim,
 };
 
-struct nfs4_state_recovery_ops nfs4_network_partition_recovery_ops = {
+struct nfs4_state_recovery_ops nfs4_nograce_recovery_ops = {
+	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
+	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
 	.recover_open	= nfs4_open_expired,
 	.recover_lock	= nfs4_lock_expired,
 };
diff --git a/fs/nfs/nfs4renewd.c b/fs/nfs/nfs4renewd.c
index 3305acbbe2a..f524e932ff7 100644
--- a/fs/nfs/nfs4renewd.c
+++ b/fs/nfs/nfs4renewd.c
@@ -65,7 +65,6 @@ nfs4_renew_state(struct work_struct *work)
 	long lease, timeout;
 	unsigned long last, now;
 
-	down_read(&clp->cl_sem);
 	dprintk("%s: start\n", __func__);
 	/* Are there any active superblocks? */
 	if (list_empty(&clp->cl_superblocks))
@@ -77,17 +76,19 @@ nfs4_renew_state(struct work_struct *work)
 	timeout = (2 * lease) / 3 + (long)last - (long)now;
 	/* Are we close to a lease timeout? */
 	if (time_after(now, last + lease/3)) {
-		cred = nfs4_get_renew_cred(clp);
+		cred = nfs4_get_renew_cred_locked(clp);
+		spin_unlock(&clp->cl_lock);
 		if (cred == NULL) {
-			set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
-			spin_unlock(&clp->cl_lock);
+			if (list_empty(&clp->cl_delegations)) {
+				set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+				goto out;
+			}
 			nfs_expire_all_delegations(clp);
-			goto out;
+		} else {
+			/* Queue an asynchronous RENEW. */
+			nfs4_proc_async_renew(clp, cred);
+			put_rpccred(cred);
 		}
-		spin_unlock(&clp->cl_lock);
-		/* Queue an asynchronous RENEW. */
-		nfs4_proc_async_renew(clp, cred);
-		put_rpccred(cred);
 		timeout = (2 * lease) / 3;
 		spin_lock(&clp->cl_lock);
 	} else
@@ -100,12 +101,11 @@ nfs4_renew_state(struct work_struct *work)
 	cancel_delayed_work(&clp->cl_renewd);
 	schedule_delayed_work(&clp->cl_renewd, timeout);
 	spin_unlock(&clp->cl_lock);
+	nfs_expire_unreferenced_delegations(clp);
 out:
-	up_read(&clp->cl_sem);
 	dprintk("%s: done\n", __func__);
 }
 
-/* Must be called with clp->cl_sem locked for writes */
 void
 nfs4_schedule_state_renewal(struct nfs_client *clp)
 {
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 401ef8b28f9..2022fe47966 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -71,14 +71,12 @@ static int nfs4_init_client(struct nfs_client *clp, struct rpc_cred *cred)
 	return status;
 }
 
-static struct rpc_cred *nfs4_get_machine_cred(struct nfs_client *clp)
+static struct rpc_cred *nfs4_get_machine_cred_locked(struct nfs_client *clp)
 {
 	struct rpc_cred *cred = NULL;
 
-	spin_lock(&clp->cl_lock);
 	if (clp->cl_machine_cred != NULL)
 		cred = get_rpccred(clp->cl_machine_cred);
-	spin_unlock(&clp->cl_lock);
 	return cred;
 }
 
@@ -94,7 +92,7 @@ static void nfs4_clear_machine_cred(struct nfs_client *clp)
 		put_rpccred(cred);
 }
 
-struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
+struct rpc_cred *nfs4_get_renew_cred_locked(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
@@ -110,13 +108,24 @@ struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
 	return cred;
 }
 
+static struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
+{
+	struct rpc_cred *cred;
+
+	spin_lock(&clp->cl_lock);
+	cred = nfs4_get_renew_cred_locked(clp);
+	spin_unlock(&clp->cl_lock);
+	return cred;
+}
+
 static struct rpc_cred *nfs4_get_setclientid_cred(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
 	struct rpc_cred *cred;
 
-	cred = nfs4_get_machine_cred(clp);
+	spin_lock(&clp->cl_lock);
+	cred = nfs4_get_machine_cred_locked(clp);
 	if (cred != NULL)
 		goto out;
 	pos = rb_first(&clp->cl_state_owners);
@@ -125,6 +134,7 @@ static struct rpc_cred *nfs4_get_setclientid_cred(struct nfs_client *clp)
 		cred = get_rpccred(sp->so_cred);
 	}
 out:
+	spin_unlock(&clp->cl_lock);
 	return cred;
 }
 
@@ -295,10 +305,6 @@ nfs4_drop_state_owner(struct nfs4_state_owner *sp)
 	}
 }
 
-/*
- * Note: must be called with clp->cl_sem held in order to prevent races
- *       with reboot recovery!
- */
 struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct rpc_cred *cred)
 {
 	struct nfs_client *clp = server->nfs_client;
@@ -327,10 +333,6 @@ struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct
 	return sp;
 }
 
-/*
- * Must be called with clp->cl_sem held in order to avoid races
- * with state recovery...
- */
 void nfs4_put_state_owner(struct nfs4_state_owner *sp)
 {
 	struct nfs_client *clp = sp->so_client;
@@ -361,18 +363,18 @@ nfs4_alloc_open_state(void)
 }
 
 void
-nfs4_state_set_mode_locked(struct nfs4_state *state, mode_t mode)
+nfs4_state_set_mode_locked(struct nfs4_state *state, fmode_t fmode)
 {
-	if (state->state == mode)
+	if (state->state == fmode)
 		return;
 	/* NB! List reordering - see the reclaim code for why.  */
-	if ((mode & FMODE_WRITE) != (state->state & FMODE_WRITE)) {
-		if (mode & FMODE_WRITE)
+	if ((fmode & FMODE_WRITE) != (state->state & FMODE_WRITE)) {
+		if (fmode & FMODE_WRITE)
 			list_move(&state->open_states, &state->owner->so_states);
 		else
 			list_move_tail(&state->open_states, &state->owner->so_states);
 	}
-	state->state = mode;
+	state->state = fmode;
 }
 
 static struct nfs4_state *
@@ -432,10 +434,6 @@ out:
 	return state;
 }
 
-/*
- * Beware! Caller must be holding exactly one
- * reference to clp->cl_sem!
- */
 void nfs4_put_open_state(struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
@@ -456,16 +454,16 @@ void nfs4_put_open_state(struct nfs4_state *state)
 /*
  * Close the current file.
  */
-static void __nfs4_close(struct path *path, struct nfs4_state *state, mode_t mode, int wait)
+static void __nfs4_close(struct path *path, struct nfs4_state *state, fmode_t fmode, int wait)
 {
 	struct nfs4_state_owner *owner = state->owner;
 	int call_close = 0;
-	int newstate;
+	fmode_t newstate;
 
 	atomic_inc(&owner->so_count);
 	/* Protect against nfs4_find_state() */
 	spin_lock(&owner->so_lock);
-	switch (mode & (FMODE_READ | FMODE_WRITE)) {
+	switch (fmode & (FMODE_READ | FMODE_WRITE)) {
 		case FMODE_READ:
 			state->n_rdonly--;
 			break;
@@ -500,14 +498,14 @@ static void __nfs4_close(struct path *path, struct nfs4_state *state, mode_t mod
 		nfs4_do_close(path, state, wait);
 }
 
-void nfs4_close_state(struct path *path, struct nfs4_state *state, mode_t mode)
+void nfs4_close_state(struct path *path, struct nfs4_state *state, fmode_t fmode)
 {
-	__nfs4_close(path, state, mode, 0);
+	__nfs4_close(path, state, fmode, 0);
 }
 
-void nfs4_close_sync(struct path *path, struct nfs4_state *state, mode_t mode)
+void nfs4_close_sync(struct path *path, struct nfs4_state *state, fmode_t fmode)
 {
-	__nfs4_close(path, state, mode, 1);
+	__nfs4_close(path, state, fmode, 1);
 }
 
 /*
@@ -568,7 +566,6 @@ static void nfs4_free_lock_state(struct nfs4_lock_state *lsp)
  * Return a compatible lock_state. If no initialized lock_state structure
  * exists, return an uninitialized one.
  *
- * The caller must be holding clp->cl_sem
  */
 static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner)
 {
@@ -770,32 +767,34 @@ unlock:
 	return status;
 }
 
-static int reclaimer(void *);
+static int nfs4_run_state_manager(void *);
 
-static inline void nfs4_clear_recover_bit(struct nfs_client *clp)
+static void nfs4_clear_state_manager_bit(struct nfs_client *clp)
 {
 	smp_mb__before_clear_bit();
-	clear_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state);
+	clear_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state);
 	smp_mb__after_clear_bit();
-	wake_up_bit(&clp->cl_state, NFS4CLNT_STATE_RECOVER);
+	wake_up_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING);
 	rpc_wake_up(&clp->cl_rpcwaitq);
 }
 
 /*
- * State recovery routine
+ * Schedule the nfs_client asynchronous state management routine
  */
-static void nfs4_recover_state(struct nfs_client *clp)
+void nfs4_schedule_state_manager(struct nfs_client *clp)
 {
 	struct task_struct *task;
 
+	if (test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) != 0)
+		return;
 	__module_get(THIS_MODULE);
 	atomic_inc(&clp->cl_count);
-	task = kthread_run(reclaimer, clp, "%s-reclaim",
+	task = kthread_run(nfs4_run_state_manager, clp, "%s-manager",
 				rpc_peeraddr2str(clp->cl_rpcclient,
 							RPC_DISPLAY_ADDR));
 	if (!IS_ERR(task))
 		return;
-	nfs4_clear_recover_bit(clp);
+	nfs4_clear_state_manager_bit(clp);
 	nfs_put_client(clp);
 	module_put(THIS_MODULE);
 }
@@ -807,16 +806,42 @@ void nfs4_schedule_state_recovery(struct nfs_client *clp)
 {
 	if (!clp)
 		return;
-	if (test_and_set_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) == 0)
-		nfs4_recover_state(clp);
+	if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
+		set_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state);
+	nfs4_schedule_state_manager(clp);
 }
 
-static int nfs4_reclaim_locks(struct nfs4_state_recovery_ops *ops, struct nfs4_state *state)
+static int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_state *state)
+{
+
+	set_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags);
+	/* Don't recover state that expired before the reboot */
+	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags)) {
+		clear_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags);
+		return 0;
+	}
+	set_bit(NFS_OWNER_RECLAIM_REBOOT, &state->owner->so_flags);
+	set_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state);
+	return 1;
+}
+
+int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_state *state)
+{
+	set_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags);
+	clear_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags);
+	set_bit(NFS_OWNER_RECLAIM_NOGRACE, &state->owner->so_flags);
+	set_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state);
+	return 1;
+}
+
+static int nfs4_reclaim_locks(struct nfs4_state *state, const struct nfs4_state_recovery_ops *ops)
 {
 	struct inode *inode = state->inode;
+	struct nfs_inode *nfsi = NFS_I(inode);
 	struct file_lock *fl;
 	int status = 0;
 
+	down_write(&nfsi->rwsem);
 	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
 		if (!(fl->fl_flags & (FL_POSIX|FL_FLOCK)))
 			continue;
@@ -839,12 +864,14 @@ static int nfs4_reclaim_locks(struct nfs4_state_recovery_ops *ops, struct nfs4_s
 				goto out_err;
 		}
 	}
+	up_write(&nfsi->rwsem);
 	return 0;
 out_err:
+	up_write(&nfsi->rwsem);
 	return status;
 }
 
-static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct nfs4_state_owner *sp)
+static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs4_state_recovery_ops *ops)
 {
 	struct nfs4_state *state;
 	struct nfs4_lock_state *lock;
@@ -858,28 +885,34 @@ static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct n
 	 * recovering after a network partition or a reboot from a
 	 * server that doesn't support a grace period.
 	 */
+restart:
+	spin_lock(&sp->so_lock);
 	list_for_each_entry(state, &sp->so_states, open_states) {
+		if (!test_and_clear_bit(ops->state_flag_bit, &state->flags))
+			continue;
 		if (state->state == 0)
 			continue;
+		atomic_inc(&state->count);
+		spin_unlock(&sp->so_lock);
 		status = ops->recover_open(sp, state);
 		if (status >= 0) {
-			status = nfs4_reclaim_locks(ops, state);
-			if (status < 0)
-				goto out_err;
-			list_for_each_entry(lock, &state->lock_states, ls_locks) {
-				if (!(lock->ls_flags & NFS_LOCK_INITIALIZED))
-					printk("%s: Lock reclaim failed!\n",
+			status = nfs4_reclaim_locks(state, ops);
+			if (status >= 0) {
+				list_for_each_entry(lock, &state->lock_states, ls_locks) {
+					if (!(lock->ls_flags & NFS_LOCK_INITIALIZED))
+						printk("%s: Lock reclaim failed!\n",
 							__func__);
+				}
+				nfs4_put_open_state(state);
+				goto restart;
 			}
-			continue;
 		}
 		switch (status) {
 			default:
 				printk(KERN_ERR "%s: unhandled error %d. Zeroing state\n",
 						__func__, status);
 			case -ENOENT:
-			case -NFS4ERR_RECLAIM_BAD:
-			case -NFS4ERR_RECLAIM_CONFLICT:
+			case -ESTALE:
 				/*
 				 * Open state on this file cannot be recovered
 				 * All we can do is revert to using the zero stateid.
@@ -889,84 +922,176 @@ static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct n
 				/* Mark the file as being 'closed' */
 				state->state = 0;
 				break;
+			case -NFS4ERR_RECLAIM_BAD:
+			case -NFS4ERR_RECLAIM_CONFLICT:
+				nfs4_state_mark_reclaim_nograce(sp->so_client, state);
+				break;
 			case -NFS4ERR_EXPIRED:
 			case -NFS4ERR_NO_GRACE:
+				nfs4_state_mark_reclaim_nograce(sp->so_client, state);
 			case -NFS4ERR_STALE_CLIENTID:
 				goto out_err;
 		}
+		nfs4_put_open_state(state);
+		goto restart;
 	}
+	spin_unlock(&sp->so_lock);
 	return 0;
 out_err:
+	nfs4_put_open_state(state);
 	return status;
 }
 
-static void nfs4_state_mark_reclaim(struct nfs_client *clp)
+static void nfs4_clear_open_state(struct nfs4_state *state)
+{
+	struct nfs4_lock_state *lock;
+
+	clear_bit(NFS_DELEGATED_STATE, &state->flags);
+	clear_bit(NFS_O_RDONLY_STATE, &state->flags);
+	clear_bit(NFS_O_WRONLY_STATE, &state->flags);
+	clear_bit(NFS_O_RDWR_STATE, &state->flags);
+	list_for_each_entry(lock, &state->lock_states, ls_locks) {
+		lock->ls_seqid.flags = 0;
+		lock->ls_flags &= ~NFS_LOCK_INITIALIZED;
+	}
+}
+
+static void nfs4_state_mark_reclaim_helper(struct nfs_client *clp, int (*mark_reclaim)(struct nfs_client *clp, struct nfs4_state *state))
 {
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
 	struct nfs4_state *state;
-	struct nfs4_lock_state *lock;
 
 	/* Reset all sequence ids to zero */
 	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
 		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
-		sp->so_seqid.counter = 0;
 		sp->so_seqid.flags = 0;
 		spin_lock(&sp->so_lock);
 		list_for_each_entry(state, &sp->so_states, open_states) {
-			clear_bit(NFS_DELEGATED_STATE, &state->flags);
-			clear_bit(NFS_O_RDONLY_STATE, &state->flags);
-			clear_bit(NFS_O_WRONLY_STATE, &state->flags);
-			clear_bit(NFS_O_RDWR_STATE, &state->flags);
-			list_for_each_entry(lock, &state->lock_states, ls_locks) {
-				lock->ls_seqid.counter = 0;
-				lock->ls_seqid.flags = 0;
-				lock->ls_flags &= ~NFS_LOCK_INITIALIZED;
-			}
+			if (mark_reclaim(clp, state))
+				nfs4_clear_open_state(state);
 		}
 		spin_unlock(&sp->so_lock);
 	}
 }
 
-static int reclaimer(void *ptr)
+static void nfs4_state_start_reclaim_reboot(struct nfs_client *clp)
+{
+	/* Mark all delegations for reclaim */
+	nfs_delegation_mark_reclaim(clp);
+	nfs4_state_mark_reclaim_helper(clp, nfs4_state_mark_reclaim_reboot);
+}
+
+static void nfs4_state_end_reclaim_reboot(struct nfs_client *clp)
 {
-	struct nfs_client *clp = ptr;
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
-	struct nfs4_state_recovery_ops *ops;
-	struct rpc_cred *cred;
+	struct nfs4_state *state;
+
+	if (!test_and_clear_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state))
+		return;
+
+	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
+		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
+		spin_lock(&sp->so_lock);
+		list_for_each_entry(state, &sp->so_states, open_states) {
+			if (!test_and_clear_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags))
+				continue;
+			nfs4_state_mark_reclaim_nograce(clp, state);
+		}
+		spin_unlock(&sp->so_lock);
+	}
+
+	nfs_delegation_reap_unclaimed(clp);
+}
+
+static void nfs_delegation_clear_all(struct nfs_client *clp)
+{
+	nfs_delegation_mark_reclaim(clp);
+	nfs_delegation_reap_unclaimed(clp);
+}
+
+static void nfs4_state_start_reclaim_nograce(struct nfs_client *clp)
+{
+	nfs_delegation_clear_all(clp);
+	nfs4_state_mark_reclaim_helper(clp, nfs4_state_mark_reclaim_nograce);
+}
+
+static void nfs4_state_end_reclaim_nograce(struct nfs_client *clp)
+{
+	clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state);
+}
+
+static void nfs4_recovery_handle_error(struct nfs_client *clp, int error)
+{
+	switch (error) {
+		case -NFS4ERR_CB_PATH_DOWN:
+			nfs_handle_cb_pathdown(clp);
+			break;
+		case -NFS4ERR_STALE_CLIENTID:
+		case -NFS4ERR_LEASE_MOVED:
+			set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+			nfs4_state_start_reclaim_reboot(clp);
+			break;
+		case -NFS4ERR_EXPIRED:
+			set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+			nfs4_state_start_reclaim_nograce(clp);
+	}
+}
+
+static int nfs4_do_reclaim(struct nfs_client *clp, const struct nfs4_state_recovery_ops *ops)
+{
+	struct rb_node *pos;
 	int status = 0;
 
-	allow_signal(SIGKILL);
+restart:
+	spin_lock(&clp->cl_lock);
+	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
+		struct nfs4_state_owner *sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
+		if (!test_and_clear_bit(ops->owner_flag_bit, &sp->so_flags))
+			continue;
+		atomic_inc(&sp->so_count);
+		spin_unlock(&clp->cl_lock);
+		status = nfs4_reclaim_open_state(sp, ops);
+		if (status < 0) {
+			set_bit(ops->owner_flag_bit, &sp->so_flags);
+			nfs4_put_state_owner(sp);
+			nfs4_recovery_handle_error(clp, status);
+			return status;
+		}
+		nfs4_put_state_owner(sp);
+		goto restart;
+	}
+	spin_unlock(&clp->cl_lock);
+	return status;
+}
 
-	/* Ensure exclusive access to NFSv4 state */
-	down_write(&clp->cl_sem);
-	/* Are there any NFS mounts out there? */
-	if (list_empty(&clp->cl_superblocks))
-		goto out;
-restart_loop:
-	ops = &nfs4_network_partition_recovery_ops;
-	/* Are there any open files on this volume? */
+static int nfs4_check_lease(struct nfs_client *clp)
+{
+	struct rpc_cred *cred;
+	int status = -NFS4ERR_EXPIRED;
+
+	/* Is the client already known to have an expired lease? */
+	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
+		return 0;
 	cred = nfs4_get_renew_cred(clp);
-	if (cred != NULL) {
-		/* Yes there are: try to renew the old lease */
-		status = nfs4_proc_renew(clp, cred);
-		put_rpccred(cred);
-		switch (status) {
-			case 0:
-			case -NFS4ERR_CB_PATH_DOWN:
-				goto out;
-			case -NFS4ERR_STALE_CLIENTID:
-			case -NFS4ERR_LEASE_MOVED:
-				ops = &nfs4_reboot_recovery_ops;
-		}
-	} else {
-		/* "reboot" to ensure we clear all state on the server */
-		clp->cl_boot_time = CURRENT_TIME;
+	if (cred == NULL) {
+		cred = nfs4_get_setclientid_cred(clp);
+		if (cred == NULL)
+			goto out;
 	}
-	/* We're going to have to re-establish a clientid */
-	nfs4_state_mark_reclaim(clp);
-	status = -ENOENT;
+	status = nfs4_proc_renew(clp, cred);
+	put_rpccred(cred);
+out:
+	nfs4_recovery_handle_error(clp, status);
+	return status;
+}
+
+static int nfs4_reclaim_lease(struct nfs_client *clp)
+{
+	struct rpc_cred *cred;
+	int status = -ENOENT;
+
 	cred = nfs4_get_setclientid_cred(clp);
 	if (cred != NULL) {
 		status = nfs4_init_client(clp, cred);
@@ -974,42 +1099,90 @@ restart_loop:
 		/* Handle case where the user hasn't set up machine creds */
 		if (status == -EACCES && cred == clp->cl_machine_cred) {
 			nfs4_clear_machine_cred(clp);
-			goto restart_loop;
+			status = -EAGAIN;
 		}
 	}
-	if (status)
-		goto out_error;
-	/* Mark all delegations for reclaim */
-	nfs_delegation_mark_reclaim(clp);
-	/* Note: list is protected by exclusive lock on cl->cl_sem */
-	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
-		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
-		status = nfs4_reclaim_open_state(ops, sp);
-		if (status < 0) {
-			if (status == -NFS4ERR_NO_GRACE) {
-				ops = &nfs4_network_partition_recovery_ops;
-				status = nfs4_reclaim_open_state(ops, sp);
+	return status;
+}
+
+static void nfs4_state_manager(struct nfs_client *clp)
+{
+	int status = 0;
+
+	/* Ensure exclusive access to NFSv4 state */
+	for(;;) {
+		if (test_and_clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state)) {
+			/* We're going to have to re-establish a clientid */
+			status = nfs4_reclaim_lease(clp);
+			if (status) {
+				set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+				if (status == -EAGAIN)
+					continue;
+				goto out_error;
 			}
+			clear_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state);
+		}
+
+		if (test_and_clear_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state)) {
+			status = nfs4_check_lease(clp);
+			if (status != 0)
+				continue;
+		}
+
+		/* First recover reboot state... */
+		if (test_and_clear_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state)) {
+			status = nfs4_do_reclaim(clp, &nfs4_reboot_recovery_ops);
 			if (status == -NFS4ERR_STALE_CLIENTID)
-				goto restart_loop;
-			if (status == -NFS4ERR_EXPIRED)
-				goto restart_loop;
+				continue;
+			nfs4_state_end_reclaim_reboot(clp);
+			continue;
+		}
+
+		/* Now recover expired state... */
+		if (test_and_clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state)) {
+			status = nfs4_do_reclaim(clp, &nfs4_nograce_recovery_ops);
+			if (status < 0) {
+				set_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state);
+				if (status == -NFS4ERR_STALE_CLIENTID)
+					continue;
+				if (status == -NFS4ERR_EXPIRED)
+					continue;
+				goto out_error;
+			} else
+				nfs4_state_end_reclaim_nograce(clp);
+			continue;
 		}
+
+		if (test_and_clear_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state)) {
+			nfs_client_return_marked_delegations(clp);
+			continue;
+		}
+
+		nfs4_clear_state_manager_bit(clp);
+		/* Did we race with an attempt to give us more work? */
+		if (clp->cl_state == 0)
+			break;
+		if (test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) != 0)
+			break;
 	}
-	nfs_delegation_reap_unclaimed(clp);
-out:
-	up_write(&clp->cl_sem);
-	if (status == -NFS4ERR_CB_PATH_DOWN)
-		nfs_handle_cb_pathdown(clp);
-	nfs4_clear_recover_bit(clp);
+	return;
+out_error:
+	printk(KERN_WARNING "Error: state manager failed on NFSv4 server %s"
+			" with error %d\n", clp->cl_hostname, -status);
+	if (test_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state))
+		nfs4_state_end_reclaim_reboot(clp);
+	nfs4_clear_state_manager_bit(clp);
+}
+
+static int nfs4_run_state_manager(void *ptr)
+{
+	struct nfs_client *clp = ptr;
+
+	allow_signal(SIGKILL);
+	nfs4_state_manager(clp);
 	nfs_put_client(clp);
 	module_put_and_exit(0);
 	return 0;
-out_error:
-	printk(KERN_WARNING "Error: state recovery failed on NFSv4 server %s"
-			" with error %d\n", clp->cl_hostname, -status);
-	set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
-	goto out;
 }
 
 /*
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index b916297d233..d1e4c8f8a0a 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -8,7 +8,7 @@
  *
  *  Kendrick Smith <kmsmith@umich.edu>
  *  Andy Adamson   <andros@umich.edu>
- * 
+ *
  *  Redistribution and use in source and binary forms, with or without
  *  modification, are permitted provided that the following conditions
  *  are met:
@@ -67,7 +67,7 @@ static int nfs4_stat_to_errno(int);
 #define NFS4_MAXTAGLEN		0
 #endif
 
-/* lock,open owner id: 
+/* lock,open owner id:
  * we currently use size 2 (u64) out of (NFS4_OPAQUE_LIMIT  >> 2)
  */
 #define open_owner_id_maxsz	(1 + 4)
@@ -541,6 +541,7 @@ static struct {
 struct compound_hdr {
 	int32_t		status;
 	uint32_t	nops;
+	__be32 *	nops_p;
 	uint32_t	taglen;
 	char *		tag;
 };
@@ -578,7 +579,7 @@ static void encode_string(struct xdr_stream *xdr, unsigned int len, const char *
 	xdr_encode_opaque(p, str, len);
 }
 
-static int encode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
+static void encode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -588,8 +589,13 @@ static int encode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
 	WRITE32(hdr->taglen);
 	WRITEMEM(hdr->tag, hdr->taglen);
 	WRITE32(NFS4_MINOR_VERSION);
+	hdr->nops_p = p;
 	WRITE32(hdr->nops);
-	return 0;
+}
+
+static void encode_nops(struct compound_hdr *hdr)
+{
+	*hdr->nops_p = htonl(hdr->nops);
 }
 
 static void encode_nfs4_verifier(struct xdr_stream *xdr, const nfs4_verifier *verf)
@@ -601,7 +607,7 @@ static void encode_nfs4_verifier(struct xdr_stream *xdr, const nfs4_verifier *ve
 	xdr_encode_opaque_fixed(p, verf->data, NFS4_VERIFIER_SIZE);
 }
 
-static int encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const struct nfs_server *server)
+static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const struct nfs_server *server)
 {
 	char owner_name[IDMAP_NAMESZ];
 	char owner_group[IDMAP_NAMESZ];
@@ -612,7 +618,6 @@ static int encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const s
 	int len;
 	uint32_t bmval0 = 0;
 	uint32_t bmval1 = 0;
-	int status;
 
 	/*
 	 * We reserve enough space to write the entire attribute buffer at once.
@@ -709,7 +714,7 @@ static int encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const s
 		bmval1 |= FATTR4_WORD1_TIME_MODIFY_SET;
 		WRITE32(NFS4_SET_TO_SERVER_TIME);
 	}
-	
+
 	/*
 	 * Now we backfill the bitmap and the attribute buffer length.
 	 */
@@ -723,23 +728,20 @@ static int encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const s
 	*q++ = htonl(bmval1);
 	*q++ = htonl(len);
 
-	status = 0;
 /* out: */
-	return status;
 }
 
-static int encode_access(struct xdr_stream *xdr, u32 access)
+static void encode_access(struct xdr_stream *xdr, u32 access, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(8);
 	WRITE32(OP_ACCESS);
 	WRITE32(access);
-	
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_close(struct xdr_stream *xdr, const struct nfs_closeargs *arg)
+static void encode_close(struct xdr_stream *xdr, const struct nfs_closeargs *arg, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -747,26 +749,24 @@ static int encode_close(struct xdr_stream *xdr, const struct nfs_closeargs *arg)
 	WRITE32(OP_CLOSE);
 	WRITE32(arg->seqid->sequence->counter);
 	WRITEMEM(arg->stateid->data, NFS4_STATEID_SIZE);
-	
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_commit(struct xdr_stream *xdr, const struct nfs_writeargs *args)
+static void encode_commit(struct xdr_stream *xdr, const struct nfs_writeargs *args, struct compound_hdr *hdr)
 {
 	__be32 *p;
-        
-        RESERVE_SPACE(16);
-        WRITE32(OP_COMMIT);
-        WRITE64(args->offset);
-        WRITE32(args->count);
 
-        return 0;
+	RESERVE_SPACE(16);
+	WRITE32(OP_COMMIT);
+	WRITE64(args->offset);
+	WRITE32(args->count);
+	hdr->nops++;
 }
 
-static int encode_create(struct xdr_stream *xdr, const struct nfs4_create_arg *create)
+static void encode_create(struct xdr_stream *xdr, const struct nfs4_create_arg *create, struct compound_hdr *hdr)
 {
 	__be32 *p;
-	
+
 	RESERVE_SPACE(8);
 	WRITE32(OP_CREATE);
 	WRITE32(create->ftype);
@@ -791,64 +791,62 @@ static int encode_create(struct xdr_stream *xdr, const struct nfs4_create_arg *c
 	RESERVE_SPACE(4 + create->name->len);
 	WRITE32(create->name->len);
 	WRITEMEM(create->name->name, create->name->len);
+	hdr->nops++;
 
-	return encode_attrs(xdr, create->attrs, create->server);
+	encode_attrs(xdr, create->attrs, create->server);
 }
 
-static int encode_getattr_one(struct xdr_stream *xdr, uint32_t bitmap)
+static void encode_getattr_one(struct xdr_stream *xdr, uint32_t bitmap, struct compound_hdr *hdr)
 {
-        __be32 *p;
+	__be32 *p;
 
-        RESERVE_SPACE(12);
-        WRITE32(OP_GETATTR);
-        WRITE32(1);
-        WRITE32(bitmap);
-        return 0;
+	RESERVE_SPACE(12);
+	WRITE32(OP_GETATTR);
+	WRITE32(1);
+	WRITE32(bitmap);
+	hdr->nops++;
 }
 
-static int encode_getattr_two(struct xdr_stream *xdr, uint32_t bm0, uint32_t bm1)
+static void encode_getattr_two(struct xdr_stream *xdr, uint32_t bm0, uint32_t bm1, struct compound_hdr *hdr)
 {
-        __be32 *p;
+	__be32 *p;
 
-        RESERVE_SPACE(16);
-        WRITE32(OP_GETATTR);
-        WRITE32(2);
-        WRITE32(bm0);
-        WRITE32(bm1);
-        return 0;
+	RESERVE_SPACE(16);
+	WRITE32(OP_GETATTR);
+	WRITE32(2);
+	WRITE32(bm0);
+	WRITE32(bm1);
+	hdr->nops++;
 }
 
-static int encode_getfattr(struct xdr_stream *xdr, const u32* bitmask)
+static void encode_getfattr(struct xdr_stream *xdr, const u32* bitmask, struct compound_hdr *hdr)
 {
-	return encode_getattr_two(xdr,
-			bitmask[0] & nfs4_fattr_bitmap[0],
-			bitmask[1] & nfs4_fattr_bitmap[1]);
+	encode_getattr_two(xdr, bitmask[0] & nfs4_fattr_bitmap[0],
+			   bitmask[1] & nfs4_fattr_bitmap[1], hdr);
 }
 
-static int encode_fsinfo(struct xdr_stream *xdr, const u32* bitmask)
+static void encode_fsinfo(struct xdr_stream *xdr, const u32* bitmask, struct compound_hdr *hdr)
 {
-	return encode_getattr_two(xdr, bitmask[0] & nfs4_fsinfo_bitmap[0],
-			bitmask[1] & nfs4_fsinfo_bitmap[1]);
+	encode_getattr_two(xdr, bitmask[0] & nfs4_fsinfo_bitmap[0],
+			   bitmask[1] & nfs4_fsinfo_bitmap[1], hdr);
 }
 
-static int encode_fs_locations(struct xdr_stream *xdr, const u32* bitmask)
+static void encode_fs_locations(struct xdr_stream *xdr, const u32* bitmask, struct compound_hdr *hdr)
 {
-	return encode_getattr_two(xdr,
-				  bitmask[0] & nfs4_fs_locations_bitmap[0],
-				  bitmask[1] & nfs4_fs_locations_bitmap[1]);
+	encode_getattr_two(xdr, bitmask[0] & nfs4_fs_locations_bitmap[0],
+			   bitmask[1] & nfs4_fs_locations_bitmap[1], hdr);
 }
 
-static int encode_getfh(struct xdr_stream *xdr)
+static void encode_getfh(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(4);
 	WRITE32(OP_GETFH);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_link(struct xdr_stream *xdr, const struct qstr *name)
+static void encode_link(struct xdr_stream *xdr, const struct qstr *name, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -856,8 +854,7 @@ static int encode_link(struct xdr_stream *xdr, const struct qstr *name)
 	WRITE32(OP_LINK);
 	WRITE32(name->len);
 	WRITEMEM(name->name, name->len);
-	
-	return 0;
+	hdr->nops++;
 }
 
 static inline int nfs4_lock_type(struct file_lock *fl, int block)
@@ -878,7 +875,7 @@ static inline uint64_t nfs4_lock_length(struct file_lock *fl)
  * opcode,type,reclaim,offset,length,new_lock_owner = 32
  * open_seqid,open_stateid,lock_seqid,lock_owner.clientid, lock_owner.id = 40
  */
-static int encode_lock(struct xdr_stream *xdr, const struct nfs_lock_args *args)
+static void encode_lock(struct xdr_stream *xdr, const struct nfs_lock_args *args, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -904,11 +901,10 @@ static int encode_lock(struct xdr_stream *xdr, const struct nfs_lock_args *args)
 		WRITEMEM(args->lock_stateid->data, NFS4_STATEID_SIZE);
 		WRITE32(args->lock_seqid->sequence->counter);
 	}
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_lockt(struct xdr_stream *xdr, const struct nfs_lockt_args *args)
+static void encode_lockt(struct xdr_stream *xdr, const struct nfs_lockt_args *args, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -921,11 +917,10 @@ static int encode_lockt(struct xdr_stream *xdr, const struct nfs_lockt_args *arg
 	WRITE32(16);
 	WRITEMEM("lock id:", 8);
 	WRITE64(args->lock_owner.id);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_locku(struct xdr_stream *xdr, const struct nfs_locku_args *args)
+static void encode_locku(struct xdr_stream *xdr, const struct nfs_locku_args *args, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -936,11 +931,10 @@ static int encode_locku(struct xdr_stream *xdr, const struct nfs_locku_args *arg
 	WRITEMEM(args->stateid->data, NFS4_STATEID_SIZE);
 	WRITE64(args->fl->fl_start);
 	WRITE64(nfs4_lock_length(args->fl));
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_lookup(struct xdr_stream *xdr, const struct qstr *name)
+static void encode_lookup(struct xdr_stream *xdr, const struct qstr *name, struct compound_hdr *hdr)
 {
 	int len = name->len;
 	__be32 *p;
@@ -949,27 +943,26 @@ static int encode_lookup(struct xdr_stream *xdr, const struct qstr *name)
 	WRITE32(OP_LOOKUP);
 	WRITE32(len);
 	WRITEMEM(name->name, len);
-
-	return 0;
+	hdr->nops++;
 }
 
-static void encode_share_access(struct xdr_stream *xdr, int open_flags)
+static void encode_share_access(struct xdr_stream *xdr, fmode_t fmode)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(8);
-	switch (open_flags & (FMODE_READ|FMODE_WRITE)) {
-		case FMODE_READ:
-			WRITE32(NFS4_SHARE_ACCESS_READ);
-			break;
-		case FMODE_WRITE:
-			WRITE32(NFS4_SHARE_ACCESS_WRITE);
-			break;
-		case FMODE_READ|FMODE_WRITE:
-			WRITE32(NFS4_SHARE_ACCESS_BOTH);
-			break;
-		default:
-			BUG();
+	switch (fmode & (FMODE_READ|FMODE_WRITE)) {
+	case FMODE_READ:
+		WRITE32(NFS4_SHARE_ACCESS_READ);
+		break;
+	case FMODE_WRITE:
+		WRITE32(NFS4_SHARE_ACCESS_WRITE);
+		break;
+	case FMODE_READ|FMODE_WRITE:
+		WRITE32(NFS4_SHARE_ACCESS_BOTH);
+		break;
+	default:
+		WRITE32(0);
 	}
 	WRITE32(0);		/* for linux, share_deny = 0 always */
 }
@@ -984,7 +977,7 @@ static inline void encode_openhdr(struct xdr_stream *xdr, const struct nfs_opena
 	RESERVE_SPACE(8);
 	WRITE32(OP_OPEN);
 	WRITE32(arg->seqid->sequence->counter);
-	encode_share_access(xdr, arg->open_flags);
+	encode_share_access(xdr, arg->fmode);
 	RESERVE_SPACE(28);
 	WRITE64(arg->clientid);
 	WRITE32(16);
@@ -998,13 +991,13 @@ static inline void encode_createmode(struct xdr_stream *xdr, const struct nfs_op
 
 	RESERVE_SPACE(4);
 	switch(arg->open_flags & O_EXCL) {
-		case 0:
-			WRITE32(NFS4_CREATE_UNCHECKED);
-			encode_attrs(xdr, arg->u.attrs, arg->server);
-			break;
-		default:
-			WRITE32(NFS4_CREATE_EXCLUSIVE);
-			encode_nfs4_verifier(xdr, &arg->u.verifier);
+	case 0:
+		WRITE32(NFS4_CREATE_UNCHECKED);
+		encode_attrs(xdr, arg->u.attrs, arg->server);
+		break;
+	default:
+		WRITE32(NFS4_CREATE_EXCLUSIVE);
+		encode_nfs4_verifier(xdr, &arg->u.verifier);
 	}
 }
 
@@ -1014,33 +1007,33 @@ static void encode_opentype(struct xdr_stream *xdr, const struct nfs_openargs *a
 
 	RESERVE_SPACE(4);
 	switch (arg->open_flags & O_CREAT) {
-		case 0:
-			WRITE32(NFS4_OPEN_NOCREATE);
-			break;
-		default:
-			BUG_ON(arg->claim != NFS4_OPEN_CLAIM_NULL);
-			WRITE32(NFS4_OPEN_CREATE);
-			encode_createmode(xdr, arg);
+	case 0:
+		WRITE32(NFS4_OPEN_NOCREATE);
+		break;
+	default:
+		BUG_ON(arg->claim != NFS4_OPEN_CLAIM_NULL);
+		WRITE32(NFS4_OPEN_CREATE);
+		encode_createmode(xdr, arg);
 	}
 }
 
-static inline void encode_delegation_type(struct xdr_stream *xdr, int delegation_type)
+static inline void encode_delegation_type(struct xdr_stream *xdr, fmode_t delegation_type)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(4);
 	switch (delegation_type) {
-		case 0:
-			WRITE32(NFS4_OPEN_DELEGATE_NONE);
-			break;
-		case FMODE_READ:
-			WRITE32(NFS4_OPEN_DELEGATE_READ);
-			break;
-		case FMODE_WRITE|FMODE_READ:
-			WRITE32(NFS4_OPEN_DELEGATE_WRITE);
-			break;
-		default:
-			BUG();
+	case 0:
+		WRITE32(NFS4_OPEN_DELEGATE_NONE);
+		break;
+	case FMODE_READ:
+		WRITE32(NFS4_OPEN_DELEGATE_READ);
+		break;
+	case FMODE_WRITE|FMODE_READ:
+		WRITE32(NFS4_OPEN_DELEGATE_WRITE);
+		break;
+	default:
+		BUG();
 	}
 }
 
@@ -1053,7 +1046,7 @@ static inline void encode_claim_null(struct xdr_stream *xdr, const struct qstr *
 	encode_string(xdr, name->len, name->name);
 }
 
-static inline void encode_claim_previous(struct xdr_stream *xdr, int type)
+static inline void encode_claim_previous(struct xdr_stream *xdr, fmode_t type)
 {
 	__be32 *p;
 
@@ -1072,27 +1065,27 @@ static inline void encode_claim_delegate_cur(struct xdr_stream *xdr, const struc
 	encode_string(xdr, name->len, name->name);
 }
 
-static int encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg)
+static void encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg, struct compound_hdr *hdr)
 {
 	encode_openhdr(xdr, arg);
 	encode_opentype(xdr, arg);
 	switch (arg->claim) {
-		case NFS4_OPEN_CLAIM_NULL:
-			encode_claim_null(xdr, arg->name);
-			break;
-		case NFS4_OPEN_CLAIM_PREVIOUS:
-			encode_claim_previous(xdr, arg->u.delegation_type);
-			break;
-		case NFS4_OPEN_CLAIM_DELEGATE_CUR:
-			encode_claim_delegate_cur(xdr, arg->name, &arg->u.delegation);
-			break;
-		default:
-			BUG();
+	case NFS4_OPEN_CLAIM_NULL:
+		encode_claim_null(xdr, arg->name);
+		break;
+	case NFS4_OPEN_CLAIM_PREVIOUS:
+		encode_claim_previous(xdr, arg->u.delegation_type);
+		break;
+	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
+		encode_claim_delegate_cur(xdr, arg->name, &arg->u.delegation);
+		break;
+	default:
+		BUG();
 	}
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_open_confirm(struct xdr_stream *xdr, const struct nfs_open_confirmargs *arg)
+static void encode_open_confirm(struct xdr_stream *xdr, const struct nfs_open_confirmargs *arg, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1100,11 +1093,10 @@ static int encode_open_confirm(struct xdr_stream *xdr, const struct nfs_open_con
 	WRITE32(OP_OPEN_CONFIRM);
 	WRITEMEM(arg->stateid->data, NFS4_STATEID_SIZE);
 	WRITE32(arg->seqid->sequence->counter);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_open_downgrade(struct xdr_stream *xdr, const struct nfs_closeargs *arg)
+static void encode_open_downgrade(struct xdr_stream *xdr, const struct nfs_closeargs *arg, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1112,12 +1104,12 @@ static int encode_open_downgrade(struct xdr_stream *xdr, const struct nfs_closea
 	WRITE32(OP_OPEN_DOWNGRADE);
 	WRITEMEM(arg->stateid->data, NFS4_STATEID_SIZE);
 	WRITE32(arg->seqid->sequence->counter);
-	encode_share_access(xdr, arg->open_flags);
-	return 0;
+	encode_share_access(xdr, arg->fmode);
+	hdr->nops++;
 }
 
-static int
-encode_putfh(struct xdr_stream *xdr, const struct nfs_fh *fh)
+static void
+encode_putfh(struct xdr_stream *xdr, const struct nfs_fh *fh, struct compound_hdr *hdr)
 {
 	int len = fh->size;
 	__be32 *p;
@@ -1126,18 +1118,16 @@ encode_putfh(struct xdr_stream *xdr, const struct nfs_fh *fh)
 	WRITE32(OP_PUTFH);
 	WRITE32(len);
 	WRITEMEM(fh->data, len);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_putrootfh(struct xdr_stream *xdr)
+static void encode_putrootfh(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
-        __be32 *p;
-        
-        RESERVE_SPACE(4);
-        WRITE32(OP_PUTROOTFH);
+	__be32 *p;
 
-        return 0;
+	RESERVE_SPACE(4);
+	WRITE32(OP_PUTROOTFH);
+	hdr->nops++;
 }
 
 static void encode_stateid(struct xdr_stream *xdr, const struct nfs_open_context *ctx)
@@ -1153,7 +1143,7 @@ static void encode_stateid(struct xdr_stream *xdr, const struct nfs_open_context
 		WRITEMEM(zero_stateid.data, NFS4_STATEID_SIZE);
 }
 
-static int encode_read(struct xdr_stream *xdr, const struct nfs_readargs *args)
+static void encode_read(struct xdr_stream *xdr, const struct nfs_readargs *args, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1165,11 +1155,10 @@ static int encode_read(struct xdr_stream *xdr, const struct nfs_readargs *args)
 	RESERVE_SPACE(12);
 	WRITE64(args->offset);
 	WRITE32(args->count);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg *readdir, struct rpc_rqst *req)
+static void encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg *readdir, struct rpc_rqst *req, struct compound_hdr *hdr)
 {
 	uint32_t attrs[2] = {
 		FATTR4_WORD0_RDATTR_ERROR|FATTR4_WORD0_FILEID,
@@ -1191,6 +1180,7 @@ static int encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg
 		attrs[1] &= ~FATTR4_WORD1_MOUNTED_ON_FILEID;
 	WRITE32(attrs[0] & readdir->bitmask[0]);
 	WRITE32(attrs[1] & readdir->bitmask[1]);
+	hdr->nops++;
 	dprintk("%s: cookie = %Lu, verifier = %08x:%08x, bitmap = %08x:%08x\n",
 			__func__,
 			(unsigned long long)readdir->cookie,
@@ -1198,21 +1188,18 @@ static int encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg
 			((u32 *)readdir->verifier.data)[1],
 			attrs[0] & readdir->bitmask[0],
 			attrs[1] & readdir->bitmask[1]);
-
-	return 0;
 }
 
-static int encode_readlink(struct xdr_stream *xdr, const struct nfs4_readlink *readlink, struct rpc_rqst *req)
+static void encode_readlink(struct xdr_stream *xdr, const struct nfs4_readlink *readlink, struct rpc_rqst *req, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(4);
 	WRITE32(OP_READLINK);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_remove(struct xdr_stream *xdr, const struct qstr *name)
+static void encode_remove(struct xdr_stream *xdr, const struct qstr *name, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1220,11 +1207,10 @@ static int encode_remove(struct xdr_stream *xdr, const struct qstr *name)
 	WRITE32(OP_REMOVE);
 	WRITE32(name->len);
 	WRITEMEM(name->name, name->len);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_rename(struct xdr_stream *xdr, const struct qstr *oldname, const struct qstr *newname)
+static void encode_rename(struct xdr_stream *xdr, const struct qstr *oldname, const struct qstr *newname, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1232,38 +1218,35 @@ static int encode_rename(struct xdr_stream *xdr, const struct qstr *oldname, con
 	WRITE32(OP_RENAME);
 	WRITE32(oldname->len);
 	WRITEMEM(oldname->name, oldname->len);
-	
+
 	RESERVE_SPACE(4 + newname->len);
 	WRITE32(newname->len);
 	WRITEMEM(newname->name, newname->len);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_renew(struct xdr_stream *xdr, const struct nfs_client *client_stateid)
+static void encode_renew(struct xdr_stream *xdr, const struct nfs_client *client_stateid, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(12);
 	WRITE32(OP_RENEW);
 	WRITE64(client_stateid->cl_clientid);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int
-encode_restorefh(struct xdr_stream *xdr)
+static void
+encode_restorefh(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(4);
 	WRITE32(OP_RESTOREFH);
-
-	return 0;
+	hdr->nops++;
 }
 
 static int
-encode_setacl(struct xdr_stream *xdr, struct nfs_setaclargs *arg)
+encode_setacl(struct xdr_stream *xdr, struct nfs_setaclargs *arg, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1278,36 +1261,32 @@ encode_setacl(struct xdr_stream *xdr, struct nfs_setaclargs *arg)
 	RESERVE_SPACE(4);
 	WRITE32(arg->acl_len);
 	xdr_write_pages(xdr, arg->acl_pages, arg->acl_pgbase, arg->acl_len);
+	hdr->nops++;
 	return 0;
 }
 
-static int
-encode_savefh(struct xdr_stream *xdr)
+static void
+encode_savefh(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
 	RESERVE_SPACE(4);
 	WRITE32(OP_SAVEFH);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_setattr(struct xdr_stream *xdr, const struct nfs_setattrargs *arg, const struct nfs_server *server)
+static void encode_setattr(struct xdr_stream *xdr, const struct nfs_setattrargs *arg, const struct nfs_server *server, struct compound_hdr *hdr)
 {
-	int status;
 	__be32 *p;
-	
-        RESERVE_SPACE(4+NFS4_STATEID_SIZE);
-        WRITE32(OP_SETATTR);
-	WRITEMEM(arg->stateid.data, NFS4_STATEID_SIZE);
 
-        if ((status = encode_attrs(xdr, arg->iap, server)))
-		return status;
-
-        return 0;
+	RESERVE_SPACE(4+NFS4_STATEID_SIZE);
+	WRITE32(OP_SETATTR);
+	WRITEMEM(arg->stateid.data, NFS4_STATEID_SIZE);
+	hdr->nops++;
+	encode_attrs(xdr, arg->iap, server);
 }
 
-static int encode_setclientid(struct xdr_stream *xdr, const struct nfs4_setclientid *setclientid)
+static void encode_setclientid(struct xdr_stream *xdr, const struct nfs4_setclientid *setclientid, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1322,23 +1301,21 @@ static int encode_setclientid(struct xdr_stream *xdr, const struct nfs4_setclien
 	encode_string(xdr, setclientid->sc_uaddr_len, setclientid->sc_uaddr);
 	RESERVE_SPACE(4);
 	WRITE32(setclientid->sc_cb_ident);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_setclientid_confirm(struct xdr_stream *xdr, const struct nfs_client *client_state)
+static void encode_setclientid_confirm(struct xdr_stream *xdr, const struct nfs_client *client_state, struct compound_hdr *hdr)
 {
-        __be32 *p;
-
-        RESERVE_SPACE(12 + NFS4_VERIFIER_SIZE);
-        WRITE32(OP_SETCLIENTID_CONFIRM);
-        WRITE64(client_state->cl_clientid);
-        WRITEMEM(client_state->cl_confirm.data, NFS4_VERIFIER_SIZE);
+	__be32 *p;
 
-        return 0;
+	RESERVE_SPACE(12 + NFS4_VERIFIER_SIZE);
+	WRITE32(OP_SETCLIENTID_CONFIRM);
+	WRITE64(client_state->cl_clientid);
+	WRITEMEM(client_state->cl_confirm.data, NFS4_VERIFIER_SIZE);
+	hdr->nops++;
 }
 
-static int encode_write(struct xdr_stream *xdr, const struct nfs_writeargs *args)
+static void encode_write(struct xdr_stream *xdr, const struct nfs_writeargs *args, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1353,11 +1330,10 @@ static int encode_write(struct xdr_stream *xdr, const struct nfs_writeargs *args
 	WRITE32(args->count);
 
 	xdr_write_pages(xdr, args->pages, args->pgbase, args->count);
-
-	return 0;
+	hdr->nops++;
 }
 
-static int encode_delegreturn(struct xdr_stream *xdr, const nfs4_stateid *stateid)
+static void encode_delegreturn(struct xdr_stream *xdr, const nfs4_stateid *stateid, struct compound_hdr *hdr)
 {
 	__be32 *p;
 
@@ -1365,8 +1341,7 @@ static int encode_delegreturn(struct xdr_stream *xdr, const nfs4_stateid *statei
 
 	WRITE32(OP_DELEGRETURN);
 	WRITEMEM(stateid->data, NFS4_STATEID_SIZE);
-	return 0;
-
+	hdr->nops++;
 }
 /*
  * END OF "GENERIC" ENCODE ROUTINES.
@@ -1379,21 +1354,16 @@ static int nfs4_xdr_enc_access(struct rpc_rqst *req, __be32 *p, const struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status != 0)
-		goto out;
-	status = encode_access(&xdr, args->access);
-	if (status != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_access(&xdr, args->access, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1403,21 +1373,17 @@ static int nfs4_xdr_enc_lookup(struct rpc_rqst *req, __be32 *p, const struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 4,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->dir_fh)) != 0)
-		goto out;
-	if ((status = encode_lookup(&xdr, args->name)) != 0)
-		goto out;
-	if ((status = encode_getfh(&xdr)) != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->dir_fh, &hdr);
+	encode_lookup(&xdr, args->name, &hdr);
+	encode_getfh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1427,18 +1393,16 @@ static int nfs4_xdr_enc_lookup_root(struct rpc_rqst *req, __be32 *p, const struc
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putrootfh(&xdr)) != 0)
-		goto out;
-	if ((status = encode_getfh(&xdr)) == 0)
-		status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putrootfh(&xdr, &hdr);
+	encode_getfh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1448,19 +1412,16 @@ static int nfs4_xdr_enc_remove(struct rpc_rqst *req, __be32 *p, const struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->fh)) != 0)
-		goto out;
-	if ((status = encode_remove(&xdr, &args->name)) != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_remove(&xdr, &args->name, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1470,27 +1431,20 @@ static int nfs4_xdr_enc_rename(struct rpc_rqst *req, __be32 *p, const struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 7,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->old_dir)) != 0)
-		goto out;
-	if ((status = encode_savefh(&xdr)) != 0)
-		goto out;
-	if ((status = encode_putfh(&xdr, args->new_dir)) != 0)
-		goto out;
-	if ((status = encode_rename(&xdr, args->old_name, args->new_name)) != 0)
-		goto out;
-	if ((status = encode_getfattr(&xdr, args->bitmask)) != 0)
-		goto out;
-	if ((status = encode_restorefh(&xdr)) != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->old_dir, &hdr);
+	encode_savefh(&xdr, &hdr);
+	encode_putfh(&xdr, args->new_dir, &hdr);
+	encode_rename(&xdr, args->old_name, args->new_name, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_restorefh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1500,27 +1454,20 @@ static int nfs4_xdr_enc_link(struct rpc_rqst *req, __be32 *p, const struct nfs4_
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 7,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->fh)) != 0)
-		goto out;
-	if ((status = encode_savefh(&xdr)) != 0)
-		goto out;
-	if ((status = encode_putfh(&xdr, args->dir_fh)) != 0)
-		goto out;
-	if ((status = encode_link(&xdr, args->name)) != 0)
-		goto out;
-	if ((status = encode_getfattr(&xdr, args->bitmask)) != 0)
-		goto out;
-	if ((status = encode_restorefh(&xdr)) != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_savefh(&xdr, &hdr);
+	encode_putfh(&xdr, args->dir_fh, &hdr);
+	encode_link(&xdr, args->name, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_restorefh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1530,27 +1477,20 @@ static int nfs4_xdr_enc_create(struct rpc_rqst *req, __be32 *p, const struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 7,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->dir_fh)) != 0)
-		goto out;
-	if ((status = encode_savefh(&xdr)) != 0)
-		goto out;
-	if ((status = encode_create(&xdr, args)) != 0)
-		goto out;
-	if ((status = encode_getfh(&xdr)) != 0)
-		goto out;
-	if ((status = encode_getfattr(&xdr, args->bitmask)) != 0)
-		goto out;
-	if ((status = encode_restorefh(&xdr)) != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->dir_fh, &hdr);
+	encode_savefh(&xdr, &hdr);
+	encode_create(&xdr, args, &hdr);
+	encode_getfh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_restorefh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1568,15 +1508,15 @@ static int nfs4_xdr_enc_getattr(struct rpc_rqst *req, __be32 *p, const struct nf
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->fh)) == 0)
-		status = encode_getfattr(&xdr, args->bitmask);
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1584,23 +1524,18 @@ static int nfs4_xdr_enc_getattr(struct rpc_rqst *req, __be32 *p, const struct nf
  */
 static int nfs4_xdr_enc_close(struct rpc_rqst *req, __be32 *p, struct nfs_closeargs *args)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr = {
-                .nops   = 3,
-        };
-        int status;
-
-        xdr_init_encode(&xdr, &req->rq_snd_buf, p);
-        encode_compound_hdr(&xdr, &hdr);
-        status = encode_putfh(&xdr, args->fh);
-        if(status)
-                goto out;
-        status = encode_close(&xdr, args);
-	if (status != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-        return status;
+	struct xdr_stream xdr;
+	struct compound_hdr hdr = {
+		.nops   = 0,
+	};
+
+	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
+	encode_compound_hdr(&xdr, &hdr);
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_close(&xdr, args, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1610,33 +1545,20 @@ static int nfs4_xdr_enc_open(struct rpc_rqst *req, __be32 *p, struct nfs_openarg
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 7,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_savefh(&xdr);
-	if (status)
-		goto out;
-	status = encode_open(&xdr, args);
-	if (status)
-		goto out;
-	status = encode_getfh(&xdr);
-	if (status)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-	if (status)
-		goto out;
-	status = encode_restorefh(&xdr);
-	if (status)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_savefh(&xdr, &hdr);
+	encode_open(&xdr, args, &hdr);
+	encode_getfh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_restorefh(&xdr, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1646,18 +1568,15 @@ static int nfs4_xdr_enc_open_confirm(struct rpc_rqst *req, __be32 *p, struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops   = 2,
+		.nops   = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if(status)
-		goto out;
-	status = encode_open_confirm(&xdr, args);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_open_confirm(&xdr, args, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1667,21 +1586,16 @@ static int nfs4_xdr_enc_open_noattr(struct rpc_rqst *req, __be32 *p, struct nfs_
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops   = 3,
+		.nops   = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_open(&xdr, args);
-	if (status)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_open(&xdr, args, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1691,21 +1605,16 @@ static int nfs4_xdr_enc_open_downgrade(struct rpc_rqst *req, __be32 *p, struct n
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops	= 3,
+		.nops	= 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_open_downgrade(&xdr, args);
-	if (status != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_open_downgrade(&xdr, args, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1715,18 +1624,15 @@ static int nfs4_xdr_enc_lock(struct rpc_rqst *req, __be32 *p, struct nfs_lock_ar
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops   = 2,
+		.nops   = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if(status)
-		goto out;
-	status = encode_lock(&xdr, args);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_lock(&xdr, args, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1736,18 +1642,15 @@ static int nfs4_xdr_enc_lockt(struct rpc_rqst *req, __be32 *p, struct nfs_lockt_
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops   = 2,
+		.nops   = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if(status)
-		goto out;
-	status = encode_lockt(&xdr, args);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_lockt(&xdr, args, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1757,18 +1660,15 @@ static int nfs4_xdr_enc_locku(struct rpc_rqst *req, __be32 *p, struct nfs_locku_
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops   = 2,
+		.nops   = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if(status)
-		goto out;
-	status = encode_locku(&xdr, args);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_locku(&xdr, args, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1778,18 +1678,15 @@ static int nfs4_xdr_enc_readlink(struct rpc_rqst *req, __be32 *p, const struct n
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
 	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if(status)
-		goto out;
-	status = encode_readlink(&xdr, args, req);
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_readlink(&xdr, args, req, &hdr);
 
 	/* set up reply kvec
 	 *    toplevel_status + taglen + rescount + OP_PUTFH + status
@@ -1798,9 +1695,8 @@ static int nfs4_xdr_enc_readlink(struct rpc_rqst *req, __be32 *p, const struct n
 	replen = (RPC_REPHDRSIZE + auth->au_rslack + NFS4_dec_readlink_sz) << 2;
 	xdr_inline_pages(&req->rq_rcv_buf, replen, args->pages,
 			args->pgbase, args->pglen);
-
-out:
-	return status;
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1810,18 +1706,15 @@ static int nfs4_xdr_enc_readdir(struct rpc_rqst *req, __be32 *p, const struct nf
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
 	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	int replen;
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if(status)
-		goto out;
-	status = encode_readdir(&xdr, args, req);
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_readdir(&xdr, args, req, &hdr);
 
 	/* set up reply kvec
 	 *    toplevel_status + taglen + rescount + OP_PUTFH + status
@@ -1833,9 +1726,8 @@ static int nfs4_xdr_enc_readdir(struct rpc_rqst *req, __be32 *p, const struct nf
 	dprintk("%s: inlined page args = (%u, %p, %u, %u)\n",
 			__func__, replen, args->pages,
 			args->pgbase, args->count);
-
-out:
-	return status;
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1846,18 +1738,14 @@ static int nfs4_xdr_enc_read(struct rpc_rqst *req, __be32 *p, struct nfs_readarg
 	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
-	int replen, status;
+	int replen;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_read(&xdr, args);
-	if (status)
-		goto out;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_read(&xdr, args, &hdr);
 
 	/* set up reply kvec
 	 *    toplevel status + taglen=0 + rescount + OP_PUTFH + status
@@ -1867,33 +1755,27 @@ static int nfs4_xdr_enc_read(struct rpc_rqst *req, __be32 *p, struct nfs_readarg
 	xdr_inline_pages(&req->rq_rcv_buf, replen,
 			 args->pages, args->pgbase, args->count);
 	req->rq_rcv_buf.flags |= XDRBUF_READ;
-out:
-	return status;
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
  * Encode an SETATTR request
  */
 static int nfs4_xdr_enc_setattr(struct rpc_rqst *req, __be32 *p, struct nfs_setattrargs *args)
-
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr = {
-                .nops   = 3,
-        };
-        int status;
-
-        xdr_init_encode(&xdr, &req->rq_snd_buf, p);
-        encode_compound_hdr(&xdr, &hdr);
-        status = encode_putfh(&xdr, args->fh);
-        if(status)
-                goto out;
-        status = encode_setattr(&xdr, args, args->server);
-        if(status)
-                goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-        return status;
+	struct xdr_stream xdr;
+	struct compound_hdr hdr = {
+		.nops   = 0,
+	};
+
+	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
+	encode_compound_hdr(&xdr, &hdr);
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_setattr(&xdr, args, args->server, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1906,22 +1788,21 @@ nfs4_xdr_enc_getacl(struct rpc_rqst *req, __be32 *p,
 	struct xdr_stream xdr;
 	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	struct compound_hdr hdr = {
-		.nops   = 2,
+		.nops   = 0,
 	};
-	int replen, status;
+	int replen;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_getattr_two(&xdr, FATTR4_WORD0_ACL, 0);
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_getattr_two(&xdr, FATTR4_WORD0_ACL, 0, &hdr);
+
 	/* set up reply buffer: */
 	replen = (RPC_REPHDRSIZE + auth->au_rslack + NFS4_dec_getacl_sz) << 2;
 	xdr_inline_pages(&req->rq_rcv_buf, replen,
 		args->acl_pages, args->acl_pgbase, args->acl_len);
-out:
-	return status;
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1931,22 +1812,17 @@ static int nfs4_xdr_enc_write(struct rpc_rqst *req, __be32 *p, struct nfs_writea
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_write(&xdr, args);
-	if (status)
-		goto out;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_write(&xdr, args, &hdr);
 	req->rq_snd_buf.flags |= XDRBUF_WRITE;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1956,21 +1832,16 @@ static int nfs4_xdr_enc_commit(struct rpc_rqst *req, __be32 *p, struct nfs_write
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status)
-		goto out;
-	status = encode_commit(&xdr, args);
-	if (status)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_commit(&xdr, args, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1980,16 +1851,15 @@ static int nfs4_xdr_enc_fsinfo(struct rpc_rqst *req, __be32 *p, struct nfs4_fsin
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops	= 2,
+		.nops	= 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (!status)
-		status = encode_fsinfo(&xdr, args->bitmask);
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_fsinfo(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -1999,17 +1869,16 @@ static int nfs4_xdr_enc_pathconf(struct rpc_rqst *req, __be32 *p, const struct n
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (!status)
-		status = encode_getattr_one(&xdr,
-				args->bitmask[0] & nfs4_pathconf_bitmap[0]);
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_getattr_one(&xdr, args->bitmask[0] & nfs4_pathconf_bitmap[0],
+			   &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2019,18 +1888,16 @@ static int nfs4_xdr_enc_statfs(struct rpc_rqst *req, __be32 *p, const struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fh);
-	if (status == 0)
-		status = encode_getattr_two(&xdr,
-				args->bitmask[0] & nfs4_statfs_bitmap[0],
-				args->bitmask[1] & nfs4_statfs_bitmap[1]);
-	return status;
+	encode_putfh(&xdr, args->fh, &hdr);
+	encode_getattr_two(&xdr, args->bitmask[0] & nfs4_statfs_bitmap[0],
+			   args->bitmask[1] & nfs4_statfs_bitmap[1], &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2040,19 +1907,18 @@ static int nfs4_xdr_enc_server_caps(struct rpc_rqst *req, __be32 *p, const struc
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 2,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, fhandle);
-	if (status == 0)
-		status = encode_getattr_one(&xdr, FATTR4_WORD0_SUPPORTED_ATTRS|
-				FATTR4_WORD0_LINK_SUPPORT|
-				FATTR4_WORD0_SYMLINK_SUPPORT|
-				FATTR4_WORD0_ACLSUPPORT);
-	return status;
+	encode_putfh(&xdr, fhandle, &hdr);
+	encode_getattr_one(&xdr, FATTR4_WORD0_SUPPORTED_ATTRS|
+			   FATTR4_WORD0_LINK_SUPPORT|
+			   FATTR4_WORD0_SYMLINK_SUPPORT|
+			   FATTR4_WORD0_ACLSUPPORT, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2062,12 +1928,14 @@ static int nfs4_xdr_enc_renew(struct rpc_rqst *req, __be32 *p, struct nfs_client
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops	= 1,
+		.nops	= 0,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	return encode_renew(&xdr, clp);
+	encode_renew(&xdr, clp, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2077,12 +1945,14 @@ static int nfs4_xdr_enc_setclientid(struct rpc_rqst *req, __be32 *p, struct nfs4
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops	= 1,
+		.nops	= 0,
 	};
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	return encode_setclientid(&xdr, sc);
+	encode_setclientid(&xdr, sc, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2092,19 +1962,17 @@ static int nfs4_xdr_enc_setclientid_confirm(struct rpc_rqst *req, __be32 *p, str
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops	= 3,
+		.nops	= 0,
 	};
 	const u32 lease_bitmap[2] = { FATTR4_WORD0_LEASE_TIME, 0 };
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_setclientid_confirm(&xdr, clp);
-	if (!status)
-		status = encode_putrootfh(&xdr);
-	if (!status)
-		status = encode_fsinfo(&xdr, lease_bitmap);
-	return status;
+	encode_setclientid_confirm(&xdr, clp, &hdr);
+	encode_putrootfh(&xdr, &hdr);
+	encode_fsinfo(&xdr, lease_bitmap, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2114,21 +1982,16 @@ static int nfs4_xdr_enc_delegreturn(struct rpc_rqst *req, __be32 *p, const struc
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	status = encode_putfh(&xdr, args->fhandle);
-	if (status != 0)
-		goto out;
-	status = encode_delegreturn(&xdr, args->stateid);
-	if (status != 0)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
-out:
-	return status;
+	encode_putfh(&xdr, args->fhandle, &hdr);
+	encode_delegreturn(&xdr, args->stateid, &hdr);
+	encode_getfattr(&xdr, args->bitmask, &hdr);
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2138,20 +2001,17 @@ static int nfs4_xdr_enc_fs_locations(struct rpc_rqst *req, __be32 *p, struct nfs
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops = 3,
+		.nops = 0,
 	};
 	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	int replen;
-	int status;
 
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
-	if ((status = encode_putfh(&xdr, args->dir_fh)) != 0)
-		goto out;
-	if ((status = encode_lookup(&xdr, args->name)) != 0)
-		goto out;
-	if ((status = encode_fs_locations(&xdr, args->bitmask)) != 0)
-		goto out;
+	encode_putfh(&xdr, args->dir_fh, &hdr);
+	encode_lookup(&xdr, args->name, &hdr);
+	encode_fs_locations(&xdr, args->bitmask, &hdr);
+
 	/* set up reply
 	 *   toplevel_status + OP_PUTFH + status
 	 *   + OP_LOOKUP + status + OP_GETATTR + status = 7
@@ -2159,8 +2019,8 @@ static int nfs4_xdr_enc_fs_locations(struct rpc_rqst *req, __be32 *p, struct nfs
 	replen = (RPC_REPHDRSIZE + auth->au_rslack + 7) << 2;
 	xdr_inline_pages(&req->rq_rcv_buf, replen, &args->page,
 			0, PAGE_SIZE);
-out:
-	return status;
+	encode_nops(&hdr);
+	return 0;
 }
 
 /*
@@ -2217,11 +2077,13 @@ static int decode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
 	READ_BUF(8);
 	READ32(hdr->status);
 	READ32(hdr->taglen);
-	
+
 	READ_BUF(hdr->taglen + 4);
 	hdr->tag = (char *)p;
 	p += XDR_QUADLEN(hdr->taglen);
 	READ32(hdr->nops);
+	if (unlikely(hdr->nops < 1))
+		return nfs4_stat_to_errno(hdr->status);
 	return 0;
 }
 
@@ -3047,8 +2909,7 @@ static int decode_create(struct xdr_stream *xdr, struct nfs4_change_info *cinfo)
 static int decode_server_caps(struct xdr_stream *xdr, struct nfs4_server_caps_res *res)
 {
 	__be32 *savep;
-	uint32_t attrlen, 
-		 bitmap[2] = {0};
+	uint32_t attrlen, bitmap[2] = {0};
 	int status;
 
 	if ((status = decode_op_hdr(xdr, OP_GETATTR)) != 0)
@@ -3070,14 +2931,13 @@ xdr_error:
 	dprintk("%s: xdr returned %d!\n", __func__, -status);
 	return status;
 }
-	
+
 static int decode_statfs(struct xdr_stream *xdr, struct nfs_fsstat *fsstat)
 {
 	__be32 *savep;
-	uint32_t attrlen, 
-		 bitmap[2] = {0};
+	uint32_t attrlen, bitmap[2] = {0};
 	int status;
-	
+
 	if ((status = decode_op_hdr(xdr, OP_GETATTR)) != 0)
 		goto xdr_error;
 	if ((status = decode_attr_bitmap(xdr, bitmap)) != 0)
@@ -3107,10 +2967,9 @@ xdr_error:
 static int decode_pathconf(struct xdr_stream *xdr, struct nfs_pathconf *pathconf)
 {
 	__be32 *savep;
-	uint32_t attrlen, 
-		 bitmap[2] = {0};
+	uint32_t attrlen, bitmap[2] = {0};
 	int status;
-	
+
 	if ((status = decode_op_hdr(xdr, OP_GETATTR)) != 0)
 		goto xdr_error;
 	if ((status = decode_attr_bitmap(xdr, bitmap)) != 0)
@@ -3256,7 +3115,7 @@ static int decode_getfh(struct xdr_stream *xdr, struct nfs_fh *fh)
 static int decode_link(struct xdr_stream *xdr, struct nfs4_change_info *cinfo)
 {
 	int status;
-	
+
 	status = decode_op_hdr(xdr, OP_LINK);
 	if (status)
 		return status;
@@ -3344,27 +3203,27 @@ static int decode_lookup(struct xdr_stream *xdr)
 /* This is too sick! */
 static int decode_space_limit(struct xdr_stream *xdr, u64 *maxsize)
 {
-        __be32 *p;
+	__be32 *p;
 	uint32_t limit_type, nblocks, blocksize;
 
 	READ_BUF(12);
 	READ32(limit_type);
 	switch (limit_type) {
-		case 1:
-			READ64(*maxsize);
-			break;
-		case 2:
-			READ32(nblocks);
-			READ32(blocksize);
-			*maxsize = (uint64_t)nblocks * (uint64_t)blocksize;
+	case 1:
+		READ64(*maxsize);
+		break;
+	case 2:
+		READ32(nblocks);
+		READ32(blocksize);
+		*maxsize = (uint64_t)nblocks * (uint64_t)blocksize;
 	}
 	return 0;
 }
 
 static int decode_delegation(struct xdr_stream *xdr, struct nfs_openres *res)
 {
-        __be32 *p;
-        uint32_t delegation_type;
+	__be32 *p;
+	uint32_t delegation_type;
 
 	READ_BUF(4);
 	READ32(delegation_type);
@@ -3375,13 +3234,14 @@ static int decode_delegation(struct xdr_stream *xdr, struct nfs_openres *res)
 	READ_BUF(NFS4_STATEID_SIZE+4);
 	COPYMEM(res->delegation.data, NFS4_STATEID_SIZE);
 	READ32(res->do_recall);
+
 	switch (delegation_type) {
-		case NFS4_OPEN_DELEGATE_READ:
-			res->delegation_type = FMODE_READ;
-			break;
-		case NFS4_OPEN_DELEGATE_WRITE:
-			res->delegation_type = FMODE_WRITE|FMODE_READ;
-			if (decode_space_limit(xdr, &res->maxsize) < 0)
+	case NFS4_OPEN_DELEGATE_READ:
+		res->delegation_type = FMODE_READ;
+		break;
+	case NFS4_OPEN_DELEGATE_WRITE:
+		res->delegation_type = FMODE_WRITE|FMODE_READ;
+		if (decode_space_limit(xdr, &res->maxsize) < 0)
 				return -EIO;
 	}
 	return decode_ace(xdr, NULL, res->server->nfs_client);
@@ -3389,27 +3249,27 @@ static int decode_delegation(struct xdr_stream *xdr, struct nfs_openres *res)
 
 static int decode_open(struct xdr_stream *xdr, struct nfs_openres *res)
 {
-        __be32 *p;
+	__be32 *p;
 	uint32_t savewords, bmlen, i;
-        int status;
+	int status;
 
-        status = decode_op_hdr(xdr, OP_OPEN);
+	status = decode_op_hdr(xdr, OP_OPEN);
 	if (status != -EIO)
 		nfs_increment_open_seqid(status, res->seqid);
-        if (status)
-                return status;
-        READ_BUF(NFS4_STATEID_SIZE);
-        COPYMEM(res->stateid.data, NFS4_STATEID_SIZE);
+	if (status)
+		return status;
+	READ_BUF(NFS4_STATEID_SIZE);
+	COPYMEM(res->stateid.data, NFS4_STATEID_SIZE);
 
-        decode_change_info(xdr, &res->cinfo);
+	decode_change_info(xdr, &res->cinfo);
 
-        READ_BUF(8);
-        READ32(res->rflags);
-        READ32(bmlen);
-        if (bmlen > 10)
-                goto xdr_error;
+	READ_BUF(8);
+	READ32(res->rflags);
+	READ32(bmlen);
+	if (bmlen > 10)
+		goto xdr_error;
 
-        READ_BUF(bmlen << 2);
+	READ_BUF(bmlen << 2);
 	savewords = min_t(uint32_t, bmlen, NFS4_BITMAP_SIZE);
 	for (i = 0; i < savewords; ++i)
 		READ32(res->attrset[i]);
@@ -3424,17 +3284,17 @@ xdr_error:
 
 static int decode_open_confirm(struct xdr_stream *xdr, struct nfs_open_confirmres *res)
 {
-        __be32 *p;
+	__be32 *p;
 	int status;
 
-        status = decode_op_hdr(xdr, OP_OPEN_CONFIRM);
+	status = decode_op_hdr(xdr, OP_OPEN_CONFIRM);
 	if (status != -EIO)
 		nfs_increment_open_seqid(status, res->seqid);
-        if (status)
-                return status;
-        READ_BUF(NFS4_STATEID_SIZE);
-        COPYMEM(res->stateid.data, NFS4_STATEID_SIZE);
-        return 0;
+	if (status)
+		return status;
+	READ_BUF(NFS4_STATEID_SIZE);
+	COPYMEM(res->stateid.data, NFS4_STATEID_SIZE);
+	return 0;
 }
 
 static int decode_open_downgrade(struct xdr_stream *xdr, struct nfs_closeres *res)
@@ -3562,7 +3422,7 @@ static int decode_readdir(struct xdr_stream *xdr, struct rpc_rqst *req, struct n
 		dprintk("NFS: readdir reply truncated!\n");
 		entry[1] = 1;
 	}
-out:	
+out:
 	kunmap_atomic(kaddr, KM_USER0);
 	return 0;
 short_pkt:
@@ -3718,7 +3578,6 @@ static int decode_setattr(struct xdr_stream *xdr, struct nfs_setattrres *res)
 	uint32_t bmlen;
 	int status;
 
-        
 	status = decode_op_hdr(xdr, OP_SETATTR);
 	if (status)
 		return status;
@@ -3738,7 +3597,7 @@ static int decode_setclientid(struct xdr_stream *xdr, struct nfs_client *clp)
 	READ32(opnum);
 	if (opnum != OP_SETCLIENTID) {
 		dprintk("nfs: decode_setclientid: Server returned operation"
-			       	" %d\n", opnum);
+			" %d\n", opnum);
 		return -EIO;
 	}
 	READ32(nfserr);
@@ -3792,34 +3651,34 @@ static int decode_delegreturn(struct xdr_stream *xdr)
 }
 
 /*
+ * END OF "GENERIC" DECODE ROUTINES.
+ */
+
+/*
  * Decode OPEN_DOWNGRADE response
  */
 static int nfs4_xdr_dec_open_downgrade(struct rpc_rqst *rqstp, __be32 *p, struct nfs_closeres *res)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr;
-        int status;
-
-        xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
-        status = decode_compound_hdr(&xdr, &hdr);
-        if (status)
-                goto out;
-        status = decode_putfh(&xdr);
-        if (status)
-                goto out;
-        status = decode_open_downgrade(&xdr, res);
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_putfh(&xdr);
+	if (status)
+		goto out;
+	status = decode_open_downgrade(&xdr, res);
 	if (status != 0)
 		goto out;
 	decode_getfattr(&xdr, res->fattr, res->server);
 out:
-        return status;
+	return status;
 }
 
 /*
- * END OF "GENERIC" DECODE ROUTINES.
- */
-
-/*
  * Decode ACCESS response
  */
 static int nfs4_xdr_dec_access(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_accessres *res)
@@ -3827,7 +3686,7 @@ static int nfs4_xdr_dec_access(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_ac
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -3850,7 +3709,7 @@ static int nfs4_xdr_dec_lookup(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_lo
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -3873,7 +3732,7 @@ static int nfs4_xdr_dec_lookup_root(struct rpc_rqst *rqstp, __be32 *p, struct nf
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -3893,7 +3752,7 @@ static int nfs4_xdr_dec_remove(struct rpc_rqst *rqstp, __be32 *p, struct nfs_rem
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -3914,7 +3773,7 @@ static int nfs4_xdr_dec_rename(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_re
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -3944,7 +3803,7 @@ static int nfs4_xdr_dec_link(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_link
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -3977,7 +3836,7 @@ static int nfs4_xdr_dec_create(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_cr
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	if ((status = decode_compound_hdr(&xdr, &hdr)) != 0)
 		goto out;
@@ -4014,7 +3873,7 @@ static int nfs4_xdr_dec_getattr(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_g
 	struct xdr_stream xdr;
 	struct compound_hdr hdr;
 	int status;
-	
+
 	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
 	status = decode_compound_hdr(&xdr, &hdr);
 	if (status)
@@ -4025,7 +3884,6 @@ static int nfs4_xdr_dec_getattr(struct rpc_rqst *rqstp, __be32 *p, struct nfs4_g
 	status = decode_getfattr(&xdr, res->fattr, res->server);
 out:
 	return status;
-
 }
 
 /*
@@ -4034,21 +3892,20 @@ out:
 static int
 nfs4_xdr_enc_setacl(struct rpc_rqst *req, __be32 *p, struct nfs_setaclargs *args)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr = {
-                .nops   = 2,
-        };
-        int status;
-
-        xdr_init_encode(&xdr, &req->rq_snd_buf, p);
-        encode_compound_hdr(&xdr, &hdr);
-        status = encode_putfh(&xdr, args->fh);
-        if (status)
-                goto out;
-        status = encode_setacl(&xdr, args);
-out:
-        return status;
+	struct xdr_stream xdr;
+	struct compound_hdr hdr = {
+		.nops   = 0,
+	};
+	int status;
+
+	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
+	encode_compound_hdr(&xdr, &hdr);
+	encode_putfh(&xdr, args->fh, &hdr);
+	status = encode_setacl(&xdr, args, &hdr);
+	encode_nops(&hdr);
+	return status;
 }
+
 /*
  * Decode SETACL response
  */
@@ -4099,18 +3956,18 @@ out:
  */
 static int nfs4_xdr_dec_close(struct rpc_rqst *rqstp, __be32 *p, struct nfs_closeres *res)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr;
-        int status;
-
-        xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
-        status = decode_compound_hdr(&xdr, &hdr);
-        if (status)
-                goto out;
-        status = decode_putfh(&xdr);
-        if (status)
-                goto out;
-        status = decode_close(&xdr, res);
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_putfh(&xdr);
+	if (status)
+		goto out;
+	status = decode_close(&xdr, res);
 	if (status != 0)
 		goto out;
 	/*
@@ -4121,7 +3978,7 @@ static int nfs4_xdr_dec_close(struct rpc_rqst *rqstp, __be32 *p, struct nfs_clos
 	 */
 	decode_getfattr(&xdr, res->fattr, res->server);
 out:
-        return status;
+	return status;
 }
 
 /*
@@ -4129,23 +3986,23 @@ out:
  */
 static int nfs4_xdr_dec_open(struct rpc_rqst *rqstp, __be32 *p, struct nfs_openres *res)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr;
-        int status;
-
-        xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
-        status = decode_compound_hdr(&xdr, &hdr);
-        if (status)
-                goto out;
-        status = decode_putfh(&xdr);
-        if (status)
-                goto out;
-        status = decode_savefh(&xdr);
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_putfh(&xdr);
+	if (status)
+		goto out;
+	status = decode_savefh(&xdr);
+	if (status)
+		goto out;
+	status = decode_open(&xdr, res);
 	if (status)
 		goto out;
-        status = decode_open(&xdr, res);
-        if (status)
-                goto out;
 	if (decode_getfh(&xdr, &res->fh) != 0)
 		goto out;
 	if (decode_getfattr(&xdr, res->f_attr, res->server) != 0)
@@ -4154,7 +4011,7 @@ static int nfs4_xdr_dec_open(struct rpc_rqst *rqstp, __be32 *p, struct nfs_openr
 		goto out;
 	decode_getfattr(&xdr, res->dir_attr, res->server);
 out:
-        return status;
+	return status;
 }
 
 /*
@@ -4162,20 +4019,20 @@ out:
  */
 static int nfs4_xdr_dec_open_confirm(struct rpc_rqst *rqstp, __be32 *p, struct nfs_open_confirmres *res)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr;
-        int status;
-
-        xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
-        status = decode_compound_hdr(&xdr, &hdr);
-        if (status)
-                goto out;
-        status = decode_putfh(&xdr);
-        if (status)
-                goto out;
-        status = decode_open_confirm(&xdr, res);
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_putfh(&xdr);
+	if (status)
+		goto out;
+	status = decode_open_confirm(&xdr, res);
 out:
-        return status;
+	return status;
 }
 
 /*
@@ -4183,23 +4040,23 @@ out:
  */
 static int nfs4_xdr_dec_open_noattr(struct rpc_rqst *rqstp, __be32 *p, struct nfs_openres *res)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr;
-        int status;
-
-        xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
-        status = decode_compound_hdr(&xdr, &hdr);
-        if (status)
-                goto out;
-        status = decode_putfh(&xdr);
-        if (status)
-                goto out;
-        status = decode_open(&xdr, res);
-        if (status)
-                goto out;
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_putfh(&xdr);
+	if (status)
+		goto out;
+	status = decode_open(&xdr, res);
+	if (status)
+		goto out;
 	decode_getfattr(&xdr, res->f_attr, res->server);
 out:
-        return status;
+	return status;
 }
 
 /*
@@ -4207,25 +4064,25 @@ out:
  */
 static int nfs4_xdr_dec_setattr(struct rpc_rqst *rqstp, __be32 *p, struct nfs_setattrres *res)
 {
-        struct xdr_stream xdr;
-        struct compound_hdr hdr;
-        int status;
-
-        xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
-        status = decode_compound_hdr(&xdr, &hdr);
-        if (status)
-                goto out;
-        status = decode_putfh(&xdr);
-        if (status)
-                goto out;
-        status = decode_setattr(&xdr, res);
-        if (status)
-                goto out;
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status)
+		goto out;
+	status = decode_putfh(&xdr);
+	if (status)
+		goto out;
+	status = decode_setattr(&xdr, res);
+	if (status)
+		goto out;
 	status = decode_getfattr(&xdr, res->fattr, res->server);
 	if (status == NFS4ERR_DELAY)
 		status = 0;
 out:
-        return status;
+	return status;
 }
 
 /*
@@ -4421,8 +4278,6 @@ static int nfs4_xdr_dec_fsinfo(struct rpc_rqst *req, __be32 *p, struct nfs_fsinf
 		status = decode_putfh(&xdr);
 	if (!status)
 		status = decode_fsinfo(&xdr, fsinfo);
-	if (!status)
-		status = nfs4_stat_to_errno(hdr.status);
 	return status;
 }
 
@@ -4511,8 +4366,6 @@ static int nfs4_xdr_dec_setclientid(struct rpc_rqst *req, __be32 *p,
 	status = decode_compound_hdr(&xdr, &hdr);
 	if (!status)
 		status = decode_setclientid(&xdr, clp);
-	if (!status)
-		status = nfs4_stat_to_errno(hdr.status);
 	return status;
 }
 
@@ -4533,8 +4386,6 @@ static int nfs4_xdr_dec_setclientid_confirm(struct rpc_rqst *req, __be32 *p, str
 		status = decode_putrootfh(&xdr);
 	if (!status)
 		status = decode_fsinfo(&xdr, fsinfo);
-	if (!status)
-		status = nfs4_stat_to_errno(hdr.status);
 	return status;
 }
 
@@ -4715,7 +4566,7 @@ nfs4_stat_to_errno(int stat)
 	.p_replen = NFS4_##restype##_sz,			\
 	.p_statidx = NFSPROC4_CLNT_##proc,			\
 	.p_name   = #proc,					\
-    }
+}
 
 struct rpc_procinfo	nfs4_procedures[] = {
   PROC(READ,		enc_read,	dec_read),
diff --git a/fs/nfs/nfsroot.c b/fs/nfs/nfsroot.c
index d74d16ce0d4..d9ef602fbc5 100644
--- a/fs/nfs/nfsroot.c
+++ b/fs/nfs/nfsroot.c
@@ -86,6 +86,8 @@
 #include <net/ipconfig.h>
 #include <linux/parser.h>
 
+#include "internal.h"
+
 /* Define this to allow debugging output */
 #undef NFSROOT_DEBUG
 #define NFSDBG_FACILITY NFSDBG_ROOT
@@ -100,7 +102,7 @@ static char nfs_root_name[256] __initdata = "";
 static __be32 servaddr __initdata = 0;
 
 /* Name of directory to mount */
-static char nfs_path[NFS_MAXPATHLEN] __initdata = { 0, };
+static char nfs_export_path[NFS_MAXPATHLEN] __initdata = { 0, };
 
 /* NFS-related data */
 static struct nfs_mount_data nfs_data __initdata = { 0, };/* NFS mount info */
@@ -312,7 +314,7 @@ static int __init root_nfs_name(char *name)
 		printk(KERN_ERR "Root-NFS: Pathname for remote directory too long.\n");
 		return -1;
 	}
-	sprintf(nfs_path, buf, cp);
+	sprintf(nfs_export_path, buf, cp);
 
 	return 1;
 }
@@ -340,7 +342,7 @@ static int __init root_nfs_addr(void)
 static void __init root_nfs_print(void)
 {
 	printk(KERN_NOTICE "Root-NFS: Mounting %s on server %s as root\n",
-		nfs_path, nfs_data.hostname);
+		nfs_export_path, nfs_data.hostname);
 	printk(KERN_NOTICE "Root-NFS:     rsize = %d, wsize = %d, timeo = %d, retrans = %d\n",
 		nfs_data.rsize, nfs_data.wsize, nfs_data.timeo, nfs_data.retrans);
 	printk(KERN_NOTICE "Root-NFS:     acreg (min,max) = (%d,%d), acdir (min,max) = (%d,%d)\n",
@@ -485,18 +487,23 @@ static int __init root_nfs_get_handle(void)
 {
 	struct nfs_fh fh;
 	struct sockaddr_in sin;
+	struct nfs_mount_request request = {
+		.sap		= (struct sockaddr *)&sin,
+		.salen		= sizeof(sin),
+		.dirpath	= nfs_export_path,
+		.version	= (nfs_data.flags & NFS_MOUNT_VER3) ?
+					NFS_MNT3_VERSION : NFS_MNT_VERSION,
+		.protocol	= (nfs_data.flags & NFS_MOUNT_TCP) ?
+					XPRT_TRANSPORT_TCP : XPRT_TRANSPORT_UDP,
+		.fh		= &fh,
+	};
 	int status;
-	int protocol = (nfs_data.flags & NFS_MOUNT_TCP) ?
-					XPRT_TRANSPORT_TCP : XPRT_TRANSPORT_UDP;
-	int version = (nfs_data.flags & NFS_MOUNT_VER3) ?
-					NFS_MNT3_VERSION : NFS_MNT_VERSION;
 
 	set_sockaddr(&sin, servaddr, htons(mount_port));
-	status = nfs_mount((struct sockaddr *) &sin, sizeof(sin), NULL,
-			   nfs_path, version, protocol, &fh);
+	status = nfs_mount(&request);
 	if (status < 0)
 		printk(KERN_ERR "Root-NFS: Server returned error %d "
-				"while mounting %s\n", status, nfs_path);
+				"while mounting %s\n", status, nfs_export_path);
 	else {
 		nfs_data.root.size = fh.size;
 		memcpy(nfs_data.root.data, fh.data, fh.size);
diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index 40d17987d0e..f856004bb7f 100644
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -533,12 +533,6 @@ readpage_async_filler(void *data, struct page *page)
 	unsigned int len;
 	int error;
 
-	error = nfs_wb_page(inode, page);
-	if (error)
-		goto out_unlock;
-	if (PageUptodate(page))
-		goto out_unlock;
-
 	len = nfs_page_length(page);
 	if (len == 0)
 		return nfs_return_empty_page(page);
diff --git a/fs/nfs/super.c b/fs/nfs/super.c
index bb0313ac9e1..d6686f4786d 100644
--- a/fs/nfs/super.c
+++ b/fs/nfs/super.c
@@ -75,6 +75,7 @@ enum {
 	Opt_acl, Opt_noacl,
 	Opt_rdirplus, Opt_nordirplus,
 	Opt_sharecache, Opt_nosharecache,
+	Opt_resvport, Opt_noresvport,
 
 	/* Mount options that take integer arguments */
 	Opt_port,
@@ -129,6 +130,8 @@ static const match_table_t nfs_mount_option_tokens = {
 	{ Opt_nordirplus, "nordirplus" },
 	{ Opt_sharecache, "sharecache" },
 	{ Opt_nosharecache, "nosharecache" },
+	{ Opt_resvport, "resvport" },
+	{ Opt_noresvport, "noresvport" },
 
 	{ Opt_port, "port=%u" },
 	{ Opt_rsize, "rsize=%u" },
@@ -512,7 +515,8 @@ static void nfs_show_mount_options(struct seq_file *m, struct nfs_server *nfss,
 		{ NFS_MOUNT_NONLM, ",nolock", "" },
 		{ NFS_MOUNT_NOACL, ",noacl", "" },
 		{ NFS_MOUNT_NORDIRPLUS, ",nordirplus", "" },
-		{ NFS_MOUNT_UNSHARED, ",nosharecache", ""},
+		{ NFS_MOUNT_UNSHARED, ",nosharecache", "" },
+		{ NFS_MOUNT_NORESVPORT, ",noresvport", "" },
 		{ 0, NULL, NULL }
 	};
 	const struct proc_nfs_info *nfs_infop;
@@ -1033,6 +1037,12 @@ static int nfs_parse_mount_options(char *raw,
 		case Opt_nosharecache:
 			mnt->flags |= NFS_MOUNT_UNSHARED;
 			break;
+		case Opt_resvport:
+			mnt->flags &= ~NFS_MOUNT_NORESVPORT;
+			break;
+		case Opt_noresvport:
+			mnt->flags |= NFS_MOUNT_NORESVPORT;
+			break;
 
 		/*
 		 * options that take numeric values
@@ -1327,8 +1337,14 @@ out_security_failure:
 static int nfs_try_mount(struct nfs_parsed_mount_data *args,
 			 struct nfs_fh *root_fh)
 {
-	struct sockaddr *sap = (struct sockaddr *)&args->mount_server.address;
-	char *hostname;
+	struct nfs_mount_request request = {
+		.sap		= (struct sockaddr *)
+						&args->mount_server.address,
+		.dirpath	= args->nfs_server.export_path,
+		.protocol	= args->mount_server.protocol,
+		.fh		= root_fh,
+		.noresvport	= args->flags & NFS_MOUNT_NORESVPORT,
+	};
 	int status;
 
 	if (args->mount_server.version == 0) {
@@ -1337,42 +1353,38 @@ static int nfs_try_mount(struct nfs_parsed_mount_data *args,
 		else
 			args->mount_server.version = NFS_MNT_VERSION;
 	}
+	request.version = args->mount_server.version;
 
 	if (args->mount_server.hostname)
-		hostname = args->mount_server.hostname;
+		request.hostname = args->mount_server.hostname;
 	else
-		hostname = args->nfs_server.hostname;
+		request.hostname = args->nfs_server.hostname;
 
 	/*
 	 * Construct the mount server's address.
 	 */
 	if (args->mount_server.address.ss_family == AF_UNSPEC) {
-		memcpy(sap, &args->nfs_server.address,
+		memcpy(request.sap, &args->nfs_server.address,
 		       args->nfs_server.addrlen);
 		args->mount_server.addrlen = args->nfs_server.addrlen;
 	}
+	request.salen = args->mount_server.addrlen;
 
 	/*
 	 * autobind will be used if mount_server.port == 0
 	 */
-	nfs_set_port(sap, args->mount_server.port);
+	nfs_set_port(request.sap, args->mount_server.port);
 
 	/*
 	 * Now ask the mount server to map our export path
 	 * to a file handle.
 	 */
-	status = nfs_mount(sap,
-			   args->mount_server.addrlen,
-			   hostname,
-			   args->nfs_server.export_path,
-			   args->mount_server.version,
-			   args->mount_server.protocol,
-			   root_fh);
+	status = nfs_mount(&request);
 	if (status == 0)
 		return 0;
 
 	dfprintk(MOUNT, "NFS: unable to mount server %s, error %d\n",
-			hostname, status);
+			request.hostname, status);
 	return status;
 }
 
@@ -2419,7 +2431,7 @@ static void nfs4_kill_super(struct super_block *sb)
 {
 	struct nfs_server *server = NFS_SB(sb);
 
-	nfs_return_all_delegations(sb);
+	nfs_super_return_all_delegations(sb);
 	kill_anon_super(sb);
 
 	nfs4_renewd_prepare_shutdown(server);
diff --git a/fs/nfs_common/nfsacl.c b/fs/nfs_common/nfsacl.c
index c11f5375d7c..04133aacb1e 100644
--- a/fs/nfs_common/nfsacl.c
+++ b/fs/nfs_common/nfsacl.c
@@ -29,8 +29,8 @@
 
 MODULE_LICENSE("GPL");
 
-EXPORT_SYMBOL(nfsacl_encode);
-EXPORT_SYMBOL(nfsacl_decode);
+EXPORT_SYMBOL_GPL(nfsacl_encode);
+EXPORT_SYMBOL_GPL(nfsacl_decode);
 
 struct nfsacl_encode_desc {
 	struct xdr_array2_desc desc;
diff --git a/fs/nfsd/nfs4callback.c b/fs/nfsd/nfs4callback.c
index 094747a1227..6d7d8c02c19 100644
--- a/fs/nfsd/nfs4callback.c
+++ b/fs/nfsd/nfs4callback.c
@@ -358,6 +358,7 @@ static struct rpc_program cb_program = {
 		.nrvers		= ARRAY_SIZE(nfs_cb_version),
 		.version	= nfs_cb_version,
 		.stats		= &cb_stats,
+		.pipe_dir_name  = "/nfsd4_cb",
 };
 
 /* Reference counting, callback cleanup, etc., all look racy as heck.
@@ -382,8 +383,9 @@ static int do_probe_callback(void *data)
 		.program	= &cb_program,
 		.prognumber	= cb->cb_prog,
 		.version	= nfs_cb_version[1]->number,
-		.authflavor	= RPC_AUTH_UNIX, /* XXX: need AUTH_GSS... */
+		.authflavor	= clp->cl_flavor,
 		.flags		= (RPC_CLNT_CREATE_NOPING | RPC_CLNT_CREATE_QUIET),
+		.client_name    = clp->cl_principal,
 	};
 	struct rpc_message msg = {
 		.rpc_proc       = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
@@ -392,6 +394,11 @@ static int do_probe_callback(void *data)
 	struct rpc_clnt *client;
 	int status;
 
+	if (!clp->cl_principal && (clp->cl_flavor >= RPC_AUTH_GSS_KRB5)) {
+		status = nfserr_cb_path_down;
+		goto out_err;
+	}
+
 	/* Initialize address */
 	memset(&addr, 0, sizeof(addr));
 	addr.sin_family = AF_INET;
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index bf4cd46a5a1..13e0e074dbb 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -54,6 +54,7 @@
 #include <linux/mutex.h>
 #include <linux/lockd/bind.h>
 #include <linux/module.h>
+#include <linux/sunrpc/svcauth_gss.h>
 
 #define NFSDDBG_FACILITY                NFSDDBG_PROC
 
@@ -377,6 +378,7 @@ free_client(struct nfs4_client *clp)
 	shutdown_callback_client(clp);
 	if (clp->cl_cred.cr_group_info)
 		put_group_info(clp->cl_cred.cr_group_info);
+	kfree(clp->cl_principal);
 	kfree(clp->cl_name.data);
 	kfree(clp);
 }
@@ -696,6 +698,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	unsigned int 		strhashval;
 	struct nfs4_client	*conf, *unconf, *new;
 	__be32 			status;
+	char			*princ;
 	char                    dname[HEXDIR_LEN];
 	
 	if (!check_name(clname))
@@ -783,6 +786,15 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	}
 	copy_verf(new, &clverifier);
 	new->cl_addr = sin->sin_addr.s_addr;
+	new->cl_flavor = rqstp->rq_flavor;
+	princ = svc_gss_principal(rqstp);
+	if (princ) {
+		new->cl_principal = kstrdup(princ, GFP_KERNEL);
+		if (new->cl_principal == NULL) {
+			free_client(new);
+			goto out;
+		}
+	}
 	copy_cred(&new->cl_cred, &rqstp->rq_cred);
 	gen_confirm(new);
 	gen_callback(new, setclid);
diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 81904f07679..3bb1cf1e742 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -44,10 +44,13 @@ static int show_stat(struct seq_file *p, void *v)
 		softirq = cputime64_add(softirq, kstat_cpu(i).cpustat.softirq);
 		steal = cputime64_add(steal, kstat_cpu(i).cpustat.steal);
 		guest = cputime64_add(guest, kstat_cpu(i).cpustat.guest);
-
-		for_each_irq_nr(j)
+		for_each_irq_nr(j) {
+#ifdef CONFIG_SPARSE_IRQ
+			if (!irq_to_desc(j))
+				continue;
+#endif
 			sum += kstat_irqs_cpu(j, i);
-
+		}
 		sum += arch_irq_stat_cpu(i);
 	}
 	sum += arch_irq_stat();
@@ -92,7 +95,12 @@ static int show_stat(struct seq_file *p, void *v)
 	/* sum again ? it could be updated? */
 	for_each_irq_nr(j) {
 		per_irq_sum = 0;
-
+#ifdef CONFIG_SPARSE_IRQ
+		if (!irq_to_desc(j)) {
+			seq_printf(p, " %u", per_irq_sum);
+			continue;
+		}
+#endif
 		for_each_possible_cpu(i)
 			per_irq_sum += kstat_irqs_cpu(j, i);
 
diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 737c9a42536..c3dc491fff8 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -85,13 +85,13 @@ xfs-y				+= xfs_alloc.o \
 				   xfs_trans_inode.o \
 				   xfs_trans_item.o \
 				   xfs_utils.o \
-				   xfs_vfsops.o \
 				   xfs_vnodeops.o \
 				   xfs_rw.o \
 				   xfs_dmops.o \
 				   xfs_qmops.o
 
-xfs-$(CONFIG_XFS_TRACE)		+= xfs_dir2_trace.o
+xfs-$(CONFIG_XFS_TRACE)		+= xfs_btree_trace.o \
+				   xfs_dir2_trace.o
 
 # Objects in linux/
 xfs-y				+= $(addprefix $(XFS_LINUX)/, \
@@ -106,7 +106,7 @@ xfs-y				+= $(addprefix $(XFS_LINUX)/, \
 				   xfs_iops.o \
 				   xfs_lrw.o \
 				   xfs_super.o \
-				   xfs_vnode.o \
+				   xfs_sync.o \
 				   xfs_xattr.o)
 
 # Objects in support/
diff --git a/fs/xfs/linux-2.6/sv.h b/fs/xfs/linux-2.6/sv.h
index 351a8f454bd..4dfc7c37081 100644
--- a/fs/xfs/linux-2.6/sv.h
+++ b/fs/xfs/linux-2.6/sv.h
@@ -32,23 +32,15 @@ typedef struct sv_s {
 	wait_queue_head_t waiters;
 } sv_t;
 
-#define SV_FIFO		0x0		/* sv_t is FIFO type */
-#define SV_LIFO		0x2		/* sv_t is LIFO type */
-#define SV_PRIO		0x4		/* sv_t is PRIO type */
-#define SV_KEYED	0x6		/* sv_t is KEYED type */
-#define SV_DEFAULT      SV_FIFO
-
-
-static inline void _sv_wait(sv_t *sv, spinlock_t *lock, int state,
-			     unsigned long timeout)
+static inline void _sv_wait(sv_t *sv, spinlock_t *lock)
 {
 	DECLARE_WAITQUEUE(wait, current);
 
 	add_wait_queue_exclusive(&sv->waiters, &wait);
-	__set_current_state(state);
+	__set_current_state(TASK_UNINTERRUPTIBLE);
 	spin_unlock(lock);
 
-	schedule_timeout(timeout);
+	schedule();
 
 	remove_wait_queue(&sv->waiters, &wait);
 }
@@ -58,13 +50,7 @@ static inline void _sv_wait(sv_t *sv, spinlock_t *lock, int state,
 #define sv_destroy(sv) \
 	/*NOTHING*/
 #define sv_wait(sv, pri, lock, s) \
-	_sv_wait(sv, lock, TASK_UNINTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT)
-#define sv_wait_sig(sv, pri, lock, s)   \
-	_sv_wait(sv, lock, TASK_INTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT)
-#define sv_timedwait(sv, pri, lock, s, svf, ts, rts) \
-	_sv_wait(sv, lock, TASK_UNINTERRUPTIBLE, timespec_to_jiffies(ts))
-#define sv_timedwait_sig(sv, pri, lock, s, svf, ts, rts) \
-	_sv_wait(sv, lock, TASK_INTERRUPTIBLE, timespec_to_jiffies(ts))
+	_sv_wait(sv, lock)
 #define sv_signal(sv) \
 	wake_up(&(sv)->waiters)
 #define sv_broadcast(sv) \
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index a44d68eb50b..de3a198f771 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -42,6 +42,40 @@
 #include <linux/pagevec.h>
 #include <linux/writeback.h>
 
+
+/*
+ * Prime number of hash buckets since address is used as the key.
+ */
+#define NVSYNC		37
+#define to_ioend_wq(v)	(&xfs_ioend_wq[((unsigned long)v) % NVSYNC])
+static wait_queue_head_t xfs_ioend_wq[NVSYNC];
+
+void __init
+xfs_ioend_init(void)
+{
+	int i;
+
+	for (i = 0; i < NVSYNC; i++)
+		init_waitqueue_head(&xfs_ioend_wq[i]);
+}
+
+void
+xfs_ioend_wait(
+	xfs_inode_t	*ip)
+{
+	wait_queue_head_t *wq = to_ioend_wq(ip);
+
+	wait_event(*wq, (atomic_read(&ip->i_iocount) == 0));
+}
+
+STATIC void
+xfs_ioend_wake(
+	xfs_inode_t	*ip)
+{
+	if (atomic_dec_and_test(&ip->i_iocount))
+		wake_up(to_ioend_wq(ip));
+}
+
 STATIC void
 xfs_count_page_state(
 	struct page		*page,
@@ -146,16 +180,25 @@ xfs_destroy_ioend(
 	xfs_ioend_t		*ioend)
 {
 	struct buffer_head	*bh, *next;
+	struct xfs_inode	*ip = XFS_I(ioend->io_inode);
 
 	for (bh = ioend->io_buffer_head; bh; bh = next) {
 		next = bh->b_private;
 		bh->b_end_io(bh, !ioend->io_error);
 	}
-	if (unlikely(ioend->io_error)) {
-		vn_ioerror(XFS_I(ioend->io_inode), ioend->io_error,
-				__FILE__,__LINE__);
+
+	/*
+	 * Volume managers supporting multiple paths can send back ENODEV
+	 * when the final path disappears.  In this case continuing to fill
+	 * the page cache with dirty data which cannot be written out is
+	 * evil, so prevent that.
+	 */
+	if (unlikely(ioend->io_error == -ENODEV)) {
+		xfs_do_force_shutdown(ip->i_mount, SHUTDOWN_DEVICE_REQ,
+				      __FILE__, __LINE__);
 	}
-	vn_iowake(XFS_I(ioend->io_inode));
+
+	xfs_ioend_wake(ip);
 	mempool_free(ioend, xfs_ioend_pool);
 }
 
@@ -191,7 +234,7 @@ xfs_setfilesize(
 		ip->i_d.di_size = isize;
 		ip->i_update_core = 1;
 		ip->i_update_size = 1;
-		mark_inode_dirty_sync(ioend->io_inode);
+		xfs_mark_inode_dirty_sync(ip);
 	}
 
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
@@ -317,14 +360,9 @@ xfs_map_blocks(
 	xfs_iomap_t		*mapp,
 	int			flags)
 {
-	xfs_inode_t		*ip = XFS_I(inode);
-	int			error, nmaps = 1;
-
-	error = xfs_iomap(ip, offset, count,
-				flags, mapp, &nmaps);
-	if (!error && (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)))
-		xfs_iflags_set(ip, XFS_IMODIFIED);
-	return -error;
+	int			nmaps = 1;
+
+	return -xfs_iomap(XFS_I(inode), offset, count, flags, mapp, &nmaps);
 }
 
 STATIC_INLINE int
@@ -512,7 +550,7 @@ xfs_cancel_ioend(
 			unlock_buffer(bh);
 		} while ((bh = next_bh) != NULL);
 
-		vn_iowake(XFS_I(ioend->io_inode));
+		xfs_ioend_wake(XFS_I(ioend->io_inode));
 		mempool_free(ioend, xfs_ioend_pool);
 	} while ((ioend = next) != NULL);
 }
diff --git a/fs/xfs/linux-2.6/xfs_aops.h b/fs/xfs/linux-2.6/xfs_aops.h
index 3ba0631a381..7b26f5ff969 100644
--- a/fs/xfs/linux-2.6/xfs_aops.h
+++ b/fs/xfs/linux-2.6/xfs_aops.h
@@ -43,4 +43,7 @@ typedef struct xfs_ioend {
 extern const struct address_space_operations xfs_address_space_operations;
 extern int xfs_get_blocks(struct inode *, sector_t, struct buffer_head *, int);
 
+extern void xfs_ioend_init(void);
+extern void xfs_ioend_wait(struct xfs_inode *);
+
 #endif /* __XFS_AOPS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c
index 36d5fcd3f59..cb329edc925 100644
--- a/fs/xfs/linux-2.6/xfs_buf.c
+++ b/fs/xfs/linux-2.6/xfs_buf.c
@@ -630,6 +630,29 @@ xfs_buf_get_flags(
 	return NULL;
 }
 
+STATIC int
+_xfs_buf_read(
+	xfs_buf_t		*bp,
+	xfs_buf_flags_t		flags)
+{
+	int			status;
+
+	XB_TRACE(bp, "_xfs_buf_read", (unsigned long)flags);
+
+	ASSERT(!(flags & (XBF_DELWRI|XBF_WRITE)));
+	ASSERT(bp->b_bn != XFS_BUF_DADDR_NULL);
+
+	bp->b_flags &= ~(XBF_WRITE | XBF_ASYNC | XBF_DELWRI | \
+			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+	bp->b_flags |= flags & (XBF_READ | XBF_ASYNC | \
+			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+
+	status = xfs_buf_iorequest(bp);
+	if (!status && !(flags & XBF_ASYNC))
+		status = xfs_buf_iowait(bp);
+	return status;
+}
+
 xfs_buf_t *
 xfs_buf_read_flags(
 	xfs_buftarg_t		*target,
@@ -646,7 +669,7 @@ xfs_buf_read_flags(
 		if (!XFS_BUF_ISDONE(bp)) {
 			XB_TRACE(bp, "read", (unsigned long)flags);
 			XFS_STATS_INC(xb_get_read);
-			xfs_buf_iostart(bp, flags);
+			_xfs_buf_read(bp, flags);
 		} else if (flags & XBF_ASYNC) {
 			XB_TRACE(bp, "read_async", (unsigned long)flags);
 			/*
@@ -1048,50 +1071,39 @@ xfs_buf_ioerror(
 	XB_TRACE(bp, "ioerror", (unsigned long)error);
 }
 
-/*
- *	Initiate I/O on a buffer, based on the flags supplied.
- *	The b_iodone routine in the buffer supplied will only be called
- *	when all of the subsidiary I/O requests, if any, have been completed.
- */
 int
-xfs_buf_iostart(
-	xfs_buf_t		*bp,
-	xfs_buf_flags_t		flags)
+xfs_bawrite(
+	void			*mp,
+	struct xfs_buf		*bp)
 {
-	int			status = 0;
+	XB_TRACE(bp, "bawrite", 0);
 
-	XB_TRACE(bp, "iostart", (unsigned long)flags);
+	ASSERT(bp->b_bn != XFS_BUF_DADDR_NULL);
 
-	if (flags & XBF_DELWRI) {
-		bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_ASYNC);
-		bp->b_flags |= flags & (XBF_DELWRI | XBF_ASYNC);
-		xfs_buf_delwri_queue(bp, 1);
-		return 0;
-	}
+	xfs_buf_delwri_dequeue(bp);
 
-	bp->b_flags &= ~(XBF_READ | XBF_WRITE | XBF_ASYNC | XBF_DELWRI | \
-			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
-	bp->b_flags |= flags & (XBF_READ | XBF_WRITE | XBF_ASYNC | \
-			XBF_READ_AHEAD | _XBF_RUN_QUEUES);
+	bp->b_flags &= ~(XBF_READ | XBF_DELWRI | XBF_READ_AHEAD);
+	bp->b_flags |= (XBF_WRITE | XBF_ASYNC | _XBF_RUN_QUEUES);
+
+	bp->b_mount = mp;
+	bp->b_strat = xfs_bdstrat_cb;
+	return xfs_bdstrat_cb(bp);
+}
 
-	BUG_ON(bp->b_bn == XFS_BUF_DADDR_NULL);
+void
+xfs_bdwrite(
+	void			*mp,
+	struct xfs_buf		*bp)
+{
+	XB_TRACE(bp, "bdwrite", 0);
 
-	/* For writes allow an alternate strategy routine to precede
-	 * the actual I/O request (which may not be issued at all in
-	 * a shutdown situation, for example).
-	 */
-	status = (flags & XBF_WRITE) ?
-		xfs_buf_iostrategy(bp) : xfs_buf_iorequest(bp);
+	bp->b_strat = xfs_bdstrat_cb;
+	bp->b_mount = mp;
 
-	/* Wait for I/O if we are not an async request.
-	 * Note: async I/O request completion will release the buffer,
-	 * and that can already be done by this point.  So using the
-	 * buffer pointer from here on, after async I/O, is invalid.
-	 */
-	if (!status && !(flags & XBF_ASYNC))
-		status = xfs_buf_iowait(bp);
+	bp->b_flags &= ~XBF_READ;
+	bp->b_flags |= (XBF_DELWRI | XBF_ASYNC);
 
-	return status;
+	xfs_buf_delwri_queue(bp, 1);
 }
 
 STATIC_INLINE void
@@ -1114,8 +1126,7 @@ xfs_buf_bio_end_io(
 	unsigned int		blocksize = bp->b_target->bt_bsize;
 	struct bio_vec		*bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
 
-	if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
-		bp->b_error = EIO;
+	xfs_buf_ioerror(bp, -error);
 
 	do {
 		struct page	*page = bvec->bv_page;
diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h
index 456519a088c..288ae7c4c80 100644
--- a/fs/xfs/linux-2.6/xfs_buf.h
+++ b/fs/xfs/linux-2.6/xfs_buf.h
@@ -168,7 +168,7 @@ typedef struct xfs_buf {
 	struct completion	b_iowait;	/* queue for I/O waiters */
 	void			*b_fspriv;
 	void			*b_fspriv2;
-	void			*b_fspriv3;
+	struct xfs_mount	*b_mount;
 	unsigned short		b_error;	/* error code on I/O */
 	unsigned int		b_page_count;	/* size of page array */
 	unsigned int		b_offset;	/* page offset in first page */
@@ -214,9 +214,10 @@ extern void xfs_buf_lock(xfs_buf_t *);
 extern void xfs_buf_unlock(xfs_buf_t *);
 
 /* Buffer Read and Write Routines */
+extern int xfs_bawrite(void *mp, xfs_buf_t *bp);
+extern void xfs_bdwrite(void *mp, xfs_buf_t *bp);
 extern void xfs_buf_ioend(xfs_buf_t *,	int);
 extern void xfs_buf_ioerror(xfs_buf_t *, int);
-extern int xfs_buf_iostart(xfs_buf_t *, xfs_buf_flags_t);
 extern int xfs_buf_iorequest(xfs_buf_t *);
 extern int xfs_buf_iowait(xfs_buf_t *);
 extern void xfs_buf_iomove(xfs_buf_t *, size_t, size_t, xfs_caddr_t,
@@ -311,10 +312,6 @@ extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
 #define XFS_BUF_UNORDERED(bp)	((bp)->b_flags &= ~XBF_ORDERED)
 #define XFS_BUF_ISORDERED(bp)	((bp)->b_flags & XBF_ORDERED)
 
-#define XFS_BUF_SHUT(bp)	do { } while (0)
-#define XFS_BUF_UNSHUT(bp)	do { } while (0)
-#define XFS_BUF_ISSHUT(bp)	(0)
-
 #define XFS_BUF_HOLD(bp)	xfs_buf_hold(bp)
 #define XFS_BUF_READ(bp)	((bp)->b_flags |= XBF_READ)
 #define XFS_BUF_UNREAD(bp)	((bp)->b_flags &= ~XBF_READ)
@@ -334,8 +331,6 @@ extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
 #define XFS_BUF_SET_FSPRIVATE(bp, val)		((bp)->b_fspriv = (void*)(val))
 #define XFS_BUF_FSPRIVATE2(bp, type)		((type)(bp)->b_fspriv2)
 #define XFS_BUF_SET_FSPRIVATE2(bp, val)		((bp)->b_fspriv2 = (void*)(val))
-#define XFS_BUF_FSPRIVATE3(bp, type)		((type)(bp)->b_fspriv3)
-#define XFS_BUF_SET_FSPRIVATE3(bp, val)		((bp)->b_fspriv3 = (void*)(val))
 #define XFS_BUF_SET_START(bp)			do { } while (0)
 #define XFS_BUF_SET_BRELSE_FUNC(bp, func)	((bp)->b_relse = (func))
 
@@ -366,14 +361,6 @@ extern void xfs_buf_trace(xfs_buf_t *, char *, void *, void *);
 #define XFS_BUF_TARGET(bp)		((bp)->b_target)
 #define XFS_BUFTARG_NAME(target)	xfs_buf_target_name(target)
 
-static inline int xfs_bawrite(void *mp, xfs_buf_t *bp)
-{
-	bp->b_fspriv3 = mp;
-	bp->b_strat = xfs_bdstrat_cb;
-	xfs_buf_delwri_dequeue(bp);
-	return xfs_buf_iostart(bp, XBF_WRITE | XBF_ASYNC | _XBF_RUN_QUEUES);
-}
-
 static inline void xfs_buf_relse(xfs_buf_t *bp)
 {
 	if (!bp->b_relse)
@@ -414,17 +401,6 @@ static inline int XFS_bwrite(xfs_buf_t *bp)
 	return error;
 }
 
-/*
- * No error can be returned from xfs_buf_iostart for delwri
- * buffers as they are queued and no I/O is issued.
- */
-static inline void xfs_bdwrite(void *mp, xfs_buf_t *bp)
-{
-	bp->b_strat = xfs_bdstrat_cb;
-	bp->b_fspriv3 = mp;
-	(void)xfs_buf_iostart(bp, XBF_DELWRI | XBF_ASYNC);
-}
-
 #define XFS_bdstrat(bp) xfs_buf_iorequest(bp)
 
 #define xfs_iowait(bp)	xfs_buf_iowait(bp)
diff --git a/fs/xfs/linux-2.6/xfs_cred.h b/fs/xfs/linux-2.6/xfs_cred.h
index 8c022cd0ad6..55bddf3b609 100644
--- a/fs/xfs/linux-2.6/xfs_cred.h
+++ b/fs/xfs/linux-2.6/xfs_cred.h
@@ -25,12 +25,4 @@
  */
 typedef const struct cred cred_t;
 
-extern cred_t *sys_cred;
-
-/* this is a hack.. (assumes sys_cred is the only cred_t in the system) */
-static inline int capable_cred(cred_t *cr, int cid)
-{
-	return (cr == sys_cred) ? 1 : capable(cid);
-}
-
 #endif  /* __XFS_CRED_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_export.c b/fs/xfs/linux-2.6/xfs_export.c
index 7f7abec25e1..595751f7835 100644
--- a/fs/xfs/linux-2.6/xfs_export.c
+++ b/fs/xfs/linux-2.6/xfs_export.c
@@ -29,7 +29,6 @@
 #include "xfs_vnodeops.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_inode.h"
-#include "xfs_vfsops.h"
 
 /*
  * Note that we only accept fileids which are long enough rather than allow
diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index 3fee790f138..e14c4e3aea0 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -36,89 +36,54 @@
 #include "xfs_inode.h"
 #include "xfs_error.h"
 #include "xfs_rw.h"
-#include "xfs_ioctl32.h"
 #include "xfs_vnodeops.h"
+#include "xfs_da_btree.h"
+#include "xfs_ioctl.h"
 
 #include <linux/dcache.h>
 #include <linux/smp_lock.h>
 
 static struct vm_operations_struct xfs_file_vm_ops;
 
-STATIC_INLINE ssize_t
-__xfs_file_read(
+STATIC ssize_t
+xfs_file_aio_read(
 	struct kiocb		*iocb,
 	const struct iovec	*iov,
 	unsigned long		nr_segs,
-	int			ioflags,
 	loff_t			pos)
 {
 	struct file		*file = iocb->ki_filp;
+	int			ioflags = IO_ISAIO;
 
 	BUG_ON(iocb->ki_pos != pos);
 	if (unlikely(file->f_flags & O_DIRECT))
 		ioflags |= IO_ISDIRECT;
+	if (file->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
 	return xfs_read(XFS_I(file->f_path.dentry->d_inode), iocb, iov,
 				nr_segs, &iocb->ki_pos, ioflags);
 }
 
 STATIC ssize_t
-xfs_file_aio_read(
-	struct kiocb		*iocb,
-	const struct iovec	*iov,
-	unsigned long		nr_segs,
-	loff_t			pos)
-{
-	return __xfs_file_read(iocb, iov, nr_segs, IO_ISAIO, pos);
-}
-
-STATIC ssize_t
-xfs_file_aio_read_invis(
-	struct kiocb		*iocb,
-	const struct iovec	*iov,
-	unsigned long		nr_segs,
-	loff_t			pos)
-{
-	return __xfs_file_read(iocb, iov, nr_segs, IO_ISAIO|IO_INVIS, pos);
-}
-
-STATIC_INLINE ssize_t
-__xfs_file_write(
+xfs_file_aio_write(
 	struct kiocb		*iocb,
 	const struct iovec	*iov,
 	unsigned long		nr_segs,
-	int			ioflags,
 	loff_t			pos)
 {
-	struct file	*file = iocb->ki_filp;
+	struct file		*file = iocb->ki_filp;
+	int			ioflags = IO_ISAIO;
 
 	BUG_ON(iocb->ki_pos != pos);
 	if (unlikely(file->f_flags & O_DIRECT))
 		ioflags |= IO_ISDIRECT;
+	if (file->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
 	return xfs_write(XFS_I(file->f_mapping->host), iocb, iov, nr_segs,
 				&iocb->ki_pos, ioflags);
 }
 
 STATIC ssize_t
-xfs_file_aio_write(
-	struct kiocb		*iocb,
-	const struct iovec	*iov,
-	unsigned long		nr_segs,
-	loff_t			pos)
-{
-	return __xfs_file_write(iocb, iov, nr_segs, IO_ISAIO, pos);
-}
-
-STATIC ssize_t
-xfs_file_aio_write_invis(
-	struct kiocb		*iocb,
-	const struct iovec	*iov,
-	unsigned long		nr_segs,
-	loff_t			pos)
-{
-	return __xfs_file_write(iocb, iov, nr_segs, IO_ISAIO|IO_INVIS, pos);
-}
-
-STATIC ssize_t
 xfs_file_splice_read(
 	struct file		*infilp,
 	loff_t			*ppos,
@@ -126,20 +91,13 @@ xfs_file_splice_read(
 	size_t			len,
 	unsigned int		flags)
 {
-	return xfs_splice_read(XFS_I(infilp->f_path.dentry->d_inode),
-				   infilp, ppos, pipe, len, flags, 0);
-}
+	int			ioflags = 0;
+
+	if (infilp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
 
-STATIC ssize_t
-xfs_file_splice_read_invis(
-	struct file		*infilp,
-	loff_t			*ppos,
-	struct pipe_inode_info	*pipe,
-	size_t			len,
-	unsigned int		flags)
-{
 	return xfs_splice_read(XFS_I(infilp->f_path.dentry->d_inode),
-				   infilp, ppos, pipe, len, flags, IO_INVIS);
+				   infilp, ppos, pipe, len, flags, ioflags);
 }
 
 STATIC ssize_t
@@ -150,30 +108,49 @@ xfs_file_splice_write(
 	size_t			len,
 	unsigned int		flags)
 {
-	return xfs_splice_write(XFS_I(outfilp->f_path.dentry->d_inode),
-				    pipe, outfilp, ppos, len, flags, 0);
-}
+	int			ioflags = 0;
+
+	if (outfilp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
 
-STATIC ssize_t
-xfs_file_splice_write_invis(
-	struct pipe_inode_info	*pipe,
-	struct file		*outfilp,
-	loff_t			*ppos,
-	size_t			len,
-	unsigned int		flags)
-{
 	return xfs_splice_write(XFS_I(outfilp->f_path.dentry->d_inode),
-				    pipe, outfilp, ppos, len, flags, IO_INVIS);
+				    pipe, outfilp, ppos, len, flags, ioflags);
 }
 
 STATIC int
 xfs_file_open(
 	struct inode	*inode,
-	struct file	*filp)
+	struct file	*file)
 {
-	if (!(filp->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
+	if (!(file->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
 		return -EFBIG;
-	return -xfs_open(XFS_I(inode));
+	if (XFS_FORCED_SHUTDOWN(XFS_M(inode->i_sb)))
+		return -EIO;
+	return 0;
+}
+
+STATIC int
+xfs_dir_open(
+	struct inode	*inode,
+	struct file	*file)
+{
+	struct xfs_inode *ip = XFS_I(inode);
+	int		mode;
+	int		error;
+
+	error = xfs_file_open(inode, file);
+	if (error)
+		return error;
+
+	/*
+	 * If there are any blocks, read-ahead block 0 as we're almost
+	 * certain to have the next operation be a read there.
+	 */
+	mode = xfs_ilock_map_shared(ip);
+	if (ip->i_d.di_nextents > 0)
+		xfs_da_reada_buf(NULL, ip, 0, XFS_DATA_FORK);
+	xfs_iunlock(ip, mode);
+	return 0;
 }
 
 STATIC int
@@ -227,7 +204,7 @@ xfs_file_readdir(
 	 * point we can change the ->readdir prototype to include the
 	 * buffer size.
 	 */
-	bufsize = (size_t)min_t(loff_t, PAGE_SIZE, inode->i_size);
+	bufsize = (size_t)min_t(loff_t, PAGE_SIZE, ip->i_d.di_size);
 
 	error = xfs_readdir(ip, dirent, bufsize,
 				(xfs_off_t *)&filp->f_pos, filldir);
@@ -248,48 +225,6 @@ xfs_file_mmap(
 	return 0;
 }
 
-STATIC long
-xfs_file_ioctl(
-	struct file	*filp,
-	unsigned int	cmd,
-	unsigned long	p)
-{
-	int		error;
-	struct inode	*inode = filp->f_path.dentry->d_inode;
-
-	error = xfs_ioctl(XFS_I(inode), filp, 0, cmd, (void __user *)p);
-	xfs_iflags_set(XFS_I(inode), XFS_IMODIFIED);
-
-	/* NOTE:  some of the ioctl's return positive #'s as a
-	 *	  byte count indicating success, such as
-	 *	  readlink_by_handle.  So we don't "sign flip"
-	 *	  like most other routines.  This means true
-	 *	  errors need to be returned as a negative value.
-	 */
-	return error;
-}
-
-STATIC long
-xfs_file_ioctl_invis(
-	struct file	*filp,
-	unsigned int	cmd,
-	unsigned long	p)
-{
-	int		error;
-	struct inode	*inode = filp->f_path.dentry->d_inode;
-
-	error = xfs_ioctl(XFS_I(inode), filp, IO_INVIS, cmd, (void __user *)p);
-	xfs_iflags_set(XFS_I(inode), XFS_IMODIFIED);
-
-	/* NOTE:  some of the ioctl's return positive #'s as a
-	 *	  byte count indicating success, such as
-	 *	  readlink_by_handle.  So we don't "sign flip"
-	 *	  like most other routines.  This means true
-	 *	  errors need to be returned as a negative value.
-	 */
-	return error;
-}
-
 /*
  * mmap()d file has taken write protection fault and is being made
  * writable. We can set the page state up correctly for a writable
@@ -325,26 +260,8 @@ const struct file_operations xfs_file_operations = {
 #endif
 };
 
-const struct file_operations xfs_invis_file_operations = {
-	.llseek		= generic_file_llseek,
-	.read		= do_sync_read,
-	.write		= do_sync_write,
-	.aio_read	= xfs_file_aio_read_invis,
-	.aio_write	= xfs_file_aio_write_invis,
-	.splice_read	= xfs_file_splice_read_invis,
-	.splice_write	= xfs_file_splice_write_invis,
-	.unlocked_ioctl	= xfs_file_ioctl_invis,
-#ifdef CONFIG_COMPAT
-	.compat_ioctl	= xfs_file_compat_invis_ioctl,
-#endif
-	.mmap		= xfs_file_mmap,
-	.open		= xfs_file_open,
-	.release	= xfs_file_release,
-	.fsync		= xfs_file_fsync,
-};
-
-
 const struct file_operations xfs_dir_file_operations = {
+	.open		= xfs_dir_open,
 	.read		= generic_read_dir,
 	.readdir	= xfs_file_readdir,
 	.llseek		= generic_file_llseek,
diff --git a/fs/xfs/linux-2.6/xfs_fs_subr.c b/fs/xfs/linux-2.6/xfs_fs_subr.c
index 36caa6d957d..5aeb7777696 100644
--- a/fs/xfs/linux-2.6/xfs_fs_subr.c
+++ b/fs/xfs/linux-2.6/xfs_fs_subr.c
@@ -24,6 +24,10 @@ int  fs_noerr(void) { return 0; }
 int  fs_nosys(void) { return ENOSYS; }
 void fs_noval(void) { return; }
 
+/*
+ * note: all filemap functions return negative error codes. These
+ * need to be inverted before returning to the xfs core functions.
+ */
 void
 xfs_tosspages(
 	xfs_inode_t	*ip,
@@ -53,7 +57,7 @@ xfs_flushinval_pages(
 		if (!ret)
 			truncate_inode_pages(mapping, first);
 	}
-	return ret;
+	return -ret;
 }
 
 int
@@ -72,10 +76,23 @@ xfs_flush_pages(
 		xfs_iflags_clear(ip, XFS_ITRUNCATED);
 		ret = filemap_fdatawrite(mapping);
 		if (flags & XFS_B_ASYNC)
-			return ret;
+			return -ret;
 		ret2 = filemap_fdatawait(mapping);
 		if (!ret)
 			ret = ret2;
 	}
-	return ret;
+	return -ret;
+}
+
+int
+xfs_wait_on_pages(
+	xfs_inode_t	*ip,
+	xfs_off_t	first,
+	xfs_off_t	last)
+{
+	struct address_space *mapping = VFS_I(ip)->i_mapping;
+
+	if (mapping_tagged(mapping, PAGECACHE_TAG_WRITEBACK))
+		return -filemap_fdatawait(mapping);
+	return 0;
 }
diff --git a/fs/xfs/linux-2.6/xfs_globals.c b/fs/xfs/linux-2.6/xfs_globals.c
index ef90e64641e..2ae8b1ccb02 100644
--- a/fs/xfs/linux-2.6/xfs_globals.c
+++ b/fs/xfs/linux-2.6/xfs_globals.c
@@ -26,7 +26,6 @@
  */
 xfs_param_t xfs_params = {
 			  /*	MIN		DFLT		MAX	*/
-	.restrict_chown	= {	0,		1,		1	},
 	.sgid_inherit	= {	0,		0,		1	},
 	.symlink_mode	= {	0,		0,		1	},
 	.panic_mask	= {	0,		0,		255	},
@@ -43,10 +42,3 @@ xfs_param_t xfs_params = {
 	.inherit_nodfrg	= {	0,		1,		1	},
 	.fstrm_timer	= {	1,		30*100,		3600*100},
 };
-
-/*
- * Global system credential structure.
- */
-static cred_t sys_cred_val;
-cred_t *sys_cred = &sys_cred_val;
-
diff --git a/fs/xfs/linux-2.6/xfs_globals.h b/fs/xfs/linux-2.6/xfs_globals.h
index 6eda8a3eb6f..69f71caf061 100644
--- a/fs/xfs/linux-2.6/xfs_globals.h
+++ b/fs/xfs/linux-2.6/xfs_globals.h
@@ -19,6 +19,5 @@
 #define __XFS_GLOBALS_H__
 
 extern uint64_t	xfs_panic_mask;		/* set to cause more panics */
-extern cred_t *sys_cred;
 
 #endif	/* __XFS_GLOBALS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c
index 281cbd5a25c..67205f6198b 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl.c
@@ -68,26 +68,22 @@
  * XFS_IOC_PATH_TO_HANDLE
  *    returns full handle for a path
  */
-STATIC int
+int
 xfs_find_handle(
 	unsigned int		cmd,
-	void			__user *arg)
+	xfs_fsop_handlereq_t	*hreq)
 {
 	int			hsize;
 	xfs_handle_t		handle;
-	xfs_fsop_handlereq_t	hreq;
 	struct inode		*inode;
 
-	if (copy_from_user(&hreq, arg, sizeof(hreq)))
-		return -XFS_ERROR(EFAULT);
-
 	memset((char *)&handle, 0, sizeof(handle));
 
 	switch (cmd) {
 	case XFS_IOC_PATH_TO_FSHANDLE:
 	case XFS_IOC_PATH_TO_HANDLE: {
 		struct path path;
-		int error = user_lpath((const char __user *)hreq.path, &path);
+		int error = user_lpath((const char __user *)hreq->path, &path);
 		if (error)
 			return error;
 
@@ -101,7 +97,7 @@ xfs_find_handle(
 	case XFS_IOC_FD_TO_HANDLE: {
 		struct file	*file;
 
-		file = fget(hreq.fd);
+		file = fget(hreq->fd);
 		if (!file)
 		    return -EBADF;
 
@@ -158,8 +154,8 @@ xfs_find_handle(
 	}
 
 	/* now copy our handle into the user buffer & write out the size */
-	if (copy_to_user(hreq.ohandle, &handle, hsize) ||
-	    copy_to_user(hreq.ohandlen, &hsize, sizeof(__s32))) {
+	if (copy_to_user(hreq->ohandle, &handle, hsize) ||
+	    copy_to_user(hreq->ohandlen, &hsize, sizeof(__s32))) {
 		iput(inode);
 		return -XFS_ERROR(EFAULT);
 	}
@@ -249,10 +245,10 @@ xfs_vget_fsop_handlereq(
 	return 0;
 }
 
-STATIC int
+int
 xfs_open_by_handle(
 	xfs_mount_t		*mp,
-	void			__user *arg,
+	xfs_fsop_handlereq_t	*hreq,
 	struct file		*parfilp,
 	struct inode		*parinode)
 {
@@ -263,14 +259,11 @@ xfs_open_by_handle(
 	struct file		*filp;
 	struct inode		*inode;
 	struct dentry		*dentry;
-	xfs_fsop_handlereq_t	hreq;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -XFS_ERROR(EPERM);
-	if (copy_from_user(&hreq, arg, sizeof(xfs_fsop_handlereq_t)))
-		return -XFS_ERROR(EFAULT);
 
-	error = xfs_vget_fsop_handlereq(mp, parinode, &hreq, &inode);
+	error = xfs_vget_fsop_handlereq(mp, parinode, hreq, &inode);
 	if (error)
 		return -error;
 
@@ -281,10 +274,10 @@ xfs_open_by_handle(
 	}
 
 #if BITS_PER_LONG != 32
-	hreq.oflags |= O_LARGEFILE;
+	hreq->oflags |= O_LARGEFILE;
 #endif
 	/* Put open permission in namei format. */
-	permflag = hreq.oflags;
+	permflag = hreq->oflags;
 	if ((permflag+1) & O_ACCMODE)
 		permflag++;
 	if (permflag & O_TRUNC)
@@ -322,15 +315,16 @@ xfs_open_by_handle(
 	mntget(parfilp->f_path.mnt);
 
 	/* Create file pointer. */
-	filp = dentry_open(dentry, parfilp->f_path.mnt, hreq.oflags, cred);
+	filp = dentry_open(dentry, parfilp->f_path.mnt, hreq->oflags, cred);
 	if (IS_ERR(filp)) {
 		put_unused_fd(new_fd);
 		return -XFS_ERROR(-PTR_ERR(filp));
 	}
+
 	if (inode->i_mode & S_IFREG) {
 		/* invisible operation should not change atime */
 		filp->f_flags |= O_NOATIME;
-		filp->f_op = &xfs_invis_file_operations;
+		filp->f_mode |= FMODE_NOCMTIME;
 	}
 
 	fd_install(new_fd, filp);
@@ -363,24 +357,21 @@ do_readlink(
 }
 
 
-STATIC int
+int
 xfs_readlink_by_handle(
 	xfs_mount_t		*mp,
-	void			__user *arg,
+	xfs_fsop_handlereq_t	*hreq,
 	struct inode		*parinode)
 {
 	struct inode		*inode;
-	xfs_fsop_handlereq_t	hreq;
 	__u32			olen;
 	void			*link;
 	int			error;
 
 	if (!capable(CAP_SYS_ADMIN))
 		return -XFS_ERROR(EPERM);
-	if (copy_from_user(&hreq, arg, sizeof(xfs_fsop_handlereq_t)))
-		return -XFS_ERROR(EFAULT);
 
-	error = xfs_vget_fsop_handlereq(mp, parinode, &hreq, &inode);
+	error = xfs_vget_fsop_handlereq(mp, parinode, hreq, &inode);
 	if (error)
 		return -error;
 
@@ -390,7 +381,7 @@ xfs_readlink_by_handle(
 		goto out_iput;
 	}
 
-	if (copy_from_user(&olen, hreq.ohandlen, sizeof(__u32))) {
+	if (copy_from_user(&olen, hreq->ohandlen, sizeof(__u32))) {
 		error = -XFS_ERROR(EFAULT);
 		goto out_iput;
 	}
@@ -402,7 +393,7 @@ xfs_readlink_by_handle(
 	error = -xfs_readlink(XFS_I(inode), link);
 	if (error)
 		goto out_kfree;
-	error = do_readlink(hreq.ohandle, olen, link);
+	error = do_readlink(hreq->ohandle, olen, link);
 	if (error)
 		goto out_kfree;
 
@@ -501,7 +492,7 @@ xfs_attrlist_by_handle(
 	return -error;
 }
 
-STATIC int
+int
 xfs_attrmulti_attr_get(
 	struct inode		*inode,
 	char			*name,
@@ -530,7 +521,7 @@ xfs_attrmulti_attr_get(
 	return error;
 }
 
-STATIC int
+int
 xfs_attrmulti_attr_set(
 	struct inode		*inode,
 	char			*name,
@@ -560,7 +551,7 @@ xfs_attrmulti_attr_set(
 	return error;
 }
 
-STATIC int
+int
 xfs_attrmulti_attr_remove(
 	struct inode		*inode,
 	char			*name,
@@ -662,19 +653,26 @@ xfs_attrmulti_by_handle(
 	return -error;
 }
 
-STATIC int
+int
 xfs_ioc_space(
 	struct xfs_inode	*ip,
 	struct inode		*inode,
 	struct file		*filp,
 	int			ioflags,
 	unsigned int		cmd,
-	void			__user *arg)
+	xfs_flock64_t		*bf)
 {
-	xfs_flock64_t		bf;
 	int			attr_flags = 0;
 	int			error;
 
+	/*
+	 * Only allow the sys admin to reserve space unless
+	 * unwritten extents are enabled.
+	 */
+	if (!xfs_sb_version_hasextflgbit(&ip->i_mount->m_sb) &&
+	    !capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+
 	if (inode->i_flags & (S_IMMUTABLE|S_APPEND))
 		return -XFS_ERROR(EPERM);
 
@@ -684,16 +682,12 @@ xfs_ioc_space(
 	if (!S_ISREG(inode->i_mode))
 		return -XFS_ERROR(EINVAL);
 
-	if (copy_from_user(&bf, arg, sizeof(bf)))
-		return -XFS_ERROR(EFAULT);
-
 	if (filp->f_flags & (O_NDELAY|O_NONBLOCK))
 		attr_flags |= XFS_ATTR_NONBLOCK;
 	if (ioflags & IO_INVIS)
 		attr_flags |= XFS_ATTR_DMI;
 
-	error = xfs_change_file_space(ip, cmd, &bf, filp->f_pos,
-					      NULL, attr_flags);
+	error = xfs_change_file_space(ip, cmd, bf, filp->f_pos, attr_flags);
 	return -error;
 }
 
@@ -1105,10 +1099,6 @@ xfs_ioctl_setattr(
 
 	/*
 	 * Change file ownership.  Must be the owner or privileged.
-	 * If the system was configured with the "restricted_chown"
-	 * option, the owner is not permitted to give away the file,
-	 * and can change the group id only to a group of which he
-	 * or she is a member.
 	 */
 	if (mask & FSX_PROJID) {
 		/*
@@ -1137,7 +1127,7 @@ xfs_ioctl_setattr(
 			 * the superblock version number since projids didn't
 			 * exist before DINODE_VERSION_2 and SB_VERSION_NLINK.
 			 */
-			if (ip->i_d.di_version == XFS_DINODE_VERSION_1)
+			if (ip->i_d.di_version == 1)
 				xfs_bump_ino_vers2(tp, ip);
 		}
 
@@ -1256,43 +1246,67 @@ xfs_ioc_setxflags(
 }
 
 STATIC int
+xfs_getbmap_format(void **ap, struct getbmapx *bmv, int *full)
+{
+	struct getbmap __user	*base = *ap;
+
+	/* copy only getbmap portion (not getbmapx) */
+	if (copy_to_user(base, bmv, sizeof(struct getbmap)))
+		return XFS_ERROR(EFAULT);
+
+	*ap += sizeof(struct getbmap);
+	return 0;
+}
+
+STATIC int
 xfs_ioc_getbmap(
 	struct xfs_inode	*ip,
 	int			ioflags,
 	unsigned int		cmd,
 	void			__user *arg)
 {
-	struct getbmap		bm;
-	int			iflags;
+	struct getbmapx		bmx;
 	int			error;
 
-	if (copy_from_user(&bm, arg, sizeof(bm)))
+	if (copy_from_user(&bmx, arg, sizeof(struct getbmapx)))
 		return -XFS_ERROR(EFAULT);
 
-	if (bm.bmv_count < 2)
+	if (bmx.bmv_count < 2)
 		return -XFS_ERROR(EINVAL);
 
-	iflags = (cmd == XFS_IOC_GETBMAPA ? BMV_IF_ATTRFORK : 0);
+	bmx.bmv_iflags = (cmd == XFS_IOC_GETBMAPA ? BMV_IF_ATTRFORK : 0);
 	if (ioflags & IO_INVIS)
-		iflags |= BMV_IF_NO_DMAPI_READ;
+		bmx.bmv_iflags |= BMV_IF_NO_DMAPI_READ;
 
-	error = xfs_getbmap(ip, &bm, (struct getbmap __user *)arg+1, iflags);
+	error = xfs_getbmap(ip, &bmx, xfs_getbmap_format,
+			    (struct getbmap *)arg+1);
 	if (error)
 		return -error;
 
-	if (copy_to_user(arg, &bm, sizeof(bm)))
+	/* copy back header - only size of getbmap */
+	if (copy_to_user(arg, &bmx, sizeof(struct getbmap)))
 		return -XFS_ERROR(EFAULT);
 	return 0;
 }
 
 STATIC int
+xfs_getbmapx_format(void **ap, struct getbmapx *bmv, int *full)
+{
+	struct getbmapx __user	*base = *ap;
+
+	if (copy_to_user(base, bmv, sizeof(struct getbmapx)))
+		return XFS_ERROR(EFAULT);
+
+	*ap += sizeof(struct getbmapx);
+	return 0;
+}
+
+STATIC int
 xfs_ioc_getbmapx(
 	struct xfs_inode	*ip,
 	void			__user *arg)
 {
 	struct getbmapx		bmx;
-	struct getbmap		bm;
-	int			iflags;
 	int			error;
 
 	if (copy_from_user(&bmx, arg, sizeof(bmx)))
@@ -1301,46 +1315,46 @@ xfs_ioc_getbmapx(
 	if (bmx.bmv_count < 2)
 		return -XFS_ERROR(EINVAL);
 
-	/*
-	 * Map input getbmapx structure to a getbmap
-	 * structure for xfs_getbmap.
-	 */
-	GETBMAP_CONVERT(bmx, bm);
-
-	iflags = bmx.bmv_iflags;
-
-	if (iflags & (~BMV_IF_VALID))
+	if (bmx.bmv_iflags & (~BMV_IF_VALID))
 		return -XFS_ERROR(EINVAL);
 
-	iflags |= BMV_IF_EXTENDED;
-
-	error = xfs_getbmap(ip, &bm, (struct getbmapx __user *)arg+1, iflags);
+	error = xfs_getbmap(ip, &bmx, xfs_getbmapx_format,
+			    (struct getbmapx *)arg+1);
 	if (error)
 		return -error;
 
-	GETBMAP_CONVERT(bm, bmx);
-
-	if (copy_to_user(arg, &bmx, sizeof(bmx)))
+	/* copy back header */
+	if (copy_to_user(arg, &bmx, sizeof(struct getbmapx)))
 		return -XFS_ERROR(EFAULT);
 
 	return 0;
 }
 
-int
-xfs_ioctl(
-	xfs_inode_t		*ip,
+/*
+ * Note: some of the ioctl's return positive numbers as a
+ * byte count indicating success, such as readlink_by_handle.
+ * So we don't "sign flip" like most other routines.  This means
+ * true errors need to be returned as a negative value.
+ */
+long
+xfs_file_ioctl(
 	struct file		*filp,
-	int			ioflags,
 	unsigned int		cmd,
-	void			__user *arg)
+	unsigned long		p)
 {
 	struct inode		*inode = filp->f_path.dentry->d_inode;
-	xfs_mount_t		*mp = ip->i_mount;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	void			__user *arg = (void __user *)p;
+	int			ioflags = 0;
 	int			error;
 
-	xfs_itrace_entry(XFS_I(inode));
-	switch (cmd) {
+	if (filp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
 
+	xfs_itrace_entry(ip);
+
+	switch (cmd) {
 	case XFS_IOC_ALLOCSP:
 	case XFS_IOC_FREESP:
 	case XFS_IOC_RESVSP:
@@ -1348,17 +1362,13 @@ xfs_ioctl(
 	case XFS_IOC_ALLOCSP64:
 	case XFS_IOC_FREESP64:
 	case XFS_IOC_RESVSP64:
-	case XFS_IOC_UNRESVSP64:
-		/*
-		 * Only allow the sys admin to reserve space unless
-		 * unwritten extents are enabled.
-		 */
-		if (!xfs_sb_version_hasextflgbit(&mp->m_sb) &&
-		    !capable(CAP_SYS_ADMIN))
-			return -EPERM;
-
-		return xfs_ioc_space(ip, inode, filp, ioflags, cmd, arg);
+	case XFS_IOC_UNRESVSP64: {
+		xfs_flock64_t		bf;
 
+		if (copy_from_user(&bf, arg, sizeof(bf)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_ioc_space(ip, inode, filp, ioflags, cmd, &bf);
+	}
 	case XFS_IOC_DIOINFO: {
 		struct dioattr	da;
 		xfs_buftarg_t	*target =
@@ -1418,18 +1428,30 @@ xfs_ioctl(
 
 	case XFS_IOC_FD_TO_HANDLE:
 	case XFS_IOC_PATH_TO_HANDLE:
-	case XFS_IOC_PATH_TO_FSHANDLE:
-		return xfs_find_handle(cmd, arg);
+	case XFS_IOC_PATH_TO_FSHANDLE: {
+		xfs_fsop_handlereq_t	hreq;
 
-	case XFS_IOC_OPEN_BY_HANDLE:
-		return xfs_open_by_handle(mp, arg, filp, inode);
+		if (copy_from_user(&hreq, arg, sizeof(hreq)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_find_handle(cmd, &hreq);
+	}
+	case XFS_IOC_OPEN_BY_HANDLE: {
+		xfs_fsop_handlereq_t	hreq;
 
+		if (copy_from_user(&hreq, arg, sizeof(xfs_fsop_handlereq_t)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_open_by_handle(mp, &hreq, filp, inode);
+	}
 	case XFS_IOC_FSSETDM_BY_HANDLE:
 		return xfs_fssetdm_by_handle(mp, arg, inode);
 
-	case XFS_IOC_READLINK_BY_HANDLE:
-		return xfs_readlink_by_handle(mp, arg, inode);
+	case XFS_IOC_READLINK_BY_HANDLE: {
+		xfs_fsop_handlereq_t	hreq;
 
+		if (copy_from_user(&hreq, arg, sizeof(xfs_fsop_handlereq_t)))
+			return -XFS_ERROR(EFAULT);
+		return xfs_readlink_by_handle(mp, &hreq, inode);
+	}
 	case XFS_IOC_ATTRLIST_BY_HANDLE:
 		return xfs_attrlist_by_handle(mp, arg, inode);
 
@@ -1437,7 +1459,11 @@ xfs_ioctl(
 		return xfs_attrmulti_by_handle(mp, arg, filp, inode);
 
 	case XFS_IOC_SWAPEXT: {
-		error = xfs_swapext((struct xfs_swapext __user *)arg);
+		struct xfs_swapext	sxp;
+
+		if (copy_from_user(&sxp, arg, sizeof(xfs_swapext_t)))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_swapext(&sxp);
 		return -error;
 	}
 
@@ -1493,9 +1519,6 @@ xfs_ioctl(
 	case XFS_IOC_FSGROWFSDATA: {
 		xfs_growfs_data_t in;
 
-		if (!capable(CAP_SYS_ADMIN))
-			return -EPERM;
-
 		if (copy_from_user(&in, arg, sizeof(in)))
 			return -XFS_ERROR(EFAULT);
 
@@ -1506,9 +1529,6 @@ xfs_ioctl(
 	case XFS_IOC_FSGROWFSLOG: {
 		xfs_growfs_log_t in;
 
-		if (!capable(CAP_SYS_ADMIN))
-			return -EPERM;
-
 		if (copy_from_user(&in, arg, sizeof(in)))
 			return -XFS_ERROR(EFAULT);
 
@@ -1519,9 +1539,6 @@ xfs_ioctl(
 	case XFS_IOC_FSGROWFSRT: {
 		xfs_growfs_rt_t in;
 
-		if (!capable(CAP_SYS_ADMIN))
-			return -EPERM;
-
 		if (copy_from_user(&in, arg, sizeof(in)))
 			return -XFS_ERROR(EFAULT);
 
diff --git a/fs/xfs/linux-2.6/xfs_ioctl.h b/fs/xfs/linux-2.6/xfs_ioctl.h
new file mode 100644
index 00000000000..8c16bf2d7e0
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_ioctl.h
@@ -0,0 +1,82 @@
+/*
+ * Copyright (c) 2008 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_IOCTL_H__
+#define __XFS_IOCTL_H__
+
+extern int
+xfs_ioc_space(
+	struct xfs_inode	*ip,
+	struct inode		*inode,
+	struct file		*filp,
+	int			ioflags,
+	unsigned int		cmd,
+	xfs_flock64_t		*bf);
+
+extern int
+xfs_find_handle(
+	unsigned int		cmd,
+	xfs_fsop_handlereq_t	*hreq);
+
+extern int
+xfs_open_by_handle(
+	xfs_mount_t		*mp,
+	xfs_fsop_handlereq_t	*hreq,
+	struct file		*parfilp,
+	struct inode		*parinode);
+
+extern int
+xfs_readlink_by_handle(
+	xfs_mount_t		*mp,
+	xfs_fsop_handlereq_t	*hreq,
+	struct inode		*parinode);
+
+extern int
+xfs_attrmulti_attr_get(
+	struct inode		*inode,
+	char			*name,
+	char			__user *ubuf,
+	__uint32_t		*len,
+	__uint32_t		flags);
+
+extern int
+	xfs_attrmulti_attr_set(
+	struct inode		*inode,
+	char			*name,
+	const char		__user *ubuf,
+	__uint32_t		len,
+	__uint32_t		flags);
+
+extern int
+xfs_attrmulti_attr_remove(
+	struct inode		*inode,
+	char			*name,
+	__uint32_t		flags);
+
+extern long
+xfs_file_ioctl(
+	struct file		*filp,
+	unsigned int		cmd,
+	unsigned long		p);
+
+extern long
+xfs_file_compat_ioctl(
+	struct file		*file,
+	unsigned int		cmd,
+	unsigned long		arg);
+
+#endif
diff --git a/fs/xfs/linux-2.6/xfs_ioctl32.c b/fs/xfs/linux-2.6/xfs_ioctl32.c
index a4b254eb43b..0504cece9f6 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl32.c
+++ b/fs/xfs/linux-2.6/xfs_ioctl32.c
@@ -16,11 +16,7 @@
  * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  */
 #include <linux/compat.h>
-#include <linux/init.h>
 #include <linux/ioctl.h>
-#include <linux/syscalls.h>
-#include <linux/types.h>
-#include <linux/fs.h>
 #include <asm/uaccess.h>
 #include "xfs.h"
 #include "xfs_fs.h"
@@ -36,7 +32,6 @@
 #include "xfs_bmap_btree.h"
 #include "xfs_attr_sf.h"
 #include "xfs_dir2_sf.h"
-#include "xfs_vfs.h"
 #include "xfs_vnode.h"
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
@@ -44,221 +39,219 @@
 #include "xfs_error.h"
 #include "xfs_dfrag.h"
 #include "xfs_vnodeops.h"
+#include "xfs_fsops.h"
+#include "xfs_alloc.h"
+#include "xfs_rtalloc.h"
+#include "xfs_attr.h"
+#include "xfs_ioctl.h"
 #include "xfs_ioctl32.h"
 
 #define  _NATIVE_IOC(cmd, type) \
 	  _IOC(_IOC_DIR(cmd), _IOC_TYPE(cmd), _IOC_NR(cmd), sizeof(type))
 
-#if defined(CONFIG_IA64) || defined(CONFIG_X86_64)
-#define BROKEN_X86_ALIGNMENT
-#define _PACKED __attribute__((packed))
-/* on ia32 l_start is on a 32-bit boundary */
-typedef struct xfs_flock64_32 {
-	__s16		l_type;
-	__s16		l_whence;
-	__s64		l_start	__attribute__((packed));
-			/* len == 0 means until end of file */
-	__s64		l_len __attribute__((packed));
-	__s32		l_sysid;
-	__u32		l_pid;
-	__s32		l_pad[4];	/* reserve area */
-} xfs_flock64_32_t;
-
-#define XFS_IOC_ALLOCSP_32	_IOW ('X', 10, struct xfs_flock64_32)
-#define XFS_IOC_FREESP_32	_IOW ('X', 11, struct xfs_flock64_32)
-#define XFS_IOC_ALLOCSP64_32	_IOW ('X', 36, struct xfs_flock64_32)
-#define XFS_IOC_FREESP64_32	_IOW ('X', 37, struct xfs_flock64_32)
-#define XFS_IOC_RESVSP_32	_IOW ('X', 40, struct xfs_flock64_32)
-#define XFS_IOC_UNRESVSP_32	_IOW ('X', 41, struct xfs_flock64_32)
-#define XFS_IOC_RESVSP64_32	_IOW ('X', 42, struct xfs_flock64_32)
-#define XFS_IOC_UNRESVSP64_32	_IOW ('X', 43, struct xfs_flock64_32)
-
-/* just account for different alignment */
-STATIC unsigned long
-xfs_ioctl32_flock(
-	unsigned long		arg)
+#ifdef BROKEN_X86_ALIGNMENT
+STATIC int
+xfs_compat_flock64_copyin(
+	xfs_flock64_t		*bf,
+	compat_xfs_flock64_t	__user *arg32)
 {
-	xfs_flock64_32_t	__user *p32 = (void __user *)arg;
-	xfs_flock64_t		__user *p = compat_alloc_user_space(sizeof(*p));
-
-	if (copy_in_user(&p->l_type,	&p32->l_type,	sizeof(s16)) ||
-	    copy_in_user(&p->l_whence,	&p32->l_whence, sizeof(s16)) ||
-	    copy_in_user(&p->l_start,	&p32->l_start,	sizeof(s64)) ||
-	    copy_in_user(&p->l_len,	&p32->l_len,	sizeof(s64)) ||
-	    copy_in_user(&p->l_sysid,	&p32->l_sysid,	sizeof(s32)) ||
-	    copy_in_user(&p->l_pid,	&p32->l_pid,	sizeof(u32)) ||
-	    copy_in_user(&p->l_pad,	&p32->l_pad,	4*sizeof(u32)))
-		return -EFAULT;
-
-	return (unsigned long)p;
+	if (get_user(bf->l_type,	&arg32->l_type) ||
+	    get_user(bf->l_whence,	&arg32->l_whence) ||
+	    get_user(bf->l_start,	&arg32->l_start) ||
+	    get_user(bf->l_len,		&arg32->l_len) ||
+	    get_user(bf->l_sysid,	&arg32->l_sysid) ||
+	    get_user(bf->l_pid,		&arg32->l_pid) ||
+	    copy_from_user(bf->l_pad,	&arg32->l_pad,	4*sizeof(u32)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
 }
 
-typedef struct compat_xfs_fsop_geom_v1 {
-	__u32		blocksize;	/* filesystem (data) block size */
-	__u32		rtextsize;	/* realtime extent size		*/
-	__u32		agblocks;	/* fsblocks in an AG		*/
-	__u32		agcount;	/* number of allocation groups	*/
-	__u32		logblocks;	/* fsblocks in the log		*/
-	__u32		sectsize;	/* (data) sector size, bytes	*/
-	__u32		inodesize;	/* inode size in bytes		*/
-	__u32		imaxpct;	/* max allowed inode space(%)	*/
-	__u64		datablocks;	/* fsblocks in data subvolume	*/
-	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
-	__u64		rtextents;	/* rt extents in realtime subvol*/
-	__u64		logstart;	/* starting fsblock of the log	*/
-	unsigned char	uuid[16];	/* unique id of the filesystem	*/
-	__u32		sunit;		/* stripe unit, fsblocks	*/
-	__u32		swidth;		/* stripe width, fsblocks	*/
-	__s32		version;	/* structure version		*/
-	__u32		flags;		/* superblock version flags	*/
-	__u32		logsectsize;	/* log sector size, bytes	*/
-	__u32		rtsectsize;	/* realtime sector size, bytes	*/
-	__u32		dirblocksize;	/* directory block size, bytes	*/
-} __attribute__((packed)) compat_xfs_fsop_geom_v1_t;
-
-#define XFS_IOC_FSGEOMETRY_V1_32  \
-	_IOR ('X', 100, struct compat_xfs_fsop_geom_v1)
-
-STATIC unsigned long xfs_ioctl32_geom_v1(unsigned long arg)
+STATIC int
+xfs_compat_ioc_fsgeometry_v1(
+	struct xfs_mount	  *mp,
+	compat_xfs_fsop_geom_v1_t __user *arg32)
 {
-	compat_xfs_fsop_geom_v1_t __user *p32 = (void __user *)arg;
-	xfs_fsop_geom_v1_t __user *p = compat_alloc_user_space(sizeof(*p));
+	xfs_fsop_geom_t		  fsgeo;
+	int			  error;
 
-	if (copy_in_user(p, p32, sizeof(*p32)))
-		return -EFAULT;
-	return (unsigned long)p;
+	error = xfs_fs_geometry(mp, &fsgeo, 3);
+	if (error)
+		return -error;
+	/* The 32-bit variant simply has some padding at the end */
+	if (copy_to_user(arg32, &fsgeo, sizeof(struct compat_xfs_fsop_geom_v1)))
+		return -XFS_ERROR(EFAULT);
+	return 0;
 }
 
-typedef struct compat_xfs_inogrp {
-	__u64		xi_startino;	/* starting inode number	*/
-	__s32		xi_alloccount;	/* # bits set in allocmask	*/
-	__u64		xi_allocmask;	/* mask of allocated inodes	*/
-} __attribute__((packed)) compat_xfs_inogrp_t;
-
-STATIC int xfs_inumbers_fmt_compat(
-	void __user *ubuffer,
-	const xfs_inogrp_t *buffer,
-	long count,
-	long *written)
+STATIC int
+xfs_compat_growfs_data_copyin(
+	struct xfs_growfs_data	 *in,
+	compat_xfs_growfs_data_t __user *arg32)
 {
-	compat_xfs_inogrp_t __user *p32 = ubuffer;
-	long i;
+	if (get_user(in->newblocks, &arg32->newblocks) ||
+	    get_user(in->imaxpct,   &arg32->imaxpct))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_compat_growfs_rt_copyin(
+	struct xfs_growfs_rt	 *in,
+	compat_xfs_growfs_rt_t	__user *arg32)
+{
+	if (get_user(in->newblocks, &arg32->newblocks) ||
+	    get_user(in->extsize,   &arg32->extsize))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
+
+STATIC int
+xfs_inumbers_fmt_compat(
+	void			__user *ubuffer,
+	const xfs_inogrp_t	*buffer,
+	long			count,
+	long			*written)
+{
+	compat_xfs_inogrp_t	__user *p32 = ubuffer;
+	long			i;
 
 	for (i = 0; i < count; i++) {
 		if (put_user(buffer[i].xi_startino,   &p32[i].xi_startino) ||
 		    put_user(buffer[i].xi_alloccount, &p32[i].xi_alloccount) ||
 		    put_user(buffer[i].xi_allocmask,  &p32[i].xi_allocmask))
-			return -EFAULT;
+			return -XFS_ERROR(EFAULT);
 	}
 	*written = count * sizeof(*p32);
 	return 0;
 }
 
 #else
-
 #define xfs_inumbers_fmt_compat xfs_inumbers_fmt
-#define _PACKED
+#endif	/* BROKEN_X86_ALIGNMENT */
 
-#endif
+STATIC int
+xfs_ioctl32_bstime_copyin(
+	xfs_bstime_t		*bstime,
+	compat_xfs_bstime_t	__user *bstime32)
+{
+	compat_time_t		sec32;	/* tv_sec differs on 64 vs. 32 */
 
-/* XFS_IOC_FSBULKSTAT and friends */
+	if (get_user(sec32,		&bstime32->tv_sec)	||
+	    get_user(bstime->tv_nsec,	&bstime32->tv_nsec))
+		return -XFS_ERROR(EFAULT);
+	bstime->tv_sec = sec32;
+	return 0;
+}
+
+/* xfs_bstat_t has differing alignment on intel, & bstime_t sizes everywhere */
+STATIC int
+xfs_ioctl32_bstat_copyin(
+	xfs_bstat_t		*bstat,
+	compat_xfs_bstat_t	__user *bstat32)
+{
+	if (get_user(bstat->bs_ino,	&bstat32->bs_ino)	||
+	    get_user(bstat->bs_mode,	&bstat32->bs_mode)	||
+	    get_user(bstat->bs_nlink,	&bstat32->bs_nlink)	||
+	    get_user(bstat->bs_uid,	&bstat32->bs_uid)	||
+	    get_user(bstat->bs_gid,	&bstat32->bs_gid)	||
+	    get_user(bstat->bs_rdev,	&bstat32->bs_rdev)	||
+	    get_user(bstat->bs_blksize,	&bstat32->bs_blksize)	||
+	    get_user(bstat->bs_size,	&bstat32->bs_size)	||
+	    xfs_ioctl32_bstime_copyin(&bstat->bs_atime, &bstat32->bs_atime) ||
+	    xfs_ioctl32_bstime_copyin(&bstat->bs_mtime, &bstat32->bs_mtime) ||
+	    xfs_ioctl32_bstime_copyin(&bstat->bs_ctime, &bstat32->bs_ctime) ||
+	    get_user(bstat->bs_blocks,	&bstat32->bs_size)	||
+	    get_user(bstat->bs_xflags,	&bstat32->bs_size)	||
+	    get_user(bstat->bs_extsize,	&bstat32->bs_extsize)	||
+	    get_user(bstat->bs_extents,	&bstat32->bs_extents)	||
+	    get_user(bstat->bs_gen,	&bstat32->bs_gen)	||
+	    get_user(bstat->bs_projid,	&bstat32->bs_projid)	||
+	    get_user(bstat->bs_dmevmask, &bstat32->bs_dmevmask)	||
+	    get_user(bstat->bs_dmstate,	&bstat32->bs_dmstate)	||
+	    get_user(bstat->bs_aextents, &bstat32->bs_aextents))
+		return -XFS_ERROR(EFAULT);
+	return 0;
+}
 
-typedef struct compat_xfs_bstime {
-	__s32		tv_sec;		/* seconds		*/
-	__s32		tv_nsec;	/* and nanoseconds	*/
-} compat_xfs_bstime_t;
+/* XFS_IOC_FSBULKSTAT and friends */
 
-STATIC int xfs_bstime_store_compat(
-	compat_xfs_bstime_t __user *p32,
-	const xfs_bstime_t *p)
+STATIC int
+xfs_bstime_store_compat(
+	compat_xfs_bstime_t	__user *p32,
+	const xfs_bstime_t	*p)
 {
-	__s32 sec32;
+	__s32			sec32;
 
 	sec32 = p->tv_sec;
 	if (put_user(sec32, &p32->tv_sec) ||
 	    put_user(p->tv_nsec, &p32->tv_nsec))
-		return -EFAULT;
+		return -XFS_ERROR(EFAULT);
 	return 0;
 }
 
-typedef struct compat_xfs_bstat {
-	__u64		bs_ino;		/* inode number			*/
-	__u16		bs_mode;	/* type and mode		*/
-	__u16		bs_nlink;	/* number of links		*/
-	__u32		bs_uid;		/* user id			*/
-	__u32		bs_gid;		/* group id			*/
-	__u32		bs_rdev;	/* device value			*/
-	__s32		bs_blksize;	/* block size			*/
-	__s64		bs_size;	/* file size			*/
-	compat_xfs_bstime_t bs_atime;	/* access time			*/
-	compat_xfs_bstime_t bs_mtime;	/* modify time			*/
-	compat_xfs_bstime_t bs_ctime;	/* inode change time		*/
-	int64_t		bs_blocks;	/* number of blocks		*/
-	__u32		bs_xflags;	/* extended flags		*/
-	__s32		bs_extsize;	/* extent size			*/
-	__s32		bs_extents;	/* number of extents		*/
-	__u32		bs_gen;		/* generation count		*/
-	__u16		bs_projid;	/* project id			*/
-	unsigned char	bs_pad[14];	/* pad space, unused		*/
-	__u32		bs_dmevmask;	/* DMIG event mask		*/
-	__u16		bs_dmstate;	/* DMIG state info		*/
-	__u16		bs_aextents;	/* attribute number of extents	*/
-} _PACKED compat_xfs_bstat_t;
-
-STATIC int xfs_bulkstat_one_fmt_compat(
+/* Return 0 on success or positive error (to xfs_bulkstat()) */
+STATIC int
+xfs_bulkstat_one_fmt_compat(
 	void			__user *ubuffer,
+	int			ubsize,
+	int			*ubused,
 	const xfs_bstat_t	*buffer)
 {
-	compat_xfs_bstat_t __user *p32 = ubuffer;
-
-	if (put_user(buffer->bs_ino, &p32->bs_ino) ||
-	    put_user(buffer->bs_mode, &p32->bs_mode) ||
-	    put_user(buffer->bs_nlink, &p32->bs_nlink) ||
-	    put_user(buffer->bs_uid, &p32->bs_uid) ||
-	    put_user(buffer->bs_gid, &p32->bs_gid) ||
-	    put_user(buffer->bs_rdev, &p32->bs_rdev) ||
-	    put_user(buffer->bs_blksize, &p32->bs_blksize) ||
-	    put_user(buffer->bs_size, &p32->bs_size) ||
+	compat_xfs_bstat_t	__user *p32 = ubuffer;
+
+	if (ubsize < sizeof(*p32))
+		return XFS_ERROR(ENOMEM);
+
+	if (put_user(buffer->bs_ino,	  &p32->bs_ino)		||
+	    put_user(buffer->bs_mode,	  &p32->bs_mode)	||
+	    put_user(buffer->bs_nlink,	  &p32->bs_nlink)	||
+	    put_user(buffer->bs_uid,	  &p32->bs_uid)		||
+	    put_user(buffer->bs_gid,	  &p32->bs_gid)		||
+	    put_user(buffer->bs_rdev,	  &p32->bs_rdev)	||
+	    put_user(buffer->bs_blksize,  &p32->bs_blksize)	||
+	    put_user(buffer->bs_size,	  &p32->bs_size)	||
 	    xfs_bstime_store_compat(&p32->bs_atime, &buffer->bs_atime) ||
 	    xfs_bstime_store_compat(&p32->bs_mtime, &buffer->bs_mtime) ||
 	    xfs_bstime_store_compat(&p32->bs_ctime, &buffer->bs_ctime) ||
-	    put_user(buffer->bs_blocks, &p32->bs_blocks) ||
-	    put_user(buffer->bs_xflags, &p32->bs_xflags) ||
-	    put_user(buffer->bs_extsize, &p32->bs_extsize) ||
-	    put_user(buffer->bs_extents, &p32->bs_extents) ||
-	    put_user(buffer->bs_gen, &p32->bs_gen) ||
-	    put_user(buffer->bs_projid, &p32->bs_projid) ||
-	    put_user(buffer->bs_dmevmask, &p32->bs_dmevmask) ||
-	    put_user(buffer->bs_dmstate, &p32->bs_dmstate) ||
+	    put_user(buffer->bs_blocks,	  &p32->bs_blocks)	||
+	    put_user(buffer->bs_xflags,	  &p32->bs_xflags)	||
+	    put_user(buffer->bs_extsize,  &p32->bs_extsize)	||
+	    put_user(buffer->bs_extents,  &p32->bs_extents)	||
+	    put_user(buffer->bs_gen,	  &p32->bs_gen)		||
+	    put_user(buffer->bs_projid,	  &p32->bs_projid)	||
+	    put_user(buffer->bs_dmevmask, &p32->bs_dmevmask)	||
+	    put_user(buffer->bs_dmstate,  &p32->bs_dmstate)	||
 	    put_user(buffer->bs_aextents, &p32->bs_aextents))
-		return -EFAULT;
-	return sizeof(*p32);
+		return XFS_ERROR(EFAULT);
+	if (ubused)
+		*ubused = sizeof(*p32);
+	return 0;
 }
 
-
-
-typedef struct compat_xfs_fsop_bulkreq {
-	compat_uptr_t	lastip;		/* last inode # pointer		*/
-	__s32		icount;		/* count of entries in buffer	*/
-	compat_uptr_t	ubuffer;	/* user buffer for inode desc.	*/
-	compat_uptr_t	ocount;		/* output count pointer		*/
-} compat_xfs_fsop_bulkreq_t;
-
-#define XFS_IOC_FSBULKSTAT_32 \
-	_IOWR('X', 101, struct compat_xfs_fsop_bulkreq)
-#define XFS_IOC_FSBULKSTAT_SINGLE_32 \
-	_IOWR('X', 102, struct compat_xfs_fsop_bulkreq)
-#define XFS_IOC_FSINUMBERS_32 \
-	_IOWR('X', 103, struct compat_xfs_fsop_bulkreq)
+STATIC int
+xfs_bulkstat_one_compat(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	ino,		/* inode number to get data for */
+	void		__user *buffer,	/* buffer to place output in */
+	int		ubsize,		/* size of buffer */
+	void		*private_data,	/* my private data */
+	xfs_daddr_t	bno,		/* starting bno of inode cluster */
+	int		*ubused,	/* bytes used by me */
+	void		*dibuff,	/* on-disk inode buffer */
+	int		*stat)		/* BULKSTAT_RV_... */
+{
+	return xfs_bulkstat_one_int(mp, ino, buffer, ubsize,
+				    xfs_bulkstat_one_fmt_compat, bno,
+				    ubused, dibuff, stat);
+}
 
 /* copied from xfs_ioctl.c */
 STATIC int
-xfs_ioc_bulkstat_compat(
-	xfs_mount_t		*mp,
-	unsigned int		cmd,
-	void			__user *arg)
+xfs_compat_ioc_bulkstat(
+	xfs_mount_t		  *mp,
+	unsigned int		  cmd,
+	compat_xfs_fsop_bulkreq_t __user *p32)
 {
-	compat_xfs_fsop_bulkreq_t __user *p32 = (void __user *)arg;
 	u32			addr;
 	xfs_fsop_bulkreq_t	bulkreq;
 	int			count;	/* # of records returned */
@@ -270,20 +263,20 @@ xfs_ioc_bulkstat_compat(
 	/* should be called again (unused here, but used in dmapi) */
 
 	if (!capable(CAP_SYS_ADMIN))
-		return -EPERM;
+		return -XFS_ERROR(EPERM);
 
 	if (XFS_FORCED_SHUTDOWN(mp))
 		return -XFS_ERROR(EIO);
 
 	if (get_user(addr, &p32->lastip))
-		return -EFAULT;
+		return -XFS_ERROR(EFAULT);
 	bulkreq.lastip = compat_ptr(addr);
 	if (get_user(bulkreq.icount, &p32->icount) ||
 	    get_user(addr, &p32->ubuffer))
-		return -EFAULT;
+		return -XFS_ERROR(EFAULT);
 	bulkreq.ubuffer = compat_ptr(addr);
 	if (get_user(addr, &p32->ocount))
-		return -EFAULT;
+		return -XFS_ERROR(EFAULT);
 	bulkreq.ocount = compat_ptr(addr);
 
 	if (copy_from_user(&inlast, bulkreq.lastip, sizeof(__s64)))
@@ -295,17 +288,22 @@ xfs_ioc_bulkstat_compat(
 	if (bulkreq.ubuffer == NULL)
 		return -XFS_ERROR(EINVAL);
 
-	if (cmd == XFS_IOC_FSINUMBERS)
+	if (cmd == XFS_IOC_FSINUMBERS_32) {
 		error = xfs_inumbers(mp, &inlast, &count,
 				bulkreq.ubuffer, xfs_inumbers_fmt_compat);
-	else {
-		/* declare a var to get a warning in case the type changes */
-		bulkstat_one_fmt_pf formatter = xfs_bulkstat_one_fmt_compat;
+	} else if (cmd == XFS_IOC_FSBULKSTAT_SINGLE_32) {
+		int res;
+
+		error = xfs_bulkstat_one_compat(mp, inlast, bulkreq.ubuffer,
+				sizeof(compat_xfs_bstat_t),
+				NULL, 0, NULL, NULL, &res);
+	} else if (cmd == XFS_IOC_FSBULKSTAT_32) {
 		error = xfs_bulkstat(mp, &inlast, &count,
-			xfs_bulkstat_one, formatter,
+			xfs_bulkstat_one_compat, NULL,
 			sizeof(compat_xfs_bstat_t), bulkreq.ubuffer,
 			BULKSTAT_FG_QUICK, &done);
-	}
+	} else
+		error = XFS_ERROR(EINVAL);
 	if (error)
 		return -error;
 
@@ -321,63 +319,306 @@ xfs_ioc_bulkstat_compat(
 	return 0;
 }
 
+STATIC int
+xfs_compat_handlereq_copyin(
+	xfs_fsop_handlereq_t		*hreq,
+	compat_xfs_fsop_handlereq_t	__user *arg32)
+{
+	compat_xfs_fsop_handlereq_t	hreq32;
+
+	if (copy_from_user(&hreq32, arg32, sizeof(compat_xfs_fsop_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	hreq->fd = hreq32.fd;
+	hreq->path = compat_ptr(hreq32.path);
+	hreq->oflags = hreq32.oflags;
+	hreq->ihandle = compat_ptr(hreq32.ihandle);
+	hreq->ihandlen = hreq32.ihandlen;
+	hreq->ohandle = compat_ptr(hreq32.ohandle);
+	hreq->ohandlen = compat_ptr(hreq32.ohandlen);
 
+	return 0;
+}
 
-typedef struct compat_xfs_fsop_handlereq {
-	__u32		fd;		/* fd for FD_TO_HANDLE		*/
-	compat_uptr_t	path;		/* user pathname		*/
-	__u32		oflags;		/* open flags			*/
-	compat_uptr_t	ihandle;	/* user supplied handle		*/
-	__u32		ihandlen;	/* user supplied length		*/
-	compat_uptr_t	ohandle;	/* user buffer for handle	*/
-	compat_uptr_t	ohandlen;	/* user buffer length		*/
-} compat_xfs_fsop_handlereq_t;
-
-#define XFS_IOC_PATH_TO_FSHANDLE_32 \
-	_IOWR('X', 104, struct compat_xfs_fsop_handlereq)
-#define XFS_IOC_PATH_TO_HANDLE_32 \
-	_IOWR('X', 105, struct compat_xfs_fsop_handlereq)
-#define XFS_IOC_FD_TO_HANDLE_32 \
-	_IOWR('X', 106, struct compat_xfs_fsop_handlereq)
-#define XFS_IOC_OPEN_BY_HANDLE_32 \
-	_IOWR('X', 107, struct compat_xfs_fsop_handlereq)
-#define XFS_IOC_READLINK_BY_HANDLE_32 \
-	_IOWR('X', 108, struct compat_xfs_fsop_handlereq)
-
-STATIC unsigned long xfs_ioctl32_fshandle(unsigned long arg)
+/*
+ * Convert userspace handle data into inode.
+ *
+ * We use the fact that all the fsop_handlereq ioctl calls have a data
+ * structure argument whose first component is always a xfs_fsop_handlereq_t,
+ * so we can pass that sub structure into this handy, shared routine.
+ *
+ * If no error, caller must always iput the returned inode.
+ */
+STATIC int
+xfs_vget_fsop_handlereq_compat(
+	xfs_mount_t		*mp,
+	struct inode		*parinode,	/* parent inode pointer    */
+	compat_xfs_fsop_handlereq_t	*hreq,
+	struct inode		**inode)
 {
-	compat_xfs_fsop_handlereq_t __user *p32 = (void __user *)arg;
-	xfs_fsop_handlereq_t __user *p = compat_alloc_user_space(sizeof(*p));
-	u32 addr;
-
-	if (copy_in_user(&p->fd, &p32->fd, sizeof(__u32)) ||
-	    get_user(addr, &p32->path) ||
-	    put_user(compat_ptr(addr), &p->path) ||
-	    copy_in_user(&p->oflags, &p32->oflags, sizeof(__u32)) ||
-	    get_user(addr, &p32->ihandle) ||
-	    put_user(compat_ptr(addr), &p->ihandle) ||
-	    copy_in_user(&p->ihandlen, &p32->ihandlen, sizeof(__u32)) ||
-	    get_user(addr, &p32->ohandle) ||
-	    put_user(compat_ptr(addr), &p->ohandle) ||
-	    get_user(addr, &p32->ohandlen) ||
-	    put_user(compat_ptr(addr), &p->ohandlen))
-		return -EFAULT;
-
-	return (unsigned long)p;
+	void			__user *hanp;
+	size_t			hlen;
+	xfs_fid_t		*xfid;
+	xfs_handle_t		*handlep;
+	xfs_handle_t		handle;
+	xfs_inode_t		*ip;
+	xfs_ino_t		ino;
+	__u32			igen;
+	int			error;
+
+	/*
+	 * Only allow handle opens under a directory.
+	 */
+	if (!S_ISDIR(parinode->i_mode))
+		return XFS_ERROR(ENOTDIR);
+
+	hanp = compat_ptr(hreq->ihandle);
+	hlen = hreq->ihandlen;
+	handlep = &handle;
+
+	if (hlen < sizeof(handlep->ha_fsid) || hlen > sizeof(*handlep))
+		return XFS_ERROR(EINVAL);
+	if (copy_from_user(handlep, hanp, hlen))
+		return XFS_ERROR(EFAULT);
+	if (hlen < sizeof(*handlep))
+		memset(((char *)handlep) + hlen, 0, sizeof(*handlep) - hlen);
+	if (hlen > sizeof(handlep->ha_fsid)) {
+		if (handlep->ha_fid.fid_len !=
+		    (hlen - sizeof(handlep->ha_fsid) -
+			    sizeof(handlep->ha_fid.fid_len)) ||
+		    handlep->ha_fid.fid_pad)
+			return XFS_ERROR(EINVAL);
+	}
+
+	/*
+	 * Crack the handle, obtain the inode # & generation #
+	 */
+	xfid = (struct xfs_fid *)&handlep->ha_fid;
+	if (xfid->fid_len == sizeof(*xfid) - sizeof(xfid->fid_len)) {
+		ino  = xfid->fid_ino;
+		igen = xfid->fid_gen;
+	} else {
+		return XFS_ERROR(EINVAL);
+	}
+
+	/*
+	 * Get the XFS inode, building a Linux inode to go with it.
+	 */
+	error = xfs_iget(mp, NULL, ino, 0, XFS_ILOCK_SHARED, &ip, 0);
+	if (error)
+		return error;
+	if (ip == NULL)
+		return XFS_ERROR(EIO);
+	if (ip->i_d.di_gen != igen) {
+		xfs_iput_new(ip, XFS_ILOCK_SHARED);
+		return XFS_ERROR(ENOENT);
+	}
+
+	xfs_iunlock(ip, XFS_ILOCK_SHARED);
+
+	*inode = VFS_I(ip);
+	return 0;
 }
 
+STATIC int
+xfs_compat_attrlist_by_handle(
+	xfs_mount_t		*mp,
+	void			__user *arg,
+	struct inode		*parinode)
+{
+	int			error;
+	attrlist_cursor_kern_t	*cursor;
+	compat_xfs_fsop_attrlist_handlereq_t al_hreq;
+	struct inode		*inode;
+	char			*kbuf;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&al_hreq, arg,
+			   sizeof(compat_xfs_fsop_attrlist_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+	if (al_hreq.buflen > XATTR_LIST_MAX)
+		return -XFS_ERROR(EINVAL);
+
+	/*
+	 * Reject flags, only allow namespaces.
+	 */
+	if (al_hreq.flags & ~(ATTR_ROOT | ATTR_SECURE))
+		return -XFS_ERROR(EINVAL);
+
+	error = xfs_vget_fsop_handlereq_compat(mp, parinode, &al_hreq.hreq,
+					       &inode);
+	if (error)
+		goto out;
+
+	kbuf = kmalloc(al_hreq.buflen, GFP_KERNEL);
+	if (!kbuf)
+		goto out_vn_rele;
+
+	cursor = (attrlist_cursor_kern_t *)&al_hreq.pos;
+	error = xfs_attr_list(XFS_I(inode), kbuf, al_hreq.buflen,
+					al_hreq.flags, cursor);
+	if (error)
+		goto out_kfree;
+
+	if (copy_to_user(compat_ptr(al_hreq.buffer), kbuf, al_hreq.buflen))
+		error = -EFAULT;
+
+ out_kfree:
+	kfree(kbuf);
+ out_vn_rele:
+	iput(inode);
+ out:
+	return -error;
+}
 
-STATIC long
-xfs_compat_ioctl(
-	int		mode,
-	struct file	*file,
-	unsigned	cmd,
-	unsigned long	arg)
+STATIC int
+xfs_compat_attrmulti_by_handle(
+	xfs_mount_t				*mp,
+	void					__user *arg,
+	struct inode				*parinode)
+{
+	int					error;
+	compat_xfs_attr_multiop_t		*ops;
+	compat_xfs_fsop_attrmulti_handlereq_t	am_hreq;
+	struct inode				*inode;
+	unsigned int				i, size;
+	char					*attr_name;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&am_hreq, arg,
+			   sizeof(compat_xfs_fsop_attrmulti_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	error = xfs_vget_fsop_handlereq_compat(mp, parinode, &am_hreq.hreq,
+					       &inode);
+	if (error)
+		goto out;
+
+	error = E2BIG;
+	size = am_hreq.opcount * sizeof(compat_xfs_attr_multiop_t);
+	if (!size || size > 16 * PAGE_SIZE)
+		goto out_vn_rele;
+
+	error = ENOMEM;
+	ops = kmalloc(size, GFP_KERNEL);
+	if (!ops)
+		goto out_vn_rele;
+
+	error = EFAULT;
+	if (copy_from_user(ops, compat_ptr(am_hreq.ops), size))
+		goto out_kfree_ops;
+
+	attr_name = kmalloc(MAXNAMELEN, GFP_KERNEL);
+	if (!attr_name)
+		goto out_kfree_ops;
+
+
+	error = 0;
+	for (i = 0; i < am_hreq.opcount; i++) {
+		ops[i].am_error = strncpy_from_user(attr_name,
+				compat_ptr(ops[i].am_attrname),
+				MAXNAMELEN);
+		if (ops[i].am_error == 0 || ops[i].am_error == MAXNAMELEN)
+			error = -ERANGE;
+		if (ops[i].am_error < 0)
+			break;
+
+		switch (ops[i].am_opcode) {
+		case ATTR_OP_GET:
+			ops[i].am_error = xfs_attrmulti_attr_get(inode,
+					attr_name,
+					compat_ptr(ops[i].am_attrvalue),
+					&ops[i].am_length, ops[i].am_flags);
+			break;
+		case ATTR_OP_SET:
+			ops[i].am_error = xfs_attrmulti_attr_set(inode,
+					attr_name,
+					compat_ptr(ops[i].am_attrvalue),
+					ops[i].am_length, ops[i].am_flags);
+			break;
+		case ATTR_OP_REMOVE:
+			ops[i].am_error = xfs_attrmulti_attr_remove(inode,
+					attr_name, ops[i].am_flags);
+			break;
+		default:
+			ops[i].am_error = EINVAL;
+		}
+	}
+
+	if (copy_to_user(compat_ptr(am_hreq.ops), ops, size))
+		error = XFS_ERROR(EFAULT);
+
+	kfree(attr_name);
+ out_kfree_ops:
+	kfree(ops);
+ out_vn_rele:
+	iput(inode);
+ out:
+	return -error;
+}
+
+STATIC int
+xfs_compat_fssetdm_by_handle(
+	xfs_mount_t		*mp,
+	void			__user *arg,
+	struct inode		*parinode)
+{
+	int			error;
+	struct fsdmidata	fsd;
+	compat_xfs_fsop_setdm_handlereq_t dmhreq;
+	struct inode		*inode;
+
+	if (!capable(CAP_MKNOD))
+		return -XFS_ERROR(EPERM);
+	if (copy_from_user(&dmhreq, arg,
+			   sizeof(compat_xfs_fsop_setdm_handlereq_t)))
+		return -XFS_ERROR(EFAULT);
+
+	error = xfs_vget_fsop_handlereq_compat(mp, parinode, &dmhreq.hreq,
+					       &inode);
+	if (error)
+		return -error;
+
+	if (IS_IMMUTABLE(inode) || IS_APPEND(inode)) {
+		error = -XFS_ERROR(EPERM);
+		goto out;
+	}
+
+	if (copy_from_user(&fsd, compat_ptr(dmhreq.data), sizeof(fsd))) {
+		error = -XFS_ERROR(EFAULT);
+		goto out;
+	}
+
+	error = -xfs_set_dmattrs(XFS_I(inode), fsd.fsd_dmevmask,
+				 fsd.fsd_dmstate);
+
+out:
+	iput(inode);
+	return error;
+}
+
+long
+xfs_file_compat_ioctl(
+	struct file		*filp,
+	unsigned		cmd,
+	unsigned long		p)
 {
-	struct inode	*inode = file->f_path.dentry->d_inode;
-	int		error;
+	struct inode		*inode = filp->f_path.dentry->d_inode;
+	struct xfs_inode	*ip = XFS_I(inode);
+	struct xfs_mount	*mp = ip->i_mount;
+	void			__user *arg = (void __user *)p;
+	int			ioflags = 0;
+	int			error;
+
+	if (filp->f_mode & FMODE_NOCMTIME)
+		ioflags |= IO_INVIS;
+
+	xfs_itrace_entry(ip);
 
 	switch (cmd) {
+	/* No size or alignment issues on any arch */
 	case XFS_IOC_DIOINFO:
 	case XFS_IOC_FSGEOMETRY:
 	case XFS_IOC_FSGETXATTR:
@@ -387,48 +628,18 @@ xfs_compat_ioctl(
 	case XFS_IOC_GETBMAP:
 	case XFS_IOC_GETBMAPA:
 	case XFS_IOC_GETBMAPX:
-/* not handled
-	case XFS_IOC_FSSETDM_BY_HANDLE:
-	case XFS_IOC_ATTRLIST_BY_HANDLE:
-	case XFS_IOC_ATTRMULTI_BY_HANDLE:
-*/
 	case XFS_IOC_FSCOUNTS:
 	case XFS_IOC_SET_RESBLKS:
 	case XFS_IOC_GET_RESBLKS:
-	case XFS_IOC_FSGROWFSDATA:
 	case XFS_IOC_FSGROWFSLOG:
-	case XFS_IOC_FSGROWFSRT:
 	case XFS_IOC_FREEZE:
 	case XFS_IOC_THAW:
 	case XFS_IOC_GOINGDOWN:
 	case XFS_IOC_ERROR_INJECTION:
 	case XFS_IOC_ERROR_CLEARALL:
-		break;
-
-	case XFS_IOC32_GETXFLAGS:
-	case XFS_IOC32_SETXFLAGS:
-	case XFS_IOC32_GETVERSION:
-		cmd = _NATIVE_IOC(cmd, long);
-		break;
-#ifdef BROKEN_X86_ALIGNMENT
-	/* xfs_flock_t has wrong u32 vs u64 alignment */
-	case XFS_IOC_ALLOCSP_32:
-	case XFS_IOC_FREESP_32:
-	case XFS_IOC_ALLOCSP64_32:
-	case XFS_IOC_FREESP64_32:
-	case XFS_IOC_RESVSP_32:
-	case XFS_IOC_UNRESVSP_32:
-	case XFS_IOC_RESVSP64_32:
-	case XFS_IOC_UNRESVSP64_32:
-		arg = xfs_ioctl32_flock(arg);
-		cmd = _NATIVE_IOC(cmd, struct xfs_flock64);
-		break;
-	case XFS_IOC_FSGEOMETRY_V1_32:
-		arg = xfs_ioctl32_geom_v1(arg);
-		cmd = _NATIVE_IOC(cmd, struct xfs_fsop_geom_v1);
-		break;
-
-#else /* These are handled fine if no alignment issues */
+		return xfs_file_ioctl(filp, cmd, p);
+#ifndef BROKEN_X86_ALIGNMENT
+	/* These are handled fine if no alignment issues */
 	case XFS_IOC_ALLOCSP:
 	case XFS_IOC_FREESP:
 	case XFS_IOC_RESVSP:
@@ -438,51 +649,97 @@ xfs_compat_ioctl(
 	case XFS_IOC_RESVSP64:
 	case XFS_IOC_UNRESVSP64:
 	case XFS_IOC_FSGEOMETRY_V1:
-		break;
+	case XFS_IOC_FSGROWFSDATA:
+	case XFS_IOC_FSGROWFSRT:
+		return xfs_file_ioctl(filp, cmd, p);
+#else
+	case XFS_IOC_ALLOCSP_32:
+	case XFS_IOC_FREESP_32:
+	case XFS_IOC_ALLOCSP64_32:
+	case XFS_IOC_FREESP64_32:
+	case XFS_IOC_RESVSP_32:
+	case XFS_IOC_UNRESVSP_32:
+	case XFS_IOC_RESVSP64_32:
+	case XFS_IOC_UNRESVSP64_32: {
+		struct xfs_flock64	bf;
 
-	/* xfs_bstat_t still has wrong u32 vs u64 alignment */
-	case XFS_IOC_SWAPEXT:
-		break;
+		if (xfs_compat_flock64_copyin(&bf, arg))
+			return -XFS_ERROR(EFAULT);
+		cmd = _NATIVE_IOC(cmd, struct xfs_flock64);
+		return xfs_ioc_space(ip, inode, filp, ioflags, cmd, &bf);
+	}
+	case XFS_IOC_FSGEOMETRY_V1_32:
+		return xfs_compat_ioc_fsgeometry_v1(mp, arg);
+	case XFS_IOC_FSGROWFSDATA_32: {
+		struct xfs_growfs_data	in;
+
+		if (xfs_compat_growfs_data_copyin(&in, arg))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_growfs_data(mp, &in);
+		return -error;
+	}
+	case XFS_IOC_FSGROWFSRT_32: {
+		struct xfs_growfs_rt	in;
 
+		if (xfs_compat_growfs_rt_copyin(&in, arg))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_growfs_rt(mp, &in);
+		return -error;
+	}
 #endif
+	/* long changes size, but xfs only copiese out 32 bits */
+	case XFS_IOC_GETXFLAGS_32:
+	case XFS_IOC_SETXFLAGS_32:
+	case XFS_IOC_GETVERSION_32:
+		cmd = _NATIVE_IOC(cmd, long);
+		return xfs_file_ioctl(filp, cmd, p);
+	case XFS_IOC_SWAPEXT: {
+		struct xfs_swapext	  sxp;
+		struct compat_xfs_swapext __user *sxu = arg;
+
+		/* Bulk copy in up to the sx_stat field, then copy bstat */
+		if (copy_from_user(&sxp, sxu,
+				   offsetof(struct xfs_swapext, sx_stat)) ||
+		    xfs_ioctl32_bstat_copyin(&sxp.sx_stat, &sxu->sx_stat))
+			return -XFS_ERROR(EFAULT);
+		error = xfs_swapext(&sxp);
+		return -error;
+	}
 	case XFS_IOC_FSBULKSTAT_32:
 	case XFS_IOC_FSBULKSTAT_SINGLE_32:
 	case XFS_IOC_FSINUMBERS_32:
-		cmd = _NATIVE_IOC(cmd, struct xfs_fsop_bulkreq);
-		return xfs_ioc_bulkstat_compat(XFS_I(inode)->i_mount,
-				cmd, (void __user*)arg);
+		return xfs_compat_ioc_bulkstat(mp, cmd, arg);
 	case XFS_IOC_FD_TO_HANDLE_32:
 	case XFS_IOC_PATH_TO_HANDLE_32:
-	case XFS_IOC_PATH_TO_FSHANDLE_32:
-	case XFS_IOC_OPEN_BY_HANDLE_32:
-	case XFS_IOC_READLINK_BY_HANDLE_32:
-		arg = xfs_ioctl32_fshandle(arg);
+	case XFS_IOC_PATH_TO_FSHANDLE_32: {
+		struct xfs_fsop_handlereq	hreq;
+
+		if (xfs_compat_handlereq_copyin(&hreq, arg))
+			return -XFS_ERROR(EFAULT);
 		cmd = _NATIVE_IOC(cmd, struct xfs_fsop_handlereq);
-		break;
-	default:
-		return -ENOIOCTLCMD;
+		return xfs_find_handle(cmd, &hreq);
 	}
+	case XFS_IOC_OPEN_BY_HANDLE_32: {
+		struct xfs_fsop_handlereq	hreq;
 
-	error = xfs_ioctl(XFS_I(inode), file, mode, cmd, (void __user *)arg);
-	xfs_iflags_set(XFS_I(inode), XFS_IMODIFIED);
-
-	return error;
-}
-
-long
-xfs_file_compat_ioctl(
-	struct file		*file,
-	unsigned		cmd,
-	unsigned long		arg)
-{
-	return xfs_compat_ioctl(0, file, cmd, arg);
-}
+		if (xfs_compat_handlereq_copyin(&hreq, arg))
+			return -XFS_ERROR(EFAULT);
+		return xfs_open_by_handle(mp, &hreq, filp, inode);
+	}
+	case XFS_IOC_READLINK_BY_HANDLE_32: {
+		struct xfs_fsop_handlereq	hreq;
 
-long
-xfs_file_compat_invis_ioctl(
-	struct file		*file,
-	unsigned		cmd,
-	unsigned long		arg)
-{
-	return xfs_compat_ioctl(IO_INVIS, file, cmd, arg);
+		if (xfs_compat_handlereq_copyin(&hreq, arg))
+			return -XFS_ERROR(EFAULT);
+		return xfs_readlink_by_handle(mp, &hreq, inode);
+	}
+	case XFS_IOC_ATTRLIST_BY_HANDLE_32:
+		return xfs_compat_attrlist_by_handle(mp, arg, inode);
+	case XFS_IOC_ATTRMULTI_BY_HANDLE_32:
+		return xfs_compat_attrmulti_by_handle(mp, arg, inode);
+	case XFS_IOC_FSSETDM_BY_HANDLE_32:
+		return xfs_compat_fssetdm_by_handle(mp, arg, inode);
+	default:
+		return -XFS_ERROR(ENOIOCTLCMD);
+	}
 }
diff --git a/fs/xfs/linux-2.6/xfs_ioctl32.h b/fs/xfs/linux-2.6/xfs_ioctl32.h
index 02de6e62ee3..1024c4f8ba0 100644
--- a/fs/xfs/linux-2.6/xfs_ioctl32.h
+++ b/fs/xfs/linux-2.6/xfs_ioctl32.h
@@ -18,7 +18,217 @@
 #ifndef __XFS_IOCTL32_H__
 #define __XFS_IOCTL32_H__
 
-extern long xfs_file_compat_ioctl(struct file *, unsigned, unsigned long);
-extern long xfs_file_compat_invis_ioctl(struct file *, unsigned, unsigned long);
+#include <linux/compat.h>
+
+/*
+ * on 32-bit arches, ioctl argument structures may have different sizes
+ * and/or alignment.  We define compat structures which match the
+ * 32-bit sizes/alignments here, and their associated ioctl numbers.
+ *
+ * xfs_ioctl32.c contains routines to copy these structures in and out.
+ */
+
+/* stock kernel-level ioctls we support */
+#define XFS_IOC_GETXFLAGS_32	FS_IOC32_GETFLAGS
+#define XFS_IOC_SETXFLAGS_32	FS_IOC32_SETFLAGS
+#define XFS_IOC_GETVERSION_32	FS_IOC32_GETVERSION
+
+/*
+ * On intel, even if sizes match, alignment and/or padding may differ.
+ */
+#if defined(CONFIG_IA64) || defined(CONFIG_X86_64)
+#define BROKEN_X86_ALIGNMENT
+#define __compat_packed __attribute__((packed))
+#else
+#define __compat_packed
+#endif
+
+typedef struct compat_xfs_bstime {
+	compat_time_t	tv_sec;		/* seconds		*/
+	__s32		tv_nsec;	/* and nanoseconds	*/
+} compat_xfs_bstime_t;
+
+typedef struct compat_xfs_bstat {
+	__u64		bs_ino;		/* inode number			*/
+	__u16		bs_mode;	/* type and mode		*/
+	__u16		bs_nlink;	/* number of links		*/
+	__u32		bs_uid;		/* user id			*/
+	__u32		bs_gid;		/* group id			*/
+	__u32		bs_rdev;	/* device value			*/
+	__s32		bs_blksize;	/* block size			*/
+	__s64		bs_size;	/* file size			*/
+	compat_xfs_bstime_t bs_atime;	/* access time			*/
+	compat_xfs_bstime_t bs_mtime;	/* modify time			*/
+	compat_xfs_bstime_t bs_ctime;	/* inode change time		*/
+	int64_t		bs_blocks;	/* number of blocks		*/
+	__u32		bs_xflags;	/* extended flags		*/
+	__s32		bs_extsize;	/* extent size			*/
+	__s32		bs_extents;	/* number of extents		*/
+	__u32		bs_gen;		/* generation count		*/
+	__u16		bs_projid;	/* project id			*/
+	unsigned char	bs_pad[14];	/* pad space, unused		*/
+	__u32		bs_dmevmask;	/* DMIG event mask		*/
+	__u16		bs_dmstate;	/* DMIG state info		*/
+	__u16		bs_aextents;	/* attribute number of extents	*/
+} __compat_packed compat_xfs_bstat_t;
+
+typedef struct compat_xfs_fsop_bulkreq {
+	compat_uptr_t	lastip;		/* last inode # pointer		*/
+	__s32		icount;		/* count of entries in buffer	*/
+	compat_uptr_t	ubuffer;	/* user buffer for inode desc.	*/
+	compat_uptr_t	ocount;		/* output count pointer		*/
+} compat_xfs_fsop_bulkreq_t;
+
+#define XFS_IOC_FSBULKSTAT_32 \
+	_IOWR('X', 101, struct compat_xfs_fsop_bulkreq)
+#define XFS_IOC_FSBULKSTAT_SINGLE_32 \
+	_IOWR('X', 102, struct compat_xfs_fsop_bulkreq)
+#define XFS_IOC_FSINUMBERS_32 \
+	_IOWR('X', 103, struct compat_xfs_fsop_bulkreq)
+
+typedef struct compat_xfs_fsop_handlereq {
+	__u32		fd;		/* fd for FD_TO_HANDLE		*/
+	compat_uptr_t	path;		/* user pathname		*/
+	__u32		oflags;		/* open flags			*/
+	compat_uptr_t	ihandle;	/* user supplied handle		*/
+	__u32		ihandlen;	/* user supplied length		*/
+	compat_uptr_t	ohandle;	/* user buffer for handle	*/
+	compat_uptr_t	ohandlen;	/* user buffer length		*/
+} compat_xfs_fsop_handlereq_t;
+
+#define XFS_IOC_PATH_TO_FSHANDLE_32 \
+	_IOWR('X', 104, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_PATH_TO_HANDLE_32 \
+	_IOWR('X', 105, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_FD_TO_HANDLE_32 \
+	_IOWR('X', 106, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_OPEN_BY_HANDLE_32 \
+	_IOWR('X', 107, struct compat_xfs_fsop_handlereq)
+#define XFS_IOC_READLINK_BY_HANDLE_32 \
+	_IOWR('X', 108, struct compat_xfs_fsop_handlereq)
+
+/* The bstat field in the swapext struct needs translation */
+typedef struct compat_xfs_swapext {
+	__int64_t		sx_version;	/* version */
+	__int64_t		sx_fdtarget;	/* fd of target file */
+	__int64_t		sx_fdtmp;	/* fd of tmp file */
+	xfs_off_t		sx_offset;	/* offset into file */
+	xfs_off_t		sx_length;	/* leng from offset */
+	char			sx_pad[16];	/* pad space, unused */
+	compat_xfs_bstat_t	sx_stat;	/* stat of target b4 copy */
+} __compat_packed compat_xfs_swapext_t;
+
+#define XFS_IOC_SWAPEXT_32	_IOWR('X', 109, struct compat_xfs_swapext)
+
+typedef struct compat_xfs_fsop_attrlist_handlereq {
+	struct compat_xfs_fsop_handlereq hreq; /* handle interface structure */
+	struct xfs_attrlist_cursor	pos; /* opaque cookie, list offset */
+	__u32				flags;	/* which namespace to use */
+	__u32				buflen;	/* length of buffer supplied */
+	compat_uptr_t			buffer;	/* returned names */
+} __compat_packed compat_xfs_fsop_attrlist_handlereq_t;
+
+/* Note: actually this is read/write */
+#define XFS_IOC_ATTRLIST_BY_HANDLE_32 \
+	_IOW('X', 122, struct compat_xfs_fsop_attrlist_handlereq)
+
+/* am_opcodes defined in xfs_fs.h */
+typedef struct compat_xfs_attr_multiop {
+	__u32		am_opcode;
+	__s32		am_error;
+	compat_uptr_t	am_attrname;
+	compat_uptr_t	am_attrvalue;
+	__u32		am_length;
+	__u32		am_flags;
+} compat_xfs_attr_multiop_t;
+
+typedef struct compat_xfs_fsop_attrmulti_handlereq {
+	struct compat_xfs_fsop_handlereq hreq; /* handle interface structure */
+	__u32				opcount;/* count of following multiop */
+	/* ptr to compat_xfs_attr_multiop */
+	compat_uptr_t			ops; /* attr_multi data */
+} compat_xfs_fsop_attrmulti_handlereq_t;
+
+#define XFS_IOC_ATTRMULTI_BY_HANDLE_32 \
+	_IOW('X', 123, struct compat_xfs_fsop_attrmulti_handlereq)
+
+typedef struct compat_xfs_fsop_setdm_handlereq {
+	struct compat_xfs_fsop_handlereq hreq;	/* handle information   */
+	/* ptr to struct fsdmidata */
+	compat_uptr_t			data;	/* DMAPI data   */
+} compat_xfs_fsop_setdm_handlereq_t;
+
+#define XFS_IOC_FSSETDM_BY_HANDLE_32 \
+	_IOW('X', 121, struct compat_xfs_fsop_setdm_handlereq)
+
+#ifdef BROKEN_X86_ALIGNMENT
+/* on ia32 l_start is on a 32-bit boundary */
+typedef struct compat_xfs_flock64 {
+	__s16		l_type;
+	__s16		l_whence;
+	__s64		l_start	__attribute__((packed));
+			/* len == 0 means until end of file */
+	__s64		l_len __attribute__((packed));
+	__s32		l_sysid;
+	__u32		l_pid;
+	__s32		l_pad[4];	/* reserve area */
+} compat_xfs_flock64_t;
+
+#define XFS_IOC_ALLOCSP_32	_IOW('X', 10, struct compat_xfs_flock64)
+#define XFS_IOC_FREESP_32	_IOW('X', 11, struct compat_xfs_flock64)
+#define XFS_IOC_ALLOCSP64_32	_IOW('X', 36, struct compat_xfs_flock64)
+#define XFS_IOC_FREESP64_32	_IOW('X', 37, struct compat_xfs_flock64)
+#define XFS_IOC_RESVSP_32	_IOW('X', 40, struct compat_xfs_flock64)
+#define XFS_IOC_UNRESVSP_32	_IOW('X', 41, struct compat_xfs_flock64)
+#define XFS_IOC_RESVSP64_32	_IOW('X', 42, struct compat_xfs_flock64)
+#define XFS_IOC_UNRESVSP64_32	_IOW('X', 43, struct compat_xfs_flock64)
+
+typedef struct compat_xfs_fsop_geom_v1 {
+	__u32		blocksize;	/* filesystem (data) block size */
+	__u32		rtextsize;	/* realtime extent size		*/
+	__u32		agblocks;	/* fsblocks in an AG		*/
+	__u32		agcount;	/* number of allocation groups	*/
+	__u32		logblocks;	/* fsblocks in the log		*/
+	__u32		sectsize;	/* (data) sector size, bytes	*/
+	__u32		inodesize;	/* inode size in bytes		*/
+	__u32		imaxpct;	/* max allowed inode space(%)	*/
+	__u64		datablocks;	/* fsblocks in data subvolume	*/
+	__u64		rtblocks;	/* fsblocks in realtime subvol	*/
+	__u64		rtextents;	/* rt extents in realtime subvol*/
+	__u64		logstart;	/* starting fsblock of the log	*/
+	unsigned char	uuid[16];	/* unique id of the filesystem	*/
+	__u32		sunit;		/* stripe unit, fsblocks	*/
+	__u32		swidth;		/* stripe width, fsblocks	*/
+	__s32		version;	/* structure version		*/
+	__u32		flags;		/* superblock version flags	*/
+	__u32		logsectsize;	/* log sector size, bytes	*/
+	__u32		rtsectsize;	/* realtime sector size, bytes	*/
+	__u32		dirblocksize;	/* directory block size, bytes	*/
+} __attribute__((packed)) compat_xfs_fsop_geom_v1_t;
+
+#define XFS_IOC_FSGEOMETRY_V1_32  \
+	_IOR('X', 100, struct compat_xfs_fsop_geom_v1)
+
+typedef struct compat_xfs_inogrp {
+	__u64		xi_startino;	/* starting inode number	*/
+	__s32		xi_alloccount;	/* # bits set in allocmask	*/
+	__u64		xi_allocmask;	/* mask of allocated inodes	*/
+} __attribute__((packed)) compat_xfs_inogrp_t;
+
+/* These growfs input structures have padding on the end, so must translate */
+typedef struct compat_xfs_growfs_data {
+	__u64		newblocks;	/* new data subvol size, fsblocks */
+	__u32		imaxpct;	/* new inode space percentage limit */
+} __attribute__((packed)) compat_xfs_growfs_data_t;
+
+typedef struct compat_xfs_growfs_rt {
+	__u64		newblocks;	/* new realtime size, fsblocks */
+	__u32		extsize;	/* new realtime extent size, fsblocks */
+} __attribute__((packed)) compat_xfs_growfs_rt_t;
+
+#define XFS_IOC_FSGROWFSDATA_32 _IOW('X', 110, struct compat_xfs_growfs_data)
+#define XFS_IOC_FSGROWFSRT_32   _IOW('X', 112, struct compat_xfs_growfs_rt)
+
+#endif /* BROKEN_X86_ALIGNMENT */
 
 #endif /* __XFS_IOCTL32_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
index 095d271f343..7aa53fefc67 100644
--- a/fs/xfs/linux-2.6/xfs_iops.c
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -53,6 +53,7 @@
 #include <linux/namei.h>
 #include <linux/security.h>
 #include <linux/falloc.h>
+#include <linux/fiemap.h>
 
 /*
  * Bring the atime in the XFS inode uptodate.
@@ -64,14 +65,14 @@ xfs_synchronize_atime(
 {
 	struct inode	*inode = VFS_I(ip);
 
-	if (inode) {
+	if (!(inode->i_state & I_CLEAR)) {
 		ip->i_d.di_atime.t_sec = (__int32_t)inode->i_atime.tv_sec;
 		ip->i_d.di_atime.t_nsec = (__int32_t)inode->i_atime.tv_nsec;
 	}
 }
 
 /*
- * If the linux inode exists, mark it dirty.
+ * If the linux inode is valid, mark it dirty.
  * Used when commiting a dirty inode into a transaction so that
  * the inode will get written back by the linux code
  */
@@ -81,7 +82,7 @@ xfs_mark_inode_dirty_sync(
 {
 	struct inode	*inode = VFS_I(ip);
 
-	if (inode)
+	if (!(inode->i_state & (I_WILL_FREE|I_FREEING|I_CLEAR)))
 		mark_inode_dirty_sync(inode);
 }
 
@@ -128,7 +129,7 @@ xfs_ichgtime(
 	if (sync_it) {
 		SYNCHRONIZE();
 		ip->i_update_core = 1;
-		mark_inode_dirty_sync(inode);
+		xfs_mark_inode_dirty_sync(ip);
 	}
 }
 
@@ -158,8 +159,6 @@ xfs_init_security(
 	}
 
 	error = xfs_attr_set(ip, name, value, length, ATTR_SECURE);
-	if (!error)
-		xfs_iflags_set(ip, XFS_IMODIFIED);
 
 	kfree(name);
 	kfree(value);
@@ -260,7 +259,6 @@ xfs_vn_mknod(
 		error = _ACL_INHERIT(inode, mode, default_acl);
 		if (unlikely(error))
 			goto out_cleanup_inode;
-		xfs_iflags_set(ip, XFS_IMODIFIED);
 		_ACL_FREE(default_acl);
 	}
 
@@ -366,21 +364,17 @@ xfs_vn_link(
 	struct inode	*dir,
 	struct dentry	*dentry)
 {
-	struct inode	*inode;	/* inode of guy being linked to */
+	struct inode	*inode = old_dentry->d_inode;
 	struct xfs_name	name;
 	int		error;
 
-	inode = old_dentry->d_inode;
 	xfs_dentry_to_name(&name, dentry);
 
-	igrab(inode);
 	error = xfs_link(XFS_I(dir), XFS_I(inode), &name);
-	if (unlikely(error)) {
-		iput(inode);
+	if (unlikely(error))
 		return -error;
-	}
 
-	xfs_iflags_set(XFS_I(dir), XFS_IMODIFIED);
+	atomic_inc(&inode->i_count);
 	d_instantiate(dentry, inode);
 	return 0;
 }
@@ -601,7 +595,7 @@ xfs_vn_setattr(
 	struct dentry	*dentry,
 	struct iattr	*iattr)
 {
-	return -xfs_setattr(XFS_I(dentry->d_inode), iattr, 0, NULL);
+	return -xfs_setattr(XFS_I(dentry->d_inode), iattr, 0);
 }
 
 /*
@@ -642,7 +636,7 @@ xfs_vn_fallocate(
 
 	xfs_ilock(ip, XFS_IOLOCK_EXCL);
 	error = xfs_change_file_space(ip, XFS_IOC_RESVSP, &bf,
-				      0, NULL, XFS_ATTR_NOLOCK);
+				      0, XFS_ATTR_NOLOCK);
 	if (!error && !(mode & FALLOC_FL_KEEP_SIZE) &&
 	    offset + len > i_size_read(inode))
 		new_size = offset + len;
@@ -653,7 +647,7 @@ xfs_vn_fallocate(
 
 		iattr.ia_valid = ATTR_SIZE;
 		iattr.ia_size = new_size;
-		error = xfs_setattr(ip, &iattr, XFS_ATTR_NOLOCK, NULL);
+		error = xfs_setattr(ip, &iattr, XFS_ATTR_NOLOCK);
 	}
 
 	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
@@ -661,6 +655,88 @@ out_error:
 	return error;
 }
 
+#define XFS_FIEMAP_FLAGS	(FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR)
+
+/*
+ * Call fiemap helper to fill in user data.
+ * Returns positive errors to xfs_getbmap.
+ */
+STATIC int
+xfs_fiemap_format(
+	void			**arg,
+	struct getbmapx		*bmv,
+	int			*full)
+{
+	int			error;
+	struct fiemap_extent_info *fieinfo = *arg;
+	u32			fiemap_flags = 0;
+	u64			logical, physical, length;
+
+	/* Do nothing for a hole */
+	if (bmv->bmv_block == -1LL)
+		return 0;
+
+	logical = BBTOB(bmv->bmv_offset);
+	physical = BBTOB(bmv->bmv_block);
+	length = BBTOB(bmv->bmv_length);
+
+	if (bmv->bmv_oflags & BMV_OF_PREALLOC)
+		fiemap_flags |= FIEMAP_EXTENT_UNWRITTEN;
+	else if (bmv->bmv_oflags & BMV_OF_DELALLOC) {
+		fiemap_flags |= FIEMAP_EXTENT_DELALLOC;
+		physical = 0;   /* no block yet */
+	}
+	if (bmv->bmv_oflags & BMV_OF_LAST)
+		fiemap_flags |= FIEMAP_EXTENT_LAST;
+
+	error = fiemap_fill_next_extent(fieinfo, logical, physical,
+					length, fiemap_flags);
+	if (error > 0) {
+		error = 0;
+		*full = 1;	/* user array now full */
+	}
+
+	return -error;
+}
+
+STATIC int
+xfs_vn_fiemap(
+	struct inode		*inode,
+	struct fiemap_extent_info *fieinfo,
+	u64			start,
+	u64			length)
+{
+	xfs_inode_t		*ip = XFS_I(inode);
+	struct getbmapx		bm;
+	int			error;
+
+	error = fiemap_check_flags(fieinfo, XFS_FIEMAP_FLAGS);
+	if (error)
+		return error;
+
+	/* Set up bmap header for xfs internal routine */
+	bm.bmv_offset = BTOBB(start);
+	/* Special case for whole file */
+	if (length == FIEMAP_MAX_OFFSET)
+		bm.bmv_length = -1LL;
+	else
+		bm.bmv_length = BTOBB(length);
+
+	/* our formatter will tell xfs_getbmap when to stop. */
+	bm.bmv_count = MAXEXTNUM;
+	bm.bmv_iflags = BMV_IF_PREALLOC;
+	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR)
+		bm.bmv_iflags |= BMV_IF_ATTRFORK;
+	if (!(fieinfo->fi_flags & FIEMAP_FLAG_SYNC))
+		bm.bmv_iflags |= BMV_IF_DELALLOC;
+
+	error = xfs_getbmap(ip, &bm, xfs_fiemap_format, fieinfo);
+	if (error)
+		return -error;
+
+	return 0;
+}
+
 static const struct inode_operations xfs_inode_operations = {
 	.permission		= xfs_vn_permission,
 	.truncate		= xfs_vn_truncate,
@@ -671,6 +747,7 @@ static const struct inode_operations xfs_inode_operations = {
 	.removexattr		= generic_removexattr,
 	.listxattr		= xfs_vn_listxattr,
 	.fallocate		= xfs_vn_fallocate,
+	.fiemap			= xfs_vn_fiemap,
 };
 
 static const struct inode_operations xfs_dir_inode_operations = {
@@ -766,12 +843,20 @@ xfs_diflags_to_iflags(
  * When reading existing inodes from disk this is called directly
  * from xfs_iget, when creating a new inode it is called from
  * xfs_ialloc after setting up the inode.
+ *
+ * We are always called with an uninitialised linux inode here.
+ * We need to initialise the necessary fields and take a reference
+ * on it.
  */
 void
 xfs_setup_inode(
 	struct xfs_inode	*ip)
 {
-	struct inode		*inode = ip->i_vnode;
+	struct inode		*inode = &ip->i_vnode;
+
+	inode->i_ino = ip->i_ino;
+	inode->i_state = I_NEW|I_LOCK;
+	inode_add_to_lists(ip->i_mount->m_super, inode);
 
 	inode->i_mode	= ip->i_d.di_mode;
 	inode->i_nlink	= ip->i_d.di_nlink;
@@ -799,7 +884,6 @@ xfs_setup_inode(
 	inode->i_ctime.tv_sec	= ip->i_d.di_ctime.t_sec;
 	inode->i_ctime.tv_nsec	= ip->i_d.di_ctime.t_nsec;
 	xfs_diflags_to_iflags(inode, ip);
-	xfs_iflags_clear(ip, XFS_IMODIFIED);
 
 	switch (inode->i_mode & S_IFMT) {
 	case S_IFREG:
diff --git a/fs/xfs/linux-2.6/xfs_iops.h b/fs/xfs/linux-2.6/xfs_iops.h
index 8b1a1e31dc2..ef41c92ce66 100644
--- a/fs/xfs/linux-2.6/xfs_iops.h
+++ b/fs/xfs/linux-2.6/xfs_iops.h
@@ -22,7 +22,6 @@ struct xfs_inode;
 
 extern const struct file_operations xfs_file_operations;
 extern const struct file_operations xfs_dir_file_operations;
-extern const struct file_operations xfs_invis_file_operations;
 
 extern ssize_t xfs_vn_listxattr(struct dentry *, char *data, size_t size);
 
diff --git a/fs/xfs/linux-2.6/xfs_linux.h b/fs/xfs/linux-2.6/xfs_linux.h
index cc0f7b3a979..507492d6dcc 100644
--- a/fs/xfs/linux-2.6/xfs_linux.h
+++ b/fs/xfs/linux-2.6/xfs_linux.h
@@ -21,18 +21,12 @@
 #include <linux/types.h>
 
 /*
- * Some types are conditional depending on the target system.
  * XFS_BIG_BLKNOS needs block layer disk addresses to be 64 bits.
- * XFS_BIG_INUMS needs the VFS inode number to be 64 bits, as well
- * as requiring XFS_BIG_BLKNOS to be set.
+ * XFS_BIG_INUMS requires XFS_BIG_BLKNOS to be set.
  */
 #if defined(CONFIG_LBD) || (BITS_PER_LONG == 64)
 # define XFS_BIG_BLKNOS	1
-# if BITS_PER_LONG == 64
-#  define XFS_BIG_INUMS	1
-# else
-#  define XFS_BIG_INUMS	0
-# endif
+# define XFS_BIG_INUMS	1
 #else
 # define XFS_BIG_BLKNOS	0
 # define XFS_BIG_INUMS	0
@@ -77,6 +71,7 @@
 #include <linux/spinlock.h>
 #include <linux/random.h>
 #include <linux/ctype.h>
+#include <linux/writeback.h>
 
 #include <asm/page.h>
 #include <asm/div64.h>
@@ -85,7 +80,6 @@
 #include <asm/byteorder.h>
 #include <asm/unaligned.h>
 
-#include <xfs_vfs.h>
 #include <xfs_cred.h>
 #include <xfs_vnode.h>
 #include <xfs_stats.h>
@@ -107,7 +101,6 @@
 #undef  HAVE_PERCPU_SB	/* per cpu superblock counters are a 2.6 feature */
 #endif
 
-#define restricted_chown	xfs_params.restrict_chown.val
 #define irix_sgid_inherit	xfs_params.sgid_inherit.val
 #define irix_symlink_mode	xfs_params.symlink_mode.val
 #define xfs_panic_mask		xfs_params.panic_mask.val
diff --git a/fs/xfs/linux-2.6/xfs_lrw.c b/fs/xfs/linux-2.6/xfs_lrw.c
index 1957e5357d0..7e90daa0d1d 100644
--- a/fs/xfs/linux-2.6/xfs_lrw.c
+++ b/fs/xfs/linux-2.6/xfs_lrw.c
@@ -51,7 +51,6 @@
 #include "xfs_vnodeops.h"
 
 #include <linux/capability.h>
-#include <linux/mount.h>
 #include <linux/writeback.h>
 
 
@@ -243,7 +242,7 @@ xfs_read(
 
 	if (unlikely(ioflags & IO_ISDIRECT)) {
 		if (inode->i_mapping->nrpages)
-			ret = xfs_flushinval_pages(ip, (*offset & PAGE_CACHE_MASK),
+			ret = -xfs_flushinval_pages(ip, (*offset & PAGE_CACHE_MASK),
 						    -1, FI_REMAPF_LOCKED);
 		mutex_unlock(&inode->i_mutex);
 		if (ret) {
@@ -668,15 +667,8 @@ start:
 	if (new_size > xip->i_size)
 		xip->i_new_size = new_size;
 
-	/*
-	 * We're not supposed to change timestamps in readonly-mounted
-	 * filesystems.  Throw it away if anyone asks us.
-	 */
-	if (likely(!(ioflags & IO_INVIS) &&
-		   !mnt_want_write(file->f_path.mnt))) {
+	if (likely(!(ioflags & IO_INVIS)))
 		xfs_ichgtime(xip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
-		mnt_drop_write(file->f_path.mnt);
-	}
 
 	/*
 	 * If the offset is beyond the size of the file, we have a couple
@@ -715,7 +707,6 @@ start:
 		}
 	}
 
-retry:
 	/* We can write back this queue in page reclaim */
 	current->backing_dev_info = mapping->backing_dev_info;
 
@@ -771,6 +762,17 @@ retry:
 	if (ret == -EIOCBQUEUED && !(ioflags & IO_ISAIO))
 		ret = wait_on_sync_kiocb(iocb);
 
+	isize = i_size_read(inode);
+	if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize))
+		*offset = isize;
+
+	if (*offset > xip->i_size) {
+		xfs_ilock(xip, XFS_ILOCK_EXCL);
+		if (*offset > xip->i_size)
+			xip->i_size = *offset;
+		xfs_iunlock(xip, XFS_ILOCK_EXCL);
+	}
+
 	if (ret == -ENOSPC &&
 	    DM_EVENT_ENABLED(xip, DM_EVENT_NOSPACE) && !(ioflags & IO_INVIS)) {
 		xfs_iunlock(xip, iolock);
@@ -784,20 +786,7 @@ retry:
 		xfs_ilock(xip, iolock);
 		if (error)
 			goto out_unlock_internal;
-		pos = xip->i_size;
-		ret = 0;
-		goto retry;
-	}
-
-	isize = i_size_read(inode);
-	if (unlikely(ret < 0 && ret != -EFAULT && *offset > isize))
-		*offset = isize;
-
-	if (*offset > xip->i_size) {
-		xfs_ilock(xip, XFS_ILOCK_EXCL);
-		if (*offset > xip->i_size)
-			xip->i_size = *offset;
-		xfs_iunlock(xip, XFS_ILOCK_EXCL);
+		goto start;
 	}
 
 	error = -ret;
@@ -855,13 +844,7 @@ retry:
 int
 xfs_bdstrat_cb(struct xfs_buf *bp)
 {
-	xfs_mount_t	*mp;
-
-	mp = XFS_BUF_FSPRIVATE3(bp, xfs_mount_t *);
-	if (!XFS_FORCED_SHUTDOWN(mp)) {
-		xfs_buf_iorequest(bp);
-		return 0;
-	} else {
+	if (XFS_FORCED_SHUTDOWN(bp->b_mount)) {
 		xfs_buftrace("XFS__BDSTRAT IOERROR", bp);
 		/*
 		 * Metadata write that didn't get logged but
@@ -874,6 +857,9 @@ xfs_bdstrat_cb(struct xfs_buf *bp)
 		else
 			return (xfs_bioerror(bp));
 	}
+
+	xfs_buf_iorequest(bp);
+	return 0;
 }
 
 /*
diff --git a/fs/xfs/linux-2.6/xfs_stats.c b/fs/xfs/linux-2.6/xfs_stats.c
index 3d5b67c075c..c3526d445f6 100644
--- a/fs/xfs/linux-2.6/xfs_stats.c
+++ b/fs/xfs/linux-2.6/xfs_stats.c
@@ -53,11 +53,15 @@ xfs_read_xfsstats(
 		{ "icluster",		XFSSTAT_END_INODE_CLUSTER	},
 		{ "vnodes",		XFSSTAT_END_VNODE_OPS		},
 		{ "buf",		XFSSTAT_END_BUF			},
+		{ "abtb2",		XFSSTAT_END_ABTB_V2		},
+		{ "abtc2",		XFSSTAT_END_ABTC_V2		},
+		{ "bmbt2",		XFSSTAT_END_BMBT_V2		},
+		{ "ibt2",		XFSSTAT_END_IBT_V2		},
 	};
 
 	/* Loop over all stats groups */
 	for (i=j=len = 0; i < ARRAY_SIZE(xstats); i++) {
-		len += sprintf(buffer + len, xstats[i].desc);
+		len += sprintf(buffer + len, "%s", xstats[i].desc);
 		/* inner loop does each group */
 		while (j < xstats[i].endpoint) {
 			val = 0;
diff --git a/fs/xfs/linux-2.6/xfs_stats.h b/fs/xfs/linux-2.6/xfs_stats.h
index e83820febc9..736854b1ca1 100644
--- a/fs/xfs/linux-2.6/xfs_stats.h
+++ b/fs/xfs/linux-2.6/xfs_stats.h
@@ -118,6 +118,71 @@ struct xfsstats {
 	__uint32_t		xb_page_retries;
 	__uint32_t		xb_page_found;
 	__uint32_t		xb_get_read;
+/* Version 2 btree counters */
+#define XFSSTAT_END_ABTB_V2		(XFSSTAT_END_BUF+15)
+	__uint32_t		xs_abtb_2_lookup;
+	__uint32_t		xs_abtb_2_compare;
+	__uint32_t		xs_abtb_2_insrec;
+	__uint32_t		xs_abtb_2_delrec;
+	__uint32_t		xs_abtb_2_newroot;
+	__uint32_t		xs_abtb_2_killroot;
+	__uint32_t		xs_abtb_2_increment;
+	__uint32_t		xs_abtb_2_decrement;
+	__uint32_t		xs_abtb_2_lshift;
+	__uint32_t		xs_abtb_2_rshift;
+	__uint32_t		xs_abtb_2_split;
+	__uint32_t		xs_abtb_2_join;
+	__uint32_t		xs_abtb_2_alloc;
+	__uint32_t		xs_abtb_2_free;
+	__uint32_t		xs_abtb_2_moves;
+#define XFSSTAT_END_ABTC_V2		(XFSSTAT_END_ABTB_V2+15)
+	__uint32_t		xs_abtc_2_lookup;
+	__uint32_t		xs_abtc_2_compare;
+	__uint32_t		xs_abtc_2_insrec;
+	__uint32_t		xs_abtc_2_delrec;
+	__uint32_t		xs_abtc_2_newroot;
+	__uint32_t		xs_abtc_2_killroot;
+	__uint32_t		xs_abtc_2_increment;
+	__uint32_t		xs_abtc_2_decrement;
+	__uint32_t		xs_abtc_2_lshift;
+	__uint32_t		xs_abtc_2_rshift;
+	__uint32_t		xs_abtc_2_split;
+	__uint32_t		xs_abtc_2_join;
+	__uint32_t		xs_abtc_2_alloc;
+	__uint32_t		xs_abtc_2_free;
+	__uint32_t		xs_abtc_2_moves;
+#define XFSSTAT_END_BMBT_V2		(XFSSTAT_END_ABTC_V2+15)
+	__uint32_t		xs_bmbt_2_lookup;
+	__uint32_t		xs_bmbt_2_compare;
+	__uint32_t		xs_bmbt_2_insrec;
+	__uint32_t		xs_bmbt_2_delrec;
+	__uint32_t		xs_bmbt_2_newroot;
+	__uint32_t		xs_bmbt_2_killroot;
+	__uint32_t		xs_bmbt_2_increment;
+	__uint32_t		xs_bmbt_2_decrement;
+	__uint32_t		xs_bmbt_2_lshift;
+	__uint32_t		xs_bmbt_2_rshift;
+	__uint32_t		xs_bmbt_2_split;
+	__uint32_t		xs_bmbt_2_join;
+	__uint32_t		xs_bmbt_2_alloc;
+	__uint32_t		xs_bmbt_2_free;
+	__uint32_t		xs_bmbt_2_moves;
+#define XFSSTAT_END_IBT_V2		(XFSSTAT_END_BMBT_V2+15)
+	__uint32_t		xs_ibt_2_lookup;
+	__uint32_t		xs_ibt_2_compare;
+	__uint32_t		xs_ibt_2_insrec;
+	__uint32_t		xs_ibt_2_delrec;
+	__uint32_t		xs_ibt_2_newroot;
+	__uint32_t		xs_ibt_2_killroot;
+	__uint32_t		xs_ibt_2_increment;
+	__uint32_t		xs_ibt_2_decrement;
+	__uint32_t		xs_ibt_2_lshift;
+	__uint32_t		xs_ibt_2_rshift;
+	__uint32_t		xs_ibt_2_split;
+	__uint32_t		xs_ibt_2_join;
+	__uint32_t		xs_ibt_2_alloc;
+	__uint32_t		xs_ibt_2_free;
+	__uint32_t		xs_ibt_2_moves;
 /* Extra precision counters */
 	__uint64_t		xs_xstrat_bytes;
 	__uint64_t		xs_write_bytes;
diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c
index 37ebe36056e..36f6cc703ef 100644
--- a/fs/xfs/linux-2.6/xfs_super.c
+++ b/fs/xfs/linux-2.6/xfs_super.c
@@ -18,7 +18,6 @@
 #include "xfs.h"
 #include "xfs_bit.h"
 #include "xfs_log.h"
-#include "xfs_clnt.h"
 #include "xfs_inum.h"
 #include "xfs_trans.h"
 #include "xfs_sb.h"
@@ -36,6 +35,7 @@
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
 #include "xfs_btree.h"
+#include "xfs_btree_trace.h"
 #include "xfs_ialloc.h"
 #include "xfs_bmap.h"
 #include "xfs_rtalloc.h"
@@ -48,7 +48,6 @@
 #include "xfs_buf_item.h"
 #include "xfs_utils.h"
 #include "xfs_vnodeops.h"
-#include "xfs_vfsops.h"
 #include "xfs_version.h"
 #include "xfs_log_priv.h"
 #include "xfs_trans_priv.h"
@@ -58,6 +57,7 @@
 #include "xfs_extfree_item.h"
 #include "xfs_mru_cache.h"
 #include "xfs_inode_item.h"
+#include "xfs_sync.h"
 
 #include <linux/namei.h>
 #include <linux/init.h>
@@ -70,36 +70,9 @@
 
 static struct quotactl_ops xfs_quotactl_operations;
 static struct super_operations xfs_super_operations;
-static kmem_zone_t *xfs_vnode_zone;
 static kmem_zone_t *xfs_ioend_zone;
 mempool_t *xfs_ioend_pool;
 
-STATIC struct xfs_mount_args *
-xfs_args_allocate(
-	struct super_block	*sb,
-	int			silent)
-{
-	struct xfs_mount_args	*args;
-
-	args = kzalloc(sizeof(struct xfs_mount_args), GFP_KERNEL);
-	if (!args)
-		return NULL;
-
-	args->logbufs = args->logbufsize = -1;
-	strncpy(args->fsname, sb->s_id, MAXNAMELEN);
-
-	/* Copy the already-parsed mount(2) flags we're interested in */
-	if (sb->s_flags & MS_DIRSYNC)
-		args->flags |= XFSMNT_DIRSYNC;
-	if (sb->s_flags & MS_SYNCHRONOUS)
-		args->flags |= XFSMNT_WSYNC;
-	if (silent)
-		args->flags |= XFSMNT_QUIET;
-	args->flags |= XFSMNT_32BITINODES;
-
-	return args;
-}
-
 #define MNTOPT_LOGBUFS	"logbufs"	/* number of XFS log buffers */
 #define MNTOPT_LOGBSIZE	"logbsize"	/* size of XFS log buffers */
 #define MNTOPT_LOGDEV	"logdev"	/* log device */
@@ -188,26 +161,54 @@ suffix_strtoul(char *s, char **endp, unsigned int base)
 	return simple_strtoul((const char *)s, endp, base) << shift_left_factor;
 }
 
+/*
+ * This function fills in xfs_mount_t fields based on mount args.
+ * Note: the superblock has _not_ yet been read in.
+ *
+ * Note that this function leaks the various device name allocations on
+ * failure.  The caller takes care of them.
+ */
 STATIC int
 xfs_parseargs(
 	struct xfs_mount	*mp,
 	char			*options,
-	struct xfs_mount_args	*args,
-	int			update)
+	char			**mtpt)
 {
+	struct super_block	*sb = mp->m_super;
 	char			*this_char, *value, *eov;
-	int			dsunit, dswidth, vol_dsunit, vol_dswidth;
-	int			iosize;
+	int			dsunit = 0;
+	int			dswidth = 0;
+	int			iosize = 0;
 	int			dmapi_implies_ikeep = 1;
+	uchar_t			iosizelog = 0;
+
+	/*
+	 * Copy binary VFS mount flags we are interested in.
+	 */
+	if (sb->s_flags & MS_RDONLY)
+		mp->m_flags |= XFS_MOUNT_RDONLY;
+	if (sb->s_flags & MS_DIRSYNC)
+		mp->m_flags |= XFS_MOUNT_DIRSYNC;
+	if (sb->s_flags & MS_SYNCHRONOUS)
+		mp->m_flags |= XFS_MOUNT_WSYNC;
+
+	/*
+	 * Set some default flags that could be cleared by the mount option
+	 * parsing.
+	 */
+	mp->m_flags |= XFS_MOUNT_BARRIER;
+	mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
+	mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
 
-	args->flags |= XFSMNT_BARRIER;
-	args->flags2 |= XFSMNT2_COMPAT_IOSIZE;
+	/*
+	 * These can be overridden by the mount option parsing.
+	 */
+	mp->m_logbufs = -1;
+	mp->m_logbsize = -1;
 
 	if (!options)
 		goto done;
 
-	iosize = dsunit = dswidth = vol_dsunit = vol_dswidth = 0;
-
 	while ((this_char = strsep(&options, ",")) != NULL) {
 		if (!*this_char)
 			continue;
@@ -221,7 +222,7 @@ xfs_parseargs(
 					this_char);
 				return EINVAL;
 			}
-			args->logbufs = simple_strtoul(value, &eov, 10);
+			mp->m_logbufs = simple_strtoul(value, &eov, 10);
 		} else if (!strcmp(this_char, MNTOPT_LOGBSIZE)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -229,7 +230,7 @@ xfs_parseargs(
 					this_char);
 				return EINVAL;
 			}
-			args->logbufsize = suffix_strtoul(value, &eov, 10);
+			mp->m_logbsize = suffix_strtoul(value, &eov, 10);
 		} else if (!strcmp(this_char, MNTOPT_LOGDEV)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -237,7 +238,9 @@ xfs_parseargs(
 					this_char);
 				return EINVAL;
 			}
-			strncpy(args->logname, value, MAXNAMELEN);
+			mp->m_logname = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
+			if (!mp->m_logname)
+				return ENOMEM;
 		} else if (!strcmp(this_char, MNTOPT_MTPT)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -245,7 +248,9 @@ xfs_parseargs(
 					this_char);
 				return EINVAL;
 			}
-			strncpy(args->mtpt, value, MAXNAMELEN);
+			*mtpt = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
+			if (!*mtpt)
+				return ENOMEM;
 		} else if (!strcmp(this_char, MNTOPT_RTDEV)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -253,7 +258,9 @@ xfs_parseargs(
 					this_char);
 				return EINVAL;
 			}
-			strncpy(args->rtname, value, MAXNAMELEN);
+			mp->m_rtname = kstrndup(value, MAXNAMELEN, GFP_KERNEL);
+			if (!mp->m_rtname)
+				return ENOMEM;
 		} else if (!strcmp(this_char, MNTOPT_BIOSIZE)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -262,8 +269,7 @@ xfs_parseargs(
 				return EINVAL;
 			}
 			iosize = simple_strtoul(value, &eov, 10);
-			args->flags |= XFSMNT_IOSIZE;
-			args->iosizelog = (uint8_t) iosize;
+			iosizelog = ffs(iosize) - 1;
 		} else if (!strcmp(this_char, MNTOPT_ALLOCSIZE)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -272,8 +278,7 @@ xfs_parseargs(
 				return EINVAL;
 			}
 			iosize = suffix_strtoul(value, &eov, 10);
-			args->flags |= XFSMNT_IOSIZE;
-			args->iosizelog = ffs(iosize) - 1;
+			iosizelog = ffs(iosize) - 1;
 		} else if (!strcmp(this_char, MNTOPT_GRPID) ||
 			   !strcmp(this_char, MNTOPT_BSDGROUPS)) {
 			mp->m_flags |= XFS_MOUNT_GRPID;
@@ -281,23 +286,25 @@ xfs_parseargs(
 			   !strcmp(this_char, MNTOPT_SYSVGROUPS)) {
 			mp->m_flags &= ~XFS_MOUNT_GRPID;
 		} else if (!strcmp(this_char, MNTOPT_WSYNC)) {
-			args->flags |= XFSMNT_WSYNC;
+			mp->m_flags |= XFS_MOUNT_WSYNC;
 		} else if (!strcmp(this_char, MNTOPT_OSYNCISOSYNC)) {
-			args->flags |= XFSMNT_OSYNCISOSYNC;
+			mp->m_flags |= XFS_MOUNT_OSYNCISOSYNC;
 		} else if (!strcmp(this_char, MNTOPT_NORECOVERY)) {
-			args->flags |= XFSMNT_NORECOVERY;
+			mp->m_flags |= XFS_MOUNT_NORECOVERY;
 		} else if (!strcmp(this_char, MNTOPT_INO64)) {
-			args->flags |= XFSMNT_INO64;
-#if !XFS_BIG_INUMS
+#if XFS_BIG_INUMS
+			mp->m_flags |= XFS_MOUNT_INO64;
+			mp->m_inoadd = XFS_INO64_OFFSET;
+#else
 			cmn_err(CE_WARN,
 				"XFS: %s option not allowed on this system",
 				this_char);
 			return EINVAL;
 #endif
 		} else if (!strcmp(this_char, MNTOPT_NOALIGN)) {
-			args->flags |= XFSMNT_NOALIGN;
+			mp->m_flags |= XFS_MOUNT_NOALIGN;
 		} else if (!strcmp(this_char, MNTOPT_SWALLOC)) {
-			args->flags |= XFSMNT_SWALLOC;
+			mp->m_flags |= XFS_MOUNT_SWALLOC;
 		} else if (!strcmp(this_char, MNTOPT_SUNIT)) {
 			if (!value || !*value) {
 				cmn_err(CE_WARN,
@@ -315,7 +322,7 @@ xfs_parseargs(
 			}
 			dswidth = simple_strtoul(value, &eov, 10);
 		} else if (!strcmp(this_char, MNTOPT_64BITINODE)) {
-			args->flags &= ~XFSMNT_32BITINODES;
+			mp->m_flags &= ~XFS_MOUNT_SMALL_INUMS;
 #if !XFS_BIG_INUMS
 			cmn_err(CE_WARN,
 				"XFS: %s option not allowed on this system",
@@ -323,56 +330,61 @@ xfs_parseargs(
 			return EINVAL;
 #endif
 		} else if (!strcmp(this_char, MNTOPT_NOUUID)) {
-			args->flags |= XFSMNT_NOUUID;
+			mp->m_flags |= XFS_MOUNT_NOUUID;
 		} else if (!strcmp(this_char, MNTOPT_BARRIER)) {
-			args->flags |= XFSMNT_BARRIER;
+			mp->m_flags |= XFS_MOUNT_BARRIER;
 		} else if (!strcmp(this_char, MNTOPT_NOBARRIER)) {
-			args->flags &= ~XFSMNT_BARRIER;
+			mp->m_flags &= ~XFS_MOUNT_BARRIER;
 		} else if (!strcmp(this_char, MNTOPT_IKEEP)) {
-			args->flags |= XFSMNT_IKEEP;
+			mp->m_flags |= XFS_MOUNT_IKEEP;
 		} else if (!strcmp(this_char, MNTOPT_NOIKEEP)) {
 			dmapi_implies_ikeep = 0;
-			args->flags &= ~XFSMNT_IKEEP;
+			mp->m_flags &= ~XFS_MOUNT_IKEEP;
 		} else if (!strcmp(this_char, MNTOPT_LARGEIO)) {
-			args->flags2 &= ~XFSMNT2_COMPAT_IOSIZE;
+			mp->m_flags &= ~XFS_MOUNT_COMPAT_IOSIZE;
 		} else if (!strcmp(this_char, MNTOPT_NOLARGEIO)) {
-			args->flags2 |= XFSMNT2_COMPAT_IOSIZE;
+			mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
 		} else if (!strcmp(this_char, MNTOPT_ATTR2)) {
-			args->flags |= XFSMNT_ATTR2;
+			mp->m_flags |= XFS_MOUNT_ATTR2;
 		} else if (!strcmp(this_char, MNTOPT_NOATTR2)) {
-			args->flags &= ~XFSMNT_ATTR2;
-			args->flags |= XFSMNT_NOATTR2;
+			mp->m_flags &= ~XFS_MOUNT_ATTR2;
+			mp->m_flags |= XFS_MOUNT_NOATTR2;
 		} else if (!strcmp(this_char, MNTOPT_FILESTREAM)) {
-			args->flags2 |= XFSMNT2_FILESTREAMS;
+			mp->m_flags |= XFS_MOUNT_FILESTREAMS;
 		} else if (!strcmp(this_char, MNTOPT_NOQUOTA)) {
-			args->flags &= ~(XFSMNT_UQUOTAENF|XFSMNT_UQUOTA);
-			args->flags &= ~(XFSMNT_GQUOTAENF|XFSMNT_GQUOTA);
+			mp->m_qflags &= ~(XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE |
+					  XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE |
+					  XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE |
+					  XFS_UQUOTA_ENFD | XFS_OQUOTA_ENFD);
 		} else if (!strcmp(this_char, MNTOPT_QUOTA) ||
 			   !strcmp(this_char, MNTOPT_UQUOTA) ||
 			   !strcmp(this_char, MNTOPT_USRQUOTA)) {
-			args->flags |= XFSMNT_UQUOTA | XFSMNT_UQUOTAENF;
+			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE |
+					 XFS_UQUOTA_ENFD);
 		} else if (!strcmp(this_char, MNTOPT_QUOTANOENF) ||
 			   !strcmp(this_char, MNTOPT_UQUOTANOENF)) {
-			args->flags |= XFSMNT_UQUOTA;
-			args->flags &= ~XFSMNT_UQUOTAENF;
+			mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE);
+			mp->m_qflags &= ~XFS_UQUOTA_ENFD;
 		} else if (!strcmp(this_char, MNTOPT_PQUOTA) ||
 			   !strcmp(this_char, MNTOPT_PRJQUOTA)) {
-			args->flags |= XFSMNT_PQUOTA | XFSMNT_PQUOTAENF;
+			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE |
+					 XFS_OQUOTA_ENFD);
 		} else if (!strcmp(this_char, MNTOPT_PQUOTANOENF)) {
-			args->flags |= XFSMNT_PQUOTA;
-			args->flags &= ~XFSMNT_PQUOTAENF;
+			mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE);
+			mp->m_qflags &= ~XFS_OQUOTA_ENFD;
 		} else if (!strcmp(this_char, MNTOPT_GQUOTA) ||
 			   !strcmp(this_char, MNTOPT_GRPQUOTA)) {
-			args->flags |= XFSMNT_GQUOTA | XFSMNT_GQUOTAENF;
+			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE |
+					 XFS_OQUOTA_ENFD);
 		} else if (!strcmp(this_char, MNTOPT_GQUOTANOENF)) {
-			args->flags |= XFSMNT_GQUOTA;
-			args->flags &= ~XFSMNT_GQUOTAENF;
+			mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE);
+			mp->m_qflags &= ~XFS_OQUOTA_ENFD;
 		} else if (!strcmp(this_char, MNTOPT_DMAPI)) {
-			args->flags |= XFSMNT_DMAPI;
+			mp->m_flags |= XFS_MOUNT_DMAPI;
 		} else if (!strcmp(this_char, MNTOPT_XDSM)) {
-			args->flags |= XFSMNT_DMAPI;
+			mp->m_flags |= XFS_MOUNT_DMAPI;
 		} else if (!strcmp(this_char, MNTOPT_DMI)) {
-			args->flags |= XFSMNT_DMAPI;
+			mp->m_flags |= XFS_MOUNT_DMAPI;
 		} else if (!strcmp(this_char, "ihashsize")) {
 			cmn_err(CE_WARN,
 	"XFS: ihashsize no longer used, option is deprecated.");
@@ -390,27 +402,29 @@ xfs_parseargs(
 		}
 	}
 
-	if (args->flags & XFSMNT_NORECOVERY) {
-		if ((mp->m_flags & XFS_MOUNT_RDONLY) == 0) {
-			cmn_err(CE_WARN,
-				"XFS: no-recovery mounts must be read-only.");
-			return EINVAL;
-		}
+	/*
+	 * no recovery flag requires a read-only mount
+	 */
+	if ((mp->m_flags & XFS_MOUNT_NORECOVERY) &&
+	    !(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		cmn_err(CE_WARN, "XFS: no-recovery mounts must be read-only.");
+		return EINVAL;
 	}
 
-	if ((args->flags & XFSMNT_NOALIGN) && (dsunit || dswidth)) {
+	if ((mp->m_flags & XFS_MOUNT_NOALIGN) && (dsunit || dswidth)) {
 		cmn_err(CE_WARN,
 	"XFS: sunit and swidth options incompatible with the noalign option");
 		return EINVAL;
 	}
 
-	if ((args->flags & XFSMNT_GQUOTA) && (args->flags & XFSMNT_PQUOTA)) {
+	if ((mp->m_qflags & (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE)) &&
+	    (mp->m_qflags & (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE))) {
 		cmn_err(CE_WARN,
 			"XFS: cannot mount with both project and group quota");
 		return EINVAL;
 	}
 
-	if ((args->flags & XFSMNT_DMAPI) && *args->mtpt == '\0') {
+	if ((mp->m_flags & XFS_MOUNT_DMAPI) && (!*mtpt || *mtpt[0] == '\0')) {
 		printk("XFS: %s option needs the mount point option as well\n",
 			MNTOPT_DMAPI);
 		return EINVAL;
@@ -438,27 +452,66 @@ xfs_parseargs(
 	 * Note that if "ikeep" or "noikeep" mount options are
 	 * supplied, then they are honored.
 	 */
-	if ((args->flags & XFSMNT_DMAPI) && dmapi_implies_ikeep)
-		args->flags |= XFSMNT_IKEEP;
+	if ((mp->m_flags & XFS_MOUNT_DMAPI) && dmapi_implies_ikeep)
+		mp->m_flags |= XFS_MOUNT_IKEEP;
 
-	if ((args->flags & XFSMNT_NOALIGN) != XFSMNT_NOALIGN) {
+done:
+	if (!(mp->m_flags & XFS_MOUNT_NOALIGN)) {
+		/*
+		 * At this point the superblock has not been read
+		 * in, therefore we do not know the block size.
+		 * Before the mount call ends we will convert
+		 * these to FSBs.
+		 */
 		if (dsunit) {
-			args->sunit = dsunit;
-			args->flags |= XFSMNT_RETERR;
-		} else {
-			args->sunit = vol_dsunit;
+			mp->m_dalign = dsunit;
+			mp->m_flags |= XFS_MOUNT_RETERR;
 		}
-		dswidth ? (args->swidth = dswidth) :
-			  (args->swidth = vol_dswidth);
-	} else {
-		args->sunit = args->swidth = 0;
+
+		if (dswidth)
+			mp->m_swidth = dswidth;
+	}
+
+	if (mp->m_logbufs != -1 &&
+	    mp->m_logbufs != 0 &&
+	    (mp->m_logbufs < XLOG_MIN_ICLOGS ||
+	     mp->m_logbufs > XLOG_MAX_ICLOGS)) {
+		cmn_err(CE_WARN,
+			"XFS: invalid logbufs value: %d [not %d-%d]",
+			mp->m_logbufs, XLOG_MIN_ICLOGS, XLOG_MAX_ICLOGS);
+		return XFS_ERROR(EINVAL);
+	}
+	if (mp->m_logbsize != -1 &&
+	    mp->m_logbsize !=  0 &&
+	    (mp->m_logbsize < XLOG_MIN_RECORD_BSIZE ||
+	     mp->m_logbsize > XLOG_MAX_RECORD_BSIZE ||
+	     !is_power_of_2(mp->m_logbsize))) {
+		cmn_err(CE_WARN,
+	"XFS: invalid logbufsize: %d [not 16k,32k,64k,128k or 256k]",
+			mp->m_logbsize);
+		return XFS_ERROR(EINVAL);
+	}
+
+	mp->m_fsname = kstrndup(sb->s_id, MAXNAMELEN, GFP_KERNEL);
+	if (!mp->m_fsname)
+		return ENOMEM;
+	mp->m_fsname_len = strlen(mp->m_fsname) + 1;
+
+	if (iosizelog) {
+		if (iosizelog > XFS_MAX_IO_LOG ||
+		    iosizelog < XFS_MIN_IO_LOG) {
+			cmn_err(CE_WARN,
+		"XFS: invalid log iosize: %d [not %d-%d]",
+				iosizelog, XFS_MIN_IO_LOG,
+				XFS_MAX_IO_LOG);
+			return XFS_ERROR(EINVAL);
+		}
+
+		mp->m_flags |= XFS_MOUNT_DFLT_IOSIZE;
+		mp->m_readio_log = iosizelog;
+		mp->m_writeio_log = iosizelog;
 	}
 
-done:
-	if (args->flags & XFSMNT_32BITINODES)
-		mp->m_flags |= XFS_MOUNT_SMALL_INUMS;
-	if (args->flags2)
-		args->flags |= XFSMNT_FLAGS2;
 	return 0;
 }
 
@@ -704,8 +757,7 @@ xfs_close_devices(
  */
 STATIC int
 xfs_open_devices(
-	struct xfs_mount	*mp,
-	struct xfs_mount_args	*args)
+	struct xfs_mount	*mp)
 {
 	struct block_device	*ddev = mp->m_super->s_bdev;
 	struct block_device	*logdev = NULL, *rtdev = NULL;
@@ -714,14 +766,14 @@ xfs_open_devices(
 	/*
 	 * Open real time and log devices - order is important.
 	 */
-	if (args->logname[0]) {
-		error = xfs_blkdev_get(mp, args->logname, &logdev);
+	if (mp->m_logname) {
+		error = xfs_blkdev_get(mp, mp->m_logname, &logdev);
 		if (error)
 			goto out;
 	}
 
-	if (args->rtname[0]) {
-		error = xfs_blkdev_get(mp, args->rtname, &rtdev);
+	if (mp->m_rtname) {
+		error = xfs_blkdev_get(mp, mp->m_rtname, &rtdev);
 		if (error)
 			goto out_close_logdev;
 
@@ -813,18 +865,18 @@ xfs_setup_devices(
  */
 void
 xfsaild_wakeup(
-	xfs_mount_t		*mp,
+	struct xfs_ail		*ailp,
 	xfs_lsn_t		threshold_lsn)
 {
-	mp->m_ail.xa_target = threshold_lsn;
-	wake_up_process(mp->m_ail.xa_task);
+	ailp->xa_target = threshold_lsn;
+	wake_up_process(ailp->xa_task);
 }
 
 int
 xfsaild(
 	void	*data)
 {
-	xfs_mount_t	*mp = (xfs_mount_t *)data;
+	struct xfs_ail	*ailp = data;
 	xfs_lsn_t	last_pushed_lsn = 0;
 	long		tout = 0;
 
@@ -836,11 +888,11 @@ xfsaild(
 		/* swsusp */
 		try_to_freeze();
 
-		ASSERT(mp->m_log);
-		if (XFS_FORCED_SHUTDOWN(mp))
+		ASSERT(ailp->xa_mount->m_log);
+		if (XFS_FORCED_SHUTDOWN(ailp->xa_mount))
 			continue;
 
-		tout = xfsaild_push(mp, &last_pushed_lsn);
+		tout = xfsaild_push(ailp, &last_pushed_lsn);
 	}
 
 	return 0;
@@ -848,43 +900,82 @@ xfsaild(
 
 int
 xfsaild_start(
-	xfs_mount_t	*mp)
+	struct xfs_ail	*ailp)
 {
-	mp->m_ail.xa_target = 0;
-	mp->m_ail.xa_task = kthread_run(xfsaild, mp, "xfsaild");
-	if (IS_ERR(mp->m_ail.xa_task))
-		return -PTR_ERR(mp->m_ail.xa_task);
+	ailp->xa_target = 0;
+	ailp->xa_task = kthread_run(xfsaild, ailp, "xfsaild");
+	if (IS_ERR(ailp->xa_task))
+		return -PTR_ERR(ailp->xa_task);
 	return 0;
 }
 
 void
 xfsaild_stop(
-	xfs_mount_t	*mp)
+	struct xfs_ail	*ailp)
 {
-	kthread_stop(mp->m_ail.xa_task);
+	kthread_stop(ailp->xa_task);
 }
 
 
-
+/* Catch misguided souls that try to use this interface on XFS */
 STATIC struct inode *
 xfs_fs_alloc_inode(
 	struct super_block	*sb)
 {
-	return kmem_zone_alloc(xfs_vnode_zone, KM_SLEEP);
+	BUG();
+	return NULL;
 }
 
+/*
+ * Now that the generic code is guaranteed not to be accessing
+ * the linux inode, we can reclaim the inode.
+ */
 STATIC void
 xfs_fs_destroy_inode(
-	struct inode		*inode)
+	struct inode	*inode)
 {
-	kmem_zone_free(xfs_vnode_zone, inode);
+	xfs_inode_t		*ip = XFS_I(inode);
+
+	XFS_STATS_INC(vn_reclaim);
+	if (xfs_reclaim(ip))
+		panic("%s: cannot reclaim 0x%p\n", __func__, inode);
 }
 
+/*
+ * Slab object creation initialisation for the XFS inode.
+ * This covers only the idempotent fields in the XFS inode;
+ * all other fields need to be initialised on allocation
+ * from the slab. This avoids the need to repeatedly intialise
+ * fields in the xfs inode that left in the initialise state
+ * when freeing the inode.
+ */
 STATIC void
 xfs_fs_inode_init_once(
-	void			*vnode)
+	void			*inode)
 {
-	inode_init_once((struct inode *)vnode);
+	struct xfs_inode	*ip = inode;
+
+	memset(ip, 0, sizeof(struct xfs_inode));
+
+	/* vfs inode */
+	inode_init_once(VFS_I(ip));
+
+	/* xfs inode */
+	atomic_set(&ip->i_iocount, 0);
+	atomic_set(&ip->i_pincount, 0);
+	spin_lock_init(&ip->i_flags_lock);
+	init_waitqueue_head(&ip->i_ipin_wait);
+	/*
+	 * Because we want to use a counting completion, complete
+	 * the flush completion once to allow a single access to
+	 * the flush completion without blocking.
+	 */
+	init_completion(&ip->i_flush);
+	complete(&ip->i_flush);
+
+	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
+		     "xfsino", ip->i_ino);
+	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
 }
 
 /*
@@ -898,21 +989,26 @@ xfs_fs_write_inode(
 	struct inode		*inode,
 	int			sync)
 {
+	struct xfs_inode	*ip = XFS_I(inode);
 	int			error = 0;
 	int			flags = 0;
 
-	xfs_itrace_entry(XFS_I(inode));
+	xfs_itrace_entry(ip);
 	if (sync) {
-		filemap_fdatawait(inode->i_mapping);
+		error = xfs_wait_on_pages(ip, 0, -1);
+		if (error)
+			goto out_error;
 		flags |= FLUSH_SYNC;
 	}
-	error = xfs_inode_flush(XFS_I(inode), flags);
+	error = xfs_inode_flush(ip, flags);
+
+out_error:
 	/*
 	 * if we failed to write out the inode then mark
 	 * it dirty again so we'll try again later.
 	 */
 	if (error)
-		mark_inode_dirty_sync(inode);
+		xfs_mark_inode_dirty_sync(ip);
 
 	return -error;
 }
@@ -923,164 +1019,12 @@ xfs_fs_clear_inode(
 {
 	xfs_inode_t		*ip = XFS_I(inode);
 
-	/*
-	 * ip can be null when xfs_iget_core calls xfs_idestroy if we
-	 * find an inode with di_mode == 0 but without IGET_CREATE set.
-	 */
-	if (ip) {
-		xfs_itrace_entry(ip);
-		XFS_STATS_INC(vn_rele);
-		XFS_STATS_INC(vn_remove);
-		XFS_STATS_INC(vn_reclaim);
-		XFS_STATS_DEC(vn_active);
-
-		xfs_inactive(ip);
-		xfs_iflags_clear(ip, XFS_IMODIFIED);
-		if (xfs_reclaim(ip))
-			panic("%s: cannot reclaim 0x%p\n", __func__, inode);
-	}
-
-	ASSERT(XFS_I(inode) == NULL);
-}
+	xfs_itrace_entry(ip);
+	XFS_STATS_INC(vn_rele);
+	XFS_STATS_INC(vn_remove);
+	XFS_STATS_DEC(vn_active);
 
-/*
- * Enqueue a work item to be picked up by the vfs xfssyncd thread.
- * Doing this has two advantages:
- * - It saves on stack space, which is tight in certain situations
- * - It can be used (with care) as a mechanism to avoid deadlocks.
- * Flushing while allocating in a full filesystem requires both.
- */
-STATIC void
-xfs_syncd_queue_work(
-	struct xfs_mount *mp,
-	void		*data,
-	void		(*syncer)(struct xfs_mount *, void *))
-{
-	struct bhv_vfs_sync_work *work;
-
-	work = kmem_alloc(sizeof(struct bhv_vfs_sync_work), KM_SLEEP);
-	INIT_LIST_HEAD(&work->w_list);
-	work->w_syncer = syncer;
-	work->w_data = data;
-	work->w_mount = mp;
-	spin_lock(&mp->m_sync_lock);
-	list_add_tail(&work->w_list, &mp->m_sync_list);
-	spin_unlock(&mp->m_sync_lock);
-	wake_up_process(mp->m_sync_task);
-}
-
-/*
- * Flush delayed allocate data, attempting to free up reserved space
- * from existing allocations.  At this point a new allocation attempt
- * has failed with ENOSPC and we are in the process of scratching our
- * heads, looking about for more room...
- */
-STATIC void
-xfs_flush_inode_work(
-	struct xfs_mount *mp,
-	void		*arg)
-{
-	struct inode	*inode = arg;
-	filemap_flush(inode->i_mapping);
-	iput(inode);
-}
-
-void
-xfs_flush_inode(
-	xfs_inode_t	*ip)
-{
-	struct inode	*inode = VFS_I(ip);
-
-	igrab(inode);
-	xfs_syncd_queue_work(ip->i_mount, inode, xfs_flush_inode_work);
-	delay(msecs_to_jiffies(500));
-}
-
-/*
- * This is the "bigger hammer" version of xfs_flush_inode_work...
- * (IOW, "If at first you don't succeed, use a Bigger Hammer").
- */
-STATIC void
-xfs_flush_device_work(
-	struct xfs_mount *mp,
-	void		*arg)
-{
-	struct inode	*inode = arg;
-	sync_blockdev(mp->m_super->s_bdev);
-	iput(inode);
-}
-
-void
-xfs_flush_device(
-	xfs_inode_t	*ip)
-{
-	struct inode	*inode = VFS_I(ip);
-
-	igrab(inode);
-	xfs_syncd_queue_work(ip->i_mount, inode, xfs_flush_device_work);
-	delay(msecs_to_jiffies(500));
-	xfs_log_force(ip->i_mount, (xfs_lsn_t)0, XFS_LOG_FORCE|XFS_LOG_SYNC);
-}
-
-STATIC void
-xfs_sync_worker(
-	struct xfs_mount *mp,
-	void		*unused)
-{
-	int		error;
-
-	if (!(mp->m_flags & XFS_MOUNT_RDONLY))
-		error = xfs_sync(mp, SYNC_FSDATA | SYNC_BDFLUSH | SYNC_ATTR);
-	mp->m_sync_seq++;
-	wake_up(&mp->m_wait_single_sync_task);
-}
-
-STATIC int
-xfssyncd(
-	void			*arg)
-{
-	struct xfs_mount	*mp = arg;
-	long			timeleft;
-	bhv_vfs_sync_work_t	*work, *n;
-	LIST_HEAD		(tmp);
-
-	set_freezable();
-	timeleft = xfs_syncd_centisecs * msecs_to_jiffies(10);
-	for (;;) {
-		timeleft = schedule_timeout_interruptible(timeleft);
-		/* swsusp */
-		try_to_freeze();
-		if (kthread_should_stop() && list_empty(&mp->m_sync_list))
-			break;
-
-		spin_lock(&mp->m_sync_lock);
-		/*
-		 * We can get woken by laptop mode, to do a sync -
-		 * that's the (only!) case where the list would be
-		 * empty with time remaining.
-		 */
-		if (!timeleft || list_empty(&mp->m_sync_list)) {
-			if (!timeleft)
-				timeleft = xfs_syncd_centisecs *
-							msecs_to_jiffies(10);
-			INIT_LIST_HEAD(&mp->m_sync_work.w_list);
-			list_add_tail(&mp->m_sync_work.w_list,
-					&mp->m_sync_list);
-		}
-		list_for_each_entry_safe(work, n, &mp->m_sync_list, w_list)
-			list_move(&work->w_list, &tmp);
-		spin_unlock(&mp->m_sync_lock);
-
-		list_for_each_entry_safe(work, n, &tmp, w_list) {
-			(*work->w_syncer)(mp, work->w_data);
-			list_del(&work->w_list);
-			if (work == &mp->m_sync_work)
-				continue;
-			kmem_free(work);
-		}
-	}
-
-	return 0;
+	xfs_inactive(ip);
 }
 
 STATIC void
@@ -1099,11 +1043,9 @@ xfs_fs_put_super(
 	struct xfs_mount	*mp = XFS_M(sb);
 	struct xfs_inode	*rip = mp->m_rootip;
 	int			unmount_event_flags = 0;
-	int			error;
 
-	kthread_stop(mp->m_sync_task);
-
-	xfs_sync(mp, SYNC_ATTR | SYNC_DELWRI);
+	xfs_syncd_stop(mp);
+	xfs_sync_inodes(mp, SYNC_ATTR|SYNC_DELWRI);
 
 #ifdef HAVE_DMAPI
 	if (mp->m_flags & XFS_MOUNT_DMAPI) {
@@ -1128,18 +1070,6 @@ xfs_fs_put_super(
 	xfs_filestream_unmount(mp);
 
 	XFS_bflush(mp->m_ddev_targp);
-	error = xfs_unmount_flush(mp, 0);
-	WARN_ON(error);
-
-	/*
-	 * If we're forcing a shutdown, typically because of a media error,
-	 * we want to make sure we invalidate dirty pages that belong to
-	 * referenced vnodes as well.
-	 */
-	if (XFS_FORCED_SHUTDOWN(mp)) {
-		error = xfs_sync(mp, SYNC_WAIT | SYNC_CLOSE);
-		ASSERT(error != EFSCORRUPTED);
-	}
 
 	if (mp->m_flags & XFS_MOUNT_DMAPI) {
 		XFS_SEND_UNMOUNT(mp, rip, DM_RIGHT_NULL, 0, 0,
@@ -1161,7 +1091,7 @@ xfs_fs_write_super(
 	struct super_block	*sb)
 {
 	if (!(sb->s_flags & MS_RDONLY))
-		xfs_sync(XFS_M(sb), SYNC_FSDATA);
+		xfs_sync_fsdata(XFS_M(sb), 0);
 	sb->s_dirt = 0;
 }
 
@@ -1172,7 +1102,6 @@ xfs_fs_sync_super(
 {
 	struct xfs_mount	*mp = XFS_M(sb);
 	int			error;
-	int			flags;
 
 	/*
 	 * Treat a sync operation like a freeze.  This is to work
@@ -1186,20 +1115,10 @@ xfs_fs_sync_super(
 	 * dirty the Linux inode until after the transaction I/O
 	 * completes.
 	 */
-	if (wait || unlikely(sb->s_frozen == SB_FREEZE_WRITE)) {
-		/*
-		 * First stage of freeze - no more writers will make progress
-		 * now we are here, so we flush delwri and delalloc buffers
-		 * here, then wait for all I/O to complete.  Data is frozen at
-		 * that point. Metadata is not frozen, transactions can still
-		 * occur here so don't bother flushing the buftarg (i.e
-		 * SYNC_QUIESCE) because it'll just get dirty again.
-		 */
-		flags = SYNC_DATA_QUIESCE;
-	} else
-		flags = SYNC_FSDATA;
-
-	error = xfs_sync(mp, flags);
+	if (wait || unlikely(sb->s_frozen == SB_FREEZE_WRITE))
+		error = xfs_quiesce_data(mp);
+	else
+		error = xfs_sync_fsdata(mp, 0);
 	sb->s_dirt = 0;
 
 	if (unlikely(laptop_mode)) {
@@ -1337,9 +1256,8 @@ xfs_fs_remount(
 
 	/* rw -> ro */
 	if (!(mp->m_flags & XFS_MOUNT_RDONLY) && (*flags & MS_RDONLY)) {
-		xfs_filestream_flush(mp);
-		xfs_sync(mp, SYNC_DATA_QUIESCE);
-		xfs_attr_quiesce(mp);
+		xfs_quiesce_data(mp);
+		xfs_quiesce_attr(mp);
 		mp->m_flags |= XFS_MOUNT_RDONLY;
 	}
 
@@ -1348,7 +1266,7 @@ xfs_fs_remount(
 
 /*
  * Second stage of a freeze. The data is already frozen so we only
- * need to take care of themetadata. Once that's done write a dummy
+ * need to take care of the metadata. Once that's done write a dummy
  * record to dirty the log in case of a crash while frozen.
  */
 STATIC void
@@ -1357,7 +1275,7 @@ xfs_fs_lockfs(
 {
 	struct xfs_mount	*mp = XFS_M(sb);
 
-	xfs_attr_quiesce(mp);
+	xfs_quiesce_attr(mp);
 	xfs_fs_log_dummy(mp);
 }
 
@@ -1422,175 +1340,28 @@ xfs_fs_setxquota(
 
 /*
  * This function fills in xfs_mount_t fields based on mount args.
- * Note: the superblock has _not_ yet been read in.
- */
-STATIC int
-xfs_start_flags(
-	struct xfs_mount_args	*ap,
-	struct xfs_mount	*mp)
-{
-	int			error;
-
-	/* Values are in BBs */
-	if ((ap->flags & XFSMNT_NOALIGN) != XFSMNT_NOALIGN) {
-		/*
-		 * At this point the superblock has not been read
-		 * in, therefore we do not know the block size.
-		 * Before the mount call ends we will convert
-		 * these to FSBs.
-		 */
-		mp->m_dalign = ap->sunit;
-		mp->m_swidth = ap->swidth;
-	}
-
-	if (ap->logbufs != -1 &&
-	    ap->logbufs != 0 &&
-	    (ap->logbufs < XLOG_MIN_ICLOGS ||
-	     ap->logbufs > XLOG_MAX_ICLOGS)) {
-		cmn_err(CE_WARN,
-			"XFS: invalid logbufs value: %d [not %d-%d]",
-			ap->logbufs, XLOG_MIN_ICLOGS, XLOG_MAX_ICLOGS);
-		return XFS_ERROR(EINVAL);
-	}
-	mp->m_logbufs = ap->logbufs;
-	if (ap->logbufsize != -1 &&
-	    ap->logbufsize !=  0 &&
-	    (ap->logbufsize < XLOG_MIN_RECORD_BSIZE ||
-	     ap->logbufsize > XLOG_MAX_RECORD_BSIZE ||
-	     !is_power_of_2(ap->logbufsize))) {
-		cmn_err(CE_WARN,
-	"XFS: invalid logbufsize: %d [not 16k,32k,64k,128k or 256k]",
-			ap->logbufsize);
-		return XFS_ERROR(EINVAL);
-	}
-
-	error = ENOMEM;
-
-	mp->m_logbsize = ap->logbufsize;
-	mp->m_fsname_len = strlen(ap->fsname) + 1;
-
-	mp->m_fsname = kstrdup(ap->fsname, GFP_KERNEL);
-	if (!mp->m_fsname)
-		goto out;
-
-	if (ap->rtname[0]) {
-		mp->m_rtname = kstrdup(ap->rtname, GFP_KERNEL);
-		if (!mp->m_rtname)
-			goto out_free_fsname;
-
-	}
-
-	if (ap->logname[0]) {
-		mp->m_logname = kstrdup(ap->logname, GFP_KERNEL);
-		if (!mp->m_logname)
-			goto out_free_rtname;
-	}
-
-	if (ap->flags & XFSMNT_WSYNC)
-		mp->m_flags |= XFS_MOUNT_WSYNC;
-#if XFS_BIG_INUMS
-	if (ap->flags & XFSMNT_INO64) {
-		mp->m_flags |= XFS_MOUNT_INO64;
-		mp->m_inoadd = XFS_INO64_OFFSET;
-	}
-#endif
-	if (ap->flags & XFSMNT_RETERR)
-		mp->m_flags |= XFS_MOUNT_RETERR;
-	if (ap->flags & XFSMNT_NOALIGN)
-		mp->m_flags |= XFS_MOUNT_NOALIGN;
-	if (ap->flags & XFSMNT_SWALLOC)
-		mp->m_flags |= XFS_MOUNT_SWALLOC;
-	if (ap->flags & XFSMNT_OSYNCISOSYNC)
-		mp->m_flags |= XFS_MOUNT_OSYNCISOSYNC;
-	if (ap->flags & XFSMNT_32BITINODES)
-		mp->m_flags |= XFS_MOUNT_32BITINODES;
-
-	if (ap->flags & XFSMNT_IOSIZE) {
-		if (ap->iosizelog > XFS_MAX_IO_LOG ||
-		    ap->iosizelog < XFS_MIN_IO_LOG) {
-			cmn_err(CE_WARN,
-		"XFS: invalid log iosize: %d [not %d-%d]",
-				ap->iosizelog, XFS_MIN_IO_LOG,
-				XFS_MAX_IO_LOG);
-			return XFS_ERROR(EINVAL);
-		}
-
-		mp->m_flags |= XFS_MOUNT_DFLT_IOSIZE;
-		mp->m_readio_log = mp->m_writeio_log = ap->iosizelog;
-	}
-
-	if (ap->flags & XFSMNT_IKEEP)
-		mp->m_flags |= XFS_MOUNT_IKEEP;
-	if (ap->flags & XFSMNT_DIRSYNC)
-		mp->m_flags |= XFS_MOUNT_DIRSYNC;
-	if (ap->flags & XFSMNT_ATTR2)
-		mp->m_flags |= XFS_MOUNT_ATTR2;
-	if (ap->flags & XFSMNT_NOATTR2)
-		mp->m_flags |= XFS_MOUNT_NOATTR2;
-
-	if (ap->flags2 & XFSMNT2_COMPAT_IOSIZE)
-		mp->m_flags |= XFS_MOUNT_COMPAT_IOSIZE;
-
-	/*
-	 * no recovery flag requires a read-only mount
-	 */
-	if (ap->flags & XFSMNT_NORECOVERY) {
-		if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
-			cmn_err(CE_WARN,
-	"XFS: tried to mount a FS read-write without recovery!");
-			return XFS_ERROR(EINVAL);
-		}
-		mp->m_flags |= XFS_MOUNT_NORECOVERY;
-	}
-
-	if (ap->flags & XFSMNT_NOUUID)
-		mp->m_flags |= XFS_MOUNT_NOUUID;
-	if (ap->flags & XFSMNT_BARRIER)
-		mp->m_flags |= XFS_MOUNT_BARRIER;
-	else
-		mp->m_flags &= ~XFS_MOUNT_BARRIER;
-
-	if (ap->flags2 & XFSMNT2_FILESTREAMS)
-		mp->m_flags |= XFS_MOUNT_FILESTREAMS;
-
-	if (ap->flags & XFSMNT_DMAPI)
-		mp->m_flags |= XFS_MOUNT_DMAPI;
-	return 0;
-
-
- out_free_rtname:
-	kfree(mp->m_rtname);
- out_free_fsname:
-	kfree(mp->m_fsname);
- out:
-	return error;
-}
-
-/*
- * This function fills in xfs_mount_t fields based on mount args.
  * Note: the superblock _has_ now been read in.
  */
 STATIC int
 xfs_finish_flags(
-	struct xfs_mount_args	*ap,
 	struct xfs_mount	*mp)
 {
 	int			ronly = (mp->m_flags & XFS_MOUNT_RDONLY);
 
 	/* Fail a mount where the logbuf is smaller then the log stripe */
 	if (xfs_sb_version_haslogv2(&mp->m_sb)) {
-		if ((ap->logbufsize <= 0) &&
-		    (mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE)) {
+		if (mp->m_logbsize <= 0 &&
+		    mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE) {
 			mp->m_logbsize = mp->m_sb.sb_logsunit;
-		} else if (ap->logbufsize > 0 &&
-			   ap->logbufsize < mp->m_sb.sb_logsunit) {
+		} else if (mp->m_logbsize > 0 &&
+			   mp->m_logbsize < mp->m_sb.sb_logsunit) {
 			cmn_err(CE_WARN,
 	"XFS: logbuf size must be greater than or equal to log stripe size");
 			return XFS_ERROR(EINVAL);
 		}
 	} else {
 		/* Fail a mount if the logbuf is larger than 32K */
-		if (ap->logbufsize > XLOG_BIG_RECORD_BSIZE) {
+		if (mp->m_logbsize > XLOG_BIG_RECORD_BSIZE) {
 			cmn_err(CE_WARN,
 	"XFS: logbuf size for version 1 logs must be 16K or 32K");
 			return XFS_ERROR(EINVAL);
@@ -1602,7 +1373,7 @@ xfs_finish_flags(
 	 * told by noattr2 to turn it off
 	 */
 	if (xfs_sb_version_hasattr2(&mp->m_sb) &&
-	    !(ap->flags & XFSMNT_NOATTR2))
+	    !(mp->m_flags & XFS_MOUNT_NOATTR2))
 		mp->m_flags |= XFS_MOUNT_ATTR2;
 
 	/*
@@ -1614,48 +1385,6 @@ xfs_finish_flags(
 		return XFS_ERROR(EROFS);
 	}
 
-	/*
-	 * check for shared mount.
-	 */
-	if (ap->flags & XFSMNT_SHARED) {
-		if (!xfs_sb_version_hasshared(&mp->m_sb))
-			return XFS_ERROR(EINVAL);
-
-		/*
-		 * For IRIX 6.5, shared mounts must have the shared
-		 * version bit set, have the persistent readonly
-		 * field set, must be version 0 and can only be mounted
-		 * read-only.
-		 */
-		if (!ronly || !(mp->m_sb.sb_flags & XFS_SBF_READONLY) ||
-		     (mp->m_sb.sb_shared_vn != 0))
-			return XFS_ERROR(EINVAL);
-
-		mp->m_flags |= XFS_MOUNT_SHARED;
-
-		/*
-		 * Shared XFS V0 can't deal with DMI.  Return EINVAL.
-		 */
-		if (mp->m_sb.sb_shared_vn == 0 && (ap->flags & XFSMNT_DMAPI))
-			return XFS_ERROR(EINVAL);
-	}
-
-	if (ap->flags & XFSMNT_UQUOTA) {
-		mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ACTIVE);
-		if (ap->flags & XFSMNT_UQUOTAENF)
-			mp->m_qflags |= XFS_UQUOTA_ENFD;
-	}
-
-	if (ap->flags & XFSMNT_GQUOTA) {
-		mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ACTIVE);
-		if (ap->flags & XFSMNT_GQUOTAENF)
-			mp->m_qflags |= XFS_OQUOTA_ENFD;
-	} else if (ap->flags & XFSMNT_PQUOTA) {
-		mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ACTIVE);
-		if (ap->flags & XFSMNT_PQUOTAENF)
-			mp->m_qflags |= XFS_OQUOTA_ENFD;
-	}
-
 	return 0;
 }
 
@@ -1667,19 +1396,14 @@ xfs_fs_fill_super(
 {
 	struct inode		*root;
 	struct xfs_mount	*mp = NULL;
-	struct xfs_mount_args	*args;
 	int			flags = 0, error = ENOMEM;
-
-	args = xfs_args_allocate(sb, silent);
-	if (!args)
-		return -ENOMEM;
+	char			*mtpt = NULL;
 
 	mp = kzalloc(sizeof(struct xfs_mount), GFP_KERNEL);
 	if (!mp)
-		goto out_free_args;
+		goto out;
 
 	spin_lock_init(&mp->m_sb_lock);
-	mutex_init(&mp->m_ilock);
 	mutex_init(&mp->m_growlock);
 	atomic_set(&mp->m_active_trans, 0);
 	INIT_LIST_HEAD(&mp->m_sync_list);
@@ -1689,12 +1413,9 @@ xfs_fs_fill_super(
 	mp->m_super = sb;
 	sb->s_fs_info = mp;
 
-	if (sb->s_flags & MS_RDONLY)
-		mp->m_flags |= XFS_MOUNT_RDONLY;
-
-	error = xfs_parseargs(mp, (char *)data, args, 0);
+	error = xfs_parseargs(mp, (char *)data, &mtpt);
 	if (error)
-		goto out_free_mp;
+		goto out_free_fsname;
 
 	sb_min_blocksize(sb, BBSIZE);
 	sb->s_xattr = xfs_xattr_handlers;
@@ -1702,33 +1423,28 @@ xfs_fs_fill_super(
 	sb->s_qcop = &xfs_quotactl_operations;
 	sb->s_op = &xfs_super_operations;
 
-	error = xfs_dmops_get(mp, args);
+	error = xfs_dmops_get(mp);
 	if (error)
-		goto out_free_mp;
-	error = xfs_qmops_get(mp, args);
+		goto out_free_fsname;
+	error = xfs_qmops_get(mp);
 	if (error)
 		goto out_put_dmops;
 
-	if (args->flags & XFSMNT_QUIET)
+	if (silent)
 		flags |= XFS_MFSI_QUIET;
 
-	error = xfs_open_devices(mp, args);
+	error = xfs_open_devices(mp);
 	if (error)
 		goto out_put_qmops;
 
 	if (xfs_icsb_init_counters(mp))
 		mp->m_flags |= XFS_MOUNT_NO_PERCPU_SB;
 
-	/*
-	 * Setup flags based on mount(2) options and then the superblock
-	 */
-	error = xfs_start_flags(args, mp);
-	if (error)
-		goto out_free_fsname;
 	error = xfs_readsb(mp, flags);
 	if (error)
-		goto out_free_fsname;
-	error = xfs_finish_flags(args, mp);
+		goto out_destroy_counters;
+
+	error = xfs_finish_flags(mp);
 	if (error)
 		goto out_free_sb;
 
@@ -1747,7 +1463,7 @@ xfs_fs_fill_super(
 	if (error)
 		goto out_filestream_unmount;
 
-	XFS_SEND_MOUNT(mp, DM_RIGHT_NULL, args->mtpt, args->fsname);
+	XFS_SEND_MOUNT(mp, DM_RIGHT_NULL, mtpt, mp->m_fsname);
 
 	sb->s_dirt = 1;
 	sb->s_magic = XFS_SB_MAGIC;
@@ -1772,35 +1488,31 @@ xfs_fs_fill_super(
 		goto fail_vnrele;
 	}
 
-	mp->m_sync_work.w_syncer = xfs_sync_worker;
-	mp->m_sync_work.w_mount = mp;
-	mp->m_sync_task = kthread_run(xfssyncd, mp, "xfssyncd");
-	if (IS_ERR(mp->m_sync_task)) {
-		error = -PTR_ERR(mp->m_sync_task);
+	error = xfs_syncd_init(mp);
+	if (error)
 		goto fail_vnrele;
-	}
 
-	xfs_itrace_exit(XFS_I(sb->s_root->d_inode));
+	kfree(mtpt);
 
-	kfree(args);
+	xfs_itrace_exit(XFS_I(sb->s_root->d_inode));
 	return 0;
 
  out_filestream_unmount:
 	xfs_filestream_unmount(mp);
  out_free_sb:
 	xfs_freesb(mp);
- out_free_fsname:
-	xfs_free_fsname(mp);
+ out_destroy_counters:
 	xfs_icsb_destroy_counters(mp);
 	xfs_close_devices(mp);
  out_put_qmops:
 	xfs_qmops_put(mp);
  out_put_dmops:
 	xfs_dmops_put(mp);
- out_free_mp:
+ out_free_fsname:
+	xfs_free_fsname(mp);
+	kfree(mtpt);
 	kfree(mp);
- out_free_args:
-	kfree(args);
+ out:
 	return -error;
 
  fail_vnrele:
@@ -1820,8 +1532,6 @@ xfs_fs_fill_super(
 	xfs_filestream_unmount(mp);
 
 	XFS_bflush(mp->m_ddev_targp);
-	error = xfs_unmount_flush(mp, 0);
-	WARN_ON(error);
 
 	xfs_unmountfs(mp);
 	goto out_free_sb;
@@ -1882,10 +1592,19 @@ xfs_alloc_trace_bufs(void)
 	if (!xfs_bmap_trace_buf)
 		goto out_free_alloc_trace;
 #endif
-#ifdef XFS_BMBT_TRACE
+#ifdef XFS_BTREE_TRACE
+	xfs_allocbt_trace_buf = ktrace_alloc(XFS_ALLOCBT_TRACE_SIZE,
+					     KM_MAYFAIL);
+	if (!xfs_allocbt_trace_buf)
+		goto out_free_bmap_trace;
+
+	xfs_inobt_trace_buf = ktrace_alloc(XFS_INOBT_TRACE_SIZE, KM_MAYFAIL);
+	if (!xfs_inobt_trace_buf)
+		goto out_free_allocbt_trace;
+
 	xfs_bmbt_trace_buf = ktrace_alloc(XFS_BMBT_TRACE_SIZE, KM_MAYFAIL);
 	if (!xfs_bmbt_trace_buf)
-		goto out_free_bmap_trace;
+		goto out_free_inobt_trace;
 #endif
 #ifdef XFS_ATTR_TRACE
 	xfs_attr_trace_buf = ktrace_alloc(XFS_ATTR_TRACE_SIZE, KM_MAYFAIL);
@@ -1907,8 +1626,12 @@ xfs_alloc_trace_bufs(void)
 	ktrace_free(xfs_attr_trace_buf);
  out_free_bmbt_trace:
 #endif
-#ifdef XFS_BMBT_TRACE
+#ifdef XFS_BTREE_TRACE
 	ktrace_free(xfs_bmbt_trace_buf);
+ out_free_inobt_trace:
+	ktrace_free(xfs_inobt_trace_buf);
+ out_free_allocbt_trace:
+	ktrace_free(xfs_allocbt_trace_buf);
  out_free_bmap_trace:
 #endif
 #ifdef XFS_BMAP_TRACE
@@ -1931,8 +1654,10 @@ xfs_free_trace_bufs(void)
 #ifdef XFS_ATTR_TRACE
 	ktrace_free(xfs_attr_trace_buf);
 #endif
-#ifdef XFS_BMBT_TRACE
+#ifdef XFS_BTREE_TRACE
 	ktrace_free(xfs_bmbt_trace_buf);
+	ktrace_free(xfs_inobt_trace_buf);
+	ktrace_free(xfs_allocbt_trace_buf);
 #endif
 #ifdef XFS_BMAP_TRACE
 	ktrace_free(xfs_bmap_trace_buf);
@@ -1945,16 +1670,10 @@ xfs_free_trace_bufs(void)
 STATIC int __init
 xfs_init_zones(void)
 {
-	xfs_vnode_zone = kmem_zone_init_flags(sizeof(struct inode), "xfs_vnode",
-					KM_ZONE_HWALIGN | KM_ZONE_RECLAIM |
-					KM_ZONE_SPREAD,
-					xfs_fs_inode_init_once);
-	if (!xfs_vnode_zone)
-		goto out;
 
 	xfs_ioend_zone = kmem_zone_init(sizeof(xfs_ioend_t), "xfs_ioend");
 	if (!xfs_ioend_zone)
-		goto out_destroy_vnode_zone;
+		goto out;
 
 	xfs_ioend_pool = mempool_create_slab_pool(4 * MAX_BUF_PER_PAGE,
 						  xfs_ioend_zone);
@@ -1970,6 +1689,7 @@ xfs_init_zones(void)
 						"xfs_bmap_free_item");
 	if (!xfs_bmap_free_item_zone)
 		goto out_destroy_log_ticket_zone;
+
 	xfs_btree_cur_zone = kmem_zone_init(sizeof(xfs_btree_cur_t),
 						"xfs_btree_cur");
 	if (!xfs_btree_cur_zone)
@@ -2017,8 +1737,8 @@ xfs_init_zones(void)
 
 	xfs_inode_zone =
 		kmem_zone_init_flags(sizeof(xfs_inode_t), "xfs_inode",
-					KM_ZONE_HWALIGN | KM_ZONE_RECLAIM |
-					KM_ZONE_SPREAD, NULL);
+			KM_ZONE_HWALIGN | KM_ZONE_RECLAIM | KM_ZONE_SPREAD,
+			xfs_fs_inode_init_once);
 	if (!xfs_inode_zone)
 		goto out_destroy_efi_zone;
 
@@ -2066,8 +1786,6 @@ xfs_init_zones(void)
 	mempool_destroy(xfs_ioend_pool);
  out_destroy_ioend_zone:
 	kmem_zone_destroy(xfs_ioend_zone);
- out_destroy_vnode_zone:
-	kmem_zone_destroy(xfs_vnode_zone);
  out:
 	return -ENOMEM;
 }
@@ -2092,7 +1810,6 @@ xfs_destroy_zones(void)
 	kmem_zone_destroy(xfs_log_ticket_zone);
 	mempool_destroy(xfs_ioend_pool);
 	kmem_zone_destroy(xfs_ioend_zone);
-	kmem_zone_destroy(xfs_vnode_zone);
 
 }
 
@@ -2100,13 +1817,12 @@ STATIC int __init
 init_xfs_fs(void)
 {
 	int			error;
-	static char		message[] __initdata = KERN_INFO \
-		XFS_VERSION_STRING " with " XFS_BUILD_OPTIONS " enabled\n";
 
-	printk(message);
+	printk(KERN_INFO XFS_VERSION_STRING " with "
+			 XFS_BUILD_OPTIONS " enabled\n");
 
 	ktrace_init(64);
-	vn_init();
+	xfs_ioend_init();
 	xfs_dir_startup();
 
 	error = xfs_init_zones();
diff --git a/fs/xfs/linux-2.6/xfs_super.h b/fs/xfs/linux-2.6/xfs_super.h
index fe2ef4e6a0f..d5d776d4cd6 100644
--- a/fs/xfs/linux-2.6/xfs_super.h
+++ b/fs/xfs/linux-2.6/xfs_super.h
@@ -20,24 +20,12 @@
 
 #include <linux/exportfs.h>
 
-#ifdef CONFIG_XFS_DMAPI
-# define vfs_insertdmapi(vfs)	vfs_insertops(vfsp, &xfs_dmops)
-# define vfs_initdmapi()	dmapi_init()
-# define vfs_exitdmapi()	dmapi_uninit()
-#else
-# define vfs_insertdmapi(vfs)	do { } while (0)
-# define vfs_initdmapi()	do { } while (0)
-# define vfs_exitdmapi()	do { } while (0)
-#endif
-
 #ifdef CONFIG_XFS_QUOTA
-# define vfs_insertquota(vfs)	vfs_insertops(vfsp, &xfs_qmops)
 extern void xfs_qm_init(void);
 extern void xfs_qm_exit(void);
 # define vfs_initquota()	xfs_qm_init()
 # define vfs_exitquota()	xfs_qm_exit()
 #else
-# define vfs_insertquota(vfs)	do { } while (0)
 # define vfs_initquota()	do { } while (0)
 # define vfs_exitquota()	do { } while (0)
 #endif
@@ -101,9 +89,6 @@ struct block_device;
 
 extern __uint64_t xfs_max_file_offset(unsigned int);
 
-extern void xfs_flush_inode(struct xfs_inode *);
-extern void xfs_flush_device(struct xfs_inode *);
-
 extern void xfs_blkdev_issue_flush(struct xfs_buftarg *);
 
 extern const struct export_operations xfs_export_operations;
diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c
new file mode 100644
index 00000000000..2ed035354c2
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_sync.c
@@ -0,0 +1,762 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_types.h"
+#include "xfs_bit.h"
+#include "xfs_log.h"
+#include "xfs_inum.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_dmapi.h"
+#include "xfs_mount.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_btree.h"
+#include "xfs_dir2_sf.h"
+#include "xfs_attr_sf.h"
+#include "xfs_inode.h"
+#include "xfs_dinode.h"
+#include "xfs_error.h"
+#include "xfs_mru_cache.h"
+#include "xfs_filestream.h"
+#include "xfs_vnodeops.h"
+#include "xfs_utils.h"
+#include "xfs_buf_item.h"
+#include "xfs_inode_item.h"
+#include "xfs_rw.h"
+
+#include <linux/kthread.h>
+#include <linux/freezer.h>
+
+/*
+ * Sync all the inodes in the given AG according to the
+ * direction given by the flags.
+ */
+STATIC int
+xfs_sync_inodes_ag(
+	xfs_mount_t	*mp,
+	int		ag,
+	int		flags)
+{
+	xfs_perag_t	*pag = &mp->m_perag[ag];
+	int		nr_found;
+	uint32_t	first_index = 0;
+	int		error = 0;
+	int		last_error = 0;
+	int		fflag = XFS_B_ASYNC;
+
+	if (flags & SYNC_DELWRI)
+		fflag = XFS_B_DELWRI;
+	if (flags & SYNC_WAIT)
+		fflag = 0;		/* synchronous overrides all */
+
+	do {
+		struct inode	*inode;
+		xfs_inode_t	*ip = NULL;
+		int		lock_flags = XFS_ILOCK_SHARED;
+
+		/*
+		 * use a gang lookup to find the next inode in the tree
+		 * as the tree is sparse and a gang lookup walks to find
+		 * the number of objects requested.
+		 */
+		read_lock(&pag->pag_ici_lock);
+		nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
+				(void**)&ip, first_index, 1);
+
+		if (!nr_found) {
+			read_unlock(&pag->pag_ici_lock);
+			break;
+		}
+
+		/*
+		 * Update the index for the next lookup. Catch overflows
+		 * into the next AG range which can occur if we have inodes
+		 * in the last block of the AG and we are currently
+		 * pointing to the last inode.
+		 */
+		first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
+		if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino)) {
+			read_unlock(&pag->pag_ici_lock);
+			break;
+		}
+
+		/* nothing to sync during shutdown */
+		if (XFS_FORCED_SHUTDOWN(mp)) {
+			read_unlock(&pag->pag_ici_lock);
+			return 0;
+		}
+
+		/*
+		 * If we can't get a reference on the inode, it must be
+		 * in reclaim. Leave it for the reclaim code to flush.
+		 */
+		inode = VFS_I(ip);
+		if (!igrab(inode)) {
+			read_unlock(&pag->pag_ici_lock);
+			continue;
+		}
+		read_unlock(&pag->pag_ici_lock);
+
+		/* avoid new or bad inodes */
+		if (is_bad_inode(inode) ||
+		    xfs_iflags_test(ip, XFS_INEW)) {
+			IRELE(ip);
+			continue;
+		}
+
+		/*
+		 * If we have to flush data or wait for I/O completion
+		 * we need to hold the iolock.
+		 */
+		if ((flags & SYNC_DELWRI) && VN_DIRTY(inode)) {
+			xfs_ilock(ip, XFS_IOLOCK_SHARED);
+			lock_flags |= XFS_IOLOCK_SHARED;
+			error = xfs_flush_pages(ip, 0, -1, fflag, FI_NONE);
+			if (flags & SYNC_IOWAIT)
+				xfs_ioend_wait(ip);
+		}
+		xfs_ilock(ip, XFS_ILOCK_SHARED);
+
+		if ((flags & SYNC_ATTR) && !xfs_inode_clean(ip)) {
+			if (flags & SYNC_WAIT) {
+				xfs_iflock(ip);
+				if (!xfs_inode_clean(ip))
+					error = xfs_iflush(ip, XFS_IFLUSH_SYNC);
+				else
+					xfs_ifunlock(ip);
+			} else if (xfs_iflock_nowait(ip)) {
+				if (!xfs_inode_clean(ip))
+					error = xfs_iflush(ip, XFS_IFLUSH_DELWRI);
+				else
+					xfs_ifunlock(ip);
+			}
+		}
+		xfs_iput(ip, lock_flags);
+
+		if (error)
+			last_error = error;
+		/*
+		 * bail out if the filesystem is corrupted.
+		 */
+		if (error == EFSCORRUPTED)
+			return XFS_ERROR(error);
+
+	} while (nr_found);
+
+	return last_error;
+}
+
+int
+xfs_sync_inodes(
+	xfs_mount_t	*mp,
+	int		flags)
+{
+	int		error;
+	int		last_error;
+	int		i;
+	int		lflags = XFS_LOG_FORCE;
+
+	if (mp->m_flags & XFS_MOUNT_RDONLY)
+		return 0;
+	error = 0;
+	last_error = 0;
+
+	if (flags & SYNC_WAIT)
+		lflags |= XFS_LOG_SYNC;
+
+	for (i = 0; i < mp->m_sb.sb_agcount; i++) {
+		if (!mp->m_perag[i].pag_ici_init)
+			continue;
+		error = xfs_sync_inodes_ag(mp, i, flags);
+		if (error)
+			last_error = error;
+		if (error == EFSCORRUPTED)
+			break;
+	}
+	if (flags & SYNC_DELWRI)
+		xfs_log_force(mp, 0, lflags);
+
+	return XFS_ERROR(last_error);
+}
+
+STATIC int
+xfs_commit_dummy_trans(
+	struct xfs_mount	*mp,
+	uint			log_flags)
+{
+	struct xfs_inode	*ip = mp->m_rootip;
+	struct xfs_trans	*tp;
+	int			error;
+
+	/*
+	 * Put a dummy transaction in the log to tell recovery
+	 * that all others are OK.
+	 */
+	tp = xfs_trans_alloc(mp, XFS_TRANS_DUMMY1);
+	error = xfs_trans_reserve(tp, 0, XFS_ICHANGE_LOG_RES(mp), 0, 0, 0);
+	if (error) {
+		xfs_trans_cancel(tp, 0);
+		return error;
+	}
+
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
+	xfs_trans_ihold(tp, ip);
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	/* XXX(hch): ignoring the error here.. */
+	error = xfs_trans_commit(tp, 0);
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+
+	xfs_log_force(mp, 0, log_flags);
+	return 0;
+}
+
+int
+xfs_sync_fsdata(
+	struct xfs_mount	*mp,
+	int			flags)
+{
+	struct xfs_buf		*bp;
+	struct xfs_buf_log_item	*bip;
+	int			error = 0;
+
+	/*
+	 * If this is xfssyncd() then only sync the superblock if we can
+	 * lock it without sleeping and it is not pinned.
+	 */
+	if (flags & SYNC_BDFLUSH) {
+		ASSERT(!(flags & SYNC_WAIT));
+
+		bp = xfs_getsb(mp, XFS_BUF_TRYLOCK);
+		if (!bp)
+			goto out;
+
+		bip = XFS_BUF_FSPRIVATE(bp, struct xfs_buf_log_item *);
+		if (!bip || !xfs_buf_item_dirty(bip) || XFS_BUF_ISPINNED(bp))
+			goto out_brelse;
+	} else {
+		bp = xfs_getsb(mp, 0);
+
+		/*
+		 * If the buffer is pinned then push on the log so we won't
+		 * get stuck waiting in the write for someone, maybe
+		 * ourselves, to flush the log.
+		 *
+		 * Even though we just pushed the log above, we did not have
+		 * the superblock buffer locked at that point so it can
+		 * become pinned in between there and here.
+		 */
+		if (XFS_BUF_ISPINNED(bp))
+			xfs_log_force(mp, 0, XFS_LOG_FORCE);
+	}
+
+
+	if (flags & SYNC_WAIT)
+		XFS_BUF_UNASYNC(bp);
+	else
+		XFS_BUF_ASYNC(bp);
+
+	return xfs_bwrite(mp, bp);
+
+ out_brelse:
+	xfs_buf_relse(bp);
+ out:
+	return error;
+}
+
+/*
+ * When remounting a filesystem read-only or freezing the filesystem, we have
+ * two phases to execute. This first phase is syncing the data before we
+ * quiesce the filesystem, and the second is flushing all the inodes out after
+ * we've waited for all the transactions created by the first phase to
+ * complete. The second phase ensures that the inodes are written to their
+ * location on disk rather than just existing in transactions in the log. This
+ * means after a quiesce there is no log replay required to write the inodes to
+ * disk (this is the main difference between a sync and a quiesce).
+ */
+/*
+ * First stage of freeze - no writers will make progress now we are here,
+ * so we flush delwri and delalloc buffers here, then wait for all I/O to
+ * complete.  Data is frozen at that point. Metadata is not frozen,
+ * transactions can still occur here so don't bother flushing the buftarg
+ * because it'll just get dirty again.
+ */
+int
+xfs_quiesce_data(
+	struct xfs_mount	*mp)
+{
+	int error;
+
+	/* push non-blocking */
+	xfs_sync_inodes(mp, SYNC_DELWRI|SYNC_BDFLUSH);
+	XFS_QM_DQSYNC(mp, SYNC_BDFLUSH);
+	xfs_filestream_flush(mp);
+
+	/* push and block */
+	xfs_sync_inodes(mp, SYNC_DELWRI|SYNC_WAIT|SYNC_IOWAIT);
+	XFS_QM_DQSYNC(mp, SYNC_WAIT);
+
+	/* write superblock and hoover up shutdown errors */
+	error = xfs_sync_fsdata(mp, 0);
+
+	/* flush data-only devices */
+	if (mp->m_rtdev_targp)
+		XFS_bflush(mp->m_rtdev_targp);
+
+	return error;
+}
+
+STATIC void
+xfs_quiesce_fs(
+	struct xfs_mount	*mp)
+{
+	int	count = 0, pincount;
+
+	xfs_flush_buftarg(mp->m_ddev_targp, 0);
+	xfs_reclaim_inodes(mp, 0, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
+
+	/*
+	 * This loop must run at least twice.  The first instance of the loop
+	 * will flush most meta data but that will generate more meta data
+	 * (typically directory updates).  Which then must be flushed and
+	 * logged before we can write the unmount record.
+	 */
+	do {
+		xfs_sync_inodes(mp, SYNC_ATTR|SYNC_WAIT);
+		pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
+		if (!pincount) {
+			delay(50);
+			count++;
+		}
+	} while (count < 2);
+}
+
+/*
+ * Second stage of a quiesce. The data is already synced, now we have to take
+ * care of the metadata. New transactions are already blocked, so we need to
+ * wait for any remaining transactions to drain out before proceding.
+ */
+void
+xfs_quiesce_attr(
+	struct xfs_mount	*mp)
+{
+	int	error = 0;
+
+	/* wait for all modifications to complete */
+	while (atomic_read(&mp->m_active_trans) > 0)
+		delay(100);
+
+	/* flush inodes and push all remaining buffers out to disk */
+	xfs_quiesce_fs(mp);
+
+	ASSERT_ALWAYS(atomic_read(&mp->m_active_trans) == 0);
+
+	/* Push the superblock and write an unmount record */
+	error = xfs_log_sbcount(mp, 1);
+	if (error)
+		xfs_fs_cmn_err(CE_WARN, mp,
+				"xfs_attr_quiesce: failed to log sb changes. "
+				"Frozen image may not be consistent.");
+	xfs_log_unmount_write(mp);
+	xfs_unmountfs_writesb(mp);
+}
+
+/*
+ * Enqueue a work item to be picked up by the vfs xfssyncd thread.
+ * Doing this has two advantages:
+ * - It saves on stack space, which is tight in certain situations
+ * - It can be used (with care) as a mechanism to avoid deadlocks.
+ * Flushing while allocating in a full filesystem requires both.
+ */
+STATIC void
+xfs_syncd_queue_work(
+	struct xfs_mount *mp,
+	void		*data,
+	void		(*syncer)(struct xfs_mount *, void *))
+{
+	struct bhv_vfs_sync_work *work;
+
+	work = kmem_alloc(sizeof(struct bhv_vfs_sync_work), KM_SLEEP);
+	INIT_LIST_HEAD(&work->w_list);
+	work->w_syncer = syncer;
+	work->w_data = data;
+	work->w_mount = mp;
+	spin_lock(&mp->m_sync_lock);
+	list_add_tail(&work->w_list, &mp->m_sync_list);
+	spin_unlock(&mp->m_sync_lock);
+	wake_up_process(mp->m_sync_task);
+}
+
+/*
+ * Flush delayed allocate data, attempting to free up reserved space
+ * from existing allocations.  At this point a new allocation attempt
+ * has failed with ENOSPC and we are in the process of scratching our
+ * heads, looking about for more room...
+ */
+STATIC void
+xfs_flush_inode_work(
+	struct xfs_mount *mp,
+	void		*arg)
+{
+	struct inode	*inode = arg;
+	filemap_flush(inode->i_mapping);
+	iput(inode);
+}
+
+void
+xfs_flush_inode(
+	xfs_inode_t	*ip)
+{
+	struct inode	*inode = VFS_I(ip);
+
+	igrab(inode);
+	xfs_syncd_queue_work(ip->i_mount, inode, xfs_flush_inode_work);
+	delay(msecs_to_jiffies(500));
+}
+
+/*
+ * This is the "bigger hammer" version of xfs_flush_inode_work...
+ * (IOW, "If at first you don't succeed, use a Bigger Hammer").
+ */
+STATIC void
+xfs_flush_device_work(
+	struct xfs_mount *mp,
+	void		*arg)
+{
+	struct inode	*inode = arg;
+	sync_blockdev(mp->m_super->s_bdev);
+	iput(inode);
+}
+
+void
+xfs_flush_device(
+	xfs_inode_t	*ip)
+{
+	struct inode	*inode = VFS_I(ip);
+
+	igrab(inode);
+	xfs_syncd_queue_work(ip->i_mount, inode, xfs_flush_device_work);
+	delay(msecs_to_jiffies(500));
+	xfs_log_force(ip->i_mount, (xfs_lsn_t)0, XFS_LOG_FORCE|XFS_LOG_SYNC);
+}
+
+/*
+ * Every sync period we need to unpin all items, reclaim inodes, sync
+ * quota and write out the superblock. We might need to cover the log
+ * to indicate it is idle.
+ */
+STATIC void
+xfs_sync_worker(
+	struct xfs_mount *mp,
+	void		*unused)
+{
+	int		error;
+
+	if (!(mp->m_flags & XFS_MOUNT_RDONLY)) {
+		xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE);
+		xfs_reclaim_inodes(mp, 0, XFS_IFLUSH_DELWRI_ELSE_ASYNC);
+		/* dgc: errors ignored here */
+		error = XFS_QM_DQSYNC(mp, SYNC_BDFLUSH);
+		error = xfs_sync_fsdata(mp, SYNC_BDFLUSH);
+		if (xfs_log_need_covered(mp))
+			error = xfs_commit_dummy_trans(mp, XFS_LOG_FORCE);
+	}
+	mp->m_sync_seq++;
+	wake_up(&mp->m_wait_single_sync_task);
+}
+
+STATIC int
+xfssyncd(
+	void			*arg)
+{
+	struct xfs_mount	*mp = arg;
+	long			timeleft;
+	bhv_vfs_sync_work_t	*work, *n;
+	LIST_HEAD		(tmp);
+
+	set_freezable();
+	timeleft = xfs_syncd_centisecs * msecs_to_jiffies(10);
+	for (;;) {
+		timeleft = schedule_timeout_interruptible(timeleft);
+		/* swsusp */
+		try_to_freeze();
+		if (kthread_should_stop() && list_empty(&mp->m_sync_list))
+			break;
+
+		spin_lock(&mp->m_sync_lock);
+		/*
+		 * We can get woken by laptop mode, to do a sync -
+		 * that's the (only!) case where the list would be
+		 * empty with time remaining.
+		 */
+		if (!timeleft || list_empty(&mp->m_sync_list)) {
+			if (!timeleft)
+				timeleft = xfs_syncd_centisecs *
+							msecs_to_jiffies(10);
+			INIT_LIST_HEAD(&mp->m_sync_work.w_list);
+			list_add_tail(&mp->m_sync_work.w_list,
+					&mp->m_sync_list);
+		}
+		list_for_each_entry_safe(work, n, &mp->m_sync_list, w_list)
+			list_move(&work->w_list, &tmp);
+		spin_unlock(&mp->m_sync_lock);
+
+		list_for_each_entry_safe(work, n, &tmp, w_list) {
+			(*work->w_syncer)(mp, work->w_data);
+			list_del(&work->w_list);
+			if (work == &mp->m_sync_work)
+				continue;
+			kmem_free(work);
+		}
+	}
+
+	return 0;
+}
+
+int
+xfs_syncd_init(
+	struct xfs_mount	*mp)
+{
+	mp->m_sync_work.w_syncer = xfs_sync_worker;
+	mp->m_sync_work.w_mount = mp;
+	mp->m_sync_task = kthread_run(xfssyncd, mp, "xfssyncd");
+	if (IS_ERR(mp->m_sync_task))
+		return -PTR_ERR(mp->m_sync_task);
+	return 0;
+}
+
+void
+xfs_syncd_stop(
+	struct xfs_mount	*mp)
+{
+	kthread_stop(mp->m_sync_task);
+}
+
+int
+xfs_reclaim_inode(
+	xfs_inode_t	*ip,
+	int		locked,
+	int		sync_mode)
+{
+	xfs_perag_t	*pag = xfs_get_perag(ip->i_mount, ip->i_ino);
+
+	/* The hash lock here protects a thread in xfs_iget_core from
+	 * racing with us on linking the inode back with a vnode.
+	 * Once we have the XFS_IRECLAIM flag set it will not touch
+	 * us.
+	 */
+	write_lock(&pag->pag_ici_lock);
+	spin_lock(&ip->i_flags_lock);
+	if (__xfs_iflags_test(ip, XFS_IRECLAIM) ||
+	    !__xfs_iflags_test(ip, XFS_IRECLAIMABLE)) {
+		spin_unlock(&ip->i_flags_lock);
+		write_unlock(&pag->pag_ici_lock);
+		if (locked) {
+			xfs_ifunlock(ip);
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+		}
+		return 1;
+	}
+	__xfs_iflags_set(ip, XFS_IRECLAIM);
+	spin_unlock(&ip->i_flags_lock);
+	write_unlock(&pag->pag_ici_lock);
+	xfs_put_perag(ip->i_mount, pag);
+
+	/*
+	 * If the inode is still dirty, then flush it out.  If the inode
+	 * is not in the AIL, then it will be OK to flush it delwri as
+	 * long as xfs_iflush() does not keep any references to the inode.
+	 * We leave that decision up to xfs_iflush() since it has the
+	 * knowledge of whether it's OK to simply do a delwri flush of
+	 * the inode or whether we need to wait until the inode is
+	 * pulled from the AIL.
+	 * We get the flush lock regardless, though, just to make sure
+	 * we don't free it while it is being flushed.
+	 */
+	if (!locked) {
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
+		xfs_iflock(ip);
+	}
+
+	/*
+	 * In the case of a forced shutdown we rely on xfs_iflush() to
+	 * wait for the inode to be unpinned before returning an error.
+	 */
+	if (!is_bad_inode(VFS_I(ip)) && xfs_iflush(ip, sync_mode) == 0) {
+		/* synchronize with xfs_iflush_done */
+		xfs_iflock(ip);
+		xfs_ifunlock(ip);
+	}
+
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	xfs_ireclaim(ip);
+	return 0;
+}
+
+/*
+ * We set the inode flag atomically with the radix tree tag.
+ * Once we get tag lookups on the radix tree, this inode flag
+ * can go away.
+ */
+void
+xfs_inode_set_reclaim_tag(
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_perag_t	*pag = xfs_get_perag(mp, ip->i_ino);
+
+	read_lock(&pag->pag_ici_lock);
+	spin_lock(&ip->i_flags_lock);
+	radix_tree_tag_set(&pag->pag_ici_root,
+			XFS_INO_TO_AGINO(mp, ip->i_ino), XFS_ICI_RECLAIM_TAG);
+	__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
+	spin_unlock(&ip->i_flags_lock);
+	read_unlock(&pag->pag_ici_lock);
+	xfs_put_perag(mp, pag);
+}
+
+void
+__xfs_inode_clear_reclaim_tag(
+	xfs_mount_t	*mp,
+	xfs_perag_t	*pag,
+	xfs_inode_t	*ip)
+{
+	radix_tree_tag_clear(&pag->pag_ici_root,
+			XFS_INO_TO_AGINO(mp, ip->i_ino), XFS_ICI_RECLAIM_TAG);
+}
+
+void
+xfs_inode_clear_reclaim_tag(
+	xfs_inode_t	*ip)
+{
+	xfs_mount_t	*mp = ip->i_mount;
+	xfs_perag_t	*pag = xfs_get_perag(mp, ip->i_ino);
+
+	read_lock(&pag->pag_ici_lock);
+	spin_lock(&ip->i_flags_lock);
+	__xfs_inode_clear_reclaim_tag(mp, pag, ip);
+	spin_unlock(&ip->i_flags_lock);
+	read_unlock(&pag->pag_ici_lock);
+	xfs_put_perag(mp, pag);
+}
+
+
+STATIC void
+xfs_reclaim_inodes_ag(
+	xfs_mount_t	*mp,
+	int		ag,
+	int		noblock,
+	int		mode)
+{
+	xfs_inode_t	*ip = NULL;
+	xfs_perag_t	*pag = &mp->m_perag[ag];
+	int		nr_found;
+	uint32_t	first_index;
+	int		skipped;
+
+restart:
+	first_index = 0;
+	skipped = 0;
+	do {
+		/*
+		 * use a gang lookup to find the next inode in the tree
+		 * as the tree is sparse and a gang lookup walks to find
+		 * the number of objects requested.
+		 */
+		read_lock(&pag->pag_ici_lock);
+		nr_found = radix_tree_gang_lookup_tag(&pag->pag_ici_root,
+					(void**)&ip, first_index, 1,
+					XFS_ICI_RECLAIM_TAG);
+
+		if (!nr_found) {
+			read_unlock(&pag->pag_ici_lock);
+			break;
+		}
+
+		/*
+		 * Update the index for the next lookup. Catch overflows
+		 * into the next AG range which can occur if we have inodes
+		 * in the last block of the AG and we are currently
+		 * pointing to the last inode.
+		 */
+		first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
+		if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino)) {
+			read_unlock(&pag->pag_ici_lock);
+			break;
+		}
+
+		/* ignore if already under reclaim */
+		if (xfs_iflags_test(ip, XFS_IRECLAIM)) {
+			read_unlock(&pag->pag_ici_lock);
+			continue;
+		}
+
+		if (noblock) {
+			if (!xfs_ilock_nowait(ip, XFS_ILOCK_EXCL)) {
+				read_unlock(&pag->pag_ici_lock);
+				continue;
+			}
+			if (xfs_ipincount(ip) ||
+			    !xfs_iflock_nowait(ip)) {
+				xfs_iunlock(ip, XFS_ILOCK_EXCL);
+				read_unlock(&pag->pag_ici_lock);
+				continue;
+			}
+		}
+		read_unlock(&pag->pag_ici_lock);
+
+		/*
+		 * hmmm - this is an inode already in reclaim. Do
+		 * we even bother catching it here?
+		 */
+		if (xfs_reclaim_inode(ip, noblock, mode))
+			skipped++;
+	} while (nr_found);
+
+	if (skipped) {
+		delay(1);
+		goto restart;
+	}
+	return;
+
+}
+
+int
+xfs_reclaim_inodes(
+	xfs_mount_t	*mp,
+	int		 noblock,
+	int		mode)
+{
+	int		i;
+
+	for (i = 0; i < mp->m_sb.sb_agcount; i++) {
+		if (!mp->m_perag[i].pag_ici_init)
+			continue;
+		xfs_reclaim_inodes_ag(mp, i, noblock, mode);
+	}
+	return 0;
+}
+
+
diff --git a/fs/xfs/linux-2.6/xfs_sync.h b/fs/xfs/linux-2.6/xfs_sync.h
new file mode 100644
index 00000000000..5f6de1efe1f
--- /dev/null
+++ b/fs/xfs/linux-2.6/xfs_sync.h
@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) 2000-2006 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef XFS_SYNC_H
+#define XFS_SYNC_H 1
+
+struct xfs_mount;
+
+typedef struct bhv_vfs_sync_work {
+	struct list_head	w_list;
+	struct xfs_mount	*w_mount;
+	void			*w_data;	/* syncer routine argument */
+	void			(*w_syncer)(struct xfs_mount *, void *);
+} bhv_vfs_sync_work_t;
+
+#define SYNC_ATTR		0x0001	/* sync attributes */
+#define SYNC_DELWRI		0x0002	/* look at delayed writes */
+#define SYNC_WAIT		0x0004	/* wait for i/o to complete */
+#define SYNC_BDFLUSH		0x0008	/* BDFLUSH is calling -- don't block */
+#define SYNC_IOWAIT		0x0010  /* wait for all I/O to complete */
+
+int xfs_syncd_init(struct xfs_mount *mp);
+void xfs_syncd_stop(struct xfs_mount *mp);
+
+int xfs_sync_inodes(struct xfs_mount *mp, int flags);
+int xfs_sync_fsdata(struct xfs_mount *mp, int flags);
+
+int xfs_quiesce_data(struct xfs_mount *mp);
+void xfs_quiesce_attr(struct xfs_mount *mp);
+
+void xfs_flush_inode(struct xfs_inode *ip);
+void xfs_flush_device(struct xfs_inode *ip);
+
+int xfs_reclaim_inode(struct xfs_inode *ip, int locked, int sync_mode);
+int xfs_reclaim_inodes(struct xfs_mount *mp, int noblock, int mode);
+
+void xfs_inode_set_reclaim_tag(struct xfs_inode *ip);
+void xfs_inode_clear_reclaim_tag(struct xfs_inode *ip);
+void __xfs_inode_clear_reclaim_tag(struct xfs_mount *mp, struct xfs_perag *pag,
+				struct xfs_inode *ip);
+#endif
diff --git a/fs/xfs/linux-2.6/xfs_sysctl.c b/fs/xfs/linux-2.6/xfs_sysctl.c
index 7dacb5bbde3..916c0ffb608 100644
--- a/fs/xfs/linux-2.6/xfs_sysctl.c
+++ b/fs/xfs/linux-2.6/xfs_sysctl.c
@@ -56,17 +56,6 @@ xfs_stats_clear_proc_handler(
 
 static ctl_table xfs_table[] = {
 	{
-		.ctl_name	= XFS_RESTRICT_CHOWN,
-		.procname	= "restrict_chown",
-		.data		= &xfs_params.restrict_chown.val,
-		.maxlen		= sizeof(int),
-		.mode		= 0644,
-		.proc_handler	= &proc_dointvec_minmax,
-		.strategy	= &sysctl_intvec,
-		.extra1		= &xfs_params.restrict_chown.min,
-		.extra2		= &xfs_params.restrict_chown.max
-	},
-	{
 		.ctl_name	= XFS_SGID_INHERIT,
 		.procname	= "irix_sgid_inherit",
 		.data		= &xfs_params.sgid_inherit.val,
diff --git a/fs/xfs/linux-2.6/xfs_sysctl.h b/fs/xfs/linux-2.6/xfs_sysctl.h
index 4aadb8056c3..b9937d450f8 100644
--- a/fs/xfs/linux-2.6/xfs_sysctl.h
+++ b/fs/xfs/linux-2.6/xfs_sysctl.h
@@ -31,7 +31,6 @@ typedef struct xfs_sysctl_val {
 } xfs_sysctl_val_t;
 
 typedef struct xfs_param {
-	xfs_sysctl_val_t restrict_chown;/* Root/non-root can give away files.*/
 	xfs_sysctl_val_t sgid_inherit;	/* Inherit S_ISGID if process' GID is
 					 * not a member of parent dir GID. */
 	xfs_sysctl_val_t symlink_mode;	/* Link creat mode affected by umask */
@@ -68,7 +67,7 @@ typedef struct xfs_param {
 enum {
 	/* XFS_REFCACHE_SIZE = 1 */
 	/* XFS_REFCACHE_PURGE = 2 */
-	XFS_RESTRICT_CHOWN = 3,
+	/* XFS_RESTRICT_CHOWN = 3 */
 	XFS_SGID_INHERIT = 4,
 	XFS_SYMLINK_MODE = 5,
 	XFS_PANIC_MASK = 6,
diff --git a/fs/xfs/linux-2.6/xfs_vfs.h b/fs/xfs/linux-2.6/xfs_vfs.h
deleted file mode 100644
index 7e60c7776b1..00000000000
--- a/fs/xfs/linux-2.6/xfs_vfs.h
+++ /dev/null
@@ -1,77 +0,0 @@
-/*
- * Copyright (c) 2000-2006 Silicon Graphics, Inc.
- * All Rights Reserved.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it would be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write the Free Software Foundation,
- * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#ifndef __XFS_VFS_H__
-#define __XFS_VFS_H__
-
-#include <linux/vfs.h>
-#include "xfs_fs.h"
-
-struct inode;
-
-struct fid;
-struct cred;
-struct seq_file;
-struct super_block;
-struct xfs_inode;
-struct xfs_mount;
-struct xfs_mount_args;
-
-typedef struct kstatfs	bhv_statvfs_t;
-
-typedef struct bhv_vfs_sync_work {
-	struct list_head	w_list;
-	struct xfs_mount	*w_mount;
-	void			*w_data;	/* syncer routine argument */
-	void			(*w_syncer)(struct xfs_mount *, void *);
-} bhv_vfs_sync_work_t;
-
-#define SYNC_ATTR		0x0001	/* sync attributes */
-#define SYNC_CLOSE		0x0002	/* close file system down */
-#define SYNC_DELWRI		0x0004	/* look at delayed writes */
-#define SYNC_WAIT		0x0008	/* wait for i/o to complete */
-#define SYNC_BDFLUSH		0x0010	/* BDFLUSH is calling -- don't block */
-#define SYNC_FSDATA		0x0020	/* flush fs data (e.g. superblocks) */
-#define SYNC_REFCACHE		0x0040  /* prune some of the nfs ref cache */
-#define SYNC_REMOUNT		0x0080  /* remount readonly, no dummy LRs */
-#define SYNC_IOWAIT		0x0100  /* wait for all I/O to complete */
-
-/*
- * When remounting a filesystem read-only or freezing the filesystem,
- * we have two phases to execute. This first phase is syncing the data
- * before we quiesce the fielsystem, and the second is flushing all the
- * inodes out after we've waited for all the transactions created by
- * the first phase to complete. The second phase uses SYNC_INODE_QUIESCE
- * to ensure that the inodes are written to their location on disk
- * rather than just existing in transactions in the log. This means
- * after a quiesce there is no log replay required to write the inodes
- * to disk (this is the main difference between a sync and a quiesce).
- */
-#define SYNC_DATA_QUIESCE	(SYNC_DELWRI|SYNC_FSDATA|SYNC_WAIT|SYNC_IOWAIT)
-#define SYNC_INODE_QUIESCE	(SYNC_REMOUNT|SYNC_ATTR|SYNC_WAIT)
-
-#define SHUTDOWN_META_IO_ERROR	0x0001	/* write attempt to metadata failed */
-#define SHUTDOWN_LOG_IO_ERROR	0x0002	/* write attempt to the log failed */
-#define SHUTDOWN_FORCE_UMOUNT	0x0004	/* shutdown from a forced unmount */
-#define SHUTDOWN_CORRUPT_INCORE	0x0008	/* corrupt in-memory data structures */
-#define SHUTDOWN_REMOTE_REQ	0x0010	/* shutdown came from remote cell */
-#define SHUTDOWN_DEVICE_REQ	0x0020	/* failed all paths to the device */
-
-#define xfs_test_for_freeze(mp)		((mp)->m_super->s_frozen)
-#define xfs_wait_for_freeze(mp,l)	vfs_check_frozen((mp)->m_super, (l))
-
-#endif	/* __XFS_VFS_H__ */
diff --git a/fs/xfs/linux-2.6/xfs_vnode.c b/fs/xfs/linux-2.6/xfs_vnode.c
deleted file mode 100644
index b52528bbbff..00000000000
--- a/fs/xfs/linux-2.6/xfs_vnode.c
+++ /dev/null
@@ -1,145 +0,0 @@
-/*
- * Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
- * All Rights Reserved.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it would be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write the Free Software Foundation,
- * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#include "xfs.h"
-#include "xfs_vnodeops.h"
-#include "xfs_bmap_btree.h"
-#include "xfs_inode.h"
-
-/*
- * And this gunk is needed for xfs_mount.h"
- */
-#include "xfs_log.h"
-#include "xfs_trans.h"
-#include "xfs_sb.h"
-#include "xfs_dmapi.h"
-#include "xfs_inum.h"
-#include "xfs_ag.h"
-#include "xfs_mount.h"
-
-
-/*
- * Dedicated vnode inactive/reclaim sync wait queues.
- * Prime number of hash buckets since address is used as the key.
- */
-#define NVSYNC                  37
-#define vptosync(v)             (&vsync[((unsigned long)v) % NVSYNC])
-static wait_queue_head_t vsync[NVSYNC];
-
-void __init
-vn_init(void)
-{
-	int i;
-
-	for (i = 0; i < NVSYNC; i++)
-		init_waitqueue_head(&vsync[i]);
-}
-
-void
-vn_iowait(
-	xfs_inode_t	*ip)
-{
-	wait_queue_head_t *wq = vptosync(ip);
-
-	wait_event(*wq, (atomic_read(&ip->i_iocount) == 0));
-}
-
-void
-vn_iowake(
-	xfs_inode_t	*ip)
-{
-	if (atomic_dec_and_test(&ip->i_iocount))
-		wake_up(vptosync(ip));
-}
-
-/*
- * Volume managers supporting multiple paths can send back ENODEV when the
- * final path disappears.  In this case continuing to fill the page cache
- * with dirty data which cannot be written out is evil, so prevent that.
- */
-void
-vn_ioerror(
-	xfs_inode_t	*ip,
-	int		error,
-	char		*f,
-	int		l)
-{
-	if (unlikely(error == -ENODEV))
-		xfs_do_force_shutdown(ip->i_mount, SHUTDOWN_DEVICE_REQ, f, l);
-}
-
-#ifdef	XFS_INODE_TRACE
-
-/*
- * Reference count of Linux inode if present, -1 if the xfs_inode
- * has no associated Linux inode.
- */
-static inline int xfs_icount(struct xfs_inode *ip)
-{
-	struct inode *vp = VFS_I(ip);
-
-	if (vp)
-		return vn_count(vp);
-	return -1;
-}
-
-#define KTRACE_ENTER(ip, vk, s, line, ra)			\
-	ktrace_enter(	(ip)->i_trace,				\
-/*  0 */		(void *)(__psint_t)(vk),		\
-/*  1 */		(void *)(s),				\
-/*  2 */		(void *)(__psint_t) line,		\
-/*  3 */		(void *)(__psint_t)xfs_icount(ip),	\
-/*  4 */		(void *)(ra),				\
-/*  5 */		NULL,					\
-/*  6 */		(void *)(__psint_t)current_cpu(),	\
-/*  7 */		(void *)(__psint_t)current_pid(),	\
-/*  8 */		(void *)__return_address,		\
-/*  9 */		NULL, NULL, NULL, NULL, NULL, NULL, NULL)
-
-/*
- * Vnode tracing code.
- */
-void
-_xfs_itrace_entry(xfs_inode_t *ip, const char *func, inst_t *ra)
-{
-	KTRACE_ENTER(ip, INODE_KTRACE_ENTRY, func, 0, ra);
-}
-
-void
-_xfs_itrace_exit(xfs_inode_t *ip, const char *func, inst_t *ra)
-{
-	KTRACE_ENTER(ip, INODE_KTRACE_EXIT, func, 0, ra);
-}
-
-void
-xfs_itrace_hold(xfs_inode_t *ip, char *file, int line, inst_t *ra)
-{
-	KTRACE_ENTER(ip, INODE_KTRACE_HOLD, file, line, ra);
-}
-
-void
-_xfs_itrace_ref(xfs_inode_t *ip, char *file, int line, inst_t *ra)
-{
-	KTRACE_ENTER(ip, INODE_KTRACE_REF, file, line, ra);
-}
-
-void
-xfs_itrace_rele(xfs_inode_t *ip, char *file, int line, inst_t *ra)
-{
-	KTRACE_ENTER(ip, INODE_KTRACE_RELE, file, line, ra);
-}
-#endif	/* XFS_INODE_TRACE */
diff --git a/fs/xfs/linux-2.6/xfs_vnode.h b/fs/xfs/linux-2.6/xfs_vnode.h
index 683ce16210f..f65983a230d 100644
--- a/fs/xfs/linux-2.6/xfs_vnode.h
+++ b/fs/xfs/linux-2.6/xfs_vnode.h
@@ -18,7 +18,10 @@
 #ifndef __XFS_VNODE_H__
 #define __XFS_VNODE_H__
 
+#include "xfs_fs.h"
+
 struct file;
+struct xfs_inode;
 struct xfs_iomap;
 struct attrlist_cursor_kern;
 
@@ -51,40 +54,6 @@ struct attrlist_cursor_kern;
 					   Prevent VM access to the pages until
 					   the operation completes. */
 
-
-extern void	vn_init(void);
-
-/*
- * Yeah, these don't take vnode anymore at all, all this should be
- * cleaned up at some point.
- */
-extern void	vn_iowait(struct xfs_inode *ip);
-extern void	vn_iowake(struct xfs_inode *ip);
-extern void	vn_ioerror(struct xfs_inode *ip, int error, char *f, int l);
-
-static inline int vn_count(struct inode *vp)
-{
-	return atomic_read(&vp->i_count);
-}
-
-#define IHOLD(ip) \
-do { \
-	ASSERT(atomic_read(&VFS_I(ip)->i_count) > 0) ; \
-	atomic_inc(&(VFS_I(ip)->i_count)); \
-	xfs_itrace_hold((ip), __FILE__, __LINE__, (inst_t *)__return_address); \
-} while (0)
-
-#define IRELE(ip) \
-do { \
-	xfs_itrace_rele((ip), __FILE__, __LINE__, (inst_t *)__return_address); \
-	iput(VFS_I(ip)); \
-} while (0)
-
-static inline struct inode *vn_grab(struct inode *vp)
-{
-	return igrab(vp);
-}
-
 /*
  * Dealing with bad inodes
  */
@@ -121,39 +90,4 @@ static inline void vn_atime_to_time_t(struct inode *vp, time_t *tt)
 					PAGECACHE_TAG_DIRTY)
 
 
-/*
- * Tracking vnode activity.
- */
-#if defined(XFS_INODE_TRACE)
-
-#define	INODE_TRACE_SIZE	16		/* number of trace entries */
-#define	INODE_KTRACE_ENTRY	1
-#define	INODE_KTRACE_EXIT	2
-#define	INODE_KTRACE_HOLD	3
-#define	INODE_KTRACE_REF	4
-#define	INODE_KTRACE_RELE	5
-
-extern void _xfs_itrace_entry(struct xfs_inode *, const char *, inst_t *);
-extern void _xfs_itrace_exit(struct xfs_inode *, const char *, inst_t *);
-extern void xfs_itrace_hold(struct xfs_inode *, char *, int, inst_t *);
-extern void _xfs_itrace_ref(struct xfs_inode *, char *, int, inst_t *);
-extern void xfs_itrace_rele(struct xfs_inode *, char *, int, inst_t *);
-#define xfs_itrace_entry(ip)	\
-	_xfs_itrace_entry(ip, __func__, (inst_t *)__return_address)
-#define xfs_itrace_exit(ip)	\
-	_xfs_itrace_exit(ip, __func__, (inst_t *)__return_address)
-#define xfs_itrace_exit_tag(ip, tag)	\
-	_xfs_itrace_exit(ip, tag, (inst_t *)__return_address)
-#define xfs_itrace_ref(ip)	\
-	_xfs_itrace_ref(ip, __FILE__, __LINE__, (inst_t *)__return_address)
-
-#else
-#define	xfs_itrace_entry(a)
-#define	xfs_itrace_exit(a)
-#define	xfs_itrace_exit_tag(a, b)
-#define	xfs_itrace_hold(a, b, c, d)
-#define	xfs_itrace_ref(a)
-#define	xfs_itrace_rele(a, b, c, d)
-#endif
-
 #endif	/* __XFS_VNODE_H__ */
diff --git a/fs/xfs/quota/xfs_dquot.c b/fs/xfs/quota/xfs_dquot.c
index f2705f2fd43..591ca6602bf 100644
--- a/fs/xfs/quota/xfs_dquot.c
+++ b/fs/xfs/quota/xfs_dquot.c
@@ -101,7 +101,7 @@ xfs_qm_dqinit(
 	if (brandnewdquot) {
 		dqp->dq_flnext = dqp->dq_flprev = dqp;
 		mutex_init(&dqp->q_qlock);
-		sv_init(&dqp->q_pinwait, SV_DEFAULT, "pdq");
+		init_waitqueue_head(&dqp->q_pinwait);
 
 		/*
 		 * Because we want to use a counting completion, complete
@@ -131,7 +131,7 @@ xfs_qm_dqinit(
 		 dqp->q_res_bcount = 0;
 		 dqp->q_res_icount = 0;
 		 dqp->q_res_rtbcount = 0;
-		 dqp->q_pincount = 0;
+		 atomic_set(&dqp->q_pincount, 0);
 		 dqp->q_hash = NULL;
 		 ASSERT(dqp->dq_flnext == dqp->dq_flprev);
 
@@ -1221,16 +1221,14 @@ xfs_qm_dqflush(
 	xfs_dqtrace_entry(dqp, "DQFLUSH");
 
 	/*
-	 * If not dirty, nada.
+	 * If not dirty, or it's pinned and we are not supposed to
+	 * block, nada.
 	 */
-	if (!XFS_DQ_IS_DIRTY(dqp)) {
+	if (!XFS_DQ_IS_DIRTY(dqp) ||
+	    (!(flags & XFS_QMOPT_SYNC) && atomic_read(&dqp->q_pincount) > 0)) {
 		xfs_dqfunlock(dqp);
-		return (0);
+		return 0;
 	}
-
-	/*
-	 * Cant flush a pinned dquot. Wait for it.
-	 */
 	xfs_qm_dqunpin_wait(dqp);
 
 	/*
@@ -1274,10 +1272,8 @@ xfs_qm_dqflush(
 	dqp->dq_flags &= ~(XFS_DQ_DIRTY);
 	mp = dqp->q_mount;
 
-	/* lsn is 64 bits */
-	spin_lock(&mp->m_ail_lock);
-	dqp->q_logitem.qli_flush_lsn = dqp->q_logitem.qli_item.li_lsn;
-	spin_unlock(&mp->m_ail_lock);
+	xfs_trans_ail_copy_lsn(mp->m_ail, &dqp->q_logitem.qli_flush_lsn,
+					&dqp->q_logitem.qli_item.li_lsn);
 
 	/*
 	 * Attach an iodone routine so that we can remove this dquot from the
@@ -1323,8 +1319,10 @@ xfs_qm_dqflush_done(
 	xfs_dq_logitem_t	*qip)
 {
 	xfs_dquot_t		*dqp;
+	struct xfs_ail		*ailp;
 
 	dqp = qip->qli_dquot;
+	ailp = qip->qli_item.li_ailp;
 
 	/*
 	 * We only want to pull the item from the AIL if its
@@ -1337,15 +1335,12 @@ xfs_qm_dqflush_done(
 	if ((qip->qli_item.li_flags & XFS_LI_IN_AIL) &&
 	    qip->qli_item.li_lsn == qip->qli_flush_lsn) {
 
-		spin_lock(&dqp->q_mount->m_ail_lock);
-		/*
-		 * xfs_trans_delete_ail() drops the AIL lock.
-		 */
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		spin_lock(&ailp->xa_lock);
 		if (qip->qli_item.li_lsn == qip->qli_flush_lsn)
-			xfs_trans_delete_ail(dqp->q_mount,
-					     (xfs_log_item_t*)qip);
+			xfs_trans_ail_delete(ailp, (xfs_log_item_t*)qip);
 		else
-			spin_unlock(&dqp->q_mount->m_ail_lock);
+			spin_unlock(&ailp->xa_lock);
 	}
 
 	/*
@@ -1375,7 +1370,7 @@ xfs_dqunlock(
 	mutex_unlock(&(dqp->q_qlock));
 	if (dqp->q_logitem.qli_dquot == dqp) {
 		/* Once was dqp->q_mount, but might just have been cleared */
-		xfs_trans_unlocked_item(dqp->q_logitem.qli_item.li_mountp,
+		xfs_trans_unlocked_item(dqp->q_logitem.qli_item.li_ailp,
 					(xfs_log_item_t*)&(dqp->q_logitem));
 	}
 }
@@ -1489,7 +1484,7 @@ xfs_qm_dqpurge(
 				"xfs_qm_dqpurge: dquot %p flush failed", dqp);
 		xfs_dqflock(dqp);
 	}
-	ASSERT(dqp->q_pincount == 0);
+	ASSERT(atomic_read(&dqp->q_pincount) == 0);
 	ASSERT(XFS_FORCED_SHUTDOWN(mp) ||
 	       !(dqp->q_logitem.qli_item.li_flags & XFS_LI_IN_AIL));
 
diff --git a/fs/xfs/quota/xfs_dquot.h b/fs/xfs/quota/xfs_dquot.h
index 8958d0faf8d..7e455337e2b 100644
--- a/fs/xfs/quota/xfs_dquot.h
+++ b/fs/xfs/quota/xfs_dquot.h
@@ -83,8 +83,8 @@ typedef struct xfs_dquot {
 	xfs_qcnt_t	 q_res_rtbcount;/* total realtime blks used+reserved */
 	mutex_t		 q_qlock;	/* quota lock */
 	struct completion q_flush;	/* flush completion queue */
-	uint		 q_pincount;	/* pin count for this dquot */
-	sv_t		 q_pinwait;	/* sync var for pinning */
+	atomic_t          q_pincount;	/* dquot pin count */
+	wait_queue_head_t q_pinwait;	/* dquot pinning wait queue */
 #ifdef XFS_DQUOT_TRACE
 	struct ktrace	*q_trace;	/* trace header structure */
 #endif
diff --git a/fs/xfs/quota/xfs_dquot_item.c b/fs/xfs/quota/xfs_dquot_item.c
index f028644caa5..1728f6a7c4f 100644
--- a/fs/xfs/quota/xfs_dquot_item.c
+++ b/fs/xfs/quota/xfs_dquot_item.c
@@ -88,25 +88,22 @@ xfs_qm_dquot_logitem_format(
 
 /*
  * Increment the pin count of the given dquot.
- * This value is protected by pinlock spinlock in the xQM structure.
  */
 STATIC void
 xfs_qm_dquot_logitem_pin(
 	xfs_dq_logitem_t *logitem)
 {
-	xfs_dquot_t *dqp;
+	xfs_dquot_t *dqp = logitem->qli_dquot;
 
-	dqp = logitem->qli_dquot;
 	ASSERT(XFS_DQ_IS_LOCKED(dqp));
-	spin_lock(&(XFS_DQ_TO_QINF(dqp)->qi_pinlock));
-	dqp->q_pincount++;
-	spin_unlock(&(XFS_DQ_TO_QINF(dqp)->qi_pinlock));
+	atomic_inc(&dqp->q_pincount);
 }
 
 /*
  * Decrement the pin count of the given dquot, and wake up
  * anyone in xfs_dqwait_unpin() if the count goes to 0.	 The
- * dquot must have been previously pinned with a call to xfs_dqpin().
+ * dquot must have been previously pinned with a call to
+ * xfs_qm_dquot_logitem_pin().
  */
 /* ARGSUSED */
 STATIC void
@@ -114,16 +111,11 @@ xfs_qm_dquot_logitem_unpin(
 	xfs_dq_logitem_t *logitem,
 	int		  stale)
 {
-	xfs_dquot_t *dqp;
+	xfs_dquot_t *dqp = logitem->qli_dquot;
 
-	dqp = logitem->qli_dquot;
-	ASSERT(dqp->q_pincount > 0);
-	spin_lock(&(XFS_DQ_TO_QINF(dqp)->qi_pinlock));
-	dqp->q_pincount--;
-	if (dqp->q_pincount == 0) {
-		sv_broadcast(&dqp->q_pinwait);
-	}
-	spin_unlock(&(XFS_DQ_TO_QINF(dqp)->qi_pinlock));
+	ASSERT(atomic_read(&dqp->q_pincount) > 0);
+	if (atomic_dec_and_test(&dqp->q_pincount))
+		wake_up(&dqp->q_pinwait);
 }
 
 /* ARGSUSED */
@@ -193,21 +185,14 @@ xfs_qm_dqunpin_wait(
 	xfs_dquot_t	*dqp)
 {
 	ASSERT(XFS_DQ_IS_LOCKED(dqp));
-	if (dqp->q_pincount == 0) {
+	if (atomic_read(&dqp->q_pincount) == 0)
 		return;
-	}
 
 	/*
 	 * Give the log a push so we don't wait here too long.
 	 */
 	xfs_log_force(dqp->q_mount, (xfs_lsn_t)0, XFS_LOG_FORCE);
-	spin_lock(&(XFS_DQ_TO_QINF(dqp)->qi_pinlock));
-	if (dqp->q_pincount == 0) {
-		spin_unlock(&(XFS_DQ_TO_QINF(dqp)->qi_pinlock));
-		return;
-	}
-	sv_wait(&(dqp->q_pinwait), PINOD,
-		&(XFS_DQ_TO_QINF(dqp)->qi_pinlock), s);
+	wait_event(dqp->q_pinwait, (atomic_read(&dqp->q_pincount) == 0));
 }
 
 /*
@@ -310,7 +295,7 @@ xfs_qm_dquot_logitem_trylock(
 	uint			retval;
 
 	dqp = qip->qli_dquot;
-	if (dqp->q_pincount > 0)
+	if (atomic_read(&dqp->q_pincount) > 0)
 		return (XFS_ITEM_PINNED);
 
 	if (! xfs_qm_dqlock_nowait(dqp))
@@ -568,14 +553,16 @@ xfs_qm_qoffend_logitem_committed(
 	xfs_lsn_t lsn)
 {
 	xfs_qoff_logitem_t	*qfs;
+	struct xfs_ail		*ailp;
 
 	qfs = qfe->qql_start_lip;
-	spin_lock(&qfs->qql_item.li_mountp->m_ail_lock);
+	ailp = qfs->qql_item.li_ailp;
+	spin_lock(&ailp->xa_lock);
 	/*
 	 * Delete the qoff-start logitem from the AIL.
-	 * xfs_trans_delete_ail() drops the AIL lock.
+	 * xfs_trans_ail_delete() drops the AIL lock.
 	 */
-	xfs_trans_delete_ail(qfs->qql_item.li_mountp, (xfs_log_item_t *)qfs);
+	xfs_trans_ail_delete(ailp, (xfs_log_item_t *)qfs);
 	kmem_free(qfs);
 	kmem_free(qfe);
 	return (xfs_lsn_t)-1;
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index df0ffef9775..6b13960cf31 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -20,7 +20,6 @@
 #include "xfs_bit.h"
 #include "xfs_log.h"
 #include "xfs_inum.h"
-#include "xfs_clnt.h"
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
@@ -396,13 +395,10 @@ xfs_qm_mount_quotas(
 /*
  * Called from the vfsops layer.
  */
-int
+void
 xfs_qm_unmount_quotas(
 	xfs_mount_t	*mp)
 {
-	xfs_inode_t	*uqp, *gqp;
-	int		error = 0;
-
 	/*
 	 * Release the dquots that root inode, et al might be holding,
 	 * before we flush quotas and blow away the quotainfo structure.
@@ -415,43 +411,18 @@ xfs_qm_unmount_quotas(
 		xfs_qm_dqdetach(mp->m_rsumip);
 
 	/*
-	 * Flush out the quota inodes.
+	 * Release the quota inodes.
 	 */
-	uqp = gqp = NULL;
 	if (mp->m_quotainfo) {
-		if ((uqp = mp->m_quotainfo->qi_uquotaip) != NULL) {
-			xfs_ilock(uqp, XFS_ILOCK_EXCL);
-			xfs_iflock(uqp);
-			error = xfs_iflush(uqp, XFS_IFLUSH_SYNC);
-			xfs_iunlock(uqp, XFS_ILOCK_EXCL);
-			if (unlikely(error == EFSCORRUPTED)) {
-				XFS_ERROR_REPORT("xfs_qm_unmount_quotas(1)",
-						 XFS_ERRLEVEL_LOW, mp);
-				goto out;
-			}
+		if (mp->m_quotainfo->qi_uquotaip) {
+			IRELE(mp->m_quotainfo->qi_uquotaip);
+			mp->m_quotainfo->qi_uquotaip = NULL;
 		}
-		if ((gqp = mp->m_quotainfo->qi_gquotaip) != NULL) {
-			xfs_ilock(gqp, XFS_ILOCK_EXCL);
-			xfs_iflock(gqp);
-			error = xfs_iflush(gqp, XFS_IFLUSH_SYNC);
-			xfs_iunlock(gqp, XFS_ILOCK_EXCL);
-			if (unlikely(error == EFSCORRUPTED)) {
-				XFS_ERROR_REPORT("xfs_qm_unmount_quotas(2)",
-						 XFS_ERRLEVEL_LOW, mp);
-				goto out;
-			}
+		if (mp->m_quotainfo->qi_gquotaip) {
+			IRELE(mp->m_quotainfo->qi_gquotaip);
+			mp->m_quotainfo->qi_gquotaip = NULL;
 		}
 	}
-	if (uqp) {
-		 IRELE(uqp);
-		 mp->m_quotainfo->qi_uquotaip = NULL;
-	}
-	if (gqp) {
-		IRELE(gqp);
-		mp->m_quotainfo->qi_gquotaip = NULL;
-	}
-out:
-	return XFS_ERROR(error);
 }
 
 /*
@@ -987,14 +958,10 @@ xfs_qm_dqdetach(
 }
 
 /*
- * This is called by VFS_SYNC and flags arg determines the caller,
- * and its motives, as done in xfs_sync.
- *
- * vfs_sync: SYNC_FSDATA|SYNC_ATTR|SYNC_BDFLUSH 0x31
- * syscall sync: SYNC_FSDATA|SYNC_ATTR|SYNC_DELWRI 0x25
- * umountroot : SYNC_WAIT | SYNC_CLOSE | SYNC_ATTR | SYNC_FSDATA
+ * This is called to sync quotas. We can be told to use non-blocking
+ * semantics by either the SYNC_BDFLUSH flag or the absence of the
+ * SYNC_WAIT flag.
  */
-
 int
 xfs_qm_sync(
 	xfs_mount_t	*mp,
@@ -1137,7 +1104,6 @@ xfs_qm_init_quotainfo(
 		return error;
 	}
 
-	spin_lock_init(&qinf->qi_pinlock);
 	xfs_qm_list_init(&qinf->qi_dqlist, "mpdqlist", 0);
 	qinf->qi_dqreclaims = 0;
 
@@ -1234,7 +1200,6 @@ xfs_qm_destroy_quotainfo(
 	 */
 	xfs_qm_rele_quotafs_ref(mp);
 
-	spinlock_destroy(&qi->qi_pinlock);
 	xfs_qm_list_destroy(&qi->qi_dqlist);
 
 	if (qi->qi_uquotaip) {
diff --git a/fs/xfs/quota/xfs_qm.h b/fs/xfs/quota/xfs_qm.h
index 44f25349e47..ddf09166387 100644
--- a/fs/xfs/quota/xfs_qm.h
+++ b/fs/xfs/quota/xfs_qm.h
@@ -106,7 +106,6 @@ typedef struct xfs_qm {
 typedef struct xfs_quotainfo {
 	xfs_inode_t	*qi_uquotaip;	 /* user quota inode */
 	xfs_inode_t	*qi_gquotaip;	 /* group quota inode */
-	spinlock_t	 qi_pinlock;	 /* dquot pinning lock */
 	xfs_dqlist_t	 qi_dqlist;	 /* all dquots in filesys */
 	int		 qi_dqreclaims;	 /* a change here indicates
 					    a removal in the dqlist */
@@ -168,7 +167,7 @@ extern void		xfs_qm_destroy_quotainfo(xfs_mount_t *);
 extern void		xfs_qm_mount_quotas(xfs_mount_t *);
 extern int		xfs_qm_quotacheck(xfs_mount_t *);
 extern void		xfs_qm_unmount_quotadestroy(xfs_mount_t *);
-extern int		xfs_qm_unmount_quotas(xfs_mount_t *);
+extern void		xfs_qm_unmount_quotas(xfs_mount_t *);
 extern int		xfs_qm_write_sb_changes(xfs_mount_t *, __int64_t);
 extern int		xfs_qm_sync(xfs_mount_t *, int);
 
diff --git a/fs/xfs/quota/xfs_qm_bhv.c b/fs/xfs/quota/xfs_qm_bhv.c
index eea2e60b456..bc6c5cca3e1 100644
--- a/fs/xfs/quota/xfs_qm_bhv.c
+++ b/fs/xfs/quota/xfs_qm_bhv.c
@@ -20,7 +20,6 @@
 #include "xfs_bit.h"
 #include "xfs_log.h"
 #include "xfs_inum.h"
-#include "xfs_clnt.h"
 #include "xfs_trans.h"
 #include "xfs_sb.h"
 #include "xfs_ag.h"
@@ -51,7 +50,7 @@
 
 STATIC void
 xfs_fill_statvfs_from_dquot(
-	bhv_statvfs_t		*statp,
+	struct kstatfs		*statp,
 	xfs_disk_dquot_t	*dp)
 {
 	__uint64_t		limit;
@@ -88,7 +87,7 @@ xfs_fill_statvfs_from_dquot(
 STATIC void
 xfs_qm_statvfs(
 	xfs_inode_t		*ip,
-	bhv_statvfs_t		*statp)
+	struct kstatfs		*statp)
 {
 	xfs_mount_t		*mp = ip->i_mount;
 	xfs_dquot_t		*dqp;
diff --git a/fs/xfs/quota/xfs_qm_syscalls.c b/fs/xfs/quota/xfs_qm_syscalls.c
index 1a3b803dfa5..68139b38aed 100644
--- a/fs/xfs/quota/xfs_qm_syscalls.c
+++ b/fs/xfs/quota/xfs_qm_syscalls.c
@@ -127,7 +127,7 @@ xfs_qm_quotactl(
 		break;
 
 	case Q_XQUOTASYNC:
-		return (xfs_sync_inodes(mp, SYNC_DELWRI, NULL));
+		return xfs_sync_inodes(mp, SYNC_DELWRI);
 
 	default:
 		break;
@@ -1022,101 +1022,104 @@ xfs_qm_export_flags(
 
 
 /*
- * Go thru all the inodes in the file system, releasing their dquots.
- * Note that the mount structure gets modified to indicate that quotas are off
- * AFTER this, in the case of quotaoff. This also gets called from
- * xfs_rootumount.
+ * Release all the dquots on the inodes in an AG.
  */
-void
-xfs_qm_dqrele_all_inodes(
-	struct xfs_mount *mp,
-	uint		 flags)
+STATIC void
+xfs_qm_dqrele_inodes_ag(
+	xfs_mount_t	*mp,
+	int		ag,
+	uint		flags)
 {
-	xfs_inode_t	*ip, *topino;
-	uint		ireclaims;
-	struct inode	*vp;
-	boolean_t	vnode_refd;
+	xfs_inode_t	*ip = NULL;
+	xfs_perag_t	*pag = &mp->m_perag[ag];
+	int		first_index = 0;
+	int		nr_found;
 
-	ASSERT(mp->m_quotainfo);
-
-	XFS_MOUNT_ILOCK(mp);
-again:
-	ip = mp->m_inodes;
-	if (ip == NULL) {
-		XFS_MOUNT_IUNLOCK(mp);
-		return;
-	}
 	do {
-		/* Skip markers inserted by xfs_sync */
-		if (ip->i_mount == NULL) {
-			ip = ip->i_mnext;
-			continue;
+		/*
+		 * use a gang lookup to find the next inode in the tree
+		 * as the tree is sparse and a gang lookup walks to find
+		 * the number of objects requested.
+		 */
+		read_lock(&pag->pag_ici_lock);
+		nr_found = radix_tree_gang_lookup(&pag->pag_ici_root,
+				(void**)&ip, first_index, 1);
+
+		if (!nr_found) {
+			read_unlock(&pag->pag_ici_lock);
+			break;
 		}
-		/* Root inode, rbmip and rsumip have associated blocks */
+
+		/*
+		 * Update the index for the next lookup. Catch overflows
+		 * into the next AG range which can occur if we have inodes
+		 * in the last block of the AG and we are currently
+		 * pointing to the last inode.
+		 */
+		first_index = XFS_INO_TO_AGINO(mp, ip->i_ino + 1);
+		if (first_index < XFS_INO_TO_AGINO(mp, ip->i_ino)) {
+			read_unlock(&pag->pag_ici_lock);
+			break;
+		}
+
+		/* skip quota inodes */
 		if (ip == XFS_QI_UQIP(mp) || ip == XFS_QI_GQIP(mp)) {
 			ASSERT(ip->i_udquot == NULL);
 			ASSERT(ip->i_gdquot == NULL);
-			ip = ip->i_mnext;
+			read_unlock(&pag->pag_ici_lock);
 			continue;
 		}
-		vp = VFS_I(ip);
-		if (!vp) {
-			ASSERT(ip->i_udquot == NULL);
-			ASSERT(ip->i_gdquot == NULL);
-			ip = ip->i_mnext;
+
+		/*
+		 * If we can't get a reference on the inode, it must be
+		 * in reclaim. Leave it for the reclaim code to flush.
+		 */
+		if (!igrab(VFS_I(ip))) {
+			read_unlock(&pag->pag_ici_lock);
 			continue;
 		}
-		vnode_refd = B_FALSE;
-		if (xfs_ilock_nowait(ip, XFS_ILOCK_EXCL) == 0) {
-			ireclaims = mp->m_ireclaims;
-			topino = mp->m_inodes;
-			vp = vn_grab(vp);
-			if (!vp)
-				goto again;
-
-			XFS_MOUNT_IUNLOCK(mp);
-			/* XXX restart limit ? */
-			xfs_ilock(ip, XFS_ILOCK_EXCL);
-			vnode_refd = B_TRUE;
-		} else {
-			ireclaims = mp->m_ireclaims;
-			topino = mp->m_inodes;
-			XFS_MOUNT_IUNLOCK(mp);
+		read_unlock(&pag->pag_ici_lock);
+
+		/* avoid new inodes though we shouldn't find any here */
+		if (xfs_iflags_test(ip, XFS_INEW)) {
+			IRELE(ip);
+			continue;
 		}
 
-		/*
-		 * We don't keep the mountlock across the dqrele() call,
-		 * since it can take a while..
-		 */
+		xfs_ilock(ip, XFS_ILOCK_EXCL);
 		if ((flags & XFS_UQUOTA_ACCT) && ip->i_udquot) {
 			xfs_qm_dqrele(ip->i_udquot);
 			ip->i_udquot = NULL;
 		}
-		if (flags & (XFS_PQUOTA_ACCT|XFS_GQUOTA_ACCT) && ip->i_gdquot) {
+		if (flags & (XFS_PQUOTA_ACCT|XFS_GQUOTA_ACCT) &&
+		    ip->i_gdquot) {
 			xfs_qm_dqrele(ip->i_gdquot);
 			ip->i_gdquot = NULL;
 		}
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-		/*
-		 * Wait until we've dropped the ilock and mountlock to
-		 * do the vn_rele. Or be condemned to an eternity in the
-		 * inactive code in hell.
-		 */
-		if (vnode_refd)
-			IRELE(ip);
-		XFS_MOUNT_ILOCK(mp);
-		/*
-		 * If an inode was inserted or removed, we gotta
-		 * start over again.
-		 */
-		if (topino != mp->m_inodes || mp->m_ireclaims != ireclaims) {
-			/* XXX use a sentinel */
-			goto again;
-		}
-		ip = ip->i_mnext;
-	} while (ip != mp->m_inodes);
+		xfs_iput(ip, XFS_ILOCK_EXCL);
+
+	} while (nr_found);
+}
+
+/*
+ * Go thru all the inodes in the file system, releasing their dquots.
+ * Note that the mount structure gets modified to indicate that quotas are off
+ * AFTER this, in the case of quotaoff. This also gets called from
+ * xfs_rootumount.
+ */
+void
+xfs_qm_dqrele_all_inodes(
+	struct xfs_mount *mp,
+	uint		 flags)
+{
+	int		i;
 
-	XFS_MOUNT_IUNLOCK(mp);
+	ASSERT(mp->m_quotainfo);
+	for (i = 0; i < mp->m_sb.sb_agcount; i++) {
+		if (!mp->m_perag[i].pag_ici_init)
+			continue;
+		xfs_qm_dqrele_inodes_ag(mp, i, flags);
+	}
 }
 
 /*------------------------------------------------------------------------*/
diff --git a/fs/xfs/support/debug.c b/fs/xfs/support/debug.c
index c27abef7b84..ae548296542 100644
--- a/fs/xfs/support/debug.c
+++ b/fs/xfs/support/debug.c
@@ -18,6 +18,13 @@
 #include <xfs.h>
 #include "debug.h"
 
+/* xfs_mount.h drags a lot of crap in, sorry.. */
+#include "xfs_sb.h"
+#include "xfs_inum.h"
+#include "xfs_ag.h"
+#include "xfs_dmapi.h"
+#include "xfs_mount.h"
+
 static char		message[1024];	/* keep it off the stack */
 static DEFINE_SPINLOCK(xfs_err_lock);
 
@@ -55,22 +62,42 @@ cmn_err(register int level, char *fmt, ...)
 }
 
 void
-icmn_err(register int level, char *fmt, va_list ap)
+xfs_fs_vcmn_err(
+	int			level,
+	struct xfs_mount	*mp,
+	char			*fmt,
+	va_list			ap)
 {
-	ulong	flags;
-	int	len;
+	unsigned long		flags;
+	int			len = 0;
 
 	level &= XFS_ERR_MASK;
-	if(level > XFS_MAX_ERR_LEVEL)
+	if (level > XFS_MAX_ERR_LEVEL)
 		level = XFS_MAX_ERR_LEVEL;
+
 	spin_lock_irqsave(&xfs_err_lock,flags);
-	len = vsnprintf(message, sizeof(message), fmt, ap);
+
+	if (mp) {
+		len = sprintf(message, "Filesystem \"%s\": ", mp->m_fsname);
+
+		/*
+		 * Skip the printk if we can't print anything useful
+		 * due to an over-long device name.
+		 */
+		if (len >= sizeof(message))
+			goto out;
+	}
+
+	len = vsnprintf(message + len, sizeof(message) - len, fmt, ap);
 	if (len >= sizeof(message))
 		len = sizeof(message) - 1;
 	if (message[len-1] == '\n')
 		message[len-1] = 0;
+
 	printk("%s%s\n", err_level[level], message);
+ out:
 	spin_unlock_irqrestore(&xfs_err_lock,flags);
+
 	BUG_ON(level == CE_PANIC);
 }
 
@@ -84,5 +111,5 @@ assfail(char *expr, char *file, int line)
 void
 xfs_hex_dump(void *p, int length)
 {
-	print_hex_dump(KERN_ALERT, "", DUMP_PREFIX_OFFSET, 16, 1, p, length, 1);
+	print_hex_dump(KERN_ALERT, "", DUMP_PREFIX_ADDRESS, 16, 1, p, length, 1);
 }
diff --git a/fs/xfs/support/debug.h b/fs/xfs/support/debug.h
index 75845f95081..6f4fd37c67a 100644
--- a/fs/xfs/support/debug.h
+++ b/fs/xfs/support/debug.h
@@ -27,8 +27,6 @@
 #define CE_ALERT        1               /* alert        */
 #define CE_PANIC        0               /* panic        */
 
-extern void icmn_err(int, char *, va_list)
-	__attribute__ ((format (printf, 2, 0)));
 extern void cmn_err(int, char *, ...)
 	__attribute__ ((format (printf, 2, 3)));
 extern void assfail(char *expr, char *f, int l);
diff --git a/fs/xfs/support/ktrace.c b/fs/xfs/support/ktrace.c
index a34ef05489b..2d494c26717 100644
--- a/fs/xfs/support/ktrace.c
+++ b/fs/xfs/support/ktrace.c
@@ -113,21 +113,16 @@ ktrace_alloc(int nentries, unsigned int __nocast sleep)
 void
 ktrace_free(ktrace_t *ktp)
 {
-	int     entries_size;
-
 	if (ktp == (ktrace_t *)NULL)
 		return;
 
 	/*
 	 * Special treatment for the Vnode trace buffer.
 	 */
-	if (ktp->kt_nentries == ktrace_zentries) {
+	if (ktp->kt_nentries == ktrace_zentries)
 		kmem_zone_free(ktrace_ent_zone, ktp->kt_entries);
-	} else {
-		entries_size = (int)(ktp->kt_nentries * sizeof(ktrace_entry_t));
-
+	else
 		kmem_free(ktp->kt_entries);
-	}
 
 	kmem_zone_free(ktrace_hdr_zone, ktp);
 }
diff --git a/fs/xfs/xfs.h b/fs/xfs/xfs.h
index 540e4c98982..17254b529c5 100644
--- a/fs/xfs/xfs.h
+++ b/fs/xfs/xfs.h
@@ -30,7 +30,7 @@
 #define XFS_ATTR_TRACE 1
 #define XFS_BLI_TRACE 1
 #define XFS_BMAP_TRACE 1
-#define XFS_BMBT_TRACE 1
+#define XFS_BTREE_TRACE 1
 #define XFS_DIR2_TRACE 1
 #define XFS_DQUOT_TRACE 1
 #define XFS_ILOCK_TRACE 1
diff --git a/fs/xfs/xfs_acl.c b/fs/xfs/xfs_acl.c
index 91d69338d3b..a8cdd73999a 100644
--- a/fs/xfs/xfs_acl.c
+++ b/fs/xfs/xfs_acl.c
@@ -758,7 +758,7 @@ xfs_acl_setmode(
 	if (gap && nomask)
 		iattr.ia_mode |= gap->ae_perm << 3;
 
-	return xfs_setattr(XFS_I(vp), &iattr, 0, sys_cred);
+	return xfs_setattr(XFS_I(vp), &iattr, 0);
 }
 
 /*
diff --git a/fs/xfs/xfs_ag.h b/fs/xfs/xfs_ag.h
index 61b292a9fb4..f2e21817a22 100644
--- a/fs/xfs/xfs_ag.h
+++ b/fs/xfs/xfs_ag.h
@@ -91,6 +91,8 @@ typedef struct xfs_agf {
 #define	XFS_AGF_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGF_DADDR(mp))
 #define	XFS_BUF_TO_AGF(bp)	((xfs_agf_t *)XFS_BUF_PTR(bp))
 
+extern int xfs_read_agf(struct xfs_mount *mp, struct xfs_trans *tp,
+			xfs_agnumber_t agno, int flags, struct xfs_buf **bpp);
 
 /*
  * Size of the unlinked inode hash table in the agi.
@@ -142,6 +144,9 @@ typedef struct xfs_agi {
 #define	XFS_AGI_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGI_DADDR(mp))
 #define	XFS_BUF_TO_AGI(bp)	((xfs_agi_t *)XFS_BUF_PTR(bp))
 
+extern int xfs_read_agi(struct xfs_mount *mp, struct xfs_trans *tp,
+				xfs_agnumber_t agno, struct xfs_buf **bpp);
+
 /*
  * The third a.g. block contains the a.g. freelist, an array
  * of block pointers to blocks owned by the allocation btree code.
@@ -192,17 +197,23 @@ typedef struct xfs_perag
 	xfs_agino_t	pagi_freecount;	/* number of free inodes */
 	xfs_agino_t	pagi_count;	/* number of allocated inodes */
 	int		pagb_count;	/* pagb slots in use */
+	xfs_perag_busy_t *pagb_list;	/* unstable blocks */
 #ifdef __KERNEL__
 	spinlock_t	pagb_lock;	/* lock for pagb_list */
-#endif
-	xfs_perag_busy_t *pagb_list;	/* unstable blocks */
+
 	atomic_t        pagf_fstrms;    /* # of filestreams active in this AG */
 
 	int		pag_ici_init;	/* incore inode cache initialised */
 	rwlock_t	pag_ici_lock;	/* incore inode lock */
 	struct radix_tree_root pag_ici_root;	/* incore inode cache root */
+#endif
 } xfs_perag_t;
 
+/*
+ * tags for inode radix tree
+ */
+#define XFS_ICI_RECLAIM_TAG	0	/* inode is to be reclaimed */
+
 #define	XFS_AG_MAXLEVELS(mp)		((mp)->m_ag_maxlevels)
 #define	XFS_MIN_FREELIST_RAW(bl,cl,mp)	\
 	(MIN(bl + 1, XFS_AG_MAXLEVELS(mp)) + MIN(cl + 1, XFS_AG_MAXLEVELS(mp)))
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index 1956f83489f..028e44e58ea 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -90,6 +90,92 @@ STATIC int xfs_alloc_ag_vextent_small(xfs_alloc_arg_t *,
  */
 
 /*
+ * Lookup the record equal to [bno, len] in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_alloc_lookup_eq(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+/*
+ * Lookup the first record greater than or equal to [bno, len]
+ * in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_alloc_lookup_ge(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+ * Lookup the first record less than or equal to [bno, len]
+ * in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_alloc_lookup_le(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len,	/* length of extent */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.a.ar_startblock = bno;
+	cur->bc_rec.a.ar_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
+}
+
+/*
+ * Update the record referred to by cur to the value given
+ * by [bno, len].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int				/* error */
+xfs_alloc_update(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* starting block of extent */
+	xfs_extlen_t		len)	/* length of extent */
+{
+	union xfs_btree_rec	rec;
+
+	rec.alloc.ar_startblock = cpu_to_be32(bno);
+	rec.alloc.ar_blockcount = cpu_to_be32(len);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+STATIC int				/* error */
+xfs_alloc_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		*bno,	/* output: starting block of extent */
+	xfs_extlen_t		*len,	/* output: length of extent */
+	int			*stat)	/* output: success/failure */
+{
+	union xfs_btree_rec	*rec;
+	int			error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (!error && *stat == 1) {
+		*bno = be32_to_cpu(rec->alloc.ar_startblock);
+		*len = be32_to_cpu(rec->alloc.ar_blockcount);
+	}
+	return error;
+}
+
+/*
  * Compute aligned version of the found extent.
  * Takes alignment and min length into account.
  */
@@ -294,21 +380,20 @@ xfs_alloc_fixup_trees(
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 1);
 	}
+
 #ifdef DEBUG
-	{
-		xfs_alloc_block_t	*bnoblock;
-		xfs_alloc_block_t	*cntblock;
-
-		if (bno_cur->bc_nlevels == 1 &&
-		    cnt_cur->bc_nlevels == 1) {
-			bnoblock = XFS_BUF_TO_ALLOC_BLOCK(bno_cur->bc_bufs[0]);
-			cntblock = XFS_BUF_TO_ALLOC_BLOCK(cnt_cur->bc_bufs[0]);
-			XFS_WANT_CORRUPTED_RETURN(
-				be16_to_cpu(bnoblock->bb_numrecs) ==
-				be16_to_cpu(cntblock->bb_numrecs));
-		}
+	if (bno_cur->bc_nlevels == 1 && cnt_cur->bc_nlevels == 1) {
+		struct xfs_btree_block	*bnoblock;
+		struct xfs_btree_block	*cntblock;
+
+		bnoblock = XFS_BUF_TO_BLOCK(bno_cur->bc_bufs[0]);
+		cntblock = XFS_BUF_TO_BLOCK(cnt_cur->bc_bufs[0]);
+
+		XFS_WANT_CORRUPTED_RETURN(
+			bnoblock->bb_numrecs == cntblock->bb_numrecs);
 	}
 #endif
+
 	/*
 	 * Deal with all four cases: the allocated record is contained
 	 * within the freespace record, so we can have new freespace
@@ -333,7 +418,7 @@ xfs_alloc_fixup_trees(
 	/*
 	 * Delete the entry from the by-size btree.
 	 */
-	if ((error = xfs_alloc_delete(cnt_cur, &i)))
+	if ((error = xfs_btree_delete(cnt_cur, &i)))
 		return error;
 	XFS_WANT_CORRUPTED_RETURN(i == 1);
 	/*
@@ -343,7 +428,7 @@ xfs_alloc_fixup_trees(
 		if ((error = xfs_alloc_lookup_eq(cnt_cur, nfbno1, nflen1, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 0);
-		if ((error = xfs_alloc_insert(cnt_cur, &i)))
+		if ((error = xfs_btree_insert(cnt_cur, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 1);
 	}
@@ -351,7 +436,7 @@ xfs_alloc_fixup_trees(
 		if ((error = xfs_alloc_lookup_eq(cnt_cur, nfbno2, nflen2, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 0);
-		if ((error = xfs_alloc_insert(cnt_cur, &i)))
+		if ((error = xfs_btree_insert(cnt_cur, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 1);
 	}
@@ -362,7 +447,7 @@ xfs_alloc_fixup_trees(
 		/*
 		 * No remaining freespace, just delete the by-block tree entry.
 		 */
-		if ((error = xfs_alloc_delete(bno_cur, &i)))
+		if ((error = xfs_btree_delete(bno_cur, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 1);
 	} else {
@@ -379,7 +464,7 @@ xfs_alloc_fixup_trees(
 		if ((error = xfs_alloc_lookup_eq(bno_cur, nfbno2, nflen2, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 0);
-		if ((error = xfs_alloc_insert(bno_cur, &i)))
+		if ((error = xfs_btree_insert(bno_cur, &i)))
 			return error;
 		XFS_WANT_CORRUPTED_RETURN(i == 1);
 	}
@@ -640,8 +725,8 @@ xfs_alloc_ag_vextent_exact(
 	/*
 	 * Allocate/initialize a cursor for the by-number freespace btree.
 	 */
-	bno_cur = xfs_btree_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_BNO, NULL, 0);
+	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_BNO);
 	/*
 	 * Lookup bno and minlen in the btree (minlen is irrelevant, really).
 	 * Look for the closest free block <= bno, it must contain bno
@@ -696,8 +781,8 @@ xfs_alloc_ag_vextent_exact(
 	 * We are allocating agbno for rlen [agbno .. end]
 	 * Allocate/initialize a cursor for the by-size btree.
 	 */
-	cnt_cur = xfs_btree_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_CNT, NULL, 0);
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_CNT);
 	ASSERT(args->agbno + args->len <=
 		be32_to_cpu(XFS_BUF_TO_AGF(args->agbp)->agf_length));
 	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen,
@@ -759,8 +844,8 @@ xfs_alloc_ag_vextent_near(
 	/*
 	 * Get a cursor for the by-size btree.
 	 */
-	cnt_cur = xfs_btree_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_CNT, NULL, 0);
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_CNT);
 	ltlen = 0;
 	bno_cur_lt = bno_cur_gt = NULL;
 	/*
@@ -818,7 +903,7 @@ xfs_alloc_ag_vextent_near(
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 				if (ltlen >= args->minlen)
 					break;
-				if ((error = xfs_alloc_increment(cnt_cur, 0, &i)))
+				if ((error = xfs_btree_increment(cnt_cur, 0, &i)))
 					goto error0;
 			} while (i);
 			ASSERT(ltlen >= args->minlen);
@@ -828,7 +913,7 @@ xfs_alloc_ag_vextent_near(
 		i = cnt_cur->bc_ptrs[0];
 		for (j = 1, blen = 0, bdiff = 0;
 		     !error && j && (blen < args->maxlen || bdiff > 0);
-		     error = xfs_alloc_increment(cnt_cur, 0, &j)) {
+		     error = xfs_btree_increment(cnt_cur, 0, &j)) {
 			/*
 			 * For each entry, decide if it's better than
 			 * the previous best entry.
@@ -886,8 +971,8 @@ xfs_alloc_ag_vextent_near(
 		/*
 		 * Set up a cursor for the by-bno tree.
 		 */
-		bno_cur_lt = xfs_btree_init_cursor(args->mp, args->tp,
-			args->agbp, args->agno, XFS_BTNUM_BNO, NULL, 0);
+		bno_cur_lt = xfs_allocbt_init_cursor(args->mp, args->tp,
+			args->agbp, args->agno, XFS_BTNUM_BNO);
 		/*
 		 * Fix up the btree entries.
 		 */
@@ -914,8 +999,8 @@ xfs_alloc_ag_vextent_near(
 	/*
 	 * Allocate and initialize the cursor for the leftward search.
 	 */
-	bno_cur_lt = xfs_btree_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_BNO, NULL, 0);
+	bno_cur_lt = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_BNO);
 	/*
 	 * Lookup <= bno to find the leftward search's starting point.
 	 */
@@ -938,7 +1023,7 @@ xfs_alloc_ag_vextent_near(
 	 * Increment the cursor, so we will point at the entry just right
 	 * of the leftward entry if any, or to the leftmost entry.
 	 */
-	if ((error = xfs_alloc_increment(bno_cur_gt, 0, &i)))
+	if ((error = xfs_btree_increment(bno_cur_gt, 0, &i)))
 		goto error0;
 	if (!i) {
 		/*
@@ -961,7 +1046,7 @@ xfs_alloc_ag_vextent_near(
 					args->minlen, &ltbnoa, &ltlena);
 			if (ltlena >= args->minlen)
 				break;
-			if ((error = xfs_alloc_decrement(bno_cur_lt, 0, &i)))
+			if ((error = xfs_btree_decrement(bno_cur_lt, 0, &i)))
 				goto error0;
 			if (!i) {
 				xfs_btree_del_cursor(bno_cur_lt,
@@ -977,7 +1062,7 @@ xfs_alloc_ag_vextent_near(
 					args->minlen, &gtbnoa, &gtlena);
 			if (gtlena >= args->minlen)
 				break;
-			if ((error = xfs_alloc_increment(bno_cur_gt, 0, &i)))
+			if ((error = xfs_btree_increment(bno_cur_gt, 0, &i)))
 				goto error0;
 			if (!i) {
 				xfs_btree_del_cursor(bno_cur_gt,
@@ -1066,7 +1151,7 @@ xfs_alloc_ag_vextent_near(
 					/*
 					 * Fell off the right end.
 					 */
-					if ((error = xfs_alloc_increment(
+					if ((error = xfs_btree_increment(
 							bno_cur_gt, 0, &i)))
 						goto error0;
 					if (!i) {
@@ -1162,7 +1247,7 @@ xfs_alloc_ag_vextent_near(
 					/*
 					 * Fell off the left end.
 					 */
-					if ((error = xfs_alloc_decrement(
+					if ((error = xfs_btree_decrement(
 							bno_cur_lt, 0, &i)))
 						goto error0;
 					if (!i) {
@@ -1267,8 +1352,8 @@ xfs_alloc_ag_vextent_size(
 	/*
 	 * Allocate and initialize a cursor for the by-size btree.
 	 */
-	cnt_cur = xfs_btree_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_CNT, NULL, 0);
+	cnt_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_CNT);
 	bno_cur = NULL;
 	/*
 	 * Look for an entry >= maxlen+alignment-1 blocks.
@@ -1321,7 +1406,7 @@ xfs_alloc_ag_vextent_size(
 		bestflen = flen;
 		bestfbno = fbno;
 		for (;;) {
-			if ((error = xfs_alloc_decrement(cnt_cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cnt_cur, 0, &i)))
 				goto error0;
 			if (i == 0)
 				break;
@@ -1372,8 +1457,8 @@ xfs_alloc_ag_vextent_size(
 	/*
 	 * Allocate and initialize a cursor for the by-block tree.
 	 */
-	bno_cur = xfs_btree_init_cursor(args->mp, args->tp, args->agbp,
-		args->agno, XFS_BTNUM_BNO, NULL, 0);
+	bno_cur = xfs_allocbt_init_cursor(args->mp, args->tp, args->agbp,
+		args->agno, XFS_BTNUM_BNO);
 	if ((error = xfs_alloc_fixup_trees(cnt_cur, bno_cur, fbno, flen,
 			rbno, rlen, XFSA_FIXUP_CNT_OK)))
 		goto error0;
@@ -1416,7 +1501,7 @@ xfs_alloc_ag_vextent_small(
 	xfs_extlen_t	flen;
 	int		i;
 
-	if ((error = xfs_alloc_decrement(ccur, 0, &i)))
+	if ((error = xfs_btree_decrement(ccur, 0, &i)))
 		goto error0;
 	if (i) {
 		if ((error = xfs_alloc_get_rec(ccur, &fbno, &flen, &i)))
@@ -1515,8 +1600,7 @@ xfs_free_ag_extent(
 	/*
 	 * Allocate and initialize a cursor for the by-block btree.
 	 */
-	bno_cur = xfs_btree_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_BNO, NULL,
-		0);
+	bno_cur = xfs_allocbt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_BNO);
 	cnt_cur = NULL;
 	/*
 	 * Look for a neighboring block on the left (lower block numbers)
@@ -1549,7 +1633,7 @@ xfs_free_ag_extent(
 	 * Look for a neighboring block on the right (higher block numbers)
 	 * that is contiguous with this space.
 	 */
-	if ((error = xfs_alloc_increment(bno_cur, 0, &haveright)))
+	if ((error = xfs_btree_increment(bno_cur, 0, &haveright)))
 		goto error0;
 	if (haveright) {
 		/*
@@ -1575,8 +1659,7 @@ xfs_free_ag_extent(
 	/*
 	 * Now allocate and initialize a cursor for the by-size tree.
 	 */
-	cnt_cur = xfs_btree_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_CNT, NULL,
-		0);
+	cnt_cur = xfs_allocbt_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_CNT);
 	/*
 	 * Have both left and right contiguous neighbors.
 	 * Merge all three into a single free block.
@@ -1588,7 +1671,7 @@ xfs_free_ag_extent(
 		if ((error = xfs_alloc_lookup_eq(cnt_cur, ltbno, ltlen, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_alloc_delete(cnt_cur, &i)))
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		/*
@@ -1597,19 +1680,19 @@ xfs_free_ag_extent(
 		if ((error = xfs_alloc_lookup_eq(cnt_cur, gtbno, gtlen, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_alloc_delete(cnt_cur, &i)))
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		/*
 		 * Delete the old by-block entry for the right block.
 		 */
-		if ((error = xfs_alloc_delete(bno_cur, &i)))
+		if ((error = xfs_btree_delete(bno_cur, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		/*
 		 * Move the by-block cursor back to the left neighbor.
 		 */
-		if ((error = xfs_alloc_decrement(bno_cur, 0, &i)))
+		if ((error = xfs_btree_decrement(bno_cur, 0, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 #ifdef DEBUG
@@ -1648,14 +1731,14 @@ xfs_free_ag_extent(
 		if ((error = xfs_alloc_lookup_eq(cnt_cur, ltbno, ltlen, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_alloc_delete(cnt_cur, &i)))
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		/*
 		 * Back up the by-block cursor to the left neighbor, and
 		 * update its length.
 		 */
-		if ((error = xfs_alloc_decrement(bno_cur, 0, &i)))
+		if ((error = xfs_btree_decrement(bno_cur, 0, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		nbno = ltbno;
@@ -1674,7 +1757,7 @@ xfs_free_ag_extent(
 		if ((error = xfs_alloc_lookup_eq(cnt_cur, gtbno, gtlen, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_alloc_delete(cnt_cur, &i)))
+		if ((error = xfs_btree_delete(cnt_cur, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 		/*
@@ -1693,7 +1776,7 @@ xfs_free_ag_extent(
 	else {
 		nbno = bno;
 		nlen = len;
-		if ((error = xfs_alloc_insert(bno_cur, &i)))
+		if ((error = xfs_btree_insert(bno_cur, &i)))
 			goto error0;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 	}
@@ -1705,7 +1788,7 @@ xfs_free_ag_extent(
 	if ((error = xfs_alloc_lookup_eq(cnt_cur, nbno, nlen, &i)))
 		goto error0;
 	XFS_WANT_CORRUPTED_GOTO(i == 0, error0);
-	if ((error = xfs_alloc_insert(cnt_cur, &i)))
+	if ((error = xfs_btree_insert(cnt_cur, &i)))
 		goto error0;
 	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 	xfs_btree_del_cursor(cnt_cur, XFS_BTREE_NOERROR);
@@ -2150,51 +2233,83 @@ xfs_alloc_put_freelist(
  * Read in the allocation group header (free/alloc section).
  */
 int					/* error */
-xfs_alloc_read_agf(
-	xfs_mount_t	*mp,		/* mount point structure */
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_agnumber_t	agno,		/* allocation group number */
-	int		flags,		/* XFS_ALLOC_FLAG_... */
-	xfs_buf_t	**bpp)		/* buffer for the ag freelist header */
+xfs_read_agf(
+	struct xfs_mount	*mp,	/* mount point structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	int			flags,	/* XFS_BUF_ */
+	struct xfs_buf		**bpp)	/* buffer for the ag freelist header */
 {
-	xfs_agf_t	*agf;		/* ag freelist header */
+	struct xfs_agf	*agf;		/* ag freelist header */
 	int		agf_ok;		/* set if agf is consistent */
-	xfs_buf_t	*bp;		/* return value */
-	xfs_perag_t	*pag;		/* per allocation group data */
 	int		error;
 
 	ASSERT(agno != NULLAGNUMBER);
 	error = xfs_trans_read_buf(
 			mp, tp, mp->m_ddev_targp,
 			XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
-			XFS_FSS_TO_BB(mp, 1),
-			(flags & XFS_ALLOC_FLAG_TRYLOCK) ? XFS_BUF_TRYLOCK : 0U,
-			&bp);
+			XFS_FSS_TO_BB(mp, 1), flags, bpp);
 	if (error)
 		return error;
-	ASSERT(!bp || !XFS_BUF_GETERROR(bp));
-	if (!bp) {
-		*bpp = NULL;
+	if (!*bpp)
 		return 0;
-	}
+
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+	agf = XFS_BUF_TO_AGF(*bpp);
+
 	/*
 	 * Validate the magic number of the agf block.
 	 */
-	agf = XFS_BUF_TO_AGF(bp);
 	agf_ok =
 		be32_to_cpu(agf->agf_magicnum) == XFS_AGF_MAGIC &&
 		XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
 		be32_to_cpu(agf->agf_freeblks) <= be32_to_cpu(agf->agf_length) &&
 		be32_to_cpu(agf->agf_flfirst) < XFS_AGFL_SIZE(mp) &&
 		be32_to_cpu(agf->agf_fllast) < XFS_AGFL_SIZE(mp) &&
-		be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp);
+		be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp) &&
+		be32_to_cpu(agf->agf_seqno) == agno;
+	if (xfs_sb_version_haslazysbcount(&mp->m_sb))
+		agf_ok = agf_ok && be32_to_cpu(agf->agf_btreeblks) <=
+						be32_to_cpu(agf->agf_length);
 	if (unlikely(XFS_TEST_ERROR(!agf_ok, mp, XFS_ERRTAG_ALLOC_READ_AGF,
 			XFS_RANDOM_ALLOC_READ_AGF))) {
 		XFS_CORRUPTION_ERROR("xfs_alloc_read_agf",
 				     XFS_ERRLEVEL_LOW, mp, agf);
-		xfs_trans_brelse(tp, bp);
+		xfs_trans_brelse(tp, *bpp);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
+
+	XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_AGF, XFS_AGF_REF);
+	return 0;
+}
+
+/*
+ * Read in the allocation group header (free/alloc section).
+ */
+int					/* error */
+xfs_alloc_read_agf(
+	struct xfs_mount	*mp,	/* mount point structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	int			flags,	/* XFS_ALLOC_FLAG_... */
+	struct xfs_buf		**bpp)	/* buffer for the ag freelist header */
+{
+	struct xfs_agf		*agf;		/* ag freelist header */
+	struct xfs_perag	*pag;		/* per allocation group data */
+	int			error;
+
+	ASSERT(agno != NULLAGNUMBER);
+
+	error = xfs_read_agf(mp, tp, agno,
+			(flags & XFS_ALLOC_FLAG_TRYLOCK) ? XFS_BUF_TRYLOCK : 0,
+			bpp);
+	if (error)
+		return error;
+	if (!*bpp)
+		return 0;
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+
+	agf = XFS_BUF_TO_AGF(*bpp);
 	pag = &mp->m_perag[agno];
 	if (!pag->pagf_init) {
 		pag->pagf_freeblks = be32_to_cpu(agf->agf_freeblks);
@@ -2213,6 +2328,7 @@ xfs_alloc_read_agf(
 #ifdef DEBUG
 	else if (!XFS_FORCED_SHUTDOWN(mp)) {
 		ASSERT(pag->pagf_freeblks == be32_to_cpu(agf->agf_freeblks));
+		ASSERT(pag->pagf_btreeblks == be32_to_cpu(agf->agf_btreeblks));
 		ASSERT(pag->pagf_flcount == be32_to_cpu(agf->agf_flcount));
 		ASSERT(pag->pagf_longest == be32_to_cpu(agf->agf_longest));
 		ASSERT(pag->pagf_levels[XFS_BTNUM_BNOi] ==
@@ -2221,8 +2337,6 @@ xfs_alloc_read_agf(
 		       be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]));
 	}
 #endif
-	XFS_BUF_SET_VTYPE_REF(bp, B_FS_AGF, XFS_AGF_REF);
-	*bpp = bp;
 	return 0;
 }
 
diff --git a/fs/xfs/xfs_alloc.h b/fs/xfs/xfs_alloc.h
index 5aec15d0651..588172796f7 100644
--- a/fs/xfs/xfs_alloc.h
+++ b/fs/xfs/xfs_alloc.h
@@ -121,6 +121,19 @@ extern ktrace_t *xfs_alloc_trace_buf;
 #define	XFS_ALLOC_KTRACE_BUSYSEARCH	6
 #endif
 
+void
+xfs_alloc_mark_busy(xfs_trans_t *tp,
+		xfs_agnumber_t agno,
+		xfs_agblock_t bno,
+		xfs_extlen_t len);
+
+void
+xfs_alloc_clear_busy(xfs_trans_t *tp,
+		xfs_agnumber_t ag,
+		int idx);
+
+#endif	/* __KERNEL__ */
+
 /*
  * Compute and fill in value of m_ag_maxlevels.
  */
@@ -196,18 +209,4 @@ xfs_free_extent(
 	xfs_fsblock_t	bno,	/* starting block number of extent */
 	xfs_extlen_t	len);	/* length of extent */
 
-void
-xfs_alloc_mark_busy(xfs_trans_t *tp,
-		xfs_agnumber_t agno,
-		xfs_agblock_t bno,
-		xfs_extlen_t len);
-
-void
-xfs_alloc_clear_busy(xfs_trans_t *tp,
-		xfs_agnumber_t ag,
-		int idx);
-
-
-#endif	/* __KERNEL__ */
-
 #endif	/* __XFS_ALLOC_H__ */
diff --git a/fs/xfs/xfs_alloc_btree.c b/fs/xfs/xfs_alloc_btree.c
index 3ce2645508a..733cb75a8c5 100644
--- a/fs/xfs/xfs_alloc_btree.c
+++ b/fs/xfs/xfs_alloc_btree.c
@@ -35,2177 +35,464 @@
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
 #include "xfs_btree.h"
+#include "xfs_btree_trace.h"
 #include "xfs_ialloc.h"
 #include "xfs_alloc.h"
 #include "xfs_error.h"
 
-/*
- * Prototypes for internal functions.
- */
 
-STATIC void xfs_alloc_log_block(xfs_trans_t *, xfs_buf_t *, int);
-STATIC void xfs_alloc_log_keys(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC void xfs_alloc_log_ptrs(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC void xfs_alloc_log_recs(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC int xfs_alloc_lshift(xfs_btree_cur_t *, int, int *);
-STATIC int xfs_alloc_newroot(xfs_btree_cur_t *, int *);
-STATIC int xfs_alloc_rshift(xfs_btree_cur_t *, int, int *);
-STATIC int xfs_alloc_split(xfs_btree_cur_t *, int, xfs_agblock_t *,
-		xfs_alloc_key_t *, xfs_btree_cur_t **, int *);
-STATIC int xfs_alloc_updkey(xfs_btree_cur_t *, xfs_alloc_key_t *, int);
+STATIC struct xfs_btree_cur *
+xfs_allocbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_allocbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_private.a.agbp, cur->bc_private.a.agno,
+			cur->bc_btnum);
+}
 
-/*
- * Internal functions.
- */
+STATIC void
+xfs_allocbt_set_root(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			inc)
+{
+	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
+	int			btnum = cur->bc_btnum;
 
-/*
- * Single level of the xfs_alloc_delete record deletion routine.
- * Delete record pointed to by cur/level.
- * Remove the record from its block then rebalance the tree.
- * Return 0 for error, 1 for done, 2 to go on to the next level.
- */
-STATIC int				/* error */
-xfs_alloc_delrec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level removing record from */
-	int			*stat)	/* fail/done/go-on */
+	ASSERT(ptr->s != 0);
+
+	agf->agf_roots[btnum] = ptr->s;
+	be32_add_cpu(&agf->agf_levels[btnum], inc);
+	cur->bc_mp->m_perag[seqno].pagf_levels[btnum] += inc;
+
+	xfs_alloc_log_agf(cur->bc_tp, agbp, XFS_AGF_ROOTS | XFS_AGF_LEVELS);
+}
+
+STATIC int
+xfs_allocbt_alloc_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*start,
+	union xfs_btree_ptr	*new,
+	int			length,
+	int			*stat)
 {
-	xfs_agf_t		*agf;	/* allocation group freelist header */
-	xfs_alloc_block_t	*block;	/* btree block record/key lives in */
-	xfs_agblock_t		bno;	/* btree block number */
-	xfs_buf_t		*bp;	/* buffer for block */
-	int			error;	/* error return value */
-	int			i;	/* loop index */
-	xfs_alloc_key_t		key;	/* kp points here if block is level 0 */
-	xfs_agblock_t		lbno;	/* left block's block number */
-	xfs_buf_t		*lbp;	/* left block's buffer pointer */
-	xfs_alloc_block_t	*left;	/* left btree block */
-	xfs_alloc_key_t		*lkp=NULL;	/* left block key pointer */
-	xfs_alloc_ptr_t		*lpp=NULL;	/* left block address pointer */
-	int			lrecs=0;	/* number of records in left block */
-	xfs_alloc_rec_t		*lrp;	/* left block record pointer */
-	xfs_mount_t		*mp;	/* mount structure */
-	int			ptr;	/* index in btree block for this rec */
-	xfs_agblock_t		rbno;	/* right block's block number */
-	xfs_buf_t		*rbp;	/* right block's buffer pointer */
-	xfs_alloc_block_t	*right;	/* right btree block */
-	xfs_alloc_key_t		*rkp;	/* right block key pointer */
-	xfs_alloc_ptr_t		*rpp;	/* right block address pointer */
-	int			rrecs=0;	/* number of records in right block */
-	int			numrecs;
-	xfs_alloc_rec_t		*rrp;	/* right block record pointer */
-	xfs_btree_cur_t		*tcur;	/* temporary btree cursor */
+	int			error;
+	xfs_agblock_t		bno;
 
-	/*
-	 * Get the index of the entry being deleted, check for nothing there.
-	 */
-	ptr = cur->bc_ptrs[level];
-	if (ptr == 0) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Get the buffer & block containing the record or key/ptr.
-	 */
-	bp = cur->bc_bufs[level];
-	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	/* Allocate the new block from the freelist. If we can't, give up.  */
+	error = xfs_alloc_get_freelist(cur->bc_tp, cur->bc_private.a.agbp,
+				       &bno, 1);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
 		return error;
-#endif
-	/*
-	 * Fail if we're off the end of the block.
-	 */
-	numrecs = be16_to_cpu(block->bb_numrecs);
-	if (ptr > numrecs) {
+	}
+
+	if (bno == NULLAGBLOCK) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
 		*stat = 0;
 		return 0;
 	}
-	XFS_STATS_INC(xs_abt_delrec);
-	/*
-	 * It's a nonleaf.  Excise the key and ptr being deleted, by
-	 * sliding the entries past them down one.
-	 * Log the changed areas of the block.
-	 */
-	if (level > 0) {
-		lkp = XFS_ALLOC_KEY_ADDR(block, 1, cur);
-		lpp = XFS_ALLOC_PTR_ADDR(block, 1, cur);
-#ifdef DEBUG
-		for (i = ptr; i < numrecs; i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(lpp[i]), level)))
-				return error;
-		}
-#endif
-		if (ptr < numrecs) {
-			memmove(&lkp[ptr - 1], &lkp[ptr],
-				(numrecs - ptr) * sizeof(*lkp));
-			memmove(&lpp[ptr - 1], &lpp[ptr],
-				(numrecs - ptr) * sizeof(*lpp));
-			xfs_alloc_log_ptrs(cur, bp, ptr, numrecs - 1);
-			xfs_alloc_log_keys(cur, bp, ptr, numrecs - 1);
-		}
-	}
-	/*
-	 * It's a leaf.  Excise the record being deleted, by sliding the
-	 * entries past it down one.  Log the changed areas of the block.
-	 */
-	else {
-		lrp = XFS_ALLOC_REC_ADDR(block, 1, cur);
-		if (ptr < numrecs) {
-			memmove(&lrp[ptr - 1], &lrp[ptr],
-				(numrecs - ptr) * sizeof(*lrp));
-			xfs_alloc_log_recs(cur, bp, ptr, numrecs - 1);
-		}
-		/*
-		 * If it's the first record in the block, we'll need a key
-		 * structure to pass up to the next level (updkey).
-		 */
-		if (ptr == 1) {
-			key.ar_startblock = lrp->ar_startblock;
-			key.ar_blockcount = lrp->ar_blockcount;
-			lkp = &key;
-		}
-	}
-	/*
-	 * Decrement and log the number of entries in the block.
-	 */
-	numrecs--;
-	block->bb_numrecs = cpu_to_be16(numrecs);
-	xfs_alloc_log_block(cur->bc_tp, bp, XFS_BB_NUMRECS);
-	/*
-	 * See if the longest free extent in the allocation group was
-	 * changed by this operation.  True if it's the by-size btree, and
-	 * this is the leaf level, and there is no right sibling block,
-	 * and this was the last record.
-	 */
-	agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-	mp = cur->bc_mp;
 
-	if (level == 0 &&
-	    cur->bc_btnum == XFS_BTNUM_CNT &&
-	    be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK &&
-	    ptr > numrecs) {
-		ASSERT(ptr == numrecs + 1);
-		/*
-		 * There are still records in the block.  Grab the size
-		 * from the last one.
-		 */
-		if (numrecs) {
-			rrp = XFS_ALLOC_REC_ADDR(block, numrecs, cur);
-			agf->agf_longest = rrp->ar_blockcount;
-		}
-		/*
-		 * No free extents left.
-		 */
-		else
-			agf->agf_longest = 0;
-		mp->m_perag[be32_to_cpu(agf->agf_seqno)].pagf_longest =
-			be32_to_cpu(agf->agf_longest);
-		xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp,
-			XFS_AGF_LONGEST);
-	}
-	/*
-	 * Is this the root level?  If so, we're almost done.
-	 */
-	if (level == cur->bc_nlevels - 1) {
-		/*
-		 * If this is the root level,
-		 * and there's only one entry left,
-		 * and it's NOT the leaf level,
-		 * then we can get rid of this level.
-		 */
-		if (numrecs == 1 && level > 0) {
-			/*
-			 * lpp is still set to the first pointer in the block.
-			 * Make it the new root of the btree.
-			 */
-			bno = be32_to_cpu(agf->agf_roots[cur->bc_btnum]);
-			agf->agf_roots[cur->bc_btnum] = *lpp;
-			be32_add_cpu(&agf->agf_levels[cur->bc_btnum], -1);
-			mp->m_perag[be32_to_cpu(agf->agf_seqno)].pagf_levels[cur->bc_btnum]--;
-			/*
-			 * Put this buffer/block on the ag's freelist.
-			 */
-			error = xfs_alloc_put_freelist(cur->bc_tp,
-					cur->bc_private.a.agbp, NULL, bno, 1);
-			if (error)
-				return error;
-			/*
-			 * Since blocks move to the free list without the
-			 * coordination used in xfs_bmap_finish, we can't allow
-			 * block to be available for reallocation and
-			 * non-transaction writing (user data) until we know
-			 * that the transaction that moved it to the free list
-			 * is permanently on disk. We track the blocks by
-			 * declaring these blocks as "busy"; the busy list is
-			 * maintained on a per-ag basis and each transaction
-			 * records which entries should be removed when the
-			 * iclog commits to disk. If a busy block is
-			 * allocated, the iclog is pushed up to the LSN
-			 * that freed the block.
-			 */
-			xfs_alloc_mark_busy(cur->bc_tp,
-				be32_to_cpu(agf->agf_seqno), bno, 1);
+	xfs_trans_agbtree_delta(cur->bc_tp, 1);
+	new->s = cpu_to_be32(bno);
 
-			xfs_trans_agbtree_delta(cur->bc_tp, -1);
-			xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp,
-				XFS_AGF_ROOTS | XFS_AGF_LEVELS);
-			/*
-			 * Update the cursor so there's one fewer level.
-			 */
-			xfs_btree_setbuf(cur, level, NULL);
-			cur->bc_nlevels--;
-		} else if (level > 0 &&
-			   (error = xfs_alloc_decrement(cur, level, &i)))
-			return error;
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * If we deleted the leftmost entry in the block, update the
-	 * key values above us in the tree.
-	 */
-	if (ptr == 1 && (error = xfs_alloc_updkey(cur, lkp, level + 1)))
-		return error;
-	/*
-	 * If the number of records remaining in the block is at least
-	 * the minimum, we're done.
-	 */
-	if (numrecs >= XFS_ALLOC_BLOCK_MINRECS(level, cur)) {
-		if (level > 0 && (error = xfs_alloc_decrement(cur, level, &i)))
-			return error;
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * Otherwise, we have to move some records around to keep the
-	 * tree balanced.  Look at the left and right sibling blocks to
-	 * see if we can re-balance by moving only one record.
-	 */
-	rbno = be32_to_cpu(block->bb_rightsib);
-	lbno = be32_to_cpu(block->bb_leftsib);
-	bno = NULLAGBLOCK;
-	ASSERT(rbno != NULLAGBLOCK || lbno != NULLAGBLOCK);
-	/*
-	 * Duplicate the cursor so our btree manipulations here won't
-	 * disrupt the next level up.
-	 */
-	if ((error = xfs_btree_dup_cursor(cur, &tcur)))
-		return error;
-	/*
-	 * If there's a right sibling, see if it's ok to shift an entry
-	 * out of it.
-	 */
-	if (rbno != NULLAGBLOCK) {
-		/*
-		 * Move the temp cursor to the last entry in the next block.
-		 * Actually any entry but the first would suffice.
-		 */
-		i = xfs_btree_lastrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_alloc_increment(tcur, level, &i)))
-			goto error0;
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		i = xfs_btree_lastrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		/*
-		 * Grab a pointer to the block.
-		 */
-		rbp = tcur->bc_bufs[level];
-		right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-			goto error0;
-#endif
-		/*
-		 * Grab the current block number, for future use.
-		 */
-		bno = be32_to_cpu(right->bb_leftsib);
-		/*
-		 * If right block is full enough so that removing one entry
-		 * won't make it too empty, and left-shifting an entry out
-		 * of right to us works, we're done.
-		 */
-		if (be16_to_cpu(right->bb_numrecs) - 1 >=
-		     XFS_ALLOC_BLOCK_MINRECS(level, cur)) {
-			if ((error = xfs_alloc_lshift(tcur, level, &i)))
-				goto error0;
-			if (i) {
-				ASSERT(be16_to_cpu(block->bb_numrecs) >=
-				       XFS_ALLOC_BLOCK_MINRECS(level, cur));
-				xfs_btree_del_cursor(tcur,
-						     XFS_BTREE_NOERROR);
-				if (level > 0 &&
-				    (error = xfs_alloc_decrement(cur, level,
-					    &i)))
-					return error;
-				*stat = 1;
-				return 0;
-			}
-		}
-		/*
-		 * Otherwise, grab the number of records in right for
-		 * future reference, and fix up the temp cursor to point
-		 * to our block again (last record).
-		 */
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		if (lbno != NULLAGBLOCK) {
-			i = xfs_btree_firstrec(tcur, level);
-			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-			if ((error = xfs_alloc_decrement(tcur, level, &i)))
-				goto error0;
-			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		}
-	}
-	/*
-	 * If there's a left sibling, see if it's ok to shift an entry
-	 * out of it.
-	 */
-	if (lbno != NULLAGBLOCK) {
-		/*
-		 * Move the temp cursor to the first entry in the
-		 * previous block.
-		 */
-		i = xfs_btree_firstrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_alloc_decrement(tcur, level, &i)))
-			goto error0;
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		xfs_btree_firstrec(tcur, level);
-		/*
-		 * Grab a pointer to the block.
-		 */
-		lbp = tcur->bc_bufs[level];
-		left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-			goto error0;
-#endif
-		/*
-		 * Grab the current block number, for future use.
-		 */
-		bno = be32_to_cpu(left->bb_rightsib);
-		/*
-		 * If left block is full enough so that removing one entry
-		 * won't make it too empty, and right-shifting an entry out
-		 * of left to us works, we're done.
-		 */
-		if (be16_to_cpu(left->bb_numrecs) - 1 >=
-		     XFS_ALLOC_BLOCK_MINRECS(level, cur)) {
-			if ((error = xfs_alloc_rshift(tcur, level, &i)))
-				goto error0;
-			if (i) {
-				ASSERT(be16_to_cpu(block->bb_numrecs) >=
-				       XFS_ALLOC_BLOCK_MINRECS(level, cur));
-				xfs_btree_del_cursor(tcur,
-						     XFS_BTREE_NOERROR);
-				if (level == 0)
-					cur->bc_ptrs[0]++;
-				*stat = 1;
-				return 0;
-			}
-		}
-		/*
-		 * Otherwise, grab the number of records in right for
-		 * future reference.
-		 */
-		lrecs = be16_to_cpu(left->bb_numrecs);
-	}
-	/*
-	 * Delete the temp cursor, we're done with it.
-	 */
-	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-	/*
-	 * If here, we need to do a join to keep the tree balanced.
-	 */
-	ASSERT(bno != NULLAGBLOCK);
-	/*
-	 * See if we can join with the left neighbor block.
-	 */
-	if (lbno != NULLAGBLOCK &&
-	    lrecs + numrecs <= XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
-		/*
-		 * Set "right" to be the starting block,
-		 * "left" to be the left neighbor.
-		 */
-		rbno = bno;
-		right = block;
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		rbp = bp;
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, lbno, 0, &lbp,
-				XFS_ALLOC_BTREE_REF)))
-			return error;
-		left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-		lrecs = be16_to_cpu(left->bb_numrecs);
-		if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-			return error;
-	}
-	/*
-	 * If that won't work, see if we can join with the right neighbor block.
-	 */
-	else if (rbno != NULLAGBLOCK &&
-		 rrecs + numrecs <= XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
-		/*
-		 * Set "left" to be the starting block,
-		 * "right" to be the right neighbor.
-		 */
-		lbno = bno;
-		left = block;
-		lrecs = be16_to_cpu(left->bb_numrecs);
-		lbp = bp;
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, rbno, 0, &rbp,
-				XFS_ALLOC_BTREE_REF)))
-			return error;
-		right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-			return error;
-	}
-	/*
-	 * Otherwise, we can't fix the imbalance.
-	 * Just return.  This is probably a logic error, but it's not fatal.
-	 */
-	else {
-		if (level > 0 && (error = xfs_alloc_decrement(cur, level, &i)))
-			return error;
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * We're now going to join "left" and "right" by moving all the stuff
-	 * in "right" to "left" and deleting "right".
-	 */
-	if (level > 0) {
-		/*
-		 * It's a non-leaf.  Move keys and pointers.
-		 */
-		lkp = XFS_ALLOC_KEY_ADDR(left, lrecs + 1, cur);
-		lpp = XFS_ALLOC_PTR_ADDR(left, lrecs + 1, cur);
-		rkp = XFS_ALLOC_KEY_ADDR(right, 1, cur);
-		rpp = XFS_ALLOC_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = 0; i < rrecs; i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i]), level)))
-				return error;
-		}
-#endif
-		memcpy(lkp, rkp, rrecs * sizeof(*lkp));
-		memcpy(lpp, rpp, rrecs * sizeof(*lpp));
-		xfs_alloc_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
-		xfs_alloc_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
-	} else {
-		/*
-		 * It's a leaf.  Move records.
-		 */
-		lrp = XFS_ALLOC_REC_ADDR(left, lrecs + 1, cur);
-		rrp = XFS_ALLOC_REC_ADDR(right, 1, cur);
-		memcpy(lrp, rrp, rrecs * sizeof(*lrp));
-		xfs_alloc_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
-	}
-	/*
-	 * If we joined with the left neighbor, set the buffer in the
-	 * cursor to the left block, and fix up the index.
-	 */
-	if (bp != lbp) {
-		xfs_btree_setbuf(cur, level, lbp);
-		cur->bc_ptrs[level] += lrecs;
-	}
-	/*
-	 * If we joined with the right neighbor and there's a level above
-	 * us, increment the cursor at that level.
-	 */
-	else if (level + 1 < cur->bc_nlevels &&
-		 (error = xfs_alloc_increment(cur, level + 1, &i)))
-		return error;
-	/*
-	 * Fix up the number of records in the surviving block.
-	 */
-	lrecs += rrecs;
-	left->bb_numrecs = cpu_to_be16(lrecs);
-	/*
-	 * Fix up the right block pointer in the surviving block, and log it.
-	 */
-	left->bb_rightsib = right->bb_rightsib;
-	xfs_alloc_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
-	/*
-	 * If there is a right sibling now, make it point to the
-	 * remaining block.
-	 */
-	if (be32_to_cpu(left->bb_rightsib) != NULLAGBLOCK) {
-		xfs_alloc_block_t	*rrblock;
-		xfs_buf_t		*rrbp;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+}
 
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, be32_to_cpu(left->bb_rightsib), 0,
-				&rrbp, XFS_ALLOC_BTREE_REF)))
-			return error;
-		rrblock = XFS_BUF_TO_ALLOC_BLOCK(rrbp);
-		if ((error = xfs_btree_check_sblock(cur, rrblock, level, rrbp)))
-			return error;
-		rrblock->bb_leftsib = cpu_to_be32(lbno);
-		xfs_alloc_log_block(cur->bc_tp, rrbp, XFS_BB_LEFTSIB);
-	}
-	/*
-	 * Free the deleting block by putting it on the freelist.
-	 */
-	error = xfs_alloc_put_freelist(cur->bc_tp,
-					 cur->bc_private.a.agbp, NULL, rbno, 1);
+STATIC int
+xfs_allocbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	xfs_agblock_t		bno;
+	int			error;
+
+	bno = XFS_DADDR_TO_AGBNO(cur->bc_mp, XFS_BUF_ADDR(bp));
+	error = xfs_alloc_put_freelist(cur->bc_tp, agbp, NULL, bno, 1);
 	if (error)
 		return error;
+
 	/*
-	 * Since blocks move to the free list without the coordination
-	 * used in xfs_bmap_finish, we can't allow block to be available
-	 * for reallocation and non-transaction writing (user data)
-	 * until we know that the transaction that moved it to the free
-	 * list is permanently on disk. We track the blocks by declaring
-	 * these blocks as "busy"; the busy list is maintained on a
-	 * per-ag basis and each transaction records which entries
-	 * should be removed when the iclog commits to disk. If a
-	 * busy block is allocated, the iclog is pushed up to the
+	 * Since blocks move to the free list without the coordination used in
+	 * xfs_bmap_finish, we can't allow block to be available for
+	 * reallocation and non-transaction writing (user data) until we know
+	 * that the transaction that moved it to the free list is permanently
+	 * on disk. We track the blocks by declaring these blocks as "busy";
+	 * the busy list is maintained on a per-ag basis and each transaction
+	 * records which entries should be removed when the iclog commits to
+	 * disk. If a busy block is allocated, the iclog is pushed up to the
 	 * LSN that freed the block.
 	 */
 	xfs_alloc_mark_busy(cur->bc_tp, be32_to_cpu(agf->agf_seqno), bno, 1);
 	xfs_trans_agbtree_delta(cur->bc_tp, -1);
-
-	/*
-	 * Adjust the current level's cursor so that we're left referring
-	 * to the right node, after we're done.
-	 * If this leaves the ptr value 0 our caller will fix it up.
-	 */
-	if (level > 0)
-		cur->bc_ptrs[level]--;
-	/*
-	 * Return value means the next level up has something to do.
-	 */
-	*stat = 2;
 	return 0;
-
-error0:
-	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
-	return error;
 }
 
 /*
- * Insert one record/level.  Return information to the caller
- * allowing the next level up to proceed if necessary.
+ * Update the longest extent in the AGF
  */
-STATIC int				/* error */
-xfs_alloc_insrec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to insert record at */
-	xfs_agblock_t		*bnop,	/* i/o: block number inserted */
-	xfs_alloc_rec_t		*recp,	/* i/o: record data inserted */
-	xfs_btree_cur_t		**curp,	/* output: new cursor replacing cur */
-	int			*stat)	/* output: success/failure */
+STATIC void
+xfs_allocbt_update_lastrec(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_rec	*rec,
+	int			ptr,
+	int			reason)
 {
-	xfs_agf_t		*agf;	/* allocation group freelist header */
-	xfs_alloc_block_t	*block;	/* btree block record/key lives in */
-	xfs_buf_t		*bp;	/* buffer for block */
-	int			error;	/* error return value */
-	int			i;	/* loop index */
-	xfs_alloc_key_t		key;	/* key value being inserted */
-	xfs_alloc_key_t		*kp;	/* pointer to btree keys */
-	xfs_agblock_t		nbno;	/* block number of allocated block */
-	xfs_btree_cur_t		*ncur;	/* new cursor to be used at next lvl */
-	xfs_alloc_key_t		nkey;	/* new key value, from split */
-	xfs_alloc_rec_t		nrec;	/* new record value, for caller */
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
+	xfs_agnumber_t		seqno = be32_to_cpu(agf->agf_seqno);
+	__be32			len;
 	int			numrecs;
-	int			optr;	/* old ptr value */
-	xfs_alloc_ptr_t		*pp;	/* pointer to btree addresses */
-	int			ptr;	/* index in btree block for this rec */
-	xfs_alloc_rec_t		*rp;	/* pointer to btree records */
 
-	ASSERT(be32_to_cpu(recp->ar_blockcount) > 0);
+	ASSERT(cur->bc_btnum == XFS_BTNUM_CNT);
+
+	switch (reason) {
+	case LASTREC_UPDATE:
+		/*
+		 * If this is the last leaf block and it's the last record,
+		 * then update the size of the longest extent in the AG.
+		 */
+		if (ptr != xfs_btree_get_numrecs(block))
+			return;
+		len = rec->alloc.ar_blockcount;
+		break;
+	case LASTREC_INSREC:
+		if (be32_to_cpu(rec->alloc.ar_blockcount) <=
+		    be32_to_cpu(agf->agf_longest))
+			return;
+		len = rec->alloc.ar_blockcount;
+		break;
+	case LASTREC_DELREC:
+		numrecs = xfs_btree_get_numrecs(block);
+		if (ptr <= numrecs)
+			return;
+		ASSERT(ptr == numrecs + 1);
 
-	/*
-	 * GCC doesn't understand the (arguably complex) control flow in
-	 * this function and complains about uninitialized structure fields
-	 * without this.
-	 */
-	memset(&nrec, 0, sizeof(nrec));
+		if (numrecs) {
+			xfs_alloc_rec_t *rrp;
 
-	/*
-	 * If we made it to the root level, allocate a new root block
-	 * and we're done.
-	 */
-	if (level >= cur->bc_nlevels) {
-		XFS_STATS_INC(xs_abt_insrec);
-		if ((error = xfs_alloc_newroot(cur, &i)))
-			return error;
-		*bnop = NULLAGBLOCK;
-		*stat = i;
-		return 0;
-	}
-	/*
-	 * Make a key out of the record data to be inserted, and save it.
-	 */
-	key.ar_startblock = recp->ar_startblock;
-	key.ar_blockcount = recp->ar_blockcount;
-	optr = ptr = cur->bc_ptrs[level];
-	/*
-	 * If we're off the left edge, return failure.
-	 */
-	if (ptr == 0) {
-		*stat = 0;
-		return 0;
-	}
-	XFS_STATS_INC(xs_abt_insrec);
-	/*
-	 * Get pointers to the btree buffer and block.
-	 */
-	bp = cur->bc_bufs[level];
-	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-	numrecs = be16_to_cpu(block->bb_numrecs);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-		return error;
-	/*
-	 * Check that the new entry is being inserted in the right place.
-	 */
-	if (ptr <= numrecs) {
-		if (level == 0) {
-			rp = XFS_ALLOC_REC_ADDR(block, ptr, cur);
-			xfs_btree_check_rec(cur->bc_btnum, recp, rp);
+			rrp = XFS_ALLOC_REC_ADDR(cur->bc_mp, block, numrecs);
+			len = rrp->ar_blockcount;
 		} else {
-			kp = XFS_ALLOC_KEY_ADDR(block, ptr, cur);
-			xfs_btree_check_key(cur->bc_btnum, &key, kp);
-		}
-	}
-#endif
-	nbno = NULLAGBLOCK;
-	ncur = NULL;
-	/*
-	 * If the block is full, we can't insert the new entry until we
-	 * make the block un-full.
-	 */
-	if (numrecs == XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
-		/*
-		 * First, try shifting an entry to the right neighbor.
-		 */
-		if ((error = xfs_alloc_rshift(cur, level, &i)))
-			return error;
-		if (i) {
-			/* nothing */
-		}
-		/*
-		 * Next, try shifting an entry to the left neighbor.
-		 */
-		else {
-			if ((error = xfs_alloc_lshift(cur, level, &i)))
-				return error;
-			if (i)
-				optr = ptr = cur->bc_ptrs[level];
-			else {
-				/*
-				 * Next, try splitting the current block in
-				 * half. If this works we have to re-set our
-				 * variables because we could be in a
-				 * different block now.
-				 */
-				if ((error = xfs_alloc_split(cur, level, &nbno,
-						&nkey, &ncur, &i)))
-					return error;
-				if (i) {
-					bp = cur->bc_bufs[level];
-					block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-#ifdef DEBUG
-					if ((error =
-						xfs_btree_check_sblock(cur,
-							block, level, bp)))
-						return error;
-#endif
-					ptr = cur->bc_ptrs[level];
-					nrec.ar_startblock = nkey.ar_startblock;
-					nrec.ar_blockcount = nkey.ar_blockcount;
-				}
-				/*
-				 * Otherwise the insert fails.
-				 */
-				else {
-					*stat = 0;
-					return 0;
-				}
-			}
-		}
-	}
-	/*
-	 * At this point we know there's room for our new entry in the block
-	 * we're pointing at.
-	 */
-	numrecs = be16_to_cpu(block->bb_numrecs);
-	if (level > 0) {
-		/*
-		 * It's a non-leaf entry.  Make a hole for the new data
-		 * in the key and ptr regions of the block.
-		 */
-		kp = XFS_ALLOC_KEY_ADDR(block, 1, cur);
-		pp = XFS_ALLOC_PTR_ADDR(block, 1, cur);
-#ifdef DEBUG
-		for (i = numrecs; i >= ptr; i--) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(pp[i - 1]), level)))
-				return error;
+			len = 0;
 		}
-#endif
-		memmove(&kp[ptr], &kp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*kp));
-		memmove(&pp[ptr], &pp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*pp));
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sptr(cur, *bnop, level)))
-			return error;
-#endif
-		/*
-		 * Now stuff the new data in, bump numrecs and log the new data.
-		 */
-		kp[ptr - 1] = key;
-		pp[ptr - 1] = cpu_to_be32(*bnop);
-		numrecs++;
-		block->bb_numrecs = cpu_to_be16(numrecs);
-		xfs_alloc_log_keys(cur, bp, ptr, numrecs);
-		xfs_alloc_log_ptrs(cur, bp, ptr, numrecs);
-#ifdef DEBUG
-		if (ptr < numrecs)
-			xfs_btree_check_key(cur->bc_btnum, kp + ptr - 1,
-				kp + ptr);
-#endif
-	} else {
-		/*
-		 * It's a leaf entry.  Make a hole for the new record.
-		 */
-		rp = XFS_ALLOC_REC_ADDR(block, 1, cur);
-		memmove(&rp[ptr], &rp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*rp));
-		/*
-		 * Now stuff the new record in, bump numrecs
-		 * and log the new data.
-		 */
-		rp[ptr - 1] = *recp;
-		numrecs++;
-		block->bb_numrecs = cpu_to_be16(numrecs);
-		xfs_alloc_log_recs(cur, bp, ptr, numrecs);
-#ifdef DEBUG
-		if (ptr < numrecs)
-			xfs_btree_check_rec(cur->bc_btnum, rp + ptr - 1,
-				rp + ptr);
-#endif
-	}
-	/*
-	 * Log the new number of records in the btree header.
-	 */
-	xfs_alloc_log_block(cur->bc_tp, bp, XFS_BB_NUMRECS);
-	/*
-	 * If we inserted at the start of a block, update the parents' keys.
-	 */
-	if (optr == 1 && (error = xfs_alloc_updkey(cur, &key, level + 1)))
-		return error;
-	/*
-	 * Look to see if the longest extent in the allocation group
-	 * needs to be updated.
-	 */
 
-	agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-	if (level == 0 &&
-	    cur->bc_btnum == XFS_BTNUM_CNT &&
-	    be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK &&
-	    be32_to_cpu(recp->ar_blockcount) > be32_to_cpu(agf->agf_longest)) {
-		/*
-		 * If this is a leaf in the by-size btree and there
-		 * is no right sibling block and this block is bigger
-		 * than the previous longest block, update it.
-		 */
-		agf->agf_longest = recp->ar_blockcount;
-		cur->bc_mp->m_perag[be32_to_cpu(agf->agf_seqno)].pagf_longest
-			= be32_to_cpu(recp->ar_blockcount);
-		xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp,
-			XFS_AGF_LONGEST);
+		break;
+	default:
+		ASSERT(0);
+		return;
 	}
-	/*
-	 * Return the new block number, if any.
-	 * If there is one, give back a record value and a cursor too.
-	 */
-	*bnop = nbno;
-	if (nbno != NULLAGBLOCK) {
-		*recp = nrec;
-		*curp = ncur;
-	}
-	*stat = 1;
-	return 0;
+
+	agf->agf_longest = len;
+	cur->bc_mp->m_perag[seqno].pagf_longest = be32_to_cpu(len);
+	xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp, XFS_AGF_LONGEST);
 }
 
-/*
- * Log header fields from a btree block.
- */
-STATIC void
-xfs_alloc_log_block(
-	xfs_trans_t		*tp,	/* transaction pointer */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			fields)	/* mask of fields: XFS_BB_... */
+STATIC int
+xfs_allocbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
 {
-	int			first;	/* first byte offset logged */
-	int			last;	/* last byte offset logged */
-	static const short	offsets[] = {	/* table of offsets */
-		offsetof(xfs_alloc_block_t, bb_magic),
-		offsetof(xfs_alloc_block_t, bb_level),
-		offsetof(xfs_alloc_block_t, bb_numrecs),
-		offsetof(xfs_alloc_block_t, bb_leftsib),
-		offsetof(xfs_alloc_block_t, bb_rightsib),
-		sizeof(xfs_alloc_block_t)
-	};
+	return cur->bc_mp->m_alloc_mnr[level != 0];
+}
 
-	xfs_btree_offsets(fields, offsets, XFS_BB_NUM_BITS, &first, &last);
-	xfs_trans_log_buf(tp, bp, first, last);
+STATIC int
+xfs_allocbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	return cur->bc_mp->m_alloc_mxr[level != 0];
 }
 
-/*
- * Log keys from a btree block (nonleaf).
- */
 STATIC void
-xfs_alloc_log_keys(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			kfirst,	/* index of first key to log */
-	int			klast)	/* index of last key to log */
+xfs_allocbt_init_key_from_rec(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
 {
-	xfs_alloc_block_t	*block;	/* btree block to log from */
-	int			first;	/* first byte offset logged */
-	xfs_alloc_key_t		*kp;	/* key pointer in btree block */
-	int			last;	/* last byte offset logged */
+	ASSERT(rec->alloc.ar_startblock != 0);
 
-	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-	kp = XFS_ALLOC_KEY_ADDR(block, 1, cur);
-	first = (int)((xfs_caddr_t)&kp[kfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&kp[klast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	key->alloc.ar_startblock = rec->alloc.ar_startblock;
+	key->alloc.ar_blockcount = rec->alloc.ar_blockcount;
 }
 
-/*
- * Log block pointer fields from a btree block (nonleaf).
- */
 STATIC void
-xfs_alloc_log_ptrs(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			pfirst,	/* index of first pointer to log */
-	int			plast)	/* index of last pointer to log */
+xfs_allocbt_init_rec_from_key(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
 {
-	xfs_alloc_block_t	*block;	/* btree block to log from */
-	int			first;	/* first byte offset logged */
-	int			last;	/* last byte offset logged */
-	xfs_alloc_ptr_t		*pp;	/* block-pointer pointer in btree blk */
+	ASSERT(key->alloc.ar_startblock != 0);
 
-	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-	pp = XFS_ALLOC_PTR_ADDR(block, 1, cur);
-	first = (int)((xfs_caddr_t)&pp[pfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&pp[plast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	rec->alloc.ar_startblock = key->alloc.ar_startblock;
+	rec->alloc.ar_blockcount = key->alloc.ar_blockcount;
 }
 
-/*
- * Log records from a btree block (leaf).
- */
 STATIC void
-xfs_alloc_log_recs(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			rfirst,	/* index of first record to log */
-	int			rlast)	/* index of last record to log */
+xfs_allocbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
 {
-	xfs_alloc_block_t	*block;	/* btree block to log from */
-	int			first;	/* first byte offset logged */
-	int			last;	/* last byte offset logged */
-	xfs_alloc_rec_t		*rp;	/* record pointer for btree block */
-
+	ASSERT(cur->bc_rec.a.ar_startblock != 0);
 
-	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-	rp = XFS_ALLOC_REC_ADDR(block, 1, cur);
-#ifdef DEBUG
-	{
-		xfs_agf_t	*agf;
-		xfs_alloc_rec_t	*p;
-
-		agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-		for (p = &rp[rfirst - 1]; p <= &rp[rlast - 1]; p++)
-			ASSERT(be32_to_cpu(p->ar_startblock) +
-			       be32_to_cpu(p->ar_blockcount) <=
-			       be32_to_cpu(agf->agf_length));
-	}
-#endif
-	first = (int)((xfs_caddr_t)&rp[rfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&rp[rlast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	rec->alloc.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
+	rec->alloc.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
 }
 
-/*
- * Lookup the record.  The cursor is made to point to it, based on dir.
- * Return 0 if can't find any such record, 1 for success.
- */
-STATIC int				/* error */
-xfs_alloc_lookup(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_lookup_t		dir,	/* <=, ==, or >= */
-	int			*stat)	/* success/failure */
+STATIC void
+xfs_allocbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
 {
-	xfs_agblock_t		agbno;	/* a.g. relative btree block number */
-	xfs_agnumber_t		agno;	/* allocation group number */
-	xfs_alloc_block_t	*block=NULL;	/* current btree block */
-	int			diff;	/* difference for the current key */
-	int			error;	/* error return value */
-	int			keyno=0;	/* current key number */
-	int			level;	/* level in the btree */
-	xfs_mount_t		*mp;	/* file system mount point */
-
-	XFS_STATS_INC(xs_abt_lookup);
-	/*
-	 * Get the allocation group header, and the root block number.
-	 */
-	mp = cur->bc_mp;
-
-	{
-		xfs_agf_t	*agf;	/* a.g. freespace header */
-
-		agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-		agno = be32_to_cpu(agf->agf_seqno);
-		agbno = be32_to_cpu(agf->agf_roots[cur->bc_btnum]);
-	}
-	/*
-	 * Iterate over each level in the btree, starting at the root.
-	 * For each level above the leaves, find the key we need, based
-	 * on the lookup record, then follow the corresponding block
-	 * pointer down to the next level.
-	 */
-	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
-		xfs_buf_t	*bp;	/* buffer pointer for btree block */
-		xfs_daddr_t	d;	/* disk address of btree block */
-
-		/*
-		 * Get the disk address we're looking for.
-		 */
-		d = XFS_AGB_TO_DADDR(mp, agno, agbno);
-		/*
-		 * If the old buffer at this level is for a different block,
-		 * throw it away, otherwise just use it.
-		 */
-		bp = cur->bc_bufs[level];
-		if (bp && XFS_BUF_ADDR(bp) != d)
-			bp = NULL;
-		if (!bp) {
-			/*
-			 * Need to get a new buffer.  Read it, then
-			 * set it in the cursor, releasing the old one.
-			 */
-			if ((error = xfs_btree_read_bufs(mp, cur->bc_tp, agno,
-					agbno, 0, &bp, XFS_ALLOC_BTREE_REF)))
-				return error;
-			xfs_btree_setbuf(cur, level, bp);
-			/*
-			 * Point to the btree block, now that we have the buffer
-			 */
-			block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-			if ((error = xfs_btree_check_sblock(cur, block, level,
-					bp)))
-				return error;
-		} else
-			block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-		/*
-		 * If we already had a key match at a higher level, we know
-		 * we need to use the first entry in this block.
-		 */
-		if (diff == 0)
-			keyno = 1;
-		/*
-		 * Otherwise we need to search this block.  Do a binary search.
-		 */
-		else {
-			int		high;	/* high entry number */
-			xfs_alloc_key_t	*kkbase=NULL;/* base of keys in block */
-			xfs_alloc_rec_t	*krbase=NULL;/* base of records in block */
-			int		low;	/* low entry number */
-
-			/*
-			 * Get a pointer to keys or records.
-			 */
-			if (level > 0)
-				kkbase = XFS_ALLOC_KEY_ADDR(block, 1, cur);
-			else
-				krbase = XFS_ALLOC_REC_ADDR(block, 1, cur);
-			/*
-			 * Set low and high entry numbers, 1-based.
-			 */
-			low = 1;
-			if (!(high = be16_to_cpu(block->bb_numrecs))) {
-				/*
-				 * If the block is empty, the tree must
-				 * be an empty leaf.
-				 */
-				ASSERT(level == 0 && cur->bc_nlevels == 1);
-				cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
-				*stat = 0;
-				return 0;
-			}
-			/*
-			 * Binary search the block.
-			 */
-			while (low <= high) {
-				xfs_extlen_t	blockcount;	/* key value */
-				xfs_agblock_t	startblock;	/* key value */
-
-				XFS_STATS_INC(xs_abt_compare);
-				/*
-				 * keyno is average of low and high.
-				 */
-				keyno = (low + high) >> 1;
-				/*
-				 * Get startblock & blockcount.
-				 */
-				if (level > 0) {
-					xfs_alloc_key_t	*kkp;
-
-					kkp = kkbase + keyno - 1;
-					startblock = be32_to_cpu(kkp->ar_startblock);
-					blockcount = be32_to_cpu(kkp->ar_blockcount);
-				} else {
-					xfs_alloc_rec_t	*krp;
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
 
-					krp = krbase + keyno - 1;
-					startblock = be32_to_cpu(krp->ar_startblock);
-					blockcount = be32_to_cpu(krp->ar_blockcount);
-				}
-				/*
-				 * Compute difference to get next direction.
-				 */
-				if (cur->bc_btnum == XFS_BTNUM_BNO)
-					diff = (int)startblock -
-					       (int)cur->bc_rec.a.ar_startblock;
-				else if (!(diff = (int)blockcount -
-					    (int)cur->bc_rec.a.ar_blockcount))
-					diff = (int)startblock -
-					    (int)cur->bc_rec.a.ar_startblock;
-				/*
-				 * Less than, move right.
-				 */
-				if (diff < 0)
-					low = keyno + 1;
-				/*
-				 * Greater than, move left.
-				 */
-				else if (diff > 0)
-					high = keyno - 1;
-				/*
-				 * Equal, we're done.
-				 */
-				else
-					break;
-			}
-		}
-		/*
-		 * If there are more levels, set up for the next level
-		 * by getting the block number and filling in the cursor.
-		 */
-		if (level > 0) {
-			/*
-			 * If we moved left, need the previous key number,
-			 * unless there isn't one.
-			 */
-			if (diff > 0 && --keyno < 1)
-				keyno = 1;
-			agbno = be32_to_cpu(*XFS_ALLOC_PTR_ADDR(block, keyno, cur));
-#ifdef DEBUG
-			if ((error = xfs_btree_check_sptr(cur, agbno, level)))
-				return error;
-#endif
-			cur->bc_ptrs[level] = keyno;
-		}
-	}
-	/*
-	 * Done with the search.
-	 * See if we need to adjust the results.
-	 */
-	if (dir != XFS_LOOKUP_LE && diff < 0) {
-		keyno++;
-		/*
-		 * If ge search and we went off the end of the block, but it's
-		 * not the last block, we're in the wrong block.
-		 */
-		if (dir == XFS_LOOKUP_GE &&
-		    keyno > be16_to_cpu(block->bb_numrecs) &&
-		    be32_to_cpu(block->bb_rightsib) != NULLAGBLOCK) {
-			int	i;
+	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agf->agf_seqno));
+	ASSERT(agf->agf_roots[cur->bc_btnum] != 0);
 
-			cur->bc_ptrs[0] = keyno;
-			if ((error = xfs_alloc_increment(cur, 0, &i)))
-				return error;
-			XFS_WANT_CORRUPTED_RETURN(i == 1);
-			*stat = 1;
-			return 0;
-		}
-	}
-	else if (dir == XFS_LOOKUP_LE && diff > 0)
-		keyno--;
-	cur->bc_ptrs[0] = keyno;
-	/*
-	 * Return if we succeeded or not.
-	 */
-	if (keyno == 0 || keyno > be16_to_cpu(block->bb_numrecs))
-		*stat = 0;
-	else
-		*stat = ((dir != XFS_LOOKUP_EQ) || (diff == 0));
-	return 0;
+	ptr->s = agf->agf_roots[cur->bc_btnum];
 }
 
-/*
- * Move 1 record left from cur/level if possible.
- * Update cur to reflect the new path.
- */
-STATIC int				/* error */
-xfs_alloc_lshift(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to shift record on */
-	int			*stat)	/* success/failure */
+STATIC __int64_t
+xfs_allocbt_key_diff(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
 {
-	int			error;	/* error return value */
-#ifdef DEBUG
-	int			i;	/* loop index */
-#endif
-	xfs_alloc_key_t		key;	/* key value for leaf level upward */
-	xfs_buf_t		*lbp;	/* buffer for left neighbor block */
-	xfs_alloc_block_t	*left;	/* left neighbor btree block */
-	int			nrec;	/* new number of left block entries */
-	xfs_buf_t		*rbp;	/* buffer for right (current) block */
-	xfs_alloc_block_t	*right;	/* right (current) btree block */
-	xfs_alloc_key_t		*rkp=NULL;	/* key pointer for right block */
-	xfs_alloc_ptr_t		*rpp=NULL;	/* address pointer for right block */
-	xfs_alloc_rec_t		*rrp=NULL;	/* record pointer for right block */
+	xfs_alloc_rec_incore_t	*rec = &cur->bc_rec.a;
+	xfs_alloc_key_t		*kp = &key->alloc;
+	__int64_t		diff;
 
-	/*
-	 * Set up variables for this block as "right".
-	 */
-	rbp = cur->bc_bufs[level];
-	right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-		return error;
-#endif
-	/*
-	 * If we've got no left sibling then we can't shift an entry left.
-	 */
-	if (be32_to_cpu(right->bb_leftsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * If the cursor entry is the one that would be moved, don't
-	 * do it... it's too complicated.
-	 */
-	if (cur->bc_ptrs[level] <= 1) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Set up the left neighbor as "left".
-	 */
-	if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-			cur->bc_private.a.agno, be32_to_cpu(right->bb_leftsib),
-			0, &lbp, XFS_ALLOC_BTREE_REF)))
-		return error;
-	left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-	if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-		return error;
-	/*
-	 * If it's full, it can't take another entry.
-	 */
-	if (be16_to_cpu(left->bb_numrecs) == XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
-		*stat = 0;
-		return 0;
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		return (__int64_t)be32_to_cpu(kp->ar_startblock) -
+				rec->ar_startblock;
 	}
-	nrec = be16_to_cpu(left->bb_numrecs) + 1;
-	/*
-	 * If non-leaf, copy a key and a ptr to the left block.
-	 */
-	if (level > 0) {
-		xfs_alloc_key_t	*lkp;	/* key pointer for left block */
-		xfs_alloc_ptr_t	*lpp;	/* address pointer for left block */
 
-		lkp = XFS_ALLOC_KEY_ADDR(left, nrec, cur);
-		rkp = XFS_ALLOC_KEY_ADDR(right, 1, cur);
-		*lkp = *rkp;
-		xfs_alloc_log_keys(cur, lbp, nrec, nrec);
-		lpp = XFS_ALLOC_PTR_ADDR(left, nrec, cur);
-		rpp = XFS_ALLOC_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*rpp), level)))
-			return error;
-#endif
-		*lpp = *rpp;
-		xfs_alloc_log_ptrs(cur, lbp, nrec, nrec);
-		xfs_btree_check_key(cur->bc_btnum, lkp - 1, lkp);
-	}
-	/*
-	 * If leaf, copy a record to the left block.
-	 */
-	else {
-		xfs_alloc_rec_t	*lrp;	/* record pointer for left block */
+	diff = (__int64_t)be32_to_cpu(kp->ar_blockcount) - rec->ar_blockcount;
+	if (diff)
+		return diff;
 
-		lrp = XFS_ALLOC_REC_ADDR(left, nrec, cur);
-		rrp = XFS_ALLOC_REC_ADDR(right, 1, cur);
-		*lrp = *rrp;
-		xfs_alloc_log_recs(cur, lbp, nrec, nrec);
-		xfs_btree_check_rec(cur->bc_btnum, lrp - 1, lrp);
-	}
-	/*
-	 * Bump and log left's numrecs, decrement and log right's numrecs.
-	 */
-	be16_add_cpu(&left->bb_numrecs, 1);
-	xfs_alloc_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS);
-	be16_add_cpu(&right->bb_numrecs, -1);
-	xfs_alloc_log_block(cur->bc_tp, rbp, XFS_BB_NUMRECS);
-	/*
-	 * Slide the contents of right down one entry.
-	 */
-	if (level > 0) {
-#ifdef DEBUG
-		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i + 1]),
-					level)))
-				return error;
-		}
-#endif
-		memmove(rkp, rkp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memmove(rpp, rpp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-		xfs_alloc_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		xfs_alloc_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-	} else {
-		memmove(rrp, rrp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		xfs_alloc_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		key.ar_startblock = rrp->ar_startblock;
-		key.ar_blockcount = rrp->ar_blockcount;
-		rkp = &key;
-	}
-	/*
-	 * Update the parent key values of right.
-	 */
-	if ((error = xfs_alloc_updkey(cur, rkp, level + 1)))
-		return error;
-	/*
-	 * Slide the cursor value left one.
-	 */
-	cur->bc_ptrs[level]--;
-	*stat = 1;
-	return 0;
+	return (__int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
 }
 
-/*
- * Allocate a new root block, fill it in.
- */
-STATIC int				/* error */
-xfs_alloc_newroot(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			*stat)	/* success/failure */
+STATIC int
+xfs_allocbt_kill_root(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			level,
+	union xfs_btree_ptr	*newroot)
 {
-	int			error;	/* error return value */
-	xfs_agblock_t		lbno;	/* left block number */
-	xfs_buf_t		*lbp;	/* left btree buffer */
-	xfs_alloc_block_t	*left;	/* left btree block */
-	xfs_mount_t		*mp;	/* mount structure */
-	xfs_agblock_t		nbno;	/* new block number */
-	xfs_buf_t		*nbp;	/* new (root) buffer */
-	xfs_alloc_block_t	*new;	/* new (root) btree block */
-	int			nptr;	/* new value for key index, 1 or 2 */
-	xfs_agblock_t		rbno;	/* right block number */
-	xfs_buf_t		*rbp;	/* right btree buffer */
-	xfs_alloc_block_t	*right;	/* right btree block */
-
-	mp = cur->bc_mp;
+	int			error;
 
-	ASSERT(cur->bc_nlevels < XFS_AG_MAXLEVELS(mp));
-	/*
-	 * Get a buffer from the freelist blocks, for the new root.
-	 */
-	error = xfs_alloc_get_freelist(cur->bc_tp,
-					cur->bc_private.a.agbp, &nbno, 1);
-	if (error)
-		return error;
-	/*
-	 * None available, we fail.
-	 */
-	if (nbno == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	xfs_trans_agbtree_delta(cur->bc_tp, 1);
-	nbp = xfs_btree_get_bufs(mp, cur->bc_tp, cur->bc_private.a.agno, nbno,
-		0);
-	new = XFS_BUF_TO_ALLOC_BLOCK(nbp);
-	/*
-	 * Set the root data in the a.g. freespace structure.
-	 */
-	{
-		xfs_agf_t	*agf;	/* a.g. freespace header */
-		xfs_agnumber_t	seqno;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, killroot);
 
-		agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-		agf->agf_roots[cur->bc_btnum] = cpu_to_be32(nbno);
-		be32_add_cpu(&agf->agf_levels[cur->bc_btnum], 1);
-		seqno = be32_to_cpu(agf->agf_seqno);
-		mp->m_perag[seqno].pagf_levels[cur->bc_btnum]++;
-		xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp,
-			XFS_AGF_ROOTS | XFS_AGF_LEVELS);
-	}
 	/*
-	 * At the previous root level there are now two blocks: the old
-	 * root, and the new block generated when it was split.
-	 * We don't know which one the cursor is pointing at, so we
-	 * set up variables "left" and "right" for each case.
+	 * Update the root pointer, decreasing the level by 1 and then
+	 * free the old root.
 	 */
-	lbp = cur->bc_bufs[cur->bc_nlevels - 1];
-	left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, left, cur->bc_nlevels - 1, lbp)))
+	xfs_allocbt_set_root(cur, newroot, -1);
+	error = xfs_allocbt_free_block(cur, bp);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
 		return error;
-#endif
-	if (be32_to_cpu(left->bb_rightsib) != NULLAGBLOCK) {
-		/*
-		 * Our block is left, pick up the right block.
-		 */
-		lbno = XFS_DADDR_TO_AGBNO(mp, XFS_BUF_ADDR(lbp));
-		rbno = be32_to_cpu(left->bb_rightsib);
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, rbno, 0, &rbp,
-				XFS_ALLOC_BTREE_REF)))
-			return error;
-		right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
-		if ((error = xfs_btree_check_sblock(cur, right,
-				cur->bc_nlevels - 1, rbp)))
-			return error;
-		nptr = 1;
-	} else {
-		/*
-		 * Our block is right, pick up the left block.
-		 */
-		rbp = lbp;
-		right = left;
-		rbno = XFS_DADDR_TO_AGBNO(mp, XFS_BUF_ADDR(rbp));
-		lbno = be32_to_cpu(right->bb_leftsib);
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, lbno, 0, &lbp,
-				XFS_ALLOC_BTREE_REF)))
-			return error;
-		left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-		if ((error = xfs_btree_check_sblock(cur, left,
-				cur->bc_nlevels - 1, lbp)))
-			return error;
-		nptr = 2;
 	}
-	/*
-	 * Fill in the new block's btree header and log it.
-	 */
-	new->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
-	new->bb_level = cpu_to_be16(cur->bc_nlevels);
-	new->bb_numrecs = cpu_to_be16(2);
-	new->bb_leftsib = cpu_to_be32(NULLAGBLOCK);
-	new->bb_rightsib = cpu_to_be32(NULLAGBLOCK);
-	xfs_alloc_log_block(cur->bc_tp, nbp, XFS_BB_ALL_BITS);
-	ASSERT(lbno != NULLAGBLOCK && rbno != NULLAGBLOCK);
-	/*
-	 * Fill in the key data in the new root.
-	 */
-	{
-		xfs_alloc_key_t		*kp;	/* btree key pointer */
 
-		kp = XFS_ALLOC_KEY_ADDR(new, 1, cur);
-		if (be16_to_cpu(left->bb_level) > 0) {
-			kp[0] = *XFS_ALLOC_KEY_ADDR(left, 1, cur);
-			kp[1] = *XFS_ALLOC_KEY_ADDR(right, 1, cur);
-		} else {
-			xfs_alloc_rec_t	*rp;	/* btree record pointer */
+	XFS_BTREE_STATS_INC(cur, free);
 
-			rp = XFS_ALLOC_REC_ADDR(left, 1, cur);
-			kp[0].ar_startblock = rp->ar_startblock;
-			kp[0].ar_blockcount = rp->ar_blockcount;
-			rp = XFS_ALLOC_REC_ADDR(right, 1, cur);
-			kp[1].ar_startblock = rp->ar_startblock;
-			kp[1].ar_blockcount = rp->ar_blockcount;
-		}
-	}
-	xfs_alloc_log_keys(cur, nbp, 1, 2);
-	/*
-	 * Fill in the pointer data in the new root.
-	 */
-	{
-		xfs_alloc_ptr_t		*pp;	/* btree address pointer */
+	xfs_btree_setbuf(cur, level, NULL);
+	cur->bc_nlevels--;
 
-		pp = XFS_ALLOC_PTR_ADDR(new, 1, cur);
-		pp[0] = cpu_to_be32(lbno);
-		pp[1] = cpu_to_be32(rbno);
-	}
-	xfs_alloc_log_ptrs(cur, nbp, 1, 2);
-	/*
-	 * Fix up the cursor.
-	 */
-	xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
-	cur->bc_ptrs[cur->bc_nlevels] = nptr;
-	cur->bc_nlevels++;
-	*stat = 1;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
 	return 0;
 }
 
-/*
- * Move 1 record right from cur/level if possible.
- * Update cur to reflect the new path.
- */
-STATIC int				/* error */
-xfs_alloc_rshift(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to shift record on */
-	int			*stat)	/* success/failure */
-{
-	int			error;	/* error return value */
-	int			i;	/* loop index */
-	xfs_alloc_key_t		key;	/* key value for leaf level upward */
-	xfs_buf_t		*lbp;	/* buffer for left (current) block */
-	xfs_alloc_block_t	*left;	/* left (current) btree block */
-	xfs_buf_t		*rbp;	/* buffer for right neighbor block */
-	xfs_alloc_block_t	*right;	/* right neighbor btree block */
-	xfs_alloc_key_t		*rkp;	/* key pointer for right block */
-	xfs_btree_cur_t		*tcur;	/* temporary cursor */
-
-	/*
-	 * Set up variables for this block as "left".
-	 */
-	lbp = cur->bc_bufs[level];
-	left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-		return error;
-#endif
-	/*
-	 * If we've got no right sibling then we can't shift an entry right.
-	 */
-	if (be32_to_cpu(left->bb_rightsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * If the cursor entry is the one that would be moved, don't
-	 * do it... it's too complicated.
-	 */
-	if (cur->bc_ptrs[level] >= be16_to_cpu(left->bb_numrecs)) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Set up the right neighbor as "right".
-	 */
-	if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-			cur->bc_private.a.agno, be32_to_cpu(left->bb_rightsib),
-			0, &rbp, XFS_ALLOC_BTREE_REF)))
-		return error;
-	right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
-	if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-		return error;
-	/*
-	 * If it's full, it can't take another entry.
-	 */
-	if (be16_to_cpu(right->bb_numrecs) == XFS_ALLOC_BLOCK_MAXRECS(level, cur)) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Make a hole at the start of the right neighbor block, then
-	 * copy the last left block entry to the hole.
-	 */
-	if (level > 0) {
-		xfs_alloc_key_t	*lkp;	/* key pointer for left block */
-		xfs_alloc_ptr_t	*lpp;	/* address pointer for left block */
-		xfs_alloc_ptr_t	*rpp;	/* address pointer for right block */
-
-		lkp = XFS_ALLOC_KEY_ADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		lpp = XFS_ALLOC_PTR_ADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		rkp = XFS_ALLOC_KEY_ADDR(right, 1, cur);
-		rpp = XFS_ALLOC_PTR_ADDR(right, 1, cur);
 #ifdef DEBUG
-		for (i = be16_to_cpu(right->bb_numrecs) - 1; i >= 0; i--) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i]), level)))
-				return error;
-		}
-#endif
-		memmove(rkp + 1, rkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memmove(rpp + 1, rpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*lpp), level)))
-			return error;
-#endif
-		*rkp = *lkp;
-		*rpp = *lpp;
-		xfs_alloc_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		xfs_alloc_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		xfs_btree_check_key(cur->bc_btnum, rkp, rkp + 1);
+STATIC int
+xfs_allocbt_keys_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*k1,
+	union xfs_btree_key	*k2)
+{
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		return be32_to_cpu(k1->alloc.ar_startblock) <
+		       be32_to_cpu(k2->alloc.ar_startblock);
 	} else {
-		xfs_alloc_rec_t	*lrp;	/* record pointer for left block */
-		xfs_alloc_rec_t	*rrp;	/* record pointer for right block */
-
-		lrp = XFS_ALLOC_REC_ADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		rrp = XFS_ALLOC_REC_ADDR(right, 1, cur);
-		memmove(rrp + 1, rrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		*rrp = *lrp;
-		xfs_alloc_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		key.ar_startblock = rrp->ar_startblock;
-		key.ar_blockcount = rrp->ar_blockcount;
-		rkp = &key;
-		xfs_btree_check_rec(cur->bc_btnum, rrp, rrp + 1);
+		return be32_to_cpu(k1->alloc.ar_blockcount) <
+			be32_to_cpu(k2->alloc.ar_blockcount) ||
+			(k1->alloc.ar_blockcount == k2->alloc.ar_blockcount &&
+			 be32_to_cpu(k1->alloc.ar_startblock) <
+			 be32_to_cpu(k2->alloc.ar_startblock));
 	}
-	/*
-	 * Decrement and log left's numrecs, bump and log right's numrecs.
-	 */
-	be16_add_cpu(&left->bb_numrecs, -1);
-	xfs_alloc_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS);
-	be16_add_cpu(&right->bb_numrecs, 1);
-	xfs_alloc_log_block(cur->bc_tp, rbp, XFS_BB_NUMRECS);
-	/*
-	 * Using a temporary cursor, update the parent key values of the
-	 * block on the right.
-	 */
-	if ((error = xfs_btree_dup_cursor(cur, &tcur)))
-		return error;
-	i = xfs_btree_lastrec(tcur, level);
-	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-	if ((error = xfs_alloc_increment(tcur, level, &i)) ||
-	    (error = xfs_alloc_updkey(tcur, rkp, level + 1)))
-		goto error0;
-	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-	*stat = 1;
-	return 0;
-error0:
-	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
-	return error;
 }
 
-/*
- * Split cur/level block in half.
- * Return new block number and its first record (to be inserted into parent).
- */
-STATIC int				/* error */
-xfs_alloc_split(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to split */
-	xfs_agblock_t		*bnop,	/* output: block number allocated */
-	xfs_alloc_key_t		*keyp,	/* output: first key of new block */
-	xfs_btree_cur_t		**curp,	/* output: new cursor */
-	int			*stat)	/* success/failure */
+STATIC int
+xfs_allocbt_recs_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*r1,
+	union xfs_btree_rec	*r2)
 {
-	int			error;	/* error return value */
-	int			i;	/* loop index/record number */
-	xfs_agblock_t		lbno;	/* left (current) block number */
-	xfs_buf_t		*lbp;	/* buffer for left block */
-	xfs_alloc_block_t	*left;	/* left (current) btree block */
-	xfs_agblock_t		rbno;	/* right (new) block number */
-	xfs_buf_t		*rbp;	/* buffer for right block */
-	xfs_alloc_block_t	*right;	/* right (new) btree block */
-
-	/*
-	 * Allocate the new block from the freelist.
-	 * If we can't do it, we're toast.  Give up.
-	 */
-	error = xfs_alloc_get_freelist(cur->bc_tp,
-					 cur->bc_private.a.agbp, &rbno, 1);
-	if (error)
-		return error;
-	if (rbno == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	xfs_trans_agbtree_delta(cur->bc_tp, 1);
-	rbp = xfs_btree_get_bufs(cur->bc_mp, cur->bc_tp, cur->bc_private.a.agno,
-		rbno, 0);
-	/*
-	 * Set up the new block as "right".
-	 */
-	right = XFS_BUF_TO_ALLOC_BLOCK(rbp);
-	/*
-	 * "Left" is the current (according to the cursor) block.
-	 */
-	lbp = cur->bc_bufs[level];
-	left = XFS_BUF_TO_ALLOC_BLOCK(lbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-		return error;
-#endif
-	/*
-	 * Fill in the btree header for the new block.
-	 */
-	right->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
-	right->bb_level = left->bb_level;
-	right->bb_numrecs = cpu_to_be16(be16_to_cpu(left->bb_numrecs) / 2);
-	/*
-	 * Make sure that if there's an odd number of entries now, that
-	 * each new block will have the same number of entries.
-	 */
-	if ((be16_to_cpu(left->bb_numrecs) & 1) &&
-	    cur->bc_ptrs[level] <= be16_to_cpu(right->bb_numrecs) + 1)
-		be16_add_cpu(&right->bb_numrecs, 1);
-	i = be16_to_cpu(left->bb_numrecs) - be16_to_cpu(right->bb_numrecs) + 1;
-	/*
-	 * For non-leaf blocks, copy keys and addresses over to the new block.
-	 */
-	if (level > 0) {
-		xfs_alloc_key_t	*lkp;	/* left btree key pointer */
-		xfs_alloc_ptr_t	*lpp;	/* left btree address pointer */
-		xfs_alloc_key_t	*rkp;	/* right btree key pointer */
-		xfs_alloc_ptr_t	*rpp;	/* right btree address pointer */
-
-		lkp = XFS_ALLOC_KEY_ADDR(left, i, cur);
-		lpp = XFS_ALLOC_PTR_ADDR(left, i, cur);
-		rkp = XFS_ALLOC_KEY_ADDR(right, 1, cur);
-		rpp = XFS_ALLOC_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(lpp[i]), level)))
-				return error;
-		}
-#endif
-		memcpy(rkp, lkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memcpy(rpp, lpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-		xfs_alloc_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		xfs_alloc_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		*keyp = *rkp;
+	if (cur->bc_btnum == XFS_BTNUM_BNO) {
+		return be32_to_cpu(r1->alloc.ar_startblock) +
+			be32_to_cpu(r1->alloc.ar_blockcount) <=
+			be32_to_cpu(r2->alloc.ar_startblock);
+	} else {
+		return be32_to_cpu(r1->alloc.ar_blockcount) <
+			be32_to_cpu(r2->alloc.ar_blockcount) ||
+			(r1->alloc.ar_blockcount == r2->alloc.ar_blockcount &&
+			 be32_to_cpu(r1->alloc.ar_startblock) <
+			 be32_to_cpu(r2->alloc.ar_startblock));
 	}
-	/*
-	 * For leaf blocks, copy records over to the new block.
-	 */
-	else {
-		xfs_alloc_rec_t	*lrp;	/* left btree record pointer */
-		xfs_alloc_rec_t	*rrp;	/* right btree record pointer */
+}
+#endif	/* DEBUG */
 
-		lrp = XFS_ALLOC_REC_ADDR(left, i, cur);
-		rrp = XFS_ALLOC_REC_ADDR(right, 1, cur);
-		memcpy(rrp, lrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		xfs_alloc_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		keyp->ar_startblock = rrp->ar_startblock;
-		keyp->ar_blockcount = rrp->ar_blockcount;
-	}
-	/*
-	 * Find the left block number by looking in the buffer.
-	 * Adjust numrecs, sibling pointers.
-	 */
-	lbno = XFS_DADDR_TO_AGBNO(cur->bc_mp, XFS_BUF_ADDR(lbp));
-	be16_add_cpu(&left->bb_numrecs, -(be16_to_cpu(right->bb_numrecs)));
-	right->bb_rightsib = left->bb_rightsib;
-	left->bb_rightsib = cpu_to_be32(rbno);
-	right->bb_leftsib = cpu_to_be32(lbno);
-	xfs_alloc_log_block(cur->bc_tp, rbp, XFS_BB_ALL_BITS);
-	xfs_alloc_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
-	/*
-	 * If there's a block to the new block's right, make that block
-	 * point back to right instead of to left.
-	 */
-	if (be32_to_cpu(right->bb_rightsib) != NULLAGBLOCK) {
-		xfs_alloc_block_t	*rrblock;	/* rr btree block */
-		xfs_buf_t		*rrbp;		/* buffer for rrblock */
+#ifdef XFS_BTREE_TRACE
+ktrace_t	*xfs_allocbt_trace_buf;
 
-		if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-				cur->bc_private.a.agno, be32_to_cpu(right->bb_rightsib), 0,
-				&rrbp, XFS_ALLOC_BTREE_REF)))
-			return error;
-		rrblock = XFS_BUF_TO_ALLOC_BLOCK(rrbp);
-		if ((error = xfs_btree_check_sblock(cur, rrblock, level, rrbp)))
-			return error;
-		rrblock->bb_leftsib = cpu_to_be32(rbno);
-		xfs_alloc_log_block(cur->bc_tp, rrbp, XFS_BB_LEFTSIB);
-	}
-	/*
-	 * If the cursor is really in the right block, move it there.
-	 * If it's just pointing past the last entry in left, then we'll
-	 * insert there, so don't change anything in that case.
-	 */
-	if (cur->bc_ptrs[level] > be16_to_cpu(left->bb_numrecs) + 1) {
-		xfs_btree_setbuf(cur, level, rbp);
-		cur->bc_ptrs[level] -= be16_to_cpu(left->bb_numrecs);
-	}
-	/*
-	 * If there are more levels, we'll need another cursor which refers to
-	 * the right block, no matter where this cursor was.
-	 */
-	if (level + 1 < cur->bc_nlevels) {
-		if ((error = xfs_btree_dup_cursor(cur, curp)))
-			return error;
-		(*curp)->bc_ptrs[level + 1]++;
-	}
-	*bnop = rbno;
-	*stat = 1;
-	return 0;
+STATIC void
+xfs_allocbt_trace_enter(
+	struct xfs_btree_cur	*cur,
+	const char		*func,
+	char			*s,
+	int			type,
+	int			line,
+	__psunsigned_t		a0,
+	__psunsigned_t		a1,
+	__psunsigned_t		a2,
+	__psunsigned_t		a3,
+	__psunsigned_t		a4,
+	__psunsigned_t		a5,
+	__psunsigned_t		a6,
+	__psunsigned_t		a7,
+	__psunsigned_t		a8,
+	__psunsigned_t		a9,
+	__psunsigned_t		a10)
+{
+	ktrace_enter(xfs_allocbt_trace_buf, (void *)(__psint_t)type,
+		(void *)func, (void *)s, NULL, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
 }
 
-/*
- * Update keys at all levels from here to the root along the cursor's path.
- */
-STATIC int				/* error */
-xfs_alloc_updkey(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_alloc_key_t		*keyp,	/* new key value to update to */
-	int			level)	/* starting level for update */
+STATIC void
+xfs_allocbt_trace_cursor(
+	struct xfs_btree_cur	*cur,
+	__uint32_t		*s0,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
 {
-	int			ptr;	/* index of key in block */
-
-	/*
-	 * Go up the tree from this level toward the root.
-	 * At each level, update the key value to the value input.
-	 * Stop when we reach a level where the cursor isn't pointing
-	 * at the first entry in the block.
-	 */
-	for (ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
-		xfs_alloc_block_t	*block;	/* btree block */
-		xfs_buf_t		*bp;	/* buffer for block */
-#ifdef DEBUG
-		int			error;	/* error return value */
-#endif
-		xfs_alloc_key_t		*kp;	/* ptr to btree block keys */
-
-		bp = cur->bc_bufs[level];
-		block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-			return error;
-#endif
-		ptr = cur->bc_ptrs[level];
-		kp = XFS_ALLOC_KEY_ADDR(block, ptr, cur);
-		*kp = *keyp;
-		xfs_alloc_log_keys(cur, bp, ptr, ptr);
-	}
-	return 0;
+	*s0 = cur->bc_private.a.agno;
+	*l0 = cur->bc_rec.a.ar_startblock;
+	*l1 = cur->bc_rec.a.ar_blockcount;
 }
 
-/*
- * Externally visible routines.
- */
-
-/*
- * Decrement cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-int					/* error */
-xfs_alloc_decrement(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level in btree, 0 is leaf */
-	int			*stat)	/* success/failure */
+STATIC void
+xfs_allocbt_trace_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
 {
-	xfs_alloc_block_t	*block;	/* btree block */
-	int			error;	/* error return value */
-	int			lev;	/* btree level */
-
-	ASSERT(level < cur->bc_nlevels);
-	/*
-	 * Read-ahead to the left at this level.
-	 */
-	xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
-	/*
-	 * Decrement the ptr at this level.  If we're still in the block
-	 * then we're done.
-	 */
-	if (--cur->bc_ptrs[level] > 0) {
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * Get a pointer to the btree block.
-	 */
-	block = XFS_BUF_TO_ALLOC_BLOCK(cur->bc_bufs[level]);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level,
-			cur->bc_bufs[level])))
-		return error;
-#endif
-	/*
-	 * If we just went off the left edge of the tree, return failure.
-	 */
-	if (be32_to_cpu(block->bb_leftsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * March up the tree decrementing pointers.
-	 * Stop when we don't go off the left edge of a block.
-	 */
-	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
-		if (--cur->bc_ptrs[lev] > 0)
-			break;
-		/*
-		 * Read-ahead the left block, we're going to read it
-		 * in the next loop.
-		 */
-		xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
-	}
-	/*
-	 * If we went off the root then we are seriously confused.
-	 */
-	ASSERT(lev < cur->bc_nlevels);
-	/*
-	 * Now walk back down the tree, fixing up the cursor's buffer
-	 * pointers and key numbers.
-	 */
-	for (block = XFS_BUF_TO_ALLOC_BLOCK(cur->bc_bufs[lev]); lev > level; ) {
-		xfs_agblock_t	agbno;	/* block number of btree block */
-		xfs_buf_t	*bp;	/* buffer pointer for block */
-
-		agbno = be32_to_cpu(*XFS_ALLOC_PTR_ADDR(block, cur->bc_ptrs[lev], cur));
-		if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-				cur->bc_private.a.agno, agbno, 0, &bp,
-				XFS_ALLOC_BTREE_REF)))
-			return error;
-		lev--;
-		xfs_btree_setbuf(cur, lev, bp);
-		block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-		if ((error = xfs_btree_check_sblock(cur, block, lev, bp)))
-			return error;
-		cur->bc_ptrs[lev] = be16_to_cpu(block->bb_numrecs);
-	}
-	*stat = 1;
-	return 0;
+	*l0 = be32_to_cpu(key->alloc.ar_startblock);
+	*l1 = be32_to_cpu(key->alloc.ar_blockcount);
 }
 
-/*
- * Delete the record pointed to by cur.
- * The cursor refers to the place where the record was (could be inserted)
- * when the operation returns.
- */
-int					/* error */
-xfs_alloc_delete(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	int		*stat)		/* success/failure */
+STATIC void
+xfs_allocbt_trace_record(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	__uint64_t		*l0,
+	__uint64_t		*l1,
+	__uint64_t		*l2)
 {
-	int		error;		/* error return value */
-	int		i;		/* result code */
-	int		level;		/* btree level */
-
-	/*
-	 * Go up the tree, starting at leaf level.
-	 * If 2 is returned then a join was done; go to the next level.
-	 * Otherwise we are done.
-	 */
-	for (level = 0, i = 2; i == 2; level++) {
-		if ((error = xfs_alloc_delrec(cur, level, &i)))
-			return error;
-	}
-	if (i == 0) {
-		for (level = 1; level < cur->bc_nlevels; level++) {
-			if (cur->bc_ptrs[level] == 0) {
-				if ((error = xfs_alloc_decrement(cur, level, &i)))
-					return error;
-				break;
-			}
-		}
-	}
-	*stat = i;
-	return 0;
+	*l0 = be32_to_cpu(rec->alloc.ar_startblock);
+	*l1 = be32_to_cpu(rec->alloc.ar_blockcount);
+	*l2 = 0;
 }
+#endif /* XFS_BTREE_TRACE */
+
+static const struct xfs_btree_ops xfs_allocbt_ops = {
+	.rec_len		= sizeof(xfs_alloc_rec_t),
+	.key_len		= sizeof(xfs_alloc_key_t),
+
+	.dup_cursor		= xfs_allocbt_dup_cursor,
+	.set_root		= xfs_allocbt_set_root,
+	.kill_root		= xfs_allocbt_kill_root,
+	.alloc_block		= xfs_allocbt_alloc_block,
+	.free_block		= xfs_allocbt_free_block,
+	.update_lastrec		= xfs_allocbt_update_lastrec,
+	.get_minrecs		= xfs_allocbt_get_minrecs,
+	.get_maxrecs		= xfs_allocbt_get_maxrecs,
+	.init_key_from_rec	= xfs_allocbt_init_key_from_rec,
+	.init_rec_from_key	= xfs_allocbt_init_rec_from_key,
+	.init_rec_from_cur	= xfs_allocbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_allocbt_init_ptr_from_cur,
+	.key_diff		= xfs_allocbt_key_diff,
 
-/*
- * Get the data from the pointed-to record.
- */
-int					/* error */
-xfs_alloc_get_rec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_agblock_t		*bno,	/* output: starting block of extent */
-	xfs_extlen_t		*len,	/* output: length of extent */
-	int			*stat)	/* output: success/failure */
-{
-	xfs_alloc_block_t	*block;	/* btree block */
 #ifdef DEBUG
-	int			error;	/* error return value */
+	.keys_inorder		= xfs_allocbt_keys_inorder,
+	.recs_inorder		= xfs_allocbt_recs_inorder,
 #endif
-	int			ptr;	/* record number */
 
-	ptr = cur->bc_ptrs[0];
-	block = XFS_BUF_TO_ALLOC_BLOCK(cur->bc_bufs[0]);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, 0, cur->bc_bufs[0])))
-		return error;
+#ifdef XFS_BTREE_TRACE
+	.trace_enter		= xfs_allocbt_trace_enter,
+	.trace_cursor		= xfs_allocbt_trace_cursor,
+	.trace_key		= xfs_allocbt_trace_key,
+	.trace_record		= xfs_allocbt_trace_record,
 #endif
-	/*
-	 * Off the right end or left end, return failure.
-	 */
-	if (ptr > be16_to_cpu(block->bb_numrecs) || ptr <= 0) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Point to the record and extract its data.
-	 */
-	{
-		xfs_alloc_rec_t		*rec;	/* record data */
-
-		rec = XFS_ALLOC_REC_ADDR(block, ptr, cur);
-		*bno = be32_to_cpu(rec->ar_startblock);
-		*len = be32_to_cpu(rec->ar_blockcount);
-	}
-	*stat = 1;
-	return 0;
-}
+};
 
 /*
- * Increment cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
+ * Allocate a new allocation btree cursor.
  */
-int					/* error */
-xfs_alloc_increment(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level in btree, 0 is leaf */
-	int			*stat)	/* success/failure */
+struct xfs_btree_cur *			/* new alloc btree cursor */
+xfs_allocbt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_buf		*agbp,		/* buffer for agf structure */
+	xfs_agnumber_t		agno,		/* allocation group number */
+	xfs_btnum_t		btnum)		/* btree identifier */
 {
-	xfs_alloc_block_t	*block;	/* btree block */
-	xfs_buf_t		*bp;	/* tree block buffer */
-	int			error;	/* error return value */
-	int			lev;	/* btree level */
-
-	ASSERT(level < cur->bc_nlevels);
-	/*
-	 * Read-ahead to the right at this level.
-	 */
-	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
-	/*
-	 * Get a pointer to the btree block.
-	 */
-	bp = cur->bc_bufs[level];
-	block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-		return error;
-#endif
-	/*
-	 * Increment the ptr at this level.  If we're still in the block
-	 * then we're done.
-	 */
-	if (++cur->bc_ptrs[level] <= be16_to_cpu(block->bb_numrecs)) {
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * If we just went off the right edge of the tree, return failure.
-	 */
-	if (be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * March up the tree incrementing pointers.
-	 * Stop when we don't go off the right edge of a block.
-	 */
-	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
-		bp = cur->bc_bufs[lev];
-		block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, block, lev, bp)))
-			return error;
-#endif
-		if (++cur->bc_ptrs[lev] <= be16_to_cpu(block->bb_numrecs))
-			break;
-		/*
-		 * Read-ahead the right block, we're going to read it
-		 * in the next loop.
-		 */
-		xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
-	}
-	/*
-	 * If we went off the root then we are seriously confused.
-	 */
-	ASSERT(lev < cur->bc_nlevels);
-	/*
-	 * Now walk back down the tree, fixing up the cursor's buffer
-	 * pointers and key numbers.
-	 */
-	for (bp = cur->bc_bufs[lev], block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-	     lev > level; ) {
-		xfs_agblock_t	agbno;	/* block number of btree block */
+	struct xfs_agf		*agf = XFS_BUF_TO_AGF(agbp);
+	struct xfs_btree_cur	*cur;
 
-		agbno = be32_to_cpu(*XFS_ALLOC_PTR_ADDR(block, cur->bc_ptrs[lev], cur));
-		if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-				cur->bc_private.a.agno, agbno, 0, &bp,
-				XFS_ALLOC_BTREE_REF)))
-			return error;
-		lev--;
-		xfs_btree_setbuf(cur, lev, bp);
-		block = XFS_BUF_TO_ALLOC_BLOCK(bp);
-		if ((error = xfs_btree_check_sblock(cur, block, lev, bp)))
-			return error;
-		cur->bc_ptrs[lev] = 1;
-	}
-	*stat = 1;
-	return 0;
-}
+	ASSERT(btnum == XFS_BTNUM_BNO || btnum == XFS_BTNUM_CNT);
 
-/*
- * Insert the current record at the point referenced by cur.
- * The cursor may be inconsistent on return if splits have been done.
- */
-int					/* error */
-xfs_alloc_insert(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	int		*stat)		/* success/failure */
-{
-	int		error;		/* error return value */
-	int		i;		/* result value, 0 for failure */
-	int		level;		/* current level number in btree */
-	xfs_agblock_t	nbno;		/* new block number (split result) */
-	xfs_btree_cur_t	*ncur;		/* new cursor (split result) */
-	xfs_alloc_rec_t	nrec;		/* record being inserted this level */
-	xfs_btree_cur_t	*pcur;		/* previous level's cursor */
+	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
 
-	level = 0;
-	nbno = NULLAGBLOCK;
-	nrec.ar_startblock = cpu_to_be32(cur->bc_rec.a.ar_startblock);
-	nrec.ar_blockcount = cpu_to_be32(cur->bc_rec.a.ar_blockcount);
-	ncur = NULL;
-	pcur = cur;
-	/*
-	 * Loop going up the tree, starting at the leaf level.
-	 * Stop when we don't get a split block, that must mean that
-	 * the insert is finished with this level.
-	 */
-	do {
-		/*
-		 * Insert nrec/nbno into this level of the tree.
-		 * Note if we fail, nbno will be null.
-		 */
-		if ((error = xfs_alloc_insrec(pcur, level++, &nbno, &nrec, &ncur,
-				&i))) {
-			if (pcur != cur)
-				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
-			return error;
-		}
-		/*
-		 * See if the cursor we just used is trash.
-		 * Can't trash the caller's cursor, but otherwise we should
-		 * if ncur is a new cursor or we're about to be done.
-		 */
-		if (pcur != cur && (ncur || nbno == NULLAGBLOCK)) {
-			cur->bc_nlevels = pcur->bc_nlevels;
-			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
-		}
-		/*
-		 * If we got a new cursor, switch to it.
-		 */
-		if (ncur) {
-			pcur = ncur;
-			ncur = NULL;
-		}
-	} while (nbno != NULLAGBLOCK);
-	*stat = i;
-	return 0;
-}
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_nlevels = be32_to_cpu(agf->agf_levels[btnum]);
+	cur->bc_btnum = btnum;
+	cur->bc_blocklog = mp->m_sb.sb_blocklog;
 
-/*
- * Lookup the record equal to [bno, len] in the btree given by cur.
- */
-int					/* error */
-xfs_alloc_lookup_eq(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agblock_t	bno,		/* starting block of extent */
-	xfs_extlen_t	len,		/* length of extent */
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.a.ar_startblock = bno;
-	cur->bc_rec.a.ar_blockcount = len;
-	return xfs_alloc_lookup(cur, XFS_LOOKUP_EQ, stat);
-}
+	cur->bc_ops = &xfs_allocbt_ops;
+	if (btnum == XFS_BTNUM_CNT)
+		cur->bc_flags = XFS_BTREE_LASTREC_UPDATE;
 
-/*
- * Lookup the first record greater than or equal to [bno, len]
- * in the btree given by cur.
- */
-int					/* error */
-xfs_alloc_lookup_ge(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agblock_t	bno,		/* starting block of extent */
-	xfs_extlen_t	len,		/* length of extent */
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.a.ar_startblock = bno;
-	cur->bc_rec.a.ar_blockcount = len;
-	return xfs_alloc_lookup(cur, XFS_LOOKUP_GE, stat);
-}
+	cur->bc_private.a.agbp = agbp;
+	cur->bc_private.a.agno = agno;
 
-/*
- * Lookup the first record less than or equal to [bno, len]
- * in the btree given by cur.
- */
-int					/* error */
-xfs_alloc_lookup_le(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agblock_t	bno,		/* starting block of extent */
-	xfs_extlen_t	len,		/* length of extent */
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.a.ar_startblock = bno;
-	cur->bc_rec.a.ar_blockcount = len;
-	return xfs_alloc_lookup(cur, XFS_LOOKUP_LE, stat);
+	return cur;
 }
 
 /*
- * Update the record referred to by cur, to the value given by [bno, len].
- * This either works (return 0) or gets an EFSCORRUPTED error.
+ * Calculate number of records in an alloc btree block.
  */
-int					/* error */
-xfs_alloc_update(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_agblock_t		bno,	/* starting block of extent */
-	xfs_extlen_t		len)	/* length of extent */
+int
+xfs_allocbt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
 {
-	xfs_alloc_block_t	*block;	/* btree block to update */
-	int			error;	/* error return value */
-	int			ptr;	/* current record number (updating) */
+	blocklen -= XFS_ALLOC_BLOCK_LEN(mp);
 
-	ASSERT(len > 0);
-	/*
-	 * Pick up the a.g. freelist struct and the current block.
-	 */
-	block = XFS_BUF_TO_ALLOC_BLOCK(cur->bc_bufs[0]);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, 0, cur->bc_bufs[0])))
-		return error;
-#endif
-	/*
-	 * Get the address of the rec to be updated.
-	 */
-	ptr = cur->bc_ptrs[0];
-	{
-		xfs_alloc_rec_t		*rp;	/* pointer to updated record */
-
-		rp = XFS_ALLOC_REC_ADDR(block, ptr, cur);
-		/*
-		 * Fill in the new contents and log them.
-		 */
-		rp->ar_startblock = cpu_to_be32(bno);
-		rp->ar_blockcount = cpu_to_be32(len);
-		xfs_alloc_log_recs(cur, cur->bc_bufs[0], ptr, ptr);
-	}
-	/*
-	 * If it's the by-size btree and it's the last leaf block and
-	 * it's the last record... then update the size of the longest
-	 * extent in the a.g., which we cache in the a.g. freelist header.
-	 */
-	if (cur->bc_btnum == XFS_BTNUM_CNT &&
-	    be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK &&
-	    ptr == be16_to_cpu(block->bb_numrecs)) {
-		xfs_agf_t	*agf;	/* a.g. freespace header */
-		xfs_agnumber_t	seqno;
-
-		agf = XFS_BUF_TO_AGF(cur->bc_private.a.agbp);
-		seqno = be32_to_cpu(agf->agf_seqno);
-		cur->bc_mp->m_perag[seqno].pagf_longest = len;
-		agf->agf_longest = cpu_to_be32(len);
-		xfs_alloc_log_agf(cur->bc_tp, cur->bc_private.a.agbp,
-			XFS_AGF_LONGEST);
-	}
-	/*
-	 * Updating first record in leaf. Pass new key value up to our parent.
-	 */
-	if (ptr == 1) {
-		xfs_alloc_key_t	key;	/* key containing [bno, len] */
-
-		key.ar_startblock = cpu_to_be32(bno);
-		key.ar_blockcount = cpu_to_be32(len);
-		if ((error = xfs_alloc_updkey(cur, &key, 1)))
-			return error;
-	}
-	return 0;
+	if (leaf)
+		return blocklen / sizeof(xfs_alloc_rec_t);
+	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
 }
diff --git a/fs/xfs/xfs_alloc_btree.h b/fs/xfs/xfs_alloc_btree.h
index 5bd1a2c8bd0..a6caa0022c9 100644
--- a/fs/xfs/xfs_alloc_btree.h
+++ b/fs/xfs/xfs_alloc_btree.h
@@ -24,7 +24,6 @@
 
 struct xfs_buf;
 struct xfs_btree_cur;
-struct xfs_btree_sblock;
 struct xfs_mount;
 
 /*
@@ -50,16 +49,6 @@ typedef struct xfs_alloc_rec_incore {
 
 /* btree pointer type */
 typedef __be32 xfs_alloc_ptr_t;
-/* btree block header type */
-typedef	struct xfs_btree_sblock xfs_alloc_block_t;
-
-#define	XFS_BUF_TO_ALLOC_BLOCK(bp)	((xfs_alloc_block_t *)XFS_BUF_PTR(bp))
-
-/*
- * Real block structures have a size equal to the disk block size.
- */
-#define	XFS_ALLOC_BLOCK_MAXRECS(lev,cur) ((cur)->bc_mp->m_alloc_mxr[lev != 0])
-#define	XFS_ALLOC_BLOCK_MINRECS(lev,cur) ((cur)->bc_mp->m_alloc_mnr[lev != 0])
 
 /*
  * Minimum and maximum blocksize and sectorsize.
@@ -83,73 +72,39 @@ typedef	struct xfs_btree_sblock xfs_alloc_block_t;
 #define	XFS_CNT_BLOCK(mp)	((xfs_agblock_t)(XFS_BNO_BLOCK(mp) + 1))
 
 /*
- * Record, key, and pointer address macros for btree blocks.
- */
-#define	XFS_ALLOC_REC_ADDR(bb,i,cur)	\
-	XFS_BTREE_REC_ADDR(xfs_alloc, bb, i)
-
-#define	XFS_ALLOC_KEY_ADDR(bb,i,cur)	\
-	XFS_BTREE_KEY_ADDR(xfs_alloc, bb, i)
-
-#define	XFS_ALLOC_PTR_ADDR(bb,i,cur)	\
-	XFS_BTREE_PTR_ADDR(xfs_alloc, bb, i, XFS_ALLOC_BLOCK_MAXRECS(1, cur))
-
-/*
- * Decrement cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-extern int xfs_alloc_decrement(struct xfs_btree_cur *cur, int level, int *stat);
-
-/*
- * Delete the record pointed to by cur.
- * The cursor refers to the place where the record was (could be inserted)
- * when the operation returns.
- */
-extern int xfs_alloc_delete(struct xfs_btree_cur *cur, int *stat);
-
-/*
- * Get the data from the pointed-to record.
- */
-extern int xfs_alloc_get_rec(struct xfs_btree_cur *cur,	xfs_agblock_t *bno,
-				xfs_extlen_t *len, int *stat);
-
-/*
- * Increment cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-extern int xfs_alloc_increment(struct xfs_btree_cur *cur, int level, int *stat);
-
-/*
- * Insert the current record at the point referenced by cur.
- * The cursor may be inconsistent on return if splits have been done.
- */
-extern int xfs_alloc_insert(struct xfs_btree_cur *cur, int *stat);
-
-/*
- * Lookup the record equal to [bno, len] in the btree given by cur.
- */
-extern int xfs_alloc_lookup_eq(struct xfs_btree_cur *cur, xfs_agblock_t bno,
-				xfs_extlen_t len, int *stat);
-
-/*
- * Lookup the first record greater than or equal to [bno, len]
- * in the btree given by cur.
- */
-extern int xfs_alloc_lookup_ge(struct xfs_btree_cur *cur, xfs_agblock_t bno,
-				xfs_extlen_t len, int *stat);
-
-/*
- * Lookup the first record less than or equal to [bno, len]
- * in the btree given by cur.
+ * Btree block header size depends on a superblock flag.
+ *
+ * (not quite yet, but soon)
  */
-extern int xfs_alloc_lookup_le(struct xfs_btree_cur *cur, xfs_agblock_t bno,
-				xfs_extlen_t len, int *stat);
+#define XFS_ALLOC_BLOCK_LEN(mp)	XFS_BTREE_SBLOCK_LEN
 
 /*
- * Update the record referred to by cur, to the value given by [bno, len].
- * This either works (return 0) or gets an EFSCORRUPTED error.
- */
-extern int xfs_alloc_update(struct xfs_btree_cur *cur, xfs_agblock_t bno,
-				xfs_extlen_t len);
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_ALLOC_REC_ADDR(mp, block, index) \
+	((xfs_alloc_rec_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 (((index) - 1) * sizeof(xfs_alloc_rec_t))))
+
+#define XFS_ALLOC_KEY_ADDR(mp, block, index) \
+	((xfs_alloc_key_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_alloc_key_t)))
+
+#define XFS_ALLOC_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_alloc_ptr_t *) \
+		((char *)(block) + \
+		 XFS_ALLOC_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_alloc_key_t) + \
+		 ((index) - 1) * sizeof(xfs_alloc_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_allocbt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_buf *,
+		xfs_agnumber_t, xfs_btnum_t);
+extern int xfs_allocbt_maxrecs(struct xfs_mount *, int, int);
 
 #endif	/* __XFS_ALLOC_BTREE_H__ */
diff --git a/fs/xfs/xfs_arch.h b/fs/xfs/xfs_arch.h
index 0b3b5efe848..53d5e70d136 100644
--- a/fs/xfs/xfs_arch.h
+++ b/fs/xfs/xfs_arch.h
@@ -41,21 +41,36 @@
 #endif
 
 #ifdef XFS_NATIVE_HOST
-#define cpu_to_be16(val)	((__be16)(val))
-#define cpu_to_be32(val)	((__be32)(val))
-#define cpu_to_be64(val)	((__be64)(val))
-#define be16_to_cpu(val)	((__uint16_t)(val))
-#define be32_to_cpu(val)	((__uint32_t)(val))
-#define be64_to_cpu(val)	((__uint64_t)(val))
+#define cpu_to_be16(val)	((__force __be16)(__u16)(val))
+#define cpu_to_be32(val)	((__force __be32)(__u32)(val))
+#define cpu_to_be64(val)	((__force __be64)(__u64)(val))
+#define be16_to_cpu(val)	((__force __u16)(__be16)(val))
+#define be32_to_cpu(val)	((__force __u32)(__be32)(val))
+#define be64_to_cpu(val)	((__force __u64)(__be64)(val))
 #else
-#define cpu_to_be16(val)	(__swab16((__uint16_t)(val)))
-#define cpu_to_be32(val)	(__swab32((__uint32_t)(val)))
-#define cpu_to_be64(val)	(__swab64((__uint64_t)(val)))
-#define be16_to_cpu(val)	(__swab16((__be16)(val)))
-#define be32_to_cpu(val)	(__swab32((__be32)(val)))
-#define be64_to_cpu(val)	(__swab64((__be64)(val)))
+#define cpu_to_be16(val)	((__force __be16)__swab16((__u16)(val)))
+#define cpu_to_be32(val)	((__force __be32)__swab32((__u32)(val)))
+#define cpu_to_be64(val)	((__force __be64)__swab64((__u64)(val)))
+#define be16_to_cpu(val)	(__swab16((__force __u16)(__be16)(val)))
+#define be32_to_cpu(val)	(__swab32((__force __u32)(__be32)(val)))
+#define be64_to_cpu(val)	(__swab64((__force __u64)(__be64)(val)))
 #endif
 
+static inline void be16_add_cpu(__be16 *a, __s16 b)
+{
+	*a = cpu_to_be16(be16_to_cpu(*a) + b);
+}
+
+static inline void be32_add_cpu(__be32 *a, __s32 b)
+{
+	*a = cpu_to_be32(be32_to_cpu(*a) + b);
+}
+
+static inline void be64_add_cpu(__be64 *a, __s64 b)
+{
+	*a = cpu_to_be64(be64_to_cpu(*a) + b);
+}
+
 #endif	/* __KERNEL__ */
 
 /* do we need conversion? */
diff --git a/fs/xfs/xfs_bit.h b/fs/xfs/xfs_bit.h
index 8e0e463dae2..bca7b243c31 100644
--- a/fs/xfs/xfs_bit.h
+++ b/fs/xfs/xfs_bit.h
@@ -61,8 +61,7 @@ static inline int xfs_highbit64(__uint64_t v)
 /* Get low bit set out of 32-bit argument, -1 if none set */
 static inline int xfs_lowbit32(__uint32_t v)
 {
-	unsigned long	t = v;
-	return (v) ? find_first_bit(&t, 32) : -1;
+	return ffs(v) - 1;
 }
 
 /* Get low bit set out of 64-bit argument, -1 if none set */
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index a1aab9275d5..138308e70d1 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -393,8 +393,8 @@ xfs_bmap_count_leaves(
 
 STATIC void
 xfs_bmap_disk_count_leaves(
-	xfs_extnum_t		idx,
-	xfs_bmbt_block_t	*block,
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*block,
 	int			numrecs,
 	int			*count);
 
@@ -402,6 +402,53 @@ xfs_bmap_disk_count_leaves(
  * Bmap internal routines.
  */
 
+STATIC int				/* error */
+xfs_bmbt_lookup_eq(
+	struct xfs_btree_cur	*cur,
+	xfs_fileoff_t		off,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		len,
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.b.br_startoff = off;
+	cur->bc_rec.b.br_startblock = bno;
+	cur->bc_rec.b.br_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+STATIC int				/* error */
+xfs_bmbt_lookup_ge(
+	struct xfs_btree_cur	*cur,
+	xfs_fileoff_t		off,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		len,
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.b.br_startoff = off;
+	cur->bc_rec.b.br_startblock = bno;
+	cur->bc_rec.b.br_blockcount = len;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+* Update the record referred to by cur to the value given
+ * by [off, bno, len, state].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int
+xfs_bmbt_update(
+	struct xfs_btree_cur	*cur,
+	xfs_fileoff_t		off,
+	xfs_fsblock_t		bno,
+	xfs_filblks_t		len,
+	xfs_exntst_t		state)
+{
+	union xfs_btree_rec	rec;
+
+	xfs_bmbt_disk_set_allf(&rec.bmbt, off, bno, len, state);
+	return xfs_btree_update(cur, &rec);
+}
+
 /*
  * Called from xfs_bmap_add_attrfork to handle btree format files.
  */
@@ -422,15 +469,14 @@ xfs_bmap_add_attrfork_btree(
 	if (ip->i_df.if_broot_bytes <= XFS_IFORK_DSIZE(ip))
 		*flags |= XFS_ILOG_DBROOT;
 	else {
-		cur = xfs_btree_init_cursor(mp, tp, NULL, 0, XFS_BTNUM_BMAP, ip,
-			XFS_DATA_FORK);
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, XFS_DATA_FORK);
 		cur->bc_private.b.flist = flist;
 		cur->bc_private.b.firstblock = *firstblock;
 		if ((error = xfs_bmbt_lookup_ge(cur, 0, 0, 0, &stat)))
 			goto error0;
 		/* must be at least one entry */
 		XFS_WANT_CORRUPTED_GOTO(stat == 1, error0);
-		if ((error = xfs_bmbt_newroot(cur, flags, &stat)))
+		if ((error = xfs_btree_new_iroot(cur, flags, &stat)))
 			goto error0;
 		if (stat == 0) {
 			xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
@@ -818,10 +864,10 @@ xfs_bmap_add_extent_delay_real(
 					RIGHT.br_blockcount, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_delete(cur, &i)))
+			if ((error = xfs_btree_delete(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
@@ -931,7 +977,7 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1007,7 +1053,7 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1097,7 +1143,7 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1152,7 +1198,7 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1379,16 +1425,16 @@ xfs_bmap_add_extent_unwritten_real(
 					RIGHT.br_blockcount, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_delete(cur, &i)))
+			if ((error = xfs_btree_delete(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_delete(cur, &i)))
+			if ((error = xfs_btree_delete(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
@@ -1428,10 +1474,10 @@ xfs_bmap_add_extent_unwritten_real(
 					&i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_delete(cur, &i)))
+			if ((error = xfs_btree_delete(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
@@ -1471,10 +1517,10 @@ xfs_bmap_add_extent_unwritten_real(
 					RIGHT.br_blockcount, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_delete(cur, &i)))
+			if ((error = xfs_btree_delete(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 			if ((error = xfs_bmbt_update(cur, new->br_startoff,
@@ -1557,7 +1603,7 @@ xfs_bmap_add_extent_unwritten_real(
 				PREV.br_blockcount - new->br_blockcount,
 				oldext)))
 				goto done;
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			if (xfs_bmbt_update(cur, LEFT.br_startoff,
 				LEFT.br_startblock,
@@ -1605,7 +1651,7 @@ xfs_bmap_add_extent_unwritten_real(
 				oldext)))
 				goto done;
 			cur->bc_rec.b = *new;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1647,7 +1693,7 @@ xfs_bmap_add_extent_unwritten_real(
 				PREV.br_blockcount - new->br_blockcount,
 				oldext)))
 				goto done;
-			if ((error = xfs_bmbt_increment(cur, 0, &i)))
+			if ((error = xfs_btree_increment(cur, 0, &i)))
 				goto done;
 			if ((error = xfs_bmbt_update(cur, new->br_startoff,
 				new->br_startblock,
@@ -1695,7 +1741,7 @@ xfs_bmap_add_extent_unwritten_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1743,7 +1789,7 @@ xfs_bmap_add_extent_unwritten_real(
 			cur->bc_rec.b = PREV;
 			cur->bc_rec.b.br_blockcount =
 				new->br_startoff - PREV.br_startoff;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 			/*
@@ -1758,7 +1804,7 @@ xfs_bmap_add_extent_unwritten_real(
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			/* new middle extent - newext */
 			cur->bc_rec.b.br_state = new->br_state;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -2106,10 +2152,10 @@ xfs_bmap_add_extent_hole_real(
 					right.br_blockcount, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_delete(cur, &i)))
+			if ((error = xfs_btree_delete(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_decrement(cur, 0, &i)))
+			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 			if ((error = xfs_bmbt_update(cur, left.br_startoff,
@@ -2218,7 +2264,7 @@ xfs_bmap_add_extent_hole_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
 			cur->bc_rec.b.br_state = new->br_state;
-			if ((error = xfs_bmbt_insert(cur, &i)))
+			if ((error = xfs_btree_insert(cur, &i)))
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -2996,24 +3042,24 @@ xfs_bmap_btree_to_extents(
 	int			whichfork)  /* data or attr fork */
 {
 	/* REFERENCED */
-	xfs_bmbt_block_t	*cblock;/* child btree block */
+	struct xfs_btree_block	*cblock;/* child btree block */
 	xfs_fsblock_t		cbno;	/* child block number */
 	xfs_buf_t		*cbp;	/* child block's buffer */
 	int			error;	/* error return value */
 	xfs_ifork_t		*ifp;	/* inode fork data */
 	xfs_mount_t		*mp;	/* mount point structure */
 	__be64			*pp;	/* ptr to block address */
-	xfs_bmbt_block_t	*rblock;/* root btree block */
+	struct xfs_btree_block	*rblock;/* root btree block */
 
+	mp = ip->i_mount;
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	ASSERT(ifp->if_flags & XFS_IFEXTENTS);
 	ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE);
 	rblock = ifp->if_broot;
 	ASSERT(be16_to_cpu(rblock->bb_level) == 1);
 	ASSERT(be16_to_cpu(rblock->bb_numrecs) == 1);
-	ASSERT(XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes) == 1);
-	mp = ip->i_mount;
-	pp = XFS_BMAP_BROOT_PTR_ADDR(rblock, 1, ifp->if_broot_bytes);
+	ASSERT(xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0) == 1);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, ifp->if_broot_bytes);
 	cbno = be64_to_cpu(*pp);
 	*logflagsp = 0;
 #ifdef DEBUG
@@ -3023,8 +3069,8 @@ xfs_bmap_btree_to_extents(
 	if ((error = xfs_btree_read_bufl(mp, tp, cbno, 0, &cbp,
 			XFS_BMAP_BTREE_REF)))
 		return error;
-	cblock = XFS_BUF_TO_BMBT_BLOCK(cbp);
-	if ((error = xfs_btree_check_lblock(cur, cblock, 0, cbp)))
+	cblock = XFS_BUF_TO_BLOCK(cbp);
+	if ((error = xfs_btree_check_block(cur, cblock, 0, cbp)))
 		return error;
 	xfs_bmap_add_free(cbno, 1, cur->bc_private.b.flist, mp);
 	ip->i_d.di_nblocks--;
@@ -3170,7 +3216,7 @@ xfs_bmap_del_extent(
 			flags |= XFS_ILOG_FEXT(whichfork);
 			break;
 		}
-		if ((error = xfs_bmbt_delete(cur, &i)))
+		if ((error = xfs_btree_delete(cur, &i)))
 			goto done;
 		XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		break;
@@ -3254,10 +3300,10 @@ xfs_bmap_del_extent(
 						got.br_startblock, temp,
 						got.br_state)))
 					goto done;
-				if ((error = xfs_bmbt_increment(cur, 0, &i)))
+				if ((error = xfs_btree_increment(cur, 0, &i)))
 					goto done;
 				cur->bc_rec.b = new;
-				error = xfs_bmbt_insert(cur, &i);
+				error = xfs_btree_insert(cur, &i);
 				if (error && error != ENOSPC)
 					goto done;
 				/*
@@ -3404,11 +3450,11 @@ xfs_bmap_extents_to_btree(
 	int			*logflagsp,	/* inode logging flags */
 	int			whichfork)	/* data or attr fork */
 {
-	xfs_bmbt_block_t	*ablock;	/* allocated (child) bt block */
+	struct xfs_btree_block	*ablock;	/* allocated (child) bt block */
 	xfs_buf_t		*abp;		/* buffer for ablock */
 	xfs_alloc_arg_t		args;		/* allocation arguments */
 	xfs_bmbt_rec_t		*arp;		/* child record pointer */
-	xfs_bmbt_block_t	*block;		/* btree root block */
+	struct xfs_btree_block	*block;		/* btree root block */
 	xfs_btree_cur_t		*cur;		/* bmap btree cursor */
 	xfs_bmbt_rec_host_t	*ep;		/* extent record pointer */
 	int			error;		/* error return value */
@@ -3428,6 +3474,7 @@ xfs_bmap_extents_to_btree(
 	 */
 	xfs_iroot_realloc(ip, 1, whichfork);
 	ifp->if_flags |= XFS_IFBROOT;
+
 	/*
 	 * Fill in the root.
 	 */
@@ -3435,14 +3482,14 @@ xfs_bmap_extents_to_btree(
 	block->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
 	block->bb_level = cpu_to_be16(1);
 	block->bb_numrecs = cpu_to_be16(1);
-	block->bb_leftsib = cpu_to_be64(NULLDFSBNO);
-	block->bb_rightsib = cpu_to_be64(NULLDFSBNO);
+	block->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+	block->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+
 	/*
 	 * Need a cursor.  Can't allocate until bb_level is filled in.
 	 */
 	mp = ip->i_mount;
-	cur = xfs_btree_init_cursor(mp, tp, NULL, 0, XFS_BTNUM_BMAP, ip,
-		whichfork);
+	cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
 	cur->bc_private.b.firstblock = *firstblock;
 	cur->bc_private.b.flist = flist;
 	cur->bc_private.b.flags = wasdel ? XFS_BTCUR_BPRV_WASDEL : 0;
@@ -3489,12 +3536,12 @@ xfs_bmap_extents_to_btree(
 	/*
 	 * Fill in the child block.
 	 */
-	ablock = XFS_BUF_TO_BMBT_BLOCK(abp);
+	ablock = XFS_BUF_TO_BLOCK(abp);
 	ablock->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
 	ablock->bb_level = 0;
-	ablock->bb_leftsib = cpu_to_be64(NULLDFSBNO);
-	ablock->bb_rightsib = cpu_to_be64(NULLDFSBNO);
-	arp = XFS_BMAP_REC_IADDR(ablock, 1, cur);
+	ablock->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+	ablock->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+	arp = XFS_BMBT_REC_ADDR(mp, ablock, 1);
 	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	for (cnt = i = 0; i < nextents; i++) {
 		ep = xfs_iext_get_ext(ifp, i);
@@ -3505,21 +3552,24 @@ xfs_bmap_extents_to_btree(
 		}
 	}
 	ASSERT(cnt == XFS_IFORK_NEXTENTS(ip, whichfork));
-	ablock->bb_numrecs = cpu_to_be16(cnt);
+	xfs_btree_set_numrecs(ablock, cnt);
+
 	/*
 	 * Fill in the root key and pointer.
 	 */
-	kp = XFS_BMAP_KEY_IADDR(block, 1, cur);
-	arp = XFS_BMAP_REC_IADDR(ablock, 1, cur);
+	kp = XFS_BMBT_KEY_ADDR(mp, block, 1);
+	arp = XFS_BMBT_REC_ADDR(mp, ablock, 1);
 	kp->br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(arp));
-	pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
+	pp = XFS_BMBT_PTR_ADDR(mp, block, 1, xfs_bmbt_get_maxrecs(cur,
+						be16_to_cpu(block->bb_level)));
 	*pp = cpu_to_be64(args.fsbno);
+
 	/*
 	 * Do all this logging at the end so that
 	 * the root is at the right level.
 	 */
-	xfs_bmbt_log_block(cur, abp, XFS_BB_ALL_BITS);
-	xfs_bmbt_log_recs(cur, abp, 1, be16_to_cpu(ablock->bb_numrecs));
+	xfs_btree_log_block(cur, abp, XFS_BB_ALL_BITS);
+	xfs_btree_log_recs(cur, abp, 1, be16_to_cpu(ablock->bb_numrecs));
 	ASSERT(*curp == NULL);
 	*curp = cur;
 	*logflagsp = XFS_ILOG_CORE | XFS_ILOG_FBROOT(whichfork);
@@ -4176,7 +4226,7 @@ xfs_bmap_compute_maxlevels(
 		maxleafents = MAXAEXTNUM;
 		sz = XFS_BMDR_SPACE_CALC(MINABTPTRS);
 	}
-	maxrootrecs = (int)XFS_BTREE_BLOCK_MAXRECS(sz, xfs_bmdr, 0);
+	maxrootrecs = xfs_bmdr_maxrecs(mp, sz, 0);
 	minleafrecs = mp->m_bmap_dmnr[0];
 	minnoderecs = mp->m_bmap_dmnr[1];
 	maxblocks = (maxleafents + minleafrecs - 1) / minleafrecs;
@@ -4242,9 +4292,15 @@ xfs_bmap_finish(
 	 * We have a new transaction, so we should return committed=1,
 	 * even though we're returning an error.
 	 */
-	if (error) {
+	if (error)
 		return error;
-	}
+
+	/*
+	 * transaction commit worked ok so we can drop the extra ticket
+	 * reference that we gained in xfs_trans_dup()
+	 */
+	xfs_log_ticket_put(ntp->t_ticket);
+
 	if ((error = xfs_trans_reserve(ntp, 0, logres, 0, XFS_TRANS_PERM_LOG_RES,
 			logcount)))
 		return error;
@@ -4474,6 +4530,22 @@ xfs_bmap_one_block(
 	return rval;
 }
 
+STATIC int
+xfs_bmap_sanity_check(
+	struct xfs_mount	*mp,
+	struct xfs_buf		*bp,
+	int			level)
+{
+	struct xfs_btree_block  *block = XFS_BUF_TO_BLOCK(bp);
+
+	if (be32_to_cpu(block->bb_magic) != XFS_BMAP_MAGIC ||
+	    be16_to_cpu(block->bb_level) != level ||
+	    be16_to_cpu(block->bb_numrecs) == 0 ||
+	    be16_to_cpu(block->bb_numrecs) > mp->m_bmap_dmxr[level != 0])
+		return 0;
+	return 1;
+}
+
 /*
  * Read in the extents to if_extents.
  * All inode fields are set up by caller, we just traverse the btree
@@ -4486,7 +4558,7 @@ xfs_bmap_read_extents(
 	xfs_inode_t		*ip,	/* incore inode */
 	int			whichfork) /* data or attr fork */
 {
-	xfs_bmbt_block_t	*block;	/* current btree block */
+	struct xfs_btree_block	*block;	/* current btree block */
 	xfs_fsblock_t		bno;	/* block # of "block" */
 	xfs_buf_t		*bp;	/* buffer for "block" */
 	int			error;	/* error return value */
@@ -4510,7 +4582,7 @@ xfs_bmap_read_extents(
 	 */
 	level = be16_to_cpu(block->bb_level);
 	ASSERT(level > 0);
-	pp = XFS_BMAP_BROOT_PTR_ADDR(block, 1, ifp->if_broot_bytes);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
 	bno = be64_to_cpu(*pp);
 	ASSERT(bno != NULLDFSBNO);
 	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
@@ -4523,13 +4595,13 @@ xfs_bmap_read_extents(
 		if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp,
 				XFS_BMAP_BTREE_REF)))
 			return error;
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 		XFS_WANT_CORRUPTED_GOTO(
-			XFS_BMAP_SANITY_CHECK(mp, block, level),
+			xfs_bmap_sanity_check(mp, bp, level),
 			error0);
 		if (level == 0)
 			break;
-		pp = XFS_BTREE_PTR_ADDR(xfs_bmbt, block, 1, mp->m_bmap_dmxr[1]);
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
 		bno = be64_to_cpu(*pp);
 		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, bno), error0);
 		xfs_trans_brelse(tp, bp);
@@ -4549,7 +4621,7 @@ xfs_bmap_read_extents(
 		xfs_extnum_t	start;
 
 
-		num_recs = be16_to_cpu(block->bb_numrecs);
+		num_recs = xfs_btree_get_numrecs(block);
 		if (unlikely(i + num_recs > room)) {
 			ASSERT(i + num_recs <= room);
 			xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
@@ -4561,18 +4633,18 @@ xfs_bmap_read_extents(
 			goto error0;
 		}
 		XFS_WANT_CORRUPTED_GOTO(
-			XFS_BMAP_SANITY_CHECK(mp, block, 0),
+			xfs_bmap_sanity_check(mp, bp, 0),
 			error0);
 		/*
 		 * Read-ahead the next leaf block, if any.
 		 */
-		nextbno = be64_to_cpu(block->bb_rightsib);
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
 		if (nextbno != NULLFSBLOCK)
 			xfs_btree_reada_bufl(mp, nextbno, 1);
 		/*
 		 * Copy records into the extent records.
 		 */
-		frp = XFS_BTREE_REC_ADDR(xfs_bmbt, block, 1);
+		frp = XFS_BMBT_REC_ADDR(mp, block, 1);
 		start = i;
 		for (j = 0; j < num_recs; j++, i++, frp++) {
 			xfs_bmbt_rec_host_t *trp = xfs_iext_get_ext(ifp, i);
@@ -4603,7 +4675,7 @@ xfs_bmap_read_extents(
 		if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp,
 				XFS_BMAP_BTREE_REF)))
 			return error;
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 	}
 	ASSERT(i == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)));
 	ASSERT(i == XFS_IFORK_NEXTENTS(ip, whichfork));
@@ -5029,8 +5101,7 @@ xfs_bmapi(
 				if (abno == NULLFSBLOCK)
 					break;
 				if ((ifp->if_flags & XFS_IFBROOT) && !cur) {
-					cur = xfs_btree_init_cursor(mp,
-						tp, NULL, 0, XFS_BTNUM_BMAP,
+					cur = xfs_bmbt_init_cursor(mp, tp,
 						ip, whichfork);
 					cur->bc_private.b.firstblock =
 						*firstblock;
@@ -5147,9 +5218,8 @@ xfs_bmapi(
 			 */
 			ASSERT(mval->br_blockcount <= len);
 			if ((ifp->if_flags & XFS_IFBROOT) && !cur) {
-				cur = xfs_btree_init_cursor(mp,
-					tp, NULL, 0, XFS_BTNUM_BMAP,
-					ip, whichfork);
+				cur = xfs_bmbt_init_cursor(mp,
+					tp, ip, whichfork);
 				cur->bc_private.b.firstblock =
 					*firstblock;
 				cur->bc_private.b.flist = flist;
@@ -5440,8 +5510,7 @@ xfs_bunmapi(
 	logflags = 0;
 	if (ifp->if_flags & XFS_IFBROOT) {
 		ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE);
-		cur = xfs_btree_init_cursor(mp, tp, NULL, 0, XFS_BTNUM_BMAP, ip,
-			whichfork);
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
 		cur->bc_private.b.firstblock = *firstblock;
 		cur->bc_private.b.flist = flist;
 		cur->bc_private.b.flags = 0;
@@ -5742,14 +5811,17 @@ error0:
 STATIC int
 xfs_getbmapx_fix_eof_hole(
 	xfs_inode_t		*ip,		/* xfs incore inode pointer */
-	struct getbmap		*out,		/* output structure */
+	struct getbmapx		*out,		/* output structure */
 	int			prealloced,	/* this is a file with
-						* preallocated data space */
+						 * preallocated data space */
 	__int64_t		end,		/* last block requested */
 	xfs_fsblock_t		startblock)
 {
 	__int64_t		fixlen;
 	xfs_mount_t		*mp;		/* file system mount point */
+	xfs_ifork_t		*ifp;		/* inode fork pointer */
+	xfs_extnum_t		lastx;		/* last extent pointer */
+	xfs_fileoff_t		fileblock;
 
 	if (startblock == HOLESTARTBLOCK) {
 		mp = ip->i_mount;
@@ -5763,21 +5835,33 @@ xfs_getbmapx_fix_eof_hole(
 			out->bmv_length = fixlen;
 		}
 	} else {
-		out->bmv_block = XFS_FSB_TO_DB(ip, startblock);
+		if (startblock == DELAYSTARTBLOCK)
+			out->bmv_block = -2;
+		else
+			out->bmv_block = XFS_FSB_TO_DB(ip, startblock);
+		fileblock = XFS_BB_TO_FSB(ip->i_mount, out->bmv_offset);
+		ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+		if (xfs_iext_bno_to_ext(ifp, fileblock, &lastx) &&
+		   (lastx == (ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t))-1))
+			out->bmv_oflags |= BMV_OF_LAST;
 	}
 
 	return 1;
 }
 
 /*
- * Fcntl interface to xfs_bmapi.
+ * Get inode's extents as described in bmv, and format for output.
+ * Calls formatter to fill the user's buffer until all extents
+ * are mapped, until the passed-in bmv->bmv_count slots have
+ * been filled, or until the formatter short-circuits the loop,
+ * if it is tracking filled-in extents on its own.
  */
 int						/* error code */
 xfs_getbmap(
 	xfs_inode_t		*ip,
-	struct getbmap		*bmv,		/* user bmap structure */
-	void			__user *ap,	/* pointer to user's array */
-	int			interface)	/* interface flags */
+	struct getbmapx		*bmv,		/* user bmap structure */
+	xfs_bmap_format_t	formatter,	/* format to user */
+	void			*arg)		/* formatter arg */
 {
 	__int64_t		bmvend;		/* last block requested */
 	int			error;		/* return value */
@@ -5790,19 +5874,17 @@ xfs_getbmap(
 	int			nexleft;	/* # of user extents left */
 	int			subnex;		/* # of bmapi's can do */
 	int			nmap;		/* number of map entries */
-	struct getbmap		out;		/* output structure */
+	struct getbmapx		out;		/* output structure */
 	int			whichfork;	/* data or attr fork */
 	int			prealloced;	/* this is a file with
 						 * preallocated data space */
-	int			sh_unwritten;	/* true, if unwritten */
-						/* extents listed separately */
+	int			iflags;		/* interface flags */
 	int			bmapi_flags;	/* flags for xfs_bmapi */
-	__int32_t		oflags;		/* getbmapx bmv_oflags field */
 
 	mp = ip->i_mount;
+	iflags = bmv->bmv_iflags;
 
-	whichfork = interface & BMV_IF_ATTRFORK ? XFS_ATTR_FORK : XFS_DATA_FORK;
-	sh_unwritten = (interface & BMV_IF_PREALLOC) != 0;
+	whichfork = iflags & BMV_IF_ATTRFORK ? XFS_ATTR_FORK : XFS_DATA_FORK;
 
 	/*	If the BMV_IF_NO_DMAPI_READ interface bit specified, do not
 	 *	generate a DMAPI read event.  Otherwise, if the DM_EVENT_READ
@@ -5817,7 +5899,7 @@ xfs_getbmap(
 	 *	could misinterpret holes in a DMAPI file as true holes,
 	 *	when in fact they may represent offline user data.
 	 */
-	if ((interface & BMV_IF_NO_DMAPI_READ) == 0 &&
+	if ((iflags & BMV_IF_NO_DMAPI_READ) == 0 &&
 	    DM_EVENT_ENABLED(ip, DM_EVENT_READ) &&
 	    whichfork == XFS_DATA_FORK) {
 		error = XFS_SEND_DATA(mp, DM_EVENT_READ, ip, 0, 0, 0, NULL);
@@ -5873,8 +5955,9 @@ xfs_getbmap(
 
 	xfs_ilock(ip, XFS_IOLOCK_SHARED);
 
-	if (whichfork == XFS_DATA_FORK &&
-		(ip->i_delayed_blks || ip->i_size > ip->i_d.di_size)) {
+	if (((iflags & BMV_IF_DELALLOC) == 0) &&
+	    (whichfork == XFS_DATA_FORK) &&
+	    (ip->i_delayed_blks || ip->i_size > ip->i_d.di_size)) {
 		/* xfs_fsize_t last_byte = xfs_file_last_byte(ip); */
 		error = xfs_flush_pages(ip, (xfs_off_t)0,
 					       -1, 0, FI_REMAPF);
@@ -5884,7 +5967,8 @@ xfs_getbmap(
 		}
 	}
 
-	ASSERT(whichfork == XFS_ATTR_FORK || ip->i_delayed_blks == 0);
+	ASSERT(whichfork == XFS_ATTR_FORK || (iflags & BMV_IF_DELALLOC) ||
+	       ip->i_delayed_blks == 0);
 
 	lock = xfs_ilock_map_shared(ip);
 
@@ -5896,7 +5980,7 @@ xfs_getbmap(
 		nex = XFS_IFORK_NEXTENTS(ip, whichfork) * 2 + 1;
 
 	bmapi_flags = XFS_BMAPI_AFLAG(whichfork) |
-			((sh_unwritten) ? 0 : XFS_BMAPI_IGSTATE);
+			((iflags & BMV_IF_PREALLOC) ? 0 : XFS_BMAPI_IGSTATE);
 
 	/*
 	 * Allocate enough space to handle "subnex" maps at a time.
@@ -5906,9 +5990,12 @@ xfs_getbmap(
 
 	bmv->bmv_entries = 0;
 
-	if (XFS_IFORK_NEXTENTS(ip, whichfork) == 0) {
-		error = 0;
-		goto unlock_and_return;
+	if ((XFS_IFORK_NEXTENTS(ip, whichfork) == 0)) {
+		if (((iflags & BMV_IF_DELALLOC) == 0) ||
+		    whichfork == XFS_ATTR_FORK) {
+			error = 0;
+			goto unlock_and_return;
+		}
 	}
 
 	nexleft = nex;
@@ -5924,52 +6011,40 @@ xfs_getbmap(
 		ASSERT(nmap <= subnex);
 
 		for (i = 0; i < nmap && nexleft && bmv->bmv_length; i++) {
-			nexleft--;
-			oflags = (map[i].br_state == XFS_EXT_UNWRITTEN) ?
-					BMV_OF_PREALLOC : 0;
+			out.bmv_oflags = 0;
+			if (map[i].br_state == XFS_EXT_UNWRITTEN)
+				out.bmv_oflags |= BMV_OF_PREALLOC;
+			else if (map[i].br_startblock == DELAYSTARTBLOCK)
+				out.bmv_oflags |= BMV_OF_DELALLOC;
 			out.bmv_offset = XFS_FSB_TO_BB(mp, map[i].br_startoff);
 			out.bmv_length = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
-			ASSERT(map[i].br_startblock != DELAYSTARTBLOCK);
+			out.bmv_unused1 = out.bmv_unused2 = 0;
+			ASSERT(((iflags & BMV_IF_DELALLOC) != 0) ||
+			      (map[i].br_startblock != DELAYSTARTBLOCK));
                         if (map[i].br_startblock == HOLESTARTBLOCK &&
 			    whichfork == XFS_ATTR_FORK) {
 				/* came to the end of attribute fork */
+				out.bmv_oflags |= BMV_OF_LAST;
 				goto unlock_and_return;
 			} else {
+				int full = 0;	/* user array is full */
+
 				if (!xfs_getbmapx_fix_eof_hole(ip, &out,
 							prealloced, bmvend,
 							map[i].br_startblock)) {
 					goto unlock_and_return;
 				}
 
-				/* return either getbmap/getbmapx structure. */
-				if (interface & BMV_IF_EXTENDED) {
-					struct	getbmapx	outx;
-
-					GETBMAP_CONVERT(out,outx);
-					outx.bmv_oflags = oflags;
-					outx.bmv_unused1 = outx.bmv_unused2 = 0;
-					if (copy_to_user(ap, &outx,
-							sizeof(outx))) {
-						error = XFS_ERROR(EFAULT);
-						goto unlock_and_return;
-					}
-				} else {
-					if (copy_to_user(ap, &out,
-							sizeof(out))) {
-						error = XFS_ERROR(EFAULT);
-						goto unlock_and_return;
-					}
-				}
+				/* format results & advance arg */
+				error = formatter(&arg, &out, &full);
+				if (error || full)
+					goto unlock_and_return;
+				nexleft--;
 				bmv->bmv_offset =
 					out.bmv_offset + out.bmv_length;
 				bmv->bmv_length = MAX((__int64_t)0,
 					(__int64_t)(bmvend - bmv->bmv_offset));
 				bmv->bmv_entries++;
-				ap = (interface & BMV_IF_EXTENDED) ?
-						(void __user *)
-					((struct getbmapx __user *)ap + 1) :
-						(void __user *)
-					((struct getbmap __user *)ap + 1);
 			}
 		}
 	} while (nmap && nexleft && bmv->bmv_length);
@@ -6131,7 +6206,7 @@ xfs_bmap_get_bp(
 
 void
 xfs_check_block(
-	xfs_bmbt_block_t        *block,
+	struct xfs_btree_block	*block,
 	xfs_mount_t		*mp,
 	int			root,
 	short			sz)
@@ -6143,36 +6218,29 @@ xfs_check_block(
 	ASSERT(be16_to_cpu(block->bb_level) > 0);
 
 	prevp = NULL;
-	for( i = 1; i <= be16_to_cpu(block->bb_numrecs); i++) {
+	for( i = 1; i <= xfs_btree_get_numrecs(block); i++) {
 		dmxr = mp->m_bmap_dmxr[0];
-
-		if (root) {
-			keyp = XFS_BMAP_BROOT_KEY_ADDR(block, i, sz);
-		} else {
-			keyp = XFS_BTREE_KEY_ADDR(xfs_bmbt, block, i);
-		}
+		keyp = XFS_BMBT_KEY_ADDR(mp, block, i);
 
 		if (prevp) {
-			xfs_btree_check_key(XFS_BTNUM_BMAP, prevp, keyp);
+			ASSERT(be64_to_cpu(prevp->br_startoff) <
+			       be64_to_cpu(keyp->br_startoff));
 		}
 		prevp = keyp;
 
 		/*
 		 * Compare the block numbers to see if there are dups.
 		 */
+		if (root)
+			pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, i, sz);
+		else
+			pp = XFS_BMBT_PTR_ADDR(mp, block, i, dmxr);
 
-		if (root) {
-			pp = XFS_BMAP_BROOT_PTR_ADDR(block, i, sz);
-		} else {
-			pp = XFS_BTREE_PTR_ADDR(xfs_bmbt, block, i, dmxr);
-		}
 		for (j = i+1; j <= be16_to_cpu(block->bb_numrecs); j++) {
-			if (root) {
-				thispa = XFS_BMAP_BROOT_PTR_ADDR(block, j, sz);
-			} else {
-				thispa = XFS_BTREE_PTR_ADDR(xfs_bmbt, block, j,
-							    dmxr);
-			}
+			if (root)
+				thispa = XFS_BMAP_BROOT_PTR_ADDR(mp, block, j, sz);
+			else
+				thispa = XFS_BMBT_PTR_ADDR(mp, block, j, dmxr);
 			if (*thispa == *pp) {
 				cmn_err(CE_WARN, "%s: thispa(%d) == pp(%d) %Ld",
 					__func__, j, i,
@@ -6195,7 +6263,7 @@ xfs_bmap_check_leaf_extents(
 	xfs_inode_t		*ip,		/* incore inode pointer */
 	int			whichfork)	/* data or attr fork */
 {
-	xfs_bmbt_block_t	*block;	/* current btree block */
+	struct xfs_btree_block	*block;	/* current btree block */
 	xfs_fsblock_t		bno;	/* block # of "block" */
 	xfs_buf_t		*bp;	/* buffer for "block" */
 	int			error;	/* error return value */
@@ -6223,7 +6291,7 @@ xfs_bmap_check_leaf_extents(
 	level = be16_to_cpu(block->bb_level);
 	ASSERT(level > 0);
 	xfs_check_block(block, mp, 1, ifp->if_broot_bytes);
-	pp = XFS_BMAP_BROOT_PTR_ADDR(block, 1, ifp->if_broot_bytes);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
 	bno = be64_to_cpu(*pp);
 
 	ASSERT(bno != NULLDFSBNO);
@@ -6245,9 +6313,9 @@ xfs_bmap_check_leaf_extents(
 		if (!bp && (error = xfs_btree_read_bufl(mp, NULL, bno, 0, &bp,
 				XFS_BMAP_BTREE_REF)))
 			goto error_norelse;
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 		XFS_WANT_CORRUPTED_GOTO(
-			XFS_BMAP_SANITY_CHECK(mp, block, level),
+			xfs_bmap_sanity_check(mp, bp, level),
 			error0);
 		if (level == 0)
 			break;
@@ -6258,7 +6326,7 @@ xfs_bmap_check_leaf_extents(
 		 */
 
 		xfs_check_block(block, mp, 0, 0);
-		pp = XFS_BTREE_PTR_ADDR(xfs_bmbt, block, 1, mp->m_bmap_dmxr[1]);
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
 		bno = be64_to_cpu(*pp);
 		XFS_WANT_CORRUPTED_GOTO(XFS_FSB_SANITY_CHECK(mp, bno), error0);
 		if (bp_release) {
@@ -6280,13 +6348,13 @@ xfs_bmap_check_leaf_extents(
 		xfs_extnum_t	num_recs;
 
 
-		num_recs = be16_to_cpu(block->bb_numrecs);
+		num_recs = xfs_btree_get_numrecs(block);
 
 		/*
 		 * Read-ahead the next leaf block, if any.
 		 */
 
-		nextbno = be64_to_cpu(block->bb_rightsib);
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
 
 		/*
 		 * Check all the extents to make sure they are OK.
@@ -6294,13 +6362,17 @@ xfs_bmap_check_leaf_extents(
 		 * conform with the first entry in this one.
 		 */
 
-		ep = XFS_BTREE_REC_ADDR(xfs_bmbt, block, 1);
+		ep = XFS_BMBT_REC_ADDR(mp, block, 1);
 		if (i) {
-			xfs_btree_check_rec(XFS_BTNUM_BMAP, &last, ep);
+			ASSERT(xfs_bmbt_disk_get_startoff(&last) +
+			       xfs_bmbt_disk_get_blockcount(&last) <=
+			       xfs_bmbt_disk_get_startoff(ep));
 		}
 		for (j = 1; j < num_recs; j++) {
-			nextp = XFS_BTREE_REC_ADDR(xfs_bmbt, block, j + 1);
-			xfs_btree_check_rec(XFS_BTNUM_BMAP, ep, nextp);
+			nextp = XFS_BMBT_REC_ADDR(mp, block, j + 1);
+			ASSERT(xfs_bmbt_disk_get_startoff(ep) +
+			       xfs_bmbt_disk_get_blockcount(ep) <=
+			       xfs_bmbt_disk_get_startoff(nextp));
 			ep = nextp;
 		}
 
@@ -6326,7 +6398,7 @@ xfs_bmap_check_leaf_extents(
 		if (!bp && (error = xfs_btree_read_bufl(mp, NULL, bno, 0, &bp,
 				XFS_BMAP_BTREE_REF)))
 			goto error_norelse;
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 	}
 	if (bp_release) {
 		bp_release = 0;
@@ -6356,7 +6428,7 @@ xfs_bmap_count_blocks(
 	int			whichfork,	/* data or attr fork */
 	int			*count)		/* out: count of blocks */
 {
-	xfs_bmbt_block_t	*block;	/* current btree block */
+	struct xfs_btree_block	*block;	/* current btree block */
 	xfs_fsblock_t		bno;	/* block # of "block" */
 	xfs_ifork_t		*ifp;	/* fork structure */
 	int			level;	/* btree level, for checking */
@@ -6379,7 +6451,7 @@ xfs_bmap_count_blocks(
 	block = ifp->if_broot;
 	level = be16_to_cpu(block->bb_level);
 	ASSERT(level > 0);
-	pp = XFS_BMAP_BROOT_PTR_ADDR(block, 1, ifp->if_broot_bytes);
+	pp = XFS_BMAP_BROOT_PTR_ADDR(mp, block, 1, ifp->if_broot_bytes);
 	bno = be64_to_cpu(*pp);
 	ASSERT(bno != NULLDFSBNO);
 	ASSERT(XFS_FSB_TO_AGNO(mp, bno) < mp->m_sb.sb_agcount);
@@ -6413,29 +6485,29 @@ xfs_bmap_count_tree(
 	__be64			*pp;
 	xfs_fsblock_t           bno = blockno;
 	xfs_fsblock_t		nextbno;
-	xfs_bmbt_block_t        *block, *nextblock;
+	struct xfs_btree_block	*block, *nextblock;
 	int			numrecs;
 
 	if ((error = xfs_btree_read_bufl(mp, tp, bno, 0, &bp, XFS_BMAP_BTREE_REF)))
 		return error;
 	*count += 1;
-	block = XFS_BUF_TO_BMBT_BLOCK(bp);
+	block = XFS_BUF_TO_BLOCK(bp);
 
 	if (--level) {
 		/* Not at node above leafs, count this level of nodes */
-		nextbno = be64_to_cpu(block->bb_rightsib);
+		nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
 		while (nextbno != NULLFSBLOCK) {
 			if ((error = xfs_btree_read_bufl(mp, tp, nextbno,
 				0, &nbp, XFS_BMAP_BTREE_REF)))
 				return error;
 			*count += 1;
-			nextblock = XFS_BUF_TO_BMBT_BLOCK(nbp);
-			nextbno = be64_to_cpu(nextblock->bb_rightsib);
+			nextblock = XFS_BUF_TO_BLOCK(nbp);
+			nextbno = be64_to_cpu(nextblock->bb_u.l.bb_rightsib);
 			xfs_trans_brelse(tp, nbp);
 		}
 
 		/* Dive to the next level */
-		pp = XFS_BTREE_PTR_ADDR(xfs_bmbt, block, 1, mp->m_bmap_dmxr[1]);
+		pp = XFS_BMBT_PTR_ADDR(mp, block, 1, mp->m_bmap_dmxr[1]);
 		bno = be64_to_cpu(*pp);
 		if (unlikely((error =
 		     xfs_bmap_count_tree(mp, tp, ifp, bno, level, count)) < 0)) {
@@ -6448,9 +6520,9 @@ xfs_bmap_count_tree(
 	} else {
 		/* count all level 1 nodes and their leaves */
 		for (;;) {
-			nextbno = be64_to_cpu(block->bb_rightsib);
+			nextbno = be64_to_cpu(block->bb_u.l.bb_rightsib);
 			numrecs = be16_to_cpu(block->bb_numrecs);
-			xfs_bmap_disk_count_leaves(0, block, numrecs, count);
+			xfs_bmap_disk_count_leaves(mp, block, numrecs, count);
 			xfs_trans_brelse(tp, bp);
 			if (nextbno == NULLFSBLOCK)
 				break;
@@ -6459,7 +6531,7 @@ xfs_bmap_count_tree(
 				XFS_BMAP_BTREE_REF)))
 				return error;
 			*count += 1;
-			block = XFS_BUF_TO_BMBT_BLOCK(bp);
+			block = XFS_BUF_TO_BLOCK(bp);
 		}
 	}
 	return 0;
@@ -6489,8 +6561,8 @@ xfs_bmap_count_leaves(
  */
 STATIC void
 xfs_bmap_disk_count_leaves(
-	xfs_extnum_t		idx,
-	xfs_bmbt_block_t	*block,
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*block,
 	int			numrecs,
 	int			*count)
 {
@@ -6498,7 +6570,7 @@ xfs_bmap_disk_count_leaves(
 	xfs_bmbt_rec_t	*frp;
 
 	for (b = 1; b <= numrecs; b++) {
-		frp = XFS_BTREE_REC_ADDR(xfs_bmbt, block, idx + b);
+		frp = XFS_BMBT_REC_ADDR(mp, block, b);
 		*count += xfs_bmbt_disk_get_blockcount(frp);
 	}
 }
diff --git a/fs/xfs/xfs_bmap.h b/fs/xfs/xfs_bmap.h
index 9f3e3a836d1..284571c05ed 100644
--- a/fs/xfs/xfs_bmap.h
+++ b/fs/xfs/xfs_bmap.h
@@ -137,9 +137,7 @@ typedef struct xfs_bmalloca {
 	char			conv;	/* overwriting unwritten extents */
 } xfs_bmalloca_t;
 
-#ifdef __KERNEL__
-
-#if defined(XFS_BMAP_TRACE)
+#if defined(__KERNEL__) && defined(XFS_BMAP_TRACE)
 /*
  * Trace operations for bmap extent tracing
  */
@@ -163,9 +161,12 @@ xfs_bmap_trace_exlist(
 	int			whichfork);	/* data or attr fork */
 #define	XFS_BMAP_TRACE_EXLIST(ip,c,w)	\
 	xfs_bmap_trace_exlist(__func__,ip,c,w)
-#else
+
+#else	/* __KERNEL__ && XFS_BMAP_TRACE */
+
 #define	XFS_BMAP_TRACE_EXLIST(ip,c,w)
-#endif
+
+#endif	/* __KERNEL__ && XFS_BMAP_TRACE */
 
 /*
  * Convert inode from non-attributed to attributed.
@@ -206,20 +207,6 @@ xfs_bmap_compute_maxlevels(
 	int			whichfork);	/* data or attr fork */
 
 /*
- * Routine to be called at transaction's end by xfs_bmapi, xfs_bunmapi
- * caller.  Frees all the extents that need freeing, which must be done
- * last due to locking considerations.
- *
- * Return 1 if the given transaction was committed and a new one allocated,
- * and 0 otherwise.
- */
-int						/* error */
-xfs_bmap_finish(
-	struct xfs_trans	**tp,		/* transaction pointer addr */
-	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
-	int			*committed);	/* xact committed or not */
-
-/*
  * Returns the file-relative block number of the first unused block in the file.
  * This is the lowest-address hole if the file has holes, else the first block
  * past the end of file.
@@ -344,14 +331,43 @@ xfs_bunmapi(
 	int			*done);		/* set if not done yet */
 
 /*
- * Fcntl interface to xfs_bmapi.
+ * Check an extent list, which has just been read, for
+ * any bit in the extent flag field.
+ */
+int
+xfs_check_nostate_extents(
+	struct xfs_ifork	*ifp,
+	xfs_extnum_t		idx,
+	xfs_extnum_t		num);
+
+#ifdef __KERNEL__
+
+/*
+ * Routine to be called at transaction's end by xfs_bmapi, xfs_bunmapi
+ * caller.  Frees all the extents that need freeing, which must be done
+ * last due to locking considerations.
+ *
+ * Return 1 if the given transaction was committed and a new one allocated,
+ * and 0 otherwise.
+ */
+int						/* error */
+xfs_bmap_finish(
+	struct xfs_trans	**tp,		/* transaction pointer addr */
+	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
+	int			*committed);	/* xact committed or not */
+
+/* bmap to userspace formatter - copy to user & advance pointer */
+typedef int (*xfs_bmap_format_t)(void **, struct getbmapx *, int *);
+
+/*
+ * Get inode's extents as described in bmv, and format for output.
  */
 int						/* error code */
 xfs_getbmap(
 	xfs_inode_t		*ip,
-	struct getbmap		*bmv,		/* user bmap structure */
-	void			__user *ap,	/* pointer to user's array */
-	int			iflags);	/* interface flags */
+	struct getbmapx		*bmv,		/* user bmap structure */
+	xfs_bmap_format_t	formatter,	/* format to user */
+	void			*arg);		/* formatter arg */
 
 /*
  * Check if the endoff is outside the last extent. If so the caller will grow
@@ -375,16 +391,6 @@ xfs_bmap_count_blocks(
 	int			*count);
 
 /*
- * Check an extent list, which has just been read, for
- * any bit in the extent flag field.
- */
-int
-xfs_check_nostate_extents(
-	struct xfs_ifork	*ifp,
-	xfs_extnum_t		idx,
-	xfs_extnum_t		num);
-
-/*
  * Search the extent records for the entry containing block bno.
  * If bno lies in a hole, point to the next entry.  If bno lies
  * past eof, *eofp will be set, and *prevp will contain the last
diff --git a/fs/xfs/xfs_bmap_btree.c b/fs/xfs/xfs_bmap_btree.c
index 23efad29a5c..8f1ec73725d 100644
--- a/fs/xfs/xfs_bmap_btree.c
+++ b/fs/xfs/xfs_bmap_btree.c
@@ -37,1406 +37,13 @@
 #include "xfs_inode_item.h"
 #include "xfs_alloc.h"
 #include "xfs_btree.h"
+#include "xfs_btree_trace.h"
 #include "xfs_ialloc.h"
 #include "xfs_itable.h"
 #include "xfs_bmap.h"
 #include "xfs_error.h"
 #include "xfs_quota.h"
 
-#if defined(XFS_BMBT_TRACE)
-ktrace_t	*xfs_bmbt_trace_buf;
-#endif
-
-/*
- * Prototypes for internal btree functions.
- */
-
-
-STATIC int xfs_bmbt_killroot(xfs_btree_cur_t *);
-STATIC void xfs_bmbt_log_keys(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC void xfs_bmbt_log_ptrs(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC int xfs_bmbt_lshift(xfs_btree_cur_t *, int, int *);
-STATIC int xfs_bmbt_rshift(xfs_btree_cur_t *, int, int *);
-STATIC int xfs_bmbt_split(xfs_btree_cur_t *, int, xfs_fsblock_t *,
-		__uint64_t *, xfs_btree_cur_t **, int *);
-STATIC int xfs_bmbt_updkey(xfs_btree_cur_t *, xfs_bmbt_key_t *, int);
-
-
-#if defined(XFS_BMBT_TRACE)
-
-static char	ARGS[] = "args";
-static char	ENTRY[] = "entry";
-static char	ERROR[] = "error";
-#undef EXIT
-static char	EXIT[] = "exit";
-
-/*
- * Add a trace buffer entry for the arguments given to the routine,
- * generic form.
- */
-STATIC void
-xfs_bmbt_trace_enter(
-	const char	*func,
-	xfs_btree_cur_t	*cur,
-	char		*s,
-	int		type,
-	int		line,
-	__psunsigned_t	a0,
-	__psunsigned_t	a1,
-	__psunsigned_t	a2,
-	__psunsigned_t	a3,
-	__psunsigned_t	a4,
-	__psunsigned_t	a5,
-	__psunsigned_t	a6,
-	__psunsigned_t	a7,
-	__psunsigned_t	a8,
-	__psunsigned_t	a9,
-	__psunsigned_t	a10)
-{
-	xfs_inode_t	*ip;
-	int		whichfork;
-
-	ip = cur->bc_private.b.ip;
-	whichfork = cur->bc_private.b.whichfork;
-	ktrace_enter(xfs_bmbt_trace_buf,
-		(void *)((__psint_t)type | (whichfork << 8) | (line << 16)),
-		(void *)func, (void *)s, (void *)ip, (void *)cur,
-		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
-		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
-		(void *)a8, (void *)a9, (void *)a10);
-	ASSERT(ip->i_btrace);
-	ktrace_enter(ip->i_btrace,
-		(void *)((__psint_t)type | (whichfork << 8) | (line << 16)),
-		(void *)func, (void *)s, (void *)ip, (void *)cur,
-		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
-		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
-		(void *)a8, (void *)a9, (void *)a10);
-}
-/*
- * Add a trace buffer entry for arguments, for a buffer & 1 integer arg.
- */
-STATIC void
-xfs_bmbt_trace_argbi(
-	const char	*func,
-	xfs_btree_cur_t	*cur,
-	xfs_buf_t	*b,
-	int		i,
-	int		line)
-{
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGBI, line,
-		(__psunsigned_t)b, i, 0, 0,
-		0, 0, 0, 0,
-		0, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for arguments, for a buffer & 2 integer args.
- */
-STATIC void
-xfs_bmbt_trace_argbii(
-	const char	*func,
-	xfs_btree_cur_t	*cur,
-	xfs_buf_t	*b,
-	int		i0,
-	int		i1,
-	int		line)
-{
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGBII, line,
-		(__psunsigned_t)b, i0, i1, 0,
-		0, 0, 0, 0,
-		0, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for arguments, for 3 block-length args
- * and an integer arg.
- */
-STATIC void
-xfs_bmbt_trace_argfffi(
-	const char		*func,
-	xfs_btree_cur_t		*cur,
-	xfs_dfiloff_t		o,
-	xfs_dfsbno_t		b,
-	xfs_dfilblks_t		i,
-	int			j,
-	int			line)
-{
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGFFFI, line,
-		o >> 32, (int)o, b >> 32, (int)b,
-		i >> 32, (int)i, (int)j, 0,
-		0, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for arguments, for one integer arg.
- */
-STATIC void
-xfs_bmbt_trace_argi(
-	const char	*func,
-	xfs_btree_cur_t	*cur,
-	int		i,
-	int		line)
-{
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGI, line,
-		i, 0, 0, 0,
-		0, 0, 0, 0,
-		0, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for arguments, for int, fsblock, key.
- */
-STATIC void
-xfs_bmbt_trace_argifk(
-	const char		*func,
-	xfs_btree_cur_t		*cur,
-	int			i,
-	xfs_fsblock_t		f,
-	xfs_dfiloff_t		o,
-	int			line)
-{
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGIFK, line,
-		i, (xfs_dfsbno_t)f >> 32, (int)f, o >> 32,
-		(int)o, 0, 0, 0,
-		0, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for arguments, for int, fsblock, rec.
- */
-STATIC void
-xfs_bmbt_trace_argifr(
-	const char		*func,
-	xfs_btree_cur_t		*cur,
-	int			i,
-	xfs_fsblock_t		f,
-	xfs_bmbt_rec_t		*r,
-	int			line)
-{
-	xfs_dfsbno_t		b;
-	xfs_dfilblks_t		c;
-	xfs_dfsbno_t		d;
-	xfs_dfiloff_t		o;
-	xfs_bmbt_irec_t		s;
-
-	d = (xfs_dfsbno_t)f;
-	xfs_bmbt_disk_get_all(r, &s);
-	o = (xfs_dfiloff_t)s.br_startoff;
-	b = (xfs_dfsbno_t)s.br_startblock;
-	c = s.br_blockcount;
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGIFR, line,
-		i, d >> 32, (int)d, o >> 32,
-		(int)o, b >> 32, (int)b, c >> 32,
-		(int)c, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for arguments, for int, key.
- */
-STATIC void
-xfs_bmbt_trace_argik(
-	const char		*func,
-	xfs_btree_cur_t		*cur,
-	int			i,
-	xfs_bmbt_key_t		*k,
-	int			line)
-{
-	xfs_dfiloff_t		o;
-
-	o = be64_to_cpu(k->br_startoff);
-	xfs_bmbt_trace_enter(func, cur, ARGS, XFS_BMBT_KTRACE_ARGIFK, line,
-		i, o >> 32, (int)o, 0,
-		0, 0, 0, 0,
-		0, 0, 0);
-}
-
-/*
- * Add a trace buffer entry for the cursor/operation.
- */
-STATIC void
-xfs_bmbt_trace_cursor(
-	const char	*func,
-	xfs_btree_cur_t	*cur,
-	char		*s,
-	int		line)
-{
-	xfs_bmbt_rec_host_t	r;
-
-	xfs_bmbt_set_all(&r, &cur->bc_rec.b);
-	xfs_bmbt_trace_enter(func, cur, s, XFS_BMBT_KTRACE_CUR, line,
-		(cur->bc_nlevels << 24) | (cur->bc_private.b.flags << 16) |
-		cur->bc_private.b.allocated,
-		r.l0 >> 32, (int)r.l0,
-		r.l1 >> 32, (int)r.l1,
-		(unsigned long)cur->bc_bufs[0], (unsigned long)cur->bc_bufs[1],
-		(unsigned long)cur->bc_bufs[2], (unsigned long)cur->bc_bufs[3],
-		(cur->bc_ptrs[0] << 16) | cur->bc_ptrs[1],
-		(cur->bc_ptrs[2] << 16) | cur->bc_ptrs[3]);
-}
-
-#define	XFS_BMBT_TRACE_ARGBI(c,b,i)	\
-	xfs_bmbt_trace_argbi(__func__, c, b, i, __LINE__)
-#define	XFS_BMBT_TRACE_ARGBII(c,b,i,j)	\
-	xfs_bmbt_trace_argbii(__func__, c, b, i, j, __LINE__)
-#define	XFS_BMBT_TRACE_ARGFFFI(c,o,b,i,j)	\
-	xfs_bmbt_trace_argfffi(__func__, c, o, b, i, j, __LINE__)
-#define	XFS_BMBT_TRACE_ARGI(c,i)	\
-	xfs_bmbt_trace_argi(__func__, c, i, __LINE__)
-#define	XFS_BMBT_TRACE_ARGIFK(c,i,f,s)	\
-	xfs_bmbt_trace_argifk(__func__, c, i, f, s, __LINE__)
-#define	XFS_BMBT_TRACE_ARGIFR(c,i,f,r)	\
-	xfs_bmbt_trace_argifr(__func__, c, i, f, r, __LINE__)
-#define	XFS_BMBT_TRACE_ARGIK(c,i,k)	\
-	xfs_bmbt_trace_argik(__func__, c, i, k, __LINE__)
-#define	XFS_BMBT_TRACE_CURSOR(c,s)	\
-	xfs_bmbt_trace_cursor(__func__, c, s, __LINE__)
-#else
-#define	XFS_BMBT_TRACE_ARGBI(c,b,i)
-#define	XFS_BMBT_TRACE_ARGBII(c,b,i,j)
-#define	XFS_BMBT_TRACE_ARGFFFI(c,o,b,i,j)
-#define	XFS_BMBT_TRACE_ARGI(c,i)
-#define	XFS_BMBT_TRACE_ARGIFK(c,i,f,s)
-#define	XFS_BMBT_TRACE_ARGIFR(c,i,f,r)
-#define	XFS_BMBT_TRACE_ARGIK(c,i,k)
-#define	XFS_BMBT_TRACE_CURSOR(c,s)
-#endif	/* XFS_BMBT_TRACE */
-
-
-/*
- * Internal functions.
- */
-
-/*
- * Delete record pointed to by cur/level.
- */
-STATIC int					/* error */
-xfs_bmbt_delrec(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	int			*stat)		/* success/failure */
-{
-	xfs_bmbt_block_t	*block;		/* bmap btree block */
-	xfs_fsblock_t		bno;		/* fs-relative block number */
-	xfs_buf_t		*bp;		/* buffer for block */
-	int			error;		/* error return value */
-	int			i;		/* loop counter */
-	int			j;		/* temp state */
-	xfs_bmbt_key_t		key;		/* bmap btree key */
-	xfs_bmbt_key_t		*kp=NULL;	/* pointer to bmap btree key */
-	xfs_fsblock_t		lbno;		/* left sibling block number */
-	xfs_buf_t		*lbp;		/* left buffer pointer */
-	xfs_bmbt_block_t	*left;		/* left btree block */
-	xfs_bmbt_key_t		*lkp;		/* left btree key */
-	xfs_bmbt_ptr_t		*lpp;		/* left address pointer */
-	int			lrecs=0;	/* left record count */
-	xfs_bmbt_rec_t		*lrp;		/* left record pointer */
-	xfs_mount_t		*mp;		/* file system mount point */
-	xfs_bmbt_ptr_t		*pp;		/* pointer to bmap block addr */
-	int			ptr;		/* key/record index */
-	xfs_fsblock_t		rbno;		/* right sibling block number */
-	xfs_buf_t		*rbp;		/* right buffer pointer */
-	xfs_bmbt_block_t	*right;		/* right btree block */
-	xfs_bmbt_key_t		*rkp;		/* right btree key */
-	xfs_bmbt_rec_t		*rp;		/* pointer to bmap btree rec */
-	xfs_bmbt_ptr_t		*rpp;		/* right address pointer */
-	xfs_bmbt_block_t	*rrblock;	/* right-right btree block */
-	xfs_buf_t		*rrbp;		/* right-right buffer pointer */
-	int			rrecs=0;	/* right record count */
-	xfs_bmbt_rec_t		*rrp;		/* right record pointer */
-	xfs_btree_cur_t		*tcur;		/* temporary btree cursor */
-	int			numrecs;	/* temporary numrec count */
-	int			numlrecs, numrrecs;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGI(cur, level);
-	ptr = cur->bc_ptrs[level];
-	tcur = NULL;
-	if (ptr == 0) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	block = xfs_bmbt_get_block(cur, level, &bp);
-	numrecs = be16_to_cpu(block->bb_numrecs);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, block, level, bp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		goto error0;
-	}
-#endif
-	if (ptr > numrecs) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	XFS_STATS_INC(xs_bmbt_delrec);
-	if (level > 0) {
-		kp = XFS_BMAP_KEY_IADDR(block, 1, cur);
-		pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
-#ifdef DEBUG
-		for (i = ptr; i < numrecs; i++) {
-			if ((error = xfs_btree_check_lptr_disk(cur, pp[i], level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				goto error0;
-			}
-		}
-#endif
-		if (ptr < numrecs) {
-			memmove(&kp[ptr - 1], &kp[ptr],
-				(numrecs - ptr) * sizeof(*kp));
-			memmove(&pp[ptr - 1], &pp[ptr],
-				(numrecs - ptr) * sizeof(*pp));
-			xfs_bmbt_log_ptrs(cur, bp, ptr, numrecs - 1);
-			xfs_bmbt_log_keys(cur, bp, ptr, numrecs - 1);
-		}
-	} else {
-		rp = XFS_BMAP_REC_IADDR(block, 1, cur);
-		if (ptr < numrecs) {
-			memmove(&rp[ptr - 1], &rp[ptr],
-				(numrecs - ptr) * sizeof(*rp));
-			xfs_bmbt_log_recs(cur, bp, ptr, numrecs - 1);
-		}
-		if (ptr == 1) {
-			key.br_startoff =
-				cpu_to_be64(xfs_bmbt_disk_get_startoff(rp));
-			kp = &key;
-		}
-	}
-	numrecs--;
-	block->bb_numrecs = cpu_to_be16(numrecs);
-	xfs_bmbt_log_block(cur, bp, XFS_BB_NUMRECS);
-	/*
-	 * We're at the root level.
-	 * First, shrink the root block in-memory.
-	 * Try to get rid of the next level down.
-	 * If we can't then there's nothing left to do.
-	 */
-	if (level == cur->bc_nlevels - 1) {
-		xfs_iroot_realloc(cur->bc_private.b.ip, -1,
-			cur->bc_private.b.whichfork);
-		if ((error = xfs_bmbt_killroot(cur))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		if (level > 0 && (error = xfs_bmbt_decrement(cur, level, &j))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 1;
-		return 0;
-	}
-	if (ptr == 1 && (error = xfs_bmbt_updkey(cur, kp, level + 1))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		goto error0;
-	}
-	if (numrecs >= XFS_BMAP_BLOCK_IMINRECS(level, cur)) {
-		if (level > 0 && (error = xfs_bmbt_decrement(cur, level, &j))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 1;
-		return 0;
-	}
-	rbno = be64_to_cpu(block->bb_rightsib);
-	lbno = be64_to_cpu(block->bb_leftsib);
-	/*
-	 * One child of root, need to get a chance to copy its contents
-	 * into the root and delete it. Can't go up to next level,
-	 * there's nothing to delete there.
-	 */
-	if (lbno == NULLFSBLOCK && rbno == NULLFSBLOCK &&
-	    level == cur->bc_nlevels - 2) {
-		if ((error = xfs_bmbt_killroot(cur))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		if (level > 0 && (error = xfs_bmbt_decrement(cur, level, &i))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 1;
-		return 0;
-	}
-	ASSERT(rbno != NULLFSBLOCK || lbno != NULLFSBLOCK);
-	if ((error = xfs_btree_dup_cursor(cur, &tcur))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		goto error0;
-	}
-	bno = NULLFSBLOCK;
-	if (rbno != NULLFSBLOCK) {
-		i = xfs_btree_lastrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_bmbt_increment(tcur, level, &i))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		i = xfs_btree_lastrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		rbp = tcur->bc_bufs[level];
-		right = XFS_BUF_TO_BMBT_BLOCK(rbp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lblock(cur, right, level, rbp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-#endif
-		bno = be64_to_cpu(right->bb_leftsib);
-		if (be16_to_cpu(right->bb_numrecs) - 1 >=
-		    XFS_BMAP_BLOCK_IMINRECS(level, cur)) {
-			if ((error = xfs_bmbt_lshift(tcur, level, &i))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				goto error0;
-			}
-			if (i) {
-				ASSERT(be16_to_cpu(block->bb_numrecs) >=
-				       XFS_BMAP_BLOCK_IMINRECS(level, tcur));
-				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-				tcur = NULL;
-				if (level > 0) {
-					if ((error = xfs_bmbt_decrement(cur,
-							level, &i))) {
-						XFS_BMBT_TRACE_CURSOR(cur,
-							ERROR);
-						goto error0;
-					}
-				}
-				XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-				*stat = 1;
-				return 0;
-			}
-		}
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		if (lbno != NULLFSBLOCK) {
-			i = xfs_btree_firstrec(tcur, level);
-			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-			if ((error = xfs_bmbt_decrement(tcur, level, &i))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				goto error0;
-			}
-			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		}
-	}
-	if (lbno != NULLFSBLOCK) {
-		i = xfs_btree_firstrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		/*
-		 * decrement to last in block
-		 */
-		if ((error = xfs_bmbt_decrement(tcur, level, &i))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		i = xfs_btree_firstrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		lbp = tcur->bc_bufs[level];
-		left = XFS_BUF_TO_BMBT_BLOCK(lbp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lblock(cur, left, level, lbp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-#endif
-		bno = be64_to_cpu(left->bb_rightsib);
-		if (be16_to_cpu(left->bb_numrecs) - 1 >=
-		    XFS_BMAP_BLOCK_IMINRECS(level, cur)) {
-			if ((error = xfs_bmbt_rshift(tcur, level, &i))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				goto error0;
-			}
-			if (i) {
-				ASSERT(be16_to_cpu(block->bb_numrecs) >=
-				       XFS_BMAP_BLOCK_IMINRECS(level, tcur));
-				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-				tcur = NULL;
-				if (level == 0)
-					cur->bc_ptrs[0]++;
-				XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-				*stat = 1;
-				return 0;
-			}
-		}
-		lrecs = be16_to_cpu(left->bb_numrecs);
-	}
-	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-	tcur = NULL;
-	mp = cur->bc_mp;
-	ASSERT(bno != NULLFSBLOCK);
-	if (lbno != NULLFSBLOCK &&
-	    lrecs + be16_to_cpu(block->bb_numrecs) <= XFS_BMAP_BLOCK_IMAXRECS(level, cur)) {
-		rbno = bno;
-		right = block;
-		rbp = bp;
-		if ((error = xfs_btree_read_bufl(mp, cur->bc_tp, lbno, 0, &lbp,
-				XFS_BMAP_BTREE_REF))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		left = XFS_BUF_TO_BMBT_BLOCK(lbp);
-		if ((error = xfs_btree_check_lblock(cur, left, level, lbp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-	} else if (rbno != NULLFSBLOCK &&
-		   rrecs + be16_to_cpu(block->bb_numrecs) <=
-		   XFS_BMAP_BLOCK_IMAXRECS(level, cur)) {
-		lbno = bno;
-		left = block;
-		lbp = bp;
-		if ((error = xfs_btree_read_bufl(mp, cur->bc_tp, rbno, 0, &rbp,
-				XFS_BMAP_BTREE_REF))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		right = XFS_BUF_TO_BMBT_BLOCK(rbp);
-		if ((error = xfs_btree_check_lblock(cur, right, level, rbp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		lrecs = be16_to_cpu(left->bb_numrecs);
-	} else {
-		if (level > 0 && (error = xfs_bmbt_decrement(cur, level, &i))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 1;
-		return 0;
-	}
-	numlrecs = be16_to_cpu(left->bb_numrecs);
-	numrrecs = be16_to_cpu(right->bb_numrecs);
-	if (level > 0) {
-		lkp = XFS_BMAP_KEY_IADDR(left, numlrecs + 1, cur);
-		lpp = XFS_BMAP_PTR_IADDR(left, numlrecs + 1, cur);
-		rkp = XFS_BMAP_KEY_IADDR(right, 1, cur);
-		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = 0; i < numrrecs; i++) {
-			if ((error = xfs_btree_check_lptr_disk(cur, rpp[i], level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				goto error0;
-			}
-		}
-#endif
-		memcpy(lkp, rkp, numrrecs * sizeof(*lkp));
-		memcpy(lpp, rpp, numrrecs * sizeof(*lpp));
-		xfs_bmbt_log_keys(cur, lbp, numlrecs + 1, numlrecs + numrrecs);
-		xfs_bmbt_log_ptrs(cur, lbp, numlrecs + 1, numlrecs + numrrecs);
-	} else {
-		lrp = XFS_BMAP_REC_IADDR(left, numlrecs + 1, cur);
-		rrp = XFS_BMAP_REC_IADDR(right, 1, cur);
-		memcpy(lrp, rrp, numrrecs * sizeof(*lrp));
-		xfs_bmbt_log_recs(cur, lbp, numlrecs + 1, numlrecs + numrrecs);
-	}
-	be16_add_cpu(&left->bb_numrecs, numrrecs);
-	left->bb_rightsib = right->bb_rightsib;
-	xfs_bmbt_log_block(cur, lbp, XFS_BB_RIGHTSIB | XFS_BB_NUMRECS);
-	if (be64_to_cpu(left->bb_rightsib) != NULLDFSBNO) {
-		if ((error = xfs_btree_read_bufl(mp, cur->bc_tp,
-				be64_to_cpu(left->bb_rightsib),
-				0, &rrbp, XFS_BMAP_BTREE_REF))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		rrblock = XFS_BUF_TO_BMBT_BLOCK(rrbp);
-		if ((error = xfs_btree_check_lblock(cur, rrblock, level, rrbp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			goto error0;
-		}
-		rrblock->bb_leftsib = cpu_to_be64(lbno);
-		xfs_bmbt_log_block(cur, rrbp, XFS_BB_LEFTSIB);
-	}
-	xfs_bmap_add_free(XFS_DADDR_TO_FSB(mp, XFS_BUF_ADDR(rbp)), 1,
-		cur->bc_private.b.flist, mp);
-	cur->bc_private.b.ip->i_d.di_nblocks--;
-	xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip, XFS_ILOG_CORE);
-	XFS_TRANS_MOD_DQUOT_BYINO(mp, cur->bc_tp, cur->bc_private.b.ip,
-			XFS_TRANS_DQ_BCOUNT, -1L);
-	xfs_trans_binval(cur->bc_tp, rbp);
-	if (bp != lbp) {
-		cur->bc_bufs[level] = lbp;
-		cur->bc_ptrs[level] += lrecs;
-		cur->bc_ra[level] = 0;
-	} else if ((error = xfs_bmbt_increment(cur, level + 1, &i))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		goto error0;
-	}
-	if (level > 0)
-		cur->bc_ptrs[level]--;
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 2;
-	return 0;
-
-error0:
-	if (tcur)
-		xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
-	return error;
-}
-
-/*
- * Insert one record/level.  Return information to the caller
- * allowing the next level up to proceed if necessary.
- */
-STATIC int					/* error */
-xfs_bmbt_insrec(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	xfs_fsblock_t		*bnop,
-	xfs_bmbt_rec_t		*recp,
-	xfs_btree_cur_t		**curp,
-	int			*stat)		/* no-go/done/continue */
-{
-	xfs_bmbt_block_t	*block;		/* bmap btree block */
-	xfs_buf_t		*bp;		/* buffer for block */
-	int			error;		/* error return value */
-	int			i;		/* loop index */
-	xfs_bmbt_key_t		key;		/* bmap btree key */
-	xfs_bmbt_key_t		*kp=NULL;	/* pointer to bmap btree key */
-	int			logflags;	/* inode logging flags */
-	xfs_fsblock_t		nbno;		/* new block number */
-	struct xfs_btree_cur	*ncur;		/* new btree cursor */
-	__uint64_t		startoff;	/* new btree key value */
-	xfs_bmbt_rec_t		nrec;		/* new record count */
-	int			optr;		/* old key/record index */
-	xfs_bmbt_ptr_t		*pp;		/* pointer to bmap block addr */
-	int			ptr;		/* key/record index */
-	xfs_bmbt_rec_t		*rp=NULL;	/* pointer to bmap btree rec */
-	int			numrecs;
-
-	ASSERT(level < cur->bc_nlevels);
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGIFR(cur, level, *bnop, recp);
-	ncur = NULL;
-	key.br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(recp));
-	optr = ptr = cur->bc_ptrs[level];
-	if (ptr == 0) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	XFS_STATS_INC(xs_bmbt_insrec);
-	block = xfs_bmbt_get_block(cur, level, &bp);
-	numrecs = be16_to_cpu(block->bb_numrecs);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, block, level, bp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	if (ptr <= numrecs) {
-		if (level == 0) {
-			rp = XFS_BMAP_REC_IADDR(block, ptr, cur);
-			xfs_btree_check_rec(XFS_BTNUM_BMAP, recp, rp);
-		} else {
-			kp = XFS_BMAP_KEY_IADDR(block, ptr, cur);
-			xfs_btree_check_key(XFS_BTNUM_BMAP, &key, kp);
-		}
-	}
-#endif
-	nbno = NULLFSBLOCK;
-	if (numrecs == XFS_BMAP_BLOCK_IMAXRECS(level, cur)) {
-		if (numrecs < XFS_BMAP_BLOCK_DMAXRECS(level, cur)) {
-			/*
-			 * A root block, that can be made bigger.
-			 */
-			xfs_iroot_realloc(cur->bc_private.b.ip, 1,
-				cur->bc_private.b.whichfork);
-			block = xfs_bmbt_get_block(cur, level, &bp);
-		} else if (level == cur->bc_nlevels - 1) {
-			if ((error = xfs_bmbt_newroot(cur, &logflags, stat)) ||
-			    *stat == 0) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-			xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
-				logflags);
-			block = xfs_bmbt_get_block(cur, level, &bp);
-		} else {
-			if ((error = xfs_bmbt_rshift(cur, level, &i))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-			if (i) {
-				/* nothing */
-			} else {
-				if ((error = xfs_bmbt_lshift(cur, level, &i))) {
-					XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-					return error;
-				}
-				if (i) {
-					optr = ptr = cur->bc_ptrs[level];
-				} else {
-					if ((error = xfs_bmbt_split(cur, level,
-							&nbno, &startoff, &ncur,
-							&i))) {
-						XFS_BMBT_TRACE_CURSOR(cur,
-							ERROR);
-						return error;
-					}
-					if (i) {
-						block = xfs_bmbt_get_block(
-							    cur, level, &bp);
-#ifdef DEBUG
-						if ((error =
-						    xfs_btree_check_lblock(cur,
-							    block, level, bp))) {
-							XFS_BMBT_TRACE_CURSOR(
-								cur, ERROR);
-							return error;
-						}
-#endif
-						ptr = cur->bc_ptrs[level];
-						xfs_bmbt_disk_set_allf(&nrec,
-							startoff, 0, 0,
-							XFS_EXT_NORM);
-					} else {
-						XFS_BMBT_TRACE_CURSOR(cur,
-							EXIT);
-						*stat = 0;
-						return 0;
-					}
-				}
-			}
-		}
-	}
-	numrecs = be16_to_cpu(block->bb_numrecs);
-	if (level > 0) {
-		kp = XFS_BMAP_KEY_IADDR(block, 1, cur);
-		pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
-#ifdef DEBUG
-		for (i = numrecs; i >= ptr; i--) {
-			if ((error = xfs_btree_check_lptr_disk(cur, pp[i - 1],
-					level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-		}
-#endif
-		memmove(&kp[ptr], &kp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*kp));
-		memmove(&pp[ptr], &pp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*pp));
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lptr(cur, *bnop, level))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-#endif
-		kp[ptr - 1] = key;
-		pp[ptr - 1] = cpu_to_be64(*bnop);
-		numrecs++;
-		block->bb_numrecs = cpu_to_be16(numrecs);
-		xfs_bmbt_log_keys(cur, bp, ptr, numrecs);
-		xfs_bmbt_log_ptrs(cur, bp, ptr, numrecs);
-	} else {
-		rp = XFS_BMAP_REC_IADDR(block, 1, cur);
-		memmove(&rp[ptr], &rp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*rp));
-		rp[ptr - 1] = *recp;
-		numrecs++;
-		block->bb_numrecs = cpu_to_be16(numrecs);
-		xfs_bmbt_log_recs(cur, bp, ptr, numrecs);
-	}
-	xfs_bmbt_log_block(cur, bp, XFS_BB_NUMRECS);
-#ifdef DEBUG
-	if (ptr < numrecs) {
-		if (level == 0)
-			xfs_btree_check_rec(XFS_BTNUM_BMAP, rp + ptr - 1,
-				rp + ptr);
-		else
-			xfs_btree_check_key(XFS_BTNUM_BMAP, kp + ptr - 1,
-				kp + ptr);
-	}
-#endif
-	if (optr == 1 && (error = xfs_bmbt_updkey(cur, &key, level + 1))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	*bnop = nbno;
-	if (nbno != NULLFSBLOCK) {
-		*recp = nrec;
-		*curp = ncur;
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 1;
-	return 0;
-}
-
-STATIC int
-xfs_bmbt_killroot(
-	xfs_btree_cur_t		*cur)
-{
-	xfs_bmbt_block_t	*block;
-	xfs_bmbt_block_t	*cblock;
-	xfs_buf_t		*cbp;
-	xfs_bmbt_key_t		*ckp;
-	xfs_bmbt_ptr_t		*cpp;
-#ifdef DEBUG
-	int			error;
-#endif
-	int			i;
-	xfs_bmbt_key_t		*kp;
-	xfs_inode_t		*ip;
-	xfs_ifork_t		*ifp;
-	int			level;
-	xfs_bmbt_ptr_t		*pp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	level = cur->bc_nlevels - 1;
-	ASSERT(level >= 1);
-	/*
-	 * Don't deal with the root block needs to be a leaf case.
-	 * We're just going to turn the thing back into extents anyway.
-	 */
-	if (level == 1) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		return 0;
-	}
-	block = xfs_bmbt_get_block(cur, level, &cbp);
-	/*
-	 * Give up if the root has multiple children.
-	 */
-	if (be16_to_cpu(block->bb_numrecs) != 1) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		return 0;
-	}
-	/*
-	 * Only do this if the next level will fit.
-	 * Then the data must be copied up to the inode,
-	 * instead of freeing the root you free the next level.
-	 */
-	cbp = cur->bc_bufs[level - 1];
-	cblock = XFS_BUF_TO_BMBT_BLOCK(cbp);
-	if (be16_to_cpu(cblock->bb_numrecs) > XFS_BMAP_BLOCK_DMAXRECS(level, cur)) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		return 0;
-	}
-	ASSERT(be64_to_cpu(cblock->bb_leftsib) == NULLDFSBNO);
-	ASSERT(be64_to_cpu(cblock->bb_rightsib) == NULLDFSBNO);
-	ip = cur->bc_private.b.ip;
-	ifp = XFS_IFORK_PTR(ip, cur->bc_private.b.whichfork);
-	ASSERT(XFS_BMAP_BLOCK_IMAXRECS(level, cur) ==
-	       XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes));
-	i = (int)(be16_to_cpu(cblock->bb_numrecs) - XFS_BMAP_BLOCK_IMAXRECS(level, cur));
-	if (i) {
-		xfs_iroot_realloc(ip, i, cur->bc_private.b.whichfork);
-		block = ifp->if_broot;
-	}
-	be16_add_cpu(&block->bb_numrecs, i);
-	ASSERT(block->bb_numrecs == cblock->bb_numrecs);
-	kp = XFS_BMAP_KEY_IADDR(block, 1, cur);
-	ckp = XFS_BMAP_KEY_IADDR(cblock, 1, cur);
-	memcpy(kp, ckp, be16_to_cpu(block->bb_numrecs) * sizeof(*kp));
-	pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
-	cpp = XFS_BMAP_PTR_IADDR(cblock, 1, cur);
-#ifdef DEBUG
-	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
-		if ((error = xfs_btree_check_lptr_disk(cur, cpp[i], level - 1))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-	}
-#endif
-	memcpy(pp, cpp, be16_to_cpu(block->bb_numrecs) * sizeof(*pp));
-	xfs_bmap_add_free(XFS_DADDR_TO_FSB(cur->bc_mp, XFS_BUF_ADDR(cbp)), 1,
-			cur->bc_private.b.flist, cur->bc_mp);
-	ip->i_d.di_nblocks--;
-	XFS_TRANS_MOD_DQUOT_BYINO(cur->bc_mp, cur->bc_tp, ip,
-			XFS_TRANS_DQ_BCOUNT, -1L);
-	xfs_trans_binval(cur->bc_tp, cbp);
-	cur->bc_bufs[level - 1] = NULL;
-	be16_add_cpu(&block->bb_level, -1);
-	xfs_trans_log_inode(cur->bc_tp, ip,
-		XFS_ILOG_CORE | XFS_ILOG_FBROOT(cur->bc_private.b.whichfork));
-	cur->bc_nlevels--;
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	return 0;
-}
-
-/*
- * Log key values from the btree block.
- */
-STATIC void
-xfs_bmbt_log_keys(
-	xfs_btree_cur_t	*cur,
-	xfs_buf_t	*bp,
-	int		kfirst,
-	int		klast)
-{
-	xfs_trans_t	*tp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGBII(cur, bp, kfirst, klast);
-	tp = cur->bc_tp;
-	if (bp) {
-		xfs_bmbt_block_t	*block;
-		int			first;
-		xfs_bmbt_key_t		*kp;
-		int			last;
-
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
-		kp = XFS_BMAP_KEY_DADDR(block, 1, cur);
-		first = (int)((xfs_caddr_t)&kp[kfirst - 1] - (xfs_caddr_t)block);
-		last = (int)(((xfs_caddr_t)&kp[klast] - 1) - (xfs_caddr_t)block);
-		xfs_trans_log_buf(tp, bp, first, last);
-	} else {
-		xfs_inode_t		 *ip;
-
-		ip = cur->bc_private.b.ip;
-		xfs_trans_log_inode(tp, ip,
-			XFS_ILOG_FBROOT(cur->bc_private.b.whichfork));
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-}
-
-/*
- * Log pointer values from the btree block.
- */
-STATIC void
-xfs_bmbt_log_ptrs(
-	xfs_btree_cur_t	*cur,
-	xfs_buf_t	*bp,
-	int		pfirst,
-	int		plast)
-{
-	xfs_trans_t	*tp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGBII(cur, bp, pfirst, plast);
-	tp = cur->bc_tp;
-	if (bp) {
-		xfs_bmbt_block_t	*block;
-		int			first;
-		int			last;
-		xfs_bmbt_ptr_t		*pp;
-
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
-		pp = XFS_BMAP_PTR_DADDR(block, 1, cur);
-		first = (int)((xfs_caddr_t)&pp[pfirst - 1] - (xfs_caddr_t)block);
-		last = (int)(((xfs_caddr_t)&pp[plast] - 1) - (xfs_caddr_t)block);
-		xfs_trans_log_buf(tp, bp, first, last);
-	} else {
-		xfs_inode_t		*ip;
-
-		ip = cur->bc_private.b.ip;
-		xfs_trans_log_inode(tp, ip,
-			XFS_ILOG_FBROOT(cur->bc_private.b.whichfork));
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-}
-
-/*
- * Lookup the record.  The cursor is made to point to it, based on dir.
- */
-STATIC int				/* error */
-xfs_bmbt_lookup(
-	xfs_btree_cur_t		*cur,
-	xfs_lookup_t		dir,
-	int			*stat)		/* success/failure */
-{
-	xfs_bmbt_block_t	*block=NULL;
-	xfs_buf_t		*bp;
-	xfs_daddr_t		d;
-	xfs_sfiloff_t		diff;
-	int			error;		/* error return value */
-	xfs_fsblock_t		fsbno=0;
-	int			high;
-	int			i;
-	int			keyno=0;
-	xfs_bmbt_key_t		*kkbase=NULL;
-	xfs_bmbt_key_t		*kkp;
-	xfs_bmbt_rec_t		*krbase=NULL;
-	xfs_bmbt_rec_t		*krp;
-	int			level;
-	int			low;
-	xfs_mount_t		*mp;
-	xfs_bmbt_ptr_t		*pp;
-	xfs_bmbt_irec_t		*rp;
-	xfs_fileoff_t		startoff;
-	xfs_trans_t		*tp;
-
-	XFS_STATS_INC(xs_bmbt_lookup);
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGI(cur, (int)dir);
-	tp = cur->bc_tp;
-	mp = cur->bc_mp;
-	rp = &cur->bc_rec.b;
-	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
-		if (level < cur->bc_nlevels - 1) {
-			d = XFS_FSB_TO_DADDR(mp, fsbno);
-			bp = cur->bc_bufs[level];
-			if (bp && XFS_BUF_ADDR(bp) != d)
-				bp = NULL;
-			if (!bp) {
-				if ((error = xfs_btree_read_bufl(mp, tp, fsbno,
-						0, &bp, XFS_BMAP_BTREE_REF))) {
-					XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-					return error;
-				}
-				xfs_btree_setbuf(cur, level, bp);
-				block = XFS_BUF_TO_BMBT_BLOCK(bp);
-				if ((error = xfs_btree_check_lblock(cur, block,
-						level, bp))) {
-					XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-					return error;
-				}
-			} else
-				block = XFS_BUF_TO_BMBT_BLOCK(bp);
-		} else
-			block = xfs_bmbt_get_block(cur, level, &bp);
-		if (diff == 0)
-			keyno = 1;
-		else {
-			if (level > 0)
-				kkbase = XFS_BMAP_KEY_IADDR(block, 1, cur);
-			else
-				krbase = XFS_BMAP_REC_IADDR(block, 1, cur);
-			low = 1;
-			if (!(high = be16_to_cpu(block->bb_numrecs))) {
-				ASSERT(level == 0);
-				cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
-				XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-				*stat = 0;
-				return 0;
-			}
-			while (low <= high) {
-				XFS_STATS_INC(xs_bmbt_compare);
-				keyno = (low + high) >> 1;
-				if (level > 0) {
-					kkp = kkbase + keyno - 1;
-					startoff = be64_to_cpu(kkp->br_startoff);
-				} else {
-					krp = krbase + keyno - 1;
-					startoff = xfs_bmbt_disk_get_startoff(krp);
-				}
-				diff = (xfs_sfiloff_t)
-						(startoff - rp->br_startoff);
-				if (diff < 0)
-					low = keyno + 1;
-				else if (diff > 0)
-					high = keyno - 1;
-				else
-					break;
-			}
-		}
-		if (level > 0) {
-			if (diff > 0 && --keyno < 1)
-				keyno = 1;
-			pp = XFS_BMAP_PTR_IADDR(block, keyno, cur);
-			fsbno = be64_to_cpu(*pp);
-#ifdef DEBUG
-			if ((error = xfs_btree_check_lptr(cur, fsbno, level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-#endif
-			cur->bc_ptrs[level] = keyno;
-		}
-	}
-	if (dir != XFS_LOOKUP_LE && diff < 0) {
-		keyno++;
-		/*
-		 * If ge search and we went off the end of the block, but it's
-		 * not the last block, we're in the wrong block.
-		 */
-		if (dir == XFS_LOOKUP_GE && keyno > be16_to_cpu(block->bb_numrecs) &&
-		    be64_to_cpu(block->bb_rightsib) != NULLDFSBNO) {
-			cur->bc_ptrs[0] = keyno;
-			if ((error = xfs_bmbt_increment(cur, 0, &i))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-			XFS_WANT_CORRUPTED_RETURN(i == 1);
-			XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-			*stat = 1;
-			return 0;
-		}
-	}
-	else if (dir == XFS_LOOKUP_LE && diff > 0)
-		keyno--;
-	cur->bc_ptrs[0] = keyno;
-	if (keyno == 0 || keyno > be16_to_cpu(block->bb_numrecs)) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-	} else {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = ((dir != XFS_LOOKUP_EQ) || (diff == 0));
-	}
-	return 0;
-}
-
-/*
- * Move 1 record left from cur/level if possible.
- * Update cur to reflect the new path.
- */
-STATIC int					/* error */
-xfs_bmbt_lshift(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	int			*stat)		/* success/failure */
-{
-	int			error;		/* error return value */
-#ifdef DEBUG
-	int			i;		/* loop counter */
-#endif
-	xfs_bmbt_key_t		key;		/* bmap btree key */
-	xfs_buf_t		*lbp;		/* left buffer pointer */
-	xfs_bmbt_block_t	*left;		/* left btree block */
-	xfs_bmbt_key_t		*lkp=NULL;	/* left btree key */
-	xfs_bmbt_ptr_t		*lpp;		/* left address pointer */
-	int			lrecs;		/* left record count */
-	xfs_bmbt_rec_t		*lrp=NULL;	/* left record pointer */
-	xfs_mount_t		*mp;		/* file system mount point */
-	xfs_buf_t		*rbp;		/* right buffer pointer */
-	xfs_bmbt_block_t	*right;		/* right btree block */
-	xfs_bmbt_key_t		*rkp=NULL;	/* right btree key */
-	xfs_bmbt_ptr_t		*rpp=NULL;	/* right address pointer */
-	xfs_bmbt_rec_t		*rrp=NULL;	/* right record pointer */
-	int			rrecs;		/* right record count */
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGI(cur, level);
-	if (level == cur->bc_nlevels - 1) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	rbp = cur->bc_bufs[level];
-	right = XFS_BUF_TO_BMBT_BLOCK(rbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, right, level, rbp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	if (be64_to_cpu(right->bb_leftsib) == NULLDFSBNO) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	if (cur->bc_ptrs[level] <= 1) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	mp = cur->bc_mp;
-	if ((error = xfs_btree_read_bufl(mp, cur->bc_tp, be64_to_cpu(right->bb_leftsib), 0,
-			&lbp, XFS_BMAP_BTREE_REF))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	left = XFS_BUF_TO_BMBT_BLOCK(lbp);
-	if ((error = xfs_btree_check_lblock(cur, left, level, lbp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	if (be16_to_cpu(left->bb_numrecs) == XFS_BMAP_BLOCK_IMAXRECS(level, cur)) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	lrecs = be16_to_cpu(left->bb_numrecs) + 1;
-	if (level > 0) {
-		lkp = XFS_BMAP_KEY_IADDR(left, lrecs, cur);
-		rkp = XFS_BMAP_KEY_IADDR(right, 1, cur);
-		*lkp = *rkp;
-		xfs_bmbt_log_keys(cur, lbp, lrecs, lrecs);
-		lpp = XFS_BMAP_PTR_IADDR(left, lrecs, cur);
-		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lptr_disk(cur, *rpp, level))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-#endif
-		*lpp = *rpp;
-		xfs_bmbt_log_ptrs(cur, lbp, lrecs, lrecs);
-	} else {
-		lrp = XFS_BMAP_REC_IADDR(left, lrecs, cur);
-		rrp = XFS_BMAP_REC_IADDR(right, 1, cur);
-		*lrp = *rrp;
-		xfs_bmbt_log_recs(cur, lbp, lrecs, lrecs);
-	}
-	left->bb_numrecs = cpu_to_be16(lrecs);
-	xfs_bmbt_log_block(cur, lbp, XFS_BB_NUMRECS);
-#ifdef DEBUG
-	if (level > 0)
-		xfs_btree_check_key(XFS_BTNUM_BMAP, lkp - 1, lkp);
-	else
-		xfs_btree_check_rec(XFS_BTNUM_BMAP, lrp - 1, lrp);
-#endif
-	rrecs = be16_to_cpu(right->bb_numrecs) - 1;
-	right->bb_numrecs = cpu_to_be16(rrecs);
-	xfs_bmbt_log_block(cur, rbp, XFS_BB_NUMRECS);
-	if (level > 0) {
-#ifdef DEBUG
-		for (i = 0; i < rrecs; i++) {
-			if ((error = xfs_btree_check_lptr_disk(cur, rpp[i + 1],
-					level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-		}
-#endif
-		memmove(rkp, rkp + 1, rrecs * sizeof(*rkp));
-		memmove(rpp, rpp + 1, rrecs * sizeof(*rpp));
-		xfs_bmbt_log_keys(cur, rbp, 1, rrecs);
-		xfs_bmbt_log_ptrs(cur, rbp, 1, rrecs);
-	} else {
-		memmove(rrp, rrp + 1, rrecs * sizeof(*rrp));
-		xfs_bmbt_log_recs(cur, rbp, 1, rrecs);
-		key.br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(rrp));
-		rkp = &key;
-	}
-	if ((error = xfs_bmbt_updkey(cur, rkp, level + 1))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	cur->bc_ptrs[level]--;
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 1;
-	return 0;
-}
-
-/*
- * Move 1 record right from cur/level if possible.
- * Update cur to reflect the new path.
- */
-STATIC int					/* error */
-xfs_bmbt_rshift(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	int			*stat)		/* success/failure */
-{
-	int			error;		/* error return value */
-	int			i;		/* loop counter */
-	xfs_bmbt_key_t		key;		/* bmap btree key */
-	xfs_buf_t		*lbp;		/* left buffer pointer */
-	xfs_bmbt_block_t	*left;		/* left btree block */
-	xfs_bmbt_key_t		*lkp;		/* left btree key */
-	xfs_bmbt_ptr_t		*lpp;		/* left address pointer */
-	xfs_bmbt_rec_t		*lrp;		/* left record pointer */
-	xfs_mount_t		*mp;		/* file system mount point */
-	xfs_buf_t		*rbp;		/* right buffer pointer */
-	xfs_bmbt_block_t	*right;		/* right btree block */
-	xfs_bmbt_key_t		*rkp;		/* right btree key */
-	xfs_bmbt_ptr_t		*rpp;		/* right address pointer */
-	xfs_bmbt_rec_t		*rrp=NULL;	/* right record pointer */
-	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGI(cur, level);
-	if (level == cur->bc_nlevels - 1) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	lbp = cur->bc_bufs[level];
-	left = XFS_BUF_TO_BMBT_BLOCK(lbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, left, level, lbp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	if (be64_to_cpu(left->bb_rightsib) == NULLDFSBNO) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	if (cur->bc_ptrs[level] >= be16_to_cpu(left->bb_numrecs)) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	mp = cur->bc_mp;
-	if ((error = xfs_btree_read_bufl(mp, cur->bc_tp, be64_to_cpu(left->bb_rightsib), 0,
-			&rbp, XFS_BMAP_BTREE_REF))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	right = XFS_BUF_TO_BMBT_BLOCK(rbp);
-	if ((error = xfs_btree_check_lblock(cur, right, level, rbp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	if (be16_to_cpu(right->bb_numrecs) == XFS_BMAP_BLOCK_IMAXRECS(level, cur)) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	if (level > 0) {
-		lkp = XFS_BMAP_KEY_IADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		lpp = XFS_BMAP_PTR_IADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		rkp = XFS_BMAP_KEY_IADDR(right, 1, cur);
-		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = be16_to_cpu(right->bb_numrecs) - 1; i >= 0; i--) {
-			if ((error = xfs_btree_check_lptr_disk(cur, rpp[i], level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-		}
-#endif
-		memmove(rkp + 1, rkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memmove(rpp + 1, rpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lptr_disk(cur, *lpp, level))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-#endif
-		*rkp = *lkp;
-		*rpp = *lpp;
-		xfs_bmbt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		xfs_bmbt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-	} else {
-		lrp = XFS_BMAP_REC_IADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		rrp = XFS_BMAP_REC_IADDR(right, 1, cur);
-		memmove(rrp + 1, rrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		*rrp = *lrp;
-		xfs_bmbt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		key.br_startoff = cpu_to_be64(xfs_bmbt_disk_get_startoff(rrp));
-		rkp = &key;
-	}
-	be16_add_cpu(&left->bb_numrecs, -1);
-	xfs_bmbt_log_block(cur, lbp, XFS_BB_NUMRECS);
-	be16_add_cpu(&right->bb_numrecs, 1);
-#ifdef DEBUG
-	if (level > 0)
-		xfs_btree_check_key(XFS_BTNUM_BMAP, rkp, rkp + 1);
-	else
-		xfs_btree_check_rec(XFS_BTNUM_BMAP, rrp, rrp + 1);
-#endif
-	xfs_bmbt_log_block(cur, rbp, XFS_BB_NUMRECS);
-	if ((error = xfs_btree_dup_cursor(cur, &tcur))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	i = xfs_btree_lastrec(tcur, level);
-	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-	if ((error = xfs_bmbt_increment(tcur, level, &i))) {
-		XFS_BMBT_TRACE_CURSOR(tcur, ERROR);
-		goto error1;
-	}
-	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-	if ((error = xfs_bmbt_updkey(tcur, rkp, level + 1))) {
-		XFS_BMBT_TRACE_CURSOR(tcur, ERROR);
-		goto error1;
-	}
-	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 1;
-	return 0;
-error0:
-	XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-error1:
-	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
-	return error;
-}
-
 /*
  * Determine the extent state.
  */
@@ -1453,229 +60,15 @@ xfs_extent_state(
 	return XFS_EXT_NORM;
 }
 
-
-/*
- * Split cur/level block in half.
- * Return new block number and its first record (to be inserted into parent).
- */
-STATIC int					/* error */
-xfs_bmbt_split(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	xfs_fsblock_t		*bnop,
-	__uint64_t		*startoff,
-	xfs_btree_cur_t		**curp,
-	int			*stat)		/* success/failure */
-{
-	xfs_alloc_arg_t		args;		/* block allocation args */
-	int			error;		/* error return value */
-	int			i;		/* loop counter */
-	xfs_fsblock_t		lbno;		/* left sibling block number */
-	xfs_buf_t		*lbp;		/* left buffer pointer */
-	xfs_bmbt_block_t	*left;		/* left btree block */
-	xfs_bmbt_key_t		*lkp;		/* left btree key */
-	xfs_bmbt_ptr_t		*lpp;		/* left address pointer */
-	xfs_bmbt_rec_t		*lrp;		/* left record pointer */
-	xfs_buf_t		*rbp;		/* right buffer pointer */
-	xfs_bmbt_block_t	*right;		/* right btree block */
-	xfs_bmbt_key_t		*rkp;		/* right btree key */
-	xfs_bmbt_ptr_t		*rpp;		/* right address pointer */
-	xfs_bmbt_block_t	*rrblock;	/* right-right btree block */
-	xfs_buf_t		*rrbp;		/* right-right buffer pointer */
-	xfs_bmbt_rec_t		*rrp;		/* right record pointer */
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGIFK(cur, level, *bnop, *startoff);
-	args.tp = cur->bc_tp;
-	args.mp = cur->bc_mp;
-	lbp = cur->bc_bufs[level];
-	lbno = XFS_DADDR_TO_FSB(args.mp, XFS_BUF_ADDR(lbp));
-	left = XFS_BUF_TO_BMBT_BLOCK(lbp);
-	args.fsbno = cur->bc_private.b.firstblock;
-	args.firstblock = args.fsbno;
-	args.minleft = 0;
-	if (args.fsbno == NULLFSBLOCK) {
-		args.fsbno = lbno;
-		args.type = XFS_ALLOCTYPE_START_BNO;
-		/*
-		 * Make sure there is sufficient room left in the AG to
-		 * complete a full tree split for an extent insert.  If
-		 * we are converting the middle part of an extent then
-		 * we may need space for two tree splits.
-		 *
-		 * We are relying on the caller to make the correct block
-		 * reservation for this operation to succeed.  If the
-		 * reservation amount is insufficient then we may fail a
-		 * block allocation here and corrupt the filesystem.
-		 */
-		args.minleft = xfs_trans_get_block_res(args.tp);
-	} else if (cur->bc_private.b.flist->xbf_low)
-		args.type = XFS_ALLOCTYPE_START_BNO;
-	else
-		args.type = XFS_ALLOCTYPE_NEAR_BNO;
-	args.mod = args.alignment = args.total = args.isfl =
-		args.userdata = args.minalignslop = 0;
-	args.minlen = args.maxlen = args.prod = 1;
-	args.wasdel = cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL;
-	if (!args.wasdel && xfs_trans_get_block_res(args.tp) == 0) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return XFS_ERROR(ENOSPC);
-	}
-	if ((error = xfs_alloc_vextent(&args))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	if (args.fsbno == NULLFSBLOCK && args.minleft) {
-		/*
-		 * Could not find an AG with enough free space to satisfy
-		 * a full btree split.  Try again without minleft and if
-		 * successful activate the lowspace algorithm.
-		 */
-		args.fsbno = 0;
-		args.type = XFS_ALLOCTYPE_FIRST_AG;
-		args.minleft = 0;
-		if ((error = xfs_alloc_vextent(&args))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		cur->bc_private.b.flist->xbf_low = 1;
-	}
-	if (args.fsbno == NULLFSBLOCK) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	ASSERT(args.len == 1);
-	cur->bc_private.b.firstblock = args.fsbno;
-	cur->bc_private.b.allocated++;
-	cur->bc_private.b.ip->i_d.di_nblocks++;
-	xfs_trans_log_inode(args.tp, cur->bc_private.b.ip, XFS_ILOG_CORE);
-	XFS_TRANS_MOD_DQUOT_BYINO(args.mp, args.tp, cur->bc_private.b.ip,
-			XFS_TRANS_DQ_BCOUNT, 1L);
-	rbp = xfs_btree_get_bufl(args.mp, args.tp, args.fsbno, 0);
-	right = XFS_BUF_TO_BMBT_BLOCK(rbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, left, level, rbp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	right->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
-	right->bb_level = left->bb_level;
-	right->bb_numrecs = cpu_to_be16(be16_to_cpu(left->bb_numrecs) / 2);
-	if ((be16_to_cpu(left->bb_numrecs) & 1) &&
-	    cur->bc_ptrs[level] <= be16_to_cpu(right->bb_numrecs) + 1)
-		be16_add_cpu(&right->bb_numrecs, 1);
-	i = be16_to_cpu(left->bb_numrecs) - be16_to_cpu(right->bb_numrecs) + 1;
-	if (level > 0) {
-		lkp = XFS_BMAP_KEY_IADDR(left, i, cur);
-		lpp = XFS_BMAP_PTR_IADDR(left, i, cur);
-		rkp = XFS_BMAP_KEY_IADDR(right, 1, cur);
-		rpp = XFS_BMAP_PTR_IADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
-			if ((error = xfs_btree_check_lptr_disk(cur, lpp[i], level))) {
-				XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-				return error;
-			}
-		}
-#endif
-		memcpy(rkp, lkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memcpy(rpp, lpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-		xfs_bmbt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		xfs_bmbt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		*startoff = be64_to_cpu(rkp->br_startoff);
-	} else {
-		lrp = XFS_BMAP_REC_IADDR(left, i, cur);
-		rrp = XFS_BMAP_REC_IADDR(right, 1, cur);
-		memcpy(rrp, lrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		xfs_bmbt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		*startoff = xfs_bmbt_disk_get_startoff(rrp);
-	}
-	be16_add_cpu(&left->bb_numrecs, -(be16_to_cpu(right->bb_numrecs)));
-	right->bb_rightsib = left->bb_rightsib;
-	left->bb_rightsib = cpu_to_be64(args.fsbno);
-	right->bb_leftsib = cpu_to_be64(lbno);
-	xfs_bmbt_log_block(cur, rbp, XFS_BB_ALL_BITS);
-	xfs_bmbt_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
-	if (be64_to_cpu(right->bb_rightsib) != NULLDFSBNO) {
-		if ((error = xfs_btree_read_bufl(args.mp, args.tp,
-				be64_to_cpu(right->bb_rightsib), 0, &rrbp,
-				XFS_BMAP_BTREE_REF))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		rrblock = XFS_BUF_TO_BMBT_BLOCK(rrbp);
-		if ((error = xfs_btree_check_lblock(cur, rrblock, level, rrbp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		rrblock->bb_leftsib = cpu_to_be64(args.fsbno);
-		xfs_bmbt_log_block(cur, rrbp, XFS_BB_LEFTSIB);
-	}
-	if (cur->bc_ptrs[level] > be16_to_cpu(left->bb_numrecs) + 1) {
-		xfs_btree_setbuf(cur, level, rbp);
-		cur->bc_ptrs[level] -= be16_to_cpu(left->bb_numrecs);
-	}
-	if (level + 1 < cur->bc_nlevels) {
-		if ((error = xfs_btree_dup_cursor(cur, curp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		(*curp)->bc_ptrs[level + 1]++;
-	}
-	*bnop = args.fsbno;
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 1;
-	return 0;
-}
-
-
-/*
- * Update keys for the record.
- */
-STATIC int
-xfs_bmbt_updkey(
-	xfs_btree_cur_t		*cur,
-	xfs_bmbt_key_t		*keyp,	/* on-disk format */
-	int			level)
-{
-	xfs_bmbt_block_t	*block;
-	xfs_buf_t		*bp;
-#ifdef DEBUG
-	int			error;
-#endif
-	xfs_bmbt_key_t		*kp;
-	int			ptr;
-
-	ASSERT(level >= 1);
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGIK(cur, level, keyp);
-	for (ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
-		block = xfs_bmbt_get_block(cur, level, &bp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lblock(cur, block, level, bp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-#endif
-		ptr = cur->bc_ptrs[level];
-		kp = XFS_BMAP_KEY_IADDR(block, ptr, cur);
-		*kp = *keyp;
-		xfs_bmbt_log_keys(cur, bp, ptr, ptr);
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	return 0;
-}
-
 /*
  * Convert on-disk form of btree root to in-memory form.
  */
 void
 xfs_bmdr_to_bmbt(
+	struct xfs_mount	*mp,
 	xfs_bmdr_block_t	*dblock,
 	int			dblocklen,
-	xfs_bmbt_block_t	*rblock,
+	struct xfs_btree_block	*rblock,
 	int			rblocklen)
 {
 	int			dmxr;
@@ -1688,129 +81,19 @@ xfs_bmdr_to_bmbt(
 	rblock->bb_level = dblock->bb_level;
 	ASSERT(be16_to_cpu(rblock->bb_level) > 0);
 	rblock->bb_numrecs = dblock->bb_numrecs;
-	rblock->bb_leftsib = cpu_to_be64(NULLDFSBNO);
-	rblock->bb_rightsib = cpu_to_be64(NULLDFSBNO);
-	dmxr = (int)XFS_BTREE_BLOCK_MAXRECS(dblocklen, xfs_bmdr, 0);
-	fkp = XFS_BTREE_KEY_ADDR(xfs_bmdr, dblock, 1);
-	tkp = XFS_BMAP_BROOT_KEY_ADDR(rblock, 1, rblocklen);
-	fpp = XFS_BTREE_PTR_ADDR(xfs_bmdr, dblock, 1, dmxr);
-	tpp = XFS_BMAP_BROOT_PTR_ADDR(rblock, 1, rblocklen);
+	rblock->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+	rblock->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+	dmxr = xfs_bmdr_maxrecs(mp, dblocklen, 0);
+	fkp = XFS_BMDR_KEY_ADDR(dblock, 1);
+	tkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
+	fpp = XFS_BMDR_PTR_ADDR(dblock, 1, dmxr);
+	tpp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, rblocklen);
 	dmxr = be16_to_cpu(dblock->bb_numrecs);
 	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
 	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
 }
 
 /*
- * Decrement cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-int						/* error */
-xfs_bmbt_decrement(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	int			*stat)		/* success/failure */
-{
-	xfs_bmbt_block_t	*block;
-	xfs_buf_t		*bp;
-	int			error;		/* error return value */
-	xfs_fsblock_t		fsbno;
-	int			lev;
-	xfs_mount_t		*mp;
-	xfs_trans_t		*tp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGI(cur, level);
-	ASSERT(level < cur->bc_nlevels);
-	if (level < cur->bc_nlevels - 1)
-		xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
-	if (--cur->bc_ptrs[level] > 0) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 1;
-		return 0;
-	}
-	block = xfs_bmbt_get_block(cur, level, &bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, block, level, bp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	if (be64_to_cpu(block->bb_leftsib) == NULLDFSBNO) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
-		if (--cur->bc_ptrs[lev] > 0)
-			break;
-		if (lev < cur->bc_nlevels - 1)
-			xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
-	}
-	if (lev == cur->bc_nlevels) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	tp = cur->bc_tp;
-	mp = cur->bc_mp;
-	for (block = xfs_bmbt_get_block(cur, lev, &bp); lev > level; ) {
-		fsbno = be64_to_cpu(*XFS_BMAP_PTR_IADDR(block, cur->bc_ptrs[lev], cur));
-		if ((error = xfs_btree_read_bufl(mp, tp, fsbno, 0, &bp,
-				XFS_BMAP_BTREE_REF))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		lev--;
-		xfs_btree_setbuf(cur, lev, bp);
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
-		if ((error = xfs_btree_check_lblock(cur, block, lev, bp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		cur->bc_ptrs[lev] = be16_to_cpu(block->bb_numrecs);
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 1;
-	return 0;
-}
-
-/*
- * Delete the record pointed to by cur.
- */
-int					/* error */
-xfs_bmbt_delete(
-	xfs_btree_cur_t	*cur,
-	int		*stat)		/* success/failure */
-{
-	int		error;		/* error return value */
-	int		i;
-	int		level;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	for (level = 0, i = 2; i == 2; level++) {
-		if ((error = xfs_bmbt_delrec(cur, level, &i))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-	}
-	if (i == 0) {
-		for (level = 1; level < cur->bc_nlevels; level++) {
-			if (cur->bc_ptrs[level] == 0) {
-				if ((error = xfs_bmbt_decrement(cur, level,
-						&i))) {
-					XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-					return error;
-				}
-				break;
-			}
-		}
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = i;
-	return 0;
-}
-
-/*
  * Convert a compressed bmap extent record to an uncompressed form.
  * This code must be in sync with the routines xfs_bmbt_get_startoff,
  * xfs_bmbt_get_startblock, xfs_bmbt_get_blockcount and xfs_bmbt_get_state.
@@ -1864,31 +147,6 @@ xfs_bmbt_get_all(
 }
 
 /*
- * Get the block pointer for the given level of the cursor.
- * Fill in the buffer pointer, if applicable.
- */
-xfs_bmbt_block_t *
-xfs_bmbt_get_block(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	xfs_buf_t		**bpp)
-{
-	xfs_ifork_t		*ifp;
-	xfs_bmbt_block_t	*rval;
-
-	if (level < cur->bc_nlevels - 1) {
-		*bpp = cur->bc_bufs[level];
-		rval = XFS_BUF_TO_BMBT_BLOCK(*bpp);
-	} else {
-		*bpp = NULL;
-		ifp = XFS_IFORK_PTR(cur->bc_private.b.ip,
-			cur->bc_private.b.whichfork);
-		rval = ifp->if_broot;
-	}
-	return rval;
-}
-
-/*
  * Extract the blockcount field from an in memory bmap extent record.
  */
 xfs_filblks_t
@@ -1950,7 +208,8 @@ xfs_bmbt_disk_get_all(
 	xfs_bmbt_rec_t	*r,
 	xfs_bmbt_irec_t *s)
 {
-	__xfs_bmbt_get_all(be64_to_cpu(r->l0), be64_to_cpu(r->l1), s);
+	__xfs_bmbt_get_all(get_unaligned_be64(&r->l0),
+				get_unaligned_be64(&r->l1), s);
 }
 
 /*
@@ -1974,348 +233,6 @@ xfs_bmbt_disk_get_startoff(
 		 XFS_MASK64LO(64 - BMBT_EXNTFLAG_BITLEN)) >> 9;
 }
 
-/*
- * Increment cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-int						/* error */
-xfs_bmbt_increment(
-	xfs_btree_cur_t		*cur,
-	int			level,
-	int			*stat)		/* success/failure */
-{
-	xfs_bmbt_block_t	*block;
-	xfs_buf_t		*bp;
-	int			error;		/* error return value */
-	xfs_fsblock_t		fsbno;
-	int			lev;
-	xfs_mount_t		*mp;
-	xfs_trans_t		*tp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGI(cur, level);
-	ASSERT(level < cur->bc_nlevels);
-	if (level < cur->bc_nlevels - 1)
-		xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
-	block = xfs_bmbt_get_block(cur, level, &bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, block, level, bp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	if (++cur->bc_ptrs[level] <= be16_to_cpu(block->bb_numrecs)) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 1;
-		return 0;
-	}
-	if (be64_to_cpu(block->bb_rightsib) == NULLDFSBNO) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
-		block = xfs_bmbt_get_block(cur, lev, &bp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lblock(cur, block, lev, bp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-#endif
-		if (++cur->bc_ptrs[lev] <= be16_to_cpu(block->bb_numrecs))
-			break;
-		if (lev < cur->bc_nlevels - 1)
-			xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
-	}
-	if (lev == cur->bc_nlevels) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	tp = cur->bc_tp;
-	mp = cur->bc_mp;
-	for (block = xfs_bmbt_get_block(cur, lev, &bp); lev > level; ) {
-		fsbno = be64_to_cpu(*XFS_BMAP_PTR_IADDR(block, cur->bc_ptrs[lev], cur));
-		if ((error = xfs_btree_read_bufl(mp, tp, fsbno, 0, &bp,
-				XFS_BMAP_BTREE_REF))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		lev--;
-		xfs_btree_setbuf(cur, lev, bp);
-		block = XFS_BUF_TO_BMBT_BLOCK(bp);
-		if ((error = xfs_btree_check_lblock(cur, block, lev, bp))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		cur->bc_ptrs[lev] = 1;
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = 1;
-	return 0;
-}
-
-/*
- * Insert the current record at the point referenced by cur.
- *
- * A multi-level split of the tree on insert will invalidate the original
- * cursor.  All callers of this function should assume that the cursor is
- * no longer valid and revalidate it.
- */
-int					/* error */
-xfs_bmbt_insert(
-	xfs_btree_cur_t	*cur,
-	int		*stat)		/* success/failure */
-{
-	int		error;		/* error return value */
-	int		i;
-	int		level;
-	xfs_fsblock_t	nbno;
-	xfs_btree_cur_t	*ncur;
-	xfs_bmbt_rec_t	nrec;
-	xfs_btree_cur_t	*pcur;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	level = 0;
-	nbno = NULLFSBLOCK;
-	xfs_bmbt_disk_set_all(&nrec, &cur->bc_rec.b);
-	ncur = NULL;
-	pcur = cur;
-	do {
-		if ((error = xfs_bmbt_insrec(pcur, level++, &nbno, &nrec, &ncur,
-				&i))) {
-			if (pcur != cur)
-				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if (pcur != cur && (ncur || nbno == NULLFSBLOCK)) {
-			cur->bc_nlevels = pcur->bc_nlevels;
-			cur->bc_private.b.allocated +=
-				pcur->bc_private.b.allocated;
-			pcur->bc_private.b.allocated = 0;
-			ASSERT((cur->bc_private.b.firstblock != NULLFSBLOCK) ||
-			       XFS_IS_REALTIME_INODE(cur->bc_private.b.ip));
-			cur->bc_private.b.firstblock =
-				pcur->bc_private.b.firstblock;
-			ASSERT(cur->bc_private.b.flist ==
-			       pcur->bc_private.b.flist);
-			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
-		}
-		if (ncur) {
-			pcur = ncur;
-			ncur = NULL;
-		}
-	} while (nbno != NULLFSBLOCK);
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*stat = i;
-	return 0;
-error0:
-	XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-	return error;
-}
-
-/*
- * Log fields from the btree block header.
- */
-void
-xfs_bmbt_log_block(
-	xfs_btree_cur_t		*cur,
-	xfs_buf_t		*bp,
-	int			fields)
-{
-	int			first;
-	int			last;
-	xfs_trans_t		*tp;
-	static const short	offsets[] = {
-		offsetof(xfs_bmbt_block_t, bb_magic),
-		offsetof(xfs_bmbt_block_t, bb_level),
-		offsetof(xfs_bmbt_block_t, bb_numrecs),
-		offsetof(xfs_bmbt_block_t, bb_leftsib),
-		offsetof(xfs_bmbt_block_t, bb_rightsib),
-		sizeof(xfs_bmbt_block_t)
-	};
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGBI(cur, bp, fields);
-	tp = cur->bc_tp;
-	if (bp) {
-		xfs_btree_offsets(fields, offsets, XFS_BB_NUM_BITS, &first,
-				  &last);
-		xfs_trans_log_buf(tp, bp, first, last);
-	} else
-		xfs_trans_log_inode(tp, cur->bc_private.b.ip,
-			XFS_ILOG_FBROOT(cur->bc_private.b.whichfork));
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-}
-
-/*
- * Log record values from the btree block.
- */
-void
-xfs_bmbt_log_recs(
-	xfs_btree_cur_t		*cur,
-	xfs_buf_t		*bp,
-	int			rfirst,
-	int			rlast)
-{
-	xfs_bmbt_block_t	*block;
-	int			first;
-	int			last;
-	xfs_bmbt_rec_t		*rp;
-	xfs_trans_t		*tp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGBII(cur, bp, rfirst, rlast);
-	ASSERT(bp);
-	tp = cur->bc_tp;
-	block = XFS_BUF_TO_BMBT_BLOCK(bp);
-	rp = XFS_BMAP_REC_DADDR(block, 1, cur);
-	first = (int)((xfs_caddr_t)&rp[rfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&rp[rlast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(tp, bp, first, last);
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-}
-
-int					/* error */
-xfs_bmbt_lookup_eq(
-	xfs_btree_cur_t	*cur,
-	xfs_fileoff_t	off,
-	xfs_fsblock_t	bno,
-	xfs_filblks_t	len,
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.b.br_startoff = off;
-	cur->bc_rec.b.br_startblock = bno;
-	cur->bc_rec.b.br_blockcount = len;
-	return xfs_bmbt_lookup(cur, XFS_LOOKUP_EQ, stat);
-}
-
-int					/* error */
-xfs_bmbt_lookup_ge(
-	xfs_btree_cur_t	*cur,
-	xfs_fileoff_t	off,
-	xfs_fsblock_t	bno,
-	xfs_filblks_t	len,
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.b.br_startoff = off;
-	cur->bc_rec.b.br_startblock = bno;
-	cur->bc_rec.b.br_blockcount = len;
-	return xfs_bmbt_lookup(cur, XFS_LOOKUP_GE, stat);
-}
-
-/*
- * Give the bmap btree a new root block.  Copy the old broot contents
- * down into a real block and make the broot point to it.
- */
-int						/* error */
-xfs_bmbt_newroot(
-	xfs_btree_cur_t		*cur,		/* btree cursor */
-	int			*logflags,	/* logging flags for inode */
-	int			*stat)		/* return status - 0 fail */
-{
-	xfs_alloc_arg_t		args;		/* allocation arguments */
-	xfs_bmbt_block_t	*block;		/* bmap btree block */
-	xfs_buf_t		*bp;		/* buffer for block */
-	xfs_bmbt_block_t	*cblock;	/* child btree block */
-	xfs_bmbt_key_t		*ckp;		/* child key pointer */
-	xfs_bmbt_ptr_t		*cpp;		/* child ptr pointer */
-	int			error;		/* error return code */
-#ifdef DEBUG
-	int			i;		/* loop counter */
-#endif
-	xfs_bmbt_key_t		*kp;		/* pointer to bmap btree key */
-	int			level;		/* btree level */
-	xfs_bmbt_ptr_t		*pp;		/* pointer to bmap block addr */
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	level = cur->bc_nlevels - 1;
-	block = xfs_bmbt_get_block(cur, level, &bp);
-	/*
-	 * Copy the root into a real block.
-	 */
-	args.mp = cur->bc_mp;
-	pp = XFS_BMAP_PTR_IADDR(block, 1, cur);
-	args.tp = cur->bc_tp;
-	args.fsbno = cur->bc_private.b.firstblock;
-	args.mod = args.minleft = args.alignment = args.total = args.isfl =
-		args.userdata = args.minalignslop = 0;
-	args.minlen = args.maxlen = args.prod = 1;
-	args.wasdel = cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL;
-	args.firstblock = args.fsbno;
-	if (args.fsbno == NULLFSBLOCK) {
-#ifdef DEBUG
-		if ((error = xfs_btree_check_lptr_disk(cur, *pp, level))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-#endif
-		args.fsbno = be64_to_cpu(*pp);
-		args.type = XFS_ALLOCTYPE_START_BNO;
-	} else if (cur->bc_private.b.flist->xbf_low)
-		args.type = XFS_ALLOCTYPE_START_BNO;
-	else
-		args.type = XFS_ALLOCTYPE_NEAR_BNO;
-	if ((error = xfs_alloc_vextent(&args))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	if (args.fsbno == NULLFSBLOCK) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		*stat = 0;
-		return 0;
-	}
-	ASSERT(args.len == 1);
-	cur->bc_private.b.firstblock = args.fsbno;
-	cur->bc_private.b.allocated++;
-	cur->bc_private.b.ip->i_d.di_nblocks++;
-	XFS_TRANS_MOD_DQUOT_BYINO(args.mp, args.tp, cur->bc_private.b.ip,
-			  XFS_TRANS_DQ_BCOUNT, 1L);
-	bp = xfs_btree_get_bufl(args.mp, cur->bc_tp, args.fsbno, 0);
-	cblock = XFS_BUF_TO_BMBT_BLOCK(bp);
-	*cblock = *block;
-	be16_add_cpu(&block->bb_level, 1);
-	block->bb_numrecs = cpu_to_be16(1);
-	cur->bc_nlevels++;
-	cur->bc_ptrs[level + 1] = 1;
-	kp = XFS_BMAP_KEY_IADDR(block, 1, cur);
-	ckp = XFS_BMAP_KEY_IADDR(cblock, 1, cur);
-	memcpy(ckp, kp, be16_to_cpu(cblock->bb_numrecs) * sizeof(*kp));
-	cpp = XFS_BMAP_PTR_IADDR(cblock, 1, cur);
-#ifdef DEBUG
-	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
-		if ((error = xfs_btree_check_lptr_disk(cur, pp[i], level))) {
-			XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-			return error;
-		}
-	}
-#endif
-	memcpy(cpp, pp, be16_to_cpu(cblock->bb_numrecs) * sizeof(*pp));
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lptr(cur, args.fsbno, level))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	*pp = cpu_to_be64(args.fsbno);
-	xfs_iroot_realloc(cur->bc_private.b.ip, 1 - be16_to_cpu(cblock->bb_numrecs),
-		cur->bc_private.b.whichfork);
-	xfs_btree_setbuf(cur, level, bp);
-	/*
-	 * Do all this logging at the end so that
-	 * the root is at the right level.
-	 */
-	xfs_bmbt_log_block(cur, bp, XFS_BB_ALL_BITS);
-	xfs_bmbt_log_keys(cur, bp, 1, be16_to_cpu(cblock->bb_numrecs));
-	xfs_bmbt_log_ptrs(cur, bp, 1, be16_to_cpu(cblock->bb_numrecs));
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	*logflags |=
-		XFS_ILOG_CORE | XFS_ILOG_FBROOT(cur->bc_private.b.whichfork);
-	*stat = 1;
-	return 0;
-}
 
 /*
  * Set all the fields in a bmap extent record from the arguments.
@@ -2512,7 +429,8 @@ xfs_bmbt_set_state(
  */
 void
 xfs_bmbt_to_bmdr(
-	xfs_bmbt_block_t	*rblock,
+	struct xfs_mount	*mp,
+	struct xfs_btree_block	*rblock,
 	int			rblocklen,
 	xfs_bmdr_block_t	*dblock,
 	int			dblocklen)
@@ -2524,67 +442,22 @@ xfs_bmbt_to_bmdr(
 	__be64			*tpp;
 
 	ASSERT(be32_to_cpu(rblock->bb_magic) == XFS_BMAP_MAGIC);
-	ASSERT(be64_to_cpu(rblock->bb_leftsib) == NULLDFSBNO);
-	ASSERT(be64_to_cpu(rblock->bb_rightsib) == NULLDFSBNO);
+	ASSERT(be64_to_cpu(rblock->bb_u.l.bb_leftsib) == NULLDFSBNO);
+	ASSERT(be64_to_cpu(rblock->bb_u.l.bb_rightsib) == NULLDFSBNO);
 	ASSERT(be16_to_cpu(rblock->bb_level) > 0);
 	dblock->bb_level = rblock->bb_level;
 	dblock->bb_numrecs = rblock->bb_numrecs;
-	dmxr = (int)XFS_BTREE_BLOCK_MAXRECS(dblocklen, xfs_bmdr, 0);
-	fkp = XFS_BMAP_BROOT_KEY_ADDR(rblock, 1, rblocklen);
-	tkp = XFS_BTREE_KEY_ADDR(xfs_bmdr, dblock, 1);
-	fpp = XFS_BMAP_BROOT_PTR_ADDR(rblock, 1, rblocklen);
-	tpp = XFS_BTREE_PTR_ADDR(xfs_bmdr, dblock, 1, dmxr);
+	dmxr = xfs_bmdr_maxrecs(mp, dblocklen, 0);
+	fkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
+	tkp = XFS_BMDR_KEY_ADDR(dblock, 1);
+	fpp = XFS_BMAP_BROOT_PTR_ADDR(mp, rblock, 1, rblocklen);
+	tpp = XFS_BMDR_PTR_ADDR(dblock, 1, dmxr);
 	dmxr = be16_to_cpu(dblock->bb_numrecs);
 	memcpy(tkp, fkp, sizeof(*fkp) * dmxr);
 	memcpy(tpp, fpp, sizeof(*fpp) * dmxr);
 }
 
 /*
- * Update the record to the passed values.
- */
-int
-xfs_bmbt_update(
-	xfs_btree_cur_t		*cur,
-	xfs_fileoff_t		off,
-	xfs_fsblock_t		bno,
-	xfs_filblks_t		len,
-	xfs_exntst_t		state)
-{
-	xfs_bmbt_block_t	*block;
-	xfs_buf_t		*bp;
-	int			error;
-	xfs_bmbt_key_t		key;
-	int			ptr;
-	xfs_bmbt_rec_t		*rp;
-
-	XFS_BMBT_TRACE_CURSOR(cur, ENTRY);
-	XFS_BMBT_TRACE_ARGFFFI(cur, (xfs_dfiloff_t)off, (xfs_dfsbno_t)bno,
-		(xfs_dfilblks_t)len, (int)state);
-	block = xfs_bmbt_get_block(cur, 0, &bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_lblock(cur, block, 0, bp))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-#endif
-	ptr = cur->bc_ptrs[0];
-	rp = XFS_BMAP_REC_IADDR(block, ptr, cur);
-	xfs_bmbt_disk_set_allf(rp, off, bno, len, state);
-	xfs_bmbt_log_recs(cur, bp, ptr, ptr);
-	if (ptr > 1) {
-		XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-		return 0;
-	}
-	key.br_startoff = cpu_to_be64(off);
-	if ((error = xfs_bmbt_updkey(cur, &key, 1))) {
-		XFS_BMBT_TRACE_CURSOR(cur, ERROR);
-		return error;
-	}
-	XFS_BMBT_TRACE_CURSOR(cur, EXIT);
-	return 0;
-}
-
-/*
  * Check extent records, which have just been read, for
  * any bit in the extent flag field. ASSERT on debug
  * kernels, as this condition should not occur.
@@ -2608,3 +481,451 @@ xfs_check_nostate_extents(
 	}
 	return 0;
 }
+
+
+STATIC struct xfs_btree_cur *
+xfs_bmbt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	struct xfs_btree_cur	*new;
+
+	new = xfs_bmbt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_private.b.ip, cur->bc_private.b.whichfork);
+
+	/*
+	 * Copy the firstblock, flist, and flags values,
+	 * since init cursor doesn't get them.
+	 */
+	new->bc_private.b.firstblock = cur->bc_private.b.firstblock;
+	new->bc_private.b.flist = cur->bc_private.b.flist;
+	new->bc_private.b.flags = cur->bc_private.b.flags;
+
+	return new;
+}
+
+STATIC void
+xfs_bmbt_update_cursor(
+	struct xfs_btree_cur	*src,
+	struct xfs_btree_cur	*dst)
+{
+	ASSERT((dst->bc_private.b.firstblock != NULLFSBLOCK) ||
+	       (dst->bc_private.b.ip->i_d.di_flags & XFS_DIFLAG_REALTIME));
+	ASSERT(dst->bc_private.b.flist == src->bc_private.b.flist);
+
+	dst->bc_private.b.allocated += src->bc_private.b.allocated;
+	dst->bc_private.b.firstblock = src->bc_private.b.firstblock;
+
+	src->bc_private.b.allocated = 0;
+}
+
+STATIC int
+xfs_bmbt_alloc_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*start,
+	union xfs_btree_ptr	*new,
+	int			length,
+	int			*stat)
+{
+	xfs_alloc_arg_t		args;		/* block allocation args */
+	int			error;		/* error return value */
+
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.fsbno = cur->bc_private.b.firstblock;
+	args.firstblock = args.fsbno;
+
+	if (args.fsbno == NULLFSBLOCK) {
+		args.fsbno = be64_to_cpu(start->l);
+		args.type = XFS_ALLOCTYPE_START_BNO;
+		/*
+		 * Make sure there is sufficient room left in the AG to
+		 * complete a full tree split for an extent insert.  If
+		 * we are converting the middle part of an extent then
+		 * we may need space for two tree splits.
+		 *
+		 * We are relying on the caller to make the correct block
+		 * reservation for this operation to succeed.  If the
+		 * reservation amount is insufficient then we may fail a
+		 * block allocation here and corrupt the filesystem.
+		 */
+		args.minleft = xfs_trans_get_block_res(args.tp);
+	} else if (cur->bc_private.b.flist->xbf_low) {
+		args.type = XFS_ALLOCTYPE_START_BNO;
+	} else {
+		args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	}
+
+	args.minlen = args.maxlen = args.prod = 1;
+	args.wasdel = cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL;
+	if (!args.wasdel && xfs_trans_get_block_res(args.tp) == 0) {
+		error = XFS_ERROR(ENOSPC);
+		goto error0;
+	}
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		goto error0;
+
+	if (args.fsbno == NULLFSBLOCK && args.minleft) {
+		/*
+		 * Could not find an AG with enough free space to satisfy
+		 * a full btree split.  Try again without minleft and if
+		 * successful activate the lowspace algorithm.
+		 */
+		args.fsbno = 0;
+		args.type = XFS_ALLOCTYPE_FIRST_AG;
+		args.minleft = 0;
+		error = xfs_alloc_vextent(&args);
+		if (error)
+			goto error0;
+		cur->bc_private.b.flist->xbf_low = 1;
+	}
+	if (args.fsbno == NULLFSBLOCK) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+	ASSERT(args.len == 1);
+	cur->bc_private.b.firstblock = args.fsbno;
+	cur->bc_private.b.allocated++;
+	cur->bc_private.b.ip->i_d.di_nblocks++;
+	xfs_trans_log_inode(args.tp, cur->bc_private.b.ip, XFS_ILOG_CORE);
+	XFS_TRANS_MOD_DQUOT_BYINO(args.mp, args.tp, cur->bc_private.b.ip,
+			XFS_TRANS_DQ_BCOUNT, 1L);
+
+	new->l = cpu_to_be64(args.fsbno);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+ error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+STATIC int
+xfs_bmbt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_inode	*ip = cur->bc_private.b.ip;
+	struct xfs_trans	*tp = cur->bc_tp;
+	xfs_fsblock_t		fsbno = XFS_DADDR_TO_FSB(mp, XFS_BUF_ADDR(bp));
+
+	xfs_bmap_add_free(fsbno, 1, cur->bc_private.b.flist, mp);
+	ip->i_d.di_nblocks--;
+
+	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+	XFS_TRANS_MOD_DQUOT_BYINO(mp, tp, ip, XFS_TRANS_DQ_BCOUNT, -1L);
+	xfs_trans_binval(tp, bp);
+	return 0;
+}
+
+STATIC int
+xfs_bmbt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp;
+
+		ifp = XFS_IFORK_PTR(cur->bc_private.b.ip,
+				    cur->bc_private.b.whichfork);
+
+		return xfs_bmbt_maxrecs(cur->bc_mp,
+					ifp->if_broot_bytes, level == 0) / 2;
+	}
+
+	return cur->bc_mp->m_bmap_dmnr[level != 0];
+}
+
+int
+xfs_bmbt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level == cur->bc_nlevels - 1) {
+		struct xfs_ifork	*ifp;
+
+		ifp = XFS_IFORK_PTR(cur->bc_private.b.ip,
+				    cur->bc_private.b.whichfork);
+
+		return xfs_bmbt_maxrecs(cur->bc_mp,
+					ifp->if_broot_bytes, level == 0);
+	}
+
+	return cur->bc_mp->m_bmap_dmxr[level != 0];
+
+}
+
+/*
+ * Get the maximum records we could store in the on-disk format.
+ *
+ * For non-root nodes this is equivalent to xfs_bmbt_get_maxrecs, but
+ * for the root node this checks the available space in the dinode fork
+ * so that we can resize the in-memory buffer to match it.  After a
+ * resize to the maximum size this function returns the same value
+ * as xfs_bmbt_get_maxrecs for the root node, too.
+ */
+STATIC int
+xfs_bmbt_get_dmaxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
+{
+	if (level != cur->bc_nlevels - 1)
+		return cur->bc_mp->m_bmap_dmxr[level != 0];
+	return xfs_bmdr_maxrecs(cur->bc_mp, cur->bc_private.b.forksize,
+				level == 0);
+}
+
+STATIC void
+xfs_bmbt_init_key_from_rec(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	key->bmbt.br_startoff =
+		cpu_to_be64(xfs_bmbt_disk_get_startoff(&rec->bmbt));
+}
+
+STATIC void
+xfs_bmbt_init_rec_from_key(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
+{
+	ASSERT(key->bmbt.br_startoff != 0);
+
+	xfs_bmbt_disk_set_allf(&rec->bmbt, be64_to_cpu(key->bmbt.br_startoff),
+			       0, 0, XFS_EXT_NORM);
+}
+
+STATIC void
+xfs_bmbt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	xfs_bmbt_disk_set_all(&rec->bmbt, &cur->bc_rec.b);
+}
+
+STATIC void
+xfs_bmbt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	ptr->l = 0;
+}
+
+STATIC __int64_t
+xfs_bmbt_key_diff(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
+{
+	return (__int64_t)be64_to_cpu(key->bmbt.br_startoff) -
+				      cur->bc_rec.b.br_startoff;
+}
+
+#ifdef DEBUG
+STATIC int
+xfs_bmbt_keys_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*k1,
+	union xfs_btree_key	*k2)
+{
+	return be64_to_cpu(k1->bmbt.br_startoff) <
+		be64_to_cpu(k2->bmbt.br_startoff);
+}
+
+STATIC int
+xfs_bmbt_recs_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*r1,
+	union xfs_btree_rec	*r2)
+{
+	return xfs_bmbt_disk_get_startoff(&r1->bmbt) +
+		xfs_bmbt_disk_get_blockcount(&r1->bmbt) <=
+		xfs_bmbt_disk_get_startoff(&r2->bmbt);
+}
+#endif	/* DEBUG */
+
+#ifdef XFS_BTREE_TRACE
+ktrace_t	*xfs_bmbt_trace_buf;
+
+STATIC void
+xfs_bmbt_trace_enter(
+	struct xfs_btree_cur	*cur,
+	const char		*func,
+	char			*s,
+	int			type,
+	int			line,
+	__psunsigned_t		a0,
+	__psunsigned_t		a1,
+	__psunsigned_t		a2,
+	__psunsigned_t		a3,
+	__psunsigned_t		a4,
+	__psunsigned_t		a5,
+	__psunsigned_t		a6,
+	__psunsigned_t		a7,
+	__psunsigned_t		a8,
+	__psunsigned_t		a9,
+	__psunsigned_t		a10)
+{
+	struct xfs_inode	*ip = cur->bc_private.b.ip;
+	int			whichfork = cur->bc_private.b.whichfork;
+
+	ktrace_enter(xfs_bmbt_trace_buf,
+		(void *)((__psint_t)type | (whichfork << 8) | (line << 16)),
+		(void *)func, (void *)s, (void *)ip, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
+	ktrace_enter(ip->i_btrace,
+		(void *)((__psint_t)type | (whichfork << 8) | (line << 16)),
+		(void *)func, (void *)s, (void *)ip, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
+}
+
+STATIC void
+xfs_bmbt_trace_cursor(
+	struct xfs_btree_cur	*cur,
+	__uint32_t		*s0,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	struct xfs_bmbt_rec_host r;
+
+	xfs_bmbt_set_all(&r, &cur->bc_rec.b);
+
+	*s0 = (cur->bc_nlevels << 24) |
+	      (cur->bc_private.b.flags << 16) |
+	       cur->bc_private.b.allocated;
+	*l0 = r.l0;
+	*l1 = r.l1;
+}
+
+STATIC void
+xfs_bmbt_trace_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
+{
+	*l0 = be64_to_cpu(key->bmbt.br_startoff);
+	*l1 = 0;
+}
+
+STATIC void
+xfs_bmbt_trace_record(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	__uint64_t		*l0,
+	__uint64_t		*l1,
+	__uint64_t		*l2)
+{
+	struct xfs_bmbt_irec	irec;
+
+	xfs_bmbt_disk_get_all(&rec->bmbt, &irec);
+	*l0 = irec.br_startoff;
+	*l1 = irec.br_startblock;
+	*l2 = irec.br_blockcount;
+}
+#endif /* XFS_BTREE_TRACE */
+
+static const struct xfs_btree_ops xfs_bmbt_ops = {
+	.rec_len		= sizeof(xfs_bmbt_rec_t),
+	.key_len		= sizeof(xfs_bmbt_key_t),
+
+	.dup_cursor		= xfs_bmbt_dup_cursor,
+	.update_cursor		= xfs_bmbt_update_cursor,
+	.alloc_block		= xfs_bmbt_alloc_block,
+	.free_block		= xfs_bmbt_free_block,
+	.get_maxrecs		= xfs_bmbt_get_maxrecs,
+	.get_minrecs		= xfs_bmbt_get_minrecs,
+	.get_dmaxrecs		= xfs_bmbt_get_dmaxrecs,
+	.init_key_from_rec	= xfs_bmbt_init_key_from_rec,
+	.init_rec_from_key	= xfs_bmbt_init_rec_from_key,
+	.init_rec_from_cur	= xfs_bmbt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_bmbt_init_ptr_from_cur,
+	.key_diff		= xfs_bmbt_key_diff,
+
+#ifdef DEBUG
+	.keys_inorder		= xfs_bmbt_keys_inorder,
+	.recs_inorder		= xfs_bmbt_recs_inorder,
+#endif
+
+#ifdef XFS_BTREE_TRACE
+	.trace_enter		= xfs_bmbt_trace_enter,
+	.trace_cursor		= xfs_bmbt_trace_cursor,
+	.trace_key		= xfs_bmbt_trace_key,
+	.trace_record		= xfs_bmbt_trace_record,
+#endif
+};
+
+/*
+ * Allocate a new bmap btree cursor.
+ */
+struct xfs_btree_cur *				/* new bmap btree cursor */
+xfs_bmbt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* inode owning the btree */
+	int			whichfork)	/* data or attr fork */
+{
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
+	struct xfs_btree_cur	*cur;
+
+	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
+
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_nlevels = be16_to_cpu(ifp->if_broot->bb_level) + 1;
+	cur->bc_btnum = XFS_BTNUM_BMAP;
+	cur->bc_blocklog = mp->m_sb.sb_blocklog;
+
+	cur->bc_ops = &xfs_bmbt_ops;
+	cur->bc_flags = XFS_BTREE_LONG_PTRS | XFS_BTREE_ROOT_IN_INODE;
+
+	cur->bc_private.b.forksize = XFS_IFORK_SIZE(ip, whichfork);
+	cur->bc_private.b.ip = ip;
+	cur->bc_private.b.firstblock = NULLFSBLOCK;
+	cur->bc_private.b.flist = NULL;
+	cur->bc_private.b.allocated = 0;
+	cur->bc_private.b.flags = 0;
+	cur->bc_private.b.whichfork = whichfork;
+
+	return cur;
+}
+
+/*
+ * Calculate number of records in a bmap btree block.
+ */
+int
+xfs_bmbt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= XFS_BMBT_BLOCK_LEN(mp);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_bmbt_rec_t);
+	return blocklen / (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t));
+}
+
+/*
+ * Calculate number of records in a bmap btree inode root.
+ */
+int
+xfs_bmdr_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
+{
+	blocklen -= sizeof(xfs_bmdr_block_t);
+
+	if (leaf)
+		return blocklen / sizeof(xfs_bmdr_rec_t);
+	return blocklen / (sizeof(xfs_bmdr_key_t) + sizeof(xfs_bmdr_ptr_t));
+}
diff --git a/fs/xfs/xfs_bmap_btree.h b/fs/xfs/xfs_bmap_btree.h
index cd0d4b4bb81..a4555abb662 100644
--- a/fs/xfs/xfs_bmap_btree.h
+++ b/fs/xfs/xfs_bmap_btree.h
@@ -21,9 +21,10 @@
 #define XFS_BMAP_MAGIC	0x424d4150	/* 'BMAP' */
 
 struct xfs_btree_cur;
-struct xfs_btree_lblock;
+struct xfs_btree_block;
 struct xfs_mount;
 struct xfs_inode;
+struct xfs_trans;
 
 /*
  * Bmap root header, on-disk form only.
@@ -145,71 +146,60 @@ typedef struct xfs_bmbt_key {
 /* btree pointer type */
 typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
 
-/* btree block header type */
-typedef struct xfs_btree_lblock xfs_bmbt_block_t;
-
-#define XFS_BUF_TO_BMBT_BLOCK(bp)	((xfs_bmbt_block_t *)XFS_BUF_PTR(bp))
-
-#define XFS_BMAP_RBLOCK_DSIZE(lev,cur)	((cur)->bc_private.b.forksize)
-#define XFS_BMAP_RBLOCK_ISIZE(lev,cur)	\
-	((int)XFS_IFORK_PTR((cur)->bc_private.b.ip, \
-		    (cur)->bc_private.b.whichfork)->if_broot_bytes)
-
-#define XFS_BMAP_BLOCK_DMAXRECS(lev,cur) \
-	(((lev) == (cur)->bc_nlevels - 1 ? \
-		XFS_BTREE_BLOCK_MAXRECS(XFS_BMAP_RBLOCK_DSIZE(lev,cur), \
-			xfs_bmdr, (lev) == 0) : \
-		((cur)->bc_mp->m_bmap_dmxr[(lev) != 0])))
-#define XFS_BMAP_BLOCK_IMAXRECS(lev,cur) \
-	(((lev) == (cur)->bc_nlevels - 1 ? \
-			XFS_BTREE_BLOCK_MAXRECS(XFS_BMAP_RBLOCK_ISIZE(lev,cur),\
-				xfs_bmbt, (lev) == 0) : \
-			((cur)->bc_mp->m_bmap_dmxr[(lev) != 0])))
-
-#define XFS_BMAP_BLOCK_DMINRECS(lev,cur) \
-	(((lev) == (cur)->bc_nlevels - 1 ? \
-			XFS_BTREE_BLOCK_MINRECS(XFS_BMAP_RBLOCK_DSIZE(lev,cur),\
-				xfs_bmdr, (lev) == 0) : \
-			((cur)->bc_mp->m_bmap_dmnr[(lev) != 0])))
-#define XFS_BMAP_BLOCK_IMINRECS(lev,cur) \
-	(((lev) == (cur)->bc_nlevels - 1 ? \
-			XFS_BTREE_BLOCK_MINRECS(XFS_BMAP_RBLOCK_ISIZE(lev,cur),\
-				xfs_bmbt, (lev) == 0) : \
-			((cur)->bc_mp->m_bmap_dmnr[(lev) != 0])))
-
-#define XFS_BMAP_REC_DADDR(bb,i,cur)	(XFS_BTREE_REC_ADDR(xfs_bmbt, bb, i))
-
-#define XFS_BMAP_REC_IADDR(bb,i,cur)	(XFS_BTREE_REC_ADDR(xfs_bmbt, bb, i))
-
-#define XFS_BMAP_KEY_DADDR(bb,i,cur)	\
-	(XFS_BTREE_KEY_ADDR(xfs_bmbt, bb, i))
-
-#define XFS_BMAP_KEY_IADDR(bb,i,cur)	\
-	(XFS_BTREE_KEY_ADDR(xfs_bmbt, bb, i))
-
-#define XFS_BMAP_PTR_DADDR(bb,i,cur)	\
-	(XFS_BTREE_PTR_ADDR(xfs_bmbt, bb, i, XFS_BMAP_BLOCK_DMAXRECS(	\
-				be16_to_cpu((bb)->bb_level), cur)))
-#define XFS_BMAP_PTR_IADDR(bb,i,cur)	\
-	(XFS_BTREE_PTR_ADDR(xfs_bmbt, bb, i, XFS_BMAP_BLOCK_IMAXRECS(	\
-				be16_to_cpu((bb)->bb_level), cur)))
+/*
+ * Btree block header size depends on a superblock flag.
+ *
+ * (not quite yet, but soon)
+ */
+#define XFS_BMBT_BLOCK_LEN(mp)	XFS_BTREE_LBLOCK_LEN
+
+#define XFS_BMBT_REC_ADDR(mp, block, index) \
+	((xfs_bmbt_rec_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_rec_t)))
+
+#define XFS_BMBT_KEY_ADDR(mp, block, index) \
+	((xfs_bmbt_key_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_key_t)))
+
+#define XFS_BMBT_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_bmbt_ptr_t *) \
+		((char *)(block) + \
+		 XFS_BMBT_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_bmbt_key_t) + \
+		 ((index) - 1) * sizeof(xfs_bmbt_ptr_t)))
+
+#define XFS_BMDR_REC_ADDR(block, index) \
+	((xfs_bmdr_rec_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+	         ((index) - 1) * sizeof(xfs_bmdr_rec_t)))
+
+#define XFS_BMDR_KEY_ADDR(block, index) \
+	((xfs_bmdr_key_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+		 ((index) - 1) * sizeof(xfs_bmdr_key_t)))
+
+#define XFS_BMDR_PTR_ADDR(block, index, maxrecs) \
+	((xfs_bmdr_ptr_t *) \
+		((char *)(block) + \
+		 sizeof(struct xfs_bmdr_block) + \
+		 (maxrecs) * sizeof(xfs_bmdr_key_t) + \
+		 ((index) - 1) * sizeof(xfs_bmdr_ptr_t)))
 
 /*
  * These are to be used when we know the size of the block and
  * we don't have a cursor.
  */
-#define XFS_BMAP_BROOT_REC_ADDR(bb,i,sz) \
-	(XFS_BTREE_REC_ADDR(xfs_bmbt,bb,i))
-#define XFS_BMAP_BROOT_KEY_ADDR(bb,i,sz) \
-	(XFS_BTREE_KEY_ADDR(xfs_bmbt,bb,i))
-#define XFS_BMAP_BROOT_PTR_ADDR(bb,i,sz) \
-	(XFS_BTREE_PTR_ADDR(xfs_bmbt,bb,i,XFS_BMAP_BROOT_MAXRECS(sz)))
-
-#define XFS_BMAP_BROOT_NUMRECS(bb)	be16_to_cpu((bb)->bb_numrecs)
-#define XFS_BMAP_BROOT_MAXRECS(sz)	XFS_BTREE_BLOCK_MAXRECS(sz,xfs_bmbt,0)
+#define XFS_BMAP_BROOT_PTR_ADDR(mp, bb, i, sz) \
+	XFS_BMBT_PTR_ADDR(mp, bb, i, xfs_bmbt_maxrecs(mp, sz, 0))
 
 #define XFS_BMAP_BROOT_SPACE_CALC(nrecs) \
-	(int)(sizeof(xfs_bmbt_block_t) + \
+	(int)(XFS_BTREE_LBLOCK_LEN + \
 	       ((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
 
 #define XFS_BMAP_BROOT_SPACE(bb) \
@@ -223,42 +213,12 @@ typedef struct xfs_btree_lblock xfs_bmbt_block_t;
  */
 #define XFS_BM_MAXLEVELS(mp,w)		((mp)->m_bm_maxlevels[(w)])
 
-#define XFS_BMAP_SANITY_CHECK(mp,bb,level) \
-	(be32_to_cpu((bb)->bb_magic) == XFS_BMAP_MAGIC && \
-	 be16_to_cpu((bb)->bb_level) == level && \
-	 be16_to_cpu((bb)->bb_numrecs) > 0 && \
-	 be16_to_cpu((bb)->bb_numrecs) <= (mp)->m_bmap_dmxr[(level) != 0])
-
-
-#ifdef __KERNEL__
-
-#if defined(XFS_BMBT_TRACE)
-/*
- * Trace buffer entry types.
- */
-#define XFS_BMBT_KTRACE_ARGBI	1
-#define XFS_BMBT_KTRACE_ARGBII	2
-#define XFS_BMBT_KTRACE_ARGFFFI 3
-#define XFS_BMBT_KTRACE_ARGI	4
-#define XFS_BMBT_KTRACE_ARGIFK	5
-#define XFS_BMBT_KTRACE_ARGIFR	6
-#define XFS_BMBT_KTRACE_ARGIK	7
-#define XFS_BMBT_KTRACE_CUR	8
-
-#define XFS_BMBT_TRACE_SIZE	4096	/* size of global trace buffer */
-#define XFS_BMBT_KTRACE_SIZE	32	/* size of per-inode trace buffer */
-extern ktrace_t	*xfs_bmbt_trace_buf;
-#endif
-
 /*
  * Prototypes for xfs_bmap.c to call.
  */
-extern void xfs_bmdr_to_bmbt(xfs_bmdr_block_t *, int, xfs_bmbt_block_t *, int);
-extern int xfs_bmbt_decrement(struct xfs_btree_cur *, int, int *);
-extern int xfs_bmbt_delete(struct xfs_btree_cur *, int *);
+extern void xfs_bmdr_to_bmbt(struct xfs_mount *, xfs_bmdr_block_t *, int,
+			struct xfs_btree_block *, int);
 extern void xfs_bmbt_get_all(xfs_bmbt_rec_host_t *r, xfs_bmbt_irec_t *s);
-extern xfs_bmbt_block_t *xfs_bmbt_get_block(struct xfs_btree_cur *cur,
-						int, struct xfs_buf **bpp);
 extern xfs_filblks_t xfs_bmbt_get_blockcount(xfs_bmbt_rec_host_t *r);
 extern xfs_fsblock_t xfs_bmbt_get_startblock(xfs_bmbt_rec_host_t *r);
 extern xfs_fileoff_t xfs_bmbt_get_startoff(xfs_bmbt_rec_host_t *r);
@@ -268,22 +228,6 @@ extern void xfs_bmbt_disk_get_all(xfs_bmbt_rec_t *r, xfs_bmbt_irec_t *s);
 extern xfs_filblks_t xfs_bmbt_disk_get_blockcount(xfs_bmbt_rec_t *r);
 extern xfs_fileoff_t xfs_bmbt_disk_get_startoff(xfs_bmbt_rec_t *r);
 
-extern int xfs_bmbt_increment(struct xfs_btree_cur *, int, int *);
-extern int xfs_bmbt_insert(struct xfs_btree_cur *, int *);
-extern void xfs_bmbt_log_block(struct xfs_btree_cur *, struct xfs_buf *, int);
-extern void xfs_bmbt_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int,
-				int);
-extern int xfs_bmbt_lookup_eq(struct xfs_btree_cur *, xfs_fileoff_t,
-				xfs_fsblock_t, xfs_filblks_t, int *);
-extern int xfs_bmbt_lookup_ge(struct xfs_btree_cur *, xfs_fileoff_t,
-				xfs_fsblock_t, xfs_filblks_t, int *);
-
-/*
- * Give the bmap btree a new root block.  Copy the old broot contents
- * down into a real block and make the broot point to it.
- */
-extern int xfs_bmbt_newroot(struct xfs_btree_cur *cur, int *lflags, int *stat);
-
 extern void xfs_bmbt_set_all(xfs_bmbt_rec_host_t *r, xfs_bmbt_irec_t *s);
 extern void xfs_bmbt_set_allf(xfs_bmbt_rec_host_t *r, xfs_fileoff_t o,
 			xfs_fsblock_t b, xfs_filblks_t c, xfs_exntst_t v);
@@ -296,10 +240,15 @@ extern void xfs_bmbt_disk_set_all(xfs_bmbt_rec_t *r, xfs_bmbt_irec_t *s);
 extern void xfs_bmbt_disk_set_allf(xfs_bmbt_rec_t *r, xfs_fileoff_t o,
 			xfs_fsblock_t b, xfs_filblks_t c, xfs_exntst_t v);
 
-extern void xfs_bmbt_to_bmdr(xfs_bmbt_block_t *, int, xfs_bmdr_block_t *, int);
-extern int xfs_bmbt_update(struct xfs_btree_cur *, xfs_fileoff_t,
-				xfs_fsblock_t, xfs_filblks_t, xfs_exntst_t);
+extern void xfs_bmbt_to_bmdr(struct xfs_mount *, struct xfs_btree_block *, int,
+			xfs_bmdr_block_t *, int);
+
+extern int xfs_bmbt_get_maxrecs(struct xfs_btree_cur *, int level);
+extern int xfs_bmdr_maxrecs(struct xfs_mount *, int blocklen, int leaf);
+extern int xfs_bmbt_maxrecs(struct xfs_mount *, int blocklen, int leaf);
+
+extern struct xfs_btree_cur *xfs_bmbt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_inode *, int);
 
-#endif	/* __KERNEL__ */
 
 #endif	/* __XFS_BMAP_BTREE_H__ */
diff --git a/fs/xfs/xfs_btree.c b/fs/xfs/xfs_btree.c
index cc593a84c34..7ed59267420 100644
--- a/fs/xfs/xfs_btree.c
+++ b/fs/xfs/xfs_btree.c
@@ -34,7 +34,9 @@
 #include "xfs_attr_sf.h"
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
+#include "xfs_inode_item.h"
 #include "xfs_btree.h"
+#include "xfs_btree_trace.h"
 #include "xfs_ialloc.h"
 #include "xfs_error.h"
 
@@ -50,135 +52,33 @@ const __uint32_t xfs_magics[XFS_BTNUM_MAX] = {
 	XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, XFS_BMAP_MAGIC, XFS_IBT_MAGIC
 };
 
-/*
- * Checking routine: return maxrecs for the block.
- */
-STATIC int				/* number of records fitting in block */
-xfs_btree_maxrecs(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_block_t	*block)	/* generic btree block pointer */
-{
-	switch (cur->bc_btnum) {
-	case XFS_BTNUM_BNO:
-	case XFS_BTNUM_CNT:
-		return (int)XFS_ALLOC_BLOCK_MAXRECS(
-				be16_to_cpu(block->bb_h.bb_level), cur);
-	case XFS_BTNUM_BMAP:
-		return (int)XFS_BMAP_BLOCK_IMAXRECS(
-				be16_to_cpu(block->bb_h.bb_level), cur);
-	case XFS_BTNUM_INO:
-		return (int)XFS_INOBT_BLOCK_MAXRECS(
-				be16_to_cpu(block->bb_h.bb_level), cur);
-	default:
-		ASSERT(0);
-		return 0;
-	}
-}
-
-/*
- * External routines.
- */
-
-#ifdef DEBUG
-/*
- * Debug routine: check that block header is ok.
- */
-void
-xfs_btree_check_block(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_block_t	*block,	/* generic btree block pointer */
-	int			level,	/* level of the btree block */
-	xfs_buf_t		*bp)	/* buffer containing block, if any */
-{
-	if (XFS_BTREE_LONG_PTRS(cur->bc_btnum))
-		xfs_btree_check_lblock(cur, (xfs_btree_lblock_t *)block, level,
-			bp);
-	else
-		xfs_btree_check_sblock(cur, (xfs_btree_sblock_t *)block, level,
-			bp);
-}
-
-/*
- * Debug routine: check that keys are in the right order.
- */
-void
-xfs_btree_check_key(
-	xfs_btnum_t	btnum,		/* btree identifier */
-	void		*ak1,		/* pointer to left (lower) key */
-	void		*ak2)		/* pointer to right (higher) key */
-{
-	switch (btnum) {
-	case XFS_BTNUM_BNO: {
-		xfs_alloc_key_t	*k1;
-		xfs_alloc_key_t	*k2;
-
-		k1 = ak1;
-		k2 = ak2;
-		ASSERT(be32_to_cpu(k1->ar_startblock) < be32_to_cpu(k2->ar_startblock));
-		break;
-	    }
-	case XFS_BTNUM_CNT: {
-		xfs_alloc_key_t	*k1;
-		xfs_alloc_key_t	*k2;
-
-		k1 = ak1;
-		k2 = ak2;
-		ASSERT(be32_to_cpu(k1->ar_blockcount) < be32_to_cpu(k2->ar_blockcount) ||
-		       (k1->ar_blockcount == k2->ar_blockcount &&
-			be32_to_cpu(k1->ar_startblock) < be32_to_cpu(k2->ar_startblock)));
-		break;
-	    }
-	case XFS_BTNUM_BMAP: {
-		xfs_bmbt_key_t	*k1;
-		xfs_bmbt_key_t	*k2;
-
-		k1 = ak1;
-		k2 = ak2;
-		ASSERT(be64_to_cpu(k1->br_startoff) < be64_to_cpu(k2->br_startoff));
-		break;
-	    }
-	case XFS_BTNUM_INO: {
-		xfs_inobt_key_t	*k1;
-		xfs_inobt_key_t	*k2;
-
-		k1 = ak1;
-		k2 = ak2;
-		ASSERT(be32_to_cpu(k1->ir_startino) < be32_to_cpu(k2->ir_startino));
-		break;
-	    }
-	default:
-		ASSERT(0);
-	}
-}
-#endif	/* DEBUG */
 
-/*
- * Checking routine: check that long form block header is ok.
- */
-/* ARGSUSED */
-int					/* error (0 or EFSCORRUPTED) */
+STATIC int				/* error (0 or EFSCORRUPTED) */
 xfs_btree_check_lblock(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_lblock_t	*block,	/* btree long form block pointer */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* btree long form block pointer */
 	int			level,	/* level of the btree block */
-	xfs_buf_t		*bp)	/* buffer for block, if any */
+	struct xfs_buf		*bp)	/* buffer for block, if any */
 {
 	int			lblock_ok; /* block passes checks */
-	xfs_mount_t		*mp;	/* file system mount point */
+	struct xfs_mount	*mp;	/* file system mount point */
 
 	mp = cur->bc_mp;
 	lblock_ok =
 		be32_to_cpu(block->bb_magic) == xfs_magics[cur->bc_btnum] &&
 		be16_to_cpu(block->bb_level) == level &&
 		be16_to_cpu(block->bb_numrecs) <=
-			xfs_btree_maxrecs(cur, (xfs_btree_block_t *)block) &&
-		block->bb_leftsib &&
-		(be64_to_cpu(block->bb_leftsib) == NULLDFSBNO ||
-		 XFS_FSB_SANITY_CHECK(mp, be64_to_cpu(block->bb_leftsib))) &&
-		block->bb_rightsib &&
-		(be64_to_cpu(block->bb_rightsib) == NULLDFSBNO ||
-		 XFS_FSB_SANITY_CHECK(mp, be64_to_cpu(block->bb_rightsib)));
-	if (unlikely(XFS_TEST_ERROR(!lblock_ok, mp, XFS_ERRTAG_BTREE_CHECK_LBLOCK,
+			cur->bc_ops->get_maxrecs(cur, level) &&
+		block->bb_u.l.bb_leftsib &&
+		(be64_to_cpu(block->bb_u.l.bb_leftsib) == NULLDFSBNO ||
+		 XFS_FSB_SANITY_CHECK(mp,
+		 	be64_to_cpu(block->bb_u.l.bb_leftsib))) &&
+		block->bb_u.l.bb_rightsib &&
+		(be64_to_cpu(block->bb_u.l.bb_rightsib) == NULLDFSBNO ||
+		 XFS_FSB_SANITY_CHECK(mp,
+		 	be64_to_cpu(block->bb_u.l.bb_rightsib)));
+	if (unlikely(XFS_TEST_ERROR(!lblock_ok, mp,
+			XFS_ERRTAG_BTREE_CHECK_LBLOCK,
 			XFS_RANDOM_BTREE_CHECK_LBLOCK))) {
 		if (bp)
 			xfs_buftrace("LBTREE ERROR", bp);
@@ -189,98 +89,15 @@ xfs_btree_check_lblock(
 	return 0;
 }
 
-/*
- * Checking routine: check that (long) pointer is ok.
- */
-int					/* error (0 or EFSCORRUPTED) */
-xfs_btree_check_lptr(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_dfsbno_t	ptr,		/* btree block disk address */
-	int		level)		/* btree block level */
-{
-	xfs_mount_t	*mp;		/* file system mount point */
-
-	mp = cur->bc_mp;
-	XFS_WANT_CORRUPTED_RETURN(
-		level > 0 &&
-		ptr != NULLDFSBNO &&
-		XFS_FSB_SANITY_CHECK(mp, ptr));
-	return 0;
-}
-
-#ifdef DEBUG
-/*
- * Debug routine: check that records are in the right order.
- */
-void
-xfs_btree_check_rec(
-	xfs_btnum_t	btnum,		/* btree identifier */
-	void		*ar1,		/* pointer to left (lower) record */
-	void		*ar2)		/* pointer to right (higher) record */
-{
-	switch (btnum) {
-	case XFS_BTNUM_BNO: {
-		xfs_alloc_rec_t	*r1;
-		xfs_alloc_rec_t	*r2;
-
-		r1 = ar1;
-		r2 = ar2;
-		ASSERT(be32_to_cpu(r1->ar_startblock) +
-		       be32_to_cpu(r1->ar_blockcount) <=
-		       be32_to_cpu(r2->ar_startblock));
-		break;
-	    }
-	case XFS_BTNUM_CNT: {
-		xfs_alloc_rec_t	*r1;
-		xfs_alloc_rec_t	*r2;
-
-		r1 = ar1;
-		r2 = ar2;
-		ASSERT(be32_to_cpu(r1->ar_blockcount) < be32_to_cpu(r2->ar_blockcount) ||
-		       (r1->ar_blockcount == r2->ar_blockcount &&
-			be32_to_cpu(r1->ar_startblock) < be32_to_cpu(r2->ar_startblock)));
-		break;
-	    }
-	case XFS_BTNUM_BMAP: {
-		xfs_bmbt_rec_t	*r1;
-		xfs_bmbt_rec_t	*r2;
-
-		r1 = ar1;
-		r2 = ar2;
-		ASSERT(xfs_bmbt_disk_get_startoff(r1) +
-		       xfs_bmbt_disk_get_blockcount(r1) <=
-		       xfs_bmbt_disk_get_startoff(r2));
-		break;
-	    }
-	case XFS_BTNUM_INO: {
-		xfs_inobt_rec_t	*r1;
-		xfs_inobt_rec_t	*r2;
-
-		r1 = ar1;
-		r2 = ar2;
-		ASSERT(be32_to_cpu(r1->ir_startino) + XFS_INODES_PER_CHUNK <=
-		       be32_to_cpu(r2->ir_startino));
-		break;
-	    }
-	default:
-		ASSERT(0);
-	}
-}
-#endif	/* DEBUG */
-
-/*
- * Checking routine: check that block header is ok.
- */
-/* ARGSUSED */
-int					/* error (0 or EFSCORRUPTED) */
+STATIC int				/* error (0 or EFSCORRUPTED) */
 xfs_btree_check_sblock(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_sblock_t	*block,	/* btree short form block pointer */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* btree short form block pointer */
 	int			level,	/* level of the btree block */
-	xfs_buf_t		*bp)	/* buffer containing block */
+	struct xfs_buf		*bp)	/* buffer containing block */
 {
-	xfs_buf_t		*agbp;	/* buffer for ag. freespace struct */
-	xfs_agf_t		*agf;	/* ag. freespace structure */
+	struct xfs_buf		*agbp;	/* buffer for ag. freespace struct */
+	struct xfs_agf		*agf;	/* ag. freespace structure */
 	xfs_agblock_t		agflen;	/* native ag. freespace length */
 	int			sblock_ok; /* block passes checks */
 
@@ -291,13 +108,13 @@ xfs_btree_check_sblock(
 		be32_to_cpu(block->bb_magic) == xfs_magics[cur->bc_btnum] &&
 		be16_to_cpu(block->bb_level) == level &&
 		be16_to_cpu(block->bb_numrecs) <=
-			xfs_btree_maxrecs(cur, (xfs_btree_block_t *)block) &&
-		(be32_to_cpu(block->bb_leftsib) == NULLAGBLOCK ||
-		 be32_to_cpu(block->bb_leftsib) < agflen) &&
-		block->bb_leftsib &&
-		(be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK ||
-		 be32_to_cpu(block->bb_rightsib) < agflen) &&
-		block->bb_rightsib;
+			cur->bc_ops->get_maxrecs(cur, level) &&
+		(be32_to_cpu(block->bb_u.s.bb_leftsib) == NULLAGBLOCK ||
+		 be32_to_cpu(block->bb_u.s.bb_leftsib) < agflen) &&
+		block->bb_u.s.bb_leftsib &&
+		(be32_to_cpu(block->bb_u.s.bb_rightsib) == NULLAGBLOCK ||
+		 be32_to_cpu(block->bb_u.s.bb_rightsib) < agflen) &&
+		block->bb_u.s.bb_rightsib;
 	if (unlikely(XFS_TEST_ERROR(!sblock_ok, cur->bc_mp,
 			XFS_ERRTAG_BTREE_CHECK_SBLOCK,
 			XFS_RANDOM_BTREE_CHECK_SBLOCK))) {
@@ -311,27 +128,78 @@ xfs_btree_check_sblock(
 }
 
 /*
- * Checking routine: check that (short) pointer is ok.
+ * Debug routine: check that block header is ok.
+ */
+int
+xfs_btree_check_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* generic btree block pointer */
+	int			level,	/* level of the btree block */
+	struct xfs_buf		*bp)	/* buffer containing block, if any */
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return xfs_btree_check_lblock(cur, block, level, bp);
+	else
+		return xfs_btree_check_sblock(cur, block, level, bp);
+}
+
+/*
+ * Check that (long) pointer is ok.
  */
 int					/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_lptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_dfsbno_t		bno,	/* btree block disk address */
+	int			level)	/* btree block level */
+{
+	XFS_WANT_CORRUPTED_RETURN(
+		level > 0 &&
+		bno != NULLDFSBNO &&
+		XFS_FSB_SANITY_CHECK(cur->bc_mp, bno));
+	return 0;
+}
+
+#ifdef DEBUG
+/*
+ * Check that (short) pointer is ok.
+ */
+STATIC int				/* error (0 or EFSCORRUPTED) */
 xfs_btree_check_sptr(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agblock_t	ptr,		/* btree block disk address */
-	int		level)		/* btree block level */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agblock_t		bno,	/* btree block disk address */
+	int			level)	/* btree block level */
 {
-	xfs_buf_t	*agbp;		/* buffer for ag. freespace struct */
-	xfs_agf_t	*agf;		/* ag. freespace structure */
+	xfs_agblock_t		agblocks = cur->bc_mp->m_sb.sb_agblocks;
 
-	agbp = cur->bc_private.a.agbp;
-	agf = XFS_BUF_TO_AGF(agbp);
 	XFS_WANT_CORRUPTED_RETURN(
 		level > 0 &&
-		ptr != NULLAGBLOCK && ptr != 0 &&
-		ptr < be32_to_cpu(agf->agf_length));
+		bno != NULLAGBLOCK &&
+		bno != 0 &&
+		bno < agblocks);
 	return 0;
 }
 
 /*
+ * Check that block ptr is ok.
+ */
+STATIC int				/* error (0 or EFSCORRUPTED) */
+xfs_btree_check_ptr(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	union xfs_btree_ptr	*ptr,	/* btree block disk address */
+	int			index,	/* offset from ptr to check */
+	int			level)	/* btree block level */
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		return xfs_btree_check_lptr(cur,
+				be64_to_cpu((&ptr->l)[index]), level);
+	} else {
+		return xfs_btree_check_sptr(cur,
+				be32_to_cpu((&ptr->s)[index]), level);
+	}
+}
+#endif
+
+/*
  * Delete the btree cursor.
  */
 void
@@ -387,16 +255,17 @@ xfs_btree_dup_cursor(
 
 	tp = cur->bc_tp;
 	mp = cur->bc_mp;
+
 	/*
 	 * Allocate a new cursor like the old one.
 	 */
-	new = xfs_btree_init_cursor(mp, tp, cur->bc_private.a.agbp,
-		cur->bc_private.a.agno, cur->bc_btnum, cur->bc_private.b.ip,
-		cur->bc_private.b.whichfork);
+	new = cur->bc_ops->dup_cursor(cur);
+
 	/*
 	 * Copy the record currently in the cursor.
 	 */
 	new->bc_rec = cur->bc_rec;
+
 	/*
 	 * For each level current, re-get the buffer and copy the ptr value.
 	 */
@@ -416,46 +285,174 @@ xfs_btree_dup_cursor(
 		} else
 			new->bc_bufs[i] = NULL;
 	}
-	/*
-	 * For bmap btrees, copy the firstblock, flist, and flags values,
-	 * since init cursor doesn't get them.
-	 */
-	if (new->bc_btnum == XFS_BTNUM_BMAP) {
-		new->bc_private.b.firstblock = cur->bc_private.b.firstblock;
-		new->bc_private.b.flist = cur->bc_private.b.flist;
-		new->bc_private.b.flags = cur->bc_private.b.flags;
-	}
 	*ncur = new;
 	return 0;
 }
 
 /*
+ * XFS btree block layout and addressing:
+ *
+ * There are two types of blocks in the btree: leaf and non-leaf blocks.
+ *
+ * The leaf record start with a header then followed by records containing
+ * the values.  A non-leaf block also starts with the same header, and
+ * then first contains lookup keys followed by an equal number of pointers
+ * to the btree blocks at the previous level.
+ *
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ * Leaf:	| header | rec 1 | rec 2 | rec 3 | rec 4 | rec 5 | rec N |
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ *
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ * Non-Leaf:	| header | key 1 | key 2 | key N | ptr 1 | ptr 2 | ptr N |
+ *		+--------+-------+-------+-------+-------+-------+-------+
+ *
+ * The header is called struct xfs_btree_block for reasons better left unknown
+ * and comes in different versions for short (32bit) and long (64bit) block
+ * pointers.  The record and key structures are defined by the btree instances
+ * and opaque to the btree core.  The block pointers are simple disk endian
+ * integers, available in a short (32bit) and long (64bit) variant.
+ *
+ * The helpers below calculate the offset of a given record, key or pointer
+ * into a btree block (xfs_btree_*_offset) or return a pointer to the given
+ * record, key or pointer (xfs_btree_*_addr).  Note that all addressing
+ * inside the btree block is done using indices starting at one, not zero!
+ */
+
+/*
+ * Return size of the btree block header for this btree instance.
+ */
+static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
+{
+	return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+		XFS_BTREE_LBLOCK_LEN :
+		XFS_BTREE_SBLOCK_LEN;
+}
+
+/*
+ * Return size of btree block pointers for this btree instance.
+ */
+static inline size_t xfs_btree_ptr_len(struct xfs_btree_cur *cur)
+{
+	return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+		sizeof(__be64) : sizeof(__be32);
+}
+
+/*
+ * Calculate offset of the n-th record in a btree block.
+ */
+STATIC size_t
+xfs_btree_rec_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->rec_len;
+}
+
+/*
+ * Calculate offset of the n-th key in a btree block.
+ */
+STATIC size_t
+xfs_btree_key_offset(
+	struct xfs_btree_cur	*cur,
+	int			n)
+{
+	return xfs_btree_block_len(cur) +
+		(n - 1) * cur->bc_ops->key_len;
+}
+
+/*
+ * Calculate offset of the n-th block pointer in a btree block.
+ */
+STATIC size_t
+xfs_btree_ptr_offset(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	int			level)
+{
+	return xfs_btree_block_len(cur) +
+		cur->bc_ops->get_maxrecs(cur, level) * cur->bc_ops->key_len +
+		(n - 1) * xfs_btree_ptr_len(cur);
+}
+
+/*
+ * Return a pointer to the n-th record in the btree block.
+ */
+STATIC union xfs_btree_rec *
+xfs_btree_rec_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_rec *)
+		((char *)block + xfs_btree_rec_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th key in the btree block.
+ */
+STATIC union xfs_btree_key *
+xfs_btree_key_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	return (union xfs_btree_key *)
+		((char *)block + xfs_btree_key_offset(cur, n));
+}
+
+/*
+ * Return a pointer to the n-th block pointer in the btree block.
+ */
+STATIC union xfs_btree_ptr *
+xfs_btree_ptr_addr(
+	struct xfs_btree_cur	*cur,
+	int			n,
+	struct xfs_btree_block	*block)
+{
+	int			level = xfs_btree_get_level(block);
+
+	ASSERT(block->bb_level != 0);
+
+	return (union xfs_btree_ptr *)
+		((char *)block + xfs_btree_ptr_offset(cur, n, level));
+}
+
+/*
+ * Get a the root block which is stored in the inode.
+ *
+ * For now this btree implementation assumes the btree root is always
+ * stored in the if_broot field of an inode fork.
+ */
+STATIC struct xfs_btree_block *
+xfs_btree_get_iroot(
+       struct xfs_btree_cur    *cur)
+{
+       struct xfs_ifork        *ifp;
+
+       ifp = XFS_IFORK_PTR(cur->bc_private.b.ip, cur->bc_private.b.whichfork);
+       return (struct xfs_btree_block *)ifp->if_broot;
+}
+
+/*
  * Retrieve the block pointer from the cursor at the given level.
- * This may be a bmap btree root or from a buffer.
+ * This may be an inode btree root or from a buffer.
  */
-STATIC xfs_btree_block_t *		/* generic btree block pointer */
+STATIC struct xfs_btree_block *		/* generic btree block pointer */
 xfs_btree_get_block(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
 	int			level,	/* level in btree */
-	xfs_buf_t		**bpp)	/* buffer containing the block */
-{
-	xfs_btree_block_t	*block;	/* return value */
-	xfs_buf_t		*bp;	/* return buffer */
-	xfs_ifork_t		*ifp;	/* inode fork pointer */
-	int			whichfork; /* data or attr fork */
-
-	if (cur->bc_btnum == XFS_BTNUM_BMAP && level == cur->bc_nlevels - 1) {
-		whichfork = cur->bc_private.b.whichfork;
-		ifp = XFS_IFORK_PTR(cur->bc_private.b.ip, whichfork);
-		block = (xfs_btree_block_t *)ifp->if_broot;
-		bp = NULL;
-	} else {
-		bp = cur->bc_bufs[level];
-		block = XFS_BUF_TO_BLOCK(bp);
+	struct xfs_buf		**bpp)	/* buffer containing the block */
+{
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1)) {
+		*bpp = NULL;
+		return xfs_btree_get_iroot(cur);
 	}
-	ASSERT(block != NULL);
-	*bpp = bp;
-	return block;
+
+	*bpp = cur->bc_bufs[level];
+	return XFS_BUF_TO_BLOCK(*bpp);
 }
 
 /*
@@ -505,97 +502,6 @@ xfs_btree_get_bufs(
 }
 
 /*
- * Allocate a new btree cursor.
- * The cursor is either for allocation (A) or bmap (B) or inodes (I).
- */
-xfs_btree_cur_t *			/* new btree cursor */
-xfs_btree_init_cursor(
-	xfs_mount_t	*mp,		/* file system mount point */
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_buf_t	*agbp,		/* (A only) buffer for agf structure */
-					/* (I only) buffer for agi structure */
-	xfs_agnumber_t	agno,		/* (AI only) allocation group number */
-	xfs_btnum_t	btnum,		/* btree identifier */
-	xfs_inode_t	*ip,		/* (B only) inode owning the btree */
-	int		whichfork)	/* (B only) data or attr fork */
-{
-	xfs_agf_t	*agf;		/* (A) allocation group freespace */
-	xfs_agi_t	*agi;		/* (I) allocation group inodespace */
-	xfs_btree_cur_t	*cur;		/* return value */
-	xfs_ifork_t	*ifp;		/* (I) inode fork pointer */
-	int		nlevels=0;	/* number of levels in the btree */
-
-	ASSERT(xfs_btree_cur_zone != NULL);
-	/*
-	 * Allocate a new cursor.
-	 */
-	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
-	/*
-	 * Deduce the number of btree levels from the arguments.
-	 */
-	switch (btnum) {
-	case XFS_BTNUM_BNO:
-	case XFS_BTNUM_CNT:
-		agf = XFS_BUF_TO_AGF(agbp);
-		nlevels = be32_to_cpu(agf->agf_levels[btnum]);
-		break;
-	case XFS_BTNUM_BMAP:
-		ifp = XFS_IFORK_PTR(ip, whichfork);
-		nlevels = be16_to_cpu(ifp->if_broot->bb_level) + 1;
-		break;
-	case XFS_BTNUM_INO:
-		agi = XFS_BUF_TO_AGI(agbp);
-		nlevels = be32_to_cpu(agi->agi_level);
-		break;
-	default:
-		ASSERT(0);
-	}
-	/*
-	 * Fill in the common fields.
-	 */
-	cur->bc_tp = tp;
-	cur->bc_mp = mp;
-	cur->bc_nlevels = nlevels;
-	cur->bc_btnum = btnum;
-	cur->bc_blocklog = mp->m_sb.sb_blocklog;
-	/*
-	 * Fill in private fields.
-	 */
-	switch (btnum) {
-	case XFS_BTNUM_BNO:
-	case XFS_BTNUM_CNT:
-		/*
-		 * Allocation btree fields.
-		 */
-		cur->bc_private.a.agbp = agbp;
-		cur->bc_private.a.agno = agno;
-		break;
-	case XFS_BTNUM_INO:
-		/*
-		 * Inode allocation btree fields.
-		 */
-		cur->bc_private.a.agbp = agbp;
-		cur->bc_private.a.agno = agno;
-		break;
-	case XFS_BTNUM_BMAP:
-		/*
-		 * Bmap btree fields.
-		 */
-		cur->bc_private.b.forksize = XFS_IFORK_SIZE(ip, whichfork);
-		cur->bc_private.b.ip = ip;
-		cur->bc_private.b.firstblock = NULLFSBLOCK;
-		cur->bc_private.b.flist = NULL;
-		cur->bc_private.b.allocated = 0;
-		cur->bc_private.b.flags = 0;
-		cur->bc_private.b.whichfork = whichfork;
-		break;
-	default:
-		ASSERT(0);
-	}
-	return cur;
-}
-
-/*
  * Check for the cursor referring to the last block at the given level.
  */
 int					/* 1=is last block, 0=not last block */
@@ -603,12 +509,12 @@ xfs_btree_islastblock(
 	xfs_btree_cur_t		*cur,	/* btree cursor */
 	int			level)	/* level to check */
 {
-	xfs_btree_block_t	*block;	/* generic btree block pointer */
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
 	xfs_buf_t		*bp;	/* buffer containing block */
 
 	block = xfs_btree_get_block(cur, level, &bp);
 	xfs_btree_check_block(cur, block, level, bp);
-	if (XFS_BTREE_LONG_PTRS(cur->bc_btnum))
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
 		return be64_to_cpu(block->bb_u.l.bb_rightsib) == NULLDFSBNO;
 	else
 		return be32_to_cpu(block->bb_u.s.bb_rightsib) == NULLAGBLOCK;
@@ -618,12 +524,12 @@ xfs_btree_islastblock(
  * Change the cursor to point to the first record at the given level.
  * Other levels are unaffected.
  */
-int					/* success=1, failure=0 */
+STATIC int				/* success=1, failure=0 */
 xfs_btree_firstrec(
 	xfs_btree_cur_t		*cur,	/* btree cursor */
 	int			level)	/* level to change */
 {
-	xfs_btree_block_t	*block;	/* generic btree block pointer */
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
 	xfs_buf_t		*bp;	/* buffer containing block */
 
 	/*
@@ -634,7 +540,7 @@ xfs_btree_firstrec(
 	/*
 	 * It's empty, there is no such record.
 	 */
-	if (!block->bb_h.bb_numrecs)
+	if (!block->bb_numrecs)
 		return 0;
 	/*
 	 * Set the ptr value to 1, that's the first record/key.
@@ -647,12 +553,12 @@ xfs_btree_firstrec(
  * Change the cursor to point to the last record in the current block
  * at the given level.  Other levels are unaffected.
  */
-int					/* success=1, failure=0 */
+STATIC int				/* success=1, failure=0 */
 xfs_btree_lastrec(
 	xfs_btree_cur_t		*cur,	/* btree cursor */
 	int			level)	/* level to change */
 {
-	xfs_btree_block_t	*block;	/* generic btree block pointer */
+	struct xfs_btree_block	*block;	/* generic btree block pointer */
 	xfs_buf_t		*bp;	/* buffer containing block */
 
 	/*
@@ -663,12 +569,12 @@ xfs_btree_lastrec(
 	/*
 	 * It's empty, there is no such record.
 	 */
-	if (!block->bb_h.bb_numrecs)
+	if (!block->bb_numrecs)
 		return 0;
 	/*
 	 * Set the ptr value to numrecs, that's the last record/key.
 	 */
-	cur->bc_ptrs[level] = be16_to_cpu(block->bb_h.bb_numrecs);
+	cur->bc_ptrs[level] = be16_to_cpu(block->bb_numrecs);
 	return 1;
 }
 
@@ -817,66 +723,84 @@ xfs_btree_reada_bufs(
 	xfs_baread(mp->m_ddev_targp, d, mp->m_bsize * count);
 }
 
+STATIC int
+xfs_btree_readahead_lblock(
+	struct xfs_btree_cur	*cur,
+	int			lr,
+	struct xfs_btree_block	*block)
+{
+	int			rval = 0;
+	xfs_fsblock_t		left = be64_to_cpu(block->bb_u.l.bb_leftsib);
+	xfs_fsblock_t		right = be64_to_cpu(block->bb_u.l.bb_rightsib);
+
+	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLDFSBNO) {
+		xfs_btree_reada_bufl(cur->bc_mp, left, 1);
+		rval++;
+	}
+
+	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLDFSBNO) {
+		xfs_btree_reada_bufl(cur->bc_mp, right, 1);
+		rval++;
+	}
+
+	return rval;
+}
+
+STATIC int
+xfs_btree_readahead_sblock(
+	struct xfs_btree_cur	*cur,
+	int			lr,
+	struct xfs_btree_block *block)
+{
+	int			rval = 0;
+	xfs_agblock_t		left = be32_to_cpu(block->bb_u.s.bb_leftsib);
+	xfs_agblock_t		right = be32_to_cpu(block->bb_u.s.bb_rightsib);
+
+
+	if ((lr & XFS_BTCUR_LEFTRA) && left != NULLAGBLOCK) {
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
+				     left, 1);
+		rval++;
+	}
+
+	if ((lr & XFS_BTCUR_RIGHTRA) && right != NULLAGBLOCK) {
+		xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
+				     right, 1);
+		rval++;
+	}
+
+	return rval;
+}
+
 /*
  * Read-ahead btree blocks, at the given level.
  * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
  */
-int
-xfs_btree_readahead_core(
-	xfs_btree_cur_t		*cur,		/* btree cursor */
+STATIC int
+xfs_btree_readahead(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
 	int			lev,		/* level in btree */
 	int			lr)		/* left/right bits */
 {
-	xfs_alloc_block_t	*a;
-	xfs_bmbt_block_t	*b;
-	xfs_inobt_block_t	*i;
-	int			rval = 0;
+	struct xfs_btree_block	*block;
+
+	/*
+	 * No readahead needed if we are at the root level and the
+	 * btree root is stored in the inode.
+	 */
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (lev == cur->bc_nlevels - 1))
+		return 0;
+
+	if ((cur->bc_ra[lev] | lr) == cur->bc_ra[lev])
+		return 0;
 
-	ASSERT(cur->bc_bufs[lev] != NULL);
 	cur->bc_ra[lev] |= lr;
-	switch (cur->bc_btnum) {
-	case XFS_BTNUM_BNO:
-	case XFS_BTNUM_CNT:
-		a = XFS_BUF_TO_ALLOC_BLOCK(cur->bc_bufs[lev]);
-		if ((lr & XFS_BTCUR_LEFTRA) && be32_to_cpu(a->bb_leftsib) != NULLAGBLOCK) {
-			xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
-				be32_to_cpu(a->bb_leftsib), 1);
-			rval++;
-		}
-		if ((lr & XFS_BTCUR_RIGHTRA) && be32_to_cpu(a->bb_rightsib) != NULLAGBLOCK) {
-			xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
-				be32_to_cpu(a->bb_rightsib), 1);
-			rval++;
-		}
-		break;
-	case XFS_BTNUM_BMAP:
-		b = XFS_BUF_TO_BMBT_BLOCK(cur->bc_bufs[lev]);
-		if ((lr & XFS_BTCUR_LEFTRA) && be64_to_cpu(b->bb_leftsib) != NULLDFSBNO) {
-			xfs_btree_reada_bufl(cur->bc_mp, be64_to_cpu(b->bb_leftsib), 1);
-			rval++;
-		}
-		if ((lr & XFS_BTCUR_RIGHTRA) && be64_to_cpu(b->bb_rightsib) != NULLDFSBNO) {
-			xfs_btree_reada_bufl(cur->bc_mp, be64_to_cpu(b->bb_rightsib), 1);
-			rval++;
-		}
-		break;
-	case XFS_BTNUM_INO:
-		i = XFS_BUF_TO_INOBT_BLOCK(cur->bc_bufs[lev]);
-		if ((lr & XFS_BTCUR_LEFTRA) && be32_to_cpu(i->bb_leftsib) != NULLAGBLOCK) {
-			xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
-				be32_to_cpu(i->bb_leftsib), 1);
-			rval++;
-		}
-		if ((lr & XFS_BTCUR_RIGHTRA) && be32_to_cpu(i->bb_rightsib) != NULLAGBLOCK) {
-			xfs_btree_reada_bufs(cur->bc_mp, cur->bc_private.a.agno,
-				be32_to_cpu(i->bb_rightsib), 1);
-			rval++;
-		}
-		break;
-	default:
-		ASSERT(0);
-	}
-	return rval;
+	block = XFS_BUF_TO_BLOCK(cur->bc_bufs[lev]);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return xfs_btree_readahead_lblock(cur, lr, block);
+	return xfs_btree_readahead_sblock(cur, lr, block);
 }
 
 /*
@@ -889,7 +813,7 @@ xfs_btree_setbuf(
 	int			lev,	/* level in btree */
 	xfs_buf_t		*bp)	/* new buffer to set */
 {
-	xfs_btree_block_t	*b;	/* btree block */
+	struct xfs_btree_block	*b;	/* btree block */
 	xfs_buf_t		*obp;	/* old buffer pointer */
 
 	obp = cur->bc_bufs[lev];
@@ -900,7 +824,7 @@ xfs_btree_setbuf(
 	if (!bp)
 		return;
 	b = XFS_BUF_TO_BLOCK(bp);
-	if (XFS_BTREE_LONG_PTRS(cur->bc_btnum)) {
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
 		if (be64_to_cpu(b->bb_u.l.bb_leftsib) == NULLDFSBNO)
 			cur->bc_ra[lev] |= XFS_BTCUR_LEFTRA;
 		if (be64_to_cpu(b->bb_u.l.bb_rightsib) == NULLDFSBNO)
@@ -912,3 +836,2855 @@ xfs_btree_setbuf(
 			cur->bc_ra[lev] |= XFS_BTCUR_RIGHTRA;
 	}
 }
+
+STATIC int
+xfs_btree_ptr_is_null(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return be64_to_cpu(ptr->l) == NULLFSBLOCK;
+	else
+		return be32_to_cpu(ptr->s) == NULLAGBLOCK;
+}
+
+STATIC void
+xfs_btree_set_ptr_null(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		ptr->l = cpu_to_be64(NULLFSBLOCK);
+	else
+		ptr->s = cpu_to_be32(NULLAGBLOCK);
+}
+
+/*
+ * Get/set/init sibling pointers
+ */
+STATIC void
+xfs_btree_get_sibling(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_ptr	*ptr,
+	int			lr)
+{
+	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (lr == XFS_BB_RIGHTSIB)
+			ptr->l = block->bb_u.l.bb_rightsib;
+		else
+			ptr->l = block->bb_u.l.bb_leftsib;
+	} else {
+		if (lr == XFS_BB_RIGHTSIB)
+			ptr->s = block->bb_u.s.bb_rightsib;
+		else
+			ptr->s = block->bb_u.s.bb_leftsib;
+	}
+}
+
+STATIC void
+xfs_btree_set_sibling(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	union xfs_btree_ptr	*ptr,
+	int			lr)
+{
+	ASSERT(lr == XFS_BB_LEFTSIB || lr == XFS_BB_RIGHTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		if (lr == XFS_BB_RIGHTSIB)
+			block->bb_u.l.bb_rightsib = ptr->l;
+		else
+			block->bb_u.l.bb_leftsib = ptr->l;
+	} else {
+		if (lr == XFS_BB_RIGHTSIB)
+			block->bb_u.s.bb_rightsib = ptr->s;
+		else
+			block->bb_u.s.bb_leftsib = ptr->s;
+	}
+}
+
+STATIC void
+xfs_btree_init_block(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			numrecs,
+	struct xfs_btree_block	*new)	/* new block */
+{
+	new->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
+	new->bb_level = cpu_to_be16(level);
+	new->bb_numrecs = cpu_to_be16(numrecs);
+
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		new->bb_u.l.bb_leftsib = cpu_to_be64(NULLFSBLOCK);
+		new->bb_u.l.bb_rightsib = cpu_to_be64(NULLFSBLOCK);
+	} else {
+		new->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		new->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+	}
+}
+
+/*
+ * Return true if ptr is the last record in the btree and
+ * we need to track updateѕ to this record.  The decision
+ * will be further refined in the update_lastrec method.
+ */
+STATIC int
+xfs_btree_is_lastrec(
+	struct xfs_btree_cur	*cur,
+	struct xfs_btree_block	*block,
+	int			level)
+{
+	union xfs_btree_ptr	ptr;
+
+	if (level > 0)
+		return 0;
+	if (!(cur->bc_flags & XFS_BTREE_LASTREC_UPDATE))
+		return 0;
+
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &ptr))
+		return 0;
+	return 1;
+}
+
+STATIC void
+xfs_btree_buf_to_ptr(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		ptr->l = cpu_to_be64(XFS_DADDR_TO_FSB(cur->bc_mp,
+					XFS_BUF_ADDR(bp)));
+	else {
+		ptr->s = cpu_to_be32(XFS_DADDR_TO_AGBNO(cur->bc_mp,
+					XFS_BUF_ADDR(bp)));
+	}
+}
+
+STATIC xfs_daddr_t
+xfs_btree_ptr_to_daddr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		ASSERT(be64_to_cpu(ptr->l) != NULLFSBLOCK);
+
+		return XFS_FSB_TO_DADDR(cur->bc_mp, be64_to_cpu(ptr->l));
+	} else {
+		ASSERT(cur->bc_private.a.agno != NULLAGNUMBER);
+		ASSERT(be32_to_cpu(ptr->s) != NULLAGBLOCK);
+
+		return XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_private.a.agno,
+					be32_to_cpu(ptr->s));
+	}
+}
+
+STATIC void
+xfs_btree_set_refs(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
+{
+	switch (cur->bc_btnum) {
+	case XFS_BTNUM_BNO:
+	case XFS_BTNUM_CNT:
+		XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_MAP, XFS_ALLOC_BTREE_REF);
+		break;
+	case XFS_BTNUM_INO:
+		XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_INOMAP, XFS_INO_BTREE_REF);
+		break;
+	case XFS_BTNUM_BMAP:
+		XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_MAP, XFS_BMAP_BTREE_REF);
+		break;
+	default:
+		ASSERT(0);
+	}
+}
+
+STATIC int
+xfs_btree_get_buf_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			flags,
+	struct xfs_btree_block	**block,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_daddr_t		d;
+
+	/* need to sort out how callers deal with failures first */
+	ASSERT(!(flags & XFS_BUF_TRYLOCK));
+
+	d = xfs_btree_ptr_to_daddr(cur, ptr);
+	*bpp = xfs_trans_get_buf(cur->bc_tp, mp->m_ddev_targp, d,
+				 mp->m_bsize, flags);
+
+	ASSERT(*bpp);
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+
+	*block = XFS_BUF_TO_BLOCK(*bpp);
+	return 0;
+}
+
+/*
+ * Read in the buffer at the given ptr and return the buffer and
+ * the block pointer within the buffer.
+ */
+STATIC int
+xfs_btree_read_buf_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			level,
+	int			flags,
+	struct xfs_btree_block	**block,
+	struct xfs_buf		**bpp)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	xfs_daddr_t		d;
+	int			error;
+
+	/* need to sort out how callers deal with failures first */
+	ASSERT(!(flags & XFS_BUF_TRYLOCK));
+
+	d = xfs_btree_ptr_to_daddr(cur, ptr);
+	error = xfs_trans_read_buf(mp, cur->bc_tp, mp->m_ddev_targp, d,
+				   mp->m_bsize, flags, bpp);
+	if (error)
+		return error;
+
+	ASSERT(*bpp != NULL);
+	ASSERT(!XFS_BUF_GETERROR(*bpp));
+
+	xfs_btree_set_refs(cur, *bpp);
+	*block = XFS_BUF_TO_BLOCK(*bpp);
+
+	error = xfs_btree_check_block(cur, *block, level, *bpp);
+	if (error)
+		xfs_trans_brelse(cur->bc_tp, *bpp);
+	return error;
+}
+
+/*
+ * Copy keys from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_keys(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*dst_key,
+	union xfs_btree_key	*src_key,
+	int			numkeys)
+{
+	ASSERT(numkeys >= 0);
+	memcpy(dst_key, src_key, numkeys * cur->bc_ops->key_len);
+}
+
+/*
+ * Copy records from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_recs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*dst_rec,
+	union xfs_btree_rec	*src_rec,
+	int			numrecs)
+{
+	ASSERT(numrecs >= 0);
+	memcpy(dst_rec, src_rec, numrecs * cur->bc_ops->rec_len);
+}
+
+/*
+ * Copy block pointers from one btree block to another.
+ */
+STATIC void
+xfs_btree_copy_ptrs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*dst_ptr,
+	union xfs_btree_ptr	*src_ptr,
+	int			numptrs)
+{
+	ASSERT(numptrs >= 0);
+	memcpy(dst_ptr, src_ptr, numptrs * xfs_btree_ptr_len(cur));
+}
+
+/*
+ * Shift keys one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_keys(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	int			dir,
+	int			numkeys)
+{
+	char			*dst_key;
+
+	ASSERT(numkeys >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_key = (char *)key + (dir * cur->bc_ops->key_len);
+	memmove(dst_key, key, numkeys * cur->bc_ops->key_len);
+}
+
+/*
+ * Shift records one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_recs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	int			dir,
+	int			numrecs)
+{
+	char			*dst_rec;
+
+	ASSERT(numrecs >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_rec = (char *)rec + (dir * cur->bc_ops->rec_len);
+	memmove(dst_rec, rec, numrecs * cur->bc_ops->rec_len);
+}
+
+/*
+ * Shift block pointers one index left/right inside a single btree block.
+ */
+STATIC void
+xfs_btree_shift_ptrs(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	int			dir,
+	int			numptrs)
+{
+	char			*dst_ptr;
+
+	ASSERT(numptrs >= 0);
+	ASSERT(dir == 1 || dir == -1);
+
+	dst_ptr = (char *)ptr + (dir * xfs_btree_ptr_len(cur));
+	memmove(dst_ptr, ptr, numptrs * xfs_btree_ptr_len(cur));
+}
+
+/*
+ * Log key values from the btree block.
+ */
+STATIC void
+xfs_btree_log_keys(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+
+	if (bp) {
+		xfs_trans_log_buf(cur->bc_tp, bp,
+				  xfs_btree_key_offset(cur, first),
+				  xfs_btree_key_offset(cur, last + 1) - 1);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
+				xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Log record values from the btree block.
+ */
+void
+xfs_btree_log_recs(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			first,
+	int			last)
+{
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+
+	xfs_trans_log_buf(cur->bc_tp, bp,
+			  xfs_btree_rec_offset(cur, first),
+			  xfs_btree_rec_offset(cur, last + 1) - 1);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Log block pointer fields from a btree block (nonleaf).
+ */
+STATIC void
+xfs_btree_log_ptrs(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_buf		*bp,	/* buffer containing btree block */
+	int			first,	/* index of first pointer to log */
+	int			last)	/* index of last pointer to log */
+{
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+
+	if (bp) {
+		struct xfs_btree_block	*block = XFS_BUF_TO_BLOCK(bp);
+		int			level = xfs_btree_get_level(block);
+
+		xfs_trans_log_buf(cur->bc_tp, bp,
+				xfs_btree_ptr_offset(cur, first, level),
+				xfs_btree_ptr_offset(cur, last + 1, level) - 1);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
+			xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Log fields from a btree block header.
+ */
+void
+xfs_btree_log_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_buf		*bp,	/* buffer containing btree block */
+	int			fields)	/* mask of fields: XFS_BB_... */
+{
+	int			first;	/* first byte offset logged */
+	int			last;	/* last byte offset logged */
+	static const short	soffsets[] = {	/* table of offsets (short) */
+		offsetof(struct xfs_btree_block, bb_magic),
+		offsetof(struct xfs_btree_block, bb_level),
+		offsetof(struct xfs_btree_block, bb_numrecs),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
+		offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
+		XFS_BTREE_SBLOCK_LEN
+	};
+	static const short	loffsets[] = {	/* table of offsets (long) */
+		offsetof(struct xfs_btree_block, bb_magic),
+		offsetof(struct xfs_btree_block, bb_level),
+		offsetof(struct xfs_btree_block, bb_numrecs),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
+		offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
+		XFS_BTREE_LBLOCK_LEN
+	};
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGBI(cur, bp, fields);
+
+	if (bp) {
+		xfs_btree_offsets(fields,
+				  (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
+					loffsets : soffsets,
+				  XFS_BB_NUM_BITS, &first, &last);
+		xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	} else {
+		xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
+			xfs_ilog_fbroot(cur->bc_private.b.whichfork));
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+}
+
+/*
+ * Increment cursor by one record at the level.
+ * For nonzero levels the leaf-ward information is untouched.
+ */
+int						/* error */
+xfs_btree_increment(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_block	*block;
+	union xfs_btree_ptr	ptr;
+	struct xfs_buf		*bp;
+	int			error;		/* error return value */
+	int			lev;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	ASSERT(level < cur->bc_nlevels);
+
+	/* Read-ahead to the right at this level. */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
+
+	/* Get a pointer to the btree block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* We're done if we remain in the block after the increment. */
+	if (++cur->bc_ptrs[level] <= xfs_btree_get_numrecs(block))
+		goto out1;
+
+	/* Fail if we just went off the right edge of the tree. */
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &ptr))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, increment);
+
+	/*
+	 * March up the tree incrementing pointers.
+	 * Stop when we don't go off the right edge of a block.
+	 */
+	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
+		block = xfs_btree_get_block(cur, lev, &bp);
+
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, lev, bp);
+		if (error)
+			goto error0;
+#endif
+
+		if (++cur->bc_ptrs[lev] <= xfs_btree_get_numrecs(block))
+			break;
+
+		/* Read-ahead the right block for the next loop. */
+		xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
+	}
+
+	/*
+	 * If we went off the root then we are either seriously
+	 * confused or have the tree root in an inode.
+	 */
+	if (lev == cur->bc_nlevels) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+			goto out0;
+		ASSERT(0);
+		error = EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(lev < cur->bc_nlevels);
+
+	/*
+	 * Now walk back down the tree, fixing up the cursor's buffer
+	 * pointers and key numbers.
+	 */
+	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
+		union xfs_btree_ptr	*ptrp;
+
+		ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
+		error = xfs_btree_read_buf_block(cur, ptrp, --lev,
+							0, &block, &bp);
+		if (error)
+			goto error0;
+
+		xfs_btree_setbuf(cur, lev, bp);
+		cur->bc_ptrs[lev] = 1;
+	}
+out1:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Decrement cursor by one record at the level.
+ * For nonzero levels the leaf-ward information is untouched.
+ */
+int						/* error */
+xfs_btree_decrement(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	struct xfs_btree_block	*block;
+	xfs_buf_t		*bp;
+	int			error;		/* error return value */
+	int			lev;
+	union xfs_btree_ptr	ptr;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	ASSERT(level < cur->bc_nlevels);
+
+	/* Read-ahead to the left at this level. */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
+
+	/* We're done if we remain in the block after the decrement. */
+	if (--cur->bc_ptrs[level] > 0)
+		goto out1;
+
+	/* Get a pointer to the btree block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* Fail if we just went off the left edge of the tree. */
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
+	if (xfs_btree_ptr_is_null(cur, &ptr))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, decrement);
+
+	/*
+	 * March up the tree decrementing pointers.
+	 * Stop when we don't go off the left edge of a block.
+	 */
+	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
+		if (--cur->bc_ptrs[lev] > 0)
+			break;
+		/* Read-ahead the left block for the next loop. */
+		xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
+	}
+
+	/*
+	 * If we went off the root then we are seriously confused.
+	 * or the root of the tree is in an inode.
+	 */
+	if (lev == cur->bc_nlevels) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE)
+			goto out0;
+		ASSERT(0);
+		error = EFSCORRUPTED;
+		goto error0;
+	}
+	ASSERT(lev < cur->bc_nlevels);
+
+	/*
+	 * Now walk back down the tree, fixing up the cursor's buffer
+	 * pointers and key numbers.
+	 */
+	for (block = xfs_btree_get_block(cur, lev, &bp); lev > level; ) {
+		union xfs_btree_ptr	*ptrp;
+
+		ptrp = xfs_btree_ptr_addr(cur, cur->bc_ptrs[lev], block);
+		error = xfs_btree_read_buf_block(cur, ptrp, --lev,
+							0, &block, &bp);
+		if (error)
+			goto error0;
+		xfs_btree_setbuf(cur, lev, bp);
+		cur->bc_ptrs[lev] = xfs_btree_get_numrecs(block);
+	}
+out1:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+STATIC int
+xfs_btree_lookup_get_block(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level in the btree */
+	union xfs_btree_ptr	*pp,	/* ptr to btree block */
+	struct xfs_btree_block	**blkp) /* return btree block */
+{
+	struct xfs_buf		*bp;	/* buffer pointer for btree block */
+	int			error = 0;
+
+	/* special case the root block if in an inode */
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1)) {
+		*blkp = xfs_btree_get_iroot(cur);
+		return 0;
+	}
+
+	/*
+	 * If the old buffer at this level for the disk address we are
+	 * looking for re-use it.
+	 *
+	 * Otherwise throw it away and get a new one.
+	 */
+	bp = cur->bc_bufs[level];
+	if (bp && XFS_BUF_ADDR(bp) == xfs_btree_ptr_to_daddr(cur, pp)) {
+		*blkp = XFS_BUF_TO_BLOCK(bp);
+		return 0;
+	}
+
+	error = xfs_btree_read_buf_block(cur, pp, level, 0, blkp, &bp);
+	if (error)
+		return error;
+
+	xfs_btree_setbuf(cur, level, bp);
+	return 0;
+}
+
+/*
+ * Get current search key.  For level 0 we don't actually have a key
+ * structure so we make one up from the record.  For all other levels
+ * we just return the right key.
+ */
+STATIC union xfs_btree_key *
+xfs_lookup_get_search_key(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			keyno,
+	struct xfs_btree_block	*block,
+	union xfs_btree_key	*kp)
+{
+	if (level == 0) {
+		cur->bc_ops->init_key_from_rec(kp,
+				xfs_btree_rec_addr(cur, keyno, block));
+		return kp;
+	}
+
+	return xfs_btree_key_addr(cur, keyno, block);
+}
+
+/*
+ * Lookup the record.  The cursor is made to point to it, based on dir.
+ * Return 0 if can't find any such record, 1 for success.
+ */
+int					/* error */
+xfs_btree_lookup(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_lookup_t		dir,	/* <=, ==, or >= */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* current btree block */
+	__int64_t		diff;	/* difference for the current key */
+	int			error;	/* error return value */
+	int			keyno;	/* current key number */
+	int			level;	/* level in the btree */
+	union xfs_btree_ptr	*pp;	/* ptr to btree block */
+	union xfs_btree_ptr	ptr;	/* ptr to btree block */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, dir);
+
+	XFS_BTREE_STATS_INC(cur, lookup);
+
+	block = NULL;
+	keyno = 0;
+
+	/* initialise start pointer from cursor */
+	cur->bc_ops->init_ptr_from_cur(cur, &ptr);
+	pp = &ptr;
+
+	/*
+	 * Iterate over each level in the btree, starting at the root.
+	 * For each level above the leaves, find the key we need, based
+	 * on the lookup record, then follow the corresponding block
+	 * pointer down to the next level.
+	 */
+	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
+		/* Get the block we need to do the lookup on. */
+		error = xfs_btree_lookup_get_block(cur, level, pp, &block);
+		if (error)
+			goto error0;
+
+		if (diff == 0) {
+			/*
+			 * If we already had a key match at a higher level, we
+			 * know we need to use the first entry in this block.
+			 */
+			keyno = 1;
+		} else {
+			/* Otherwise search this block. Do a binary search. */
+
+			int	high;	/* high entry number */
+			int	low;	/* low entry number */
+
+			/* Set low and high entry numbers, 1-based. */
+			low = 1;
+			high = xfs_btree_get_numrecs(block);
+			if (!high) {
+				/* Block is empty, must be an empty leaf. */
+				ASSERT(level == 0 && cur->bc_nlevels == 1);
+
+				cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
+				XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+				*stat = 0;
+				return 0;
+			}
+
+			/* Binary search the block. */
+			while (low <= high) {
+				union xfs_btree_key	key;
+				union xfs_btree_key	*kp;
+
+				XFS_BTREE_STATS_INC(cur, compare);
+
+				/* keyno is average of low and high. */
+				keyno = (low + high) >> 1;
+
+				/* Get current search key */
+				kp = xfs_lookup_get_search_key(cur, level,
+						keyno, block, &key);
+
+				/*
+				 * Compute difference to get next direction:
+				 *  - less than, move right
+				 *  - greater than, move left
+				 *  - equal, we're done
+				 */
+				diff = cur->bc_ops->key_diff(cur, kp);
+				if (diff < 0)
+					low = keyno + 1;
+				else if (diff > 0)
+					high = keyno - 1;
+				else
+					break;
+			}
+		}
+
+		/*
+		 * If there are more levels, set up for the next level
+		 * by getting the block number and filling in the cursor.
+		 */
+		if (level > 0) {
+			/*
+			 * If we moved left, need the previous key number,
+			 * unless there isn't one.
+			 */
+			if (diff > 0 && --keyno < 1)
+				keyno = 1;
+			pp = xfs_btree_ptr_addr(cur, keyno, block);
+
+#ifdef DEBUG
+			error = xfs_btree_check_ptr(cur, pp, 0, level);
+			if (error)
+				goto error0;
+#endif
+			cur->bc_ptrs[level] = keyno;
+		}
+	}
+
+	/* Done with the search. See if we need to adjust the results. */
+	if (dir != XFS_LOOKUP_LE && diff < 0) {
+		keyno++;
+		/*
+		 * If ge search and we went off the end of the block, but it's
+		 * not the last block, we're in the wrong block.
+		 */
+		xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+		if (dir == XFS_LOOKUP_GE &&
+		    keyno > xfs_btree_get_numrecs(block) &&
+		    !xfs_btree_ptr_is_null(cur, &ptr)) {
+			int	i;
+
+			cur->bc_ptrs[0] = keyno;
+			error = xfs_btree_increment(cur, 0, &i);
+			if (error)
+				goto error0;
+			XFS_WANT_CORRUPTED_RETURN(i == 1);
+			XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+			*stat = 1;
+			return 0;
+		}
+	} else if (dir == XFS_LOOKUP_LE && diff > 0)
+		keyno--;
+	cur->bc_ptrs[0] = keyno;
+
+	/* Return if we succeeded or not. */
+	if (keyno == 0 || keyno > xfs_btree_get_numrecs(block))
+		*stat = 0;
+	else if (dir != XFS_LOOKUP_EQ || diff == 0)
+		*stat = 1;
+	else
+		*stat = 0;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Update keys at all levels from here to the root along the cursor's path.
+ */
+STATIC int
+xfs_btree_updkey(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*keyp,
+	int			level)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	union xfs_btree_key	*kp;
+	int			ptr;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGIK(cur, level, keyp);
+
+	ASSERT(!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) || level >= 1);
+
+	/*
+	 * Go up the tree from this level toward the root.
+	 * At each level, update the key value to the value input.
+	 * Stop when we reach a level where the cursor isn't pointing
+	 * at the first entry in the block.
+	 */
+	for (ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
+#ifdef DEBUG
+		int		error;
+#endif
+		block = xfs_btree_get_block(cur, level, &bp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, block, level, bp);
+		if (error) {
+			XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+			return error;
+		}
+#endif
+		ptr = cur->bc_ptrs[level];
+		kp = xfs_btree_key_addr(cur, ptr, block);
+		xfs_btree_copy_keys(cur, kp, keyp, 1);
+		xfs_btree_log_keys(cur, bp, ptr, ptr);
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+}
+
+/*
+ * Update the record referred to by cur to the value in the
+ * given record. This either works (return 0) or gets an
+ * EFSCORRUPTED error.
+ */
+int
+xfs_btree_update(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	int			error;
+	int			ptr;
+	union xfs_btree_rec	*rp;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGR(cur, rec);
+
+	/* Pick up the current block. */
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, 0, bp);
+	if (error)
+		goto error0;
+#endif
+	/* Get the address of the rec to be updated. */
+	ptr = cur->bc_ptrs[0];
+	rp = xfs_btree_rec_addr(cur, ptr, block);
+
+	/* Fill in the new contents and log them. */
+	xfs_btree_copy_recs(cur, rp, rec, 1);
+	xfs_btree_log_recs(cur, bp, ptr, ptr);
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, 0)) {
+		cur->bc_ops->update_lastrec(cur, block, rec,
+					    ptr, LASTREC_UPDATE);
+	}
+
+	/* Updating first rec in leaf. Pass new key value up to our parent. */
+	if (ptr == 1) {
+		union xfs_btree_key	key;
+
+		cur->bc_ops->init_key_from_rec(&key, rec);
+		error = xfs_btree_updkey(cur, &key, 1);
+		if (error)
+			goto error0;
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Move 1 record left from cur/level if possible.
+ * Update cur to reflect the new path.
+ */
+STATIC int					/* error */
+xfs_btree_lshift(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_key	key;		/* btree key */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	int			lrecs;		/* left record count */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	int			rrecs;		/* right record count */
+	union xfs_btree_ptr	lptr;		/* left btree pointer */
+	union xfs_btree_key	*rkp = NULL;	/* right btree key */
+	union xfs_btree_ptr	*rpp = NULL;	/* right address pointer */
+	union xfs_btree_rec	*rrp = NULL;	/* right record pointer */
+	int			error;		/* error return value */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1)
+		goto out0;
+
+	/* Set up variables for this block as "right". */
+	right = xfs_btree_get_block(cur, level, &rbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, right, level, rbp);
+	if (error)
+		goto error0;
+#endif
+
+	/* If we've got no left sibling then we can't shift an entry left. */
+	xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+	if (xfs_btree_ptr_is_null(cur, &lptr))
+		goto out0;
+
+	/*
+	 * If the cursor entry is the one that would be moved, don't
+	 * do it... it's too complicated.
+	 */
+	if (cur->bc_ptrs[level] <= 1)
+		goto out0;
+
+	/* Set up the left neighbor as "left". */
+	error = xfs_btree_read_buf_block(cur, &lptr, level, 0, &left, &lbp);
+	if (error)
+		goto error0;
+
+	/* If it's full, it can't take another entry. */
+	lrecs = xfs_btree_get_numrecs(left);
+	if (lrecs == cur->bc_ops->get_maxrecs(cur, level))
+		goto out0;
+
+	rrecs = xfs_btree_get_numrecs(right);
+
+	/*
+	 * We add one entry to the left side and remove one for the right side.
+	 * Accout for it here, the changes will be updated on disk and logged
+	 * later.
+	 */
+	lrecs++;
+	rrecs--;
+
+	XFS_BTREE_STATS_INC(cur, lshift);
+	XFS_BTREE_STATS_ADD(cur, moves, 1);
+
+	/*
+	 * If non-leaf, copy a key and a ptr to the left block.
+	 * Log the changes to the left block.
+	 */
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+
+		lkp = xfs_btree_key_addr(cur, lrecs, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+
+		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+#ifdef DEBUG
+		error = xfs_btree_check_ptr(cur, rpp, 0, level);
+		if (error)
+			goto error0;
+#endif
+		xfs_btree_copy_keys(cur, lkp, rkp, 1);
+		xfs_btree_copy_ptrs(cur, lpp, rpp, 1);
+
+		xfs_btree_log_keys(cur, lbp, lrecs, lrecs);
+		xfs_btree_log_ptrs(cur, lbp, lrecs, lrecs);
+
+		ASSERT(cur->bc_ops->keys_inorder(cur,
+			xfs_btree_key_addr(cur, lrecs - 1, left), lkp));
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, lrecs, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, lrp, rrp, 1);
+		xfs_btree_log_recs(cur, lbp, lrecs, lrecs);
+
+		ASSERT(cur->bc_ops->recs_inorder(cur,
+			xfs_btree_rec_addr(cur, lrecs - 1, left), lrp));
+	}
+
+	xfs_btree_set_numrecs(left, lrecs);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
+
+	xfs_btree_set_numrecs(right, rrecs);
+	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
+
+	/*
+	 * Slide the contents of right down one entry.
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs - 1);
+	if (level > 0) {
+		/* It's a nonleaf. operate on keys and ptrs */
+#ifdef DEBUG
+		int			i;		/* loop index */
+
+		for (i = 0; i < rrecs; i++) {
+			error = xfs_btree_check_ptr(cur, rpp, i + 1, level);
+			if (error)
+				goto error0;
+		}
+#endif
+		xfs_btree_shift_keys(cur,
+				xfs_btree_key_addr(cur, 2, right),
+				-1, rrecs);
+		xfs_btree_shift_ptrs(cur,
+				xfs_btree_ptr_addr(cur, 2, right),
+				-1, rrecs);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
+	} else {
+		/* It's a leaf. operate on records */
+		xfs_btree_shift_recs(cur,
+			xfs_btree_rec_addr(cur, 2, right),
+			-1, rrecs);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs);
+
+		/*
+		 * If it's the first record in the block, we'll need a key
+		 * structure to pass up to the next level (updkey).
+		 */
+		cur->bc_ops->init_key_from_rec(&key,
+			xfs_btree_rec_addr(cur, 1, right));
+		rkp = &key;
+	}
+
+	/* Update the parent key values of right. */
+	error = xfs_btree_updkey(cur, rkp, level + 1);
+	if (error)
+		goto error0;
+
+	/* Slide the cursor value left one. */
+	cur->bc_ptrs[level]--;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Move 1 record right from cur/level if possible.
+ * Update cur to reflect the new path.
+ */
+STATIC int					/* error */
+xfs_btree_rshift(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_key	key;		/* btree key */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	union xfs_btree_ptr	rptr;		/* right block pointer */
+	union xfs_btree_key	*rkp;		/* right btree key */
+	int			rrecs;		/* right record count */
+	int			lrecs;		/* left record count */
+	int			error;		/* error return value */
+	int			i;		/* loop counter */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level == cur->bc_nlevels - 1))
+		goto out0;
+
+	/* Set up variables for this block as "left". */
+	left = xfs_btree_get_block(cur, level, &lbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, left, level, lbp);
+	if (error)
+		goto error0;
+#endif
+
+	/* If we've got no right sibling then we can't shift an entry right. */
+	xfs_btree_get_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &rptr))
+		goto out0;
+
+	/*
+	 * If the cursor entry is the one that would be moved, don't
+	 * do it... it's too complicated.
+	 */
+	lrecs = xfs_btree_get_numrecs(left);
+	if (cur->bc_ptrs[level] >= lrecs)
+		goto out0;
+
+	/* Set up the right neighbor as "right". */
+	error = xfs_btree_read_buf_block(cur, &rptr, level, 0, &right, &rbp);
+	if (error)
+		goto error0;
+
+	/* If it's full, it can't take another entry. */
+	rrecs = xfs_btree_get_numrecs(right);
+	if (rrecs == cur->bc_ops->get_maxrecs(cur, level))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, rshift);
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+
+	/*
+	 * Make a hole at the start of the right neighbor block, then
+	 * copy the last left block entry to the hole.
+	 */
+	if (level > 0) {
+		/* It's a nonleaf. make a hole in the keys and ptrs */
+		union xfs_btree_key	*lkp;
+		union xfs_btree_ptr	*lpp;
+		union xfs_btree_ptr	*rpp;
+
+		lkp = xfs_btree_key_addr(cur, lrecs, left);
+		lpp = xfs_btree_ptr_addr(cur, lrecs, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+#ifdef DEBUG
+		for (i = rrecs - 1; i >= 0; i--) {
+			error = xfs_btree_check_ptr(cur, rpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+
+		xfs_btree_shift_keys(cur, rkp, 1, rrecs);
+		xfs_btree_shift_ptrs(cur, rpp, 1, rrecs);
+
+#ifdef DEBUG
+		error = xfs_btree_check_ptr(cur, lpp, 0, level);
+		if (error)
+			goto error0;
+#endif
+
+		/* Now put the new data in, and log it. */
+		xfs_btree_copy_keys(cur, rkp, lkp, 1);
+		xfs_btree_copy_ptrs(cur, rpp, lpp, 1);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs + 1);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs + 1);
+
+		ASSERT(cur->bc_ops->keys_inorder(cur, rkp,
+			xfs_btree_key_addr(cur, 2, right)));
+	} else {
+		/* It's a leaf. make a hole in the records */
+		union xfs_btree_rec	*lrp;
+		union xfs_btree_rec	*rrp;
+
+		lrp = xfs_btree_rec_addr(cur, lrecs, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_shift_recs(cur, rrp, 1, rrecs);
+
+		/* Now put the new data in, and log it. */
+		xfs_btree_copy_recs(cur, rrp, lrp, 1);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs + 1);
+
+		cur->bc_ops->init_key_from_rec(&key, rrp);
+		rkp = &key;
+
+		ASSERT(cur->bc_ops->recs_inorder(cur, rrp,
+			xfs_btree_rec_addr(cur, 2, right)));
+	}
+
+	/*
+	 * Decrement and log left's numrecs, bump and log right's numrecs.
+	 */
+	xfs_btree_set_numrecs(left, --lrecs);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS);
+
+	xfs_btree_set_numrecs(right, ++rrecs);
+	xfs_btree_log_block(cur, rbp, XFS_BB_NUMRECS);
+
+	/*
+	 * Using a temporary cursor, update the parent key values of the
+	 * block on the right.
+	 */
+	error = xfs_btree_dup_cursor(cur, &tcur);
+	if (error)
+		goto error0;
+	i = xfs_btree_lastrec(tcur, level);
+	XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+	error = xfs_btree_increment(tcur, level, &i);
+	if (error)
+		goto error1;
+
+	error = xfs_btree_updkey(tcur, rkp, level + 1);
+	if (error)
+		goto error1;
+
+	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+
+error1:
+	XFS_BTREE_TRACE_CURSOR(tcur, XBT_ERROR);
+	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Split cur/level block in half.
+ * Return new block number and the key to its first
+ * record (to be inserted into parent).
+ */
+STATIC int					/* error */
+xfs_btree_split(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	union xfs_btree_ptr	*ptrp,
+	union xfs_btree_key	*key,
+	struct xfs_btree_cur	**curp,
+	int			*stat)		/* success/failure */
+{
+	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	union xfs_btree_ptr	rrptr;		/* right-right sibling ptr */
+	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
+	struct xfs_btree_block	*rrblock;	/* right-right btree block */
+	int			lrecs;
+	int			rrecs;
+	int			src_index;
+	int			error;		/* error return value */
+#ifdef DEBUG
+	int			i;
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGIPK(cur, level, *ptrp, key);
+
+	XFS_BTREE_STATS_INC(cur, split);
+
+	/* Set up left block (current one). */
+	left = xfs_btree_get_block(cur, level, &lbp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, left, level, lbp);
+	if (error)
+		goto error0;
+#endif
+
+	xfs_btree_buf_to_ptr(cur, lbp, &lptr);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, &lptr, &rptr, 1, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		goto out0;
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Set up the new block as "right". */
+	error = xfs_btree_get_buf_block(cur, &rptr, 0, &right, &rbp);
+	if (error)
+		goto error0;
+
+	/* Fill in the btree header for the new right block. */
+	xfs_btree_init_block(cur, xfs_btree_get_level(left), 0, right);
+
+	/*
+	 * Split the entries between the old and the new block evenly.
+	 * Make sure that if there's an odd number of entries now, that
+	 * each new block will have the same number of entries.
+	 */
+	lrecs = xfs_btree_get_numrecs(left);
+	rrecs = lrecs / 2;
+	if ((lrecs & 1) && cur->bc_ptrs[level] <= rrecs + 1)
+		rrecs++;
+	src_index = (lrecs - rrecs + 1);
+
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+
+	/*
+	 * Copy btree block entries from the left block over to the
+	 * new block, the right. Update the right block and log the
+	 * changes.
+	 */
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+		union xfs_btree_key	*rkp;	/* right btree key */
+		union xfs_btree_ptr	*rpp;	/* right address pointer */
+
+		lkp = xfs_btree_key_addr(cur, src_index, left);
+		lpp = xfs_btree_ptr_addr(cur, src_index, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+
+#ifdef DEBUG
+		for (i = src_index; i < rrecs; i++) {
+			error = xfs_btree_check_ptr(cur, lpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+
+		xfs_btree_copy_keys(cur, rkp, lkp, rrecs);
+		xfs_btree_copy_ptrs(cur, rpp, lpp, rrecs);
+
+		xfs_btree_log_keys(cur, rbp, 1, rrecs);
+		xfs_btree_log_ptrs(cur, rbp, 1, rrecs);
+
+		/* Grab the keys to the entries moved to the right block */
+		xfs_btree_copy_keys(cur, key, rkp, 1);
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+		union xfs_btree_rec	*rrp;	/* right record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, src_index, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, rrp, lrp, rrecs);
+		xfs_btree_log_recs(cur, rbp, 1, rrecs);
+
+		cur->bc_ops->init_key_from_rec(key,
+			xfs_btree_rec_addr(cur, 1, right));
+	}
+
+
+	/*
+	 * Find the left block number by looking in the buffer.
+	 * Adjust numrecs, sibling pointers.
+	 */
+	xfs_btree_get_sibling(cur, left, &rrptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, right, &rrptr, XFS_BB_RIGHTSIB);
+	xfs_btree_set_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+	xfs_btree_set_sibling(cur, left, &rptr, XFS_BB_RIGHTSIB);
+
+	lrecs -= rrecs;
+	xfs_btree_set_numrecs(left, lrecs);
+	xfs_btree_set_numrecs(right, xfs_btree_get_numrecs(right) + rrecs);
+
+	xfs_btree_log_block(cur, rbp, XFS_BB_ALL_BITS);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
+
+	/*
+	 * If there's a block to the new block's right, make that block
+	 * point back to right instead of to left.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &rrptr)) {
+		error = xfs_btree_read_buf_block(cur, &rrptr, level,
+							0, &rrblock, &rrbp);
+		if (error)
+			goto error0;
+		xfs_btree_set_sibling(cur, rrblock, &rptr, XFS_BB_LEFTSIB);
+		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
+	}
+	/*
+	 * If the cursor is really in the right block, move it there.
+	 * If it's just pointing past the last entry in left, then we'll
+	 * insert there, so don't change anything in that case.
+	 */
+	if (cur->bc_ptrs[level] > lrecs + 1) {
+		xfs_btree_setbuf(cur, level, rbp);
+		cur->bc_ptrs[level] -= lrecs;
+	}
+	/*
+	 * If there are more levels, we'll need another cursor which refers
+	 * the right block, no matter where this cursor was.
+	 */
+	if (level + 1 < cur->bc_nlevels) {
+		error = xfs_btree_dup_cursor(cur, curp);
+		if (error)
+			goto error0;
+		(*curp)->bc_ptrs[level + 1]++;
+	}
+	*ptrp = rptr;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Copy the old inode root contents into a real block and make the
+ * broot point to it.
+ */
+int						/* error */
+xfs_btree_new_iroot(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			*logflags,	/* logging flags for inode */
+	int			*stat)		/* return status - 0 fail */
+{
+	struct xfs_buf		*cbp;		/* buffer for cblock */
+	struct xfs_btree_block	*block;		/* btree block */
+	struct xfs_btree_block	*cblock;	/* child btree block */
+	union xfs_btree_key	*ckp;		/* child key pointer */
+	union xfs_btree_ptr	*cpp;		/* child ptr pointer */
+	union xfs_btree_key	*kp;		/* pointer to btree key */
+	union xfs_btree_ptr	*pp;		/* pointer to block addr */
+	union xfs_btree_ptr	nptr;		/* new block addr */
+	int			level;		/* btree level */
+	int			error;		/* error return code */
+#ifdef DEBUG
+	int			i;		/* loop counter */
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, newroot);
+
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+
+	level = cur->bc_nlevels - 1;
+
+	block = xfs_btree_get_iroot(cur);
+	pp = xfs_btree_ptr_addr(cur, 1, block);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, pp, &nptr, 1, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		return 0;
+	}
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Copy the root into a real block. */
+	error = xfs_btree_get_buf_block(cur, &nptr, 0, &cblock, &cbp);
+	if (error)
+		goto error0;
+
+	memcpy(cblock, block, xfs_btree_block_len(cur));
+
+	be16_add_cpu(&block->bb_level, 1);
+	xfs_btree_set_numrecs(block, 1);
+	cur->bc_nlevels++;
+	cur->bc_ptrs[level + 1] = 1;
+
+	kp = xfs_btree_key_addr(cur, 1, block);
+	ckp = xfs_btree_key_addr(cur, 1, cblock);
+	xfs_btree_copy_keys(cur, ckp, kp, xfs_btree_get_numrecs(cblock));
+
+	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
+#ifdef DEBUG
+	for (i = 0; i < be16_to_cpu(cblock->bb_numrecs); i++) {
+		error = xfs_btree_check_ptr(cur, pp, i, level);
+		if (error)
+			goto error0;
+	}
+#endif
+	xfs_btree_copy_ptrs(cur, cpp, pp, xfs_btree_get_numrecs(cblock));
+
+#ifdef DEBUG
+	error = xfs_btree_check_ptr(cur, &nptr, 0, level);
+	if (error)
+		goto error0;
+#endif
+	xfs_btree_copy_ptrs(cur, pp, &nptr, 1);
+
+	xfs_iroot_realloc(cur->bc_private.b.ip,
+			  1 - xfs_btree_get_numrecs(cblock),
+			  cur->bc_private.b.whichfork);
+
+	xfs_btree_setbuf(cur, level, cbp);
+
+	/*
+	 * Do all this logging at the end so that
+	 * the root is at the right level.
+	 */
+	xfs_btree_log_block(cur, cbp, XFS_BB_ALL_BITS);
+	xfs_btree_log_keys(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
+	xfs_btree_log_ptrs(cur, cbp, 1, be16_to_cpu(cblock->bb_numrecs));
+
+	*logflags |=
+		XFS_ILOG_CORE | XFS_ILOG_FBROOT(cur->bc_private.b.whichfork);
+	*stat = 1;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Allocate a new root block, fill it in.
+ */
+STATIC int				/* error */
+xfs_btree_new_root(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* one half of the old root block */
+	struct xfs_buf		*bp;	/* buffer containing block */
+	int			error;	/* error return value */
+	struct xfs_buf		*lbp;	/* left buffer pointer */
+	struct xfs_btree_block	*left;	/* left btree block */
+	struct xfs_buf		*nbp;	/* new (root) buffer */
+	struct xfs_btree_block	*new;	/* new (root) btree block */
+	int			nptr;	/* new value for key index, 1 or 2 */
+	struct xfs_buf		*rbp;	/* right buffer pointer */
+	struct xfs_btree_block	*right;	/* right btree block */
+	union xfs_btree_ptr	rptr;
+	union xfs_btree_ptr	lptr;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, newroot);
+
+	/* initialise our start point from the cursor */
+	cur->bc_ops->init_ptr_from_cur(cur, &rptr);
+
+	/* Allocate the new block. If we can't do it, we're toast. Give up. */
+	error = cur->bc_ops->alloc_block(cur, &rptr, &lptr, 1, stat);
+	if (error)
+		goto error0;
+	if (*stat == 0)
+		goto out0;
+	XFS_BTREE_STATS_INC(cur, alloc);
+
+	/* Set up the new block. */
+	error = xfs_btree_get_buf_block(cur, &lptr, 0, &new, &nbp);
+	if (error)
+		goto error0;
+
+	/* Set the root in the holding structure  increasing the level by 1. */
+	cur->bc_ops->set_root(cur, &lptr, 1);
+
+	/*
+	 * At the previous root level there are now two blocks: the old root,
+	 * and the new block generated when it was split.  We don't know which
+	 * one the cursor is pointing at, so we set up variables "left" and
+	 * "right" for each case.
+	 */
+	block = xfs_btree_get_block(cur, cur->bc_nlevels - 1, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, cur->bc_nlevels - 1, bp);
+	if (error)
+		goto error0;
+#endif
+
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
+		/* Our block is left, pick up the right block. */
+		lbp = bp;
+		xfs_btree_buf_to_ptr(cur, lbp, &lptr);
+		left = block;
+		error = xfs_btree_read_buf_block(cur, &rptr,
+					cur->bc_nlevels - 1, 0, &right, &rbp);
+		if (error)
+			goto error0;
+		bp = rbp;
+		nptr = 1;
+	} else {
+		/* Our block is right, pick up the left block. */
+		rbp = bp;
+		xfs_btree_buf_to_ptr(cur, rbp, &rptr);
+		right = block;
+		xfs_btree_get_sibling(cur, right, &lptr, XFS_BB_LEFTSIB);
+		error = xfs_btree_read_buf_block(cur, &lptr,
+					cur->bc_nlevels - 1, 0, &left, &lbp);
+		if (error)
+			goto error0;
+		bp = lbp;
+		nptr = 2;
+	}
+	/* Fill in the new block's btree header and log it. */
+	xfs_btree_init_block(cur, cur->bc_nlevels, 2, new);
+	xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
+	ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
+			!xfs_btree_ptr_is_null(cur, &rptr));
+
+	/* Fill in the key data in the new root. */
+	if (xfs_btree_get_level(left) > 0) {
+		xfs_btree_copy_keys(cur,
+				xfs_btree_key_addr(cur, 1, new),
+				xfs_btree_key_addr(cur, 1, left), 1);
+		xfs_btree_copy_keys(cur,
+				xfs_btree_key_addr(cur, 2, new),
+				xfs_btree_key_addr(cur, 1, right), 1);
+	} else {
+		cur->bc_ops->init_key_from_rec(
+				xfs_btree_key_addr(cur, 1, new),
+				xfs_btree_rec_addr(cur, 1, left));
+		cur->bc_ops->init_key_from_rec(
+				xfs_btree_key_addr(cur, 2, new),
+				xfs_btree_rec_addr(cur, 1, right));
+	}
+	xfs_btree_log_keys(cur, nbp, 1, 2);
+
+	/* Fill in the pointer data in the new root. */
+	xfs_btree_copy_ptrs(cur,
+		xfs_btree_ptr_addr(cur, 1, new), &lptr, 1);
+	xfs_btree_copy_ptrs(cur,
+		xfs_btree_ptr_addr(cur, 2, new), &rptr, 1);
+	xfs_btree_log_ptrs(cur, nbp, 1, 2);
+
+	/* Fix up the cursor. */
+	xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
+	cur->bc_ptrs[cur->bc_nlevels] = nptr;
+	cur->bc_nlevels++;
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 0;
+	return 0;
+}
+
+STATIC int
+xfs_btree_make_block_unfull(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* btree level */
+	int			numrecs,/* # of recs in block */
+	int			*oindex,/* old tree index */
+	int			*index,	/* new tree index */
+	union xfs_btree_ptr	*nptr,	/* new btree ptr */
+	struct xfs_btree_cur	**ncur,	/* new btree cursor */
+	union xfs_btree_rec	*nrec,	/* new record */
+	int			*stat)
+{
+	union xfs_btree_key	key;	/* new btree key value */
+	int			error = 0;
+
+	if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    level == cur->bc_nlevels - 1) {
+	    	struct xfs_inode *ip = cur->bc_private.b.ip;
+
+		if (numrecs < cur->bc_ops->get_dmaxrecs(cur, level)) {
+			/* A root block that can be made bigger. */
+
+			xfs_iroot_realloc(ip, 1, cur->bc_private.b.whichfork);
+		} else {
+			/* A root block that needs replacing */
+			int	logflags = 0;
+
+			error = xfs_btree_new_iroot(cur, &logflags, stat);
+			if (error || *stat == 0)
+				return error;
+
+			xfs_trans_log_inode(cur->bc_tp, ip, logflags);
+		}
+
+		return 0;
+	}
+
+	/* First, try shifting an entry to the right neighbor. */
+	error = xfs_btree_rshift(cur, level, stat);
+	if (error || *stat)
+		return error;
+
+	/* Next, try shifting an entry to the left neighbor. */
+	error = xfs_btree_lshift(cur, level, stat);
+	if (error)
+		return error;
+
+	if (*stat) {
+		*oindex = *index = cur->bc_ptrs[level];
+		return 0;
+	}
+
+	/*
+	 * Next, try splitting the current block in half.
+	 *
+	 * If this works we have to re-set our variables because we
+	 * could be in a different block now.
+	 */
+	error = xfs_btree_split(cur, level, nptr, &key, ncur, stat);
+	if (error || *stat == 0)
+		return error;
+
+
+	*index = cur->bc_ptrs[level];
+	cur->bc_ops->init_rec_from_key(&key, nrec);
+	return 0;
+}
+
+/*
+ * Insert one record/level.  Return information to the caller
+ * allowing the next level up to proceed if necessary.
+ */
+STATIC int
+xfs_btree_insrec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	int			level,	/* level to insert record at */
+	union xfs_btree_ptr	*ptrp,	/* i/o: block number inserted */
+	union xfs_btree_rec	*recp,	/* i/o: record data inserted */
+	struct xfs_btree_cur	**curp,	/* output: new cursor replacing cur */
+	int			*stat)	/* success/failure */
+{
+	struct xfs_btree_block	*block;	/* btree block */
+	struct xfs_buf		*bp;	/* buffer for block */
+	union xfs_btree_key	key;	/* btree key */
+	union xfs_btree_ptr	nptr;	/* new block ptr */
+	struct xfs_btree_cur	*ncur;	/* new btree cursor */
+	union xfs_btree_rec	nrec;	/* new record count */
+	int			optr;	/* old key/record index */
+	int			ptr;	/* key/record index */
+	int			numrecs;/* number of records */
+	int			error;	/* error return value */
+#ifdef DEBUG
+	int			i;
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGIPR(cur, level, *ptrp, recp);
+
+	ncur = NULL;
+
+	/*
+	 * If we have an external root pointer, and we've made it to the
+	 * root level, allocate a new root block and we're done.
+	 */
+	if (!(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) &&
+	    (level >= cur->bc_nlevels)) {
+		error = xfs_btree_new_root(cur, stat);
+		xfs_btree_set_ptr_null(cur, ptrp);
+
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		return error;
+	}
+
+	/* If we're off the left edge, return failure. */
+	ptr = cur->bc_ptrs[level];
+	if (ptr == 0) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	/* Make a key out of the record data to be inserted, and save it. */
+	cur->bc_ops->init_key_from_rec(&key, recp);
+
+	optr = ptr;
+
+	XFS_BTREE_STATS_INC(cur, insrec);
+
+	/* Get pointers to the btree buffer and block. */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+
+	/* Check that the new entry is being inserted in the right place. */
+	if (ptr <= numrecs) {
+		if (level == 0) {
+			ASSERT(cur->bc_ops->recs_inorder(cur, recp,
+				xfs_btree_rec_addr(cur, ptr, block)));
+		} else {
+			ASSERT(cur->bc_ops->keys_inorder(cur, &key,
+				xfs_btree_key_addr(cur, ptr, block)));
+		}
+	}
+#endif
+
+	/*
+	 * If the block is full, we can't insert the new entry until we
+	 * make the block un-full.
+	 */
+	xfs_btree_set_ptr_null(cur, &nptr);
+	if (numrecs == cur->bc_ops->get_maxrecs(cur, level)) {
+		error = xfs_btree_make_block_unfull(cur, level, numrecs,
+					&optr, &ptr, &nptr, &ncur, &nrec, stat);
+		if (error || *stat == 0)
+			goto error0;
+	}
+
+	/*
+	 * The current block may have changed if the block was
+	 * previously full and we have just made space in it.
+	 */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		return error;
+#endif
+
+	/*
+	 * At this point we know there's room for our new entry in the block
+	 * we're pointing at.
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr + 1);
+
+	if (level > 0) {
+		/* It's a nonleaf. make a hole in the keys and ptrs */
+		union xfs_btree_key	*kp;
+		union xfs_btree_ptr	*pp;
+
+		kp = xfs_btree_key_addr(cur, ptr, block);
+		pp = xfs_btree_ptr_addr(cur, ptr, block);
+
+#ifdef DEBUG
+		for (i = numrecs - ptr; i >= 0; i--) {
+			error = xfs_btree_check_ptr(cur, pp, i, level);
+			if (error)
+				return error;
+		}
+#endif
+
+		xfs_btree_shift_keys(cur, kp, 1, numrecs - ptr + 1);
+		xfs_btree_shift_ptrs(cur, pp, 1, numrecs - ptr + 1);
+
+#ifdef DEBUG
+		error = xfs_btree_check_ptr(cur, ptrp, 0, level);
+		if (error)
+			goto error0;
+#endif
+
+		/* Now put the new data in, bump numrecs and log it. */
+		xfs_btree_copy_keys(cur, kp, &key, 1);
+		xfs_btree_copy_ptrs(cur, pp, ptrp, 1);
+		numrecs++;
+		xfs_btree_set_numrecs(block, numrecs);
+		xfs_btree_log_ptrs(cur, bp, ptr, numrecs);
+		xfs_btree_log_keys(cur, bp, ptr, numrecs);
+#ifdef DEBUG
+		if (ptr < numrecs) {
+			ASSERT(cur->bc_ops->keys_inorder(cur, kp,
+				xfs_btree_key_addr(cur, ptr + 1, block)));
+		}
+#endif
+	} else {
+		/* It's a leaf. make a hole in the records */
+		union xfs_btree_rec             *rp;
+
+		rp = xfs_btree_rec_addr(cur, ptr, block);
+
+		xfs_btree_shift_recs(cur, rp, 1, numrecs - ptr + 1);
+
+		/* Now put the new data in, bump numrecs and log it. */
+		xfs_btree_copy_recs(cur, rp, recp, 1);
+		xfs_btree_set_numrecs(block, ++numrecs);
+		xfs_btree_log_recs(cur, bp, ptr, numrecs);
+#ifdef DEBUG
+		if (ptr < numrecs) {
+			ASSERT(cur->bc_ops->recs_inorder(cur, rp,
+				xfs_btree_rec_addr(cur, ptr + 1, block)));
+		}
+#endif
+	}
+
+	/* Log the new number of records in the btree header. */
+	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
+
+	/* If we inserted at the start of a block, update the parents' keys. */
+	if (optr == 1) {
+		error = xfs_btree_updkey(cur, &key, level + 1);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, level)) {
+		cur->bc_ops->update_lastrec(cur, block, recp,
+					    ptr, LASTREC_INSREC);
+	}
+
+	/*
+	 * Return the new block number, if any.
+	 * If there is one, give back a record value and a cursor too.
+	 */
+	*ptrp = nptr;
+	if (!xfs_btree_ptr_is_null(cur, &nptr)) {
+		*recp = nrec;
+		*curp = ncur;
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Insert the record at the point referenced by cur.
+ *
+ * A multi-level split of the tree on insert will invalidate the original
+ * cursor.  All callers of this function should assume that the cursor is
+ * no longer valid and revalidate it.
+ */
+int
+xfs_btree_insert(
+	struct xfs_btree_cur	*cur,
+	int			*stat)
+{
+	int			error;	/* error return value */
+	int			i;	/* result value, 0 for failure */
+	int			level;	/* current level number in btree */
+	union xfs_btree_ptr	nptr;	/* new block number (split result) */
+	struct xfs_btree_cur	*ncur;	/* new cursor (split result) */
+	struct xfs_btree_cur	*pcur;	/* previous level's cursor */
+	union xfs_btree_rec	rec;	/* record to insert */
+
+	level = 0;
+	ncur = NULL;
+	pcur = cur;
+
+	xfs_btree_set_ptr_null(cur, &nptr);
+	cur->bc_ops->init_rec_from_cur(cur, &rec);
+
+	/*
+	 * Loop going up the tree, starting at the leaf level.
+	 * Stop when we don't get a split block, that must mean that
+	 * the insert is finished with this level.
+	 */
+	do {
+		/*
+		 * Insert nrec/nptr into this level of the tree.
+		 * Note if we fail, nptr will be null.
+		 */
+		error = xfs_btree_insrec(pcur, level, &nptr, &rec, &ncur, &i);
+		if (error) {
+			if (pcur != cur)
+				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
+			goto error0;
+		}
+
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		level++;
+
+		/*
+		 * See if the cursor we just used is trash.
+		 * Can't trash the caller's cursor, but otherwise we should
+		 * if ncur is a new cursor or we're about to be done.
+		 */
+		if (pcur != cur &&
+		    (ncur || xfs_btree_ptr_is_null(cur, &nptr))) {
+			/* Save the state from the cursor before we trash it */
+			if (cur->bc_ops->update_cursor)
+				cur->bc_ops->update_cursor(pcur, cur);
+			cur->bc_nlevels = pcur->bc_nlevels;
+			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
+		}
+		/* If we got a new cursor, switch to it. */
+		if (ncur) {
+			pcur = ncur;
+			ncur = NULL;
+		}
+	} while (!xfs_btree_ptr_is_null(cur, &nptr));
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = i;
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Try to merge a non-leaf block back into the inode root.
+ *
+ * Note: the killroot names comes from the fact that we're effectively
+ * killing the old root block.  But because we can't just delete the
+ * inode we have to copy the single block it was pointing to into the
+ * inode.
+ */
+int
+xfs_btree_kill_iroot(
+	struct xfs_btree_cur	*cur)
+{
+	int			whichfork = cur->bc_private.b.whichfork;
+	struct xfs_inode	*ip = cur->bc_private.b.ip;
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
+	struct xfs_btree_block	*block;
+	struct xfs_btree_block	*cblock;
+	union xfs_btree_key	*kp;
+	union xfs_btree_key	*ckp;
+	union xfs_btree_ptr	*pp;
+	union xfs_btree_ptr	*cpp;
+	struct xfs_buf		*cbp;
+	int			level;
+	int			index;
+	int			numrecs;
+#ifdef DEBUG
+	union xfs_btree_ptr	ptr;
+	int			i;
+#endif
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+	ASSERT(cur->bc_nlevels > 1);
+
+	/*
+	 * Don't deal with the root block needs to be a leaf case.
+	 * We're just going to turn the thing back into extents anyway.
+	 */
+	level = cur->bc_nlevels - 1;
+	if (level == 1)
+		goto out0;
+
+	/*
+	 * Give up if the root has multiple children.
+	 */
+	block = xfs_btree_get_iroot(cur);
+	if (xfs_btree_get_numrecs(block) != 1)
+		goto out0;
+
+	cblock = xfs_btree_get_block(cur, level - 1, &cbp);
+	numrecs = xfs_btree_get_numrecs(cblock);
+
+	/*
+	 * Only do this if the next level will fit.
+	 * Then the data must be copied up to the inode,
+	 * instead of freeing the root you free the next level.
+	 */
+	if (numrecs > cur->bc_ops->get_dmaxrecs(cur, level))
+		goto out0;
+
+	XFS_BTREE_STATS_INC(cur, killroot);
+
+#ifdef DEBUG
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_LEFTSIB);
+	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+	xfs_btree_get_sibling(cur, block, &ptr, XFS_BB_RIGHTSIB);
+	ASSERT(xfs_btree_ptr_is_null(cur, &ptr));
+#endif
+
+	index = numrecs - cur->bc_ops->get_maxrecs(cur, level);
+	if (index) {
+		xfs_iroot_realloc(cur->bc_private.b.ip, index,
+				  cur->bc_private.b.whichfork);
+		block = ifp->if_broot;
+	}
+
+	be16_add_cpu(&block->bb_numrecs, index);
+	ASSERT(block->bb_numrecs == cblock->bb_numrecs);
+
+	kp = xfs_btree_key_addr(cur, 1, block);
+	ckp = xfs_btree_key_addr(cur, 1, cblock);
+	xfs_btree_copy_keys(cur, kp, ckp, numrecs);
+
+	pp = xfs_btree_ptr_addr(cur, 1, block);
+	cpp = xfs_btree_ptr_addr(cur, 1, cblock);
+#ifdef DEBUG
+	for (i = 0; i < numrecs; i++) {
+		int		error;
+
+		error = xfs_btree_check_ptr(cur, cpp, i, level - 1);
+		if (error) {
+			XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+			return error;
+		}
+	}
+#endif
+	xfs_btree_copy_ptrs(cur, pp, cpp, numrecs);
+
+	cur->bc_ops->free_block(cur, cbp);
+	XFS_BTREE_STATS_INC(cur, free);
+
+	cur->bc_bufs[level - 1] = NULL;
+	be16_add_cpu(&block->bb_level, -1);
+	xfs_trans_log_inode(cur->bc_tp, ip,
+		XFS_ILOG_CORE | XFS_ILOG_FBROOT(cur->bc_private.b.whichfork));
+	cur->bc_nlevels--;
+out0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	return 0;
+}
+
+STATIC int
+xfs_btree_dec_cursor(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	int			*stat)
+{
+	int			error;
+	int			i;
+
+	if (level > 0) {
+		error = xfs_btree_decrement(cur, level, &i);
+		if (error)
+			return error;
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = 1;
+	return 0;
+}
+
+/*
+ * Single level of the btree record deletion routine.
+ * Delete record pointed to by cur/level.
+ * Remove the record from its block then rebalance the tree.
+ * Return 0 for error, 1 for done, 2 to go on to the next level.
+ */
+STATIC int					/* error */
+xfs_btree_delrec(
+	struct xfs_btree_cur	*cur,		/* btree cursor */
+	int			level,		/* level removing record from */
+	int			*stat)		/* fail/done/go-on */
+{
+	struct xfs_btree_block	*block;		/* btree block */
+	union xfs_btree_ptr	cptr;		/* current block ptr */
+	struct xfs_buf		*bp;		/* buffer for block */
+	int			error;		/* error return value */
+	int			i;		/* loop counter */
+	union xfs_btree_key	key;		/* storage for keyp */
+	union xfs_btree_key	*keyp = &key;	/* passed to the next level */
+	union xfs_btree_ptr	lptr;		/* left sibling block ptr */
+	struct xfs_buf		*lbp;		/* left buffer pointer */
+	struct xfs_btree_block	*left;		/* left btree block */
+	int			lrecs = 0;	/* left record count */
+	int			ptr;		/* key/record index */
+	union xfs_btree_ptr	rptr;		/* right sibling block ptr */
+	struct xfs_buf		*rbp;		/* right buffer pointer */
+	struct xfs_btree_block	*right;		/* right btree block */
+	struct xfs_btree_block	*rrblock;	/* right-right btree block */
+	struct xfs_buf		*rrbp;		/* right-right buffer pointer */
+	int			rrecs = 0;	/* right record count */
+	struct xfs_btree_cur	*tcur;		/* temporary btree cursor */
+	int			numrecs;	/* temporary numrec count */
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_TRACE_ARGI(cur, level);
+
+	tcur = NULL;
+
+	/* Get the index of the entry being deleted, check for nothing there. */
+	ptr = cur->bc_ptrs[level];
+	if (ptr == 0) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	/* Get the buffer & block containing the record or key/ptr. */
+	block = xfs_btree_get_block(cur, level, &bp);
+	numrecs = xfs_btree_get_numrecs(block);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, level, bp);
+	if (error)
+		goto error0;
+#endif
+
+	/* Fail if we're off the end of the block. */
+	if (ptr > numrecs) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+		*stat = 0;
+		return 0;
+	}
+
+	XFS_BTREE_STATS_INC(cur, delrec);
+	XFS_BTREE_STATS_ADD(cur, moves, numrecs - ptr);
+
+	/* Excise the entries being deleted. */
+	if (level > 0) {
+		/* It's a nonleaf. operate on keys and ptrs */
+		union xfs_btree_key	*lkp;
+		union xfs_btree_ptr	*lpp;
+
+		lkp = xfs_btree_key_addr(cur, ptr + 1, block);
+		lpp = xfs_btree_ptr_addr(cur, ptr + 1, block);
+
+#ifdef DEBUG
+		for (i = 0; i < numrecs - ptr; i++) {
+			error = xfs_btree_check_ptr(cur, lpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+
+		if (ptr < numrecs) {
+			xfs_btree_shift_keys(cur, lkp, -1, numrecs - ptr);
+			xfs_btree_shift_ptrs(cur, lpp, -1, numrecs - ptr);
+			xfs_btree_log_keys(cur, bp, ptr, numrecs - 1);
+			xfs_btree_log_ptrs(cur, bp, ptr, numrecs - 1);
+		}
+
+		/*
+		 * If it's the first record in the block, we'll need to pass a
+		 * key up to the next level (updkey).
+		 */
+		if (ptr == 1)
+			keyp = xfs_btree_key_addr(cur, 1, block);
+	} else {
+		/* It's a leaf. operate on records */
+		if (ptr < numrecs) {
+			xfs_btree_shift_recs(cur,
+				xfs_btree_rec_addr(cur, ptr + 1, block),
+				-1, numrecs - ptr);
+			xfs_btree_log_recs(cur, bp, ptr, numrecs - 1);
+		}
+
+		/*
+		 * If it's the first record in the block, we'll need a key
+		 * structure to pass up to the next level (updkey).
+		 */
+		if (ptr == 1) {
+			cur->bc_ops->init_key_from_rec(&key,
+					xfs_btree_rec_addr(cur, 1, block));
+			keyp = &key;
+		}
+	}
+
+	/*
+	 * Decrement and log the number of entries in the block.
+	 */
+	xfs_btree_set_numrecs(block, --numrecs);
+	xfs_btree_log_block(cur, bp, XFS_BB_NUMRECS);
+
+	/*
+	 * If we are tracking the last record in the tree and
+	 * we are at the far right edge of the tree, update it.
+	 */
+	if (xfs_btree_is_lastrec(cur, block, level)) {
+		cur->bc_ops->update_lastrec(cur, block, NULL,
+					    ptr, LASTREC_DELREC);
+	}
+
+	/*
+	 * We're at the root level.  First, shrink the root block in-memory.
+	 * Try to get rid of the next level down.  If we can't then there's
+	 * nothing left to do.
+	 */
+	if (level == cur->bc_nlevels - 1) {
+		if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+			xfs_iroot_realloc(cur->bc_private.b.ip, -1,
+					  cur->bc_private.b.whichfork);
+
+			error = xfs_btree_kill_iroot(cur);
+			if (error)
+				goto error0;
+
+			error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+			*stat = 1;
+			return 0;
+		}
+
+		/*
+		 * If this is the root level, and there's only one entry left,
+		 * and it's NOT the leaf level, then we can get rid of this
+		 * level.
+		 */
+		if (numrecs == 1 && level > 0) {
+			union xfs_btree_ptr	*pp;
+			/*
+			 * pp is still set to the first pointer in the block.
+			 * Make it the new root of the btree.
+			 */
+			pp = xfs_btree_ptr_addr(cur, 1, block);
+			error = cur->bc_ops->kill_root(cur, bp, level, pp);
+			if (error)
+				goto error0;
+		} else if (level > 0) {
+			error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+		}
+		*stat = 1;
+		return 0;
+	}
+
+	/*
+	 * If we deleted the leftmost entry in the block, update the
+	 * key values above us in the tree.
+	 */
+	if (ptr == 1) {
+		error = xfs_btree_updkey(cur, keyp, level + 1);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * If the number of records remaining in the block is at least
+	 * the minimum, we're done.
+	 */
+	if (numrecs >= cur->bc_ops->get_minrecs(cur, level)) {
+		error = xfs_btree_dec_cursor(cur, level, stat);
+		if (error)
+			goto error0;
+		return 0;
+	}
+
+	/*
+	 * Otherwise, we have to move some records around to keep the
+	 * tree balanced.  Look at the left and right sibling blocks to
+	 * see if we can re-balance by moving only one record.
+	 */
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	xfs_btree_get_sibling(cur, block, &lptr, XFS_BB_LEFTSIB);
+
+	if (cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) {
+		/*
+		 * One child of root, need to get a chance to copy its contents
+		 * into the root and delete it. Can't go up to next level,
+		 * there's nothing to delete there.
+		 */
+		if (xfs_btree_ptr_is_null(cur, &rptr) &&
+		    xfs_btree_ptr_is_null(cur, &lptr) &&
+		    level == cur->bc_nlevels - 2) {
+			error = xfs_btree_kill_iroot(cur);
+			if (!error)
+				error = xfs_btree_dec_cursor(cur, level, stat);
+			if (error)
+				goto error0;
+			return 0;
+		}
+	}
+
+	ASSERT(!xfs_btree_ptr_is_null(cur, &rptr) ||
+	       !xfs_btree_ptr_is_null(cur, &lptr));
+
+	/*
+	 * Duplicate the cursor so our btree manipulations here won't
+	 * disrupt the next level up.
+	 */
+	error = xfs_btree_dup_cursor(cur, &tcur);
+	if (error)
+		goto error0;
+
+	/*
+	 * If there's a right sibling, see if it's ok to shift an entry
+	 * out of it.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &rptr)) {
+		/*
+		 * Move the temp cursor to the last entry in the next block.
+		 * Actually any entry but the first would suffice.
+		 */
+		i = xfs_btree_lastrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		error = xfs_btree_increment(tcur, level, &i);
+		if (error)
+			goto error0;
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		i = xfs_btree_lastrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		/* Grab a pointer to the block. */
+		right = xfs_btree_get_block(tcur, level, &rbp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(tcur, right, level, rbp);
+		if (error)
+			goto error0;
+#endif
+		/* Grab the current block number, for future use. */
+		xfs_btree_get_sibling(tcur, right, &cptr, XFS_BB_LEFTSIB);
+
+		/*
+		 * If right block is full enough so that removing one entry
+		 * won't make it too empty, and left-shifting an entry out
+		 * of right to us works, we're done.
+		 */
+		if (xfs_btree_get_numrecs(right) - 1 >=
+		    cur->bc_ops->get_minrecs(tcur, level)) {
+			error = xfs_btree_lshift(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (i) {
+				ASSERT(xfs_btree_get_numrecs(block) >=
+				       cur->bc_ops->get_minrecs(tcur, level));
+
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				tcur = NULL;
+
+				error = xfs_btree_dec_cursor(cur, level, stat);
+				if (error)
+					goto error0;
+				return 0;
+			}
+		}
+
+		/*
+		 * Otherwise, grab the number of records in right for
+		 * future reference, and fix up the temp cursor to point
+		 * to our block again (last record).
+		 */
+		rrecs = xfs_btree_get_numrecs(right);
+		if (!xfs_btree_ptr_is_null(cur, &lptr)) {
+			i = xfs_btree_firstrec(tcur, level);
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+			error = xfs_btree_decrement(tcur, level, &i);
+			if (error)
+				goto error0;
+			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+		}
+	}
+
+	/*
+	 * If there's a left sibling, see if it's ok to shift an entry
+	 * out of it.
+	 */
+	if (!xfs_btree_ptr_is_null(cur, &lptr)) {
+		/*
+		 * Move the temp cursor to the first entry in the
+		 * previous block.
+		 */
+		i = xfs_btree_firstrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		error = xfs_btree_decrement(tcur, level, &i);
+		if (error)
+			goto error0;
+		i = xfs_btree_firstrec(tcur, level);
+		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
+
+		/* Grab a pointer to the block. */
+		left = xfs_btree_get_block(tcur, level, &lbp);
+#ifdef DEBUG
+		error = xfs_btree_check_block(cur, left, level, lbp);
+		if (error)
+			goto error0;
+#endif
+		/* Grab the current block number, for future use. */
+		xfs_btree_get_sibling(tcur, left, &cptr, XFS_BB_RIGHTSIB);
+
+		/*
+		 * If left block is full enough so that removing one entry
+		 * won't make it too empty, and right-shifting an entry out
+		 * of left to us works, we're done.
+		 */
+		if (xfs_btree_get_numrecs(left) - 1 >=
+		    cur->bc_ops->get_minrecs(tcur, level)) {
+			error = xfs_btree_rshift(tcur, level, &i);
+			if (error)
+				goto error0;
+			if (i) {
+				ASSERT(xfs_btree_get_numrecs(block) >=
+				       cur->bc_ops->get_minrecs(tcur, level));
+				xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+				tcur = NULL;
+				if (level == 0)
+					cur->bc_ptrs[0]++;
+				XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+				*stat = 1;
+				return 0;
+			}
+		}
+
+		/*
+		 * Otherwise, grab the number of records in right for
+		 * future reference.
+		 */
+		lrecs = xfs_btree_get_numrecs(left);
+	}
+
+	/* Delete the temp cursor, we're done with it. */
+	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
+	tcur = NULL;
+
+	/* If here, we need to do a join to keep the tree balanced. */
+	ASSERT(!xfs_btree_ptr_is_null(cur, &cptr));
+
+	if (!xfs_btree_ptr_is_null(cur, &lptr) &&
+	    lrecs + xfs_btree_get_numrecs(block) <=
+			cur->bc_ops->get_maxrecs(cur, level)) {
+		/*
+		 * Set "right" to be the starting block,
+		 * "left" to be the left neighbor.
+		 */
+		rptr = cptr;
+		right = block;
+		rbp = bp;
+		error = xfs_btree_read_buf_block(cur, &lptr, level,
+							0, &left, &lbp);
+		if (error)
+			goto error0;
+
+	/*
+	 * If that won't work, see if we can join with the right neighbor block.
+	 */
+	} else if (!xfs_btree_ptr_is_null(cur, &rptr) &&
+		   rrecs + xfs_btree_get_numrecs(block) <=
+			cur->bc_ops->get_maxrecs(cur, level)) {
+		/*
+		 * Set "left" to be the starting block,
+		 * "right" to be the right neighbor.
+		 */
+		lptr = cptr;
+		left = block;
+		lbp = bp;
+		error = xfs_btree_read_buf_block(cur, &rptr, level,
+							0, &right, &rbp);
+		if (error)
+			goto error0;
+
+	/*
+	 * Otherwise, we can't fix the imbalance.
+	 * Just return.  This is probably a logic error, but it's not fatal.
+	 */
+	} else {
+		error = xfs_btree_dec_cursor(cur, level, stat);
+		if (error)
+			goto error0;
+		return 0;
+	}
+
+	rrecs = xfs_btree_get_numrecs(right);
+	lrecs = xfs_btree_get_numrecs(left);
+
+	/*
+	 * We're now going to join "left" and "right" by moving all the stuff
+	 * in "right" to "left" and deleting "right".
+	 */
+	XFS_BTREE_STATS_ADD(cur, moves, rrecs);
+	if (level > 0) {
+		/* It's a non-leaf.  Move keys and pointers. */
+		union xfs_btree_key	*lkp;	/* left btree key */
+		union xfs_btree_ptr	*lpp;	/* left address pointer */
+		union xfs_btree_key	*rkp;	/* right btree key */
+		union xfs_btree_ptr	*rpp;	/* right address pointer */
+
+		lkp = xfs_btree_key_addr(cur, lrecs + 1, left);
+		lpp = xfs_btree_ptr_addr(cur, lrecs + 1, left);
+		rkp = xfs_btree_key_addr(cur, 1, right);
+		rpp = xfs_btree_ptr_addr(cur, 1, right);
+#ifdef DEBUG
+		for (i = 1; i < rrecs; i++) {
+			error = xfs_btree_check_ptr(cur, rpp, i, level);
+			if (error)
+				goto error0;
+		}
+#endif
+		xfs_btree_copy_keys(cur, lkp, rkp, rrecs);
+		xfs_btree_copy_ptrs(cur, lpp, rpp, rrecs);
+
+		xfs_btree_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
+		xfs_btree_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
+	} else {
+		/* It's a leaf.  Move records.  */
+		union xfs_btree_rec	*lrp;	/* left record pointer */
+		union xfs_btree_rec	*rrp;	/* right record pointer */
+
+		lrp = xfs_btree_rec_addr(cur, lrecs + 1, left);
+		rrp = xfs_btree_rec_addr(cur, 1, right);
+
+		xfs_btree_copy_recs(cur, lrp, rrp, rrecs);
+		xfs_btree_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
+	}
+
+	XFS_BTREE_STATS_INC(cur, join);
+
+	/*
+	 * Fix up the the number of records and right block pointer in the
+	 * surviving block, and log it.
+	 */
+	xfs_btree_set_numrecs(left, lrecs + rrecs);
+	xfs_btree_get_sibling(cur, right, &cptr, XFS_BB_RIGHTSIB),
+	xfs_btree_set_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
+	xfs_btree_log_block(cur, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
+
+	/* If there is a right sibling, point it to the remaining block. */
+	xfs_btree_get_sibling(cur, left, &cptr, XFS_BB_RIGHTSIB);
+	if (!xfs_btree_ptr_is_null(cur, &cptr)) {
+		error = xfs_btree_read_buf_block(cur, &cptr, level,
+							0, &rrblock, &rrbp);
+		if (error)
+			goto error0;
+		xfs_btree_set_sibling(cur, rrblock, &lptr, XFS_BB_LEFTSIB);
+		xfs_btree_log_block(cur, rrbp, XFS_BB_LEFTSIB);
+	}
+
+	/* Free the deleted block. */
+	error = cur->bc_ops->free_block(cur, rbp);
+	if (error)
+		goto error0;
+	XFS_BTREE_STATS_INC(cur, free);
+
+	/*
+	 * If we joined with the left neighbor, set the buffer in the
+	 * cursor to the left block, and fix up the index.
+	 */
+	if (bp != lbp) {
+		cur->bc_bufs[level] = lbp;
+		cur->bc_ptrs[level] += lrecs;
+		cur->bc_ra[level] = 0;
+	}
+	/*
+	 * If we joined with the right neighbor and there's a level above
+	 * us, increment the cursor at that level.
+	 */
+	else if ((cur->bc_flags & XFS_BTREE_ROOT_IN_INODE) ||
+		   (level + 1 < cur->bc_nlevels)) {
+		error = xfs_btree_increment(cur, level + 1, &i);
+		if (error)
+			goto error0;
+	}
+
+	/*
+	 * Readjust the ptr at this level if it's not a leaf, since it's
+	 * still pointing at the deletion point, which makes the cursor
+	 * inconsistent.  If this makes the ptr 0, the caller fixes it up.
+	 * We can't use decrement because it would change the next level up.
+	 */
+	if (level > 0)
+		cur->bc_ptrs[level]--;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	/* Return value means the next level up has something to do. */
+	*stat = 2;
+	return 0;
+
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	if (tcur)
+		xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/*
+ * Delete the record pointed to by cur.
+ * The cursor refers to the place where the record was (could be inserted)
+ * when the operation returns.
+ */
+int					/* error */
+xfs_btree_delete(
+	struct xfs_btree_cur	*cur,
+	int			*stat)	/* success/failure */
+{
+	int			error;	/* error return value */
+	int			level;
+	int			i;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+
+	/*
+	 * Go up the tree, starting at leaf level.
+	 *
+	 * If 2 is returned then a join was done; go to the next level.
+	 * Otherwise we are done.
+	 */
+	for (level = 0, i = 2; i == 2; level++) {
+		error = xfs_btree_delrec(cur, level, &i);
+		if (error)
+			goto error0;
+	}
+
+	if (i == 0) {
+		for (level = 1; level < cur->bc_nlevels; level++) {
+			if (cur->bc_ptrs[level] == 0) {
+				error = xfs_btree_decrement(cur, level, &i);
+				if (error)
+					goto error0;
+				break;
+			}
+		}
+	}
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+	*stat = i;
+	return 0;
+error0:
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
+	return error;
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_btree_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	union xfs_btree_rec	**recp,	/* output: btree record */
+	int			*stat)	/* output: success/failure */
+{
+	struct xfs_btree_block	*block;	/* btree block */
+	struct xfs_buf		*bp;	/* buffer pointer */
+	int			ptr;	/* record number */
+#ifdef DEBUG
+	int			error;	/* error return value */
+#endif
+
+	ptr = cur->bc_ptrs[0];
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+#ifdef DEBUG
+	error = xfs_btree_check_block(cur, block, 0, bp);
+	if (error)
+		return error;
+#endif
+
+	/*
+	 * Off the right end or left end, return failure.
+	 */
+	if (ptr > xfs_btree_get_numrecs(block) || ptr <= 0) {
+		*stat = 0;
+		return 0;
+	}
+
+	/*
+	 * Point to the record and extract its data.
+	 */
+	*recp = xfs_btree_rec_addr(cur, ptr, block);
+	*stat = 1;
+	return 0;
+}
diff --git a/fs/xfs/xfs_btree.h b/fs/xfs/xfs_btree.h
index 1f528a2a375..789fffdf8b2 100644
--- a/fs/xfs/xfs_btree.h
+++ b/fs/xfs/xfs_btree.h
@@ -39,39 +39,19 @@ extern kmem_zone_t	*xfs_btree_cur_zone;
 #define	XFS_BTNUM_INO	((xfs_btnum_t)XFS_BTNUM_INOi)
 
 /*
- * Short form header: space allocation btrees.
- */
-typedef struct xfs_btree_sblock {
-	__be32		bb_magic;	/* magic number for block type */
-	__be16		bb_level;	/* 0 is a leaf */
-	__be16		bb_numrecs;	/* current # of data records */
-	__be32		bb_leftsib;	/* left sibling block or NULLAGBLOCK */
-	__be32		bb_rightsib;	/* right sibling block or NULLAGBLOCK */
-} xfs_btree_sblock_t;
-
-/*
- * Long form header: bmap btrees.
- */
-typedef struct xfs_btree_lblock {
-	__be32		bb_magic;	/* magic number for block type */
-	__be16		bb_level;	/* 0 is a leaf */
-	__be16		bb_numrecs;	/* current # of data records */
-	__be64		bb_leftsib;	/* left sibling block or NULLDFSBNO */
-	__be64		bb_rightsib;	/* right sibling block or NULLDFSBNO */
-} xfs_btree_lblock_t;
-
-/*
- * Combined header and structure, used by common code.
+ * Generic btree header.
+ *
+ * This is a comination of the actual format used on disk for short and long
+ * format btrees.  The first three fields are shared by both format, but
+ * the pointers are different and should be used with care.
+ *
+ * To get the size of the actual short or long form headers please use
+ * the size macros below.  Never use sizeof(xfs_btree_block).
  */
-typedef struct xfs_btree_hdr
-{
+struct xfs_btree_block {
 	__be32		bb_magic;	/* magic number for block type */
 	__be16		bb_level;	/* 0 is a leaf */
 	__be16		bb_numrecs;	/* current # of data records */
-} xfs_btree_hdr_t;
-
-typedef struct xfs_btree_block {
-	xfs_btree_hdr_t	bb_h;		/* header */
 	union {
 		struct {
 			__be32		bb_leftsib;
@@ -82,7 +62,36 @@ typedef struct xfs_btree_block {
 			__be64		bb_rightsib;
 		} l;			/* long form pointers */
 	} bb_u;				/* rest */
-} xfs_btree_block_t;
+};
+
+#define XFS_BTREE_SBLOCK_LEN	16	/* size of a short form block */
+#define XFS_BTREE_LBLOCK_LEN	24	/* size of a long form block */
+
+
+/*
+ * Generic key, ptr and record wrapper structures.
+ *
+ * These are disk format structures, and are converted where necessary
+ * by the btree specific code that needs to interpret them.
+ */
+union xfs_btree_ptr {
+	__be32			s;	/* short form ptr */
+	__be64			l;	/* long form ptr */
+};
+
+union xfs_btree_key {
+	xfs_bmbt_key_t		bmbt;
+	xfs_bmdr_key_t		bmbr;	/* bmbt root block */
+	xfs_alloc_key_t		alloc;
+	xfs_inobt_key_t		inobt;
+};
+
+union xfs_btree_rec {
+	xfs_bmbt_rec_t		bmbt;
+	xfs_bmdr_rec_t		bmbr;	/* bmbt root block */
+	xfs_alloc_rec_t		alloc;
+	xfs_inobt_rec_t		inobt;
+};
 
 /*
  * For logging record fields.
@@ -96,46 +105,131 @@ typedef struct xfs_btree_block {
 #define	XFS_BB_ALL_BITS		((1 << XFS_BB_NUM_BITS) - 1)
 
 /*
- * Boolean to select which form of xfs_btree_block_t.bb_u to use.
- */
-#define	XFS_BTREE_LONG_PTRS(btnum)	((btnum) == XFS_BTNUM_BMAP)
-
-/*
  * Magic numbers for btree blocks.
  */
 extern const __uint32_t	xfs_magics[];
 
 /*
- * Maximum and minimum records in a btree block.
- * Given block size, type prefix, and leaf flag (0 or 1).
- * The divisor below is equivalent to lf ? (e1) : (e2) but that produces
- * compiler warnings.
- */
-#define	XFS_BTREE_BLOCK_MAXRECS(bsz,t,lf)	\
-	((int)(((bsz) - (uint)sizeof(t ## _block_t)) / \
-	 (((lf) * (uint)sizeof(t ## _rec_t)) + \
-	  ((1 - (lf)) * \
-	   ((uint)sizeof(t ## _key_t) + (uint)sizeof(t ## _ptr_t))))))
-#define	XFS_BTREE_BLOCK_MINRECS(bsz,t,lf)	\
-	(XFS_BTREE_BLOCK_MAXRECS(bsz,t,lf) / 2)
-
-/*
- * Record, key, and pointer address calculation macros.
- * Given block size, type prefix, block pointer, and index of requested entry
- * (first entry numbered 1).
- */
-#define	XFS_BTREE_REC_ADDR(t,bb,i)	\
-	((t ## _rec_t *)((char *)(bb) + sizeof(t ## _block_t) + \
-	 ((i) - 1) * sizeof(t ## _rec_t)))
-#define	XFS_BTREE_KEY_ADDR(t,bb,i)	\
-	((t ## _key_t *)((char *)(bb) + sizeof(t ## _block_t) + \
-	 ((i) - 1) * sizeof(t ## _key_t)))
-#define	XFS_BTREE_PTR_ADDR(t,bb,i,mxr)	\
-	((t ## _ptr_t *)((char *)(bb) + sizeof(t ## _block_t) + \
-	 (mxr) * sizeof(t ## _key_t) + ((i) - 1) * sizeof(t ## _ptr_t)))
+ * Generic stats interface
+ */
+#define __XFS_BTREE_STATS_INC(type, stat) \
+	XFS_STATS_INC(xs_ ## type ## _2_ ## stat)
+#define XFS_BTREE_STATS_INC(cur, stat)  \
+do {    \
+	switch (cur->bc_btnum) {  \
+	case XFS_BTNUM_BNO: __XFS_BTREE_STATS_INC(abtb, stat); break;	\
+	case XFS_BTNUM_CNT: __XFS_BTREE_STATS_INC(abtc, stat); break;	\
+	case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_INC(bmbt, stat); break;	\
+	case XFS_BTNUM_INO: __XFS_BTREE_STATS_INC(ibt, stat); break;	\
+	case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break;	\
+	}       \
+} while (0)
+
+#define __XFS_BTREE_STATS_ADD(type, stat, val) \
+	XFS_STATS_ADD(xs_ ## type ## _2_ ## stat, val)
+#define XFS_BTREE_STATS_ADD(cur, stat, val)  \
+do {    \
+	switch (cur->bc_btnum) {  \
+	case XFS_BTNUM_BNO: __XFS_BTREE_STATS_ADD(abtb, stat, val); break; \
+	case XFS_BTNUM_CNT: __XFS_BTREE_STATS_ADD(abtc, stat, val); break; \
+	case XFS_BTNUM_BMAP: __XFS_BTREE_STATS_ADD(bmbt, stat, val); break; \
+	case XFS_BTNUM_INO: __XFS_BTREE_STATS_ADD(ibt, stat, val); break; \
+	case XFS_BTNUM_MAX: ASSERT(0); /* fucking gcc */ ; break;	\
+	}       \
+} while (0)
 
 #define	XFS_BTREE_MAXLEVELS	8	/* max of all btrees */
 
+struct xfs_btree_ops {
+	/* size of the key and record structures */
+	size_t	key_len;
+	size_t	rec_len;
+
+	/* cursor operations */
+	struct xfs_btree_cur *(*dup_cursor)(struct xfs_btree_cur *);
+	void	(*update_cursor)(struct xfs_btree_cur *src,
+				 struct xfs_btree_cur *dst);
+
+	/* update btree root pointer */
+	void	(*set_root)(struct xfs_btree_cur *cur,
+				union xfs_btree_ptr *nptr, int level_change);
+	int	(*kill_root)(struct xfs_btree_cur *cur, struct xfs_buf *bp,
+				int level, union xfs_btree_ptr *newroot);
+
+	/* block allocation / freeing */
+	int	(*alloc_block)(struct xfs_btree_cur *cur,
+			       union xfs_btree_ptr *start_bno,
+			       union xfs_btree_ptr *new_bno,
+			       int length, int *stat);
+	int	(*free_block)(struct xfs_btree_cur *cur, struct xfs_buf *bp);
+
+	/* update last record information */
+	void	(*update_lastrec)(struct xfs_btree_cur *cur,
+				  struct xfs_btree_block *block,
+				  union xfs_btree_rec *rec,
+				  int ptr, int reason);
+
+	/* records in block/level */
+	int	(*get_minrecs)(struct xfs_btree_cur *cur, int level);
+	int	(*get_maxrecs)(struct xfs_btree_cur *cur, int level);
+
+	/* records on disk.  Matter for the root in inode case. */
+	int	(*get_dmaxrecs)(struct xfs_btree_cur *cur, int level);
+
+	/* init values of btree structures */
+	void	(*init_key_from_rec)(union xfs_btree_key *key,
+				     union xfs_btree_rec *rec);
+	void	(*init_rec_from_key)(union xfs_btree_key *key,
+				     union xfs_btree_rec *rec);
+	void	(*init_rec_from_cur)(struct xfs_btree_cur *cur,
+				     union xfs_btree_rec *rec);
+	void	(*init_ptr_from_cur)(struct xfs_btree_cur *cur,
+				     union xfs_btree_ptr *ptr);
+
+	/* difference between key value and cursor value */
+	__int64_t (*key_diff)(struct xfs_btree_cur *cur,
+			      union xfs_btree_key *key);
+
+#ifdef DEBUG
+	/* check that k1 is lower than k2 */
+	int	(*keys_inorder)(struct xfs_btree_cur *cur,
+				union xfs_btree_key *k1,
+				union xfs_btree_key *k2);
+
+	/* check that r1 is lower than r2 */
+	int	(*recs_inorder)(struct xfs_btree_cur *cur,
+				union xfs_btree_rec *r1,
+				union xfs_btree_rec *r2);
+#endif
+
+	/* btree tracing */
+#ifdef XFS_BTREE_TRACE
+	void		(*trace_enter)(struct xfs_btree_cur *, const char *,
+				       char *, int, int, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t,
+				       __psunsigned_t, __psunsigned_t);
+	void		(*trace_cursor)(struct xfs_btree_cur *, __uint32_t *,
+					__uint64_t *, __uint64_t *);
+	void		(*trace_key)(struct xfs_btree_cur *,
+				     union xfs_btree_key *, __uint64_t *,
+				     __uint64_t *);
+	void		(*trace_record)(struct xfs_btree_cur *,
+					union xfs_btree_rec *, __uint64_t *,
+					__uint64_t *, __uint64_t *);
+#endif
+};
+
+/*
+ * Reasons for the update_lastrec method to be called.
+ */
+#define LASTREC_UPDATE	0
+#define LASTREC_INSREC	1
+#define LASTREC_DELREC	2
+
+
 /*
  * Btree cursor structure.
  * This collects all information needed by the btree code in one place.
@@ -144,6 +238,8 @@ typedef struct xfs_btree_cur
 {
 	struct xfs_trans	*bc_tp;	/* transaction we're in, if any */
 	struct xfs_mount	*bc_mp;	/* file system mount struct */
+	const struct xfs_btree_ops *bc_ops;
+	uint			bc_flags; /* btree features - below */
 	union {
 		xfs_alloc_rec_incore_t	a;
 		xfs_bmbt_irec_t		b;
@@ -175,94 +271,40 @@ typedef struct xfs_btree_cur
 	}		bc_private;	/* per-btree type data */
 } xfs_btree_cur_t;
 
+/* cursor flags */
+#define XFS_BTREE_LONG_PTRS		(1<<0)	/* pointers are 64bits long */
+#define XFS_BTREE_ROOT_IN_INODE		(1<<1)	/* root may be variable size */
+#define XFS_BTREE_LASTREC_UPDATE	(1<<2)	/* track last rec externally */
+
+
 #define	XFS_BTREE_NOERROR	0
 #define	XFS_BTREE_ERROR		1
 
 /*
  * Convert from buffer to btree block header.
  */
-#define	XFS_BUF_TO_BLOCK(bp)	((xfs_btree_block_t *)XFS_BUF_PTR(bp))
-#define	XFS_BUF_TO_LBLOCK(bp)	((xfs_btree_lblock_t *)XFS_BUF_PTR(bp))
-#define	XFS_BUF_TO_SBLOCK(bp)	((xfs_btree_sblock_t *)XFS_BUF_PTR(bp))
+#define	XFS_BUF_TO_BLOCK(bp)	((struct xfs_btree_block *)XFS_BUF_PTR(bp))
 
 
-#ifdef __KERNEL__
-
-#ifdef DEBUG
 /*
- * Debug routine: check that block header is ok.
+ * Check that block header is ok.
  */
-void
+int
 xfs_btree_check_block(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_block_t	*block,	/* generic btree block pointer */
-	int			level,	/* level of the btree block */
-	struct xfs_buf		*bp);	/* buffer containing block, if any */
-
-/*
- * Debug routine: check that keys are in the right order.
- */
-void
-xfs_btree_check_key(
-	xfs_btnum_t		btnum,	/* btree identifier */
-	void			*ak1,	/* pointer to left (lower) key */
-	void			*ak2);	/* pointer to right (higher) key */
-
-/*
- * Debug routine: check that records are in the right order.
- */
-void
-xfs_btree_check_rec(
-	xfs_btnum_t		btnum,	/* btree identifier */
-	void			*ar1,	/* pointer to left (lower) record */
-	void			*ar2);	/* pointer to right (higher) record */
-#else
-#define	xfs_btree_check_block(a,b,c,d)
-#define	xfs_btree_check_key(a,b,c)
-#define	xfs_btree_check_rec(a,b,c)
-#endif	/* DEBUG */
-
-/*
- * Checking routine: check that long form block header is ok.
- */
-int					/* error (0 or EFSCORRUPTED) */
-xfs_btree_check_lblock(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_lblock_t	*block,	/* btree long form block pointer */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	struct xfs_btree_block	*block,	/* generic btree block pointer */
 	int			level,	/* level of the btree block */
 	struct xfs_buf		*bp);	/* buffer containing block, if any */
 
 /*
- * Checking routine: check that (long) pointer is ok.
+ * Check that (long) pointer is ok.
  */
 int					/* error (0 or EFSCORRUPTED) */
 xfs_btree_check_lptr(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
+	struct xfs_btree_cur	*cur,	/* btree cursor */
 	xfs_dfsbno_t		ptr,	/* btree block disk address */
 	int			level);	/* btree block level */
 
-#define xfs_btree_check_lptr_disk(cur, ptr, level) \
-	xfs_btree_check_lptr(cur, be64_to_cpu(ptr), level)
-
-/*
- * Checking routine: check that short form block header is ok.
- */
-int					/* error (0 or EFSCORRUPTED) */
-xfs_btree_check_sblock(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_btree_sblock_t	*block,	/* btree short form block pointer */
-	int			level,	/* level of the btree block */
-	struct xfs_buf		*bp);	/* buffer containing block */
-
-/*
- * Checking routine: check that (short) pointer is ok.
- */
-int					/* error (0 or EFSCORRUPTED) */
-xfs_btree_check_sptr(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_agblock_t		ptr,	/* btree block disk address */
-	int			level);	/* btree block level */
-
 /*
  * Delete the btree cursor.
  */
@@ -281,15 +323,6 @@ xfs_btree_dup_cursor(
 	xfs_btree_cur_t		**ncur);/* output cursor */
 
 /*
- * Change the cursor to point to the first record in the current block
- * at the given level.  Other levels are unaffected.
- */
-int					/* success=1, failure=0 */
-xfs_btree_firstrec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level);	/* level to change */
-
-/*
  * Get a buffer for the block, return it with no data read.
  * Long-form addressing.
  */
@@ -313,20 +346,6 @@ xfs_btree_get_bufs(
 	uint			lock);	/* lock flags for get_buf */
 
 /*
- * Allocate a new btree cursor.
- * The cursor is either for allocation (A) or bmap (B).
- */
-xfs_btree_cur_t *			/* new btree cursor */
-xfs_btree_init_cursor(
-	struct xfs_mount	*mp,	/* file system mount point */
-	struct xfs_trans	*tp,	/* transaction pointer */
-	struct xfs_buf		*agbp,	/* (A only) buffer for agf structure */
-	xfs_agnumber_t		agno,	/* (A only) allocation group number */
-	xfs_btnum_t		btnum,	/* btree identifier */
-	struct xfs_inode	*ip,	/* (B only) inode owning the btree */
-	int			whichfork); /* (B only) data/attr fork */
-
-/*
  * Check for the cursor referring to the last block at the given level.
  */
 int					/* 1=is last block, 0=not last block */
@@ -335,15 +354,6 @@ xfs_btree_islastblock(
 	int			level);	/* level to check */
 
 /*
- * Change the cursor to point to the last record in the current block
- * at the given level.  Other levels are unaffected.
- */
-int					/* success=1, failure=0 */
-xfs_btree_lastrec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level);	/* level to change */
-
-/*
  * Compute first and last byte offsets for the fields given.
  * Interprets the offsets table, which contains struct field offsets.
  */
@@ -404,39 +414,53 @@ xfs_btree_reada_bufs(
 	xfs_extlen_t		count);	/* count of filesystem blocks */
 
 /*
- * Read-ahead btree blocks, at the given level.
- * Bits in lr are set from XFS_BTCUR_{LEFT,RIGHT}RA.
+ * Set the buffer for level "lev" in the cursor to bp, releasing
+ * any previous buffer.
  */
-int					/* readahead block count */
-xfs_btree_readahead_core(
+void
+xfs_btree_setbuf(
 	xfs_btree_cur_t		*cur,	/* btree cursor */
 	int			lev,	/* level in btree */
-	int			lr);	/* left/right bits */
+	struct xfs_buf		*bp);	/* new buffer to set */
 
-static inline int			/* readahead block count */
-xfs_btree_readahead(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			lev,	/* level in btree */
-	int			lr)	/* left/right bits */
-{
-	if ((cur->bc_ra[lev] | lr) == cur->bc_ra[lev])
-		return 0;
 
-	return xfs_btree_readahead_core(cur, lev, lr);
-}
+/*
+ * Common btree core entry points.
+ */
+int xfs_btree_increment(struct xfs_btree_cur *, int, int *);
+int xfs_btree_decrement(struct xfs_btree_cur *, int, int *);
+int xfs_btree_lookup(struct xfs_btree_cur *, xfs_lookup_t, int *);
+int xfs_btree_update(struct xfs_btree_cur *, union xfs_btree_rec *);
+int xfs_btree_new_iroot(struct xfs_btree_cur *, int *, int *);
+int xfs_btree_kill_iroot(struct xfs_btree_cur *);
+int xfs_btree_insert(struct xfs_btree_cur *, int *);
+int xfs_btree_delete(struct xfs_btree_cur *, int *);
+int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
 
+/*
+ * Internal btree helpers also used by xfs_bmap.c.
+ */
+void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, int);
+void xfs_btree_log_recs(struct xfs_btree_cur *, struct xfs_buf *, int, int);
 
 /*
- * Set the buffer for level "lev" in the cursor to bp, releasing
- * any previous buffer.
+ * Helpers.
  */
-void
-xfs_btree_setbuf(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			lev,	/* level in btree */
-	struct xfs_buf		*bp);	/* new buffer to set */
+static inline int xfs_btree_get_numrecs(struct xfs_btree_block *block)
+{
+	return be16_to_cpu(block->bb_numrecs);
+}
+
+static inline void xfs_btree_set_numrecs(struct xfs_btree_block *block,
+		__uint16_t numrecs)
+{
+	block->bb_numrecs = cpu_to_be16(numrecs);
+}
 
-#endif	/* __KERNEL__ */
+static inline int xfs_btree_get_level(struct xfs_btree_block *block)
+{
+	return be16_to_cpu(block->bb_level);
+}
 
 
 /*
diff --git a/fs/xfs/xfs_btree_trace.c b/fs/xfs/xfs_btree_trace.c
new file mode 100644
index 00000000000..44ff942a0fd
--- /dev/null
+++ b/fs/xfs/xfs_btree_trace.c
@@ -0,0 +1,249 @@
+/*
+ * Copyright (c) 2008 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#include "xfs.h"
+#include "xfs_types.h"
+#include "xfs_inum.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_inode.h"
+#include "xfs_btree.h"
+#include "xfs_btree_trace.h"
+
+STATIC void
+xfs_btree_trace_ptr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	ptr,
+	__psunsigned_t		*high,
+	__psunsigned_t		*low)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		__u64 val = be64_to_cpu(ptr.l);
+		*high = val >> 32;
+		*low = (int)val;
+	} else {
+		*high = 0;
+		*low = be32_to_cpu(ptr.s);
+	}
+}
+
+/*
+ * Add a trace buffer entry for arguments, for a buffer & 1 integer arg.
+ */
+void
+xfs_btree_trace_argbi(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*b,
+	int			i,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGBI,
+				 line, (__psunsigned_t)b, i, 0, 0, 0, 0, 0,
+				 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for a buffer & 2 integer args.
+ */
+void
+xfs_btree_trace_argbii(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*b,
+	int			i0,
+	int			i1,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGBII,
+				 line, (__psunsigned_t)b, i0, i1, 0, 0, 0, 0,
+				 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for 3 block-length args
+ * and an integer arg.
+ */
+void
+xfs_btree_trace_argfffi(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	xfs_dfiloff_t		o,
+	xfs_dfsbno_t		b,
+	xfs_dfilblks_t		i,
+	int			j,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGFFFI,
+				 line,
+				 o >> 32, (int)o,
+				 b >> 32, (int)b,
+				 i >> 32, (int)i,
+				 (int)j, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for one integer arg.
+ */
+void
+xfs_btree_trace_argi(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	int			line)
+{
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGI,
+				 line, i, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for int, fsblock, key.
+ */
+void
+xfs_btree_trace_argipk(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	union xfs_btree_ptr	ptr,
+	union xfs_btree_key	*key,
+	int			line)
+{
+	__psunsigned_t		high, low;
+	__uint64_t		l0, l1;
+
+	xfs_btree_trace_ptr(cur, ptr, &high, &low);
+	cur->bc_ops->trace_key(cur, key, &l0, &l1);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGIPK,
+				 line, i, high, low,
+				 l0 >> 32, (int)l0,
+				 l1 >> 32, (int)l1,
+				 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for int, fsblock, rec.
+ */
+void
+xfs_btree_trace_argipr(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	union xfs_btree_ptr	ptr,
+	union xfs_btree_rec	*rec,
+	int			line)
+{
+	__psunsigned_t		high, low;
+	__uint64_t		l0, l1, l2;
+
+	xfs_btree_trace_ptr(cur, ptr, &high, &low);
+	cur->bc_ops->trace_record(cur, rec, &l0, &l1, &l2);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGIPR,
+			      line, i,
+			      high, low,
+			      l0 >> 32, (int)l0,
+			      l1 >> 32, (int)l1,
+			      l2 >> 32, (int)l2,
+			      0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for int, key.
+ */
+void
+xfs_btree_trace_argik(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			i,
+	union xfs_btree_key	*key,
+	int			line)
+{
+	__uint64_t		l0, l1;
+
+	cur->bc_ops->trace_key(cur, key, &l0, &l1);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGIK,
+				 line, i,
+				 l0 >> 32, (int)l0,
+				 l1 >> 32, (int)l1,
+				 0, 0, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for arguments, for record.
+ */
+void
+xfs_btree_trace_argr(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	int			line)
+{
+	__uint64_t		l0, l1, l2;
+
+	cur->bc_ops->trace_record(cur, rec, &l0, &l1, &l2);
+	cur->bc_ops->trace_enter(cur, func, XBT_ARGS, XFS_BTREE_KTRACE_ARGR,
+			      line,
+			      l0 >> 32, (int)l0,
+			      l1 >> 32, (int)l1,
+			      l2 >> 32, (int)l2,
+			      0, 0, 0, 0, 0);
+}
+
+/*
+ * Add a trace buffer entry for the cursor/operation.
+ */
+void
+xfs_btree_trace_cursor(
+	const char		*func,
+	struct xfs_btree_cur	*cur,
+	int			type,
+	int			line)
+{
+	__uint32_t		s0;
+	__uint64_t		l0, l1;
+	char			*s;
+
+	switch (type) {
+	case XBT_ARGS:
+		s = "args";
+		break;
+	case XBT_ENTRY:
+		s = "entry";
+		break;
+	case XBT_ERROR:
+		s = "error";
+		break;
+	case XBT_EXIT:
+		s = "exit";
+		break;
+	default:
+		s = "unknown";
+		break;
+	}
+
+	cur->bc_ops->trace_cursor(cur, &s0, &l0, &l1);
+	cur->bc_ops->trace_enter(cur, func, s, XFS_BTREE_KTRACE_CUR, line,
+				 s0,
+				 l0 >> 32, (int)l0,
+				 l1 >> 32, (int)l1,
+				 (__psunsigned_t)cur->bc_bufs[0],
+				 (__psunsigned_t)cur->bc_bufs[1],
+				 (__psunsigned_t)cur->bc_bufs[2],
+				 (__psunsigned_t)cur->bc_bufs[3],
+				 (cur->bc_ptrs[0] << 16) | cur->bc_ptrs[1],
+				 (cur->bc_ptrs[2] << 16) | cur->bc_ptrs[3]);
+}
diff --git a/fs/xfs/xfs_btree_trace.h b/fs/xfs/xfs_btree_trace.h
new file mode 100644
index 00000000000..b3f5eb3c3c6
--- /dev/null
+++ b/fs/xfs/xfs_btree_trace.h
@@ -0,0 +1,116 @@
+/*
+ * Copyright (c) 2008 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+#ifndef __XFS_BTREE_TRACE_H__
+#define	__XFS_BTREE_TRACE_H__
+
+struct xfs_btree_cur;
+struct xfs_buf;
+
+
+/*
+ * Trace hooks.
+ * i,j = integer (32 bit)
+ * b = btree block buffer (xfs_buf_t)
+ * p = btree ptr
+ * r = btree record
+ * k = btree key
+ */
+
+#ifdef XFS_BTREE_TRACE
+
+/*
+ * Trace buffer entry types.
+ */
+#define XFS_BTREE_KTRACE_ARGBI   1
+#define XFS_BTREE_KTRACE_ARGBII  2
+#define XFS_BTREE_KTRACE_ARGFFFI 3
+#define XFS_BTREE_KTRACE_ARGI    4
+#define XFS_BTREE_KTRACE_ARGIPK  5
+#define XFS_BTREE_KTRACE_ARGIPR  6
+#define XFS_BTREE_KTRACE_ARGIK   7
+#define XFS_BTREE_KTRACE_ARGR	 8
+#define XFS_BTREE_KTRACE_CUR     9
+
+/*
+ * Sub-types for cursor traces.
+ */
+#define XBT_ARGS	0
+#define XBT_ENTRY	1
+#define XBT_ERROR	2
+#define XBT_EXIT	3
+
+void xfs_btree_trace_argbi(const char *, struct xfs_btree_cur *,
+		struct xfs_buf *, int, int);
+void xfs_btree_trace_argbii(const char *, struct xfs_btree_cur *,
+		struct xfs_buf *, int, int, int);
+void xfs_btree_trace_argfffi(const char *, struct xfs_btree_cur *,
+		xfs_dfiloff_t, xfs_dfsbno_t, xfs_dfilblks_t, int, int);
+void xfs_btree_trace_argi(const char *, struct xfs_btree_cur *, int, int);
+void xfs_btree_trace_argipk(const char *, struct xfs_btree_cur *, int,
+		union xfs_btree_ptr, union xfs_btree_key *, int);
+void xfs_btree_trace_argipr(const char *, struct xfs_btree_cur *, int,
+		union xfs_btree_ptr, union xfs_btree_rec *, int);
+void xfs_btree_trace_argik(const char *, struct xfs_btree_cur *, int,
+		union xfs_btree_key *, int);
+void xfs_btree_trace_argr(const char *, struct xfs_btree_cur *,
+		union xfs_btree_rec *, int);
+void xfs_btree_trace_cursor(const char *, struct xfs_btree_cur *, int, int);
+
+
+#define XFS_ALLOCBT_TRACE_SIZE	4096	/* size of global trace buffer */
+extern ktrace_t	*xfs_allocbt_trace_buf;
+
+#define XFS_INOBT_TRACE_SIZE	4096	/* size of global trace buffer */
+extern ktrace_t	*xfs_inobt_trace_buf;
+
+#define XFS_BMBT_TRACE_SIZE	4096	/* size of global trace buffer */
+#define XFS_BMBT_KTRACE_SIZE	32	/* size of per-inode trace buffer */
+extern ktrace_t	*xfs_bmbt_trace_buf;
+
+
+#define	XFS_BTREE_TRACE_ARGBI(c, b, i)	\
+	xfs_btree_trace_argbi(__func__, c, b, i, __LINE__)
+#define	XFS_BTREE_TRACE_ARGBII(c, b, i, j)	\
+	xfs_btree_trace_argbii(__func__, c, b, i, j, __LINE__)
+#define	XFS_BTREE_TRACE_ARGFFFI(c, o, b, i, j)	\
+	xfs_btree_trace_argfffi(__func__, c, o, b, i, j, __LINE__)
+#define	XFS_BTREE_TRACE_ARGI(c, i)	\
+	xfs_btree_trace_argi(__func__, c, i, __LINE__)
+#define	XFS_BTREE_TRACE_ARGIPK(c, i, p, k)	\
+	xfs_btree_trace_argipk(__func__, c, i, p, k, __LINE__)
+#define	XFS_BTREE_TRACE_ARGIPR(c, i, p, r)	\
+	xfs_btree_trace_argipr(__func__, c, i, p, r, __LINE__)
+#define	XFS_BTREE_TRACE_ARGIK(c, i, k)	\
+	xfs_btree_trace_argik(__func__, c, i, k, __LINE__)
+#define XFS_BTREE_TRACE_ARGR(c, r)	\
+	xfs_btree_trace_argr(__func__, c, r, __LINE__)
+#define	XFS_BTREE_TRACE_CURSOR(c, t)	\
+	xfs_btree_trace_cursor(__func__, c, t, __LINE__)
+#else
+#define	XFS_BTREE_TRACE_ARGBI(c, b, i)
+#define	XFS_BTREE_TRACE_ARGBII(c, b, i, j)
+#define	XFS_BTREE_TRACE_ARGFFFI(c, o, b, i, j)
+#define	XFS_BTREE_TRACE_ARGI(c, i)
+#define	XFS_BTREE_TRACE_ARGIPK(c, i, p, s)
+#define	XFS_BTREE_TRACE_ARGIPR(c, i, p, r)
+#define	XFS_BTREE_TRACE_ARGIK(c, i, k)
+#define XFS_BTREE_TRACE_ARGR(c, r)
+#define	XFS_BTREE_TRACE_CURSOR(c, t)
+#endif	/* XFS_BTREE_TRACE */
+
+#endif /* __XFS_BTREE_TRACE_H__ */
diff --git a/fs/xfs/xfs_buf_item.c b/fs/xfs/xfs_buf_item.c
index 002fc2617c8..92af4098c7e 100644
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -375,7 +375,7 @@ xfs_buf_item_unpin(
 	xfs_buf_log_item_t	*bip,
 	int			stale)
 {
-	xfs_mount_t	*mp;
+	struct xfs_ail	*ailp;
 	xfs_buf_t	*bp;
 	int		freed;
 
@@ -387,7 +387,7 @@ xfs_buf_item_unpin(
 	xfs_buftrace("XFS_UNPIN", bp);
 
 	freed = atomic_dec_and_test(&bip->bli_refcount);
-	mp = bip->bli_item.li_mountp;
+	ailp = bip->bli_item.li_ailp;
 	xfs_bunpin(bp);
 	if (freed && stale) {
 		ASSERT(bip->bli_flags & XFS_BLI_STALE);
@@ -399,17 +399,17 @@ xfs_buf_item_unpin(
 		xfs_buftrace("XFS_UNPIN STALE", bp);
 		/*
 		 * If we get called here because of an IO error, we may
-		 * or may not have the item on the AIL. xfs_trans_delete_ail()
+		 * or may not have the item on the AIL. xfs_trans_ail_delete()
 		 * will take care of that situation.
-		 * xfs_trans_delete_ail() drops the AIL lock.
+		 * xfs_trans_ail_delete() drops the AIL lock.
 		 */
 		if (bip->bli_flags & XFS_BLI_STALE_INODE) {
 			xfs_buf_do_callbacks(bp, (xfs_log_item_t *)bip);
 			XFS_BUF_SET_FSPRIVATE(bp, NULL);
 			XFS_BUF_CLR_IODONE_FUNC(bp);
 		} else {
-			spin_lock(&mp->m_ail_lock);
-			xfs_trans_delete_ail(mp, (xfs_log_item_t *)bip);
+			spin_lock(&ailp->xa_lock);
+			xfs_trans_ail_delete(ailp, (xfs_log_item_t *)bip);
 			xfs_buf_item_relse(bp);
 			ASSERT(XFS_BUF_FSPRIVATE(bp, void *) == NULL);
 		}
@@ -707,8 +707,8 @@ xfs_buf_item_init(
 	 * the first.  If we do already have one, there is
 	 * nothing to do here so return.
 	 */
-	if (XFS_BUF_FSPRIVATE3(bp, xfs_mount_t *) != mp)
-		XFS_BUF_SET_FSPRIVATE3(bp, mp);
+	if (bp->b_mount != mp)
+		bp->b_mount = mp;
 	XFS_BUF_SET_BDSTRAT_FUNC(bp, xfs_bdstrat_cb);
 	if (XFS_BUF_FSPRIVATE(bp, void *) != NULL) {
 		lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
@@ -731,6 +731,7 @@ xfs_buf_item_init(
 	bip->bli_item.li_type = XFS_LI_BUF;
 	bip->bli_item.li_ops = &xfs_buf_item_ops;
 	bip->bli_item.li_mountp = mp;
+	bip->bli_item.li_ailp = mp->m_ail;
 	bip->bli_buf = bp;
 	xfs_buf_hold(bp);
 	bip->bli_format.blf_type = XFS_LI_BUF;
@@ -997,21 +998,7 @@ xfs_buf_iodone_callbacks(
 			xfs_buf_do_callbacks(bp, lip);
 			XFS_BUF_SET_FSPRIVATE(bp, NULL);
 			XFS_BUF_CLR_IODONE_FUNC(bp);
-
-			/*
-			 * XFS_SHUT flag gets set when we go thru the
-			 * entire buffer cache and deliberately start
-			 * throwing away delayed write buffers.
-			 * Since there's no biowait done on those,
-			 * we should just brelse them.
-			 */
-			if (XFS_BUF_ISSHUT(bp)) {
-			    XFS_BUF_UNSHUT(bp);
-				xfs_buf_relse(bp);
-			} else {
-				xfs_biodone(bp);
-			}
-
+			xfs_biodone(bp);
 			return;
 		}
 
@@ -1122,27 +1109,23 @@ xfs_buf_iodone(
 	xfs_buf_t		*bp,
 	xfs_buf_log_item_t	*bip)
 {
-	struct xfs_mount	*mp;
+	struct xfs_ail		*ailp = bip->bli_item.li_ailp;
 
 	ASSERT(bip->bli_buf == bp);
 
 	xfs_buf_rele(bp);
-	mp = bip->bli_item.li_mountp;
 
 	/*
 	 * If we are forcibly shutting down, this may well be
 	 * off the AIL already. That's because we simulate the
 	 * log-committed callbacks to unpin these buffers. Or we may never
 	 * have put this item on AIL because of the transaction was
-	 * aborted forcibly. xfs_trans_delete_ail() takes care of these.
+	 * aborted forcibly. xfs_trans_ail_delete() takes care of these.
 	 *
 	 * Either way, AIL is useless if we're forcing a shutdown.
 	 */
-	spin_lock(&mp->m_ail_lock);
-	/*
-	 * xfs_trans_delete_ail() drops the AIL lock.
-	 */
-	xfs_trans_delete_ail(mp, (xfs_log_item_t *)bip);
+	spin_lock(&ailp->xa_lock);
+	xfs_trans_ail_delete(ailp, (xfs_log_item_t *)bip);
 	xfs_buf_item_free(bip);
 }
 
diff --git a/fs/xfs/xfs_clnt.h b/fs/xfs/xfs_clnt.h
deleted file mode 100644
index d2ce5dd70d8..00000000000
--- a/fs/xfs/xfs_clnt.h
+++ /dev/null
@@ -1,105 +0,0 @@
-/*
- * Copyright (c) 2000-2005 Silicon Graphics, Inc.
- * All Rights Reserved.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it would be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write the Free Software Foundation,
- * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#ifndef __XFS_CLNT_H__
-#define __XFS_CLNT_H__
-
-/*
- * XFS arguments structure, constructed from the arguments we
- * are passed via the mount system call.
- *
- * NOTE: The mount system call is handled differently between
- * Linux and IRIX.  In IRIX we worked work with a binary data
- * structure coming in across the syscall interface from user
- * space (the mount userspace knows about each filesystem type
- * and the set of valid options for it, and converts the users
- * argument string into a binary structure _before_ making the
- * system call), and the ABI issues that this implies.
- *
- * In Linux, we are passed a comma separated set of options;
- * ie. a NULL terminated string of characters.  Userspace mount
- * code does not have any knowledge of mount options expected by
- * each filesystem type and so each filesystem parses its mount
- * options in kernel space.
- *
- * For the Linux port, we kept this structure pretty much intact
- * and use it internally (because the existing code groks it).
- */
-struct xfs_mount_args {
-	int	flags;		/* flags -> see XFSMNT_... macros below */
-	int	flags2;		/* flags -> see XFSMNT2_... macros below */
-	int	logbufs;	/* Number of log buffers, -1 to default */
-	int	logbufsize;	/* Size of log buffers, -1 to default */
-	char	fsname[MAXNAMELEN+1];	/* data device name */
-	char	rtname[MAXNAMELEN+1];	/* realtime device filename */
-	char	logname[MAXNAMELEN+1];	/* journal device filename */
-	char	mtpt[MAXNAMELEN+1];	/* filesystem mount point */
-	int	sunit;		/* stripe unit (BBs) */
-	int	swidth;		/* stripe width (BBs), multiple of sunit */
-	uchar_t iosizelog;	/* log2 of the preferred I/O size */
-	int	ihashsize;	/* inode hash table size (buckets) */
-};
-
-/*
- * XFS mount option flags -- args->flags1
- */
-#define	XFSMNT_ATTR2		0x00000001	/* allow ATTR2 EA format */
-#define	XFSMNT_WSYNC		0x00000002	/* safe mode nfs mount
-						 * compatible */
-#define	XFSMNT_INO64		0x00000004	/* move inode numbers up
-						 * past 2^32 */
-#define XFSMNT_UQUOTA		0x00000008	/* user quota accounting */
-#define XFSMNT_PQUOTA		0x00000010	/* IRIX prj quota accounting */
-#define XFSMNT_UQUOTAENF	0x00000020	/* user quota limit
-						 * enforcement */
-#define XFSMNT_PQUOTAENF	0x00000040	/* IRIX project quota limit
-						 * enforcement */
-#define XFSMNT_QUIET		0x00000080	/* don't report mount errors */
-#define XFSMNT_NOALIGN		0x00000200	/* don't allocate at
-						 * stripe boundaries*/
-#define XFSMNT_RETERR		0x00000400	/* return error to user */
-#define XFSMNT_NORECOVERY	0x00000800	/* no recovery, implies
-						 * read-only mount */
-#define XFSMNT_SHARED		0x00001000	/* shared XFS mount */
-#define XFSMNT_IOSIZE		0x00002000	/* optimize for I/O size */
-#define XFSMNT_OSYNCISOSYNC	0x00004000	/* o_sync is REALLY o_sync */
-						/* (osyncisdsync is default) */
-#define XFSMNT_NOATTR2		0x00008000	/* turn off ATTR2 EA format */
-#define XFSMNT_32BITINODES	0x00200000	/* restrict inodes to 32
-						 * bits of address space */
-#define XFSMNT_GQUOTA		0x00400000	/* group quota accounting */
-#define XFSMNT_GQUOTAENF	0x00800000	/* group quota limit
-						 * enforcement */
-#define XFSMNT_NOUUID		0x01000000	/* Ignore fs uuid */
-#define XFSMNT_DMAPI		0x02000000	/* enable dmapi/xdsm */
-#define XFSMNT_BARRIER		0x04000000	/* use write barriers */
-#define XFSMNT_IKEEP		0x08000000	/* inode cluster delete */
-#define XFSMNT_SWALLOC		0x10000000	/* turn on stripe width
-						 * allocation */
-#define XFSMNT_DIRSYNC		0x40000000	/* sync creat,link,unlink,rename
-						 * symlink,mkdir,rmdir,mknod */
-#define XFSMNT_FLAGS2		0x80000000	/* more flags set in flags2 */
-
-/*
- * XFS mount option flags -- args->flags2
- */
-#define XFSMNT2_COMPAT_IOSIZE	0x00000001	/* don't report large preferred
-						 * I/O size in stat(2) */
-#define XFSMNT2_FILESTREAMS	0x00000002	/* enable the filestreams
-						 * allocator */
-
-#endif	/* __XFS_CLNT_H__ */
diff --git a/fs/xfs/xfs_da_btree.h b/fs/xfs/xfs_da_btree.h
index 8be0b00ede9..70b710c1792 100644
--- a/fs/xfs/xfs_da_btree.h
+++ b/fs/xfs/xfs_da_btree.h
@@ -72,27 +72,7 @@ typedef struct xfs_da_intnode {
 typedef struct xfs_da_node_hdr xfs_da_node_hdr_t;
 typedef struct xfs_da_node_entry xfs_da_node_entry_t;
 
-#define XFS_DA_MAXHASH	((xfs_dahash_t)-1) /* largest valid hash value */
-
 #define	XFS_LBSIZE(mp)	(mp)->m_sb.sb_blocksize
-#define	XFS_LBLOG(mp)	(mp)->m_sb.sb_blocklog
-
-#define	XFS_DA_MAKE_BNOENTRY(mp,bno,entry)	\
-	(((bno) << (mp)->m_dircook_elog) | (entry))
-#define	XFS_DA_MAKE_COOKIE(mp,bno,entry,hash)	\
-	(((xfs_off_t)XFS_DA_MAKE_BNOENTRY(mp, bno, entry) << 32) | (hash))
-#define	XFS_DA_COOKIE_HASH(mp,cookie)		((xfs_dahash_t)cookie)
-#define	XFS_DA_COOKIE_BNO(mp,cookie)		\
-	((((xfs_off_t)(cookie) >> 31) == -1LL ? \
-		(xfs_dablk_t)0 : \
-		(xfs_dablk_t)((xfs_off_t)(cookie) >> \
-				((mp)->m_dircook_elog + 32))))
-#define	XFS_DA_COOKIE_ENTRY(mp,cookie)		\
-	((((xfs_off_t)(cookie) >> 31) == -1LL ?	\
-		(xfs_dablk_t)0 : \
-		(xfs_dablk_t)(((xfs_off_t)(cookie) >> 32) & \
-				((1 << (mp)->m_dircook_elog) - 1))))
-
 
 /*========================================================================
  * Btree searching and modification structure definitions.
@@ -226,9 +206,8 @@ struct xfs_nameops {
 };
 
 
-#ifdef __KERNEL__
 /*========================================================================
- * Function prototypes for the kernel.
+ * Function prototypes.
  *========================================================================*/
 
 /*
@@ -289,6 +268,5 @@ xfs_daddr_t xfs_da_blkno(xfs_dabuf_t *dabuf);
 
 extern struct kmem_zone *xfs_da_state_zone;
 extern struct kmem_zone *xfs_dabuf_zone;
-#endif	/* __KERNEL__ */
 
 #endif	/* __XFS_DA_BTREE_H__ */
diff --git a/fs/xfs/xfs_dfrag.c b/fs/xfs/xfs_dfrag.c
index 75b0cd4da0e..b4c1ee71349 100644
--- a/fs/xfs/xfs_dfrag.c
+++ b/fs/xfs/xfs_dfrag.c
@@ -49,9 +49,8 @@
  */
 int
 xfs_swapext(
-	xfs_swapext_t	__user *sxu)
+	xfs_swapext_t	*sxp)
 {
-	xfs_swapext_t	*sxp;
 	xfs_inode_t     *ip, *tip;
 	struct file	*file, *target_file;
 	int		error = 0;
@@ -62,11 +61,6 @@ xfs_swapext(
 		goto out;
 	}
 
-	if (copy_from_user(sxp, sxu, sizeof(xfs_swapext_t))) {
-		error = XFS_ERROR(EFAULT);
-		goto out_free_sxp;
-	}
-
 	/* Pull information for the target fd */
 	file = fget((int)sxp->sx_fdtarget);
 	if (!file) {
diff --git a/fs/xfs/xfs_dfrag.h b/fs/xfs/xfs_dfrag.h
index da178205be6..4f55a630655 100644
--- a/fs/xfs/xfs_dfrag.h
+++ b/fs/xfs/xfs_dfrag.h
@@ -46,7 +46,7 @@ typedef struct xfs_swapext
 /*
  * Syscall interface for xfs_swapext
  */
-int	xfs_swapext(struct xfs_swapext __user *sx);
+int	xfs_swapext(struct xfs_swapext *sx);
 
 int	xfs_swap_extents(struct xfs_inode *ip, struct xfs_inode *tip,
 		struct xfs_swapext *sxp);
diff --git a/fs/xfs/xfs_dinode.h b/fs/xfs/xfs_dinode.h
index c9065eaf2a4..162e8726df5 100644
--- a/fs/xfs/xfs_dinode.h
+++ b/fs/xfs/xfs_dinode.h
@@ -18,32 +18,29 @@
 #ifndef __XFS_DINODE_H__
 #define	__XFS_DINODE_H__
 
-struct xfs_buf;
-struct xfs_mount;
+#define	XFS_DINODE_MAGIC		0x494e	/* 'IN' */
+#define XFS_DINODE_GOOD_VERSION(v)	(((v) == 1 || (v) == 2))
 
-#define	XFS_DINODE_VERSION_1	1
-#define	XFS_DINODE_VERSION_2	2
-#define XFS_DINODE_GOOD_VERSION(v)	\
-	(((v) == XFS_DINODE_VERSION_1 || (v) == XFS_DINODE_VERSION_2))
-#define	XFS_DINODE_MAGIC	0x494e	/* 'IN' */
-
-/*
- * Disk inode structure.
- * This is just the header; the inode is expanded to fill a variable size
- * with the last field expanding.  It is split into the core and "other"
- * because we only need the core part in the in-core inode.
- */
 typedef struct xfs_timestamp {
 	__be32		t_sec;		/* timestamp seconds */
 	__be32		t_nsec;		/* timestamp nanoseconds */
 } xfs_timestamp_t;
 
 /*
- * Note: Coordinate changes to this structure with the XFS_DI_* #defines
- * below, the offsets table in xfs_ialloc_log_di() and struct xfs_icdinode
- * in xfs_inode.h.
+ * On-disk inode structure.
+ *
+ * This is just the header or "dinode core", the inode is expanded to fill a
+ * variable size the leftover area split into a data and an attribute fork.
+ * The format of the data and attribute fork depends on the format of the
+ * inode as indicated by di_format and di_aformat.  To access the data and
+ * attribute use the XFS_DFORK_PTR, XFS_DFORK_DPTR, and XFS_DFORK_PTR macros
+ * below.
+ *
+ * There is a very similar struct icdinode in xfs_inode which matches the
+ * layout of the first 96 bytes of this structure, but is kept in native
+ * format instead of big endian.
  */
-typedef struct xfs_dinode_core {
+typedef struct xfs_dinode {
 	__be16		di_magic;	/* inode magic # = XFS_DINODE_MAGIC */
 	__be16		di_mode;	/* mode and type of file */
 	__u8		di_version;	/* inode version */
@@ -69,34 +66,12 @@ typedef struct xfs_dinode_core {
 	__be16		di_dmstate;	/* DMIG state info */
 	__be16		di_flags;	/* random flags, XFS_DIFLAG_... */
 	__be32		di_gen;		/* generation number */
-} xfs_dinode_core_t;
 
-#define DI_MAX_FLUSH 0xffff
+	/* di_next_unlinked is the only non-core field in the old dinode */
+	__be32		di_next_unlinked;/* agi unlinked list ptr */
+} __attribute__((packed)) xfs_dinode_t;
 
-typedef struct xfs_dinode
-{
-	xfs_dinode_core_t	di_core;
-	/*
-	 * In adding anything between the core and the union, be
-	 * sure to update the macros like XFS_LITINO below and
-	 * XFS_BMAP_RBLOCK_DSIZE in xfs_bmap_btree.h.
-	 */
-	__be32			di_next_unlinked;/* agi unlinked list ptr */
-	union {
-		xfs_bmdr_block_t di_bmbt;	/* btree root block */
-		xfs_bmbt_rec_32_t di_bmx[1];	/* extent list */
-		xfs_dir2_sf_t	di_dir2sf;	/* shortform directory v2 */
-		char		di_c[1];	/* local contents */
-		__be32		di_dev;		/* device for S_IFCHR/S_IFBLK */
-		uuid_t		di_muuid;	/* mount point value */
-		char		di_symlink[1];	/* local symbolic link */
-	}		di_u;
-	union {
-		xfs_bmdr_block_t di_abmbt;	/* btree root block */
-		xfs_bmbt_rec_32_t di_abmx[1];	/* extent list */
-		xfs_attr_shortform_t di_attrsf;	/* shortform attribute list */
-	}		di_a;
-} xfs_dinode_t;
+#define DI_MAX_FLUSH 0xffff
 
 /*
  * The 32 bit link count in the inode theoretically maxes out at UINT_MAX.
@@ -107,50 +82,14 @@ typedef struct xfs_dinode
 #define	XFS_MAXLINK_1		65535U
 
 /*
- * Bit names for logging disk inodes only
- */
-#define	XFS_DI_MAGIC		0x0000001
-#define	XFS_DI_MODE		0x0000002
-#define	XFS_DI_VERSION		0x0000004
-#define	XFS_DI_FORMAT		0x0000008
-#define	XFS_DI_ONLINK		0x0000010
-#define	XFS_DI_UID		0x0000020
-#define	XFS_DI_GID		0x0000040
-#define	XFS_DI_NLINK		0x0000080
-#define	XFS_DI_PROJID		0x0000100
-#define	XFS_DI_PAD		0x0000200
-#define	XFS_DI_ATIME		0x0000400
-#define	XFS_DI_MTIME		0x0000800
-#define	XFS_DI_CTIME		0x0001000
-#define	XFS_DI_SIZE		0x0002000
-#define	XFS_DI_NBLOCKS		0x0004000
-#define	XFS_DI_EXTSIZE		0x0008000
-#define	XFS_DI_NEXTENTS		0x0010000
-#define	XFS_DI_NAEXTENTS	0x0020000
-#define	XFS_DI_FORKOFF		0x0040000
-#define	XFS_DI_AFORMAT		0x0080000
-#define	XFS_DI_DMEVMASK		0x0100000
-#define	XFS_DI_DMSTATE		0x0200000
-#define	XFS_DI_FLAGS		0x0400000
-#define	XFS_DI_GEN		0x0800000
-#define	XFS_DI_NEXT_UNLINKED	0x1000000
-#define	XFS_DI_U		0x2000000
-#define	XFS_DI_A		0x4000000
-#define	XFS_DI_NUM_BITS		27
-#define	XFS_DI_ALL_BITS		((1 << XFS_DI_NUM_BITS) - 1)
-#define	XFS_DI_CORE_BITS	(XFS_DI_ALL_BITS & ~(XFS_DI_U|XFS_DI_A))
-
-/*
  * Values for di_format
  */
-typedef enum xfs_dinode_fmt
-{
-	XFS_DINODE_FMT_DEV,		/* CHR, BLK: di_dev */
-	XFS_DINODE_FMT_LOCAL,		/* DIR, REG: di_c */
-					/* LNK: di_symlink */
-	XFS_DINODE_FMT_EXTENTS,		/* DIR, REG, LNK: di_bmx */
-	XFS_DINODE_FMT_BTREE,		/* DIR, REG, LNK: di_bmbt */
-	XFS_DINODE_FMT_UUID		/* MNT: di_uuid */
+typedef enum xfs_dinode_fmt {
+	XFS_DINODE_FMT_DEV,		/* xfs_dev_t */
+	XFS_DINODE_FMT_LOCAL,		/* bulk data */
+	XFS_DINODE_FMT_EXTENTS,		/* struct xfs_bmbt_rec */
+	XFS_DINODE_FMT_BTREE,		/* struct xfs_bmdr_block */
+	XFS_DINODE_FMT_UUID		/* uuid_t */
 } xfs_dinode_fmt_t;
 
 /*
@@ -166,13 +105,13 @@ typedef enum xfs_dinode_fmt
  */
 #define	XFS_LITINO(mp)	((mp)->m_litino)
 #define	XFS_BROOT_SIZE_ADJ	\
-	(sizeof(xfs_bmbt_block_t) - sizeof(xfs_bmdr_block_t))
+	(XFS_BTREE_LBLOCK_LEN - sizeof(xfs_bmdr_block_t))
 
 /*
  * Inode data & attribute fork sizes, per inode.
  */
-#define XFS_DFORK_Q(dip)		((dip)->di_core.di_forkoff != 0)
-#define XFS_DFORK_BOFF(dip)		((int)((dip)->di_core.di_forkoff << 3))
+#define XFS_DFORK_Q(dip)		((dip)->di_forkoff != 0)
+#define XFS_DFORK_BOFF(dip)		((int)((dip)->di_forkoff << 3))
 
 #define XFS_DFORK_DSIZE(dip,mp) \
 	(XFS_DFORK_Q(dip) ? \
@@ -187,23 +126,42 @@ typedef enum xfs_dinode_fmt
 		XFS_DFORK_DSIZE(dip, mp) : \
 		XFS_DFORK_ASIZE(dip, mp))
 
-#define XFS_DFORK_DPTR(dip)		    ((dip)->di_u.di_c)
+/*
+ * Return pointers to the data or attribute forks.
+ */
+#define XFS_DFORK_DPTR(dip) \
+	((char *)(dip) + sizeof(struct xfs_dinode))
 #define XFS_DFORK_APTR(dip)	\
-	((dip)->di_u.di_c + XFS_DFORK_BOFF(dip))
+	(XFS_DFORK_DPTR(dip) + XFS_DFORK_BOFF(dip))
 #define XFS_DFORK_PTR(dip,w)	\
 	((w) == XFS_DATA_FORK ? XFS_DFORK_DPTR(dip) : XFS_DFORK_APTR(dip))
+
 #define XFS_DFORK_FORMAT(dip,w) \
 	((w) == XFS_DATA_FORK ? \
-		(dip)->di_core.di_format : \
-		(dip)->di_core.di_aformat)
+		(dip)->di_format : \
+		(dip)->di_aformat)
 #define XFS_DFORK_NEXTENTS(dip,w) \
 	((w) == XFS_DATA_FORK ? \
-	 	be32_to_cpu((dip)->di_core.di_nextents) : \
-	 	be16_to_cpu((dip)->di_core.di_anextents))
+		be32_to_cpu((dip)->di_nextents) : \
+		be16_to_cpu((dip)->di_anextents))
 
 #define	XFS_BUF_TO_DINODE(bp)	((xfs_dinode_t *)XFS_BUF_PTR(bp))
 
 /*
+ * For block and character special files the 32bit dev_t is stored at the
+ * beginning of the data fork.
+ */
+static inline xfs_dev_t xfs_dinode_get_rdev(struct xfs_dinode *dip)
+{
+	return be32_to_cpu(*(__be32 *)XFS_DFORK_DPTR(dip));
+}
+
+static inline void xfs_dinode_put_rdev(struct xfs_dinode *dip, xfs_dev_t rdev)
+{
+	*(__be32 *)XFS_DFORK_DPTR(dip) = cpu_to_be32(rdev);
+}
+
+/*
  * Values for di_flags
  * There should be a one-to-one correspondence between these flags and the
  * XFS_XFLAG_s.
diff --git a/fs/xfs/xfs_dir2_sf.h b/fs/xfs/xfs_dir2_sf.h
index deecc9d238f..6ac44b550d3 100644
--- a/fs/xfs/xfs_dir2_sf.h
+++ b/fs/xfs/xfs_dir2_sf.h
@@ -34,13 +34,6 @@ struct xfs_mount;
 struct xfs_trans;
 
 /*
- * Maximum size of a shortform directory.
- */
-#define	XFS_DIR2_SF_MAX_SIZE	\
-	(XFS_DINODE_MAX_SIZE - (uint)sizeof(xfs_dinode_core_t) - \
-	 (uint)sizeof(xfs_agino_t))
-
-/*
  * Inode number stored as 8 8-bit values.
  */
 typedef	struct { __uint8_t i[8]; } xfs_dir2_ino8_t;
diff --git a/fs/xfs/xfs_dmops.c b/fs/xfs/xfs_dmops.c
index a1e55fb9d5d..e71e2581c0c 100644
--- a/fs/xfs/xfs_dmops.c
+++ b/fs/xfs/xfs_dmops.c
@@ -25,7 +25,6 @@
 #include "xfs_inum.h"
 #include "xfs_ag.h"
 #include "xfs_mount.h"
-#include "xfs_clnt.h"
 
 
 static struct xfs_dmops xfs_dmcore_stub = {
@@ -38,9 +37,9 @@ static struct xfs_dmops xfs_dmcore_stub = {
 };
 
 int
-xfs_dmops_get(struct xfs_mount *mp, struct xfs_mount_args *args)
+xfs_dmops_get(struct xfs_mount *mp)
 {
-	if (args->flags & XFSMNT_DMAPI) {
+	if (mp->m_flags & XFS_MOUNT_DMAPI) {
 		cmn_err(CE_WARN,
 			"XFS: dmapi support not available in this kernel.");
 		return EINVAL;
diff --git a/fs/xfs/xfs_error.c b/fs/xfs/xfs_error.c
index f227ecd1a29..92d5cd5bf4f 100644
--- a/fs/xfs/xfs_error.c
+++ b/fs/xfs/xfs_error.c
@@ -153,21 +153,6 @@ xfs_errortag_clearall(xfs_mount_t *mp, int loud)
 }
 #endif /* DEBUG */
 
-static void
-xfs_fs_vcmn_err(int level, xfs_mount_t *mp, char *fmt, va_list ap)
-{
-	if (mp != NULL) {
-		char	*newfmt;
-		int	len = 16 + mp->m_fsname_len + strlen(fmt);
-
-		newfmt = kmem_alloc(len, KM_SLEEP);
-		sprintf(newfmt, "Filesystem \"%s\": %s", mp->m_fsname, fmt);
-		icmn_err(level, newfmt, ap);
-		kmem_free(newfmt);
-	} else {
-		icmn_err(level, fmt, ap);
-	}
-}
 
 void
 xfs_fs_cmn_err(int level, xfs_mount_t *mp, char *fmt, ...)
diff --git a/fs/xfs/xfs_error.h b/fs/xfs/xfs_error.h
index 11543f10b0c..0c93051c465 100644
--- a/fs/xfs/xfs_error.h
+++ b/fs/xfs/xfs_error.h
@@ -159,11 +159,15 @@ extern int xfs_errortag_clearall(xfs_mount_t *mp, int loud);
 #define		XFS_PTAG_FSBLOCK_ZERO		0x00000080
 
 struct xfs_mount;
-/* PRINTFLIKE4 */
+
+extern void xfs_fs_vcmn_err(int level, struct xfs_mount *mp,
+		char *fmt, va_list ap)
+	__attribute__ ((format (printf, 3, 0)));
 extern void xfs_cmn_err(int panic_tag, int level, struct xfs_mount *mp,
-			char *fmt, ...);
-/* PRINTFLIKE3 */
-extern void xfs_fs_cmn_err(int level, struct xfs_mount *mp, char *fmt, ...);
+			char *fmt, ...)
+	__attribute__ ((format (printf, 4, 5)));
+extern void xfs_fs_cmn_err(int level, struct xfs_mount *mp, char *fmt, ...)
+	__attribute__ ((format (printf, 3, 4)));
 
 extern void xfs_hex_dump(void *p, int length);
 
diff --git a/fs/xfs/xfs_extfree_item.c b/fs/xfs/xfs_extfree_item.c
index 8aa28f751b2..05a4bdd4be3 100644
--- a/fs/xfs/xfs_extfree_item.c
+++ b/fs/xfs/xfs_extfree_item.c
@@ -108,19 +108,16 @@ xfs_efi_item_pin(xfs_efi_log_item_t *efip)
 STATIC void
 xfs_efi_item_unpin(xfs_efi_log_item_t *efip, int stale)
 {
-	xfs_mount_t	*mp;
+	struct xfs_ail		*ailp = efip->efi_item.li_ailp;
 
-	mp = efip->efi_item.li_mountp;
-	spin_lock(&mp->m_ail_lock);
+	spin_lock(&ailp->xa_lock);
 	if (efip->efi_flags & XFS_EFI_CANCELED) {
-		/*
-		 * xfs_trans_delete_ail() drops the AIL lock.
-		 */
-		xfs_trans_delete_ail(mp, (xfs_log_item_t *)efip);
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		xfs_trans_ail_delete(ailp, (xfs_log_item_t *)efip);
 		xfs_efi_item_free(efip);
 	} else {
 		efip->efi_flags |= XFS_EFI_COMMITTED;
-		spin_unlock(&mp->m_ail_lock);
+		spin_unlock(&ailp->xa_lock);
 	}
 }
 
@@ -134,26 +131,23 @@ xfs_efi_item_unpin(xfs_efi_log_item_t *efip, int stale)
 STATIC void
 xfs_efi_item_unpin_remove(xfs_efi_log_item_t *efip, xfs_trans_t *tp)
 {
-	xfs_mount_t	*mp;
+	struct xfs_ail		*ailp = efip->efi_item.li_ailp;
 	xfs_log_item_desc_t	*lidp;
 
-	mp = efip->efi_item.li_mountp;
-	spin_lock(&mp->m_ail_lock);
+	spin_lock(&ailp->xa_lock);
 	if (efip->efi_flags & XFS_EFI_CANCELED) {
 		/*
 		 * free the xaction descriptor pointing to this item
 		 */
 		lidp = xfs_trans_find_item(tp, (xfs_log_item_t *) efip);
 		xfs_trans_free_item(tp, lidp);
-		/*
-		 * pull the item off the AIL.
-		 * xfs_trans_delete_ail() drops the AIL lock.
-		 */
-		xfs_trans_delete_ail(mp, (xfs_log_item_t *)efip);
+
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		xfs_trans_ail_delete(ailp, (xfs_log_item_t *)efip);
 		xfs_efi_item_free(efip);
 	} else {
 		efip->efi_flags |= XFS_EFI_COMMITTED;
-		spin_unlock(&mp->m_ail_lock);
+		spin_unlock(&ailp->xa_lock);
 	}
 }
 
@@ -268,6 +262,7 @@ xfs_efi_init(xfs_mount_t	*mp,
 	efip->efi_item.li_type = XFS_LI_EFI;
 	efip->efi_item.li_ops = &xfs_efi_item_ops;
 	efip->efi_item.li_mountp = mp;
+	efip->efi_item.li_ailp = mp->m_ail;
 	efip->efi_format.efi_nextents = nextents;
 	efip->efi_format.efi_id = (__psint_t)(void*)efip;
 
@@ -345,25 +340,22 @@ void
 xfs_efi_release(xfs_efi_log_item_t	*efip,
 		uint			nextents)
 {
-	xfs_mount_t	*mp;
-	int		extents_left;
+	struct xfs_ail		*ailp = efip->efi_item.li_ailp;
+	int			extents_left;
 
-	mp = efip->efi_item.li_mountp;
 	ASSERT(efip->efi_next_extent > 0);
 	ASSERT(efip->efi_flags & XFS_EFI_COMMITTED);
 
-	spin_lock(&mp->m_ail_lock);
+	spin_lock(&ailp->xa_lock);
 	ASSERT(efip->efi_next_extent >= nextents);
 	efip->efi_next_extent -= nextents;
 	extents_left = efip->efi_next_extent;
 	if (extents_left == 0) {
-		/*
-		 * xfs_trans_delete_ail() drops the AIL lock.
-		 */
-		xfs_trans_delete_ail(mp, (xfs_log_item_t *)efip);
+		/* xfs_trans_ail_delete() drops the AIL lock. */
+		xfs_trans_ail_delete(ailp, (xfs_log_item_t *)efip);
 		xfs_efi_item_free(efip);
 	} else {
-		spin_unlock(&mp->m_ail_lock);
+		spin_unlock(&ailp->xa_lock);
 	}
 }
 
@@ -565,6 +557,7 @@ xfs_efd_init(xfs_mount_t	*mp,
 	efdp->efd_item.li_type = XFS_LI_EFD;
 	efdp->efd_item.li_ops = &xfs_efd_item_ops;
 	efdp->efd_item.li_mountp = mp;
+	efdp->efd_item.li_ailp = mp->m_ail;
 	efdp->efd_efip = efip;
 	efdp->efd_format.efd_nextents = nextents;
 	efdp->efd_format.efd_efi_id = efip->efi_format.efi_id;
diff --git a/fs/xfs/xfs_fs.h b/fs/xfs/xfs_fs.h
index 01c0cc88d3f..589c41c3844 100644
--- a/fs/xfs/xfs_fs.h
+++ b/fs/xfs/xfs_fs.h
@@ -113,22 +113,14 @@ struct getbmapx {
 #define BMV_IF_ATTRFORK		0x1	/* return attr fork rather than data */
 #define BMV_IF_NO_DMAPI_READ	0x2	/* Do not generate DMAPI read event  */
 #define BMV_IF_PREALLOC		0x4	/* rtn status BMV_OF_PREALLOC if req */
-#define BMV_IF_VALID	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC)
-#ifdef __KERNEL__
-#define BMV_IF_EXTENDED 0x40000000	/* getpmapx if set */
-#endif
+#define BMV_IF_DELALLOC		0x8	/* rtn status BMV_OF_DELALLOC if req */
+#define BMV_IF_VALID	\
+	(BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|BMV_IF_DELALLOC)
 
 /*	bmv_oflags values - returned for for each non-header segment */
 #define BMV_OF_PREALLOC		0x1	/* segment = unwritten pre-allocation */
-
-/*	Convert getbmap <-> getbmapx - move fields from p1 to p2. */
-#define GETBMAP_CONVERT(p1,p2) {	\
-	p2.bmv_offset = p1.bmv_offset;	\
-	p2.bmv_block = p1.bmv_block;	\
-	p2.bmv_length = p1.bmv_length;	\
-	p2.bmv_count = p1.bmv_count;	\
-	p2.bmv_entries = p1.bmv_entries;  }
-
+#define BMV_OF_DELALLOC		0x2	/* segment = delayed allocation */
+#define BMV_OF_LAST		0x4	/* segment is the last in the file */
 
 /*
  * Structure for XFS_IOC_FSSETDM.
@@ -426,10 +418,6 @@ typedef struct xfs_handle {
 #define XFS_IOC_GETXFLAGS	FS_IOC_GETFLAGS
 #define XFS_IOC_SETXFLAGS	FS_IOC_SETFLAGS
 #define XFS_IOC_GETVERSION	FS_IOC_GETVERSION
-/* 32-bit compat counterparts */
-#define XFS_IOC32_GETXFLAGS	FS_IOC32_GETFLAGS
-#define XFS_IOC32_SETXFLAGS	FS_IOC32_SETFLAGS
-#define XFS_IOC32_GETVERSION	FS_IOC32_GETVERSION
 
 /*
  * ioctl commands that replace IRIX fcntl()'s
diff --git a/fs/xfs/xfs_fsops.c b/fs/xfs/xfs_fsops.c
index 84583cf73db..852b6d32e8d 100644
--- a/fs/xfs/xfs_fsops.c
+++ b/fs/xfs/xfs_fsops.c
@@ -126,7 +126,7 @@ xfs_growfs_data_private(
 	xfs_extlen_t		agsize;
 	xfs_extlen_t		tmpsize;
 	xfs_alloc_rec_t		*arec;
-	xfs_btree_sblock_t	*block;
+	struct xfs_btree_block	*block;
 	xfs_buf_t		*bp;
 	int			bucket;
 	int			dpct;
@@ -251,14 +251,14 @@ xfs_growfs_data_private(
 		bp = xfs_buf_get(mp->m_ddev_targp,
 			XFS_AGB_TO_DADDR(mp, agno, XFS_BNO_BLOCK(mp)),
 			BTOBB(mp->m_sb.sb_blocksize), 0);
-		block = XFS_BUF_TO_SBLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 		memset(block, 0, mp->m_sb.sb_blocksize);
 		block->bb_magic = cpu_to_be32(XFS_ABTB_MAGIC);
 		block->bb_level = 0;
 		block->bb_numrecs = cpu_to_be16(1);
-		block->bb_leftsib = cpu_to_be32(NULLAGBLOCK);
-		block->bb_rightsib = cpu_to_be32(NULLAGBLOCK);
-		arec = XFS_BTREE_REC_ADDR(xfs_alloc, block, 1);
+		block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		arec = XFS_ALLOC_REC_ADDR(mp, block, 1);
 		arec->ar_startblock = cpu_to_be32(XFS_PREALLOC_BLOCKS(mp));
 		arec->ar_blockcount = cpu_to_be32(
 			agsize - be32_to_cpu(arec->ar_startblock));
@@ -272,14 +272,14 @@ xfs_growfs_data_private(
 		bp = xfs_buf_get(mp->m_ddev_targp,
 			XFS_AGB_TO_DADDR(mp, agno, XFS_CNT_BLOCK(mp)),
 			BTOBB(mp->m_sb.sb_blocksize), 0);
-		block = XFS_BUF_TO_SBLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 		memset(block, 0, mp->m_sb.sb_blocksize);
 		block->bb_magic = cpu_to_be32(XFS_ABTC_MAGIC);
 		block->bb_level = 0;
 		block->bb_numrecs = cpu_to_be16(1);
-		block->bb_leftsib = cpu_to_be32(NULLAGBLOCK);
-		block->bb_rightsib = cpu_to_be32(NULLAGBLOCK);
-		arec = XFS_BTREE_REC_ADDR(xfs_alloc, block, 1);
+		block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		arec = XFS_ALLOC_REC_ADDR(mp, block, 1);
 		arec->ar_startblock = cpu_to_be32(XFS_PREALLOC_BLOCKS(mp));
 		arec->ar_blockcount = cpu_to_be32(
 			agsize - be32_to_cpu(arec->ar_startblock));
@@ -294,13 +294,13 @@ xfs_growfs_data_private(
 		bp = xfs_buf_get(mp->m_ddev_targp,
 			XFS_AGB_TO_DADDR(mp, agno, XFS_IBT_BLOCK(mp)),
 			BTOBB(mp->m_sb.sb_blocksize), 0);
-		block = XFS_BUF_TO_SBLOCK(bp);
+		block = XFS_BUF_TO_BLOCK(bp);
 		memset(block, 0, mp->m_sb.sb_blocksize);
 		block->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
 		block->bb_level = 0;
 		block->bb_numrecs = 0;
-		block->bb_leftsib = cpu_to_be32(NULLAGBLOCK);
-		block->bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+		block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
 		error = xfs_bwrite(mp, bp);
 		if (error) {
 			goto error0;
@@ -435,6 +435,9 @@ xfs_growfs_data(
 	xfs_growfs_data_t	*in)
 {
 	int error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
 	if (!mutex_trylock(&mp->m_growlock))
 		return XFS_ERROR(EWOULDBLOCK);
 	error = xfs_growfs_data_private(mp, in);
@@ -448,6 +451,9 @@ xfs_growfs_log(
 	xfs_growfs_log_t	*in)
 {
 	int error;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
 	if (!mutex_trylock(&mp->m_growlock))
 		return XFS_ERROR(EWOULDBLOCK);
 	error = xfs_growfs_log_private(mp, in);
diff --git a/fs/xfs/xfs_ialloc.c b/fs/xfs/xfs_ialloc.c
index aad8c5da38a..e6ebbaeb4dc 100644
--- a/fs/xfs/xfs_ialloc.c
+++ b/fs/xfs/xfs_ialloc.c
@@ -41,68 +41,6 @@
 #include "xfs_error.h"
 #include "xfs_bmap.h"
 
-/*
- * Log specified fields for the inode given by bp and off.
- */
-STATIC void
-xfs_ialloc_log_di(
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_buf_t	*bp,		/* inode buffer */
-	int		off,		/* index of inode in buffer */
-	int		fields)		/* bitmask of fields to log */
-{
-	int			first;		/* first byte number */
-	int			ioffset;	/* off in bytes */
-	int			last;		/* last byte number */
-	xfs_mount_t		*mp;		/* mount point structure */
-	static const short	offsets[] = {	/* field offsets */
-						/* keep in sync with bits */
-		offsetof(xfs_dinode_core_t, di_magic),
-		offsetof(xfs_dinode_core_t, di_mode),
-		offsetof(xfs_dinode_core_t, di_version),
-		offsetof(xfs_dinode_core_t, di_format),
-		offsetof(xfs_dinode_core_t, di_onlink),
-		offsetof(xfs_dinode_core_t, di_uid),
-		offsetof(xfs_dinode_core_t, di_gid),
-		offsetof(xfs_dinode_core_t, di_nlink),
-		offsetof(xfs_dinode_core_t, di_projid),
-		offsetof(xfs_dinode_core_t, di_pad),
-		offsetof(xfs_dinode_core_t, di_atime),
-		offsetof(xfs_dinode_core_t, di_mtime),
-		offsetof(xfs_dinode_core_t, di_ctime),
-		offsetof(xfs_dinode_core_t, di_size),
-		offsetof(xfs_dinode_core_t, di_nblocks),
-		offsetof(xfs_dinode_core_t, di_extsize),
-		offsetof(xfs_dinode_core_t, di_nextents),
-		offsetof(xfs_dinode_core_t, di_anextents),
-		offsetof(xfs_dinode_core_t, di_forkoff),
-		offsetof(xfs_dinode_core_t, di_aformat),
-		offsetof(xfs_dinode_core_t, di_dmevmask),
-		offsetof(xfs_dinode_core_t, di_dmstate),
-		offsetof(xfs_dinode_core_t, di_flags),
-		offsetof(xfs_dinode_core_t, di_gen),
-		offsetof(xfs_dinode_t, di_next_unlinked),
-		offsetof(xfs_dinode_t, di_u),
-		offsetof(xfs_dinode_t, di_a),
-		sizeof(xfs_dinode_t)
-	};
-
-
-	ASSERT(offsetof(xfs_dinode_t, di_core) == 0);
-	ASSERT((fields & (XFS_DI_U|XFS_DI_A)) == 0);
-	mp = tp->t_mountp;
-	/*
-	 * Get the inode-relative first and last bytes for these fields
-	 */
-	xfs_btree_offsets(fields, offsets, XFS_DI_NUM_BITS, &first, &last);
-	/*
-	 * Convert to buffer offsets and log it.
-	 */
-	ioffset = off << mp->m_sb.sb_inodelog;
-	first += ioffset;
-	last += ioffset;
-	xfs_trans_log_buf(tp, bp, first, last);
-}
 
 /*
  * Allocation group level functions.
@@ -119,6 +57,102 @@ xfs_ialloc_cluster_alignment(
 }
 
 /*
+ * Lookup the record equal to ino in the btree given by cur.
+ */
+STATIC int				/* error */
+xfs_inobt_lookup_eq(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		ino,	/* starting inode of chunk */
+	__int32_t		fcnt,	/* free inode count */
+	xfs_inofree_t		free,	/* free inode mask */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.i.ir_startino = ino;
+	cur->bc_rec.i.ir_freecount = fcnt;
+	cur->bc_rec.i.ir_free = free;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
+/*
+ * Lookup the first record greater than or equal to ino
+ * in the btree given by cur.
+ */
+int					/* error */
+xfs_inobt_lookup_ge(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		ino,	/* starting inode of chunk */
+	__int32_t		fcnt,	/* free inode count */
+	xfs_inofree_t		free,	/* free inode mask */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.i.ir_startino = ino;
+	cur->bc_rec.i.ir_freecount = fcnt;
+	cur->bc_rec.i.ir_free = free;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
+}
+
+/*
+ * Lookup the first record less than or equal to ino
+ * in the btree given by cur.
+ */
+int					/* error */
+xfs_inobt_lookup_le(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		ino,	/* starting inode of chunk */
+	__int32_t		fcnt,	/* free inode count */
+	xfs_inofree_t		free,	/* free inode mask */
+	int			*stat)	/* success/failure */
+{
+	cur->bc_rec.i.ir_startino = ino;
+	cur->bc_rec.i.ir_freecount = fcnt;
+	cur->bc_rec.i.ir_free = free;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_LE, stat);
+}
+
+/*
+ * Update the record referred to by cur to the value given
+ * by [ino, fcnt, free].
+ * This either works (return 0) or gets an EFSCORRUPTED error.
+ */
+STATIC int				/* error */
+xfs_inobt_update(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		ino,	/* starting inode of chunk */
+	__int32_t		fcnt,	/* free inode count */
+	xfs_inofree_t		free)	/* free inode mask */
+{
+	union xfs_btree_rec	rec;
+
+	rec.inobt.ir_startino = cpu_to_be32(ino);
+	rec.inobt.ir_freecount = cpu_to_be32(fcnt);
+	rec.inobt.ir_free = cpu_to_be64(free);
+	return xfs_btree_update(cur, &rec);
+}
+
+/*
+ * Get the data from the pointed-to record.
+ */
+int					/* error */
+xfs_inobt_get_rec(
+	struct xfs_btree_cur	*cur,	/* btree cursor */
+	xfs_agino_t		*ino,	/* output: starting inode of chunk */
+	__int32_t		*fcnt,	/* output: number of free inodes */
+	xfs_inofree_t		*free,	/* output: free inode mask */
+	int			*stat)	/* output: success/failure */
+{
+	union xfs_btree_rec	*rec;
+	int			error;
+
+	error = xfs_btree_get_rec(cur, &rec, stat);
+	if (!error && *stat == 1) {
+		*ino = be32_to_cpu(rec->inobt.ir_startino);
+		*fcnt = be32_to_cpu(rec->inobt.ir_freecount);
+		*free = be64_to_cpu(rec->inobt.ir_free);
+	}
+	return error;
+}
+
+/*
  * Allocate new inodes in the allocation group specified by agbp.
  * Return 0 for success, else error code.
  */
@@ -287,9 +321,9 @@ xfs_ialloc_ag_alloc(
 	 * able to use the file system.
 	 */
 	if (xfs_sb_version_hasnlink(&args.mp->m_sb))
-		version = XFS_DINODE_VERSION_2;
+		version = 2;
 	else
-		version = XFS_DINODE_VERSION_1;
+		version = 1;
 
 	/*
 	 * Seed the new inode cluster with a random generation number. This
@@ -310,18 +344,25 @@ xfs_ialloc_ag_alloc(
 					 XFS_BUF_LOCK);
 		ASSERT(fbuf);
 		ASSERT(!XFS_BUF_GETERROR(fbuf));
+
 		/*
-		 * Set initial values for the inodes in this buffer.
+		 * Initialize all inodes in this buffer and then log them.
+		 *
+		 * XXX: It would be much better if we had just one transaction to
+		 *      log a whole cluster of inodes instead of all the indivdual
+		 *      transactions causing a lot of log traffic.
 		 */
 		xfs_biozero(fbuf, 0, ninodes << args.mp->m_sb.sb_inodelog);
 		for (i = 0; i < ninodes; i++) {
+			int	ioffset = i << args.mp->m_sb.sb_inodelog;
+			uint	isize = sizeof(struct xfs_dinode);
+
 			free = XFS_MAKE_IPTR(args.mp, fbuf, i);
-			free->di_core.di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
-			free->di_core.di_version = version;
-			free->di_core.di_gen = cpu_to_be32(gen);
+			free->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
+			free->di_version = version;
+			free->di_gen = cpu_to_be32(gen);
 			free->di_next_unlinked = cpu_to_be32(NULLAGINO);
-			xfs_ialloc_log_di(tp, fbuf, i,
-				XFS_DI_CORE_BITS | XFS_DI_NEXT_UNLINKED);
+			xfs_trans_log_buf(tp, fbuf, ioffset, ioffset + isize - 1);
 		}
 		xfs_trans_inode_alloc_buf(tp, fbuf);
 	}
@@ -335,8 +376,7 @@ xfs_ialloc_ag_alloc(
 	/*
 	 * Insert records describing the new inode chunk into the btree.
 	 */
-	cur = xfs_btree_init_cursor(args.mp, tp, agbp, agno,
-			XFS_BTNUM_INO, (xfs_inode_t *)0, 0);
+	cur = xfs_inobt_init_cursor(args.mp, tp, agbp, agno);
 	for (thisino = newino;
 	     thisino < newino + newlen;
 	     thisino += XFS_INODES_PER_CHUNK) {
@@ -346,7 +386,7 @@ xfs_ialloc_ag_alloc(
 			return error;
 		}
 		ASSERT(i == 0);
-		if ((error = xfs_inobt_insert(cur, &i))) {
+		if ((error = xfs_btree_insert(cur, &i))) {
 			xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 			return error;
 		}
@@ -676,8 +716,7 @@ nextag:
 	 */
 	agno = tagno;
 	*IO_agbp = NULL;
-	cur = xfs_btree_init_cursor(mp, tp, agbp, be32_to_cpu(agi->agi_seqno),
-				    XFS_BTNUM_INO, (xfs_inode_t *)0, 0);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, be32_to_cpu(agi->agi_seqno));
 	/*
 	 * If pagino is 0 (this is the root inode allocation) use newino.
 	 * This must work because we've just allocated some.
@@ -697,7 +736,7 @@ nextag:
 				goto error0;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 			freecount += rec.ir_freecount;
-			if ((error = xfs_inobt_increment(cur, 0, &i)))
+			if ((error = xfs_btree_increment(cur, 0, &i)))
 				goto error0;
 		} while (i == 1);
 
@@ -741,7 +780,7 @@ nextag:
 			/*
 			 * Search left with tcur, back up 1 record.
 			 */
-			if ((error = xfs_inobt_decrement(tcur, 0, &i)))
+			if ((error = xfs_btree_decrement(tcur, 0, &i)))
 				goto error1;
 			doneleft = !i;
 			if (!doneleft) {
@@ -755,7 +794,7 @@ nextag:
 			/*
 			 * Search right with cur, go forward 1 record.
 			 */
-			if ((error = xfs_inobt_increment(cur, 0, &i)))
+			if ((error = xfs_btree_increment(cur, 0, &i)))
 				goto error1;
 			doneright = !i;
 			if (!doneright) {
@@ -817,7 +856,7 @@ nextag:
 				 * further left.
 				 */
 				if (useleft) {
-					if ((error = xfs_inobt_decrement(tcur, 0,
+					if ((error = xfs_btree_decrement(tcur, 0,
 							&i)))
 						goto error1;
 					doneleft = !i;
@@ -837,7 +876,7 @@ nextag:
 				 * further right.
 				 */
 				else {
-					if ((error = xfs_inobt_increment(cur, 0,
+					if ((error = xfs_btree_increment(cur, 0,
 							&i)))
 						goto error1;
 					doneright = !i;
@@ -892,7 +931,7 @@ nextag:
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 				if (rec.ir_freecount > 0)
 					break;
-				if ((error = xfs_inobt_increment(cur, 0, &i)))
+				if ((error = xfs_btree_increment(cur, 0, &i)))
 					goto error0;
 				XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 			}
@@ -926,7 +965,7 @@ nextag:
 				goto error0;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
 			freecount += rec.ir_freecount;
-			if ((error = xfs_inobt_increment(cur, 0, &i)))
+			if ((error = xfs_btree_increment(cur, 0, &i)))
 				goto error0;
 		} while (i == 1);
 		ASSERT(freecount == be32_to_cpu(agi->agi_freecount) ||
@@ -1022,8 +1061,7 @@ xfs_difree(
 	/*
 	 * Initialize the cursor.
 	 */
-	cur = xfs_btree_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO,
-		(xfs_inode_t *)0, 0);
+	cur = xfs_inobt_init_cursor(mp, tp, agbp, agno);
 #ifdef DEBUG
 	if (cur->bc_nlevels == 1) {
 		int freecount = 0;
@@ -1036,7 +1074,7 @@ xfs_difree(
 				goto error0;
 			if (i) {
 				freecount += rec.ir_freecount;
-				if ((error = xfs_inobt_increment(cur, 0, &i)))
+				if ((error = xfs_btree_increment(cur, 0, &i)))
 					goto error0;
 			}
 		} while (i == 1);
@@ -1098,8 +1136,8 @@ xfs_difree(
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_ICOUNT, -ilen);
 		xfs_trans_mod_sb(tp, XFS_TRANS_SB_IFREE, -(ilen - 1));
 
-		if ((error = xfs_inobt_delete(cur, &i))) {
-			cmn_err(CE_WARN, "xfs_difree: xfs_inobt_delete returned an error %d on %s.\n",
+		if ((error = xfs_btree_delete(cur, &i))) {
+			cmn_err(CE_WARN, "xfs_difree: xfs_btree_delete returned an error %d on %s.\n",
 				error, mp->m_fsname);
 			goto error0;
 		}
@@ -1141,7 +1179,7 @@ xfs_difree(
 				goto error0;
 			if (i) {
 				freecount += rec.ir_freecount;
-				if ((error = xfs_inobt_increment(cur, 0, &i)))
+				if ((error = xfs_btree_increment(cur, 0, &i)))
 					goto error0;
 			}
 		} while (i == 1);
@@ -1158,36 +1196,28 @@ error0:
 }
 
 /*
- * Return the location of the inode in bno/off, for mapping it into a buffer.
+ * Return the location of the inode in imap, for mapping it into a buffer.
  */
-/*ARGSUSED*/
 int
-xfs_dilocate(
-	xfs_mount_t	*mp,	/* file system mount structure */
-	xfs_trans_t	*tp,	/* transaction pointer */
+xfs_imap(
+	xfs_mount_t	 *mp,	/* file system mount structure */
+	xfs_trans_t	 *tp,	/* transaction pointer */
 	xfs_ino_t	ino,	/* inode to locate */
-	xfs_fsblock_t	*bno,	/* output: block containing inode */
-	int		*len,	/* output: num blocks in inode cluster */
-	int		*off,	/* output: index in block of inode */
-	uint		flags)	/* flags concerning inode lookup */
+	struct xfs_imap	*imap,	/* location map structure */
+	uint		flags)	/* flags for inode btree lookup */
 {
 	xfs_agblock_t	agbno;	/* block number of inode in the alloc group */
-	xfs_buf_t	*agbp;	/* agi buffer */
 	xfs_agino_t	agino;	/* inode number within alloc group */
 	xfs_agnumber_t	agno;	/* allocation group number */
 	int		blks_per_cluster; /* num blocks per inode cluster */
 	xfs_agblock_t	chunk_agbno;	/* first block in inode chunk */
-	xfs_agino_t	chunk_agino;	/* first agino in inode chunk */
-	__int32_t	chunk_cnt;	/* count of free inodes in chunk */
-	xfs_inofree_t	chunk_free;	/* mask of free inodes in chunk */
 	xfs_agblock_t	cluster_agbno;	/* first block in inode cluster */
-	xfs_btree_cur_t	*cur;	/* inode btree cursor */
 	int		error;	/* error code */
-	int		i;	/* temp state */
 	int		offset;	/* index of inode in its buffer */
 	int		offset_agbno;	/* blks from chunk start to inode */
 
 	ASSERT(ino != NULLFSINO);
+
 	/*
 	 * Split up the inode number into its parts.
 	 */
@@ -1198,24 +1228,24 @@ xfs_dilocate(
 	    ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
 #ifdef DEBUG
 		/* no diagnostics for bulkstat, ino comes from userspace */
-		if (flags & XFS_IMAP_BULKSTAT)
+		if (flags & XFS_IGET_BULKSTAT)
 			return XFS_ERROR(EINVAL);
 		if (agno >= mp->m_sb.sb_agcount) {
 			xfs_fs_cmn_err(CE_ALERT, mp,
-					"xfs_dilocate: agno (%d) >= "
+					"xfs_imap: agno (%d) >= "
 					"mp->m_sb.sb_agcount (%d)",
 					agno,  mp->m_sb.sb_agcount);
 		}
 		if (agbno >= mp->m_sb.sb_agblocks) {
 			xfs_fs_cmn_err(CE_ALERT, mp,
-					"xfs_dilocate: agbno (0x%llx) >= "
+					"xfs_imap: agbno (0x%llx) >= "
 					"mp->m_sb.sb_agblocks (0x%lx)",
 					(unsigned long long) agbno,
 					(unsigned long) mp->m_sb.sb_agblocks);
 		}
 		if (ino != XFS_AGINO_TO_INO(mp, agno, agino)) {
 			xfs_fs_cmn_err(CE_ALERT, mp,
-					"xfs_dilocate: ino (0x%llx) != "
+					"xfs_imap: ino (0x%llx) != "
 					"XFS_AGINO_TO_INO(mp, agno, agino) "
 					"(0x%llx)",
 					ino, XFS_AGINO_TO_INO(mp, agno, agino));
@@ -1224,65 +1254,89 @@ xfs_dilocate(
 #endif /* DEBUG */
 		return XFS_ERROR(EINVAL);
 	}
-	if ((mp->m_sb.sb_blocksize >= XFS_INODE_CLUSTER_SIZE(mp)) ||
-	    !(flags & XFS_IMAP_LOOKUP)) {
+
+	/*
+	 * If the inode cluster size is the same as the blocksize or
+	 * smaller we get to the buffer by simple arithmetics.
+	 */
+	if (XFS_INODE_CLUSTER_SIZE(mp) <= mp->m_sb.sb_blocksize) {
 		offset = XFS_INO_TO_OFFSET(mp, ino);
 		ASSERT(offset < mp->m_sb.sb_inopblock);
-		*bno = XFS_AGB_TO_FSB(mp, agno, agbno);
-		*off = offset;
-		*len = 1;
+
+		imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, agbno);
+		imap->im_len = XFS_FSB_TO_BB(mp, 1);
+		imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
 		return 0;
 	}
+
 	blks_per_cluster = XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_blocklog;
-	if (*bno != NULLFSBLOCK) {
+
+	/*
+	 * If we get a block number passed from bulkstat we can use it to
+	 * find the buffer easily.
+	 */
+	if (imap->im_blkno) {
 		offset = XFS_INO_TO_OFFSET(mp, ino);
 		ASSERT(offset < mp->m_sb.sb_inopblock);
-		cluster_agbno = XFS_FSB_TO_AGBNO(mp, *bno);
-		*off = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
-			offset;
-		*len = blks_per_cluster;
+
+		cluster_agbno = XFS_DADDR_TO_AGBNO(mp, imap->im_blkno);
+		offset += (agbno - cluster_agbno) * mp->m_sb.sb_inopblock;
+
+		imap->im_len = XFS_FSB_TO_BB(mp, blks_per_cluster);
+		imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
 		return 0;
 	}
+
+	/*
+	 * If the inode chunks are aligned then use simple maths to
+	 * find the location. Otherwise we have to do a btree
+	 * lookup to find the location.
+	 */
 	if (mp->m_inoalign_mask) {
 		offset_agbno = agbno & mp->m_inoalign_mask;
 		chunk_agbno = agbno - offset_agbno;
 	} else {
+		xfs_btree_cur_t	*cur;	/* inode btree cursor */
+		xfs_agino_t	chunk_agino; /* first agino in inode chunk */
+		__int32_t	chunk_cnt; /* count of free inodes in chunk */
+		xfs_inofree_t	chunk_free; /* mask of free inodes in chunk */
+		xfs_buf_t	*agbp;	/* agi buffer */
+		int		i;	/* temp state */
+
 		down_read(&mp->m_peraglock);
 		error = xfs_ialloc_read_agi(mp, tp, agno, &agbp);
 		up_read(&mp->m_peraglock);
 		if (error) {
-#ifdef DEBUG
-			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
+			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
 					"xfs_ialloc_read_agi() returned "
 					"error %d, agno %d",
 					error, agno);
-#endif /* DEBUG */
 			return error;
 		}
-		cur = xfs_btree_init_cursor(mp, tp, agbp, agno, XFS_BTNUM_INO,
-			(xfs_inode_t *)0, 0);
-		if ((error = xfs_inobt_lookup_le(cur, agino, 0, 0, &i))) {
-#ifdef DEBUG
-			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
+
+		cur = xfs_inobt_init_cursor(mp, tp, agbp, agno);
+		error = xfs_inobt_lookup_le(cur, agino, 0, 0, &i);
+		if (error) {
+			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
 					"xfs_inobt_lookup_le() failed");
-#endif /* DEBUG */
 			goto error0;
 		}
-		if ((error = xfs_inobt_get_rec(cur, &chunk_agino, &chunk_cnt,
-				&chunk_free, &i))) {
-#ifdef DEBUG
-			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
+
+		error = xfs_inobt_get_rec(cur, &chunk_agino, &chunk_cnt,
+				&chunk_free, &i);
+		if (error) {
+			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
 					"xfs_inobt_get_rec() failed");
-#endif /* DEBUG */
 			goto error0;
 		}
 		if (i == 0) {
 #ifdef DEBUG
-			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_dilocate: "
+			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
 					"xfs_inobt_get_rec() failed");
 #endif /* DEBUG */
 			error = XFS_ERROR(EINVAL);
 		}
+ error0:
 		xfs_trans_brelse(tp, agbp);
 		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
 		if (error)
@@ -1290,19 +1344,35 @@ xfs_dilocate(
 		chunk_agbno = XFS_AGINO_TO_AGBNO(mp, chunk_agino);
 		offset_agbno = agbno - chunk_agbno;
 	}
+
 	ASSERT(agbno >= chunk_agbno);
 	cluster_agbno = chunk_agbno +
 		((offset_agbno / blks_per_cluster) * blks_per_cluster);
 	offset = ((agbno - cluster_agbno) * mp->m_sb.sb_inopblock) +
 		XFS_INO_TO_OFFSET(mp, ino);
-	*bno = XFS_AGB_TO_FSB(mp, agno, cluster_agbno);
-	*off = offset;
-	*len = blks_per_cluster;
+
+	imap->im_blkno = XFS_AGB_TO_DADDR(mp, agno, cluster_agbno);
+	imap->im_len = XFS_FSB_TO_BB(mp, blks_per_cluster);
+	imap->im_boffset = (ushort)(offset << mp->m_sb.sb_inodelog);
+
+	/*
+	 * If the inode number maps to a block outside the bounds
+	 * of the file system then return NULL rather than calling
+	 * read_buf and panicing when we get an error from the
+	 * driver.
+	 */
+	if ((imap->im_blkno + imap->im_len) >
+	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
+		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
+			"(imap->im_blkno (0x%llx) + imap->im_len (0x%llx)) > "
+			" XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) (0x%llx)",
+			(unsigned long long) imap->im_blkno,
+			(unsigned long long) imap->im_len,
+			XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
+		return XFS_ERROR(EINVAL);
+	}
+
 	return 0;
-error0:
-	xfs_trans_brelse(tp, agbp);
-	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
-	return error;
 }
 
 /*
@@ -1370,70 +1440,95 @@ xfs_ialloc_log_agi(
 	xfs_trans_log_buf(tp, bp, first, last);
 }
 
+#ifdef DEBUG
+STATIC void
+xfs_check_agi_unlinked(
+	struct xfs_agi		*agi)
+{
+	int			i;
+
+	for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++)
+		ASSERT(agi->agi_unlinked[i]);
+}
+#else
+#define xfs_check_agi_unlinked(agi)
+#endif
+
 /*
  * Read in the allocation group header (inode allocation section)
  */
 int
-xfs_ialloc_read_agi(
-	xfs_mount_t	*mp,		/* file system mount structure */
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_agnumber_t	agno,		/* allocation group number */
-	xfs_buf_t	**bpp)		/* allocation group hdr buf */
+xfs_read_agi(
+	struct xfs_mount	*mp,	/* file system mount structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	struct xfs_buf		**bpp)	/* allocation group hdr buf */
 {
-	xfs_agi_t	*agi;		/* allocation group header */
-	int		agi_ok;		/* agi is consistent */
-	xfs_buf_t	*bp;		/* allocation group hdr buf */
-	xfs_perag_t	*pag;		/* per allocation group data */
-	int		error;
+	struct xfs_agi		*agi;	/* allocation group header */
+	int			agi_ok;	/* agi is consistent */
+	int			error;
 
 	ASSERT(agno != NULLAGNUMBER);
-	error = xfs_trans_read_buf(
-			mp, tp, mp->m_ddev_targp,
+
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
 			XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
-			XFS_FSS_TO_BB(mp, 1), 0, &bp);
+			XFS_FSS_TO_BB(mp, 1), 0, bpp);
 	if (error)
 		return error;
-	ASSERT(bp && !XFS_BUF_GETERROR(bp));
+
+	ASSERT(*bpp && !XFS_BUF_GETERROR(*bpp));
+	agi = XFS_BUF_TO_AGI(*bpp);
 
 	/*
 	 * Validate the magic number of the agi block.
 	 */
-	agi = XFS_BUF_TO_AGI(bp);
-	agi_ok =
-		be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
-		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
+	agi_ok = be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
+		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum)) &&
+		be32_to_cpu(agi->agi_seqno) == agno;
 	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
 			XFS_RANDOM_IALLOC_READ_AGI))) {
-		XFS_CORRUPTION_ERROR("xfs_ialloc_read_agi", XFS_ERRLEVEL_LOW,
+		XFS_CORRUPTION_ERROR("xfs_read_agi", XFS_ERRLEVEL_LOW,
 				     mp, agi);
-		xfs_trans_brelse(tp, bp);
+		xfs_trans_brelse(tp, *bpp);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
+
+	XFS_BUF_SET_VTYPE_REF(*bpp, B_FS_AGI, XFS_AGI_REF);
+
+	xfs_check_agi_unlinked(agi);
+	return 0;
+}
+
+int
+xfs_ialloc_read_agi(
+	struct xfs_mount	*mp,	/* file system mount structure */
+	struct xfs_trans	*tp,	/* transaction pointer */
+	xfs_agnumber_t		agno,	/* allocation group number */
+	struct xfs_buf		**bpp)	/* allocation group hdr buf */
+{
+	struct xfs_agi		*agi;	/* allocation group header */
+	struct xfs_perag	*pag;	/* per allocation group data */
+	int			error;
+
+	error = xfs_read_agi(mp, tp, agno, bpp);
+	if (error)
+		return error;
+
+	agi = XFS_BUF_TO_AGI(*bpp);
 	pag = &mp->m_perag[agno];
+
 	if (!pag->pagi_init) {
 		pag->pagi_freecount = be32_to_cpu(agi->agi_freecount);
 		pag->pagi_count = be32_to_cpu(agi->agi_count);
 		pag->pagi_init = 1;
-	} else {
-		/*
-		 * It's possible for these to be out of sync if
-		 * we are in the middle of a forced shutdown.
-		 */
-		ASSERT(pag->pagi_freecount == be32_to_cpu(agi->agi_freecount) ||
-			XFS_FORCED_SHUTDOWN(mp));
 	}
 
-#ifdef DEBUG
-	{
-		int	i;
-
-		for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++)
-			ASSERT(agi->agi_unlinked[i]);
-	}
-#endif
-
-	XFS_BUF_SET_VTYPE_REF(bp, B_FS_AGI, XFS_AGI_REF);
-	*bpp = bp;
+	/*
+	 * It's possible for these to be out of sync if
+	 * we are in the middle of a forced shutdown.
+	 */
+	ASSERT(pag->pagi_freecount == be32_to_cpu(agi->agi_freecount) ||
+		XFS_FORCED_SHUTDOWN(mp));
 	return 0;
 }
 
diff --git a/fs/xfs/xfs_ialloc.h b/fs/xfs/xfs_ialloc.h
index 4e30ec1d13b..50f558a4e0a 100644
--- a/fs/xfs/xfs_ialloc.h
+++ b/fs/xfs/xfs_ialloc.h
@@ -20,6 +20,7 @@
 
 struct xfs_buf;
 struct xfs_dinode;
+struct xfs_imap;
 struct xfs_mount;
 struct xfs_trans;
 
@@ -56,7 +57,6 @@ static inline int xfs_ialloc_find_free(xfs_inofree_t *fp)
 }
 
 
-#ifdef __KERNEL__
 /*
  * Allocate an inode on disk.
  * Mode is used to tell whether the new inode will need space, and whether
@@ -105,17 +105,14 @@ xfs_difree(
 	xfs_ino_t	*first_ino);	/* first inode in deleted cluster */
 
 /*
- * Return the location of the inode in bno/len/off,
- * for mapping it into a buffer.
+ * Return the location of the inode in imap, for mapping it into a buffer.
  */
 int
-xfs_dilocate(
+xfs_imap(
 	struct xfs_mount *mp,		/* file system mount structure */
 	struct xfs_trans *tp,		/* transaction pointer */
 	xfs_ino_t	ino,		/* inode to locate */
-	xfs_fsblock_t	*bno,		/* output: block containing inode */
-	int		*len,		/* output: num blocks in cluster*/
-	int		*off,		/* output: index in block of inode */
+	struct xfs_imap	*imap,		/* location map structure */
 	uint		flags);		/* flags for inode btree lookup */
 
 /*
@@ -154,6 +151,24 @@ xfs_ialloc_pagi_init(
 	struct xfs_trans *tp,		/* transaction pointer */
         xfs_agnumber_t  agno);		/* allocation group number */
 
-#endif	/* __KERNEL__ */
+/*
+ * Lookup the first record greater than or equal to ino
+ * in the btree given by cur.
+ */
+int xfs_inobt_lookup_ge(struct xfs_btree_cur *cur, xfs_agino_t ino,
+		__int32_t fcnt,	xfs_inofree_t free, int *stat);
+
+/*
+ * Lookup the first record less than or equal to ino
+ * in the btree given by cur.
+ */
+int xfs_inobt_lookup_le(struct xfs_btree_cur *cur, xfs_agino_t ino,
+		__int32_t fcnt,	xfs_inofree_t free, int *stat);
+
+/*
+ * Get the data from the pointed-to record.
+ */
+extern int xfs_inobt_get_rec(struct xfs_btree_cur *cur, xfs_agino_t *ino,
+			     __int32_t *fcnt, xfs_inofree_t *free, int *stat);
 
 #endif	/* __XFS_IALLOC_H__ */
diff --git a/fs/xfs/xfs_ialloc_btree.c b/fs/xfs/xfs_ialloc_btree.c
index 83502f3edef..99f2408e8d8 100644
--- a/fs/xfs/xfs_ialloc_btree.c
+++ b/fs/xfs/xfs_ialloc_btree.c
@@ -35,2044 +35,349 @@
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
 #include "xfs_btree.h"
+#include "xfs_btree_trace.h"
 #include "xfs_ialloc.h"
 #include "xfs_alloc.h"
 #include "xfs_error.h"
 
-STATIC void xfs_inobt_log_block(xfs_trans_t *, xfs_buf_t *, int);
-STATIC void xfs_inobt_log_keys(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC void xfs_inobt_log_ptrs(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC void xfs_inobt_log_recs(xfs_btree_cur_t *, xfs_buf_t *, int, int);
-STATIC int xfs_inobt_lshift(xfs_btree_cur_t *, int, int *);
-STATIC int xfs_inobt_newroot(xfs_btree_cur_t *, int *);
-STATIC int xfs_inobt_rshift(xfs_btree_cur_t *, int, int *);
-STATIC int xfs_inobt_split(xfs_btree_cur_t *, int, xfs_agblock_t *,
-		xfs_inobt_key_t *, xfs_btree_cur_t **, int *);
-STATIC int xfs_inobt_updkey(xfs_btree_cur_t *, xfs_inobt_key_t *, int);
 
-/*
- * Single level of the xfs_inobt_delete record deletion routine.
- * Delete record pointed to by cur/level.
- * Remove the record from its block then rebalance the tree.
- * Return 0 for error, 1 for done, 2 to go on to the next level.
- */
-STATIC int				/* error */
-xfs_inobt_delrec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level removing record from */
-	int			*stat)	/* fail/done/go-on */
+STATIC int
+xfs_inobt_get_minrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
 {
-	xfs_buf_t		*agbp;	/* buffer for a.g. inode header */
-	xfs_mount_t		*mp;	/* mount structure */
-	xfs_agi_t		*agi;	/* allocation group inode header */
-	xfs_inobt_block_t	*block;	/* btree block record/key lives in */
-	xfs_agblock_t		bno;	/* btree block number */
-	xfs_buf_t		*bp;	/* buffer for block */
-	int			error;	/* error return value */
-	int			i;	/* loop index */
-	xfs_inobt_key_t		key;	/* kp points here if block is level 0 */
-	xfs_inobt_key_t		*kp = NULL;	/* pointer to btree keys */
-	xfs_agblock_t		lbno;	/* left block's block number */
-	xfs_buf_t		*lbp;	/* left block's buffer pointer */
-	xfs_inobt_block_t	*left;	/* left btree block */
-	xfs_inobt_key_t		*lkp;	/* left block key pointer */
-	xfs_inobt_ptr_t		*lpp;	/* left block address pointer */
-	int			lrecs = 0;	/* number of records in left block */
-	xfs_inobt_rec_t		*lrp;	/* left block record pointer */
-	xfs_inobt_ptr_t		*pp = NULL;	/* pointer to btree addresses */
-	int			ptr;	/* index in btree block for this rec */
-	xfs_agblock_t		rbno;	/* right block's block number */
-	xfs_buf_t		*rbp;	/* right block's buffer pointer */
-	xfs_inobt_block_t	*right;	/* right btree block */
-	xfs_inobt_key_t		*rkp;	/* right block key pointer */
-	xfs_inobt_rec_t		*rp;	/* pointer to btree records */
-	xfs_inobt_ptr_t		*rpp;	/* right block address pointer */
-	int			rrecs = 0;	/* number of records in right block */
-	int			numrecs;
-	xfs_inobt_rec_t		*rrp;	/* right block record pointer */
-	xfs_btree_cur_t		*tcur;	/* temporary btree cursor */
-
-	mp = cur->bc_mp;
-
-	/*
-	 * Get the index of the entry being deleted, check for nothing there.
-	 */
-	ptr = cur->bc_ptrs[level];
-	if (ptr == 0) {
-		*stat = 0;
-		return 0;
-	}
-
-	/*
-	 * Get the buffer & block containing the record or key/ptr.
-	 */
-	bp = cur->bc_bufs[level];
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-		return error;
-#endif
-	/*
-	 * Fail if we're off the end of the block.
-	 */
+	return cur->bc_mp->m_inobt_mnr[level != 0];
+}
 
-	numrecs = be16_to_cpu(block->bb_numrecs);
-	if (ptr > numrecs) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * It's a nonleaf.  Excise the key and ptr being deleted, by
-	 * sliding the entries past them down one.
-	 * Log the changed areas of the block.
-	 */
-	if (level > 0) {
-		kp = XFS_INOBT_KEY_ADDR(block, 1, cur);
-		pp = XFS_INOBT_PTR_ADDR(block, 1, cur);
-#ifdef DEBUG
-		for (i = ptr; i < numrecs; i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(pp[i]), level)))
-				return error;
-		}
-#endif
-		if (ptr < numrecs) {
-			memmove(&kp[ptr - 1], &kp[ptr],
-				(numrecs - ptr) * sizeof(*kp));
-			memmove(&pp[ptr - 1], &pp[ptr],
-				(numrecs - ptr) * sizeof(*kp));
-			xfs_inobt_log_keys(cur, bp, ptr, numrecs - 1);
-			xfs_inobt_log_ptrs(cur, bp, ptr, numrecs - 1);
-		}
-	}
-	/*
-	 * It's a leaf.  Excise the record being deleted, by sliding the
-	 * entries past it down one.  Log the changed areas of the block.
-	 */
-	else {
-		rp = XFS_INOBT_REC_ADDR(block, 1, cur);
-		if (ptr < numrecs) {
-			memmove(&rp[ptr - 1], &rp[ptr],
-				(numrecs - ptr) * sizeof(*rp));
-			xfs_inobt_log_recs(cur, bp, ptr, numrecs - 1);
-		}
-		/*
-		 * If it's the first record in the block, we'll need a key
-		 * structure to pass up to the next level (updkey).
-		 */
-		if (ptr == 1) {
-			key.ir_startino = rp->ir_startino;
-			kp = &key;
-		}
-	}
-	/*
-	 * Decrement and log the number of entries in the block.
-	 */
-	numrecs--;
-	block->bb_numrecs = cpu_to_be16(numrecs);
-	xfs_inobt_log_block(cur->bc_tp, bp, XFS_BB_NUMRECS);
-	/*
-	 * Is this the root level?  If so, we're almost done.
-	 */
-	if (level == cur->bc_nlevels - 1) {
-		/*
-		 * If this is the root level,
-		 * and there's only one entry left,
-		 * and it's NOT the leaf level,
-		 * then we can get rid of this level.
-		 */
-		if (numrecs == 1 && level > 0) {
-			agbp = cur->bc_private.a.agbp;
-			agi = XFS_BUF_TO_AGI(agbp);
-			/*
-			 * pp is still set to the first pointer in the block.
-			 * Make it the new root of the btree.
-			 */
-			bno = be32_to_cpu(agi->agi_root);
-			agi->agi_root = *pp;
-			be32_add_cpu(&agi->agi_level, -1);
-			/*
-			 * Free the block.
-			 */
-			if ((error = xfs_free_extent(cur->bc_tp,
-				XFS_AGB_TO_FSB(mp, cur->bc_private.a.agno, bno), 1)))
-				return error;
-			xfs_trans_binval(cur->bc_tp, bp);
-			xfs_ialloc_log_agi(cur->bc_tp, agbp,
-				XFS_AGI_ROOT | XFS_AGI_LEVEL);
-			/*
-			 * Update the cursor so there's one fewer level.
-			 */
-			cur->bc_bufs[level] = NULL;
-			cur->bc_nlevels--;
-		} else if (level > 0 &&
-			   (error = xfs_inobt_decrement(cur, level, &i)))
-			return error;
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * If we deleted the leftmost entry in the block, update the
-	 * key values above us in the tree.
-	 */
-	if (ptr == 1 && (error = xfs_inobt_updkey(cur, kp, level + 1)))
-		return error;
-	/*
-	 * If the number of records remaining in the block is at least
-	 * the minimum, we're done.
-	 */
-	if (numrecs >= XFS_INOBT_BLOCK_MINRECS(level, cur)) {
-		if (level > 0 &&
-		    (error = xfs_inobt_decrement(cur, level, &i)))
-			return error;
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * Otherwise, we have to move some records around to keep the
-	 * tree balanced.  Look at the left and right sibling blocks to
-	 * see if we can re-balance by moving only one record.
-	 */
-	rbno = be32_to_cpu(block->bb_rightsib);
-	lbno = be32_to_cpu(block->bb_leftsib);
-	bno = NULLAGBLOCK;
-	ASSERT(rbno != NULLAGBLOCK || lbno != NULLAGBLOCK);
-	/*
-	 * Duplicate the cursor so our btree manipulations here won't
-	 * disrupt the next level up.
-	 */
-	if ((error = xfs_btree_dup_cursor(cur, &tcur)))
-		return error;
-	/*
-	 * If there's a right sibling, see if it's ok to shift an entry
-	 * out of it.
-	 */
-	if (rbno != NULLAGBLOCK) {
-		/*
-		 * Move the temp cursor to the last entry in the next block.
-		 * Actually any entry but the first would suffice.
-		 */
-		i = xfs_btree_lastrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		if ((error = xfs_inobt_increment(tcur, level, &i)))
-			goto error0;
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		i = xfs_btree_lastrec(tcur, level);
-		XFS_WANT_CORRUPTED_GOTO(i == 1, error0);
-		/*
-		 * Grab a pointer to the block.
-		 */
-		rbp = tcur->bc_bufs[level];
-		right = XFS_BUF_TO_INOBT_BLOCK(rbp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-			goto error0;
-#endif
-		/*
-		 * Grab the current block number, for future use.
-		 */
-		bno = be32_to_cpu(right->bb_leftsib);
-		/*
-		 * If right block is full enough so that removing one entry
-		 * won't make it too empty, and left-shifting an entry out
-		 * of right to us works, we're done.
-		 */
-		if (be16_to_cpu(right->bb_numrecs) - 1 >=
-		     XFS_INOBT_BLOCK_MINRECS(level, cur)) {
-			if ((error = xfs_inobt_lshift(tcur, level, &i)))
-				goto error0;
-			if (i) {
-				ASSERT(be16_to_cpu(block->bb_numrecs) >=
-				       XFS_INOBT_BLOCK_MINRECS(level, cur));
-				xfs_btree_del_cursor(tcur,
-						     XFS_BTREE_NOERROR);
-				if (level > 0 &&
-				    (error = xfs_inobt_decrement(cur, level,
-						&i)))
-					return error;
-				*stat = 1;
-				return 0;
-			}
-		}
-		/*
-		 * Otherwise, grab the number of records in right for
-		 * future reference, and fix up the temp cursor to point
-		 * to our block again (last record).
-		 */
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		if (lbno != NULLAGBLOCK) {
-			xfs_btree_firstrec(tcur, level);
-			if ((error = xfs_inobt_decrement(tcur, level, &i)))
-				goto error0;
-		}
-	}
-	/*
-	 * If there's a left sibling, see if it's ok to shift an entry
-	 * out of it.
-	 */
-	if (lbno != NULLAGBLOCK) {
-		/*
-		 * Move the temp cursor to the first entry in the
-		 * previous block.
-		 */
-		xfs_btree_firstrec(tcur, level);
-		if ((error = xfs_inobt_decrement(tcur, level, &i)))
-			goto error0;
-		xfs_btree_firstrec(tcur, level);
-		/*
-		 * Grab a pointer to the block.
-		 */
-		lbp = tcur->bc_bufs[level];
-		left = XFS_BUF_TO_INOBT_BLOCK(lbp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-			goto error0;
-#endif
-		/*
-		 * Grab the current block number, for future use.
-		 */
-		bno = be32_to_cpu(left->bb_rightsib);
-		/*
-		 * If left block is full enough so that removing one entry
-		 * won't make it too empty, and right-shifting an entry out
-		 * of left to us works, we're done.
-		 */
-		if (be16_to_cpu(left->bb_numrecs) - 1 >=
-		     XFS_INOBT_BLOCK_MINRECS(level, cur)) {
-			if ((error = xfs_inobt_rshift(tcur, level, &i)))
-				goto error0;
-			if (i) {
-				ASSERT(be16_to_cpu(block->bb_numrecs) >=
-				       XFS_INOBT_BLOCK_MINRECS(level, cur));
-				xfs_btree_del_cursor(tcur,
-						     XFS_BTREE_NOERROR);
-				if (level == 0)
-					cur->bc_ptrs[0]++;
-				*stat = 1;
-				return 0;
-			}
-		}
-		/*
-		 * Otherwise, grab the number of records in right for
-		 * future reference.
-		 */
-		lrecs = be16_to_cpu(left->bb_numrecs);
-	}
-	/*
-	 * Delete the temp cursor, we're done with it.
-	 */
-	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-	/*
-	 * If here, we need to do a join to keep the tree balanced.
-	 */
-	ASSERT(bno != NULLAGBLOCK);
-	/*
-	 * See if we can join with the left neighbor block.
-	 */
-	if (lbno != NULLAGBLOCK &&
-	    lrecs + numrecs <= XFS_INOBT_BLOCK_MAXRECS(level, cur)) {
-		/*
-		 * Set "right" to be the starting block,
-		 * "left" to be the left neighbor.
-		 */
-		rbno = bno;
-		right = block;
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		rbp = bp;
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, lbno, 0, &lbp,
-				XFS_INO_BTREE_REF)))
-			return error;
-		left = XFS_BUF_TO_INOBT_BLOCK(lbp);
-		lrecs = be16_to_cpu(left->bb_numrecs);
-		if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-			return error;
-	}
-	/*
-	 * If that won't work, see if we can join with the right neighbor block.
-	 */
-	else if (rbno != NULLAGBLOCK &&
-		 rrecs + numrecs <= XFS_INOBT_BLOCK_MAXRECS(level, cur)) {
-		/*
-		 * Set "left" to be the starting block,
-		 * "right" to be the right neighbor.
-		 */
-		lbno = bno;
-		left = block;
-		lrecs = be16_to_cpu(left->bb_numrecs);
-		lbp = bp;
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, rbno, 0, &rbp,
-				XFS_INO_BTREE_REF)))
-			return error;
-		right = XFS_BUF_TO_INOBT_BLOCK(rbp);
-		rrecs = be16_to_cpu(right->bb_numrecs);
-		if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-			return error;
-	}
-	/*
-	 * Otherwise, we can't fix the imbalance.
-	 * Just return.  This is probably a logic error, but it's not fatal.
-	 */
-	else {
-		if (level > 0 && (error = xfs_inobt_decrement(cur, level, &i)))
-			return error;
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * We're now going to join "left" and "right" by moving all the stuff
-	 * in "right" to "left" and deleting "right".
-	 */
-	if (level > 0) {
-		/*
-		 * It's a non-leaf.  Move keys and pointers.
-		 */
-		lkp = XFS_INOBT_KEY_ADDR(left, lrecs + 1, cur);
-		lpp = XFS_INOBT_PTR_ADDR(left, lrecs + 1, cur);
-		rkp = XFS_INOBT_KEY_ADDR(right, 1, cur);
-		rpp = XFS_INOBT_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = 0; i < rrecs; i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i]), level)))
-				return error;
-		}
-#endif
-		memcpy(lkp, rkp, rrecs * sizeof(*lkp));
-		memcpy(lpp, rpp, rrecs * sizeof(*lpp));
-		xfs_inobt_log_keys(cur, lbp, lrecs + 1, lrecs + rrecs);
-		xfs_inobt_log_ptrs(cur, lbp, lrecs + 1, lrecs + rrecs);
-	} else {
-		/*
-		 * It's a leaf.  Move records.
-		 */
-		lrp = XFS_INOBT_REC_ADDR(left, lrecs + 1, cur);
-		rrp = XFS_INOBT_REC_ADDR(right, 1, cur);
-		memcpy(lrp, rrp, rrecs * sizeof(*lrp));
-		xfs_inobt_log_recs(cur, lbp, lrecs + 1, lrecs + rrecs);
-	}
-	/*
-	 * If we joined with the left neighbor, set the buffer in the
-	 * cursor to the left block, and fix up the index.
-	 */
-	if (bp != lbp) {
-		xfs_btree_setbuf(cur, level, lbp);
-		cur->bc_ptrs[level] += lrecs;
-	}
-	/*
-	 * If we joined with the right neighbor and there's a level above
-	 * us, increment the cursor at that level.
-	 */
-	else if (level + 1 < cur->bc_nlevels &&
-		 (error = xfs_alloc_increment(cur, level + 1, &i)))
-		return error;
-	/*
-	 * Fix up the number of records in the surviving block.
-	 */
-	lrecs += rrecs;
-	left->bb_numrecs = cpu_to_be16(lrecs);
-	/*
-	 * Fix up the right block pointer in the surviving block, and log it.
-	 */
-	left->bb_rightsib = right->bb_rightsib;
-	xfs_inobt_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
-	/*
-	 * If there is a right sibling now, make it point to the
-	 * remaining block.
-	 */
-	if (be32_to_cpu(left->bb_rightsib) != NULLAGBLOCK) {
-		xfs_inobt_block_t	*rrblock;
-		xfs_buf_t		*rrbp;
+STATIC struct xfs_btree_cur *
+xfs_inobt_dup_cursor(
+	struct xfs_btree_cur	*cur)
+{
+	return xfs_inobt_init_cursor(cur->bc_mp, cur->bc_tp,
+			cur->bc_private.a.agbp, cur->bc_private.a.agno);
+}
 
-		if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-				cur->bc_private.a.agno, be32_to_cpu(left->bb_rightsib), 0,
-				&rrbp, XFS_INO_BTREE_REF)))
-			return error;
-		rrblock = XFS_BUF_TO_INOBT_BLOCK(rrbp);
-		if ((error = xfs_btree_check_sblock(cur, rrblock, level, rrbp)))
-			return error;
-		rrblock->bb_leftsib = cpu_to_be32(lbno);
-		xfs_inobt_log_block(cur->bc_tp, rrbp, XFS_BB_LEFTSIB);
-	}
-	/*
-	 * Free the deleting block.
-	 */
-	if ((error = xfs_free_extent(cur->bc_tp, XFS_AGB_TO_FSB(mp,
-				     cur->bc_private.a.agno, rbno), 1)))
-		return error;
-	xfs_trans_binval(cur->bc_tp, rbp);
-	/*
-	 * Readjust the ptr at this level if it's not a leaf, since it's
-	 * still pointing at the deletion point, which makes the cursor
-	 * inconsistent.  If this makes the ptr 0, the caller fixes it up.
-	 * We can't use decrement because it would change the next level up.
-	 */
-	if (level > 0)
-		cur->bc_ptrs[level]--;
-	/*
-	 * Return value means the next level up has something to do.
-	 */
-	*stat = 2;
-	return 0;
+STATIC void
+xfs_inobt_set_root(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*nptr,
+	int			inc)	/* level change */
+{
+	struct xfs_buf		*agbp = cur->bc_private.a.agbp;
+	struct xfs_agi		*agi = XFS_BUF_TO_AGI(agbp);
 
-error0:
-	xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
-	return error;
+	agi->agi_root = nptr->s;
+	be32_add_cpu(&agi->agi_level, inc);
+	xfs_ialloc_log_agi(cur->bc_tp, agbp, XFS_AGI_ROOT | XFS_AGI_LEVEL);
 }
 
-/*
- * Insert one record/level.  Return information to the caller
- * allowing the next level up to proceed if necessary.
- */
-STATIC int				/* error */
-xfs_inobt_insrec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to insert record at */
-	xfs_agblock_t		*bnop,	/* i/o: block number inserted */
-	xfs_inobt_rec_t		*recp,	/* i/o: record data inserted */
-	xfs_btree_cur_t		**curp,	/* output: new cursor replacing cur */
-	int			*stat)	/* success/failure */
+STATIC int
+xfs_inobt_alloc_block(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*start,
+	union xfs_btree_ptr	*new,
+	int			length,
+	int			*stat)
 {
-	xfs_inobt_block_t	*block;	/* btree block record/key lives in */
-	xfs_buf_t		*bp;	/* buffer for block */
-	int			error;	/* error return value */
-	int			i;	/* loop index */
-	xfs_inobt_key_t		key;	/* key value being inserted */
-	xfs_inobt_key_t		*kp=NULL;	/* pointer to btree keys */
-	xfs_agblock_t		nbno;	/* block number of allocated block */
-	xfs_btree_cur_t		*ncur;	/* new cursor to be used at next lvl */
-	xfs_inobt_key_t		nkey;	/* new key value, from split */
-	xfs_inobt_rec_t		nrec;	/* new record value, for caller */
-	int			numrecs;
-	int			optr;	/* old ptr value */
-	xfs_inobt_ptr_t		*pp;	/* pointer to btree addresses */
-	int			ptr;	/* index in btree block for this rec */
-	xfs_inobt_rec_t		*rp=NULL;	/* pointer to btree records */
+	xfs_alloc_arg_t		args;		/* block allocation args */
+	int			error;		/* error return value */
+	xfs_agblock_t		sbno = be32_to_cpu(start->s);
 
-	/*
-	 * GCC doesn't understand the (arguably complex) control flow in
-	 * this function and complains about uninitialized structure fields
-	 * without this.
-	 */
-	memset(&nrec, 0, sizeof(nrec));
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
 
-	/*
-	 * If we made it to the root level, allocate a new root block
-	 * and we're done.
-	 */
-	if (level >= cur->bc_nlevels) {
-		error = xfs_inobt_newroot(cur, &i);
-		*bnop = NULLAGBLOCK;
-		*stat = i;
+	memset(&args, 0, sizeof(args));
+	args.tp = cur->bc_tp;
+	args.mp = cur->bc_mp;
+	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_private.a.agno, sbno);
+	args.minlen = 1;
+	args.maxlen = 1;
+	args.prod = 1;
+	args.type = XFS_ALLOCTYPE_NEAR_BNO;
+
+	error = xfs_alloc_vextent(&args);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
 		return error;
 	}
-	/*
-	 * Make a key out of the record data to be inserted, and save it.
-	 */
-	key.ir_startino = recp->ir_startino;
-	optr = ptr = cur->bc_ptrs[level];
-	/*
-	 * If we're off the left edge, return failure.
-	 */
-	if (ptr == 0) {
+	if (args.fsbno == NULLFSBLOCK) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
 		*stat = 0;
 		return 0;
 	}
-	/*
-	 * Get pointers to the btree buffer and block.
-	 */
-	bp = cur->bc_bufs[level];
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-	numrecs = be16_to_cpu(block->bb_numrecs);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-		return error;
-	/*
-	 * Check that the new entry is being inserted in the right place.
-	 */
-	if (ptr <= numrecs) {
-		if (level == 0) {
-			rp = XFS_INOBT_REC_ADDR(block, ptr, cur);
-			xfs_btree_check_rec(cur->bc_btnum, recp, rp);
-		} else {
-			kp = XFS_INOBT_KEY_ADDR(block, ptr, cur);
-			xfs_btree_check_key(cur->bc_btnum, &key, kp);
-		}
-	}
-#endif
-	nbno = NULLAGBLOCK;
-	ncur = NULL;
-	/*
-	 * If the block is full, we can't insert the new entry until we
-	 * make the block un-full.
-	 */
-	if (numrecs == XFS_INOBT_BLOCK_MAXRECS(level, cur)) {
-		/*
-		 * First, try shifting an entry to the right neighbor.
-		 */
-		if ((error = xfs_inobt_rshift(cur, level, &i)))
-			return error;
-		if (i) {
-			/* nothing */
-		}
-		/*
-		 * Next, try shifting an entry to the left neighbor.
-		 */
-		else {
-			if ((error = xfs_inobt_lshift(cur, level, &i)))
-				return error;
-			if (i) {
-				optr = ptr = cur->bc_ptrs[level];
-			} else {
-				/*
-				 * Next, try splitting the current block
-				 * in half. If this works we have to
-				 * re-set our variables because
-				 * we could be in a different block now.
-				 */
-				if ((error = xfs_inobt_split(cur, level, &nbno,
-						&nkey, &ncur, &i)))
-					return error;
-				if (i) {
-					bp = cur->bc_bufs[level];
-					block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-					if ((error = xfs_btree_check_sblock(cur,
-							block, level, bp)))
-						return error;
-#endif
-					ptr = cur->bc_ptrs[level];
-					nrec.ir_startino = nkey.ir_startino;
-				} else {
-					/*
-					 * Otherwise the insert fails.
-					 */
-					*stat = 0;
-					return 0;
-				}
-			}
-		}
-	}
-	/*
-	 * At this point we know there's room for our new entry in the block
-	 * we're pointing at.
-	 */
-	numrecs = be16_to_cpu(block->bb_numrecs);
-	if (level > 0) {
-		/*
-		 * It's a non-leaf entry.  Make a hole for the new data
-		 * in the key and ptr regions of the block.
-		 */
-		kp = XFS_INOBT_KEY_ADDR(block, 1, cur);
-		pp = XFS_INOBT_PTR_ADDR(block, 1, cur);
-#ifdef DEBUG
-		for (i = numrecs; i >= ptr; i--) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(pp[i - 1]), level)))
-				return error;
-		}
-#endif
-		memmove(&kp[ptr], &kp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*kp));
-		memmove(&pp[ptr], &pp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*pp));
-		/*
-		 * Now stuff the new data in, bump numrecs and log the new data.
-		 */
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sptr(cur, *bnop, level)))
-			return error;
-#endif
-		kp[ptr - 1] = key;
-		pp[ptr - 1] = cpu_to_be32(*bnop);
-		numrecs++;
-		block->bb_numrecs = cpu_to_be16(numrecs);
-		xfs_inobt_log_keys(cur, bp, ptr, numrecs);
-		xfs_inobt_log_ptrs(cur, bp, ptr, numrecs);
-	} else {
-		/*
-		 * It's a leaf entry.  Make a hole for the new record.
-		 */
-		rp = XFS_INOBT_REC_ADDR(block, 1, cur);
-		memmove(&rp[ptr], &rp[ptr - 1],
-			(numrecs - ptr + 1) * sizeof(*rp));
-		/*
-		 * Now stuff the new record in, bump numrecs
-		 * and log the new data.
-		 */
-		rp[ptr - 1] = *recp;
-		numrecs++;
-		block->bb_numrecs = cpu_to_be16(numrecs);
-		xfs_inobt_log_recs(cur, bp, ptr, numrecs);
-	}
-	/*
-	 * Log the new number of records in the btree header.
-	 */
-	xfs_inobt_log_block(cur->bc_tp, bp, XFS_BB_NUMRECS);
-#ifdef DEBUG
-	/*
-	 * Check that the key/record is in the right place, now.
-	 */
-	if (ptr < numrecs) {
-		if (level == 0)
-			xfs_btree_check_rec(cur->bc_btnum, rp + ptr - 1,
-				rp + ptr);
-		else
-			xfs_btree_check_key(cur->bc_btnum, kp + ptr - 1,
-				kp + ptr);
-	}
-#endif
-	/*
-	 * If we inserted at the start of a block, update the parents' keys.
-	 */
-	if (optr == 1 && (error = xfs_inobt_updkey(cur, &key, level + 1)))
-		return error;
-	/*
-	 * Return the new block number, if any.
-	 * If there is one, give back a record value and a cursor too.
-	 */
-	*bnop = nbno;
-	if (nbno != NULLAGBLOCK) {
-		*recp = nrec;
-		*curp = ncur;
-	}
+	ASSERT(args.len == 1);
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
+
+	new->s = cpu_to_be32(XFS_FSB_TO_AGBNO(args.mp, args.fsbno));
 	*stat = 1;
 	return 0;
 }
 
-/*
- * Log header fields from a btree block.
- */
-STATIC void
-xfs_inobt_log_block(
-	xfs_trans_t		*tp,	/* transaction pointer */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			fields)	/* mask of fields: XFS_BB_... */
+STATIC int
+xfs_inobt_free_block(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp)
 {
-	int			first;	/* first byte offset logged */
-	int			last;	/* last byte offset logged */
-	static const short	offsets[] = {	/* table of offsets */
-		offsetof(xfs_inobt_block_t, bb_magic),
-		offsetof(xfs_inobt_block_t, bb_level),
-		offsetof(xfs_inobt_block_t, bb_numrecs),
-		offsetof(xfs_inobt_block_t, bb_leftsib),
-		offsetof(xfs_inobt_block_t, bb_rightsib),
-		sizeof(xfs_inobt_block_t)
-	};
+	xfs_fsblock_t		fsbno;
+	int			error;
 
-	xfs_btree_offsets(fields, offsets, XFS_BB_NUM_BITS, &first, &last);
-	xfs_trans_log_buf(tp, bp, first, last);
+	fsbno = XFS_DADDR_TO_FSB(cur->bc_mp, XFS_BUF_ADDR(bp));
+	error = xfs_free_extent(cur->bc_tp, fsbno, 1);
+	if (error)
+		return error;
+
+	xfs_trans_binval(cur->bc_tp, bp);
+	return error;
 }
 
-/*
- * Log keys from a btree block (nonleaf).
- */
-STATIC void
-xfs_inobt_log_keys(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			kfirst,	/* index of first key to log */
-	int			klast)	/* index of last key to log */
+STATIC int
+xfs_inobt_get_maxrecs(
+	struct xfs_btree_cur	*cur,
+	int			level)
 {
-	xfs_inobt_block_t	*block;	/* btree block to log from */
-	int			first;	/* first byte offset logged */
-	xfs_inobt_key_t		*kp;	/* key pointer in btree block */
-	int			last;	/* last byte offset logged */
-
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-	kp = XFS_INOBT_KEY_ADDR(block, 1, cur);
-	first = (int)((xfs_caddr_t)&kp[kfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&kp[klast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	return cur->bc_mp->m_inobt_mxr[level != 0];
 }
 
-/*
- * Log block pointer fields from a btree block (nonleaf).
- */
 STATIC void
-xfs_inobt_log_ptrs(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			pfirst,	/* index of first pointer to log */
-	int			plast)	/* index of last pointer to log */
+xfs_inobt_init_key_from_rec(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
 {
-	xfs_inobt_block_t	*block;	/* btree block to log from */
-	int			first;	/* first byte offset logged */
-	int			last;	/* last byte offset logged */
-	xfs_inobt_ptr_t		*pp;	/* block-pointer pointer in btree blk */
-
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-	pp = XFS_INOBT_PTR_ADDR(block, 1, cur);
-	first = (int)((xfs_caddr_t)&pp[pfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&pp[plast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+	key->inobt.ir_startino = rec->inobt.ir_startino;
 }
 
-/*
- * Log records from a btree block (leaf).
- */
 STATIC void
-xfs_inobt_log_recs(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_buf_t		*bp,	/* buffer containing btree block */
-	int			rfirst,	/* index of first record to log */
-	int			rlast)	/* index of last record to log */
+xfs_inobt_init_rec_from_key(
+	union xfs_btree_key	*key,
+	union xfs_btree_rec	*rec)
 {
-	xfs_inobt_block_t	*block;	/* btree block to log from */
-	int			first;	/* first byte offset logged */
-	int			last;	/* last byte offset logged */
-	xfs_inobt_rec_t		*rp;	/* record pointer for btree block */
+	rec->inobt.ir_startino = key->inobt.ir_startino;
+}
 
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-	rp = XFS_INOBT_REC_ADDR(block, 1, cur);
-	first = (int)((xfs_caddr_t)&rp[rfirst - 1] - (xfs_caddr_t)block);
-	last = (int)(((xfs_caddr_t)&rp[rlast] - 1) - (xfs_caddr_t)block);
-	xfs_trans_log_buf(cur->bc_tp, bp, first, last);
+STATIC void
+xfs_inobt_init_rec_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec)
+{
+	rec->inobt.ir_startino = cpu_to_be32(cur->bc_rec.i.ir_startino);
+	rec->inobt.ir_freecount = cpu_to_be32(cur->bc_rec.i.ir_freecount);
+	rec->inobt.ir_free = cpu_to_be64(cur->bc_rec.i.ir_free);
 }
 
 /*
- * Lookup the record.  The cursor is made to point to it, based on dir.
- * Return 0 if can't find any such record, 1 for success.
+ * intial value of ptr for lookup
  */
-STATIC int				/* error */
-xfs_inobt_lookup(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_lookup_t		dir,	/* <=, ==, or >= */
-	int			*stat)	/* success/failure */
+STATIC void
+xfs_inobt_init_ptr_from_cur(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
 {
-	xfs_agblock_t		agbno;	/* a.g. relative btree block number */
-	xfs_agnumber_t		agno;	/* allocation group number */
-	xfs_inobt_block_t	*block=NULL;	/* current btree block */
-	__int64_t		diff;	/* difference for the current key */
-	int			error;	/* error return value */
-	int			keyno=0;	/* current key number */
-	int			level;	/* level in the btree */
-	xfs_mount_t		*mp;	/* file system mount point */
-
-	/*
-	 * Get the allocation group header, and the root block number.
-	 */
-	mp = cur->bc_mp;
-	{
-		xfs_agi_t	*agi;	/* a.g. inode header */
-
-		agi = XFS_BUF_TO_AGI(cur->bc_private.a.agbp);
-		agno = be32_to_cpu(agi->agi_seqno);
-		agbno = be32_to_cpu(agi->agi_root);
-	}
-	/*
-	 * Iterate over each level in the btree, starting at the root.
-	 * For each level above the leaves, find the key we need, based
-	 * on the lookup record, then follow the corresponding block
-	 * pointer down to the next level.
-	 */
-	for (level = cur->bc_nlevels - 1, diff = 1; level >= 0; level--) {
-		xfs_buf_t	*bp;	/* buffer pointer for btree block */
-		xfs_daddr_t	d;	/* disk address of btree block */
-
-		/*
-		 * Get the disk address we're looking for.
-		 */
-		d = XFS_AGB_TO_DADDR(mp, agno, agbno);
-		/*
-		 * If the old buffer at this level is for a different block,
-		 * throw it away, otherwise just use it.
-		 */
-		bp = cur->bc_bufs[level];
-		if (bp && XFS_BUF_ADDR(bp) != d)
-			bp = NULL;
-		if (!bp) {
-			/*
-			 * Need to get a new buffer.  Read it, then
-			 * set it in the cursor, releasing the old one.
-			 */
-			if ((error = xfs_btree_read_bufs(mp, cur->bc_tp,
-					agno, agbno, 0, &bp, XFS_INO_BTREE_REF)))
-				return error;
-			xfs_btree_setbuf(cur, level, bp);
-			/*
-			 * Point to the btree block, now that we have the buffer
-			 */
-			block = XFS_BUF_TO_INOBT_BLOCK(bp);
-			if ((error = xfs_btree_check_sblock(cur, block, level,
-					bp)))
-				return error;
-		} else
-			block = XFS_BUF_TO_INOBT_BLOCK(bp);
-		/*
-		 * If we already had a key match at a higher level, we know
-		 * we need to use the first entry in this block.
-		 */
-		if (diff == 0)
-			keyno = 1;
-		/*
-		 * Otherwise we need to search this block.  Do a binary search.
-		 */
-		else {
-			int		high;	/* high entry number */
-			xfs_inobt_key_t	*kkbase=NULL;/* base of keys in block */
-			xfs_inobt_rec_t	*krbase=NULL;/* base of records in block */
-			int		low;	/* low entry number */
+	struct xfs_agi		*agi = XFS_BUF_TO_AGI(cur->bc_private.a.agbp);
 
-			/*
-			 * Get a pointer to keys or records.
-			 */
-			if (level > 0)
-				kkbase = XFS_INOBT_KEY_ADDR(block, 1, cur);
-			else
-				krbase = XFS_INOBT_REC_ADDR(block, 1, cur);
-			/*
-			 * Set low and high entry numbers, 1-based.
-			 */
-			low = 1;
-			if (!(high = be16_to_cpu(block->bb_numrecs))) {
-				/*
-				 * If the block is empty, the tree must
-				 * be an empty leaf.
-				 */
-				ASSERT(level == 0 && cur->bc_nlevels == 1);
-				cur->bc_ptrs[0] = dir != XFS_LOOKUP_LE;
-				*stat = 0;
-				return 0;
-			}
-			/*
-			 * Binary search the block.
-			 */
-			while (low <= high) {
-				xfs_agino_t	startino;	/* key value */
-
-				/*
-				 * keyno is average of low and high.
-				 */
-				keyno = (low + high) >> 1;
-				/*
-				 * Get startino.
-				 */
-				if (level > 0) {
-					xfs_inobt_key_t	*kkp;
-
-					kkp = kkbase + keyno - 1;
-					startino = be32_to_cpu(kkp->ir_startino);
-				} else {
-					xfs_inobt_rec_t	*krp;
-
-					krp = krbase + keyno - 1;
-					startino = be32_to_cpu(krp->ir_startino);
-				}
-				/*
-				 * Compute difference to get next direction.
-				 */
-				diff = (__int64_t)
-					startino - cur->bc_rec.i.ir_startino;
-				/*
-				 * Less than, move right.
-				 */
-				if (diff < 0)
-					low = keyno + 1;
-				/*
-				 * Greater than, move left.
-				 */
-				else if (diff > 0)
-					high = keyno - 1;
-				/*
-				 * Equal, we're done.
-				 */
-				else
-					break;
-			}
-		}
-		/*
-		 * If there are more levels, set up for the next level
-		 * by getting the block number and filling in the cursor.
-		 */
-		if (level > 0) {
-			/*
-			 * If we moved left, need the previous key number,
-			 * unless there isn't one.
-			 */
-			if (diff > 0 && --keyno < 1)
-				keyno = 1;
-			agbno = be32_to_cpu(*XFS_INOBT_PTR_ADDR(block, keyno, cur));
-#ifdef DEBUG
-			if ((error = xfs_btree_check_sptr(cur, agbno, level)))
-				return error;
-#endif
-			cur->bc_ptrs[level] = keyno;
-		}
-	}
-	/*
-	 * Done with the search.
-	 * See if we need to adjust the results.
-	 */
-	if (dir != XFS_LOOKUP_LE && diff < 0) {
-		keyno++;
-		/*
-		 * If ge search and we went off the end of the block, but it's
-		 * not the last block, we're in the wrong block.
-		 */
-		if (dir == XFS_LOOKUP_GE &&
-		    keyno > be16_to_cpu(block->bb_numrecs) &&
-		    be32_to_cpu(block->bb_rightsib) != NULLAGBLOCK) {
-			int	i;
+	ASSERT(cur->bc_private.a.agno == be32_to_cpu(agi->agi_seqno));
 
-			cur->bc_ptrs[0] = keyno;
-			if ((error = xfs_inobt_increment(cur, 0, &i)))
-				return error;
-			ASSERT(i == 1);
-			*stat = 1;
-			return 0;
-		}
-	}
-	else if (dir == XFS_LOOKUP_LE && diff > 0)
-		keyno--;
-	cur->bc_ptrs[0] = keyno;
-	/*
-	 * Return if we succeeded or not.
-	 */
-	if (keyno == 0 || keyno > be16_to_cpu(block->bb_numrecs))
-		*stat = 0;
-	else
-		*stat = ((dir != XFS_LOOKUP_EQ) || (diff == 0));
-	return 0;
+	ptr->s = agi->agi_root;
 }
 
-/*
- * Move 1 record left from cur/level if possible.
- * Update cur to reflect the new path.
- */
-STATIC int				/* error */
-xfs_inobt_lshift(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to shift record on */
-	int			*stat)	/* success/failure */
+STATIC __int64_t
+xfs_inobt_key_diff(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key)
 {
-	int			error;	/* error return value */
-#ifdef DEBUG
-	int			i;	/* loop index */
-#endif
-	xfs_inobt_key_t		key;	/* key value for leaf level upward */
-	xfs_buf_t		*lbp;	/* buffer for left neighbor block */
-	xfs_inobt_block_t	*left;	/* left neighbor btree block */
-	xfs_inobt_key_t		*lkp=NULL;	/* key pointer for left block */
-	xfs_inobt_ptr_t		*lpp;	/* address pointer for left block */
-	xfs_inobt_rec_t		*lrp=NULL;	/* record pointer for left block */
-	int			nrec;	/* new number of left block entries */
-	xfs_buf_t		*rbp;	/* buffer for right (current) block */
-	xfs_inobt_block_t	*right;	/* right (current) btree block */
-	xfs_inobt_key_t		*rkp=NULL;	/* key pointer for right block */
-	xfs_inobt_ptr_t		*rpp=NULL;	/* address pointer for right block */
-	xfs_inobt_rec_t		*rrp=NULL;	/* record pointer for right block */
-
-	/*
-	 * Set up variables for this block as "right".
-	 */
-	rbp = cur->bc_bufs[level];
-	right = XFS_BUF_TO_INOBT_BLOCK(rbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-		return error;
-#endif
-	/*
-	 * If we've got no left sibling then we can't shift an entry left.
-	 */
-	if (be32_to_cpu(right->bb_leftsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * If the cursor entry is the one that would be moved, don't
-	 * do it... it's too complicated.
-	 */
-	if (cur->bc_ptrs[level] <= 1) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Set up the left neighbor as "left".
-	 */
-	if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-			cur->bc_private.a.agno, be32_to_cpu(right->bb_leftsib),
-			0, &lbp, XFS_INO_BTREE_REF)))
-		return error;
-	left = XFS_BUF_TO_INOBT_BLOCK(lbp);
-	if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-		return error;
-	/*
-	 * If it's full, it can't take another entry.
-	 */
-	if (be16_to_cpu(left->bb_numrecs) == XFS_INOBT_BLOCK_MAXRECS(level, cur)) {
-		*stat = 0;
-		return 0;
-	}
-	nrec = be16_to_cpu(left->bb_numrecs) + 1;
-	/*
-	 * If non-leaf, copy a key and a ptr to the left block.
-	 */
-	if (level > 0) {
-		lkp = XFS_INOBT_KEY_ADDR(left, nrec, cur);
-		rkp = XFS_INOBT_KEY_ADDR(right, 1, cur);
-		*lkp = *rkp;
-		xfs_inobt_log_keys(cur, lbp, nrec, nrec);
-		lpp = XFS_INOBT_PTR_ADDR(left, nrec, cur);
-		rpp = XFS_INOBT_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*rpp), level)))
-			return error;
-#endif
-		*lpp = *rpp;
-		xfs_inobt_log_ptrs(cur, lbp, nrec, nrec);
-	}
-	/*
-	 * If leaf, copy a record to the left block.
-	 */
-	else {
-		lrp = XFS_INOBT_REC_ADDR(left, nrec, cur);
-		rrp = XFS_INOBT_REC_ADDR(right, 1, cur);
-		*lrp = *rrp;
-		xfs_inobt_log_recs(cur, lbp, nrec, nrec);
-	}
-	/*
-	 * Bump and log left's numrecs, decrement and log right's numrecs.
-	 */
-	be16_add_cpu(&left->bb_numrecs, 1);
-	xfs_inobt_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS);
-#ifdef DEBUG
-	if (level > 0)
-		xfs_btree_check_key(cur->bc_btnum, lkp - 1, lkp);
-	else
-		xfs_btree_check_rec(cur->bc_btnum, lrp - 1, lrp);
-#endif
-	be16_add_cpu(&right->bb_numrecs, -1);
-	xfs_inobt_log_block(cur->bc_tp, rbp, XFS_BB_NUMRECS);
-	/*
-	 * Slide the contents of right down one entry.
-	 */
-	if (level > 0) {
-#ifdef DEBUG
-		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i + 1]),
-					level)))
-				return error;
-		}
-#endif
-		memmove(rkp, rkp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memmove(rpp, rpp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-		xfs_inobt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		xfs_inobt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-	} else {
-		memmove(rrp, rrp + 1, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		xfs_inobt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		key.ir_startino = rrp->ir_startino;
-		rkp = &key;
-	}
-	/*
-	 * Update the parent key values of right.
-	 */
-	if ((error = xfs_inobt_updkey(cur, rkp, level + 1)))
-		return error;
-	/*
-	 * Slide the cursor value left one.
-	 */
-	cur->bc_ptrs[level]--;
-	*stat = 1;
-	return 0;
+	return (__int64_t)be32_to_cpu(key->inobt.ir_startino) -
+			  cur->bc_rec.i.ir_startino;
 }
 
-/*
- * Allocate a new root block, fill it in.
- */
-STATIC int				/* error */
-xfs_inobt_newroot(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			*stat)	/* success/failure */
+STATIC int
+xfs_inobt_kill_root(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	int			level,
+	union xfs_btree_ptr	*newroot)
 {
-	xfs_agi_t		*agi;	/* a.g. inode header */
-	xfs_alloc_arg_t		args;	/* allocation argument structure */
-	xfs_inobt_block_t	*block;	/* one half of the old root block */
-	xfs_buf_t		*bp;	/* buffer containing block */
-	int			error;	/* error return value */
-	xfs_inobt_key_t		*kp;	/* btree key pointer */
-	xfs_agblock_t		lbno;	/* left block number */
-	xfs_buf_t		*lbp;	/* left buffer pointer */
-	xfs_inobt_block_t	*left;	/* left btree block */
-	xfs_buf_t		*nbp;	/* new (root) buffer */
-	xfs_inobt_block_t	*new;	/* new (root) btree block */
-	int			nptr;	/* new value for key index, 1 or 2 */
-	xfs_inobt_ptr_t		*pp;	/* btree address pointer */
-	xfs_agblock_t		rbno;	/* right block number */
-	xfs_buf_t		*rbp;	/* right buffer pointer */
-	xfs_inobt_block_t	*right;	/* right btree block */
-	xfs_inobt_rec_t		*rp;	/* btree record pointer */
+	int			error;
 
-	ASSERT(cur->bc_nlevels < XFS_IN_MAXLEVELS(cur->bc_mp));
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
+	XFS_BTREE_STATS_INC(cur, killroot);
 
 	/*
-	 * Get a block & a buffer.
+	 * Update the root pointer, decreasing the level by 1 and then
+	 * free the old root.
 	 */
-	agi = XFS_BUF_TO_AGI(cur->bc_private.a.agbp);
-	args.tp = cur->bc_tp;
-	args.mp = cur->bc_mp;
-	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_private.a.agno,
-		be32_to_cpu(agi->agi_root));
-	args.mod = args.minleft = args.alignment = args.total = args.wasdel =
-		args.isfl = args.userdata = args.minalignslop = 0;
-	args.minlen = args.maxlen = args.prod = 1;
-	args.type = XFS_ALLOCTYPE_NEAR_BNO;
-	if ((error = xfs_alloc_vextent(&args)))
+	xfs_inobt_set_root(cur, newroot, -1);
+	error = xfs_inobt_free_block(cur, bp);
+	if (error) {
+		XFS_BTREE_TRACE_CURSOR(cur, XBT_ERROR);
 		return error;
-	/*
-	 * None available, we fail.
-	 */
-	if (args.fsbno == NULLFSBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	ASSERT(args.len == 1);
-	nbp = xfs_btree_get_bufs(args.mp, args.tp, args.agno, args.agbno, 0);
-	new = XFS_BUF_TO_INOBT_BLOCK(nbp);
-	/*
-	 * Set the root data in the a.g. inode structure.
-	 */
-	agi->agi_root = cpu_to_be32(args.agbno);
-	be32_add_cpu(&agi->agi_level, 1);
-	xfs_ialloc_log_agi(args.tp, cur->bc_private.a.agbp,
-		XFS_AGI_ROOT | XFS_AGI_LEVEL);
-	/*
-	 * At the previous root level there are now two blocks: the old
-	 * root, and the new block generated when it was split.
-	 * We don't know which one the cursor is pointing at, so we
-	 * set up variables "left" and "right" for each case.
-	 */
-	bp = cur->bc_bufs[cur->bc_nlevels - 1];
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, cur->bc_nlevels - 1, bp)))
-		return error;
-#endif
-	if (be32_to_cpu(block->bb_rightsib) != NULLAGBLOCK) {
-		/*
-		 * Our block is left, pick up the right block.
-		 */
-		lbp = bp;
-		lbno = XFS_DADDR_TO_AGBNO(args.mp, XFS_BUF_ADDR(lbp));
-		left = block;
-		rbno = be32_to_cpu(left->bb_rightsib);
-		if ((error = xfs_btree_read_bufs(args.mp, args.tp, args.agno,
-				rbno, 0, &rbp, XFS_INO_BTREE_REF)))
-			return error;
-		bp = rbp;
-		right = XFS_BUF_TO_INOBT_BLOCK(rbp);
-		if ((error = xfs_btree_check_sblock(cur, right,
-				cur->bc_nlevels - 1, rbp)))
-			return error;
-		nptr = 1;
-	} else {
-		/*
-		 * Our block is right, pick up the left block.
-		 */
-		rbp = bp;
-		rbno = XFS_DADDR_TO_AGBNO(args.mp, XFS_BUF_ADDR(rbp));
-		right = block;
-		lbno = be32_to_cpu(right->bb_leftsib);
-		if ((error = xfs_btree_read_bufs(args.mp, args.tp, args.agno,
-				lbno, 0, &lbp, XFS_INO_BTREE_REF)))
-			return error;
-		bp = lbp;
-		left = XFS_BUF_TO_INOBT_BLOCK(lbp);
-		if ((error = xfs_btree_check_sblock(cur, left,
-				cur->bc_nlevels - 1, lbp)))
-			return error;
-		nptr = 2;
-	}
-	/*
-	 * Fill in the new block's btree header and log it.
-	 */
-	new->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
-	new->bb_level = cpu_to_be16(cur->bc_nlevels);
-	new->bb_numrecs = cpu_to_be16(2);
-	new->bb_leftsib = cpu_to_be32(NULLAGBLOCK);
-	new->bb_rightsib = cpu_to_be32(NULLAGBLOCK);
-	xfs_inobt_log_block(args.tp, nbp, XFS_BB_ALL_BITS);
-	ASSERT(lbno != NULLAGBLOCK && rbno != NULLAGBLOCK);
-	/*
-	 * Fill in the key data in the new root.
-	 */
-	kp = XFS_INOBT_KEY_ADDR(new, 1, cur);
-	if (be16_to_cpu(left->bb_level) > 0) {
-		kp[0] = *XFS_INOBT_KEY_ADDR(left, 1, cur);
-		kp[1] = *XFS_INOBT_KEY_ADDR(right, 1, cur);
-	} else {
-		rp = XFS_INOBT_REC_ADDR(left, 1, cur);
-		kp[0].ir_startino = rp->ir_startino;
-		rp = XFS_INOBT_REC_ADDR(right, 1, cur);
-		kp[1].ir_startino = rp->ir_startino;
 	}
-	xfs_inobt_log_keys(cur, nbp, 1, 2);
-	/*
-	 * Fill in the pointer data in the new root.
-	 */
-	pp = XFS_INOBT_PTR_ADDR(new, 1, cur);
-	pp[0] = cpu_to_be32(lbno);
-	pp[1] = cpu_to_be32(rbno);
-	xfs_inobt_log_ptrs(cur, nbp, 1, 2);
-	/*
-	 * Fix up the cursor.
-	 */
-	xfs_btree_setbuf(cur, cur->bc_nlevels, nbp);
-	cur->bc_ptrs[cur->bc_nlevels] = nptr;
-	cur->bc_nlevels++;
-	*stat = 1;
-	return 0;
-}
 
-/*
- * Move 1 record right from cur/level if possible.
- * Update cur to reflect the new path.
- */
-STATIC int				/* error */
-xfs_inobt_rshift(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to shift record on */
-	int			*stat)	/* success/failure */
-{
-	int			error;	/* error return value */
-	int			i;	/* loop index */
-	xfs_inobt_key_t		key;	/* key value for leaf level upward */
-	xfs_buf_t		*lbp;	/* buffer for left (current) block */
-	xfs_inobt_block_t	*left;	/* left (current) btree block */
-	xfs_inobt_key_t		*lkp;	/* key pointer for left block */
-	xfs_inobt_ptr_t		*lpp;	/* address pointer for left block */
-	xfs_inobt_rec_t		*lrp;	/* record pointer for left block */
-	xfs_buf_t		*rbp;	/* buffer for right neighbor block */
-	xfs_inobt_block_t	*right;	/* right neighbor btree block */
-	xfs_inobt_key_t		*rkp;	/* key pointer for right block */
-	xfs_inobt_ptr_t		*rpp;	/* address pointer for right block */
-	xfs_inobt_rec_t		*rrp=NULL;	/* record pointer for right block */
-	xfs_btree_cur_t		*tcur;	/* temporary cursor */
+	XFS_BTREE_STATS_INC(cur, free);
 
-	/*
-	 * Set up variables for this block as "left".
-	 */
-	lbp = cur->bc_bufs[level];
-	left = XFS_BUF_TO_INOBT_BLOCK(lbp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-		return error;
-#endif
-	/*
-	 * If we've got no right sibling then we can't shift an entry right.
-	 */
-	if (be32_to_cpu(left->bb_rightsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * If the cursor entry is the one that would be moved, don't
-	 * do it... it's too complicated.
-	 */
-	if (cur->bc_ptrs[level] >= be16_to_cpu(left->bb_numrecs)) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Set up the right neighbor as "right".
-	 */
-	if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-			cur->bc_private.a.agno, be32_to_cpu(left->bb_rightsib),
-			0, &rbp, XFS_INO_BTREE_REF)))
-		return error;
-	right = XFS_BUF_TO_INOBT_BLOCK(rbp);
-	if ((error = xfs_btree_check_sblock(cur, right, level, rbp)))
-		return error;
-	/*
-	 * If it's full, it can't take another entry.
-	 */
-	if (be16_to_cpu(right->bb_numrecs) == XFS_INOBT_BLOCK_MAXRECS(level, cur)) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Make a hole at the start of the right neighbor block, then
-	 * copy the last left block entry to the hole.
-	 */
-	if (level > 0) {
-		lkp = XFS_INOBT_KEY_ADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		lpp = XFS_INOBT_PTR_ADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		rkp = XFS_INOBT_KEY_ADDR(right, 1, cur);
-		rpp = XFS_INOBT_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = be16_to_cpu(right->bb_numrecs) - 1; i >= 0; i--) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(rpp[i]), level)))
-				return error;
-		}
-#endif
-		memmove(rkp + 1, rkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memmove(rpp + 1, rpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(*lpp), level)))
-			return error;
-#endif
-		*rkp = *lkp;
-		*rpp = *lpp;
-		xfs_inobt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		xfs_inobt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-	} else {
-		lrp = XFS_INOBT_REC_ADDR(left, be16_to_cpu(left->bb_numrecs), cur);
-		rrp = XFS_INOBT_REC_ADDR(right, 1, cur);
-		memmove(rrp + 1, rrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		*rrp = *lrp;
-		xfs_inobt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs) + 1);
-		key.ir_startino = rrp->ir_startino;
-		rkp = &key;
-	}
-	/*
-	 * Decrement and log left's numrecs, bump and log right's numrecs.
-	 */
-	be16_add_cpu(&left->bb_numrecs, -1);
-	xfs_inobt_log_block(cur->bc_tp, lbp, XFS_BB_NUMRECS);
-	be16_add_cpu(&right->bb_numrecs, 1);
-#ifdef DEBUG
-	if (level > 0)
-		xfs_btree_check_key(cur->bc_btnum, rkp, rkp + 1);
-	else
-		xfs_btree_check_rec(cur->bc_btnum, rrp, rrp + 1);
-#endif
-	xfs_inobt_log_block(cur->bc_tp, rbp, XFS_BB_NUMRECS);
-	/*
-	 * Using a temporary cursor, update the parent key values of the
-	 * block on the right.
-	 */
-	if ((error = xfs_btree_dup_cursor(cur, &tcur)))
-		return error;
-	xfs_btree_lastrec(tcur, level);
-	if ((error = xfs_inobt_increment(tcur, level, &i)) ||
-	    (error = xfs_inobt_updkey(tcur, rkp, level + 1))) {
-		xfs_btree_del_cursor(tcur, XFS_BTREE_ERROR);
-		return error;
-	}
-	xfs_btree_del_cursor(tcur, XFS_BTREE_NOERROR);
-	*stat = 1;
+	cur->bc_bufs[level] = NULL;
+	cur->bc_nlevels--;
+
+	XFS_BTREE_TRACE_CURSOR(cur, XBT_EXIT);
 	return 0;
 }
 
-/*
- * Split cur/level block in half.
- * Return new block number and its first record (to be inserted into parent).
- */
-STATIC int				/* error */
-xfs_inobt_split(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level to split */
-	xfs_agblock_t		*bnop,	/* output: block number allocated */
-	xfs_inobt_key_t		*keyp,	/* output: first key of new block */
-	xfs_btree_cur_t		**curp,	/* output: new cursor */
-	int			*stat)	/* success/failure */
-{
-	xfs_alloc_arg_t		args;	/* allocation argument structure */
-	int			error;	/* error return value */
-	int			i;	/* loop index/record number */
-	xfs_agblock_t		lbno;	/* left (current) block number */
-	xfs_buf_t		*lbp;	/* buffer for left block */
-	xfs_inobt_block_t	*left;	/* left (current) btree block */
-	xfs_inobt_key_t		*lkp;	/* left btree key pointer */
-	xfs_inobt_ptr_t		*lpp;	/* left btree address pointer */
-	xfs_inobt_rec_t		*lrp;	/* left btree record pointer */
-	xfs_buf_t		*rbp;	/* buffer for right block */
-	xfs_inobt_block_t	*right;	/* right (new) btree block */
-	xfs_inobt_key_t		*rkp;	/* right btree key pointer */
-	xfs_inobt_ptr_t		*rpp;	/* right btree address pointer */
-	xfs_inobt_rec_t		*rrp;	/* right btree record pointer */
-
-	/*
-	 * Set up left block (current one).
-	 */
-	lbp = cur->bc_bufs[level];
-	args.tp = cur->bc_tp;
-	args.mp = cur->bc_mp;
-	lbno = XFS_DADDR_TO_AGBNO(args.mp, XFS_BUF_ADDR(lbp));
-	/*
-	 * Allocate the new block.
-	 * If we can't do it, we're toast.  Give up.
-	 */
-	args.fsbno = XFS_AGB_TO_FSB(args.mp, cur->bc_private.a.agno, lbno);
-	args.mod = args.minleft = args.alignment = args.total = args.wasdel =
-		args.isfl = args.userdata = args.minalignslop = 0;
-	args.minlen = args.maxlen = args.prod = 1;
-	args.type = XFS_ALLOCTYPE_NEAR_BNO;
-	if ((error = xfs_alloc_vextent(&args)))
-		return error;
-	if (args.fsbno == NULLFSBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	ASSERT(args.len == 1);
-	rbp = xfs_btree_get_bufs(args.mp, args.tp, args.agno, args.agbno, 0);
-	/*
-	 * Set up the new block as "right".
-	 */
-	right = XFS_BUF_TO_INOBT_BLOCK(rbp);
-	/*
-	 * "Left" is the current (according to the cursor) block.
-	 */
-	left = XFS_BUF_TO_INOBT_BLOCK(lbp);
 #ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, left, level, lbp)))
-		return error;
-#endif
-	/*
-	 * Fill in the btree header for the new block.
-	 */
-	right->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
-	right->bb_level = left->bb_level;
-	right->bb_numrecs = cpu_to_be16(be16_to_cpu(left->bb_numrecs) / 2);
-	/*
-	 * Make sure that if there's an odd number of entries now, that
-	 * each new block will have the same number of entries.
-	 */
-	if ((be16_to_cpu(left->bb_numrecs) & 1) &&
-	    cur->bc_ptrs[level] <= be16_to_cpu(right->bb_numrecs) + 1)
-		be16_add_cpu(&right->bb_numrecs, 1);
-	i = be16_to_cpu(left->bb_numrecs) - be16_to_cpu(right->bb_numrecs) + 1;
-	/*
-	 * For non-leaf blocks, copy keys and addresses over to the new block.
-	 */
-	if (level > 0) {
-		lkp = XFS_INOBT_KEY_ADDR(left, i, cur);
-		lpp = XFS_INOBT_PTR_ADDR(left, i, cur);
-		rkp = XFS_INOBT_KEY_ADDR(right, 1, cur);
-		rpp = XFS_INOBT_PTR_ADDR(right, 1, cur);
-#ifdef DEBUG
-		for (i = 0; i < be16_to_cpu(right->bb_numrecs); i++) {
-			if ((error = xfs_btree_check_sptr(cur, be32_to_cpu(lpp[i]), level)))
-				return error;
-		}
-#endif
-		memcpy(rkp, lkp, be16_to_cpu(right->bb_numrecs) * sizeof(*rkp));
-		memcpy(rpp, lpp, be16_to_cpu(right->bb_numrecs) * sizeof(*rpp));
-		xfs_inobt_log_keys(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		xfs_inobt_log_ptrs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		*keyp = *rkp;
-	}
-	/*
-	 * For leaf blocks, copy records over to the new block.
-	 */
-	else {
-		lrp = XFS_INOBT_REC_ADDR(left, i, cur);
-		rrp = XFS_INOBT_REC_ADDR(right, 1, cur);
-		memcpy(rrp, lrp, be16_to_cpu(right->bb_numrecs) * sizeof(*rrp));
-		xfs_inobt_log_recs(cur, rbp, 1, be16_to_cpu(right->bb_numrecs));
-		keyp->ir_startino = rrp->ir_startino;
-	}
-	/*
-	 * Find the left block number by looking in the buffer.
-	 * Adjust numrecs, sibling pointers.
-	 */
-	be16_add_cpu(&left->bb_numrecs, -(be16_to_cpu(right->bb_numrecs)));
-	right->bb_rightsib = left->bb_rightsib;
-	left->bb_rightsib = cpu_to_be32(args.agbno);
-	right->bb_leftsib = cpu_to_be32(lbno);
-	xfs_inobt_log_block(args.tp, rbp, XFS_BB_ALL_BITS);
-	xfs_inobt_log_block(args.tp, lbp, XFS_BB_NUMRECS | XFS_BB_RIGHTSIB);
-	/*
-	 * If there's a block to the new block's right, make that block
-	 * point back to right instead of to left.
-	 */
-	if (be32_to_cpu(right->bb_rightsib) != NULLAGBLOCK) {
-		xfs_inobt_block_t	*rrblock;	/* rr btree block */
-		xfs_buf_t		*rrbp;		/* buffer for rrblock */
-
-		if ((error = xfs_btree_read_bufs(args.mp, args.tp, args.agno,
-				be32_to_cpu(right->bb_rightsib), 0, &rrbp,
-				XFS_INO_BTREE_REF)))
-			return error;
-		rrblock = XFS_BUF_TO_INOBT_BLOCK(rrbp);
-		if ((error = xfs_btree_check_sblock(cur, rrblock, level, rrbp)))
-			return error;
-		rrblock->bb_leftsib = cpu_to_be32(args.agbno);
-		xfs_inobt_log_block(args.tp, rrbp, XFS_BB_LEFTSIB);
-	}
-	/*
-	 * If the cursor is really in the right block, move it there.
-	 * If it's just pointing past the last entry in left, then we'll
-	 * insert there, so don't change anything in that case.
-	 */
-	if (cur->bc_ptrs[level] > be16_to_cpu(left->bb_numrecs) + 1) {
-		xfs_btree_setbuf(cur, level, rbp);
-		cur->bc_ptrs[level] -= be16_to_cpu(left->bb_numrecs);
-	}
-	/*
-	 * If there are more levels, we'll need another cursor which refers
-	 * the right block, no matter where this cursor was.
-	 */
-	if (level + 1 < cur->bc_nlevels) {
-		if ((error = xfs_btree_dup_cursor(cur, curp)))
-			return error;
-		(*curp)->bc_ptrs[level + 1]++;
-	}
-	*bnop = args.agbno;
-	*stat = 1;
-	return 0;
+STATIC int
+xfs_inobt_keys_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*k1,
+	union xfs_btree_key	*k2)
+{
+	return be32_to_cpu(k1->inobt.ir_startino) <
+		be32_to_cpu(k2->inobt.ir_startino);
 }
 
-/*
- * Update keys at all levels from here to the root along the cursor's path.
- */
-STATIC int				/* error */
-xfs_inobt_updkey(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_inobt_key_t		*keyp,	/* new key value to update to */
-	int			level)	/* starting level for update */
+STATIC int
+xfs_inobt_recs_inorder(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*r1,
+	union xfs_btree_rec	*r2)
 {
-	int			ptr;	/* index of key in block */
-
-	/*
-	 * Go up the tree from this level toward the root.
-	 * At each level, update the key value to the value input.
-	 * Stop when we reach a level where the cursor isn't pointing
-	 * at the first entry in the block.
-	 */
-	for (ptr = 1; ptr == 1 && level < cur->bc_nlevels; level++) {
-		xfs_buf_t		*bp;	/* buffer for block */
-		xfs_inobt_block_t	*block;	/* btree block */
-#ifdef DEBUG
-		int			error;	/* error return value */
-#endif
-		xfs_inobt_key_t		*kp;	/* ptr to btree block keys */
-
-		bp = cur->bc_bufs[level];
-		block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-			return error;
-#endif
-		ptr = cur->bc_ptrs[level];
-		kp = XFS_INOBT_KEY_ADDR(block, ptr, cur);
-		*kp = *keyp;
-		xfs_inobt_log_keys(cur, bp, ptr, ptr);
-	}
-	return 0;
+	return be32_to_cpu(r1->inobt.ir_startino) + XFS_INODES_PER_CHUNK <=
+		be32_to_cpu(r2->inobt.ir_startino);
 }
+#endif	/* DEBUG */
 
-/*
- * Externally visible routines.
- */
+#ifdef XFS_BTREE_TRACE
+ktrace_t	*xfs_inobt_trace_buf;
 
-/*
- * Decrement cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-int					/* error */
-xfs_inobt_decrement(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level in btree, 0 is leaf */
-	int			*stat)	/* success/failure */
+STATIC void
+xfs_inobt_trace_enter(
+	struct xfs_btree_cur	*cur,
+	const char		*func,
+	char			*s,
+	int			type,
+	int			line,
+	__psunsigned_t		a0,
+	__psunsigned_t		a1,
+	__psunsigned_t		a2,
+	__psunsigned_t		a3,
+	__psunsigned_t		a4,
+	__psunsigned_t		a5,
+	__psunsigned_t		a6,
+	__psunsigned_t		a7,
+	__psunsigned_t		a8,
+	__psunsigned_t		a9,
+	__psunsigned_t		a10)
 {
-	xfs_inobt_block_t	*block;	/* btree block */
-	int			error;
-	int			lev;	/* btree level */
-
-	ASSERT(level < cur->bc_nlevels);
-	/*
-	 * Read-ahead to the left at this level.
-	 */
-	xfs_btree_readahead(cur, level, XFS_BTCUR_LEFTRA);
-	/*
-	 * Decrement the ptr at this level.  If we're still in the block
-	 * then we're done.
-	 */
-	if (--cur->bc_ptrs[level] > 0) {
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * Get a pointer to the btree block.
-	 */
-	block = XFS_BUF_TO_INOBT_BLOCK(cur->bc_bufs[level]);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level,
-			cur->bc_bufs[level])))
-		return error;
-#endif
-	/*
-	 * If we just went off the left edge of the tree, return failure.
-	 */
-	if (be32_to_cpu(block->bb_leftsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * March up the tree decrementing pointers.
-	 * Stop when we don't go off the left edge of a block.
-	 */
-	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
-		if (--cur->bc_ptrs[lev] > 0)
-			break;
-		/*
-		 * Read-ahead the left block, we're going to read it
-		 * in the next loop.
-		 */
-		xfs_btree_readahead(cur, lev, XFS_BTCUR_LEFTRA);
-	}
-	/*
-	 * If we went off the root then we are seriously confused.
-	 */
-	ASSERT(lev < cur->bc_nlevels);
-	/*
-	 * Now walk back down the tree, fixing up the cursor's buffer
-	 * pointers and key numbers.
-	 */
-	for (block = XFS_BUF_TO_INOBT_BLOCK(cur->bc_bufs[lev]); lev > level; ) {
-		xfs_agblock_t	agbno;	/* block number of btree block */
-		xfs_buf_t	*bp;	/* buffer containing btree block */
-
-		agbno = be32_to_cpu(*XFS_INOBT_PTR_ADDR(block, cur->bc_ptrs[lev], cur));
-		if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-				cur->bc_private.a.agno, agbno, 0, &bp,
-				XFS_INO_BTREE_REF)))
-			return error;
-		lev--;
-		xfs_btree_setbuf(cur, lev, bp);
-		block = XFS_BUF_TO_INOBT_BLOCK(bp);
-		if ((error = xfs_btree_check_sblock(cur, block, lev, bp)))
-			return error;
-		cur->bc_ptrs[lev] = be16_to_cpu(block->bb_numrecs);
-	}
-	*stat = 1;
-	return 0;
+	ktrace_enter(xfs_inobt_trace_buf, (void *)(__psint_t)type,
+		(void *)func, (void *)s, NULL, (void *)cur,
+		(void *)a0, (void *)a1, (void *)a2, (void *)a3,
+		(void *)a4, (void *)a5, (void *)a6, (void *)a7,
+		(void *)a8, (void *)a9, (void *)a10);
 }
 
-/*
- * Delete the record pointed to by cur.
- * The cursor refers to the place where the record was (could be inserted)
- * when the operation returns.
- */
-int					/* error */
-xfs_inobt_delete(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	int		*stat)		/* success/failure */
+STATIC void
+xfs_inobt_trace_cursor(
+	struct xfs_btree_cur	*cur,
+	__uint32_t		*s0,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
 {
-	int		error;
-	int		i;		/* result code */
-	int		level;		/* btree level */
-
-	/*
-	 * Go up the tree, starting at leaf level.
-	 * If 2 is returned then a join was done; go to the next level.
-	 * Otherwise we are done.
-	 */
-	for (level = 0, i = 2; i == 2; level++) {
-		if ((error = xfs_inobt_delrec(cur, level, &i)))
-			return error;
-	}
-	if (i == 0) {
-		for (level = 1; level < cur->bc_nlevels; level++) {
-			if (cur->bc_ptrs[level] == 0) {
-				if ((error = xfs_inobt_decrement(cur, level, &i)))
-					return error;
-				break;
-			}
-		}
-	}
-	*stat = i;
-	return 0;
+	*s0 = cur->bc_private.a.agno;
+	*l0 = cur->bc_rec.i.ir_startino;
+	*l1 = cur->bc_rec.i.ir_free;
 }
 
-
-/*
- * Get the data from the pointed-to record.
- */
-int					/* error */
-xfs_inobt_get_rec(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_agino_t		*ino,	/* output: starting inode of chunk */
-	__int32_t		*fcnt,	/* output: number of free inodes */
-	xfs_inofree_t		*free,	/* output: free inode mask */
-	int			*stat)	/* output: success/failure */
+STATIC void
+xfs_inobt_trace_key(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_key	*key,
+	__uint64_t		*l0,
+	__uint64_t		*l1)
 {
-	xfs_inobt_block_t	*block;	/* btree block */
-	xfs_buf_t		*bp;	/* buffer containing btree block */
-#ifdef DEBUG
-	int			error;	/* error return value */
-#endif
-	int			ptr;	/* record number */
-	xfs_inobt_rec_t		*rec;	/* record data */
-
-	bp = cur->bc_bufs[0];
-	ptr = cur->bc_ptrs[0];
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, 0, bp)))
-		return error;
-#endif
-	/*
-	 * Off the right end or left end, return failure.
-	 */
-	if (ptr > be16_to_cpu(block->bb_numrecs) || ptr <= 0) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * Point to the record and extract its data.
-	 */
-	rec = XFS_INOBT_REC_ADDR(block, ptr, cur);
-	*ino = be32_to_cpu(rec->ir_startino);
-	*fcnt = be32_to_cpu(rec->ir_freecount);
-	*free = be64_to_cpu(rec->ir_free);
-	*stat = 1;
-	return 0;
+	*l0 = be32_to_cpu(key->inobt.ir_startino);
+	*l1 = 0;
 }
 
-/*
- * Increment cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-int					/* error */
-xfs_inobt_increment(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	int			level,	/* level in btree, 0 is leaf */
-	int			*stat)	/* success/failure */
+STATIC void
+xfs_inobt_trace_record(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_rec	*rec,
+	__uint64_t		*l0,
+	__uint64_t		*l1,
+	__uint64_t		*l2)
 {
-	xfs_inobt_block_t	*block;	/* btree block */
-	xfs_buf_t		*bp;	/* buffer containing btree block */
-	int			error;	/* error return value */
-	int			lev;	/* btree level */
+	*l0 = be32_to_cpu(rec->inobt.ir_startino);
+	*l1 = be32_to_cpu(rec->inobt.ir_freecount);
+	*l2 = be64_to_cpu(rec->inobt.ir_free);
+}
+#endif /* XFS_BTREE_TRACE */
+
+static const struct xfs_btree_ops xfs_inobt_ops = {
+	.rec_len		= sizeof(xfs_inobt_rec_t),
+	.key_len		= sizeof(xfs_inobt_key_t),
+
+	.dup_cursor		= xfs_inobt_dup_cursor,
+	.set_root		= xfs_inobt_set_root,
+	.kill_root		= xfs_inobt_kill_root,
+	.alloc_block		= xfs_inobt_alloc_block,
+	.free_block		= xfs_inobt_free_block,
+	.get_minrecs		= xfs_inobt_get_minrecs,
+	.get_maxrecs		= xfs_inobt_get_maxrecs,
+	.init_key_from_rec	= xfs_inobt_init_key_from_rec,
+	.init_rec_from_key	= xfs_inobt_init_rec_from_key,
+	.init_rec_from_cur	= xfs_inobt_init_rec_from_cur,
+	.init_ptr_from_cur	= xfs_inobt_init_ptr_from_cur,
+	.key_diff		= xfs_inobt_key_diff,
 
-	ASSERT(level < cur->bc_nlevels);
-	/*
-	 * Read-ahead to the right at this level.
-	 */
-	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
-	/*
-	 * Get a pointer to the btree block.
-	 */
-	bp = cur->bc_bufs[level];
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, level, bp)))
-		return error;
-#endif
-	/*
-	 * Increment the ptr at this level.  If we're still in the block
-	 * then we're done.
-	 */
-	if (++cur->bc_ptrs[level] <= be16_to_cpu(block->bb_numrecs)) {
-		*stat = 1;
-		return 0;
-	}
-	/*
-	 * If we just went off the right edge of the tree, return failure.
-	 */
-	if (be32_to_cpu(block->bb_rightsib) == NULLAGBLOCK) {
-		*stat = 0;
-		return 0;
-	}
-	/*
-	 * March up the tree incrementing pointers.
-	 * Stop when we don't go off the right edge of a block.
-	 */
-	for (lev = level + 1; lev < cur->bc_nlevels; lev++) {
-		bp = cur->bc_bufs[lev];
-		block = XFS_BUF_TO_INOBT_BLOCK(bp);
 #ifdef DEBUG
-		if ((error = xfs_btree_check_sblock(cur, block, lev, bp)))
-			return error;
+	.keys_inorder		= xfs_inobt_keys_inorder,
+	.recs_inorder		= xfs_inobt_recs_inorder,
 #endif
-		if (++cur->bc_ptrs[lev] <= be16_to_cpu(block->bb_numrecs))
-			break;
-		/*
-		 * Read-ahead the right block, we're going to read it
-		 * in the next loop.
-		 */
-		xfs_btree_readahead(cur, lev, XFS_BTCUR_RIGHTRA);
-	}
-	/*
-	 * If we went off the root then we are seriously confused.
-	 */
-	ASSERT(lev < cur->bc_nlevels);
-	/*
-	 * Now walk back down the tree, fixing up the cursor's buffer
-	 * pointers and key numbers.
-	 */
-	for (bp = cur->bc_bufs[lev], block = XFS_BUF_TO_INOBT_BLOCK(bp);
-	     lev > level; ) {
-		xfs_agblock_t	agbno;	/* block number of btree block */
 
-		agbno = be32_to_cpu(*XFS_INOBT_PTR_ADDR(block, cur->bc_ptrs[lev], cur));
-		if ((error = xfs_btree_read_bufs(cur->bc_mp, cur->bc_tp,
-				cur->bc_private.a.agno, agbno, 0, &bp,
-				XFS_INO_BTREE_REF)))
-			return error;
-		lev--;
-		xfs_btree_setbuf(cur, lev, bp);
-		block = XFS_BUF_TO_INOBT_BLOCK(bp);
-		if ((error = xfs_btree_check_sblock(cur, block, lev, bp)))
-			return error;
-		cur->bc_ptrs[lev] = 1;
-	}
-	*stat = 1;
-	return 0;
-}
+#ifdef XFS_BTREE_TRACE
+	.trace_enter		= xfs_inobt_trace_enter,
+	.trace_cursor		= xfs_inobt_trace_cursor,
+	.trace_key		= xfs_inobt_trace_key,
+	.trace_record		= xfs_inobt_trace_record,
+#endif
+};
 
 /*
- * Insert the current record at the point referenced by cur.
- * The cursor may be inconsistent on return if splits have been done.
+ * Allocate a new inode btree cursor.
  */
-int					/* error */
-xfs_inobt_insert(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	int		*stat)		/* success/failure */
+struct xfs_btree_cur *				/* new inode btree cursor */
+xfs_inobt_init_cursor(
+	struct xfs_mount	*mp,		/* file system mount point */
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_buf		*agbp,		/* buffer for agi structure */
+	xfs_agnumber_t		agno)		/* allocation group number */
 {
-	int		error;		/* error return value */
-	int		i;		/* result value, 0 for failure */
-	int		level;		/* current level number in btree */
-	xfs_agblock_t	nbno;		/* new block number (split result) */
-	xfs_btree_cur_t	*ncur;		/* new cursor (split result) */
-	xfs_inobt_rec_t	nrec;		/* record being inserted this level */
-	xfs_btree_cur_t	*pcur;		/* previous level's cursor */
+	struct xfs_agi		*agi = XFS_BUF_TO_AGI(agbp);
+	struct xfs_btree_cur	*cur;
 
-	level = 0;
-	nbno = NULLAGBLOCK;
-	nrec.ir_startino = cpu_to_be32(cur->bc_rec.i.ir_startino);
-	nrec.ir_freecount = cpu_to_be32(cur->bc_rec.i.ir_freecount);
-	nrec.ir_free = cpu_to_be64(cur->bc_rec.i.ir_free);
-	ncur = NULL;
-	pcur = cur;
-	/*
-	 * Loop going up the tree, starting at the leaf level.
-	 * Stop when we don't get a split block, that must mean that
-	 * the insert is finished with this level.
-	 */
-	do {
-		/*
-		 * Insert nrec/nbno into this level of the tree.
-		 * Note if we fail, nbno will be null.
-		 */
-		if ((error = xfs_inobt_insrec(pcur, level++, &nbno, &nrec, &ncur,
-				&i))) {
-			if (pcur != cur)
-				xfs_btree_del_cursor(pcur, XFS_BTREE_ERROR);
-			return error;
-		}
-		/*
-		 * See if the cursor we just used is trash.
-		 * Can't trash the caller's cursor, but otherwise we should
-		 * if ncur is a new cursor or we're about to be done.
-		 */
-		if (pcur != cur && (ncur || nbno == NULLAGBLOCK)) {
-			cur->bc_nlevels = pcur->bc_nlevels;
-			xfs_btree_del_cursor(pcur, XFS_BTREE_NOERROR);
-		}
-		/*
-		 * If we got a new cursor, switch to it.
-		 */
-		if (ncur) {
-			pcur = ncur;
-			ncur = NULL;
-		}
-	} while (nbno != NULLAGBLOCK);
-	*stat = i;
-	return 0;
-}
+	cur = kmem_zone_zalloc(xfs_btree_cur_zone, KM_SLEEP);
 
-/*
- * Lookup the record equal to ino in the btree given by cur.
- */
-int					/* error */
-xfs_inobt_lookup_eq(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agino_t	ino,		/* starting inode of chunk */
-	__int32_t	fcnt,		/* free inode count */
-	xfs_inofree_t	free,		/* free inode mask */
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.i.ir_startino = ino;
-	cur->bc_rec.i.ir_freecount = fcnt;
-	cur->bc_rec.i.ir_free = free;
-	return xfs_inobt_lookup(cur, XFS_LOOKUP_EQ, stat);
-}
+	cur->bc_tp = tp;
+	cur->bc_mp = mp;
+	cur->bc_nlevels = be32_to_cpu(agi->agi_level);
+	cur->bc_btnum = XFS_BTNUM_INO;
+	cur->bc_blocklog = mp->m_sb.sb_blocklog;
 
-/*
- * Lookup the first record greater than or equal to ino
- * in the btree given by cur.
- */
-int					/* error */
-xfs_inobt_lookup_ge(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agino_t	ino,		/* starting inode of chunk */
-	__int32_t	fcnt,		/* free inode count */
-	xfs_inofree_t	free,		/* free inode mask */
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.i.ir_startino = ino;
-	cur->bc_rec.i.ir_freecount = fcnt;
-	cur->bc_rec.i.ir_free = free;
-	return xfs_inobt_lookup(cur, XFS_LOOKUP_GE, stat);
-}
+	cur->bc_ops = &xfs_inobt_ops;
 
-/*
- * Lookup the first record less than or equal to ino
- * in the btree given by cur.
- */
-int					/* error */
-xfs_inobt_lookup_le(
-	xfs_btree_cur_t	*cur,		/* btree cursor */
-	xfs_agino_t	ino,		/* starting inode of chunk */
-	__int32_t	fcnt,		/* free inode count */
-	xfs_inofree_t	free,		/* free inode mask */
-	int		*stat)		/* success/failure */
-{
-	cur->bc_rec.i.ir_startino = ino;
-	cur->bc_rec.i.ir_freecount = fcnt;
-	cur->bc_rec.i.ir_free = free;
-	return xfs_inobt_lookup(cur, XFS_LOOKUP_LE, stat);
+	cur->bc_private.a.agbp = agbp;
+	cur->bc_private.a.agno = agno;
+
+	return cur;
 }
 
 /*
- * Update the record referred to by cur, to the value given
- * by [ino, fcnt, free].
- * This either works (return 0) or gets an EFSCORRUPTED error.
+ * Calculate number of records in an inobt btree block.
  */
-int					/* error */
-xfs_inobt_update(
-	xfs_btree_cur_t		*cur,	/* btree cursor */
-	xfs_agino_t		ino,	/* starting inode of chunk */
-	__int32_t		fcnt,	/* free inode count */
-	xfs_inofree_t		free)	/* free inode mask */
+int
+xfs_inobt_maxrecs(
+	struct xfs_mount	*mp,
+	int			blocklen,
+	int			leaf)
 {
-	xfs_inobt_block_t	*block;	/* btree block to update */
-	xfs_buf_t		*bp;	/* buffer containing btree block */
-	int			error;	/* error return value */
-	int			ptr;	/* current record number (updating) */
-	xfs_inobt_rec_t		*rp;	/* pointer to updated record */
+	blocklen -= XFS_INOBT_BLOCK_LEN(mp);
 
-	/*
-	 * Pick up the current block.
-	 */
-	bp = cur->bc_bufs[0];
-	block = XFS_BUF_TO_INOBT_BLOCK(bp);
-#ifdef DEBUG
-	if ((error = xfs_btree_check_sblock(cur, block, 0, bp)))
-		return error;
-#endif
-	/*
-	 * Get the address of the rec to be updated.
-	 */
-	ptr = cur->bc_ptrs[0];
-	rp = XFS_INOBT_REC_ADDR(block, ptr, cur);
-	/*
-	 * Fill in the new contents and log them.
-	 */
-	rp->ir_startino = cpu_to_be32(ino);
-	rp->ir_freecount = cpu_to_be32(fcnt);
-	rp->ir_free = cpu_to_be64(free);
-	xfs_inobt_log_recs(cur, bp, ptr, ptr);
-	/*
-	 * Updating first record in leaf. Pass new key value up to our parent.
-	 */
-	if (ptr == 1) {
-		xfs_inobt_key_t	key;	/* key containing [ino] */
-
-		key.ir_startino = cpu_to_be32(ino);
-		if ((error = xfs_inobt_updkey(cur, &key, 1)))
-			return error;
-	}
-	return 0;
+	if (leaf)
+		return blocklen / sizeof(xfs_inobt_rec_t);
+	return blocklen / (sizeof(xfs_inobt_key_t) + sizeof(xfs_inobt_ptr_t));
 }
diff --git a/fs/xfs/xfs_ialloc_btree.h b/fs/xfs/xfs_ialloc_btree.h
index 8efc4a5b8b9..37e5dd01a57 100644
--- a/fs/xfs/xfs_ialloc_btree.h
+++ b/fs/xfs/xfs_ialloc_btree.h
@@ -24,7 +24,6 @@
 
 struct xfs_buf;
 struct xfs_btree_cur;
-struct xfs_btree_sblock;
 struct xfs_mount;
 
 /*
@@ -70,11 +69,6 @@ typedef struct xfs_inobt_key {
 /* btree pointer type */
 typedef __be32 xfs_inobt_ptr_t;
 
-/* btree block header type */
-typedef	struct xfs_btree_sblock xfs_inobt_block_t;
-
-#define	XFS_BUF_TO_INOBT_BLOCK(bp)	((xfs_inobt_block_t *)XFS_BUF_PTR(bp))
-
 /*
  * Bit manipulations for ir_free.
  */
@@ -85,14 +79,6 @@ typedef	struct xfs_btree_sblock xfs_inobt_block_t;
 #define	XFS_INOBT_CLR_FREE(rp,i)	((rp)->ir_free &= ~XFS_INOBT_MASK(i))
 
 /*
- * Real block structures have a size equal to the disk block size.
- */
-#define	XFS_INOBT_BLOCK_MAXRECS(lev,cur) ((cur)->bc_mp->m_inobt_mxr[lev != 0])
-#define	XFS_INOBT_BLOCK_MINRECS(lev,cur) ((cur)->bc_mp->m_inobt_mnr[lev != 0])
-#define	XFS_INOBT_IS_LAST_REC(cur)	\
-	((cur)->bc_ptrs[0] == be16_to_cpu(XFS_BUF_TO_INOBT_BLOCK((cur)->bc_bufs[0])->bb_numrecs))
-
-/*
  * Maximum number of inode btree levels.
  */
 #define	XFS_IN_MAXLEVELS(mp)		((mp)->m_in_maxlevels)
@@ -104,75 +90,38 @@ typedef	struct xfs_btree_sblock xfs_inobt_block_t;
 #define	XFS_PREALLOC_BLOCKS(mp)		((xfs_agblock_t)(XFS_IBT_BLOCK(mp) + 1))
 
 /*
- * Record, key, and pointer address macros for btree blocks.
- */
-#define XFS_INOBT_REC_ADDR(bb,i,cur) \
-	(XFS_BTREE_REC_ADDR(xfs_inobt, bb, i))
-
-#define	XFS_INOBT_KEY_ADDR(bb,i,cur) \
-	(XFS_BTREE_KEY_ADDR(xfs_inobt, bb, i))
-
-#define	XFS_INOBT_PTR_ADDR(bb,i,cur) \
-	(XFS_BTREE_PTR_ADDR(xfs_inobt, bb, \
-				i, XFS_INOBT_BLOCK_MAXRECS(1, cur)))
-
-/*
- * Decrement cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-extern int xfs_inobt_decrement(struct xfs_btree_cur *cur, int level, int *stat);
-
-/*
- * Delete the record pointed to by cur.
- * The cursor refers to the place where the record was (could be inserted)
- * when the operation returns.
- */
-extern int xfs_inobt_delete(struct xfs_btree_cur *cur, int *stat);
-
-/*
- * Get the data from the pointed-to record.
- */
-extern int xfs_inobt_get_rec(struct xfs_btree_cur *cur, xfs_agino_t *ino,
-			     __int32_t *fcnt, xfs_inofree_t *free, int *stat);
-
-/*
- * Increment cursor by one record at the level.
- * For nonzero levels the leaf-ward information is untouched.
- */
-extern int xfs_inobt_increment(struct xfs_btree_cur *cur, int level, int *stat);
-
-/*
- * Insert the current record at the point referenced by cur.
- * The cursor may be inconsistent on return if splits have been done.
- */
-extern int xfs_inobt_insert(struct xfs_btree_cur *cur, int *stat);
-
-/*
- * Lookup the record equal to ino in the btree given by cur.
- */
-extern int xfs_inobt_lookup_eq(struct xfs_btree_cur *cur, xfs_agino_t ino,
-				__int32_t fcnt, xfs_inofree_t free, int *stat);
-
-/*
- * Lookup the first record greater than or equal to ino
- * in the btree given by cur.
- */
-extern int xfs_inobt_lookup_ge(struct xfs_btree_cur *cur, xfs_agino_t ino,
-				__int32_t fcnt,	xfs_inofree_t free, int *stat);
-
-/*
- * Lookup the first record less than or equal to ino
- * in the btree given by cur.
+ * Btree block header size depends on a superblock flag.
+ *
+ * (not quite yet, but soon)
  */
-extern int xfs_inobt_lookup_le(struct xfs_btree_cur *cur, xfs_agino_t ino,
-				__int32_t fcnt,	xfs_inofree_t free, int *stat);
+#define XFS_INOBT_BLOCK_LEN(mp)	XFS_BTREE_SBLOCK_LEN
 
 /*
- * Update the record referred to by cur, to the value given
- * by [ino, fcnt, free].
- * This either works (return 0) or gets an EFSCORRUPTED error.
- */
-extern int xfs_inobt_update(struct xfs_btree_cur *cur, xfs_agino_t ino,
-				__int32_t fcnt, xfs_inofree_t free);
+ * Record, key, and pointer address macros for btree blocks.
+ *
+ * (note that some of these may appear unused, but they are used in userspace)
+ */
+#define XFS_INOBT_REC_ADDR(mp, block, index) \
+	((xfs_inobt_rec_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 (((index) - 1) * sizeof(xfs_inobt_rec_t))))
+
+#define XFS_INOBT_KEY_ADDR(mp, block, index) \
+	((xfs_inobt_key_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 ((index) - 1) * sizeof(xfs_inobt_key_t)))
+
+#define XFS_INOBT_PTR_ADDR(mp, block, index, maxrecs) \
+	((xfs_inobt_ptr_t *) \
+		((char *)(block) + \
+		 XFS_INOBT_BLOCK_LEN(mp) + \
+		 (maxrecs) * sizeof(xfs_inobt_key_t) + \
+		 ((index) - 1) * sizeof(xfs_inobt_ptr_t)))
+
+extern struct xfs_btree_cur *xfs_inobt_init_cursor(struct xfs_mount *,
+		struct xfs_trans *, struct xfs_buf *, xfs_agnumber_t);
+extern int xfs_inobt_maxrecs(struct xfs_mount *, int, int);
 
 #endif	/* __XFS_IALLOC_BTREE_H__ */
diff --git a/fs/xfs/xfs_iget.c b/fs/xfs/xfs_iget.c
index e229e9e001c..e2fb6210d4c 100644
--- a/fs/xfs/xfs_iget.c
+++ b/fs/xfs/xfs_iget.c
@@ -38,281 +38,283 @@
 #include "xfs_ialloc.h"
 #include "xfs_quota.h"
 #include "xfs_utils.h"
+#include "xfs_trans_priv.h"
+#include "xfs_inode_item.h"
+#include "xfs_bmap.h"
+#include "xfs_btree_trace.h"
+#include "xfs_dir2_trace.h"
+
 
 /*
- * Look up an inode by number in the given file system.
- * The inode is looked up in the cache held in each AG.
- * If the inode is found in the cache, attach it to the provided
- * vnode.
- *
- * If it is not in core, read it in from the file system's device,
- * add it to the cache and attach the provided vnode.
- *
- * The inode is locked according to the value of the lock_flags parameter.
- * This flag parameter indicates how and if the inode's IO lock and inode lock
- * should be taken.
- *
- * mp -- the mount point structure for the current file system.  It points
- *       to the inode hash table.
- * tp -- a pointer to the current transaction if there is one.  This is
- *       simply passed through to the xfs_iread() call.
- * ino -- the number of the inode desired.  This is the unique identifier
- *        within the file system for the inode being requested.
- * lock_flags -- flags indicating how to lock the inode.  See the comment
- *		 for xfs_ilock() for a list of valid values.
- * bno -- the block number starting the buffer containing the inode,
- *	  if known (as by bulkstat), else 0.
+ * Allocate and initialise an xfs_inode.
  */
-STATIC int
-xfs_iget_core(
-	struct inode	*inode,
-	xfs_mount_t	*mp,
-	xfs_trans_t	*tp,
-	xfs_ino_t	ino,
-	uint		flags,
-	uint		lock_flags,
-	xfs_inode_t	**ipp,
-	xfs_daddr_t	bno)
+STATIC struct xfs_inode *
+xfs_inode_alloc(
+	struct xfs_mount	*mp,
+	xfs_ino_t		ino)
 {
-	struct inode	*old_inode;
-	xfs_inode_t	*ip;
-	xfs_inode_t	*iq;
-	int		error;
-	unsigned long	first_index, mask;
-	xfs_perag_t	*pag;
-	xfs_agino_t	agino;
+	struct xfs_inode	*ip;
 
-	/* the radix tree exists only in inode capable AGs */
-	if (XFS_INO_TO_AGNO(mp, ino) >= mp->m_maxagi)
-		return EINVAL;
+	/*
+	 * if this didn't occur in transactions, we could use
+	 * KM_MAYFAIL and return NULL here on ENOMEM. Set the
+	 * code up to do this anyway.
+	 */
+	ip = kmem_zone_alloc(xfs_inode_zone, KM_SLEEP);
+	if (!ip)
+		return NULL;
 
-	/* get the perag structure and ensure that it's inode capable */
-	pag = xfs_get_perag(mp, ino);
-	if (!pag->pagi_inodeok)
-		return EINVAL;
-	ASSERT(pag->pag_ici_init);
-	agino = XFS_INO_TO_AGINO(mp, ino);
+	ASSERT(atomic_read(&ip->i_iocount) == 0);
+	ASSERT(atomic_read(&ip->i_pincount) == 0);
+	ASSERT(!spin_is_locked(&ip->i_flags_lock));
+	ASSERT(completion_done(&ip->i_flush));
 
-again:
-	read_lock(&pag->pag_ici_lock);
-	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+	/*
+	 * initialise the VFS inode here to get failures
+	 * out of the way early.
+	 */
+	if (!inode_init_always(mp->m_super, VFS_I(ip))) {
+		kmem_zone_free(xfs_inode_zone, ip);
+		return NULL;
+	}
+
+	/* initialise the xfs inode */
+	ip->i_ino = ino;
+	ip->i_mount = mp;
+	memset(&ip->i_imap, 0, sizeof(struct xfs_imap));
+	ip->i_afp = NULL;
+	memset(&ip->i_df, 0, sizeof(xfs_ifork_t));
+	ip->i_flags = 0;
+	ip->i_update_core = 0;
+	ip->i_update_size = 0;
+	ip->i_delayed_blks = 0;
+	memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
+	ip->i_size = 0;
+	ip->i_new_size = 0;
+
+	/*
+	 * Initialize inode's trace buffers.
+	 */
+#ifdef	XFS_INODE_TRACE
+	ip->i_trace = ktrace_alloc(INODE_TRACE_SIZE, KM_NOFS);
+#endif
+#ifdef XFS_BMAP_TRACE
+	ip->i_xtrace = ktrace_alloc(XFS_BMAP_KTRACE_SIZE, KM_NOFS);
+#endif
+#ifdef XFS_BTREE_TRACE
+	ip->i_btrace = ktrace_alloc(XFS_BMBT_KTRACE_SIZE, KM_NOFS);
+#endif
+#ifdef XFS_RW_TRACE
+	ip->i_rwtrace = ktrace_alloc(XFS_RW_KTRACE_SIZE, KM_NOFS);
+#endif
+#ifdef XFS_ILOCK_TRACE
+	ip->i_lock_trace = ktrace_alloc(XFS_ILOCK_KTRACE_SIZE, KM_NOFS);
+#endif
+#ifdef XFS_DIR2_TRACE
+	ip->i_dir_trace = ktrace_alloc(XFS_DIR2_KTRACE_SIZE, KM_NOFS);
+#endif
+
+	return ip;
+}
+
+/*
+ * Check the validity of the inode we just found it the cache
+ */
+static int
+xfs_iget_cache_hit(
+	struct xfs_perag	*pag,
+	struct xfs_inode	*ip,
+	int			flags,
+	int			lock_flags) __releases(pag->pag_ici_lock)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	int			error = EAGAIN;
+
+	/*
+	 * If INEW is set this inode is being set up
+	 * If IRECLAIM is set this inode is being torn down
+	 * Pause and try again.
+	 */
+	if (xfs_iflags_test(ip, (XFS_INEW|XFS_IRECLAIM))) {
+		XFS_STATS_INC(xs_ig_frecycle);
+		goto out_error;
+	}
+
+	/* If IRECLAIMABLE is set, we've torn down the vfs inode part */
+	if (xfs_iflags_test(ip, XFS_IRECLAIMABLE)) {
 
-	if (ip != NULL) {
 		/*
-		 * If INEW is set this inode is being set up
-		 * we need to pause and try again.
+		 * If lookup is racing with unlink, then we should return an
+		 * error immediately so we don't remove it from the reclaim
+		 * list and potentially leak the inode.
 		 */
-		if (xfs_iflags_test(ip, XFS_INEW)) {
-			read_unlock(&pag->pag_ici_lock);
-			delay(1);
-			XFS_STATS_INC(xs_ig_frecycle);
-
-			goto again;
+		if ((ip->i_d.di_mode == 0) && !(flags & XFS_IGET_CREATE)) {
+			error = ENOENT;
+			goto out_error;
 		}
 
-		old_inode = ip->i_vnode;
-		if (old_inode == NULL) {
-			/*
-			 * If IRECLAIM is set this inode is
-			 * on its way out of the system,
-			 * we need to pause and try again.
-			 */
-			if (xfs_iflags_test(ip, XFS_IRECLAIM)) {
-				read_unlock(&pag->pag_ici_lock);
-				delay(1);
-				XFS_STATS_INC(xs_ig_frecycle);
-
-				goto again;
-			}
-			ASSERT(xfs_iflags_test(ip, XFS_IRECLAIMABLE));
-
-			/*
-			 * If lookup is racing with unlink, then we
-			 * should return an error immediately so we
-			 * don't remove it from the reclaim list and
-			 * potentially leak the inode.
-			 */
-			if ((ip->i_d.di_mode == 0) &&
-			    !(flags & XFS_IGET_CREATE)) {
-				read_unlock(&pag->pag_ici_lock);
-				xfs_put_perag(mp, pag);
-				return ENOENT;
-			}
-
-			xfs_itrace_exit_tag(ip, "xfs_iget.alloc");
-
-			XFS_STATS_INC(xs_ig_found);
-			xfs_iflags_clear(ip, XFS_IRECLAIMABLE);
-			read_unlock(&pag->pag_ici_lock);
-
-			XFS_MOUNT_ILOCK(mp);
-			list_del_init(&ip->i_reclaim);
-			XFS_MOUNT_IUNLOCK(mp);
-
-			goto finish_inode;
-
-		} else if (inode != old_inode) {
-			/* The inode is being torn down, pause and
-			 * try again.
-			 */
-			if (old_inode->i_state & (I_FREEING | I_CLEAR)) {
-				read_unlock(&pag->pag_ici_lock);
-				delay(1);
-				XFS_STATS_INC(xs_ig_frecycle);
-
-				goto again;
-			}
-/* Chances are the other vnode (the one in the inode) is being torn
-* down right now, and we landed on top of it. Question is, what do
-* we do? Unhook the old inode and hook up the new one?
-*/
-			cmn_err(CE_PANIC,
-		"xfs_iget_core: ambiguous vns: vp/0x%p, invp/0x%p",
-					old_inode, inode);
-		}
+		xfs_itrace_exit_tag(ip, "xfs_iget.alloc");
 
 		/*
-		 * Inode cache hit
+		 * We need to re-initialise the VFS inode as it has been
+		 * 'freed' by the VFS. Do this here so we can deal with
+		 * errors cleanly, then tag it so it can be set up correctly
+		 * later.
 		 */
-		read_unlock(&pag->pag_ici_lock);
-		XFS_STATS_INC(xs_ig_found);
-
-finish_inode:
-		if (ip->i_d.di_mode == 0 && !(flags & XFS_IGET_CREATE)) {
-			xfs_put_perag(mp, pag);
-			return ENOENT;
+		if (!inode_init_always(mp->m_super, VFS_I(ip))) {
+			error = ENOMEM;
+			goto out_error;
 		}
 
-		if (lock_flags != 0)
-			xfs_ilock(ip, lock_flags);
+		/*
+		 * We must set the XFS_INEW flag before clearing the
+		 * XFS_IRECLAIMABLE flag so that if a racing lookup does
+		 * not find the XFS_IRECLAIMABLE above but has the igrab()
+		 * below succeed we can safely check XFS_INEW to detect
+		 * that this inode is still being initialised.
+		 */
+		xfs_iflags_set(ip, XFS_INEW);
+		xfs_iflags_clear(ip, XFS_IRECLAIMABLE);
+
+		/* clear the radix tree reclaim flag as well. */
+		__xfs_inode_clear_reclaim_tag(mp, pag, ip);
+	} else if (!igrab(VFS_I(ip))) {
+		/* If the VFS inode is being torn down, pause and try again. */
+		XFS_STATS_INC(xs_ig_frecycle);
+		goto out_error;
+	} else if (xfs_iflags_test(ip, XFS_INEW)) {
+		/*
+		 * We are racing with another cache hit that is
+		 * currently recycling this inode out of the XFS_IRECLAIMABLE
+		 * state. Wait for the initialisation to complete before
+		 * continuing.
+		 */
+		wait_on_inode(VFS_I(ip));
+	}
 
-		xfs_iflags_clear(ip, XFS_ISTALE);
-		xfs_itrace_exit_tag(ip, "xfs_iget.found");
-		goto return_ip;
+	if (ip->i_d.di_mode == 0 && !(flags & XFS_IGET_CREATE)) {
+		error = ENOENT;
+		iput(VFS_I(ip));
+		goto out_error;
 	}
 
-	/*
-	 * Inode cache miss
-	 */
+	/* We've got a live one. */
 	read_unlock(&pag->pag_ici_lock);
-	XFS_STATS_INC(xs_ig_missed);
 
-	/*
-	 * Read the disk inode attributes into a new inode structure and get
-	 * a new vnode for it. This should also initialize i_ino and i_mount.
-	 */
-	error = xfs_iread(mp, tp, ino, &ip, bno,
-			  (flags & XFS_IGET_BULKSTAT) ? XFS_IMAP_BULKSTAT : 0);
-	if (error) {
-		xfs_put_perag(mp, pag);
-		return error;
-	}
+	if (lock_flags != 0)
+		xfs_ilock(ip, lock_flags);
 
-	xfs_itrace_exit_tag(ip, "xfs_iget.alloc");
+	xfs_iflags_clear(ip, XFS_ISTALE);
+	xfs_itrace_exit_tag(ip, "xfs_iget.found");
+	XFS_STATS_INC(xs_ig_found);
+	return 0;
+
+out_error:
+	read_unlock(&pag->pag_ici_lock);
+	return error;
+}
 
 
-	mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
-		     "xfsino", ip->i_ino);
-	mrlock_init(&ip->i_iolock, MRLOCK_BARRIER, "xfsio", ip->i_ino);
-	init_waitqueue_head(&ip->i_ipin_wait);
-	atomic_set(&ip->i_pincount, 0);
+static int
+xfs_iget_cache_miss(
+	struct xfs_mount	*mp,
+	struct xfs_perag	*pag,
+	xfs_trans_t		*tp,
+	xfs_ino_t		ino,
+	struct xfs_inode	**ipp,
+	xfs_daddr_t		bno,
+	int			flags,
+	int			lock_flags) __releases(pag->pag_ici_lock)
+{
+	struct xfs_inode	*ip;
+	int			error;
+	unsigned long		first_index, mask;
+	xfs_agino_t		agino = XFS_INO_TO_AGINO(mp, ino);
 
-	/*
-	 * Because we want to use a counting completion, complete
-	 * the flush completion once to allow a single access to
-	 * the flush completion without blocking.
-	 */
-	init_completion(&ip->i_flush);
-	complete(&ip->i_flush);
+	ip = xfs_inode_alloc(mp, ino);
+	if (!ip)
+		return ENOMEM;
 
-	if (lock_flags)
-		xfs_ilock(ip, lock_flags);
+	error = xfs_iread(mp, tp, ip, bno, flags);
+	if (error)
+		goto out_destroy;
+
+	xfs_itrace_exit_tag(ip, "xfs_iget.alloc");
 
 	if ((ip->i_d.di_mode == 0) && !(flags & XFS_IGET_CREATE)) {
-		xfs_idestroy(ip);
-		xfs_put_perag(mp, pag);
-		return ENOENT;
+		error = ENOENT;
+		goto out_destroy;
 	}
 
+	if (lock_flags)
+		xfs_ilock(ip, lock_flags);
+
 	/*
 	 * Preload the radix tree so we can insert safely under the
-	 * write spinlock.
+	 * write spinlock. Note that we cannot sleep inside the preload
+	 * region.
 	 */
 	if (radix_tree_preload(GFP_KERNEL)) {
-		xfs_idestroy(ip);
-		delay(1);
-		goto again;
+		error = EAGAIN;
+		goto out_unlock;
 	}
+
 	mask = ~(((XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog)) - 1);
 	first_index = agino & mask;
 	write_lock(&pag->pag_ici_lock);
-	/*
-	 * insert the new inode
-	 */
+
+	/* insert the new inode */
 	error = radix_tree_insert(&pag->pag_ici_root, agino, ip);
 	if (unlikely(error)) {
-		BUG_ON(error != -EEXIST);
-		write_unlock(&pag->pag_ici_lock);
-		radix_tree_preload_end();
-		xfs_idestroy(ip);
+		WARN_ON(error != -EEXIST);
 		XFS_STATS_INC(xs_ig_dup);
-		goto again;
+		error = EAGAIN;
+		goto out_preload_end;
 	}
 
-	/*
-	 * These values _must_ be set before releasing the radix tree lock!
-	 */
+	/* These values _must_ be set before releasing the radix tree lock! */
 	ip->i_udquot = ip->i_gdquot = NULL;
 	xfs_iflags_set(ip, XFS_INEW);
 
 	write_unlock(&pag->pag_ici_lock);
 	radix_tree_preload_end();
-
-	/*
-	 * Link ip to its mount and thread it on the mount's inode list.
-	 */
-	XFS_MOUNT_ILOCK(mp);
-	if ((iq = mp->m_inodes)) {
-		ASSERT(iq->i_mprev->i_mnext == iq);
-		ip->i_mprev = iq->i_mprev;
-		iq->i_mprev->i_mnext = ip;
-		iq->i_mprev = ip;
-		ip->i_mnext = iq;
-	} else {
-		ip->i_mnext = ip;
-		ip->i_mprev = ip;
-	}
-	mp->m_inodes = ip;
-
-	XFS_MOUNT_IUNLOCK(mp);
-	xfs_put_perag(mp, pag);
-
- return_ip:
-	ASSERT(ip->i_df.if_ext_max ==
-	       XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t));
-
-	xfs_iflags_set(ip, XFS_IMODIFIED);
 	*ipp = ip;
-
-	/*
-	 * Set up the Linux with the Linux inode.
-	 */
-	ip->i_vnode = inode;
-	inode->i_private = ip;
-
-	/*
-	 * If we have a real type for an on-disk inode, we can set ops(&unlock)
-	 * now.	 If it's a new inode being created, xfs_ialloc will handle it.
-	 */
-	if (ip->i_d.di_mode != 0)
-		xfs_setup_inode(ip);
 	return 0;
-}
 
+out_preload_end:
+	write_unlock(&pag->pag_ici_lock);
+	radix_tree_preload_end();
+out_unlock:
+	if (lock_flags)
+		xfs_iunlock(ip, lock_flags);
+out_destroy:
+	xfs_destroy_inode(ip);
+	return error;
+}
 
 /*
- * The 'normal' internal xfs_iget, if needed it will
- * 'allocate', or 'get', the vnode.
+ * Look up an inode by number in the given file system.
+ * The inode is looked up in the cache held in each AG.
+ * If the inode is found in the cache, initialise the vfs inode
+ * if necessary.
+ *
+ * If it is not in core, read it in from the file system's device,
+ * add it to the cache and initialise the vfs inode.
+ *
+ * The inode is locked according to the value of the lock_flags parameter.
+ * This flag parameter indicates how and if the inode's IO lock and inode lock
+ * should be taken.
+ *
+ * mp -- the mount point structure for the current file system.  It points
+ *       to the inode hash table.
+ * tp -- a pointer to the current transaction if there is one.  This is
+ *       simply passed through to the xfs_iread() call.
+ * ino -- the number of the inode desired.  This is the unique identifier
+ *        within the file system for the inode being requested.
+ * lock_flags -- flags indicating how to lock the inode.  See the comment
+ *		 for xfs_ilock() for a list of valid values.
+ * bno -- the block number starting the buffer containing the inode,
+ *	  if known (as by bulkstat), else 0.
  */
 int
 xfs_iget(
@@ -324,61 +326,64 @@ xfs_iget(
 	xfs_inode_t	**ipp,
 	xfs_daddr_t	bno)
 {
-	struct inode	*inode;
 	xfs_inode_t	*ip;
 	int		error;
+	xfs_perag_t	*pag;
+	xfs_agino_t	agino;
 
-	XFS_STATS_INC(xs_ig_attempts);
+	/* the radix tree exists only in inode capable AGs */
+	if (XFS_INO_TO_AGNO(mp, ino) >= mp->m_maxagi)
+		return EINVAL;
 
-retry:
-	inode = iget_locked(mp->m_super, ino);
-	if (!inode)
-		/* If we got no inode we are out of memory */
-		return ENOMEM;
+	/* get the perag structure and ensure that it's inode capable */
+	pag = xfs_get_perag(mp, ino);
+	if (!pag->pagi_inodeok)
+		return EINVAL;
+	ASSERT(pag->pag_ici_init);
+	agino = XFS_INO_TO_AGINO(mp, ino);
 
-	if (inode->i_state & I_NEW) {
-		XFS_STATS_INC(vn_active);
-		XFS_STATS_INC(vn_alloc);
-
-		error = xfs_iget_core(inode, mp, tp, ino, flags,
-				lock_flags, ipp, bno);
-		if (error) {
-			make_bad_inode(inode);
-			if (inode->i_state & I_NEW)
-				unlock_new_inode(inode);
-			iput(inode);
-		}
-		return error;
+again:
+	error = 0;
+	read_lock(&pag->pag_ici_lock);
+	ip = radix_tree_lookup(&pag->pag_ici_root, agino);
+
+	if (ip) {
+		error = xfs_iget_cache_hit(pag, ip, flags, lock_flags);
+		if (error)
+			goto out_error_or_again;
+	} else {
+		read_unlock(&pag->pag_ici_lock);
+		XFS_STATS_INC(xs_ig_missed);
+
+		error = xfs_iget_cache_miss(mp, pag, tp, ino, &ip, bno,
+							flags, lock_flags);
+		if (error)
+			goto out_error_or_again;
 	}
+	xfs_put_perag(mp, pag);
 
+	*ipp = ip;
+
+	ASSERT(ip->i_df.if_ext_max ==
+	       XFS_IFORK_DSIZE(ip) / sizeof(xfs_bmbt_rec_t));
 	/*
-	 * If the inode is not fully constructed due to
-	 * filehandle mismatches wait for the inode to go
-	 * away and try again.
-	 *
-	 * iget_locked will call __wait_on_freeing_inode
-	 * to wait for the inode to go away.
+	 * If we have a real type for an on-disk inode, we can set ops(&unlock)
+	 * now.	 If it's a new inode being created, xfs_ialloc will handle it.
 	 */
-	if (is_bad_inode(inode)) {
-		iput(inode);
-		delay(1);
-		goto retry;
-	}
+	if (xfs_iflags_test(ip, XFS_INEW) && ip->i_d.di_mode != 0)
+		xfs_setup_inode(ip);
+	return 0;
 
-	ip = XFS_I(inode);
-	if (!ip) {
-		iput(inode);
+out_error_or_again:
+	if (error == EAGAIN) {
 		delay(1);
-		goto retry;
+		goto again;
 	}
-
-	if (lock_flags != 0)
-		xfs_ilock(ip, lock_flags);
-	XFS_STATS_INC(xs_ig_found);
-	*ipp = ip;
-	return 0;
+	xfs_put_perag(mp, pag);
+	return error;
 }
 
+
 /*
  * Look for the inode corresponding to the given ino in the hash table.
  * If it is there and its i_transp pointer matches tp, return it.
@@ -444,99 +449,109 @@ xfs_iput_new(
 	IRELE(ip);
 }
 
-
 /*
- * This routine embodies the part of the reclaim code that pulls
- * the inode from the inode hash table and the mount structure's
- * inode list.
- * This should only be called from xfs_reclaim().
+ * This is called free all the memory associated with an inode.
+ * It must free the inode itself and any buffers allocated for
+ * if_extents/if_data and if_broot.  It must also free the lock
+ * associated with the inode.
+ *
+ * Note: because we don't initialise everything on reallocation out
+ * of the zone, we must ensure we nullify everything correctly before
+ * freeing the structure.
  */
 void
-xfs_ireclaim(xfs_inode_t *ip)
+xfs_ireclaim(
+	struct xfs_inode	*ip)
 {
-	/*
-	 * Remove from old hash list and mount list.
-	 */
-	XFS_STATS_INC(xs_ig_reclaims);
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_perag	*pag;
 
-	xfs_iextract(ip);
-
-	/*
-	 * Here we do a spurious inode lock in order to coordinate with
-	 * xfs_sync().  This is because xfs_sync() references the inodes
-	 * in the mount list without taking references on the corresponding
-	 * vnodes.  We make that OK here by ensuring that we wait until
-	 * the inode is unlocked in xfs_sync() before we go ahead and
-	 * free it.  We get both the regular lock and the io lock because
-	 * the xfs_sync() code may need to drop the regular one but will
-	 * still hold the io lock.
-	 */
-	xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
-
-	/*
-	 * Release dquots (and their references) if any. An inode may escape
-	 * xfs_inactive and get here via vn_alloc->vn_reclaim path.
-	 */
-	XFS_QM_DQDETACH(ip->i_mount, ip);
-
-	/*
-	 * Pull our behavior descriptor from the vnode chain.
-	 */
-	if (ip->i_vnode) {
-		ip->i_vnode->i_private = NULL;
-		ip->i_vnode = NULL;
-	}
+	XFS_STATS_INC(xs_ig_reclaims);
 
 	/*
-	 * Free all memory associated with the inode.
+	 * Remove the inode from the per-AG radix tree.  It doesn't matter
+	 * if it was never added to it because radix_tree_delete can deal
+	 * with that case just fine.
 	 */
-	xfs_iunlock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
-	xfs_idestroy(ip);
-}
-
-/*
- * This routine removes an about-to-be-destroyed inode from
- * all of the lists in which it is located with the exception
- * of the behavior chain.
- */
-void
-xfs_iextract(
-	xfs_inode_t	*ip)
-{
-	xfs_mount_t	*mp = ip->i_mount;
-	xfs_perag_t	*pag = xfs_get_perag(mp, ip->i_ino);
-	xfs_inode_t	*iq;
-
+	pag = xfs_get_perag(mp, ip->i_ino);
 	write_lock(&pag->pag_ici_lock);
 	radix_tree_delete(&pag->pag_ici_root, XFS_INO_TO_AGINO(mp, ip->i_ino));
 	write_unlock(&pag->pag_ici_lock);
 	xfs_put_perag(mp, pag);
 
 	/*
-	 * Remove from mount's inode list.
+	 * Here we do an (almost) spurious inode lock in order to coordinate
+	 * with inode cache radix tree lookups.  This is because the lookup
+	 * can reference the inodes in the cache without taking references.
+	 *
+	 * We make that OK here by ensuring that we wait until the inode is
+	 * unlocked after the lookup before we go ahead and free it.  We get
+	 * both the ilock and the iolock because the code may need to drop the
+	 * ilock one but will still hold the iolock.
 	 */
-	XFS_MOUNT_ILOCK(mp);
-	ASSERT((ip->i_mnext != NULL) && (ip->i_mprev != NULL));
-	iq = ip->i_mnext;
-	iq->i_mprev = ip->i_mprev;
-	ip->i_mprev->i_mnext = iq;
-
+	xfs_ilock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 	/*
-	 * Fix up the head pointer if it points to the inode being deleted.
+	 * Release dquots (and their references) if any.
 	 */
-	if (mp->m_inodes == ip) {
-		if (ip == iq) {
-			mp->m_inodes = NULL;
-		} else {
-			mp->m_inodes = iq;
-		}
+	XFS_QM_DQDETACH(ip->i_mount, ip);
+	xfs_iunlock(ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+
+	switch (ip->i_d.di_mode & S_IFMT) {
+	case S_IFREG:
+	case S_IFDIR:
+	case S_IFLNK:
+		xfs_idestroy_fork(ip, XFS_DATA_FORK);
+		break;
 	}
 
-	/* Deal with the deleted inodes list */
-	list_del_init(&ip->i_reclaim);
+	if (ip->i_afp)
+		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
 
-	mp->m_ireclaims++;
-	XFS_MOUNT_IUNLOCK(mp);
+#ifdef XFS_INODE_TRACE
+	ktrace_free(ip->i_trace);
+#endif
+#ifdef XFS_BMAP_TRACE
+	ktrace_free(ip->i_xtrace);
+#endif
+#ifdef XFS_BTREE_TRACE
+	ktrace_free(ip->i_btrace);
+#endif
+#ifdef XFS_RW_TRACE
+	ktrace_free(ip->i_rwtrace);
+#endif
+#ifdef XFS_ILOCK_TRACE
+	ktrace_free(ip->i_lock_trace);
+#endif
+#ifdef XFS_DIR2_TRACE
+	ktrace_free(ip->i_dir_trace);
+#endif
+	if (ip->i_itemp) {
+		/*
+		 * Only if we are shutting down the fs will we see an
+		 * inode still in the AIL. If it is there, we should remove
+		 * it to prevent a use-after-free from occurring.
+		 */
+		xfs_log_item_t	*lip = &ip->i_itemp->ili_item;
+		struct xfs_ail	*ailp = lip->li_ailp;
+
+		ASSERT(((lip->li_flags & XFS_LI_IN_AIL) == 0) ||
+				       XFS_FORCED_SHUTDOWN(ip->i_mount));
+		if (lip->li_flags & XFS_LI_IN_AIL) {
+			spin_lock(&ailp->xa_lock);
+			if (lip->li_flags & XFS_LI_IN_AIL)
+				xfs_trans_ail_delete(ailp, lip);
+			else
+				spin_unlock(&ailp->xa_lock);
+		}
+		xfs_inode_item_destroy(ip);
+		ip->i_itemp = NULL;
+	}
+	/* asserts to verify all state is correct here */
+	ASSERT(atomic_read(&ip->i_iocount) == 0);
+	ASSERT(atomic_read(&ip->i_pincount) == 0);
+	ASSERT(!spin_is_locked(&ip->i_flags_lock));
+	ASSERT(completion_done(&ip->i_flush));
+	kmem_zone_free(xfs_inode_zone, ip);
 }
 
 /*
@@ -737,7 +752,7 @@ xfs_iunlock(
 		 * it is in the AIL and anyone is waiting on it.  Don't do
 		 * this if the caller has asked us not to.
 		 */
-		xfs_trans_unlocked_item(ip->i_mount,
+		xfs_trans_unlocked_item(ip->i_itemp->ili_item.li_ailp,
 					(xfs_log_item_t*)(ip->i_itemp));
 	}
 	xfs_ilock_trace(ip, 3, lock_flags, (inst_t *)__return_address);
@@ -790,3 +805,51 @@ xfs_isilocked(
 }
 #endif
 
+#ifdef	XFS_INODE_TRACE
+
+#define KTRACE_ENTER(ip, vk, s, line, ra)			\
+	ktrace_enter((ip)->i_trace,				\
+/*  0 */		(void *)(__psint_t)(vk),		\
+/*  1 */		(void *)(s),				\
+/*  2 */		(void *)(__psint_t) line,		\
+/*  3 */		(void *)(__psint_t)atomic_read(&VFS_I(ip)->i_count), \
+/*  4 */		(void *)(ra),				\
+/*  5 */		NULL,					\
+/*  6 */		(void *)(__psint_t)current_cpu(),	\
+/*  7 */		(void *)(__psint_t)current_pid(),	\
+/*  8 */		(void *)__return_address,		\
+/*  9 */		NULL, NULL, NULL, NULL, NULL, NULL, NULL)
+
+/*
+ * Vnode tracing code.
+ */
+void
+_xfs_itrace_entry(xfs_inode_t *ip, const char *func, inst_t *ra)
+{
+	KTRACE_ENTER(ip, INODE_KTRACE_ENTRY, func, 0, ra);
+}
+
+void
+_xfs_itrace_exit(xfs_inode_t *ip, const char *func, inst_t *ra)
+{
+	KTRACE_ENTER(ip, INODE_KTRACE_EXIT, func, 0, ra);
+}
+
+void
+xfs_itrace_hold(xfs_inode_t *ip, char *file, int line, inst_t *ra)
+{
+	KTRACE_ENTER(ip, INODE_KTRACE_HOLD, file, line, ra);
+}
+
+void
+_xfs_itrace_ref(xfs_inode_t *ip, char *file, int line, inst_t *ra)
+{
+	KTRACE_ENTER(ip, INODE_KTRACE_REF, file, line, ra);
+}
+
+void
+xfs_itrace_rele(xfs_inode_t *ip, char *file, int line, inst_t *ra)
+{
+	KTRACE_ENTER(ip, INODE_KTRACE_RELE, file, line, ra);
+}
+#endif	/* XFS_INODE_TRACE */
diff --git a/fs/xfs/xfs_imap.h b/fs/xfs/xfs_imap.h
deleted file mode 100644
index d3645000398..00000000000
--- a/fs/xfs/xfs_imap.h
+++ /dev/null
@@ -1,40 +0,0 @@
-/*
- * Copyright (c) 2000,2005 Silicon Graphics, Inc.
- * All Rights Reserved.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it would be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write the Free Software Foundation,
- * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#ifndef __XFS_IMAP_H__
-#define	__XFS_IMAP_H__
-
-/*
- * This is the structure passed to xfs_imap() to map
- * an inode number to its on disk location.
- */
-typedef struct xfs_imap {
-	xfs_daddr_t	im_blkno;	/* starting BB of inode chunk */
-	uint		im_len;		/* length in BBs of inode chunk */
-	xfs_agblock_t	im_agblkno;	/* logical block of inode chunk in ag */
-	ushort		im_ioffset;	/* inode offset in block in "inodes" */
-	ushort		im_boffset;	/* inode offset in block in bytes */
-} xfs_imap_t;
-
-#ifdef __KERNEL__
-struct xfs_mount;
-struct xfs_trans;
-int	xfs_imap(struct xfs_mount *, struct xfs_trans *, xfs_ino_t,
-		 xfs_imap_t *, uint);
-#endif
-
-#endif	/* __XFS_IMAP_H__ */
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index a391b955df0..5a5e035e5d3 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -23,7 +23,6 @@
 #include "xfs_bit.h"
 #include "xfs_log.h"
 #include "xfs_inum.h"
-#include "xfs_imap.h"
 #include "xfs_trans.h"
 #include "xfs_trans_priv.h"
 #include "xfs_sb.h"
@@ -41,6 +40,7 @@
 #include "xfs_buf_item.h"
 #include "xfs_inode_item.h"
 #include "xfs_btree.h"
+#include "xfs_btree_trace.h"
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_bmap.h"
@@ -133,10 +133,10 @@ STATIC int
 xfs_imap_to_bp(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
-	xfs_imap_t	*imap,
+	struct xfs_imap	*imap,
 	xfs_buf_t	**bpp,
 	uint		buf_flags,
-	uint		imap_flags)
+	uint		iget_flags)
 {
 	int		error;
 	int		i;
@@ -173,12 +173,12 @@ xfs_imap_to_bp(
 
 		dip = (xfs_dinode_t *)xfs_buf_offset(bp,
 					(i << mp->m_sb.sb_inodelog));
-		di_ok = be16_to_cpu(dip->di_core.di_magic) == XFS_DINODE_MAGIC &&
-			    XFS_DINODE_GOOD_VERSION(dip->di_core.di_version);
+		di_ok = be16_to_cpu(dip->di_magic) == XFS_DINODE_MAGIC &&
+			    XFS_DINODE_GOOD_VERSION(dip->di_version);
 		if (unlikely(XFS_TEST_ERROR(!di_ok, mp,
 						XFS_ERRTAG_ITOBP_INOTOBP,
 						XFS_RANDOM_ITOBP_INOTOBP))) {
-			if (imap_flags & XFS_IMAP_BULKSTAT) {
+			if (iget_flags & XFS_IGET_BULKSTAT) {
 				xfs_trans_brelse(tp, bp);
 				return XFS_ERROR(EINVAL);
 			}
@@ -190,7 +190,7 @@ xfs_imap_to_bp(
 					"daddr %lld #%d (magic=%x)",
 				XFS_BUFTARG_NAME(mp->m_ddev_targp),
 				(unsigned long long)imap->im_blkno, i,
-				be16_to_cpu(dip->di_core.di_magic));
+				be16_to_cpu(dip->di_magic));
 #endif
 			xfs_trans_brelse(tp, bp);
 			return XFS_ERROR(EFSCORRUPTED);
@@ -221,25 +221,26 @@ xfs_imap_to_bp(
  * Use xfs_imap() to determine the size and location of the
  * buffer to read from disk.
  */
-STATIC int
+int
 xfs_inotobp(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
 	xfs_ino_t	ino,
 	xfs_dinode_t	**dipp,
 	xfs_buf_t	**bpp,
-	int		*offset)
+	int		*offset,
+	uint		imap_flags)
 {
-	xfs_imap_t	imap;
+	struct xfs_imap	imap;
 	xfs_buf_t	*bp;
 	int		error;
 
 	imap.im_blkno = 0;
-	error = xfs_imap(mp, tp, ino, &imap, XFS_IMAP_LOOKUP);
+	error = xfs_imap(mp, tp, ino, &imap, imap_flags);
 	if (error)
 		return error;
 
-	error = xfs_imap_to_bp(mp, tp, &imap, &bp, XFS_BUF_LOCK, 0);
+	error = xfs_imap_to_bp(mp, tp, &imap, &bp, XFS_BUF_LOCK, imap_flags);
 	if (error)
 		return error;
 
@@ -260,15 +261,11 @@ xfs_inotobp(
  * If a non-zero error is returned, then the contents of bpp and
  * dipp are undefined.
  *
- * If the inode is new and has not yet been initialized, use xfs_imap()
- * to determine the size and location of the buffer to read from disk.
- * If the inode has already been mapped to its buffer and read in once,
- * then use the mapping information stored in the inode rather than
- * calling xfs_imap().  This allows us to avoid the overhead of looking
- * at the inode btree for small block file systems (see xfs_dilocate()).
- * We can tell whether the inode has been mapped in before by comparing
- * its disk block address to 0.  Only uninitialized inodes will have
- * 0 for the disk block address.
+ * The inode is expected to already been mapped to its buffer and read
+ * in once, thus we can use the mapping information stored in the inode
+ * rather than calling xfs_imap().  This allows us to avoid the overhead
+ * of looking at the inode btree for small block file systems
+ * (see xfs_imap()).
  */
 int
 xfs_itobp(
@@ -277,40 +274,14 @@ xfs_itobp(
 	xfs_inode_t	*ip,
 	xfs_dinode_t	**dipp,
 	xfs_buf_t	**bpp,
-	xfs_daddr_t	bno,
-	uint		imap_flags,
 	uint		buf_flags)
 {
-	xfs_imap_t	imap;
 	xfs_buf_t	*bp;
 	int		error;
 
-	if (ip->i_blkno == (xfs_daddr_t)0) {
-		imap.im_blkno = bno;
-		error = xfs_imap(mp, tp, ip->i_ino, &imap,
-					XFS_IMAP_LOOKUP | imap_flags);
-		if (error)
-			return error;
+	ASSERT(ip->i_imap.im_blkno != 0);
 
-		/*
-		 * Fill in the fields in the inode that will be used to
-		 * map the inode to its buffer from now on.
-		 */
-		ip->i_blkno = imap.im_blkno;
-		ip->i_len = imap.im_len;
-		ip->i_boffset = imap.im_boffset;
-	} else {
-		/*
-		 * We've already mapped the inode once, so just use the
-		 * mapping that we saved the first time.
-		 */
-		imap.im_blkno = ip->i_blkno;
-		imap.im_len = ip->i_len;
-		imap.im_boffset = ip->i_boffset;
-	}
-	ASSERT(bno == 0 || bno == imap.im_blkno);
-
-	error = xfs_imap_to_bp(mp, tp, &imap, &bp, buf_flags, imap_flags);
+	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &bp, buf_flags, 0);
 	if (error)
 		return error;
 
@@ -321,7 +292,7 @@ xfs_itobp(
 		return EAGAIN;
 	}
 
-	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
+	*dipp = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
 	*bpp = bp;
 	return 0;
 }
@@ -348,26 +319,26 @@ xfs_iformat(
 		XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
 	error = 0;
 
-	if (unlikely(be32_to_cpu(dip->di_core.di_nextents) +
-		     be16_to_cpu(dip->di_core.di_anextents) >
-		     be64_to_cpu(dip->di_core.di_nblocks))) {
+	if (unlikely(be32_to_cpu(dip->di_nextents) +
+		     be16_to_cpu(dip->di_anextents) >
+		     be64_to_cpu(dip->di_nblocks))) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt dinode %Lu, extent total = %d, nblocks = %Lu.",
 			(unsigned long long)ip->i_ino,
-			(int)(be32_to_cpu(dip->di_core.di_nextents) +
-			      be16_to_cpu(dip->di_core.di_anextents)),
+			(int)(be32_to_cpu(dip->di_nextents) +
+			      be16_to_cpu(dip->di_anextents)),
 			(unsigned long long)
-				be64_to_cpu(dip->di_core.di_nblocks));
+				be64_to_cpu(dip->di_nblocks));
 		XFS_CORRUPTION_ERROR("xfs_iformat(1)", XFS_ERRLEVEL_LOW,
 				     ip->i_mount, dip);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
 
-	if (unlikely(dip->di_core.di_forkoff > ip->i_mount->m_sb.sb_inodesize)) {
+	if (unlikely(dip->di_forkoff > ip->i_mount->m_sb.sb_inodesize)) {
 		xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 			"corrupt dinode %Lu, forkoff = 0x%x.",
 			(unsigned long long)ip->i_ino,
-			dip->di_core.di_forkoff);
+			dip->di_forkoff);
 		XFS_CORRUPTION_ERROR("xfs_iformat(2)", XFS_ERRLEVEL_LOW,
 				     ip->i_mount, dip);
 		return XFS_ERROR(EFSCORRUPTED);
@@ -378,25 +349,25 @@ xfs_iformat(
 	case S_IFCHR:
 	case S_IFBLK:
 	case S_IFSOCK:
-		if (unlikely(dip->di_core.di_format != XFS_DINODE_FMT_DEV)) {
+		if (unlikely(dip->di_format != XFS_DINODE_FMT_DEV)) {
 			XFS_CORRUPTION_ERROR("xfs_iformat(3)", XFS_ERRLEVEL_LOW,
 					      ip->i_mount, dip);
 			return XFS_ERROR(EFSCORRUPTED);
 		}
 		ip->i_d.di_size = 0;
 		ip->i_size = 0;
-		ip->i_df.if_u2.if_rdev = be32_to_cpu(dip->di_u.di_dev);
+		ip->i_df.if_u2.if_rdev = xfs_dinode_get_rdev(dip);
 		break;
 
 	case S_IFREG:
 	case S_IFLNK:
 	case S_IFDIR:
-		switch (dip->di_core.di_format) {
+		switch (dip->di_format) {
 		case XFS_DINODE_FMT_LOCAL:
 			/*
 			 * no local regular files yet
 			 */
-			if (unlikely((be16_to_cpu(dip->di_core.di_mode) & S_IFMT) == S_IFREG)) {
+			if (unlikely((be16_to_cpu(dip->di_mode) & S_IFMT) == S_IFREG)) {
 				xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 					"corrupt inode %Lu "
 					"(local format for regular file).",
@@ -407,7 +378,7 @@ xfs_iformat(
 				return XFS_ERROR(EFSCORRUPTED);
 			}
 
-			di_size = be64_to_cpu(dip->di_core.di_size);
+			di_size = be64_to_cpu(dip->di_size);
 			if (unlikely(di_size > XFS_DFORK_DSIZE(dip, ip->i_mount))) {
 				xfs_fs_repair_cmn_err(CE_WARN, ip->i_mount,
 					"corrupt inode %Lu "
@@ -449,7 +420,7 @@ xfs_iformat(
 	ip->i_afp = kmem_zone_zalloc(xfs_ifork_zone, KM_SLEEP);
 	ip->i_afp->if_ext_max =
 		XFS_IFORK_ASIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
-	switch (dip->di_core.di_aformat) {
+	switch (dip->di_aformat) {
 	case XFS_DINODE_FMT_LOCAL:
 		atp = (xfs_attr_shortform_t *)XFS_DFORK_APTR(dip);
 		size = be16_to_cpu(atp->hdr.totsize);
@@ -621,7 +592,7 @@ xfs_iformat_btree(
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	dfp = (xfs_bmdr_block_t *)XFS_DFORK_PTR(dip, whichfork);
 	size = XFS_BMAP_BROOT_SPACE(dfp);
-	nrecs = XFS_BMAP_BROOT_NUMRECS(dfp);
+	nrecs = be16_to_cpu(dfp->bb_numrecs);
 
 	/*
 	 * blow out if -- fork has less extents than can fit in
@@ -649,8 +620,9 @@ xfs_iformat_btree(
 	 * Copy and convert from the on-disk structure
 	 * to the in-memory structure.
 	 */
-	xfs_bmdr_to_bmbt(dfp, XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
-		ifp->if_broot, size);
+	xfs_bmdr_to_bmbt(ip->i_mount, dfp,
+			 XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
+			 ifp->if_broot, size);
 	ifp->if_flags &= ~XFS_IFEXTENTS;
 	ifp->if_flags |= XFS_IFBROOT;
 
@@ -660,7 +632,7 @@ xfs_iformat_btree(
 void
 xfs_dinode_from_disk(
 	xfs_icdinode_t		*to,
-	xfs_dinode_core_t	*from)
+	xfs_dinode_t		*from)
 {
 	to->di_magic = be16_to_cpu(from->di_magic);
 	to->di_mode = be16_to_cpu(from->di_mode);
@@ -694,7 +666,7 @@ xfs_dinode_from_disk(
 
 void
 xfs_dinode_to_disk(
-	xfs_dinode_core_t	*to,
+	xfs_dinode_t		*to,
 	xfs_icdinode_t		*from)
 {
 	to->di_magic = cpu_to_be16(from->di_magic);
@@ -781,93 +753,57 @@ uint
 xfs_dic2xflags(
 	xfs_dinode_t		*dip)
 {
-	xfs_dinode_core_t	*dic = &dip->di_core;
-
-	return _xfs_dic2xflags(be16_to_cpu(dic->di_flags)) |
+	return _xfs_dic2xflags(be16_to_cpu(dip->di_flags)) |
 				(XFS_DFORK_Q(dip) ? XFS_XFLAG_HASATTR : 0);
 }
 
 /*
- * Given a mount structure and an inode number, return a pointer
- * to a newly allocated in-core inode corresponding to the given
- * inode number.
- *
- * Initialize the inode's attributes and extent pointers if it
- * already has them (it will not if the inode has no links).
+ * Read the disk inode attributes into the in-core inode structure.
  */
 int
 xfs_iread(
 	xfs_mount_t	*mp,
 	xfs_trans_t	*tp,
-	xfs_ino_t	ino,
-	xfs_inode_t	**ipp,
+	xfs_inode_t	*ip,
 	xfs_daddr_t	bno,
-	uint		imap_flags)
+	uint		iget_flags)
 {
 	xfs_buf_t	*bp;
 	xfs_dinode_t	*dip;
-	xfs_inode_t	*ip;
 	int		error;
 
-	ASSERT(xfs_inode_zone != NULL);
-
-	ip = kmem_zone_zalloc(xfs_inode_zone, KM_SLEEP);
-	ip->i_ino = ino;
-	ip->i_mount = mp;
-	atomic_set(&ip->i_iocount, 0);
-	spin_lock_init(&ip->i_flags_lock);
-
 	/*
-	 * Get pointer's to the on-disk inode and the buffer containing it.
-	 * If the inode number refers to a block outside the file system
-	 * then xfs_itobp() will return NULL.  In this case we should
-	 * return NULL as well.  Set i_blkno to 0 so that xfs_itobp() will
-	 * know that this is a new incore inode.
+	 * Fill in the location information in the in-core inode.
 	 */
-	error = xfs_itobp(mp, tp, ip, &dip, &bp, bno, imap_flags, XFS_BUF_LOCK);
-	if (error) {
-		kmem_zone_free(xfs_inode_zone, ip);
+	ip->i_imap.im_blkno = bno;
+	error = xfs_imap(mp, tp, ip->i_ino, &ip->i_imap, iget_flags);
+	if (error)
 		return error;
-	}
+	ASSERT(bno == 0 || bno == ip->i_imap.im_blkno);
 
 	/*
-	 * Initialize inode's trace buffers.
-	 * Do this before xfs_iformat in case it adds entries.
+	 * Get pointers to the on-disk inode and the buffer containing it.
 	 */
-#ifdef	XFS_INODE_TRACE
-	ip->i_trace = ktrace_alloc(INODE_TRACE_SIZE, KM_NOFS);
-#endif
-#ifdef XFS_BMAP_TRACE
-	ip->i_xtrace = ktrace_alloc(XFS_BMAP_KTRACE_SIZE, KM_NOFS);
-#endif
-#ifdef XFS_BMBT_TRACE
-	ip->i_btrace = ktrace_alloc(XFS_BMBT_KTRACE_SIZE, KM_NOFS);
-#endif
-#ifdef XFS_RW_TRACE
-	ip->i_rwtrace = ktrace_alloc(XFS_RW_KTRACE_SIZE, KM_NOFS);
-#endif
-#ifdef XFS_ILOCK_TRACE
-	ip->i_lock_trace = ktrace_alloc(XFS_ILOCK_KTRACE_SIZE, KM_NOFS);
-#endif
-#ifdef XFS_DIR2_TRACE
-	ip->i_dir_trace = ktrace_alloc(XFS_DIR2_KTRACE_SIZE, KM_NOFS);
-#endif
+	error = xfs_imap_to_bp(mp, tp, &ip->i_imap, &bp,
+			       XFS_BUF_LOCK, iget_flags);
+	if (error)
+		return error;
+	dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
 
 	/*
 	 * If we got something that isn't an inode it means someone
 	 * (nfs or dmi) has a stale handle.
 	 */
-	if (be16_to_cpu(dip->di_core.di_magic) != XFS_DINODE_MAGIC) {
-		kmem_zone_free(xfs_inode_zone, ip);
-		xfs_trans_brelse(tp, bp);
+	if (be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC) {
 #ifdef DEBUG
 		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_iread: "
-				"dip->di_core.di_magic (0x%x) != "
+				"dip->di_magic (0x%x) != "
 				"XFS_DINODE_MAGIC (0x%x)",
-				be16_to_cpu(dip->di_core.di_magic),
+				be16_to_cpu(dip->di_magic),
 				XFS_DINODE_MAGIC);
 #endif /* DEBUG */
-		return XFS_ERROR(EINVAL);
+		error = XFS_ERROR(EINVAL);
+		goto out_brelse;
 	}
 
 	/*
@@ -877,24 +813,22 @@ xfs_iread(
 	 * specific information.
 	 * Otherwise, just get the truly permanent information.
 	 */
-	if (dip->di_core.di_mode) {
-		xfs_dinode_from_disk(&ip->i_d, &dip->di_core);
+	if (dip->di_mode) {
+		xfs_dinode_from_disk(&ip->i_d, dip);
 		error = xfs_iformat(ip, dip);
 		if (error)  {
-			kmem_zone_free(xfs_inode_zone, ip);
-			xfs_trans_brelse(tp, bp);
 #ifdef DEBUG
 			xfs_fs_cmn_err(CE_ALERT, mp, "xfs_iread: "
 					"xfs_iformat() returned error %d",
 					error);
 #endif /* DEBUG */
-			return error;
+			goto out_brelse;
 		}
 	} else {
-		ip->i_d.di_magic = be16_to_cpu(dip->di_core.di_magic);
-		ip->i_d.di_version = dip->di_core.di_version;
-		ip->i_d.di_gen = be32_to_cpu(dip->di_core.di_gen);
-		ip->i_d.di_flushiter = be16_to_cpu(dip->di_core.di_flushiter);
+		ip->i_d.di_magic = be16_to_cpu(dip->di_magic);
+		ip->i_d.di_version = dip->di_version;
+		ip->i_d.di_gen = be32_to_cpu(dip->di_gen);
+		ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter);
 		/*
 		 * Make sure to pull in the mode here as well in
 		 * case the inode is released without being used.
@@ -911,8 +845,6 @@ xfs_iread(
 			XFS_IFORK_DSIZE(ip) / (uint)sizeof(xfs_bmbt_rec_t);
 	}
 
-	INIT_LIST_HEAD(&ip->i_reclaim);
-
 	/*
 	 * The inode format changed when we moved the link count and
 	 * made it 32 bits long.  If this is an old format inode,
@@ -924,7 +856,7 @@ xfs_iread(
 	 * the new format. We don't change the version number so that we
 	 * can distinguish this from a real new format inode.
 	 */
-	if (ip->i_d.di_version == XFS_DINODE_VERSION_1) {
+	if (ip->i_d.di_version == 1) {
 		ip->i_d.di_nlink = ip->i_d.di_onlink;
 		ip->i_d.di_onlink = 0;
 		ip->i_d.di_projid = 0;
@@ -938,7 +870,7 @@ xfs_iread(
 	 * around for a while.  This helps to keep recently accessed
 	 * meta-data in-core longer.
 	 */
-	 XFS_BUF_SET_REF(bp, XFS_INO_REF);
+	XFS_BUF_SET_REF(bp, XFS_INO_REF);
 
 	/*
 	 * Use xfs_trans_brelse() to release the buffer containing the
@@ -953,9 +885,9 @@ xfs_iread(
 	 * to worry about the inode being changed just because we released
 	 * the buffer.
 	 */
+ out_brelse:
 	xfs_trans_brelse(tp, bp);
-	*ipp = ip;
-	return 0;
+	return error;
 }
 
 /*
@@ -1049,6 +981,7 @@ xfs_ialloc(
 	uint		flags;
 	int		error;
 	timespec_t	tv;
+	int		filestreams = 0;
 
 	/*
 	 * Call the space management code to pick
@@ -1056,9 +989,8 @@ xfs_ialloc(
 	 */
 	error = xfs_dialloc(tp, pip ? pip->i_ino : 0, mode, okalloc,
 			    ialloc_context, call_again, &ino);
-	if (error != 0) {
+	if (error)
 		return error;
-	}
 	if (*call_again || ino == NULLFSINO) {
 		*ipp = NULL;
 		return 0;
@@ -1072,9 +1004,8 @@ xfs_ialloc(
 	 */
 	error = xfs_trans_iget(tp->t_mountp, tp, ino,
 				XFS_IGET_CREATE, XFS_ILOCK_EXCL, &ip);
-	if (error != 0) {
+	if (error)
 		return error;
-	}
 	ASSERT(ip != NULL);
 
 	ip->i_d.di_mode = (__uint16_t)mode;
@@ -1093,8 +1024,8 @@ xfs_ialloc(
 	 * here rather than here and in the flush/logging code.
 	 */
 	if (xfs_sb_version_hasnlink(&tp->t_mountp->m_sb) &&
-	    ip->i_d.di_version == XFS_DINODE_VERSION_1) {
-		ip->i_d.di_version = XFS_DINODE_VERSION_2;
+	    ip->i_d.di_version == 1) {
+		ip->i_d.di_version = 2;
 		/*
 		 * We've already zeroed the old link count, the projid field,
 		 * and the pad field.
@@ -1104,7 +1035,7 @@ xfs_ialloc(
 	/*
 	 * Project ids won't be stored on disk if we are using a version 1 inode.
 	 */
-	if ((prid != 0) && (ip->i_d.di_version == XFS_DINODE_VERSION_1))
+	if ((prid != 0) && (ip->i_d.di_version == 1))
 		xfs_bump_ino_vers2(tp, ip);
 
 	if (pip && XFS_INHERIT_GID(pip)) {
@@ -1155,13 +1086,12 @@ xfs_ialloc(
 		flags |= XFS_ILOG_DEV;
 		break;
 	case S_IFREG:
-		if (pip && xfs_inode_is_filestream(pip)) {
-			error = xfs_filestream_associate(pip, ip);
-			if (error < 0)
-				return -error;
-			if (!error)
-				xfs_iflags_set(ip, XFS_IFILESTREAM);
-		}
+		/*
+		 * we can't set up filestreams until after the VFS inode
+		 * is set up properly.
+		 */
+		if (pip && xfs_inode_is_filestream(pip))
+			filestreams = 1;
 		/* fall through */
 	case S_IFDIR:
 		if (pip && (pip->i_d.di_flags & XFS_DIFLAG_ANY)) {
@@ -1227,6 +1157,15 @@ xfs_ialloc(
 	/* now that we have an i_mode we can setup inode ops and unlock */
 	xfs_setup_inode(ip);
 
+	/* now we have set up the vfs inode we can associate the filestream */
+	if (filestreams) {
+		error = xfs_filestream_associate(pip, ip);
+		if (error < 0)
+			return -error;
+		if (!error)
+			xfs_iflags_set(ip, XFS_IFILESTREAM);
+	}
+
 	*ipp = ip;
 	return 0;
 }
@@ -1383,8 +1322,8 @@ xfs_itrunc_trace(
  * direct I/O with the truncate operation.  Also, because we hold
  * the IOLOCK in exclusive mode, we prevent new direct I/Os from being
  * started until the truncate completes and drops the lock. Essentially,
- * the vn_iowait() call forms an I/O barrier that provides strict ordering
- * between direct I/Os and the truncate operation.
+ * the xfs_ioend_wait() call forms an I/O barrier that provides strict
+ * ordering between direct I/Os and the truncate operation.
  *
  * The flags parameter can have either the value XFS_ITRUNC_DEFINITE
  * or XFS_ITRUNC_MAYBE.  The XFS_ITRUNC_MAYBE value should be used
@@ -1415,7 +1354,7 @@ xfs_itruncate_start(
 
 	/* wait for the completion of any pending DIOs */
 	if (new_size == 0 || new_size < ip->i_size)
-		vn_iowait(ip);
+		xfs_ioend_wait(ip);
 
 	/*
 	 * Call toss_pages or flushinval_pages to get rid of pages
@@ -1726,8 +1665,14 @@ xfs_itruncate_finish(
 		xfs_trans_ijoin(ntp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 		xfs_trans_ihold(ntp, ip);
 
-		if (!error)
-			error = xfs_trans_reserve(ntp, 0,
+		if (error)
+			return error;
+		/*
+		 * transaction commit worked ok so we can drop the extra ticket
+		 * reference that we gained in xfs_trans_dup()
+		 */
+		xfs_log_ticket_put(ntp->t_ticket);
+		error = xfs_trans_reserve(ntp, 0,
 					XFS_ITRUNCATE_LOG_RES(mp), 0,
 					XFS_TRANS_PERM_LOG_RES,
 					XFS_ITRUNCATE_LOG_COUNT);
@@ -1781,13 +1726,10 @@ xfs_iunlink(
 	xfs_dinode_t	*dip;
 	xfs_buf_t	*agibp;
 	xfs_buf_t	*ibp;
-	xfs_agnumber_t	agno;
-	xfs_daddr_t	agdaddr;
 	xfs_agino_t	agino;
 	short		bucket_index;
 	int		offset;
 	int		error;
-	int		agi_ok;
 
 	ASSERT(ip->i_d.di_nlink == 0);
 	ASSERT(ip->i_d.di_mode != 0);
@@ -1795,31 +1737,15 @@ xfs_iunlink(
 
 	mp = tp->t_mountp;
 
-	agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
-	agdaddr = XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp));
-
 	/*
 	 * Get the agi buffer first.  It ensures lock ordering
 	 * on the list.
 	 */
-	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, agdaddr,
-				   XFS_FSS_TO_BB(mp, 1), 0, &agibp);
+	error = xfs_read_agi(mp, tp, XFS_INO_TO_AGNO(mp, ip->i_ino), &agibp);
 	if (error)
 		return error;
-
-	/*
-	 * Validate the magic number of the agi block.
-	 */
 	agi = XFS_BUF_TO_AGI(agibp);
-	agi_ok =
-		be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
-		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
-	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IUNLINK,
-			XFS_RANDOM_IUNLINK))) {
-		XFS_CORRUPTION_ERROR("xfs_iunlink", XFS_ERRLEVEL_LOW, mp, agi);
-		xfs_trans_brelse(tp, agibp);
-		return XFS_ERROR(EFSCORRUPTED);
-	}
+
 	/*
 	 * Get the index into the agi hash table for the
 	 * list this inode will go on.
@@ -1837,14 +1763,14 @@ xfs_iunlink(
 		 * Here we put the head pointer into our next pointer,
 		 * and then we fall through to point the head at us.
 		 */
-		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, XFS_BUF_LOCK);
 		if (error)
 			return error;
 
 		ASSERT(be32_to_cpu(dip->di_next_unlinked) == NULLAGINO);
 		/* both on-disk, don't endian flip twice */
 		dip->di_next_unlinked = agi->agi_unlinked[bucket_index];
-		offset = ip->i_boffset +
+		offset = ip->i_imap.im_boffset +
 			offsetof(xfs_dinode_t, di_next_unlinked);
 		xfs_trans_inode_buf(tp, ibp);
 		xfs_trans_log_buf(tp, ibp, offset,
@@ -1879,7 +1805,6 @@ xfs_iunlink_remove(
 	xfs_buf_t	*agibp;
 	xfs_buf_t	*ibp;
 	xfs_agnumber_t	agno;
-	xfs_daddr_t	agdaddr;
 	xfs_agino_t	agino;
 	xfs_agino_t	next_agino;
 	xfs_buf_t	*last_ibp;
@@ -1887,45 +1812,20 @@ xfs_iunlink_remove(
 	short		bucket_index;
 	int		offset, last_offset = 0;
 	int		error;
-	int		agi_ok;
 
-	/*
-	 * First pull the on-disk inode from the AGI unlinked list.
-	 */
 	mp = tp->t_mountp;
-
 	agno = XFS_INO_TO_AGNO(mp, ip->i_ino);
-	agdaddr = XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp));
 
 	/*
 	 * Get the agi buffer first.  It ensures lock ordering
 	 * on the list.
 	 */
-	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, agdaddr,
-				   XFS_FSS_TO_BB(mp, 1), 0, &agibp);
-	if (error) {
-		cmn_err(CE_WARN,
-			"xfs_iunlink_remove: xfs_trans_read_buf()  returned an error %d on %s.  Returning error.",
-			error, mp->m_fsname);
+	error = xfs_read_agi(mp, tp, agno, &agibp);
+	if (error)
 		return error;
-	}
-	/*
-	 * Validate the magic number of the agi block.
-	 */
+
 	agi = XFS_BUF_TO_AGI(agibp);
-	agi_ok =
-		be32_to_cpu(agi->agi_magicnum) == XFS_AGI_MAGIC &&
-		XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
-	if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IUNLINK_REMOVE,
-			XFS_RANDOM_IUNLINK_REMOVE))) {
-		XFS_CORRUPTION_ERROR("xfs_iunlink_remove", XFS_ERRLEVEL_LOW,
-				     mp, agi);
-		xfs_trans_brelse(tp, agibp);
-		cmn_err(CE_WARN,
-			"xfs_iunlink_remove: XFS_TEST_ERROR()  returned an error on %s.  Returning EFSCORRUPTED.",
-			 mp->m_fsname);
-		return XFS_ERROR(EFSCORRUPTED);
-	}
+
 	/*
 	 * Get the index into the agi hash table for the
 	 * list this inode will go on.
@@ -1945,7 +1845,7 @@ xfs_iunlink_remove(
 		 * of dealing with the buffer when there is no need to
 		 * change it.
 		 */
-		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, XFS_BUF_LOCK);
 		if (error) {
 			cmn_err(CE_WARN,
 				"xfs_iunlink_remove: xfs_itobp()  returned an error %d on %s.  Returning error.",
@@ -1956,7 +1856,7 @@ xfs_iunlink_remove(
 		ASSERT(next_agino != 0);
 		if (next_agino != NULLAGINO) {
 			dip->di_next_unlinked = cpu_to_be32(NULLAGINO);
-			offset = ip->i_boffset +
+			offset = ip->i_imap.im_boffset +
 				offsetof(xfs_dinode_t, di_next_unlinked);
 			xfs_trans_inode_buf(tp, ibp);
 			xfs_trans_log_buf(tp, ibp, offset,
@@ -1992,7 +1892,7 @@ xfs_iunlink_remove(
 			}
 			next_ino = XFS_AGINO_TO_INO(mp, agno, next_agino);
 			error = xfs_inotobp(mp, tp, next_ino, &last_dip,
-					    &last_ibp, &last_offset);
+					    &last_ibp, &last_offset, 0);
 			if (error) {
 				cmn_err(CE_WARN,
 			"xfs_iunlink_remove: xfs_inotobp()  returned an error %d on %s.  Returning error.",
@@ -2007,7 +1907,7 @@ xfs_iunlink_remove(
 		 * Now last_ibp points to the buffer previous to us on
 		 * the unlinked list.  Pull us from the list.
 		 */
-		error = xfs_itobp(mp, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
+		error = xfs_itobp(mp, tp, ip, &dip, &ibp, XFS_BUF_LOCK);
 		if (error) {
 			cmn_err(CE_WARN,
 				"xfs_iunlink_remove: xfs_itobp()  returned an error %d on %s.  Returning error.",
@@ -2019,7 +1919,7 @@ xfs_iunlink_remove(
 		ASSERT(next_agino != agino);
 		if (next_agino != NULLAGINO) {
 			dip->di_next_unlinked = cpu_to_be32(NULLAGINO);
-			offset = ip->i_boffset +
+			offset = ip->i_imap.im_boffset +
 				offsetof(xfs_dinode_t, di_next_unlinked);
 			xfs_trans_inode_buf(tp, ibp);
 			xfs_trans_log_buf(tp, ibp, offset,
@@ -2160,9 +2060,9 @@ xfs_ifree_cluster(
 				iip = (xfs_inode_log_item_t *)lip;
 				ASSERT(iip->ili_logged == 1);
 				lip->li_cb = (void(*)(xfs_buf_t*,xfs_log_item_t*)) xfs_istale_done;
-				spin_lock(&mp->m_ail_lock);
-				iip->ili_flush_lsn = iip->ili_item.li_lsn;
-				spin_unlock(&mp->m_ail_lock);
+				xfs_trans_ail_copy_lsn(mp->m_ail,
+							&iip->ili_flush_lsn,
+							&iip->ili_item.li_lsn);
 				xfs_iflags_set(iip->ili_inode, XFS_ISTALE);
 				pre_flushed++;
 			}
@@ -2183,9 +2083,8 @@ xfs_ifree_cluster(
 			iip->ili_last_fields = iip->ili_format.ilf_fields;
 			iip->ili_format.ilf_fields = 0;
 			iip->ili_logged = 1;
-			spin_lock(&mp->m_ail_lock);
-			iip->ili_flush_lsn = iip->ili_item.li_lsn;
-			spin_unlock(&mp->m_ail_lock);
+			xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
+						&iip->ili_item.li_lsn);
 
 			xfs_buf_attach_iodone(bp,
 				(void(*)(xfs_buf_t*,xfs_log_item_t*))
@@ -2263,7 +2162,7 @@ xfs_ifree(
 
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 
-	error = xfs_itobp(ip->i_mount, tp, ip, &dip, &ibp, 0, 0, XFS_BUF_LOCK);
+	error = xfs_itobp(ip->i_mount, tp, ip, &dip, &ibp, XFS_BUF_LOCK);
 	if (error)
 		return error;
 
@@ -2279,7 +2178,7 @@ xfs_ifree(
 	* This is a temporary hack that would require a proper fix
 	* in the future.
 	*/
-	dip->di_core.di_mode = 0;
+	dip->di_mode = 0;
 
 	if (delete) {
 		xfs_ifree_cluster(ip, tp, first_ino);
@@ -2312,9 +2211,10 @@ xfs_iroot_realloc(
 	int			rec_diff,
 	int			whichfork)
 {
+	struct xfs_mount	*mp = ip->i_mount;
 	int			cur_max;
 	xfs_ifork_t		*ifp;
-	xfs_bmbt_block_t	*new_broot;
+	struct xfs_btree_block	*new_broot;
 	int			new_max;
 	size_t			new_size;
 	char			*np;
@@ -2335,8 +2235,7 @@ xfs_iroot_realloc(
 		 */
 		if (ifp->if_broot_bytes == 0) {
 			new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(rec_diff);
-			ifp->if_broot = (xfs_bmbt_block_t*)kmem_alloc(new_size,
-								     KM_SLEEP);
+			ifp->if_broot = kmem_alloc(new_size, KM_SLEEP);
 			ifp->if_broot_bytes = (int)new_size;
 			return;
 		}
@@ -2347,18 +2246,16 @@ xfs_iroot_realloc(
 		 * location.  The records don't change location because
 		 * they are kept butted up against the btree block header.
 		 */
-		cur_max = XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes);
+		cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
 		new_max = cur_max + rec_diff;
 		new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
-		ifp->if_broot = (xfs_bmbt_block_t *)
-		  kmem_realloc(ifp->if_broot,
-				new_size,
+		ifp->if_broot = kmem_realloc(ifp->if_broot, new_size,
 				(size_t)XFS_BMAP_BROOT_SPACE_CALC(cur_max), /* old size */
 				KM_SLEEP);
-		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(ifp->if_broot, 1,
-						      ifp->if_broot_bytes);
-		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(ifp->if_broot, 1,
-						      (int)new_size);
+		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     ifp->if_broot_bytes);
+		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
+						     (int)new_size);
 		ifp->if_broot_bytes = (int)new_size;
 		ASSERT(ifp->if_broot_bytes <=
 			XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
@@ -2372,7 +2269,7 @@ xfs_iroot_realloc(
 	 * records, just get rid of the root and clear the status bit.
 	 */
 	ASSERT((ifp->if_broot != NULL) && (ifp->if_broot_bytes > 0));
-	cur_max = XFS_BMAP_BROOT_MAXRECS(ifp->if_broot_bytes);
+	cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
 	new_max = cur_max + rec_diff;
 	ASSERT(new_max >= 0);
 	if (new_max > 0)
@@ -2380,11 +2277,11 @@ xfs_iroot_realloc(
 	else
 		new_size = 0;
 	if (new_size > 0) {
-		new_broot = (xfs_bmbt_block_t *)kmem_alloc(new_size, KM_SLEEP);
+		new_broot = kmem_alloc(new_size, KM_SLEEP);
 		/*
 		 * First copy over the btree block header.
 		 */
-		memcpy(new_broot, ifp->if_broot, sizeof(xfs_bmbt_block_t));
+		memcpy(new_broot, ifp->if_broot, XFS_BTREE_LBLOCK_LEN);
 	} else {
 		new_broot = NULL;
 		ifp->if_flags &= ~XFS_IFBROOT;
@@ -2397,18 +2294,16 @@ xfs_iroot_realloc(
 		/*
 		 * First copy the records.
 		 */
-		op = (char *)XFS_BMAP_BROOT_REC_ADDR(ifp->if_broot, 1,
-						     ifp->if_broot_bytes);
-		np = (char *)XFS_BMAP_BROOT_REC_ADDR(new_broot, 1,
-						     (int)new_size);
+		op = (char *)XFS_BMBT_REC_ADDR(mp, ifp->if_broot, 1);
+		np = (char *)XFS_BMBT_REC_ADDR(mp, new_broot, 1);
 		memcpy(np, op, new_max * (uint)sizeof(xfs_bmbt_rec_t));
 
 		/*
 		 * Then copy the pointers.
 		 */
-		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(ifp->if_broot, 1,
+		op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
 						     ifp->if_broot_bytes);
-		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(new_broot, 1,
+		np = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, new_broot, 1,
 						     (int)new_size);
 		memcpy(np, op, new_max * (uint)sizeof(xfs_dfsbno_t));
 	}
@@ -2511,64 +2406,6 @@ xfs_idata_realloc(
 	ASSERT(ifp->if_bytes <= XFS_IFORK_SIZE(ip, whichfork));
 }
 
-
-
-
-/*
- * Map inode to disk block and offset.
- *
- * mp -- the mount point structure for the current file system
- * tp -- the current transaction
- * ino -- the inode number of the inode to be located
- * imap -- this structure is filled in with the information necessary
- *	 to retrieve the given inode from disk
- * flags -- flags to pass to xfs_dilocate indicating whether or not
- *	 lookups in the inode btree were OK or not
- */
-int
-xfs_imap(
-	xfs_mount_t	*mp,
-	xfs_trans_t	*tp,
-	xfs_ino_t	ino,
-	xfs_imap_t	*imap,
-	uint		flags)
-{
-	xfs_fsblock_t	fsbno;
-	int		len;
-	int		off;
-	int		error;
-
-	fsbno = imap->im_blkno ?
-		XFS_DADDR_TO_FSB(mp, imap->im_blkno) : NULLFSBLOCK;
-	error = xfs_dilocate(mp, tp, ino, &fsbno, &len, &off, flags);
-	if (error)
-		return error;
-
-	imap->im_blkno = XFS_FSB_TO_DADDR(mp, fsbno);
-	imap->im_len = XFS_FSB_TO_BB(mp, len);
-	imap->im_agblkno = XFS_FSB_TO_AGBNO(mp, fsbno);
-	imap->im_ioffset = (ushort)off;
-	imap->im_boffset = (ushort)(off << mp->m_sb.sb_inodelog);
-
-	/*
-	 * If the inode number maps to a block outside the bounds
-	 * of the file system then return NULL rather than calling
-	 * read_buf and panicing when we get an error from the
-	 * driver.
-	 */
-	if ((imap->im_blkno + imap->im_len) >
-	    XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks)) {
-		xfs_fs_cmn_err(CE_ALERT, mp, "xfs_imap: "
-			"(imap->im_blkno (0x%llx) + imap->im_len (0x%llx)) > "
-			" XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks) (0x%llx)",
-			(unsigned long long) imap->im_blkno,
-			(unsigned long long) imap->im_len,
-			XFS_FSB_TO_BB(mp, mp->m_sb.sb_dblocks));
-		return EINVAL;
-	}
-	return 0;
-}
-
 void
 xfs_idestroy_fork(
 	xfs_inode_t	*ip,
@@ -2613,70 +2450,6 @@ xfs_idestroy_fork(
 }
 
 /*
- * This is called free all the memory associated with an inode.
- * It must free the inode itself and any buffers allocated for
- * if_extents/if_data and if_broot.  It must also free the lock
- * associated with the inode.
- */
-void
-xfs_idestroy(
-	xfs_inode_t	*ip)
-{
-	switch (ip->i_d.di_mode & S_IFMT) {
-	case S_IFREG:
-	case S_IFDIR:
-	case S_IFLNK:
-		xfs_idestroy_fork(ip, XFS_DATA_FORK);
-		break;
-	}
-	if (ip->i_afp)
-		xfs_idestroy_fork(ip, XFS_ATTR_FORK);
-	mrfree(&ip->i_lock);
-	mrfree(&ip->i_iolock);
-
-#ifdef XFS_INODE_TRACE
-	ktrace_free(ip->i_trace);
-#endif
-#ifdef XFS_BMAP_TRACE
-	ktrace_free(ip->i_xtrace);
-#endif
-#ifdef XFS_BMBT_TRACE
-	ktrace_free(ip->i_btrace);
-#endif
-#ifdef XFS_RW_TRACE
-	ktrace_free(ip->i_rwtrace);
-#endif
-#ifdef XFS_ILOCK_TRACE
-	ktrace_free(ip->i_lock_trace);
-#endif
-#ifdef XFS_DIR2_TRACE
-	ktrace_free(ip->i_dir_trace);
-#endif
-	if (ip->i_itemp) {
-		/*
-		 * Only if we are shutting down the fs will we see an
-		 * inode still in the AIL. If it is there, we should remove
-		 * it to prevent a use-after-free from occurring.
-		 */
-		xfs_mount_t	*mp = ip->i_mount;
-		xfs_log_item_t	*lip = &ip->i_itemp->ili_item;
-
-		ASSERT(((lip->li_flags & XFS_LI_IN_AIL) == 0) ||
-				       XFS_FORCED_SHUTDOWN(ip->i_mount));
-		if (lip->li_flags & XFS_LI_IN_AIL) {
-			spin_lock(&mp->m_ail_lock);
-			if (lip->li_flags & XFS_LI_IN_AIL)
-				xfs_trans_delete_ail(mp, lip);
-			else
-				spin_unlock(&mp->m_ail_lock);
-		}
-		xfs_inode_item_destroy(ip);
-	}
-	kmem_zone_free(xfs_inode_zone, ip);
-}
-
-
-/*
  * Increment the pin count of the given buffer.
  * This value is protected by ipinlock spinlock in the mount structure.
  */
@@ -2880,7 +2653,7 @@ xfs_iflush_fork(
 			ASSERT(ifp->if_broot_bytes <=
 			       (XFS_IFORK_SIZE(ip, whichfork) +
 				XFS_BROOT_SIZE_ADJ));
-			xfs_bmbt_to_bmdr(ifp->if_broot, ifp->if_broot_bytes,
+			xfs_bmbt_to_bmdr(mp, ifp->if_broot, ifp->if_broot_bytes,
 				(xfs_bmdr_block_t *)cp,
 				XFS_DFORK_SIZE(dip, mp, whichfork));
 		}
@@ -2889,15 +2662,16 @@ xfs_iflush_fork(
 	case XFS_DINODE_FMT_DEV:
 		if (iip->ili_format.ilf_fields & XFS_ILOG_DEV) {
 			ASSERT(whichfork == XFS_DATA_FORK);
-			dip->di_u.di_dev = cpu_to_be32(ip->i_df.if_u2.if_rdev);
+			xfs_dinode_put_rdev(dip, ip->i_df.if_u2.if_rdev);
 		}
 		break;
 
 	case XFS_DINODE_FMT_UUID:
 		if (iip->ili_format.ilf_fields & XFS_ILOG_UUID) {
 			ASSERT(whichfork == XFS_DATA_FORK);
-			memcpy(&dip->di_u.di_muuid, &ip->i_df.if_u2.if_uuid,
-				sizeof(uuid_t));
+			memcpy(XFS_DFORK_DPTR(dip),
+			       &ip->i_df.if_u2.if_uuid,
+			       sizeof(uuid_t));
 		}
 		break;
 
@@ -3030,7 +2804,6 @@ cluster_corrupt_out:
 			XFS_BUF_CLR_BDSTRAT_FUNC(bp);
 			XFS_BUF_UNDONE(bp);
 			XFS_BUF_STALE(bp);
-			XFS_BUF_SHUT(bp);
 			XFS_BUF_ERROR(bp,EIO);
 			xfs_biodone(bp);
 		} else {
@@ -3172,7 +2945,7 @@ xfs_iflush(
 	/*
 	 * Get the buffer containing the on-disk inode.
 	 */
-	error = xfs_itobp(mp, NULL, ip, &dip, &bp, 0, 0,
+	error = xfs_itobp(mp, NULL, ip, &dip, &bp,
 				noblock ? XFS_BUF_TRYLOCK : XFS_BUF_LOCK);
 	if (error || !bp) {
 		xfs_ifunlock(ip);
@@ -3253,7 +3026,7 @@ xfs_iflush_int(
 	}
 
 	/* set *dip = inode's place in the buffer */
-	dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_boffset);
+	dip = (xfs_dinode_t *)xfs_buf_offset(bp, ip->i_imap.im_boffset);
 
 	/*
 	 * Clear i_update_core before copying out the data.
@@ -3275,11 +3048,11 @@ xfs_iflush_int(
 	 */
 	xfs_synchronize_atime(ip);
 
-	if (XFS_TEST_ERROR(be16_to_cpu(dip->di_core.di_magic) != XFS_DINODE_MAGIC,
+	if (XFS_TEST_ERROR(be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC,
 			       mp, XFS_ERRTAG_IFLUSH_1, XFS_RANDOM_IFLUSH_1)) {
 		xfs_cmn_err(XFS_PTAG_IFLUSH, CE_ALERT, mp,
 		    "xfs_iflush: Bad inode %Lu magic number 0x%x, ptr 0x%p",
-			ip->i_ino, be16_to_cpu(dip->di_core.di_magic), dip);
+			ip->i_ino, be16_to_cpu(dip->di_magic), dip);
 		goto corrupt_out;
 	}
 	if (XFS_TEST_ERROR(ip->i_d.di_magic != XFS_DINODE_MAGIC,
@@ -3342,7 +3115,7 @@ xfs_iflush_int(
 	 * because if the inode is dirty at all the core must
 	 * be.
 	 */
-	xfs_dinode_to_disk(&dip->di_core, &ip->i_d);
+	xfs_dinode_to_disk(dip, &ip->i_d);
 
 	/* Wrap, we never let the log put out DI_MAX_FLUSH */
 	if (ip->i_d.di_flushiter == DI_MAX_FLUSH)
@@ -3354,28 +3127,27 @@ xfs_iflush_int(
 	 * convert back to the old inode format.  If the superblock version
 	 * has been updated, then make the conversion permanent.
 	 */
-	ASSERT(ip->i_d.di_version == XFS_DINODE_VERSION_1 ||
-	       xfs_sb_version_hasnlink(&mp->m_sb));
-	if (ip->i_d.di_version == XFS_DINODE_VERSION_1) {
+	ASSERT(ip->i_d.di_version == 1 || xfs_sb_version_hasnlink(&mp->m_sb));
+	if (ip->i_d.di_version == 1) {
 		if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
 			/*
 			 * Convert it back.
 			 */
 			ASSERT(ip->i_d.di_nlink <= XFS_MAXLINK_1);
-			dip->di_core.di_onlink = cpu_to_be16(ip->i_d.di_nlink);
+			dip->di_onlink = cpu_to_be16(ip->i_d.di_nlink);
 		} else {
 			/*
 			 * The superblock version has already been bumped,
 			 * so just make the conversion to the new inode
 			 * format permanent.
 			 */
-			ip->i_d.di_version = XFS_DINODE_VERSION_2;
-			dip->di_core.di_version =  XFS_DINODE_VERSION_2;
+			ip->i_d.di_version = 2;
+			dip->di_version = 2;
 			ip->i_d.di_onlink = 0;
-			dip->di_core.di_onlink = 0;
+			dip->di_onlink = 0;
 			memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
-			memset(&(dip->di_core.di_pad[0]), 0,
-			      sizeof(dip->di_core.di_pad));
+			memset(&(dip->di_pad[0]), 0,
+			      sizeof(dip->di_pad));
 			ASSERT(ip->i_d.di_projid == 0);
 		}
 	}
@@ -3418,10 +3190,8 @@ xfs_iflush_int(
 		iip->ili_format.ilf_fields = 0;
 		iip->ili_logged = 1;
 
-		ASSERT(sizeof(xfs_lsn_t) == 8);	/* don't lock if it shrinks */
-		spin_lock(&mp->m_ail_lock);
-		iip->ili_flush_lsn = iip->ili_item.li_lsn;
-		spin_unlock(&mp->m_ail_lock);
+		xfs_trans_ail_copy_lsn(mp->m_ail, &iip->ili_flush_lsn,
+					&iip->ili_item.li_lsn);
 
 		/*
 		 * Attach the function xfs_iflush_done to the inode's
@@ -3459,45 +3229,8 @@ corrupt_out:
 }
 
 
-/*
- * Flush all inactive inodes in mp.
- */
-void
-xfs_iflush_all(
-	xfs_mount_t	*mp)
-{
-	xfs_inode_t	*ip;
-
- again:
-	XFS_MOUNT_ILOCK(mp);
-	ip = mp->m_inodes;
-	if (ip == NULL)
-		goto out;
-
-	do {
-		/* Make sure we skip markers inserted by sync */
-		if (ip->i_mount == NULL) {
-			ip = ip->i_mnext;
-			continue;
-		}
-
-		if (!VFS_I(ip)) {
-			XFS_MOUNT_IUNLOCK(mp);
-			xfs_finish_reclaim(ip, 0, XFS_IFLUSH_ASYNC);
-			goto again;
-		}
-
-		ASSERT(vn_count(VFS_I(ip)) == 0);
-
-		ip = ip->i_mnext;
-	} while (ip != mp->m_inodes);
- out:
-	XFS_MOUNT_IUNLOCK(mp);
-}
 
 #ifdef XFS_ILOCK_TRACE
-ktrace_t	*xfs_ilock_trace_buf;
-
 void
 xfs_ilock_trace(xfs_inode_t *ip, int lock, unsigned int lockflags, inst_t *ra)
 {
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 6be310d41da..1f175fa34b2 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -19,8 +19,7 @@
 #define	__XFS_INODE_H__
 
 struct xfs_dinode;
-struct xfs_dinode_core;
-
+struct xfs_inode;
 
 /*
  * Fork identifiers.
@@ -63,7 +62,7 @@ typedef struct xfs_ext_irec {
 typedef struct xfs_ifork {
 	int			if_bytes;	/* bytes in if_u1 */
 	int			if_real_bytes;	/* bytes allocated in if_u1 */
-	xfs_bmbt_block_t	*if_broot;	/* file's incore btree root */
+	struct xfs_btree_block	*if_broot;	/* file's incore btree root */
 	short			if_broot_bytes;	/* bytes allocated for root */
 	unsigned char		if_flags;	/* per-fork flags */
 	unsigned char		if_ext_max;	/* max # of extent records */
@@ -84,52 +83,14 @@ typedef struct xfs_ifork {
 } xfs_ifork_t;
 
 /*
- * Flags for xfs_ichgtime().
+ * Inode location information.  Stored in the inode and passed to
+ * xfs_imap_to_bp() to get a buffer and dinode for a given inode.
  */
-#define	XFS_ICHGTIME_MOD	0x1	/* data fork modification timestamp */
-#define	XFS_ICHGTIME_CHG	0x2	/* inode field change timestamp */
-
-/*
- * Per-fork incore inode flags.
- */
-#define	XFS_IFINLINE	0x01	/* Inline data is read in */
-#define	XFS_IFEXTENTS	0x02	/* All extent pointers are read in */
-#define	XFS_IFBROOT	0x04	/* i_broot points to the bmap b-tree root */
-#define	XFS_IFEXTIREC	0x08	/* Indirection array of extent blocks */
-
-/*
- * Flags for xfs_itobp(), xfs_imap() and xfs_dilocate().
- */
-#define XFS_IMAP_LOOKUP		0x1
-#define XFS_IMAP_BULKSTAT	0x2
-
-#ifdef __KERNEL__
-struct bhv_desc;
-struct cred;
-struct ktrace;
-struct xfs_buf;
-struct xfs_bmap_free;
-struct xfs_bmbt_irec;
-struct xfs_bmbt_block;
-struct xfs_inode;
-struct xfs_inode_log_item;
-struct xfs_mount;
-struct xfs_trans;
-struct xfs_dquot;
-
-#if defined(XFS_ILOCK_TRACE)
-#define XFS_ILOCK_KTRACE_SIZE	32
-extern ktrace_t *xfs_ilock_trace_buf;
-extern void xfs_ilock_trace(struct xfs_inode *, int, unsigned int, inst_t *);
-#else
-#define	xfs_ilock_trace(i,n,f,ra)
-#endif
-
-typedef struct dm_attrs_s {
-	__uint32_t	da_dmevmask;	/* DMIG event mask */
-	__uint16_t	da_dmstate;	/* DMIG state info */
-	__uint16_t	da_pad;		/* DMIG extra padding */
-} dm_attrs_t;
+struct xfs_imap {
+	xfs_daddr_t	im_blkno;	/* starting BB of inode chunk */
+	ushort		im_len;		/* length in BBs of inode chunk */
+	ushort		im_boffset;	/* inode offset in block in bytes */
+};
 
 /*
  * This is the xfs in-core inode structure.
@@ -160,7 +121,7 @@ typedef struct xfs_ictimestamp {
 } xfs_ictimestamp_t;
 
 /*
- * NOTE:  This structure must be kept identical to struct xfs_dinode_core
+ * NOTE:  This structure must be kept identical to struct xfs_dinode
  * 	  in xfs_dinode.h except for the endianess annotations.
  */
 typedef struct xfs_icdinode {
@@ -191,27 +152,97 @@ typedef struct xfs_icdinode {
 	__uint32_t	di_gen;		/* generation number */
 } xfs_icdinode_t;
 
-typedef struct {
-	struct xfs_inode	*ip_mnext;	/* next inode in mount list */
-	struct xfs_inode	*ip_mprev;	/* ptr to prev inode */
-	struct xfs_mount	*ip_mount;	/* fs mount struct ptr */
-} xfs_iptr_t;
+/*
+ * Flags for xfs_ichgtime().
+ */
+#define	XFS_ICHGTIME_MOD	0x1	/* data fork modification timestamp */
+#define	XFS_ICHGTIME_CHG	0x2	/* inode field change timestamp */
+
+/*
+ * Per-fork incore inode flags.
+ */
+#define	XFS_IFINLINE	0x01	/* Inline data is read in */
+#define	XFS_IFEXTENTS	0x02	/* All extent pointers are read in */
+#define	XFS_IFBROOT	0x04	/* i_broot points to the bmap b-tree root */
+#define	XFS_IFEXTIREC	0x08	/* Indirection array of extent blocks */
+
+/*
+ * Fork handling.
+ */
+
+#define XFS_IFORK_Q(ip)			((ip)->i_d.di_forkoff != 0)
+#define XFS_IFORK_BOFF(ip)		((int)((ip)->i_d.di_forkoff << 3))
+
+#define XFS_IFORK_PTR(ip,w)		\
+	((w) == XFS_DATA_FORK ? \
+		&(ip)->i_df : \
+		(ip)->i_afp)
+#define XFS_IFORK_DSIZE(ip) \
+	(XFS_IFORK_Q(ip) ? \
+		XFS_IFORK_BOFF(ip) : \
+		XFS_LITINO((ip)->i_mount))
+#define XFS_IFORK_ASIZE(ip) \
+	(XFS_IFORK_Q(ip) ? \
+		XFS_LITINO((ip)->i_mount) - XFS_IFORK_BOFF(ip) : \
+		0)
+#define XFS_IFORK_SIZE(ip,w) \
+	((w) == XFS_DATA_FORK ? \
+		XFS_IFORK_DSIZE(ip) : \
+		XFS_IFORK_ASIZE(ip))
+#define XFS_IFORK_FORMAT(ip,w) \
+	((w) == XFS_DATA_FORK ? \
+		(ip)->i_d.di_format : \
+		(ip)->i_d.di_aformat)
+#define XFS_IFORK_FMT_SET(ip,w,n) \
+	((w) == XFS_DATA_FORK ? \
+		((ip)->i_d.di_format = (n)) : \
+		((ip)->i_d.di_aformat = (n)))
+#define XFS_IFORK_NEXTENTS(ip,w) \
+	((w) == XFS_DATA_FORK ? \
+		(ip)->i_d.di_nextents : \
+		(ip)->i_d.di_anextents)
+#define XFS_IFORK_NEXT_SET(ip,w,n) \
+	((w) == XFS_DATA_FORK ? \
+		((ip)->i_d.di_nextents = (n)) : \
+		((ip)->i_d.di_anextents = (n)))
+
+
+
+#ifdef __KERNEL__
+
+struct bhv_desc;
+struct cred;
+struct ktrace;
+struct xfs_buf;
+struct xfs_bmap_free;
+struct xfs_bmbt_irec;
+struct xfs_inode_log_item;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_dquot;
+
+#if defined(XFS_ILOCK_TRACE)
+#define XFS_ILOCK_KTRACE_SIZE	32
+extern void xfs_ilock_trace(struct xfs_inode *, int, unsigned int, inst_t *);
+#else
+#define	xfs_ilock_trace(i,n,f,ra)
+#endif
+
+typedef struct dm_attrs_s {
+	__uint32_t	da_dmevmask;	/* DMIG event mask */
+	__uint16_t	da_dmstate;	/* DMIG state info */
+	__uint16_t	da_pad;		/* DMIG extra padding */
+} dm_attrs_t;
 
 typedef struct xfs_inode {
 	/* Inode linking and identification information. */
-	struct xfs_inode	*i_mnext;	/* next inode in mount list */
-	struct xfs_inode	*i_mprev;	/* ptr to prev inode */
 	struct xfs_mount	*i_mount;	/* fs mount struct ptr */
-	struct list_head	i_reclaim;	/* reclaim list */
-	struct inode		*i_vnode;	/* vnode backpointer */
 	struct xfs_dquot	*i_udquot;	/* user dquot */
 	struct xfs_dquot	*i_gdquot;	/* group dquot */
 
 	/* Inode location stuff */
 	xfs_ino_t		i_ino;		/* inode number (agno/agino)*/
-	xfs_daddr_t		i_blkno;	/* blkno of inode buffer */
-	ushort			i_len;		/* len of inode buffer */
-	ushort			i_boffset;	/* off of inode in buffer */
+	struct xfs_imap		i_imap;		/* location for xfs_imap() */
 
 	/* Extent information. */
 	xfs_ifork_t		*i_afp;		/* attribute fork pointer */
@@ -230,7 +261,6 @@ typedef struct xfs_inode {
 	unsigned short		i_flags;	/* see defined flags below */
 	unsigned char		i_update_core;	/* timestamps/size is dirty */
 	unsigned char		i_update_size;	/* di_size field is dirty */
-	unsigned int		i_gen;		/* generation count */
 	unsigned int		i_delayed_blks;	/* count of delay alloc blks */
 
 	xfs_icdinode_t		i_d;		/* most of ondisk inode */
@@ -238,6 +268,10 @@ typedef struct xfs_inode {
 	xfs_fsize_t		i_size;		/* in-memory size */
 	xfs_fsize_t		i_new_size;	/* size when write completes */
 	atomic_t		i_iocount;	/* outstanding I/O count */
+
+	/* VFS inode */
+	struct inode		i_vnode;	/* embedded VFS inode */
+
 	/* Trace buffers per inode. */
 #ifdef XFS_INODE_TRACE
 	struct ktrace		*i_trace;	/* general inode trace */
@@ -245,7 +279,7 @@ typedef struct xfs_inode {
 #ifdef XFS_BMAP_TRACE
 	struct ktrace		*i_xtrace;	/* inode extent list trace */
 #endif
-#ifdef XFS_BMBT_TRACE
+#ifdef XFS_BTREE_TRACE
 	struct ktrace		*i_btrace;	/* inode bmap btree trace */
 #endif
 #ifdef XFS_RW_TRACE
@@ -265,13 +299,30 @@ typedef struct xfs_inode {
 /* Convert from vfs inode to xfs inode */
 static inline struct xfs_inode *XFS_I(struct inode *inode)
 {
-	return (struct xfs_inode *)inode->i_private;
+	return container_of(inode, struct xfs_inode, i_vnode);
 }
 
 /* convert from xfs inode to vfs inode */
 static inline struct inode *VFS_I(struct xfs_inode *ip)
 {
-	return (struct inode *)ip->i_vnode;
+	return &ip->i_vnode;
+}
+
+/*
+ * Get rid of a partially initialized inode.
+ *
+ * We have to go through destroy_inode to make sure allocations
+ * from init_inode_always like the security data are undone.
+ *
+ * We mark the inode bad so that it takes the short cut in
+ * the reclaim path instead of going through the flush path
+ * which doesn't make sense for an inode that has never seen the
+ * light of day.
+ */
+static inline void xfs_destroy_inode(struct xfs_inode *ip)
+{
+	make_bad_inode(VFS_I(ip));
+	return destroy_inode(VFS_I(ip));
 }
 
 /*
@@ -327,65 +378,36 @@ xfs_iflags_test_and_clear(xfs_inode_t *ip, unsigned short flags)
 	spin_unlock(&ip->i_flags_lock);
 	return ret;
 }
-#endif	/* __KERNEL__ */
-
 
 /*
- * Fork handling.
+ * Manage the i_flush queue embedded in the inode.  This completion
+ * queue synchronizes processes attempting to flush the in-core
+ * inode back to disk.
  */
+static inline void xfs_iflock(xfs_inode_t *ip)
+{
+	wait_for_completion(&ip->i_flush);
+}
 
-#define XFS_IFORK_Q(ip)			((ip)->i_d.di_forkoff != 0)
-#define XFS_IFORK_BOFF(ip)		((int)((ip)->i_d.di_forkoff << 3))
-
-#define XFS_IFORK_PTR(ip,w)		\
-	((w) == XFS_DATA_FORK ? \
-		&(ip)->i_df : \
-		(ip)->i_afp)
-#define XFS_IFORK_DSIZE(ip) \
-	(XFS_IFORK_Q(ip) ? \
-		XFS_IFORK_BOFF(ip) : \
-		XFS_LITINO((ip)->i_mount))
-#define XFS_IFORK_ASIZE(ip) \
-	(XFS_IFORK_Q(ip) ? \
-		XFS_LITINO((ip)->i_mount) - XFS_IFORK_BOFF(ip) : \
-		0)
-#define XFS_IFORK_SIZE(ip,w) \
-	((w) == XFS_DATA_FORK ? \
-		XFS_IFORK_DSIZE(ip) : \
-		XFS_IFORK_ASIZE(ip))
-#define XFS_IFORK_FORMAT(ip,w) \
-	((w) == XFS_DATA_FORK ? \
-		(ip)->i_d.di_format : \
-		(ip)->i_d.di_aformat)
-#define XFS_IFORK_FMT_SET(ip,w,n) \
-	((w) == XFS_DATA_FORK ? \
-		((ip)->i_d.di_format = (n)) : \
-		((ip)->i_d.di_aformat = (n)))
-#define XFS_IFORK_NEXTENTS(ip,w) \
-	((w) == XFS_DATA_FORK ? \
-		(ip)->i_d.di_nextents : \
-		(ip)->i_d.di_anextents)
-#define XFS_IFORK_NEXT_SET(ip,w,n) \
-	((w) == XFS_DATA_FORK ? \
-		((ip)->i_d.di_nextents = (n)) : \
-		((ip)->i_d.di_anextents = (n)))
+static inline int xfs_iflock_nowait(xfs_inode_t *ip)
+{
+	return try_wait_for_completion(&ip->i_flush);
+}
 
-#ifdef __KERNEL__
+static inline void xfs_ifunlock(xfs_inode_t *ip)
+{
+	complete(&ip->i_flush);
+}
 
 /*
  * In-core inode flags.
  */
-#define XFS_IGRIO	0x0001  /* inode used for guaranteed rate i/o */
-#define XFS_IUIOSZ	0x0002  /* inode i/o sizes have been explicitly set */
-#define XFS_IQUIESCE    0x0004  /* we have started quiescing for this inode */
-#define XFS_IRECLAIM    0x0008  /* we have started reclaiming this inode    */
-#define XFS_ISTALE	0x0010	/* inode has been staled */
-#define XFS_IRECLAIMABLE 0x0020 /* inode can be reclaimed */
-#define XFS_INEW	0x0040
-#define XFS_IFILESTREAM	0x0080	/* inode is in a filestream directory */
-#define XFS_IMODIFIED	0x0100	/* XFS inode state possibly differs */
-				/* to the Linux inode state. */
-#define XFS_ITRUNCATED	0x0200	/* truncated down so flush-on-close */
+#define XFS_IRECLAIM    0x0001  /* we have started reclaiming this inode    */
+#define XFS_ISTALE	0x0002	/* inode has been staled */
+#define XFS_IRECLAIMABLE 0x0004 /* inode can be reclaimed */
+#define XFS_INEW	0x0008	/* inode has just been allocated */
+#define XFS_IFILESTREAM	0x0010	/* inode is in a filestream directory */
+#define XFS_ITRUNCATED	0x0020	/* truncated down so flush-on-close */
 
 /*
  * Flags for inode locking.
@@ -460,16 +482,8 @@ xfs_iflags_test_and_clear(xfs_inode_t *ip, unsigned short flags)
 	 ((pip)->i_d.di_mode & S_ISGID))
 
 /*
- * Flags for xfs_iget()
- */
-#define XFS_IGET_CREATE		0x1
-#define XFS_IGET_BULKSTAT	0x2
-
-/*
  * xfs_iget.c prototypes.
  */
-void		xfs_ihash_init(struct xfs_mount *);
-void		xfs_ihash_free(struct xfs_mount *);
 xfs_inode_t	*xfs_inode_incore(struct xfs_mount *, xfs_ino_t,
 				  struct xfs_trans *);
 int		xfs_iget(struct xfs_mount *, struct xfs_trans *, xfs_ino_t,
@@ -484,25 +498,13 @@ int		xfs_isilocked(xfs_inode_t *, uint);
 uint		xfs_ilock_map_shared(xfs_inode_t *);
 void		xfs_iunlock_map_shared(xfs_inode_t *, uint);
 void		xfs_ireclaim(xfs_inode_t *);
-int		xfs_finish_reclaim(xfs_inode_t *, int, int);
-int		xfs_finish_reclaim_all(struct xfs_mount *, int);
 
 /*
  * xfs_inode.c prototypes.
  */
-int		xfs_itobp(struct xfs_mount *, struct xfs_trans *,
-			  xfs_inode_t *, struct xfs_dinode **, struct xfs_buf **,
-			  xfs_daddr_t, uint, uint);
-int		xfs_iread(struct xfs_mount *, struct xfs_trans *, xfs_ino_t,
-			  xfs_inode_t **, xfs_daddr_t, uint);
-int		xfs_iread_extents(struct xfs_trans *, xfs_inode_t *, int);
 int		xfs_ialloc(struct xfs_trans *, xfs_inode_t *, mode_t,
 			   xfs_nlink_t, xfs_dev_t, cred_t *, xfs_prid_t,
 			   int, struct xfs_buf **, boolean_t *, xfs_inode_t **);
-void		xfs_dinode_from_disk(struct xfs_icdinode *,
-				     struct xfs_dinode_core *);
-void		xfs_dinode_to_disk(struct xfs_dinode_core *,
-				   struct xfs_icdinode *);
 
 uint		xfs_ip2xflags(struct xfs_inode *);
 uint		xfs_dic2xflags(struct xfs_dinode *);
@@ -513,17 +515,10 @@ int		xfs_itruncate_finish(struct xfs_trans **, xfs_inode_t *,
 				     xfs_fsize_t, int, int);
 int		xfs_iunlink(struct xfs_trans *, xfs_inode_t *);
 
-void		xfs_idestroy_fork(xfs_inode_t *, int);
-void		xfs_idestroy(xfs_inode_t *);
-void		xfs_idata_realloc(xfs_inode_t *, int, int);
-void		xfs_iextract(xfs_inode_t *);
 void		xfs_iext_realloc(xfs_inode_t *, int, int);
-void		xfs_iroot_realloc(xfs_inode_t *, int, int);
 void		xfs_ipin(xfs_inode_t *);
 void		xfs_iunpin(xfs_inode_t *);
-int		xfs_iextents_copy(xfs_inode_t *, xfs_bmbt_rec_t *, int);
 int		xfs_iflush(xfs_inode_t *, uint);
-void		xfs_iflush_all(struct xfs_mount *);
 void		xfs_ichgtime(xfs_inode_t *, int);
 xfs_fsize_t	xfs_file_last_byte(xfs_inode_t *);
 void		xfs_lock_inodes(xfs_inode_t **, int, uint);
@@ -532,6 +527,77 @@ void		xfs_lock_two_inodes(xfs_inode_t *, xfs_inode_t *, uint);
 void		xfs_synchronize_atime(xfs_inode_t *);
 void		xfs_mark_inode_dirty_sync(xfs_inode_t *);
 
+#if defined(XFS_INODE_TRACE)
+
+#define	INODE_TRACE_SIZE	16		/* number of trace entries */
+#define	INODE_KTRACE_ENTRY	1
+#define	INODE_KTRACE_EXIT	2
+#define	INODE_KTRACE_HOLD	3
+#define	INODE_KTRACE_REF	4
+#define	INODE_KTRACE_RELE	5
+
+extern void _xfs_itrace_entry(struct xfs_inode *, const char *, inst_t *);
+extern void _xfs_itrace_exit(struct xfs_inode *, const char *, inst_t *);
+extern void xfs_itrace_hold(struct xfs_inode *, char *, int, inst_t *);
+extern void _xfs_itrace_ref(struct xfs_inode *, char *, int, inst_t *);
+extern void xfs_itrace_rele(struct xfs_inode *, char *, int, inst_t *);
+#define xfs_itrace_entry(ip)	\
+	_xfs_itrace_entry(ip, __func__, (inst_t *)__return_address)
+#define xfs_itrace_exit(ip)	\
+	_xfs_itrace_exit(ip, __func__, (inst_t *)__return_address)
+#define xfs_itrace_exit_tag(ip, tag)	\
+	_xfs_itrace_exit(ip, tag, (inst_t *)__return_address)
+#define xfs_itrace_ref(ip)	\
+	_xfs_itrace_ref(ip, __FILE__, __LINE__, (inst_t *)__return_address)
+
+#else
+#define	xfs_itrace_entry(a)
+#define	xfs_itrace_exit(a)
+#define	xfs_itrace_exit_tag(a, b)
+#define	xfs_itrace_hold(a, b, c, d)
+#define	xfs_itrace_ref(a)
+#define	xfs_itrace_rele(a, b, c, d)
+#endif
+
+#define IHOLD(ip) \
+do { \
+	ASSERT(atomic_read(&VFS_I(ip)->i_count) > 0) ; \
+	atomic_inc(&(VFS_I(ip)->i_count)); \
+	xfs_itrace_hold((ip), __FILE__, __LINE__, (inst_t *)__return_address); \
+} while (0)
+
+#define IRELE(ip) \
+do { \
+	xfs_itrace_rele((ip), __FILE__, __LINE__, (inst_t *)__return_address); \
+	iput(VFS_I(ip)); \
+} while (0)
+
+#endif /* __KERNEL__ */
+
+/*
+ * Flags for xfs_iget()
+ */
+#define XFS_IGET_CREATE		0x1
+#define XFS_IGET_BULKSTAT	0x2
+
+int		xfs_inotobp(struct xfs_mount *, struct xfs_trans *,
+			    xfs_ino_t, struct xfs_dinode **,
+			    struct xfs_buf **, int *, uint);
+int		xfs_itobp(struct xfs_mount *, struct xfs_trans *,
+			  struct xfs_inode *, struct xfs_dinode **,
+			  struct xfs_buf **, uint);
+int		xfs_iread(struct xfs_mount *, struct xfs_trans *,
+			  struct xfs_inode *, xfs_daddr_t, uint);
+void		xfs_dinode_from_disk(struct xfs_icdinode *,
+				     struct xfs_dinode *);
+void		xfs_dinode_to_disk(struct xfs_dinode *,
+				   struct xfs_icdinode *);
+void		xfs_idestroy_fork(struct xfs_inode *, int);
+void		xfs_idata_realloc(struct xfs_inode *, int, int);
+void		xfs_iroot_realloc(struct xfs_inode *, int, int);
+int		xfs_iread_extents(struct xfs_trans *, struct xfs_inode *, int);
+int		xfs_iextents_copy(struct xfs_inode *, xfs_bmbt_rec_t *, int);
+
 xfs_bmbt_rec_host_t *xfs_iext_get_ext(xfs_ifork_t *, xfs_extnum_t);
 void		xfs_iext_insert(xfs_ifork_t *, xfs_extnum_t, xfs_extnum_t,
 				xfs_bmbt_irec_t *);
@@ -561,7 +627,8 @@ void		xfs_iext_irec_update_extoffs(xfs_ifork_t *, int, int);
 #define xfs_ipincount(ip)	((unsigned int) atomic_read(&ip->i_pincount))
 
 #ifdef DEBUG
-void		xfs_isize_check(struct xfs_mount *, xfs_inode_t *, xfs_fsize_t);
+void		xfs_isize_check(struct xfs_mount *, struct xfs_inode *,
+				xfs_fsize_t);
 #else	/* DEBUG */
 #define xfs_isize_check(mp, ip, isize)
 #endif	/* DEBUG */
@@ -576,26 +643,4 @@ extern struct kmem_zone	*xfs_ifork_zone;
 extern struct kmem_zone	*xfs_inode_zone;
 extern struct kmem_zone	*xfs_ili_zone;
 
-/*
- * Manage the i_flush queue embedded in the inode.  This completion
- * queue synchronizes processes attempting to flush the in-core
- * inode back to disk.
- */
-static inline void xfs_iflock(xfs_inode_t *ip)
-{
-	wait_for_completion(&ip->i_flush);
-}
-
-static inline int xfs_iflock_nowait(xfs_inode_t *ip)
-{
-	return try_wait_for_completion(&ip->i_flush);
-}
-
-static inline void xfs_ifunlock(xfs_inode_t *ip)
-{
-	complete(&ip->i_flush);
-}
-
-#endif	/* __KERNEL__ */
-
 #endif	/* __XFS_INODE_H__ */
diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c
index 97c7452e262..977c4aec587 100644
--- a/fs/xfs/xfs_inode_item.c
+++ b/fs/xfs/xfs_inode_item.c
@@ -281,7 +281,7 @@ xfs_inode_item_format(
 	xfs_mark_inode_dirty_sync(ip);
 
 	vecp->i_addr = (xfs_caddr_t)&ip->i_d;
-	vecp->i_len  = sizeof(xfs_dinode_core_t);
+	vecp->i_len  = sizeof(struct xfs_icdinode);
 	XLOG_VEC_SET_TYPE(vecp, XLOG_REG_TYPE_ICORE);
 	vecp++;
 	nvecs++;
@@ -296,9 +296,8 @@ xfs_inode_item_format(
 	 * has a new version number, then we don't bother converting back.
 	 */
 	mp = ip->i_mount;
-	ASSERT(ip->i_d.di_version == XFS_DINODE_VERSION_1 ||
-	       xfs_sb_version_hasnlink(&mp->m_sb));
-	if (ip->i_d.di_version == XFS_DINODE_VERSION_1) {
+	ASSERT(ip->i_d.di_version == 1 || xfs_sb_version_hasnlink(&mp->m_sb));
+	if (ip->i_d.di_version == 1) {
 		if (!xfs_sb_version_hasnlink(&mp->m_sb)) {
 			/*
 			 * Convert it back.
@@ -311,7 +310,7 @@ xfs_inode_item_format(
 			 * so just make the conversion to the new inode
 			 * format permanent.
 			 */
-			ip->i_d.di_version = XFS_DINODE_VERSION_2;
+			ip->i_d.di_version = 2;
 			ip->i_d.di_onlink = 0;
 			memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
 		}
@@ -932,6 +931,7 @@ xfs_inode_item_init(
 	iip->ili_item.li_type = XFS_LI_INODE;
 	iip->ili_item.li_ops = &xfs_inode_item_ops;
 	iip->ili_item.li_mountp = mp;
+	iip->ili_item.li_ailp = mp->m_ail;
 	iip->ili_inode = ip;
 
 	/*
@@ -942,9 +942,9 @@ xfs_inode_item_init(
 
 	iip->ili_format.ilf_type = XFS_LI_INODE;
 	iip->ili_format.ilf_ino = ip->i_ino;
-	iip->ili_format.ilf_blkno = ip->i_blkno;
-	iip->ili_format.ilf_len = ip->i_len;
-	iip->ili_format.ilf_boffset = ip->i_boffset;
+	iip->ili_format.ilf_blkno = ip->i_imap.im_blkno;
+	iip->ili_format.ilf_len = ip->i_imap.im_len;
+	iip->ili_format.ilf_boffset = ip->i_imap.im_boffset;
 }
 
 /*
@@ -976,9 +976,8 @@ xfs_iflush_done(
 	xfs_buf_t		*bp,
 	xfs_inode_log_item_t	*iip)
 {
-	xfs_inode_t	*ip;
-
-	ip = iip->ili_inode;
+	xfs_inode_t		*ip = iip->ili_inode;
+	struct xfs_ail		*ailp = iip->ili_item.li_ailp;
 
 	/*
 	 * We only want to pull the item from the AIL if it is
@@ -991,15 +990,12 @@ xfs_iflush_done(
 	 */
 	if (iip->ili_logged &&
 	    (iip->ili_item.li_lsn == iip->ili_flush_lsn)) {
-		spin_lock(&ip->i_mount->m_ail_lock);
+		spin_lock(&ailp->xa_lock);
 		if (iip->ili_item.li_lsn == iip->ili_flush_lsn) {
-			/*
-			 * xfs_trans_delete_ail() drops the AIL lock.
-			 */
-			xfs_trans_delete_ail(ip->i_mount,
-					     (xfs_log_item_t*)iip);
+			/* xfs_trans_ail_delete() drops the AIL lock. */
+			xfs_trans_ail_delete(ailp, (xfs_log_item_t*)iip);
 		} else {
-			spin_unlock(&ip->i_mount->m_ail_lock);
+			spin_unlock(&ailp->xa_lock);
 		}
 	}
 
@@ -1031,21 +1027,20 @@ void
 xfs_iflush_abort(
 	xfs_inode_t		*ip)
 {
-	xfs_inode_log_item_t	*iip;
+	xfs_inode_log_item_t	*iip = ip->i_itemp;
 	xfs_mount_t		*mp;
 
 	iip = ip->i_itemp;
 	mp = ip->i_mount;
 	if (iip) {
+		struct xfs_ail	*ailp = iip->ili_item.li_ailp;
 		if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
-			spin_lock(&mp->m_ail_lock);
+			spin_lock(&ailp->xa_lock);
 			if (iip->ili_item.li_flags & XFS_LI_IN_AIL) {
-				/*
-				 * xfs_trans_delete_ail() drops the AIL lock.
-				 */
-				xfs_trans_delete_ail(mp, (xfs_log_item_t *)iip);
+				/* xfs_trans_ail_delete() drops the AIL lock. */
+				xfs_trans_ail_delete(ailp, (xfs_log_item_t *)iip);
 			} else
-				spin_unlock(&mp->m_ail_lock);
+				spin_unlock(&ailp->xa_lock);
 		}
 		iip->ili_logged = 0;
 		/*
diff --git a/fs/xfs/xfs_inode_item.h b/fs/xfs/xfs_inode_item.h
index 40513077ab3..1ff04cc323a 100644
--- a/fs/xfs/xfs_inode_item.h
+++ b/fs/xfs/xfs_inode_item.h
@@ -112,6 +112,24 @@ typedef struct xfs_inode_log_format_64 {
 #define	XFS_ILI_IOLOCKED_ANY   (XFS_ILI_IOLOCKED_EXCL | XFS_ILI_IOLOCKED_SHARED)
 
 
+#define	XFS_ILOG_FBROOT(w)	xfs_ilog_fbroot(w)
+static inline int xfs_ilog_fbroot(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DBROOT : XFS_ILOG_ABROOT);
+}
+
+#define	XFS_ILOG_FEXT(w)	xfs_ilog_fext(w)
+static inline int xfs_ilog_fext(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DEXT : XFS_ILOG_AEXT);
+}
+
+#define	XFS_ILOG_FDATA(w)	xfs_ilog_fdata(w)
+static inline int xfs_ilog_fdata(int w)
+{
+	return (w == XFS_DATA_FORK ? XFS_ILOG_DDATA : XFS_ILOG_ADATA);
+}
+
 #ifdef __KERNEL__
 
 struct xfs_buf;
@@ -148,26 +166,6 @@ typedef struct xfs_inode_log_item {
 } xfs_inode_log_item_t;
 
 
-#define	XFS_ILOG_FDATA(w)	xfs_ilog_fdata(w)
-static inline int xfs_ilog_fdata(int w)
-{
-	return (w == XFS_DATA_FORK ? XFS_ILOG_DDATA : XFS_ILOG_ADATA);
-}
-
-#endif	/* __KERNEL__ */
-
-#define	XFS_ILOG_FBROOT(w)	xfs_ilog_fbroot(w)
-static inline int xfs_ilog_fbroot(int w)
-{
-	return (w == XFS_DATA_FORK ? XFS_ILOG_DBROOT : XFS_ILOG_ABROOT);
-}
-
-#define	XFS_ILOG_FEXT(w)	xfs_ilog_fext(w)
-static inline int xfs_ilog_fext(int w)
-{
-	return (w == XFS_DATA_FORK ? XFS_ILOG_DEXT : XFS_ILOG_AEXT);
-}
-
 static inline int xfs_inode_clean(xfs_inode_t *ip)
 {
 	return (!ip->i_itemp ||
@@ -175,9 +173,6 @@ static inline int xfs_inode_clean(xfs_inode_t *ip)
 	       !ip->i_update_core;
 }
 
-
-#ifdef __KERNEL__
-
 extern void xfs_inode_item_init(struct xfs_inode *, struct xfs_mount *);
 extern void xfs_inode_item_destroy(struct xfs_inode *);
 extern void xfs_iflush_done(struct xfs_buf *, xfs_inode_log_item_t *);
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index 67f22b2b44b..911062cf73a 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -290,7 +290,6 @@ STATIC int
 xfs_iomap_eof_align_last_fsb(
 	xfs_mount_t	*mp,
 	xfs_inode_t	*ip,
-	xfs_fsize_t	isize,
 	xfs_extlen_t	extsize,
 	xfs_fileoff_t	*last_fsb)
 {
@@ -306,14 +305,14 @@ xfs_iomap_eof_align_last_fsb(
 	 * stripe width and we are allocating past the allocation eof.
 	 */
 	else if (mp->m_swidth && (mp->m_flags & XFS_MOUNT_SWALLOC) &&
-	        (isize >= XFS_FSB_TO_B(mp, mp->m_swidth)))
+	        (ip->i_size >= XFS_FSB_TO_B(mp, mp->m_swidth)))
 		new_last_fsb = roundup_64(*last_fsb, mp->m_swidth);
 	/*
 	 * Roundup the allocation request to a stripe unit (m_dalign) boundary
 	 * if the file size is >= stripe unit size, and we are allocating past
 	 * the allocation eof.
 	 */
-	else if (mp->m_dalign && (isize >= XFS_FSB_TO_B(mp, mp->m_dalign)))
+	else if (mp->m_dalign && (ip->i_size >= XFS_FSB_TO_B(mp, mp->m_dalign)))
 		new_last_fsb = roundup_64(*last_fsb, mp->m_dalign);
 
 	/*
@@ -403,7 +402,6 @@ xfs_iomap_write_direct(
 	xfs_filblks_t	count_fsb, resaligned;
 	xfs_fsblock_t	firstfsb;
 	xfs_extlen_t	extsz, temp;
-	xfs_fsize_t	isize;
 	int		nimaps;
 	int		bmapi_flag;
 	int		quota_flag;
@@ -426,15 +424,10 @@ xfs_iomap_write_direct(
 	rt = XFS_IS_REALTIME_INODE(ip);
 	extsz = xfs_get_extsz_hint(ip);
 
-	isize = ip->i_size;
-	if (ip->i_new_size > isize)
-		isize = ip->i_new_size;
-
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 	last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
-	if ((offset + count) > isize) {
-		error = xfs_iomap_eof_align_last_fsb(mp, ip, isize, extsz,
-							&last_fsb);
+	if ((offset + count) > ip->i_size) {
+		error = xfs_iomap_eof_align_last_fsb(mp, ip, extsz, &last_fsb);
 		if (error)
 			goto error_out;
 	} else {
@@ -559,7 +552,6 @@ STATIC int
 xfs_iomap_eof_want_preallocate(
 	xfs_mount_t	*mp,
 	xfs_inode_t	*ip,
-	xfs_fsize_t	isize,
 	xfs_off_t	offset,
 	size_t		count,
 	int		ioflag,
@@ -573,7 +565,7 @@ xfs_iomap_eof_want_preallocate(
 	int		n, error, imaps;
 
 	*prealloc = 0;
-	if ((ioflag & BMAPI_SYNC) || (offset + count) <= isize)
+	if ((ioflag & BMAPI_SYNC) || (offset + count) <= ip->i_size)
 		return 0;
 
 	/*
@@ -617,7 +609,6 @@ xfs_iomap_write_delay(
 	xfs_fileoff_t	ioalign;
 	xfs_fsblock_t	firstblock;
 	xfs_extlen_t	extsz;
-	xfs_fsize_t	isize;
 	int		nimaps;
 	xfs_bmbt_irec_t imap[XFS_WRITE_IMAPS];
 	int		prealloc, fsynced = 0;
@@ -637,11 +628,7 @@ xfs_iomap_write_delay(
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 
 retry:
-	isize = ip->i_size;
-	if (ip->i_new_size > isize)
-		isize = ip->i_new_size;
-
-	error = xfs_iomap_eof_want_preallocate(mp, ip, isize, offset, count,
+	error = xfs_iomap_eof_want_preallocate(mp, ip, offset, count,
 				ioflag, imap, XFS_WRITE_IMAPS, &prealloc);
 	if (error)
 		return error;
@@ -655,8 +642,7 @@ retry:
 	}
 
 	if (prealloc || extsz) {
-		error = xfs_iomap_eof_align_last_fsb(mp, ip, isize, extsz,
-							&last_fsb);
+		error = xfs_iomap_eof_align_last_fsb(mp, ip, extsz, &last_fsb);
 		if (error)
 			return error;
 	}
diff --git a/fs/xfs/xfs_itable.c b/fs/xfs/xfs_itable.c
index cf6754a3c5b..e19d0a8d561 100644
--- a/fs/xfs/xfs_itable.c
+++ b/fs/xfs/xfs_itable.c
@@ -69,7 +69,7 @@ xfs_bulkstat_one_iget(
 	}
 
 	ASSERT(ip != NULL);
-	ASSERT(ip->i_blkno != (xfs_daddr_t)0);
+	ASSERT(ip->i_imap.im_blkno != 0);
 
 	dic = &ip->i_d;
 
@@ -125,13 +125,9 @@ STATIC void
 xfs_bulkstat_one_dinode(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
 	xfs_ino_t	ino,		/* inode number to get data for */
-	xfs_dinode_t	*dip,		/* dinode inode pointer */
+	xfs_dinode_t	*dic,		/* dinode inode pointer */
 	xfs_bstat_t	*buf)		/* return buffer */
 {
-	xfs_dinode_core_t *dic;		/* dinode core info pointer */
-
-	dic = &dip->di_core;
-
 	/*
 	 * The inode format changed when we moved the link count and
 	 * made it 32 bits long.  If this is an old format inode,
@@ -143,7 +139,7 @@ xfs_bulkstat_one_dinode(
 	 * the new format. We don't change the version number so that we
 	 * can distinguish this from a real new format inode.
 	 */
-	if (dic->di_version == XFS_DINODE_VERSION_1) {
+	if (dic->di_version == 1) {
 		buf->bs_nlink = be16_to_cpu(dic->di_onlink);
 		buf->bs_projid = 0;
 	} else {
@@ -162,7 +158,7 @@ xfs_bulkstat_one_dinode(
 	buf->bs_mtime.tv_nsec = be32_to_cpu(dic->di_mtime.t_nsec);
 	buf->bs_ctime.tv_sec = be32_to_cpu(dic->di_ctime.t_sec);
 	buf->bs_ctime.tv_nsec = be32_to_cpu(dic->di_ctime.t_nsec);
-	buf->bs_xflags = xfs_dic2xflags(dip);
+	buf->bs_xflags = xfs_dic2xflags(dic);
 	buf->bs_extsize = be32_to_cpu(dic->di_extsize) << mp->m_sb.sb_blocklog;
 	buf->bs_extents = be32_to_cpu(dic->di_nextents);
 	buf->bs_gen = be32_to_cpu(dic->di_gen);
@@ -173,7 +169,7 @@ xfs_bulkstat_one_dinode(
 
 	switch (dic->di_format) {
 	case XFS_DINODE_FMT_DEV:
-		buf->bs_rdev = be32_to_cpu(dip->di_u.di_dev);
+		buf->bs_rdev = xfs_dinode_get_rdev(dic);
 		buf->bs_blksize = BLKDEV_IOSIZE;
 		buf->bs_blocks = 0;
 		break;
@@ -192,27 +188,34 @@ xfs_bulkstat_one_dinode(
 	}
 }
 
+/* Return 0 on success or positive error */
 STATIC int
 xfs_bulkstat_one_fmt(
 	void			__user *ubuffer,
+	int			ubsize,
+	int			*ubused,
 	const xfs_bstat_t	*buffer)
 {
+	if (ubsize < sizeof(*buffer))
+		return XFS_ERROR(ENOMEM);
 	if (copy_to_user(ubuffer, buffer, sizeof(*buffer)))
-		return -EFAULT;
-	return sizeof(*buffer);
+		return XFS_ERROR(EFAULT);
+	if (ubused)
+		*ubused = sizeof(*buffer);
+	return 0;
 }
 
 /*
  * Return stat information for one inode.
  * Return 0 if ok, else errno.
  */
-int		       		/* error status */
-xfs_bulkstat_one(
+int		   	    		/* error status */
+xfs_bulkstat_one_int(
 	xfs_mount_t	*mp,		/* mount point for filesystem */
 	xfs_ino_t	ino,		/* inode number to get data for */
 	void		__user *buffer,	/* buffer to place output in */
 	int		ubsize,		/* size of buffer */
-	void		*private_data,	/* my private data */
+	bulkstat_one_fmt_pf formatter,	/* formatter, copy to user */
 	xfs_daddr_t	bno,		/* starting bno of inode cluster */
 	int		*ubused,	/* bytes used by me */
 	void		*dibuff,	/* on-disk inode buffer */
@@ -221,15 +224,12 @@ xfs_bulkstat_one(
 	xfs_bstat_t	*buf;		/* return buffer */
 	int		error = 0;	/* error value */
 	xfs_dinode_t	*dip;		/* dinode inode pointer */
-	bulkstat_one_fmt_pf formatter = private_data ? : xfs_bulkstat_one_fmt;
 
 	dip = (xfs_dinode_t *)dibuff;
 	*stat = BULKSTAT_RV_NOTHING;
 
 	if (!buffer || xfs_internal_inum(mp, ino))
 		return XFS_ERROR(EINVAL);
-	if (ubsize < sizeof(*buf))
-		return XFS_ERROR(ENOMEM);
 
 	buf = kmem_alloc(sizeof(*buf), KM_SLEEP);
 
@@ -244,21 +244,34 @@ xfs_bulkstat_one(
 		xfs_bulkstat_one_dinode(mp, ino, dip, buf);
 	}
 
-	error = formatter(buffer, buf);
-	if (error < 0)  {
-		error = EFAULT;
+	error = formatter(buffer, ubsize, ubused, buf);
+	if (error)
 		goto out_free;
-	}
 
 	*stat = BULKSTAT_RV_DIDONE;
-	if (ubused)
-		*ubused = error;
 
  out_free:
 	kmem_free(buf);
 	return error;
 }
 
+int
+xfs_bulkstat_one(
+	xfs_mount_t	*mp,		/* mount point for filesystem */
+	xfs_ino_t	ino,		/* inode number to get data for */
+	void		__user *buffer,	/* buffer to place output in */
+	int		ubsize,		/* size of buffer */
+	void		*private_data,	/* my private data */
+	xfs_daddr_t	bno,		/* starting bno of inode cluster */
+	int		*ubused,	/* bytes used by me */
+	void		*dibuff,	/* on-disk inode buffer */
+	int		*stat)		/* BULKSTAT_RV_... */
+{
+	return xfs_bulkstat_one_int(mp, ino, buffer, ubsize,
+				    xfs_bulkstat_one_fmt, bno,
+				    ubused, dibuff, stat);
+}
+
 /*
  * Test to see whether we can use the ondisk inode directly, based
  * on the given bulkstat flags, filling in dipp accordingly.
@@ -287,19 +300,19 @@ xfs_bulkstat_use_dinode(
 	 * to disk yet. This is a temporary hack that would require a proper
 	 * fix in the future.
 	 */
-	if (be16_to_cpu(dip->di_core.di_magic) != XFS_DINODE_MAGIC ||
-	    !XFS_DINODE_GOOD_VERSION(dip->di_core.di_version) ||
-	    !dip->di_core.di_mode)
+	if (be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC ||
+	    !XFS_DINODE_GOOD_VERSION(dip->di_version) ||
+	    !dip->di_mode)
 		return 0;
 	if (flags & BULKSTAT_FG_QUICK) {
 		*dipp = dip;
 		return 1;
 	}
 	/* BULKSTAT_FG_INLINE: if attr fork is local, or not there, use it */
-	aformat = dip->di_core.di_aformat;
+	aformat = dip->di_aformat;
 	if ((XFS_DFORK_Q(dip) == 0) ||
 	    (aformat == XFS_DINODE_FMT_LOCAL) ||
-	    (aformat == XFS_DINODE_FMT_EXTENTS && !dip->di_core.di_anextents)) {
+	    (aformat == XFS_DINODE_FMT_EXTENTS && !dip->di_anextents)) {
 		*dipp = dip;
 		return 1;
 	}
@@ -359,7 +372,6 @@ xfs_bulkstat(
 	int			ubused;	/* bytes used by formatter */
 	xfs_buf_t		*bp;	/* ptr to on-disk inode cluster buf */
 	xfs_dinode_t		*dip;	/* ptr into bp for specific inode */
-	xfs_inode_t		*ip;	/* ptr to in-core inode struct */
 
 	/*
 	 * Get the last inode value, see if there's nothing to do.
@@ -416,8 +428,7 @@ xfs_bulkstat(
 		/*
 		 * Allocate and initialize a btree cursor for ialloc btree.
 		 */
-		cur = xfs_btree_init_cursor(mp, NULL, agbp, agno, XFS_BTNUM_INO,
-						(xfs_inode_t *)0, 0);
+		cur = xfs_inobt_init_cursor(mp, NULL, agbp, agno);
 		irbp = irbuf;
 		irbufend = irbuf + nirbuf;
 		end_of_ag = 0;
@@ -472,7 +483,7 @@ xfs_bulkstat(
 			 * In any case, increment to the next record.
 			 */
 			if (!error)
-				error = xfs_inobt_increment(cur, 0, &tmp);
+				error = xfs_btree_increment(cur, 0, &tmp);
 		} else {
 			/*
 			 * Start of ag.  Lookup the first inode chunk.
@@ -539,7 +550,7 @@ xfs_bulkstat(
 			 * Set agino to after this chunk and bump the cursor.
 			 */
 			agino = gino + XFS_INODES_PER_CHUNK;
-			error = xfs_inobt_increment(cur, 0, &tmp);
+			error = xfs_btree_increment(cur, 0, &tmp);
 			cond_resched();
 		}
 		/*
@@ -586,6 +597,8 @@ xfs_bulkstat(
 
 					if (flags & (BULKSTAT_FG_QUICK |
 						     BULKSTAT_FG_INLINE)) {
+						int offset;
+
 						ino = XFS_AGINO_TO_INO(mp, agno,
 								       agino);
 						bno = XFS_AGB_TO_DADDR(mp, agno,
@@ -594,21 +607,15 @@ xfs_bulkstat(
 						/*
 						 * Get the inode cluster buffer
 						 */
-						ASSERT(xfs_inode_zone != NULL);
-						ip = kmem_zone_zalloc(xfs_inode_zone,
-								      KM_SLEEP);
-						ip->i_ino = ino;
-						ip->i_mount = mp;
-						spin_lock_init(&ip->i_flags_lock);
 						if (bp)
 							xfs_buf_relse(bp);
-						error = xfs_itobp(mp, NULL, ip,
-								&dip, &bp, bno,
-								XFS_IMAP_BULKSTAT,
-								XFS_BUF_LOCK);
+
+						error = xfs_inotobp(mp, NULL, ino, &dip,
+								    &bp, &offset,
+								    XFS_IGET_BULKSTAT);
+
 						if (!error)
-							clustidx = ip->i_boffset / mp->m_sb.sb_inodesize;
-						kmem_zone_free(xfs_inode_zone, ip);
+							clustidx = offset / mp->m_sb.sb_inodesize;
 						if (XFS_TEST_ERROR(error != 0,
 								   mp, XFS_ERRTAG_BULKSTAT_READ_CHUNK,
 								   XFS_RANDOM_BULKSTAT_READ_CHUNK)) {
@@ -842,8 +849,7 @@ xfs_inumbers(
 				agino = 0;
 				continue;
 			}
-			cur = xfs_btree_init_cursor(mp, NULL, agbp, agno,
-				XFS_BTNUM_INO, (xfs_inode_t *)0, 0);
+			cur = xfs_inobt_init_cursor(mp, NULL, agbp, agno);
 			error = xfs_inobt_lookup_ge(cur, agino, 0, 0, &tmp);
 			if (error) {
 				xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
@@ -887,7 +893,7 @@ xfs_inumbers(
 			bufidx = 0;
 		}
 		if (left) {
-			error = xfs_inobt_increment(cur, 0, &tmp);
+			error = xfs_btree_increment(cur, 0, &tmp);
 			if (error) {
 				xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
 				cur = NULL;
diff --git a/fs/xfs/xfs_itable.h b/fs/xfs/xfs_itable.h
index a1f18fce9b7..1fb04e7deb6 100644
--- a/fs/xfs/xfs_itable.h
+++ b/fs/xfs/xfs_itable.h
@@ -71,9 +71,23 @@ xfs_bulkstat_single(
 
 typedef int (*bulkstat_one_fmt_pf)(  /* used size in bytes or negative error */
 	void			__user *ubuffer, /* buffer to write to */
+	int			ubsize,		 /* remaining user buffer sz */
+	int			*ubused,	 /* bytes used by formatter */
 	const xfs_bstat_t	*buffer);        /* buffer to read from */
 
 int
+xfs_bulkstat_one_int(
+	xfs_mount_t		*mp,
+	xfs_ino_t		ino,
+	void			__user *buffer,
+	int			ubsize,
+	bulkstat_one_fmt_pf	formatter,
+	xfs_daddr_t		bno,
+	int			*ubused,
+	void			*dibuff,
+	int			*stat);
+
+int
 xfs_bulkstat_one(
 	xfs_mount_t		*mp,
 	xfs_ino_t		ino,
diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c
index 3608a0f0a5f..f4726f702a9 100644
--- a/fs/xfs/xfs_log.c
+++ b/fs/xfs/xfs_log.c
@@ -100,12 +100,11 @@ STATIC void xlog_ungrant_log_space(xlog_t	 *log,
 
 
 /* local ticket functions */
-STATIC xlog_ticket_t	*xlog_ticket_get(xlog_t *log,
+STATIC xlog_ticket_t	*xlog_ticket_alloc(xlog_t *log,
 					 int	unit_bytes,
 					 int	count,
 					 char	clientid,
 					 uint	flags);
-STATIC void		xlog_ticket_put(xlog_t *log, xlog_ticket_t *ticket);
 
 #if defined(DEBUG)
 STATIC void	xlog_verify_dest_ptr(xlog_t *log, __psint_t ptr);
@@ -360,7 +359,7 @@ xfs_log_done(xfs_mount_t	*mp,
 		 */
 		xlog_trace_loggrant(log, ticket, "xfs_log_done: (non-permanent)");
 		xlog_ungrant_log_space(log, ticket);
-		xlog_ticket_put(log, ticket);
+		xfs_log_ticket_put(ticket);
 	} else {
 		xlog_trace_loggrant(log, ticket, "xfs_log_done: (permanent)");
 		xlog_regrant_reserve_log_space(log, ticket);
@@ -514,7 +513,7 @@ xfs_log_reserve(xfs_mount_t	 *mp,
 		retval = xlog_regrant_write_log_space(log, internal_ticket);
 	} else {
 		/* may sleep if need to allocate more tickets */
-		internal_ticket = xlog_ticket_get(log, unit_bytes, cnt,
+		internal_ticket = xlog_ticket_alloc(log, unit_bytes, cnt,
 						  client, flags);
 		if (!internal_ticket)
 			return XFS_ERROR(ENOMEM);
@@ -572,12 +571,12 @@ xfs_log_mount(
 	/*
 	 * Initialize the AIL now we have a log.
 	 */
-	spin_lock_init(&mp->m_ail_lock);
 	error = xfs_trans_ail_init(mp);
 	if (error) {
 		cmn_err(CE_WARN, "XFS: AIL initialisation failed: error %d", error);
 		goto error;
 	}
+	mp->m_log->l_ailp = mp->m_ail;
 
 	/*
 	 * skip log recovery on a norecovery mount.  pretend it all
@@ -730,8 +729,8 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		spin_lock(&log->l_icloglock);
 		iclog = log->l_iclog;
 		atomic_inc(&iclog->ic_refcnt);
-		spin_unlock(&log->l_icloglock);
 		xlog_state_want_sync(log, iclog);
+		spin_unlock(&log->l_icloglock);
 		error = xlog_state_release_iclog(log, iclog);
 
 		spin_lock(&log->l_icloglock);
@@ -749,7 +748,7 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		if (tic) {
 			xlog_trace_loggrant(log, tic, "unmount rec");
 			xlog_ungrant_log_space(log, tic);
-			xlog_ticket_put(log, tic);
+			xfs_log_ticket_put(tic);
 		}
 	} else {
 		/*
@@ -768,9 +767,9 @@ xfs_log_unmount_write(xfs_mount_t *mp)
 		spin_lock(&log->l_icloglock);
 		iclog = log->l_iclog;
 		atomic_inc(&iclog->ic_refcnt);
-		spin_unlock(&log->l_icloglock);
 
 		xlog_state_want_sync(log, iclog);
+		spin_unlock(&log->l_icloglock);
 		error =  xlog_state_release_iclog(log, iclog);
 
 		spin_lock(&log->l_icloglock);
@@ -906,7 +905,7 @@ xfs_log_move_tail(xfs_mount_t	*mp,
 int
 xfs_log_need_covered(xfs_mount_t *mp)
 {
-	int		needed = 0, gen;
+	int		needed = 0;
 	xlog_t		*log = mp->m_log;
 
 	if (!xfs_fs_writable(mp))
@@ -915,7 +914,7 @@ xfs_log_need_covered(xfs_mount_t *mp)
 	spin_lock(&log->l_icloglock);
 	if (((log->l_covered_state == XLOG_STATE_COVER_NEED) ||
 		(log->l_covered_state == XLOG_STATE_COVER_NEED2))
-			&& !xfs_trans_first_ail(mp, &gen)
+			&& !xfs_trans_ail_tail(log->l_ailp)
 			&& xlog_iclogs_empty(log)) {
 		if (log->l_covered_state == XLOG_STATE_COVER_NEED)
 			log->l_covered_state = XLOG_STATE_COVER_DONE;
@@ -952,7 +951,7 @@ xlog_assign_tail_lsn(xfs_mount_t *mp)
 	xfs_lsn_t tail_lsn;
 	xlog_t	  *log = mp->m_log;
 
-	tail_lsn = xfs_trans_tail_ail(mp);
+	tail_lsn = xfs_trans_ail_tail(mp->m_ail);
 	spin_lock(&log->l_grant_lock);
 	if (tail_lsn != 0) {
 		log->l_tail_lsn = tail_lsn;
@@ -1030,12 +1029,6 @@ xlog_iodone(xfs_buf_t *bp)
 	ASSERT(XFS_BUF_FSPRIVATE2(bp, unsigned long) == (unsigned long) 2);
 	XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)1);
 	aborted = 0;
-
-	/*
-	 * Some versions of cpp barf on the recursive definition of
-	 * ic_log -> hic_fields.ic_log and expand ic_log twice when
-	 * it is passed through two macros.  Workaround broken cpp.
-	 */
 	l = iclog->ic_log;
 
 	/*
@@ -1302,7 +1295,7 @@ xlog_alloc_log(xfs_mount_t	*mp,
 		XFS_BUF_SET_BDSTRAT_FUNC(bp, xlog_bdstrat_cb);
 		XFS_BUF_SET_FSPRIVATE2(bp, (unsigned long)1);
 		iclog->ic_bp = bp;
-		iclog->hic_data = bp->b_addr;
+		iclog->ic_data = bp->b_addr;
 #ifdef DEBUG
 		log->l_iclog_bak[i] = (xfs_caddr_t)&(iclog->ic_header);
 #endif
@@ -1322,7 +1315,7 @@ xlog_alloc_log(xfs_mount_t	*mp,
 		atomic_set(&iclog->ic_refcnt, 0);
 		spin_lock_init(&iclog->ic_callback_lock);
 		iclog->ic_callback_tail = &(iclog->ic_callback);
-		iclog->ic_datap = (char *)iclog->hic_data + log->l_iclog_hsize;
+		iclog->ic_datap = (char *)iclog->ic_data + log->l_iclog_hsize;
 
 		ASSERT(XFS_BUF_ISBUSY(iclog->ic_bp));
 		ASSERT(XFS_BUF_VALUSEMA(iclog->ic_bp) <= 0);
@@ -1446,7 +1439,7 @@ xlog_grant_push_ail(xfs_mount_t	*mp,
      */
     if (threshold_lsn &&
 	!XLOG_FORCED_SHUTDOWN(log))
-	    xfs_trans_push_ail(mp, threshold_lsn);
+	    xfs_trans_ail_push(log->l_ailp, threshold_lsn);
 }	/* xlog_grant_push_ail */
 
 
@@ -1991,7 +1984,9 @@ xlog_write(xfs_mount_t *	mp,
 		if (iclog->ic_size - log_offset <= sizeof(xlog_op_header_t)) {
 		    xlog_state_finish_copy(log, iclog, record_cnt, data_cnt);
 		    record_cnt = data_cnt = 0;
+		    spin_lock(&log->l_icloglock);
 		    xlog_state_want_sync(log, iclog);
+		    spin_unlock(&log->l_icloglock);
 		    if (commit_iclog) {
 			ASSERT(flags & XLOG_COMMIT_TRANS);
 			*commit_iclog = iclog;
@@ -3200,7 +3195,7 @@ try_again:
 STATIC void
 xlog_state_want_sync(xlog_t *log, xlog_in_core_t *iclog)
 {
-	spin_lock(&log->l_icloglock);
+	ASSERT(spin_is_locked(&log->l_icloglock));
 
 	if (iclog->ic_state == XLOG_STATE_ACTIVE) {
 		xlog_state_switch_iclogs(log, iclog, 0);
@@ -3208,10 +3203,7 @@ xlog_state_want_sync(xlog_t *log, xlog_in_core_t *iclog)
 		ASSERT(iclog->ic_state &
 			(XLOG_STATE_WANT_SYNC|XLOG_STATE_IOERROR));
 	}
-
-	spin_unlock(&log->l_icloglock);
-}	/* xlog_state_want_sync */
-
+}
 
 
 /*****************************************************************************
@@ -3222,22 +3214,33 @@ xlog_state_want_sync(xlog_t *log, xlog_in_core_t *iclog)
  */
 
 /*
- * Free a used ticket.
+ * Free a used ticket when it's refcount falls to zero.
  */
-STATIC void
-xlog_ticket_put(xlog_t		*log,
-		xlog_ticket_t	*ticket)
+void
+xfs_log_ticket_put(
+	xlog_ticket_t	*ticket)
 {
-	sv_destroy(&ticket->t_wait);
-	kmem_zone_free(xfs_log_ticket_zone, ticket);
-}	/* xlog_ticket_put */
+	ASSERT(atomic_read(&ticket->t_ref) > 0);
+	if (atomic_dec_and_test(&ticket->t_ref)) {
+		sv_destroy(&ticket->t_wait);
+		kmem_zone_free(xfs_log_ticket_zone, ticket);
+	}
+}
 
+xlog_ticket_t *
+xfs_log_ticket_get(
+	xlog_ticket_t	*ticket)
+{
+	ASSERT(atomic_read(&ticket->t_ref) > 0);
+	atomic_inc(&ticket->t_ref);
+	return ticket;
+}
 
 /*
  * Allocate and initialise a new log ticket.
  */
 STATIC xlog_ticket_t *
-xlog_ticket_get(xlog_t		*log,
+xlog_ticket_alloc(xlog_t		*log,
 		int		unit_bytes,
 		int		cnt,
 		char		client,
@@ -3308,6 +3311,7 @@ xlog_ticket_get(xlog_t		*log,
 		unit_bytes += 2*BBSIZE;
         }
 
+	atomic_set(&tic->t_ref, 1);
 	tic->t_unit_res		= unit_bytes;
 	tic->t_curr_res		= unit_bytes;
 	tic->t_cnt		= cnt;
@@ -3323,7 +3327,7 @@ xlog_ticket_get(xlog_t		*log,
 	xlog_tic_reset_res(tic);
 
 	return tic;
-}	/* xlog_ticket_get */
+}
 
 
 /******************************************************************************
@@ -3452,7 +3456,7 @@ xlog_verify_iclog(xlog_t	 *log,
 	ptr = iclog->ic_datap;
 	base_ptr = ptr;
 	ophead = (xlog_op_header_t *)ptr;
-	xhdr = (xlog_in_core_2_t *)&iclog->ic_header;
+	xhdr = iclog->ic_data;
 	for (i = 0; i < len; i++) {
 		ophead = (xlog_op_header_t *)ptr;
 
@@ -3558,7 +3562,8 @@ xfs_log_force_umount(
 	if (!log ||
 	    log->l_flags & XLOG_ACTIVE_RECOVERY) {
 		mp->m_flags |= XFS_MOUNT_FS_SHUTDOWN;
-		XFS_BUF_DONE(mp->m_sb_bp);
+		if (mp->m_sb_bp)
+			XFS_BUF_DONE(mp->m_sb_bp);
 		return 0;
 	}
 
@@ -3579,7 +3584,9 @@ xfs_log_force_umount(
 	spin_lock(&log->l_icloglock);
 	spin_lock(&log->l_grant_lock);
 	mp->m_flags |= XFS_MOUNT_FS_SHUTDOWN;
-	XFS_BUF_DONE(mp->m_sb_bp);
+	if (mp->m_sb_bp)
+		XFS_BUF_DONE(mp->m_sb_bp);
+
 	/*
 	 * This flag is sort of redundant because of the mount flag, but
 	 * it's good to maintain the separation between the log and the rest
diff --git a/fs/xfs/xfs_log.h b/fs/xfs/xfs_log.h
index d47b91f1082..8a3e84e900a 100644
--- a/fs/xfs/xfs_log.h
+++ b/fs/xfs/xfs_log.h
@@ -134,6 +134,7 @@ typedef struct xfs_log_callback {
 #ifdef __KERNEL__
 /* Log manager interfaces */
 struct xfs_mount;
+struct xlog_ticket;
 xfs_lsn_t xfs_log_done(struct xfs_mount *mp,
 		       xfs_log_ticket_t ticket,
 		       void		**iclog,
@@ -177,6 +178,9 @@ int	  xfs_log_need_covered(struct xfs_mount *mp);
 
 void	  xlog_iodone(struct xfs_buf *);
 
+struct xlog_ticket * xfs_log_ticket_get(struct xlog_ticket *ticket);
+void	  xfs_log_ticket_put(struct xlog_ticket *ticket);
+
 #endif
 
 
diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h
index e7d8f84443f..654167be0ef 100644
--- a/fs/xfs/xfs_log_priv.h
+++ b/fs/xfs/xfs_log_priv.h
@@ -245,6 +245,7 @@ typedef struct xlog_ticket {
 	struct xlog_ticket *t_next;	 /*			         :4|8 */
 	struct xlog_ticket *t_prev;	 /*				 :4|8 */
 	xlog_tid_t	   t_tid;	 /* transaction identifier	 : 4  */
+	atomic_t	   t_ref;	 /* ticket reference count       : 4  */
 	int		   t_curr_res;	 /* current reservation in bytes : 4  */
 	int		   t_unit_res;	 /* unit reservation in bytes    : 4  */
 	char		   t_ocnt;	 /* original count		 : 1  */
@@ -309,6 +310,16 @@ typedef struct xlog_rec_ext_header {
 } xlog_rec_ext_header_t;
 
 #ifdef __KERNEL__
+
+/*
+ * Quite misnamed, because this union lays out the actual on-disk log buffer.
+ */
+typedef union xlog_in_core2 {
+	xlog_rec_header_t	hic_header;
+	xlog_rec_ext_header_t	hic_xheader;
+	char			hic_sector[XLOG_HEADER_SIZE];
+} xlog_in_core_2_t;
+
 /*
  * - A log record header is 512 bytes.  There is plenty of room to grow the
  *	xlog_rec_header_t into the reserved space.
@@ -338,7 +349,7 @@ typedef struct xlog_rec_ext_header {
  * We'll put all the read-only and l_icloglock fields in the first cacheline,
  * and move everything else out to subsequent cachelines.
  */
-typedef struct xlog_iclog_fields {
+typedef struct xlog_in_core {
 	sv_t			ic_force_wait;
 	sv_t			ic_write_wait;
 	struct xlog_in_core	*ic_next;
@@ -361,41 +372,11 @@ typedef struct xlog_iclog_fields {
 
 	/* reference counts need their own cacheline */
 	atomic_t		ic_refcnt ____cacheline_aligned_in_smp;
-} xlog_iclog_fields_t;
-
-typedef union xlog_in_core2 {
-	xlog_rec_header_t	hic_header;
-	xlog_rec_ext_header_t	hic_xheader;
-	char			hic_sector[XLOG_HEADER_SIZE];
-} xlog_in_core_2_t;
-
-typedef struct xlog_in_core {
-	xlog_iclog_fields_t	hic_fields;
-	xlog_in_core_2_t	*hic_data;
+	xlog_in_core_2_t	*ic_data;
+#define ic_header	ic_data->hic_header
 } xlog_in_core_t;
 
 /*
- * Defines to save our code from this glop.
- */
-#define	ic_force_wait	hic_fields.ic_force_wait
-#define ic_write_wait	hic_fields.ic_write_wait
-#define	ic_next		hic_fields.ic_next
-#define	ic_prev		hic_fields.ic_prev
-#define	ic_bp		hic_fields.ic_bp
-#define	ic_log		hic_fields.ic_log
-#define	ic_callback	hic_fields.ic_callback
-#define	ic_callback_lock hic_fields.ic_callback_lock
-#define	ic_callback_tail hic_fields.ic_callback_tail
-#define	ic_trace	hic_fields.ic_trace
-#define	ic_size		hic_fields.ic_size
-#define	ic_offset	hic_fields.ic_offset
-#define	ic_refcnt	hic_fields.ic_refcnt
-#define	ic_bwritecnt	hic_fields.ic_bwritecnt
-#define	ic_state	hic_fields.ic_state
-#define ic_datap	hic_fields.ic_datap
-#define ic_header	hic_data->hic_header
-
-/*
  * The reservation head lsn is not made up of a cycle number and block number.
  * Instead, it uses a cycle number and byte number.  Logs don't expect to
  * overflow 31 bits worth of byte offset, so using a byte number will mean
@@ -404,6 +385,7 @@ typedef struct xlog_in_core {
 typedef struct log {
 	/* The following fields don't need locking */
 	struct xfs_mount	*l_mp;	        /* mount point */
+	struct xfs_ail		*l_ailp;	/* AIL log is working with */
 	struct xfs_buf		*l_xbuf;        /* extra buffer for log
 						 * wrapping */
 	struct xfs_buftarg	*l_targ;        /* buftarg of log */
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index 70e3ba32e6b..35cca98bd94 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -36,7 +36,6 @@
 #include "xfs_dinode.h"
 #include "xfs_inode.h"
 #include "xfs_inode_item.h"
-#include "xfs_imap.h"
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_log_priv.h"
@@ -54,10 +53,8 @@ STATIC void	xlog_recover_insert_item_backq(xlog_recover_item_t **q,
 					       xlog_recover_item_t *item);
 #if defined(DEBUG)
 STATIC void	xlog_recover_check_summary(xlog_t *);
-STATIC void	xlog_recover_check_ail(xfs_mount_t *, xfs_log_item_t *, int);
 #else
 #define	xlog_recover_check_summary(log)
-#define	xlog_recover_check_ail(mp, lip, gen)
 #endif
 
 
@@ -270,21 +267,16 @@ STATIC void
 xlog_recover_iodone(
 	struct xfs_buf	*bp)
 {
-	xfs_mount_t	*mp;
-
-	ASSERT(XFS_BUF_FSPRIVATE(bp, void *));
-
 	if (XFS_BUF_GETERROR(bp)) {
 		/*
 		 * We're not going to bother about retrying
 		 * this during recovery. One strike!
 		 */
-		mp = XFS_BUF_FSPRIVATE(bp, xfs_mount_t *);
 		xfs_ioerror_alert("xlog_recover_iodone",
-				  mp, bp, XFS_BUF_ADDR(bp));
-		xfs_force_shutdown(mp, SHUTDOWN_META_IO_ERROR);
+				  bp->b_mount, bp, XFS_BUF_ADDR(bp));
+		xfs_force_shutdown(bp->b_mount, SHUTDOWN_META_IO_ERROR);
 	}
-	XFS_BUF_SET_FSPRIVATE(bp, NULL);
+	bp->b_mount = NULL;
 	XFS_BUF_CLR_IODONE_FUNC(bp);
 	xfs_biodone(bp);
 }
@@ -2228,9 +2220,8 @@ xlog_recover_do_buffer_trans(
 		XFS_BUF_STALE(bp);
 		error = xfs_bwrite(mp, bp);
 	} else {
-		ASSERT(XFS_BUF_FSPRIVATE(bp, void *) == NULL ||
-		       XFS_BUF_FSPRIVATE(bp, xfs_mount_t *) == mp);
-		XFS_BUF_SET_FSPRIVATE(bp, mp);
+		ASSERT(bp->b_mount == NULL || bp->b_mount == mp);
+		bp->b_mount = mp;
 		XFS_BUF_SET_IODONE_FUNC(bp, xlog_recover_iodone);
 		xfs_bdwrite(mp, bp);
 	}
@@ -2247,7 +2238,6 @@ xlog_recover_do_inode_trans(
 	xfs_inode_log_format_t	*in_f;
 	xfs_mount_t		*mp;
 	xfs_buf_t		*bp;
-	xfs_imap_t		imap;
 	xfs_dinode_t		*dip;
 	xfs_ino_t		ino;
 	int			len;
@@ -2275,54 +2265,35 @@ xlog_recover_do_inode_trans(
 	}
 	ino = in_f->ilf_ino;
 	mp = log->l_mp;
-	if (ITEM_TYPE(item) == XFS_LI_INODE) {
-		imap.im_blkno = (xfs_daddr_t)in_f->ilf_blkno;
-		imap.im_len = in_f->ilf_len;
-		imap.im_boffset = in_f->ilf_boffset;
-	} else {
-		/*
-		 * It's an old inode format record.  We don't know where
-		 * its cluster is located on disk, and we can't allow
-		 * xfs_imap() to figure it out because the inode btrees
-		 * are not ready to be used.  Therefore do not pass the
-		 * XFS_IMAP_LOOKUP flag to xfs_imap().  This will give
-		 * us only the single block in which the inode lives
-		 * rather than its cluster, so we must make sure to
-		 * invalidate the buffer when we write it out below.
-		 */
-		imap.im_blkno = 0;
-		error = xfs_imap(log->l_mp, NULL, ino, &imap, 0);
-		if (error)
-			goto error;
-	}
 
 	/*
 	 * Inode buffers can be freed, look out for it,
 	 * and do not replay the inode.
 	 */
-	if (xlog_check_buffer_cancelled(log, imap.im_blkno, imap.im_len, 0)) {
+	if (xlog_check_buffer_cancelled(log, in_f->ilf_blkno,
+					in_f->ilf_len, 0)) {
 		error = 0;
 		goto error;
 	}
 
-	bp = xfs_buf_read_flags(mp->m_ddev_targp, imap.im_blkno, imap.im_len,
-								XFS_BUF_LOCK);
+	bp = xfs_buf_read_flags(mp->m_ddev_targp, in_f->ilf_blkno,
+				in_f->ilf_len, XFS_BUF_LOCK);
 	if (XFS_BUF_ISERROR(bp)) {
 		xfs_ioerror_alert("xlog_recover_do..(read#2)", mp,
-				  bp, imap.im_blkno);
+				  bp, in_f->ilf_blkno);
 		error = XFS_BUF_GETERROR(bp);
 		xfs_buf_relse(bp);
 		goto error;
 	}
 	error = 0;
 	ASSERT(in_f->ilf_fields & XFS_ILOG_CORE);
-	dip = (xfs_dinode_t *)xfs_buf_offset(bp, imap.im_boffset);
+	dip = (xfs_dinode_t *)xfs_buf_offset(bp, in_f->ilf_boffset);
 
 	/*
 	 * Make sure the place we're flushing out to really looks
 	 * like an inode!
 	 */
-	if (unlikely(be16_to_cpu(dip->di_core.di_magic) != XFS_DINODE_MAGIC)) {
+	if (unlikely(be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC)) {
 		xfs_buf_relse(bp);
 		xfs_fs_cmn_err(CE_ALERT, mp,
 			"xfs_inode_recover: Bad inode magic number, dino ptr = 0x%p, dino bp = 0x%p, ino = %Ld",
@@ -2345,12 +2316,12 @@ xlog_recover_do_inode_trans(
 	}
 
 	/* Skip replay when the on disk inode is newer than the log one */
-	if (dicp->di_flushiter < be16_to_cpu(dip->di_core.di_flushiter)) {
+	if (dicp->di_flushiter < be16_to_cpu(dip->di_flushiter)) {
 		/*
 		 * Deal with the wrap case, DI_MAX_FLUSH is less
 		 * than smaller numbers
 		 */
-		if (be16_to_cpu(dip->di_core.di_flushiter) == DI_MAX_FLUSH &&
+		if (be16_to_cpu(dip->di_flushiter) == DI_MAX_FLUSH &&
 		    dicp->di_flushiter < (DI_MAX_FLUSH >> 1)) {
 			/* do nothing */
 		} else {
@@ -2410,7 +2381,7 @@ xlog_recover_do_inode_trans(
 		error = EFSCORRUPTED;
 		goto error;
 	}
-	if (unlikely(item->ri_buf[1].i_len > sizeof(xfs_dinode_core_t))) {
+	if (unlikely(item->ri_buf[1].i_len > sizeof(struct xfs_icdinode))) {
 		XFS_CORRUPTION_ERROR("xlog_recover_do_inode_trans(7)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
 		xfs_buf_relse(bp);
@@ -2422,23 +2393,24 @@ xlog_recover_do_inode_trans(
 	}
 
 	/* The core is in in-core format */
-	xfs_dinode_to_disk(&dip->di_core,
-		(xfs_icdinode_t *)item->ri_buf[1].i_addr);
+	xfs_dinode_to_disk(dip, (xfs_icdinode_t *)item->ri_buf[1].i_addr);
 
 	/* the rest is in on-disk format */
-	if (item->ri_buf[1].i_len > sizeof(xfs_dinode_core_t)) {
-		memcpy((xfs_caddr_t) dip + sizeof(xfs_dinode_core_t),
-			item->ri_buf[1].i_addr + sizeof(xfs_dinode_core_t),
-			item->ri_buf[1].i_len  - sizeof(xfs_dinode_core_t));
+	if (item->ri_buf[1].i_len > sizeof(struct xfs_icdinode)) {
+		memcpy((xfs_caddr_t) dip + sizeof(struct xfs_icdinode),
+			item->ri_buf[1].i_addr + sizeof(struct xfs_icdinode),
+			item->ri_buf[1].i_len  - sizeof(struct xfs_icdinode));
 	}
 
 	fields = in_f->ilf_fields;
 	switch (fields & (XFS_ILOG_DEV | XFS_ILOG_UUID)) {
 	case XFS_ILOG_DEV:
-		dip->di_u.di_dev = cpu_to_be32(in_f->ilf_u.ilfu_rdev);
+		xfs_dinode_put_rdev(dip, in_f->ilf_u.ilfu_rdev);
 		break;
 	case XFS_ILOG_UUID:
-		dip->di_u.di_muuid = in_f->ilf_u.ilfu_uuid;
+		memcpy(XFS_DFORK_DPTR(dip),
+		       &in_f->ilf_u.ilfu_uuid,
+		       sizeof(uuid_t));
 		break;
 	}
 
@@ -2454,12 +2426,12 @@ xlog_recover_do_inode_trans(
 	switch (fields & XFS_ILOG_DFORK) {
 	case XFS_ILOG_DDATA:
 	case XFS_ILOG_DEXT:
-		memcpy(&dip->di_u, src, len);
+		memcpy(XFS_DFORK_DPTR(dip), src, len);
 		break;
 
 	case XFS_ILOG_DBROOT:
-		xfs_bmbt_to_bmdr((xfs_bmbt_block_t *)src, len,
-				 &(dip->di_u.di_bmbt),
+		xfs_bmbt_to_bmdr(mp, (struct xfs_btree_block *)src, len,
+				 (xfs_bmdr_block_t *)XFS_DFORK_DPTR(dip),
 				 XFS_DFORK_DSIZE(dip, mp));
 		break;
 
@@ -2496,8 +2468,8 @@ xlog_recover_do_inode_trans(
 
 		case XFS_ILOG_ABROOT:
 			dest = XFS_DFORK_APTR(dip);
-			xfs_bmbt_to_bmdr((xfs_bmbt_block_t *)src, len,
-					 (xfs_bmdr_block_t*)dest,
+			xfs_bmbt_to_bmdr(mp, (struct xfs_btree_block *)src,
+					 len, (xfs_bmdr_block_t*)dest,
 					 XFS_DFORK_ASIZE(dip, mp));
 			break;
 
@@ -2512,9 +2484,8 @@ xlog_recover_do_inode_trans(
 
 write_inode_buffer:
 	if (ITEM_TYPE(item) == XFS_LI_INODE) {
-		ASSERT(XFS_BUF_FSPRIVATE(bp, void *) == NULL ||
-		       XFS_BUF_FSPRIVATE(bp, xfs_mount_t *) == mp);
-		XFS_BUF_SET_FSPRIVATE(bp, mp);
+		ASSERT(bp->b_mount == NULL || bp->b_mount == mp);
+		bp->b_mount = mp;
 		XFS_BUF_SET_IODONE_FUNC(bp, xlog_recover_iodone);
 		xfs_bdwrite(mp, bp);
 	} else {
@@ -2645,9 +2616,8 @@ xlog_recover_do_dquot_trans(
 	memcpy(ddq, recddq, item->ri_buf[1].i_len);
 
 	ASSERT(dq_f->qlf_size == 2);
-	ASSERT(XFS_BUF_FSPRIVATE(bp, void *) == NULL ||
-	       XFS_BUF_FSPRIVATE(bp, xfs_mount_t *) == mp);
-	XFS_BUF_SET_FSPRIVATE(bp, mp);
+	ASSERT(bp->b_mount == NULL || bp->b_mount == mp);
+	bp->b_mount = mp;
 	XFS_BUF_SET_IODONE_FUNC(bp, xlog_recover_iodone);
 	xfs_bdwrite(mp, bp);
 
@@ -2689,11 +2659,11 @@ xlog_recover_do_efi_trans(
 	efip->efi_next_extent = efi_formatp->efi_nextents;
 	efip->efi_flags |= XFS_EFI_COMMITTED;
 
-	spin_lock(&mp->m_ail_lock);
+	spin_lock(&log->l_ailp->xa_lock);
 	/*
-	 * xfs_trans_update_ail() drops the AIL lock.
+	 * xfs_trans_ail_update() drops the AIL lock.
 	 */
-	xfs_trans_update_ail(mp, (xfs_log_item_t *)efip, lsn);
+	xfs_trans_ail_update(log->l_ailp, (xfs_log_item_t *)efip, lsn);
 	return 0;
 }
 
@@ -2712,12 +2682,12 @@ xlog_recover_do_efd_trans(
 	xlog_recover_item_t	*item,
 	int			pass)
 {
-	xfs_mount_t		*mp;
 	xfs_efd_log_format_t	*efd_formatp;
 	xfs_efi_log_item_t	*efip = NULL;
 	xfs_log_item_t		*lip;
-	int			gen;
 	__uint64_t		efi_id;
+	struct xfs_ail_cursor	cur;
+	struct xfs_ail		*ailp = log->l_ailp;
 
 	if (pass == XLOG_RECOVER_PASS1) {
 		return;
@@ -2734,25 +2704,26 @@ xlog_recover_do_efd_trans(
 	 * Search for the efi with the id in the efd format structure
 	 * in the AIL.
 	 */
-	mp = log->l_mp;
-	spin_lock(&mp->m_ail_lock);
-	lip = xfs_trans_first_ail(mp, &gen);
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
 	while (lip != NULL) {
 		if (lip->li_type == XFS_LI_EFI) {
 			efip = (xfs_efi_log_item_t *)lip;
 			if (efip->efi_format.efi_id == efi_id) {
 				/*
-				 * xfs_trans_delete_ail() drops the
+				 * xfs_trans_ail_delete() drops the
 				 * AIL lock.
 				 */
-				xfs_trans_delete_ail(mp, lip);
+				xfs_trans_ail_delete(ailp, lip);
 				xfs_efi_item_free(efip);
-				return;
+				spin_lock(&ailp->xa_lock);
+				break;
 			}
 		}
-		lip = xfs_trans_next_ail(mp, lip, &gen, NULL);
+		lip = xfs_trans_ail_cursor_next(ailp, &cur);
 	}
-	spin_unlock(&mp->m_ail_lock);
+	xfs_trans_ail_cursor_done(ailp, &cur);
+	spin_unlock(&ailp->xa_lock);
 }
 
 /*
@@ -3036,33 +3007,6 @@ abort_error:
 }
 
 /*
- * Verify that once we've encountered something other than an EFI
- * in the AIL that there are no more EFIs in the AIL.
- */
-#if defined(DEBUG)
-STATIC void
-xlog_recover_check_ail(
-	xfs_mount_t		*mp,
-	xfs_log_item_t		*lip,
-	int			gen)
-{
-	int			orig_gen = gen;
-
-	do {
-		ASSERT(lip->li_type != XFS_LI_EFI);
-		lip = xfs_trans_next_ail(mp, lip, &gen, NULL);
-		/*
-		 * The check will be bogus if we restart from the
-		 * beginning of the AIL, so ASSERT that we don't.
-		 * We never should since we're holding the AIL lock
-		 * the entire time.
-		 */
-		ASSERT(gen == orig_gen);
-	} while (lip != NULL);
-}
-#endif	/* DEBUG */
-
-/*
  * When this is called, all of the EFIs which did not have
  * corresponding EFDs should be in the AIL.  What we do now
  * is free the extents associated with each one.
@@ -3086,20 +3030,23 @@ xlog_recover_process_efis(
 {
 	xfs_log_item_t		*lip;
 	xfs_efi_log_item_t	*efip;
-	int			gen;
-	xfs_mount_t		*mp;
 	int			error = 0;
+	struct xfs_ail_cursor	cur;
+	struct xfs_ail		*ailp;
 
-	mp = log->l_mp;
-	spin_lock(&mp->m_ail_lock);
-
-	lip = xfs_trans_first_ail(mp, &gen);
+	ailp = log->l_ailp;
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_trans_ail_cursor_first(ailp, &cur, 0);
 	while (lip != NULL) {
 		/*
 		 * We're done when we see something other than an EFI.
+		 * There should be no EFIs left in the AIL now.
 		 */
 		if (lip->li_type != XFS_LI_EFI) {
-			xlog_recover_check_ail(mp, lip, gen);
+#ifdef DEBUG
+			for (; lip; lip = xfs_trans_ail_cursor_next(ailp, &cur))
+				ASSERT(lip->li_type != XFS_LI_EFI);
+#endif
 			break;
 		}
 
@@ -3108,18 +3055,20 @@ xlog_recover_process_efis(
 		 */
 		efip = (xfs_efi_log_item_t *)lip;
 		if (efip->efi_flags & XFS_EFI_RECOVERED) {
-			lip = xfs_trans_next_ail(mp, lip, &gen, NULL);
+			lip = xfs_trans_ail_cursor_next(ailp, &cur);
 			continue;
 		}
 
-		spin_unlock(&mp->m_ail_lock);
-		error = xlog_recover_process_efi(mp, efip);
+		spin_unlock(&ailp->xa_lock);
+		error = xlog_recover_process_efi(log->l_mp, efip);
+		spin_lock(&ailp->xa_lock);
 		if (error)
-			return error;
-		spin_lock(&mp->m_ail_lock);
-		lip = xfs_trans_next_ail(mp, lip, &gen, NULL);
+			goto out;
+		lip = xfs_trans_ail_cursor_next(ailp, &cur);
 	}
-	spin_unlock(&mp->m_ail_lock);
+out:
+	xfs_trans_ail_cursor_done(ailp, &cur);
+	spin_unlock(&ailp->xa_lock);
 	return error;
 }
 
@@ -3140,19 +3089,16 @@ xlog_recover_clear_agi_bucket(
 	int		error;
 
 	tp = xfs_trans_alloc(mp, XFS_TRANS_CLEAR_AGI_BUCKET);
-	error = xfs_trans_reserve(tp, 0, XFS_CLEAR_AGI_BUCKET_LOG_RES(mp), 0, 0, 0);
-	if (!error)
-		error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
-				   XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
-				   XFS_FSS_TO_BB(mp, 1), 0, &agibp);
+	error = xfs_trans_reserve(tp, 0, XFS_CLEAR_AGI_BUCKET_LOG_RES(mp),
+				  0, 0, 0);
 	if (error)
 		goto out_abort;
 
-	error = EINVAL;
-	agi = XFS_BUF_TO_AGI(agibp);
-	if (be32_to_cpu(agi->agi_magicnum) != XFS_AGI_MAGIC)
+	error = xfs_read_agi(mp, tp, agno, &agibp);
+	if (error)
 		goto out_abort;
 
+	agi = XFS_BUF_TO_AGI(agibp);
 	agi->agi_unlinked[bucket] = cpu_to_be32(NULLAGINO);
 	offset = offsetof(xfs_agi_t, agi_unlinked) +
 		 (sizeof(xfs_agino_t) * bucket);
@@ -3172,6 +3118,62 @@ out_error:
 	return;
 }
 
+STATIC xfs_agino_t
+xlog_recover_process_one_iunlink(
+	struct xfs_mount		*mp,
+	xfs_agnumber_t			agno,
+	xfs_agino_t			agino,
+	int				bucket)
+{
+	struct xfs_buf			*ibp;
+	struct xfs_dinode		*dip;
+	struct xfs_inode		*ip;
+	xfs_ino_t			ino;
+	int				error;
+
+	ino = XFS_AGINO_TO_INO(mp, agno, agino);
+	error = xfs_iget(mp, NULL, ino, 0, 0, &ip, 0);
+	if (error)
+		goto fail;
+
+	/*
+	 * Get the on disk inode to find the next inode in the bucket.
+	 */
+	error = xfs_itobp(mp, NULL, ip, &dip, &ibp, XFS_BUF_LOCK);
+	if (error)
+		goto fail_iput;
+
+	ASSERT(ip->i_d.di_nlink == 0);
+	ASSERT(ip->i_d.di_mode != 0);
+
+	/* setup for the next pass */
+	agino = be32_to_cpu(dip->di_next_unlinked);
+	xfs_buf_relse(ibp);
+
+	/*
+	 * Prevent any DMAPI event from being sent when the reference on
+	 * the inode is dropped.
+	 */
+	ip->i_d.di_dmevmask = 0;
+
+	IRELE(ip);
+	return agino;
+
+ fail_iput:
+	IRELE(ip);
+ fail:
+	/*
+	 * We can't read in the inode this bucket points to, or this inode
+	 * is messed up.  Just ditch this bucket of inodes.  We will lose
+	 * some inodes and space, but at least we won't hang.
+	 *
+	 * Call xlog_recover_clear_agi_bucket() to perform a transaction to
+	 * clear the inode pointer in the bucket.
+	 */
+	xlog_recover_clear_agi_bucket(mp, agno, bucket);
+	return NULLAGINO;
+}
+
 /*
  * xlog_iunlink_recover
  *
@@ -3192,11 +3194,7 @@ xlog_recover_process_iunlinks(
 	xfs_agnumber_t	agno;
 	xfs_agi_t	*agi;
 	xfs_buf_t	*agibp;
-	xfs_buf_t	*ibp;
-	xfs_dinode_t	*dip;
-	xfs_inode_t	*ip;
 	xfs_agino_t	agino;
-	xfs_ino_t	ino;
 	int		bucket;
 	int		error;
 	uint		mp_dmevmask;
@@ -3213,22 +3211,21 @@ xlog_recover_process_iunlinks(
 		/*
 		 * Find the agi for this ag.
 		 */
-		agibp = xfs_buf_read(mp->m_ddev_targp,
-				XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
-				XFS_FSS_TO_BB(mp, 1), 0);
-		if (XFS_BUF_ISERROR(agibp)) {
-			xfs_ioerror_alert("xlog_recover_process_iunlinks(#1)",
-				log->l_mp, agibp,
-				XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)));
+		error = xfs_read_agi(mp, NULL, agno, &agibp);
+		if (error) {
+			/*
+			 * AGI is b0rked. Don't process it.
+			 *
+			 * We should probably mark the filesystem as corrupt
+			 * after we've recovered all the ag's we can....
+			 */
+			continue;
 		}
 		agi = XFS_BUF_TO_AGI(agibp);
-		ASSERT(XFS_AGI_MAGIC == be32_to_cpu(agi->agi_magicnum));
 
 		for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++) {
-
 			agino = be32_to_cpu(agi->agi_unlinked[bucket]);
 			while (agino != NULLAGINO) {
-
 				/*
 				 * Release the agi buffer so that it can
 				 * be acquired in the normal course of the
@@ -3236,87 +3233,17 @@ xlog_recover_process_iunlinks(
 				 */
 				xfs_buf_relse(agibp);
 
-				ino = XFS_AGINO_TO_INO(mp, agno, agino);
-				error = xfs_iget(mp, NULL, ino, 0, 0, &ip, 0);
-				ASSERT(error || (ip != NULL));
-
-				if (!error) {
-					/*
-					 * Get the on disk inode to find the
-					 * next inode in the bucket.
-					 */
-					error = xfs_itobp(mp, NULL, ip, &dip,
-							&ibp, 0, 0,
-							XFS_BUF_LOCK);
-					ASSERT(error || (dip != NULL));
-				}
-
-				if (!error) {
-					ASSERT(ip->i_d.di_nlink == 0);
-
-					/* setup for the next pass */
-					agino = be32_to_cpu(
-							dip->di_next_unlinked);
-					xfs_buf_relse(ibp);
-					/*
-					 * Prevent any DMAPI event from
-					 * being sent when the
-					 * reference on the inode is
-					 * dropped.
-					 */
-					ip->i_d.di_dmevmask = 0;
-
-					/*
-					 * If this is a new inode, handle
-					 * it specially.  Otherwise,
-					 * just drop our reference to the
-					 * inode.  If there are no
-					 * other references, this will
-					 * send the inode to
-					 * xfs_inactive() which will
-					 * truncate the file and free
-					 * the inode.
-					 */
-					if (ip->i_d.di_mode == 0)
-						xfs_iput_new(ip, 0);
-					else
-						IRELE(ip);
-				} else {
-					/*
-					 * We can't read in the inode
-					 * this bucket points to, or
-					 * this inode is messed up.  Just
-					 * ditch this bucket of inodes.  We
-					 * will lose some inodes and space,
-					 * but at least we won't hang.  Call
-					 * xlog_recover_clear_agi_bucket()
-					 * to perform a transaction to clear
-					 * the inode pointer in the bucket.
-					 */
-					xlog_recover_clear_agi_bucket(mp, agno,
-							bucket);
-
-					agino = NULLAGINO;
-				}
+				agino = xlog_recover_process_one_iunlink(mp,
+							agno, agino, bucket);
 
 				/*
 				 * Reacquire the agibuffer and continue around
-				 * the loop.
+				 * the loop. This should never fail as we know
+				 * the buffer was good earlier on.
 				 */
-				agibp = xfs_buf_read(mp->m_ddev_targp,
-						XFS_AG_DADDR(mp, agno,
-							XFS_AGI_DADDR(mp)),
-						XFS_FSS_TO_BB(mp, 1), 0);
-				if (XFS_BUF_ISERROR(agibp)) {
-					xfs_ioerror_alert(
-				"xlog_recover_process_iunlinks(#2)",
-						log->l_mp, agibp,
-						XFS_AG_DADDR(mp, agno,
-							XFS_AGI_DADDR(mp)));
-				}
+				error = xfs_read_agi(mp, NULL, agno, &agibp);
+				ASSERT(error == 0);
 				agi = XFS_BUF_TO_AGI(agibp);
-				ASSERT(XFS_AGI_MAGIC == be32_to_cpu(
-					agi->agi_magicnum));
 			}
 		}
 
@@ -3367,7 +3294,6 @@ xlog_pack_data(
 	int			size = iclog->ic_offset + roundoff;
 	__be32			cycle_lsn;
 	xfs_caddr_t		dp;
-	xlog_in_core_2_t	*xhdr;
 
 	xlog_pack_data_checksum(log, iclog, size);
 
@@ -3382,7 +3308,8 @@ xlog_pack_data(
 	}
 
 	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
-		xhdr = (xlog_in_core_2_t *)&iclog->ic_header;
+		xlog_in_core_2_t *xhdr = iclog->ic_data;
+
 		for ( ; i < BTOBB(size); i++) {
 			j = i / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
 			k = i % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
@@ -3440,7 +3367,6 @@ xlog_unpack_data(
 	xlog_t			*log)
 {
 	int			i, j, k;
-	xlog_in_core_2_t	*xhdr;
 
 	for (i = 0; i < BTOBB(be32_to_cpu(rhead->h_len)) &&
 		  i < (XLOG_HEADER_CYCLE_SIZE / BBSIZE); i++) {
@@ -3449,7 +3375,7 @@ xlog_unpack_data(
 	}
 
 	if (xfs_sb_version_haslogv2(&log->l_mp->m_sb)) {
-		xhdr = (xlog_in_core_2_t *)rhead;
+		xlog_in_core_2_t *xhdr = (xlog_in_core_2_t *)rhead;
 		for ( ; i < BTOBB(be32_to_cpu(rhead->h_len)); i++) {
 			j = i / (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
 			k = i % (XLOG_HEADER_CYCLE_SIZE / BBSIZE);
@@ -4003,11 +3929,8 @@ xlog_recover_check_summary(
 {
 	xfs_mount_t	*mp;
 	xfs_agf_t	*agfp;
-	xfs_agi_t	*agip;
 	xfs_buf_t	*agfbp;
 	xfs_buf_t	*agibp;
-	xfs_daddr_t	agfdaddr;
-	xfs_daddr_t	agidaddr;
 	xfs_buf_t	*sbbp;
 #ifdef XFS_LOUD_RECOVERY
 	xfs_sb_t	*sbp;
@@ -4016,6 +3939,7 @@ xlog_recover_check_summary(
 	__uint64_t	freeblks;
 	__uint64_t	itotal;
 	__uint64_t	ifree;
+	int		error;
 
 	mp = log->l_mp;
 
@@ -4023,37 +3947,27 @@ xlog_recover_check_summary(
 	itotal = 0LL;
 	ifree = 0LL;
 	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
-		agfdaddr = XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp));
-		agfbp = xfs_buf_read(mp->m_ddev_targp, agfdaddr,
-				XFS_FSS_TO_BB(mp, 1), 0);
-		if (XFS_BUF_ISERROR(agfbp)) {
-			xfs_ioerror_alert("xlog_recover_check_summary(agf)",
-						mp, agfbp, agfdaddr);
-		}
-		agfp = XFS_BUF_TO_AGF(agfbp);
-		ASSERT(XFS_AGF_MAGIC == be32_to_cpu(agfp->agf_magicnum));
-		ASSERT(XFS_AGF_GOOD_VERSION(be32_to_cpu(agfp->agf_versionnum)));
-		ASSERT(be32_to_cpu(agfp->agf_seqno) == agno);
-
-		freeblks += be32_to_cpu(agfp->agf_freeblks) +
-			    be32_to_cpu(agfp->agf_flcount);
-		xfs_buf_relse(agfbp);
-
-		agidaddr = XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp));
-		agibp = xfs_buf_read(mp->m_ddev_targp, agidaddr,
-				XFS_FSS_TO_BB(mp, 1), 0);
-		if (XFS_BUF_ISERROR(agibp)) {
-			xfs_ioerror_alert("xlog_recover_check_summary(agi)",
-					  mp, agibp, agidaddr);
+		error = xfs_read_agf(mp, NULL, agno, 0, &agfbp);
+		if (error) {
+			xfs_fs_cmn_err(CE_ALERT, mp,
+					"xlog_recover_check_summary(agf)"
+					"agf read failed agno %d error %d",
+							agno, error);
+		} else {
+			agfp = XFS_BUF_TO_AGF(agfbp);
+			freeblks += be32_to_cpu(agfp->agf_freeblks) +
+				    be32_to_cpu(agfp->agf_flcount);
+			xfs_buf_relse(agfbp);
 		}
-		agip = XFS_BUF_TO_AGI(agibp);
-		ASSERT(XFS_AGI_MAGIC == be32_to_cpu(agip->agi_magicnum));
-		ASSERT(XFS_AGI_GOOD_VERSION(be32_to_cpu(agip->agi_versionnum)));
-		ASSERT(be32_to_cpu(agip->agi_seqno) == agno);
 
-		itotal += be32_to_cpu(agip->agi_count);
-		ifree += be32_to_cpu(agip->agi_freecount);
-		xfs_buf_relse(agibp);
+		error = xfs_read_agi(mp, NULL, agno, &agibp);
+		if (!error) {
+			struct xfs_agi	*agi = XFS_BUF_TO_AGI(agibp);
+
+			itotal += be32_to_cpu(agi->agi_count);
+			ifree += be32_to_cpu(agi->agi_freecount);
+			xfs_buf_relse(agibp);
+		}
 	}
 
 	sbbp = xfs_getsb(mp, 0);
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 15f5dd22fbb..3c97c6463a4 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -567,8 +567,6 @@ xfs_readsb(xfs_mount_t *mp, int flags)
 STATIC void
 xfs_mount_common(xfs_mount_t *mp, xfs_sb_t *sbp)
 {
-	int	i;
-
 	mp->m_agfrotor = mp->m_agirotor = 0;
 	spin_lock_init(&mp->m_agirotor_lock);
 	mp->m_maxagi = mp->m_sb.sb_agcount;
@@ -577,12 +575,10 @@ xfs_mount_common(xfs_mount_t *mp, xfs_sb_t *sbp)
 	mp->m_sectbb_log = sbp->sb_sectlog - BBSHIFT;
 	mp->m_agno_log = xfs_highbit32(sbp->sb_agcount - 1) + 1;
 	mp->m_agino_log = sbp->sb_inopblog + sbp->sb_agblklog;
-	mp->m_litino = sbp->sb_inodesize -
-		((uint)sizeof(xfs_dinode_core_t) + (uint)sizeof(xfs_agino_t));
+	mp->m_litino = sbp->sb_inodesize - sizeof(struct xfs_dinode);
 	mp->m_blockmask = sbp->sb_blocksize - 1;
 	mp->m_blockwsize = sbp->sb_blocksize >> XFS_WORDLOG;
 	mp->m_blockwmask = mp->m_blockwsize - 1;
-	INIT_LIST_HEAD(&mp->m_del_inodes);
 
 	/*
 	 * Setup for attributes, in case they get created.
@@ -605,24 +601,20 @@ xfs_mount_common(xfs_mount_t *mp, xfs_sb_t *sbp)
 	}
 	ASSERT(mp->m_attroffset < XFS_LITINO(mp));
 
-	for (i = 0; i < 2; i++) {
-		mp->m_alloc_mxr[i] = XFS_BTREE_BLOCK_MAXRECS(sbp->sb_blocksize,
-			xfs_alloc, i == 0);
-		mp->m_alloc_mnr[i] = XFS_BTREE_BLOCK_MINRECS(sbp->sb_blocksize,
-			xfs_alloc, i == 0);
-	}
-	for (i = 0; i < 2; i++) {
-		mp->m_bmap_dmxr[i] = XFS_BTREE_BLOCK_MAXRECS(sbp->sb_blocksize,
-			xfs_bmbt, i == 0);
-		mp->m_bmap_dmnr[i] = XFS_BTREE_BLOCK_MINRECS(sbp->sb_blocksize,
-			xfs_bmbt, i == 0);
-	}
-	for (i = 0; i < 2; i++) {
-		mp->m_inobt_mxr[i] = XFS_BTREE_BLOCK_MAXRECS(sbp->sb_blocksize,
-			xfs_inobt, i == 0);
-		mp->m_inobt_mnr[i] = XFS_BTREE_BLOCK_MINRECS(sbp->sb_blocksize,
-			xfs_inobt, i == 0);
-	}
+	mp->m_alloc_mxr[0] = xfs_allocbt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_alloc_mxr[1] = xfs_allocbt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_alloc_mnr[0] = mp->m_alloc_mxr[0] / 2;
+	mp->m_alloc_mnr[1] = mp->m_alloc_mxr[1] / 2;
+
+	mp->m_inobt_mxr[0] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_inobt_mxr[1] = xfs_inobt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_inobt_mnr[0] = mp->m_inobt_mxr[0] / 2;
+	mp->m_inobt_mnr[1] = mp->m_inobt_mxr[1] / 2;
+
+	mp->m_bmap_dmxr[0] = xfs_bmbt_maxrecs(mp, sbp->sb_blocksize, 1);
+	mp->m_bmap_dmxr[1] = xfs_bmbt_maxrecs(mp, sbp->sb_blocksize, 0);
+	mp->m_bmap_dmnr[0] = mp->m_bmap_dmxr[0] / 2;
+	mp->m_bmap_dmnr[1] = mp->m_bmap_dmxr[1] / 2;
 
 	mp->m_bsize = XFS_FSB_TO_BB(mp, 1);
 	mp->m_ialloc_inos = (int)MAX((__uint16_t)XFS_INODES_PER_CHUNK,
@@ -1228,6 +1220,16 @@ xfs_unmountfs(
 	__uint64_t		resblks;
 	int			error;
 
+	/*
+	 * Release dquot that rootinode, rbmino and rsumino might be holding,
+	 * and release the quota inodes.
+	 */
+	XFS_QM_UNMOUNT(mp);
+
+	if (mp->m_rbmip)
+		IRELE(mp->m_rbmip);
+	if (mp->m_rsumip)
+		IRELE(mp->m_rsumip);
 	IRELE(mp->m_rootip);
 
 	/*
@@ -1241,7 +1243,7 @@ xfs_unmountfs(
 	 * need to force the log first.
 	 */
 	xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE | XFS_LOG_SYNC);
-	xfs_iflush_all(mp);
+	xfs_reclaim_inodes(mp, 0, XFS_IFLUSH_ASYNC);
 
 	XFS_QM_DQPURGEALL(mp, XFS_QMOPT_QUOTALL | XFS_QMOPT_UMOUNTING);
 
@@ -1288,11 +1290,6 @@ xfs_unmountfs(
 	xfs_unmountfs_wait(mp); 		/* wait for async bufs */
 	xfs_log_unmount(mp);			/* Done! No more fs ops. */
 
-	/*
-	 * All inodes from this mount point should be freed.
-	 */
-	ASSERT(mp->m_inodes == NULL);
-
 	if ((mp->m_flags & XFS_MOUNT_NOUUID) == 0)
 		uuid_table_remove(&mp->m_sb.sb_uuid);
 
@@ -1365,24 +1362,6 @@ xfs_log_sbcount(
 	return error;
 }
 
-STATIC void
-xfs_mark_shared_ro(
-	xfs_mount_t	*mp,
-	xfs_buf_t	*bp)
-{
-	xfs_dsb_t	*sb = XFS_BUF_TO_SBP(bp);
-	__uint16_t	version;
-
-	if (!(sb->sb_flags & XFS_SBF_READONLY))
-		sb->sb_flags |= XFS_SBF_READONLY;
-
-	version = be16_to_cpu(sb->sb_versionnum);
-	if ((version & XFS_SB_VERSION_NUMBITS) != XFS_SB_VERSION_4 ||
-	    !(version & XFS_SB_VERSION_SHAREDBIT))
-		version |= XFS_SB_VERSION_SHAREDBIT;
-	sb->sb_versionnum = cpu_to_be16(version);
-}
-
 int
 xfs_unmountfs_writesb(xfs_mount_t *mp)
 {
@@ -1398,12 +1377,6 @@ xfs_unmountfs_writesb(xfs_mount_t *mp)
 
 		sbp = xfs_getsb(mp, 0);
 
-		/*
-		 * mark shared-readonly if desired
-		 */
-		if (mp->m_mk_sharedro)
-			xfs_mark_shared_ro(mp, sbp);
-
 		XFS_BUF_UNDONE(sbp);
 		XFS_BUF_UNREAD(sbp);
 		XFS_BUF_UNDELAYWRITE(sbp);
@@ -1415,8 +1388,6 @@ xfs_unmountfs_writesb(xfs_mount_t *mp)
 		if (error)
 			xfs_ioerror_alert("xfs_unmountfs_writesb",
 					  mp, sbp, XFS_BUF_ADDR(sbp));
-		if (error && mp->m_mk_sharedro)
-			xfs_fs_cmn_err(CE_ALERT, mp, "Superblock write error detected while unmounting.  Filesystem may not be marked shared readonly");
 		xfs_buf_relse(sbp);
 	}
 	return error;
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index f3c1024b124..c1e02846732 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -18,7 +18,6 @@
 #ifndef __XFS_MOUNT_H__
 #define	__XFS_MOUNT_H__
 
-
 typedef struct xfs_trans_reservations {
 	uint	tr_write;	/* extent alloc trans */
 	uint	tr_itruncate;	/* truncate trans */
@@ -44,14 +43,16 @@ typedef struct xfs_trans_reservations {
 } xfs_trans_reservations_t;
 
 #ifndef __KERNEL__
-/*
- * Moved here from xfs_ag.h to avoid reordering header files
- */
+
 #define XFS_DADDR_TO_AGNO(mp,d) \
 	((xfs_agnumber_t)(XFS_BB_TO_FSBT(mp, d) / (mp)->m_sb.sb_agblocks))
 #define XFS_DADDR_TO_AGBNO(mp,d) \
 	((xfs_agblock_t)(XFS_BB_TO_FSBT(mp, d) % (mp)->m_sb.sb_agblocks))
-#else
+
+#else /* __KERNEL__ */
+
+#include "xfs_sync.h"
+
 struct cred;
 struct log;
 struct xfs_mount_args;
@@ -62,6 +63,7 @@ struct xfs_extdelta;
 struct xfs_swapext;
 struct xfs_mru_cache;
 struct xfs_nameops;
+struct xfs_ail;
 
 /*
  * Prototypes and functions for the Data Migration subsystem.
@@ -115,7 +117,7 @@ struct xfs_quotainfo;
 
 typedef int	(*xfs_qminit_t)(struct xfs_mount *, uint *, uint *);
 typedef int	(*xfs_qmmount_t)(struct xfs_mount *, uint, uint);
-typedef int	(*xfs_qmunmount_t)(struct xfs_mount *);
+typedef void	(*xfs_qmunmount_t)(struct xfs_mount *);
 typedef void	(*xfs_qmdone_t)(struct xfs_mount *);
 typedef void	(*xfs_dqrele_t)(struct xfs_dquot *);
 typedef int	(*xfs_dqattach_t)(struct xfs_inode *, uint);
@@ -132,7 +134,7 @@ typedef struct xfs_dquot * (*xfs_dqvopchown_t)(
 			struct xfs_dquot **, struct xfs_dquot *);
 typedef int	(*xfs_dqvopchownresv_t)(struct xfs_trans *, struct xfs_inode *,
 			struct xfs_dquot *, struct xfs_dquot *, uint);
-typedef void	(*xfs_dqstatvfs_t)(struct xfs_inode *, bhv_statvfs_t *);
+typedef void	(*xfs_dqstatvfs_t)(struct xfs_inode *, struct kstatfs *);
 typedef int	(*xfs_dqsync_t)(struct xfs_mount *, int flags);
 typedef int	(*xfs_quotactl_t)(struct xfs_mount *, int, int, xfs_caddr_t);
 
@@ -223,18 +225,10 @@ extern void	xfs_icsb_sync_counters_locked(struct xfs_mount *, int);
 #define xfs_icsb_sync_counters_locked(mp, flags) do { } while (0)
 #endif
 
-typedef struct xfs_ail {
-	struct list_head	xa_ail;
-	uint			xa_gen;
-	struct task_struct	*xa_task;
-	xfs_lsn_t		xa_target;
-} xfs_ail_t;
-
 typedef struct xfs_mount {
 	struct super_block	*m_super;
 	xfs_tid_t		m_tid;		/* next unused tid for fs */
-	spinlock_t		m_ail_lock;	/* fs AIL mutex */
-	xfs_ail_t		m_ail;		/* fs active log item list */
+	struct xfs_ail		*m_ail;		/* fs active log item list */
 	xfs_sb_t		m_sb;		/* copy of fs superblock */
 	spinlock_t		m_sb_lock;	/* sb counter lock */
 	struct xfs_buf		*m_sb_bp;	/* buffer for superblock */
@@ -247,10 +241,6 @@ typedef struct xfs_mount {
 	xfs_agnumber_t		m_agirotor;	/* last ag dir inode alloced */
 	spinlock_t		m_agirotor_lock;/* .. and lock protecting it */
 	xfs_agnumber_t		m_maxagi;	/* highest inode alloc group */
-	struct xfs_inode	*m_inodes;	/* active inode list */
-	struct list_head	m_del_inodes;	/* inodes to reclaim */
-	mutex_t			m_ilock;	/* inode list mutex */
-	uint			m_ireclaims;	/* count of calls to reclaim*/
 	uint			m_readio_log;	/* min read size log bytes */
 	uint			m_readio_blocks; /* min read size blocks */
 	uint			m_writeio_log;	/* min write size log bytes */
@@ -267,7 +257,6 @@ typedef struct xfs_mount {
 	xfs_buftarg_t		*m_ddev_targp;	/* saves taking the address */
 	xfs_buftarg_t		*m_logdev_targp;/* ptr to log device */
 	xfs_buftarg_t		*m_rtdev_targp;	/* ptr to rt device */
-	__uint8_t		m_dircook_elog;	/* log d-cookie entry bits */
 	__uint8_t		m_blkbit_log;	/* blocklog + NBBY */
 	__uint8_t		m_blkbb_log;	/* blocklog - BBSHIFT */
 	__uint8_t		m_agno_log;	/* log #ag's */
@@ -276,12 +265,12 @@ typedef struct xfs_mount {
 	uint			m_blockmask;	/* sb_blocksize-1 */
 	uint			m_blockwsize;	/* sb_blocksize in words */
 	uint			m_blockwmask;	/* blockwsize-1 */
-	uint			m_alloc_mxr[2];	/* XFS_ALLOC_BLOCK_MAXRECS */
-	uint			m_alloc_mnr[2];	/* XFS_ALLOC_BLOCK_MINRECS */
-	uint			m_bmap_dmxr[2];	/* XFS_BMAP_BLOCK_DMAXRECS */
-	uint			m_bmap_dmnr[2];	/* XFS_BMAP_BLOCK_DMINRECS */
-	uint			m_inobt_mxr[2];	/* XFS_INOBT_BLOCK_MAXRECS */
-	uint			m_inobt_mnr[2];	/* XFS_INOBT_BLOCK_MINRECS */
+	uint			m_alloc_mxr[2];	/* max alloc btree records */
+	uint			m_alloc_mnr[2];	/* min alloc btree records */
+	uint			m_bmap_dmxr[2];	/* max bmap btree records */
+	uint			m_bmap_dmnr[2];	/* min bmap btree records */
+	uint			m_inobt_mxr[2];	/* max inobt btree records */
+	uint			m_inobt_mnr[2];	/* min inobt btree records */
 	uint			m_ag_maxlevels;	/* XFS_AG_MAXLEVELS */
 	uint			m_bm_maxlevels[2]; /* XFS_BM_MAXLEVELS */
 	uint			m_in_maxlevels;	/* XFS_IN_MAXLEVELS */
@@ -312,9 +301,6 @@ typedef struct xfs_mount {
 	int			m_sinoalign;	/* stripe unit inode alignment */
 	int			m_attr_magicpct;/* 37% of the blocksize */
 	int			m_dir_magicpct;	/* 37% of the dir blocksize */
-	__uint8_t		m_mk_sharedro;	/* mark shared ro on unmount */
-	__uint8_t		m_inode_quiesce;/* call quiesce on new inodes.
-						   field governed by m_ilock */
 	__uint8_t		m_sectbb_log;	/* sectlog - BBSHIFT */
 	const struct xfs_nameops *m_dirnameops;	/* vector of dir name ops */
 	int			m_dirblksize;	/* directory block sz--bytes */
@@ -362,7 +348,6 @@ typedef struct xfs_mount {
 #define XFS_MOUNT_ATTR2		(1ULL << 8)	/* allow use of attr2 format */
 #define XFS_MOUNT_GRPID		(1ULL << 9)	/* group-ID assigned from directory */
 #define XFS_MOUNT_NORECOVERY	(1ULL << 10)	/* no recovery - dirty fs */
-#define XFS_MOUNT_SHARED	(1ULL << 11)	/* shared mount */
 #define XFS_MOUNT_DFLT_IOSIZE	(1ULL << 12)	/* set default i/o size */
 #define XFS_MOUNT_OSYNCISOSYNC	(1ULL << 13)	/* o_sync is REALLY o_sync */
 						/* osyncisdsync is now default*/
@@ -439,6 +424,16 @@ void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
 #define xfs_force_shutdown(m,f)	\
 	xfs_do_force_shutdown(m, f, __FILE__, __LINE__)
 
+#define SHUTDOWN_META_IO_ERROR	0x0001	/* write attempt to metadata failed */
+#define SHUTDOWN_LOG_IO_ERROR	0x0002	/* write attempt to the log failed */
+#define SHUTDOWN_FORCE_UMOUNT	0x0004	/* shutdown from a forced unmount */
+#define SHUTDOWN_CORRUPT_INCORE	0x0008	/* corrupt in-memory data structures */
+#define SHUTDOWN_REMOTE_REQ	0x0010	/* shutdown came from remote cell */
+#define SHUTDOWN_DEVICE_REQ	0x0020	/* failed all paths to the device */
+
+#define xfs_test_for_freeze(mp)		((mp)->m_super->s_frozen)
+#define xfs_wait_for_freeze(mp,l)	vfs_check_frozen((mp)->m_super, (l))
+
 /*
  * Flags for xfs_mountfs
  */
@@ -508,14 +503,12 @@ typedef struct xfs_mod_sb {
 #define	XFS_MOUNT_ILOCK(mp)	mutex_lock(&((mp)->m_ilock))
 #define	XFS_MOUNT_IUNLOCK(mp)	mutex_unlock(&((mp)->m_ilock))
 
-extern void	xfs_mod_sb(xfs_trans_t *, __int64_t);
 extern int	xfs_log_sbcount(xfs_mount_t *, uint);
 extern int	xfs_mountfs(xfs_mount_t *mp);
 extern void	xfs_mountfs_check_barriers(xfs_mount_t *mp);
 
 extern void	xfs_unmountfs(xfs_mount_t *);
 extern int	xfs_unmountfs_writesb(xfs_mount_t *);
-extern int	xfs_unmount_flush(xfs_mount_t *, int);
 extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int64_t, int);
 extern int	xfs_mod_incore_sb_unlocked(xfs_mount_t *, xfs_sb_field_t,
 			int64_t, int);
@@ -525,20 +518,20 @@ extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);
 extern int	xfs_readsb(xfs_mount_t *, int);
 extern void	xfs_freesb(xfs_mount_t *);
 extern int	xfs_fs_writable(xfs_mount_t *);
-extern int	xfs_syncsub(xfs_mount_t *, int, int *);
-extern int	xfs_sync_inodes(xfs_mount_t *, int, int *);
-extern xfs_agnumber_t	xfs_initialize_perag(xfs_mount_t *, xfs_agnumber_t);
-extern void	xfs_sb_from_disk(struct xfs_sb *, struct xfs_dsb *);
-extern void	xfs_sb_to_disk(struct xfs_dsb *, struct xfs_sb *, __int64_t);
 extern int	xfs_sb_validate_fsb_count(struct xfs_sb *, __uint64_t);
 
-extern int	xfs_dmops_get(struct xfs_mount *, struct xfs_mount_args *);
+extern int	xfs_dmops_get(struct xfs_mount *);
 extern void	xfs_dmops_put(struct xfs_mount *);
-extern int	xfs_qmops_get(struct xfs_mount *, struct xfs_mount_args *);
+extern int	xfs_qmops_get(struct xfs_mount *);
 extern void	xfs_qmops_put(struct xfs_mount *);
 
 extern struct xfs_dmops xfs_dmcore_xfs;
 
 #endif	/* __KERNEL__ */
 
+extern void	xfs_mod_sb(struct xfs_trans *, __int64_t);
+extern xfs_agnumber_t	xfs_initialize_perag(struct xfs_mount *, xfs_agnumber_t);
+extern void	xfs_sb_from_disk(struct xfs_sb *, struct xfs_dsb *);
+extern void	xfs_sb_to_disk(struct xfs_dsb *, struct xfs_sb *, __int64_t);
+
 #endif	/* __XFS_MOUNT_H__ */
diff --git a/fs/xfs/xfs_qmops.c b/fs/xfs/xfs_qmops.c
index a294e58db8d..27f80581520 100644
--- a/fs/xfs/xfs_qmops.c
+++ b/fs/xfs/xfs_qmops.c
@@ -28,7 +28,6 @@
 #include "xfs_mount.h"
 #include "xfs_quota.h"
 #include "xfs_error.h"
-#include "xfs_clnt.h"
 
 
 STATIC struct xfs_dquot *
@@ -131,9 +130,9 @@ static struct xfs_qmops xfs_qmcore_stub = {
 };
 
 int
-xfs_qmops_get(struct xfs_mount *mp, struct xfs_mount_args *args)
+xfs_qmops_get(struct xfs_mount *mp)
 {
-	if (args->flags & (XFSMNT_UQUOTA | XFSMNT_PQUOTA | XFSMNT_GQUOTA)) {
+	if (XFS_IS_QUOTA_RUNNING(mp)) {
 #ifdef CONFIG_XFS_QUOTA
 		mp->m_qm_ops = &xfs_qmcore_xfs;
 #else
diff --git a/fs/xfs/xfs_quota.h b/fs/xfs/xfs_quota.h
index 12c4ec775af..48965ecaa15 100644
--- a/fs/xfs/xfs_quota.h
+++ b/fs/xfs/xfs_quota.h
@@ -84,11 +84,9 @@ typedef struct xfs_dqblk {
 #define XFS_DQ_USER		0x0001		/* a user quota */
 #define XFS_DQ_PROJ		0x0002		/* project quota */
 #define XFS_DQ_GROUP		0x0004		/* a group quota */
-#define XFS_DQ_FLOCKED		0x0008		/* flush lock taken */
-#define XFS_DQ_DIRTY		0x0010		/* dquot is dirty */
-#define XFS_DQ_WANT		0x0020		/* for lookup/reclaim race */
-#define XFS_DQ_INACTIVE		0x0040		/* dq off mplist & hashlist */
-#define XFS_DQ_MARKER		0x0080		/* sentinel */
+#define XFS_DQ_DIRTY		0x0008		/* dquot is dirty */
+#define XFS_DQ_WANT		0x0010		/* for lookup/reclaim race */
+#define XFS_DQ_INACTIVE		0x0020		/* dq off mplist & hashlist */
 
 #define XFS_DQ_ALLTYPES		(XFS_DQ_USER|XFS_DQ_PROJ|XFS_DQ_GROUP)
 
diff --git a/fs/xfs/xfs_rename.c b/fs/xfs/xfs_rename.c
index c903130be7f..86471bb40fd 100644
--- a/fs/xfs/xfs_rename.c
+++ b/fs/xfs/xfs_rename.c
@@ -42,31 +42,6 @@
 
 
 /*
- * Given an array of up to 4 inode pointers, unlock the pointed to inodes.
- * If there are fewer than 4 entries in the array, the empty entries will
- * be at the end and will have NULL pointers in them.
- */
-STATIC void
-xfs_rename_unlock4(
-	xfs_inode_t	**i_tab,
-	uint		lock_mode)
-{
-	int	i;
-
-	xfs_iunlock(i_tab[0], lock_mode);
-	for (i = 1; i < 4; i++) {
-		if (i_tab[i] == NULL)
-			break;
-
-		/*
-		 * Watch out for duplicate entries in the table.
-		 */
-		if (i_tab[i] != i_tab[i-1])
-			xfs_iunlock(i_tab[i], lock_mode);
-	}
-}
-
-/*
  * Enter all inodes for a rename transaction into a sorted array.
  */
 STATIC void
@@ -205,19 +180,6 @@ xfs_rename(
 	xfs_lock_inodes(inodes, num_inodes, XFS_ILOCK_EXCL);
 
 	/*
-	 * If we are using project inheritance, we only allow renames
-	 * into our tree when the project IDs are the same; else the
-	 * tree quota mechanism would be circumvented.
-	 */
-	if (unlikely((target_dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
-		     (target_dp->i_d.di_projid != src_ip->i_d.di_projid))) {
-		error = XFS_ERROR(EXDEV);
-		xfs_rename_unlock4(inodes, XFS_ILOCK_EXCL);
-		xfs_trans_cancel(tp, cancel_flags);
-		goto std_return;
-	}
-
-	/*
 	 * Join all the inodes to the transaction. From this point on,
 	 * we can rely on either trans_commit or trans_cancel to unlock
 	 * them.  Note that we need to add a vnode reference to the
@@ -242,6 +204,17 @@ xfs_rename(
 	}
 
 	/*
+	 * If we are using project inheritance, we only allow renames
+	 * into our tree when the project IDs are the same; else the
+	 * tree quota mechanism would be circumvented.
+	 */
+	if (unlikely((target_dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
+		     (target_dp->i_d.di_projid != src_ip->i_d.di_projid))) {
+		error = XFS_ERROR(EXDEV);
+		goto error_return;
+	}
+
+	/*
 	 * Set up the target.
 	 */
 	if (target_ip == NULL) {
@@ -367,19 +340,11 @@ xfs_rename(
 					&first_block, &free_list, spaceres);
 	if (error)
 		goto abort_return;
-	xfs_ichgtime(src_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
 
-	/*
-	 * Update the generation counts on all the directory inodes
-	 * that we're modifying.
-	 */
-	src_dp->i_gen++;
+	xfs_ichgtime(src_dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
 	xfs_trans_log_inode(tp, src_dp, XFS_ILOG_CORE);
-
-	if (new_parent) {
-		target_dp->i_gen++;
+	if (new_parent)
 		xfs_trans_log_inode(tp, target_dp, XFS_ILOG_CORE);
-	}
 
 	/*
 	 * If this is a synchronous mount, make sure that the
diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
index e2f68de1615..edf12c7b834 100644
--- a/fs/xfs/xfs_rtalloc.c
+++ b/fs/xfs/xfs_rtalloc.c
@@ -85,7 +85,6 @@ xfs_growfs_rt_alloc(
 {
 	xfs_fileoff_t	bno;		/* block number in file */
 	xfs_buf_t	*bp;		/* temporary buffer for zeroing */
-	int		cancelflags;	/* flags for xfs_trans_cancel */
 	int		committed;	/* transaction committed flag */
 	xfs_daddr_t	d;		/* disk block address */
 	int		error;		/* error return value */
@@ -96,15 +95,16 @@ xfs_growfs_rt_alloc(
 	xfs_bmbt_irec_t	map;		/* block map output */
 	int		nmap;		/* number of block maps */
 	int		resblks;	/* space reservation */
-	xfs_trans_t	*tp;		/* transaction pointer */
 
 	/*
 	 * Allocate space to the file, as necessary.
 	 */
 	while (oblocks < nblocks) {
+		int		cancelflags = 0;
+		xfs_trans_t	*tp;
+
 		tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFSRT_ALLOC);
 		resblks = XFS_GROWFSRT_SPACE_RES(mp, nblocks - oblocks);
-		cancelflags = 0;
 		/*
 		 * Reserve space & log for one extent added to the file.
 		 */
@@ -171,7 +171,9 @@ xfs_growfs_rt_alloc(
 				mp->m_bsize, 0);
 			if (bp == NULL) {
 				error = XFS_ERROR(EIO);
-				goto error_cancel;
+error_cancel:
+				xfs_trans_cancel(tp, cancelflags);
+				goto error;
 			}
 			memset(XFS_BUF_PTR(bp), 0, mp->m_sb.sb_blocksize);
 			xfs_trans_log_buf(tp, bp, 0, mp->m_sb.sb_blocksize - 1);
@@ -188,8 +190,6 @@ xfs_growfs_rt_alloc(
 		oblocks = map.br_startoff + map.br_blockcount;
 	}
 	return 0;
-error_cancel:
-	xfs_trans_cancel(tp, cancelflags);
 error:
 	return error;
 }
@@ -1856,7 +1856,6 @@ xfs_growfs_rt(
 {
 	xfs_rtblock_t	bmbno;		/* bitmap block number */
 	xfs_buf_t	*bp;		/* temporary buffer */
-	int		cancelflags;	/* flags for xfs_trans_cancel */
 	int		error;		/* error return value */
 	xfs_inode_t	*ip;		/* bitmap inode, used as lock */
 	xfs_mount_t	*nmp;		/* new (fake) mount structure */
@@ -1872,13 +1871,13 @@ xfs_growfs_rt(
 	xfs_extlen_t	rsumblocks;	/* current number of rt summary blks */
 	xfs_sb_t	*sbp;		/* old superblock */
 	xfs_fsblock_t	sumbno;		/* summary block number */
-	xfs_trans_t	*tp;		/* transaction pointer */
 
 	sbp = &mp->m_sb;
-	cancelflags = 0;
 	/*
 	 * Initial error checking.
 	 */
+	if (!capable(CAP_SYS_ADMIN))
+		return XFS_ERROR(EPERM);
 	if (mp->m_rtdev_targp == NULL || mp->m_rbmip == NULL ||
 	    (nrblocks = in->newblocks) <= sbp->sb_rblocks ||
 	    (sbp->sb_rblocks && (in->extsize != sbp->sb_rextsize)))
@@ -1942,6 +1941,9 @@ xfs_growfs_rt(
 		     ((sbp->sb_rextents & ((1 << mp->m_blkbit_log) - 1)) != 0);
 	     bmbno < nrbmblocks;
 	     bmbno++) {
+		xfs_trans_t	*tp;
+		int		cancelflags = 0;
+
 		*nmp = *mp;
 		nsbp = &nmp->m_sb;
 		/*
@@ -1967,16 +1969,15 @@ xfs_growfs_rt(
 		 * Start a transaction, get the log reservation.
 		 */
 		tp = xfs_trans_alloc(mp, XFS_TRANS_GROWFSRT_FREE);
-		cancelflags = 0;
 		if ((error = xfs_trans_reserve(tp, 0,
 				XFS_GROWRTFREE_LOG_RES(nmp), 0, 0, 0)))
-			break;
+			goto error_cancel;
 		/*
 		 * Lock out other callers by grabbing the bitmap inode lock.
 		 */
 		if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0,
 						XFS_ILOCK_EXCL, &ip)))
-			break;
+			goto error_cancel;
 		ASSERT(ip == mp->m_rbmip);
 		/*
 		 * Update the bitmap inode's size.
@@ -1990,7 +1991,7 @@ xfs_growfs_rt(
 		 */
 		if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rsumino, 0,
 						XFS_ILOCK_EXCL, &ip)))
-			break;
+			goto error_cancel;
 		ASSERT(ip == mp->m_rsumip);
 		/*
 		 * Update the summary inode's size.
@@ -2005,7 +2006,7 @@ xfs_growfs_rt(
 		    mp->m_rsumlevels != nmp->m_rsumlevels) {
 			error = xfs_rtcopy_summary(mp, nmp, tp);
 			if (error)
-				break;
+				goto error_cancel;
 		}
 		/*
 		 * Update superblock fields.
@@ -2031,8 +2032,11 @@ xfs_growfs_rt(
 		bp = NULL;
 		error = xfs_rtfree_range(nmp, tp, sbp->sb_rextents,
 			nsbp->sb_rextents - sbp->sb_rextents, &bp, &sumbno);
-		if (error)
+		if (error) {
+error_cancel:
+			xfs_trans_cancel(tp, cancelflags);
 			break;
+		}
 		/*
 		 * Mark more blocks free in the superblock.
 		 */
@@ -2045,15 +2049,10 @@ xfs_growfs_rt(
 		mp->m_rsumsize = nrsumsize;
 
 		error = xfs_trans_commit(tp, 0);
-		if (error) {
-			tp = NULL;
+		if (error)
 			break;
-		}
 	}
 
-	if (error && tp)
-		xfs_trans_cancel(tp, cancelflags);
-
 	/*
 	 * Free the fake mp structure.
 	 */
diff --git a/fs/xfs/xfs_rw.c b/fs/xfs/xfs_rw.c
index 3a82576dde9..36f3a21c54d 100644
--- a/fs/xfs/xfs_rw.c
+++ b/fs/xfs/xfs_rw.c
@@ -406,7 +406,7 @@ xfs_bwrite(
 	 * XXXsup how does this work for quotas.
 	 */
 	XFS_BUF_SET_BDSTRAT_FUNC(bp, xfs_bdstrat_cb);
-	XFS_BUF_SET_FSPRIVATE3(bp, mp);
+	bp->b_mount = mp;
 	XFS_BUF_WRITE(bp);
 
 	if ((error = XFS_bwrite(bp))) {
diff --git a/fs/xfs/xfs_sb.h b/fs/xfs/xfs_sb.h
index 3f8cf1587f4..1ed71916e4c 100644
--- a/fs/xfs/xfs_sb.h
+++ b/fs/xfs/xfs_sb.h
@@ -79,6 +79,7 @@ struct xfs_mount;
 #define XFS_SB_VERSION2_LAZYSBCOUNTBIT	0x00000002	/* Superblk counters */
 #define XFS_SB_VERSION2_RESERVED4BIT	0x00000004
 #define XFS_SB_VERSION2_ATTR2BIT	0x00000008	/* Inline attr rework */
+#define XFS_SB_VERSION2_PARENTBIT	0x00000010	/* parent pointers */
 
 #define	XFS_SB_VERSION2_OKREALFBITS	\
 	(XFS_SB_VERSION2_LAZYSBCOUNTBIT	| \
@@ -296,30 +297,34 @@ typedef enum {
 
 #define	XFS_SB_VERSION_NUM(sbp)	((sbp)->sb_versionnum & XFS_SB_VERSION_NUMBITS)
 
-#ifdef __KERNEL__
 static inline int xfs_sb_good_version(xfs_sb_t *sbp)
 {
-	return (((sbp->sb_versionnum >= XFS_SB_VERSION_1) && \
-		  (sbp->sb_versionnum <= XFS_SB_VERSION_3)) || \
-		   ((XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		    !((sbp->sb_versionnum & ~XFS_SB_VERSION_OKREALBITS) || \
-		      ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) && \
-		       (sbp->sb_features2 & ~XFS_SB_VERSION2_OKREALBITS))) && \
-	  	    (sbp->sb_shared_vn <= XFS_SB_MAX_SHARED_VN)));
-}
+	/* We always support version 1-3 */
+	if (sbp->sb_versionnum >= XFS_SB_VERSION_1 &&
+	    sbp->sb_versionnum <= XFS_SB_VERSION_3)
+		return 1;
+
+	/* We support version 4 if all feature bits are supported */
+	if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) {
+		if ((sbp->sb_versionnum & ~XFS_SB_VERSION_OKREALBITS) ||
+		    ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) &&
+		     (sbp->sb_features2 & ~XFS_SB_VERSION2_OKREALBITS)))
+			return 0;
+
+#ifdef __KERNEL__
+		if (sbp->sb_shared_vn > XFS_SB_MAX_SHARED_VN)
+			return 0;
 #else
-static inline int xfs_sb_good_version(xfs_sb_t *sbp)
-{
-	return (((sbp->sb_versionnum >= XFS_SB_VERSION_1) && \
-		  (sbp->sb_versionnum <= XFS_SB_VERSION_3)) || \
-		   ((XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		    !((sbp->sb_versionnum & ~XFS_SB_VERSION_OKREALBITS) || \
-		      ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) && \
-		       (sbp->sb_features2 & ~XFS_SB_VERSION2_OKREALBITS))) && \
-		  (!(sbp->sb_versionnum & XFS_SB_VERSION_SHAREDBIT) || \
-		   (sbp->sb_shared_vn <= XFS_SB_MAX_SHARED_VN))));
+		if ((sbp->sb_versionnum & XFS_SB_VERSION_SHAREDBIT) &&
+		    sbp->sb_shared_vn > XFS_SB_MAX_SHARED_VN)
+			return 0;
+#endif
+
+		return 1;
+	}
+
+	return 0;
 }
-#endif /* __KERNEL__ */
 
 /*
  * Detect a mismatched features2 field.  Older kernels read/wrote
@@ -332,123 +337,127 @@ static inline int xfs_sb_has_mismatched_features2(xfs_sb_t *sbp)
 
 static inline unsigned xfs_sb_version_tonew(unsigned v)
 {
-	return ((((v) == XFS_SB_VERSION_1) ? \
-		0 : \
-		(((v) == XFS_SB_VERSION_2) ? \
-			XFS_SB_VERSION_ATTRBIT : \
-			(XFS_SB_VERSION_ATTRBIT | XFS_SB_VERSION_NLINKBIT))) | \
-		XFS_SB_VERSION_4);
+	if (v == XFS_SB_VERSION_1)
+		return XFS_SB_VERSION_4;
+
+	if (v == XFS_SB_VERSION_2)
+		return XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT;
+
+	return XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT |
+		XFS_SB_VERSION_NLINKBIT;
 }
 
 static inline unsigned xfs_sb_version_toold(unsigned v)
 {
-	return (((v) & (XFS_SB_VERSION_QUOTABIT | XFS_SB_VERSION_ALIGNBIT)) ? \
-		0 : \
-		(((v) & XFS_SB_VERSION_NLINKBIT) ? \
-			XFS_SB_VERSION_3 : \
-			(((v) & XFS_SB_VERSION_ATTRBIT) ?  \
-				XFS_SB_VERSION_2 : \
-				XFS_SB_VERSION_1)));
+	if (v & (XFS_SB_VERSION_QUOTABIT | XFS_SB_VERSION_ALIGNBIT))
+		return 0;
+	if (v & XFS_SB_VERSION_NLINKBIT)
+		return XFS_SB_VERSION_3;
+	if (v & XFS_SB_VERSION_ATTRBIT)
+		return XFS_SB_VERSION_2;
+	return XFS_SB_VERSION_1;
 }
 
 static inline int xfs_sb_version_hasattr(xfs_sb_t *sbp)
 {
-	return ((sbp)->sb_versionnum == XFS_SB_VERSION_2) || \
-		 ((sbp)->sb_versionnum == XFS_SB_VERSION_3) || \
-		 ((XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		  ((sbp)->sb_versionnum & XFS_SB_VERSION_ATTRBIT));
+	return sbp->sb_versionnum == XFS_SB_VERSION_2 ||
+		sbp->sb_versionnum == XFS_SB_VERSION_3 ||
+		(XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		 (sbp->sb_versionnum & XFS_SB_VERSION_ATTRBIT));
 }
 
 static inline void xfs_sb_version_addattr(xfs_sb_t *sbp)
 {
-	(sbp)->sb_versionnum = (((sbp)->sb_versionnum == XFS_SB_VERSION_1) ? \
-		XFS_SB_VERSION_2 : \
-		((XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) ? \
-			((sbp)->sb_versionnum | XFS_SB_VERSION_ATTRBIT) : \
-			(XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT)));
+	if (sbp->sb_versionnum == XFS_SB_VERSION_1)
+		sbp->sb_versionnum = XFS_SB_VERSION_2;
+	else if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
+		sbp->sb_versionnum |= XFS_SB_VERSION_ATTRBIT;
+	else
+		sbp->sb_versionnum = XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT;
 }
 
 static inline int xfs_sb_version_hasnlink(xfs_sb_t *sbp)
 {
-	return ((sbp)->sb_versionnum == XFS_SB_VERSION_3) || \
-		 ((XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		  ((sbp)->sb_versionnum & XFS_SB_VERSION_NLINKBIT));
+	return sbp->sb_versionnum == XFS_SB_VERSION_3 ||
+		 (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		  (sbp->sb_versionnum & XFS_SB_VERSION_NLINKBIT));
 }
 
 static inline void xfs_sb_version_addnlink(xfs_sb_t *sbp)
 {
-	(sbp)->sb_versionnum = ((sbp)->sb_versionnum <= XFS_SB_VERSION_2 ? \
-		XFS_SB_VERSION_3 : \
-		((sbp)->sb_versionnum | XFS_SB_VERSION_NLINKBIT));
+	if (sbp->sb_versionnum <= XFS_SB_VERSION_2)
+		sbp->sb_versionnum = XFS_SB_VERSION_3;
+	else
+		sbp->sb_versionnum |= XFS_SB_VERSION_NLINKBIT;
 }
 
 static inline int xfs_sb_version_hasquota(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_QUOTABIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_QUOTABIT);
 }
 
 static inline void xfs_sb_version_addquota(xfs_sb_t *sbp)
 {
-	(sbp)->sb_versionnum = \
-		 (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 ? \
-			((sbp)->sb_versionnum | XFS_SB_VERSION_QUOTABIT) : \
-			(xfs_sb_version_tonew((sbp)->sb_versionnum) | \
-			 XFS_SB_VERSION_QUOTABIT));
+	if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
+		sbp->sb_versionnum |= XFS_SB_VERSION_QUOTABIT;
+	else
+		sbp->sb_versionnum = xfs_sb_version_tonew(sbp->sb_versionnum) |
+					XFS_SB_VERSION_QUOTABIT;
 }
 
 static inline int xfs_sb_version_hasalign(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_ALIGNBIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_ALIGNBIT);
 }
 
 static inline int xfs_sb_version_hasdalign(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_DALIGNBIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_DALIGNBIT);
 }
 
 static inline int xfs_sb_version_hasshared(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_SHAREDBIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_SHAREDBIT);
 }
 
 static inline int xfs_sb_version_hasdirv2(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_DIRV2BIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_DIRV2BIT);
 }
 
 static inline int xfs_sb_version_haslogv2(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_LOGV2BIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_LOGV2BIT);
 }
 
 static inline int xfs_sb_version_hasextflgbit(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT);
 }
 
 static inline int xfs_sb_version_hassector(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_SECTORBIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_SECTORBIT);
 }
 
 static inline int xfs_sb_version_hasasciici(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
 		(sbp->sb_versionnum & XFS_SB_VERSION_BORGBIT);
 }
 
 static inline int xfs_sb_version_hasmorebits(xfs_sb_t *sbp)
 {
-	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4) && \
-		((sbp)->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT);
+	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+		(sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT);
 }
 
 /*
@@ -463,22 +472,20 @@ static inline int xfs_sb_version_hasmorebits(xfs_sb_t *sbp)
 
 static inline int xfs_sb_version_haslazysbcount(xfs_sb_t *sbp)
 {
-	return (xfs_sb_version_hasmorebits(sbp) &&	\
-		((sbp)->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT));
+	return xfs_sb_version_hasmorebits(sbp) &&
+		(sbp->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT);
 }
 
 static inline int xfs_sb_version_hasattr2(xfs_sb_t *sbp)
 {
-	return (xfs_sb_version_hasmorebits(sbp)) &&	\
-		((sbp)->sb_features2 & XFS_SB_VERSION2_ATTR2BIT);
+	return xfs_sb_version_hasmorebits(sbp) &&
+		(sbp->sb_features2 & XFS_SB_VERSION2_ATTR2BIT);
 }
 
 static inline void xfs_sb_version_addattr2(xfs_sb_t *sbp)
 {
-	((sbp)->sb_versionnum =	\
-		((sbp)->sb_versionnum | XFS_SB_VERSION_MOREBITSBIT),	\
-	((sbp)->sb_features2 =	\
-		((sbp)->sb_features2 | XFS_SB_VERSION2_ATTR2BIT)));
+	sbp->sb_versionnum |= XFS_SB_VERSION_MOREBITSBIT;
+	sbp->sb_features2 |= XFS_SB_VERSION2_ATTR2BIT;
 }
 
 static inline void xfs_sb_version_removeattr2(xfs_sb_t *sbp)
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 4e1c22a23be..8570b826fed 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -290,7 +290,7 @@ xfs_trans_dup(
 	ASSERT(tp->t_ticket != NULL);
 
 	ntp->t_flags = XFS_TRANS_PERM_LOG_RES | (tp->t_flags & XFS_TRANS_RESERVE);
-	ntp->t_ticket = tp->t_ticket;
+	ntp->t_ticket = xfs_log_ticket_get(tp->t_ticket);
 	ntp->t_blk_res = tp->t_blk_res - tp->t_blk_res_used;
 	tp->t_blk_res = tp->t_blk_res_used;
 	ntp->t_rtx_res = tp->t_rtx_res - tp->t_rtx_res_used;
@@ -1260,6 +1260,13 @@ xfs_trans_roll(
 	trans = *tpp;
 
 	/*
+	 * transaction commit worked ok so we can drop the extra ticket
+	 * reference that we gained in xfs_trans_dup()
+	 */
+	xfs_log_ticket_put(trans->t_ticket);
+
+
+	/*
 	 * Reserve space in the log for th next transaction.
 	 * This also pushes items in the "AIL", the list of logged items,
 	 * out to disk if they are taking up space at the tail of the log
@@ -1383,11 +1390,12 @@ xfs_trans_chunk_committed(
 	xfs_log_item_desc_t	*lidp;
 	xfs_log_item_t		*lip;
 	xfs_lsn_t		item_lsn;
-	struct xfs_mount	*mp;
 	int			i;
 
 	lidp = licp->lic_descs;
 	for (i = 0; i < licp->lic_unused; i++, lidp++) {
+		struct xfs_ail		*ailp;
+
 		if (xfs_lic_isfree(licp, i)) {
 			continue;
 		}
@@ -1424,19 +1432,19 @@ xfs_trans_chunk_committed(
 		 * This would cause the earlier transaction to fail
 		 * the test below.
 		 */
-		mp = lip->li_mountp;
-		spin_lock(&mp->m_ail_lock);
+		ailp = lip->li_ailp;
+		spin_lock(&ailp->xa_lock);
 		if (XFS_LSN_CMP(item_lsn, lip->li_lsn) > 0) {
 			/*
 			 * This will set the item's lsn to item_lsn
 			 * and update the position of the item in
 			 * the AIL.
 			 *
-			 * xfs_trans_update_ail() drops the AIL lock.
+			 * xfs_trans_ail_update() drops the AIL lock.
 			 */
-			xfs_trans_update_ail(mp, lip, item_lsn);
+			xfs_trans_ail_update(ailp, lip, item_lsn);
 		} else {
-			spin_unlock(&mp->m_ail_lock);
+			spin_unlock(&ailp->xa_lock);
 		}
 
 		/*
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index 74c80bd2b0e..d6fe4a88d79 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -18,6 +18,8 @@
 #ifndef	__XFS_TRANS_H__
 #define	__XFS_TRANS_H__
 
+struct xfs_log_item;
+
 /*
  * This is the structure written in the log at the head of
  * every transaction. It identifies the type and id of the
@@ -98,76 +100,6 @@ typedef struct xfs_trans_header {
 #define	XFS_TRANS_TYPE_MAX		41
 /* new transaction types need to be reflected in xfs_logprint(8) */
 
-
-#ifdef __KERNEL__
-struct xfs_buf;
-struct xfs_buftarg;
-struct xfs_efd_log_item;
-struct xfs_efi_log_item;
-struct xfs_inode;
-struct xfs_item_ops;
-struct xfs_log_iovec;
-struct xfs_log_item;
-struct xfs_log_item_desc;
-struct xfs_mount;
-struct xfs_trans;
-struct xfs_dquot_acct;
-
-typedef struct xfs_log_item {
-	struct list_head		li_ail;		/* AIL pointers */
-	xfs_lsn_t			li_lsn;		/* last on-disk lsn */
-	struct xfs_log_item_desc	*li_desc;	/* ptr to current desc*/
-	struct xfs_mount		*li_mountp;	/* ptr to fs mount */
-	uint				li_type;	/* item type */
-	uint				li_flags;	/* misc flags */
-	struct xfs_log_item		*li_bio_list;	/* buffer item list */
-	void				(*li_cb)(struct xfs_buf *,
-						 struct xfs_log_item *);
-							/* buffer item iodone */
-							/* callback func */
-	struct xfs_item_ops		*li_ops;	/* function list */
-} xfs_log_item_t;
-
-#define	XFS_LI_IN_AIL	0x1
-#define XFS_LI_ABORTED	0x2
-
-typedef struct xfs_item_ops {
-	uint (*iop_size)(xfs_log_item_t *);
-	void (*iop_format)(xfs_log_item_t *, struct xfs_log_iovec *);
-	void (*iop_pin)(xfs_log_item_t *);
-	void (*iop_unpin)(xfs_log_item_t *, int);
-	void (*iop_unpin_remove)(xfs_log_item_t *, struct xfs_trans *);
-	uint (*iop_trylock)(xfs_log_item_t *);
-	void (*iop_unlock)(xfs_log_item_t *);
-	xfs_lsn_t (*iop_committed)(xfs_log_item_t *, xfs_lsn_t);
-	void (*iop_push)(xfs_log_item_t *);
-	void (*iop_pushbuf)(xfs_log_item_t *);
-	void (*iop_committing)(xfs_log_item_t *, xfs_lsn_t);
-} xfs_item_ops_t;
-
-#define IOP_SIZE(ip)		(*(ip)->li_ops->iop_size)(ip)
-#define IOP_FORMAT(ip,vp)	(*(ip)->li_ops->iop_format)(ip, vp)
-#define IOP_PIN(ip)		(*(ip)->li_ops->iop_pin)(ip)
-#define IOP_UNPIN(ip, flags)	(*(ip)->li_ops->iop_unpin)(ip, flags)
-#define IOP_UNPIN_REMOVE(ip,tp) (*(ip)->li_ops->iop_unpin_remove)(ip, tp)
-#define IOP_TRYLOCK(ip)		(*(ip)->li_ops->iop_trylock)(ip)
-#define IOP_UNLOCK(ip)		(*(ip)->li_ops->iop_unlock)(ip)
-#define IOP_COMMITTED(ip, lsn)	(*(ip)->li_ops->iop_committed)(ip, lsn)
-#define IOP_PUSH(ip)		(*(ip)->li_ops->iop_push)(ip)
-#define IOP_PUSHBUF(ip)		(*(ip)->li_ops->iop_pushbuf)(ip)
-#define IOP_COMMITTING(ip, lsn) (*(ip)->li_ops->iop_committing)(ip, lsn)
-
-/*
- * Return values for the IOP_TRYLOCK() routines.
- */
-#define	XFS_ITEM_SUCCESS	0
-#define	XFS_ITEM_PINNED		1
-#define	XFS_ITEM_LOCKED		2
-#define	XFS_ITEM_FLUSHING	3
-#define XFS_ITEM_PUSHBUF	4
-
-#endif	/* __KERNEL__ */
-
 /*
  * This structure is used to track log items associated with
  * a transaction.  It points to the log item and keeps some
@@ -176,7 +108,7 @@ typedef struct xfs_item_ops {
  * once we get to commit processing (see xfs_trans_commit()).
  */
 typedef struct xfs_log_item_desc {
-	xfs_log_item_t	*lid_item;
+	struct xfs_log_item	*lid_item;
 	ushort		lid_size;
 	unsigned char	lid_flags;
 	unsigned char	lid_index;
@@ -276,94 +208,6 @@ xfs_lic_desc_to_chunk(xfs_log_item_desc_t *dp)
 		(xfs_caddr_t)(((xfs_log_item_chunk_t*)0)->lic_descs));
 }
 
-#ifdef __KERNEL__
-/*
- * This structure is used to maintain a list of block ranges that have been
- * freed in the transaction.  The ranges are listed in the perag[] busy list
- * between when they're freed and the transaction is committed to disk.
- */
-
-typedef struct xfs_log_busy_slot {
-	xfs_agnumber_t		lbc_ag;
-	ushort			lbc_idx;	/* index in perag.busy[] */
-} xfs_log_busy_slot_t;
-
-#define XFS_LBC_NUM_SLOTS	31
-typedef struct xfs_log_busy_chunk {
-	struct xfs_log_busy_chunk	*lbc_next;
-	uint				lbc_free;	/* free slots bitmask */
-	ushort				lbc_unused;	/* first unused */
-	xfs_log_busy_slot_t		lbc_busy[XFS_LBC_NUM_SLOTS];
-} xfs_log_busy_chunk_t;
-
-#define	XFS_LBC_MAX_SLOT	(XFS_LBC_NUM_SLOTS - 1)
-#define	XFS_LBC_FREEMASK	((1U << XFS_LBC_NUM_SLOTS) - 1)
-
-#define	XFS_LBC_INIT(cp)	((cp)->lbc_free = XFS_LBC_FREEMASK)
-#define	XFS_LBC_CLAIM(cp, slot)	((cp)->lbc_free &= ~(1 << (slot)))
-#define	XFS_LBC_SLOT(cp, slot)	(&((cp)->lbc_busy[(slot)]))
-#define	XFS_LBC_VACANCY(cp)	(((cp)->lbc_free) & XFS_LBC_FREEMASK)
-#define	XFS_LBC_ISFREE(cp, slot) ((cp)->lbc_free & (1 << (slot)))
-
-/*
- * This is the type of function which can be given to xfs_trans_callback()
- * to be called upon the transaction's commit to disk.
- */
-typedef void (*xfs_trans_callback_t)(struct xfs_trans *, void *);
-
-/*
- * This is the structure maintained for every active transaction.
- */
-typedef struct xfs_trans {
-	unsigned int		t_magic;	/* magic number */
-	xfs_log_callback_t	t_logcb;	/* log callback struct */
-	unsigned int		t_type;		/* transaction type */
-	unsigned int		t_log_res;	/* amt of log space resvd */
-	unsigned int		t_log_count;	/* count for perm log res */
-	unsigned int		t_blk_res;	/* # of blocks resvd */
-	unsigned int		t_blk_res_used;	/* # of resvd blocks used */
-	unsigned int		t_rtx_res;	/* # of rt extents resvd */
-	unsigned int		t_rtx_res_used;	/* # of resvd rt extents used */
-	xfs_log_ticket_t	t_ticket;	/* log mgr ticket */
-	xfs_lsn_t		t_lsn;		/* log seq num of start of
-						 * transaction. */
-	xfs_lsn_t		t_commit_lsn;	/* log seq num of end of
-						 * transaction. */
-	struct xfs_mount	*t_mountp;	/* ptr to fs mount struct */
-	struct xfs_dquot_acct   *t_dqinfo;	/* acctg info for dquots */
-	xfs_trans_callback_t	t_callback;	/* transaction callback */
-	void			*t_callarg;	/* callback arg */
-	unsigned int		t_flags;	/* misc flags */
-	int64_t			t_icount_delta;	/* superblock icount change */
-	int64_t			t_ifree_delta;	/* superblock ifree change */
-	int64_t			t_fdblocks_delta; /* superblock fdblocks chg */
-	int64_t			t_res_fdblocks_delta; /* on-disk only chg */
-	int64_t			t_frextents_delta;/* superblock freextents chg*/
-	int64_t			t_res_frextents_delta; /* on-disk only chg */
-#ifdef DEBUG
-	int64_t			t_ag_freeblks_delta; /* debugging counter */
-	int64_t			t_ag_flist_delta; /* debugging counter */
-	int64_t			t_ag_btree_delta; /* debugging counter */
-#endif
-	int64_t			t_dblocks_delta;/* superblock dblocks change */
-	int64_t			t_agcount_delta;/* superblock agcount change */
-	int64_t			t_imaxpct_delta;/* superblock imaxpct change */
-	int64_t			t_rextsize_delta;/* superblock rextsize chg */
-	int64_t			t_rbmblocks_delta;/* superblock rbmblocks chg */
-	int64_t			t_rblocks_delta;/* superblock rblocks change */
-	int64_t			t_rextents_delta;/* superblocks rextents chg */
-	int64_t			t_rextslog_delta;/* superblocks rextslog chg */
-	unsigned int		t_items_free;	/* log item descs free */
-	xfs_log_item_chunk_t	t_items;	/* first log item desc chunk */
-	xfs_trans_header_t	t_header;	/* header for in-log trans */
-	unsigned int		t_busy_free;	/* busy descs free */
-	xfs_log_busy_chunk_t	t_busy;		/* busy/async free blocks */
-	unsigned long		t_pflags;	/* saved process flags state */
-} xfs_trans_t;
-
-#endif	/* __KERNEL__ */
-
-
 #define	XFS_TRANS_MAGIC		0x5452414E	/* 'TRAN' */
 /*
  * Values for t_flags.
@@ -906,6 +750,157 @@ typedef struct xfs_trans {
 #define	XFS_DQUOT_REF		1
 
 #ifdef __KERNEL__
+
+struct xfs_buf;
+struct xfs_buftarg;
+struct xfs_efd_log_item;
+struct xfs_efi_log_item;
+struct xfs_inode;
+struct xfs_item_ops;
+struct xfs_log_iovec;
+struct xfs_log_item_desc;
+struct xfs_mount;
+struct xfs_trans;
+struct xfs_dquot_acct;
+
+typedef struct xfs_log_item {
+	struct list_head		li_ail;		/* AIL pointers */
+	xfs_lsn_t			li_lsn;		/* last on-disk lsn */
+	struct xfs_log_item_desc	*li_desc;	/* ptr to current desc*/
+	struct xfs_mount		*li_mountp;	/* ptr to fs mount */
+	struct xfs_ail			*li_ailp;	/* ptr to AIL */
+	uint				li_type;	/* item type */
+	uint				li_flags;	/* misc flags */
+	struct xfs_log_item		*li_bio_list;	/* buffer item list */
+	void				(*li_cb)(struct xfs_buf *,
+						 struct xfs_log_item *);
+							/* buffer item iodone */
+							/* callback func */
+	struct xfs_item_ops		*li_ops;	/* function list */
+} xfs_log_item_t;
+
+#define	XFS_LI_IN_AIL	0x1
+#define XFS_LI_ABORTED	0x2
+
+typedef struct xfs_item_ops {
+	uint (*iop_size)(xfs_log_item_t *);
+	void (*iop_format)(xfs_log_item_t *, struct xfs_log_iovec *);
+	void (*iop_pin)(xfs_log_item_t *);
+	void (*iop_unpin)(xfs_log_item_t *, int);
+	void (*iop_unpin_remove)(xfs_log_item_t *, struct xfs_trans *);
+	uint (*iop_trylock)(xfs_log_item_t *);
+	void (*iop_unlock)(xfs_log_item_t *);
+	xfs_lsn_t (*iop_committed)(xfs_log_item_t *, xfs_lsn_t);
+	void (*iop_push)(xfs_log_item_t *);
+	void (*iop_pushbuf)(xfs_log_item_t *);
+	void (*iop_committing)(xfs_log_item_t *, xfs_lsn_t);
+} xfs_item_ops_t;
+
+#define IOP_SIZE(ip)		(*(ip)->li_ops->iop_size)(ip)
+#define IOP_FORMAT(ip,vp)	(*(ip)->li_ops->iop_format)(ip, vp)
+#define IOP_PIN(ip)		(*(ip)->li_ops->iop_pin)(ip)
+#define IOP_UNPIN(ip, flags)	(*(ip)->li_ops->iop_unpin)(ip, flags)
+#define IOP_UNPIN_REMOVE(ip,tp) (*(ip)->li_ops->iop_unpin_remove)(ip, tp)
+#define IOP_TRYLOCK(ip)		(*(ip)->li_ops->iop_trylock)(ip)
+#define IOP_UNLOCK(ip)		(*(ip)->li_ops->iop_unlock)(ip)
+#define IOP_COMMITTED(ip, lsn)	(*(ip)->li_ops->iop_committed)(ip, lsn)
+#define IOP_PUSH(ip)		(*(ip)->li_ops->iop_push)(ip)
+#define IOP_PUSHBUF(ip)		(*(ip)->li_ops->iop_pushbuf)(ip)
+#define IOP_COMMITTING(ip, lsn) (*(ip)->li_ops->iop_committing)(ip, lsn)
+
+/*
+ * Return values for the IOP_TRYLOCK() routines.
+ */
+#define	XFS_ITEM_SUCCESS	0
+#define	XFS_ITEM_PINNED		1
+#define	XFS_ITEM_LOCKED		2
+#define	XFS_ITEM_FLUSHING	3
+#define XFS_ITEM_PUSHBUF	4
+
+/*
+ * This structure is used to maintain a list of block ranges that have been
+ * freed in the transaction.  The ranges are listed in the perag[] busy list
+ * between when they're freed and the transaction is committed to disk.
+ */
+
+typedef struct xfs_log_busy_slot {
+	xfs_agnumber_t		lbc_ag;
+	ushort			lbc_idx;	/* index in perag.busy[] */
+} xfs_log_busy_slot_t;
+
+#define XFS_LBC_NUM_SLOTS	31
+typedef struct xfs_log_busy_chunk {
+	struct xfs_log_busy_chunk	*lbc_next;
+	uint				lbc_free;	/* free slots bitmask */
+	ushort				lbc_unused;	/* first unused */
+	xfs_log_busy_slot_t		lbc_busy[XFS_LBC_NUM_SLOTS];
+} xfs_log_busy_chunk_t;
+
+#define	XFS_LBC_MAX_SLOT	(XFS_LBC_NUM_SLOTS - 1)
+#define	XFS_LBC_FREEMASK	((1U << XFS_LBC_NUM_SLOTS) - 1)
+
+#define	XFS_LBC_INIT(cp)	((cp)->lbc_free = XFS_LBC_FREEMASK)
+#define	XFS_LBC_CLAIM(cp, slot)	((cp)->lbc_free &= ~(1 << (slot)))
+#define	XFS_LBC_SLOT(cp, slot)	(&((cp)->lbc_busy[(slot)]))
+#define	XFS_LBC_VACANCY(cp)	(((cp)->lbc_free) & XFS_LBC_FREEMASK)
+#define	XFS_LBC_ISFREE(cp, slot) ((cp)->lbc_free & (1 << (slot)))
+
+/*
+ * This is the type of function which can be given to xfs_trans_callback()
+ * to be called upon the transaction's commit to disk.
+ */
+typedef void (*xfs_trans_callback_t)(struct xfs_trans *, void *);
+
+/*
+ * This is the structure maintained for every active transaction.
+ */
+typedef struct xfs_trans {
+	unsigned int		t_magic;	/* magic number */
+	xfs_log_callback_t	t_logcb;	/* log callback struct */
+	unsigned int		t_type;		/* transaction type */
+	unsigned int		t_log_res;	/* amt of log space resvd */
+	unsigned int		t_log_count;	/* count for perm log res */
+	unsigned int		t_blk_res;	/* # of blocks resvd */
+	unsigned int		t_blk_res_used;	/* # of resvd blocks used */
+	unsigned int		t_rtx_res;	/* # of rt extents resvd */
+	unsigned int		t_rtx_res_used;	/* # of resvd rt extents used */
+	xfs_log_ticket_t	t_ticket;	/* log mgr ticket */
+	xfs_lsn_t		t_lsn;		/* log seq num of start of
+						 * transaction. */
+	xfs_lsn_t		t_commit_lsn;	/* log seq num of end of
+						 * transaction. */
+	struct xfs_mount	*t_mountp;	/* ptr to fs mount struct */
+	struct xfs_dquot_acct   *t_dqinfo;	/* acctg info for dquots */
+	xfs_trans_callback_t	t_callback;	/* transaction callback */
+	void			*t_callarg;	/* callback arg */
+	unsigned int		t_flags;	/* misc flags */
+	int64_t			t_icount_delta;	/* superblock icount change */
+	int64_t			t_ifree_delta;	/* superblock ifree change */
+	int64_t			t_fdblocks_delta; /* superblock fdblocks chg */
+	int64_t			t_res_fdblocks_delta; /* on-disk only chg */
+	int64_t			t_frextents_delta;/* superblock freextents chg*/
+	int64_t			t_res_frextents_delta; /* on-disk only chg */
+#ifdef DEBUG
+	int64_t			t_ag_freeblks_delta; /* debugging counter */
+	int64_t			t_ag_flist_delta; /* debugging counter */
+	int64_t			t_ag_btree_delta; /* debugging counter */
+#endif
+	int64_t			t_dblocks_delta;/* superblock dblocks change */
+	int64_t			t_agcount_delta;/* superblock agcount change */
+	int64_t			t_imaxpct_delta;/* superblock imaxpct change */
+	int64_t			t_rextsize_delta;/* superblock rextsize chg */
+	int64_t			t_rbmblocks_delta;/* superblock rbmblocks chg */
+	int64_t			t_rblocks_delta;/* superblock rblocks change */
+	int64_t			t_rextents_delta;/* superblocks rextents chg */
+	int64_t			t_rextslog_delta;/* superblocks rextslog chg */
+	unsigned int		t_items_free;	/* log item descs free */
+	xfs_log_item_chunk_t	t_items;	/* first log item desc chunk */
+	xfs_trans_header_t	t_header;	/* header for in-log trans */
+	unsigned int		t_busy_free;	/* busy descs free */
+	xfs_log_busy_chunk_t	t_busy;		/* busy/async free blocks */
+	unsigned long		t_pflags;	/* saved process flags state */
+} xfs_trans_t;
+
 /*
  * XFS transaction mechanism exported interfaces that are
  * actually macros.
@@ -928,7 +923,6 @@ typedef struct xfs_trans {
 /*
  * XFS transaction mechanism exported interfaces.
  */
-void		xfs_trans_init(struct xfs_mount *);
 xfs_trans_t	*xfs_trans_alloc(struct xfs_mount *, uint);
 xfs_trans_t	*_xfs_trans_alloc(struct xfs_mount *, uint);
 xfs_trans_t	*xfs_trans_dup(xfs_trans_t *);
@@ -975,13 +969,8 @@ int		_xfs_trans_commit(xfs_trans_t *,
 				  int *);
 #define xfs_trans_commit(tp, flags)	_xfs_trans_commit(tp, flags, NULL)
 void		xfs_trans_cancel(xfs_trans_t *, int);
-int		xfs_trans_roll(struct xfs_trans **, struct xfs_inode *);
 int		xfs_trans_ail_init(struct xfs_mount *);
 void		xfs_trans_ail_destroy(struct xfs_mount *);
-void		xfs_trans_push_ail(struct xfs_mount *, xfs_lsn_t);
-xfs_lsn_t	xfs_trans_tail_ail(struct xfs_mount *);
-void		xfs_trans_unlocked_item(struct xfs_mount *,
-					xfs_log_item_t *);
 xfs_log_busy_slot_t *xfs_trans_add_busy(xfs_trans_t *tp,
 					xfs_agnumber_t ag,
 					xfs_extlen_t idx);
@@ -990,4 +979,7 @@ extern kmem_zone_t	*xfs_trans_zone;
 
 #endif	/* __KERNEL__ */
 
+void		xfs_trans_init(struct xfs_mount *);
+int		xfs_trans_roll(struct xfs_trans **, struct xfs_inode *);
+
 #endif	/* __XFS_TRANS_H__ */
diff --git a/fs/xfs/xfs_trans_ail.c b/fs/xfs/xfs_trans_ail.c
index 1f77c00af56..2d47f10f8be 100644
--- a/fs/xfs/xfs_trans_ail.c
+++ b/fs/xfs/xfs_trans_ail.c
@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2008 Dave Chinner
  * All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or
@@ -28,13 +29,13 @@
 #include "xfs_trans_priv.h"
 #include "xfs_error.h"
 
-STATIC void xfs_ail_insert(xfs_ail_t *, xfs_log_item_t *);
-STATIC xfs_log_item_t * xfs_ail_delete(xfs_ail_t *, xfs_log_item_t *);
-STATIC xfs_log_item_t * xfs_ail_min(xfs_ail_t *);
-STATIC xfs_log_item_t * xfs_ail_next(xfs_ail_t *, xfs_log_item_t *);
+STATIC void xfs_ail_insert(struct xfs_ail *, xfs_log_item_t *);
+STATIC xfs_log_item_t * xfs_ail_delete(struct xfs_ail *, xfs_log_item_t *);
+STATIC xfs_log_item_t * xfs_ail_min(struct xfs_ail *);
+STATIC xfs_log_item_t * xfs_ail_next(struct xfs_ail *, xfs_log_item_t *);
 
 #ifdef DEBUG
-STATIC void xfs_ail_check(xfs_ail_t *, xfs_log_item_t *);
+STATIC void xfs_ail_check(struct xfs_ail *, xfs_log_item_t *);
 #else
 #define	xfs_ail_check(a,l)
 #endif /* DEBUG */
@@ -50,20 +51,20 @@ STATIC void xfs_ail_check(xfs_ail_t *, xfs_log_item_t *);
  * lsn of the last item in the AIL.
  */
 xfs_lsn_t
-xfs_trans_tail_ail(
-	xfs_mount_t	*mp)
+xfs_trans_ail_tail(
+	struct xfs_ail	*ailp)
 {
 	xfs_lsn_t	lsn;
 	xfs_log_item_t	*lip;
 
-	spin_lock(&mp->m_ail_lock);
-	lip = xfs_ail_min(&mp->m_ail);
+	spin_lock(&ailp->xa_lock);
+	lip = xfs_ail_min(ailp);
 	if (lip == NULL) {
 		lsn = (xfs_lsn_t)0;
 	} else {
 		lsn = lip->li_lsn;
 	}
-	spin_unlock(&mp->m_ail_lock);
+	spin_unlock(&ailp->xa_lock);
 
 	return lsn;
 }
@@ -85,16 +86,125 @@ xfs_trans_tail_ail(
  * any of the objects, so the lock is not needed.
  */
 void
-xfs_trans_push_ail(
-	xfs_mount_t		*mp,
-	xfs_lsn_t		threshold_lsn)
+xfs_trans_ail_push(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	threshold_lsn)
 {
-	xfs_log_item_t		*lip;
+	xfs_log_item_t	*lip;
+
+	lip = xfs_ail_min(ailp);
+	if (lip && !XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
+		if (XFS_LSN_CMP(threshold_lsn, ailp->xa_target) > 0)
+			xfsaild_wakeup(ailp, threshold_lsn);
+	}
+}
+
+/*
+ * AIL traversal cursor initialisation.
+ *
+ * The cursor keeps track of where our current traversal is up
+ * to by tracking the next ƣtem in the list for us. However, for
+ * this to be safe, removing an object from the AIL needs to invalidate
+ * any cursor that points to it. hence the traversal cursor needs to
+ * be linked to the struct xfs_ail so that deletion can search all the
+ * active cursors for invalidation.
+ *
+ * We don't link the push cursor because it is embedded in the struct
+ * xfs_ail and hence easily findable.
+ */
+STATIC void
+xfs_trans_ail_cursor_init(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur)
+{
+	cur->item = NULL;
+	if (cur == &ailp->xa_cursors)
+		return;
+
+	cur->next = ailp->xa_cursors.next;
+	ailp->xa_cursors.next = cur;
+}
+
+/*
+ * Set the cursor to the next item, because when we look
+ * up the cursor the current item may have been freed.
+ */
+STATIC void
+xfs_trans_ail_cursor_set(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	struct xfs_log_item	*lip)
+{
+	if (lip)
+		cur->item = xfs_ail_next(ailp, lip);
+}
+
+/*
+ * Get the next item in the traversal and advance the cursor.
+ * If the cursor was invalidated (inidicated by a lip of 1),
+ * restart the traversal.
+ */
+struct xfs_log_item *
+xfs_trans_ail_cursor_next(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur)
+{
+	struct xfs_log_item	*lip = cur->item;
+
+	if ((__psint_t)lip & 1)
+		lip = xfs_ail_min(ailp);
+	xfs_trans_ail_cursor_set(ailp, cur, lip);
+	return lip;
+}
+
+/*
+ * Now that the traversal is complete, we need to remove the cursor
+ * from the list of traversing cursors. Avoid removing the embedded
+ * push cursor, but use the fact it is alway present to make the
+ * list deletion simple.
+ */
+void
+xfs_trans_ail_cursor_done(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*done)
+{
+	struct xfs_ail_cursor	*prev = NULL;
+	struct xfs_ail_cursor	*cur;
+
+	done->item = NULL;
+	if (done == &ailp->xa_cursors)
+		return;
+	prev = &ailp->xa_cursors;
+	for (cur = prev->next; cur; prev = cur, cur = prev->next) {
+		if (cur == done) {
+			prev->next = cur->next;
+			break;
+		}
+	}
+	ASSERT(cur);
+}
+
+/*
+ * Invalidate any cursor that is pointing to this item. This is
+ * called when an item is removed from the AIL. Any cursor pointing
+ * to this object is now invalid and the traversal needs to be
+ * terminated so it doesn't reference a freed object. We set the
+ * cursor item to a value of 1 so we can distinguish between an
+ * invalidation and the end of the list when getting the next item
+ * from the cursor.
+ */
+STATIC void
+xfs_trans_ail_cursor_clear(
+	struct xfs_ail		*ailp,
+	struct xfs_log_item	*lip)
+{
+	struct xfs_ail_cursor	*cur;
 
-	lip = xfs_ail_min(&mp->m_ail);
-	if (lip && !XFS_FORCED_SHUTDOWN(mp)) {
-		if (XFS_LSN_CMP(threshold_lsn, mp->m_ail.xa_target) > 0)
-			xfsaild_wakeup(mp, threshold_lsn);
+	/* need to search all cursors */
+	for (cur = &ailp->xa_cursors; cur; cur = cur->next) {
+		if (cur->item == lip)
+			cur->item = (struct xfs_log_item *)
+					((__psint_t)cur->item | 1);
 	}
 }
 
@@ -103,25 +213,27 @@ xfs_trans_push_ail(
  * Return the current tree generation number for use
  * in calls to xfs_trans_next_ail().
  */
-STATIC xfs_log_item_t *
-xfs_trans_first_push_ail(
-	xfs_mount_t	*mp,
-	int		*gen,
-	xfs_lsn_t	lsn)
+xfs_log_item_t *
+xfs_trans_ail_cursor_first(
+	struct xfs_ail		*ailp,
+	struct xfs_ail_cursor	*cur,
+	xfs_lsn_t		lsn)
 {
-	xfs_log_item_t	*lip;
+	xfs_log_item_t		*lip;
 
-	lip = xfs_ail_min(&mp->m_ail);
-	*gen = (int)mp->m_ail.xa_gen;
+	xfs_trans_ail_cursor_init(ailp, cur);
+	lip = xfs_ail_min(ailp);
 	if (lsn == 0)
-		return lip;
+		goto out;
 
-	list_for_each_entry(lip, &mp->m_ail.xa_ail, li_ail) {
+	list_for_each_entry(lip, &ailp->xa_ail, li_ail) {
 		if (XFS_LSN_CMP(lip->li_lsn, lsn) >= 0)
-			return lip;
+			goto out;
 	}
-
-	return NULL;
+	lip = NULL;
+out:
+	xfs_trans_ail_cursor_set(ailp, cur, lip);
+	return lip;
 }
 
 /*
@@ -129,29 +241,29 @@ xfs_trans_first_push_ail(
  */
 long
 xfsaild_push(
-	xfs_mount_t	*mp,
+	struct xfs_ail	*ailp,
 	xfs_lsn_t	*last_lsn)
 {
 	long		tout = 1000; /* milliseconds */
 	xfs_lsn_t	last_pushed_lsn = *last_lsn;
-	xfs_lsn_t	target =  mp->m_ail.xa_target;
+	xfs_lsn_t	target =  ailp->xa_target;
 	xfs_lsn_t	lsn;
 	xfs_log_item_t	*lip;
-	int		gen;
-	int		restarts;
 	int		flush_log, count, stuck;
+	xfs_mount_t	*mp = ailp->xa_mount;
+	struct xfs_ail_cursor	*cur = &ailp->xa_cursors;
 
-#define	XFS_TRANS_PUSH_AIL_RESTARTS	10
-
-	spin_lock(&mp->m_ail_lock);
-	lip = xfs_trans_first_push_ail(mp, &gen, *last_lsn);
+	spin_lock(&ailp->xa_lock);
+	xfs_trans_ail_cursor_init(ailp, cur);
+	lip = xfs_trans_ail_cursor_first(ailp, cur, *last_lsn);
 	if (!lip || XFS_FORCED_SHUTDOWN(mp)) {
 		/*
 		 * AIL is empty or our push has reached the end.
 		 */
-		spin_unlock(&mp->m_ail_lock);
+		xfs_trans_ail_cursor_done(ailp, cur);
+		spin_unlock(&ailp->xa_lock);
 		last_pushed_lsn = 0;
-		goto out;
+		return tout;
 	}
 
 	XFS_STATS_INC(xs_push_ail);
@@ -169,7 +281,7 @@ xfsaild_push(
 	 */
 	tout = 10;
 	lsn = lip->li_lsn;
-	flush_log = stuck = count = restarts = 0;
+	flush_log = stuck = count = 0;
 	while ((XFS_LSN_CMP(lip->li_lsn, target) < 0)) {
 		int	lock_result;
 		/*
@@ -184,7 +296,7 @@ xfsaild_push(
 		 * skip to the next item in the list.
 		 */
 		lock_result = IOP_TRYLOCK(lip);
-		spin_unlock(&mp->m_ail_lock);
+		spin_unlock(&ailp->xa_lock);
 		switch (lock_result) {
 		case XFS_ITEM_SUCCESS:
 			XFS_STATS_INC(xs_push_ail_success);
@@ -221,7 +333,7 @@ xfsaild_push(
 			break;
 		}
 
-		spin_lock(&mp->m_ail_lock);
+		spin_lock(&ailp->xa_lock);
 		/* should we bother continuing? */
 		if (XFS_FORCED_SHUTDOWN(mp))
 			break;
@@ -244,14 +356,13 @@ xfsaild_push(
 		if (stuck > 100)
 			break;
 
-		lip = xfs_trans_next_ail(mp, lip, &gen, &restarts);
+		lip = xfs_trans_ail_cursor_next(ailp, cur);
 		if (lip == NULL)
 			break;
-		if (restarts > XFS_TRANS_PUSH_AIL_RESTARTS)
-			break;
 		lsn = lip->li_lsn;
 	}
-	spin_unlock(&mp->m_ail_lock);
+	xfs_trans_ail_cursor_done(ailp, cur);
+	spin_unlock(&ailp->xa_lock);
 
 	if (flush_log) {
 		/*
@@ -274,8 +385,7 @@ xfsaild_push(
 		 */
 		tout += 20;
 		last_pushed_lsn = 0;
-	} else if ((restarts > XFS_TRANS_PUSH_AIL_RESTARTS) ||
-		   ((stuck * 100) / count > 90)) {
+	} else if ((stuck * 100) / count > 90) {
 		/*
 		 * Either there is a lot of contention on the AIL or we
 		 * are stuck due to operations in progress. "Stuck" in this
@@ -287,7 +397,6 @@ xfsaild_push(
 		 */
 		tout += 10;
 	}
-out:
 	*last_lsn = last_pushed_lsn;
 	return tout;
 }	/* xfsaild_push */
@@ -303,7 +412,7 @@ out:
  */
 void
 xfs_trans_unlocked_item(
-	xfs_mount_t	*mp,
+	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
 {
 	xfs_log_item_t	*min_lip;
@@ -315,7 +424,7 @@ xfs_trans_unlocked_item(
 	 * over some potentially valid data.
 	 */
 	if (!(lip->li_flags & XFS_LI_IN_AIL) ||
-	    XFS_FORCED_SHUTDOWN(mp)) {
+	    XFS_FORCED_SHUTDOWN(ailp->xa_mount)) {
 		return;
 	}
 
@@ -331,10 +440,10 @@ xfs_trans_unlocked_item(
 	 * the call to xfs_log_move_tail() doesn't do anything if there's
 	 * not enough free space to wake people up so we're safe calling it.
 	 */
-	min_lip = xfs_ail_min(&mp->m_ail);
+	min_lip = xfs_ail_min(ailp);
 
 	if (min_lip == lip)
-		xfs_log_move_tail(mp, 1);
+		xfs_log_move_tail(ailp->xa_mount, 1);
 }	/* xfs_trans_unlocked_item */
 
 
@@ -347,41 +456,37 @@ xfs_trans_unlocked_item(
  * we move in the AIL is the minimum one, update the tail lsn in the
  * log manager.
  *
- * Increment the AIL's generation count to indicate that the tree
- * has changed.
- *
  * This function must be called with the AIL lock held.  The lock
  * is dropped before returning.
  */
 void
-xfs_trans_update_ail(
-	xfs_mount_t	*mp,
+xfs_trans_ail_update(
+	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip,
-	xfs_lsn_t	lsn) __releases(mp->m_ail_lock)
+	xfs_lsn_t	lsn) __releases(ailp->xa_lock)
 {
-	xfs_log_item_t		*dlip=NULL;
+	xfs_log_item_t		*dlip = NULL;
 	xfs_log_item_t		*mlip;	/* ptr to minimum lip */
 
-	mlip = xfs_ail_min(&mp->m_ail);
+	mlip = xfs_ail_min(ailp);
 
 	if (lip->li_flags & XFS_LI_IN_AIL) {
-		dlip = xfs_ail_delete(&mp->m_ail, lip);
+		dlip = xfs_ail_delete(ailp, lip);
 		ASSERT(dlip == lip);
+		xfs_trans_ail_cursor_clear(ailp, dlip);
 	} else {
 		lip->li_flags |= XFS_LI_IN_AIL;
 	}
 
 	lip->li_lsn = lsn;
-
-	xfs_ail_insert(&mp->m_ail, lip);
-	mp->m_ail.xa_gen++;
+	xfs_ail_insert(ailp, lip);
 
 	if (mlip == dlip) {
-		mlip = xfs_ail_min(&mp->m_ail);
-		spin_unlock(&mp->m_ail_lock);
-		xfs_log_move_tail(mp, mlip->li_lsn);
+		mlip = xfs_ail_min(ailp);
+		spin_unlock(&ailp->xa_lock);
+		xfs_log_move_tail(ailp->xa_mount, mlip->li_lsn);
 	} else {
-		spin_unlock(&mp->m_ail_lock);
+		spin_unlock(&ailp->xa_lock);
 	}
 
 
@@ -403,29 +508,30 @@ xfs_trans_update_ail(
  * is dropped before returning.
  */
 void
-xfs_trans_delete_ail(
-	xfs_mount_t	*mp,
-	xfs_log_item_t	*lip) __releases(mp->m_ail_lock)
+xfs_trans_ail_delete(
+	struct xfs_ail	*ailp,
+	xfs_log_item_t	*lip) __releases(ailp->xa_lock)
 {
 	xfs_log_item_t		*dlip;
 	xfs_log_item_t		*mlip;
 
 	if (lip->li_flags & XFS_LI_IN_AIL) {
-		mlip = xfs_ail_min(&mp->m_ail);
-		dlip = xfs_ail_delete(&mp->m_ail, lip);
+		mlip = xfs_ail_min(ailp);
+		dlip = xfs_ail_delete(ailp, lip);
 		ASSERT(dlip == lip);
+		xfs_trans_ail_cursor_clear(ailp, dlip);
 
 
 		lip->li_flags &= ~XFS_LI_IN_AIL;
 		lip->li_lsn = 0;
-		mp->m_ail.xa_gen++;
 
 		if (mlip == dlip) {
-			mlip = xfs_ail_min(&mp->m_ail);
-			spin_unlock(&mp->m_ail_lock);
-			xfs_log_move_tail(mp, (mlip ? mlip->li_lsn : 0));
+			mlip = xfs_ail_min(ailp);
+			spin_unlock(&ailp->xa_lock);
+			xfs_log_move_tail(ailp->xa_mount,
+						(mlip ? mlip->li_lsn : 0));
 		} else {
-			spin_unlock(&mp->m_ail_lock);
+			spin_unlock(&ailp->xa_lock);
 		}
 	}
 	else {
@@ -433,13 +539,13 @@ xfs_trans_delete_ail(
 		 * If the file system is not being shutdown, we are in
 		 * serious trouble if we get to this stage.
 		 */
-		if (XFS_FORCED_SHUTDOWN(mp))
-			spin_unlock(&mp->m_ail_lock);
-		else {
+		struct xfs_mount	*mp = ailp->xa_mount;
+
+		spin_unlock(&ailp->xa_lock);
+		if (!XFS_FORCED_SHUTDOWN(mp)) {
 			xfs_cmn_err(XFS_PTAG_AILDELETE, CE_ALERT, mp,
 		"%s: attempting to delete a log item that is not in the AIL",
 					__func__);
-			spin_unlock(&mp->m_ail_lock);
 			xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
 		}
 	}
@@ -448,56 +554,6 @@ xfs_trans_delete_ail(
 
 
 /*
- * Return the item in the AIL with the smallest lsn.
- * Return the current tree generation number for use
- * in calls to xfs_trans_next_ail().
- */
-xfs_log_item_t *
-xfs_trans_first_ail(
-	xfs_mount_t	*mp,
-	int		*gen)
-{
-	xfs_log_item_t	*lip;
-
-	lip = xfs_ail_min(&mp->m_ail);
-	*gen = (int)mp->m_ail.xa_gen;
-
-	return lip;
-}
-
-/*
- * If the generation count of the tree has not changed since the
- * caller last took something from the AIL, then return the elmt
- * in the tree which follows the one given.  If the count has changed,
- * then return the minimum elmt of the AIL and bump the restarts counter
- * if one is given.
- */
-xfs_log_item_t *
-xfs_trans_next_ail(
-	xfs_mount_t	*mp,
-	xfs_log_item_t	*lip,
-	int		*gen,
-	int		*restarts)
-{
-	xfs_log_item_t	*nlip;
-
-	ASSERT(mp && lip && gen);
-	if (mp->m_ail.xa_gen == *gen) {
-		nlip = xfs_ail_next(&mp->m_ail, lip);
-	} else {
-		nlip = xfs_ail_min(&mp->m_ail);
-		*gen = (int)mp->m_ail.xa_gen;
-		if (restarts != NULL) {
-			XFS_STATS_INC(xs_push_ail_restarts);
-			(*restarts)++;
-		}
-	}
-
-	return (nlip);
-}
-
-
-/*
  * The active item list (AIL) is a doubly linked list of log
  * items sorted by ascending lsn.  The base of the list is
  * a forw/back pointer pair embedded in the xfs mount structure.
@@ -515,15 +571,35 @@ int
 xfs_trans_ail_init(
 	xfs_mount_t	*mp)
 {
-	INIT_LIST_HEAD(&mp->m_ail.xa_ail);
-	return xfsaild_start(mp);
+	struct xfs_ail	*ailp;
+	int		error;
+
+	ailp = kmem_zalloc(sizeof(struct xfs_ail), KM_MAYFAIL);
+	if (!ailp)
+		return ENOMEM;
+
+	ailp->xa_mount = mp;
+	INIT_LIST_HEAD(&ailp->xa_ail);
+	spin_lock_init(&ailp->xa_lock);
+	error = xfsaild_start(ailp);
+	if (error)
+		goto out_free_ailp;
+	mp->m_ail = ailp;
+	return 0;
+
+out_free_ailp:
+	kmem_free(ailp);
+	return error;
 }
 
 void
 xfs_trans_ail_destroy(
 	xfs_mount_t	*mp)
 {
-	xfsaild_stop(mp);
+	struct xfs_ail	*ailp = mp->m_ail;
+
+	xfsaild_stop(ailp);
+	kmem_free(ailp);
 }
 
 /*
@@ -534,7 +610,7 @@ xfs_trans_ail_destroy(
  */
 STATIC void
 xfs_ail_insert(
-	xfs_ail_t	*ailp,
+	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
 /* ARGSUSED */
 {
@@ -568,7 +644,7 @@ xfs_ail_insert(
 /*ARGSUSED*/
 STATIC xfs_log_item_t *
 xfs_ail_delete(
-	xfs_ail_t	*ailp,
+	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
 /* ARGSUSED */
 {
@@ -585,7 +661,7 @@ xfs_ail_delete(
  */
 STATIC xfs_log_item_t *
 xfs_ail_min(
-	xfs_ail_t	*ailp)
+	struct xfs_ail	*ailp)
 /* ARGSUSED */
 {
 	if (list_empty(&ailp->xa_ail))
@@ -601,7 +677,7 @@ xfs_ail_min(
  */
 STATIC xfs_log_item_t *
 xfs_ail_next(
-	xfs_ail_t	*ailp,
+	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
 /* ARGSUSED */
 {
@@ -617,7 +693,7 @@ xfs_ail_next(
  */
 STATIC void
 xfs_ail_check(
-	xfs_ail_t 	*ailp,
+	struct xfs_ail	*ailp,
 	xfs_log_item_t	*lip)
 {
 	xfs_log_item_t	*prev_lip;
diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c
index 4e855b5ced6..8ee2f8c8b0a 100644
--- a/fs/xfs/xfs_trans_buf.c
+++ b/fs/xfs/xfs_trans_buf.c
@@ -527,9 +527,8 @@ xfs_trans_brelse(xfs_trans_t	*tp,
 			lip = XFS_BUF_FSPRIVATE(bp, xfs_log_item_t *);
 			if (lip->li_type == XFS_LI_BUF) {
 				bip = XFS_BUF_FSPRIVATE(bp,xfs_buf_log_item_t*);
-				xfs_trans_unlocked_item(
-						bip->bli_item.li_mountp,
-						lip);
+				xfs_trans_unlocked_item(bip->bli_item.li_ailp,
+							lip);
 			}
 		}
 		xfs_buf_relse(bp);
@@ -626,7 +625,7 @@ xfs_trans_brelse(xfs_trans_t	*tp,
 	 * tell the AIL that the buffer is being unlocked.
 	 */
 	if (bip != NULL) {
-		xfs_trans_unlocked_item(bip->bli_item.li_mountp,
+		xfs_trans_unlocked_item(bip->bli_item.li_ailp,
 					(xfs_log_item_t*)bip);
 	}
 
diff --git a/fs/xfs/xfs_trans_inode.c b/fs/xfs/xfs_trans_inode.c
index 2a1c0f071f9..23d276af2e0 100644
--- a/fs/xfs/xfs_trans_inode.c
+++ b/fs/xfs/xfs_trans_inode.c
@@ -85,7 +85,6 @@ xfs_trans_iget(
 {
 	int			error;
 	xfs_inode_t		*ip;
-	xfs_inode_log_item_t	*iip;
 
 	/*
 	 * If the transaction pointer is NULL, just call the normal
@@ -138,34 +137,7 @@ xfs_trans_iget(
 	}
 	ASSERT(ip != NULL);
 
-	/*
-	 * Get a log_item_desc to point at the new item.
-	 */
-	if (ip->i_itemp == NULL)
-		xfs_inode_item_init(ip, mp);
-	iip = ip->i_itemp;
-	(void) xfs_trans_add_item(tp, (xfs_log_item_t *)(iip));
-
-	xfs_trans_inode_broot_debug(ip);
-
-	/*
-	 * If the IO lock has been acquired, mark that in
-	 * the inode log item so we'll know to unlock it
-	 * when the transaction commits.
-	 */
-	ASSERT(iip->ili_flags == 0);
-	if (lock_flags & XFS_IOLOCK_EXCL) {
-		iip->ili_flags |= XFS_ILI_IOLOCKED_EXCL;
-	} else if (lock_flags & XFS_IOLOCK_SHARED) {
-		iip->ili_flags |= XFS_ILI_IOLOCKED_SHARED;
-	}
-
-	/*
-	 * Initialize i_transp so we can find it with xfs_inode_incore()
-	 * above.
-	 */
-	ip->i_transp = tp;
-
+	xfs_trans_ijoin(tp, ip, lock_flags);
 	*ipp = ip;
 	return 0;
 }
diff --git a/fs/xfs/xfs_trans_item.c b/fs/xfs/xfs_trans_item.c
index 3c666e8317f..e110bf57d7f 100644
--- a/fs/xfs/xfs_trans_item.c
+++ b/fs/xfs/xfs_trans_item.c
@@ -22,6 +22,14 @@
 #include "xfs_inum.h"
 #include "xfs_trans.h"
 #include "xfs_trans_priv.h"
+/* XXX: from here down needed until struct xfs_trans has it's own ailp */
+#include "xfs_bit.h"
+#include "xfs_buf_item.h"
+#include "xfs_sb.h"
+#include "xfs_ag.h"
+#include "xfs_dir2.h"
+#include "xfs_dmapi.h"
+#include "xfs_mount.h"
 
 STATIC int	xfs_trans_unlock_chunk(xfs_log_item_chunk_t *,
 					int, int, xfs_lsn_t);
@@ -79,6 +87,7 @@ xfs_trans_add_item(xfs_trans_t *tp, xfs_log_item_t *lip)
 		lidp->lid_size = 0;
 		lip->li_desc = lidp;
 		lip->li_mountp = tp->t_mountp;
+		lip->li_ailp = tp->t_mountp->m_ail;
 		return lidp;
 	}
 
@@ -120,6 +129,7 @@ xfs_trans_add_item(xfs_trans_t *tp, xfs_log_item_t *lip)
 	lidp->lid_size = 0;
 	lip->li_desc = lidp;
 	lip->li_mountp = tp->t_mountp;
+	lip->li_ailp = tp->t_mountp->m_ail;
 	return lidp;
 }
 
diff --git a/fs/xfs/xfs_trans_priv.h b/fs/xfs/xfs_trans_priv.h
index 3c748c456ed..73e2ad39743 100644
--- a/fs/xfs/xfs_trans_priv.h
+++ b/fs/xfs/xfs_trans_priv.h
@@ -44,25 +44,93 @@ xfs_log_busy_slot_t		*xfs_trans_add_busy(xfs_trans_t *tp,
 						    xfs_extlen_t idx);
 
 /*
- * From xfs_trans_ail.c
+ * AIL traversal cursor.
+ *
+ * Rather than using a generation number for detecting changes in the ail, use
+ * a cursor that is protected by the ail lock. The aild cursor exists in the
+ * struct xfs_ail, but other traversals can declare it on the stack and link it
+ * to the ail list.
+ *
+ * When an object is deleted from or moved int the AIL, the cursor list is
+ * searched to see if the object is a designated cursor item. If it is, it is
+ * deleted from the cursor so that the next time the cursor is used traversal
+ * will return to the start.
+ *
+ * This means a traversal colliding with a removal will cause a restart of the
+ * list scan, rather than any insertion or deletion anywhere in the list. The
+ * low bit of the item pointer is set if the cursor has been invalidated so
+ * that we can tell the difference between invalidation and reaching the end
+ * of the list to trigger traversal restarts.
  */
-void			xfs_trans_update_ail(struct xfs_mount *mp,
-				     struct xfs_log_item *lip, xfs_lsn_t lsn)
-				     __releases(mp->m_ail_lock);
-void			xfs_trans_delete_ail(struct xfs_mount *mp,
-				     struct xfs_log_item *lip)
-				     __releases(mp->m_ail_lock);
-struct xfs_log_item	*xfs_trans_first_ail(struct xfs_mount *, int *);
-struct xfs_log_item	*xfs_trans_next_ail(struct xfs_mount *,
-				     struct xfs_log_item *, int *, int *);
+struct xfs_ail_cursor {
+	struct xfs_ail_cursor	*next;
+	struct xfs_log_item	*item;
+};
 
+/*
+ * Private AIL structures.
+ *
+ * Eventually we need to drive the locking in here as well.
+ */
+struct xfs_ail {
+	struct xfs_mount	*xa_mount;
+	struct list_head	xa_ail;
+	uint			xa_gen;
+	struct task_struct	*xa_task;
+	xfs_lsn_t		xa_target;
+	struct xfs_ail_cursor	xa_cursors;
+	spinlock_t		xa_lock;
+};
 
 /*
- * AIL push thread support
+ * From xfs_trans_ail.c
  */
-long	xfsaild_push(struct xfs_mount *, xfs_lsn_t *);
-void	xfsaild_wakeup(struct xfs_mount *, xfs_lsn_t);
-int	xfsaild_start(struct xfs_mount *);
-void	xfsaild_stop(struct xfs_mount *);
+void			xfs_trans_ail_update(struct xfs_ail *ailp,
+					struct xfs_log_item *lip, xfs_lsn_t lsn)
+					__releases(ailp->xa_lock);
+void			xfs_trans_ail_delete(struct xfs_ail *ailp,
+					struct xfs_log_item *lip)
+					__releases(ailp->xa_lock);
+void			xfs_trans_ail_push(struct xfs_ail *, xfs_lsn_t);
+void			xfs_trans_unlocked_item(struct xfs_ail *,
+					xfs_log_item_t *);
+
+xfs_lsn_t		xfs_trans_ail_tail(struct xfs_ail *ailp);
+
+struct xfs_log_item	*xfs_trans_ail_cursor_first(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur,
+					xfs_lsn_t lsn);
+struct xfs_log_item	*xfs_trans_ail_cursor_next(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur);
+void			xfs_trans_ail_cursor_done(struct xfs_ail *ailp,
+					struct xfs_ail_cursor *cur);
+
+long	xfsaild_push(struct xfs_ail *, xfs_lsn_t *);
+void	xfsaild_wakeup(struct xfs_ail *, xfs_lsn_t);
+int	xfsaild_start(struct xfs_ail *);
+void	xfsaild_stop(struct xfs_ail *);
 
+#if BITS_PER_LONG != 64
+static inline void
+xfs_trans_ail_copy_lsn(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	*dst,
+	xfs_lsn_t	*src)
+{
+	ASSERT(sizeof(xfs_lsn_t) == 8);	/* don't lock if it shrinks */
+	spin_lock(&ailp->xa_lock);
+	*dst = *src;
+	spin_unlock(&ailp->xa_lock);
+}
+#else
+static inline void
+xfs_trans_ail_copy_lsn(
+	struct xfs_ail	*ailp,
+	xfs_lsn_t	*dst,
+	xfs_lsn_t	*src)
+{
+	ASSERT(sizeof(xfs_lsn_t) == 8);
+	*dst = *src;
+}
+#endif
 #endif	/* __XFS_TRANS_PRIV_H__ */
diff --git a/fs/xfs/xfs_utils.c b/fs/xfs/xfs_utils.c
index 35d4d414bcc..fcc2285d03e 100644
--- a/fs/xfs/xfs_utils.c
+++ b/fs/xfs/xfs_utils.c
@@ -172,6 +172,12 @@ xfs_dir_ialloc(
 			*ipp = NULL;
 			return code;
 		}
+
+		/*
+		 * transaction commit worked ok so we can drop the extra ticket
+		 * reference that we gained in xfs_trans_dup()
+		 */
+		xfs_log_ticket_put(tp->t_ticket);
 		code = xfs_trans_reserve(tp, 0, log_res, 0,
 					 XFS_TRANS_PERM_LOG_RES, log_count);
 		/*
@@ -268,9 +274,9 @@ xfs_bump_ino_vers2(
 	xfs_mount_t	*mp;
 
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
-	ASSERT(ip->i_d.di_version == XFS_DINODE_VERSION_1);
+	ASSERT(ip->i_d.di_version == 1);
 
-	ip->i_d.di_version = XFS_DINODE_VERSION_2;
+	ip->i_d.di_version = 2;
 	ip->i_d.di_onlink = 0;
 	memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
 	mp = tp->t_mountp;
@@ -302,7 +308,7 @@ xfs_bumplink(
 	ASSERT(ip->i_d.di_nlink > 0);
 	ip->i_d.di_nlink++;
 	inc_nlink(VFS_I(ip));
-	if ((ip->i_d.di_version == XFS_DINODE_VERSION_1) &&
+	if ((ip->i_d.di_version == 1) &&
 	    (ip->i_d.di_nlink > XFS_MAXLINK_1)) {
 		/*
 		 * The inode has increased its number of links beyond
diff --git a/fs/xfs/xfs_vfsops.c b/fs/xfs/xfs_vfsops.c
deleted file mode 100644
index 439dd3939dd..00000000000
--- a/fs/xfs/xfs_vfsops.c
+++ /dev/null
@@ -1,757 +0,0 @@
-/*
- * Copyright (c) 2000-2005 Silicon Graphics, Inc.
- * All Rights Reserved.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of the GNU General Public License as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it would be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write the Free Software Foundation,
- * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
- */
-#include "xfs.h"
-#include "xfs_fs.h"
-#include "xfs_types.h"
-#include "xfs_bit.h"
-#include "xfs_log.h"
-#include "xfs_inum.h"
-#include "xfs_trans.h"
-#include "xfs_sb.h"
-#include "xfs_ag.h"
-#include "xfs_dir2.h"
-#include "xfs_dmapi.h"
-#include "xfs_mount.h"
-#include "xfs_da_btree.h"
-#include "xfs_bmap_btree.h"
-#include "xfs_ialloc_btree.h"
-#include "xfs_alloc_btree.h"
-#include "xfs_dir2_sf.h"
-#include "xfs_attr_sf.h"
-#include "xfs_dinode.h"
-#include "xfs_inode.h"
-#include "xfs_inode_item.h"
-#include "xfs_btree.h"
-#include "xfs_alloc.h"
-#include "xfs_ialloc.h"
-#include "xfs_quota.h"
-#include "xfs_error.h"
-#include "xfs_bmap.h"
-#include "xfs_rw.h"
-#include "xfs_buf_item.h"
-#include "xfs_log_priv.h"
-#include "xfs_dir2_trace.h"
-#include "xfs_extfree_item.h"
-#include "xfs_acl.h"
-#include "xfs_attr.h"
-#include "xfs_clnt.h"
-#include "xfs_mru_cache.h"
-#include "xfs_filestream.h"
-#include "xfs_fsops.h"
-#include "xfs_vnodeops.h"
-#include "xfs_vfsops.h"
-#include "xfs_utils.h"
-
-
-STATIC void
-xfs_quiesce_fs(
-	xfs_mount_t		*mp)
-{
-	int			count = 0, pincount;
-
-	xfs_flush_buftarg(mp->m_ddev_targp, 0);
-	xfs_finish_reclaim_all(mp, 0);
-
-	/* This loop must run at least twice.
-	 * The first instance of the loop will flush
-	 * most meta data but that will generate more
-	 * meta data (typically directory updates).
-	 * Which then must be flushed and logged before
-	 * we can write the unmount record.
-	 */
-	do {
-		xfs_syncsub(mp, SYNC_INODE_QUIESCE, NULL);
-		pincount = xfs_flush_buftarg(mp->m_ddev_targp, 1);
-		if (!pincount) {
-			delay(50);
-			count++;
-		}
-	} while (count < 2);
-}
-
-/*
- * Second stage of a quiesce. The data is already synced, now we have to take
- * care of the metadata. New transactions are already blocked, so we need to
- * wait for any remaining transactions to drain out before proceding.
- */
-void
-xfs_attr_quiesce(
-	xfs_mount_t	*mp)
-{
-	int	error = 0;
-
-	/* wait for all modifications to complete */
-	while (atomic_read(&mp->m_active_trans) > 0)
-		delay(100);
-
-	/* flush inodes and push all remaining buffers out to disk */
-	xfs_quiesce_fs(mp);
-
-	ASSERT_ALWAYS(atomic_read(&mp->m_active_trans) == 0);
-
-	/* Push the superblock and write an unmount record */
-	error = xfs_log_sbcount(mp, 1);
-	if (error)
-		xfs_fs_cmn_err(CE_WARN, mp,
-				"xfs_attr_quiesce: failed to log sb changes. "
-				"Frozen image may not be consistent.");
-	xfs_log_unmount_write(mp);
-	xfs_unmountfs_writesb(mp);
-}
-
-/*
- * xfs_unmount_flush implements a set of flush operation on special
- * inodes, which are needed as a separate set of operations so that
- * they can be called as part of relocation process.
- */
-int
-xfs_unmount_flush(
-	xfs_mount_t	*mp,		/* Mount structure we are getting
-					   rid of. */
-	int             relocation)	/* Called from vfs relocation. */
-{
-	xfs_inode_t	*rip = mp->m_rootip;
-	xfs_inode_t	*rbmip;
-	xfs_inode_t	*rsumip = NULL;
-	int		error;
-
-	xfs_ilock(rip, XFS_ILOCK_EXCL | XFS_ILOCK_PARENT);
-	xfs_iflock(rip);
-
-	/*
-	 * Flush out the real time inodes.
-	 */
-	if ((rbmip = mp->m_rbmip) != NULL) {
-		xfs_ilock(rbmip, XFS_ILOCK_EXCL);
-		xfs_iflock(rbmip);
-		error = xfs_iflush(rbmip, XFS_IFLUSH_SYNC);
-		xfs_iunlock(rbmip, XFS_ILOCK_EXCL);
-
-		if (error == EFSCORRUPTED)
-			goto fscorrupt_out;
-
-		ASSERT(vn_count(VFS_I(rbmip)) == 1);
-
-		rsumip = mp->m_rsumip;
-		xfs_ilock(rsumip, XFS_ILOCK_EXCL);
-		xfs_iflock(rsumip);
-		error = xfs_iflush(rsumip, XFS_IFLUSH_SYNC);
-		xfs_iunlock(rsumip, XFS_ILOCK_EXCL);
-
-		if (error == EFSCORRUPTED)
-			goto fscorrupt_out;
-
-		ASSERT(vn_count(VFS_I(rsumip)) == 1);
-	}
-
-	/*
-	 * Synchronously flush root inode to disk
-	 */
-	error = xfs_iflush(rip, XFS_IFLUSH_SYNC);
-	if (error == EFSCORRUPTED)
-		goto fscorrupt_out2;
-
-	if (vn_count(VFS_I(rip)) != 1 && !relocation) {
-		xfs_iunlock(rip, XFS_ILOCK_EXCL);
-		return XFS_ERROR(EBUSY);
-	}
-
-	/*
-	 * Release dquot that rootinode, rbmino and rsumino might be holding,
-	 * flush and purge the quota inodes.
-	 */
-	error = XFS_QM_UNMOUNT(mp);
-	if (error == EFSCORRUPTED)
-		goto fscorrupt_out2;
-
-	if (rbmip) {
-		IRELE(rbmip);
-		IRELE(rsumip);
-	}
-
-	xfs_iunlock(rip, XFS_ILOCK_EXCL);
-	return 0;
-
-fscorrupt_out:
-	xfs_ifunlock(rip);
-
-fscorrupt_out2:
-	xfs_iunlock(rip, XFS_ILOCK_EXCL);
-
-	return XFS_ERROR(EFSCORRUPTED);
-}
-
-/*
- * xfs_sync flushes any pending I/O to file system vfsp.
- *
- * This routine is called by vfs_sync() to make sure that things make it
- * out to disk eventually, on sync() system calls to flush out everything,
- * and when the file system is unmounted.  For the vfs_sync() case, all
- * we really need to do is sync out the log to make all of our meta-data
- * updates permanent (except for timestamps).  For calls from pflushd(),
- * dirty pages are kept moving by calling pdflush() on the inodes
- * containing them.  We also flush the inodes that we can lock without
- * sleeping and the superblock if we can lock it without sleeping from
- * vfs_sync() so that items at the tail of the log are always moving out.
- *
- * Flags:
- *      SYNC_BDFLUSH - We're being called from vfs_sync() so we don't want
- *		       to sleep if we can help it.  All we really need
- *		       to do is ensure that the log is synced at least
- *		       periodically.  We also push the inodes and
- *		       superblock if we can lock them without sleeping
- *			and they are not pinned.
- *      SYNC_ATTR    - We need to flush the inodes.  If SYNC_BDFLUSH is not
- *		       set, then we really want to lock each inode and flush
- *		       it.
- *      SYNC_WAIT    - All the flushes that take place in this call should
- *		       be synchronous.
- *      SYNC_DELWRI  - This tells us to push dirty pages associated with
- *		       inodes.  SYNC_WAIT and SYNC_BDFLUSH are used to
- *		       determine if they should be flushed sync, async, or
- *		       delwri.
- *      SYNC_CLOSE   - This flag is passed when the system is being
- *		       unmounted.  We should sync and invalidate everything.
- *      SYNC_FSDATA  - This indicates that the caller would like to make
- *		       sure the superblock is safe on disk.  We can ensure
- *		       this by simply making sure the log gets flushed
- *		       if SYNC_BDFLUSH is set, and by actually writing it
- *		       out otherwise.
- *	SYNC_IOWAIT  - The caller wants us to wait for all data I/O to complete
- *		       before we return (including direct I/O). Forms the drain
- *		       side of the write barrier needed to safely quiesce the
- *		       filesystem.
- *
- */
-int
-xfs_sync(
-	xfs_mount_t	*mp,
-	int		flags)
-{
-	int		error;
-
-	/*
-	 * Get the Quota Manager to flush the dquots.
-	 *
-	 * If XFS quota support is not enabled or this filesystem
-	 * instance does not use quotas XFS_QM_DQSYNC will always
-	 * return zero.
-	 */
-	error = XFS_QM_DQSYNC(mp, flags);
-	if (error) {
-		/*
-		 * If we got an IO error, we will be shutting down.
-		 * So, there's nothing more for us to do here.
-		 */
-		ASSERT(error != EIO || XFS_FORCED_SHUTDOWN(mp));
-		if (XFS_FORCED_SHUTDOWN(mp))
-			return XFS_ERROR(error);
-	}
-
-	if (flags & SYNC_IOWAIT)
-		xfs_filestream_flush(mp);
-
-	return xfs_syncsub(mp, flags, NULL);
-}
-
-/*
- * xfs sync routine for internal use
- *
- * This routine supports all of the flags defined for the generic vfs_sync
- * interface as explained above under xfs_sync.
- *
- */
-int
-xfs_sync_inodes(
-	xfs_mount_t	*mp,
-	int		flags,
-	int             *bypassed)
-{
-	xfs_inode_t	*ip = NULL;
-	struct inode	*vp = NULL;
-	int		error;
-	int		last_error;
-	uint64_t	fflag;
-	uint		lock_flags;
-	uint		base_lock_flags;
-	boolean_t	mount_locked;
-	boolean_t	vnode_refed;
-	int		preempt;
-	xfs_iptr_t	*ipointer;
-#ifdef DEBUG
-	boolean_t	ipointer_in = B_FALSE;
-
-#define IPOINTER_SET	ipointer_in = B_TRUE
-#define IPOINTER_CLR	ipointer_in = B_FALSE
-#else
-#define IPOINTER_SET
-#define IPOINTER_CLR
-#endif
-
-
-/* Insert a marker record into the inode list after inode ip. The list
- * must be locked when this is called. After the call the list will no
- * longer be locked.
- */
-#define IPOINTER_INSERT(ip, mp)	{ \
-		ASSERT(ipointer_in == B_FALSE); \
-		ipointer->ip_mnext = ip->i_mnext; \
-		ipointer->ip_mprev = ip; \
-		ip->i_mnext = (xfs_inode_t *)ipointer; \
-		ipointer->ip_mnext->i_mprev = (xfs_inode_t *)ipointer; \
-		preempt = 0; \
-		XFS_MOUNT_IUNLOCK(mp); \
-		mount_locked = B_FALSE; \
-		IPOINTER_SET; \
-	}
-
-/* Remove the marker from the inode list. If the marker was the only item
- * in the list then there are no remaining inodes and we should zero out
- * the whole list. If we are the current head of the list then move the head
- * past us.
- */
-#define IPOINTER_REMOVE(ip, mp)	{ \
-		ASSERT(ipointer_in == B_TRUE); \
-		if (ipointer->ip_mnext != (xfs_inode_t *)ipointer) { \
-			ip = ipointer->ip_mnext; \
-			ip->i_mprev = ipointer->ip_mprev; \
-			ipointer->ip_mprev->i_mnext = ip; \
-			if (mp->m_inodes == (xfs_inode_t *)ipointer) { \
-				mp->m_inodes = ip; \
-			} \
-		} else { \
-			ASSERT(mp->m_inodes == (xfs_inode_t *)ipointer); \
-			mp->m_inodes = NULL; \
-			ip = NULL; \
-		} \
-		IPOINTER_CLR; \
-	}
-
-#define XFS_PREEMPT_MASK	0x7f
-
-	ASSERT(!(flags & SYNC_BDFLUSH));
-
-	if (bypassed)
-		*bypassed = 0;
-	if (mp->m_flags & XFS_MOUNT_RDONLY)
-		return 0;
-	error = 0;
-	last_error = 0;
-	preempt = 0;
-
-	/* Allocate a reference marker */
-	ipointer = (xfs_iptr_t *)kmem_zalloc(sizeof(xfs_iptr_t), KM_SLEEP);
-
-	fflag = XFS_B_ASYNC;		/* default is don't wait */
-	if (flags & SYNC_DELWRI)
-		fflag = XFS_B_DELWRI;
-	if (flags & SYNC_WAIT)
-		fflag = 0;		/* synchronous overrides all */
-
-	base_lock_flags = XFS_ILOCK_SHARED;
-	if (flags & (SYNC_DELWRI | SYNC_CLOSE)) {
-		/*
-		 * We need the I/O lock if we're going to call any of
-		 * the flush/inval routines.
-		 */
-		base_lock_flags |= XFS_IOLOCK_SHARED;
-	}
-
-	XFS_MOUNT_ILOCK(mp);
-
-	ip = mp->m_inodes;
-
-	mount_locked = B_TRUE;
-	vnode_refed  = B_FALSE;
-
-	IPOINTER_CLR;
-
-	do {
-		ASSERT(ipointer_in == B_FALSE);
-		ASSERT(vnode_refed == B_FALSE);
-
-		lock_flags = base_lock_flags;
-
-		/*
-		 * There were no inodes in the list, just break out
-		 * of the loop.
-		 */
-		if (ip == NULL) {
-			break;
-		}
-
-		/*
-		 * We found another sync thread marker - skip it
-		 */
-		if (ip->i_mount == NULL) {
-			ip = ip->i_mnext;
-			continue;
-		}
-
-		vp = VFS_I(ip);
-
-		/*
-		 * If the vnode is gone then this is being torn down,
-		 * call reclaim if it is flushed, else let regular flush
-		 * code deal with it later in the loop.
-		 */
-
-		if (vp == NULL) {
-			/* Skip ones already in reclaim */
-			if (ip->i_flags & XFS_IRECLAIM) {
-				ip = ip->i_mnext;
-				continue;
-			}
-			if (xfs_ilock_nowait(ip, XFS_ILOCK_EXCL) == 0) {
-				ip = ip->i_mnext;
-			} else if ((xfs_ipincount(ip) == 0) &&
-				    xfs_iflock_nowait(ip)) {
-				IPOINTER_INSERT(ip, mp);
-
-				xfs_finish_reclaim(ip, 1,
-						XFS_IFLUSH_DELWRI_ELSE_ASYNC);
-
-				XFS_MOUNT_ILOCK(mp);
-				mount_locked = B_TRUE;
-				IPOINTER_REMOVE(ip, mp);
-			} else {
-				xfs_iunlock(ip, XFS_ILOCK_EXCL);
-				ip = ip->i_mnext;
-			}
-			continue;
-		}
-
-		if (VN_BAD(vp)) {
-			ip = ip->i_mnext;
-			continue;
-		}
-
-		if (XFS_FORCED_SHUTDOWN(mp) && !(flags & SYNC_CLOSE)) {
-			XFS_MOUNT_IUNLOCK(mp);
-			kmem_free(ipointer);
-			return 0;
-		}
-
-		/*
-		 * Try to lock without sleeping.  We're out of order with
-		 * the inode list lock here, so if we fail we need to drop
-		 * the mount lock and try again.  If we're called from
-		 * bdflush() here, then don't bother.
-		 *
-		 * The inode lock here actually coordinates with the
-		 * almost spurious inode lock in xfs_ireclaim() to prevent
-		 * the vnode we handle here without a reference from
-		 * being freed while we reference it.  If we lock the inode
-		 * while it's on the mount list here, then the spurious inode
-		 * lock in xfs_ireclaim() after the inode is pulled from
-		 * the mount list will sleep until we release it here.
-		 * This keeps the vnode from being freed while we reference
-		 * it.
-		 */
-		if (xfs_ilock_nowait(ip, lock_flags) == 0) {
-			if (vp == NULL) {
-				ip = ip->i_mnext;
-				continue;
-			}
-
-			vp = vn_grab(vp);
-			if (vp == NULL) {
-				ip = ip->i_mnext;
-				continue;
-			}
-
-			IPOINTER_INSERT(ip, mp);
-			xfs_ilock(ip, lock_flags);
-
-			ASSERT(vp == VFS_I(ip));
-			ASSERT(ip->i_mount == mp);
-
-			vnode_refed = B_TRUE;
-		}
-
-		/* From here on in the loop we may have a marker record
-		 * in the inode list.
-		 */
-
-		/*
-		 * If we have to flush data or wait for I/O completion
-		 * we need to drop the ilock that we currently hold.
-		 * If we need to drop the lock, insert a marker if we
-		 * have not already done so.
-		 */
-		if ((flags & (SYNC_CLOSE|SYNC_IOWAIT)) ||
-		    ((flags & SYNC_DELWRI) && VN_DIRTY(vp))) {
-			if (mount_locked) {
-				IPOINTER_INSERT(ip, mp);
-			}
-			xfs_iunlock(ip, XFS_ILOCK_SHARED);
-
-			if (flags & SYNC_CLOSE) {
-				/* Shutdown case. Flush and invalidate. */
-				if (XFS_FORCED_SHUTDOWN(mp))
-					xfs_tosspages(ip, 0, -1,
-							     FI_REMAPF);
-				else
-					error = xfs_flushinval_pages(ip,
-							0, -1, FI_REMAPF);
-			} else if ((flags & SYNC_DELWRI) && VN_DIRTY(vp)) {
-				error = xfs_flush_pages(ip, 0,
-							-1, fflag, FI_NONE);
-			}
-
-			/*
-			 * When freezing, we need to wait ensure all I/O (including direct
-			 * I/O) is complete to ensure no further data modification can take
-			 * place after this point
-			 */
-			if (flags & SYNC_IOWAIT)
-				vn_iowait(ip);
-
-			xfs_ilock(ip, XFS_ILOCK_SHARED);
-		}
-
-		if ((flags & SYNC_ATTR) &&
-		    (ip->i_update_core ||
-		     (ip->i_itemp && ip->i_itemp->ili_format.ilf_fields))) {
-			if (mount_locked)
-				IPOINTER_INSERT(ip, mp);
-
-			if (flags & SYNC_WAIT) {
-				xfs_iflock(ip);
-				error = xfs_iflush(ip, XFS_IFLUSH_SYNC);
-
-			/*
-			 * If we can't acquire the flush lock, then the inode
-			 * is already being flushed so don't bother waiting.
-			 *
-			 * If we can lock it then do a delwri flush so we can
-			 * combine multiple inode flushes in each disk write.
-			 */
-			} else if (xfs_iflock_nowait(ip)) {
-				error = xfs_iflush(ip, XFS_IFLUSH_DELWRI);
-			} else if (bypassed) {
-				(*bypassed)++;
-			}
-		}
-
-		if (lock_flags != 0) {
-			xfs_iunlock(ip, lock_flags);
-		}
-
-		if (vnode_refed) {
-			/*
-			 * If we had to take a reference on the vnode
-			 * above, then wait until after we've unlocked
-			 * the inode to release the reference.  This is
-			 * because we can be already holding the inode
-			 * lock when IRELE() calls xfs_inactive().
-			 *
-			 * Make sure to drop the mount lock before calling
-			 * IRELE() so that we don't trip over ourselves if
-			 * we have to go for the mount lock again in the
-			 * inactive code.
-			 */
-			if (mount_locked) {
-				IPOINTER_INSERT(ip, mp);
-			}
-
-			IRELE(ip);
-
-			vnode_refed = B_FALSE;
-		}
-
-		if (error) {
-			last_error = error;
-		}
-
-		/*
-		 * bail out if the filesystem is corrupted.
-		 */
-		if (error == EFSCORRUPTED)  {
-			if (!mount_locked) {
-				XFS_MOUNT_ILOCK(mp);
-				IPOINTER_REMOVE(ip, mp);
-			}
-			XFS_MOUNT_IUNLOCK(mp);
-			ASSERT(ipointer_in == B_FALSE);
-			kmem_free(ipointer);
-			return XFS_ERROR(error);
-		}
-
-		/* Let other threads have a chance at the mount lock
-		 * if we have looped many times without dropping the
-		 * lock.
-		 */
-		if ((++preempt & XFS_PREEMPT_MASK) == 0) {
-			if (mount_locked) {
-				IPOINTER_INSERT(ip, mp);
-			}
-		}
-
-		if (mount_locked == B_FALSE) {
-			XFS_MOUNT_ILOCK(mp);
-			mount_locked = B_TRUE;
-			IPOINTER_REMOVE(ip, mp);
-			continue;
-		}
-
-		ASSERT(ipointer_in == B_FALSE);
-		ip = ip->i_mnext;
-
-	} while (ip != mp->m_inodes);
-
-	XFS_MOUNT_IUNLOCK(mp);
-
-	ASSERT(ipointer_in == B_FALSE);
-
-	kmem_free(ipointer);
-	return XFS_ERROR(last_error);
-}
-
-/*
- * xfs sync routine for internal use
- *
- * This routine supports all of the flags defined for the generic vfs_sync
- * interface as explained above under xfs_sync.
- *
- */
-int
-xfs_syncsub(
-	xfs_mount_t	*mp,
-	int		flags,
-	int             *bypassed)
-{
-	int		error = 0;
-	int		last_error = 0;
-	uint		log_flags = XFS_LOG_FORCE;
-	xfs_buf_t	*bp;
-	xfs_buf_log_item_t	*bip;
-
-	/*
-	 * Sync out the log.  This ensures that the log is periodically
-	 * flushed even if there is not enough activity to fill it up.
-	 */
-	if (flags & SYNC_WAIT)
-		log_flags |= XFS_LOG_SYNC;
-
-	xfs_log_force(mp, (xfs_lsn_t)0, log_flags);
-
-	if (flags & (SYNC_ATTR|SYNC_DELWRI)) {
-		if (flags & SYNC_BDFLUSH)
-			xfs_finish_reclaim_all(mp, 1);
-		else
-			error = xfs_sync_inodes(mp, flags, bypassed);
-	}
-
-	/*
-	 * Flushing out dirty data above probably generated more
-	 * log activity, so if this isn't vfs_sync() then flush
-	 * the log again.
-	 */
-	if (flags & SYNC_DELWRI) {
-		xfs_log_force(mp, (xfs_lsn_t)0, log_flags);
-	}
-
-	if (flags & SYNC_FSDATA) {
-		/*
-		 * If this is vfs_sync() then only sync the superblock
-		 * if we can lock it without sleeping and it is not pinned.
-		 */
-		if (flags & SYNC_BDFLUSH) {
-			bp = xfs_getsb(mp, XFS_BUF_TRYLOCK);
-			if (bp != NULL) {
-				bip = XFS_BUF_FSPRIVATE(bp,xfs_buf_log_item_t*);
-				if ((bip != NULL) &&
-				    xfs_buf_item_dirty(bip)) {
-					if (!(XFS_BUF_ISPINNED(bp))) {
-						XFS_BUF_ASYNC(bp);
-						error = xfs_bwrite(mp, bp);
-					} else {
-						xfs_buf_relse(bp);
-					}
-				} else {
-					xfs_buf_relse(bp);
-				}
-			}
-		} else {
-			bp = xfs_getsb(mp, 0);
-			/*
-			 * If the buffer is pinned then push on the log so
-			 * we won't get stuck waiting in the write for
-			 * someone, maybe ourselves, to flush the log.
-			 * Even though we just pushed the log above, we
-			 * did not have the superblock buffer locked at
-			 * that point so it can become pinned in between
-			 * there and here.
-			 */
-			if (XFS_BUF_ISPINNED(bp))
-				xfs_log_force(mp, (xfs_lsn_t)0, XFS_LOG_FORCE);
-			if (flags & SYNC_WAIT)
-				XFS_BUF_UNASYNC(bp);
-			else
-				XFS_BUF_ASYNC(bp);
-			error = xfs_bwrite(mp, bp);
-		}
-		if (error) {
-			last_error = error;
-		}
-	}
-
-	/*
-	 * Now check to see if the log needs a "dummy" transaction.
-	 */
-	if (!(flags & SYNC_REMOUNT) && xfs_log_need_covered(mp)) {
-		xfs_trans_t *tp;
-		xfs_inode_t *ip;
-
-		/*
-		 * Put a dummy transaction in the log to tell
-		 * recovery that all others are OK.
-		 */
-		tp = xfs_trans_alloc(mp, XFS_TRANS_DUMMY1);
-		if ((error = xfs_trans_reserve(tp, 0,
-				XFS_ICHANGE_LOG_RES(mp),
-				0, 0, 0)))  {
-			xfs_trans_cancel(tp, 0);
-			return error;
-		}
-
-		ip = mp->m_rootip;
-		xfs_ilock(ip, XFS_ILOCK_EXCL);
-
-		xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
-		xfs_trans_ihold(tp, ip);
-		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-		error = xfs_trans_commit(tp, 0);
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-		xfs_log_force(mp, (xfs_lsn_t)0, log_flags);
-	}
-
-	/*
-	 * When shutting down, we need to insure that the AIL is pushed
-	 * to disk or the filesystem can appear corrupt from the PROM.
-	 */
-	if ((flags & (SYNC_CLOSE|SYNC_WAIT)) == (SYNC_CLOSE|SYNC_WAIT)) {
-		XFS_bflush(mp->m_ddev_targp);
-		if (mp->m_rtdev_targp) {
-			XFS_bflush(mp->m_rtdev_targp);
-		}
-	}
-
-	return XFS_ERROR(last_error);
-}
diff --git a/fs/xfs/xfs_vfsops.h b/fs/xfs/xfs_vfsops.h
deleted file mode 100644
index a74b05087da..00000000000
--- a/fs/xfs/xfs_vfsops.h
+++ /dev/null
@@ -1,16 +0,0 @@
-#ifndef _XFS_VFSOPS_H
-#define _XFS_VFSOPS_H 1
-
-struct cred;
-struct xfs_fid;
-struct inode;
-struct kstatfs;
-struct xfs_mount;
-struct xfs_mount_args;
-
-int xfs_sync(struct xfs_mount *mp, int flags);
-void xfs_do_force_shutdown(struct xfs_mount *mp, int flags, char *fname,
-		int lnnum);
-void xfs_attr_quiesce(struct xfs_mount *mp);
-
-#endif /* _XFS_VFSOPS_H */
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index 8b6812f66a1..f07bf8768c3 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -54,33 +54,10 @@
 #include "xfs_vnodeops.h"
 
 int
-xfs_open(
-	xfs_inode_t	*ip)
-{
-	int		mode;
-
-	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
-		return XFS_ERROR(EIO);
-
-	/*
-	 * If it's a directory with any blocks, read-ahead block 0
-	 * as we're almost certain to have the next operation be a read there.
-	 */
-	if (S_ISDIR(ip->i_d.di_mode) && ip->i_d.di_nextents > 0) {
-		mode = xfs_ilock_map_shared(ip);
-		if (ip->i_d.di_nextents > 0)
-			(void)xfs_da_reada_buf(NULL, ip, 0, XFS_DATA_FORK);
-		xfs_iunlock(ip, mode);
-	}
-	return 0;
-}
-
-int
 xfs_setattr(
 	struct xfs_inode	*ip,
 	struct iattr		*iattr,
-	int			flags,
-	cred_t			*credp)
+	int			flags)
 {
 	xfs_mount_t		*mp = ip->i_mount;
 	struct inode		*inode = VFS_I(ip);
@@ -93,7 +70,6 @@ xfs_setattr(
 	gid_t			gid=0, igid=0;
 	int			timeflags = 0;
 	struct xfs_dquot	*udqp, *gdqp, *olddquot1, *olddquot2;
-	int			file_owner;
 	int			need_iolock = 1;
 
 	xfs_itrace_entry(ip);
@@ -104,6 +80,10 @@ xfs_setattr(
 	if (XFS_FORCED_SHUTDOWN(mp))
 		return XFS_ERROR(EIO);
 
+	code = -inode_change_ok(inode, iattr);
+	if (code)
+		return code;
+
 	olddquot1 = olddquot2 = NULL;
 	udqp = gdqp = NULL;
 
@@ -181,62 +161,8 @@ xfs_setattr(
 
 	xfs_ilock(ip, lock_flags);
 
-	/* boolean: are we the file owner? */
-	file_owner = (current_fsuid() == ip->i_d.di_uid);
-
-	/*
-	 * Change various properties of a file.
-	 * Only the owner or users with CAP_FOWNER
-	 * capability may do these things.
-	 */
-	if (mask & (ATTR_MODE|ATTR_UID|ATTR_GID)) {
-		/*
-		 * CAP_FOWNER overrides the following restrictions:
-		 *
-		 * The user ID of the calling process must be equal
-		 * to the file owner ID, except in cases where the
-		 * CAP_FSETID capability is applicable.
-		 */
-		if (!file_owner && !capable(CAP_FOWNER)) {
-			code = XFS_ERROR(EPERM);
-			goto error_return;
-		}
-
-		/*
-		 * CAP_FSETID overrides the following restrictions:
-		 *
-		 * The effective user ID of the calling process shall match
-		 * the file owner when setting the set-user-ID and
-		 * set-group-ID bits on that file.
-		 *
-		 * The effective group ID or one of the supplementary group
-		 * IDs of the calling process shall match the group owner of
-		 * the file when setting the set-group-ID bit on that file
-		 */
-		if (mask & ATTR_MODE) {
-			mode_t m = 0;
-
-			if ((iattr->ia_mode & S_ISUID) && !file_owner)
-				m |= S_ISUID;
-			if ((iattr->ia_mode & S_ISGID) &&
-			    !in_group_p((gid_t)ip->i_d.di_gid))
-				m |= S_ISGID;
-#if 0
-			/* Linux allows this, Irix doesn't. */
-			if ((iattr->ia_mode & S_ISVTX) && !S_ISDIR(ip->i_d.di_mode))
-				m |= S_ISVTX;
-#endif
-			if (m && !capable(CAP_FSETID))
-				iattr->ia_mode &= ~m;
-		}
-	}
-
 	/*
 	 * Change file ownership.  Must be the owner or privileged.
-	 * If the system was configured with the "restricted_chown"
-	 * option, the owner is not permitted to give away the file,
-	 * and can change the group id only to a group of which he
-	 * or she is a member.
 	 */
 	if (mask & (ATTR_UID|ATTR_GID)) {
 		/*
@@ -251,23 +177,6 @@ xfs_setattr(
 		uid = (mask & ATTR_UID) ? iattr->ia_uid : iuid;
 
 		/*
-		 * CAP_CHOWN overrides the following restrictions:
-		 *
-		 * If _POSIX_CHOWN_RESTRICTED is defined, this capability
-		 * shall override the restriction that a process cannot
-		 * change the user ID of a file it owns and the restriction
-		 * that the group ID supplied to the chown() function
-		 * shall be equal to either the group ID or one of the
-		 * supplementary group IDs of the calling process.
-		 */
-		if (restricted_chown &&
-		    (iuid != uid || (igid != gid &&
-				     !in_group_p((gid_t)gid))) &&
-		    !capable(CAP_CHOWN)) {
-			code = XFS_ERROR(EPERM);
-			goto error_return;
-		}
-		/*
 		 * Do a quota reservation only if uid/gid is actually
 		 * going to change.
 		 */
@@ -304,36 +213,22 @@ xfs_setattr(
 			code = XFS_ERROR(EINVAL);
 			goto error_return;
 		}
+
 		/*
 		 * Make sure that the dquots are attached to the inode.
 		 */
-		if ((code = XFS_QM_DQATTACH(mp, ip, XFS_QMOPT_ILOCKED)))
+		code = XFS_QM_DQATTACH(mp, ip, XFS_QMOPT_ILOCKED);
+		if (code)
 			goto error_return;
-	}
-
-	/*
-	 * Change file access or modified times.
-	 */
-	if (mask & (ATTR_ATIME|ATTR_MTIME)) {
-		if (!file_owner) {
-			if ((mask & (ATTR_MTIME_SET|ATTR_ATIME_SET)) &&
-			    !capable(CAP_FOWNER)) {
-				code = XFS_ERROR(EPERM);
-				goto error_return;
-			}
-		}
-	}
 
-	/*
-	 * Now we can make the changes.  Before we join the inode
-	 * to the transaction, if ATTR_SIZE is set then take care of
-	 * the part of the truncation that must be done without the
-	 * inode lock.  This needs to be done before joining the inode
-	 * to the transaction, because the inode cannot be unlocked
-	 * once it is a part of the transaction.
-	 */
-	if (mask & ATTR_SIZE) {
-		code = 0;
+		/*
+		 * Now we can make the changes.  Before we join the inode
+		 * to the transaction, if ATTR_SIZE is set then take care of
+		 * the part of the truncation that must be done without the
+		 * inode lock.  This needs to be done before joining the inode
+		 * to the transaction, because the inode cannot be unlocked
+		 * once it is a part of the transaction.
+		 */
 		if (iattr->ia_size > ip->i_size) {
 			/*
 			 * Do the first part of growing a file: zero any data
@@ -366,7 +261,7 @@ xfs_setattr(
 		}
 
 		/* wait for all I/O to complete */
-		vn_iowait(ip);
+		xfs_ioend_wait(ip);
 
 		if (!code)
 			code = xfs_itruncate_data(ip, iattr->ia_size);
@@ -388,17 +283,10 @@ xfs_setattr(
 		}
 		commit_flags = XFS_TRANS_RELEASE_LOG_RES;
 		xfs_ilock(ip, XFS_ILOCK_EXCL);
-	}
 
-	if (tp) {
 		xfs_trans_ijoin(tp, ip, lock_flags);
 		xfs_trans_ihold(tp, ip);
-	}
 
-	/*
-	 * Truncate file.  Must have write permission and not be a directory.
-	 */
-	if (mask & ATTR_SIZE) {
 		/*
 		 * Only change the c/mtime if we are changing the size
 		 * or we are explicitly asked to change it. This handles
@@ -438,28 +326,13 @@ xfs_setattr(
 			 */
 			xfs_iflags_set(ip, XFS_ITRUNCATED);
 		}
-	}
-
-	/*
-	 * Change file access modes.
-	 */
-	if (mask & ATTR_MODE) {
-		ip->i_d.di_mode &= S_IFMT;
-		ip->i_d.di_mode |= iattr->ia_mode & ~S_IFMT;
-
-		inode->i_mode &= S_IFMT;
-		inode->i_mode |= iattr->ia_mode & ~S_IFMT;
-
-		xfs_trans_log_inode (tp, ip, XFS_ILOG_CORE);
-		timeflags |= XFS_ICHGTIME_CHG;
+	} else if (tp) {
+		xfs_trans_ijoin(tp, ip, lock_flags);
+		xfs_trans_ihold(tp, ip);
 	}
 
 	/*
 	 * Change file ownership.  Must be the owner or privileged.
-	 * If the system was configured with the "restricted_chown"
-	 * option, the owner is not permitted to give away the file,
-	 * and can change the group id only to a group of which he
-	 * or she is a member.
 	 */
 	if (mask & (ATTR_UID|ATTR_GID)) {
 		/*
@@ -503,6 +376,24 @@ xfs_setattr(
 		timeflags |= XFS_ICHGTIME_CHG;
 	}
 
+	/*
+	 * Change file access modes.
+	 */
+	if (mask & ATTR_MODE) {
+		umode_t mode = iattr->ia_mode;
+
+		if (!in_group_p(inode->i_gid) && !capable(CAP_FSETID))
+			mode &= ~S_ISGID;
+
+		ip->i_d.di_mode &= S_IFMT;
+		ip->i_d.di_mode |= mode & ~S_IFMT;
+
+		inode->i_mode &= S_IFMT;
+		inode->i_mode |= mode & ~S_IFMT;
+
+		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
+		timeflags |= XFS_ICHGTIME_CHG;
+	}
 
 	/*
 	 * Change file access or modified times.
@@ -713,7 +604,7 @@ xfs_fsync(
 		return XFS_ERROR(EIO);
 
 	/* capture size updates in I/O completion before writing the inode. */
-	error = filemap_fdatawait(VFS_I(ip)->i_mapping);
+	error = xfs_wait_on_pages(ip, 0, -1);
 	if (error)
 		return XFS_ERROR(error);
 
@@ -1029,6 +920,12 @@ xfs_inactive_symlink_rmt(
 		goto error0;
 	}
 	/*
+	 * transaction commit worked ok so we can drop the extra ticket
+	 * reference that we gained in xfs_trans_dup()
+	 */
+	xfs_log_ticket_put(tp->t_ticket);
+
+	/*
 	 * Remove the memory for extent descriptions (just bookkeeping).
 	 */
 	if (ip->i_df.if_bytes)
@@ -1625,8 +1522,6 @@ xfs_create(
 		xfs_trans_set_sync(tp);
 	}
 
-	dp->i_gen++;
-
 	/*
 	 * Attach the dquot(s) to the inodes and modify them incore.
 	 * These ids of the inode couldn't have changed since the new
@@ -1993,13 +1888,6 @@ xfs_remove(
 	}
 	xfs_ichgtime(dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
 
-	/*
-	 * Bump the in memory generation count on the parent
-	 * directory so that other can know that it has changed.
-	 */
-	dp->i_gen++;
-	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
-
 	if (is_dir) {
 		/*
 		 * Drop the link from ip's "..".
@@ -2009,7 +1897,7 @@ xfs_remove(
 			goto out_bmap_cancel;
 
 		/*
-		 * Drop the link from dp to ip.
+		 * Drop the "." link from ip to self.
 		 */
 		error = xfs_droplink(tp, ip);
 		if (error)
@@ -2017,14 +1905,14 @@ xfs_remove(
 	} else {
 		/*
 		 * When removing a non-directory we need to log the parent
-		 * inode here for the i_gen update.  For a directory this is
-		 * done implicitly by the xfs_droplink call for the ".." entry.
+		 * inode here.  For a directory this is done implicitly
+		 * by the xfs_droplink call for the ".." entry.
 		 */
 		xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
 	}
 
 	/*
-	 * Drop the "." link from ip to self.
+	 * Drop the link from dp to ip.
 	 */
 	error = xfs_droplink(tp, ip);
 	if (error)
@@ -2178,7 +2066,6 @@ xfs_link(
 	if (error)
 		goto abort_return;
 	xfs_ichgtime(tdp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
-	tdp->i_gen++;
 	xfs_trans_log_inode(tp, tdp, XFS_ILOG_CORE);
 
 	error = xfs_bumplink(tp, sip);
@@ -2355,18 +2242,10 @@ xfs_mkdir(
 	}
 	xfs_ichgtime(dp, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
 
-	/*
-	 * Bump the in memory version number of the parent directory
-	 * so that other processes accessing it will recognize that
-	 * the directory has changed.
-	 */
-	dp->i_gen++;
-
 	error = xfs_dir_init(tp, cdp, dp);
 	if (error)
 		goto error2;
 
-	cdp->i_gen = 1;
 	error = xfs_bumplink(tp, dp);
 	if (error)
 		goto error2;
@@ -2653,13 +2532,6 @@ xfs_symlink(
 	xfs_trans_log_inode(tp, dp, XFS_ILOG_CORE);
 
 	/*
-	 * Bump the in memory version number of the parent directory
-	 * so that other processes accessing it will recognize that
-	 * the directory has changed.
-	 */
-	dp->i_gen++;
-
-	/*
 	 * If this is a synchronous mount, make sure that the
 	 * symlink transaction goes to disk before returning to
 	 * the user.
@@ -2809,7 +2681,7 @@ xfs_reclaim(
 		return 0;
 	}
 
-	vn_iowait(ip);
+	xfs_ioend_wait(ip);
 
 	ASSERT(XFS_FORCED_SHUTDOWN(ip->i_mount) || ip->i_delayed_blks == 0);
 
@@ -2833,122 +2705,10 @@ xfs_reclaim(
 	if (!ip->i_update_core && (ip->i_itemp == NULL)) {
 		xfs_ilock(ip, XFS_ILOCK_EXCL);
 		xfs_iflock(ip);
-		return xfs_finish_reclaim(ip, 1, XFS_IFLUSH_DELWRI_ELSE_SYNC);
-	} else {
-		xfs_mount_t	*mp = ip->i_mount;
-
-		/* Protect sync and unpin from us */
-		XFS_MOUNT_ILOCK(mp);
-		spin_lock(&ip->i_flags_lock);
-		__xfs_iflags_set(ip, XFS_IRECLAIMABLE);
-		VFS_I(ip)->i_private = NULL;
-		ip->i_vnode = NULL;
-		spin_unlock(&ip->i_flags_lock);
-		list_add_tail(&ip->i_reclaim, &mp->m_del_inodes);
-		XFS_MOUNT_IUNLOCK(mp);
-	}
-	return 0;
-}
-
-int
-xfs_finish_reclaim(
-	xfs_inode_t	*ip,
-	int		locked,
-	int		sync_mode)
-{
-	xfs_perag_t	*pag = xfs_get_perag(ip->i_mount, ip->i_ino);
-	struct inode	*vp = VFS_I(ip);
-
-	if (vp && VN_BAD(vp))
-		goto reclaim;
-
-	/* The hash lock here protects a thread in xfs_iget_core from
-	 * racing with us on linking the inode back with a vnode.
-	 * Once we have the XFS_IRECLAIM flag set it will not touch
-	 * us.
-	 */
-	write_lock(&pag->pag_ici_lock);
-	spin_lock(&ip->i_flags_lock);
-	if (__xfs_iflags_test(ip, XFS_IRECLAIM) ||
-	    (!__xfs_iflags_test(ip, XFS_IRECLAIMABLE) && vp == NULL)) {
-		spin_unlock(&ip->i_flags_lock);
-		write_unlock(&pag->pag_ici_lock);
-		if (locked) {
-			xfs_ifunlock(ip);
-			xfs_iunlock(ip, XFS_ILOCK_EXCL);
-		}
-		return 1;
-	}
-	__xfs_iflags_set(ip, XFS_IRECLAIM);
-	spin_unlock(&ip->i_flags_lock);
-	write_unlock(&pag->pag_ici_lock);
-	xfs_put_perag(ip->i_mount, pag);
-
-	/*
-	 * If the inode is still dirty, then flush it out.  If the inode
-	 * is not in the AIL, then it will be OK to flush it delwri as
-	 * long as xfs_iflush() does not keep any references to the inode.
-	 * We leave that decision up to xfs_iflush() since it has the
-	 * knowledge of whether it's OK to simply do a delwri flush of
-	 * the inode or whether we need to wait until the inode is
-	 * pulled from the AIL.
-	 * We get the flush lock regardless, though, just to make sure
-	 * we don't free it while it is being flushed.
-	 */
-	if (!locked) {
-		xfs_ilock(ip, XFS_ILOCK_EXCL);
-		xfs_iflock(ip);
+		xfs_iflags_set(ip, XFS_IRECLAIMABLE);
+		return xfs_reclaim_inode(ip, 1, XFS_IFLUSH_DELWRI_ELSE_SYNC);
 	}
-
-	/*
-	 * In the case of a forced shutdown we rely on xfs_iflush() to
-	 * wait for the inode to be unpinned before returning an error.
-	 */
-	if (xfs_iflush(ip, sync_mode) == 0) {
-		/* synchronize with xfs_iflush_done */
-		xfs_iflock(ip);
-		xfs_ifunlock(ip);
-	}
-
-	xfs_iunlock(ip, XFS_ILOCK_EXCL);
-
- reclaim:
-	xfs_ireclaim(ip);
-	return 0;
-}
-
-int
-xfs_finish_reclaim_all(xfs_mount_t *mp, int noblock)
-{
-	int		purged;
-	xfs_inode_t	*ip, *n;
-	int		done = 0;
-
-	while (!done) {
-		purged = 0;
-		XFS_MOUNT_ILOCK(mp);
-		list_for_each_entry_safe(ip, n, &mp->m_del_inodes, i_reclaim) {
-			if (noblock) {
-				if (xfs_ilock_nowait(ip, XFS_ILOCK_EXCL) == 0)
-					continue;
-				if (xfs_ipincount(ip) ||
-				    !xfs_iflock_nowait(ip)) {
-					xfs_iunlock(ip, XFS_ILOCK_EXCL);
-					continue;
-				}
-			}
-			XFS_MOUNT_IUNLOCK(mp);
-			if (xfs_finish_reclaim(ip, noblock,
-					XFS_IFLUSH_DELWRI_ELSE_ASYNC))
-				delay(1);
-			purged = 1;
-			break;
-		}
-
-		done = !purged;
-	}
-
-	XFS_MOUNT_IUNLOCK(mp);
+	xfs_inode_set_reclaim_tag(ip);
 	return 0;
 }
 
@@ -3197,6 +2957,8 @@ xfs_zero_remaining_bytes(
 	bp = xfs_buf_get_noaddr(mp->m_sb.sb_blocksize,
 				XFS_IS_REALTIME_INODE(ip) ?
 				mp->m_rtdev_targp : mp->m_ddev_targp);
+	if (!bp)
+		return XFS_ERROR(ENOMEM);
 
 	for (offset = startoff; offset <= endoff; offset = lastoffset + 1) {
 		offset_fsb = XFS_B_TO_FSBT(mp, offset);
@@ -3312,7 +3074,8 @@ xfs_free_file_space(
 		need_iolock = 0;
 	if (need_iolock) {
 		xfs_ilock(ip, XFS_IOLOCK_EXCL);
-		vn_iowait(ip);	/* wait for the completion of any pending DIOs */
+		/* wait for the completion of any pending DIOs */
+		xfs_ioend_wait(ip);
 	}
 
 	rounding = max_t(uint, 1 << mp->m_sb.sb_blocklog, PAGE_CACHE_SIZE);
@@ -3474,7 +3237,6 @@ xfs_change_file_space(
 	int		cmd,
 	xfs_flock64_t	*bf,
 	xfs_off_t	offset,
-	cred_t		*credp,
 	int		attr_flags)
 {
 	xfs_mount_t	*mp = ip->i_mount;
@@ -3562,7 +3324,7 @@ xfs_change_file_space(
 		iattr.ia_valid = ATTR_SIZE;
 		iattr.ia_size = startoffset;
 
-		error = xfs_setattr(ip, &iattr, attr_flags, credp);
+		error = xfs_setattr(ip, &iattr, attr_flags);
 
 		if (error)
 			return error;
diff --git a/fs/xfs/xfs_vnodeops.h b/fs/xfs/xfs_vnodeops.h
index 7b0c2ab8833..76df328c61b 100644
--- a/fs/xfs/xfs_vnodeops.h
+++ b/fs/xfs/xfs_vnodeops.h
@@ -14,9 +14,7 @@ struct xfs_inode;
 struct xfs_iomap;
 
 
-int xfs_open(struct xfs_inode *ip);
-int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags,
-		cred_t *credp);
+int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags);
 #define	XFS_ATTR_DMI		0x01	/* invocation from a DMI function */
 #define	XFS_ATTR_NONBLOCK	0x02	/* return EAGAIN if operation would block */
 #define XFS_ATTR_NOLOCK		0x04	/* Don't grab any conflicting locks */
@@ -44,8 +42,7 @@ int xfs_inode_flush(struct xfs_inode *ip, int flags);
 int xfs_set_dmattrs(struct xfs_inode *ip, u_int evmask, u_int16_t state);
 int xfs_reclaim(struct xfs_inode *ip);
 int xfs_change_file_space(struct xfs_inode *ip, int cmd,
-		xfs_flock64_t *bf, xfs_off_t offset,
-		cred_t *credp, int	attr_flags);
+		xfs_flock64_t *bf, xfs_off_t offset, int attr_flags);
 int xfs_rename(struct xfs_inode *src_dp, struct xfs_name *src_name,
 		struct xfs_inode *src_ip, struct xfs_inode *target_dp,
 		struct xfs_name *target_name, struct xfs_inode *target_ip);
@@ -56,8 +53,6 @@ int xfs_attr_set(struct xfs_inode *dp, const char *name, char *value,
 int xfs_attr_remove(struct xfs_inode *dp, const char *name, int flags);
 int xfs_attr_list(struct xfs_inode *dp, char *buffer, int bufsize,
 		int flags, struct attrlist_cursor_kern *cursor);
-int xfs_ioctl(struct xfs_inode *ip, struct file *filp,
-		int ioflags, unsigned int cmd, void __user *arg);
 ssize_t xfs_read(struct xfs_inode *ip, struct kiocb *iocb,
 		const struct iovec *iovp, unsigned int segs,
 		loff_t *offset, int ioflags);
@@ -78,5 +73,6 @@ int xfs_flushinval_pages(struct xfs_inode *ip, xfs_off_t first,
 		xfs_off_t last, int fiopt);
 int xfs_flush_pages(struct xfs_inode *ip, xfs_off_t first,
 		xfs_off_t last, uint64_t flags, int fiopt);
+int xfs_wait_on_pages(struct xfs_inode *ip, xfs_off_t first, xfs_off_t last);
 
 #endif /* _XFS_VNODEOPS_H */