sysfs: allow sysfs_read to detect and drop removed disks

All operations that rely on loading from an existing container (like
--add) will fail after a disk has been removed.  Provide an option to
skip missing / offline disks rather than abort.  We attempt to do this
in the load_super_{imsm,ddf}_all cases when mdmon is running i.e. we
already have a consitent version of the metadata running in the system.
Otherwise, we fail as normal and let the administrator fix up the
container.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>

4 files changed, 53 insertions, 24 deletions

diff --git a/mdadm.h b/mdadm.h
index 5be3aff..f8f3685 100644
--- a/mdadm.h
+++ b/mdadm.h
@@ -334,22 +334,24 @@ extern int map_lock(struct map_ent **melp);
 extern void map_unlock(struct map_ent **melp);
 
 /* various details can be requested */
-#define	GET_LEVEL	1
-#define	GET_LAYOUT	2
-#define	GET_COMPONENT	4
-#define	GET_CHUNK	8
-#define GET_CACHE	16
-#define	GET_MISMATCH	32
-#define	GET_VERSION	64
-#define	GET_DISKS	128
-#define	GET_DEGRADED	256
-#define	GET_SAFEMODE	512
-
-#define	GET_DEVS	1024 /* gets role, major, minor */
-#define	GET_OFFSET	2048
-#define	GET_SIZE	4096
-#define	GET_STATE	8192
-#define	GET_ERROR	16384
+enum sysfs_read_flags {
+	GET_LEVEL	= (1 << 0),
+	GET_LAYOUT	= (1 << 1),
+	GET_COMPONENT	= (1 << 2),
+	GET_CHUNK	= (1 << 3),
+	GET_CACHE	= (1 << 4),
+	GET_MISMATCH	= (1 << 5),
+	GET_VERSION	= (1 << 6),
+	GET_DISKS	= (1 << 7),
+	GET_DEGRADED	= (1 << 8),
+	GET_SAFEMODE	= (1 << 9),
+	GET_DEVS	= (1 << 10), /* gets role, major, minor */
+	GET_OFFSET	= (1 << 11),
+	GET_SIZE	= (1 << 12),
+	GET_STATE	= (1 << 13),
+	GET_ERROR	= (1 << 14),
+	SKIP_GONE_DEVS	= (1 << 15),
+};
 
 /* If fd >= 0, get the array it is open on,
  * else use devnum. >=0 -> major9. <0.....
diff --git a/super-ddf.c b/super-ddf.c
index f621f4d..6c75998 100644
--- a/super-ddf.c
+++ b/super-ddf.c
@@ -2734,8 +2734,14 @@ static int load_super_ddf_all(struct supertype *st, int fd,
 	int seq;
 	char nm[20];
 	int dfd;
+	int devnum = fd2devnum(fd);
+	enum sysfs_read_flags flags;
 
-	sra = sysfs_read(fd, 0, GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE);
+	flags = GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE;
+	if (mdmon_running(devnum))
+		flags |= SKIP_GONE_DEVS;
+
+	sra = sysfs_read(fd, 0, flags);
 	if (!sra)
 		return 1;
 	if (sra->array.major_version != -1 ||
diff --git a/super-intel.c b/super-intel.c
index f55d707..dd69cb1 100644
--- a/super-intel.c
+++ b/super-intel.c
@@ -1976,9 +1976,14 @@ static int load_super_imsm_all(struct supertype *st, int fd, void **sbp,
 	int rv;
 	int devnum = fd2devnum(fd);
 	int retry;
+	enum sysfs_read_flags flags;
 
-	/* check if this disk is a member of an active array */
-	sra = sysfs_read(fd, 0, GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE);
+	flags = GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE;
+	if (mdmon_running(devnum))
+		flags |= SKIP_GONE_DEVS;
+
+	/* check if 'fd' an opened container */
+	sra = sysfs_read(fd, 0, flags);
 	if (!sra)
 		return 1;
 
diff --git a/sysfs.c b/sysfs.c
index b9fd3da..2dad7d3 100644
--- a/sysfs.c
+++ b/sysfs.c
@@ -272,18 +272,34 @@ struct mdinfo *sysfs_read(int fd, int devnum, unsigned long options)
 			}
 			
 		}
-		dev->next = sra->devs;
-		sra->devs = dev;
-
 		strcpy(dev->sys_name, de->d_name);
 		dev->disk.raid_disk = strtoul(buf, &ep, 10);
 		if (*ep) dev->disk.raid_disk = -1;
 
 		strcpy(dbase, "block/dev");
-		if (load_sys(fname, buf))
-			goto abort;
+		if (load_sys(fname, buf)) {
+			free(dev);
+			if (options & SKIP_GONE_DEVS)
+				continue;
+			else
+				goto abort;
+		}
 		sscanf(buf, "%d:%d", &dev->disk.major, &dev->disk.minor);
 
+		/* special case check for block devices that can go 'offline' */
+		if (options & SKIP_GONE_DEVS) {
+			strcpy(dbase, "block/device/state");
+			if (load_sys(fname, buf) == 0 &&
+			    strncmp(buf, "offline", 7) == 0) {
+				free(dev);
+				continue;
+			}
+		}
+
+		/* finally add this disk to the array */
+		dev->next = sra->devs;
+		sra->devs = dev;
+
 		if (options & GET_OFFSET) {
 			strcpy(dbase, "offset");
 			if (load_sys(fname, buf))