r16945: Sync trunk -> 3.0 for 3.0.24 code. Still need

to do the upper layer directories but this is what
everyone is waiting for....

Jeremy.
(This used to be commit 9dafb7f48ca3e7af956b0a7d1720c2546fc4cfb8)

1 files changed, 200 insertions, 0 deletions

diff --git a/source3/modules/vfs_cacheprime.c b/source3/modules/vfs_cacheprime.c
new file mode 100644
index 00000000000..196441c4dda
--- /dev/null
+++ b/source3/modules/vfs_cacheprime.c
@@ -0,0 +1,200 @@
+/*
+ * Copyright (c) James Peach 2005-2006
+ *
+ * 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; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will 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 to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include "includes.h"
+
+/* Cache priming module.
+ *
+ * The purpose of this module is to do RAID stripe width reads to prime the
+ * buffer cache to do zero-copy I/O for subsequent sendfile calls. The idea is
+ * to do a single large read at the start of the file to make sure that most or
+ * all of the file is pulled into the buffer cache. Subsequent I/Os have
+ * reduced latency.
+ *
+ * Tunables.
+ *
+ *      cacheprime:rsize    Amount of readahead in bytes. This should be a
+ *                          multiple of the RAID stripe width.
+ *      cacheprime:debug    Debug level at which to emit messages.
+ */
+
+#define READAHEAD_MIN       (128 * 1024)        /* min is 128 KiB */
+#define READAHEAD_MAX       (100 * 1024 * 1024) /* max is 100 MiB */
+
+#define MODULE "cacheprime"
+
+static int module_debug;
+static ssize_t g_readsz = 0;
+static void * g_readbuf = NULL;
+
+/* Prime the kernel buffer cache with data from the specified file. We use
+ * per-fsp data to make sure we only ever do this once. If pread is being
+ * emulated by seek/read/seek, when this will suck quite a lot.
+ */
+static BOOL prime_cache(
+            struct vfs_handle_struct *  handle,
+			files_struct *		        fsp,
+			int			                fd,
+			SMB_OFF_T		            offset,
+			size_t			            count)
+{
+        SMB_OFF_T * last;
+        ssize_t nread;
+
+        last = VFS_ADD_FSP_EXTENSION(handle, fsp, SMB_OFF_T);
+        if (!last) {
+                return False;
+        }
+
+        if (*last == -1) {
+            /* Readahead disabled. */
+            return False;
+        }
+
+        if ((*last + g_readsz) > (offset + count)) {
+            /* Skip readahead ... we've already been here. */
+            return False;
+        }
+
+        DEBUG(module_debug,
+            ("%s: doing readahead of %lld bytes at %lld for %s\n",
+            MODULE, (long long)g_readsz, (long long)*last,
+            fsp->fsp_name));
+
+        nread = sys_pread(fd, g_readbuf, g_readsz, *last);
+        if (nread < 0) {
+            *last = -1;
+            return False;
+        }
+
+        *last += nread;
+        return True;
+}
+
+static int cprime_connect(
+                struct vfs_handle_struct *  handle,
+                const char *                service,
+                const char *                user)
+{
+        module_debug = lp_parm_int(SNUM(handle->conn), MODULE, "debug", 100);
+        if (g_readbuf) {
+                /* Only allocate g_readbuf once. If the config changes and
+                 * another client multiplexes onto this smbd, we don't want
+                 * to risk memory corruption.
+                 */
+                return SMB_VFS_NEXT_CONNECT(handle, service, user);
+        }
+
+        g_readsz = conv_str_size(lp_parm_const_string(SNUM(handle->conn),
+                                        MODULE, "rsize", NULL));
+
+        if (g_readsz < READAHEAD_MIN) {
+                DEBUG(module_debug, ("%s: %ld bytes of readahead "
+                            "requested, using minimum of %u\n",
+                            MODULE, (long)g_readsz, READAHEAD_MIN));
+                g_readsz = READAHEAD_MIN;
+        } else if (g_readsz > READAHEAD_MAX) {
+                DEBUG(module_debug, ("%s: %ld bytes of readahead "
+                            "requested, using maximum of %u\n",
+                            MODULE, (long)g_readsz, READAHEAD_MAX));
+                g_readsz = READAHEAD_MAX;
+        }
+
+        if ((g_readbuf = SMB_MALLOC(g_readsz)) == NULL) {
+                /* Turn off readahead if we can't get a buffer. */
+                g_readsz = 0;
+        }
+
+        return SMB_VFS_NEXT_CONNECT(handle, service, user);
+}
+
+static ssize_t cprime_sendfile(
+                struct vfs_handle_struct *  handle,
+                int                         tofd,
+                files_struct *              fsp,
+                int                         fromfd,
+                const DATA_BLOB *           header,
+                SMB_OFF_T                   offset,
+                size_t                      count)
+{
+        if (g_readbuf && offset == 0) {
+                prime_cache(handle, fsp, fromfd, offset, count);
+        }
+
+        return SMB_VFS_NEXT_SENDFILE(handle, tofd, fsp, fromfd,
+                                     header, offset, count);
+}
+
+static ssize_t cprime_read(
+                vfs_handle_struct * handle,
+                files_struct *      fsp,
+                int                 fd,
+                void *              data,
+                size_t              count)
+{
+        SMB_OFF_T offset;
+
+        offset = SMB_VFS_LSEEK(fsp, fd, 0, SEEK_CUR);
+        if (offset >= 0 && g_readbuf)  {
+                prime_cache(handle, fsp, fd, offset, count);
+                SMB_VFS_LSEEK(fsp, fd, offset, SEEK_SET);
+        }
+
+        return SMB_VFS_NEXT_READ(handle, fsp, fd, data, count);
+}
+
+static ssize_t cprime_pread(
+                vfs_handle_struct * handle,
+                files_struct *      fsp,
+                int                 fd,
+                void *              data,
+		        size_t              count,
+                SMB_OFF_T           offset)
+{
+        if (g_readbuf) {
+                prime_cache(handle, fsp, fd, offset, count);
+        }
+
+        return SMB_VFS_NEXT_PREAD(handle, fsp, fd, data, count, offset);
+}
+
+static vfs_op_tuple cprime_ops [] =
+{
+        {SMB_VFS_OP(cprime_sendfile),
+                SMB_VFS_OP_SENDFILE, SMB_VFS_LAYER_TRANSPARENT},
+        {SMB_VFS_OP(cprime_pread),
+                SMB_VFS_OP_PREAD, SMB_VFS_LAYER_TRANSPARENT},
+        {SMB_VFS_OP(cprime_read),
+                SMB_VFS_OP_READ, SMB_VFS_LAYER_TRANSPARENT},
+        {SMB_VFS_OP(cprime_connect),
+                SMB_VFS_OP_CONNECT,  SMB_VFS_LAYER_TRANSPARENT},
+
+        {SMB_VFS_OP(NULL), SMB_VFS_OP_NOOP, SMB_VFS_LAYER_NOOP}
+};
+
+/* -------------------------------------------------------------------------
+ * Samba module initialisation entry point.
+ * -------------------------------------------------------------------------
+ */
+
+NTSTATUS vfs_cacheprime_init(void)
+{
+    return smb_register_vfs(SMB_VFS_INTERFACE_VERSION, MODULE, cprime_ops);
+}
+
+/* vim: set sw=4 ts=4 tw=79 et: */