ceph: enable/disable dentry complete flags via mount option

Enable/disable use of the dentry dir 'complete' flag via a mount option.
This lets the admin control whether ceph uses the dcache to satisfy
negative lookups or readdir when it has the entire directory contents in
its cache.

This is purely a performance optimization; correctness is guaranteed
whether it is enabled or not.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sage Weil <sage@newdream.net>

4 files changed, 50 insertions, 8 deletions

diff --git a/Documentation/filesystems/ceph.txt b/Documentation/filesystems/ceph.txt
index 763d8ebbbeb..d6030aa3337 100644
--- a/Documentation/filesystems/ceph.txt
+++ b/Documentation/filesystems/ceph.txt
@@ -119,12 +119,20 @@ Mount Options
 	must rely on TCP's error correction to detect data corruption
 	in the data payload.
 
-  noasyncreaddir
-	Disable client's use its local cache to satisfy	readdir
-	requests.  (This does not change correctness; the client uses
-	cached metadata only when a lease or capability ensures it is
-	valid.)
+  dcache
+        Use the dcache contents to perform negative lookups and
+        readdir when the client has the entire directory contents in
+        its cache.  (This does not change correctness; the client uses
+        cached metadata only when a lease or capability ensures it is
+        valid.)
+
+  nodcache
+        Do not use the dcache as above.  This avoids a significant amount of
+        complex code, sacrificing performance without affecting correctness,
+        and is useful for tracking down bugs.
 
+  noasyncreaddir
+	Do not use the dcache as above for readdir.
 
 More Information
 ================
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 974ef1e4d26..5259abfb5dd 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -1094,17 +1094,36 @@ static int ceph_snapdir_d_revalidate(struct dentry *dentry,
  */
 void ceph_dir_set_complete(struct inode *inode)
 {
-	/* not yet implemented */
+	struct dentry *dentry = d_find_any_alias(inode);
+	
+	if (dentry && ceph_dentry(dentry) &&
+	    ceph_test_mount_opt(ceph_sb_to_client(dentry->d_sb), DCACHE)) {
+		dout(" marking %p (%p) complete\n", inode, dentry);
+		set_bit(CEPH_D_COMPLETE, &ceph_dentry(dentry)->flags);
+	}
+	dput(dentry);
 }
 
 void ceph_dir_clear_complete(struct inode *inode)
 {
-	/* not yet implemented */
+	struct dentry *dentry = d_find_any_alias(inode);
+
+	if (dentry && ceph_dentry(dentry)) {
+		dout(" marking %p (%p) complete\n", inode, dentry);
+		set_bit(CEPH_D_COMPLETE, &ceph_dentry(dentry)->flags);
+	}
+	dput(dentry);
 }
 
 bool ceph_dir_test_complete(struct inode *inode)
 {
-	/* not yet implemented */
+	struct dentry *dentry = d_find_any_alias(inode);
+
+	if (dentry && ceph_dentry(dentry)) {
+		dout(" marking %p (%p) NOT complete\n", inode, dentry);
+		clear_bit(CEPH_D_COMPLETE, &ceph_dentry(dentry)->flags);
+	}
+	dput(dentry);
 	return false;
 }
 
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index ec74313e901..9c62fe02ce0 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -131,6 +131,8 @@ enum {
 	Opt_rbytes,
 	Opt_norbytes,
 	Opt_noasyncreaddir,
+	Opt_dcache,
+	Opt_nodcache,
 	Opt_ino32,
 };
 
@@ -152,6 +154,8 @@ static match_table_t fsopt_tokens = {
 	{Opt_rbytes, "rbytes"},
 	{Opt_norbytes, "norbytes"},
 	{Opt_noasyncreaddir, "noasyncreaddir"},
+	{Opt_dcache, "dcache"},
+	{Opt_nodcache, "nodcache"},
 	{Opt_ino32, "ino32"},
 	{-1, NULL}
 };
@@ -231,6 +235,12 @@ static int parse_fsopt_token(char *c, void *private)
 	case Opt_noasyncreaddir:
 		fsopt->flags |= CEPH_MOUNT_OPT_NOASYNCREADDIR;
 		break;
+	case Opt_dcache:
+		fsopt->flags |= CEPH_MOUNT_OPT_DCACHE;
+		break;
+	case Opt_nodcache:
+		fsopt->flags &= ~CEPH_MOUNT_OPT_DCACHE;
+		break;
 	case Opt_ino32:
 		fsopt->flags |= CEPH_MOUNT_OPT_INO32;
 		break;
@@ -377,6 +387,10 @@ static int ceph_show_options(struct seq_file *m, struct vfsmount *mnt)
 		seq_puts(m, ",norbytes");
 	if (fsopt->flags & CEPH_MOUNT_OPT_NOASYNCREADDIR)
 		seq_puts(m, ",noasyncreaddir");
+	if (fsopt->flags & CEPH_MOUNT_OPT_DCACHE)
+		seq_puts(m, ",dcache");
+	else
+		seq_puts(m, ",nodcache");
 
 	if (fsopt->wsize)
 		seq_printf(m, ",wsize=%d", fsopt->wsize);
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index edcbf3774a5..140f99f978c 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -28,6 +28,7 @@
 #define CEPH_MOUNT_OPT_RBYTES          (1<<5) /* dir st_bytes = rbytes */
 #define CEPH_MOUNT_OPT_NOASYNCREADDIR  (1<<7) /* no dcache readdir */
 #define CEPH_MOUNT_OPT_INO32           (1<<8) /* 32 bit inos */
+#define CEPH_MOUNT_OPT_DCACHE          (1<<9) /* use dcache for readdir etc */
 
 #define CEPH_MOUNT_OPT_DEFAULT    (CEPH_MOUNT_OPT_RBYTES)