trylock_super(): replacement for grab_super_passive()

I've noticed significant locking contention in memory reclaimer around
sb_lock inside grab_super_passive(). Grab_super_passive() is called from
two places: in icache/dcache shrinkers (function super_cache_scan) and
from writeback (function __writeback_inodes_wb). Both are required for
progress in memory allocator.

Grab_super_passive() acquires sb_lock to increment sb->s_count and check
sb->s_instances. It seems sb->s_umount locked for read is enough here:
super-block deactivation always runs under sb->s_umount locked for write.
Protecting super-block itself isn't a problem: in super_cache_scan() sb
is protected by shrinker_rwsem: it cannot be freed if its slab shrinkers
are still active. Inside writeback super-block comes from inode from bdi
writeback list under wb->list_lock.

This patch removes locking sb_lock and checks s_instances under s_umount:
generic_shutdown_super() unlinks it under sb->s_umount locked for write.
New variant is called trylock_super() and since it only locks semaphore,
callers must call up_read(&sb->s_umount) instead of drop_super(sb) when
they're done.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

fs/fs-writeback.c

@@ -769,17 +769,17 @@ static long __writeback_inodes_wb(struct bdi_writeback *wb,
 		struct inode *inode = wb_inode(wb->b_io.prev);
 		struct super_block *sb = inode->i_sb;
 
-		if (!grab_super_passive(sb)) {
+		if (!trylock_super(sb)) {
 			/*
-			 * grab_super_passive() may fail consistently due to
+			 * trylock_super() may fail consistently due to
 			 * s_umount being grabbed by someone else. Don't use
 			 * requeue_io() to avoid busy retrying the inode/sb.
 			 */
 			redirty_tail(inode, wb);
 			continue;
 		}
 		wrote += writeback_sb_inodes(sb, wb, work);
-		drop_super(sb);
+		up_read(&sb->s_umount);
 
 		/* refer to the same tests at the end of writeback_sb_inodes */
 		if (wrote) {

fs/internal.h

@@ -84,7 +84,7 @@ extern struct file *get_empty_filp(void);
  * super.c
  */
 extern int do_remount_sb(struct super_block *, int, void *, int);
-extern bool grab_super_passive(struct super_block *sb);
+extern bool trylock_super(struct super_block *sb);
 extern struct dentry *mount_fs(struct file_system_type *,
 			       int, const char *, void *);
 extern struct super_block *user_get_super(dev_t);

fs/super.c

@@ -71,7 +71,7 @@ static unsigned long super_cache_scan(struct shrinker *shrink,
 	if (!(sc->gfp_mask & __GFP_FS))
 		return SHRINK_STOP;
 
-	if (!grab_super_passive(sb))
+	if (!trylock_super(sb))
 		return SHRINK_STOP;
 
 	if (sb->s_op->nr_cached_objects)
@@ -105,7 +105,7 @@ static unsigned long super_cache_scan(struct shrinker *shrink,
 		freed += sb->s_op->free_cached_objects(sb, sc);
 	}
 
-	drop_super(sb);
+	up_read(&sb->s_umount);
 	return freed;
 }
 
@@ -118,7 +118,7 @@ static unsigned long super_cache_count(struct shrinker *shrink,
 	sb = container_of(shrink, struct super_block, s_shrink);
 
 	/*
-	 * Don't call grab_super_passive as it is a potential
+	 * Don't call trylock_super as it is a potential
 	 * scalability bottleneck. The counts could get updated
 	 * between super_cache_count and super_cache_scan anyway.
 	 * Call to super_cache_count with shrinker_rwsem held
@@ -348,35 +348,31 @@ static int grab_super(struct super_block *s) __releases(sb_lock)
 }
 
 /*
- *	grab_super_passive - acquire a passive reference
+ *	trylock_super - try to grab ->s_umount shared
  *	@sb: reference we are trying to grab
  *
- *	Tries to acquire a passive reference. This is used in places where we
+ *	Try to prevent fs shutdown.  This is used in places where we
  *	cannot take an active reference but we need to ensure that the
- *	superblock does not go away while we are working on it. It returns
- *	false if a reference was not gained, and returns true with the s_umount
- *	lock held in read mode if a reference is gained. On successful return,
- *	the caller must drop the s_umount lock and the passive reference when
- *	done.
+ *	filesystem is not shut down while we are working on it. It returns
+ *	false if we cannot acquire s_umount or if we lose the race and
+ *	filesystem already got into shutdown, and returns true with the s_umount
+ *	lock held in read mode in case of success. On successful return,
+ *	the caller must drop the s_umount lock when done.
+ *
+ *	Note that unlike get_super() et.al. this one does *not* bump ->s_count.
+ *	The reason why it's safe is that we are OK with doing trylock instead
+ *	of down_read().  There's a couple of places that are OK with that, but
+ *	it's very much not a general-purpose interface.
  */
-bool grab_super_passive(struct super_block *sb)
+bool trylock_super(struct super_block *sb)
 {
-	spin_lock(&sb_lock);
-	if (hlist_unhashed(&sb->s_instances)) {
-		spin_unlock(&sb_lock);
-		return false;
-	}
-
-	sb->s_count++;
-	spin_unlock(&sb_lock);
-
 	if (down_read_trylock(&sb->s_umount)) {
-		if (sb->s_root && (sb->s_flags & MS_BORN))
+		if (!hlist_unhashed(&sb->s_instances) &&
+		    sb->s_root && (sb->s_flags & MS_BORN))
 			return true;
 		up_read(&sb->s_umount);
 	}
 
-	put_super(sb);
 	return false;
 }