We need to check for s_instances to make sure we don't bother working
against a filesystem that is beeing unmounted, and we need to call
put_super to make sure a superblock is freed when we race against
umount. Also no need to keep sb_lock after we got a reference on it.
Signed-off-by: Christoph Hellwig <hch@xxxxxx>
---
Note: this is against Jens' tree with the writeback changes applied.
The logic in mainline is the same, but the context is slightl different.
Index: linux-2.6/fs/fs-writeback.c
===================================================================
--- linux-2.6.orig/fs/fs-writeback.c 2010-06-09 13:15:29.152004893 +0200
+++ linux-2.6/fs/fs-writeback.c 2010-06-09 13:15:29.946253462 +0200
@@ -534,19 +534,21 @@ select_queue:
static bool pin_sb_for_writeback(struct super_block *sb)
{
spin_lock(&sb_lock);
+ if (list_empty(&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) {
- spin_unlock(&sb_lock);
+ if (sb->s_root)
return true;
- }
- /*
- * umounted, drop rwsem again and fall through to failure
- */
up_read(&sb->s_umount);
}
- sb->s_count--;
- spin_unlock(&sb_lock);
+
+ put_super(sb);
return false;
}
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
