Re: ext4: the list debug reports about corruptions

[Theodore Ts'o <tytso@mit.edu>]

On Thu, Aug 08, 2013 at 01:45:52PM +0400, Andrey Vagin wrote:
> Hello,
> 
> I use akpm branch of linux-next
> https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/tree/?h=akpm&id=05abc43724793827835728f4f25ba827e1c54902
> .
> This bug is reproduced, when I do the following sequence of commands:
> 
> mkdir /sys/fs/cgroup/memory/xxx
> echo 1024000000 >  /sys/fs/cgroup/memory/xxx/memory.kmem.limit_in_bytes
> echo $$ > /sys/fs/cgroup/memory/xxx/tasks
> find /proc -name abra-kadabra
> echo $$ > /sys/fs/cgroup/memory/tasks
> rmdir /sys/fs/cgroup/memory/xxx/
> echo 3 > /proc/sys/vm/drop_caches

Thanks for reporting this.  I believe I have this fixed in the updated
dev branch for the ext4 tree.  The relevant fix is additition of the
INIT_LIST_HEAD(&skipped) below.  The rest is just a spelling fix up I
made along the way.

    	       	   	       	   	    	- Ted

diff --git a/fs/ext4/extents_status.c b/fs/ext4/extents_status.c
index 00e6589..0e88a36 100644
--- a/fs/ext4/extents_status.c
+++ b/fs/ext4/extents_status.c
@@ -946,7 +946,7 @@ static int __ext4_es_shrink(struct ext4_sb_info *sbi, int nr_to_scan,
 {
 	struct ext4_inode_info *ei;
 	struct list_head *cur, *tmp;
-	LIST_HEAD(skiped);
+	LIST_HEAD(skipped);
 	int ret, nr_shrunk = 0;
 	int retried = 0, skip_precached = 1, nr_skipped = 0;
 
@@ -972,7 +972,7 @@ retry:
 		    (skip_precached && ext4_test_inode_state(&ei->vfs_inode,
 						EXT4_STATE_EXT_PRECACHED))) {
 			nr_skipped++;
-			list_move_tail(cur, &skiped);
+			list_move_tail(cur, &skipped);
 			continue;
 		}
 
@@ -992,7 +992,8 @@ retry:
 	}
 
 	/* Move the newer inodes into the tail of the LRU list. */
-	list_splice_tail(&skiped, &sbi->s_es_lru);
+	list_splice_tail(&skipped, &sbi->s_es_lru);
+	INIT_LIST_HEAD(&skipped);
 
 	/*
 	 * If we skipped any inodes, and we weren't able to make any