dm bufio: fully initialize shrinker
authorGreg Thelen <gthelen@google.com>
Thu, 31 Jul 2014 16:07:19 +0000 (09:07 -0700)
committerMike Snitzer <snitzer@redhat.com>
Fri, 1 Aug 2014 16:07:21 +0000 (12:07 -0400)
1d3d4437eae1 ("vmscan: per-node deferred work") added a flags field to
struct shrinker assuming that all shrinkers were zero filled.  The dm
bufio shrinker is not zero filled, which leaves arbitrary kmalloc() data
in flags.  So far the only defined flags bit is SHRINKER_NUMA_AWARE.
But there are proposed patches which add other bits to shrinker.flags
(e.g. memcg awareness).

Rather than simply initializing the shrinker, this patch uses kzalloc()
when allocating the dm_bufio_client to ensure that the embedded shrinker
and any other similar structures are zeroed.

This fixes theoretical over aggressive shrinking of dm bufio objects.
If the uninitialized dm_bufio_client.shrinker.flags contains
SHRINKER_NUMA_AWARE then shrink_slab() would call the dm shrinker for
each numa node rather than just once.  This has been broken since 3.12.

Signed-off-by: Greg Thelen <gthelen@google.com>
Acked-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org # v3.12+
drivers/md/dm-bufio.c

index 4e84095833dbce97972318a59841a75db0f2b241..d724459860d9a54eb1934b8762504f6fcdbb437e 100644 (file)
@@ -1541,7 +1541,7 @@ struct dm_bufio_client *dm_bufio_client_create(struct block_device *bdev, unsign
        BUG_ON(block_size < 1 << SECTOR_SHIFT ||
               (block_size & (block_size - 1)));
 
-       c = kmalloc(sizeof(*c), GFP_KERNEL);
+       c = kzalloc(sizeof(*c), GFP_KERNEL);
        if (!c) {
                r = -ENOMEM;
                goto bad_client;