dm cache: fix race affecting dirty block count

nr_dirty is updated without locking, causing it to drift so that it is
non-zero (either a small positive integer, or a very large one when an
underflow occurs) even when there are no actual dirty blocks.  This was
due to a race between the workqueue and map function accessing nr_dirty
in parallel without proper protection.

People were seeing under runs due to a race on increment/decrement of
nr_dirty, see: https://lkml.org/lkml/2014/6/3/648

Fix this by using an atomic_t for nr_dirty.

Reported-by: roma1390@gmail.com
Signed-off-by: Anssi Hannula <anssi.hannula@iki.fi>
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Cc: stable@vger.kernel.org

1 files changed, 6 insertions, 7 deletions

diff --git a/drivers/md/dm-cache-target.c b/drivers/md/dm-cache-target.c
index 5f054c44b485..2c63326638b6 100644
--- a/drivers/md/dm-cache-target.c
+++ b/drivers/md/dm-cache-target.c
@@ -231,7 +231,7 @@ struct cache {
         /*
          * cache_size entries, dirty if set
          */
-        dm_cblock_t nr_dirty;
+        atomic_t nr_dirty;
         unsigned long *dirty_bitset;
 
         /*
@@ -492,7 +492,7 @@ static bool is_dirty(struct cache *cache, dm_cblock_t b)
 static void set_dirty(struct cache *cache, dm_oblock_t oblock, dm_cblock_t cblock)
 {
         if (!test_and_set_bit(from_cblock(cblock), cache->dirty_bitset)) {
-                cache->nr_dirty = to_cblock(from_cblock(cache->nr_dirty) + 1);
+                atomic_inc(&cache->nr_dirty);
                 policy_set_dirty(cache->policy, oblock);
         }
 }
@@ -501,8 +501,7 @@ static void clear_dirty(struct cache *cache, dm_oblock_t oblock, dm_cblock_t cbl
 {
         if (test_and_clear_bit(from_cblock(cblock), cache->dirty_bitset)) {
                 policy_clear_dirty(cache->policy, oblock);
-                cache->nr_dirty = to_cblock(from_cblock(cache->nr_dirty) - 1);
-                if (!from_cblock(cache->nr_dirty))
+                if (atomic_dec_return(&cache->nr_dirty) == 0)
                         dm_table_event(cache->ti->table);
         }
 }
@@ -2269,7 +2268,7 @@ static int cache_create(struct cache_args *ca, struct cache **result)
         atomic_set(&cache->quiescing_ack, 0);
 
         r = -ENOMEM;
-        cache->nr_dirty = 0;
+        atomic_set(&cache->nr_dirty, 0);
         cache->dirty_bitset = alloc_bitset(from_cblock(cache->cache_size));
         if (!cache->dirty_bitset) {
                 *error = "could not allocate dirty bitset";
@@ -2808,7 +2807,7 @@ static void cache_status(struct dm_target *ti, status_type_t type,
 
                 residency = policy_residency(cache->policy);
 
-                DMEMIT("%u %llu/%llu %u %llu/%llu %u %u %u %u %u %u %llu ",
+                DMEMIT("%u %llu/%llu %u %llu/%llu %u %u %u %u %u %u %lu ",
                        (unsigned)(DM_CACHE_METADATA_BLOCK_SIZE >> SECTOR_SHIFT),
                        (unsigned long long)(nr_blocks_metadata - nr_free_blocks_metadata),
                        (unsigned long long)nr_blocks_metadata,
@@ -2821,7 +2820,7 @@ static void cache_status(struct dm_target *ti, status_type_t type,
                        (unsigned) atomic_read(&cache->stats.write_miss),
                        (unsigned) atomic_read(&cache->stats.demotion),
                        (unsigned) atomic_read(&cache->stats.promotion),
-                       (unsigned long long) from_cblock(cache->nr_dirty));
+                       (unsigned long) atomic_read(&cache->nr_dirty));
 
                 if (writethrough_mode(&cache->features))
                         DMEMIT("1 writethrough ");