mm/swap_slots.c: fix race conditions in swap_slots cache init

Memory allocations can happen before the swap_slots cache initialization
is completed during cpu bring up.  If we are low on memory, we could
call get_swap_page() and access swap_slots_cache before it is fully
initialized.

Add a check in get_swap_page() for initialized swap_slots_cache to
prevent this condition.  Similar check already exists in free_swap_slot.
Also annotate the checks to indicate the likely condition.

We also added a memory barrier to make sure that the locks
initialization are done before the assignment of cache->slots and
cache->slots_ret pointers.  This ensures the assumption that it is safe
to acquire the slots cache locks and use the slots cache when the
corresponding cache->slots or cache->slots_ret pointers are non null.

[akpm@linux-foundation.org: tidy up comment]
[akpm@linux-foundation.org: fix spello in comment]
Link: http://lkml.kernel.org/r/65a9d0f133f63e66bba37b53b2fd0464b7cae771.1500677066.git.tim.c.chen@linux.intel.com
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Reported-by: Wenwei Tao <wenwei.tww@alibaba-inc.com>
Acked-by: Ying Huang <ying.huang@intel.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/mm/swap_slots.c b/mm/swap_slots.c
index d81cfc5a43d5083c4d3eb7273bcf10e606c2eb79..bebc19292018f355b4e8f500e985c71e4ea3d9e4 100644 (file)
--- a/mm/swap_slots.c
+++ b/mm/swap_slots.c
@@ -149,6 +149,13 @@ static int alloc_swap_slot_cache(unsigned int cpu)
        cache->nr = 0;
        cache->cur = 0;
        cache->n_ret = 0;
+       /*
+        * We initialized alloc_lock and free_lock earlier.  We use
+        * !cache->slots or !cache->slots_ret to know if it is safe to acquire
+        * the corresponding lock and use the cache.  Memory barrier below
+        * ensures the assumption.
+        */
+       mb();
        cache->slots = slots;
        slots = NULL;
        cache->slots_ret = slots_ret;
@@ -275,7 +282,7 @@ int free_swap_slot(swp_entry_t entry)
        struct swap_slots_cache *cache;
 
        cache = raw_cpu_ptr(&swp_slots);
-       if (use_swap_slot_cache && cache->slots_ret) {
+       if (likely(use_swap_slot_cache && cache->slots_ret)) {
                spin_lock_irq(&cache->free_lock);
                /* Swap slots cache may be deactivated before acquiring lock */
                if (!use_swap_slot_cache || !cache->slots_ret) {
@@ -326,7 +333,7 @@ swp_entry_t get_swap_page(struct page *page)
         */
        cache = raw_cpu_ptr(&swp_slots);
 
-       if (check_cache_active()) {
+       if (likely(check_cache_active() && cache->slots)) {
                mutex_lock(&cache->alloc_lock);
                if (cache->slots) {
 repeat: