mm: vmpressure: fix sending wrong events on underflow
authorVinayak Menon <vinmenon@codeaurora.org>
Fri, 24 Feb 2017 22:59:39 +0000 (14:59 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 25 Feb 2017 01:46:56 +0000 (17:46 -0800)
At the end of a window period, if the reclaimed pages is greater than
scanned, an unsigned underflow can result in a huge pressure value and
thus a critical event.  Reclaimed pages is found to go higher than
scanned because of the addition of reclaimed slab pages to reclaimed in
shrink_node without a corresponding increment to scanned pages.

Minchan Kim mentioned that this can also happen in the case of a THP
page where the scanned is 1 and reclaimed could be 512.

Link: http://lkml.kernel.org/r/1486641577-11685-1-git-send-email-vinmenon@codeaurora.org
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@redhat.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Anton Vorontsov <anton.vorontsov@linaro.org>
Cc: Shiraz Hashim <shashim@codeaurora.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/vmpressure.c

index 149fdf6c5c56f927f3613538c61b3c5831c80af9..6063581f705c48b97a78a087d8c67127af9c8d68 100644 (file)
@@ -112,8 +112,15 @@ static enum vmpressure_levels vmpressure_calc_level(unsigned long scanned,
                                                    unsigned long reclaimed)
 {
        unsigned long scale = scanned + reclaimed;
-       unsigned long pressure;
+       unsigned long pressure = 0;
 
+       /*
+        * reclaimed can be greater than scanned in cases
+        * like THP, where the scanned is 1 and reclaimed
+        * could be 512
+        */
+       if (reclaimed >= scanned)
+               goto out;
        /*
         * We calculate the ratio (in percents) of how many pages were
         * scanned vs. reclaimed in a given time frame (window). Note that
@@ -124,6 +131,7 @@ static enum vmpressure_levels vmpressure_calc_level(unsigned long scanned,
        pressure = scale - (reclaimed * scale / scanned);
        pressure = pressure * 100 / scale;
 
+out:
        pr_debug("%s: %3lu  (s: %lu  r: %lu)\n", __func__, pressure,
                 scanned, reclaimed);