Folding the per-cpu counters can yield a negative value in case of accounting races between CPUs. When collecting the dirty info, the code would read those sums into an unsigned variable and then check for it being negative, which can not work. Instead, fold the counters into a signed local variable, make the check, and only then assign it. This way, the function signals correctly when there are insane values instead of leaking them out to the caller. Signed-off-by: Johannes Weiner <hannes@xxxxxxxxxxx> --- mm/memcontrol.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -1261,14 +1261,15 @@ bool mem_cgroup_dirty_info(unsigned long (dirty_param.dirty_background_ratio * available_mem) / 100; - info->nr_reclaimable = - mem_cgroup_page_stat(MEMCG_NR_RECLAIM_PAGES); - if (info->nr_reclaimable < 0) + value = mem_cgroup_page_stat(MEMCG_NR_RECLAIM_PAGES); + if (value < 0) return false; + info->nr_reclaimable = value; - info->nr_writeback = mem_cgroup_page_stat(MEMCG_NR_WRITEBACK); - if (info->nr_writeback < 0) + value = mem_cgroup_page_stat(MEMCG_NR_WRITEBACK); + if (value < 0) return false; + info->nr_writeback = value; return true; } -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxxx For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>