On Tue, Nov 06, 2018 at 10:55:24AM +0100, Michal Hocko wrote:
> From: Michal Hocko <mhocko@xxxxxxxx>
>
> Page state checks are racy. Under a heavy memory workload (e.g. stress
> -m 200 -t 2h) it is quite easy to hit a race window when the page is
> allocated but its state is not fully populated yet. A debugging patch to
> dump the struct page state shows
> : [ 476.575516] has_unmovable_pages: pfn:0x10dfec00, found:0x1, count:0x0
> : [ 476.582103] page:ffffea0437fb0000 count:1 mapcount:1 mapping:ffff880e05239841 index:0x7f26e5000 compound_mapcount: 1
> : [ 476.592645] flags: 0x5fffffc0090034(uptodate|lru|active|head|swapbacked)
>
> Note that the state has been checked for both PageLRU and PageSwapBacked
> already. Closing this race completely would require some sort of retry
> logic. This can be tricky and error prone (think of potential endless
> or long taking loops).
>
> Workaround this problem for movable zones at least. Such a zone should
> only contain movable pages. 15c30bc09085 ("mm, memory_hotplug: make
> has_unmovable_pages more robust") has told us that this is not strictly
> true though. Bootmem pages should be marked reserved though so we can
> move the original check after the PageReserved check. Pages from other
> zones are still prone to races but we even do not pretend that memory
> hotremove works for those so pre-mature failure doesn't hurt that much.
>
> Reported-and-tested-by: Baoquan He <bhe@xxxxxxxxxx>
> Acked-by: Baoquan He <bhe@xxxxxxxxxx>
> Fixes: "mm, memory_hotplug: make has_unmovable_pages more robust")
> Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
> ---
>
> Hi,
> this has been reported [1] and we have tried multiple things to address
> the issue. The only reliable way was to reintroduce the movable zone
> check into has_unmovable_pages. This time it should be safe also for
> the bug originally fixed by 15c30bc09085.
>
> [1] http://lkml.kernel.org/r/20181101091055.GA15166@MiWiFi-R3L-srv
> mm/page_alloc.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 863d46da6586..c6d900ee4982 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -7788,6 +7788,14 @@ bool has_unmovable_pages(struct zone *zone, struct page *page, int count,
> if (PageReserved(page))
> goto unmovable;
>
> + /*
> + * If the zone is movable and we have ruled out all reserved
> + * pages then it should be reasonably safe to assume the rest
> + * is movable.
> + */
> + if (zone_idx(zone) == ZONE_MOVABLE)
> + continue;
> +
> /*
There is a WARN_ON() in case of failure at the end of the routine,
is that triggered when we hit the bug? If we're adding this patch,
the WARN_ON needs to go as well.
The check seems to be quite aggressive and in a loop that iterates
pages, but has nothing to do with the page, did you mean to make
the check
zone_idx(page_zone(page)) == ZONE_MOVABLE
it also skips all checks for pinned pages and other checks
Balbir Singh.
