On 28.01.19 14:35, Michal Hocko wrote: > On Mon 28-01-19 14:22:52, David Hildenbrand wrote: >> On 28.01.19 14:21, Michal Hocko wrote: >>> On Mon 28-01-19 14:14:28, David Hildenbrand wrote: >>>> On 28.01.19 14:07, Michal Hocko wrote: >>>>> On Mon 28-01-19 13:16:09, David Hildenbrand wrote: >>>>> [...] >>>>>> My theory: >>>>>> >>>>>> In __unmap_and_move(), we lock the old and newpage and perform the >>>>>> migration. In case of vitio-balloon, the new page will become >>>>>> movable, the old page will no longer be movable. >>>>>> >>>>>> However, after unlocking newpage, I think there is nothing stopping >>>>>> the newpage from getting dequeued and freed by virtio-balloon. This >>>>>> will result in the newpage >>>>>> 1. No longer having PageMovable() >>>>>> 2. Getting moved to the local list before finally freeing it (using >>>>>> page->lru) >>>>> >>>>> Does that mean that the virtio-balloon can change the Movable state >>>>> while there are other users of the page? Can you point to the code that >>>>> does it? How come this can be safe at all? Or is the PageMovable stable >>>>> only under the page lock? >>>>> >>>> >>>> PageMovable is stable under the lock. The relevant instructions are in >>>> >>>> mm/balloon_compaction.c and include/linux/balloon_compaction.h >>> >>> OK, I have just checked __ClearPageMovable and it indeed requires >>> PageLock. Then we also have to move is_lru = __PageMovable(page) after >>> the page lock. >>> >> >> I assume that is fine as is as the page is isolated? (yes, it will be >> modified later when moving but we are interested in the original state) > > OK, I've missed that the page is indeed isolated. Then the patch makes > sense to me. > Thanks Michal. I assume this has broken ever since balloon compaction was introduced. I'll wait a little more and then resend as !RFC with a cc-stable tag. -- Thanks, David / dhildenb
