On Sun, Aug 12, 2018 at 08:41:08PM -0700, Mike Kravetz wrote: > The page migration code employs try_to_unmap() to try and unmap the > source page. This is accomplished by using rmap_walk to find all > vmas where the page is mapped. This search stops when page mapcount > is zero. For shared PMD huge pages, the page map count is always 1 > not matter the number of mappings. Shared mappings are tracked via > the reference count of the PMD page. Therefore, try_to_unmap stops > prematurely and does not completely unmap all mappings of the source > page. > > This problem can result is data corruption as writes to the original > source page can happen after contents of the page are copied to the > target page. Hence, data is lost. > > This problem was originally seen as DB corruption of shared global > areas after a huge page was soft offlined. DB developers noticed > they could reproduce the issue by (hotplug) offlining memory used > to back huge pages. A simple testcase can reproduce the problem by > creating a shared PMD mapping (note that this must be at least > PUD_SIZE in size and PUD_SIZE aligned (1GB on x86)), and using > migrate_pages() to migrate process pages between nodes. > > To fix, have the try_to_unmap_one routine check for huge PMD sharing > by calling huge_pmd_unshare for hugetlbfs huge pages. If it is a > shared mapping it will be 'unshared' which removes the page table > entry and drops reference on PMD page. After this, flush caches and > TLB. > > Signed-off-by: Mike Kravetz <mike.kravetz@xxxxxxxxxx> > --- > I am not %100 sure on the required flushing, so suggestions would be > appreciated. This also should go to stable. It has been around for > a long time so still looking for an appropriate 'fixes:'. I believe we need flushing. And huge_pmd_unshare() usage in __unmap_hugepage_range() looks suspicious: I don't see how we flush TLB in that case. -- Kirill A. Shutemov
