mm/hmm: fix bad subpage pointer in try_to_unmap_one
authorRalph Campbell <[email protected]>
Tue, 13 Aug 2019 22:37:11 +0000 (15:37 -0700)
committerLinus Torvalds <[email protected]>
Tue, 13 Aug 2019 23:06:52 +0000 (16:06 -0700)
When migrating an anonymous private page to a ZONE_DEVICE private page,
the source page->mapping and page->index fields are copied to the
destination ZONE_DEVICE struct page and the page_mapcount() is
increased.  This is so rmap_walk() can be used to unmap and migrate the
page back to system memory.

However, try_to_unmap_one() computes the subpage pointer from a swap pte
which computes an invalid page pointer and a kernel panic results such
as:

  BUG: unable to handle page fault for address: ffffea1fffffffc8

Currently, only single pages can be migrated to device private memory so
no subpage computation is needed and it can be set to "page".

[[email protected]: add comment]
Link: http://lkml.kernel.org/r/[email protected]
Link: http://lkml.kernel.org/r/[email protected]
Fixes: a5430dda8a3a1c ("mm/migrate: support un-addressable ZONE_DEVICE page in migration")
Signed-off-by: Ralph Campbell <[email protected]>
Cc: "Jérôme Glisse" <[email protected]>
Cc: "Kirill A. Shutemov" <[email protected]>
Cc: Mike Kravetz <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: Jason Gunthorpe <[email protected]>
Cc: John Hubbard <[email protected]>
Cc: Andrea Arcangeli <[email protected]>
Cc: Andrey Ryabinin <[email protected]>
Cc: Christoph Lameter <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Ira Weiny <[email protected]>
Cc: Jan Kara <[email protected]>
Cc: Lai Jiangshan <[email protected]>
Cc: Logan Gunthorpe <[email protected]>
Cc: Martin Schwidefsky <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: Mel Gorman <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: Pekka Enberg <[email protected]>
Cc: Randy Dunlap <[email protected]>
Cc: Vlastimil Babka <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
mm/rmap.c

index e5dfe2ae6b0d5dfc4a0b38c58c8e2c762b0b7696..003377e2423232614525ae2e6cbdda9695aca75b 100644 (file)
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -1475,7 +1475,15 @@ static bool try_to_unmap_one(struct page *page, struct vm_area_struct *vma,
                        /*
                         * No need to invalidate here it will synchronize on
                         * against the special swap migration pte.
+                        *
+                        * The assignment to subpage above was computed from a
+                        * swap PTE which results in an invalid pointer.
+                        * Since only PAGE_SIZE pages can currently be
+                        * migrated, just set it to page. This will need to be
+                        * changed when hugepage migrations to device private
+                        * memory are supported.
                         */
+                       subpage = page;
                        goto discard;
                }