mm/shmem: use page_mapping() to detect page cache for uffd continue
authorPeter Xu <peterx@redhat.com>
Wed, 2 Nov 2022 18:41:52 +0000 (14:41 -0400)
committerAndrew Morton <akpm@linux-foundation.org>
Tue, 8 Nov 2022 23:57:23 +0000 (15:57 -0800)
commit93b0d9178743a68723babe8448981f658aebc58e
tree9bf23391a331ec7e206e56886c02e2490bb7bd8f
parent867400af90f1f953ff9e10b1b87ecaf9369a7eb8
mm/shmem: use page_mapping() to detect page cache for uffd continue

mfill_atomic_install_pte() checks page->mapping to detect whether one page
is used in the page cache.  However as pointed out by Matthew, the page
can logically be a tail page rather than always the head in the case of
uffd minor mode with UFFDIO_CONTINUE.  It means we could wrongly install
one pte with shmem thp tail page assuming it's an anonymous page.

It's not that clear even for anonymous page, since normally anonymous
pages also have page->mapping being setup with the anon vma.  It's safe
here only because the only such caller to mfill_atomic_install_pte() is
always passing in a newly allocated page (mcopy_atomic_pte()), whose
page->mapping is not yet setup.  However that's not extremely obvious
either.

For either of above, use page_mapping() instead.

Link: https://lkml.kernel.org/r/Y2K+y7wnhC4vbnP2@x1n
Fixes: 153132571f02 ("userfaultfd/shmem: support UFFDIO_CONTINUE for shmem")
Signed-off-by: Peter Xu <peterx@redhat.com>
Reported-by: Matthew Wilcox <willy@infradead.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
mm/userfaultfd.c