tmpfs: fix highmem swapoff crash regression

Commit 778dd893ae78 ("tmpfs: fix race between umount and swapoff")
forgot the new rules for strict atomic kmap nesting, causing

  WARNING: at arch/x86/mm/highmem_32.c:81

from __kunmap_atomic(), then

  BUG: unable to handle kernel paging request at fffb9000

from shmem_swp_set() when shmem_unuse_inode() is handling swapoff with
highmem in use.  My disgrace again.

See
  https://bugzilla.kernel.org/show_bug.cgi?id=35352

Reported-by: Witold Baryluk <baryluk@smp.if.uj.edu.pl>
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/mm/shmem.c b/mm/shmem.c
index dfc7069..ba4ad28 100644 (file)
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -916,11 +916,12 @@ static int shmem_unuse_inode(struct shmem_inode_info *info, swp_entry_t entry, s
                        if (size > ENTRIES_PER_PAGE)
                                size = ENTRIES_PER_PAGE;
                        offset = shmem_find_swp(entry, ptr, ptr+size);
+                       shmem_swp_unmap(ptr);
                        if (offset >= 0) {
                                shmem_dir_unmap(dir);
+                               ptr = shmem_swp_map(subdir);
                                goto found;
                        }
-                       shmem_swp_unmap(ptr);
                }
        }
 lost1: