mm: fix hang on anon_vma->root->lock
authorHugh Dickins <hughd@google.com>
Thu, 26 Aug 2010 06:12:54 +0000 (23:12 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 28 Aug 2010 20:54:12 +0000 (13:54 -0700)
commitf18194275c39835cb84563500995e0d503a32d9a
treeec3221ba856e6580e844c4bff200c48f2618c6ac
parentd4348c678977c7093438bbbf2067c49396ae941b
mm: fix hang on anon_vma->root->lock

After several hours, kbuild tests hang with anon_vma_prepare() spinning on
a newly allocated anon_vma's lock - on a box with CONFIG_TREE_PREEMPT_RCU=y
(which makes this very much more likely, but it could happen without).

The ever-subtle page_lock_anon_vma() now needs a further twist: since
anon_vma_prepare() and anon_vma_fork() are liable to change the ->root
of a reused anon_vma structure at any moment, page_lock_anon_vma()
needs to check page_mapped() again before succeeding, otherwise
page_unlock_anon_vma() might address a different root->lock.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/rmap.c