From bc658c96037fc87463f0703ad2ea7c895344cb7e Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Sun, 29 May 2011 10:33:44 +0200 Subject: [PATCH] mm, rmap: Add yet more comments to page_get_anon_vma/page_lock_anon_vma Inspired by an analysis from Hugh on why again all this doesn't explode in our face. Signed-off-by: Peter Zijlstra Signed-off-by: Linus Torvalds --- mm/rmap.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/mm/rmap.c b/mm/rmap.c index 6bada99cd61c..0eb463ea88dd 100644 --- a/mm/rmap.c +++ b/mm/rmap.c @@ -352,6 +352,11 @@ void __init anon_vma_init(void) * The page might have been remapped to a different anon_vma or the anon_vma * returned may already be freed (and even reused). * + * In case it was remapped to a different anon_vma, the new anon_vma will be a + * child of the old anon_vma, and the anon_vma lifetime rules will therefore + * ensure that any anon_vma obtained from the page will still be valid for as + * long as we observe page_mapped() [ hence all those page_mapped() tests ]. + * * All users of this function must be very careful when walking the anon_vma * chain and verify that the page in question is indeed mapped in it * [ something equivalent to page_mapped_in_vma() ]. @@ -421,7 +426,7 @@ struct anon_vma *page_lock_anon_vma(struct page *page) /* * If the page is still mapped, then this anon_vma is still * its anon_vma, and holding the mutex ensures that it will - * not go away, see __put_anon_vma(). + * not go away, see anon_vma_free(). */ if (!page_mapped(page)) { mutex_unlock(&root_anon_vma->mutex); -- 2.34.1