Commit 52f37629fd3c7b24e1e6c125e665454cd7ac1acb
Committed by
Linus Torvalds
1 parent
f1cb08798e
Exists in
master
and in
20 other branches
THP: fix comment about memory barrier
Currently the memory barrier in __do_huge_pmd_anonymous_page doesn't work. Because lru_cache_add_lru uses pagevec so it could miss spinlock easily so above rule was broken so user might see inconsistent data. I was not first person who pointed out the problem. Mel and Peter pointed out a few months ago and Peter pointed out further that even spin_lock/unlock can't make sure of it: http://marc.info/?t=134333512700004 In particular: *A = a; LOCK UNLOCK *B = b; may occur as: LOCK, STORE *B, STORE *A, UNLOCK At last, Hugh pointed out that even we don't need memory barrier in there because __SetPageUpdate already have done it from Nick's commit 0ed361dec369 ("mm: fix PageUptodate data race") explicitly. So this patch fixes comment on THP and adds same comment for do_anonymous_page, too because everybody except Hugh was missing that. It means we need a comment about that. Signed-off-by: Minchan Kim <minchan@kernel.org> Acked-by: Andrea Arcangeli <aarcange@redhat.com> Acked-by: David Rientjes <rientjes@google.com> Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Showing 2 changed files with 10 additions and 6 deletions Side-by-side Diff
mm/huge_memory.c
... | ... | @@ -713,6 +713,11 @@ |
713 | 713 | return VM_FAULT_OOM; |
714 | 714 | |
715 | 715 | clear_huge_page(page, haddr, HPAGE_PMD_NR); |
716 | + /* | |
717 | + * The memory barrier inside __SetPageUptodate makes sure that | |
718 | + * clear_huge_page writes become visible before the set_pmd_at() | |
719 | + * write. | |
720 | + */ | |
716 | 721 | __SetPageUptodate(page); |
717 | 722 | |
718 | 723 | spin_lock(&mm->page_table_lock); |
... | ... | @@ -724,12 +729,6 @@ |
724 | 729 | } else { |
725 | 730 | pmd_t entry; |
726 | 731 | entry = mk_huge_pmd(page, vma); |
727 | - /* | |
728 | - * The spinlocking to take the lru_lock inside | |
729 | - * page_add_new_anon_rmap() acts as a full memory | |
730 | - * barrier to be sure clear_huge_page writes become | |
731 | - * visible after the set_pmd_at() write. | |
732 | - */ | |
733 | 732 | page_add_new_anon_rmap(page, vma, haddr); |
734 | 733 | set_pmd_at(mm, haddr, pmd, entry); |
735 | 734 | pgtable_trans_huge_deposit(mm, pgtable); |
mm/memory.c
... | ... | @@ -3244,6 +3244,11 @@ |
3244 | 3244 | page = alloc_zeroed_user_highpage_movable(vma, address); |
3245 | 3245 | if (!page) |
3246 | 3246 | goto oom; |
3247 | + /* | |
3248 | + * The memory barrier inside __SetPageUptodate makes sure that | |
3249 | + * preceeding stores to the page contents become visible before | |
3250 | + * the set_pte_at() write. | |
3251 | + */ | |
3247 | 3252 | __SetPageUptodate(page); |
3248 | 3253 | |
3249 | 3254 | if (mem_cgroup_newpage_charge(page, mm, GFP_KERNEL)) |