diff mbox series

[AUTOSEL,13/14] mm/thp: ClearPageDoubleMap in first page_add_file_rmap()

Message ID 20220428154222.1230793-13-gregkh@linuxfoundation.org
State New
Headers show
Series [AUTOSEL,01/14] mm: fix invalid page pointer returned with FOLL_PIN gups | expand

Commit Message

Greg KH April 28, 2022, 3:42 p.m. UTC
From: Hugh Dickins <hughd@google.com>

commit bd55b0c2d64e84a75575f548a33a3dfecc135b65 upstream.

PageDoubleMap is maintained differently for anon and for shmem+file: the
shmem+file one was never cleared, because a safe place to do so could
not be found; so it would blight future use of the cached hugepage until
evicted.

See https://lore.kernel.org/lkml/1571938066-29031-1-git-send-email-yang.shi@linux.alibaba.com/

But page_add_file_rmap() does provide a safe place to do so (though later
than one might wish): allowing testing to return to an initial state
without a damaging drop_caches.

Link: https://lkml.kernel.org/r/61c5cf99-a962-9a25-597a-53ab1bd8fbc0@google.com
Fixes: 9a73f61bdb8a ("thp, mlock: do not mlock PTE-mapped file huge pages")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 mm/rmap.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

Comments

Sasha Levin April 30, 2022, 12:27 a.m. UTC | #1
On Thu, Apr 28, 2022 at 12:27:40PM -0700, Hugh Dickins wrote:
>On Thu, 28 Apr 2022, Greg Kroah-Hartman wrote:
>> On Thu, Apr 28, 2022 at 09:51:58AM -0700, Hugh Dickins wrote:
>> > On Thu, 28 Apr 2022, Greg Kroah-Hartman wrote:
>> >
>> > > From: Hugh Dickins <hughd@google.com>
>> > >
>> > > commit bd55b0c2d64e84a75575f548a33a3dfecc135b65 upstream.
>> > >
>> > > PageDoubleMap is maintained differently for anon and for shmem+file: the
>> > > shmem+file one was never cleared, because a safe place to do so could
>> > > not be found; so it would blight future use of the cached hugepage until
>> > > evicted.
>> > >
>> > > See https://lore.kernel.org/lkml/1571938066-29031-1-git-send-email-yang.shi@linux.alibaba.com/
>> > >
>> > > But page_add_file_rmap() does provide a safe place to do so (though later
>> > > than one might wish): allowing testing to return to an initial state
>> > > without a damaging drop_caches.
>> > >
>> > > Link: https://lkml.kernel.org/r/61c5cf99-a962-9a25-597a-53ab1bd8fbc0@google.com
>> > > Fixes: 9a73f61bdb8a ("thp, mlock: do not mlock PTE-mapped file huge pages")
>> > > Signed-off-by: Hugh Dickins <hughd@google.com>
>> > > Reviewed-by: Yang Shi <shy828301@gmail.com>
>> > > Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
>> > > Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
>> > > Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
>> > > Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
>> >
>> > NAK.
>> >
>> > I thought we had a long-standing agreement that AUTOSEL does not try
>> > to add patches from akpm's tree which had not been marked for stable.

I guess it was only between myself and mm/ :p

>> True, this was my attempt at saying "hey these all look like they should
>> go to stable trees, why not?"
>
>Okay, it seems I should have read "AUTOSEL" as "Hey, GregKH here,
>these all look like they should go to stable trees, why not?",
>which would have drawn a friendlier response.

FRIENDLYGREGBOT :)

>The answer is that I considered stable at the time, and akpm did too,
>and none of my three (I've not looked through the other 11) are serious
>enough to be needed in stable; and I'm cautious about backports, because
>I know that the tree they went on top of differs thereabouts from 5.17.
>
>Of course I think the patches in 5.18-rc are good, and yes, they're
>things I've thought worthwhile enough for me personally to port forward
>over several releases until I had time to send in.  But that doesn't
>make them safe stable candidates, without someone to verify and vouch
>for the results in this or that tree - I run on a much slower clock
>than you and most around here, I do not have time for that at present
>(and would prefer not even to be having this conversation).
>
>But I'm happily overruled if any mm guys think they are worth that
>extra effort, and will verify and vouch for them.

What's the extra effort here? We're seeing so many cases where we see
issues with LTS kernels and we end up spending so much time triaging and
diagnosing them only to find out that they've already been fixed.

Honesly, having them in -stable seems like *less* effort to me.
diff mbox series

Patch

diff --git a/mm/rmap.c b/mm/rmap.c
index 9e27f9f038d3..444d0d958aff 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -1252,6 +1252,17 @@  void page_add_file_rmap(struct page *page, bool compound)
 		}
 		if (!atomic_inc_and_test(compound_mapcount_ptr(page)))
 			goto out;
+
+		/*
+		 * It is racy to ClearPageDoubleMap in page_remove_file_rmap();
+		 * but page lock is held by all page_add_file_rmap() compound
+		 * callers, and SetPageDoubleMap below warns if !PageLocked:
+		 * so here is a place that DoubleMap can be safely cleared.
+		 */
+		VM_WARN_ON_ONCE(!PageLocked(page));
+		if (nr == nr_pages && PageDoubleMap(page))
+			ClearPageDoubleMap(page);
+
 		if (PageSwapBacked(page))
 			__mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED,
 						nr_pages);