* [PATCH 0/2] mm hugetlb x86: add hugepage support to pagemap
@ 2009-12-07 7:59 Naoya Horiguchi
2009-12-08 22:39 ` Andrew Morton
0 siblings, 1 reply; 3+ messages in thread
From: Naoya Horiguchi @ 2009-12-07 7:59 UTC (permalink / raw)
To: LKML; +Cc: linux-mm, hugh.dickins, ak, Wu Fengguang
Hi,
Most callers of pmd_none_or_clear_bad() check whether the target
page is in a hugepage or not, but mincore() and walk_page_range() do
not check it. So if we read /proc/pid/pagemap for the hugepage
on x86 machine, the hugepage memory is leaked as shown below.
This patch fixes it by extending pagemap interface to support hugepages.
I split this fix into two patches. The first patch just adds the check
for hugepages, and the second patch adds a new member to struct mm_walk
to handle the hugepages.
fs/proc/task_mmu.c | 43 +++++++++++++++++++++++++++++++++++++++++++
include/linux/mm.h | 3 +++
mm/pagewalk.c | 29 ++++++++++++++++++++++++++++-
3 files changed, 74 insertions(+), 1 deletions(-)
Details
=======
My test program (leak_pagemap) works as follows:
- creat() and mmap() a file on hugetlbfs (file size is 200MB == 100 hugepages,)
- read()/write() something on it,
- call page-types with option -p,
- munmap() and unlink() the file on hugetlbfs
Without my patches
------------------
$ cat /proc/meminfo |grep "HugePage"
HugePages_Total: 1000
HugePages_Free: 1000
HugePages_Rsvd: 0
HugePages_Surp: 0
$ ./leak_pagemap
flags page-count MB symbolic-flags long-symbolic-flags
0x0000000000000000 1 0 __________________________________
0x0000000000000804 1 0 __R________M______________________ referenced,mmap
0x000000000000086c 81 0 __RU_lA____M______________________ referenced,uptodate,lru,active,mmap
0x0000000000005808 5 0 ___U_______Ma_b___________________ uptodate,mmap,anonymous,swapbacked
0x0000000000005868 12 0 ___U_lA____Ma_b___________________ uptodate,lru,active,mmap,anonymous,swapbacked
0x000000000000586c 1 0 __RU_lA____Ma_b___________________ referenced,uptodate,lru,active,mmap,anonymous,swapbacked
total 101 0
$ cat /proc/meminfo |grep "HugePage"
HugePages_Total: 1000
HugePages_Free: 900
HugePages_Rsvd: 0
HugePages_Surp: 0
$ ls /hugetlbfs/
$
The output of page-types don't show any hugepage, and 100 hugepages are
accounted as used while there is no file on hugetlbfs
With my patches
---------------
$ cat /proc/meminfo |grep "HugePage"
HugePages_Total: 1000
HugePages_Free: 1000
HugePages_Rsvd: 0
HugePages_Surp: 0
$ ./leak_pagemap
flags page-count MB symbolic-flags long-symbolic-flags
0x0000000000000000 1 0 __________________________________
0x0000000000030000 51100 199 ________________TG________________ compound_tail,huge
0x0000000000028018 100 0 ___UD__________H_G________________ uptodate,dirty,compound_head,huge
0x0000000000000804 1 0 __R________M______________________ referenced,mmap
0x000000000000080c 1 0 __RU_______M______________________ referenced,uptodate,mmap
0x000000000000086c 80 0 __RU_lA____M______________________ referenced,uptodate,lru,active,mmap
0x0000000000005808 4 0 ___U_______Ma_b___________________ uptodate,mmap,anonymous,swapbacked
0x0000000000005868 12 0 ___U_lA____Ma_b___________________ uptodate,lru,active,mmap,anonymous,swapbacked
0x000000000000586c 1 0 __RU_lA____Ma_b___________________ referenced,uptodate,lru,active,mmap,anonymous,swapbacked
total 51300 200
$ cat /proc/meminfo |grep "HugePage"
HugePages_Total: 1000
HugePages_Free: 1000
HugePages_Rsvd: 0
HugePages_Surp: 0
$ ls /hugetlbfs
$
The output of page-types shows 51200 pages contributing to hugepages,
containing 100 head pages and 51100 tail pages as expected.
And no memory leaks.
Thanks,
Naoya Horiguchi
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH 0/2] mm hugetlb x86: add hugepage support to pagemap
2009-12-07 7:59 [PATCH 0/2] mm hugetlb x86: add hugepage support to pagemap Naoya Horiguchi
@ 2009-12-08 22:39 ` Andrew Morton
2009-12-09 1:08 ` Naoya Horiguchi
0 siblings, 1 reply; 3+ messages in thread
From: Andrew Morton @ 2009-12-08 22:39 UTC (permalink / raw)
To: n-horiguchi; +Cc: LKML, linux-mm, hugh.dickins, ak, Wu Fengguang
On Mon, 07 Dec 2009 16:59:18 +0900
Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:
> Most callers of pmd_none_or_clear_bad() check whether the target
> page is in a hugepage or not, but mincore() and walk_page_range() do
> not check it. So if we read /proc/pid/pagemap for the hugepage
> on x86 machine, the hugepage memory is leaked as shown below.
> This patch fixes it by extending pagemap interface to support hugepages.
>
> I split this fix into two patches. The first patch just adds the check
> for hugepages, and the second patch adds a new member to struct mm_walk
> to handle the hugepages.
I kind of dislike the practice of putting all the changelog in patch
[0/n] and then leaving the patches themselves practically
unchangelogged. Because
a) Someone (ie: me) needs to go and shuffle all the text around so
that the information gets itself into the git record. We don't add
changelog-only commits to git!
b) Someone (ie: me) might decide to backport a subset of the patches
into -stable. Now someone (ie: me) needs to carve up the changelogs
so that the pieces which go into -stable still make standalone sense.
I'm not sure that I did this particularly well in this case. Oh well.
Please confirm that
mm-hugetlb-fix-hugepage-memory-leak-in-walk_page_range.patch is
suitable for a -stable backport without inclusion of
mm-hugetlb-add-hugepage-support-to-pagemap.patch. I think it is.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH 0/2] mm hugetlb x86: add hugepage support to pagemap
2009-12-08 22:39 ` Andrew Morton
@ 2009-12-09 1:08 ` Naoya Horiguchi
0 siblings, 0 replies; 3+ messages in thread
From: Naoya Horiguchi @ 2009-12-09 1:08 UTC (permalink / raw)
To: Andrew Morton; +Cc: LKML, linux-mm, hugh.dickins, ak, Wu Fengguang
> I kind of dislike the practice of putting all the changelog in patch
> [0/n] and then leaving the patches themselves practically
> unchangelogged. Because
Sorry, I agree.
>
> a) Someone (ie: me) needs to go and shuffle all the text around so
> that the information gets itself into the git record. We don't add
> changelog-only commits to git!
>
> b) Someone (ie: me) might decide to backport a subset of the patches
> into -stable. Now someone (ie: me) needs to carve up the changelogs
> so that the pieces which go into -stable still make standalone sense.
>
> I'm not sure that I did this particularly well in this case. Oh well.
>
>
> Please confirm that
> mm-hugetlb-fix-hugepage-memory-leak-in-walk_page_range.patch is
> suitable for a -stable backport without inclusion of
> mm-hugetlb-add-hugepage-support-to-pagemap.patch. I think it is.
>
I think that's OK.
Thanks,
Naoya Horiguchi
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2009-12-09 1:11 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-12-07 7:59 [PATCH 0/2] mm hugetlb x86: add hugepage support to pagemap Naoya Horiguchi
2009-12-08 22:39 ` Andrew Morton
2009-12-09 1:08 ` Naoya Horiguchi
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).