From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
To: Minchan Kim <minchan.kim@gmail.com>
Cc: Hiroaki Wakabayashi <primulaelatior@gmail.com>,
Andrew Morton <akpm@linux-foundation.org>,
LKML <linux-kernel@vger.kernel.org>,
linux-mm@kvack.org, Paul Menage <menage@google.com>,
Ying Han <yinghan@google.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Pekka Enberg <penberg@cs.helsinki.fi>,
Lee Schermerhorn <lee.schermerhorn@hp.com>
Subject: Re: [PATCH] mm: make munlock fast when mlock is canceled by sigkill
Date: Mon, 24 Aug 2009 10:51:39 +0900 [thread overview]
Message-ID: <20090824105139.c2ab8403.kamezawa.hiroyu@jp.fujitsu.com> (raw)
In-Reply-To: <28c262360908231844o3df95b14v15b2d4424465f033@mail.gmail.com>
On Mon, 24 Aug 2009 10:44:41 +0900
Minchan Kim <minchan.kim@gmail.com> wrote:
> On Sun, Aug 23, 2009 at 1:54 AM, Hiroaki
> Wakabayashi<primulaelatior@gmail.com> wrote:
> > From 27b2fde0222c59049026e7d0bdc4a2a68d0720f5 Mon Sep 17 00:00:00 2001
> > From: Hiroaki Wakabayashi <primulaelatior@gmail.com>
> > Date: Sat, 22 Aug 2009 19:14:53 +0900
> > Subject: [PATCH] mm: make munlock fast when mlock is canceled by sigkill
> >
> > This patch is for making commit 4779280d1e (mm: make get_user_pages()
> > interruptible) complete.
> >
> > At first, munlock() assumes that all pages in vma are pinned,
> >
> > Now, by the commit, mlock() can be interrupted by SIGKILL, etc A So, part of
> > pages are not pinned.
> > If SIGKILL, In exit() path, munlock is called for unlocking pinned pages
> > in vma.
> >
> > But, there, get_user_pages(write) is used for munlock(). Then, pages are
> > allocated via page-fault for exsiting process !!! This is problem at canceling
> > big mlock.
> > This patch tries to avoid allocating new pages at munlock().
> >
> > A mlock( big area )
> > A A A A <===== sig kill
> > A do_exit()
> > A A ->mmput()
> > A A A -> do_munlock()
> > A A A A -> get_user_pages()
> > A A A A A A A <allocate *never used* memory>
> > A A A ->.....freeing allocated memory.
> >
> > * Test program
> > % cat run.sh
> > #!/bin/sh
> >
> > ./mlock_test 2000000000 &
> > sleep 2
> > kill -9 $!
> > wait
> >
> > % cat mlock_test.c
> > #include <stdio.h>
> > #include <stdlib.h>
> > #include <string.h>
> > #include <sys/mman.h>
> > #include <sys/types.h>
> > #include <sys/stat.h>
> > #include <fcntl.h>
> > #include <errno.h>
> > #include <time.h>
> > #include <unistd.h>
> > #include <sys/time.h>
> >
> > int main(int argc, char **argv)
> > {
> > A A A A size_t length = 50 * 1024 * 1024;
> > A A A A void *addr;
> > A A A A time_t timer;
> >
> > A A A A if (argc >= 2)
> > A A A A A A A A length = strtoul(argv[1], NULL, 10);
> > A A A A printf("PID = %d\n", getpid());
> > A A A A addr = mmap(NULL, length, PROT_READ | PROT_WRITE,
> > A A A A A A A A A A A A A A A A MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
> > A A A A if (addr == MAP_FAILED) {
> > A A A A A A A A fprintf(stderr, "mmap failed: %s, length=%lu\n",
> > A A A A A A A A A A A A A A A A strerror(errno), length);
> > A A A A A A A A exit(EXIT_FAILURE);
> > A A A A }
> > A A A A printf("try mlock length=%lu\n", length);
> > A A A A timer = time(NULL);
> > A A A A if (mlock(addr, length) < 0) {
> > A A A A A A A A fprintf(stderr, "mlock failed: %s, time=%lu[sec]\n",
> > A A A A A A A A A A A A A A A A strerror(errno), time(NULL) - timer);
> > A A A A A A A A exit(EXIT_FAILURE);
> > A A A A }
> > A A A A printf("mlock succeed, time=%lu[sec]\n\n", time(NULL) - timer);
> > A A A A printf("try munlock length=%lu\n", length);
> > A A A A timer = time(NULL);
> > A A A A if (munlock(addr, length) < 0) {
> > A A A A A A A A fprintf(stderr, "munlock failed: %s, time=%lu[sec]\n",
> > A A A A A A A A A A A A A A A A strerror(errno), time(NULL)-timer);
> > A A A A A A A A exit(EXIT_FAILURE);
> > A A A A }
> > A A A A printf("munlock succeed, time=%lu[sec]\n\n", time(NULL) - timer);
> > A A A A if (munmap(addr, length) < 0) {
> > A A A A A A A A fprintf(stderr, "munmap failed: %s\n", strerror(errno));
> > A A A A A A A A exit(EXIT_FAILURE);
> > A A A A }
> > A A A A return 0;
> > }
> >
> > * Executed Result
> > -- Original executed result
> > % time ./run.sh
> >
> > PID = 2678
> > try mlock length=2000000000
> > ./run.sh: line 6: A 2678 Killed A A A A A A A A A ./mlock_test 2000000000
> > ./run.sh A 0.00s user 2.59s system 13% cpu 18.781 total
> > %
> >
> > -- After applied this patch
> > % time ./run.sh
> >
> > PID = 2512
> > try mlock length=2000000000
> > ./run.sh: line 6: A 2512 Killed A A A A A A A A A ./mlock_test 2000000000
> > ./run.sh A 0.00s user 1.15s system 45% cpu 2.507 total
> > %
> >
> > Signed-off-by: Hiroaki Wakabayashi <primulaelatior@gmail.com>
> > ---
> > A mm/internal.h | A A 1 +
> > A mm/memory.c A | A A 9 +++++++--
> > A mm/mlock.c A A | A 35 +++++++++++++++++++----------------
> > A 3 files changed, 27 insertions(+), 18 deletions(-)
> >
> > diff --git a/mm/internal.h b/mm/internal.h
> > index f290c4d..4ab5b24 100644
> > --- a/mm/internal.h
> > +++ b/mm/internal.h
> > @@ -254,6 +254,7 @@ static inline void
> > mminit_validate_memmodel_limits(unsigned long *start_pfn,
> > A #define GUP_FLAGS_FORCE A A A A A A A A A 0x2
> > A #define GUP_FLAGS_IGNORE_VMA_PERMISSIONS 0x4
> > A #define GUP_FLAGS_IGNORE_SIGKILL A A A A 0x8
> > +#define GUP_FLAGS_ALLOW_NULL A A A A A A 0x10
> >
>
> I am worried about adding new flag whenever we need it.
> But I think this case makes sense to me.
> In addition, I guess ZERO page can also use this flag.
>
> Kame. What do you think about it?
>
I do welcome this !
Then, I don't have to take care of mlock/munlock in ZERO_PAGE patch.
And without this patch, munlock() does copy-on-write just for unpinning memory.
So, this patch shows some right direction, I think.
One concern is flag name, ALLOW_NULL sounds not very good.
GUP_FLAGS_NOFAULT ?
I wonder we can remove a hack of FOLL_ANON for core-dump by this flag, too.
Thanks,
-Kame
>
> > A int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
> > A A A A A A A A A A unsigned long start, int len, int flags,
> > diff --git a/mm/memory.c b/mm/memory.c
> > index aede2ce..b41fbf9 100644
> > --- a/mm/memory.c
> > +++ b/mm/memory.c
> > @@ -1217,6 +1217,7 @@ int __get_user_pages(struct task_struct *tsk,
> > struct mm_struct *mm,
> > A A A A int force = !!(flags & GUP_FLAGS_FORCE);
> > A A A A int ignore = !!(flags & GUP_FLAGS_IGNORE_VMA_PERMISSIONS);
> > A A A A int ignore_sigkill = !!(flags & GUP_FLAGS_IGNORE_SIGKILL);
> > + A A A int allow_null = !!(flags & GUP_FLAGS_ALLOW_NULL);
> >
> > A A A A if (nr_pages <= 0)
> > A A A A A A A A return 0;
> > @@ -1312,6 +1313,8 @@ int __get_user_pages(struct task_struct *tsk,
> > struct mm_struct *mm,
> > A A A A A A A A A A A A while (!(page = follow_page(vma, start, foll_flags))) {
> > A A A A A A A A A A A A A A A A int ret;
> >
> > + A A A A A A A A A A A A A A A if (allow_null)
> > + A A A A A A A A A A A A A A A A A A A break;
> > A A A A A A A A A A A A A A A A ret = handle_mm_fault(mm, vma, start,
> > A A A A A A A A A A A A A A A A A A A A (foll_flags & FOLL_WRITE) ?
> > A A A A A A A A A A A A A A A A A A A A FAULT_FLAG_WRITE : 0);
> > @@ -1351,8 +1354,10 @@ int __get_user_pages(struct task_struct *tsk,
> > struct mm_struct *mm,
> > A A A A A A A A A A A A if (pages) {
> > A A A A A A A A A A A A A A A A pages[i] = page;
> >
> > - A A A A A A A A A A A A A A A flush_anon_page(vma, page, start);
> > - A A A A A A A A A A A A A A A flush_dcache_page(page);
> > + A A A A A A A A A A A A A A A if (page) {
> > + A A A A A A A A A A A A A A A A A A A flush_anon_page(vma, page, start);
> > + A A A A A A A A A A A A A A A A A A A flush_dcache_page(page);
> > + A A A A A A A A A A A A A A A }
> > A A A A A A A A A A A A }
> > A A A A A A A A A A A A if (vmas)
> > A A A A A A A A A A A A A A A A vmas[i] = vma;
> > diff --git a/mm/mlock.c b/mm/mlock.c
> > index 45eb650..0f5827b 100644
> > --- a/mm/mlock.c
> > +++ b/mm/mlock.c
> > @@ -178,9 +178,10 @@ static long __mlock_vma_pages_range(struct
> > vm_area_struct *vma,
> > A A A A */
> > A A A A if (!mlock)
> > A A A A A A A A gup_flags |= GUP_FLAGS_IGNORE_VMA_PERMISSIONS |
> > - A A A A A A A A A A A A A A GUP_FLAGS_IGNORE_SIGKILL;
> > + A A A A A A A A A A A A A A GUP_FLAGS_IGNORE_SIGKILL |
> > + A A A A A A A A A A A A A A GUP_FLAGS_ALLOW_NULL;
> >
> > - A A A if (vma->vm_flags & VM_WRITE)
> > + A A A if (mlock && (vma->vm_flags & VM_WRITE))
> > A A A A A A A A gup_flags |= GUP_FLAGS_WRITE;
> >
> > A A A A while (nr_pages > 0) {
> > @@ -220,21 +221,23 @@ static long __mlock_vma_pages_range(struct
> > vm_area_struct *vma,
> > A A A A A A A A for (i = 0; i < ret; i++) {
> > A A A A A A A A A A A A struct page *page = pages[i];
> >
> > - A A A A A A A A A A A lock_page(page);
> > - A A A A A A A A A A A /*
> > - A A A A A A A A A A A A * Because we lock page here and migration is blocked
> > - A A A A A A A A A A A A * by the elevated reference, we need only check for
> > - A A A A A A A A A A A A * page truncation (file-cache only).
> > - A A A A A A A A A A A A */
> > - A A A A A A A A A A A if (page->mapping) {
> > - A A A A A A A A A A A A A A A if (mlock)
> > - A A A A A A A A A A A A A A A A A A A mlock_vma_page(page);
> > - A A A A A A A A A A A A A A A else
> > - A A A A A A A A A A A A A A A A A A A munlock_vma_page(page);
> > + A A A A A A A A A A A if (page) {
> > + A A A A A A A A A A A A A A A lock_page(page);
> > + A A A A A A A A A A A A A A A /*
> > + A A A A A A A A A A A A A A A A * Because we lock page here and migration is
> > + A A A A A A A A A A A A A A A A * blocked by the elevated reference, we need
> > + A A A A A A A A A A A A A A A A * only check for page truncation
> > + A A A A A A A A A A A A A A A A * (file-cache only).
> > + A A A A A A A A A A A A A A A A */
> > + A A A A A A A A A A A A A A A if (page->mapping) {
> > + A A A A A A A A A A A A A A A A A A A if (mlock)
> > + A A A A A A A A A A A A A A A A A A A A A A A mlock_vma_page(page);
> > + A A A A A A A A A A A A A A A A A A A else
> > + A A A A A A A A A A A A A A A A A A A A A A A munlock_vma_page(page);
> > + A A A A A A A A A A A A A A A }
> > + A A A A A A A A A A A A A A A unlock_page(page);
> > + A A A A A A A A A A A A A A A put_page(page); /* ref from get_user_pages() */
> > A A A A A A A A A A A A }
> > - A A A A A A A A A A A unlock_page(page);
> > - A A A A A A A A A A A put_page(page); A A A A /* ref from get_user_pages() */
> > -
> > A A A A A A A A A A A A /*
> > A A A A A A A A A A A A * here we assume that get_user_pages() has given us
> > A A A A A A A A A A A A * a list of virtually contiguous pages.
> > --
> > 1.5.6.5
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at A http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at A http://www.tux.org/lkml/
> >
>
>
>
> --
> Kind regards,
> Minchan Kim
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2009-08-25 20:07 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-08-22 16:54 Hiroaki Wakabayashi
2009-08-24 1:44 ` Minchan Kim
2009-08-24 1:51 ` KAMEZAWA Hiroyuki [this message]
2009-08-24 2:23 ` Minchan Kim
2009-08-24 4:13 ` KOSAKI Motohiro
2009-08-25 4:46 ` Hiroaki Wakabayashi
2009-08-25 5:39 ` Minchan Kim
2009-08-25 9:03 ` Hugh Dickins
2009-08-25 9:37 ` KAMEZAWA Hiroyuki
2009-08-25 11:46 ` Minchan Kim
2009-08-26 2:32 ` KOSAKI Motohiro
2009-08-26 11:11 ` Hiroaki Wakabayashi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090824105139.c2ab8403.kamezawa.hiroyu@jp.fujitsu.com \
--to=kamezawa.hiroyu@jp.fujitsu.com \
--cc=akpm@linux-foundation.org \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=lee.schermerhorn@hp.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=menage@google.com \
--cc=minchan.kim@gmail.com \
--cc=penberg@cs.helsinki.fi \
--cc=primulaelatior@gmail.com \
--cc=yinghan@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox