All of lore.kernel.org
 help / color / mirror / Atom feed
From: Andrey Ryabinin <aryabinin@virtuozzo.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: <linux-kernel@vger.kernel.org>,
	Andrey Ryabinin <aryabinin@virtuozzo.com>,
	<penguin-kernel@I-love.SAKURA.ne.jp>, <mhocko@kernel.org>,
	<linux-mm@kvack.org>, <hpa@zytor.com>, <chris@chris-wilson.co.uk>,
	<hch@lst.de>, <mingo@elte.hu>, <jszhang@marvell.com>,
	<joelaf@google.com>, <joaodias@google.com>, <willy@infradead.org>,
	<tglx@linutronix.de>, <thellstrom@vmware.com>
Subject: [PATCH v2 5/5] mm/vmalloc: Don't spawn workers if somebody already purging
Date: Wed, 12 Apr 2017 15:49:05 +0300	[thread overview]
Message-ID: <20170412124905.25443-6-aryabinin@virtuozzo.com> (raw)
In-Reply-To: <20170412124905.25443-1-aryabinin@virtuozzo.com>

Don't schedule purge_vmap_work if mutex_is_locked(&vmap_purge_lock),
as this means that purging is already running in another thread.
There is no point to schedule extra purge_vmap_work if somebody
is already purging for us, because that extra work will not do anything
useful.

To evaluate performance impact of this change test that calls
fork() 100 000 times on the kernel with enabled CONFIG_VMAP_STACK=y
and NR_CACHED_STACK changed to 0 (so that each fork()/exit() executes
vmalloc()/vfree() call) was used.

Commands:
~ # grep try_purge /proc/kallsyms
ffffffff811d0dd0 t try_purge_vmap_area_lazy

~ # perf stat --repeat 10 -ae workqueue:workqueue_queue_work \
              --filter 'function == 0xffffffff811d0dd0' ./fork

gave me the following results:

before:
   30      workqueue:workqueue_queue_work                ( +-  1.31% )
   1.613231060 seconds time elapsed                      ( +-  0.38% )

after:
   15      workqueue:workqueue_queue_work                ( +-  0.88% )
   1.615368474 seconds time elapsed                      ( +-  0.41% )

So there is no measurable difference on the performance of the test itself,
but without the optimization we queue twice more jobs. This should save
kworkers from doing some useless job.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Suggested-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
---
 mm/vmalloc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index ee62c0a..1079555 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -737,7 +737,8 @@ static void free_vmap_area_noflush(struct vmap_area *va)
 	/* After this point, we may free va at any time */
 	llist_add(&va->purge_list, &vmap_purge_list);
 
-	if (unlikely(nr_lazy > lazy_max_pages()))
+	if (unlikely(nr_lazy > lazy_max_pages()) &&
+	    !mutex_is_locked(&vmap_purge_lock))
 		schedule_work(&purge_vmap_work);
 }
 
-- 
2.10.2

WARNING: multiple messages have this Message-ID (diff)
From: Andrey Ryabinin <aryabinin@virtuozzo.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: linux-kernel@vger.kernel.org,
	Andrey Ryabinin <aryabinin@virtuozzo.com>,
	penguin-kernel@I-love.SAKURA.ne.jp, mhocko@kernel.org,
	linux-mm@kvack.org, hpa@zytor.com, chris@chris-wilson.co.uk,
	hch@lst.de, mingo@elte.hu, jszhang@marvell.com,
	joelaf@google.com, joaodias@google.com, willy@infradead.org,
	tglx@linutronix.de, thellstrom@vmware.com
Subject: [PATCH v2 5/5] mm/vmalloc: Don't spawn workers if somebody already purging
Date: Wed, 12 Apr 2017 15:49:05 +0300	[thread overview]
Message-ID: <20170412124905.25443-6-aryabinin@virtuozzo.com> (raw)
In-Reply-To: <20170412124905.25443-1-aryabinin@virtuozzo.com>

Don't schedule purge_vmap_work if mutex_is_locked(&vmap_purge_lock),
as this means that purging is already running in another thread.
There is no point to schedule extra purge_vmap_work if somebody
is already purging for us, because that extra work will not do anything
useful.

To evaluate performance impact of this change test that calls
fork() 100 000 times on the kernel with enabled CONFIG_VMAP_STACK=y
and NR_CACHED_STACK changed to 0 (so that each fork()/exit() executes
vmalloc()/vfree() call) was used.

Commands:
~ # grep try_purge /proc/kallsyms
ffffffff811d0dd0 t try_purge_vmap_area_lazy

~ # perf stat --repeat 10 -ae workqueue:workqueue_queue_work \
              --filter 'function == 0xffffffff811d0dd0' ./fork

gave me the following results:

before:
   30      workqueue:workqueue_queue_work                ( +-  1.31% )
   1.613231060 seconds time elapsed                      ( +-  0.38% )

after:
   15      workqueue:workqueue_queue_work                ( +-  0.88% )
   1.615368474 seconds time elapsed                      ( +-  0.41% )

So there is no measurable difference on the performance of the test itself,
but without the optimization we queue twice more jobs. This should save
kworkers from doing some useless job.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Suggested-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
---
 mm/vmalloc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index ee62c0a..1079555 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -737,7 +737,8 @@ static void free_vmap_area_noflush(struct vmap_area *va)
 	/* After this point, we may free va at any time */
 	llist_add(&va->purge_list, &vmap_purge_list);
 
-	if (unlikely(nr_lazy > lazy_max_pages()))
+	if (unlikely(nr_lazy > lazy_max_pages()) &&
+	    !mutex_is_locked(&vmap_purge_lock))
 		schedule_work(&purge_vmap_work);
 }
 
-- 
2.10.2

--
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>

  parent reply	other threads:[~2017-04-12 12:49 UTC|newest]

Thread overview: 65+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-03-30 10:27 [PATCH 1/4] mm/vmalloc: allow to call vfree() in atomic context Andrey Ryabinin
2017-03-30 10:27 ` Andrey Ryabinin
2017-03-30 10:27 ` Andrey Ryabinin
2017-03-30 10:27 ` [PATCH 2/4] x86/ldt: use vfree() instead of vfree_atomic() Andrey Ryabinin
2017-03-30 10:27   ` Andrey Ryabinin
2017-03-31  8:05   ` Thomas Gleixner
2017-03-31  8:05     ` Thomas Gleixner
2017-03-30 10:27 ` [PATCH 3/4] kernel/fork: use vfree() instead of vfree_atomic() to free thread stack Andrey Ryabinin
2017-03-30 10:27   ` Andrey Ryabinin
2017-03-30 10:27 ` [PATCH 4/4] mm/vmalloc: remove vfree_atomic() Andrey Ryabinin
2017-03-30 10:27   ` Andrey Ryabinin
2017-03-30 17:18   ` Matthew Wilcox
2017-03-30 17:18     ` Matthew Wilcox
2017-03-30 15:27     ` Andrey Ryabinin
2017-03-30 15:27       ` Andrey Ryabinin
2017-04-04  9:40   ` Michal Hocko
2017-04-04  9:40     ` Michal Hocko
2017-03-30 12:00 ` [PATCH 1/4] mm/vmalloc: allow to call vfree() in atomic context Thomas Hellstrom
2017-03-30 12:00   ` Thomas Hellstrom
2017-03-30 12:00   ` Thomas Hellstrom
2017-03-30 14:48   ` Andrey Ryabinin
2017-03-30 14:48     ` Andrey Ryabinin
2017-03-30 14:48     ` Andrey Ryabinin
2017-03-30 15:04     ` Thomas Hellstrom
2017-03-30 15:04       ` Thomas Hellstrom
2017-03-30 15:04       ` Thomas Hellstrom
2017-04-04  9:41     ` Michal Hocko
2017-04-04  9:41       ` Michal Hocko
2017-04-04  9:49       ` Thomas Hellstrom
2017-04-04  9:49         ` Thomas Hellstrom
2017-04-04  9:49         ` Thomas Hellstrom
2017-04-05 10:31       ` Andrey Ryabinin
2017-04-05 10:31         ` Andrey Ryabinin
2017-04-05 10:31         ` Andrey Ryabinin
2017-04-05 10:42         ` Michal Hocko
2017-04-05 10:42           ` Michal Hocko
2017-04-05 11:42     ` Vlastimil Babka
2017-04-05 11:42       ` Vlastimil Babka
2017-04-05 12:14       ` Michal Hocko
2017-04-05 12:14         ` Michal Hocko
2017-03-30 22:22 ` Andrew Morton
2017-03-30 22:22   ` Andrew Morton
2017-03-30 22:22   ` Andrew Morton
2017-03-31  7:12   ` Joel Fernandes
2017-03-31  7:12     ` Joel Fernandes
2017-03-31  9:26   ` Andrey Ryabinin
2017-03-31  9:26     ` Andrey Ryabinin
2017-03-31  9:26     ` Andrey Ryabinin
2017-04-04  9:36   ` Michal Hocko
2017-04-04  9:36     ` Michal Hocko
2017-04-04  9:38 ` Michal Hocko
2017-04-04  9:38   ` Michal Hocko
2017-04-12 12:49 ` [PATCH v2 0/5] " Andrey Ryabinin
2017-04-12 12:49   ` Andrey Ryabinin
2017-04-12 12:49   ` [PATCH v2 1/5] mm/vmalloc: " Andrey Ryabinin
2017-04-12 12:49     ` Andrey Ryabinin
2017-04-12 12:49     ` Andrey Ryabinin
2017-04-12 12:49   ` [PATCH v2 2/5] x86/ldt: use vfree() instead of vfree_atomic() Andrey Ryabinin
2017-04-12 12:49     ` Andrey Ryabinin
2017-04-12 12:49   ` [PATCH v2 3/5] kernel/fork: use vfree() instead of vfree_atomic() to free thread stack Andrey Ryabinin
2017-04-12 12:49     ` Andrey Ryabinin
2017-04-12 12:49   ` [PATCH v2 4/5] mm/vmalloc: remove vfree_atomic() Andrey Ryabinin
2017-04-12 12:49     ` Andrey Ryabinin
2017-04-12 12:49   ` Andrey Ryabinin [this message]
2017-04-12 12:49     ` [PATCH v2 5/5] mm/vmalloc: Don't spawn workers if somebody already purging Andrey Ryabinin

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=20170412124905.25443-6-aryabinin@virtuozzo.com \
    --to=aryabinin@virtuozzo.com \
    --cc=akpm@linux-foundation.org \
    --cc=chris@chris-wilson.co.uk \
    --cc=hch@lst.de \
    --cc=hpa@zytor.com \
    --cc=joaodias@google.com \
    --cc=joelaf@google.com \
    --cc=jszhang@marvell.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=mhocko@kernel.org \
    --cc=mingo@elte.hu \
    --cc=penguin-kernel@I-love.SAKURA.ne.jp \
    --cc=tglx@linutronix.de \
    --cc=thellstrom@vmware.com \
    --cc=willy@infradead.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.