From: Shaohua Li <shaohua.li@intel.com>
To: linux-kernel@vger.kernel.org
Cc: akpm@linux-foundation.org, tj@kernel.org, eric.dumazet@gmail.com,
cl@linux.com, npiggin@kernel.dk
Subject: [patch v2 0/5] percpu_counter: bug fix and enhancement
Date: Wed, 11 May 2011 16:10:12 +0800 [thread overview]
Message-ID: <20110511081012.903869567@sli10-conroe.sh.intel.com> (raw)
The patch sets do two things.
1. fix bug for 32-bit system. percpu_counter uses s64 counter. Without any
locking reading s64 in 32-bit system isn't safe and can cause bad side effect.
2. improve scalability for __percpu_counter_add. In some cases, _add could
cause heavy lock contention (see patch 4 for detailed infomation and data).
The patches will remove the contention and speed up it a bit. Last post
(http://marc.info/?l=linux-kernel&m=130259547913607&w=2) simpliy uses
atomic64 for percpu_counter, but Tejun pointed out this could cause
deviation in __percpu_counter_sum.
The new implementation uses lglock to protect percpu data. Each cpu has its
private lock while other cpu doesn't take. In this way _add doesn't need take
global lock anymore and remove the deviation. This still gives me about
about 5x ~ 6x faster (not that faster than the original 7x faster, but still
good) with the workload mentioned in patch 4.
patch 1 fix s64 read bug for 32-bit system for UP
patch 2 convert lglock to be used by dynamaically allocated structre. Later
patch will use lglock for percpu_counter
patch 3,4 fix s64 read bug for 32-bit system for MP. And it also improve the
scalability for __percpu_counter_add.
patch 5 is from Christoph Lameter to make __percpu_counter_add fastpath
preemptless. I added it here because I converted percpu_counter to use
lglock. All bugs are from mine.
Comments and suggestions are welcomed!
Thanks,
Shaohua
next reply other threads:[~2011-05-11 15:39 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-05-11 8:10 Shaohua Li [this message]
2011-05-11 8:10 ` [patch v2 1/5] percpu_counter: fix code for 32bit systems for UP Shaohua Li
2011-05-11 8:10 ` [patch v2 2/5] lglock: convert it to work with dynamically allocated structure Shaohua Li
2011-05-11 8:10 ` [patch v2 3/5] percpu_counter: use lglock to protect percpu data Shaohua Li
2011-05-11 8:10 ` [patch v2 4/5] percpu_counter: use atomic64 for counter in SMP Shaohua Li
2011-05-11 9:34 ` Andrew Morton
2011-05-12 2:40 ` Shaohua Li
2011-05-11 8:10 ` [patch v2 5/5] percpu_counter: preemptless __per_cpu_counter_add Shaohua Li
2011-05-11 9:28 ` [patch v2 0/5] percpu_counter: bug fix and enhancement Tejun Heo
2011-05-12 2:48 ` Shaohua Li
2011-05-12 8:21 ` Tejun Heo
2011-05-12 8:55 ` Shaohua Li
2011-05-12 8:59 ` Tejun Heo
2011-05-12 9:02 ` Eric Dumazet
2011-05-12 9:03 ` Eric Dumazet
2011-05-12 9:05 ` Tejun Heo
2011-05-13 3:09 ` Shaohua Li
2011-05-13 4:37 ` Shaohua Li
2011-05-13 5:20 ` Eric Dumazet
2011-05-13 5:28 ` Shaohua Li
2011-05-13 6:34 ` Eric Dumazet
2011-05-13 7:33 ` Shaohua Li
2011-05-13 14:51 ` [patch] percpu_counter: scalability works Eric Dumazet
2011-05-13 15:39 ` Eric Dumazet
2011-05-13 16:35 ` [patch V2] " Eric Dumazet
2011-05-13 16:46 ` Eric Dumazet
2011-05-13 22:03 ` [patch V3] " Eric Dumazet
2011-05-16 0:58 ` Shaohua Li
2011-05-16 6:11 ` Eric Dumazet
2011-05-16 6:37 ` Shaohua Li
2011-05-16 6:55 ` Eric Dumazet
2011-05-16 7:15 ` Shaohua Li
2011-05-16 7:44 ` Eric Dumazet
2011-05-16 8:34 ` Shaohua Li
2011-05-16 9:35 ` Eric Dumazet
2011-05-16 14:22 ` Eric Dumazet
2011-05-17 0:55 ` Shaohua Li
2011-05-17 4:56 ` Eric Dumazet
2011-05-17 5:22 ` Shaohua Li
2011-05-17 9:01 ` Eric Dumazet
2011-05-17 9:11 ` Tejun Heo
2011-05-17 9:45 ` Eric Dumazet
2011-05-17 9:50 ` Tejun Heo
2011-05-17 12:20 ` Eric Dumazet
2011-05-17 12:45 ` Tejun Heo
2011-05-17 13:00 ` Eric Dumazet
2011-05-17 13:04 ` Tejun Heo
2011-05-17 13:55 ` Christoph Lameter
2011-05-17 14:02 ` Tejun Heo
2011-05-17 14:38 ` Christoph Lameter
2011-05-18 1:00 ` Shaohua Li
2011-05-12 14:38 ` [patch v2 0/5] percpu_counter: bug fix and enhancement Christoph Lameter
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=20110511081012.903869567@sli10-conroe.sh.intel.com \
--to=shaohua.li@intel.com \
--cc=akpm@linux-foundation.org \
--cc=cl@linux.com \
--cc=eric.dumazet@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=npiggin@kernel.dk \
--cc=tj@kernel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®