From: cl@linux-foundation.org
To: akpm@linux-foundation.org
Cc: linux-kernel@vger.kernel.org, Tejun Heo <tj@kernel.org>,
"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: mingo@elte.hu
Cc: rusty@rustcorp.com.au
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Subject: [this_cpu_xx V4 11/20] RCU: Use this_cpu operations
Date: Thu, 01 Oct 2009 17:25:32 -0400 [thread overview]
Message-ID: <20091001212559.690311062@gentwo.org> (raw)
In-Reply-To: <20091001212521.123389189@gentwo.org>
[-- Attachment #1: this_cpu_rcu --]
[-- Type: text/plain, Size: 1916 bytes --]
RCU does not do dynamic allocations but it increments per cpu variables
a lot. These instructions results in a move to a register and then back
to memory. This patch will make it use the inc/dec instructions on x86
that do not need a register.
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
---
kernel/rcutorture.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
Index: linux-2.6/kernel/rcutorture.c
===================================================================
--- linux-2.6.orig/kernel/rcutorture.c 2009-09-28 10:08:10.000000000 -0500
+++ linux-2.6/kernel/rcutorture.c 2009-09-29 09:02:00.000000000 -0500
@@ -731,13 +731,13 @@ static void rcu_torture_timer(unsigned l
/* Should not happen, but... */
pipe_count = RCU_TORTURE_PIPE_LEN;
}
- ++__get_cpu_var(rcu_torture_count)[pipe_count];
+ __this_cpu_inc(per_cpu_var(rcu_torture_count)[pipe_count]);
completed = cur_ops->completed() - completed;
if (completed > RCU_TORTURE_PIPE_LEN) {
/* Should not happen, but... */
completed = RCU_TORTURE_PIPE_LEN;
}
- ++__get_cpu_var(rcu_torture_batch)[completed];
+ __this_cpu_inc(per_cpu_var(rcu_torture_batch)[completed]);
preempt_enable();
cur_ops->readunlock(idx);
}
@@ -786,13 +786,13 @@ rcu_torture_reader(void *arg)
/* Should not happen, but... */
pipe_count = RCU_TORTURE_PIPE_LEN;
}
- ++__get_cpu_var(rcu_torture_count)[pipe_count];
+ __this_cpu_inc(per_cpu_var(rcu_torture_count)[pipe_count]);
completed = cur_ops->completed() - completed;
if (completed > RCU_TORTURE_PIPE_LEN) {
/* Should not happen, but... */
completed = RCU_TORTURE_PIPE_LEN;
}
- ++__get_cpu_var(rcu_torture_batch)[completed];
+ __this_cpu_inc(per_cpu_var(rcu_torture_batch)[completed]);
preempt_enable();
cur_ops->readunlock(idx);
schedule();
--
next prev parent reply other threads:[~2009-10-02 1:50 UTC|newest]
Thread overview: 65+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-10-01 21:25 [this_cpu_xx V4 00/20] Introduce per cpu atomic operations and avoid per cpu address arithmetic cl
2009-10-01 21:25 ` [this_cpu_xx V4 01/20] Introduce this_cpu_ptr() and generic this_cpu_* operations cl
2009-10-02 9:16 ` Tejun Heo
2009-10-02 9:34 ` Ingo Molnar
2009-10-02 17:11 ` Christoph Lameter
2009-10-06 10:04 ` Rusty Russell
2009-10-06 23:39 ` Christoph Lameter
2009-10-06 23:55 ` Tejun Heo
2009-10-08 17:57 ` [Patchs vs. percpu-next] Use this_cpu_xx to dynamically allocate counters Christoph Lameter
2009-10-13 11:51 ` Rusty Russell
2009-10-08 18:06 ` Christoph Lameter
2009-10-01 21:25 ` [this_cpu_xx V4 02/20] this_cpu: X86 optimized this_cpu operations cl
2009-10-02 9:18 ` Tejun Heo
2009-10-02 9:59 ` Ingo Molnar
2009-10-03 19:33 ` Pekka Enberg
2009-10-04 16:47 ` Ingo Molnar
2009-10-04 16:51 ` Pekka Enberg
2009-10-01 21:25 ` [this_cpu_xx V4 03/20] Use this_cpu operations for SNMP statistics cl
2009-10-01 21:25 ` [this_cpu_xx V4 04/20] Use this_cpu operations for NFS statistics cl
2009-10-01 21:25 ` [this_cpu_xx V4 05/20] use this_cpu ops for network statistics cl
2009-10-01 21:25 ` [this_cpu_xx V4 06/20] this_cpu_ptr: Straight transformations cl
2009-10-01 21:25 ` [this_cpu_xx V4 07/20] this_cpu_ptr: Eliminate get/put_cpu cl
2009-10-01 21:25 ` [this_cpu_xx V4 08/20] this_cpu_ptr: xfs_icsb_modify_counters does not need "cpu" variable cl
2009-10-01 21:25 ` [this_cpu_xx V4 09/20] Use this_cpu_ptr in crypto subsystem cl
2009-10-01 21:25 ` [this_cpu_xx V4 10/20] Use this_cpu ops for VM statistics cl
2009-10-01 21:25 ` cl [this message]
2009-10-03 10:52 ` [this_cpu_xx V4 11/20] RCU: Use this_cpu operations Tejun Heo
2009-10-01 21:25 ` [this_cpu_xx V4 12/20] Move early initialization of pagesets out of zone_wait_table_init() cl
2009-10-02 14:16 ` Mel Gorman
2009-10-02 17:30 ` Christoph Lameter
2009-10-05 9:35 ` Mel Gorman
2009-10-03 10:29 ` Tejun Heo
2009-10-05 14:39 ` Christoph Lameter
2009-10-05 15:01 ` Tejun Heo
2009-10-05 15:06 ` Christoph Lameter
2009-10-05 15:21 ` Tejun Heo
2009-10-05 15:28 ` Christoph Lameter
2009-10-05 15:41 ` Tejun Heo
2009-10-05 15:39 ` Christoph Lameter
2009-10-01 21:25 ` [this_cpu_xx V4 13/20] this_cpu_ops: page allocator conversion cl
2009-10-02 15:14 ` Mel Gorman
2009-10-02 17:39 ` Christoph Lameter
2009-10-05 9:45 ` Mel Gorman
2009-10-05 14:43 ` Christoph Lameter
2009-10-05 14:55 ` Christoph Lameter
2009-10-06 9:45 ` Mel Gorman
2009-10-06 16:34 ` Christoph Lameter
2009-10-06 17:03 ` Mel Gorman
2009-10-06 17:51 ` Christoph Lameter
2009-10-06 18:36 ` Mel Gorman
2009-10-06 19:06 ` Christoph Lameter
2009-10-07 10:42 ` Mel Gorman
2009-10-01 21:25 ` [this_cpu_xx V4 14/20] this_cpu ops: Remove pageset_notifier cl
2009-10-01 21:25 ` [this_cpu_xx V4 15/20] Use this_cpu operations in slub cl
2009-10-01 21:25 ` [this_cpu_xx V4 16/20] SLUB: Get rid of dynamic DMA kmalloc cache allocation cl
2009-10-01 21:25 ` [this_cpu_xx V4 17/20] this_cpu: Remove slub kmem_cache fields cl
2009-10-01 21:25 ` [this_cpu_xx V4 18/20] Make slub statistics use this_cpu_inc cl
2009-10-01 21:25 ` [this_cpu_xx V4 19/20] this_cpu: slub aggressive use of this_cpu operations in the hotpaths cl
2009-10-01 21:25 ` [this_cpu_xx V4 20/20] SLUB: Experimental new fastpath w/o interrupt disable cl
2009-10-02 9:30 ` [this_cpu_xx V4 00/20] Introduce per cpu atomic operations and avoid per cpu address arithmetic Tejun Heo
2009-10-02 9:54 ` Ingo Molnar
2009-10-02 17:15 ` Christoph Lameter
2009-10-02 17:32 ` Ingo Molnar
2009-10-02 17:49 ` Christoph Lameter
2009-10-02 17:10 ` 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=20091001212559.690311062@gentwo.org \
--to=cl@linux-foundation.org \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=paulmck@linux.vnet.ibm.com \
--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
Powered by JetHome