From: Christoph Lameter <clameter@sgi.com>
To: linux-kernel@vger.kernel.org
Cc: linux-mm@vger.kernel.org
Cc: suresh.b.siddha@intel.com
Cc: corey.d.gough@intel.com
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: akpm@linux-foundation.org
Subject: [patch 07/10] SLUB: Optimize cacheline use for zeroing
Date: Sat, 07 Jul 2007 20:49:59 -0700 [thread overview]
Message-ID: <20070708035017.607053208@sgi.com> (raw)
In-Reply-To: 20070708034952.022985379@sgi.com
[-- Attachment #1: slub_optimize_zeroing --]
[-- Type: text/plain, Size: 2472 bytes --]
We touch a cacheline in the kmem_cache structure for zeroing to get the
size. However, the hot paths in slab_alloc and slab_free do not reference
any other fields in kmem_cache.
Add a new field to kmem_cache_cpu that contains the object size. That
cacheline must already be used. So we save one cacheline on every
slab_alloc.
We need to update the kmem_cache_cpu object size if an aliasing operation
changes the objsize of an non debug slab.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
---
include/linux/slub_def.h | 1 +
mm/slub.c | 14 ++++++++++++--
2 files changed, 13 insertions(+), 2 deletions(-)
Index: linux-2.6.22-rc6-mm1/include/linux/slub_def.h
===================================================================
--- linux-2.6.22-rc6-mm1.orig/include/linux/slub_def.h 2007-07-07 13:56:24.000000000 -0700
+++ linux-2.6.22-rc6-mm1/include/linux/slub_def.h 2007-07-07 15:52:37.000000000 -0700
@@ -16,6 +16,7 @@ struct kmem_cache_cpu {
struct page *page;
int node;
unsigned int offset;
+ unsigned int objsize;
};
struct kmem_cache_node {
Index: linux-2.6.22-rc6-mm1/mm/slub.c
===================================================================
--- linux-2.6.22-rc6-mm1.orig/mm/slub.c 2007-07-07 13:56:24.000000000 -0700
+++ linux-2.6.22-rc6-mm1/mm/slub.c 2007-07-07 17:49:25.000000000 -0700
@@ -1571,7 +1571,7 @@ static void __always_inline *slab_alloc(
local_irq_restore(flags);
if (unlikely((gfpflags & __GFP_ZERO) && object))
- memset(object, 0, s->objsize);
+ memset(object, 0, c->objsize);
return object;
}
@@ -1864,8 +1864,9 @@ static void init_kmem_cache_cpu(struct k
{
c->page = NULL;
c->freelist = NULL;
- c->offset = s->offset / sizeof(void *);
c->node = 0;
+ c->offset = s->offset / sizeof(void *);
+ c->objsize = s->objsize;
}
static void init_kmem_cache_node(struct kmem_cache_node *n)
@@ -3173,12 +3174,21 @@ struct kmem_cache *kmem_cache_create(con
down_write(&slub_lock);
s = find_mergeable(size, align, flags, ctor, ops);
if (s) {
+ int cpu;
+
s->refcount++;
/*
* Adjust the object sizes so that we clear
* the complete object on kzalloc.
*/
s->objsize = max(s->objsize, (int)size);
+
+ /*
+ * And then we need to update the object size in the
+ * per cpu structures
+ */
+ for_each_online_cpu(cpu)
+ get_cpu_slab(s, cpu)->objsize = s->objsize;
s->inuse = max_t(int, s->inuse, ALIGN(size, sizeof(void *)));
up_write(&slub_lock);
--
next prev parent reply other threads:[~2007-07-08 3:52 UTC|newest]
Thread overview: 86+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-07-08 3:49 [patch 00/10] [RFC] SLUB patches for more functionality, performance and maintenance Christoph Lameter
2007-07-08 3:49 ` [patch 01/10] SLUB: Direct pass through of page size or higher kmalloc requests Christoph Lameter
2007-07-08 3:49 ` [patch 02/10] SLUB: Avoid page struct cacheline bouncing due to remote frees to cpu slab Christoph Lameter
2007-07-08 3:49 ` [patch 03/10] SLUB: Do not use page->mapping Christoph Lameter
2007-07-08 3:49 ` [patch 04/10] SLUB: Move page->offset to kmem_cache_cpu->offset Christoph Lameter
2007-07-08 3:49 ` [patch 05/10] SLUB: Avoid touching page struct when freeing to per cpu slab Christoph Lameter
2007-07-08 3:49 ` [patch 06/10] SLUB: Place kmem_cache_cpu structures in a NUMA aware way Christoph Lameter
2007-07-08 3:49 ` Christoph Lameter [this message]
2007-07-08 3:50 ` [patch 08/10] SLUB: Single atomic instruction alloc/free using cmpxchg Christoph Lameter
2007-07-08 3:50 ` [patch 09/10] Remove the SLOB allocator for 2.6.23 Christoph Lameter
2007-07-08 7:51 ` Ingo Molnar
2007-07-08 9:43 ` Nick Piggin
2007-07-08 9:54 ` Ingo Molnar
2007-07-08 10:23 ` Nick Piggin
2007-07-08 10:42 ` Ingo Molnar
2007-07-08 18:02 ` Andrew Morton
2007-07-09 2:57 ` Nick Piggin
2007-07-09 11:04 ` Pekka Enberg
2007-07-09 11:16 ` Nick Piggin
2007-07-09 12:47 ` Pekka Enberg
2007-07-09 13:46 ` Pekka J Enberg
2007-07-09 16:08 ` Christoph Lameter
2007-07-10 8:17 ` Pekka J Enberg
2007-07-10 8:27 ` Nick Piggin
2007-07-10 9:31 ` Pekka Enberg
2007-07-10 10:09 ` Nick Piggin
2007-07-10 12:02 ` Matt Mackall
2007-07-10 12:57 ` Pekka J Enberg
2007-07-10 22:12 ` Christoph Lameter
2007-07-10 22:40 ` Matt Mackall
2007-07-10 22:50 ` Christoph Lameter
2007-07-09 16:06 ` Christoph Lameter
2007-07-09 16:51 ` Andrew Morton
2007-07-09 17:26 ` Christoph Lameter
2007-07-09 18:00 ` Andrew Morton
2007-07-10 1:43 ` Nick Piggin
2007-07-10 1:56 ` Christoph Lameter
2007-07-10 2:02 ` Nick Piggin
2007-07-10 2:11 ` Christoph Lameter
2007-07-10 7:09 ` Nick Piggin
2007-07-10 22:09 ` Christoph Lameter
2007-07-10 23:12 ` Matt Mackall
2007-07-10 8:32 ` Matt Mackall
2007-07-10 9:01 ` Håvard Skinnemoen
2007-07-10 9:11 ` Nick Piggin
2007-07-10 9:21 ` Håvard Skinnemoen
2007-07-11 1:37 ` Christoph Lameter
2007-07-11 2:06 ` Matt Mackall
2007-07-11 18:06 ` Christoph Lameter
2007-07-11 18:25 ` Pekka J Enberg
2007-07-11 18:33 ` Christoph Lameter
2007-07-11 18:36 ` Pekka J Enberg
2007-07-12 0:33 ` Nick Piggin
2007-07-09 23:09 ` Matt Mackall
2007-07-10 1:41 ` Nick Piggin
2007-07-10 1:51 ` Christoph Lameter
2007-07-10 1:58 ` Nick Piggin
2007-07-10 6:22 ` Matt Mackall
2007-07-10 7:03 ` Nick Piggin
2007-07-10 2:32 ` Matt Mackall
2007-07-09 21:57 ` Matt Mackall
2007-07-09 12:31 ` Matthieu CASTET
2007-07-09 16:00 ` Christoph Lameter
2007-07-09 20:52 ` Matt Mackall
2007-07-08 3:50 ` [patch 10/10] Remove slab in 2.6.24 Christoph Lameter
2007-07-08 4:37 ` [patch 00/10] [RFC] SLUB patches for more functionality, performance and maintenance David Miller
2007-07-09 15:45 ` Christoph Lameter
2007-07-09 19:43 ` David Miller
2007-07-09 21:21 ` Christoph Lameter
2007-07-08 11:20 ` Andi Kleen
2007-07-09 15:50 ` Christoph Lameter
2007-07-09 15:59 ` Martin Bligh
2007-07-09 18:11 ` Christoph Lameter
2007-07-09 21:00 ` Martin Bligh
2007-07-09 21:44 ` Mathieu Desnoyers
2007-07-09 21:55 ` Christoph Lameter
2007-07-09 22:58 ` Mathieu Desnoyers
2007-07-09 23:08 ` Christoph Lameter
2007-07-10 5:16 ` [PATCH] x86_64 - Use non locked version for local_cmpxchg() Mathieu Desnoyers
2007-07-10 20:46 ` Christoph Lameter
2007-07-10 0:55 ` [patch 00/10] [RFC] SLUB patches for more functionality, performance and maintenance Christoph Lameter
2007-07-10 8:27 ` Mathieu Desnoyers
2007-07-10 18:38 ` Christoph Lameter
2007-07-10 20:59 ` Mathieu Desnoyers
2007-08-13 22:18 ` Mathieu Desnoyers
2007-08-13 22:28 ` 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=20070708035017.607053208@sgi.com \
--to=clameter@sgi.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@vger.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