From: Christoph Lameter <clameter@sgi.com>
To: Matthew Wilcox <matthew@wil.cx>
Cc: akpm@linux-foundation.org, linux-kernel@vger.kernel.org,
linux-mm@kvack.org, Pekka Enberg <penberg@cs.helsinki.fi>
Subject: [patch 07/10] SLUB: Avoid referencing kmem_cache structure in __slab_alloc
Date: Sat, 27 Oct 2007 20:32:03 -0700 [thread overview]
Message-ID: <20071028033259.992768446@sgi.com> (raw)
In-Reply-To: 20071028033156.022983073@sgi.com
[-- Attachment #1: slub_objects_in_per_cpu --]
[-- Type: text/plain, Size: 1742 bytes --]
There is the need to use the objects per slab in the first part of
__slab_alloc() which is still pretty hot. Copy the number of objects
per slab into the kmem_cache_cpu structure. That way we can get the
value from a cache line that we already need to touch. This brings
the kmem_cache_cpu structure up to 4 even words.
There is no increase in the size of kmem_cache_cpu since the size of object
is rounded to the next word.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
---
include/linux/slub_def.h | 1 +
mm/slub.c | 3 ++-
2 files changed, 3 insertions(+), 1 deletion(-)
Index: linux-2.6/include/linux/slub_def.h
===================================================================
--- linux-2.6.orig/include/linux/slub_def.h 2007-10-26 19:09:16.000000000 -0700
+++ linux-2.6/include/linux/slub_def.h 2007-10-27 07:55:12.000000000 -0700
@@ -17,6 +17,7 @@ struct kmem_cache_cpu {
int node;
unsigned int offset;
unsigned int objsize;
+ unsigned int objects;
};
struct kmem_cache_node {
Index: linux-2.6/mm/slub.c
===================================================================
--- linux-2.6.orig/mm/slub.c 2007-10-27 07:52:12.000000000 -0700
+++ linux-2.6/mm/slub.c 2007-10-27 07:55:12.000000000 -0700
@@ -1512,7 +1512,7 @@ load_freelist:
object = c->page->freelist;
c->freelist = object[c->offset];
- c->page->inuse = s->objects;
+ c->page->inuse = c->objects;
c->page->freelist = c->page->end;
c->node = page_to_nid(c->page);
unlock_out:
@@ -1896,6 +1896,7 @@ static void init_kmem_cache_cpu(struct k
c->node = 0;
c->offset = s->offset / sizeof(void *);
c->objsize = s->objsize;
+ c->objects = s->objects;
}
static void init_kmem_cache_node(struct kmem_cache_node *n)
--
next prev parent reply other threads:[~2007-10-28 3:32 UTC|newest]
Thread overview: 35+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-10-28 3:31 [patch 00/10] SLUB: SMP regression tests on Dual Xeon E5345 (8p) and new performance patches Christoph Lameter
2007-10-28 3:31 ` [patch 01/10] SLUB: Consolidate add_partial and add_partial_tail to one function Christoph Lameter
2007-10-28 13:07 ` Pekka J Enberg
2007-10-28 3:31 ` [patch 02/10] SLUB: Noinline some functions to avoid them being folded into alloc/free Christoph Lameter
2007-10-28 13:08 ` Pekka J Enberg
2007-10-29 23:25 ` Matt Mackall
2007-10-28 3:31 ` [patch 03/10] SLUB: Move kmem_cache_node determination into add_full and add_partial Christoph Lameter
2007-10-28 13:09 ` Pekka J Enberg
2007-10-28 3:32 ` [patch 04/10] SLUB: Avoid checking for a valid object before zeroing on the fast path Christoph Lameter
2007-10-28 13:10 ` Pekka J Enberg
2007-10-28 3:32 ` [patch 05/10] SLUB: __slab_alloc() exit path consolidation Christoph Lameter
2007-10-28 13:11 ` Pekka J Enberg
2007-10-28 3:32 ` [patch 06/10] SLUB: Provide unique end marker for each slab Christoph Lameter
2007-10-28 3:32 ` Christoph Lameter [this message]
2007-10-28 13:12 ` [patch 07/10] SLUB: Avoid referencing kmem_cache structure in __slab_alloc Pekka J Enberg
2007-10-30 18:38 ` Andrew Morton
2007-10-28 3:32 ` [patch 08/10] SLUB: Optional fast path using cmpxchg_local Christoph Lameter
2007-10-28 13:05 ` Pekka J Enberg
2007-10-29 2:59 ` Christoph Lameter
2007-10-29 3:34 ` Christoph Lameter
2007-10-30 18:30 ` Andrew Morton
2007-10-30 18:49 ` Andrew Morton
2007-10-30 18:58 ` Christoph Lameter
2007-10-30 19:12 ` Mathieu Desnoyers
2007-10-31 1:52 ` [PATCH] local_t Documentation update 2 Mathieu Desnoyers
2007-10-31 2:28 ` [patch 08/10] SLUB: Optional fast path using cmpxchg_local Mathieu Desnoyers
2007-10-28 3:32 ` [patch 09/10] SLUB: Do our own locking via slab_lock and slab_unlock Christoph Lameter
2007-10-28 15:10 ` Pekka J Enberg
2007-10-28 15:14 ` Pekka J Enberg
2007-10-29 3:03 ` Christoph Lameter
2007-10-29 6:30 ` Pekka Enberg
2007-10-30 4:50 ` Nick Piggin
2007-10-30 18:32 ` Christoph Lameter
2007-10-31 1:17 ` Nick Piggin
2007-10-28 3:32 ` [patch 10/10] SLUB: Restructure slab alloc 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=20071028033259.992768446@sgi.com \
--to=clameter@sgi.com \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=matthew@wil.cx \
--cc=penberg@cs.helsinki.fi \
/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;
as well as URLs for NNTP newsgroup(s).