mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Nick Piggin <npiggin@suse.de>
To: netdev@vger.kernel.org,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	yanmin_zhang@linux.intel.com, David Miller <davem@davemloft.net>,
	Christoph Lameter <clameter@engr.sgi.com>,
	Eric Dumazet <dada1@cosmosbay.com>
Subject: [rfc][patch 2/3] slab: introduce SMP alignment
Date: Mon, 3 Mar 2008 10:35:29 +0100	[thread overview]
Message-ID: <20080303093529.GB15091@wotan.suse.de> (raw)
In-Reply-To: <20080303093449.GA15091@wotan.suse.de>


Introduce SLAB_SMP_ALIGN, for allocations where false sharing must be
minimised on SMP systems.

Signed-off-by: Nick Piggin <npiggin@suse.de>
---
Index: linux-2.6/include/linux/slab.h
===================================================================
--- linux-2.6.orig/include/linux/slab.h
+++ linux-2.6/include/linux/slab.h
@@ -22,7 +22,8 @@
 #define SLAB_RED_ZONE		0x00000400UL	/* DEBUG: Red zone objs in a cache */
 #define SLAB_POISON		0x00000800UL	/* DEBUG: Poison objects */
 #define SLAB_HWCACHE_ALIGN	0x00002000UL	/* Align objs on cache lines */
-#define SLAB_CACHE_DMA		0x00004000UL	/* Use GFP_DMA memory */
+#define SLAB_SMP_ALIGN		0x00004000UL	/* Align on cachelines for SMP*/
+#define SLAB_CACHE_DMA		0x00008000UL	/* Use GFP_DMA memory */
 #define SLAB_STORE_USER		0x00010000UL	/* DEBUG: Store the last owner for bug hunting */
 #define SLAB_PANIC		0x00040000UL	/* Panic if kmem_cache_create() fails */
 #define SLAB_DESTROY_BY_RCU	0x00080000UL	/* Defer freeing slabs to RCU */
Index: linux-2.6/mm/slab.c
===================================================================
--- linux-2.6.orig/mm/slab.c
+++ linux-2.6/mm/slab.c
@@ -174,13 +174,13 @@
 /* Legal flag mask for kmem_cache_create(). */
 #if DEBUG
 # define CREATE_MASK	(SLAB_RED_ZONE | \
-			 SLAB_POISON | SLAB_HWCACHE_ALIGN | \
+			 SLAB_POISON | SLAB_HWCACHE_ALIGN | SLAB_SMP_ALIGN | \
 			 SLAB_CACHE_DMA | \
 			 SLAB_STORE_USER | \
 			 SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \
 			 SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD)
 #else
-# define CREATE_MASK	(SLAB_HWCACHE_ALIGN | \
+# define CREATE_MASK	(SLAB_HWCACHE_ALIGN | SLAB_SMP_ALIGN | \
 			 SLAB_CACHE_DMA | \
 			 SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \
 			 SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD)
@@ -2245,6 +2245,9 @@ kmem_cache_create (const char *name, siz
 		ralign = BYTES_PER_WORD;
 	}
 
+	if ((flags & SLAB_SMP_ALIGN) && num_possible_cpus() > 1)
+		ralign = max_t(unsigned long, ralign, cache_line_size());
+
 	/*
 	 * Redzoning and user store require word alignment or possibly larger.
 	 * Note this will be overridden by architecture or caller mandated
Index: linux-2.6/mm/slub.c
===================================================================
--- linux-2.6.orig/mm/slub.c
+++ linux-2.6/mm/slub.c
@@ -1903,6 +1903,9 @@ static unsigned long calculate_alignment
 		align = max(align, ralign);
 	}
 
+	if ((flags & SLAB_SMP_ALIGN) && num_possible_cpus() > 1)
+		align = max_t(unsigned long, align, cache_line_size());
+
 	if (align < ARCH_SLAB_MINALIGN)
 		align = ARCH_SLAB_MINALIGN;
 

  reply	other threads:[~2008-03-03  9:35 UTC|newest]

Thread overview: 68+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-03-03  9:34 [rfc][patch 1/3] slub: fix small HWCACHE_ALIGN alignment Nick Piggin
2008-03-03  9:35 ` Nick Piggin [this message]
2008-03-03 19:06   ` [rfc][patch 2/3] slab: introduce SMP alignment Christoph Lameter
2008-03-03 20:03     ` Nick Piggin
2008-03-03 20:09       ` Christoph Lameter
2008-03-03 20:12         ` Nick Piggin
2008-03-03 20:17           ` Christoph Lameter
2008-03-03 20:24             ` Nick Piggin
2008-03-03 20:41               ` Pekka Enberg
2008-03-03 21:23                 ` Christoph Lameter
2008-03-03 21:31               ` Christoph Lameter
2008-03-05  0:16                 ` Nick Piggin
2008-03-07  4:37                 ` Nick Piggin
2008-03-07  5:11                   ` Christoph Lameter
2008-03-07  5:19                     ` Nick Piggin
2008-03-07  5:26                       ` Christoph Lameter
2008-03-07  5:37                         ` Nick Piggin
2008-03-11  7:13                         ` Nick Piggin
2008-03-12  6:21                           ` Christoph Lameter
2008-03-03  9:36 ` [rfc][patch 3/3] use SLAB_ALIGN_SMP Nick Piggin
2008-03-03  9:53   ` Eric Dumazet
2008-03-03 12:41     ` Nick Piggin
2008-03-03 13:00       ` Eric Dumazet
2008-03-03 13:46         ` Nick Piggin
2008-03-03 13:53           ` Pekka Enberg
2008-03-03 14:15             ` Eric Dumazet
2008-03-03 19:10               ` Christoph Lameter
2008-03-03 19:09             ` Christoph Lameter
2008-03-03 20:10               ` Nick Piggin
2008-03-03 20:12                 ` Christoph Lameter
2008-03-03 20:18                   ` Nick Piggin
2008-03-03 21:14                     ` Christoph Lameter
2008-03-03  9:44 ` [rfc][patch 1/3] slub: fix small HWCACHE_ALIGN alignment Pekka Enberg
2008-03-03 12:28   ` Nick Piggin
2008-03-03 19:08 ` Christoph Lameter
2008-03-03 20:06   ` Nick Piggin
2008-03-03 20:10     ` Christoph Lameter
2008-03-03 20:17       ` Nick Piggin
2008-03-03 21:16         ` Christoph Lameter
2008-03-03 21:30           ` Pekka Enberg
2008-03-03 21:32             ` Christoph Lameter
2008-03-03 21:35               ` Pekka Enberg
2008-03-05  0:28                 ` Nick Piggin
2008-03-05 20:56                   ` Christoph Lameter
2008-03-06  2:49                     ` Nick Piggin
2008-03-06 22:53                       ` Christoph Lameter
2008-03-07  2:04                         ` Nick Piggin
2008-03-07  2:20                           ` Christoph Lameter
2008-03-07  2:25                             ` Nick Piggin
2008-03-07  2:27                               ` Christoph Lameter
2008-03-07  2:33                                 ` Nick Piggin
2008-03-07  2:33                                   ` Christoph Lameter
2008-03-07  5:23                                     ` Nick Piggin
2008-03-05  0:08               ` Nick Piggin
2008-03-05  0:06           ` Nick Piggin
2008-03-05  0:10             ` David Miller
2008-03-05 21:06               ` Christoph Lameter
2008-03-06  2:57                 ` Nick Piggin
2008-03-06 22:56                   ` Christoph Lameter
2008-03-07  2:23                     ` Nick Piggin
2008-03-07  2:26                       ` Christoph Lameter
2008-03-07  2:32                         ` Nick Piggin
2008-03-07  2:54                           ` Christoph Lameter
2008-03-07  3:10                             ` Nick Piggin
2008-03-07  3:18                               ` Christoph Lameter
2008-03-07  3:22                                 ` Nick Piggin
2008-03-07  3:58                                   ` Christoph Lameter
2008-03-07  4:05                                     ` Nick Piggin

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=20080303093529.GB15091@wotan.suse.de \
    --to=npiggin@suse.de \
    --cc=clameter@engr.sgi.com \
    --cc=dada1@cosmosbay.com \
    --cc=davem@davemloft.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=yanmin_zhang@linux.intel.com \
    /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®