linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Dave Hansen <dave@sr71.net>
To: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org, akpm@linux-foundation.org,
	penberg@kernel.org, cl@linux-foundation.org,
	Dave Hansen <dave@sr71.net>
Subject: [RFC][PATCH 2/9] mm: slub: abstract out double cmpxchg option
Date: Tue, 14 Jan 2014 10:00:46 -0800	[thread overview]
Message-ID: <20140114180046.C897727E@viggo.jf.intel.com> (raw)
In-Reply-To: <20140114180042.C1C33F78@viggo.jf.intel.com>


From: Dave Hansen <dave.hansen@linux.intel.com>

I found this useful to have in my testing.  I would like to have
it available for a bit, at least until other folks have had a
chance to do some testing with it.

We should probably just pull the help text and the description
out of this so that folks are not prompted for it instead of
ripping it out entirely.

Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
---

 b/include/linux/mm_types.h |    3 +--
 b/init/Kconfig             |   11 +++++++++++
 b/mm/slub.c                |    9 +++------
 3 files changed, 15 insertions(+), 8 deletions(-)

diff -puN include/linux/mm_types.h~abstract-out-slub-double-cmpxchg-operation include/linux/mm_types.h
--- a/include/linux/mm_types.h~abstract-out-slub-double-cmpxchg-operation	2014-01-14 09:57:56.410635912 -0800
+++ b/include/linux/mm_types.h	2014-01-14 09:57:56.417636226 -0800
@@ -73,8 +73,7 @@ struct page {
 		};
 
 		union {
-#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
-	defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+#if defined(CONFIG_SLUB_ATTEMPT_CMPXCHG_DOUBLE)
 			/* Used for cmpxchg_double in slub */
 			unsigned long counters;
 #else
diff -puN init/Kconfig~abstract-out-slub-double-cmpxchg-operation init/Kconfig
--- a/init/Kconfig~abstract-out-slub-double-cmpxchg-operation	2014-01-14 09:57:56.412636002 -0800
+++ b/init/Kconfig	2014-01-14 09:57:56.418636271 -0800
@@ -1603,6 +1603,17 @@ config SLUB_CPU_PARTIAL
 	  which requires the taking of locks that may cause latency spikes.
 	  Typically one would choose no for a realtime system.
 
+config SLUB_ATTEMPT_CMPXCHG_DOUBLE
+	default y
+	depends on SLUB && HAVE_CMPXCHG_DOUBLE && HAVE_ALIGNED_STRUCT_PAGE
+	bool "SLUB: attempt to use double-cmpxchg operations"
+	help
+	  Some CPUs support instructions that let you do a large double-word
+	  atomic cmpxchg operation.  This keeps the SLUB fastpath from
+	  needing to disable interrupts.
+
+	  If you are unsure, say y.
+
 config MMAP_ALLOW_UNINITIALIZED
 	bool "Allow mmapped anonymous memory to be uninitialized"
 	depends on EXPERT && !MMU
diff -puN mm/slub.c~abstract-out-slub-double-cmpxchg-operation mm/slub.c
--- a/mm/slub.c~abstract-out-slub-double-cmpxchg-operation	2014-01-14 09:57:56.414636092 -0800
+++ b/mm/slub.c	2014-01-14 09:57:56.420636361 -0800
@@ -362,8 +362,7 @@ static inline bool __cmpxchg_double_slab
 		const char *n)
 {
 	VM_BUG_ON(!irqs_disabled());
-#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
-    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+#if defined(CONFIG_SLUB_ATTEMPT_CMPXCHG_DOUBLE)
 	if (s->flags & __CMPXCHG_DOUBLE) {
 		if (cmpxchg_double(&page->freelist, &page->counters,
 			freelist_old, counters_old,
@@ -398,8 +397,7 @@ static inline bool cmpxchg_double_slab(s
 		void *freelist_new, unsigned long counters_new,
 		const char *n)
 {
-#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
-    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+#if defined(CONFIG_SLUB_ATTEMPT_CMPXCHG_DOUBLE)
 	if (s->flags & __CMPXCHG_DOUBLE) {
 		if (cmpxchg_double(&page->freelist, &page->counters,
 			freelist_old, counters_old,
@@ -3078,8 +3076,7 @@ static int kmem_cache_open(struct kmem_c
 		}
 	}
 
-#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
-    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
+#if defined(CONFIG_SLUB_ATTEMPT_CMPXCHG_DOUBLE)
 	if (system_has_cmpxchg_double() && (s->flags & SLAB_DEBUG_FLAGS) == 0)
 		/* Enable fast mode */
 		s->flags |= __CMPXCHG_DOUBLE;
_

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  parent reply	other threads:[~2014-01-14 18:01 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-01-14 18:00 [RFC][PATCH 0/9] re-shrink 'struct page' when SLUB is on Dave Hansen
2014-01-14 18:00 ` [RFC][PATCH 1/9] mm: slab/slub: use page->list consistently instead of page->lru Dave Hansen
2014-01-14 19:31   ` Christoph Lameter
2014-01-15  2:31   ` David Rientjes
2014-01-15  6:58     ` Dave Hansen
2014-01-15  7:16       ` David Rientjes
2014-01-16  0:11   ` Kirill A. Shutemov
2014-01-14 18:00 ` Dave Hansen [this message]
2014-01-14 19:49   ` [RFC][PATCH 2/9] mm: slub: abstract out double cmpxchg option Christoph Lameter
2014-01-14 21:41     ` Dave Hansen
2014-01-15  2:37       ` David Rientjes
2014-01-16 16:45       ` Christoph Lameter
2014-01-16 17:13         ` Dave Hansen
2014-01-14 18:00 ` [RFC][PATCH 3/9] mm: page->pfmemalloc only used by slab/skb Dave Hansen
2014-01-14 19:49   ` Christoph Lameter
2014-01-14 22:17     ` Dave Hansen
2014-01-15  2:45       ` David Rientjes
2014-01-16  0:16   ` Kirill A. Shutemov
2014-01-14 18:00 ` [RFC][PATCH 4/9] mm: slabs: reset page at free Dave Hansen
2014-01-15  2:48   ` David Rientjes
2014-01-16 18:35     ` Dave Hansen
2014-01-16 18:32   ` Christoph Lameter
2014-01-14 18:00 ` [RFC][PATCH 5/9] mm: rearrange struct page Dave Hansen
2014-01-16  0:20   ` Kirill A. Shutemov
2014-01-16 18:34   ` Christoph Lameter
2014-01-16 22:29     ` Dave Hansen
2014-01-17 14:58       ` Christoph Lameter
2014-01-14 18:01 ` [RFC][PATCH 6/9] mm: slub: rearrange 'struct page' fields Dave Hansen
2014-01-14 18:01 ` [RFC][PATCH 7/9] mm: slub: remove 'struct page' alignment restrictions Dave Hansen
2014-01-14 18:01 ` [RFC][PATCH 8/9] mm: slub: cleanups after code churn Dave Hansen
2014-01-14 18:01 ` [RFC][PATCH 9/9] mm: fix alignment checks on 32-bit Dave Hansen

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=20140114180046.C897727E@viggo.jf.intel.com \
    --to=dave@sr71.net \
    --cc=akpm@linux-foundation.org \
    --cc=cl@linux-foundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=penberg@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