linux-kernel - [PATCH 3/3] slub: fill per cpu partial only when free objects larger than one quarter

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri,  2 Dec 2011 16:23:09 +0800
From:	Alex Shi <alex.shi@...el.com>
To:	cl@...ux.com, penberg@...nel.org
Cc:	linux-kernel@...r.kernel.org, linux-mm@...ck.org
Subject: [PATCH 3/3] slub: fill per cpu partial only when free objects larger than one quarter

From: Alex Shi <alexs@...el.com>

Set selection criteria when fill per cpu partial in slow allocation path,
and check the PCP left space before filling, even maybe the data from another
CPU.
The patch can bring another 1.5% performance increase on netperf loopback
testing for our 4 or 2 sockets machines, include sandbridge, core2

Signed-off-by: Alex Shi <alex.shi@...el.com>
---
 mm/slub.c |   43 +++++++++++++++++++++++++++++++------------
 1 files changed, 31 insertions(+), 12 deletions(-)

diff --git a/mm/slub.c b/mm/slub.c
index 65d901f..72df387 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -1542,25 +1542,44 @@ static void *get_partial_node(struct kmem_cache *s,
 
 	spin_lock(&n->list_lock);
 	list_for_each_entry_safe(page, page2, &n->partial, lru) {
-		void *t = acquire_slab(s, n, page, object == NULL);
 		int available;
+		void *t;
+		struct page *oldpage;
+		int pobjects;
 
-		if (!t)
-			break;
 
 		if (!object) {
-			c->page = page;
-			c->node = page_to_nid(page);
-			stat(s, ALLOC_FROM_PARTIAL);
-			object = t;
-			available =  page->objects - page->inuse;
+			t = acquire_slab(s, n, page, object == NULL);
+			if (!t)
+				break;
+			else {
+				c->page = page;
+				c->node = page_to_nid(page);
+				stat(s, ALLOC_FROM_PARTIAL);
+				object = t;
+			}
 		} else {
-			page->freelist = t;
-			available = put_cpu_partial(s, page, 0);
+			oldpage = this_cpu_read(s->cpu_slab->partial);
+			pobjects = oldpage ? oldpage->pobjects : 0;
+
+			if (pobjects > s->cpu_partial / 2)
+				break;
+
+			available =  page->objects - page->inuse;
+			if (available >= s->cpu_partial / 4) {
+				t = acquire_slab(s, n, page, object == NULL);
+				if (!t)
+					break;
+				else {
+					page->freelist = t;
+					if (put_cpu_partial(s, page, 0) >
+							s->cpu_partial / 2)
+						break;
+				}
+			}
 		}
-		if (kmem_cache_debug(s) || available > s->cpu_partial / 2)
+		if (kmem_cache_debug(s))
 			break;
-
 	}
 	spin_unlock(&n->list_lock);
 	return object;
-- 
1.7.0.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/