summaryrefslogtreecommitdiff
path: root/arch
diff options
context:
space:
mode:
authorHeiko Carstens <heiko.carstens@de.ibm.com>2012-11-08 14:18:47 +0100
committerMartin Schwidefsky <schwidefsky@de.ibm.com>2012-11-13 11:03:09 +0100
commit4bffbb3455372a26816e364fb4448810f7014452 (patch)
treeaf64b28c6bd246a9078edb624fad787e3b4fc1f2 /arch
parent516bad44b9f3bdcb0be6be0252b7557bf7a149e4 (diff)
s390/mm: have 16 byte aligned struct pages
Select HAVE_ALIGNED_STRUCT_PAGE on s390, so that the slub allocator can make use of compare and swap double for lockless updates. This increases the size of struct page to 64 bytes (instead of 56 bytes), however the performance gain justifies the increased size: - now excactly four struct pages fit into a single cache line; the case that accessing a struct page causes two cache line loads does not exist anymore. - calculating the offset of a struct page within the memmap array is only a simple shift instead of a more expensive multiplication. A "hackbench 200 process 200" run on a 32 cpu system did show an 8% runtime improvement. Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Diffstat (limited to 'arch')
-rw-r--r--arch/s390/Kconfig1
1 files changed, 1 insertions, 0 deletions
diff --git a/arch/s390/Kconfig b/arch/s390/Kconfig
index 5dba755a43e6..d385f396dfee 100644
--- a/arch/s390/Kconfig
+++ b/arch/s390/Kconfig
@@ -96,6 +96,7 @@ config S390
select HAVE_MEMBLOCK_NODE_MAP
select HAVE_CMPXCHG_LOCAL
select HAVE_CMPXCHG_DOUBLE
+ select HAVE_ALIGNED_STRUCT_PAGE if SLUB
select HAVE_VIRT_CPU_ACCOUNTING
select VIRT_CPU_ACCOUNTING
select ARCH_DISCARD_MEMBLOCK