linux/arch
Robin Holt 1f9cc3cb6a x86, pat: Update the page flags for memtype atomically instead of using memtype_lock
While testing an application using the xpmem (out of kernel) driver, we
noticed a significant page fault rate reduction of x86_64 with respect
to ia64.  For one test running with 32 cpus, one thread per cpu, it
took 01:08 for each of the threads to vm_insert_pfn 2GB worth of pages.
For the same test running on 256 cpus, one thread per cpu, it took 14:48
to vm_insert_pfn 2 GB worth of pages.

The slowdown was tracked to lookup_memtype which acquires the
spinlock memtype_lock.  This heavily contended lock was slowing down
vm_insert_pfn().

With the cmpxchg on page->flags method, both the 32 cpu and 256 cpu
cases take approx 00:01.3 seconds to complete.

Signed-off-by: Robin Holt <holt@sgi.com>
LKML-Reference: <20100423153627.751194346@gulag1.americas.sgi.com>
Cc: Venkatesh Pallipadi <venkatesh.pallipadi@gmail.com>
Cc: Rafael Wysocki <rjw@novell.com>
Reviewed-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
2010-04-23 15:57:23 -07:00
..
alpha
arm OMAP: hsmmc: fix memory leak 2010-02-10 09:20:33 -08:00
avr32
blackfin
cris
frv
h8300
ia64
m32r
m68k
m68knommu
microblaze microblaze: Invalidate dcache before enabling it 2010-02-08 11:39:18 +01:00
mips MIPS: Don't probe reserved EntryHi bits. 2010-02-10 22:15:46 +01:00
mn10300
parisc parisc: fix tracing of signals 2010-02-12 08:51:58 -08:00
powerpc powerpc: Fix address masking bug in hpte_need_flush() 2010-02-10 13:58:06 +11:00
s390 [S390] Fix struct _lowcore layout. 2010-02-09 09:46:23 +01:00
score
sh sh: Remove superfluous setup_frame_reg call 2010-02-08 10:47:11 +09:00
sparc Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6 2010-02-11 14:00:27 -08:00
um
x86 x86, pat: Update the page flags for memtype atomically instead of using memtype_lock 2010-04-23 15:57:23 -07:00
xtensa
.gitignore
Kconfig