Messages in this thread | | | Date | Wed, 20 Jan 2010 18:16:43 +0530 | From | Balbir Singh <> | Subject | Re: [PATCH mmotm] memcg use generic percpu allocator instead of private one |
| |
On Wednesday 20 January 2010 03:17 PM, KAMEZAWA Hiroyuki wrote: > On Wed, 20 Jan 2010 15:07:52 +0530 > Balbir Singh <balbir@linux.vnet.ibm.com> wrote: > >> On Wednesday 20 January 2010 12:48 PM, KAMEZAWA Hiroyuki wrote: >>> This patch is onto mmotm Jan/15. >>> = >>> From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> >>> >>> When per-cpu counter for memcg was implemneted, dynamic percpu allocator >>> was not very good. But now, we have good one and useful macros. >>> This patch replaces memcg's private percpu counter implementation with >>> generic dynamic percpu allocator and macros. >>> >>> The benefits are >>> - We can remove private implementation. >>> - The counters will be NUMA-aware. (Current one is not...) >>> - This patch reduces sizeof(struct mem_cgroup). Then, >>> struct mem_cgroup may be fit in page size on small config. >>> >>> By this, size of text is reduced. >>> [Before] >>> [kamezawa@bluextal mmotm-2.6.33-Jan15]$ size mm/memcontrol.o >>> text data bss dec hex filename >>> 24373 2528 4132 31033 7939 mm/memcontrol.o >>> [After] >>> [kamezawa@bluextal mmotm-2.6.33-Jan15]$ size mm/memcontrol.o >>> text data bss dec hex filename >>> 23913 2528 4132 30573 776d mm/memcontrol.o >>> >>> This includes no functional changes. >>> >>> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> >> >> >> Before review, could you please post parallel pagefault data on a large >> system, since root now uses these per cpu counters and its overhead is >> now dependent on these counters. Also the data read from root cgroup is >> also dependent on these, could you make sure that is not broken. >> > No number difference before/after patch on my SMP quick test. > But I don't have NUMA. Could you test on NUMA ? > > I'll measure again tomorrow if I have machine time.
I'll do the same as well if possible.
Balbir
| |