|| ||KAMEZAWA Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>|
|| ||"containers-qjLDD68F18O7TbgM5vRIOg@public.gmane.org" <containers-qjLDD68F18O7TbgM5vRIOg@public.gmane.org>|
|| ||[RFC] memory controller : backgorund reclaim and avoid excessive
|| ||Thu, 14 Feb 2008 17:21:48 +0900|
|| ||"yamamoto-jCdQPDEk3idL9jVzuh4AOg@public.gmane.org" <yamamoto-jCdQPDEk3idL9jVzuh4AOg@public.gmane.org>, xemul-GEFAQzZX7r8dnm+yROfE0A@public.gmane.org,
Hi, these are patches on my stack, now.
I believe test is not enough. Maybe there is BUG.
Before updating patches, I'd like to hear guys' advice against my concept
before going further.
- test on NUMA (machine is not avilable in these days. but I will do)
This series inludes 2 series of patches.
One is updated "back-ground reclaiming with high-low watermark".
One is for avoid locking.
[1/5] -- high-low watermark for resource counter.
[2/5] -- background reclaim with high-low watermark.
[3/5] -- throttle # of direct reclaim
This series tries to improve page reclaiming.
[4/5] -- borrow resource for avoiding counter->lock
[5/5] -- bulk freeing page_cgroup
This series tries to reduce contention on counter->lock and mz->lru_lock.
This 2 are the top most contended lock in lock_stat.
Any commetns are welcome.