Messages in this thread | | | Date | Mon, 10 Aug 2009 09:43:44 +0900 | From | KAMEZAWA Hiroyuki <> | Subject | Re: Help Resource Counters Scale Better (v3) |
| |
On Mon, 10 Aug 2009 09:32:29 +0900 KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> 1. you use res_counter_read_positive() in force_empty. It seems force_empty can > go into infinite loop. plz check. (especially when some pages are freed or swapped-in > in other cpu while force_empry runs.) > > 2. In near future, we'll see 256 or 1024 cpus on a system, anyway. > Assume 1024cpu system, 64k*1024=64M is a tolerance. > Can't we calculate max-tolerane as following ? > > tolerance = min(64k * num_online_cpus(), limit_in_bytes/100); > tolerance /= num_online_cpus(); > per_cpu_tolerance = min(16k, tolelance); > > I think automatic runtine adjusting of tolerance will be finally necessary, > but above will not be very bad because we can guarantee 1% tolerance. >
Sorry, one more.
3. As I requested when you pushed softlimit changes to mmotom, plz consider to implement a way to check-and-notify gadget to res_counter. See: http://marc.info/?l=linux-mm&m=124753058921677&w=2
Thanks, -Kame
| |