lkml.org 
[lkml]   [2021]   [Nov]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Re: Re: Re: Re: Re: Re: Re: [PATCH v1] sched/numa: add per-process numa_balancing
On Thu, Nov 18, 2021 at 11:26:30AM +0800, Gang Li wrote:
> On 11/17/21 6:10 PM, Mel Gorman wrote:
> > On Wed, Nov 17, 2021 at 05:38:28PM +0800, Gang Li wrote:
> > > If those APIs are ok with you, I will send v2 soon.
> > >
> > > 1. prctl(PR_NUMA_BALANCING, PR_SET_THP_DISABLE);
> >
> > It would be (PR_SET_NUMAB_DISABLE, 1)
> >
> > > 2. prctl(PR_NUMA_BALANCING, PR_SET_THP_ENABLE);
> >
> > An enable prctl will have the same problems as
> > prctl(PR_NUMA_BALANCING, PR_SET_NUMA_BALANCING, 0/1) -- it should have
> > meaning if the numa_balancing sysctl is disabled.
> >
> > > 3. prctl(PR_NUMA_BALANCING, PR_GET_THP);
> > >
> >
> > PR_GET_NUMAB_DISABLE
> >
>
> How about this:
>
> 1. prctl(PR_NUMA_BALANCING, PR_SET_NUMAB_DEFAULT); //follow global
> 2. prctl(PR_NUMA_BALANCING, PR_SET_NUMAB_DISABLE); //disable
> 3. prctl(PR_NUMA_BALANCING, PR_SET_NUMAB_ENABLE); //enable

If PR_SET_NUMAB_ENABLE enables numa balancing for a task when
kernel.numa_balancing == 0 instead of returning an error then sure.

> 4. prctl(PR_NUMA_BALANCING, PR_GET_NUMAB);
>
> PR_SET_NUMAB_DISABLE/ENABLE can always have meaning whether the
> numa_balancing sysctl is disabled or not,
>
> --
> Thanks,
> Gang Li
>

--
Mel Gorman
SUSE Labs

\
 
 \ /
  Last update: 2021-11-18 09:59    [W:0.079 / U:0.088 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site