lkml.org 
[lkml]   [2014]   [Jul]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH RFC tip/core/rcu] Fix attempt to avoid offloading callbacks unless requested
On Sat, Jul 26, 2014 at 09:51:52PM -0400, Pranith Kumar wrote:
> On Fri, Jul 25, 2014 at 10:30 PM, Frederic Weisbecker
> <fweisbec@gmail.com> wrote:
>
> >> I understand that if CONFIG_NO_HZ_FULL_ALL is set then CONFIG_NOCB_CPU_ALL
> >> will also be set and there is no need for this cpumask_or().
> >>
> >> Is there any reason for the coupling between CONFIG_NO_HZ_FULL_ALL
> >> and CONFIG_NOCB_CPU_ALL?
> >
> > Yeah, for any nohz full CPU, we need the corresponding CPU to be rcu_nocb.
> > So if all CPUs are full dynticks, all CPUs must be rcunocb.
> >
> > That said with this patch, the dependency is perhaps not needed anymore.
> >
> >>
> >> I ask because a user can override CONFIG_NO_HZ_FULL_ALL=y at boot time
> >> using the nohz_full= boot time parameter.
> >
> > No, the content of nohz_full= is ignored with CONFIG_NO_HZ_FULL_ALL=y.
> >
>
> Please correct me if I am wrong but that does not seem to be the case.
> If a boot parameter is passed, we are setting up tick_nohz_full_mask
> from tick_nohz_full_setup() and marking tick_nohz_full_running as true.
> Later on we check this flag and skip the CONFIG_NO_HZ_FULL_ALL
> initialization.

You're right, I missed the tick_nohz_full_running check :)

So if nohz_full is passed, we ignore CONFIG_NO_HZ_FULL_ALL. That looks like
the right behaviour though.

Paul what do you think? If we keep that behaviour, Maybe you could blindly do
rcu_nocb_mask |= tick_nohz_full and remove the CONFIG_NO_HZ_FULL_ALL dependency
on RCU_NOCB_ALL?


\
 
 \ /
  Last update: 2014-07-28 23:41    [W:0.057 / U:0.472 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site