On Tue, May 05, 2020 at 07:56:05PM +0200, SeongJae Park wrote:
> On Tue, 5 May 2020 10:30:36 -0700 "Paul E. McKenney" <paul...@kernel.org>
> wrote:
>
> > On Tue, May 05, 2020 at 07:05:53PM +0200, SeongJae Park wrote:
> > > On Tue, 5 May 2020 09:37:42 -0700 Eric Dumazet <eric.duma...@gmail.com>
> > > wrote:
> > >
> > > >
> > > >
> > > > On 5/5/20 9:31 AM, Eric Dumazet wrote:
> > > > >
> > > > >
> > > > > On 5/5/20 9:25 AM, Eric Dumazet wrote:
> > > > >>
> > > > >>
> > > > >> On 5/5/20 9:13 AM, SeongJae Park wrote:
> > > > >>> On Tue, 5 May 2020 09:00:44 -0700 Eric Dumazet
> > > > >>> <eduma...@google.com> wrote:
> > > > >>>
> > > > >>>> On Tue, May 5, 2020 at 8:47 AM SeongJae Park <sjp...@amazon.com>
> > > > >>>> wrote:
> > > > >>>>>
> > > > >>>>> On Tue, 5 May 2020 08:20:50 -0700 Eric Dumazet
> > > > >>>>> <eric.duma...@gmail.com> wrote:
> > > > >>>>>
> > > > >>>>>>
> > > > >>>>>>
> > > > >>>>>> On 5/5/20 8:07 AM, SeongJae Park wrote:
> > > > >>>>>>> On Tue, 5 May 2020 07:53:39 -0700 Eric Dumazet
> > > > >>>>>>> <eduma...@google.com> wrote:
> > > > >>>>>>>
> > > > >>>>>>
> > > [...]
> > > > >>
> > > > >> I would ask Paul opinion on this issue, because we have many objects
> > > > >> being freed after RCU grace periods.
> > > > >>
> > > > >> If RCU subsystem can not keep-up, I guess other workloads will also
> > > > >> suffer.
> > > > >>
> > > > >> Sure, we can revert patches there and there trying to work around
> > > > >> the issue,
> > > > >> but for objects allocated from process context, we should not have
> > > > >> these problems.
> > > > >>
> > > > >
> > > > > I wonder if simply adjusting rcu_divisor to 6 or 5 would help
> > > > >
> > > > > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> > > > > index
> > > > > d9a49cd6065a20936edbda1b334136ab597cde52..fde833bac0f9f81e8536211b4dad6e7575c1219a
> > > > > 100644
> > > > > --- a/kernel/rcu/tree.c
> > > > > +++ b/kernel/rcu/tree.c
> > > > > @@ -427,7 +427,7 @@ module_param(qovld, long, 0444);
> > > > > static ulong jiffies_till_first_fqs = ULONG_MAX;
> > > > > static ulong jiffies_till_next_fqs = ULONG_MAX;
> > > > > static bool rcu_kick_kthreads;
> > > > > -static int rcu_divisor = 7;
> > > > > +static int rcu_divisor = 6;
> > > > > module_param(rcu_divisor, int, 0644);
> > > > >
> > > > > /* Force an exit from rcu_do_batch() after 3 milliseconds. */
> > > > >
> > > >
> > > > To be clear, you can adjust the value without building a new kernel.
> > > >
> > > > echo 6 >/sys/module/rcutree/parameters/rcu_divisor
> > >
> > > I tried value 6, 5, and 4, but none of those removed the problem.
> >
> > Thank you for checking this!
> >
> > Was your earlier discussion on long RCU readers speculation, or do you
> > have measurements?
>
> It was just a guess without any measurement or dedicated investigation.
OK, another thing to check is the duration of the low-memory episode.
Does this duration exceed the RCU CPU stall warning time? (21 seconds
in mainline, 60 in many distros, but check rcupdate.rcu_cpu_stall_timeout
to be sure.)
Also, any chance of a .config? Or at least the RCU portions? I am
guessing CONFIG_PREEMPT=n, for example.
Thanx, Paul
