Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
Message-ID: <167d30f439d171912b1ef584f20219e67a009de8.camel@redhat.com>
Subject: Re: [PATCH 6/6] mm/page_alloc: Remotely drain per-cpu lists
From:   Nicolas Saenz Julienne <nsaenzju@redhat.com>
To:     Mel Gorman <mgorman@techsingularity.net>,
        Andrew Morton <akpm@linux-foundation.org>
Cc:     Marcelo Tosatti <mtosatti@redhat.com>,
        Vlastimil Babka <vbabka@suse.cz>,
        Michal Hocko <mhocko@kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Linux-MM <linux-mm@kvack.org>
Date:   Fri, 13 May 2022 17:19:18 +0200
In-Reply-To: <20220513150402.GJ3441@techsingularity.net>
References: <20220512085043.5234-1-mgorman@techsingularity.net>
         <20220512085043.5234-7-mgorman@techsingularity.net>
         <20220512123743.5be26b3ad4413f20d5f46564@linux-foundation.org>
         <20220513150402.GJ3441@techsingularity.net>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.1 (3.44.1-1.fc36) 
MIME-Version: 1.0
Precedence: bulk

On Fri, 2022-05-13 at 16:04 +0100, Mel Gorman wrote:
> On Thu, May 12, 2022 at 12:37:43PM -0700, Andrew Morton wrote:
> > On Thu, 12 May 2022 09:50:43 +0100 Mel Gorman <mgorman@techsingularity.=
net> wrote:
> >=20
> > > From: Nicolas Saenz Julienne <nsaenzju@redhat.com>
> > >=20
> > > Some setups, notably NOHZ_FULL CPUs, are too busy to handle the per-c=
pu
> > > drain work queued by __drain_all_pages(). So introduce a new mechanis=
m to
> > > remotely drain the per-cpu lists. It is made possible by remotely loc=
king
> > > 'struct per_cpu_pages' new per-cpu spinlocks. A benefit of this new s=
cheme
> > > is that drain operations are now migration safe.
> > >=20
> > > There was no observed performance degradation vs. the previous scheme=
.
> > > Both netperf and hackbench were run in parallel to triggering the
> > > __drain_all_pages(NULL, true) code path around ~100 times per second.
> > > The new scheme performs a bit better (~5%), although the important po=
int
> > > here is there are no performance regressions vs. the previous mechani=
sm.
> > > Per-cpu lists draining happens only in slow paths.
> > >=20
> > > Minchan Kim tested this independently and reported;
> > >=20
> > > 	My workload is not NOHZ CPUs but run apps under heavy memory
> > > 	pressure so they goes to direct reclaim and be stuck on
> > > 	drain_all_pages until work on workqueue run.
> > >=20
> > > 	unit: nanosecond
> > > 	max(dur)        avg(dur)                count(dur)
> > > 	166713013       487511.77786438033      1283
> > >=20
> > > 	From traces, system encountered the drain_all_pages 1283 times and
> > > 	worst case was 166ms and avg was 487us.
> > >=20
> > > 	The other problem was alloc_contig_range in CMA. The PCP draining
> > > 	takes several hundred millisecond sometimes though there is no
> > > 	memory pressure or a few of pages to be migrated out but CPU were
> > > 	fully booked.
> > >=20
> > > 	Your patch perfectly removed those wasted time.
> >=20
> > I'm not getting a sense here of the overall effect upon userspace
> > performance.  As Thomas said last year in
> > https://lkml.kernel.org/r/87v92sgt3n.ffs@tglx
> >=20
> > : The changelogs and the cover letter have a distinct void vs. that whi=
ch
> > : means this is just another example of 'scratch my itch' changes w/o
> > : proper justification.
> >=20
> > Is there more to all of this than itchiness and if so, well, you know
> > the rest ;)
> >=20
>=20
> I think Minchan's example is clear-cut.  The draining operation can take
> an arbitrary amount of time waiting for the workqueue to run on each CPU
> and can cause severe delays under reclaim or CMA and the patch fixes
> it. Maybe most users won't even notice but I bet phone users do if a
> camera app takes too long to open.
>=20
> The first paragraphs was written by Nicolas and I did not want to modify
> it heavily and still put his Signed-off-by on it. Maybe it could have
> been clearer though because "too busy" is vague when the actual intent
> is to avoid interfering with RT tasks. Does this sound better to you?
>=20
> 	Some setups, notably NOHZ_FULL CPUs, may be running realtime or
> 	latency-sensitive applications that cannot tolerate interference
> 	due to per-cpu drain work queued by __drain_all_pages(). Introduce
> 	a new mechanism to remotely drain the per-cpu lists. It is made
> 	possible by remotely locking 'struct per_cpu_pages' new per-cpu
> 	spinlocks. This has two advantages, the time to drain is more
> 	predictable and other unrelated tasks are not interrupted.
>=20
> You raise a very valid point with Thomas' mail and it is a concern that
> the local_lock is no longer strictly local. We still need preemption to
> be disabled between the percpu lookup and the lock acquisition but that
> can be done with get_cpu_var() to make the scope clear.

This isn't going to work in RT :(

get_cpu_var() disables preemption hampering RT spinlock use. There is more =
to
it in Documentation/locking/locktypes.rst.

Regards,

--=20
Nicol=C3=A1s S=C3=A1enz