Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Message-ID: <86ead66e34c5c3a59b0bf773ed40767b4282b768.camel@gmx.de>
Subject: Re: [rfc/patch] mm/slub: restore/expand unfreeze_partials() local
 exclusion scope
From:   Mike Galbraith <efault@gmx.de>
To:     Vlastimil Babka <vbabka@suse.cz>,
        Thomas Gleixner <tglx@linutronix.de>,
        LKML <linux-kernel@vger.kernel.org>
Cc:     linux-rt-users@vger.kernel.org,
        Mel Gorman <mgorman@techsingularity.net>,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Date:   Mon, 26 Jul 2021 12:04:23 +0200
In-Reply-To: <b41fa4f2-8368-f33a-10c2-68554b16eb1e@suse.cz>
References: <87tul5p2fa.ffs@nanos.tec.linutronix.de>
         <8c0e0c486056b5185b58998f2cce62619ed3f05c.camel@gmx.de>
         <878s2fnv79.ffs@nanos.tec.linutronix.de>
         <6c0e20dd84084036d5068e445746c3ed7e82ec4b.camel@gmx.de>
         <7431ceb9761c566cf2d1f6f263247acd8d38c4b5.camel@gmx.de>
         <f9935c4c-078c-4b52-5297-64ee22272664@suse.cz>
         <f16b78bd3bb8fecf734017d40274e4c3294554ab.camel@gmx.de>
         <240f104fc6757d8c38fa01342511eda931632d5a.camel@gmx.de>
         <69da2ecd-a797-e264-fbfa-13108dc7a573@suse.cz>
         <84a7bd02cf109c6a5a8c7cc2bfc2898cb98270aa.camel@gmx.de>
         <5be1a703-9a0a-4115-1d69-634e5e8ecefd@suse.cz>
         <bd121f5db01404774dbecc70bd7155f8431d8046.camel@gmx.de>
         <76dedfc3-0497-1776-d006-486b9bfd88da@suse.cz>
         <72a045663bf8f091ae11dd328d5e085541d54fcd.camel@gmx.de>
         <18ca0ce9-3407-61e1-31d6-5c48e80eb5bb@suse.cz>
         <73f032c2-70f1-77b6-9fd2-9aca52fd5b4d@suse.cz>
         <b41fa4f2-8368-f33a-10c2-68554b16eb1e@suse.cz>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.40.2 
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Sun, 2021-07-25 at 21:34 +0200, Vlastimil Babka wrote:
> On 7/25/21 9:12 PM, Vlastimil Babka wrote:
> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0/*
> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 * On !RT we just want to d=
isable preemption, on RT we need
> > the lock
> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 * for real. This happens t=
o match local_lock() semantics.
> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 */
> > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0local_lock(&s->cpu_slab->lo=
ck);
>
> OK I realized (and tglx confirmed) that this will blow up on !RT +
> lockdep if interrupted by ___slab_alloc() that will do
> local_lock_irqsave(). So back to #ifdefs it is. But should work as-is
> for RT testing.

I called it off after 3 1/2 hours.  Bug blew box outta the water 10
times in about half of that, so methinks it's safe to call it dead.

	-Mike