MIME-Version: 1.0
References: <20230202223409.2812443-1-jstultz@google.com> <20230203020509.GD2948950@paulmck-ThinkPad-P17-Gen-1>
In-Reply-To: <20230203020509.GD2948950@paulmck-ThinkPad-P17-Gen-1>
From:   John Stultz <jstultz@google.com>
Date:   Thu, 2 Feb 2023 19:56:05 -0800
Message-ID: <CANDhNCrSCrCgGDaGstS9=3vEqow=iiF5qWOhbzrstMbD5afXAQ@mail.gmail.com>
Subject: Re: [RFC][PATCH 1/4] locktorture: Add nested_[un]lock() hooks and
 nlocks parameter
To:     paulmck@kernel.org
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Davidlohr Bueso <dave@stgolabs.net>,
        Josh Triplett <josh@joshtriplett.org>,
        Joel Fernandes <joel@joelfernandes.org>,
        Juri Lelli <juri.lelli@redhat.com>,
        Valentin Schneider <vschneid@redhat.com>,
        Dietmar Eggemann <dietmar.eggemann@arm.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Feb 2, 2023 at 6:05 PM Paul E. McKenney <paulmck@kernel.org> wrote:
>
> On Thu, Feb 02, 2023 at 10:34:06PM +0000, John Stultz wrote:
> > In order ot extend locktorture to support lock nesting, add
> > nested_lock() and nested_unlock() hooks to the torture ops.
> >
> > These take a 32bit lockset mask which is generated at random,
> > so some number of locks will be taken before the main lock is
> > taken and released afterwards.
> >
> > Additionally, add nlocks module parameter to allow specifying
> > the number of nested locks to be used.
> >
> > This has been helpful to uncover issues in the proxy-exec
> > series development.
> >
> > This was inspired by locktorture extensions originally implemented
> > by Connor O'Brien, for stress testing the proxy-execution series:
> >   https://lore.kernel.org/lkml/20221003214501.2050087-12-connoro@google.com/
> >
> > Comments or feedback would be greatly appreciated!
>
> I have pulled this in for testing and further review, thank you!
>
> Should either of these files have a locktorture.nlocks=<whatever>
> added?
>
> tools/testing/selftests/rcutorture/configs/lock/LOCK02.boot
> tools/testing/selftests/rcutorture/configs/lock/LOCK05.boot
>
> The first is for mutex_lock and the second for rtmutex_lock.

Potentially? I wasn't aware of these files. Is there some
documentation on the intent behind them?

While the nested locking is useful to cause different lock chains to
test boosting or proxy-execution, I worry they may cause extra noise
that could distract from just thrashing the *mutex lock primitive if
that's the existing intent.

So we might want additional config files for the nested case.

> This did pass a quick "torture.sh --do-none --do-locktorture", which is
> good, but this uses the existing .boot files.

Yeah, probably no effective change in that case.

thanks
-john