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;
MIME-Version: 1.0
References: <20210425224540.GA1312438@paulmck-ThinkPad-P17-Gen-1>
 <20210425224709.1312655-1-paulmck@kernel.org> <20210426040736.GS1401198@tassilo.jf.intel.com>
 <20210426152801.GY975577@paulmck-ThinkPad-P17-Gen-1> <20210426160008.GY1401198@tassilo.jf.intel.com>
 <20210426161425.GC975577@paulmck-ThinkPad-P17-Gen-1> <20210426175627.GZ1401198@tassilo.jf.intel.com>
 <20210426182433.GD975577@paulmck-ThinkPad-P17-Gen-1>
In-Reply-To: <20210426182433.GD975577@paulmck-ThinkPad-P17-Gen-1>
From:   Luming Yu <luming.yu@gmail.com>
Date:   Wed, 28 Apr 2021 12:49:12 +0800
Message-ID: <CAJRGBZxgJuc2OJfvT_k3Xmk_qsxzm=xX5wm4NqeRo7vkTcZk6w@mail.gmail.com>
Subject: Re: [PATCH v10 clocksource 1/7] clocksource: Provide module
 parameters to inject delays in watchdog
To:     paulmck@kernel.org
Cc:     Andi Kleen <ak@linux.intel.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        LKML <linux-kernel@vger.kernel.org>, john.stultz@linaro.org,
        sboyd@kernel.org, corbet@lwn.net, Mark.Rutland@arm.com,
        maz@kernel.org, kernel-team@fb.com, neeraju@codeaurora.org,
        feng.tang@intel.com, zhengjun.xing@intel.com,
        Chris Mason <clm@fb.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

We 'd expect to see clock_source watchdog can avoid to do wrong thing
due to the injected delay or
in real life delay by doing tsc sync-re-check by applying the patch-set.
However , the noise is still cause wrong actions and the patch doesn't
defeat the injected's delay
please correct me if  I'm wrong.

parameters]# cat *
1
1
-1
3
8

[62939.809615] clocksource: clocksource_watchdog_inject_delay():
Injecting delay.
[62939.816867] clocksource: clocksource_watchdog_inject_delay():
Injecting delay.
[62939.824094] clocksource: clocksource_watchdog_inject_delay():
Injecting delay.
[62939.831314] clocksource: clocksource_watchdog_inject_delay():
Injecting delay.
[62939.838536] clocksource: timekeeping watchdog on CPU26: hpet
read-back delay of 7220833ns, attempt 4, marking unstable
[62939.849230] tsc: Marking TSC unstable due to clocksource watchdog
[62939.855340] TSC found unstable after boot, most likely due to
broken BIOS. Use 'tsc=unstable'.
[62939.863972] sched_clock: Marking unstable (62943398530130,
-3543150114)<-(62941186607503, -1331276112)
[62939.875104] clocksource: Checking clocksource tsc synchronization
from CPU 123 to CPUs 0,6,26,62,78,97-98,137.
[62939.886518] clocksource: Switched to clocksource hpet

On Tue, Apr 27, 2021 at 2:27 AM Paul E. McKenney <paulmck@kernel.org> wrote:
>
> On Mon, Apr 26, 2021 at 10:56:27AM -0700, Andi Kleen wrote:
> > > ------------------------------------------------------------------------
> > >
> > > - module parameters
> > >
> > >   If the scope of the fault injection capability is limited to a
> > >   single kernel module, it is better to provide module parameters to
> > >   configure the fault attributes.
> > >
> > > ------------------------------------------------------------------------
> > >
> > > And in this case, the fault injection capability is in fact limited to
> > > kernel/clocksource.c.
> >
> >
> > I disagree with this recommendation because it prevents fuzzer coverage.
> >
> > Much better to have an uniform interface that can be automatically
> > explored.
>
> The permissions for these module parameters is 0644, so there is no
> reason why the fuzzers cannot use them via sysfs.
>
>                                                         Thanx, Paul