linux-kernel - Re: [RFC PATCH v2 11/14] x86/watchdog/hardlockup: Add an HPET-based hardlockup detector

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 9 Apr 2019 12:59:46 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     Ricardo Neri <ricardo.neri-calderon@...ux.intel.com>,
        Ingo Molnar <mingo@...nel.org>, Borislav Petkov <bp@...e.de>,
        Ashok Raj <ashok.raj@...el.com>,
        Andi Kleen <andi.kleen@...el.com>,
        "Ravi V. Shankar" <ravi.v.shankar@...el.com>, x86@...nel.org,
        linux-kernel@...r.kernel.org,
        Ricardo Neri <ricardo.neri@...el.com>,
        "H. Peter Anvin" <hpa@...or.com>, Tony Luck <tony.luck@...el.com>,
        Clemens Ladisch <clemens@...isch.de>,
        Arnd Bergmann <arnd@...db.de>,
        Philippe Ombredanne <pombredanne@...b.com>,
        Kate Stewart <kstewart@...uxfoundation.org>,
        "Rafael J. Wysocki" <rafael.j.wysocki@...el.com>,
        Mimi Zohar <zohar@...ux.ibm.com>,
        Jan Kiszka <jan.kiszka@...mens.com>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Nayna Jain <nayna@...ux.ibm.com>
Subject: Re: [RFC PATCH v2 11/14] x86/watchdog/hardlockup: Add an HPET-based
 hardlockup detector

On Tue, Mar 26, 2019 at 09:49:13PM +0100, Thomas Gleixner wrote:
> So way you should handle this is:
> 
> 	cpumask_set_cpu(cpu, hld_data->cpu_monitored_mask);
> 
> 	if (!hld_data->enabled_cpus++) {
> 		hld_data->handling_cpu = cpu;
> 		kick_timer();
> 		enable_timer();
> 	}
> 
> The cpu mask starts off empty and each CPU sets itself when the function is
> invoked on it.
> 
> data->enabled_cpus keeps track of the enabled cpus so you avoid
> reconfiguration just because a different cpu comes online. And it's
> required for disable as well.
> 
> > +void hardlockup_detector_hpet_disable(void)
> > +{
> > +	struct cpumask *allowed = watchdog_get_allowed_cpumask();
> > +
> > +	if (!hld_data)
> > +		return;
> > +
> > +	/* Only disable the timer if there are no more CPUs to monitor. */
> > +	if (!cpumask_weight(allowed))
> > +		disable_timer(hld_data);
> 
> Again this should be:
> 
> 	cpumask_clear_cpu(cpu, hld_data->cpu_monitored_mask);
> 	hld_data->enabled_cpus--;
> 
> 	if (hld_data->handling_cpu != cpu)
> 		return;
> 
> 	disable_timer();
> 	if (hld_data->enabled_cpus)
> 		return;

	if (!hld_data->enabled_cpus)
		return;

> 
> 	hld_data->handling_cpu = cpumask_first(hld_data->cpu_monitored_mask);
> 	enable_timer();

That said; you can do the above without ->enabled_cpus, by using
->handling_cpu == nr_cpu_ids to indicate 'empty'. But I'm not at all
sure that is worth the effort, it results in less obious code.