linux-kernel - Re: [PATCH] cpufreq: intel_pstate: Fix for HWP interrupt before driver is ready

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:   Mon, 06 Sep 2021 12:56:53 -0700
From:   Srinivas Pandruvada <srinivas.pandruvada@...ux.intel.com>
To:     "Rafael J. Wysocki" <rafael@...nel.org>
Cc:     Jens Axboe <axboe@...nel.dk>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Len Brown <lenb@...nel.org>,
        Viresh Kumar <viresh.kumar@...aro.org>,
        Linux PM <linux-pm@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] cpufreq: intel_pstate: Fix for HWP interrupt before
 driver is ready

On Mon, 2021-09-06 at 20:25 +0200, Rafael J. Wysocki wrote:
> On Mon, Sep 6, 2021 at 8:14 PM Srinivas Pandruvada
> <srinivas.pandruvada@...ux.intel.com> wrote:
> > 
> > On Mon, 2021-09-06 at 19:54 +0200, Rafael J. Wysocki wrote:
> > > 
[...]

> > > 
> > We are handling offline for other thermal interrupt sources from
> > same
> > interrupt in therm-throt.c, where we do similar in offline path (by
> > TGLX). If cpufreq offline can cause such issue of changing CPU,
> 
> This is not cpufreq offline, but intel_pstate_update_status() which
> may be triggered via sysfs.  And again, the theoretically problematic
> thing is dereferencing cpudata (which may be cleared by a remote CPU)
> from the interrupt handler without protection.

This will be a problem.

> 
> > I can call intel_pstate_disable_hwp_interrupt() via override from
> > https://elixir.bootlin.com/linux/latest/C/ident/thermal_throttle_offline
> > after masking APIC interrupt.
> 
> But why would using RCU be harder than this?
I think, this will require all_cpu_data and cpu_data to be rcu
protected. This needs to be well tested.

I think better to revert the patch for the next release.

Thanks,
Srinivas

> 
> Also please note that on RT kernels interrupt handlers are run in
> threads.