| 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
| ||
|
Message-ID: <OSBPR01MB2103B1872251C12C2D1CE75EB86D0@OSBPR01MB2103.jpnprd01.prod.outlook.com>
Date: Mon, 4 Feb 2019 09:29:01 +0000
From: Biju Das <biju.das@...renesas.com>
To: Vincent Guittot <vincent.guittot@...aro.org>
CC: "Rafael J. Wysocki" <rafael@...nel.org>,
Linux PM <linux-pm@...r.kernel.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Linux ARM <linux-arm-kernel@...ts.infradead.org>,
Linux OMAP Mailing List <linux-omap@...r.kernel.org>,
"Rafael J. Wysocki" <rjw@...ysocki.net>,
Ulf Hansson <ulf.hansson@...aro.org>,
Geert Uytterhoeven <geert@...ux-m68k.org>,
Linux-Renesas <linux-renesas-soc@...r.kernel.org>
Subject: RE: [PATCH v3] PM-runtime: fix deadlock with ktime
Hi Vincent,
Thanks for the feedback.
> Subject: Re: [PATCH v3] PM-runtime: fix deadlock with ktime
>
> On Fri, 1 Feb 2019 at 16:48, Vincent Guittot <vincent.guittot@...aro.org>
> wrote:
> >
> > On Fri, 1 Feb 2019 at 16:44, Biju Das <biju.das@...renesas.com> wrote:
> > >
> > > Hi Vincent,
> > >
> > > Thanks for the feedback. Instead of reverting. I just modified the patch
> like this and it fixed the issue.
> > >
> > > diff --git a/drivers/base/power/runtime.c
> > > b/drivers/base/power/runtime.c index 4eaf166..145a182 100644
> > > --- a/drivers/base/power/runtime.c
> > > +++ b/drivers/base/power/runtime.c
> > > @@ -66,7 +66,7 @@ static int rpm_suspend(struct device *dev, int
> rpmflags);
> > > */
> > > void update_pm_runtime_accounting(struct device *dev) {
> > > - u64 now = ktime_to_ns(ktime_get());
> > > + u64 now = ktime_get_mono_fast_ns();
> > > u64 delta;
> >
> > yes.
> > This probably also need to be changed in another place
> > (pm_runtime_enable) to be safe
> > Our email have crossed. I just sent something similar
>
> In fact after more thinking, it's probably better to revert because we are not
> sure to be monotonic with ktime_get_mono_fast_ns() which means that the
> time account can be negative but this not yet taken into account. It's part of a
> clean up patch that was on hold while working on previous problem
I confirm there is no deadlock, after reverting the commit c669560be6c8 ("PM-runtime: Replace
jiffies-based accounting with ktime-based accounting")
Regards,
Biju
> > >
> > > > -----Original Message-----
> > > > From: linux-renesas-soc-owner@...r.kernel.org <linux-renesas-soc-
> > > > owner@...r.kernel.org> On Behalf Of Vincent Guittot
> > > > Sent: 01 February 2019 15:29
> > > > To: Biju Das <biju.das@...renesas.com>
> > > > Cc: Rafael J. Wysocki <rafael@...nel.org>; Linux PM <linux-
> > > > pm@...r.kernel.org>; Linux Kernel Mailing List <linux-
> > > > kernel@...r.kernel.org>; Linux ARM <linux-arm-
> > > > kernel@...ts.infradead.org>; Linux OMAP Mailing List <linux-
> > > > omap@...r.kernel.org>; Rafael J. Wysocki <rjw@...ysocki.net>; Ulf
> > > > Hansson <ulf.hansson@...aro.org>; Geert Uytterhoeven <geert@...ux-
> > > > m68k.org>; Linux-Renesas <linux-renesas-soc@...r.kernel.org>
> > > > Subject: Re: [PATCH v3] PM-runtime: fix deadlock with ktime
> > > >
> > > > On Fri, 1 Feb 2019 at 16:02, Biju Das <biju.das@...renesas.com> wrote:
> > > > >
> > > > > Hi Vincent,
> > > > >
> > > > > I have rebased my kernel to "next-20190201". Still I am seeing dead
> lock.
> > > > >
> > > > > Am I missing any patch?
> > > >
> > > > No you don't miss anything.
> > > > I think that it's the opposite.
> > > >
> > > > Modification in time accounting in PM runtime has been queued but
> > > > it has not moved (yet) to ktime_get_mono_fast_ns()
> > > >
> > > > Can you try to revert c669560be6c8 ("PM-runtime: Replace
> > > > jiffies-based accounting with ktime-based accounting") ?
> > > >
> > > > >
> > > > > root@...74:/# echo e61e0000.timer >
> > > > > /sys/devices/system/clocksource/clocksource0/current_clocksource
> > > > > [ 193.869423]
> > > > > [ 193.870963]
> ============================================
> > > > > [ 193.876292] WARNING: possible recursive locking detected [
> > > > > 193.881625] 5.0.0-rc4-next-20190201-00007-g731346f #3 Not
> > > > > tainted [ 193.887737]
> > > > > --------------------------------------------
> > > > > [ 193.893066] migration/0/11 is trying to acquire lock:
> > > > > [ 193.898136] (____ptrval____) (tk_core.seq){----}, at:
> > > > > update_pm_runtime_accounting+0x14/0x68
> > > > > [ 193.906632]
> > > > > [ 193.906632] but task is already holding lock:
> > > > > [ 193.912483] (____ptrval____) (tk_core.seq){----}, at:
> > > > > multi_cpu_stop+0x8c/0x140 [ 193.919828] [ 193.919828] other
> > > > > info that might help us debug this:
> > > > > [ 193.926377] Possible unsafe locking scenario:
> > > > > [ 193.926377]
> > > > > [ 193.932314] CPU0
> > > > > [ 193.934765] ----
> > > > > [ 193.937216] lock(tk_core.seq);
> > > > > [ 193.940453] lock(tk_core.seq);
> > > > > [ 193.943691]
> > > > > [ 193.943691] *** DEADLOCK *** [ 193.943691] [ 193.949634]
> > > > > May be due to missing lock nesting notation [ 193.949634] [
> > > > > 193.956446] 3 locks held by migration/0/11:
> > > > > [ 193.960642] #0: (____ptrval____) (timekeeper_lock){-.-.}, at:
> > > > > change_clocksource+0x2c/0x118 [ 193.969125] #1:
> > > > > (____ptrval____) (tk_core.seq){----}, at: multi_cpu_stop+0x8c/0x140
> [ 193.976903] #2:
> > > > > (____ptrval____) (&(&dev->power.lock)->rlock){....}, at:
> > > > > __pm_runtime_resume+0x40/0x98 [ 193.986339] [ 193.986339]
> > > > > stack
> > > > backtrace:
> > > > > [ 193.990715] CPU: 0 PID: 11 Comm: migration/0 Not tainted
> > > > > 5.0.0-rc4-next-20190201-00007-g731346f #3 [ 193.999707]
> > > > > Hardware
> > > > > name: Silicon Linux RZ/G2E evaluation kit EK874 (CAT874 +
> > > > > CAT875) (DT) [
> > > > 194.008089] Call trace:
> > > > > [ 194.010553] dump_backtrace+0x0/0x178 [ 194.014227]
> > > > > show_stack+0x14/0x20 [ 194.017562] dump_stack+0xb0/0xec [
> > > > > 194.020895] __lock_acquire+0xfb4/0x1c08 [ 194.024832]
> > > > > lock_acquire+0xd0/0x268 [ 194.028420] ktime_get+0x5c/0x108 [
> > > > > 194.031747] update_pm_runtime_accounting+0x14/0x68
> > > > > [ 194.036643] rpm_resume+0x4ec/0x698 [ 194.040144]
> > > > > __pm_runtime_resume+0x50/0x98 [ 194.044264]
> > > > > sh_tmu_enable.part.1+0x24/0x50 [ 194.048462]
> > > > > sh_tmu_clocksource_enable+0x48/0x70
> > > > > [ 194.053097] change_clocksource+0x84/0x118 [ 194.057208]
> > > > > multi_cpu_stop+0x8c/0x140 [ 194.060970]
> > > > > cpu_stopper_thread+0xac/0x120 [ 194.065087]
> > > > > smpboot_thread_fn+0x1ac/0x2c8 [ 194.069198]
> > > > > kthread+0x128/0x130 [ 194.072439] ret_from_fork+0x10/0x18
> > > > >
> > > > >
> > > > > Regards,
> > > > > Biju
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Rafael J. Wysocki <rafael@...nel.org>
> > > > > > Sent: 30 January 2019 21:53
> > > > > > To: Vincent Guittot <vincent.guittot@...aro.org>
> > > > > > Cc: Linux PM <linux-pm@...r.kernel.org>; Linux Kernel Mailing
> > > > > > List
> > > > > > <linux- kernel@...r.kernel.org>; Linux ARM <linux-arm-
> > > > > > kernel@...ts.infradead.org>; Linux OMAP Mailing List <linux-
> > > > > > omap@...r.kernel.org>; Rafael J. Wysocki <rjw@...ysocki.net>;
> > > > > > Ulf Hansson <ulf.hansson@...aro.org>; Biju Das
> > > > > > <biju.das@...renesas.com>; Geert Uytterhoeven
> > > > > > <geert@...ux-m68k.org>; Linux-Renesas <linux-
> > > > > > renesas-soc@...r.kernel.org>
> > > > > > Subject: Re: [PATCH v3] PM-runtime: fix deadlock with ktime
> > > > > >
> > > > > > On Wed, Jan 30, 2019 at 6:26 PM Vincent Guittot
> > > > > > <vincent.guittot@...aro.org> wrote:
> > > > > > >
> > > > > > > A deadlock has been seen when swicthing clocksources which
> > > > > > > use PM
> > > > > > runtime.
> > > > > > > The call path is:
> > > > > > > change_clocksource
> > > > > > > ...
> > > > > > > write_seqcount_begin
> > > > > > > ...
> > > > > > > timekeeping_update
> > > > > > > ...
> > > > > > > sh_cmt_clocksource_enable
> > > > > > > ...
> > > > > > > rpm_resume
> > > > > > > pm_runtime_mark_last_busy
> > > > > > > ktime_get
> > > > > > > do
> > > > > > > read_seqcount_begin
> > > > > > > while read_seqcount_retry
> > > > > > > ....
> > > > > > > write_seqcount_end
> > > > > > >
> > > > > > > Although we should be safe because we haven't yet changed
> > > > > > > the clocksource at that time, we can't because of seqcount
> protection.
> > > > > > >
> > > > > > > Use ktime_get_mono_fast_ns() instead which is lock safe for
> > > > > > > such case
> > > > > > >
> > > > > > > With ktime_get_mono_fast_ns, the timestamp is not guaranteed
> > > > > > > to be monotonic across an update and as a result can goes
> backward.
> > > > > > > According to
> > > > > > > update_fast_timekeeper() description: "In the worst case,
> > > > > > > this can result is a slightly wrong timestamp (a few
> > > > > > > nanoseconds)". For PM runtime autosuspend, this means only
> > > > > > > that the suspend decision can be slightly sub optimal.
> > > > > > >
> > > > > > > Fixes: 8234f6734c5d ("PM-runtime: Switch autosuspend over to
> > > > > > > using
> > > > > > > hrtimers")
> > > > > > > Reported-by: Biju Das <biju.das@...renesas.com>
> > > > > > > Signed-off-by: Vincent Guittot <vincent.guittot@...aro.org>
> > > > > > > ---
> > > > > > >
> > > > > > > Hi Rafael,
> > > > > > >
> > > > > > > Sorry, I sent the version with the typo mistake that
> > > > > > > generated the compilation error reported by
> > > > > > > kbuild-test-robot
> > > > > > >
> > > > > > > This version doesn't have the typo.
> > > > > >
> > > > > > OK, I've applied this one, thanks!
> > > > >
> > > > >
> > > > >
> > > > > Renesas Electronics Europe Ltd, Dukes Meadow, Millboard Road,
> > > > > Bourne
> > > > End, Buckinghamshire, SL8 5FH, UK. Registered in England & Wales
> > > > under Registered No. 04586709.
> > >
> > >
> > >
> > > Renesas Electronics Europe Ltd, Dukes Meadow, Millboard Road, Bourne
> End, Buckinghamshire, SL8 5FH, UK. Registered in England & Wales under
> Registered No. 04586709.
Renesas Electronics Europe Ltd, Dukes Meadow, Millboard Road, Bourne End, Buckinghamshire, SL8 5FH, UK. Registered in England & Wales under Registered No. 04586709.
Powered by blists - more mailing lists