linux-kernel - Re: [PATCH 1/2] PM / wakeup: Remove timer from wakeup_source

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] [day] [month] [year] [list]

Date:   Tue, 12 Mar 2019 17:11:47 +0530
From:   Viresh Kumar <viresh.kumar@...aro.org>
To:     Pavel Machek <pavel@....cz>
Cc:     "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Len Brown <len.brown@...el.com>, linux-pm@...r.kernel.org,
        Vincent Guittot <vincent.guittot@...aro.org>,
        sunzhaosheng@...ilicon.com, jean.xupeng@...ilicon.com,
        yuwei3@...ilicon.com, gengyanping@...ilicon.com,
        peter.panshilin@...ilicon.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/2] PM / wakeup: Remove timer from wakeup_source_remove()

On 12-03-19, 10:04, Pavel Machek wrote:
> On Tue 2019-03-12 08:58:02, Viresh Kumar wrote:
> > On 11-03-19, 13:05, Rafael J. Wysocki wrote:
> > > On Friday, March 8, 2019 10:53:11 AM CET Viresh Kumar wrote:
> > > > wakeup_source_remove() is the counterpart of wakeup_source_add() helper
> > > > and must undo the initializations done by wakeup_source_add(). Currently
> > > > the timer is initialized by wakeup_source_add() but removed from
> > > > wakeup_source_drop(), which doesn't look logically correct. Also it
> > > > should be okay to call wakeup_source_add() right after calling
> > > > wakeup_source_remove(), and in that case we may end up calling
> > > > timer_setup() for a potentially scheduled timer which is surely
> > > > incorrect.
> > > > 
> > > > Move the timer removal part to wakeup_source_remove() instead.
> 
> > > 
> > > I've merged it with the [2/2], rewritten the subject and changelog and
> > > queued the result as commit d856f39ac1cc ("PM / wakeup: Rework wakeup
> > > source timer cancellation").
> > 
> > Okay, thanks. We (Android guys) want this to be backported into 4.4+
> > kernels via the stable tree. Can we mark this for stable in the commit
> > itself ? Else I would be required to send this separately for all the
> > kernels. I should have marked it for stable initially though, sorry
> > about forgetting then.
> 
> Greg is normally pretty agressive about backporting everything
> remotely looking like a fix...
> 
> But better changelog would help. How is the bug actually affecting
> users?

The Android wakelock infrastructure (not in mainline) is based on the
wakeup sources stuff and that is where we found the issue. A wakelock
was taken due to a bug in user driver and the board never attempts
going into suspend anymore.

If this patch doesn't go into the stable kernels, I would be required
to send this separately for Android kernels and the Android guys will
suggest getting it via stable tree.

And because this fixes a potential issue it was worth trying getting
it via the stable kernel :)

-- 
viresh