linux-kernel - Re: [PATCH v4] locking/mutex: Prevent lock starvation when spinning is disabled

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]

Message-ID: <20160823124519.GB10138@twins.programming.kicks-ass.net>
Date:   Tue, 23 Aug 2016 14:45:19 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Waiman Long <waiman.long@....com>, Jason Low <jason.low2@....com>,
        Ding Tianhong <dingtianhong@...wei.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Will Deacon <Will.Deacon@....com>,
        Ingo Molnar <mingo@...hat.com>,
        Imre Deak <imre.deak@...el.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Davidlohr Bueso <dave@...olabs.net>,
        Tim Chen <tim.c.chen@...ux.intel.com>,
        Terry Rudd <terry.rudd@....com>,
        "Paul E. McKenney" <paulmck@...ibm.com>,
        Jason Low <jason.low2@...com>
Subject: Re: [PATCH v4] locking/mutex: Prevent lock starvation when spinning
 is disabled

On Fri, Aug 19, 2016 at 12:45:27PM -0700, Linus Torvalds wrote:
> On Fri, Aug 19, 2016 at 12:33 PM, Peter Zijlstra <peterz@...radead.org> wrote:
> >
> > That seems more messy to me..
> 
> Can't we just say that we always spin? Are there any really valid
> cases where spinning isn't ok?
> 
> We've historically disabled spinning when mutex debugging is enabled,
> but since the spinning is limited anyway, couldn't we just spin even
> with debugging enabled?
> 
> I hate how these patches are trying to solve a problem that doesn't
> even happen under normal circumstances, and add special-case code for
> something that is already a special-case condition. So rather than
> adding even more special cases, could we look at _removing_ the
> special cases that cause problems instead?

So I think I have a bunch of patches that solves the fundamental issues,
_However_... they're quite invasive and would need some serious
benchmarking.

I'll post them in a separate thread as RFC..