linux-kernel - Re: [PATCH] rtmutex: ensure only the top waiter or higher priority task can take the lock and reduce unrelated boosting

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:	Wed, 15 Dec 2010 15:52:08 +0100 (CET)
From:	Thomas Gleixner <tglx@...utronix.de>
To:	Lai Jiangshan <eag0628@...il.com>
cc:	Lai Jiangshan <laijs@...fujitsu.com>,
	Steven Rostedt <rostedt@...dmis.org>,
	Ingo Molnar <mingo@...e.hu>,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Dave Young <hidave.darkstar@...il.com>,
	Darren Hart <dvhart@...ux.intel.com>,
	Namhyung Kim <namhyung@...il.com>,
	LKML <linux-kernel@...r.kernel.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH] rtmutex: ensure only the top waiter or higher priority
 task can take the lock and reduce unrelated boosting

On Wed, 15 Dec 2010, Lai Jiangshan wrote:
> On Wed, Dec 15, 2010 at 8:07 PM, Thomas Gleixner <tglx@...utronix.de> wrote:
> > On Wed, 15 Dec 2010, Lai Jiangshan wrote:
> >> +     if (waiter || rt_mutex_has_waiters(lock)) {
> >> +             unsigned long flags;
> >> +             struct rt_mutex_waiter *top;
> >> +
> >> +             raw_spin_lock_irqsave(&task->pi_lock, flags);
> >> +
> >> +             /* remove the queued waiter. */
> >> +             if (waiter) {
> >> +                     plist_del(&waiter->list_entry, &lock->wait_list);
> >> +                     task->pi_blocked_on = NULL;
> >> +             }
> >> +
> >> +             /*
> >> +              * We have to enqueue the top waiter(if have) into
> >> +              * task->pi_waiters list and would get boost from it.
> >
> > No, we don't get boosted from it. We just have to enqueue it into
> > pi_waiters list. There is no boosting happening at this point. Please
> > be very careful with the comments in this code.
> >
> > Otherwise this looks really interesting. Still this wants to be ported
> > to -rt and stress tested there.
> >
> 
> Hi, Thomas,
> 
> Where can I find the code of stress testcases?

RT itself will stress test the rtmutex code as much as it goes as we
convert almost every lock into a rtmutex there.

Steven, could you have a look at that ?

Thanks,

	tglx