linux-kernel - Re: [PATCH] sched/completion: completion_done() should serialize with complete()

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, 16 Feb 2015 17:51:56 +0100
From:	Oleg Nesterov <oleg@...hat.com>
To:	Peter Zijlstra <peterz@...radead.org>
Cc:	"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
	linux-kernel@...r.kernel.org, dave@...olabs.net,
	waiman.long@...com, raghavendra.kt@...ux.vnet.ibm.com,
	Nicholas Mc Guire <der.herr@...r.at>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	mingo@...nel.org
Subject: Re: [PATCH] sched/completion: completion_done() should serialize
	with complete()

On 02/16, Peter Zijlstra wrote:
>
> On Thu, Feb 12, 2015 at 08:59:13PM +0100, Oleg Nesterov wrote:
> > Commit de30ec47302c "Remove unnecessary ->wait.lock serialization when
> > reading completion state" was not correct, without lock/unlock the code
> > like stop_machine_from_inactive_cpu()
> >
> > 	while (!completion_done())
> > 		cpu_relax();
> >
> > can return before complete() finishes its spin_unlock() which writes to
> > this memory. And spin_unlock_wait().
> >
> > While at it, change try_wait_for_completion() to use READ_ONCE().
>
> So I share Davidlohrs concern

Ah. I forgot to reply to Davidlohr's email. Sorry.

> if we should not simply revert that
> change; but given we've now gone over it detail I suppose we should just
> keep the optimized version.

Yes, I was going to say that of course I won't argue if we simply revert
that commit. As he rigthly pointed the lockless check doesn't make sense
performance-wise.

However, this code needs a comment to explain why we can't simply check
->done and return, unlock_wait() is more documentation than optimization.

But,

> I did add a comment to your patch; and queued the below for
> sched/urgent.

Thanks!

Now this logic is actually documented ;) unlock_wait() alone could confuse
the reader too.

Oleg.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/