Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 1 Feb 2018 16:52:42 +0100
From:   Peter Zijlstra <peterz@infradead.org>
To:     Will Deacon <will.deacon@arm.com>
Cc:     Stafford Horne <shorne@gmail.com>,
        Paul McKenney <paulmck@linux.vnet.ibm.com>,
        Jonas Bonn <jonas@southpole.se>,
        Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>,
        David Howells <dhowells@redhat.com>,
        Arnd Bergmann <arnd@arndb.de>, linux-kernel@vger.kernel.org,
        Thomas Gleixner <tglx@linutronix.de>
Subject: Re: asm-generic: Disallow no-op mb() for SMP systems
Message-ID: <20180201155242.GL2295@hirez.programming.kicks-ass.net>
References: <20180131130034.GR2269@hirez.programming.kicks-ass.net>
 <20180131131737.GA5097@arm.com>
 <20180131132610.GT2269@hirez.programming.kicks-ass.net>
 <20180201122750.GE30895@lianli.shorne-pla.net>
 <20180201132909.GW2249@hirez.programming.kicks-ass.net>
 <20180201133229.GB9182@arm.com>
 <20180201135329.GB2269@hirez.programming.kicks-ass.net>
 <20180201153951.GG9182@arm.com>
 <20180201155007.GB2249@hirez.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180201155007.GB2249@hirez.programming.kicks-ass.net>
User-Agent: Mutt/1.9.2 (2017-12-15)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Feb 01, 2018 at 04:50:07PM +0100, Peter Zijlstra wrote:
> On Thu, Feb 01, 2018 at 03:39:51PM +0000, Will Deacon wrote:
> > > I could've gotten my brain in a twist or course, which isn't _that_
> > > unusual. I never seem to be able to quite remember the holes you have
> > > with ll/sc on arm64 :-)
> > 
> > Is smp_mb__before_atomic supposed to provide ordering guarantees if it's
> > used before a failed cmpxchg? If so, I think it's needed here because the
> > l.swa might not even execute. Or did I just invent another problem?
> 
> I think it should do so indeed (and afaik all our current archs are good
> that way).

See commit:

  34d54f3d6917 ("locking/pvqspinlock: Relax cmpxchg's to improve performance on some architectures")