Re: [PATCH 4/4] locking: Introduce smp_cond_acquire()

From: Oleg Nesterov <oleg@redhat.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Boqun Feng <boqun.feng@gmail.com>,
	mingo@kernel.org, linux-kernel@vger.kernel.org,
	paulmck@linux.vnet.ibm.com, corbet@lwn.net, mhocko@kernel.org,
	dhowells@redhat.com, torvalds@linux-foundation.org,
	will.deacon@arm.com, Michael Ellerman <mpe@ellerman.id.au>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Paul Mackerras <paulus@samba.org>
Subject: Re: [PATCH 4/4] locking: Introduce smp_cond_acquire()
Date: Thu, 12 Nov 2015 20:33:02 +0100	[thread overview]
Message-ID: <20151112193302.GA9988@redhat.com> (raw)
In-Reply-To: <20151112180203.GF17308@twins.programming.kicks-ass.net>

On 11/12, Peter Zijlstra wrote:
>
> On Thu, Nov 12, 2015 at 07:38:07PM +0100, Oleg Nesterov wrote:
> > It seems that PPC needs to define smp_mb__before_spinlock() as full mb(),
> > no?
>
> It does:
>
> arch/powerpc/include/asm/barrier.h:#define smp_mb__before_spinlock() smp_mb()

Ah, indeed, thanks.

And given that it also defines smp_mb__after_unlock_lock() as smp_mb(),
I am starting to understand how it can help to avoid the races with
spin_unlock_wait() in (for example) do_exit().

But as Boqun has already mentioned, this means that mb__after_unlock_lock()
has the new meaning which should be documented.

Hmm. And 12d560f4 "Privatize smp_mb__after_unlock_lock()" should be reverted
then ;)

Oleg.