Re: [PATCH v2] tools/memory-model: Add extra ordering for locks and remove it for ordinary release/acquire

From: Alan Stern <stern@rowland.harvard.edu>
To: Peter Zijlstra <peterz@infradead.org>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Michael Ellerman <mpe@ellerman.id.au>,
	<andrea.parri@amarulasolutions.com>,
	Will Deacon <will.deacon@arm.com>,
	Akira Yokosawa <akiyks@gmail.com>,
	Boqun Feng <boqun.feng@gmail.com>,
	Daniel Lustig <dlustig@nvidia.com>,
	David Howells <dhowells@redhat.com>,
	Jade Alglave <j.alglave@ucl.ac.uk>,
	Luc Maranget <luc.maranget@inria.fr>,
	Nick Piggin <npiggin@gmail.com>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2] tools/memory-model: Add extra ordering for locks and remove it for ordinary release/acquire
Date: Tue, 17 Jul 2018 15:47:12 -0400 (EDT)	[thread overview]
Message-ID: <Pine.LNX.4.44L0.1807171542010.1344-100000@iolanthe.rowland.org> (raw)
In-Reply-To: <20180717184255.GM2494@hirez.programming.kicks-ass.net>

On Tue, 17 Jul 2018, Peter Zijlstra wrote:

> Isn't ISYNC the instruction-sync pipeline flush instruction? That is
> used as an smp_rmb() here to, together with the control dependency from the
> LL/SC, to form a LOAD->{LOAD,STORE} (aka LOAD-ACQUIRE) ordering?

That's right.

> Where LWSYNC provides a TSO like ordering and SYNC provides a full
> transitive barrier aka. smp_mb() (althgouh I think it is strictly
> stronger than smp_mb() since it also implies completion, which smp_mb()
> does not).

What is the difference?  That is, can you give a litmus test for
"completion" in this sense?

> And since both LL/SC-CTRL + ISYNC / LWSYNC are strictly CPU local, they
> cannot be used to create RCsc ordering.

Yes, sort of -- it depends on whether you consider the store buffer to 
be part of the CPU.  But in any case, you are right that none of these 
things can create RCsc ordering.

Alan