Received-SPF: pass (google.com: domain of linux-kernel+bounces-167068-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <b67e79d4-06cb-4a45-a906-b9e0fbae22c5@paulmck-laptop>
 <20240501230130.1111603-12-paulmck@kernel.org> <1376850f47279e3a3f4f40e3de2784ae3ac30414.camel@physik.fu-berlin.de>
 <b7ae0feb-d401-43ee-8d5f-ce62ca224638@paulmck-laptop> <6f7743601fe7bd50c2855a8fd1ed8f766ef03cac.camel@physik.fu-berlin.de>
 <9a4e1928-961d-43af-9951-71786b97062a@paulmck-laptop> <20240502205345.GK2118490@ZenIV>
 <0a429959-935d-4800-8d0c-4e010951996d@paulmck-laptop> <20240502220757.GL2118490@ZenIV>
 <3dac400c-d18f-4f4e-b598-cad6948362d6@paulmck-laptop>
In-Reply-To: <3dac400c-d18f-4f4e-b598-cad6948362d6@paulmck-laptop>
From: Linus Torvalds <torvalds@linux-foundation.org>
Date: Thu, 2 May 2024 16:32:35 -0700
Message-ID: <CAHk-=whaCSxengJHP82WUwrjKjYsVeD_zEN_We+gmyHpJJayoQ@mail.gmail.com>
Subject: Re: [PATCH v2 cmpxchg 12/13] sh: Emulate one-byte cmpxchg
To: paulmck@kernel.org
Cc: Al Viro <viro@zeniv.linux.org.uk>, 
	John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>, linux-arch@vger.kernel.org, 
	linux-kernel@vger.kernel.org, elver@google.com, akpm@linux-foundation.org, 
	tglx@linutronix.de, peterz@infradead.org, dianders@chromium.org, 
	pmladek@suse.com, arnd@arndb.de, kernel-team@meta.com, 
	Andi Shyti <andi.shyti@linux.intel.com>, Palmer Dabbelt <palmer@rivosinc.com>, 
	Masami Hiramatsu <mhiramat@kernel.org>, linux-sh@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"

On Thu, 2 May 2024 at 16:12, Paul E. McKenney <paulmck@kernel.org> wrote:
>
> One of RCU's state machines uses smp_store_release() to start the
> state machine (only one task gets to do this) and cmpxchg() to update
> state beyond that point.  And the state is 8 bits so that it and other
> state fits into 32 bits to allow a single check for multiple conditions
> elsewhere.

Note that since alpha lacks the release-acquire model, it's always
going to be a full memory barrier before the store.

And then the store turns into a load-mask-store for older alphas.

So it's going to be a complete mess from a performance standpoint regardless.

Happily, I doubt anybody really cares.

I've occasionally wondered if we have situations where the
"smp_store_release()" only cares about previous *writes* being ordered
(ie a "smp_wmb()+WRITE_ONCE" would be sufficient).

It makes no difference on x86 (all stores are relases), power64 (wmb
and store_release are both LWSYNC) or arm64 (str is documentated to be
cheaper than DMB).

On alpha, smp_wmb()+WRITE_ONCE() is cheaper than smp_store_release(),
but nobody sane cares.

But *if* we have a situation where the "smp_store_release()" might be
just a "previous writes need to be visible" rather than ordering
previous reads too, we could maybe introduce that kind of op. I
_think_ the RCU writes tend to be of that kind?

                    Linus