From: David Daney <ddaney@caviumnetworks.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: <ralf@linux-mips.org>, <ddaney@caviumnetworks.com>,
<linux-kernel@vger.kernel.org>,
Paul McKenney <paulmck@linux.vnet.ibm.com>,
Will Deacon <will.deacon@arm.com>,
<torvalds@linux-foundation.org>, <boqun.feng@gmail.com>
Subject: Re: [RFC][PATCH] mips: Fix arch_spin_unlock()
Date: Thu, 12 Nov 2015 09:46:53 -0800 [thread overview]
Message-ID: <5644D08D.4080206@caviumnetworks.com> (raw)
In-Reply-To: <20151112123123.GZ17308@twins.programming.kicks-ass.net>
On 11/12/2015 04:31 AM, Peter Zijlstra wrote:
> Hi
>
> I think the MIPS arch_spin_unlock() is borken.
>
> spin_unlock() must have RELEASE semantics, these require that no LOADs
> nor STOREs leak out from the critical section.
>
> From what I know MIPS has a relaxed memory model which allows reads to
> pass stores, and as implemented arch_spin_unlock() only issues a wmb
> which doesn't order prior reads vs later stores.
>
> Therefore upgrade the wmb() to smp_mb().
>
> (Also, why the unconditional wmb, as opposed to smp_wmb() ?)
asm/spinlock.h is only used for !CONFIG_SMP. So, smp_wmb() would imply
that special handling for non-SMP is needed, when this is already only
used for the SMP build case.
>
> Maybe-Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> ---
> diff --git a/arch/mips/include/asm/spinlock.h b/arch/mips/include/asm/spinlock.h
> index 40196bebe849..b2ca13f06152 100644
> --- a/arch/mips/include/asm/spinlock.h
> +++ b/arch/mips/include/asm/spinlock.h
> @@ -140,7 +140,7 @@ static inline void arch_spin_lock(arch_spinlock_t *lock)
> static inline void arch_spin_unlock(arch_spinlock_t *lock)
> {
> unsigned int serving_now = lock->h.serving_now + 1;
> - wmb();
> + smp_mb();
That is too heavy.
It implies a full MIPS "SYNC" operation which stalls execution until all
previous writes are committed and globally visible.
We really want just release semantics, and there is no standard named
primitive that gives us that.
For CONFIG_CPU_CAVIUM_OCTEON the proper thing would be:
smp_wmb();
smp_rmb();
Which expands to exactly the same thing as wmb() because smp_rmb()
expands to nothing.
For CPUs that have out-of-order loads, smp_rmb() should expand to
something lighter weight than "SYNC"
Certainly we can load up the code with "SYNC" all over the place, but it
will kill performance on SMP systems. So, my vote would be to make it
as light weight as possible, but no lighter. That will mean inventing
the proper barrier primitives.
You yourself seem to have added smp_store_release(), so we could even do:
smp_store_release(&lock->h.serving_now, lock->h.serving_now + 1);
That would leave us to cook up a proper definition of smp_store_release().
David Daney
> lock->h.serving_now = (u16)serving_now;
> nudge_writes();
> }
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
next prev parent reply other threads:[~2015-11-12 17:47 UTC|newest]
Thread overview: 54+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-11-12 12:31 [RFC][PATCH] mips: Fix arch_spin_unlock() Peter Zijlstra
2015-11-12 12:35 ` Peter Zijlstra
2015-11-12 13:31 ` Måns Rullgård
2015-11-12 14:32 ` Paul E. McKenney
2015-11-12 14:50 ` Måns Rullgård
2015-11-12 14:59 ` Paul E. McKenney
2015-11-12 17:46 ` David Daney [this message]
2015-11-12 18:00 ` Peter Zijlstra
2015-11-12 18:13 ` Måns Rullgård
2015-11-12 18:17 ` David Daney
2016-01-27 9:57 ` Maciej W. Rozycki
2016-01-27 11:43 ` Will Deacon
2016-01-27 12:41 ` Maciej W. Rozycki
2016-01-28 1:11 ` Boqun Feng
2016-01-27 14:54 ` Peter Zijlstra
2016-01-27 15:21 ` Will Deacon
2016-01-27 23:38 ` Paul E. McKenney
2016-01-28 9:57 ` Will Deacon
2016-01-28 22:31 ` Paul E. McKenney
2016-01-29 9:59 ` Will Deacon
2016-01-29 10:22 ` Paul E. McKenney
2016-02-01 13:56 ` Will Deacon
2016-02-02 3:54 ` Paul E. McKenney
2016-02-02 5:19 ` Boqun Feng
2016-02-02 6:44 ` Paul E. McKenney
2016-02-02 8:07 ` Linus Torvalds
2016-02-02 8:19 ` Linus Torvalds
2016-02-02 9:34 ` Boqun Feng
2016-02-02 17:30 ` Linus Torvalds
2016-02-02 17:51 ` Will Deacon
2016-02-02 18:06 ` Linus Torvalds
2016-02-02 19:30 ` Will Deacon
2016-02-02 19:55 ` Linus Torvalds
2016-02-03 19:13 ` Will Deacon
2016-02-03 8:33 ` Ingo Molnar
2016-02-03 13:32 ` Will Deacon
2016-02-03 19:03 ` Will Deacon
2016-02-09 11:23 ` Ingo Molnar
2016-02-09 11:42 ` Will Deacon
2016-02-02 12:02 ` Paul E. McKenney
2016-02-02 17:56 ` Linus Torvalds
2016-02-02 22:30 ` Paul E. McKenney
2016-02-02 14:49 ` Ralf Baechle
2016-02-02 14:54 ` Måns Rullgård
2016-02-02 14:58 ` Ralf Baechle
2016-02-02 15:51 ` Måns Rullgård
2016-02-02 17:23 ` Peter Zijlstra
2016-02-02 22:38 ` Paul E. McKenney
2016-02-02 11:45 ` Will Deacon
2016-02-02 12:12 ` Boqun Feng
2016-02-02 12:20 ` Will Deacon
2016-02-02 13:18 ` Boqun Feng
2016-02-02 17:12 ` Paul E. McKenney
2016-02-02 17:37 ` Will Deacon
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5644D08D.4080206@caviumnetworks.com \
--to=ddaney@caviumnetworks.com \
--cc=boqun.feng@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=paulmck@linux.vnet.ibm.com \
--cc=peterz@infradead.org \
--cc=ralf@linux-mips.org \
--cc=torvalds@linux-foundation.org \
--cc=will.deacon@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.