From: "Michael S. Tsirkin" <mst@redhat.com>
To: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
linux-kernel@vger.kernel.org, mingo@kernel.org,
jiangshanlai@gmail.com, dipankar@in.ibm.com,
akpm@linux-foundation.org, mathieu.desnoyers@efficios.com,
josh@joshtriplett.org, tglx@linutronix.de, rostedt@goodmis.org,
dhowells@redhat.com, edumazet@google.com, fweisbec@gmail.com,
oleg@redhat.com, Jason Wang <jasowang@redhat.com>,
kvm@vger.kernel.org, virtualization@lists.linux-foundation.org,
netdev@vger.kernel.org
Subject: Re: [PATCH tip/core/rcu 21/21] drivers/vhost: Remove now-redundant read_barrier_depends()
Date: Tue, 5 Dec 2017 21:51:48 +0200 [thread overview]
Message-ID: <20171205215020-mutt-send-email-mst@kernel.org> (raw)
In-Reply-To: <20171205193339.GP7829@linux.vnet.ibm.com>
On Tue, Dec 05, 2017 at 11:33:39AM -0800, Paul E. McKenney wrote:
> On Tue, Dec 05, 2017 at 09:24:21PM +0200, Michael S. Tsirkin wrote:
> > On Tue, Dec 05, 2017 at 08:17:33PM +0100, Peter Zijlstra wrote:
> > > On Tue, Dec 05, 2017 at 08:57:46PM +0200, Michael S. Tsirkin wrote:
> > >
> > > > I don't see WRITE_ONCE inserting any barriers, release or
> > > > write.
> > >
> > > Correct, never claimed there was.
> > >
> > > Just saying that:
> > >
> > > obj = READ_ONCE(*foo);
> > > val = READ_ONCE(obj->val);
> > >
> > > Never needs a barrier (except on Alpha and we want to make that go
> > > away). Simply because a CPU needs to complete the load of @obj before it
> > > can compute the address &obj->val. Thus the second load _must_ come
> > > after the first load and we get LOAD-LOAD ordering.
> > >
> > > Alpha messing that up is a royal pain, and Alpha not being an
> > > active/living architecture is just not worth the pain of keeping this in
> > > the generic model.
> > >
> >
> > Right. What I am saying is that for writes you need
> >
> > WRITE_ONCE(obj->val, 1);
> > smp_wmb();
> > WRITE_ONCE(*foo, obj);
>
> I believe Peter was instead suggesting:
>
> WRITE_ONCE(obj->val, 1);
> smp_store_release(foo, obj);
Isn't that more expensive though?
> > and this barrier is no longer paired with anything until
> > you realize there's a dependency barrier within READ_ONCE.
> >
> > Barrier pairing was a useful tool to check code validity,
> > maybe there are other, better tools now.
>
> There are quite a few people who say that smp_store_release() is
> easier for the tools to analyze than is smp_wmb(). My experience with
> smp_read_barrier_depends() and rcu_dereference() leads me to believe
> that they are correct.
>
> Thanx, Paul
OK, but smp_store_release is still not paired with anything since we
rely on READ_ONCE to include the implicit dpendendency barrier.
--
MST
next prev parent reply other threads:[~2017-12-05 19:51 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-12-01 19:50 [PATCH tip/core/rcu 0/21] De-emphasize {smp_,}read_barrier_depends Paul E. McKenney
2017-12-01 19:50 ` [PATCH tip/core/rcu 01/21] doc: READ_ONCE() now implies smp_barrier_depends() Paul E. McKenney
2017-12-01 19:50 ` [PATCH tip/core/rcu 02/21] mn10300: READ_ONCE() now implies smp_read_barrier_depends() Paul E. McKenney
2017-12-01 19:50 ` [PATCH tip/core/rcu 03/21] drivers/net/ethernet/qlogic/qed: Fix __qed_spq_block() ordering Paul E. McKenney
2017-12-01 19:50 ` [PATCH tip/core/rcu 04/21] fs/dcache: Use release-acquire for name/length update Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 05/21] percpu: READ_ONCE() now implies smp_read_barrier_depends() Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 06/21] rcu: Adjust read-side accessor comments for READ_ONCE() Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 07/21] rtnetlink: Update now-misleading smp_read_barrier_depends() comment Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 08/21] seqlock: Remove now-redundant smp_read_barrier_depends() Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 09/21] uprobes: " Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 10/21] locking: Remove smp_read_barrier_depends() from queued_spin_lock_slowpath() Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 11/21] tracepoint: Remove smp_read_barrier_depends() from comment Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 12/21] lib/assoc_array: Remove smp_read_barrier_depends() Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 13/21] mm/ksm: Remove now-redundant smp_read_barrier_depends() Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 14/21] netfilter: " Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 15/21] keyring: " Paul E. McKenney
2017-12-04 0:59 ` James Morris
2017-12-04 18:54 ` Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 16/21] drivers/infiniband: " Paul E. McKenney
2017-12-02 0:11 ` Jason Gunthorpe
2017-12-02 1:08 ` Paul E. McKenney
2017-12-05 15:08 ` Jason Gunthorpe
2017-12-05 20:06 ` Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 17/21] doc: De-emphasize smp_read_barrier_depends Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 18/21] genetlink: Remove smp_read_barrier_depends() from comment Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 19/21] netlink: " Paul E. McKenney
2017-12-01 19:51 ` [PATCH tip/core/rcu 20/21] checkpatch: Add warnings for {smp_,}read_barrier_depends() Paul E. McKenney
2017-12-01 20:14 ` Joe Perches
2017-12-01 21:44 ` Paul E. McKenney
2017-12-02 4:45 ` Joe Perches
2017-12-04 19:06 ` Paul E. McKenney
2017-12-04 19:12 ` Joe Perches
2017-12-01 19:51 ` [PATCH tip/core/rcu 21/21] drivers/vhost: Remove now-redundant read_barrier_depends() Paul E. McKenney
2017-12-05 18:31 ` Michael S. Tsirkin
2017-12-05 18:39 ` Peter Zijlstra
2017-12-05 18:57 ` Michael S. Tsirkin
2017-12-05 19:17 ` Peter Zijlstra
2017-12-05 19:24 ` Michael S. Tsirkin
2017-12-05 19:33 ` Paul E. McKenney
2017-12-05 19:51 ` Michael S. Tsirkin [this message]
2017-12-05 19:57 ` Peter Zijlstra
2017-12-05 20:28 ` Michael S. Tsirkin
2017-12-05 21:17 ` Peter Zijlstra
2017-12-05 21:42 ` Michael S. Tsirkin
2017-12-05 20:08 ` Paul E. McKenney
2017-12-05 21:24 ` Michael S. Tsirkin
2017-12-05 21:36 ` Paul E. McKenney
2017-12-05 21:43 ` Michael S. Tsirkin
2017-12-05 22:02 ` Paul E. McKenney
2017-12-05 22:09 ` Peter Zijlstra
2017-12-05 21:57 ` Peter Zijlstra
2017-12-05 22:09 ` Michael S. Tsirkin
2017-12-05 23:39 ` Paul E. McKenney
2017-12-05 19:55 ` Peter Zijlstra
2017-12-04 15:38 ` [PATCH tip/core/rcu 01/21] doc: READ_ONCE() now implies smp_barrier_depends() David Howells
2017-12-04 18:52 ` Paul E. McKenney
2017-12-04 21:54 ` Peter Zijlstra
2017-12-04 22:15 ` Paul E. McKenney
2017-12-04 22:39 ` David Howells
2017-12-04 22:57 ` Peter Zijlstra
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=20171205215020-mutt-send-email-mst@kernel.org \
--to=mst@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=dhowells@redhat.com \
--cc=dipankar@in.ibm.com \
--cc=edumazet@google.com \
--cc=fweisbec@gmail.com \
--cc=jasowang@redhat.com \
--cc=jiangshanlai@gmail.com \
--cc=josh@joshtriplett.org \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mingo@kernel.org \
--cc=netdev@vger.kernel.org \
--cc=oleg@redhat.com \
--cc=paulmck@linux.vnet.ibm.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
--cc=virtualization@lists.linux-foundation.org \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).