From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Boqun Feng <boqun.feng@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
Peter Zijlstra <peterz@infradead.org>,
Herbert Xu <herbert@gondor.apana.org.au>,
Leonid Yegoshin <Leonid.Yegoshin@imgtec.com>,
linux-mips <linux-mips@linux-mips.org>,
"linux-ia64@vger.kernel.org" <linux-ia64@vger.kernel.org>,
"Michael S. Tsirkin" <mst@redhat.com>,
Will Deacon <will.deacon@arm.com>,
virtualization <virtualization@lists.linux-foundation.org>,
Peter Anvin <hpa@zytor.com>,
sparclinux@vger.kernel.org, Ingo Molnar <mingo@kernel.org>,
"linux-arch@vger.kernel.org" <linux-arch@vger.kernel.org>,
linux-s390 <linux-s390@vger.kernel.org>,
Russell King - ARM Linux <linux@arm.linux.org.uk>,
uml-devel <user-mode-linux-devel@lists.sourceforge.net>,
linux-sh@vger.kernel.org, Michael Ellerman <mpe@ellerman.id.au>,
the arch/x86 maintainers <x86@kernel.org>,
xen-devel@lists.xenproject.org, Ingo Molnar <mingo@elte.hu>,
linux-xtensa@linux-xtensa.org,
James Hogan <james.hogan@imgtec.com>,
Arnd Bergmann <arnd@arndb.de>,
Stefano Stabellini <stefano.stabellini@eu.citrix.com>,
adi-buildroot-devel@lists.sourceforge.net,
David Daney <ddaney.cavm@gmail.com>,
Thomas Gleixner <tglx@linutronix.de>,
linux-metag@vger.kernel.org,
"linux-arm-kernel@lists.infradead.org"
<linux-arm-kernel@lists.infradead.org>,
Andrew Cooper <andrew.cooper3@citrix.com>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Ralf Baechle <ralf@linux-mips.org>, Joe Perches <joe@perches.com>,
ppc-dev <linuxppc-dev@lists.ozlabs.org>,
David Miller <davem@davemloft.net>
Subject: Re: [v3,11/41] mips: reuse asm-generic/barrier.h
Date: Wed, 27 Jan 2016 15:30:18 -0800 [thread overview]
Message-ID: <20160127233018.GN4503@linux.vnet.ibm.com> (raw)
In-Reply-To: <20160127020447.GA1293@fixme-laptop.cn.ibm.com>
On Wed, Jan 27, 2016 at 10:04:47AM +0800, Boqun Feng wrote:
> On Tue, Jan 26, 2016 at 03:29:21PM -0800, Paul E. McKenney wrote:
> > On Tue, Jan 26, 2016 at 02:33:40PM -0800, Linus Torvalds wrote:
> > > On Tue, Jan 26, 2016 at 2:15 PM, Linus Torvalds
> > > <torvalds@linux-foundation.org> wrote:
> > > >
> > > > You might as well just write it as
> > > >
> > > > struct foo x = READ_ONCE(*ptr);
> > > > x->bar = 5;
> > > >
> > > > because that "smp_read_barrier_depends()" does NOTHING wrt the second write.
> > >
> > > Just to clarify: on alpha it adds a memory barrier, but that memory
> > > barrier is useless.
> >
> > No trailing data-dependent read, so agreed, no smp_read_barrier_depends()
> > needed. That said, I believe that we should encourage rcu_dereference*()
> > or lockless_dereference() instead of READ_ONCE() for documentation
> > reasons, though.
> >
> > > On non-alpha, it is a no-op, and obviously does nothing simply because
> > > it generates no code.
> > >
> > > So if anybody believes that the "smp_read_barrier_depends()" does
> > > something, they are *wrong*.
> >
> > The other problem with smp_read_barrier_depends() is that it is often
> > a pain figuring out which prior load it is supposed to apply to.
> > Hence my preference for rcu_dereference*() and lockless_dereference().
> >
>
> Because semantically speaking, rcu_derefence*() and
> lockless_dereference() are CONSUME(i.e. data/address dependent
> read->read and read->write pairs are ordered), whereas
> smp_read_barrier_depends() only guarantees read->read pairs with data
> dependency are ordered, right?
>
> If so, maybe we need to call it out in memory-barriers.txt, for example:
>
> diff --git a/Documentation/memory-barriers.txt b/Documentation/memory-barriers.txt
> index 904ee42..6b262c2 100644
> --- a/Documentation/memory-barriers.txt
> +++ b/Documentation/memory-barriers.txt
> @@ -1703,8 +1703,8 @@ There are some more advanced barrier functions:
>
>
> (*) lockless_dereference();
> - This can be thought of as a pointer-fetch wrapper around the
> - smp_read_barrier_depends() data-dependency barrier.
> + This is a load, and any load or store that has a data dependency on the
> + value returned by this load won't be reordered before this load.
This is a good start, but more is needed to warn people off of
smp_read_barrier_depends(). But yes, better explanation would be good.
Thanx, Paul
> This is also similar to rcu_dereference(), but in cases where
> object lifetime is handled by some mechanism other than RCU, for
>
>
> Regards,
> Boqun
>
> > > And if anybody sends out an email with that smp_read_barrier_depends()
> > > in an example, they are actively just confusing other people, which is
> > > even worse than just being wrong. Which is why I jumped in.
> > >
> > > So stop perpetuating the myth that smp_read_barrier_depends() does
> > > something here. It does not. It's a bug, and it has become this "mind
> > > virus" for some people that seem to believe that it does something.
> >
> > It looks like I should add words to memory-barriers.txt de-emphasizing
> > smp_read_barrier_depends(). I will take a look at that.
> >
> > > I had to remove this crap once from the kernel already, see commit
> > > 105ff3cbf225 ("atomic: remove all traces of READ_ONCE_CTRL() and
> > > atomic*_read_ctrl()").
> > >
> > > I don't want to ever see that broken construct again. And I want to
> > > make sure that everybody is educated about how broken it was. I'm
> > > extremely unhappy that it came up again.
> >
> > Well, if it makes you feel better, that was control dependencies and this
> > was data dependencies. So it was not -exactly- the same. ;-)
> >
> > (Sorry, couldn't resist...)
> >
> > > If it turns out that some architecture does actually need a barrier
> > > between a read and a dependent write, then that will mean that
> > >
> > > (a) we'll have to make up a _new_ barrier, because
> > > "smp_read_barrier_depends()" is not that barrier. We'll presumably
> > > then have to make that new barrier part of "rcu_derefence()" and
> > > friends.
> >
> > Agreed. We can worry about whether or not we replace the current
> > smp_read_barrier_depends() with that new barrier when and if such
> > hardware appears.
> >
> > > (b) we will have found an architecture with even worse memory
> > > ordering semantics than alpha, and we'll have to stop castigating
> > > alpha for being the worst memory ordering ever.
> >
> > ;-) ;-) ;-)
> >
> > > but I sincerely hope that we'll never find that kind of broken architecture.
> >
> > Apparently at least some hardware vendors are reading memory-barriers.txt,
> > so perhaps the odds of that kind of breakage have reduced.
> >
> > Thanx, Paul
> >
next prev parent reply other threads:[~2016-01-28 0:45 UTC|newest]
Thread overview: 155+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-01-10 14:16 [PATCH v3 00/41] arch: barrier cleanup + barriers for virt Michael S. Tsirkin
2016-01-10 14:16 ` [PATCH v3 01/41] lcoking/barriers, arch: Use smp barriers in smp_store_release() Michael S. Tsirkin
2016-01-12 16:28 ` Paul E. McKenney
2016-01-12 18:40 ` Michael S. Tsirkin
2016-01-10 14:16 ` [PATCH v3 02/41] asm-generic: guard smp_store_release/load_acquire Michael S. Tsirkin
2016-01-10 14:16 ` [PATCH v3 03/41] ia64: rename nop->iosapic_nop Michael S. Tsirkin
2016-01-10 14:17 ` [PATCH v3 04/41] ia64: reuse asm-generic/barrier.h Michael S. Tsirkin
2016-01-10 14:17 ` [PATCH v3 05/41] powerpc: " Michael S. Tsirkin
2016-01-12 16:31 ` Paul E. McKenney
2016-01-10 14:17 ` [PATCH v3 06/41] s390: " Michael S. Tsirkin
2016-01-10 14:17 ` [PATCH v3 07/41] sparc: " Michael S. Tsirkin
2016-01-10 14:17 ` [PATCH v3 08/41] arm: " Michael S. Tsirkin
2016-01-10 14:17 ` [PATCH v3 09/41] arm64: " Michael S. Tsirkin
2016-01-10 14:17 ` [PATCH v3 10/41] metag: " Michael S. Tsirkin
2016-01-10 14:18 ` [PATCH v3 11/41] mips: " Michael S. Tsirkin
2016-01-12 1:14 ` [v3,11/41] " Leonid Yegoshin
2016-01-12 8:43 ` Michael S. Tsirkin
2016-01-12 9:51 ` Peter Zijlstra
2016-01-12 9:27 ` Peter Zijlstra
2016-01-12 10:25 ` Peter Zijlstra
2016-01-12 10:40 ` Peter Zijlstra
2016-01-12 11:41 ` Will Deacon
2016-01-12 20:45 ` Leonid Yegoshin
2016-01-12 21:40 ` Peter Zijlstra
2016-01-13 0:21 ` Leonid Yegoshin
2016-01-13 10:45 ` Will Deacon
2016-01-13 19:02 ` Leonid Yegoshin
2016-01-13 20:48 ` Peter Zijlstra
2016-01-13 20:58 ` Leonid Yegoshin
2016-01-14 12:04 ` Will Deacon
2016-01-14 16:16 ` Paul E. McKenney
2016-01-14 19:42 ` Leonid Yegoshin
2016-01-14 20:15 ` Peter Zijlstra
2016-01-14 20:36 ` Paul E. McKenney
2016-01-14 20:46 ` Peter Zijlstra
2016-01-14 20:46 ` Leonid Yegoshin
2016-01-14 21:34 ` Paul E. McKenney
2016-01-14 21:45 ` Leonid Yegoshin
2016-01-14 22:24 ` Paul E. McKenney
2016-01-14 23:04 ` Leonid Yegoshin
2016-01-14 20:12 ` Leonid Yegoshin
2016-01-14 20:48 ` Paul E. McKenney
2016-01-14 21:24 ` Leonid Yegoshin
2016-01-14 22:20 ` Paul E. McKenney
2016-01-15 9:57 ` Will Deacon
2016-01-15 18:54 ` Leonid Yegoshin
2016-01-26 10:24 ` Peter Zijlstra
2016-01-26 10:32 ` Peter Zijlstra
2016-01-26 11:09 ` Will Deacon
2016-01-26 20:11 ` Paul E. McKenney
2016-01-27 8:35 ` [PATCH] documentation: Add disclaimer Peter Zijlstra
2016-01-27 10:11 ` Will Deacon
2016-04-14 21:40 ` Paul E. McKenney
2016-01-27 14:57 ` David Howells
2016-01-27 23:35 ` Paul E. McKenney
2016-01-28 20:02 ` David Howells
2016-04-14 21:40 ` Paul E. McKenney
2016-01-26 19:44 ` [v3,11/41] mips: reuse asm-generic/barrier.h Paul E. McKenney
2016-01-18 8:19 ` Herbert Xu
2016-01-18 15:46 ` Paul E. McKenney
2016-01-26 16:52 ` Boqun Feng
2016-01-26 17:22 ` Peter Zijlstra
2016-01-26 19:44 ` Linus Torvalds
2016-01-26 20:10 ` Paul E. McKenney
2016-01-26 22:15 ` Linus Torvalds
2016-01-26 22:33 ` Linus Torvalds
2016-01-26 23:29 ` Paul E. McKenney
2016-01-26 23:45 ` Linus Torvalds
2016-01-27 0:57 ` Paul E. McKenney
2016-01-27 2:04 ` Boqun Feng
2016-01-27 23:30 ` Paul E. McKenney [this message]
2016-01-27 7:51 ` Peter Zijlstra
2016-01-27 8:05 ` Linus Torvalds
2016-01-26 19:51 ` Paul E. McKenney
2016-01-13 22:26 ` Leonid Yegoshin
2016-01-14 9:24 ` Michael S. Tsirkin
2016-01-14 12:14 ` Will Deacon
2016-01-14 19:28 ` Leonid Yegoshin
2016-01-14 20:34 ` Paul E. McKenney
2016-01-14 21:01 ` Leonid Yegoshin
2016-01-14 21:29 ` Paul E. McKenney
2016-01-14 21:36 ` Leonid Yegoshin
2016-01-14 22:55 ` Paul E. McKenney
2016-01-14 23:33 ` Leonid Yegoshin
2016-01-15 0:47 ` Paul E. McKenney
2016-01-15 1:07 ` Leonid Yegoshin
2016-01-27 11:26 ` Maciej W. Rozycki
2016-01-28 0:48 ` Leonid Yegoshin
2016-01-29 13:38 ` Maciej W. Rozycki
2016-01-28 0:58 ` Leonid Yegoshin
2016-01-27 10:40 ` Ralf Baechle
2016-01-27 12:09 ` Maciej W. Rozycki
2016-01-15 10:24 ` Will Deacon
2016-01-15 17:54 ` Paul E. McKenney
2016-01-15 19:28 ` Paul E. McKenney
2016-01-25 14:41 ` Will Deacon
2016-01-26 1:06 ` Paul E. McKenney
2016-01-26 12:10 ` Will Deacon
2016-01-26 23:37 ` Paul E. McKenney
2016-01-27 10:23 ` Will Deacon
2016-01-15 8:55 ` Peter Zijlstra
2016-01-15 9:13 ` Peter Zijlstra
2016-01-15 17:46 ` Paul E. McKenney
2016-01-15 21:27 ` Peter Zijlstra
2016-01-15 21:58 ` Paul E. McKenney
2016-01-25 16:42 ` Will Deacon
2016-01-26 6:03 ` Paul E. McKenney
2016-01-26 10:19 ` Peter Zijlstra
2016-01-26 20:13 ` Paul E. McKenney
2016-01-27 8:39 ` Peter Zijlstra
2016-01-26 12:16 ` Will Deacon
2016-01-26 14:35 ` Boqun Feng
2016-01-26 19:58 ` Paul E. McKenney
2016-01-27 10:25 ` Will Deacon
2016-01-27 23:32 ` Paul E. McKenney
2016-01-15 17:39 ` Paul E. McKenney
2016-01-15 21:29 ` Peter Zijlstra
2016-01-15 22:01 ` Paul E. McKenney
2016-01-25 18:02 ` Will Deacon
2016-01-26 6:12 ` Paul E. McKenney
2016-01-26 10:15 ` Peter Zijlstra
2016-01-10 14:18 ` [PATCH v3 12/41] x86/um: " Michael S. Tsirkin
2016-01-10 14:18 ` [PATCH v3 13/41] x86: " Michael S. Tsirkin
2016-01-12 14:10 ` Thomas Gleixner
2016-01-10 14:18 ` [PATCH v3 14/41] asm-generic: add __smp_xxx wrappers Michael S. Tsirkin
2016-01-10 14:18 ` [PATCH v3 15/41] powerpc: define __smp_xxx Michael S. Tsirkin
2016-01-10 14:18 ` [PATCH v3 16/41] arm64: " Michael S. Tsirkin
2016-01-10 14:18 ` [PATCH v3 17/41] arm: " Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 18/41] blackfin: " Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 19/41] ia64: " Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 20/41] metag: " Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 21/41] mips: " Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 22/41] s390: " Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 23/41] sh: define __smp_xxx, fix smp_store_mb for !SMP Michael S. Tsirkin
2016-01-10 14:19 ` [PATCH v3 24/41] sparc: define __smp_xxx Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 25/41] tile: " Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 26/41] xtensa: " Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 27/41] x86: " Michael S. Tsirkin
2016-01-12 14:11 ` Thomas Gleixner
2016-01-10 14:20 ` [PATCH v3 28/41] asm-generic: implement virt_xxx memory barriers Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 29/41] Revert "virtio_ring: Update weak barriers to use dma_wmb/rmb" Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 30/41] virtio_ring: update weak barriers to use virt_xxx Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 31/41] sh: support 1 and 2 byte xchg Michael S. Tsirkin
2016-01-10 14:20 ` [PATCH v3 32/41] sh: move xchg_cmpxchg to a header by itself Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 33/41] virtio_ring: use virt_store_mb Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 34/41] checkpatch.pl: add missing memory barriers Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 35/41] checkpatch: check for __smp outside barrier.h Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 36/41] checkpatch: add virt barriers Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 37/41] xenbus: use virt_xxx barriers Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 38/41] xen/io: " Michael S. Tsirkin
2016-01-10 14:21 ` [PATCH v3 39/41] xen/events: " Michael S. Tsirkin
2016-01-11 11:12 ` David Vrabel
2016-01-10 14:22 ` [PATCH v3 40/41] s390: use generic memory barriers Michael S. Tsirkin
2016-01-10 14:22 ` [PATCH v3 41/41] s390: more efficient smp barriers Michael S. Tsirkin
2016-01-12 12:50 ` [PATCH v3 00/41] arch: barrier cleanup + barriers for virt 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=20160127233018.GN4503@linux.vnet.ibm.com \
--to=paulmck@linux.vnet.ibm.com \
--cc=Leonid.Yegoshin@imgtec.com \
--cc=adi-buildroot-devel@lists.sourceforge.net \
--cc=andrew.cooper3@citrix.com \
--cc=arnd@arndb.de \
--cc=boqun.feng@gmail.com \
--cc=davem@davemloft.net \
--cc=ddaney.cavm@gmail.com \
--cc=herbert@gondor.apana.org.au \
--cc=hpa@zytor.com \
--cc=james.hogan@imgtec.com \
--cc=joe@perches.com \
--cc=linux-arch@vger.kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-ia64@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-metag@vger.kernel.org \
--cc=linux-mips@linux-mips.org \
--cc=linux-s390@vger.kernel.org \
--cc=linux-sh@vger.kernel.org \
--cc=linux-xtensa@linux-xtensa.org \
--cc=linux@arm.linux.org.uk \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=mingo@elte.hu \
--cc=mingo@kernel.org \
--cc=mpe@ellerman.id.au \
--cc=mst@redhat.com \
--cc=peterz@infradead.org \
--cc=ralf@linux-mips.org \
--cc=sparclinux@vger.kernel.org \
--cc=stefano.stabellini@eu.citrix.com \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=user-mode-linux-devel@lists.sourceforge.net \
--cc=virtualization@lists.linux-foundation.org \
--cc=will.deacon@arm.com \
--cc=x86@kernel.org \
--cc=xen-devel@lists.xenproject.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).