From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Miklos Szeredi <miklos@szeredi.hu>,
James Bottomley <James.Bottomley@HansenPartnership.com>,
dhowells@redhat.com, linux-kernel@vger.kernel.org,
linux-arch@vger.kernel.org
Subject: Re: memory barrier question
Date: Fri, 17 Sep 2010 16:12:22 -0700 [thread overview]
Message-ID: <20100917231222.GA3060@linux.vnet.ibm.com> (raw)
In-Reply-To: <1284760148.30449.107.camel@pasglop>
On Sat, Sep 18, 2010 at 07:49:08AM +1000, Benjamin Herrenschmidt wrote:
>
> > Right but in the concrete namei example I can't see how a compiler
> > optimization can make a difference. The order of the loads is quite
> > clear:
> >
> > LOAD inode = next.dentry->inode
> > if (inode != NULL)
> > LOAD inode->f_op
> >
> > What is there the compiler can optimize?
>
> Those two loads depend on each other, I don't think any implementation
> can re-order them. In fact, such data dependency is typically what is
> used to avoid having barriers in some cases. The second load cannot be
> issued until the value from the first one is returned.
Sufficiently sadistic compiler and CPU implementations could do value
speculation, for example, driven by profile-feedback optimization.
Then the guess might initially incorrect, but then a store by some other
CPU could make the subsequent test decide (wrongly) that the guess had
in fact been correct.
Needless to say, I am not a fan of value speculation. But other people
do like it a lot.
Thanx, Paul
next prev parent reply other threads:[~2010-09-17 23:12 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-09-15 14:36 memory barrier question Miklos Szeredi
2010-09-15 14:36 ` Miklos Szeredi
2010-09-15 19:12 ` Rafael J. Wysocki
2010-09-16 11:55 ` David Howells
2010-09-16 13:42 ` Miklos Szeredi
2010-09-16 13:42 ` Miklos Szeredi
2010-09-16 14:30 ` David Howells
2010-09-16 15:03 ` Paul E. McKenney
2010-09-16 15:03 ` Paul E. McKenney
2010-09-16 16:06 ` Miklos Szeredi
2010-09-16 16:06 ` Miklos Szeredi
2010-09-16 16:37 ` Paul E. McKenney
2010-09-16 16:56 ` Miklos Szeredi
2010-09-16 17:09 ` James Bottomley
2010-09-16 17:17 ` Miklos Szeredi
2010-09-16 17:40 ` James Bottomley
2010-09-17 21:49 ` Benjamin Herrenschmidt
2010-09-17 23:12 ` Paul E. McKenney [this message]
2010-09-19 2:47 ` Benjamin Herrenschmidt
2010-09-19 15:26 ` Paul E. McKenney
2010-09-19 20:15 ` Miklos Szeredi
2010-09-19 21:59 ` Paul E. McKenney
2010-09-20 0:58 ` James Bottomley
2010-09-20 1:29 ` Paul E. McKenney
2010-09-20 16:01 ` Miklos Szeredi
2010-09-20 18:25 ` Paul E. McKenney
2010-09-20 18:57 ` Paul E. McKenney
2010-09-20 20:26 ` Michael Cree
2010-09-20 20:40 ` Paul E. McKenney
2010-09-21 14:59 ` Paul E. McKenney
2010-09-22 18:41 ` Paul E. McKenney
2010-09-18 1:12 ` Alan Cox
2010-09-16 16:50 ` Jamie Lokier
2010-09-16 16:18 ` Peter Zijlstra
2010-09-16 17:59 ` David Howells
-- strict thread matches above, loose matches on Subject: below --
2010-09-20 10:34 George Spelvin
2010-09-20 10:34 ` George Spelvin
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=20100917231222.GA3060@linux.vnet.ibm.com \
--to=paulmck@linux.vnet.ibm.com \
--cc=James.Bottomley@HansenPartnership.com \
--cc=benh@kernel.crashing.org \
--cc=dhowells@redhat.com \
--cc=linux-arch@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=miklos@szeredi.hu \
/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).