On Thu, Aug 21, 2008 at 02:03:26PM +0200, Peter Zijlstra wrote:
quoted text
> On Thu, 2008-08-21 at 04:47 -0700, Paul E. McKenney wrote:
> > On Thu, Aug 21, 2008 at 01:02:48PM +0200, Peter Zijlstra wrote:
> > > On Thu, 2008-08-21 at 12:57 +0200, Stefan Richter wrote:
> > > > Peter Zijlstra wrote:
> > > > > On Wed, 2008-08-20 at 20:50 -0600, jmerkey@wolfmountaingroup.com wrote:
> > > > > 
> > > > >>  volatiles left in the code due to the previously stated
> > > > >>  (and still present) severe breakage of the GNU compiler with SMP 
> > > > >>  shared data.  most of the barrier() functions are just plain broken
> > > > >>  and do not result in proper compiler behavior in this tree. 
> > > > > 
> > > > > Can you provide explicit detail?
> > > > > 
> > > > > By using barrier() the compiler should clobber all its memory and
> > > > > registers therefore forcing a write/reload of the variable.
> > > > 
> > > > I hope Jeff didn't try mere barrier()s only.  smp_wmb() and smp_rmb()
> > > > are the more relevant barrier variants for mdb, from what I remember
> > > > when I last looked at it.
> > > 
> > > Sure, but volatile isn't a replacement for memory barriers.
> > 
> > Let's face it, the C standard does not support concurrency, so we are
> > all in a state of sin in any case, forced to rely on combinations of
> > gcc-specific non-standard language extensions and assembly language.
> 
> Hehe, still, a little birdie told me they are working on it and perhaps
> someone with clue could enlighten us on their direction.

Well, I guess you guys will be the judge of that.  Or one of the judges,
at least.  ;-)

One advantage of the current c++0x approach is that it allows extremely
weak memory barriers to be used in many cases that would require smp_rmb()
in current Linux kernel.  If you are crazy enough to want to see a
sneak preview in standardese, try all 10MB of:

	http://open-std.org/jtc1/sc22/wg21/docs/papers/2008/n2691.pdf

Section 1.10 (physical page 25, logical page 11) describes the memory model.
Sections 29 and 30 describe the operations (physical page 1155, logical
page 1141).  The C and C++ guys got together ahead of time and agreed to
work together towards a compatible solution.

And rcu_dereference() would be implemented in terms of memory_order_consume,
for whatever that is worth.

quoted text
> Still, I'd like Jeff to show his C, the resulting asm and the intent for
> the volatile and barrier versions of his code (well, little snippets of
> his code obviuosly).
> 
> Either he doesn't understand barriers (nothing to be ashamed about), or
> we might have more trouble lurking in the rest of the kernel.

Sounds fair to me!

							Thanx, Paul
--
unsubscribe notice
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/