Re: perf events ring buffer memory barrier on powerpc

[Paul E. McKenney]

On Mon, Nov 04, 2013 at 09:57:17AM +0000, Will Deacon wrote:
> Hi Paul,
> 
> On Sun, Nov 03, 2013 at 10:47:12PM +0000, Paul E. McKenney wrote:
> > On Sun, Nov 03, 2013 at 05:07:59PM +0000, Will Deacon wrote:
> > > On Sun, Nov 03, 2013 at 02:40:17PM +0000, Paul E. McKenney wrote:
> > > > On Sat, Nov 02, 2013 at 10:32:39AM -0700, Paul E. McKenney wrote:
> > > > > On Fri, Nov 01, 2013 at 03:56:34PM +0100, Peter Zijlstra wrote:
> > > > > > On Wed, Oct 30, 2013 at 11:40:15PM -0700, Paul E. McKenney wrote:
> > > > > > > > Now the whole crux of the question is if we need barrier A at all, since
> > > > > > > > the STORES issued by the @buf writes are dependent on the ubuf->tail
> > > > > > > > read.
> > > > > > > 
> > > > > > > The dependency you are talking about is via the "if" statement?
> > > > > > > Even C/C++11 is not required to respect control dependencies.
> > > > > > > 
> > > > > > > This one is a bit annoying.  The x86 TSO means that you really only
> > > > > > > need barrier(), ARM (recent ARM, anyway) and Power could use a weaker
> > > > > > > barrier, and so on -- but smp_mb() emits a full barrier.
> > > > > > > 
> > > > > > > Perhaps a new smp_tmb() for TSO semantics, where reads are ordered
> > > > > > > before reads, writes before writes, and reads before writes, but not
> > > > > > > writes before reads?  Another approach would be to define a per-arch
> > > > > > > barrier for this particular case.
> > > > > > 
> > > > > > I suppose we can only introduce new barrier primitives if there's more
> > > > > > than 1 use-case.
> > > 
> > > Which barrier did you have in mind when you refer to `recent ARM' above? It
> > > seems to me like you'd need a combination if dmb ishld and dmb ishst, since
> > > the former doesn't order writes before writes.
> > 
> > I heard a rumor that ARM had recently added a new dmb variant that acted
> > similarly to PowerPC's lwsync, and it was on my list to follow up.
> > 
> > Given your response, I am guessing that there is no truth to this rumor...
> 
> I think you're talking about the -ld option to dmb, which was introduced in
> ARMv8. That option orders loads against loads and stores, but doesn't order
> writes against writes. So you could do:
> 
> 	dmb ishld
> 	dmb ishst
> 
> but it's questionable whether that performs better than a dmb ish.

If Linus's smp_store_with_release_semantics() approach works out, ARM
should be able to use its shiny new ldar and stlr instructions.

							Thanx, Paul

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/