Re: [RFC PATCH 5/7] perf, x86: use the PEBS auto reload mechanism when possible

[Peter Zijlstra]

On Wed, May 28, 2014 at 07:46:43AM -0700, Andi Kleen wrote:
> On Wed, May 28, 2014 at 09:59:37AM +0200, Peter Zijlstra wrote:
> > On Wed, May 28, 2014 at 02:18:08PM +0800, Yan, Zheng wrote:
> > > When a fixed period is specified, this patch make perf use the PEBS
> > > auto reload mechanism. This makes normal profiling faster, because
> > > it avoids one costly MSR write in the PMI handler.
> > > 
> > 
> > You can't do this unconditionally. Like I think you mention in the next
> > patch, we need the PMI to fix up lots of cases.
> 
> There are two cases:
> 
> (1) Fixed period specified with PEBS.
> The PMI still runs with threshold=1, but we can use the PEBS
> auto-reload mechanism to automatically rewrite the counter which is
> faster.
> 
> (2) Free running mode with more constraints
> 
> This is case (1)

Sounds like something that should've been made clear in the Changelog.

Attachment: pgpUZStXBGDvl.pgp
Description: PGP signature