Re: [PATCH 1/2] perf/x86: update Haswell PEBS event constraints

[Andi Kleen <ak@linux.intel.com>]

On Mon, Jun 23, 2014 at 09:35:00AM +0200, Peter Zijlstra wrote:
> On Thu, Jun 19, 2014 at 01:40:41PM -0700, Andi Kleen wrote:
> > On Thu, Jun 19, 2014 at 10:31:29PM +0200, Stephane Eranian wrote:
> > > On Thu, Jun 19, 2014 at 10:18 PM, Andi Kleen <ak@linux.intel.com> wrote:
> > > >> I don't quite understand that.
> > > >> You need to know which events support PEBS. You need a table
> > > >
> > > > We're talking about the kernel allowing things here.
> > > > Yes the user still needs to know what supports PEBS, but
> > > > that doesn't concern the kernel.
> > > >
> > > Just need to make sure you don't return bogus information.
> > 
> > GIGO. We only need to prevent security issues.
> 
> > Yes if the user specifies a bogus raw event it will not count.
> > That's fine. The important part is just that nothing ever crashes.
> 
> Right. But IIRC you were previously arguing that we can in fact crash
> the machine with raw PEBS events, as illustrated with the SNB PEBS
> cycles 'event'.

The potential problem could only happen for a recognized PEBS event/umask,
but with unsupported flag combinations. That is what the SDM warns
about in 18.8.4.

If the event is not recognized as PEBS it will just effectively 
disable the event.

> Which is where my strict_pebs patch came from; by default only allow the
> sanitized known-safe list of events, but allow the system administrator
> to disable that test and allow any PEBS event.

I don't think we need to enforce the list of events
(except for the few with special limited counters)

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only