Re: perf: PERF_EVENT_IOC_PERIOD on ARM vs everywhere else

[Peter Zijlstra <peterz@infradead.org>]

On Tue, Oct 29, 2013 at 04:28:10AM +0000, Will Deacon wrote:
> On Mon, Oct 28, 2013 at 02:07:48PM +0000, Vince Weaver wrote:
> > It's also a shame this change apprently didn't hit the linux-kernel list 
> > as far as I can tell.  I do my best to try to note all of the perf 
> > ABI-related changes there, but if things like this are going to start 
> > getting merged in architecture trees then things get that much harder 
> > to keep track of.
> 
> I can CC LKML on ARM perf patches if you think it will help, but all PMU
> backend patches go via their respective arch trees afaict.

Just those that change user visible semantics that are shared between
archs I suppose :-)

> > > I don't want to be the `oddball' architecture (at least, not more than I am
> > > already :), but I do find it tricky to follow the required semantics of perf
> > > as opposed to `it happens to work this way', especially when so much of it
> > > is buried in the various arch backends. So if somebody using the thing on
> > > ARM has (what looks to me like) a valid issue, then I usually try and fix
> > > it.
> > 
> > But it was global behavior that was common on all architectures.
> > 
> > Now any cross-platform tool like PAPI is going to have to have a mess of 
> > #ifdefs around every use of this ioctl, and it will only get worse if 
> > other architectures decide to "fix" the problem too.
> 
> What would you like me to do to fix this for you? Moving more code out of
> the backends and into the core will help maintain consistency between
> architectures, but that's a huge job.

We could start by making all archs do the same thing again; but yes
ideally we'd move some of it into generic code. Not entirely sure how
that will work out though, there's a reason its in per-arch code :/


Vince, what would you prefer to do here?