Hi -
On Sat, Mar 07, 2015 at 07:45:22AM +1100, Ken McDonell wrote:
> [...]
> I am sure the h/w supports this ... the tests run and pass on the native
> machine running a Debian derivative. This would seem to support that
> hypothesis ...
Sure, but that's the native kernel & possibly a newer version of papi.
Through KVM and an older guest kernel, bets are off.
> Since this involves several tests, seems like the _pmu_present() function
> should be factored out of all of these tests an put in common.check (or
> somewhere similar) and then generalized to accommodate the restriction
> you've suggested.
Sounds about right.
> The only issue here is that papi_avail is not installed by default,
> so we introduce yet another package dependency for QA and I can't
> even find a package containing papi_avail for a standard Debian
> distro.
Yeah, the papi-tools weren't always packaged with older debian, though
see <http://packages.debian.org/search?keywords=papi-tools>.
Another option could be to attempt to install the pmda, and check out
the papi.available.num_counters metric for non-zeroness.
Or, stretching the effort perhaps too far, qa/src could include a
minimal papi client binary - a stripped-down version of papi_avail,
that just gives us a papi-support smoke-test output.
- FChE
|