[llvm-commits] [PATCH 4/4] Profile metadata: test-suite Makefiles

Daniel Dunbar daniel at zuster.org
Wed Oct 31 10:50:09 PDT 2012


Hi Alastair,

On Tue, Oct 30, 2012 at 1:07 PM, Alastair Murray <alastairmurray42 at gmail.com
> wrote:

> Hi Manman, all,
>
> Daniel: I have also CC'ed you on this as you seem to do most of the
> test-suite work.  If you have no interest in these Makefiles or you feel
> they are not suitable for the repository then please let me know.


The test-suite portion looks fine to me. From my perspective, anyone is
free to add new TEST= flavors to the test-suite if they think they might be
useful to other people.

I defer to Manman's comments on whether or not the patches actually work.
Note that when she is referring to SPEC, she means as integrated into the
test-suite. The SPEC portion has some particularities so just because
something works for the rest of the test-suite doesn't guarantee it will
work with SPEC. However, I didn't actually try to run with these patches so
I can't comment on any particular issues.

 - Daniel


>
> On 07/09/12 20:54, Manman Ren wrote:
>
>> Alastair, I looked at your test-suite patch for PGO, it needs to be
>> modified to make it work for SPEC (Makefile.spec).
>> The modifications are attached.
>>
>
> Manman: I have included your SPEC changes in the patch.  I don't have
> access to SPEC so I can't test them, but I have visually verified that your
> previous patch is still compatible with recent test-suite changes.
>
>
>
>
> > I haven't found time to make it work for both SPEC and other benchmarks.
>
> Manman, I don't understand this statement.  The patch as attached works
> for test-suite, I assume that you have tested it on SPEC.  So it works for
> both ...
>
> Otherwise the only change from the previous patch is to switch uses of
> $(RUNTOOLSAFELY) over to $(RUNSAFELYLOCAL) to match recent test-suite
> changes.
>
> This patch is still dependent on the as-yet unreviewed "[PATCH 3/4]
> Profile metadata: runtime support" patch.  It could be trivially rewritten
> to not depend on that patch, but make -jN would not work for N>1.
>
> Regards,
> Alastair.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20121031/d5217f21/attachment.html>


More information about the llvm-commits mailing list