[llvm] r217693 - [AArch64] Enable post-RA MI scheduler.

[Chad Rosier]

Np!

On Saturday, September 13, 2014, Renato Golin <renato.golin at linaro.org>
wrote:

> Thanks Chad!
>
> On 13 September 2014 04:34, Chad Rosier <mcrosier at codeaurora.org
> <javascript:;>> wrote:
> > I had my VPN token on hand and was able to get to a machine to push the
> > patch.  Hopefully, the bots will be green after r217725.
> >
> >  Chad
> >
> >> All,
> >> If someone wouldn't mind verifying the patch and pushing the change,
> >> Sanjin and I would greatly appreciate it.  I won't have access to fix
> this
> >> until Sunday and I'm not sure Sanjin has commit access.
> >>
> >>  Thanks,
> >>   Chad
> >>
> >>> Hi Renato and Chad,
> >>>
> >>> Attached is the patch that modifies the test case to use regular
> >>> expressions
> >>> to check when the post-RA scheduler pass is being skipped, regardless
> of
> >>> the
> >>> post-RA scheduler implementation.  I tested this for both ARM (which
> >>> used
> >>> the post-RA top-down list latency scheduler) and AArch64 (which now
> uses
> >>> the
> >>> post-RA MI scheduler).
> >>>
> >>> Thanks,
> >>> Sanjin
> >>>
> >>> -----Original Message-----
> >>> From: llvm-commits-bounces at cs.uiuc.edu <javascript:;>
> >>> [mailto:llvm-commits-bounces at cs.uiuc.edu <javascript:;>] On Behalf Of
> Sanjin Sijaric
> >>> Sent: Friday, September 12, 2014 4:52 PM
> >>> To: 'Renato Golin'; 'Chad Rosier'
> >>> Cc: 'LLVM Commits'
> >>> Subject: RE: [llvm] r217693 - [AArch64] Enable post-RA MI scheduler.
> >>>
> >>> Hi Renato,
> >>>
> >>> When I did make check-all, it came clean (I see that that
> >>> Feature/optnone-llc.ll passes in my logs).   But this was because my
> >>> build
> >>> was a dual arm + aarch64 build that defaults to arm.  So the checks in
> >>> optnone-llc.ll were valid, as the default is the" post-RA top down
> >>> latency
> >>> scheduler" for the arm configuration.  Now that it's the post-RA MI
> >>> scheduler for the  aarch64 configuration, the aarch64-only build will
> >>> fail
> >>> since what shows up by default is "Skipping pass 'PostRA Machine
> >>> Instruction
> >>> Scheduler'", and not "Skipping pass 'Post RA top-down list latency
> >>> scheduler'".
> >>>
> >>> I'll see if I can change the test case to match the new behaviour.
> >>>
> >>> Sorry about this oversight.
> >>>
> >>> Sanjin
> >>>
> >>> -----Original Message-----
> >>> From: llvm-commits-bounces at cs.uiuc.edu <javascript:;>
> >>> [mailto:llvm-commits-bounces at cs.uiuc.edu <javascript:;>] On Behalf Of
> Renato Golin
> >>> Sent: Friday, September 12, 2014 3:32 PM
> >>> To: Chad Rosier
> >>> Cc: LLVM Commits
> >>> Subject: Re: [llvm] r217693 - [AArch64] Enable post-RA MI scheduler.
> >>>
> >>> On 12 September 2014 18:40, Chad Rosier <mcrosier at codeaurora.org
> <javascript:;>> wrote:
> >>>> Author: mcrosier
> >>>> Date: Fri Sep 12 12:40:39 2014
> >>>> New Revision: 217693
> >>>>
> >>>> URL: http://llvm.org/viewvc/llvm-project?rev=217693&view=rev
> >>>> Log:
> >>>> [AArch64] Enable post-RA MI scheduler.
> >>>
> >>> Hi Chad,
> >>>
> >>> Have you seen this?
> >>>
> >>> http://lab.llvm.org:8011/builders/llvm-aarch64-linux/builds/4106
> >>>
> >>> cheers,
> >>> --renato
> >>> _______________________________________________
> >>> llvm-commits mailing list
> >>> llvm-commits at cs.uiuc.edu <javascript:;>
> >>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> >>>
> >>> _______________________________________________
> >>> llvm-commits mailing list
> >>> llvm-commits at cs.uiuc.edu <javascript:;>
> >>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> >>>
> >>
> >>
> >>
> >
> >
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu <javascript:;>
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140913/c6fc930b/attachment.html>