[PATCH] ARM: optimizeSelect has to consider the previous register class Was: [PATCH] improve arm machine code (make -verify-machineinstrs happier)

[Renato Golin]

LGTM

Cheers,
Renato
On 4 Oct 2013 01:26, "Matthias Braun" <mbraun at apple.com> wrote:

> A revised patch with testcase is attached.
>
> Description: optimizeSelect sometimes folded some copies without
> respecting the previous operands register class.
>
> Greetings
> Matthias
>
>
>
> On Oct 3, 2013, at 10:49 AM, Quentin Colombet <qcolombet at apple.com> wrote:
>
> Hi Matthias,
>
> I haven’t looked into the details of the patches but I have a general
> comment:
> Could you add a test case for each case you are fixing (possibly in the
> same test file)?
>
> Thanks for working on that.
>
> Cheers,
> -Quentin
>
> On Oct 3, 2013, at 10:41 AM, Matthias Braun <mbraun at apple.com> wrote:
>
> The attached 3 patches improve some cases in the ARM backend where bad
> code (in the sense of -verify-machineinstrs) was produced. Please review,
> thanks.
>
> Greetings,
> Matthias
> <ARM-do-not-add-a-regmask-for-TAILJUMPs.patch>
> <ARM-optimizeSelect-has-to-consider-the-previous-regi.patch>
> <ARM-preserv-undef-flag-in-pseudo-instruction-expande.patch>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> 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/20131004/6583e209/attachment.html>