[PATCH] D41693: [ARM][NFC] Avoid recreating MCSubtargetInfo in ARMAsmBackend

[David Blaikie via llvm-commits]

[+echristo who did the per-function-subtarget work & might have some
suggestions on how this might be tested/testable or not]

On Tue, Jan 9, 2018 at 6:15 AM Alex Bradbury <asb at lowrisc.org> wrote:

> On 8 January 2018 at 17:10, David Blaikie via llvm-commits
> <llvm-commits at lists.llvm.org> wrote:
> > No test coverage?
> >
> > I'm assuming creating the MCSubtargetInfo with the function's feature
> string
> > (rather than an empty string) would fix bugs where function-specific
> feature
> > strings were not respected? Perhaps an example of that could be tested?
>
> That could make sense. The change is motivated primarily by code
> cleanup rather than addressing any known bug. I would think observable
> differences in behaviour could be observed in tests like
> t2-modified-immediate-fixup-error{1,2}.s, but I can't see an
> appropriate baseline -march to use.  Suggestions from those of you
> more familiar with the Arm product line would be welcome. Any ideas
> anyone?
>
> Thanks,
>
> Alex
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180109/62b02a41/attachment.html>