[llvm] r227603 - ARM: further correct .fpu directive handling

Saleem Abdulrasool compnerd at compnerd.org
Fri Jan 30 12:48:25 PST 2015

On Fri, Jan 30, 2015 at 12:36 PM, Renato Golin <renato.golin at linaro.org>

> On 30 January 2015 at 19:35, Saleem Abdulrasool <compnerd at compnerd.org>
> wrote:
> > +    {ARM::VFPV3, ARM::FeatureVFP2 | ARM::FeatureVFP3,
> > +     ARM::FeatureNEON | ARM::FeatureD16},
> Hi Saleem,
> That's wrong. vfpv3 == vfpv3-d32, that's why you have to specify
> vfpv3-d16 for the shortened variant.

Correct :-).

> The same goes for vfpv4, neon, etc. Your original patch was correct.

I hate the clang-format output here, its not very legible IMO.  The change
here actually is to the *disabled* field.  That is, if you did
-mfpu=vfpv3-d16 and then did .fpu vfpv3 ... you need to disable the D16
behavior.  I was hoping that the commit message was sufficient.  Should I
add a comment to that effect above the table?

> cheers,
> --renato

Saleem Abdulrasool
compnerd (at) compnerd (dot) org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150130/540e62aa/attachment.html>

More information about the llvm-commits mailing list