[rfc/patch][pr 22995] Swapped register order of "cmp" instructions

[Ahmed Bougacha]

On Fri, Mar 27, 2015 at 11:00 AM, Rafael Espíndola
<rafael.espindola at gmail.com> wrote:
> ping
>
> On 24 March 2015 at 12:34, Rafael Espíndola <rafael.espindola at gmail.com> wrote:
>> Now with the correct patch :-(
>>
>> On 24 March 2015 at 12:33, Rafael Espíndola <rafael.espindola at gmail.com> wrote:
>>> MC and GAS handle these registers in different order.
>>>
>>> Is there a reason to be different? If not, is something along the
>>> lines of the attached patch reasonable?

I can't say anything about reasons, but this doesn't seem like a
printing issue, rather one of default encoding?  There's 38/39 for "r/m,
r", and 3A/3B for "r, r/m":  I guess MC and GAS don't agree on that.

>>> Any suggestions for a better name for BinOpRR2? Should I just "inline"
>>> it in the single use?

That's what the _REV classes are for, no?  If you switch BaseOpc and
BaseOpc2 passed to *rr* in ArithBinOp_F, does that have the same
effect?

-Ahmed

>>> Cheers,
>>> Rafael
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>