[llvm-commits] [llvm] r73919 - in /llvm/trunk: include/llvm/ lib/Target/ARM/ lib/Target/ARM/AsmPrinter/ test/CodeGen/ARM/

[Bob Wilson]

Sandeep,

Darwin does not use the AAPCS-VFP ABI so this is a fairly low priority  
for me.  I filed a bug report (4463) to keep track of it.  It should  
be straightforward to implement if you have time to do it.  I don't  
know if it will work to refer to the "Q" registers in CCAssignToReg or  
not.  I had been thinking that we would need to define a custom  
function for this, but maybe that's not necessary for this.

On Jun 22, 2009, at 6:47 PM, Bob Wilson wrote:

> Yes, I hadn't gotten to that yet. I wanted to get this change
> committed anyway so that it will be possible to collaborate on things
> like that.
>
> If you have time to review the calling convention changes, I would
> appreciate it. I had to rework things a bit to handle vector arguments
> and return values.
>
> Sent from my iPhone
>
> On Jun 22, 2009, at 5:15 PM, Sandeep Patel <deeppatel1987 at gmail.com>
> wrote:
>
>> This is excellent!
>>
>> At first glance, I notice that the AAPCS-VFP case is wrong in
>> ARMCallingConv.td. It needs a CCIfType<[v2f64], CCAssignToReg<[Q0,
>> Q1, Q2, Q3]>> in there for both calls and rets.
>>
>> deep
>>
>> _______________________________________________
>> 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