[LLVMdev] How to pick default floating point ABI?

[Weitian Leung]

On 06/16/2015 09:10 PM, Daniel Sanders wrote:
>
> You forgot to include the compiler command but I believe I've spotted 
> the problem anyway. The problem is that the compiler isn't aware that 
> it needs to avoid using the odd-numbered single-precision registers on 
> Mips2 because (at the time) they weren't intended to be usable in this 
> way. The GNU assembler does know this and warns about it.
>
> You should be able to work around this with –mno-odd-spreg for now. 
> This option forbids the use of odd-numbered FPU registers for single 
> precision values. Could you file a bug report at llvm.org/bugs so that 
> we fix this at some point?
>
> //
> You forgot to include the compiler command 
Sorry it's my fault. Actually with nothing special flags, -std=c++11 
-mips2 -fPIC etc.

> You should be able to work around this with –mno-odd-spreg for now
Yes, the warning goes away.

> Could you file a bug report
Of course, I'll file the bug as soon as possible (I do need more testing 
anyway).

-- 
/Regards/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20150617/85ffaa64/attachment.html>