This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: [PATCHv2][MIPS] Implement O32 ABI extensions (GCC)



> -----Original Message-----
> From: Matthew Fortune [mailto:Matthew.Fortune@imgtec.com]
> Sent: Tuesday, October 28, 2014 1:13 PM
> To: Moore, Catherine; 'gcc-patches@gcc.gnu.org' (gcc-
> patches@gcc.gnu.org); Eric Christopher (echristo@gmail.com)
> Cc: Richard Sandiford; Rich Fuhler; Rozycki, Maciej; Myers, Joseph
> Subject: RE: [PATCHv2][MIPS] Implement O32 ABI extensions (GCC)
> 
> Moore, Catherine <Catherine_Moore@mentor.com> writes:
> > Review comments are attached.  I will tackle the R6 patch next.
> > Thanks,
> > Catherine
> 
> Thanks Catherine. Depending on whether the fix to the following hunk
> needs discussion I'll make the changes and commit.
> 
> >> static bool
> >> mips_function_value_regno_p (const unsigned int regno) {
> >>   if (regno == GP_RETURN
> >>       || regno == FP_RETURN
> >>+      || regno == FP_RETURN + 2
> >>       || (LONG_DOUBLE_TYPE_SIZE == 128
> >> 	  && FP_RETURN != GP_RETURN
> >> 	  && regno == FP_RETURN + 2))
> >>     return true;
> >>
> >  Is this right? I'm  not following the intent of this modification.
> 
> The intent is to cover the complex return case where both FP_RETURN and
> FP_RETURN+2 are used to return the real and imaginary parts of _Complex
> float. The hunk which follows then deals with the _Complex double case
> where the corresponding odd numbered registers are also used for FP32.
> What I noticed when you first pointed me at this code on IRC was that the
> new condition leads to the pre-existing LONG_DOUBLE_TYPE_SIZE condition
> being redundant so should be removed I think. I just want to re-read how
> these functions are used to reassure myself this is all correct.

Okay, will you please add a comment here and for the next hunk?

> 
> I'll also check on whether o32 or O32 is the consistent name and change
> appropriately.
> 
> For the R6 patch. I will be rebasing and updating that patch following
> submission of FPXX but the only significant fix it needs relates to allowing the
> use of DSP with R6. That will affect the conditions on a few patterns and
> some of the expand code. A general read through the current patch on the
> list wouldn't be wasted though.
> 
Thanks,
Catherine


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]