This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Bugs in MIPS FUNCTION_ARG_REGNO_P ?
Jim Wilson <wilson@specifixinc.com> writes:
> On Tue, 2004-03-02 at 23:01, Eric Christopher wrote:
>> > && ((N) % FP_INC == 0) && mips_abi != ABI_O64)) \
>
> I see that the introduction of FP_INC made the TARGET_FLOAT64 check
> unnecessary.
>
> So the problem is only that excluding ABI_O64 does not make sense here.
>
> The patch is here:
> http://gcc.gnu.org/ml/gcc-patches/2002-04/msg01325.html
> Unfortunately, this fails to give any justification for the change, so
> there is no easy way to tell why it was checked in now.
>
> I suggest reverting it, waiting to see what breaks, and then properly
> documenting the problem that the patch supposedly fixes.
Gets my vote too. I see no reason to test N % FP_INC or mips_abi here.
When using paired 32-bit registers, if a double argument is passed
in $f12, $f13 is used as well.
Richard