This is the mail archive of the
mailing list for the GCC project.
Re: apply_result_size vs FUNCTION_VALUE_REGNO_P
On Thu, 2005-04-07 at 15:51, David Edelsohn wrote:
> >>>>> Richard Earnshaw writes:
> >> I think the change breaks __builtin_apply/__builtin_return on every platform
> >> that can return values in several contiguous registers, because the entry for
> >> FUNCTION_VALUE_REGNO_P in the manual reads:
> Richard> That was my thoughts too. You could take a look at how I fixed it on
> Richard> ARM.
> Why should every target need to fix this individually?
Maybe an even more important question is 'do we want to fix this at
__builtin_apply and friends is, to the best of my knowledge, only used
in libobjc. And the relevant bit libobjc is supposed to be being
rewritten to use libffi.
Notwithstanding that, I did it that way on ARM because I needed to be
able to use special modeless load/store instructions on the FPA to load
and store the floating point result.