|
|
On Wed, 2005-08-24 at 14:28, Daniel Jacobowitz wrote:
> > By the way, is anybody trying to cleanup need_64bit_hwint? It's less
> > than ideal to have it in both libcpp and gcc and maintain it manually.
> > (But then I don't know if it's worth messing with the top-level
> > configure bits just for libcpp and gcc.)
>
> IMO we want to get rid of H_W_I from that interface, but I'm not sure
> how to do it without slowing down 32-bit HWI targets...
We should probably look at implementing an attribute that is similar to
ARMCC's 'return_in_regs' return type qualifier. However, I'm reluctant
to call it that since the details of that implementation aren't
especially pleasant.
R.
|
|