Patches for HP-UX support on IA64
Richard Henderson
rth@redhat.com
Thu Jan 11 17:07:00 GMT 2001
On Thu, Jan 11, 2001 at 04:23:41PM -0800, Steve Ellcey wrote:
> | #ifdef HAVE_extend_ptr
> | if (unsignedp < 0)
> | return CODE_FOR_extend_ptr;
> | else
> | #endif
This makes much more sense.
> Doing this would of course imply an extension to how
> POINTERS_EXTEND_UNSIGNED is interpreted, now it is just used as a
> zero/non-zero number but I saw that in the code the value of unsignedp
> is always hardcoded to 0 or 1 for regular conversions or to
> POINTERS_EXTEND_UNSIGNED for address conversions.
I'd support something like this if we rename POINTERS_EXTEND_UNSIGNED
so that it doesn't imply signed vs unsigned. And perhaps use an
enumeration that includes a "don't extend" value that can be used
to provide a default in some instances to avoid ifdefs.
It would also be nice to take this opportunity to clean things up a
bit so that there are fewer POINTERS_EXTEND_UNSIGNED ifdefs. Perhaps
by having convert_memory_address always exist.
r~
More information about the Gcc-patches
mailing list