This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Intrinsics for PowerPC
On Dec 22, 2003, at 5:54 PM, Ziemowit Laski wrote:
On 22 Dec 2003, at 17.24, Syd Polk wrote:
Bootstrapped on Darwin; OK to check in?
Generally looks OK, but I have a couple of comments, recently having gone thru the rs6000.c baptism of fire myself:
1) Instead of creating a new .md file, perhaps you can add 'eieio' to rs6000.md? (I'm a bit surprised it is not there already, actually...) For most of the remaining intrinsics, I would expect/hope that the corresponding instructions are already in rs6000.md.
The only problem I have with that is adding 50+ more descriptions to this file. It seemed cleaner to put it in a separate file, but I can be convinced.
2) To define the function signature for 'eieio', you may want to use what's done elsewhere in the code:
tree void_ftype_void
= build_function_type (void_type_node, void_list_node);
as I'm not sure if 'build_function_type_list' will do what you want...
Aside from that (and Andrew's scheduler comments which went over my head :-) ), I see no reason why these can't be checked in after a bootstrap, but where? :-) :-) If you find out, please let me know, and maybe I'll check in my AltiVec work there as well...
I am asking for approval in the TOT 3.4.
Syd Polk
Apple Computer
Technology EPM, Mac OS X Development Tools
+1 408 974-0577