This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PPro patch for egcs
- To: hjl at lucon dot org
- Subject: Re: PPro patch for egcs
- From: john at feith dot com (John Wehle)
- Date: Wed, 3 Jun 1998 13:30:20 -0400
- Cc: law at cygnus dot com, egcs at cygnus dot com
> * config/i386/i386.c (output_fp_conditional_move): New function
> to output floating point conditional move.
> (output_int_conditional_move): New function to output integer
> conditional move.
>
> * config/i386/i386.md (movsicci+5, movhicc+5, movdicc+5): Call
> output_int_conditional_move () to output int conditional move.
> (movsfcc+5, movdfcc+5, movxfcc+5): Call
> output_fp_conditional_move () to output floating point
> conditional move.
I don't understand why you are checking for which_alternative == 3
in output_fp_conditional_move. The define_insn for the floating
point conditional moves should only result in which_alternative
equaling 0, 1, or 2. Any other value should cause an abort.
The special case handling of DImode in output_int_conditional_move
seems more complicated than the original code associated with the
define_insn for the DImode integer conditional move. What is the
advantage of creating a complicated function?
You might consider starting with the original code associated with
the define_insn if you feel compelled to create separate functions.
-- John
-------------------------------------------------------------------------
| Feith Systems | Voice: 1-215-646-8000 | Email: john@feith.com |
| John Wehle | Fax: 1-215-540-5495 | |
-------------------------------------------------------------------------