This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Patch to fix x86 output_move_double corrupting cc0
- To: egcs-patches at egcs dot cygnus dot com
- Subject: Patch to fix x86 output_move_double corrupting cc0
- From: John Wehle <john at feith dot com>
- Date: Tue, 13 Apr 1999 01:07:37 -0400 (EDT)
The x86 port has the naive belief that register to register and
register to memory moves don't modify cc0. Unfortunately the port
has a habit of using add, dec, inc, sub, etc. when generating the
actual move. One such function that does this is output_move_double
if the operand is a non-offsettable memory operand.
ChangeLog:
Mon Apr 12 22:29:56 EDT 1999 John Wehle (john@feith.com)
* i386.c (output_move_double): Assume that complex moves
clobber cc0.
Enjoy!
-- John Wehle
------------------8<------------------------8<------------------------
*** gcc/config/i386/i386.c.ORIGINAL Mon Apr 12 22:09:28 1999
--- gcc/config/i386/i386.c Mon Apr 12 22:31:31 1999
*************** output_move_double (operands)
*** 1082,1087 ****
--- 1082,1092 ----
if (optype0 == RNDOP || optype1 == RNDOP)
abort ();
+ /* Assume that complex moves clobber cc0. */
+ if ( !(optype0 == REGOP || optype0 == OFFSOP)
+ || !(optype1 == REGOP || optype1 == CNSTOP || optype1 == OFFSOP))
+ CC_STATUS_INIT;
+
/* If one operand is decrementing and one is incrementing
decrement the former register explicitly
and change that operand into ordinary indexing. */
-------------------------------------------------------------------------
| Feith Systems | Voice: 1-215-646-8000 | Email: john@feith.com |
| John Wehle | Fax: 1-215-540-5495 | |
-------------------------------------------------------------------------