[PATCH, committed] Fix PR 40558, undo part of June 25th powerpc submission
Michael Meissner
meissner@linux.vnet.ibm.com
Wed Jul 1 22:36:00 GMT 2009
I'm committing this patch as obvious to fix the bootstrap problem for 40558.
Thanks to Andrew for figuring out what the place where the regression occurred,
and also for pointing out I had missed some of the ChangeLog entries. Sorry
about that.
2009-07-01 Michael Meissner <meissner@linux.vnet.ibm.com>
PR bootstrap/40558
* config/rs6000/rs6000.c (print_operand): Undo change that breaks
darwin9 for printing reg addresses with %y.
2009-06-25 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/rs6000.c (print_operand): Correct lossage message
for %c error. Add %x support to print VSX registers as a unified
register set, instead of separate float and altivec registers.
Switch to use VECTOR_MEM_ALTIVEC_P instead of TARGET_ALTIVEC for
%y case, and add support for VSX pre-modify addresses.
(output_toc): Add assert for CONST containing an integer constant
in the PLUS case.
(rs6000_adjust_cost): Add POWER7 support.
(insn_must_be_first_in_group): Ditto.
(insn_must_be_last_in_group): Ditto.
(rs6000_emit_popcount): Ditto.
(rs6000_vector_mode_supported_p): Ditto.
Index: gcc/config/rs6000/rs6000.c
===================================================================
--- gcc/config/rs6000/rs6000.c (revision 149150)
+++ gcc/config/rs6000/rs6000.c (working copy)
@@ -13471,7 +13471,7 @@ print_operand (FILE *file, rtx x, int co
&& GET_CODE (tmp) == PRE_MODIFY)
tmp = XEXP (tmp, 1);
if (GET_CODE (tmp) == REG)
- fprintf (file, "%s,%s", reg_names[0], reg_names[REGNO (tmp)]);
+ fprintf (file, "0,%s", reg_names[REGNO (tmp)]);
else
{
if (!GET_CODE (tmp) == PLUS
--
Michael Meissner, IBM
4 Technology Place Drive, MS 2203A, Westford, MA, 01886, USA
meissner@linux.vnet.ibm.com
More information about the Gcc-patches
mailing list