This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: buildfailure in expr.c on darwin
- From: Zack Weinberg <zack at codesourcery dot com>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: gcc at gcc dot gnu dot org
- Date: Wed, 21 Nov 2001 22:25:02 -0800
- Subject: Re: buildfailure in expr.c on darwin
- References: <200111220603.BAA26636@makai.watson.ibm.com>
On Thu, Nov 22, 2001 at 01:03:45AM -0500, David Edelsohn wrote:
>
> Why are you using AGGREGATE_TYPE_P instead of the default BLKmode?
To match what happens when DEFAULT_PCC_STRUCT_RETURN is on. At least
in theory, with the present compiler, AGGREGATE_TYPE_P (x) implies
TYPE_MODE (x) == BLKmode. In any case, this change should not have
affected Darwin at all, since ABI_DARWIN is its own thing.
Based on the error message, I strongly suspect this will fix the bug.
My tree is tangled up with other changes right now, so if it works,
please apply it for me.
zw
* rs6000.c (output_cbranch): Use $, not %$, in output template.
===================================================================
Index: config/rs6000/rs6000.c
--- config/rs6000/rs6000.c 2001/11/21 02:57:14 1.240
+++ config/rs6000/rs6000.c 2001/11/22 06:24:06
@@ -5681,7 +5681,7 @@ output_cbranch (op, label, reversed, ins
/* If the branch distance was too far, we may have to use an
unconditional branch to go the distance. */
if (need_longbranch)
- s += sprintf (s, ",%c$+8\n\tb %s", '%', label);
+ s += sprintf (s, ",$+8\n\tb %s", label);
else
s += sprintf (s, ",%s", label);
}