[PATCH] Fix aliasing bug introduced with September gcc2 merge

Franz Sirl Franz.Sirl-kernel@lauterbach.com
Sat Jan 1 17:40:00 GMT 2000


Hi,

while hunting a C++ testsuite failure with tstring.cc on powerpc-linux-gnu (why
the hell does gcc generate a different stack layout for -O1 and -O1
-fstrict-aliasing?), I stumbled over this typo. It was revealed by compiling
testsuite/blkarg.c with -O1 and using an --enable-checking=tree,gc,misc
compiled gcc, it ICE'd in alias.c/mems_in_disjoint_alias_sets_p().

The patch restores the pre-merge status.

OK to commit?

Franz.

	* expr.c (store_constructor_field): Fix typo introduced with last
	gcc2 merge.

Index: expr.c
===================================================================
RCS file: /cvs/gcc/egcs/gcc/expr.c,v
retrieving revision 1.190
diff -u -p -r1.190 expr.c
--- expr.c	1999/12/24 17:27:36	1.190
+++ expr.c	2000/01/02 01:24:34
@@ -4079,7 +4079,7 @@ store_constructor_field (target, bitsize
   else
     store_field (target, bitsize, bitpos, mode, exp, VOIDmode, 0, 
 		 (align + BITS_PER_UNIT - 1) / BITS_PER_UNIT,
-		 int_size_in_bytes (type), cleared);
+		 int_size_in_bytes (type), 0);
 }
 
 /* Store the value of constructor EXP into the rtx TARGET.


More information about the Gcc-patches mailing list