This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: arm-mmx-1.c

> I see that arm-mmx-1.c is failing on the csl-arm-branch.
> It looks like this is a spurious failure: the test uses
> -mno-apcs-frame -mcpu=iwmmxt and it looks like you've changed things
> so that those options are incompatible.  If that combination is not
> supposed to work, would you please remove the test on both the
> csl-arm-branch and the mainline?  (Consider this preapproval.)
> Or, am I misunderstanding?

I believe the proper fix is to add -mabi=iwmmxt to the dg-options line. This 
used to be implied by -mcpu=iwmmxt.

Applied as follows.


2004-04-01  Paul Brook  <>

	* gcc.dg/arm-mmx-1.c: Add -mabi=iwmmxt to dg-options.

Index: gcc.dg/arm-mmx-1.c
RCS file: /var/cvsroot/gcc-cvs/gcc/gcc/testsuite/gcc.dg/arm-mmx-1.c,v
retrieving revision 1.2
diff -u -p -r1.2 arm-mmx-1.c
--- a/gcc.dg/arm-mmx-1.c	4 Feb 2004 13:43:43 -0000	1.2
+++ b/gcc.dg/arm-mmx-1.c	1 Apr 2004 00:34:45 -0000
@@ -1,7 +1,7 @@
 /* Verify that if IP is saved to ensure stack alignment, we don't load
    it into sp.  */
 /* { dg-do compile { target arm*-*-* strongarm*-*-* xscale*-*-*} } */
-/* { dg-options "-O -mno-apcs-frame -mcpu=iwmmxt" } */
+/* { dg-options "-O -mno-apcs-frame -mcpu=iwmmxt -mabi=iwmmxt" } */
 /* { dg-final { global compiler_flags; if ![string match "*-mthumb *" 
$compiler_flags] { scan-assembler "ldmfd\[ 	]sp!.*ip,\[ ]*pc" } } } */
 /* This function uses all the call-saved registers, namely r4, r5, r6,

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]