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]

[PATH] Remove xfail in const-elim-1 for ARM, PA and AIX


While trying to see why certain tests needed XFAILing on LynxOS have
started passing now I found gcc.dg/const-elim-1.c.  The test used to
fail on certain platforms because the RTL optimizers were not able to
eliminate block moves.  The variables are no longer referenced after
tree-ssa dead code elimination so the copies are never emitted.

I checked recent ARM, HPPA and AIX testresults to check my theory and
they all XPASS this test:

Checked with seeing if test was still passing on

Is the patch below OK to remove the xfail?


2005-05-23  Adam Nemet  <> 
        * gcc.dg/const-elim-1.c: Remove xfails. 

Index: const-elim-1.c 
RCS file: /cvs/gcc/gcc/gcc/testsuite/gcc.dg/const-elim-1.c,v 
retrieving revision 1.12 
diff -u -p -r1.12 const-elim-1.c 
--- const-elim-1.c      10 May 2005 16:27:48 -0000      1.12 
+++ const-elim-1.c      23 May 2005 20:19:39 -0000 
@@ -7,13 +7,7 @@ 
 /* { dg-do compile } */ 
 /* { dg-options "-O2 -std=c99" } */ 
-/* This test fails on all processors where we use a block move to 
-   initialize "S" in test2.  The RTL optimizers are not clever enough 
-   to eliminate the block move, so the constant gets emitted. 
-   Currently known targets with this problem: all ARM; PA32 ("hppa*.*" 
-   matches "hppa2.0w" but not "hppa64"); PPC if string instructions 
-   are enabled (notably under AIX).  */ 
-/* { dg-final { scan-assembler-not "L\\\$?C\[^A-Z\]" { xfail arm-*-* strongarm\-*-* xscale-*-* hppa-*-* hppa*.*-*-* powerpc*-*-aix* } } } */ 
+/* { dg-final { scan-assembler-not "L\\\$?C\[^A-Z\]" } } */ 
 #define I (__extension__ 1.0iF) 

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