http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45989
Dominique d'Humieres <dominiq at lps dot ens.fr> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |dave.ang...@nrc-cnrc.gc.ca,
| |iains at gcc dot gnu.org,
| |pthaugen at us dot ibm.com
--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2010-12-21
18:20:40 UTC ---
> Some of the objc.dg-struct-layout-encoding-1 test currently XPASS on several
> platforms:
>
> * i386-pc-solaris2.*, 32 and 64-bit:
>
> XPASS: objc.dg-struct-layout-encoding-1/t026_main.m execution test
This is the case for all the platforms I have looked at.
> * sparc-sun-solaris2.*, 32 and 64-bit, alpha-dec-osf5.1b, mips-sgi-irix6.5:
>
> XPASS: objc.dg-struct-layout-encoding-1/t025_main.m execution test
> XPASS: objc.dg-struct-layout-encoding-1/t026_main.m execution test
> XPASS: objc.dg-struct-layout-encoding-1/t027_main.m execution test
> XPASS: objc.dg-struct-layout-encoding-1/t028_main.m execution test
> XPASS: objc.dg-struct-layout-encoding-1/t029_main.m execution test
> XPASS: objc.dg-struct-layout-encoding-1/t030_main.m execution test
> XPASS: objc.dg-struct-layout-encoding-1/t031_main.m execution test
This is also the case for
hppa2.0w-hp-hpux11.11 (default mode), see
http://gcc.gnu.org/ml/gcc-testresults/2010-12/msg01891.html
powerpc64-unknown-linux-gnu (-m32/65) see
http://gcc.gnu.org/ml/gcc-testresults/2010-12/msg01488.html
> This just adds noise to the testsuite reports, so the xfails need to be
> adjusted.
Since I shared this feeling, I have the following darwin-centric patch in my
trees:
---
../../_gcc_clean/gcc/testsuite/objc.dg/gnu-encoding/struct-layout-encoding-1_generate.c
2010-12-13 10:50:10.000000000 +0100
+++
../../gcc-4.6-work/gcc/testsuite/objc.dg/gnu-encoding/struct-layout-encoding-1_generate.c
2010-12-18 12:14:33.000000000 +0100
@@ -243,10 +243,16 @@ switchfiles (int fields)
of bugs in libobjc and the objc front-end. 25 is because
vectors are not encoded. The rest are because or zero sized
arrays are encoded as pointers. */
- else if (filecnt >= 25)
+ else if (filecnt == 25 || (filecnt >= 27 && filecnt <= 29))
{
fprintf (outfile, "\
-/* { dg-do run { xfail *-*-* } } */\n\
+/* { dg-do run { xfail { { ! powerpc-apple-darwin9 } || ilp32 } } } */\n\
+/* { dg-options \"-w -I%s -fgnu-runtime\" } */\n", srcdir);
+ }
+ else if (filecnt >= 30)
+ {
+ fprintf (outfile, "\
+/* { dg-do run { xfail { ! powerpc-apple-darwin9 } } } */\n\
/* { dg-options \"-w -I%s -fgnu-runtime\" } */\n", srcdir);
}
else
I have no way to test an extension to non-darwin platforms nor commit rights,
but it could be a starting point. It would also be nice to have a pointer to
"bugs in libobjc and the objc front-end" in the FIXME (if there is any
available).
