Re: PowerPC failures (Was: [PATCH] Add variadic support)

[David Gilbert <david.gilbert@linaro.org>]

On 23 February 2011 21:17, Anthony Green <green@redhat.com> wrote:
> "Dr. David Alan Gilbert" <david.gilbert@linaro.org> writes:
>
>>    2) I've tested it on i386 (1659 expected passes, 15 unsupported),
>>                         armel (1654 expected passes, 5 unexpecteed passes (cls_longdouble.c), 15 unsupported)
>>                         armhf (1654 expected passes, 5 unexpecteed passes (cls_longdouble.c), 15 unsupported)
>>                         s390x (1639 expected passes, 5 unexpected failures - err_bad_abi.c)
>>               Those cls_longdouble unexpected passes and err_bad_abi.c
>>               failure are there in the current head.
>>       powerpc64 seems to be broken on current head, but I tested this patch
>>       on powerpc64 on the .0.9 release and it was OK.
>
> I just tested the head on powerpc64-unknown-linux-gnu and had no
> problems.  You appear to be testing powerpc64-unknown-linux-gnu like me,
> but with a different gcc and binutils version.
>
> You wrote on the wiki...
>> Failure is src/powerpc/ffi.c:961 where cif->abi is expected to be
>> FFI_GCC_SYSV or FFI_SYSV but is getting FFI_LINUX
>
> Line 961 is only run when POWERPC64 isn't defined.
>
> POWERPC64 is defined in src/powerpc/ffitarget.h.   Could you please see
> why it's not being defined in your build environment?

It seems to be a 32 v 64 issue. It's got a 64bit kernel but by default
gcc is producing 32bit
binaries, the __powerpc64__ flag is only set if -m64 is passed.
However, config.guess is saying
it's a powerpc64-unknown-linux-gnu - and hence that's the directory
that was created.

So I guess it's actually a failure on powerpc32 not  64.


$ uname -a
Linux daves-g5 2.6.26-2-powerpc64 #1 SMP Thu Nov 25 15:01:22 UTC 2010
ppc64 GNU/Linux
$ cat /etc/issue
Debian GNU/Linux 5.0 \n \l
$ gcc -v
Using built-in specs.
Target: powerpc-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian
4.3.2-1.1' --with-bugurl=file:///usr/share/doc/gcc-4.3/README.Bugs
--enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr
--enable-shared --with-system-zlib --libexecdir=/usr/lib
--without-included-gettext --enable-threads=posix --enable-nls
--with-gxx-include-dir=/usr/include/c++/4.3 --program-suffix=-4.3
--enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc
--enable-mpfr --disable-softfloat --enable-secureplt
--enable-targets=powerpc-linux,powerpc64-linux --with-cpu=default32
--with-long-double-128 --enable-checking=release
--build=powerpc-linux-gnu --host=powerpc-linux-gnu
--target=powerpc-linux-gnu
Thread model: posix
gcc version 4.3.2 (Debian 4.3.2-1.1)

There is the opposite problem that if you pass CFLAGS=-m64 and/or run
with linux64 the make check fails because it doesn't build the tests
64bit:

Executing on host: gcc
/home/dg/libffi-3.0.10rc5/testsuite/libffi.call/closure_fn0.c  -O0 -W
-Wall  -I/home/dg/libffi-3.0.10rc5/powerpc64-unknown-linux-gnu/testsuite/../include
-I/home/dg/libffi-3.0.10rc5/testsuite/../include
-I/home/dg/libffi-3.0.10rc5/powerpc64-unknown-linux-gnu/testsuite/../include/..
-L/home/dg/libffi-3.0.10rc5/powerpc64-unknown-linux-gnu/testsuite/../.libs
 -lffi -lm   -o ./closure_fn0.exe    (timeout = 300)
/usr/bin/ld: skipping incompatible
/home/dg/libffi-3.0.10rc5/powerpc64-unknown-linux-gnu/testsuite/../.libs/libffi.so
when searching for -lffi

(here that libffi.so is 64bi)

Dave