Re: [Patch, fortran] PR40646 - ICE assigning array return value from type-bound procedure

[Richard Guenther <richard dot guenther at gmail dot com>]

On Thu, Jul 16, 2009 at 1:50 PM, Janus Weil<janus@gcc.gnu.org> wrote
> 2009/7/15 Richard Guenther <richard.guenther@gmail.com>:
>> On Mon, Jul 13, 2009 at 2:12 PM, Tobias Burnus<burnus@net-b.de> wrote:
>>> Janus Weil wrote:
>>>>> Ok, here is a rather complete patch, which (mostly) fixes the PPC part
>>>>> of the PR. Note that it fixes the attached versions of comment #2 and
>>>>> #4, but still has problems with certain variations of them.
>>>>>
>>>>
>>>> Attached is a follow-up patch, which fixes the remaining issues (see
>>>> comment #8 in the PR).
>>>>
>>>> Regression-tested on x86_64-unknown-linux-gnu. Ok for trunk?
>>>>
>>>
>>> OK. Thanks for the patch.
>>
>> This patch has broken build of SPEC 2006 481.wrf:
>>
>> /gcc/spec/sb-barbella.suse.de-head-64-2006/x86_64/install-hack/bin/gfortran
>> -c -o module_ra_sw.fppized.o -I. -I./netcdf/include ? -O3
>> -funroll-loops -fpeel-loops -ffast-math -march=native
>> -DSPEC_CPU_LP64 ? ? ? ? module_ra_sw.fppized.f90
>> module_ra_sw.fppized.f90:414.23:
>>
>> END MODULE module_ra_sw
>> ? ? ? ? ? ? ? ? ? ? ? 1
>> Internal Error at (1):
>> gfc_array_dimen_size(): Bad EXPR_ARRAY expr
>> specmake: *** [module_ra_sw.fppized.o] Error 1
>>
>> Maybe HJ can come up with a short testcase, if not I'll do it tomorrow
>> and file a bugreport.
>
> Any chance this is fixed by r149662? (cf. PR 40743, which was another
> regression introduced by my patch)

Yes, it seems to be fixed now.

Thanks,
Richard.

> Cheers,
> Janus
>