Re: [V2 15/23] test: dynamic arrays passed to subroutines.

[Keven Boell <keven dot boell at linux dot intel dot com>]

On 07.08.2014 08:58, Keven Boell wrote:
> On 01.08.2014 22:11, Jan Kratochvil wrote:
>> On Fri, 11 Jul 2014 11:21:27 +0200, Keven Boell wrote:
>>> +gdb_test "finish" \
>>> +  ".*foo\\\(sub_arr1\\\(5:10, 5:10\\\), sub_arr2\\\(10:15,10:15,10:15\\\)\\\)" \
>>> +  "finish function"
>>
>> This testcase FAILs for 32-bit targets (FAILs on both x86_64-m32 and on i686).
>>
>> finish^M
>> Run till exit from #0  foo (array1=..., array2=...) at gdb/testsuite/gdb.fortran/vla-sub.f90:31^M
>> 0x08048aa5 in vla_sub () at gdb/testsuite/gdb.fortran/vla-sub.f90:72^M
>> 72        call foo(sub_arr1, sub_arr2)^M
>> (gdb) FAIL: gdb.fortran/vla-value-sub-finish.exp: finish function
>>  8048aa0:       e8 e9 fb ff ff          call   804868e <foo_>
>>  8048aa5:       83 c4 08                add    $0x8,%esp
>> /home/jkratoch/redhat/gdb-vla-intel-noasan/gdb/testsuite/./gdb.fortran/vla-sub.f90:73
>>
>> finish^M
>> Run till exit from #0  foo (array1=..., array2=...) at gdb/testsuite/gdb.fortran/vla-sub.f90:31^M
>> vla_sub () at gdb/testsuite/gdb.fortran/vla-sub.f90:73^M
>> 73        call foo(sub_arr1(5:10, 5:10), sub_arr2(10:15,10:15,10:15))^M
>> (gdb) PASS: gdb.fortran/vla-value-sub-finish.exp: finish function
>>   400d34:       e8 23 fb ff ff          callq  40085c <foo_>
>> /home/jkratoch/redhat/gdb-vla-intel-noasan/gdb/testsuite/./gdb.fortran/vla-sub.f90:73
>>
>> It is understandable, "finish" command sometimes ends up on the caller source
>> line and sometimes on the next source line after caller, depending on the arch.
>>
>> At least this way it happens on:
>> 	gcc-4.9.1-2.fc21.1.x86_64
>>
> 
> I'll try to get this reproduced with gcc 4.9.x. I was using 4.8.

I was able to reproduce the fail on FC20 with gcc 4.9 and adapted the
regular expression accordingly.
You can find an update on https://github.com/intel-gdb/vla/tree/vla-fortran

> 
>>
>> Thanks,
>> Jan
>>
> 
> Thanks,
> Keven
> 

Keven