[Patch, Fortran] PR 38536 ICE with C_LOC on consecutive array and substring references.

[Mikael Morin]

Daniel Kraft wrote:
> Mikael Morin wrote:
>> Mikael Morin wrote:
>>> Hello,
>>>
>>> this fixes part of PR 38536, where the following was leading to an ICE:
>>>
>>> f_ptr = C_LOC(chrScalar%c(1)(1:1))
>>>
>>>
>>> The code checking for pointerness of the argument (which is where the
>>> ICE occurs) has been "rewritten" using gfc_is_data_pointer.
>>>
>>> One of the testcase is that of the PR.
>>> The other one is a newly accepted program (and valid I think).
>>>
>>> regression-tested on x86-64-unknown-linux-gnu.
>>>
>>> OK for trunk (and later 4.3?), or should I wait for 4.5?
> 
> Ok, I think trunk and backporting after some wait should be ok.
> 
> Daniel
> 
Committed as revision 143050.