[patch, fortran] Fix PR 85781, ICE on valid

[Tobias Burnus]

Hi Thomas,

On 1/19/20 7:21 PM, Thomas König wrote:
> the attached patch fixes an ICE which could occur for empty
> substrings (see test case).

I think one should rather fix the following issue. – While on x86-64 it 
does not seem to cause problems, it might for other platforms. 
Additionally, it is a missed optimization to use "A123" instead of "66" 
in the call.

Namely, gfortran generates a different code when using kind=c_type 
instead of kind=1 – although the kind value is the same! (Without 
BIND(C) or without VALUE, the generated code is the same.) The dump is:

two (character(kind=1)[1:1] x) vs.  three (character(kind=1) x)

Expected is the latter. See attached test case.

Tobias

-------------- next part --------------
A non-text attachment was scrubbed...
Name: bind_c_char.f90
Type: text/x-fortran
Size: 1758 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20200120/3c09b316/attachment.bin>