Re: [PATCH] fortran: Fix creation of reference to C_FUNLOC argument [PR117303]

Hi Kirill,

On 2/2/26 16:05, Kirill Chilikin wrote:
> 
> Dear maintainers,
> 
> The attached patch fixes PR 117303. The reference returned by C_FUNLOC
> is assigned to a variable in the added code. Without that,
> no reference from the calling subprogram to the argument of C_FUNLOC
> was created in the call graph, resulting in an undefined-reference error
> with link-time optimization.
> 
> The patch was tested using with "check-fortran" on a x86_64-pc-linux-gnu
> system. One existing test from gfortran.dg/c_funloc_test_7.f90 failed
> as expected due to code-generation change. The generated code was
> 
>     cfp = (void (*<T6d>) (void)) nocsub;
> 
> but now it became
> 
>       void * D.4685;
>       D.4685 = (void *) nocsub;
>       cfp = (void (*<T6f>) (void)) D.4685;
> 
> The test has been modified in accordance with the changes.
> Another test, gfortran.dg/c_funloc_test_9.f90, has been converted from
> the test case in PR.
> 
>      PR fortran/117303
> 
> gcc/fortran/ChangeLog:
> 
>      * trans-intrinsic.cc (conv_isocbinding_function):
>           Assign the reference returned by C_FUNLOC to a variable.
> 
> gcc/testsuite/ChangeLog:
> 
>      * gfortran.dg/c_funloc_tests_7.f90:
>           Updated test due to changed code generation.
>      * gfortran.dg/c_funloc_tests_9.f90: New test.
> 
> Signed-off-by: Kirill Chilikin <[email protected]>
> ---
> 
> Please see the attached file for the actual patch.
> If accepted, please commit the change as I do not have write access.
> 
> Regards,
> Kirill

this looks good to me, but I will leave 24 h for others to have a
look or comment on it, as I am not too familiar with LTO subtleties.
Will commit then unless someone else beats me to it.

Thanks,
Harald