[Bug fortran/101304] New: Bind(C): CONTIGUOUS attribute not handled correctly in Fortran routines called from C with discontiguous argument

[Bug fortran/101304] New: Bind(C): CONTIGUOUS attribute not handled correctly in Fortran routines called from C with discontiguous argument

[sandra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101304

            Bug ID: 101304
           Summary: Bind(C): CONTIGUOUS attribute not handled correctly in
                    Fortran routines called from C with discontiguous
                    argument
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: sandra at gcc dot gnu.org
  Target Milestone: ---

Section 18.3.6 of the Fortran 2018 standard says:

"When an interoperable Fortran procedure that is invoked from C has a dummy
argument with the CONTIGUOUS attribute or that is an assumed-length CHARACTER
explicit-shape or assumed-size array, and the actual argument is the address of
a C descriptor for a discontiguous object, the Fortran processor shall handle
the
difference in contiguity."

This is not working properly.  The bug is exposed by test cases
interoperability/contiguous-2.f90 and interoperability/contiguous-3.f90 from
the WIP TS29113 testsuite:

https://gcc.gnu.org/pipermail/gcc-patches/2021-July/574115.html

It looks like ftest1 is getting an array that starts at the correct offset of
the discontiguous section but has lost the non-unit sm from the base array.

Calling ftest1 and ftest2 directly from Fortran with a discontiguous array
section works correctly.

[Bug fortran/101304] Bind(C): CONTIGUOUS attribute not handled correctly in Fortran routines called from C with discontiguous argument

[sandra at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101304

sandra at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #1 from sandra at gcc dot gnu.org ---
Although it wasn't explicitly tagged with this issue, Tobias's recent big patch
to re-implement CFI <-> C descriptor conversions (commit
64f9623765da3306b0ab6a47997dc5d62c2ea261) fixed the testcases mentioned here
and removed the XFAILs on them, so I am marking this fixed.