[Patch, fortran] PR59198 - [4.8/4.9/5 Regression] ICE on cyclically dependent polymorphic types

[Paul Richard Thomas <paul dot richard dot thomas at gmail dot com>]

Dear All,

As will be apparent from the PR, I have spent a silly amount of time
on this one :-(  Once I became 'de-obsessed' with the fact that the
reduced testcase worked, when 'rng' was made a pointer and
concentrated on the procedure pointer component 'obs1_int', finding
the problem was rather more straightforward, even if not 'obvious'.

The ChangeLog says it all. If the check is not done for components
that are not procedure pointers,
typebound_operator_9.f03 breaks. I am not entirely sure why this is
the case but the fix works fine.

Bootstraps and regtests on FC21/x86_64 - OK for 4.8, 4.9 and 5.0?

Paul

2014-03-15  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/59198
    * trans-types.c (gfc_get_derived_type): If an abstract derived
    type with procedure pointer components has no other type of
    component, return the backend_decl directly. Otherwise build
    the components.

2014-03-15  Paul Thomas  <pault@gcc.gnu.org>

    PR fortran/59198
    * gfortran.dg/proc_ptr_comp_44.f90 : New test
    * gfortran.dg/proc_ptr_comp_45.f90 : New test

Attachment: submit.diff
Description: Text document