[Bug debug/69705] [5/6 Regression] segfault in libgomp.fortran/task1.f90 with -flto -fno-use-linker-plugin -fno-toplevel-reorder -O1 -g

["jakub at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69705

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 37785
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=37785&action=edit
gcc6-pr69705.patch

Is this really a regression?  It is a long standing LTO debug info problem,
which I hope will be fixed by early debug, but that hasn't made it for LTO yet.
If you look at the debug info, with LTO for inline functions one incorrectly
gets almost empty DW_TAG_subroutine for the inlined subroutines (free lang data
or what strips all of their DECL_INITIAL block tree), but instead all the vars
etc. duplicated in all the places where the function is inlined (instead of
just pointing to DW_AT_abstract_origin if say it is needed to add location info
for them).  The attached patch provides ugly workaround which just emits the
Fortran commons the same way as it emits other stuff.