[Bug target/82005] [8 regression] early lto debug creates invalid assembly on Darwin

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82005

--- Comment #31 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Dominique d'Humieres from comment #30)
> > However I have restore my testing of the gfortran test suite with -g -flto
> > and I see the following new failures:
> >
> > FAIL: gfortran.dg/namelist_14.f90   -g -flto  (internal compiler error)
> > FAIL: gfortran.dg/namelist_14.f90   -g -flto  (test for excess errors)
> > FAIL: gfortran.dg/namelist_69.f90   -g -flto  (internal compiler error)
> > FAIL: gfortran.dg/namelist_69.f90   -g -flto  (test for excess errors)
> > FAIL: gfortran.dg/namelist_70.f90   -g -flto  (internal compiler error)
> > FAIL: gfortran.dg/namelist_70.f90   -g -flto  (test for excess errors)
> 
> This is caused by revision r251448 and "fixed" by restoring the gcc_assert
> in r251447.
> 
> I also see
> 
> FAIL: libgomp.fortran/taskloop3.f90   -g -flto  (internal compiler error)
> 
> which appeared between r251187 and r251447.

Ok, I can reproduce those when compiling with -flto -g0 and linking with
-flto -g which is what the patch in comment 26 ends up forcing (-g0 at
compile-time).  Let's split this out to a separate bug.  PR84645.

> IMO the patch in comment 26 should be committed.

Patch posted, waiting for approval.