[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2015-05-04
                 CC|                            |burnus at gcc dot gnu.org,
                   |                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |5.2
            Summary|[5.1 Regression] gfortran   |[5 Regression] gfortran ICE
                   |ICE with -dH                |with -dH
     Ever confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Most likely started with r217600.

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
The following test (from pr58334)

program foo
implicit none
character(len=16) :: a,b,c
a="XXX"
b="&
  &XXX"
c="XXX &
  & XXX"
write(0,*) 'a=',a,' b=',b,' c=',c 
endprogram foo

shows the same behavior.

BTW what -dH is supposed to do (I did not find it in the manual)?

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Though, looking at the implementation and documentation of -dH, it sounds like
it is intentional.  -dH is supposed to stand for abort the compiler after
issuing an error.  That is not an internal compiler error, but just abort
requested by the user.  So NOTABUG IMHO.

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[Melven.Roehrig-Zoellner at DLR dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

Melven.Roehrig-Zoellner at DLR dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |minor

--- Comment #4 from Melven.Roehrig-Zoellner at DLR dot de ---
> man gcc | grep -- -dH
           -dH Produce a core dump whenever an error occurs.
(This refers to a *compiler error* - so I should only need/use it to report a
gcc bug.)

But this flag shouldn't "trigger" an error when there is none without it or am
I missing something here?

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I don't have mpi installed, so it is hard to find out if you get an error or
not - your testcase is clearly not self-contained; I certainly do get an error
and correct abort after that.

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

--- Comment #6 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> I don't have mpi installed, so it is hard to find out if you get an error
> or not - your testcase is clearly not self-contained; I certainly do get
> an error and correct abort after that.

Self contained test in comment 2.

[Bug fortran/65996] [5 Regression] gfortran ICE with -dH

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
So supposedly gfc_error needs to for buffered_p case also save and temporarily
clear global_dc->abort_on_error.  Maybe various other changes needed for that.

Wonder what is -dH good for, don't really see when it would be really useful.

[Bug fortran/65996] [5/6 Regression] gfortran ICE with -dH

[Melven.Roehrig-Zoellner at DLR dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

--- Comment #8 from Melven.Roehrig-Zoellner at DLR dot de ---
Thanks for the quick replies and the discussion about the "not-so-usefulness"
of the -dH switch (at least for our project).
We most probably used this switch by mistake - you might possibly improve the
documentation by adding
"-dH Produce a core dump whenever a *compilation* error occurs."
to prevent further misunderstandings for people who don't read the full
paragraph.

[Bug fortran/65996] [5/6 Regression] gfortran ICE with -dH

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|5.2                         |5.3

--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 5.2 is being released, adjusting target milestone to 5.3.

[Bug fortran/65996] [5/6 Regression] gfortran ICE with -dH

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65996

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4