interaction with gprof

[Janus Weil]

> I vaguely remember problems with getting proper backtraces into
> internal subroutines (which occurred to me a couple of times, but
> never bothered me enough to take action). That's very probably
> related.

OTOH, at least a simple example seems to work with current gfortran
(see below), so either the backtrace problem has been fixed or I was
just fantasizing (or maybe the problem was with ifort?) ...

Cheers,
Janus



$ cat bt.f90
module m

contains

  subroutine s1
    call s2()
  contains
    subroutine s2
      call abort()
    end
  end

end

program p
use m
call s1()
end

$ gfortran-4.9 -g3 bt.f90
$ ./a.out

Program aborted. Backtrace:
#0  0x7F7A9CE947D7
#1  0x7F7A9CE95F32
#2  0x7F7A9CF61588
#3  0x4006CA in s2 at bt.f90:9
#4  0x4006D3 in __m_MOD_s1 at bt.f90:6
#5  0x4006DE in p at bt.f90:17
Aborted (core dumped)