[Bug libfortran/21787] New: libfortran - better error message needed

[Bug libfortran/21787] New: libfortran - better error message needed

[dir at lanl dot gov]

When I compile and try to run one of my programs with gfortran, all I get as
output from libfortran is "Abort" - not very informative -

 row  col          x           y                          xg          yg          zg
   1    1          0.0000      0.0000                     0.0000      0.0000   
  0.0000
   1   11          0.0000     50.0000                     0.0000     50.0000   
  0.0000
   2    1          1.0000      0.0000                     1.0000Abort
END OF END OF END OF END OF END OF  test1.in

-- 
           Summary: libfortran - better error message needed
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: libfortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dir at lanl dot gov
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: powerpc-apple-darwin7.9.0
  GCC host triplet: powerpc-apple-darwin7.9.0
GCC target triplet: powerpc-apple-darwin7.9.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-05-27 22:05 -------
Unless you can show us your code, the're not much we can do. Especially since
there is no occurence of "Abort" (case respected) in the libgfortran code!

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-27 22:08 -------
(In reply to comment #1)
> Unless you can show us your code, the're not much we can do. Especially since
> there is no occurence of "Abort" (case respected) in the libgfortran code!

Abort just comes from calling abort by the way. 

giving a backtrace might be a good idea here.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[dir at lanl dot gov]

------- Additional Comments From dir at lanl dot gov  2005-05-31 12:54 -------
It is 60,000 line program, including the 3 small libraries, that executes in two
steps - a bit long to post.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-06-01 12:48 -------
A few things you can do:

  1. Run this program inside gdb: Compile all files with "-g" flag, then run
"gdb ./myprogram" and enter the command "run"). When program aborts, signal
should be caught by gdb and you should have a prompt; enter the "where" command
and post here what it says.

  2. Trim it down if you have time.

  3. Send it to me by private mail, with instructions to build it. I'll be glad
to trim it down (and the code will not be posted publicly) and debug.


PS: sorry if point 1 sounds trivial to you. I just don't know if you are
familiar with gdb.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[fxcoudert at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fxcoudert at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[dir at lanl dot gov]

------- Additional Comments From dir at lanl dot gov  2005-06-02 13:31 -------
As gdb mostly does not work on the Macintosh (as I reported - Bug 20112) - I did
not think to try it. In this case it showed that abort does not dump the file io
buffers on the Macintosh - that is why there was no clue as to what the problem
was. Here is a simple example of that problem (nothing was written to fort.3) -


[dranta:~/tests/gfortran] dir% gfortran -o callabort callabort.f
[dranta:~/tests/gfortran] dir% callabort
Abort
[dranta:~/tests/gfortran] dir% cat callabort.f
      program test
      write(3,*)' error in program'
      call abort
      end
[dranta:~/tests/gfortran] dir% cat fort.3
[dranta:~/tests/gfortran] dir% 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-06-02 13:40 -------
Confirm (on i386-linux) at least that part of the bug: we don't flush I/O
buffers before sending SIGABRT, but we should.

As for the other part of the bug, you still have options 2 and 3.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-06-02 13:40:54
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] libfortran - better error message needed

[dir at lanl dot gov]

------- Additional Comments From dir at lanl dot gov  2005-06-02 14:50 -------
I knew that the program would not likely run with out hitting errors. It
compiles and runs correctly using several different FORTRANS including Absoft
and Lahey , but it needs options like "-fno-automatic -finit-local-zero
-fdefault-real-8 -fdefault-integer-8 -fdefault-double-8" to all work correctly.
Currectly, I do not think that gfortran has anything like the "-fno-automatic
-finit-local-zero" options working and there are still several problems with
"-fdefault-integer-8". None the less, I would be happy to pack it up and send it
(fxcoudert@gcc.gnu.org ?) with test problems and instruction for future testing.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] flush I/O before calling abort()

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-07-30 21:18 -------
For the bug noted in comments 5 and 6:

[dranta:~/tests/gfortran] dir% gfortran -o callabort callabort.f
[dranta:~/tests/gfortran] dir% callabort
Abort
[dranta:~/tests/gfortran] dir% cat callabort.f
      program test
      write(3,*)' error in program'
      call abort
      end
[dranta:~/tests/gfortran] dir% cat fort.3
[dranta:~/tests/gfortran] dir% 

A fix is easy to design: calling close_units() before abort() in file 
intrinsics/abort.c does the trick (no time to submit a real patch now, just 
noting this in case I forget about it).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |fxcoudert at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
           Keywords|                            |patch
   Last reconfirmed|2005-06-02 13:40:54         |2005-07-30 21:18:15
               date|                            |
            Summary|libfortran - better error   |flush I/O before calling
                   |message needed              |abort()


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] flush I/O before calling abort()

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-07-31 07:30 -------
Subject: Bug 21787

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	fxcoudert@gcc.gnu.org	2005-07-31 07:30:34

Modified files:
	libgfortran    : ChangeLog 
	libgfortran/intrinsics: abort.c 

Log message:
	PR libfortran/21787
	* intrinsics/abort.c (abort): Close units before aborting.
	Updated copyright years.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/ChangeLog.diff?cvsroot=gcc&r1=1.273&r2=1.274
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/libgfortran/intrinsics/abort.c.diff?cvsroot=gcc&r1=1.6&r2=1.7



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787

[Bug libfortran/21787] flush I/O before calling abort()

[fxcoudert at gcc dot gnu dot org]

------- Additional Comments From fxcoudert at gcc dot gnu dot org  2005-07-31 07:45 -------
This one is fixed (at least, the part concerning flushing units before calling
abort). For the other parts, we will file a new bug if we can produce a reduced
testcase.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21787