[Bug fortran/35627] New: namelist read error

[Bug fortran/35627] New: namelist read error

[pletzer at txcorp dot com]

Hi: The following program fails to read the namelist file below.

test program:
program test
  implicit none
    LOGICAL :: nlco(200)  ! (1:nbeam)
    REAL*8 :: xlbtna(200)  ! (1:nbeam)
  NAMELIST/nbdrive_naml/ nlco,xlbtna
    INTEGER :: nbshapa(200)  ! (1:nbeam)
  NAMELIST/nbdrive_naml/ nbshapa
  open(10, file='t.nml')
  read(10, nbdrive_naml)
  write(*,nbdrive_naml)
  close(10)
end program test

file t.nml:
&nbdrive_naml
nlco = 4*.TRUE.
xlbtna = 802.8, 802.8, 802.8, 802.8
nbshapa = 4*1
/

error message:
[pletzer@quartic namelist2]$ /contrib/gcc-4.3/bin/gfortran t.f90
[pletzer@quartic namelist2]$ ./a.out
At line 9 of file t.f90 (unit = 10, file = 't.nml')
Fortran runtime error: Bad data for namelist object xlbtna

Thanks for your help.


-- 
           Summary: namelist read error
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pletzer at txcorp dot com
 GCC build triplet: GNU Fortran (GCC) 4.3.0
  GCC host triplet: Linux quartic.txcorp.com 2.6.19-1.2288.fc5
GCC target triplet: x86_64 x86_64 x86_64 GNU/Linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[burnus at gcc dot gnu dot org]

------- Comment #1 from burnus at gcc dot gnu dot org  2008-03-18 21:50 -------
Confirmed. Again a GCC 4.3.0/4.4.0 regression.

Seemingly caused between 2007-12-07-r130671 and 2007-12-10-r130736. Probably
caused by my NaN/INF patch (PR fortran/34319 - or, less likely, fortran/34404).


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jvdelisle at gcc dot gnu dot
                   |                            |org, burnus at gcc dot gnu
                   |                            |dot org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
  GCC build triplet|GNU Fortran (GCC) 4.3.0     |
   GCC host triplet|Linux quartic.txcorp.com    |
                   |2.6.19-1.2288.fc5           |
 GCC target triplet|x86_64 x86_64 x86_64        |
                   |GNU/Linux                   |
           Keywords|                            |wrong-code
      Known to fail|                            |4.3.0 4.4.0
      Known to work|                            |4.2.2 4.1.3
   Last reconfirmed|0000-00-00 00:00:00         |2008-03-18 21:50:59
               date|                            |
            Summary|namelist read error         |[4.3, 4.4 regression]
                   |                            |namelist read error


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[burnus at gcc dot gnu dot org]

------- Comment #2 from burnus at gcc dot gnu dot org  2008-03-18 22:02 -------
The problem is that we don't properly rewind properly for "nbshapa"; the parser
sees 'n'; it then try whether "NaN" matches. It checks the next character 'b'
and fails. The problem is that only 'b' and not 'nb' are put back:

    } /* Match NaN. Up to here, c contains either 'N' or 'n'. */
  else if (((c = next_char (dtp)) == 'a' || c == 'A')
           && ((c = next_char (dtp)) == 'n' || c == 'N')
           && (c = next_char (dtp)))
    { ... }
 bad:
  if (nml_bad_return (dtp, c))
    return 0;


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #3 from jvdelisle at gcc dot gnu dot org  2008-03-18 23:57 -------
Namelist bugs are just a tradition. :) Assigning myself.


-- 

jvdelisle at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jvdelisle at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2008-03-18 21:50:59         |2008-03-18 23:57:49
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[toon at moene dot indiv dot nluug dot nl]

------- Comment #4 from toon at moene dot indiv dot nluug dot nl  2008-03-19 08:37 -------
I'm hit by this, too - don't know when it started (it's in a namelist I've been
using for years).


-- 

toon at moene dot indiv dot nluug dot nl changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |toon at moene dot indiv dot
                   |                            |nluug dot nl


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #5 from jvdelisle at gcc dot gnu dot org  2008-03-19 12:35 -------
I believe the problem is actual;y in read_logical and occurs on a short read. 
By short, I mean the case where we provide fewer data items in the namelist
file then there are in the logical array.  I am curious what has triggered
this.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[burnus at gcc dot gnu dot org]

------- Comment #6 from burnus at gcc dot gnu dot org  2008-03-19 15:54 -------
(In reply to comment #5)
> I believe the problem is actual;y in read_logical and occurs on a short read. 
> By short, I mean the case where we provide fewer data items in the namelist
> file then there are in the logical array.  I am curious what has triggered
> this.

I think it has something to do with read_real, though different from what I
thought. In any case, we end up in read_real, trying to match "NaN" and go to
unwind. However, afterwards it shows an error for "xlbtna" (for which read_real
is called) and not for the item that comes next and is not NaN, namely
"nbshapa".

Just before the return, line_buffer contains in "unwind:" of read_real:
  line_buffer[0] == '\0'
  line_buffer[1] == 'n'
  line_buffer[2] == 'b'

The problem is that the first element is '\0' and thus the rewinding does not
happen. If one changes "nbshapa" into "nnnbshapa" it works. Thus the question
is : Where does the "\0" come from?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #7 from jvdelisle at gcc dot gnu dot org  2008-03-19 17:55 -------
What I observed is that if you change the namelist file to nlco = 200*.TRUE.
the subsequent real reads work fine.  Likewise if you move the logical to the
end after the real reads, it works fine. So this is why I suspect the
read_logical may have an off by one leftover in line_buffer or we are not
cleaning up correctly on that.  This may be where the spurious \0 is coming
from.  I will have more time to look into it in the next few days. 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[burnus at gcc dot gnu dot org]

------- Comment #8 from burnus at gcc dot gnu dot org  2008-03-19 18:29 -------
> So this is why I suspect the read_logical may have an off by one leftover
I think you are right. In that sense it is an only bug which was only exposed
by the new NaN support.

Not very elegent, but the following works. One should probably do an audit of
all l_push_char. The following works here, but I do not understand why element
[0] was '\0' and not something else.

Index: libgfortran/io/list_read.c
===================================================================
--- libgfortran/io/list_read.c  (Revision 133342)
+++ libgfortran/io/list_read.c  (Arbeitskopie)
@@ -680,6 +680,7 @@ read_logical (st_parameter_dt *dtp, int
       break;

     CASE_SEPARATORS:
+      dtp->u.p.item_count--;
       unget_char (dtp, c);
       eat_separator (dtp);
       return;                  /* Null value.  */
@@ -741,6 +742,7 @@ read_logical (st_parameter_dt *dtp, int

  bad_logical:

+  dtp->u.p.item_count--;
   free_line (dtp);

   if (nml_bad_return (dtp, c))


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #9 from jvdelisle at gcc dot gnu dot org  2008-03-19 19:37 -------
I agree and plan to look at l_push_char.  We just should never read a "\0"
character.  We may be ungetting it because we use "\0" to indicate that the
line_buffer is empty and to not read from it, but maybe unget_char is not quite
right.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #10 from jvdelisle at gcc dot gnu dot org  2008-03-20 01:17 -------
Created an attachment (id=15348)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15348&action=view)
A final patch

l_push_char assumes the index which is item_count has been set to zero and
after saving a character it bumps the index.  This bug has been latent for a
long time, Forgot to set item_count back to zero when when freeing the
line_buffer.

The simplest fix is this:

@@ -741,6 +742,7 @@ read_logical (st_parameter_dt *dtp, int

  bad_logical:

+  dtp->u.p.item_count = 0;
   free_line (dtp);

   if (nml_bad_return (dtp, c))

However, I am going to move this reseting into free_line where it ought to be. 
The attachment is the final fix with some cleanup.  Regression tested on
x86-64.  I will commit as obvious with the new test case.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #11 from jvdelisle at gcc dot gnu dot org  2008-03-20 02:05 -------
Subject: Bug 35627

Author: jvdelisle
Date: Thu Mar 20 02:05:05 2008
New Revision: 133360

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133360
Log:
2008-03-19  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR libfortran/35627
        * io/list_read.c (free_line): Clear the line buffer enable flag and
        reset the index into line_buffer, aka item_count.
        (next_char): Cleanup whitespace.
        (read_logical): Use unget_char to assure that the first character of
the
        bad logical is saved in case it is part of an object name. Remove the
        clearing of index and flag that is now in free_line.
        (read_real): Likewise.

Modified:
    trunk/libgfortran/ChangeLog
    trunk/libgfortran/io/list_read.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #12 from jvdelisle at gcc dot gnu dot org  2008-03-20 02:08 -------
Subject: Bug 35627

Author: jvdelisle
Date: Thu Mar 20 02:07:38 2008
New Revision: 133361

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133361
Log:
2008-03-19  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR libfortran/35627
        * gfortran.dg/namelist_46.f90: New test.

Added:
    trunk/gcc/testsuite/gfortran.dg/namelist_46.f90
Modified:
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3, 4.4 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #13 from jvdelisle at gcc dot gnu dot org  2008-03-20 02:09 -------
Fixed on trunk.  Will backport to 4.3 in a day or so.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3 regression] namelist read error

[burnus at gcc dot gnu dot org]

-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|4.3.0 4.4.0                 |4.3.0
      Known to work|4.2.2 4.1.3                 |4.2.2 4.1.3 4.4.0
           Priority|P3                          |P4
            Summary|[4.3, 4.4 regression]       |[4.3 regression] namelist
                   |namelist read error         |read error
   Target Milestone|---                         |4.3.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #14 from jvdelisle at gcc dot gnu dot org  2008-03-21 01:30 -------
Subject: Bug 35627

Author: jvdelisle
Date: Fri Mar 21 01:29:30 2008
New Revision: 133411

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133411
Log:
2008-03-20  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR libfortran/35627
        Backport from trunk.
        * io/list_read.c (free_line): Clear the line buffer enable flag and
        reset the index into line_buffer, aka item_count.
        (next_char): Cleanup whitespace.
        (read_logical): Use unget_char to assure that the first character of
the
        bad logical is saved in case it is part of an object name. Remove the
        clearing of index and flag that is now in free_line.
        (read_real): Likewise.

        PR libfortran/35617
        Backport from trunk.
        * io/list_read.c (eat_separator): If next character after eatline is
'!'
        then eatline again. 

Modified:
    branches/gcc-4_3-branch/libgfortran/ChangeLog
    branches/gcc-4_3-branch/libgfortran/io/list_read.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #15 from jvdelisle at gcc dot gnu dot org  2008-03-21 01:38 -------
Subject: Bug 35627

Author: jvdelisle
Date: Fri Mar 21 01:37:38 2008
New Revision: 133413

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133413
Log:
2008-03-20  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        Backport from trunk:
        PR libfortran/35627
        * gfortran.dg/namelist_46.f90: New test.

        PR libfortran/35617
        * gfortran.dg/namelist_45.f90: New test.

Added:
    branches/gcc-4_3-branch/gcc/testsuite/gfortran.dg/namelist_45.f90
Modified:
    branches/gcc-4_3-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627

[Bug fortran/35627] [4.3 regression] namelist read error

[jvdelisle at gcc dot gnu dot org]

------- Comment #16 from jvdelisle at gcc dot gnu dot org  2008-03-21 03:13 -------
I typoed the file name on the commit of namelist_46.f90, but it is there.

Fixed on 4.3.


-- 

jvdelisle at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35627