[Bug debug/19192] New: Current development gcc generates incorrect line info for example code

[Bug debug/19192] New: Current development gcc generates incorrect line info for example code

[gcc-bugzilla at gcc dot gnu dot org]

For the supplied example, the current development gcc generates
incorrect line number information for one instruction from the inlined
subroutine.  The Fedora Core 3 gcc, which identifies itself as "gcc
version 3.4.2 20041017 (Red Hat 3.4.2-6.fc3)", generates correct line
information.

Environment:
System: Linux toadfish.ninemoons.com 2.6.9-1.681_FC3.stk16 #1 Mon Nov 22 16:44:32 EST 2004 i686 i686 i386 GNU/Linux
Architecture: i686

	
host: i686-pc-linux-gnu
build: i686-pc-linux-gnu
target: i686-pc-linux-gnu
configured with: /src/sourceware/gcc/gcc/configure -v --prefix=/opt/local/sourceware --with-gcc-version-trigger=/src/sourceware/gcc/gcc/gcc/version.c --enable-languages=c,c++

How-To-Repeat:

For the follow code:

  /*  1 */  int add2 (int a, int b)
  /*  2 */  {
  /*  3 */    return (a + b);
  /*  4 */  }
  /*  5 */  
  /*  6 */  inline int add3 (int a, int b, int c)
  /*  7 */  {
  /*  8 */    return (a + add2 (b , c));
  /*  9 */  }
  /* 10 */  
  /* 11 */  main ()
  /* 12 */  {
  /* 13 */    printf ("%d\n", add3 (2, 3, 4));
  /* 14 */  }

Run:

  gcc -g -O2 --save-temps -o t t.c
  readelf --debug-dump t >t.dwarf
  objdump --disassemble t >t.dis

There relevant line table entries from t.dwarf are:

  Special opcode 36: advance Address by 2 to 0x80483a0 and Line by 3 to 12
  Special opcode 15: advance Address by 1 to 0x80483a1 and Line by -4 to 8
  Special opcode 79: advance Address by 5 to 0x80483a6 and Line by 4 to 12
  Special opcode 75: advance Address by 5 to 0x80483ab and Line by 0 to 12
  Special opcode 85: advance Address by 6 to 0x80483b1 and Line by -4 to 8
  Special opcode 234: advance Address by 16 to 0x80483c1 and Line by 5 to 13
  Advance PC by constant 17 to 0x80483d2
  Special opcode 34: advance Address by 2 to 0x80483d4 and Line by 1 to 14
  Advance PC by 2 to 80483d6

I've annotated the disassembly of main() with the line numbers given by
the above entries:

    12      080483a0 <main>:
    12       80483a0:       55                      push   %ebp
     8       80483a1:       b8 04 00 00 00          mov    $0x4,%eax
    12       80483a6:       89 e5                   mov    %esp,%ebp
    12       80483a8:       83 ec 08                sub    $0x8,%esp
    12       80483ab:       83 e4 f0                and    $0xfffffff0,%esp
    12       80483ae:       83 ec 10                sub    $0x10,%esp
     8       80483b1:       89 44 24 04             mov    %eax,0x4(%esp)
     8       80483b5:       c7 04 24 03 00 00 00    movl   $0x3,(%esp)
     8       80483bc:       e8 af ff ff ff          call   8048370 <add2>
    13       80483c1:       c7 04 24 c8 84 04 08    movl   $0x80484c8,(%esp)
 ** 13       80483c8:       83 c0 02                add    $0x2,%eax
    13       80483cb:       89 44 24 04             mov    %eax,0x4(%esp)
    13       80483cf:       e8 dc fe ff ff          call   80482b0 <printf@plt>
    14       80483d4:       c9                      leave  
    14       80483d5:       c3                      ret    

The "add $0x2,%eax" instruction, marked with '**', is supposed to be at
line 8, not line 13.      

If you go through the same exercise with the Fedora Core 3 compiler, it does
correctly identify this instruction as being at line 8.

-- 
           Summary: Current development gcc generates incorrect line info
                    for example code
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: fnf at specifixinc dot com
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0 Regression] Current development gcc generates incorrect line info for example code

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-29 17:56 -------
This is caused by ter.

Confirmed, easier way to find this out is to look at the tree dump with -lineno as an option.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amacleod at redhat dot com
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |wrong-debug
   Last reconfirmed|0000-00-00 00:00:00         |2004-12-29 17:56:33
               date|                            |
            Summary|Current development gcc     |[4.0 Regression] Current
                   |generates incorrect line    |development gcc generates
                   |info for example code       |incorrect line info for
                   |                            |example code
   Target Milestone|---                         |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0 Regression] Current development gcc generates incorrect line info for example code

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-01-02 06:08 -------
With TER:
  [t.c : 8] D.1144 = [t.c : 8] add2 (3, 4);
  [t.c : 13] printf (&"%d\n"[0], D.1144 + 2) [tail call];
  [t.c : 14] return;
Without:
  [t.c : 8] D.1144 = [t.c : 8] add2 (3, 4);
  [t.c : 8] D.1124 = D.1144 + 2;
  [t.c : 13] printf (&"%d\n"[0], D.1124) [tail call];
  [t.c : 14] return;

In a way we cannot say the addition is on line 8 because we already optimizated before.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0 Regression] Current development gcc generates incorrect line info for example code

[amacleod at redhat dot com]

------- Additional Comments From amacleod at redhat dot com  2005-01-12 17:46 -------
Created an attachment (id=7940)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7940&action=view)
proposed patch

Give this a try. I havent checked it in, Im running it through the testsuites.
I'm unlikely to be able to check it in until next week. This is likely to
change the debug info quite a bit, but it should be more detailed. Let me know
if you think it is OK, and if not, what it is doing wrong.

for the specified testcase I get:
  [a.c : 8] D.1150 = [a.c : 8] add2 (3, 4);
  [a.c : 13] printf (&"%d\n"[0], [a.c : 8] D.1150 + 2) [tail call];
  [a.c : 14] return;

so that looks reasonable. something artificial and slightly more complex gives
me:
    x = a + b;
    y = x * 3;
    z = y << 2 + x;
    x = z * 10;
    printf ("%d\n", x * 44);
  [d.c : 6] x = a + b;
  [d.c : 10] printf (&"%d\n"[0], ([d.c : 8] [d.c : 7] x * 3 << x + 2) * 440)
[tail call];
  [d.c : 13] return;

which looks like it ought to be about right, but is a little tricky to read :-)


anyway, give it a shot and let me know if you think this is better.

Andrew

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |amacleod at redhat dot com
                   |dot org                     |
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0 Regression] Current development gcc generates incorrect line info for example code

[fnf at specifixinc dot com]

------- Additional Comments From fnf at specifixinc dot com  2005-01-13 20:56 -------
Subject: Re:  [4.0 Regression] Current development gcc generates incorrect line info for example code

On Wednesday 12 January 2005 11:46, amacleod at redhat dot com wrote:
> Give this a try.

Unfortunately, for my original example, this seems to have made things worse.
The differences are:

  $ diff -w ../old/z.m ../new/z.m
  6,8c6,7
  <   Special opcode 234: advance Address by 16 to 0x80483c1 and Line by 5 to 13
  <   Advance PC by constant 17 to 0x80483d2
  <   Special opcode 34: advance Address by 2 to 0x80483d4 and Line by 1 to 14
  ---
  >   Advance PC by 35 to 80483d4
  >   Special opcode 11: advance Address by 0 to 0x80483d4 and Line by 6 to 14
  20,23c19,22
  < 13     80483c1: c7 04 24 c8 84 04 08    movl   $0x80484c8,(%esp)
  < 13     80483c8: 83 c0 02                add    $0x2,%eax
  < 13     80483cb: 89 44 24 04             mov    %eax,0x4(%esp)
  < 13     80483cf: e8 dc fe ff ff          call   80482b0 <printf@plt>
  ---
  >  8     80483c1:  c7 04 24 c8 84 04 08    movl   $0x80484c8,(%esp)
  >  8     80483c8:  83 c0 02                add    $0x2,%eax
  >  8     80483cb:  89 44 24 04             mov    %eax,0x4(%esp)
  >  8     80483cf:  e8 dc fe ff ff          call   80482b0 <printf@plt>
  
I.E. gcc now associates the code to print the results with line 8 instead
of line 13.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0/4.1 Regression] Current development gcc generates inaccurate line info for example code

[mmitchel at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.0                       |4.0.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0/4.1 Regression] Current development gcc generates inaccurate line info for example code

[mmitchel at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.1                       |4.0.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192

[Bug debug/19192] [4.0/4.1 Regression] Current development gcc generates inaccurate line info for example code

[mmitchel at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.2                       |4.0.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19192