Bug in Fortran: DW_AT_name

Hi,

I try to debug a Fortan binary produced by gcc 4.3.  I get:

Breakpoint 1, MAIN__ () at 
:1
1       
: No such file or directory.
        in 

(gdb) info frame
Stack level 0, frame at 0xbfcec3e0:
 eip = 0x80485bd in MAIN__ (
:1); saved eip 0x80486c9

objdump is able to read the DW_AT_name

 <0><b>: Abbrev Number: 1 (DW_TAG_compile_unit)
     DW_AT_producer    : (indirect string, offset: 0x41):
GNU F95 4.3.0­
     DW_AT_language    : 14­····(Fortran 95)
     DW_AT_name        : (indirect string, offset: 0x0):
./neg_array.F­·
     DW_AT_comp_dir    : (indirect string, offset: 0x1b):
/home/deuling/fortran­

but shows following error:

objdump: Error: Location lists in .debug_info section aren't
in ascending order!
objdump: Warning: There is an overlap [0x2c - 0xfffffffe] in
.debug_loc section.
objdump: Warning: Offset 0xfffffffe is bigger than
.debug_loc section size.
objdump: Warning: There is a hole [0xfffffffe - 0x2c] in
.debug_loc section.

C-code works fine.

For my opinion this is a compiler bug in fortran. But as
objdump is able to recognize 
DW_AT_name maybe GDB should, too? What do you think ?

-- 
  Markus Deuling
  GNU Toolchain for Linux on Cell BE
  deulingde.ibm.com

Markus Deuling schrieb:
> Hi,
> 
> I try to debug a Fortan binary produced by gcc 4.3.  I
get:
> 
> Breakpoint 1, MAIN__ () at 
:1
> 1       
: No such file or directory.
>        in 

> (gdb) info frame
> Stack level 0, frame at 0xbfcec3e0:
> eip = 0x80485bd in MAIN__ (
:1); saved eip 0x80486c9
> 
> objdump is able to read the DW_AT_name
> 
> <0><b>: Abbrev Number: 1
(DW_TAG_compile_unit)
>     DW_AT_producer    : (indirect string, offset:
0x41): GNU F95 4.3.0­
>     DW_AT_language    : 14­····(Fortran 95)
>     DW_AT_name        : (indirect string, offset: 0x0):
./neg_array.F­·
>     DW_AT_comp_dir    : (indirect string, offset:
0x1b): 
> /home/deuling/fortran­
> 
> but shows following error:
> 
> objdump: Error: Location lists in .debug_info section
aren't in 
> ascending order!
> objdump: Warning: There is an overlap [0x2c -
0xfffffffe] in .debug_loc 
> section.
> objdump: Warning: Offset 0xfffffffe is bigger than
.debug_loc section size.
> objdump: Warning: There is a hole [0xfffffffe - 0x2c]
in .debug_loc 
> section.
> 
> C-code works fine.
> 
> For my opinion this is a compiler bug in fortran. But
as objdump is able 
> to recognize DW_AT_name maybe GDB should, too? What do
you think ?
> 

Hm, readelf does not show overlaps and no errors/warnings as
well:

  [25] .debug_aranges    PROGBITS        00000000 0009e2
000040 00      0   0  1
  [26] .debug_pubnames   PROGBITS        00000000 000a22
000038 00      0   0  1
  [27] .debug_info       PROGBITS        00000000 000a5a
000361 00      0   0  1
  [28] .debug_abbrev     PROGBITS        00000000 000dbb
000141 00      0   0  1
  [29] .debug_line       PROGBITS        00000000 000efc
0000e3 00      0   0  1
  [30] .debug_frame      PROGBITS        00000000 000fe0
00006c 00      0   0  4
  [31] .debug_str        PROGBITS        00000000 00104c
0002db 01  MS  0   0  1
  [32] .debug_loc        PROGBITS        00000000 001327
00006f 00      0   0  1
  [33] .shstrtab         STRTAB          00000000 001396
00013f 00      0   0  1
  [34] .symtab           SYMTAB          00000000 001a78
000550 10     35  58  4
  [35] .strtab           STRTAB          00000000 001fc8
000369 00      0   0  1

So it might be a Bug in BFD library? 



-- 
  Markus Deuling
  GNU Toolchain for Linux on Cell BE
  deulingde.ibm.com

Hi,

just fyi:

Markus Deuling schrieb:
>> For my opinion this is a compiler bug in fortran.
But as objdump is 
>> able to recognize DW_AT_name maybe GDB should, too?
What do you think ?
>>

it seems this is a gfortran bug. I opened a bugzilla:
htt
p://gcc.gnu.org/bugzilla/show_bug.cgi?id=35615

Regards,
Markus

-- 
  Markus Deuling
  GNU Toolchain for Linux on Cell BE
  deulingde.ibm.com