[Bug fortran/23057] New: Wrong DWARF output for Fortran common block

[Bug fortran/23057] New: Wrong DWARF output for Fortran common block

[woodzltc at sources dot redhat dot com]

In section 4.2 of DWARF STD, it says that:

A Fortran common block may be described by a debugging information entry 
with the tag DW_TAG_common_block. 

But it seems that both gfortran and g77 didn't conform to this.  This makes
trouble for debugging common block.  Here is a testcase to verify this: 

        program info_common
        implicit none
          integer :: a,b
          common /group1/ a,b
          a=1
          b=2
          call ShowCommon()
          stop
        end

        subroutine ShowCommon()
        implicit none
          integer :: num1, num2
          common /group1/ num1, num2
          write(*,*) num1, num2
          return
        end

group1 is a named common block.  After using gfortran to build this case, I use
"readelf -wi" to get the debuginfo of the executable:

 <2><9e>: Abbrev Number: 3 (DW_TAG_variable)
     DW_AT_name        : group1
     DW_AT_decl_file   : 1
     DW_AT_decl_line   : 4
     DW_AT_MIPS_linkage_name: group1_
     DW_AT_type        : <e1>        =====>  This point to the type of group1
     DW_AT_external    : 1
     DW_AT_location    : 5 byte block: 3 d0 99 4 8      (DW_OP_addr: 80499d0)
 <1><e1>: Abbrev Number: 5 (DW_TAG_structure_type)
     DW_AT_sibling     : <100>
     DW_AT_byte_size   : 8
 <2><e7>: Abbrev Number: 6 (DW_TAG_member)
     DW_AT_name        : a
     DW_AT_decl_file   : 1
     DW_AT_decl_line   : 3
     DW_AT_type        : <100>
     DW_AT_data_member_location: 2 byte block: 23 0     (DW_OP_plus_uconst: 0)
 <2><f3>: Abbrev Number: 6 (DW_TAG_member)
     DW_AT_name        : b
     DW_AT_decl_file   : 1
     DW_AT_decl_line   : 3
     DW_AT_type        : <100>
     DW_AT_data_member_location: 2 byte block: 23 4     (DW_OP_plus_uconst: 4)

It tells us that group1 is treated as a two-members structure.

g77 handles this in another way, it treats group1 as an eight-character array.
And eight is exactly the size of common block group1.  Here is the debuginfo for
g77-generated executable:

 <1><12b>: Abbrev Number: 6 (DW_TAG_base_type)
     DW_AT_name        : (indirect string, offset: 0x3b): char
     DW_AT_byte_size   : 1
     DW_AT_encoding    : 8      (unsigned char)
 <1><132>: Abbrev Number: 15 (DW_TAG_array_type)
     DW_AT_sibling     : <143>
     DW_AT_type        : <12b>
 <2><13b>: Abbrev Number: 16 (DW_TAG_subrange_type)
     DW_AT_type        : <70>
     DW_AT_lower_bound : 0
     DW_AT_upper_bound : 7
 <1><143>: Abbrev Number: 17 (DW_TAG_variable)
     DW_AT_name        : (indirect string, offset: 0xbb): group1_
     DW_AT_decl_file   : 1
     DW_AT_decl_line   : 3
     DW_AT_type        : <132>   =====>  This point to the type of group1
     DW_AT_external    : 1
     DW_AT_location    : 5 byte block: 3 a0 99 4 8      (DW_OP_addr: 80499a0)


P.S: I reported this problem some days before.  But there is no response ever
since.  Here is the link:  

  http://gcc.gnu.org/ml/fortran/2005-07/msg00310.html

Thanks.

-- 
           Summary: Wrong DWARF output for Fortran common block
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: woodzltc at sources dot redhat dot com
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23057

[Bug fortran/23057] Wrong DWARF output for Fortran common block

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-debug


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23057

[Bug fortran/23057] Fortran common blocks are not described with DW_TAG_common_block

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-25 05:48 -------
Confiremd, since the standard says "may", this is an enhancement.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-07-25 05:48:05
               date|                            |
            Summary|Wrong DWARF output for      |Fortran common blocks are
                   |Fortran common block        |not described with
                   |                            |DW_TAG_common_block


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23057

[Bug fortran/23057] Fortran common blocks are not described with DW_TAG_common_block

[howarth at nitro dot med dot uc dot edu]

------- Additional Comments From howarth at nitro dot med dot uc dot edu  2005-08-03 20:21 -------
Andrew,
     Do you think you might be interested in tackling the implementation of the
DW_TAG_common_block tag if we could get HJ Lu and/or Jakub Jelinek to provide
technical assistance? Having the ability to debug COMMON blocks in gfortran would
be quite helpful for programs like xplor-nih that heavily use them.
                   Jack

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23057

[Bug fortran/23057] Fortran common blocks are not described with DW_TAG_common_block

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-08-03 20:25 -------
(In reply to comment #2)
Jack,
Actually on ppc-darwin, GCC still uses stabs so I doubt this is the bug you are hitting.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |howarth at nitro dot med dot
                   |                            |uc dot edu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23057