[Bug debug/47946] New: Dwarf uses 64-bits to refer to a structure offset unnecessarily

[Bug debug/47946] New: Dwarf uses 64-bits to refer to a structure offset unnecessarily

[hariharans at picochip dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

           Summary: Dwarf uses 64-bits to refer to a structure offset
                    unnecessarily
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hariharans@picochip.com


Created attachment 23503
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23503
The source code

Hello,
In the attached testcase, our dwarf reader reports the structure as

Abbrevation 2: DW_TAG_structure_type [has children]
  DW_AT_name            DW_FORM_string
  DW_AT_byte_size               DW_FORM_data1
  DW_AT_decl_file               DW_FORM_data1
  DW_AT_decl_line               DW_FORM_data1
  DW_AT_sibling         DW_FORM_ref4
Abbrevation 3: DW_TAG_member
  DW_AT_name            DW_FORM_string
  DW_AT_decl_file               DW_FORM_data1
  DW_AT_decl_line               DW_FORM_data1
  DW_AT_type            DW_FORM_ref4
  DW_AT_byte_size               DW_FORM_data1
  DW_AT_bit_size                DW_FORM_data1
  DW_AT_bit_offset              DW_FORM_data1
  DW_AT_data_member_location            DW_FORM_block1
Abbrevation 4: DW_TAG_member
  DW_AT_name            DW_FORM_string
  DW_AT_decl_file               DW_FORM_data1
  DW_AT_decl_line               DW_FORM_data1
  DW_AT_type            DW_FORM_ref4
  DW_AT_byte_size               DW_FORM_data1
  DW_AT_bit_size                DW_FORM_data1
  DW_AT_bit_offset              DW_FORM_data8
  DW_AT_data_member_location            DW_FORM_block1

Note that the first element (c1) takes a DW_FORM_data1 to represent the offset,
whereas the second (i) uses DW_FORM_data8. I am fairly certain it doesn't need
64-bits to represent the offset, so something is wrong here.

I tested this with picochip port in the mainline GCC (as of 03-feb-2011).

[Bug debug/47946] Dwarf uses 64-bits to refer to a structure offset unnecessarily

[hariharans at picochip dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

--- Comment #1 from hariharans at picochip dot com 2011-03-01 15:31:47 UTC ---
Created attachment 23504
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23504
The assembly code

[Bug debug/47946] Dwarf uses 64-bits to refer to a structure offset unnecessarily

[matz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

Michael Matz <matz at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |matz at gcc dot gnu.org

--- Comment #2 from Michael Matz <matz at gcc dot gnu.org> 2011-03-01 17:06:42 UTC ---
Due to the definition of dwarf2/3 these offsets are negative on little-endian
machines.  This was fixed in dwarf4, see
  http://dwarfstd.org/ShowIssue.php?issue=081130.1&type=closed3

GCC doesn't yet support this part of dwarf4.  But even for dwarf2 we could
emit this signed number as sleb128, not as unsigned number (that is what
causes us to use FORM_data8).

[Bug debug/47946] Dwarf uses 64-bits to refer to a structure offset unnecessarily

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-01 18:39:17 UTC ---
Created attachment 23506
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23506
gcc46-pr47946.patch

Untested fix.

[Bug debug/47946] Dwarf uses 64-bits to refer to a structure offset unnecessarily

[hariharans at picochip dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

--- Comment #4 from hariharans at picochip dot com 2011-03-02 10:11:19 UTC ---
Hello Jakub,
I tested the patch and it fixes the problem. Thanks for looking into this.

Cheers
Hari

[Bug debug/47946] Dwarf uses 64-bits to refer to a structure offset unnecessarily

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-14 20:07:29 UTC ---
Author: jakub
Date: Mon Mar 14 20:07:22 2011
New Revision: 170956

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=170956
Log:
    PR debug/47946
    * dwarf2out.c (add_bit_offset_attribute): If bit_offset is negative,
    emit it as add_AT_int instead of add_AT_unsigned.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/dwarf2out.c

[Bug debug/47946] Dwarf uses 64-bits to refer to a structure offset unnecessarily

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47946

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |jakub at gcc dot gnu.org
         Resolution|                            |FIXED

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-03-14 20:13:29 UTC ---
Fixed for 4.7+.