Location lists again

[Michal Ludvig]

Daniel Berlin wrote:
> 
> On Thursday, February 13, 2003, at 11:28  AM, Michal Ludvig wrote:
>>
>>         .section        .debug_loc,"",@progbits
>> .LLST0:
>>         .quad   .LVL0   # Location list begin address (*.LLST0)
>>         .quad   .LVL1   # Location list end address (*.LLST0)
>>         .value  0x1     # Location expression size
>>         .byte   0x55    # DW_OP_reg5
>> [...]
>>
>> This way the addresses would be correctly updated on linking.
> 
> They are correct, though readelf -w will only show the offsets, because 
> it doesn't know what CU they go with (IE debug_loc parsing is completely 
> removed from debug_info parsing).
> 
> When a debugger actually loads them, it is supposed to add the base 
> address of the CU to each one.

I've been looking to gcc/dwarf2out.c (on rtlopt-branch) and in 
output_loc_list() there is a code that could output addresses instead of 
deltas:

[...]
       if (separate_line_info_table_in_use == 0)
         {
           dw2_asm_output_delta (DWARF2_ADDR_SIZE, curr->begin, 
curr->section,
                                 "Location list begin address (%s)",
                                 list_head->ll_symbol);
           dw2_asm_output_delta (DWARF2_ADDR_SIZE, curr->end, curr->section,
                                 "Location list end address (%s)",
                                 list_head->ll_symbol);
         }
       else
         {
           dw2_asm_output_addr (DWARF2_ADDR_SIZE, curr->begin,
                                "Location list begin address (%s)",
                                list_head->ll_symbol);
           dw2_asm_output_addr (DWARF2_ADDR_SIZE, curr->end,
                                "Location list end address (%s)",
                                list_head->ll_symbol);
         }
[...]

Under what circumstances can this happen? How do I convince GCC to go 
through the second branch? I just want to handle both cases in GDB...

And how about this piece:
[...]
   /* ??? This shouldn't be needed now that we've forced the
      compilation unit base address to zero when there is code
      in more than one section.  */
   if (strcmp (curr->section, ".text") != 0)
     {
       /* dw2_asm_output_data will mask off any extra bits in the ~0.  */
       dw2_asm_output_data (DWARF2_ADDR_SIZE, ~(unsigned HOST_WIDE_INT) 0,
                   "Location list base address specifier fake entry");
       dw2_asm_output_offset (DWARF2_ADDR_SIZE, curr->section,
                   "Location list base address specifier base");
     }
[...]

Imagine there is a non-null base address in Compilation unit DIE, and 
the appropriate location lists have headers in the form
.quad 0xffffff...
.quad .text[.label] # loclist base address

Are both base addresses the same? Well, as far as I could see they are, 
but is it ensured? Or does one of them have a higher "preference" in the 
case they differ?

Thanks

Michal Ludvig
-- 
* SuSE CR, s.r.o     * mludvig@suse.cz
* (+420) 296.545.373 * http://www.suse.cz