Re: [PATCH] Fix -gdwarf-5 -gsplit-dwarf ICEs (PR debug/89498)

[Jason Merrill <jason at redhat dot com>]

On 3/4/19 5:35 PM, Jakub Jelinek wrote:
> Hi!
>
> output_view_list_offset does:
>    if (dwarf_split_debug_info)
>      dw2_asm_output_delta (DWARF_OFFSET_SIZE, sym, loc_section_label,
>                            "%s", dwarf_attr_name (a->dw_attr));
>    else
>      dw2_asm_output_offset (DWARF_OFFSET_SIZE, sym, debug_loc_section,
>                             "%s", dwarf_attr_name (a->dw_attr));
> while output_loc_list_offset does:
>    if (!dwarf_split_debug_info)
>      dw2_asm_output_offset (DWARF_OFFSET_SIZE, sym, debug_loc_section,
>                             "%s", dwarf_attr_name (a->dw_attr));
>    else if (dwarf_version >= 5)
>      {
>        gcc_assert (AT_loc_list (a)->num_assigned);
>        dw2_asm_output_data_uleb128 (AT_loc_list (a)->hash, "%s (%s)",
>                                     dwarf_attr_name (a->dw_attr),
>                                     sym);
>      }
>    else
>      dw2_asm_output_delta (DWARF_OFFSET_SIZE, sym, loc_section_label,
>                            "%s", dwarf_attr_name (a->dw_attr));
> but both size_of_die and value_format handle both the same as loc_list,
> so for -gdwarf-5 -gsplit-dwarf we just ICE, as e.g. AT_loc_list is not
> valid on a view list.
>
> Assuming output_view_list_offset is correct, the following patch adjusts
> size_of_die/value_format accordingly.

I would guess that omitting the handling from output_view_list_offset 
was an oversight in the view work.  Alex, which is right? :)

Jason