Advice on fixing gdb/12528
Tom Tromey
tromey@redhat.com
Fri Mar 11 18:27:00 GMT 2011
>>>>> "Paul" == Paul Pluzhnikov <ppluzhnikov@google.com> writes:
Paul> I know that in general GDB can not ignore code at location 0,
Paul> but it appears inevitable that it must do so on platforms where
Paul> - it is known that no code can execute there, and
Paul> - linker relocates debug info to address 0 to indicate that
Paul> the corresponding code has been discarded.
Paul> So would something like:
Paul> case DW_LNE_set_address:
Paul> address = read_address (abfd, line_ptr, cu, &bytes_read);
Paul> if (!target->to_valid_code_address (address)) {
Paul> // this debug line info corresponds to function that has
Paul> // been GCd by the linker. Skip to end_sequence.
Paul> }
Paul> in dwarf_decode_lines() ?
Can you check the has_section_at_zero flag on dwarf2_per_objfile?
Tom
More information about the Gdb-patches
mailing list