On Fri, Jan 29, 2021 at 4:46 PM 'Nick Desaulniers' via Clang Built Linux <clang-built-linux@xxxxxxxxxxxxxxxx> wrote: > > On Fri, Jan 29, 2021 at 4:08 PM Sedat Dilek <sedat.dilek@xxxxxxxxx> wrote: > > > > On Fri, Jan 29, 2021 at 8:43 PM Nick Desaulniers > > <ndesaulniers@xxxxxxxxxx> wrote: > > > > > > DWARF v5 is the latest standard of the DWARF debug info format. > > > > > > DWARF5 wins significantly in terms of size and especially so when mixed > > > with compression (CONFIG_DEBUG_INFO_COMPRESSED). > > > > > > Link: http://www.dwarfstd.org/doc/DWARF5.pdf > > > > > > Patch 1 is a cleanup that lays the ground work and isn't DWARF > > > v5 specific. > > > Patch 2 implements Kconfig and Kbuild support for DWARFv5. > > > > > > > When you will do a v7... > > > > Can you look also at places where we have hardcoded DWARF-2 handling... > > Ah, sorry, I just saw this now, after sending v7. Can we wait to > purge DWARF v2 until after we have DWARF v5? > > In fact, if they are orthogonal like I suspect, why don't you send > some patches and I will help you test them? > -- > Thanks, > ~Nick Desaulniers Basically the distinction is just between DWARF v2 .debug_line and DWARF v5 .debug_line . (-gdwarf-4 adds an extra maximum_operations_per_instruction (constant 1) compared with -gdwarf-2 but that can mostly be ignored). Refinement among -gdwarf-[234] just clarifies things and is not going to affect debugging experience in any case. So I agree with Nick that it can be done separately. Note: such clarification can make things a bit ugly because binutils before 2020 does not recognize -gdwarf-[345]. -- 宋方睿
