On Wed, Apr 25, 2012 at 6:47 AM, Jakub Jelinek <ja...@redhat.com> wrote:
> Hi!
>
> For reasonable debugging experience recent GCC versions need
> GDB >= 7.0 for quite some time, and DWARF4 is almost 2 years old now,
> and offers lots of improvements over DWARF2 we still default to.
>
> So, I'd like to make -gdwarf-4 (just the version, of course -g
> is needed to enable debug info generation) the default, together
> with -fno-debug-types-section (as .debug_types isn't right now always a win,
> see the data in the dwz-0.1 announcement plus not all DWARF consumers can
> handle it yet or are gaining support only very recently (e.g. valgrind))
> and -grecord-gcc-switches which is IMHO worth the few extra bytes per CU
> (unless every CU is compiled with different code generation affecting
> options usually just 4 extra bytes).  In Fedora we default to this combo
> already for some time.  Targets that have tools that are upset by
> any extensions that defaulted to -gno-strict-dwarf previously will now
> default to -gdwarf-2 as before.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
>
> 2012-04-25  Jakub Jelinek  <ja...@redhat.com>
>
>        * common.opt (flag_debug_types_section): Default to 0.
>        (dwarf_version): Default to 4.
>        (dwarf_record_gcc_switches): Default to 1.
>        (dwarf_strict): Default to 0.
>        * toplev.c (process_options): Don't handle dwarf_strict
>        or dwarf_version here.
>        * config/vxworks.c (vxworks_override_options): Don't
>        test whether dwarf_strict or dwarf_version are negative,
>        instead test !global_options_set.x_dwarf_*.
>        * config/darwin.c (darwin_override_options): Default to
>        dwarf_version 2.
>        * doc/invoke.texi: Note that -gdwarf-4, -grecord-gcc-switches
>        and -fno-debug-types-section are now the default.
>

This caused:

FAIL: gcc.dg/pch/save-temps-1.c  -O0 -g assembly comparison
FAIL: gcc.dg/pch/save-temps-1.c   -O3 -g  assembly comparison

on Linux/x86-64 with --target_board='unix{-m32}'.

-- 
H.J.

Reply via email to