On Fri, Mar 4, 2022 at 2:12 PM Martin Liška <mli...@suse.cz> wrote: > > On 3/1/22 09:48, Richard Biener wrote: > > I think moving flag_gtoggle handling before the flag_syntax_only handling > > is a good thing. But I don't quite understand the flag_var_tracking > > disabling > > or how it worked before. > > Well, as you know, the debugging options are a can of worms. During GCC 12 > development I moved > most of the option logic to finish_options which is a place that is used both > for command line > option processing and optimize/target pragma/attribute processing. > > That's why we see this problem. OPT_LEVELS_1_PLUS enables flag_var_tracking > but we have to drop > debug debug_info_level == DINFO_LEVEL_NONE. > > > At least I think you want to check for > > debug_info_level == NONE, no? Why should DINFO_LEVEL_TERSE be > > special? > > No, sending updated version of the patch. > > Patch can bootstrap on x86_64-linux-gnu and survives regression tests. > > Ready to be installed?
OK. > Thanks, > Martin