[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: gcc STILL broken

On Fri, Apr 30, 1999 at 11:31:27PM -0500, John Goerzen wrote:
> Hi,
> gcc with -O2 can still not properly compile Tcl -- it generates
> invalid code, that manifests itself with zircon.  
> When's our compiler going to be fixed?  Knowing that -O2 produces
> invalid code makes me nervous about the stability of our dist.
  Please try to pinpoint it to an compiler option and a sample source-code.
  Then submit a bug to egcs. Only specific problems make it to solutions :)

  And for the stability... I haven't had problems since my switch some weeks
  ago (right after release of slink).

optimize 1
      flag_defer_pop                    -fdefer-pop
      flag_thread_jumps                 -fthread-jumps
      flag_delayed_branch               -fdelayed-branch
      flag_omit_frame_pointer           -fomit-frame-pointer

optimize 2
      flag_cse_follow_jumps             -fcse-follow-jumps
      flag_cse_skip_blocks              -fcse-skip-blocks
      flag_gcse                         -fgcse
      flag_expensive_optimizations      -fexpensive-optimizations
      flag_strength_reduce              -fstrength-reduce
      flag_rerun_cse_after_loop         -frerun-cse-after-loop
      flag_rerun_loop_opt               -frerun-loop-opt
      flag_caller_saves                 -fcaller-saves
      flag_force_mem                    -fforce-mem
      flag_schedule_insns               -fschedule-insns
      flag_schedule_insns_after_reload  -fschedule-insns2

optimize 3
      flag_inline_functions             -finline-functions

  -O2 == optimize 1 + optimize 2

B. Warmerdam                                              GNU/Debian Linux
bartw@xs4all.nl, bartw@debian.org (Keyid: 10A0FDD1)       ----------------

Reply to: