c/5830: The compiler crashes

Gwenole Beauchesne gbeauchesne@mandrakesoft.com
Fri Mar 8 10:36:00 GMT 2002


The following reply was made to PR c/5830; it has been noted by GNATS.

From: Gwenole Beauchesne <gbeauchesne@mandrakesoft.com>
To: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org, Manuel.Serrano@inria.fr,
        gcc-bugs@gcc.gnu.org, nobody@gcc.gnu.org
Cc: jh@suse.cz
Subject: Re: c/5830: The compiler crashes
Date: Fri, 08 Mar 2002 19:29:30 +0100

 Hi,
 
 It doesn't appear to be fixed on the gcc-3.0-branch. On gcc-3.1-branch, 
 that appears to have been fixed with Jan's crossjumping patch of 2001/07/16:
 <http://gcc.gnu.org/ml/gcc-patches/2001-07/msg01030.html>
 
 On 3.0-branch, gcc ICEs with the testcase of that PR and -O2 level. 
 However, it doesn't ICE with -O2 and -fno-reorder-blocks.
 
 With the following patch, derived from Jan's, 3.0-branch bootstrapped 
 and regtested OK. Note that if I make cleanup_cfg() after 
 reorder_basic_blocks() too, bootstrap fails. So I don't think that patch 
 is good enough.
 
 2002-03-08  Gwenole Beauchesne  <gbeauchesne@mandrakesoft.com>
    
     PR c/5830
     * toplev.c (rest_of_compilation): Do cleanup_cfg before
     bb-reorder. [From Jan Hubicka for gcc-3.1, 2001/07/16]
 
 --- gcc-3.0.4/gcc/toplev.c.reorder-blocks    Fri Mar  8 14:10:00 2002
 +++ gcc-3.0.4/gcc/toplev.c    Fri Mar  8 15:45:46 2002
 @@ -3636,12 +3636,16 @@
      = optimize > 0 && only_leaf_regs_used () && leaf_function_p ();
  #endif
  
 -  if (optimize > 0 && flag_reorder_blocks)
 +  if (optimize > 0)
      {
        timevar_push (TV_REORDER_BLOCKS);
        open_dump_file (DFI_bbro, decl);
 -
 -      reorder_basic_blocks ();
 +     
 +      /* Last attempt to optimize CFG, as life analyzis possibly removed
 +         some instructions.  */
 +      cleanup_cfg ();
 +      if (flag_reorder_blocks)
 +          reorder_basic_blocks ();
  
        close_dump_file (DFI_bbro, print_rtl_with_bb, insns);
        timevar_pop (TV_REORDER_BLOCKS);
 
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5830 
 
 



More information about the Gcc-prs mailing list