middle-end/8502: [3.2 regression] wrong code generated for switch statement

[Richard Henderson]

The following reply was made to PR middle-end/8502; it has been noted by GNATS.

From: Richard Henderson <rth@redhat.com>
To: Volker Reichelt <reichelt@igpm.rwth-aachen.de>
Cc: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org, rveldema@cs.vu.nl,
   gcc-patches@gcc.gnu.org
Subject: Re: middle-end/8502: [3.2 regression] wrong code generated for switch statement
Date: Thu, 27 Mar 2003 14:28:07 -0800

 On Tue, Mar 25, 2003 at 02:05:43PM +0100, Volker Reichelt wrote:
 > I strongly suspect your patch
 > 
 >    http://gcc.gnu.org/ml/gcc-cvs/2003-03/msg01201.html
 > 
 > is responsible for the new failure of this PR because it modifies
 > cfgrtl.c where the original failure was fixed.
 
 Blah.  Test case broke again only at -O0.  Fixed thus.
 
 
 r~
 
         * cfgrtl.c (try_redirect_by_replacing_jump): Disable removing
         tablejumps with no optimization.
 
 Index: cfgrtl.c
 ===================================================================
 RCS file: /cvs/gcc/gcc/gcc/cfgrtl.c,v
 retrieving revision 1.29.2.6.2.2
 diff -c -p -d -u -r1.29.2.6.2.2 cfgrtl.c
 --- cfgrtl.c	23 Mar 2003 20:41:22 -0000	1.29.2.6.2.2
 +++ cfgrtl.c	27 Mar 2003 22:23:00 -0000
 @@ -665,7 +665,6 @@ try_redirect_by_replacing_jump (e, targe
    edge tmp;
    rtx set;
    int fallthru = 0;
 -  rtx table;
  
    /* Verify that all targets will be TARGET.  */
    for (tmp = src->succ; tmp; tmp = tmp->succ_next)
 @@ -675,11 +674,7 @@ try_redirect_by_replacing_jump (e, targe
    if (tmp || !onlyjump_p (insn))
      return false;
  
 -  if (flow2_completed && JUMP_LABEL (insn)
 -      && (table = NEXT_INSN (JUMP_LABEL (insn))) != NULL_RTX
 -      && GET_CODE (table) == JUMP_INSN
 -      && (GET_CODE (PATTERN (table)) == ADDR_VEC
 -	  || GET_CODE (PATTERN (table)) == ADDR_DIFF_VEC))
 +  if ((!optimize || flow2_completed) && tablejump_p (insn))
      return false;
  
    /* Avoid removing branch with side effects.  */