This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[PATCH] Fix regressed patch





The following patch is an alternative way of fixing the problem
http://gcc.gnu.org/ml/gcc-patches/2004-01/msg03236.html addressed, which
was inadvertently lost when tree-ssa was merged into mainline (see
discussion in thread http://gcc.gnu.org/ml/gcc/2004-05/msg01281.html).  I'm
proposing the alternative since it is making use of a function that was
already availabe instead of adding new code.

Bootstrapped/regtested on PowerPC.

OK for mainline?

-Pat


2004-05-28  Pat Haugen  <pthaugen@us.ibm.com>

      * cfghooks.c (split_block): Call make_single_succ_edge so that edge
      count/probability are set correctly.

Index: gcc/cfghooks.c
===================================================================
RCS file: /cvsroot/gcc/gcc/gcc/cfghooks.c,v
retrieving revision 1.11
diff -c -3 -p -r1.11 cfghooks.c
*** gcc/cfghooks.c      13 May 2004 06:39:32 -0000    1.11
--- gcc/cfghooks.c      28 May 2004 17:25:27 -0000
*************** split_block (basic_block bb, void *i)
*** 321,327 ****
        set_immediate_dominator (CDI_DOMINATORS, new_bb, bb);
      }

!   return make_edge (bb, new_bb, EDGE_FALLTHRU);
  }

  /* Splits block BB just after labels.  The newly created edge is returned.  */
--- 321,327 ----
        set_immediate_dominator (CDI_DOMINATORS, new_bb, bb);
      }

!   return make_single_succ_edge (bb, new_bb, EDGE_FALLTHRU);
  }

  /* Splits block BB just after labels.  The newly created edge is returned.  */


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]