cause of all the mainline Sparc regressions

[Richard Henderson]

On Thu, May 23, 2002 at 07:02:35PM +0200, Jan Hubicka wrote:
> delete_unrechable_blocks is now unconditonally called before doing the
> INSN_NOTE_PROBABILITY lowering pass, so there is no need to re-cleanup
> the cfg when eh code is invoked I blieve.

Hmm, I see.  You may be right that we don't need any cleanup_cfg
calls in there at all any more.

If someone has some cycles to spare, it'd be good to see what
the following patch does to the size of libstdc++ and libjava.
The theory is that nothing will change.


r~



Index: except.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/except.c,v
retrieving revision 1.219
diff -c -p -d -u -r1.219 except.c
--- except.c	17 Apr 2002 01:52:37 -0000	1.219
+++ except.c	23 May 2002 17:30:11 -0000
@@ -2499,10 +2499,6 @@ finish_eh_generation ()
      connect many of the handlers, and then type information will not
      be effective.  Still, this is a win over previous implementations.  */
 
-  rebuild_jump_labels (get_insns ());
-  find_basic_blocks (get_insns (), max_reg_num (), 0);
-  cleanup_cfg (CLEANUP_PRE_LOOP);
-
   /* These registers are used by the landing pads.  Make sure they
      have been generated.  */
   get_exception_pointer (cfun);
@@ -2524,7 +2520,6 @@ finish_eh_generation ()
   find_exception_handler_labels ();
   rebuild_jump_labels (get_insns ());
   find_basic_blocks (get_insns (), max_reg_num (), 0);
-  cleanup_cfg (CLEANUP_PRE_LOOP);
 }
 
 static hashval_t