Re: [PATCH] Don't bypass blocks with multiple latch edges (PR middle-end/54838)

[Marek Polacek]

On Thu, Nov 29, 2012 at 09:34:31AM +0100, Richard Biener wrote:
> Definitely not - that means to not preserve loops until after cprop.  The goal
> is to preserve loops everywhere!

Yikes, sorry, it wasn't clear to me what PROP_loops really does.  Anyway,
I think I have a better fix now.  The problem is just that when removing
BB 4 (which was a header), we have to zap ->header and ->latch.  We
already have code for this:

  if (current_loops != NULL
      && e->src->loop_father->latch == e->src)
    {
      /* ???  Now we are creating (or may create) a loop
         with multiple entries.  Simply mark it for
         removal.  Alternatively we could not do this
         threading.  */
      e->src->loop_father->header = NULL;
      e->src->loop_father->latch = NULL;
    }

but the thing is that when there are multiple latch edges, then
->latch is NULL.  So we need to keep track of how many latch edges
the header has.  Regtested/bootstrapped on x86_64, ok for trunk?

Can I get rid of may_be_loop_header (and just use n_latch_edges > 0
instead at that one place) in a followup?

2012-11-29  Marek Polacek  <pola...@redhat.com>

        PR middle-end/54838
        * cprop.c (bypass_block): Set header and latch to NULL when
        BB has more than one latch edge.
        (n_latches): New variable.

        * gcc.dg/pr54838.c: New test.

--- gcc/cprop.c.mp      2012-11-29 15:49:53.120524295 +0100
+++ gcc/cprop.c 2012-11-29 15:50:01.421547832 +0100
@@ -1499,6 +1499,7 @@ bypass_block (basic_block bb, rtx setcc,
   int may_be_loop_header;
   unsigned removed_p;
   unsigned i;
+  unsigned n_latch_edges;
   edge_iterator ei;
 
   insn = (setcc != NULL) ? setcc : jump;
@@ -1510,13 +1511,12 @@ bypass_block (basic_block bb, rtx setcc,
   if (note)
     find_used_regs (&XEXP (note, 0), NULL);
 
-  may_be_loop_header = false;
+  n_latch_edges = 0;
   FOR_EACH_EDGE (e, ei, bb->preds)
     if (e->flags & EDGE_DFS_BACK)
-      {
-       may_be_loop_header = true;
-       break;
-      }
+      n_latch_edges++;
+
+  may_be_loop_header = n_latch_edges > 0;
 
   change = 0;
   for (ei = ei_start (bb->preds); (e = ei_safe_edge (ei)); )
@@ -1605,7 +1605,8 @@ bypass_block (basic_block bb, rtx setcc,
              && dest != EXIT_BLOCK_PTR)
             {
              if (current_loops != NULL
-                 && e->src->loop_father->latch == e->src)
+                 && (e->src->loop_father->latch == e->src
+                     || n_latch_edges > 1))
                {
                  /* ???  Now we are creating (or may create) a loop
                     with multiple entries.  Simply mark it for
--- gcc/testsuite/gcc.dg/pr54838.c.mp   2012-11-26 14:48:43.783980854 +0100
+++ gcc/testsuite/gcc.dg/pr54838.c      2012-11-26 14:49:51.051158719 +0100
@@ -0,0 +1,24 @@
+/* PR middle-end/54838 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -fno-forward-propagate -ftracer" } */
+
+void bar (void);
+
+void
+foo (void *b, int *c)
+{
+again:
+  switch (*c)
+    {
+    case 1:
+      if (!b)
+       {
+         bar ();
+         return;
+       }
+      goto again;
+    case 3:
+      if (!b)
+       goto again;
+    }
+}

        Marek