tree-optimization/104543 - fix unroll-and-jam precondition

We have to make sure that outer loop exits come after the inner
loop since we otherwise will put it into the fused loop body.

2022-02-15  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/104543
	* gimple-loop-jam.cc (unroll_jam_possible_p): Check outer loop exits
	come after the inner loop.

	* gcc.dg/torture/pr104543.c: New testcase.

diff --git a/gcc/gimple-loop-jam.cc b/gcc/gimple-loop-jam.cc
index d9a7772..e33dd90 100644 (file)
--- a/gcc/gimple-loop-jam.cc
+++ b/gcc/gimple-loop-jam.cc
@@ -198,7 +198,8 @@ unroll_jam_possible_p (class loop *outer, class loop *loop)
   if (!empty_block_p (loop->latch))
     return false;
 
-  if (!single_exit (loop))
+  edge exit;
+  if (!(exit = single_exit (loop)))
     return false;
 
   /* We need a perfect nest.  Quick check for adjacent inner loops.  */
@@ -259,7 +260,12 @@ unroll_jam_possible_p (class loop *outer, class loop *loop)
   n = get_loop_body_with_size (outer, bbs, n_basic_blocks_for_fn (cfun));
 
   for (i = 0; i < n; i++)
-    if (bbs[i]->loop_father == outer && bb_prevents_fusion_p (bbs[i]))
+    if (bbs[i]->loop_father == outer
+       && (bb_prevents_fusion_p (bbs[i])
+           /* Outer loop exits must come after the inner loop, otherwise
+              we'll put the outer loop exit into the fused inner loop.  */
+           || (loop_exits_from_bb_p (outer, bbs[i])
+               && !dominated_by_p (CDI_DOMINATORS, bbs[i], exit->src))))
       break;
   free (bbs);
   if (i != n)

diff --git a/gcc/testsuite/gcc.dg/torture/pr104543.c b/gcc/testsuite/gcc.dg/torture/pr104543.c
new file mode 100644 (file)
index 0000000..4896351
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/torture/pr104543.c
@@ -0,0 +1,21 @@
+/* { dg-do run } */
+/* { dg-additional-options "-floop-unroll-and-jam" } */
+
+int a[3], b, c;
+static int e()
+{
+  if (!c) {
+    for (b = 0; b < 3; b++)
+      for (c = 0; c < 3; c++)
+        a[c] ^= 1;
+    return -1;
+  }
+  return 0;
+}
+int main()
+{
+  e();
+  if (a[1] != 1)
+    __builtin_abort();
+  return 0;
+}