From: Jakub Jelinek <jakub@redhat.com>
Date: Fri, 16 Dec 2022 09:19:22 +0000 (+0100)
Subject: loop-invariant: Split preheader edge if the preheader bb ends with jump [PR106751]
X-Git-Tag: basepoints/gcc-14~2468
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=ddcaa60983b50378bde1b7e327086fe0ce101795;p=thirdparty%2Fgcc.git

loop-invariant: Split preheader edge if the preheader bb ends with jump [PR106751]

The RTL loop passes only request simple preheaders, but don't require
fallthru preheaders, while move_invariant_reg apparently assumes the
latter, that it can just append instruction(s) to the end of the preheader
basic block.

The following patch fixes that by splitting the preheader edge if
the preheader bb ends with a JUMP_INSN (asm goto in this case).
Without that we get control flow in the middle of a bb.

2022-12-16  Jakub Jelinek  <jakub@redhat.com>

	PR rtl-optimization/106751
	* loop-invariant.cc (move_invariant_reg): If preheader bb ends
	with a JUMP_INSN, split the preheader edge and emit invariants
	into the new preheader basic block.

	* gcc.c-torture/compile/pr106751.c: New test.
---

diff --git a/gcc/loop-invariant.cc b/gcc/loop-invariant.cc
index f3248540fb15..9b5f6cd3223c 100644
--- a/gcc/loop-invariant.cc
+++ b/gcc/loop-invariant.cc
@@ -1837,6 +1837,8 @@ move_invariant_reg (class loop *loop, unsigned invno)
       else if (dump_file)
 	fprintf (dump_file, "Invariant %d moved without introducing a new "
 			    "temporary register\n", invno);
+      if (JUMP_P (BB_END (preheader)))
+	preheader = split_edge (loop_preheader_edge (loop));
       reorder_insns (inv->insn, inv->insn, BB_END (preheader));
       df_recompute_luids (preheader);
 
diff --git a/gcc/testsuite/gcc.c-torture/compile/pr106751.c b/gcc/testsuite/gcc.c-torture/compile/pr106751.c
new file mode 100644
index 000000000000..5fbf93bd332e
--- /dev/null
+++ b/gcc/testsuite/gcc.c-torture/compile/pr106751.c
@@ -0,0 +1,17 @@
+/* PR rtl-optimization/106751 */
+
+int *foo (void);
+
+void
+bar (void)
+{
+  asm goto ("" : : : : lab);
+  __builtin_unreachable ();
+lab:
+  while (1)
+    {
+      int o;
+      asm ("" : "=r" (o) : "g" (1));
+      *foo () = o;
+    }
+}