public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [COMMITTED] Make sure exported range for SSA post-dominates the DEF in set_global_ranges_from_unreachable_edges.
@ 2022-10-17 18:56 Aldy Hernandez
  0 siblings, 0 replies; only message in thread
From: Aldy Hernandez @ 2022-10-17 18:56 UTC (permalink / raw)
  To: GCC patches; +Cc: Andrew MacLeod, Aldy Hernandez

The problem here is that we're exporting a range for an SSA range that
happens on the other side of a __builtin_unreachable, but the SSA does
not post-dominate the definition point.  This is causing ivcanon to
unroll things incorrectly.

This was a snafu when converting the code from evrp.

	PR tree-optimization/107293

gcc/ChangeLog:

	* tree-ssa-dom.cc
	(dom_opt_dom_walker::set_global_ranges_from_unreachable_edges):
	Check that condition post-dominates the definition point.

gcc/testsuite/ChangeLog:

	* gcc.dg/tree-ssa/pr107293.c: New test.
---
 gcc/testsuite/gcc.dg/tree-ssa/pr107293.c | 32 ++++++++++++++++++++++++
 gcc/tree-ssa-dom.cc                      |  6 ++++-
 2 files changed, 37 insertions(+), 1 deletion(-)
 create mode 100644 gcc/testsuite/gcc.dg/tree-ssa/pr107293.c

diff --git a/gcc/testsuite/gcc.dg/tree-ssa/pr107293.c b/gcc/testsuite/gcc.dg/tree-ssa/pr107293.c
new file mode 100644
index 00000000000..724c31a11e6
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/pr107293.c
@@ -0,0 +1,32 @@
+// { dg-do run }
+// { dg-options "-w -Os" }
+
+short a;
+int b[1];
+
+int c(int p) {
+  return (p < 0) ? 0 : 10 + ((p / 100 - 16) / 4);
+}
+
+void f(int n) {
+  while (1) {
+    int m = n;
+    while ((m ) )
+      m /= 2;
+    break;
+  }
+}
+
+void g() {
+  int h = a = 0;
+  for (; h + a <= 0; a++) {
+    if (b[c(a - 6)])
+      break;
+    f(a);
+  }
+}
+int main() {
+  g();
+  if (a != 1)
+    __builtin_abort ();
+}
diff --git a/gcc/tree-ssa-dom.cc b/gcc/tree-ssa-dom.cc
index e6b8dace5e9..c7f095d79fc 100644
--- a/gcc/tree-ssa-dom.cc
+++ b/gcc/tree-ssa-dom.cc
@@ -1367,7 +1367,11 @@ dom_opt_dom_walker::set_global_ranges_from_unreachable_edges (basic_block bb)
   tree name;
   gori_compute &gori = m_ranger->gori ();
   FOR_EACH_GORI_EXPORT_NAME (gori, pred_e->src, name)
-    if (all_uses_feed_or_dominated_by_stmt (name, stmt))
+    if (all_uses_feed_or_dominated_by_stmt (name, stmt)
+	// The condition must post-dominate the definition point.
+	&& (SSA_NAME_IS_DEFAULT_DEF (name)
+	    || (gimple_bb (SSA_NAME_DEF_STMT (name))
+		== pred_e->src)))
       {
 	Value_Range r (TREE_TYPE (name));
 
-- 
2.37.3


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2022-10-17 18:56 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-17 18:56 [COMMITTED] Make sure exported range for SSA post-dominates the DEF in set_global_ranges_from_unreachable_edges Aldy Hernandez

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).