Hi!
The following testcase ICEs, because for parallel combined with some
other construct we initialize the omp_parallel_combined_clauses pointer
and expect the construct combined with it to clear it after it no longer
needs it, but OMP_MASTER didn't do that.
Bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk,
queued for backporting.
2020-04-04 Jakub Jelinek <ja...@redhat.com>
PR c++/94477
* pt.c (tsubst_expr) <case OMP_MASTER>: Clear
omp_parallel_combined_clauses.
* g++.dg/gomp/pr94477.C: New test.
--- gcc/cp/pt.c.jj 2020-04-03 20:28:50.989754551 +0200
+++ gcc/cp/pt.c 2020-04-03 21:54:29.754337004 +0200
@@ -18467,8 +18467,10 @@ tsubst_expr (tree t, tree args, tsubst_f
add_stmt (t);
break;
- case OMP_SECTION:
case OMP_MASTER:
+ omp_parallel_combined_clauses = NULL;
+ /* FALLTHRU */
+ case OMP_SECTION:
stmt = push_stmt_list ();
RECUR (OMP_BODY (t));
stmt = pop_stmt_list (stmt);
--- gcc/testsuite/g++.dg/gomp/pr94477.C.jj 2020-04-03 21:55:10.575740787
+0200
+++ gcc/testsuite/g++.dg/gomp/pr94477.C 2020-04-03 21:54:52.388006422 +0200
@@ -0,0 +1,18 @@
+// PR c++/94477
+// { dg-do compile }
+
+void foo ();
+
+template <int>
+void
+bar ()
+{
+ #pragma omp parallel master
+ foo ();
+}
+
+void
+baz ()
+{
+ bar<0> ();
+}
Jakub
