[committed] openmp: Fix handling of allocate clause on taskloop

[Jakub Jelinek]

Hi!

This patch fixes gimplification of allocate clause on taskloop - puts
allocate on inner taskloop only if there is allocate clause, because otherwise
the data sharing clauses are only on the task construct in the construct sandwich.

Bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk.

2020-10-30  Jakub Jelinek  <jakub@redhat.com>

	* gimplify.c (gimplify_scan_omp_clauses): Force
	OMP_CLAUSE_ALLOCATE_ALLOCATOR into a temporary if it is non-NULL and
	non-constant.
	(gimplify_omp_for): Only put allocate on inner taskloop if lastprivate
	for the same variable is going to be put there, and in that case
	if the OMP_CLAUSE_ALLOCATE_ALLOCATOR is non-NULL non-constant, make
	the allocator firstprivate on task.

	* c-c++-common/gomp/allocate-3.c: New test.

--- gcc/gimplify.c.jj	2020-10-28 10:37:50.488608373 +0100
+++ gcc/gimplify.c	2020-10-29 20:05:23.869378065 +0100
@@ -9721,6 +9721,13 @@ gimplify_scan_omp_clauses (tree *list_p,
 	      remove = true;
 	      break;
 	    }
+	  else if (code == OMP_TASKLOOP
+		   && OMP_CLAUSE_ALLOCATE_ALLOCATOR (c)
+		   && (TREE_CODE (OMP_CLAUSE_ALLOCATE_ALLOCATOR (c))
+		       != INTEGER_CST))
+	    OMP_CLAUSE_ALLOCATE_ALLOCATOR (c)
+	      = get_initialized_tmp_var (OMP_CLAUSE_ALLOCATE_ALLOCATOR (c),
+					 pre_p, NULL, false);
 	  break;
 
 	case OMP_CLAUSE_DEFAULT:
@@ -12120,6 +12127,20 @@ gimplify_omp_for (tree *expr_p, gimple_s
       tree *gtask_clauses_ptr = &task_clauses;
       tree outer_for_clauses = NULL_TREE;
       tree *gforo_clauses_ptr = &outer_for_clauses;
+      bitmap lastprivate_uids = NULL;
+      if (omp_find_clause (c, OMP_CLAUSE_ALLOCATE))
+	{
+	  c = omp_find_clause (c, OMP_CLAUSE_LASTPRIVATE);
+	  if (c)
+	    {
+	      lastprivate_uids = BITMAP_ALLOC (NULL);
+	      for (; c; c = omp_find_clause (OMP_CLAUSE_CHAIN (c),
+					     OMP_CLAUSE_LASTPRIVATE))
+		bitmap_set_bit (lastprivate_uids,
+				DECL_UID (OMP_CLAUSE_DECL (c)));
+	    }
+	  c = *gfor_clauses_ptr;
+	}
       for (; c; c = OMP_CLAUSE_CHAIN (c))
 	switch (OMP_CLAUSE_CODE (c))
 	  {
@@ -12207,12 +12228,35 @@ gimplify_omp_for (tree *expr_p, gimple_s
 	    gtask_clauses_ptr
 	      = &OMP_CLAUSE_CHAIN (*gtask_clauses_ptr);
 	    break;
-	  /* Allocate clause we duplicate on task and inner taskloop.  */
+	  /* Allocate clause we duplicate on task and inner taskloop
+	     if the decl is lastprivate, otherwise just put on task.  */
 	  case OMP_CLAUSE_ALLOCATE:
-	    *gfor_clauses_ptr = c;
-	    gfor_clauses_ptr = &OMP_CLAUSE_CHAIN (c);
-	    *gtask_clauses_ptr = copy_node (c);
-	    gtask_clauses_ptr = &OMP_CLAUSE_CHAIN (*gtask_clauses_ptr);
+	    if (lastprivate_uids
+		&& bitmap_bit_p (lastprivate_uids,
+				 DECL_UID (OMP_CLAUSE_DECL (c))))
+	      {
+		if (OMP_CLAUSE_ALLOCATE_ALLOCATOR (c)
+		    && DECL_P (OMP_CLAUSE_ALLOCATE_ALLOCATOR (c)))
+		  {
+		    /* Additionally, put firstprivate clause on task
+		       for the allocator if it is not constant.  */
+		    *gtask_clauses_ptr
+		      = build_omp_clause (OMP_CLAUSE_LOCATION (c),
+					  OMP_CLAUSE_FIRSTPRIVATE);
+		    OMP_CLAUSE_DECL (*gtask_clauses_ptr)
+		      = OMP_CLAUSE_ALLOCATE_ALLOCATOR (c);
+		    gtask_clauses_ptr = &OMP_CLAUSE_CHAIN (*gtask_clauses_ptr);
+		  }
+		*gfor_clauses_ptr = c;
+		gfor_clauses_ptr = &OMP_CLAUSE_CHAIN (c);
+		*gtask_clauses_ptr = copy_node (c);
+		gtask_clauses_ptr = &OMP_CLAUSE_CHAIN (*gtask_clauses_ptr);
+	      }
+	    else
+	      {
+		*gtask_clauses_ptr = c;
+		gtask_clauses_ptr = &OMP_CLAUSE_CHAIN (c);
+	      }
 	    break;
 	  default:
 	    gcc_unreachable ();
@@ -12220,6 +12264,7 @@ gimplify_omp_for (tree *expr_p, gimple_s
       *gfor_clauses_ptr = NULL_TREE;
       *gtask_clauses_ptr = NULL_TREE;
       *gforo_clauses_ptr = NULL_TREE;
+      BITMAP_FREE (lastprivate_uids);
       g = gimple_build_bind (NULL_TREE, gfor, NULL_TREE);
       g = gimple_build_omp_task (g, task_clauses, NULL_TREE, NULL_TREE,
 				 NULL_TREE, NULL_TREE, NULL_TREE);
--- gcc/testsuite/c-c++-common/gomp/allocate-3.c.jj	2020-10-29 20:12:58.675241433 +0100
+++ gcc/testsuite/c-c++-common/gomp/allocate-3.c	2020-10-29 18:47:45.400924987 +0100
@@ -0,0 +1,38 @@
+typedef enum omp_allocator_handle_t
+#if __cplusplus >= 201103L
+: __UINTPTR_TYPE__
+#endif
+{
+  omp_null_allocator = 0,
+  omp_default_mem_alloc = 1,
+  omp_large_cap_mem_alloc = 2,
+  omp_const_mem_alloc = 3,
+  omp_high_bw_mem_alloc = 4,
+  omp_low_lat_mem_alloc = 5,
+  omp_cgroup_mem_alloc = 6,
+  omp_pteam_mem_alloc = 7,
+  omp_thread_mem_alloc = 8,
+  __omp_allocator_handle_t_max__ = __UINTPTR_MAX__
+} omp_allocator_handle_t;
+
+omp_allocator_handle_t baz (int);
+
+int
+foo (omp_allocator_handle_t h1, omp_allocator_handle_t h2, int y)
+{
+  int x;
+  #pragma omp taskloop default(none) lastprivate (x) allocate (h1:x) firstprivate(y) allocate (h2:y)
+  for (int i = 0; i < 64; i++)
+    x = y + i;
+  return x;
+}
+
+int
+bar (int y)
+{
+  int x;
+  #pragma omp taskloop default(none) lastprivate (x) allocate (baz (0):x) allocate (baz (1):y) firstprivate(y)
+  for (int i = 0; i < 64; i++)
+    x = y + i;
+  return x;
+}

	Jakub