Optimize _CALL_LEN

Author: tomasr8

Lib/test/test_capi/test_opt.py

@@ -1953,6 +1953,61 @@ def testfunc(n):
         self.assertNotIn("_GUARD_NOS_INT", uops)
         self.assertNotIn("_GUARD_TOS_INT", uops)
 
+    def test_call_len_known_length_small_int(self):
+        def testfunc(n):
+            x = 0
+            for _ in range(n):
+                a = (1, 2, 3, 4, 5)
+                if len(a) == 5:
+                    x += 1
+            return x
+
+        res, ex = self._run_with_optimizer(testfunc, TIER2_THRESHOLD)
+        self.assertEqual(res, TIER2_THRESHOLD)
+        self.assertIsNotNone(ex)
+        uops = get_opnames(ex)
+        # When the length is < _PY_NSMALLPOSINTS, the len() call is replaced
+        # with just an inline load.
+        self.assertNotIn("_CALL_LEN", uops)
+        self.assertNotIn("_POP_CALL_ONE_LOAD_CONST_INLINE_BORROW", uops)
+        self.assertNotIn("_POP_CALL_LOAD_CONST_INLINE_BORROW", uops)
+        self.assertNotIn("_POP_TOP_LOAD_CONST_INLINE_BORROW", uops)
+
+    def test_call_len_known_length(self):
+        def testfunc(n):
+            x = 0
+            for _ in range(n):
+                a = (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1)
+                if len(a) == 300:  # comparison + guard removed
+                    x += 1
+            return x
+
+        res, ex = self._run_with_optimizer(testfunc, TIER2_THRESHOLD)
+        self.assertEqual(res, TIER2_THRESHOLD)
+        self.assertIsNotNone(ex)
+        uops = get_opnames(ex)
+        # When the length is >= _PY_NSMALLPOSINTS, we cannot replace
+        # the len() call with an inline load, but knowing the exact
+        # length allows us to optimize more code, such as conditionals
+        # in this case
+        self.assertIn("_CALL_LEN", uops)
+        self.assertNotIn("_COMPARE_OP_INT", uops)
+        self.assertNotIn("_GUARD_IS_TRUE_POP", uops)
+
     def test_get_len_with_const_tuple(self):
         def testfunc(n):
             x = 0.0

Python/optimizer_bytecodes.c

@@ -1190,8 +1190,21 @@ dummy_func(void) {
         sym_set_const(callable, (PyObject *)&PyUnicode_Type);
     }
 
-    op(_CALL_LEN, (unused, unused, unused -- res)) {
+    op(_CALL_LEN, (callable, null, arg -- res)) {
         res = sym_new_type(ctx, &PyLong_Type);
+        int tuple_length = sym_tuple_length(arg);
+        if (tuple_length >= 0) {
+            PyObject *temp = PyLong_FromLong(tuple_length);
+            if (temp == NULL) {
+                goto error;
+            }
+            if (_Py_IsImmortal(temp)) {
+                REPLACE_OP(this_instr, _POP_CALL_ONE_LOAD_CONST_INLINE_BORROW,
+                           0, (uintptr_t)temp);
+            }
+            res = sym_new_const(ctx, temp);
+            Py_DECREF(temp);
+        }
     }
 
     op(_GET_LEN, (obj -- obj, len)) {