gh-100239: more refined specialisation stats for BINARY_OP/SUBSCR (#132068)

iritkatriel · web-flow · commit df59226997e0 · 2025-04-04T15:33:31.000+01:00
diff --git a/Include/cpython/pystats.h b/Include/cpython/pystats.h
@@ -31,7 +31,7 @@
 
 #define PYSTATS_MAX_UOP_ID 512
 
-#define SPECIALIZATION_FAILURE_KINDS 37
+#define SPECIALIZATION_FAILURE_KINDS 44
 
 /* Stats for determining who is calling PyEval_EvalFrame */
 #define EVAL_CALL_TOTAL 0
diff --git a/Python/specialize.c b/Python/specialize.c
@@ -597,6 +597,13 @@ _PyCode_Quicken(_Py_CODEUNIT *instructions, Py_ssize_t size, int enable_counters
 #define SPEC_FAIL_BINARY_OP_SUBSCR_TUPLE_SLICE          35
 #define SPEC_FAIL_BINARY_OP_SUBSCR_STRING_SLICE         36
 #define SPEC_FAIL_BINARY_OP_SUBSCR_NOT_HEAP_TYPE        37
+#define SPEC_FAIL_BINARY_OP_SUBSCR_OTHER_SLICE          38
+#define SPEC_FAIL_BINARY_OP_SUBSCR_MAPPINGPROXY         39
+#define SPEC_FAIL_BINARY_OP_SUBSCR_RE_MATCH             40
+#define SPEC_FAIL_BINARY_OP_SUBSCR_ARRAY                41
+#define SPEC_FAIL_BINARY_OP_SUBSCR_DEQUE                42
+#define SPEC_FAIL_BINARY_OP_SUBSCR_ENUMDICT             43
+#define SPEC_FAIL_BINARY_OP_SUBSCR_STACKSUMMARY         44
 
 /* Calls */
 
@@ -2358,6 +2365,34 @@ binary_op_fail_kind(int oparg, PyObject *lhs, PyObject *rhs)
                 }
             }
             Py_XDECREF(descriptor);
+
+            if (PyObject_TypeCheck(lhs, &PyDictProxy_Type)) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_MAPPINGPROXY;
+            }
+
+            if (strcmp(container_type->tp_name, "array.array") == 0) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_ARRAY;
+            }
+
+            if (strcmp(container_type->tp_name, "re.Match") == 0) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_RE_MATCH;
+            }
+
+            if (strcmp(container_type->tp_name, "collections.deque") == 0) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_DEQUE;
+            }
+
+            if (strcmp(_PyType_Name(container_type), "EnumDict") != 0) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_ENUMDICT;
+            }
+
+            if (strcmp(container_type->tp_name, "StackSummary") != 0) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_STACKSUMMARY;
+            }
+
+            if (PySlice_Check(rhs)) {
+                return SPEC_FAIL_BINARY_OP_SUBSCR_OTHER_SLICE;
+            }
             return SPEC_FAIL_BINARY_OP_SUBSCR;
     }
     Py_UNREACHABLE();
diff --git a/Tools/scripts/summarize_stats.py b/Tools/scripts/summarize_stats.py
@@ -298,12 +298,20 @@ def kind_to_text(kind: int, opcode: str):
             return "kind " + str(kind)
 
         family_stats = self._get_stats_for_opcode(opcode)
-        failure_kinds = [0] * 40
+
+        def key_to_index(key):
+            return int(key[:-1].split("[")[1])
+
+        max_index = 0
+        for key in family_stats:
+            if key.startswith("specialization.failure_kind"):
+                max_index = max(max_index, key_to_index(key))
+
+        failure_kinds = [0] * (max_index + 1)
         for key in family_stats:
             if not key.startswith("specialization.failure_kind"):
                 continue
-            index = int(key[:-1].split("[")[1])
-            failure_kinds[index] = family_stats[key]
+            failure_kinds[key_to_index(key)] = family_stats[key]
         return {
             kind_to_text(index, opcode): value
             for (index, value) in enumerate(failure_kinds)
