gh-104909: Split some more insts into ops (#109943)

These are the most popular specializations of `LOAD_ATTR` and `STORE_ATTR`
that weren't already viable uops:

* Split LOAD_ATTR_METHOD_WITH_VALUES
* Split LOAD_ATTR_METHOD_NO_DICT
* Split LOAD_ATTR_SLOT
* Split STORE_ATTR_SLOT
* Split STORE_ATTR_INSTANCE_VALUE

Also:

* Add `-v` flag to code generator which prints a list of non-viable uops
  (easter-egg: it can print execution counts -- see source)
* Double _Py_UOP_MAX_TRACE_LENGTH to 128



I had dropped one of the DEOPT_IF() calls! :-(

Python/executor_cases.c.h

@@ -1714,6 +1714,83 @@
             break;
         }
 
+        case _LOAD_ATTR_SLOT: {
+            PyObject *owner;
+            PyObject *attr;
+            PyObject *null = NULL;
+            owner = stack_pointer[-1];
+            uint16_t index = (uint16_t)operand;
+            char *addr = (char *)owner + index;
+            attr = *(PyObject **)addr;
+            DEOPT_IF(attr == NULL, LOAD_ATTR);
+            STAT_INC(LOAD_ATTR, hit);
+            Py_INCREF(attr);
+            null = NULL;
+            Py_DECREF(owner);
+            STACK_GROW(((oparg & 1) ? 1 : 0));
+            stack_pointer[-1 - (oparg & 1 ? 1 : 0)] = attr;
+            if (oparg & 1) { stack_pointer[-(oparg & 1 ? 1 : 0)] = null; }
+            break;
+        }
+
+        case _GUARD_DORV_VALUES: {
+            PyObject *owner;
+            owner = stack_pointer[-1];
+            PyTypeObject *tp = Py_TYPE(owner);
+            assert(tp->tp_flags & Py_TPFLAGS_MANAGED_DICT);
+            PyDictOrValues dorv = *_PyObject_DictOrValuesPointer(owner);
+            DEOPT_IF(!_PyDictOrValues_IsValues(dorv), STORE_ATTR);
+            break;
+        }
+
+        case _STORE_ATTR_INSTANCE_VALUE: {
+            PyObject *owner;
+            PyObject *value;
+            owner = stack_pointer[-1];
+            value = stack_pointer[-2];
+            uint16_t index = (uint16_t)operand;
+            PyDictOrValues dorv = *_PyObject_DictOrValuesPointer(owner);
+            STAT_INC(STORE_ATTR, hit);
+            PyDictValues *values = _PyDictOrValues_GetValues(dorv);
+            PyObject *old_value = values->values[index];
+            values->values[index] = value;
+            if (old_value == NULL) {
+                _PyDictValues_AddToInsertionOrder(values, index);
+            }
+            else {
+                Py_DECREF(old_value);
+            }
+            Py_DECREF(owner);
+            STACK_SHRINK(2);
+            break;
+        }
+
+        case _GUARD_TYPE_VERSION_STORE: {
+            PyObject *owner;
+            owner = stack_pointer[-1];
+            uint32_t type_version = (uint32_t)operand;
+            PyTypeObject *tp = Py_TYPE(owner);
+            assert(type_version != 0);
+            DEOPT_IF(tp->tp_version_tag != type_version, STORE_ATTR);
+            break;
+        }
+
+        case _STORE_ATTR_SLOT: {
+            PyObject *owner;
+            PyObject *value;
+            owner = stack_pointer[-1];
+            value = stack_pointer[-2];
+            uint16_t index = (uint16_t)operand;
+            char *addr = (char *)owner + index;
+            STAT_INC(STORE_ATTR, hit);
+            PyObject *old_value = *(PyObject **)addr;
+            *(PyObject **)addr = value;
+            Py_XDECREF(old_value);
+            Py_DECREF(owner);
+            STACK_SHRINK(2);
+            break;
+        }
+
         case COMPARE_OP: {
             PyObject *right;
             PyObject *left;
@@ -2219,6 +2296,68 @@
             break;
         }
 
+        case _GUARD_DORV_VALUES_INST_ATTR_FROM_DICT: {
+            PyObject *owner;
+            owner = stack_pointer[-1];
+            PyTypeObject *owner_cls = Py_TYPE(owner);
+            assert(owner_cls->tp_flags & Py_TPFLAGS_MANAGED_DICT);
+            PyDictOrValues *dorv = _PyObject_DictOrValuesPointer(owner);
+            DEOPT_IF(!_PyDictOrValues_IsValues(*dorv) &&
+                     !_PyObject_MakeInstanceAttributesFromDict(owner, dorv),
+                     LOAD_ATTR);
+            break;
+        }
+
+        case _GUARD_KEYS_VERSION: {
+            PyObject *owner;
+            owner = stack_pointer[-1];
+            uint32_t keys_version = (uint32_t)operand;
+            PyTypeObject *owner_cls = Py_TYPE(owner);
+            PyHeapTypeObject *owner_heap_type = (PyHeapTypeObject *)owner_cls;
+            DEOPT_IF(owner_heap_type->ht_cached_keys->dk_version !=
+                     keys_version, LOAD_ATTR);
+            break;
+        }
+
+        case _LOAD_ATTR_METHOD_WITH_VALUES: {
+            PyObject *owner;
+            PyObject *attr;
+            PyObject *self;
+            owner = stack_pointer[-1];
+            PyObject *descr = (PyObject *)operand;
+            assert(oparg & 1);
+            /* Cached method object */
+            STAT_INC(LOAD_ATTR, hit);
+            assert(descr != NULL);
+            attr = Py_NewRef(descr);
+            assert(_PyType_HasFeature(Py_TYPE(attr), Py_TPFLAGS_METHOD_DESCRIPTOR));
+            self = owner;
+            STACK_GROW(1);
+            stack_pointer[-2] = attr;
+            stack_pointer[-1] = self;
+            break;
+        }
+
+        case _LOAD_ATTR_METHOD_NO_DICT: {
+            PyObject *owner;
+            PyObject *attr;
+            PyObject *self;
+            owner = stack_pointer[-1];
+            PyObject *descr = (PyObject *)operand;
+            assert(oparg & 1);
+            PyTypeObject *owner_cls = Py_TYPE(owner);
+            assert(owner_cls->tp_dictoffset == 0);
+            STAT_INC(LOAD_ATTR, hit);
+            assert(descr != NULL);
+            assert(_PyType_HasFeature(Py_TYPE(descr), Py_TPFLAGS_METHOD_DESCRIPTOR));
+            attr = Py_NewRef(descr);
+            self = owner;
+            STACK_GROW(1);
+            stack_pointer[-2] = attr;
+            stack_pointer[-1] = self;
+            break;
+        }
+
         case _CHECK_CALL_BOUND_METHOD_EXACT_ARGS: {
             PyObject *null;
             PyObject *callable;