gh-105481: add OPCODE_IS_INSTRUMENTED (generated from bytecodes.c) to…

… replace MIN_INSTRUMENTED_OPCODE (defined in opcode.py)
python · iritkatriel · Jul 25, 2023 · Jul 25, 2023 · Jul 26, 2023 · Jul 26, 2023
commit d64208367eadd0c36fc93097819c7bdacfc51df6
diff --git a/Include/cpython/compile.h b/Include/cpython/compile.h
@@ -76,6 +76,7 @@ PyAPI_FUNC(int) PyUnstable_OpcodeHasJump(int opcode);
 PyAPI_FUNC(int) PyUnstable_OpcodeHasFree(int opcode);
 PyAPI_FUNC(int) PyUnstable_OpcodeHasLocal(int opcode);
 PyAPI_FUNC(int) PyUnstable_OpcodeHasExc(int opcode);
+PyAPI_FUNC(int) PyUnstable_OpcodeIsInstrumented(int opcode);

 PyAPI_FUNC(PyObject*) _PyUnstable_GetUnaryIntrinsicName(int index);
 PyAPI_FUNC(PyObject*) _PyUnstable_GetBinaryIntrinsicName(int index);
diff --git a/Include/internal/pycore_opcode_metadata.h b/Include/internal/pycore_opcode_metadata.h
@@ -957,12 +957,14 @@ enum InstructionFormat { INSTR_FMT_IB, INSTR_FMT_IBC, INSTR_FMT_IBC00, INSTR_FMT
 #define HAS_JUMP_FLAG (8)
 #define HAS_FREE_FLAG (16)
 #define HAS_LOCAL_FLAG (32)
+#define IS_INSTRUMENTED_FLAG (64)
 #define OPCODE_HAS_ARG(OP) (_PyOpcode_opcode_metadata[OP].flags & (HAS_ARG_FLAG))
 #define OPCODE_HAS_CONST(OP) (_PyOpcode_opcode_metadata[OP].flags & (HAS_CONST_FLAG))
 #define OPCODE_HAS_NAME(OP) (_PyOpcode_opcode_metadata[OP].flags & (HAS_NAME_FLAG))
 #define OPCODE_HAS_JUMP(OP) (_PyOpcode_opcode_metadata[OP].flags & (HAS_JUMP_FLAG))
 #define OPCODE_HAS_FREE(OP) (_PyOpcode_opcode_metadata[OP].flags & (HAS_FREE_FLAG))
 #define OPCODE_HAS_LOCAL(OP) (_PyOpcode_opcode_metadata[OP].flags & (HAS_LOCAL_FLAG))
+#define OPCODE_IS_INSTRUMENTED(OP) (_PyOpcode_opcode_metadata[OP].flags & (IS_INSTRUMENTED_FLAG))

 struct opcode_metadata {
    bool valid_entry;
@@ -998,7 +1000,7 @@ extern const char * const _PyOpcode_uop_name[OPCODE_UOP_NAME_SIZE];
 const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [NOP] = { true, INSTR_FMT_IX, 0 },
    [RESUME] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_RESUME] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
+    [INSTRUMENTED_RESUME] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
    [LOAD_CLOSURE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG },
    [LOAD_FAST_CHECK] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG },
    [LOAD_FAST] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_LOCAL_FLAG },
@@ -1012,9 +1014,9 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [POP_TOP] = { true, INSTR_FMT_IX, 0 },
    [PUSH_NULL] = { true, INSTR_FMT_IX, 0 },
    [END_FOR] = { true, INSTR_FMT_IB, 0 },
-    [INSTRUMENTED_END_FOR] = { true, INSTR_FMT_IX, 0 },
+    [INSTRUMENTED_END_FOR] = { true, INSTR_FMT_IX, IS_INSTRUMENTED_FLAG },
    [END_SEND] = { true, INSTR_FMT_IX, 0 },
-    [INSTRUMENTED_END_SEND] = { true, INSTR_FMT_IX, 0 },
+    [INSTRUMENTED_END_SEND] = { true, INSTR_FMT_IX, IS_INSTRUMENTED_FLAG },
    [UNARY_NEGATIVE] = { true, INSTR_FMT_IX, 0 },
    [UNARY_NOT] = { true, INSTR_FMT_IX, 0 },
    [TO_BOOL] = { true, INSTR_FMT_IXC00, 0 },
@@ -1051,15 +1053,15 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [RAISE_VARARGS] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [INTERPRETER_EXIT] = { true, INSTR_FMT_IX, 0 },
    [RETURN_VALUE] = { true, INSTR_FMT_IX, 0 },
-    [INSTRUMENTED_RETURN_VALUE] = { true, INSTR_FMT_IX, 0 },
+    [INSTRUMENTED_RETURN_VALUE] = { true, INSTR_FMT_IX, IS_INSTRUMENTED_FLAG },
    [RETURN_CONST] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_CONST_FLAG },
-    [INSTRUMENTED_RETURN_CONST] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_CONST_FLAG },
+    [INSTRUMENTED_RETURN_CONST] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_CONST_FLAG | IS_INSTRUMENTED_FLAG },
    [GET_AITER] = { true, INSTR_FMT_IX, 0 },
    [GET_ANEXT] = { true, INSTR_FMT_IX, 0 },
    [GET_AWAITABLE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [SEND] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_JUMP_FLAG },
    [SEND_GEN] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG },
-    [INSTRUMENTED_YIELD_VALUE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
+    [INSTRUMENTED_YIELD_VALUE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
    [YIELD_VALUE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [POP_EXCEPT] = { true, INSTR_FMT_IX, 0 },
    [RERAISE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
@@ -1103,7 +1105,7 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [DICT_UPDATE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [DICT_MERGE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [MAP_ADD] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_LOAD_SUPER_ATTR] = { true, INSTR_FMT_IBC00000000, HAS_ARG_FLAG },
+    [INSTRUMENTED_LOAD_SUPER_ATTR] = { true, INSTR_FMT_IBC00000000, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
    [LOAD_SUPER_ATTR] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_NAME_FLAG },
    [LOAD_SUPER_METHOD] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_NAME_FLAG },
    [LOAD_ZERO_SUPER_METHOD] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_NAME_FLAG },
@@ -1150,7 +1152,7 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [GET_ITER] = { true, INSTR_FMT_IX, 0 },
    [GET_YIELD_FROM_ITER] = { true, INSTR_FMT_IX, 0 },
    [FOR_ITER] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_JUMP_FLAG },
-    [INSTRUMENTED_FOR_ITER] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
+    [INSTRUMENTED_FOR_ITER] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
    [FOR_ITER_LIST] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_JUMP_FLAG },
    [FOR_ITER_TUPLE] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_JUMP_FLAG },
    [FOR_ITER_RANGE] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG | HAS_JUMP_FLAG },
@@ -1169,7 +1171,7 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [LOAD_ATTR_NONDESCRIPTOR_NO_DICT] = { true, INSTR_FMT_IBC00000000, HAS_ARG_FLAG },
    [LOAD_ATTR_METHOD_LAZY_DICT] = { true, INSTR_FMT_IBC00000000, HAS_ARG_FLAG },
    [KW_NAMES] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_CONST_FLAG },
-    [INSTRUMENTED_CALL] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
+    [INSTRUMENTED_CALL] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
    [CALL] = { true, INSTR_FMT_IBC00, HAS_ARG_FLAG },
    [CALL_BOUND_METHOD_EXACT_ARGS] = { true, INSTR_FMT_IBC00, HAS_ARG_FLAG },
    [CALL_PY_EXACT_ARGS] = { true, INSTR_FMT_IBC00, HAS_ARG_FLAG },
@@ -1190,7 +1192,7 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS] = { true, INSTR_FMT_IBC00, HAS_ARG_FLAG },
    [CALL_NO_KW_METHOD_DESCRIPTOR_NOARGS] = { true, INSTR_FMT_IBC00, HAS_ARG_FLAG },
    [CALL_NO_KW_METHOD_DESCRIPTOR_FAST] = { true, INSTR_FMT_IBC00, HAS_ARG_FLAG },
-    [INSTRUMENTED_CALL_FUNCTION_EX] = { true, INSTR_FMT_IX, 0 },
+    [INSTRUMENTED_CALL_FUNCTION_EX] = { true, INSTR_FMT_IX, IS_INSTRUMENTED_FLAG },
    [CALL_FUNCTION_EX] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [MAKE_FUNCTION] = { true, INSTR_FMT_IX, 0 },
    [SET_FUNCTION_ATTRIBUTE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
@@ -1202,13 +1204,13 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[OPCODE_METADATA_SIZE] = {
    [COPY] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [BINARY_OP] = { true, INSTR_FMT_IBC, HAS_ARG_FLAG },
    [SWAP] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_INSTRUCTION] = { true, INSTR_FMT_IX, 0 },
-    [INSTRUMENTED_JUMP_FORWARD] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_JUMP_BACKWARD] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_POP_JUMP_IF_TRUE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_POP_JUMP_IF_FALSE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_POP_JUMP_IF_NONE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
-    [INSTRUMENTED_POP_JUMP_IF_NOT_NONE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
+    [INSTRUMENTED_INSTRUCTION] = { true, INSTR_FMT_IX, IS_INSTRUMENTED_FLAG },
+    [INSTRUMENTED_JUMP_FORWARD] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
+    [INSTRUMENTED_JUMP_BACKWARD] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
+    [INSTRUMENTED_POP_JUMP_IF_TRUE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
+    [INSTRUMENTED_POP_JUMP_IF_FALSE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
+    [INSTRUMENTED_POP_JUMP_IF_NONE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
+    [INSTRUMENTED_POP_JUMP_IF_NOT_NONE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | IS_INSTRUMENTED_FLAG },
    [EXTENDED_ARG] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
    [CACHE] = { true, INSTR_FMT_IX, 0 },
    [RESERVED] = { true, INSTR_FMT_IX, 0 },

diff --git a/Include/opcode.h b/Include/opcode.h
diff --git a/Lib/opcode.py b/Lib/opcode.py
@@ -192,7 +192,6 @@ def pseudo_op(name, op, real_ops):
 def_op('ENTER_EXECUTOR', 230)

 # Instrumented instructions
-MIN_INSTRUMENTED_OPCODE = 237

 def_op('INSTRUMENTED_LOAD_SUPER_ATTR', 237)
 def_op('INSTRUMENTED_POP_JUMP_IF_NONE', 238)

diff --git a/Lib/test/test__opcode.py b/Lib/test/test__opcode.py
@@ -18,6 +18,7 @@ def test_invalid_opcodes(self):
        self.check_bool_function_result(_opcode.has_free, invalid, False)
        self.check_bool_function_result(_opcode.has_local, invalid, False)
        self.check_bool_function_result(_opcode.has_exc, invalid, False)
+        self.check_bool_function_result(_opcode.is_instrumented, invalid, False)

    def test_is_valid(self):
        names = [

diff --git a/Modules/_opcode.c b/Modules/_opcode.c
@@ -203,6 +203,24 @@ _opcode_has_exc_impl(PyObject *module, int opcode)
           PyUnstable_OpcodeHasExc(opcode);
 }

+/*[clinic input]
+
+_opcode.is_instrumented -> bool
+
+  opcode: int
+
+Return True if the opcode is instrumented, False otherwise.
+[clinic start generated code]*/
+
+static int
+_opcode_is_instrumented_impl(PyObject *module, int opcode)
+/*[clinic end generated code: output=dbfb4dce2bef87c8 input=5122645f4009417a]*/
+{
+    return PyUnstable_OpcodeIsValid(opcode) &&
+           PyUnstable_OpcodeIsInstrumented(opcode);
+}
+
+
 /*[clinic input]

 _opcode.get_specialization_stats
@@ -286,6 +304,7 @@ opcode_functions[] =  {
    _OPCODE_HAS_FREE_METHODDEF
    _OPCODE_HAS_LOCAL_METHODDEF
    _OPCODE_HAS_EXC_METHODDEF
+    _OPCODE_IS_INSTRUMENTED_METHODDEF
    _OPCODE_GET_SPECIALIZATION_STATS_METHODDEF
    _OPCODE_GET_INTRINSIC1_DESCS_METHODDEF
    _OPCODE_GET_INTRINSIC2_DESCS_METHODDEF

diff --git a/Modules/clinic/_opcode.c.h b/Modules/clinic/_opcode.c.h
@@ -915,6 +915,12 @@ PyUnstable_OpcodeHasExc(int opcode)
    return IS_BLOCK_PUSH_OPCODE(opcode);
 }

+int
+PyUnstable_OpcodeIsInstrumented(int opcode)
+{
+    return OPCODE_IS_INSTRUMENTED(opcode);
+}
+
 static int
 codegen_addop_noarg(instr_sequence *seq, int opcode, location loc)
 {

diff --git a/Python/instrumentation.c b/Python/instrumentation.c
@@ -3,12 +3,13 @@
 #include "pycore_frame.h"
 #include "pycore_interp.h"
 #include "pycore_long.h"
-#include "pycore_modsupport.h"    // _PyModule_CreateInitialized()
+#include "pycore_modsupport.h"       // _PyModule_CreateInitialized()
 #include "pycore_namespace.h"
 #include "pycore_object.h"
 #include "pycore_opcode.h"
+#include "pycore_opcode_metadata.h"  // OPCODE_IS_INSTRUMENTED
 #include "pycore_pyerrors.h"
-#include "pycore_pystate.h"       // _PyInterpreterState_GET()
+#include "pycore_pystate.h"          // _PyInterpreterState_GET()

 /* Uncomment this to dump debugging output when assertions fail */
 // #define INSTRUMENT_DEBUG 1
@@ -130,7 +131,7 @@ is_instrumented(int opcode)
 {
    assert(opcode != 0);
    assert(opcode != RESERVED);
-    return opcode >= MIN_INSTRUMENTED_OPCODE;
+    return OPCODE_IS_INSTRUMENTED(opcode);
 }

 #ifndef NDEBUG

diff --git a/Python/specialize.c b/Python/specialize.c
@@ -7,8 +7,9 @@
 #include "pycore_long.h"
 #include "pycore_moduleobject.h"
 #include "pycore_object.h"
-#include "pycore_opcode.h"        // _PyOpcode_Caches
-#include "pycore_pylifecycle.h"   // _PyOS_URandomNonblock()
+#include "pycore_opcode.h"          // _PyOpcode_Caches
+#include "pycore_opcode_metadata.h" // OPCODE_IS_INSTRUMENTED
+#include "pycore_pylifecycle.h"     // _PyOS_URandomNonblock()


 #include <stdlib.h> // rand()
@@ -279,7 +280,7 @@ _PyCode_Quicken(PyCodeObject *code)
    _Py_CODEUNIT *instructions = _PyCode_CODE(code);
    for (int i = 0; i < Py_SIZE(code); i++) {
        opcode = _Py_GetBaseOpcode(code, i);
-        assert(opcode < MIN_INSTRUMENTED_OPCODE);
+        assert(!OPCODE_IS_INSTRUMENTED(opcode));
        int caches = _PyOpcode_Caches[opcode];
        if (caches) {
            instructions[i + 1].cache = adaptive_counter_warmup();

diff --git a/Tools/build/generate_opcode_h.py b/Tools/build/generate_opcode_h.py
@@ -76,7 +76,6 @@ def main(opcode_py,
    ENABLE_SPECIALIZATION = opcode["ENABLE_SPECIALIZATION"]
    MIN_PSEUDO_OPCODE = opcode["MIN_PSEUDO_OPCODE"]
    MAX_PSEUDO_OPCODE = opcode["MAX_PSEUDO_OPCODE"]
-    MIN_INSTRUMENTED_OPCODE = opcode["MIN_INSTRUMENTED_OPCODE"]

    NUM_OPCODES = len(opname)
    used = [ False ] * len(opname)
@@ -103,8 +102,6 @@ def main(opcode_py,
                op = opmap[name]
                if op == MIN_PSEUDO_OPCODE:
                    fobj.write(DEFINE.format("MIN_PSEUDO_OPCODE", MIN_PSEUDO_OPCODE))
-                if op == MIN_INSTRUMENTED_OPCODE:
-                    fobj.write(DEFINE.format("MIN_INSTRUMENTED_OPCODE", MIN_INSTRUMENTED_OPCODE))

                fobj.write(DEFINE.format(name, op))


@@ -15,6 +15,7 @@ class InstructionFlags:
    HAS_JUMP_FLAG: bool
    HAS_FREE_FLAG: bool
    HAS_LOCAL_FLAG: bool
+    IS_INSTRUMENTED_FLAG: bool

    def __post_init__(self):
        self.bitmask = {name: (1 << i) for i, name in enumerate(self.names())}
@@ -38,11 +39,12 @@ def fromInstruction(instr: parsing.Node):
                variable_used(instr, "GETLOCAL") or variable_used(instr, "SETLOCAL")
            )
            and not has_free,
+            IS_INSTRUMENTED_FLAG=instr.name.startswith("INSTRUMENTED_"),
        )

    @staticmethod
    def newEmpty():
-        return InstructionFlags(False, False, False, False, False, False)
+        return InstructionFlags(False, False, False, False, False, False, False)

    def add(self, other: "InstructionFlags") -> None:
        for name, value in dataclasses.asdict(other).items():