GH-103082: Implementation of PEP 669: Low Impact Monitoring for CPython (GH-103083)

* The majority of the monitoring code is in instrumentation.c

* The new instrumentation bytecodes are in bytecodes.c

* legacy_tracing.c adapts the new API to the old sys.setrace and sys.setprofile APIs

9 files changed, 325 insertions, 179 deletions

diff --git a/Include/cpython/code.h b/Include/cpython/code.h
index abcf125..6bead36 100644
--- a/Include/cpython/code.h
+++ b/Include/cpython/code.h
@@ -3,10 +3,22 @@
 #ifndef Py_LIMITED_API
 #ifndef Py_CODE_H
 #define Py_CODE_H
+
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+
+/* Count of all "real" monitoring events (not derived from other events) */
+#define PY_MONITORING_UNGROUPED_EVENTS 14
+/* Count of all  monitoring events */
+#define PY_MONITORING_EVENTS 16
+
+/* Table of which tools are active for each monitored event. */
+typedef struct _Py_Monitors {
+    uint8_t tools[PY_MONITORING_UNGROUPED_EVENTS];
+} _Py_Monitors;
+
 /* Each instruction in a code object is a fixed-width value,
  * currently 2 bytes: 1-byte opcode + 1-byte oparg.  The EXTENDED_ARG
  * opcode allows for larger values but the current limit is 3 uses
@@ -56,6 +68,35 @@ typedef struct {
     PyObject *_co_freevars;
 } _PyCoCached;
 
+/* Ancilliary data structure used for instrumentation.
+   Line instrumentation creates an array of
+   these. One entry per code unit.*/
+typedef struct {
+    uint8_t original_opcode;
+    int8_t line_delta;
+} _PyCoLineInstrumentationData;
+
+/* Main data structure used for instrumentation.
+ * This is allocated when needed for instrumentation
+ */
+typedef struct {
+    /* Monitoring specific to this code object */
+    _Py_Monitors local_monitors;
+    /* Monitoring that is active on this code object */
+    _Py_Monitors active_monitors;
+    /* The tools that are to be notified for events for the matching code unit */
+    uint8_t *tools;
+    /* Information to support line events */
+    _PyCoLineInstrumentationData *lines;
+    /* The tools that are to be notified for line events for the matching code unit */
+    uint8_t *line_tools;
+    /* Information to support instruction events */
+    /* The underlying instructions, which can themselves be instrumented */
+    uint8_t *per_instruction_opcodes;
+    /* The tools that are to be notified for instruction events for the matching code unit */
+    uint8_t *per_instruction_tools;
+} _PyCoMonitoringData;
+
 // To avoid repeating ourselves in deepfreeze.py, all PyCodeObject members are
 // defined in this macro:
 #define _PyCode_DEF(SIZE) {                                                    \
@@ -87,7 +128,6 @@ typedef struct {
     PyObject *co_exceptiontable;   /* Byte string encoding exception handling  \
                                       table */                                 \
     int co_flags;                  /* CO_..., see below */                     \
-    short _co_linearray_entry_size;  /* Size of each entry in _co_linearray */ \
                                                                                \
     /* The rest are not so impactful on performance. */                        \
     int co_argcount;              /* #arguments, except *args */               \
@@ -114,8 +154,9 @@ typedef struct {
     PyObject *co_linetable;       /* bytes object that holds location info */  \
     PyObject *co_weakreflist;     /* to support weakrefs to code objects */    \
     _PyCoCached *_co_cached;      /* cached co_* attributes */                 \
+    uint64_t _co_instrumentation_version; /* current instrumentation version */  \
+    _PyCoMonitoringData *_co_monitoring; /* Monitoring data */                 \
     int _co_firsttraceable;       /* index of first traceable instruction */   \
-    char *_co_linearray;          /* array of line offsets */                  \
     /* Scratch space for extra data relating to the code object.               \
        Type is a void* to keep the format private in codeobject.c to force     \
        people to go through the proper APIs. */                                \
diff --git a/Include/cpython/pystate.h b/Include/cpython/pystate.h
index 3efb241..ea6ed8d 100644
--- a/Include/cpython/pystate.h
+++ b/Include/cpython/pystate.h
@@ -58,12 +58,6 @@ typedef int (*Py_tracefunc)(PyObject *, PyFrameObject *, int, PyObject *);
 #define PyTrace_C_RETURN 6
 #define PyTrace_OPCODE 7
 
-
-typedef struct {
-    PyCodeObject *code; // The code object for the bounds. May be NULL.
-    PyCodeAddressRange bounds; // Only valid if code != NULL.
-} PyTraceInfo;
-
 // Internal structure: you should not use it directly, but use public functions
 // like PyThreadState_EnterTracing() and PyThreadState_LeaveTracing().
 typedef struct _PyCFrame {
@@ -77,7 +71,6 @@ typedef struct _PyCFrame {
      * discipline and make sure that instances of this struct cannot
      * accessed outside of their lifetime.
      */
-    uint8_t use_tracing;  // 0 or 255 (or'ed into opcode, hence 8-bit type)
     /* Pointer to the currently executing frame (it can be NULL) */
     struct _PyInterpreterFrame *current_frame;
     struct _PyCFrame *previous;
@@ -157,7 +150,7 @@ struct _ts {
        This is to prevent the actual trace/profile code from being recorded in
        the trace/profile. */
     int tracing;
-    int tracing_what; /* The event currently being traced, if any. */
+    int what_event; /* The event currently being monitored, if any. */
 
     /* Pointer to current _PyCFrame in the C stack frame of the currently,
      * or most recently, executing _PyEval_EvalFrameDefault. */
@@ -228,8 +221,6 @@ struct _ts {
     /* Unique thread state id. */
     uint64_t id;
 
-    PyTraceInfo trace_info;
-
     _PyStackChunk *datastack_chunk;
     PyObject **datastack_top;
     PyObject **datastack_limit;
diff --git a/Include/internal/pycore_code.h b/Include/internal/pycore_code.h
index faf1be5..d32f37a 100644
--- a/Include/internal/pycore_code.h
+++ b/Include/internal/pycore_code.h
@@ -441,32 +441,6 @@ adaptive_counter_backoff(uint16_t counter) {
 
 /* Line array cache for tracing */
 
-extern int _PyCode_CreateLineArray(PyCodeObject *co);
-
-static inline int
-_PyCode_InitLineArray(PyCodeObject *co)
-{
-    if (co->_co_linearray) {
-        return 0;
-    }
-    return _PyCode_CreateLineArray(co);
-}
-
-static inline int
-_PyCode_LineNumberFromArray(PyCodeObject *co, int index)
-{
-    assert(co->_co_linearray != NULL);
-    assert(index >= 0);
-    assert(index < Py_SIZE(co));
-    if (co->_co_linearray_entry_size == 2) {
-        return ((int16_t *)co->_co_linearray)[index];
-    }
-    else {
-        assert(co->_co_linearray_entry_size == 4);
-        return ((int32_t *)co->_co_linearray)[index];
-    }
-}
-
 typedef struct _PyShimCodeDef {
     const uint8_t *code;
     int codelen;
@@ -500,6 +474,10 @@ extern uint32_t _Py_next_func_version;
 
 #define COMPARISON_NOT_EQUALS (COMPARISON_UNORDERED | COMPARISON_LESS_THAN | COMPARISON_GREATER_THAN)
 
+extern int _Py_Instrument(PyCodeObject *co, PyInterpreterState *interp);
+
+extern int _Py_GetBaseOpcode(PyCodeObject *code, int offset);
+
 
 #ifdef __cplusplus
 }
diff --git a/Include/internal/pycore_frame.h b/Include/internal/pycore_frame.h
index 5806cf0..856297a 100644
--- a/Include/internal/pycore_frame.h
+++ b/Include/internal/pycore_frame.h
@@ -19,6 +19,7 @@ struct _frame {
     struct _PyInterpreterFrame *f_frame; /* points to the frame data */
     PyObject *f_trace;          /* Trace function */
     int f_lineno;               /* Current line number. Only valid if non-zero */
+    int f_last_traced_line;     /* The last line traced for this frame */
     char f_trace_lines;         /* Emit per-line trace events? */
     char f_trace_opcodes;       /* Emit per-opcode trace events? */
     char f_fast_as_locals;      /* Have the fast locals of this frame been converted to a dict? */
@@ -137,10 +138,16 @@ _PyFrame_GetLocalsArray(_PyInterpreterFrame *frame)
     return frame->localsplus;
 }
 
+/* Fetches the stack pointer, and sets stacktop to -1.
+    Having stacktop <= 0 ensures that invalid
+    values are not visible to the cycle GC.
+    We choose -1 rather than 0 to assist debugging. */
 static inline PyObject**
 _PyFrame_GetStackPointer(_PyInterpreterFrame *frame)
 {
-    return frame->localsplus+frame->stacktop;
+    PyObject **sp = frame->localsplus + frame->stacktop;
+    frame->stacktop = -1;
+    return sp;
 }
 
 static inline void
diff --git a/Include/internal/pycore_instruments.h b/Include/internal/pycore_instruments.h
new file mode 100644
index 0000000..e94d875
--- /dev/null
+++ b/Include/internal/pycore_instruments.h
@@ -0,0 +1,107 @@
+
+#ifndef Py_INTERNAL_INSTRUMENT_H
+#define Py_INTERNAL_INSTRUMENT_H
+
+
+#include "pycore_bitutils.h"      // _Py_popcount32
+#include "pycore_frame.h"
+
+#include "cpython/code.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define PY_MONITORING_TOOL_IDS 8
+
+/* Local events.
+ * These require bytecode instrumentation */
+
+#define PY_MONITORING_EVENT_PY_START 0
+#define PY_MONITORING_EVENT_PY_RESUME 1
+#define PY_MONITORING_EVENT_PY_RETURN 2
+#define PY_MONITORING_EVENT_PY_YIELD 3
+#define PY_MONITORING_EVENT_CALL 4
+#define PY_MONITORING_EVENT_LINE 5
+#define PY_MONITORING_EVENT_INSTRUCTION 6
+#define PY_MONITORING_EVENT_JUMP 7
+#define PY_MONITORING_EVENT_BRANCH 8
+#define PY_MONITORING_EVENT_STOP_ITERATION 9
+
+#define PY_MONITORING_INSTRUMENTED_EVENTS 10
+
+/* Other events, mainly exceptions */
+
+#define PY_MONITORING_EVENT_RAISE 10
+#define PY_MONITORING_EVENT_EXCEPTION_HANDLED 11
+#define PY_MONITORING_EVENT_PY_UNWIND 12
+#define PY_MONITORING_EVENT_PY_THROW 13
+
+
+/* Ancilliary events */
+
+#define PY_MONITORING_EVENT_C_RETURN 14
+#define PY_MONITORING_EVENT_C_RAISE 15
+
+
+typedef uint32_t _PyMonitoringEventSet;
+
+/* Tool IDs */
+
+/* These are defined in PEP 669 for convenience to avoid clashes */
+#define PY_MONITORING_DEBUGGER_ID 0
+#define PY_MONITORING_COVERAGE_ID 1
+#define PY_MONITORING_PROFILER_ID 2
+#define PY_MONITORING_OPTIMIZER_ID 5
+
+/* Internal IDs used to suuport sys.setprofile() and sys.settrace() */
+#define PY_MONITORING_SYS_PROFILE_ID 6
+#define PY_MONITORING_SYS_TRACE_ID 7
+
+
+PyObject *_PyMonitoring_RegisterCallback(int tool_id, int event_id, PyObject *obj);
+
+int _PyMonitoring_SetEvents(int tool_id, _PyMonitoringEventSet events);
+
+extern int
+_Py_call_instrumentation(PyThreadState *tstate, int event,
+    _PyInterpreterFrame *frame, _Py_CODEUNIT *instr);
+
+extern int
+_Py_call_instrumentation_line(PyThreadState *tstate, _PyInterpreterFrame* frame,
+                              _Py_CODEUNIT *instr);
+
+extern int
+_Py_call_instrumentation_instruction(
+    PyThreadState *tstate, _PyInterpreterFrame* frame, _Py_CODEUNIT *instr);
+
+int
+_Py_call_instrumentation_jump(
+    PyThreadState *tstate, int event,
+    _PyInterpreterFrame *frame, _Py_CODEUNIT *instr, _Py_CODEUNIT *target);
+
+extern int
+_Py_call_instrumentation_arg(PyThreadState *tstate, int event,
+    _PyInterpreterFrame *frame, _Py_CODEUNIT *instr, PyObject *arg);
+
+extern int
+_Py_call_instrumentation_2args(PyThreadState *tstate, int event,
+    _PyInterpreterFrame *frame, _Py_CODEUNIT *instr, PyObject *arg0, PyObject *arg1);
+
+extern void
+_Py_call_instrumentation_exc0(PyThreadState *tstate, int event,
+    _PyInterpreterFrame *frame, _Py_CODEUNIT *instr);
+
+extern void
+_Py_call_instrumentation_exc2(PyThreadState *tstate, int event,
+    _PyInterpreterFrame *frame, _Py_CODEUNIT *instr, PyObject *arg0, PyObject *arg1);
+
+extern int
+_Py_Instrumentation_GetLine(PyCodeObject *code, int index);
+
+extern PyObject _PyInstrumentation_MISSING;
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* !Py_INTERNAL_INSTRUMENT_H */
diff --git a/Include/internal/pycore_interp.h b/Include/internal/pycore_interp.h
index d64a68c..86ae3d8 100644
--- a/Include/internal/pycore_interp.h
+++ b/Include/internal/pycore_interp.h
@@ -24,6 +24,7 @@ extern "C" {
 #include "pycore_genobject.h"     // struct _Py_async_gen_state
 #include "pycore_gc.h"            // struct _gc_runtime_state
 #include "pycore_import.h"        // struct _import_state
+#include "pycore_instruments.h"   // PY_MONITORING_EVENTS
 #include "pycore_list.h"          // struct _Py_list_state
 #include "pycore_global_objects.h"  // struct _Py_interp_static_objects
 #include "pycore_object_state.h"   // struct _py_object_state
@@ -37,7 +38,6 @@ struct _Py_long_state {
     int max_str_digits;
 };
 
-
 /* interpreter state */
 
 /* PyInterpreterState holds the global state for one of the runtime's
@@ -49,6 +49,9 @@ struct _is {
 
     PyInterpreterState *next;
 
+    uint64_t monitoring_version;
+    uint64_t last_restart_version;
+
     struct pythreads {
         uint64_t next_unique_id;
         /* The linked list of threads, newest first. */
@@ -148,6 +151,15 @@ struct _is {
     struct callable_cache callable_cache;
     PyCodeObject *interpreter_trampoline;
 
+    _Py_Monitors monitors;
+    bool f_opcode_trace_set;
+    bool sys_profile_initialized;
+    bool sys_trace_initialized;
+    Py_ssize_t sys_profiling_threads; /* Count of threads with c_profilefunc set */
+    Py_ssize_t sys_tracing_threads; /* Count of threads with c_tracefunc set */
+    PyObject *monitoring_callables[PY_MONITORING_TOOL_IDS][PY_MONITORING_EVENTS];
+    PyObject *monitoring_tool_names[PY_MONITORING_TOOL_IDS];
+
     struct _Py_interp_cached_objects cached_objects;
     struct _Py_interp_static_objects static_objects;
 
diff --git a/Include/internal/pycore_opcode.h b/Include/internal/pycore_opcode.h
index f5e9176..c039d71 100644
--- a/Include/internal/pycore_opcode.h
+++ b/Include/internal/pycore_opcode.h
@@ -112,6 +112,7 @@ const uint8_t _PyOpcode_Deopt[256] = {
     [DICT_UPDATE] = DICT_UPDATE,
     [END_ASYNC_FOR] = END_ASYNC_FOR,
     [END_FOR] = END_FOR,
+    [END_SEND] = END_SEND,
     [EXTENDED_ARG] = EXTENDED_ARG,
     [FORMAT_VALUE] = FORMAT_VALUE,
     [FOR_ITER] = FOR_ITER,
@@ -127,6 +128,23 @@ const uint8_t _PyOpcode_Deopt[256] = {
     [GET_YIELD_FROM_ITER] = GET_YIELD_FROM_ITER,
     [IMPORT_FROM] = IMPORT_FROM,
     [IMPORT_NAME] = IMPORT_NAME,
+    [INSTRUMENTED_CALL] = INSTRUMENTED_CALL,
+    [INSTRUMENTED_CALL_FUNCTION_EX] = INSTRUMENTED_CALL_FUNCTION_EX,
+    [INSTRUMENTED_END_FOR] = INSTRUMENTED_END_FOR,
+    [INSTRUMENTED_END_SEND] = INSTRUMENTED_END_SEND,
+    [INSTRUMENTED_FOR_ITER] = INSTRUMENTED_FOR_ITER,
+    [INSTRUMENTED_INSTRUCTION] = INSTRUMENTED_INSTRUCTION,
+    [INSTRUMENTED_JUMP_BACKWARD] = INSTRUMENTED_JUMP_BACKWARD,
+    [INSTRUMENTED_JUMP_FORWARD] = INSTRUMENTED_JUMP_FORWARD,
+    [INSTRUMENTED_LINE] = INSTRUMENTED_LINE,
+    [INSTRUMENTED_POP_JUMP_IF_FALSE] = INSTRUMENTED_POP_JUMP_IF_FALSE,
+    [INSTRUMENTED_POP_JUMP_IF_NONE] = INSTRUMENTED_POP_JUMP_IF_NONE,
+    [INSTRUMENTED_POP_JUMP_IF_NOT_NONE] = INSTRUMENTED_POP_JUMP_IF_NOT_NONE,
+    [INSTRUMENTED_POP_JUMP_IF_TRUE] = INSTRUMENTED_POP_JUMP_IF_TRUE,
+    [INSTRUMENTED_RESUME] = INSTRUMENTED_RESUME,
+    [INSTRUMENTED_RETURN_CONST] = INSTRUMENTED_RETURN_CONST,
+    [INSTRUMENTED_RETURN_VALUE] = INSTRUMENTED_RETURN_VALUE,
+    [INSTRUMENTED_YIELD_VALUE] = INSTRUMENTED_YIELD_VALUE,
     [INTERPRETER_EXIT] = INTERPRETER_EXIT,
     [IS_OP] = IS_OP,
     [JUMP_BACKWARD] = JUMP_BACKWARD,
@@ -179,6 +197,7 @@ const uint8_t _PyOpcode_Deopt[256] = {
     [PUSH_NULL] = PUSH_NULL,
     [RAISE_VARARGS] = RAISE_VARARGS,
     [RERAISE] = RERAISE,
+    [RESERVED] = RESERVED,
     [RESUME] = RESUME,
     [RETURN_CONST] = RETURN_CONST,
     [RETURN_GENERATOR] = RETURN_GENERATOR,
@@ -223,17 +242,19 @@ static const char *const _PyOpcode_OpName[263] = {
     [PUSH_NULL] = "PUSH_NULL",
     [INTERPRETER_EXIT] = "INTERPRETER_EXIT",
     [END_FOR] = "END_FOR",
+    [END_SEND] = "END_SEND",
     [BINARY_OP_ADD_FLOAT] = "BINARY_OP_ADD_FLOAT",
     [BINARY_OP_ADD_INT] = "BINARY_OP_ADD_INT",
     [BINARY_OP_ADD_UNICODE] = "BINARY_OP_ADD_UNICODE",
-    [BINARY_OP_INPLACE_ADD_UNICODE] = "BINARY_OP_INPLACE_ADD_UNICODE",
     [NOP] = "NOP",
-    [BINARY_OP_MULTIPLY_FLOAT] = "BINARY_OP_MULTIPLY_FLOAT",
+    [BINARY_OP_INPLACE_ADD_UNICODE] = "BINARY_OP_INPLACE_ADD_UNICODE",
     [UNARY_NEGATIVE] = "UNARY_NEGATIVE",
     [UNARY_NOT] = "UNARY_NOT",
+    [BINARY_OP_MULTIPLY_FLOAT] = "BINARY_OP_MULTIPLY_FLOAT",
     [BINARY_OP_MULTIPLY_INT] = "BINARY_OP_MULTIPLY_INT",
-    [BINARY_OP_SUBTRACT_FLOAT] = "BINARY_OP_SUBTRACT_FLOAT",
     [UNARY_INVERT] = "UNARY_INVERT",
+    [BINARY_OP_SUBTRACT_FLOAT] = "BINARY_OP_SUBTRACT_FLOAT",
+    [RESERVED] = "RESERVED",
     [BINARY_OP_SUBTRACT_INT] = "BINARY_OP_SUBTRACT_INT",
     [BINARY_SUBSCR_DICT] = "BINARY_SUBSCR_DICT",
     [BINARY_SUBSCR_GETITEM] = "BINARY_SUBSCR_GETITEM",
@@ -241,21 +262,21 @@ static const char *const _PyOpcode_OpName[263] = {
     [BINARY_SUBSCR_TUPLE_INT] = "BINARY_SUBSCR_TUPLE_INT",
     [CALL_PY_EXACT_ARGS] = "CALL_PY_EXACT_ARGS",
     [CALL_PY_WITH_DEFAULTS] = "CALL_PY_WITH_DEFAULTS",
-    [CALL_BOUND_METHOD_EXACT_ARGS] = "CALL_BOUND_METHOD_EXACT_ARGS",
-    [CALL_BUILTIN_CLASS] = "CALL_BUILTIN_CLASS",
     [BINARY_SUBSCR] = "BINARY_SUBSCR",
     [BINARY_SLICE] = "BINARY_SLICE",
     [STORE_SLICE] = "STORE_SLICE",
-    [CALL_BUILTIN_FAST_WITH_KEYWORDS] = "CALL_BUILTIN_FAST_WITH_KEYWORDS",
-    [CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS] = "CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS",
+    [CALL_BOUND_METHOD_EXACT_ARGS] = "CALL_BOUND_METHOD_EXACT_ARGS",
+    [CALL_BUILTIN_CLASS] = "CALL_BUILTIN_CLASS",
     [GET_LEN] = "GET_LEN",
     [MATCH_MAPPING] = "MATCH_MAPPING",
     [MATCH_SEQUENCE] = "MATCH_SEQUENCE",
     [MATCH_KEYS] = "MATCH_KEYS",
-    [CALL_NO_KW_BUILTIN_FAST] = "CALL_NO_KW_BUILTIN_FAST",
+    [CALL_BUILTIN_FAST_WITH_KEYWORDS] = "CALL_BUILTIN_FAST_WITH_KEYWORDS",
     [PUSH_EXC_INFO] = "PUSH_EXC_INFO",
     [CHECK_EXC_MATCH] = "CHECK_EXC_MATCH",
     [CHECK_EG_MATCH] = "CHECK_EG_MATCH",
+    [CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS] = "CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS",
+    [CALL_NO_KW_BUILTIN_FAST] = "CALL_NO_KW_BUILTIN_FAST",
     [CALL_NO_KW_BUILTIN_O] = "CALL_NO_KW_BUILTIN_O",
     [CALL_NO_KW_ISINSTANCE] = "CALL_NO_KW_ISINSTANCE",
     [CALL_NO_KW_LEN] = "CALL_NO_KW_LEN",
@@ -265,8 +286,6 @@ static const char *const _PyOpcode_OpName[263] = {
     [CALL_NO_KW_METHOD_DESCRIPTOR_O] = "CALL_NO_KW_METHOD_DESCRIPTOR_O",
     [CALL_NO_KW_STR_1] = "CALL_NO_KW_STR_1",
     [CALL_NO_KW_TUPLE_1] = "CALL_NO_KW_TUPLE_1",
-    [CALL_NO_KW_TYPE_1] = "CALL_NO_KW_TYPE_1",
-    [COMPARE_OP_FLOAT] = "COMPARE_OP_FLOAT",
     [WITH_EXCEPT_START] = "WITH_EXCEPT_START",
     [GET_AITER] = "GET_AITER",
     [GET_ANEXT] = "GET_ANEXT",
@@ -274,39 +293,39 @@ static const char *const _PyOpcode_OpName[263] = {
     [BEFORE_WITH] = "BEFORE_WITH",
     [END_ASYNC_FOR] = "END_ASYNC_FOR",
     [CLEANUP_THROW] = "CLEANUP_THROW",
+    [CALL_NO_KW_TYPE_1] = "CALL_NO_KW_TYPE_1",
+    [COMPARE_OP_FLOAT] = "COMPARE_OP_FLOAT",
     [COMPARE_OP_INT] = "COMPARE_OP_INT",
     [COMPARE_OP_STR] = "COMPARE_OP_STR",
-    [FOR_ITER_LIST] = "FOR_ITER_LIST",
-    [FOR_ITER_TUPLE] = "FOR_ITER_TUPLE",
     [STORE_SUBSCR] = "STORE_SUBSCR",
     [DELETE_SUBSCR] = "DELETE_SUBSCR",
+    [FOR_ITER_LIST] = "FOR_ITER_LIST",
+    [FOR_ITER_TUPLE] = "FOR_ITER_TUPLE",
     [FOR_ITER_RANGE] = "FOR_ITER_RANGE",
     [FOR_ITER_GEN] = "FOR_ITER_GEN",
     [LOAD_ATTR_CLASS] = "LOAD_ATTR_CLASS",
     [LOAD_ATTR_GETATTRIBUTE_OVERRIDDEN] = "LOAD_ATTR_GETATTRIBUTE_OVERRIDDEN",
-    [LOAD_ATTR_INSTANCE_VALUE] = "LOAD_ATTR_INSTANCE_VALUE",
-    [LOAD_ATTR_MODULE] = "LOAD_ATTR_MODULE",
     [GET_ITER] = "GET_ITER",
     [GET_YIELD_FROM_ITER] = "GET_YIELD_FROM_ITER",
-    [LOAD_ATTR_PROPERTY] = "LOAD_ATTR_PROPERTY",
+    [LOAD_ATTR_INSTANCE_VALUE] = "LOAD_ATTR_INSTANCE_VALUE",
     [LOAD_BUILD_CLASS] = "LOAD_BUILD_CLASS",
-    [LOAD_ATTR_SLOT] = "LOAD_ATTR_SLOT",
-    [LOAD_ATTR_WITH_HINT] = "LOAD_ATTR_WITH_HINT",
+    [LOAD_ATTR_MODULE] = "LOAD_ATTR_MODULE",
+    [LOAD_ATTR_PROPERTY] = "LOAD_ATTR_PROPERTY",
     [LOAD_ASSERTION_ERROR] = "LOAD_ASSERTION_ERROR",
     [RETURN_GENERATOR] = "RETURN_GENERATOR",
+    [LOAD_ATTR_SLOT] = "LOAD_ATTR_SLOT",
+    [LOAD_ATTR_WITH_HINT] = "LOAD_ATTR_WITH_HINT",
     [LOAD_ATTR_METHOD_LAZY_DICT] = "LOAD_ATTR_METHOD_LAZY_DICT",
     [LOAD_ATTR_METHOD_NO_DICT] = "LOAD_ATTR_METHOD_NO_DICT",
     [LOAD_ATTR_METHOD_WITH_VALUES] = "LOAD_ATTR_METHOD_WITH_VALUES",
     [LOAD_CONST__LOAD_FAST] = "LOAD_CONST__LOAD_FAST",
     [LOAD_FAST__LOAD_CONST] = "LOAD_FAST__LOAD_CONST",
+    [RETURN_VALUE] = "RETURN_VALUE",
     [LOAD_FAST__LOAD_FAST] = "LOAD_FAST__LOAD_FAST",
+    [SETUP_ANNOTATIONS] = "SETUP_ANNOTATIONS",
     [LOAD_GLOBAL_BUILTIN] = "LOAD_GLOBAL_BUILTIN",
-    [RETURN_VALUE] = "RETURN_VALUE",
     [LOAD_GLOBAL_MODULE] = "LOAD_GLOBAL_MODULE",
-    [SETUP_ANNOTATIONS] = "SETUP_ANNOTATIONS",
     [STORE_ATTR_INSTANCE_VALUE] = "STORE_ATTR_INSTANCE_VALUE",
-    [STORE_ATTR_SLOT] = "STORE_ATTR_SLOT",
-    [STORE_ATTR_WITH_HINT] = "STORE_ATTR_WITH_HINT",
     [POP_EXCEPT] = "POP_EXCEPT",
     [STORE_NAME] = "STORE_NAME",
     [DELETE_NAME] = "DELETE_NAME",
@@ -329,9 +348,9 @@ static const char *const _PyOpcode_OpName[263] = {
     [IMPORT_NAME] = "IMPORT_NAME",
     [IMPORT_FROM] = "IMPORT_FROM",
     [JUMP_FORWARD] = "JUMP_FORWARD",
+    [STORE_ATTR_SLOT] = "STORE_ATTR_SLOT",
+    [STORE_ATTR_WITH_HINT] = "STORE_ATTR_WITH_HINT",
     [STORE_FAST__LOAD_FAST] = "STORE_FAST__LOAD_FAST",
-    [STORE_FAST__STORE_FAST] = "STORE_FAST__STORE_FAST",
-    [STORE_SUBSCR_DICT] = "STORE_SUBSCR_DICT",
     [POP_JUMP_IF_FALSE] = "POP_JUMP_IF_FALSE",
     [POP_JUMP_IF_TRUE] = "POP_JUMP_IF_TRUE",
     [LOAD_GLOBAL] = "LOAD_GLOBAL",
@@ -359,9 +378,9 @@ static const char *const _PyOpcode_OpName[263] = {
     [STORE_DEREF] = "STORE_DEREF",
     [DELETE_DEREF] = "DELETE_DEREF",
     [JUMP_BACKWARD] = "JUMP_BACKWARD",
-    [STORE_SUBSCR_LIST_INT] = "STORE_SUBSCR_LIST_INT",
+    [STORE_FAST__STORE_FAST] = "STORE_FAST__STORE_FAST",
     [CALL_FUNCTION_EX] = "CALL_FUNCTION_EX",
-    [UNPACK_SEQUENCE_LIST] = "UNPACK_SEQUENCE_LIST",
+    [STORE_SUBSCR_DICT] = "STORE_SUBSCR_DICT",
     [EXTENDED_ARG] = "EXTENDED_ARG",
     [LIST_APPEND] = "LIST_APPEND",
     [SET_ADD] = "SET_ADD",
@@ -371,14 +390,14 @@ static const char *const _PyOpcode_OpName[263] = {
     [YIELD_VALUE] = "YIELD_VALUE",
     [RESUME] = "RESUME",
     [MATCH_CLASS] = "MATCH_CLASS",
-    [UNPACK_SEQUENCE_TUPLE] = "UNPACK_SEQUENCE_TUPLE",
-    [UNPACK_SEQUENCE_TWO_TUPLE] = "UNPACK_SEQUENCE_TWO_TUPLE",
+    [STORE_SUBSCR_LIST_INT] = "STORE_SUBSCR_LIST_INT",
+    [UNPACK_SEQUENCE_LIST] = "UNPACK_SEQUENCE_LIST",
     [FORMAT_VALUE] = "FORMAT_VALUE",
     [BUILD_CONST_KEY_MAP] = "BUILD_CONST_KEY_MAP",
     [BUILD_STRING] = "BUILD_STRING",
+    [UNPACK_SEQUENCE_TUPLE] = "UNPACK_SEQUENCE_TUPLE",
+    [UNPACK_SEQUENCE_TWO_TUPLE] = "UNPACK_SEQUENCE_TWO_TUPLE",
     [SEND_GEN] = "SEND_GEN",
-    [159] = "<159>",
-    [160] = "<160>",
     [161] = "<161>",
     [LIST_EXTEND] = "LIST_EXTEND",
     [SET_UPDATE] = "SET_UPDATE",
@@ -456,24 +475,24 @@ static const char *const _PyOpcode_OpName[263] = {
     [235] = "<235>",
     [236] = "<236>",
     [237] = "<237>",
-    [238] = "<238>",
-    [239] = "<239>",
-    [240] = "<240>",
-    [241] = "<241>",
-    [242] = "<242>",
-    [243] = "<243>",
-    [244] = "<244>",
-    [245] = "<245>",
-    [246] = "<246>",
-    [247] = "<247>",
-    [248] = "<248>",
-    [249] = "<249>",
-    [250] = "<250>",
-    [251] = "<251>",
-    [252] = "<252>",
-    [253] = "<253>",
-    [254] = "<254>",
-    [DO_TRACING] = "DO_TRACING",
+    [INSTRUMENTED_POP_JUMP_IF_NONE] = "INSTRUMENTED_POP_JUMP_IF_NONE",
+    [INSTRUMENTED_POP_JUMP_IF_NOT_NONE] = "INSTRUMENTED_POP_JUMP_IF_NOT_NONE",
+    [INSTRUMENTED_RESUME] = "INSTRUMENTED_RESUME",
+    [INSTRUMENTED_CALL] = "INSTRUMENTED_CALL",
+    [INSTRUMENTED_RETURN_VALUE] = "INSTRUMENTED_RETURN_VALUE",
+    [INSTRUMENTED_YIELD_VALUE] = "INSTRUMENTED_YIELD_VALUE",
+    [INSTRUMENTED_CALL_FUNCTION_EX] = "INSTRUMENTED_CALL_FUNCTION_EX",
+    [INSTRUMENTED_JUMP_FORWARD] = "INSTRUMENTED_JUMP_FORWARD",
+    [INSTRUMENTED_JUMP_BACKWARD] = "INSTRUMENTED_JUMP_BACKWARD",
+    [INSTRUMENTED_RETURN_CONST] = "INSTRUMENTED_RETURN_CONST",
+    [INSTRUMENTED_FOR_ITER] = "INSTRUMENTED_FOR_ITER",
+    [INSTRUMENTED_POP_JUMP_IF_FALSE] = "INSTRUMENTED_POP_JUMP_IF_FALSE",
+    [INSTRUMENTED_POP_JUMP_IF_TRUE] = "INSTRUMENTED_POP_JUMP_IF_TRUE",
+    [INSTRUMENTED_END_FOR] = "INSTRUMENTED_END_FOR",
+    [INSTRUMENTED_END_SEND] = "INSTRUMENTED_END_SEND",
+    [INSTRUMENTED_INSTRUCTION] = "INSTRUMENTED_INSTRUCTION",
+    [INSTRUMENTED_LINE] = "INSTRUMENTED_LINE",
+    [255] = "<255>",
     [SETUP_FINALLY] = "SETUP_FINALLY",
     [SETUP_CLEANUP] = "SETUP_CLEANUP",
     [SETUP_WITH] = "SETUP_WITH",
@@ -485,8 +504,6 @@ static const char *const _PyOpcode_OpName[263] = {
 #endif
 
 #define EXTRA_CASES \
-    case 159: \
-    case 160: \
     case 161: \
     case 166: \
     case 167: \
@@ -556,23 +573,7 @@ static const char *const _PyOpcode_OpName[263] = {
     case 235: \
     case 236: \
     case 237: \
-    case 238: \
-    case 239: \
-    case 240: \
-    case 241: \
-    case 242: \
-    case 243: \
-    case 244: \
-    case 245: \
-    case 246: \
-    case 247: \
-    case 248: \
-    case 249: \
-    case 250: \
-    case 251: \
-    case 252: \
-    case 253: \
-    case 254: \
+    case 255: \
         ;
 
 #ifdef __cplusplus
diff --git a/Include/internal/pycore_pystate.h b/Include/internal/pycore_pystate.h
index b540862..6e5f228 100644
--- a/Include/internal/pycore_pystate.h
+++ b/Include/internal/pycore_pystate.h
@@ -133,16 +133,6 @@ extern void _PyThreadState_BindDetached(PyThreadState *);
 extern void _PyThreadState_UnbindDetached(PyThreadState *);
 
 
-static inline void
-_PyThreadState_UpdateTracingState(PyThreadState *tstate)
-{
-    bool use_tracing =
-        (tstate->tracing == 0) &&
-        (tstate->c_tracefunc != NULL || tstate->c_profilefunc != NULL);
-    tstate->cframe->use_tracing = (use_tracing ? 255 : 0);
-}
-
-
 /* Other */
 
 PyAPI_FUNC(PyThreadState *) _PyThreadState_Swap(
diff --git a/Include/opcode.h b/Include/opcode.h
index 0ff84dc..aa8716e 100644
--- a/Include/opcode.h
+++ b/Include/opcode.h
@@ -13,10 +13,12 @@ extern "C" {
 #define PUSH_NULL                                2
 #define INTERPRETER_EXIT                         3
 #define END_FOR                                  4
+#define END_SEND                                 5
 #define NOP                                      9
 #define UNARY_NEGATIVE                          11
 #define UNARY_NOT                               12
 #define UNARY_INVERT                            15
+#define RESERVED                                17
 #define BINARY_SUBSCR                           25
 #define BINARY_SLICE                            26
 #define STORE_SLICE                             27
@@ -114,6 +116,24 @@ extern "C" {
 #define KW_NAMES                               172
 #define CALL_INTRINSIC_1                       173
 #define CALL_INTRINSIC_2                       174
+#define MIN_INSTRUMENTED_OPCODE                238
+#define INSTRUMENTED_POP_JUMP_IF_NONE          238
+#define INSTRUMENTED_POP_JUMP_IF_NOT_NONE      239
+#define INSTRUMENTED_RESUME                    240
+#define INSTRUMENTED_CALL                      241
+#define INSTRUMENTED_RETURN_VALUE              242
+#define INSTRUMENTED_YIELD_VALUE               243
+#define INSTRUMENTED_CALL_FUNCTION_EX          244
+#define INSTRUMENTED_JUMP_FORWARD              245
+#define INSTRUMENTED_JUMP_BACKWARD             246
+#define INSTRUMENTED_RETURN_CONST              247
+#define INSTRUMENTED_FOR_ITER                  248
+#define INSTRUMENTED_POP_JUMP_IF_FALSE         249
+#define INSTRUMENTED_POP_JUMP_IF_TRUE          250
+#define INSTRUMENTED_END_FOR                   251
+#define INSTRUMENTED_END_SEND                  252
+#define INSTRUMENTED_INSTRUCTION               253
+#define INSTRUMENTED_LINE                      254
 #define MIN_PSEUDO_OPCODE                      256
 #define SETUP_FINALLY                          256
 #define SETUP_CLEANUP                          257
@@ -123,69 +143,68 @@ extern "C" {
 #define JUMP_NO_INTERRUPT                      261
 #define LOAD_METHOD                            262
 #define MAX_PSEUDO_OPCODE                      262
-#define BINARY_OP_ADD_FLOAT                      5
-#define BINARY_OP_ADD_INT                        6
-#define BINARY_OP_ADD_UNICODE                    7
-#define BINARY_OP_INPLACE_ADD_UNICODE            8
-#define BINARY_OP_MULTIPLY_FLOAT                10
-#define BINARY_OP_MULTIPLY_INT                  13
-#define BINARY_OP_SUBTRACT_FLOAT                14
-#define BINARY_OP_SUBTRACT_INT                  16
-#define BINARY_SUBSCR_DICT                      17
-#define BINARY_SUBSCR_GETITEM                   18
-#define BINARY_SUBSCR_LIST_INT                  19
-#define BINARY_SUBSCR_TUPLE_INT                 20
-#define CALL_PY_EXACT_ARGS                      21
-#define CALL_PY_WITH_DEFAULTS                   22
-#define CALL_BOUND_METHOD_EXACT_ARGS            23
-#define CALL_BUILTIN_CLASS                      24
-#define CALL_BUILTIN_FAST_WITH_KEYWORDS         28
-#define CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS  29
-#define CALL_NO_KW_BUILTIN_FAST                 34
-#define CALL_NO_KW_BUILTIN_O                    38
-#define CALL_NO_KW_ISINSTANCE                   39
-#define CALL_NO_KW_LEN                          40
-#define CALL_NO_KW_LIST_APPEND                  41
-#define CALL_NO_KW_METHOD_DESCRIPTOR_FAST       42
-#define CALL_NO_KW_METHOD_DESCRIPTOR_NOARGS     43
-#define CALL_NO_KW_METHOD_DESCRIPTOR_O          44
-#define CALL_NO_KW_STR_1                        45
-#define CALL_NO_KW_TUPLE_1                      46
-#define CALL_NO_KW_TYPE_1                       47
-#define COMPARE_OP_FLOAT                        48
-#define COMPARE_OP_INT                          56
-#define COMPARE_OP_STR                          57
-#define FOR_ITER_LIST                           58
-#define FOR_ITER_TUPLE                          59
-#define FOR_ITER_RANGE                          62
-#define FOR_ITER_GEN                            63
-#define LOAD_ATTR_CLASS                         64
-#define LOAD_ATTR_GETATTRIBUTE_OVERRIDDEN       65
-#define LOAD_ATTR_INSTANCE_VALUE                66
-#define LOAD_ATTR_MODULE                        67
-#define LOAD_ATTR_PROPERTY                      70
-#define LOAD_ATTR_SLOT                          72
-#define LOAD_ATTR_WITH_HINT                     73
-#define LOAD_ATTR_METHOD_LAZY_DICT              76
-#define LOAD_ATTR_METHOD_NO_DICT                77
-#define LOAD_ATTR_METHOD_WITH_VALUES            78
-#define LOAD_CONST__LOAD_FAST                   79
-#define LOAD_FAST__LOAD_CONST                   80
-#define LOAD_FAST__LOAD_FAST                    81
-#define LOAD_GLOBAL_BUILTIN                     82
-#define LOAD_GLOBAL_MODULE                      84
-#define STORE_ATTR_INSTANCE_VALUE               86
-#define STORE_ATTR_SLOT                         87
-#define STORE_ATTR_WITH_HINT                    88
-#define STORE_FAST__LOAD_FAST                  111
-#define STORE_FAST__STORE_FAST                 112
-#define STORE_SUBSCR_DICT                      113
-#define STORE_SUBSCR_LIST_INT                  141
-#define UNPACK_SEQUENCE_LIST                   143
-#define UNPACK_SEQUENCE_TUPLE                  153
-#define UNPACK_SEQUENCE_TWO_TUPLE              154
-#define SEND_GEN                               158
-#define DO_TRACING                             255
+#define BINARY_OP_ADD_FLOAT                      6
+#define BINARY_OP_ADD_INT                        7
+#define BINARY_OP_ADD_UNICODE                    8
+#define BINARY_OP_INPLACE_ADD_UNICODE           10
+#define BINARY_OP_MULTIPLY_FLOAT                13
+#define BINARY_OP_MULTIPLY_INT                  14
+#define BINARY_OP_SUBTRACT_FLOAT                16
+#define BINARY_OP_SUBTRACT_INT                  18
+#define BINARY_SUBSCR_DICT                      19
+#define BINARY_SUBSCR_GETITEM                   20
+#define BINARY_SUBSCR_LIST_INT                  21
+#define BINARY_SUBSCR_TUPLE_INT                 22
+#define CALL_PY_EXACT_ARGS                      23
+#define CALL_PY_WITH_DEFAULTS                   24
+#define CALL_BOUND_METHOD_EXACT_ARGS            28
+#define CALL_BUILTIN_CLASS                      29
+#define CALL_BUILTIN_FAST_WITH_KEYWORDS         34
+#define CALL_METHOD_DESCRIPTOR_FAST_WITH_KEYWORDS  38
+#define CALL_NO_KW_BUILTIN_FAST                 39
+#define CALL_NO_KW_BUILTIN_O                    40
+#define CALL_NO_KW_ISINSTANCE                   41
+#define CALL_NO_KW_LEN                          42
+#define CALL_NO_KW_LIST_APPEND                  43
+#define CALL_NO_KW_METHOD_DESCRIPTOR_FAST       44
+#define CALL_NO_KW_METHOD_DESCRIPTOR_NOARGS     45
+#define CALL_NO_KW_METHOD_DESCRIPTOR_O          46
+#define CALL_NO_KW_STR_1                        47
+#define CALL_NO_KW_TUPLE_1                      48
+#define CALL_NO_KW_TYPE_1                       56
+#define COMPARE_OP_FLOAT                        57
+#define COMPARE_OP_INT                          58
+#define COMPARE_OP_STR                          59
+#define FOR_ITER_LIST                           62
+#define FOR_ITER_TUPLE                          63
+#define FOR_ITER_RANGE                          64
+#define FOR_ITER_GEN                            65
+#define LOAD_ATTR_CLASS                         66
+#define LOAD_ATTR_GETATTRIBUTE_OVERRIDDEN       67
+#define LOAD_ATTR_INSTANCE_VALUE                70
+#define LOAD_ATTR_MODULE                        72
+#define LOAD_ATTR_PROPERTY                      73
+#define LOAD_ATTR_SLOT                          76
+#define LOAD_ATTR_WITH_HINT                     77
+#define LOAD_ATTR_METHOD_LAZY_DICT              78
+#define LOAD_ATTR_METHOD_NO_DICT                79
+#define LOAD_ATTR_METHOD_WITH_VALUES            80
+#define LOAD_CONST__LOAD_FAST                   81
+#define LOAD_FAST__LOAD_CONST                   82
+#define LOAD_FAST__LOAD_FAST                    84
+#define LOAD_GLOBAL_BUILTIN                     86
+#define LOAD_GLOBAL_MODULE                      87
+#define STORE_ATTR_INSTANCE_VALUE               88
+#define STORE_ATTR_SLOT                        111
+#define STORE_ATTR_WITH_HINT                   112
+#define STORE_FAST__LOAD_FAST                  113
+#define STORE_FAST__STORE_FAST                 141
+#define STORE_SUBSCR_DICT                      143
+#define STORE_SUBSCR_LIST_INT                  153
+#define UNPACK_SEQUENCE_LIST                   154
+#define UNPACK_SEQUENCE_TUPLE                  158
+#define UNPACK_SEQUENCE_TWO_TUPLE              159
+#define SEND_GEN                               160
 
 #define HAS_ARG(op) ((((op) >= HAVE_ARGUMENT) && (!IS_PSEUDO_OPCODE(op)))\
     || ((op) == JUMP) \