pythongh-108753: Enhance pystats

* Add PYTHONSTATS environment variable.
* Add PyConfig._pystats member.
* Add tests on sys functions and on setting PyConfig._pystats to 1.
* At exit, only dump statistics if PyConfig._pystats is non-zero.
* Add Include/cpython/pystats.h and Include/internal/pycore_pystats.h
  header files.
* Rename '_py_stats' variable to '_Py_stats'.
* Exclude Include/cpython/pystats.h from the Py_LIMITED_API.
* Move pystats.h include from object.h to Python.h.
* Add _Py_StatsOn() and _Py_StatsOff() functions. Remove
  '_py_stats_struct' variable from the API: make it static in
  specialize.c.
* Document API in Include/pystats.h and Include/cpython/pystats.h.
* Complete pystats documentation in Doc/using/configure.rst.

Doc/using/configure.rst

@@ -192,14 +192,66 @@ General Options
 
 .. cmdoption:: --enable-pystats
 
-   Turn on internal statistics gathering.
+   Turn on internal Python performance statistics gathering.
+
+   By default, statistics gathering is off. Use ``python3 -X pystats`` command
+   or set ``PYTHONSTATS=1`` environment variable to turn on statistics
+   gathering at Python startup and to dump statistics at Python exit.
+
+   Effects:
+
+   * Add :option:`-X pystats <-X>` command line option.
+   * Add :envvar:`PYTHONSTATS` environment variable.
+   * Define the ``Py_STATS`` macro.
+   * Add functions to the :mod:`sys` module:
+
+     * :func:`sys._stats_on`: Turns on statistics gathering.
+     * :func:`sys._stats_off`: Turns off statistics gathering.
+     * :func:`sys._stats_clear`: Clears the statistics.
+     * :func:`sys._stats_dump`: Dump statistics to file, and clears the statistics.
 
    The statistics will be dumped to a arbitrary (probably unique) file in
-   ``/tmp/py_stats/``, or ``C:\temp\py_stats\`` on Windows. If that directory
-   does not exist, results will be printed on stdout.
+   ``/tmp/py_stats/`` (Unix) or ``C:\temp\py_stats\`` (Windows). If that
+   directory does not exist, results will be printed on stdout.
 
    Use ``Tools/scripts/summarize_stats.py`` to read the stats.
 
+   Statistics:
+
+   * Opcode:
+
+     * Specialization: success, failure, hit, deferred, miss, deopt, failures;
+     * Execution count;
+     * Pair count.
+
+   * Call:
+
+     * Inlined Python calls;
+     * PyEval calls;
+     * Frames pushed;
+     * Frame object created;
+     * Eval calls: vector, generator, legacy, function VECTORCALL, build class,
+       slot, function "ex", API, method.
+
+   * Object:
+
+     * incref and decref;
+     * interpreter incref and decref;
+     * allocations: all, 512 bytes, 4 kiB, big;
+     * free;
+     * to/from free lists;
+     * dictionary materialized/dematerialized;
+     * type cache;
+     * optimization attemps;
+     * optimization traces created/executed;
+     * uops executed.
+
+   * Garbage collector:
+
+     * Garbage collections;
+     * Objects visited;
+     * Objects collected.
+
    .. versionadded:: 3.11
 
 .. cmdoption:: --disable-gil

Include/Python.h

@@ -41,6 +41,7 @@
 #include "pymem.h"
 #include "pytypedefs.h"
 #include "pybuffer.h"
+#include "pystats.h"
 #include "object.h"
 #include "objimpl.h"
 #include "typeslots.h"

Include/cpython/initconfig.h

@@ -215,6 +215,11 @@ typedef struct PyConfig {
 
     // If non-zero, we believe we're running from a source tree.
     int _is_python_build;
+
+#ifdef Py_STATS
+    // If non-zero, turns on statistics gathering.
+    int _pystats;
+#endif
 } PyConfig;
 
 PyAPI_FUNC(void) PyConfig_InitPythonConfig(PyConfig *config);

Include/cpython/pystats.h

@@ -0,0 +1,120 @@
+// Statistics on Python performance.
+//
+// API:
+//
+// - _Py_INCREF_STAT_INC() and _Py_DECREF_STAT_INC() used by Py_INCREF()
+//   and Py_DECREF().
+// - _Py_stats variable
+//
+// Functions of the sys module:
+//
+// - sys._stats_on()
+// - sys._stats_off()
+// - sys._stats_clear()
+// - sys._stats_dump()
+//
+// Python must be built with ./configure --enable-pystats to define the
+// Py_STATS macro.
+//
+// Define _PY_INTERPRETER macro to increment interpreter_increfs and
+// interpreter_decrefs. Otherwise, increment increfs and decrefs.
+
+#ifndef Py_CPYTHON_PYSTATS_H
+#  error "this header file must not be included directly"
+#endif
+
+#define SPECIALIZATION_FAILURE_KINDS 36
+
+/* Stats for determining who is calling PyEval_EvalFrame */
+#define EVAL_CALL_TOTAL 0
+#define EVAL_CALL_VECTOR 1
+#define EVAL_CALL_GENERATOR 2
+#define EVAL_CALL_LEGACY 3
+#define EVAL_CALL_FUNCTION_VECTORCALL 4
+#define EVAL_CALL_BUILD_CLASS 5
+#define EVAL_CALL_SLOT 6
+#define EVAL_CALL_FUNCTION_EX 7
+#define EVAL_CALL_API 8
+#define EVAL_CALL_METHOD 9
+
+#define EVAL_CALL_KINDS 10
+
+typedef struct _specialization_stats {
+    uint64_t success;
+    uint64_t failure;
+    uint64_t hit;
+    uint64_t deferred;
+    uint64_t miss;
+    uint64_t deopt;
+    uint64_t failure_kinds[SPECIALIZATION_FAILURE_KINDS];
+} SpecializationStats;
+
+typedef struct _opcode_stats {
+    SpecializationStats specialization;
+    uint64_t execution_count;
+    uint64_t pair_count[256];
+} OpcodeStats;
+
+typedef struct _call_stats {
+    uint64_t inlined_py_calls;
+    uint64_t pyeval_calls;
+    uint64_t frames_pushed;
+    uint64_t frame_objects_created;
+    uint64_t eval_calls[EVAL_CALL_KINDS];
+} CallStats;
+
+typedef struct _object_stats {
+    uint64_t increfs;
+    uint64_t decrefs;
+    uint64_t interpreter_increfs;
+    uint64_t interpreter_decrefs;
+    uint64_t allocations;
+    uint64_t allocations512;
+    uint64_t allocations4k;
+    uint64_t allocations_big;
+    uint64_t frees;
+    uint64_t to_freelist;
+    uint64_t from_freelist;
+    uint64_t new_values;
+    uint64_t dict_materialized_on_request;
+    uint64_t dict_materialized_new_key;
+    uint64_t dict_materialized_too_big;
+    uint64_t dict_materialized_str_subclass;
+    uint64_t dict_dematerialized;
+    uint64_t type_cache_hits;
+    uint64_t type_cache_misses;
+    uint64_t type_cache_dunder_hits;
+    uint64_t type_cache_dunder_misses;
+    uint64_t type_cache_collisions;
+    uint64_t optimization_attempts;
+    uint64_t optimization_traces_created;
+    uint64_t optimization_traces_executed;
+    uint64_t optimization_uops_executed;
+    /* Temporary value used during GC */
+    uint64_t object_visits;
+} ObjectStats;
+
+typedef struct _gc_stats {
+    uint64_t collections;
+    uint64_t object_visits;
+    uint64_t objects_collected;
+} GCStats;
+
+typedef struct _stats {
+    OpcodeStats opcode_stats[256];
+    CallStats call_stats;
+    ObjectStats object_stats;
+    GCStats *gc_stats;
+} PyStats;
+
+
+// Export for shared extensions like 'math'
+PyAPI_DATA(PyStats*) _Py_stats;
+
+#ifdef _PY_INTERPRETER
+#  define _Py_INCREF_STAT_INC() do { if (_Py_stats) _Py_stats->object_stats.interpreter_increfs++; } while (0)
+#  define _Py_DECREF_STAT_INC() do { if (_Py_stats) _Py_stats->object_stats.interpreter_decrefs++; } while (0)
+#else
+#  define _Py_INCREF_STAT_INC() do { if (_Py_stats) _Py_stats->object_stats.increfs++; } while (0)
+#  define _Py_DECREF_STAT_INC() do { if (_Py_stats) _Py_stats->object_stats.decrefs++; } while (0)
+#endif

Include/internal/pycore_ceval.h

@@ -69,7 +69,7 @@ extern int _PyEval_SetAsyncGenFinalizer(PyObject *);
 extern int _PyEval_GetCoroutineOriginTrackingDepth(void);
 extern int _PyEval_SetCoroutineOriginTrackingDepth(int depth);
 
-extern void _PyEval_Fini(void);
+extern void _PyEval_Fini(PyThreadState *tstate);
 
 
 extern PyObject* _PyEval_GetBuiltins(PyThreadState *tstate);

Include/internal/pycore_code.h

@@ -268,17 +268,17 @@ extern int _PyStaticCode_Init(PyCodeObject *co);
 
 #ifdef Py_STATS
 
-#define STAT_INC(opname, name) do { if (_py_stats) _py_stats->opcode_stats[opname].specialization.name++; } while (0)
-#define STAT_DEC(opname, name) do { if (_py_stats) _py_stats->opcode_stats[opname].specialization.name--; } while (0)
-#define OPCODE_EXE_INC(opname) do { if (_py_stats) _py_stats->opcode_stats[opname].execution_count++; } while (0)
-#define CALL_STAT_INC(name) do { if (_py_stats) _py_stats->call_stats.name++; } while (0)
-#define OBJECT_STAT_INC(name) do { if (_py_stats) _py_stats->object_stats.name++; } while (0)
+#define STAT_INC(opname, name) do { if (_Py_stats) _Py_stats->opcode_stats[opname].specialization.name++; } while (0)
+#define STAT_DEC(opname, name) do { if (_Py_stats) _Py_stats->opcode_stats[opname].specialization.name--; } while (0)
+#define OPCODE_EXE_INC(opname) do { if (_Py_stats) _Py_stats->opcode_stats[opname].execution_count++; } while (0)
+#define CALL_STAT_INC(name) do { if (_Py_stats) _Py_stats->call_stats.name++; } while (0)
+#define OBJECT_STAT_INC(name) do { if (_Py_stats) _Py_stats->object_stats.name++; } while (0)
 #define OBJECT_STAT_INC_COND(name, cond) \
-    do { if (_py_stats && cond) _py_stats->object_stats.name++; } while (0)
-#define EVAL_CALL_STAT_INC(name) do { if (_py_stats) _py_stats->call_stats.eval_calls[name]++; } while (0)
+    do { if (_Py_stats && cond) _Py_stats->object_stats.name++; } while (0)
+#define EVAL_CALL_STAT_INC(name) do { if (_Py_stats) _Py_stats->call_stats.eval_calls[name]++; } while (0)
 #define EVAL_CALL_STAT_INC_IF_FUNCTION(name, callable) \
-    do { if (_py_stats && PyFunction_Check(callable)) _py_stats->call_stats.eval_calls[name]++; } while (0)
-#define GC_STAT_ADD(gen, name, n) do { if (_py_stats) _py_stats->gc_stats[(gen)].name += (n); } while (0)
+    do { if (_Py_stats && PyFunction_Check(callable)) _Py_stats->call_stats.eval_calls[name]++; } while (0)
+#define GC_STAT_ADD(gen, name, n) do { if (_Py_stats) _Py_stats->gc_stats[(gen)].name += (n); } while (0)
 
 // Export for '_opcode' shared extension
 PyAPI_FUNC(PyObject*) _Py_GetSpecializationStats(void);

Include/internal/pycore_pystats.h

@@ -0,0 +1,21 @@
+#ifndef Py_INTERNAL_PYSTATS_H
+#define Py_INTERNAL_PYSTATS_H
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef Py_BUILD_CORE
+#  error "this header requires Py_BUILD_CORE define"
+#endif
+
+#ifdef Py_STATS
+extern void _Py_StatsOn(void);
+extern void _Py_StatsOff(void);
+extern void _Py_StatsClear(void);
+extern void _Py_PrintSpecializationStats(int to_file);
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+#endif  // !Py_INTERNAL_PYSTATS_H

Include/object.h

@@ -51,8 +51,6 @@ A standard interface exists for objects that contain an array of items
 whose size is determined when the object is allocated.
 */
 
-#include "pystats.h"
-
 /* Py_DEBUG implies Py_REF_DEBUG. */
 #if defined(Py_DEBUG) && !defined(Py_REF_DEBUG)
 #  define Py_REF_DEBUG