Merge pull request #187 from P403n1x87/feat/python312

feat: add support for Python 3.12

.github/workflows/checks.yml

@@ -156,8 +156,7 @@ jobs:
           sudo apt-get update
           sudo apt-get -y install \
             valgrind \
-            python2.7 \
-            python3.{5..11} \
+            python3.{8..12} \
             python3.10-full python3.10-dev
 
           python3.10 -m venv .venv
@@ -175,8 +174,8 @@ jobs:
         run: |
           echo "core.%p" | sudo tee /proc/sys/kernel/core_pattern
           ulimit -c unlimited
-          .venv/bin/pytest --pastebin=failed -sr fE -n auto || true
-          sudo -E env PATH="$PATH" .venv/bin/pytest --pastebin=failed -sr fE -n auto || true
+          .venv/bin/pytest -sr fE -n auto || true
+          sudo -E env PATH="$PATH" .venv/bin/pytest -sr fE -n auto || true
 
       - name: Generate Cobertura report
         run: gcovr --xml ./cobertura.xml -r src/

.github/workflows/tests.yml

@@ -75,7 +75,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     env:
       AUSTIN_TESTS_PYTHON_VERSIONS: ${{ matrix.python-version }}
@@ -251,7 +251,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     env:
       AUSTIN_TESTS_PYTHON_VERSIONS: ${{ matrix.python-version }}
@@ -377,7 +377,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     env:
       AUSTIN_TESTS_PYTHON_VERSIONS: ${{ matrix.python-version }}
@@ -466,7 +466,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     env:
       AUSTIN_TESTS_PYTHON_VERSIONS: ${{ matrix.python-version }}

scripts/build-wheel.py

@@ -20,6 +20,7 @@
         "Programming Language :: Python :: 3.9",
         "Programming Language :: Python :: 3.10",
         "Programming Language :: Python :: 3.11",
+        "Programming Language :: Python :: 3.12",
     ],
     "Project-URL": [
         "Homepage, https://github.com/P403n1x87/austin",

src/argparse.c

@@ -45,7 +45,7 @@
 #else
 #define DEFAULT_SAMPLING_INTERVAL    100
 #endif
-#define DEFAULT_INIT_TIMEOUT_MS     1000  // 1 seconds
+#define DEFAULT_INIT_TIMEOUT_MS     1000  // 1 second
 #define DEFAULT_HEAP_SIZE              0
 
 const char SAMPLE_FORMAT_NORMAL[]      = ";%s:%s:%d";

src/py_proc.c

@@ -1157,13 +1157,25 @@ py_proc__sample(py_proc_t * self) {
 
     if (pargs.memory) {
       // Use the current thread to determine which thread is manipulating memory
-      current_thread = _py_proc__get_current_thread_state_raddr(self);
+      if (V_MIN(3, 12)) {
+        void * gil_state_raddr = V_FIELD(void *, is, py_is, o_gil_state);
+        if (!isvalid(gil_state_raddr))
+          SUCCESS;
+        gil_state_t gil_state;
+        if (fail(copy_datatype(self->proc_ref, gil_state_raddr, gil_state))) {
+          log_ie("Failed to copy GIL state");
+          FAIL;
+        }
+        current_thread = (void *) gil_state.last_holder._value;
+      }
+      else
+        current_thread = _py_proc__get_current_thread_state_raddr(self);
     }
 
     do {
       if (pargs.memory) {
         mem_delta = 0;
-        if (self->symbols[DYNSYM_RUNTIME] != NULL && current_thread == (void *) -1) {
+        if (V_MAX(3, 11) && self->symbols[DYNSYM_RUNTIME] != NULL && current_thread == (void *) -1) {
           if (_py_proc__find_current_thread_offset(self, py_thread.raddr.addr))
             continue;
           else

src/py_string.h

@@ -35,7 +35,7 @@
 
 #define MAGIC_TINY                            7
 #define MAGIC_BIG                       1000003
-#define p_ascii_data(raddr)                    (raddr + sizeof(PyASCIIObject))
+#define p_ascii_data(raddr, size)              (raddr + size)
 
 
 // ----------------------------------------------------------------------------
@@ -57,7 +57,7 @@ string__hash(char * string) {
 // ----------------------------------------------------------------------------
 static inline char *
 _string_from_raddr(proc_ref_t pref, void * raddr, python_v * py_v) {
-  PyUnicodeObject3   unicode;
+  PyUnicodeObject    unicode;
   char             * buffer = NULL;
   ssize_t            len = 0;
 
@@ -66,15 +66,22 @@ _string_from_raddr(proc_ref_t pref, void * raddr, python_v * py_v) {
     goto failed;
   }
 
-  PyASCIIObject ascii = unicode._base._base;
+  PyASCIIObject ascii = unicode.v3._base._base;
 
   if (ascii.state.kind != 1) {
     set_error(ECODEFMT);
     goto failed;
   }
-
-  void * data = ascii.state.compact ? p_ascii_data(raddr) : unicode._base.utf8;
-  len = ascii.state.compact ? ascii.length : unicode._base.utf8_length;
+
+  // Because changes to PyASCIIObject are rare, we handle the version manually
+  // instead of using a version offset descriptor.
+  ssize_t ascii_size = V_MIN(3, 12) ? sizeof(unicode.v3_12._base._base) : sizeof(unicode.v3._base._base);
+  void * data = ascii.state.compact
+    ? p_ascii_data(raddr, ascii_size)
+    : (V_MIN(3, 12) ? unicode.v3_12._base.utf8 : unicode.v3._base.utf8);
+  len = ascii.state.compact
+    ? ascii.length
+    : (V_MIN(3, 12) ? unicode.v3_12._base.utf8_length : unicode.v3._base.utf8_length);
 
   if (len < 0 || len > 4096) {
     log_e("Invalid string length");

src/py_thread.c

@@ -317,14 +317,24 @@ _py_thread__push_iframe_from_addr(py_thread_t * self, PyInterpreterFrame * ifram
     FAIL;
   }
 
+  if (V_MIN(3, 12) && V_FIELD_PTR(char, iframe, py_iframe, o_owner) == FRAME_OWNED_BY_CSTACK) {
+    // This is a shim frame that we can ignore
+    #ifdef NATIVE
+    // In native mode we take this as the marker for the beginning of the stack
+    // for a call to PyEval_EvalFrameDefault.
+    stack_py_push_cframe();
+    #endif
+    SUCCESS;
+  }
+
   stack_py_push(
     origin,
     code_raddr,
     (((int)(V_FIELD_PTR(void *, iframe, py_iframe, o_prev_instr) - code_raddr)) - py_v->py_code.o_code) / sizeof(_Py_CODEUNIT)
   );
 
   #ifdef NATIVE
-  if (V_MIN(3, 11) && V_FIELD_PTR(int, iframe, py_iframe, o_is_entry)) {
+  if (V_EQ(3, 11) && V_FIELD_PTR(int, iframe, py_iframe, o_is_entry)) {
     // This marks the end of a CFrame
     stack_py_push_cframe();
   }
@@ -797,7 +807,9 @@ py_thread__fill_from_raddr(py_thread_t * self, raddr_t * raddr, py_proc_t * proc
   self->raddr = *raddr;
 
   self->top_frame = V_FIELD(void*, ts, py_thread, o_frame);
-
+
+  self->status = V_FIELD(tstate_status_t, ts, py_thread, o_status);
+
   self->next_raddr = (raddr_t) {
     raddr->pref,
     V_FIELD(void*, ts, py_thread, o_next) == raddr->addr \

src/py_thread.h

@@ -61,6 +61,8 @@ typedef struct thread {
   /* The per-thread datastack was introduced in Python 3.11 */
   void           * stack;
   size_t           stack_size;
+
+  tstate_status_t  status;
 } py_thread_t;
 
 

src/python/cframe.h

@@ -40,8 +40,16 @@ typedef struct _PyCFrame3_11 {
 } _PyCFrame3_11;
 
 
+typedef struct {
+    /* Pointer to the currently executing frame (it can be NULL) */
+    void *current_frame;
+    void *previous;
+} _PyCFrame3_12;
+
+
 typedef union {
     _PyCFrame3_11 v3_11;
+    _PyCFrame3_12 v3_12;
 } PyCFrame;
 
 #endif

src/python/code.h

@@ -92,9 +92,57 @@ typedef struct {
 typedef struct _PyCode_DEF_311(1) PyCodeObject3_11;
 
 
+#define _PyCode_DEF_312(SIZE) {                                                \
+    PyObject_VAR_HEAD                                                          \
+                                                                               \
+    PyObject *co_consts;           /* list (constants used) */                 \
+    PyObject *co_names;            /* list of strings (names used) */          \
+    PyObject *co_exceptiontable;   /* Byte string encoding exception handling  \
+                                      table */                                 \
+    int co_flags;                  /* CO_..., see below */                     \
+                                                                               \
+    /* The rest are not so impactful on performance. */                        \
+    int co_argcount;              /* #arguments, except *args */               \
+    int co_posonlyargcount;       /* #positional only arguments */             \
+    int co_kwonlyargcount;        /* #keyword only arguments */                \
+    int co_stacksize;             /* #entries needed for evaluation stack */   \
+    int co_firstlineno;           /* first source line number */               \
+                                                                               \
+    /* redundant values (derived from co_localsplusnames and                   \
+       co_localspluskinds) */                                                  \
+    int co_nlocalsplus;           /* number of local + cell + free variables */ \
+    int co_framesize;             /* Size of frame in words */                 \
+    int co_nlocals;               /* number of local variables */              \
+    int co_ncellvars;             /* total number of cell variables */         \
+    int co_nfreevars;             /* number of free variables */               \
+    uint32_t co_version;          /* version number */                         \
+                                                                               \
+    PyObject *co_localsplusnames; /* tuple mapping offsets to names */         \
+    PyObject *co_localspluskinds; /* Bytes mapping to local kinds (one byte    \
+                                     per variable) */                          \
+    PyObject *co_filename;        /* unicode (where it was loaded from) */     \
+    PyObject *co_name;            /* unicode (name, for reference) */          \
+    PyObject *co_qualname;        /* unicode (qualname, for reference) */      \
+    PyObject *co_linetable;       /* bytes object that holds location info */  \
+    PyObject *co_weakreflist;     /* to support weakrefs to code objects */    \
+    void *co_executors;      /* executors from optimizer */                    \
+    void *_co_cached;      /* cached co_* attributes */                        \
+    uint64_t _co_instrumentation_version; /* current instrumentation version */  \
+    void *_co_monitoring; /* Monitoring data */                                \
+    int _co_firsttraceable;       /* index of first traceable instruction */   \
+    /* 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. */                                \
+    void *co_extra;                                                            \
+    char co_code_adaptive[(SIZE)];                                             \
+}
+typedef struct _PyCode_DEF_312(1) PyCodeObject3_12;
+
+
 typedef union {
   PyCodeObject3_8  v3_8;
   PyCodeObject3_11 v3_11;
+  PyCodeObject3_12 v3_12;
 } PyCodeObject;
 
 #endif

src/python/gc.h

@@ -70,8 +70,46 @@ struct _gc_runtime_state3_8 {
     int collecting;
 };
 
+
+struct _gc_runtime_state3_12 {
+    /* List of objects that still need to be cleaned up, singly linked
+     * via their gc headers' gc_prev pointers.  */
+    PyObject *trash_delete_later;
+    /* Current call-stack depth of tp_dealloc calls. */
+    int trash_delete_nesting;
+
+    /* Is automatic collection enabled? */
+    int enabled;
+    int debug;
+    /* linked lists of container objects */
+    struct gc_generation3_8 generations[NUM_GENERATIONS];
+    void *generation0;
+    /* a permanent generation which won't be collected */
+    struct gc_generation3_8 permanent_generation;
+    struct gc_generation_stats generation_stats[NUM_GENERATIONS];
+    /* true if we are currently running the collector */
+    int collecting;
+    /* list of uncollectable objects */
+    PyObject *garbage;
+    /* a list of callbacks to be invoked when collection is performed */
+    PyObject *callbacks;
+    /* This is the number of objects that survived the last full
+       collection. It approximates the number of long lived objects
+       tracked by the GC.
+
+       (by "full collection", we mean a collection of the oldest
+       generation). */
+    Py_ssize_t long_lived_total;
+    /* This is the number of objects that survived all "non-full"
+       collections, and are awaiting to undergo a full collection for
+       the first time. */
+    Py_ssize_t long_lived_pending;
+};
+
+
 typedef union {
-    struct _gc_runtime_state3_8 v3_8;
+    struct _gc_runtime_state3_8  v3_8;
+    struct _gc_runtime_state3_12 v3_12;
 } GCRuntimeState;
 
 #endif

src/python/gil.h

@@ -42,15 +42,8 @@ struct _gil_runtime_state3_11 {
   unsigned long interval;
   _Py_atomic_address last_holder;
   _Py_atomic_int locked;
-  unsigned long switch_number;
-  PyCOND_T cond;
-  PyMUTEX_T mutex;
-#ifdef FORCE_SWITCHING
-  /* This condition variable helps the GIL-releasing thread wait for
-     a GIL-awaiting thread to be scheduled and take the GIL. */
-  PyCOND_T switch_cond;
-  PyMUTEX_T switch_mutex;
-#endif
 };
 
+typedef struct _gil_runtime_state3_11 gil_state_t;
+
 #endif

src/python/iframe.h

@@ -57,8 +57,38 @@ typedef struct _PyInterpreterFrame3_11 {
 } _PyInterpreterFrame3_11;
 
 
+#define FRAME_OWNED_BY_CSTACK 3
+
+typedef struct _PyInterpreterFrame3_12 {
+    PyCodeObject *f_code; /* Strong reference */
+    struct _PyInterpreterFrame3_12 *previous;
+    PyObject *f_funcobj; /* Strong reference. Only valid if not on C stack */
+    PyObject *f_globals; /* Borrowed reference. Only valid if not on C stack */
+    PyObject *f_builtins; /* Borrowed reference. Only valid if not on C stack */
+    PyObject *f_locals; /* Strong reference, may be NULL. Only valid if not on C stack */
+    PyFrameObject *frame_obj; /* Strong reference, may be NULL. Only valid if not on C stack */
+    // NOTE: This is not necessarily the last instruction started in the given
+    // frame. Rather, it is the code unit *prior to* the *next* instruction. For
+    // example, it may be an inline CACHE entry, an instruction we just jumped
+    // over, or (in the case of a newly-created frame) a totally invalid value:
+    _Py_CODEUNIT *prev_instr;
+    int stacktop;  /* Offset of TOS from localsplus  */
+    /* The return_offset determines where a `RETURN` should go in the caller,
+     * relative to `prev_instr`.
+     * It is only meaningful to the callee,
+     * so it needs to be set in any CALL (to a Python function)
+     * or SEND (to a coroutine or generator).
+     * If there is no callee, then it is meaningless. */
+    uint16_t return_offset;
+    char owner;
+    /* Locals and stack */
+    PyObject *localsplus[1];
+} _PyInterpreterFrame3_12;
+
+
 typedef union {
     _PyInterpreterFrame3_11 v3_11;
+    _PyInterpreterFrame3_12 v3_12;
 } PyInterpreterFrame;
 
 #endif