6
*****************************************
7
Initialization, Finalization, and Threads
8
*****************************************
11
.. cfunction:: void Py_Initialize()
14
single: Py_SetProgramName()
15
single: PyEval_InitThreads()
16
single: PyEval_ReleaseLock()
17
single: PyEval_AcquireLock()
18
single: modules (in module sys)
19
single: path (in module sys)
23
triple: module; search; path
24
single: PySys_SetArgv()
27
Initialize the Python interpreter. In an application embedding Python, this
28
should be called before using any other Python/C API functions; with the
29
exception of :cfunc:`Py_SetProgramName`, :cfunc:`PyEval_InitThreads`,
30
:cfunc:`PyEval_ReleaseLock`, and :cfunc:`PyEval_AcquireLock`. This initializes
31
the table of loaded modules (``sys.modules``), and creates the fundamental
32
modules :mod:`builtins`, :mod:`__main__` and :mod:`sys`. It also initializes
33
the module search path (``sys.path``). It does not set ``sys.argv``; use
34
:cfunc:`PySys_SetArgv` for that. This is a no-op when called for a second time
35
(without calling :cfunc:`Py_Finalize` first). There is no return value; it is a
36
fatal error if the initialization fails.
39
.. cfunction:: void Py_InitializeEx(int initsigs)
41
This function works like :cfunc:`Py_Initialize` if *initsigs* is 1. If
42
*initsigs* is 0, it skips initialization registration of signal handlers, which
43
might be useful when Python is embedded.
46
.. cfunction:: int Py_IsInitialized()
48
Return true (nonzero) when the Python interpreter has been initialized, false
49
(zero) if not. After :cfunc:`Py_Finalize` is called, this returns false until
50
:cfunc:`Py_Initialize` is called again.
53
.. cfunction:: void Py_Finalize()
55
Undo all initializations made by :cfunc:`Py_Initialize` and subsequent use of
56
Python/C API functions, and destroy all sub-interpreters (see
57
:cfunc:`Py_NewInterpreter` below) that were created and not yet destroyed since
58
the last call to :cfunc:`Py_Initialize`. Ideally, this frees all memory
59
allocated by the Python interpreter. This is a no-op when called for a second
60
time (without calling :cfunc:`Py_Initialize` again first). There is no return
61
value; errors during finalization are ignored.
63
This function is provided for a number of reasons. An embedding application
64
might want to restart Python without having to restart the application itself.
65
An application that has loaded the Python interpreter from a dynamically
66
loadable library (or DLL) might want to free all memory allocated by Python
67
before unloading the DLL. During a hunt for memory leaks in an application a
68
developer might want to free all memory allocated by Python before exiting from
71
**Bugs and caveats:** The destruction of modules and objects in modules is done
72
in random order; this may cause destructors (:meth:`__del__` methods) to fail
73
when they depend on other objects (even functions) or modules. Dynamically
74
loaded extension modules loaded by Python are not unloaded. Small amounts of
75
memory allocated by the Python interpreter may not be freed (if you find a leak,
76
please report it). Memory tied up in circular references between objects is not
77
freed. Some memory allocated by extension modules may not be freed. Some
78
extensions may not work properly if their initialization routine is called more
79
than once; this can happen if an application calls :cfunc:`Py_Initialize` and
80
:cfunc:`Py_Finalize` more than once.
83
.. cfunction:: PyThreadState* Py_NewInterpreter()
89
single: stdout (in module sys)
90
single: stderr (in module sys)
91
single: stdin (in module sys)
93
Create a new sub-interpreter. This is an (almost) totally separate environment
94
for the execution of Python code. In particular, the new interpreter has
95
separate, independent versions of all imported modules, including the
96
fundamental modules :mod:`builtins`, :mod:`__main__` and :mod:`sys`. The
97
table of loaded modules (``sys.modules``) and the module search path
98
(``sys.path``) are also separate. The new environment has no ``sys.argv``
99
variable. It has new standard I/O stream file objects ``sys.stdin``,
100
``sys.stdout`` and ``sys.stderr`` (however these refer to the same underlying
101
:ctype:`FILE` structures in the C library).
103
The return value points to the first thread state created in the new
104
sub-interpreter. This thread state is made in the current thread state.
105
Note that no actual thread is created; see the discussion of thread states
106
below. If creation of the new interpreter is unsuccessful, *NULL* is
107
returned; no exception is set since the exception state is stored in the
108
current thread state and there may not be a current thread state. (Like all
109
other Python/C API functions, the global interpreter lock must be held before
110
calling this function and is still held when it returns; however, unlike most
111
other Python/C API functions, there needn't be a current thread state on
115
single: Py_Finalize()
116
single: Py_Initialize()
118
Extension modules are shared between (sub-)interpreters as follows: the first
119
time a particular extension is imported, it is initialized normally, and a
120
(shallow) copy of its module's dictionary is squirreled away. When the same
121
extension is imported by another (sub-)interpreter, a new module is initialized
122
and filled with the contents of this copy; the extension's ``init`` function is
123
not called. Note that this is different from what happens when an extension is
124
imported after the interpreter has been completely re-initialized by calling
125
:cfunc:`Py_Finalize` and :cfunc:`Py_Initialize`; in that case, the extension's
126
``initmodule`` function *is* called again.
128
.. index:: single: close() (in module os)
130
**Bugs and caveats:** Because sub-interpreters (and the main interpreter) are
131
part of the same process, the insulation between them isn't perfect --- for
132
example, using low-level file operations like :func:`os.close` they can
133
(accidentally or maliciously) affect each other's open files. Because of the
134
way extensions are shared between (sub-)interpreters, some extensions may not
135
work properly; this is especially likely when the extension makes use of
136
(static) global variables, or when the extension manipulates its module's
137
dictionary after its initialization. It is possible to insert objects created
138
in one sub-interpreter into a namespace of another sub-interpreter; this should
139
be done with great care to avoid sharing user-defined functions, methods,
140
instances or classes between sub-interpreters, since import operations executed
141
by such objects may affect the wrong (sub-)interpreter's dictionary of loaded
142
modules. (XXX This is a hard-to-fix bug that will be addressed in a future
145
Also note that the use of this functionality is incompatible with extension
146
modules such as PyObjC and ctypes that use the :cfunc:`PyGILState_\*` APIs (and
147
this is inherent in the way the :cfunc:`PyGILState_\*` functions work). Simple
148
things may work, but confusing behavior will always be near.
151
.. cfunction:: void Py_EndInterpreter(PyThreadState *tstate)
153
.. index:: single: Py_Finalize()
155
Destroy the (sub-)interpreter represented by the given thread state. The given
156
thread state must be the current thread state. See the discussion of thread
157
states below. When the call returns, the current thread state is *NULL*. All
158
thread states associated with this interpreter are destroyed. (The global
159
interpreter lock must be held before calling this function and is still held
160
when it returns.) :cfunc:`Py_Finalize` will destroy all sub-interpreters that
161
haven't been explicitly destroyed at that point.
164
.. cfunction:: void Py_SetProgramName(wchar_t *name)
167
single: Py_Initialize()
171
This function should be called before :cfunc:`Py_Initialize` is called for
172
the first time, if it is called at all. It tells the interpreter the value
173
of the ``argv[0]`` argument to the :cfunc:`main` function of the program
174
(converted to wide characters).
175
This is used by :cfunc:`Py_GetPath` and some other functions below to find
176
the Python run-time libraries relative to the interpreter executable. The
177
default value is ``'python'``. The argument should point to a
178
zero-terminated wide character string in static storage whose contents will not
179
change for the duration of the program's execution. No code in the Python
180
interpreter will change the contents of this storage.
183
.. cfunction:: wchar* Py_GetProgramName()
185
.. index:: single: Py_SetProgramName()
187
Return the program name set with :cfunc:`Py_SetProgramName`, or the default.
188
The returned string points into static storage; the caller should not modify its
192
.. cfunction:: wchar_t* Py_GetPrefix()
194
Return the *prefix* for installed platform-independent files. This is derived
195
through a number of complicated rules from the program name set with
196
:cfunc:`Py_SetProgramName` and some environment variables; for example, if the
197
program name is ``'/usr/local/bin/python'``, the prefix is ``'/usr/local'``. The
198
returned string points into static storage; the caller should not modify its
199
value. This corresponds to the :makevar:`prefix` variable in the top-level
200
:file:`Makefile` and the :option:`--prefix` argument to the :program:`configure`
201
script at build time. The value is available to Python code as ``sys.prefix``.
202
It is only useful on Unix. See also the next function.
205
.. cfunction:: wchar_t* Py_GetExecPrefix()
207
Return the *exec-prefix* for installed platform-*dependent* files. This is
208
derived through a number of complicated rules from the program name set with
209
:cfunc:`Py_SetProgramName` and some environment variables; for example, if the
210
program name is ``'/usr/local/bin/python'``, the exec-prefix is
211
``'/usr/local'``. The returned string points into static storage; the caller
212
should not modify its value. This corresponds to the :makevar:`exec_prefix`
213
variable in the top-level :file:`Makefile` and the :option:`--exec-prefix`
214
argument to the :program:`configure` script at build time. The value is
215
available to Python code as ``sys.exec_prefix``. It is only useful on Unix.
217
Background: The exec-prefix differs from the prefix when platform dependent
218
files (such as executables and shared libraries) are installed in a different
219
directory tree. In a typical installation, platform dependent files may be
220
installed in the :file:`/usr/local/plat` subtree while platform independent may
221
be installed in :file:`/usr/local`.
223
Generally speaking, a platform is a combination of hardware and software
224
families, e.g. Sparc machines running the Solaris 2.x operating system are
225
considered the same platform, but Intel machines running Solaris 2.x are another
226
platform, and Intel machines running Linux are yet another platform. Different
227
major revisions of the same operating system generally also form different
228
platforms. Non-Unix operating systems are a different story; the installation
229
strategies on those systems are so different that the prefix and exec-prefix are
230
meaningless, and set to the empty string. Note that compiled Python bytecode
231
files are platform independent (but not independent from the Python version by
232
which they were compiled!).
234
System administrators will know how to configure the :program:`mount` or
235
:program:`automount` programs to share :file:`/usr/local` between platforms
236
while having :file:`/usr/local/plat` be a different filesystem for each
240
.. cfunction:: wchar_t* Py_GetProgramFullPath()
243
single: Py_SetProgramName()
244
single: executable (in module sys)
246
Return the full program name of the Python executable; this is computed as a
247
side-effect of deriving the default module search path from the program name
248
(set by :cfunc:`Py_SetProgramName` above). The returned string points into
249
static storage; the caller should not modify its value. The value is available
250
to Python code as ``sys.executable``.
253
.. cfunction:: wchar_t* Py_GetPath()
256
triple: module; search; path
257
single: path (in module sys)
259
Return the default module search path; this is computed from the program name
260
(set by :cfunc:`Py_SetProgramName` above) and some environment variables. The
261
returned string consists of a series of directory names separated by a platform
262
dependent delimiter character. The delimiter character is ``':'`` on Unix and
263
Mac OS X, ``';'`` on Windows. The returned string points into static storage;
264
the caller should not modify its value. The value is available to Python code
265
as the list ``sys.path``, which may be modified to change the future search path
268
.. XXX should give the exact rules
271
.. cfunction:: const char* Py_GetVersion()
273
Return the version of this Python interpreter. This is a string that looks
276
"3.0a5+ (py3k:63103M, May 12 2008, 00:53:55) \n[GCC 4.2.3]"
278
.. index:: single: version (in module sys)
280
The first word (up to the first space character) is the current Python version;
281
the first three characters are the major and minor version separated by a
282
period. The returned string points into static storage; the caller should not
283
modify its value. The value is available to Python code as :data:`sys.version`.
286
.. cfunction:: const char* Py_GetBuildNumber()
288
Return a string representing the Subversion revision that this Python executable
289
was built from. This number is a string because it may contain a trailing 'M'
290
if Python was built from a mixed revision source tree.
293
.. cfunction:: const char* Py_GetPlatform()
295
.. index:: single: platform (in module sys)
297
Return the platform identifier for the current platform. On Unix, this is
298
formed from the "official" name of the operating system, converted to lower
299
case, followed by the major revision number; e.g., for Solaris 2.x, which is
300
also known as SunOS 5.x, the value is ``'sunos5'``. On Mac OS X, it is
301
``'darwin'``. On Windows, it is ``'win'``. The returned string points into
302
static storage; the caller should not modify its value. The value is available
303
to Python code as ``sys.platform``.
306
.. cfunction:: const char* Py_GetCopyright()
308
Return the official copyright string for the current Python version, for example
310
``'Copyright 1991-1995 Stichting Mathematisch Centrum, Amsterdam'``
312
.. index:: single: copyright (in module sys)
314
The returned string points into static storage; the caller should not modify its
315
value. The value is available to Python code as ``sys.copyright``.
318
.. cfunction:: const char* Py_GetCompiler()
320
Return an indication of the compiler used to build the current Python version,
321
in square brackets, for example::
325
.. index:: single: version (in module sys)
327
The returned string points into static storage; the caller should not modify its
328
value. The value is available to Python code as part of the variable
332
.. cfunction:: const char* Py_GetBuildInfo()
334
Return information about the sequence number and build date and time of the
335
current Python interpreter instance, for example ::
337
"#67, Aug 1 1997, 22:34:28"
339
.. index:: single: version (in module sys)
341
The returned string points into static storage; the caller should not modify its
342
value. The value is available to Python code as part of the variable
346
.. cfunction:: void PySys_SetArgv(int argc, wchar_t **argv)
350
single: Py_FatalError()
351
single: argv (in module sys)
353
Set :data:`sys.argv` based on *argc* and *argv*. These parameters are
354
similar to those passed to the program's :cfunc:`main` function with the
355
difference that the first entry should refer to the script file to be
356
executed rather than the executable hosting the Python interpreter. If there
357
isn't a script that will be run, the first entry in *argv* can be an empty
358
string. If this function fails to initialize :data:`sys.argv`, a fatal
359
condition is signalled using :cfunc:`Py_FatalError`.
361
This function also prepends the executed script's path to :data:`sys.path`.
362
If no script is executed (in the case of calling ``python -c`` or just the
363
interactive interpreter), the empty string is used instead.
365
.. XXX impl. doesn't seem consistent in allowing 0/NULL for the params;
369
.. cfunction:: void Py_SetPythonHome(char *home)
371
Set the default "home" directory, that is, the location of the standard
372
Python libraries. The libraries are searched in
373
:file:`{home}/lib/python{version}` and :file:`{home}/lib/python{version}`.
376
.. cfunction:: char* Py_GetPythonHome()
378
Return the default "home", that is, the value set by a previous call to
379
:cfunc:`Py_SetPythonHome`, or the value of the :envvar:`PYTHONHOME`
380
environment variable if it is set.
385
Thread State and the Global Interpreter Lock
386
============================================
389
single: global interpreter lock
390
single: interpreter lock
391
single: lock, interpreter
393
The Python interpreter is not fully thread safe. In order to support
394
multi-threaded Python programs, there's a global lock that must be held by the
395
current thread before it can safely access Python objects. Without the lock,
396
even the simplest operations could cause problems in a multi-threaded program:
397
for example, when two threads simultaneously increment the reference count of
398
the same object, the reference count could end up being incremented only once
401
.. index:: single: setcheckinterval() (in module sys)
403
Therefore, the rule exists that only the thread that has acquired the global
404
interpreter lock may operate on Python objects or call Python/C API functions.
405
In order to support multi-threaded Python programs, the interpreter regularly
406
releases and reacquires the lock --- by default, every 100 bytecode instructions
407
(this can be changed with :func:`sys.setcheckinterval`). The lock is also
408
released and reacquired around potentially blocking I/O operations like reading
409
or writing a file, so that other threads can run while the thread that requests
410
the I/O is waiting for the I/O operation to complete.
413
single: PyThreadState
414
single: PyThreadState
416
The Python interpreter needs to keep some bookkeeping information separate per
417
thread --- for this it uses a data structure called :ctype:`PyThreadState`.
418
There's one global variable, however: the pointer to the current
419
:ctype:`PyThreadState` structure. While most thread packages have a way to
420
store "per-thread global data," Python's internal platform independent thread
421
abstraction doesn't support this yet. Therefore, the current thread state must
422
be manipulated explicitly.
424
This is easy enough in most cases. Most code manipulating the global
425
interpreter lock has the following simple structure::
427
Save the thread state in a local variable.
428
Release the interpreter lock.
429
...Do some blocking I/O operation...
430
Reacquire the interpreter lock.
431
Restore the thread state from the local variable.
433
This is so common that a pair of macros exists to simplify it::
435
Py_BEGIN_ALLOW_THREADS
436
...Do some blocking I/O operation...
440
single: Py_BEGIN_ALLOW_THREADS
441
single: Py_END_ALLOW_THREADS
443
The :cmacro:`Py_BEGIN_ALLOW_THREADS` macro opens a new block and declares a
444
hidden local variable; the :cmacro:`Py_END_ALLOW_THREADS` macro closes the
445
block. Another advantage of using these two macros is that when Python is
446
compiled without thread support, they are defined empty, thus saving the thread
447
state and lock manipulations.
449
When thread support is enabled, the block above expands to the following code::
451
PyThreadState *_save;
453
_save = PyEval_SaveThread();
454
...Do some blocking I/O operation...
455
PyEval_RestoreThread(_save);
457
Using even lower level primitives, we can get roughly the same effect as
460
PyThreadState *_save;
462
_save = PyThreadState_Swap(NULL);
463
PyEval_ReleaseLock();
464
...Do some blocking I/O operation...
465
PyEval_AcquireLock();
466
PyThreadState_Swap(_save);
469
single: PyEval_RestoreThread()
471
single: PyEval_SaveThread()
472
single: PyEval_ReleaseLock()
473
single: PyEval_AcquireLock()
475
There are some subtle differences; in particular, :cfunc:`PyEval_RestoreThread`
476
saves and restores the value of the global variable :cdata:`errno`, since the
477
lock manipulation does not guarantee that :cdata:`errno` is left alone. Also,
478
when thread support is disabled, :cfunc:`PyEval_SaveThread` and
479
:cfunc:`PyEval_RestoreThread` don't manipulate the lock; in this case,
480
:cfunc:`PyEval_ReleaseLock` and :cfunc:`PyEval_AcquireLock` are not available.
481
This is done so that dynamically loaded extensions compiled with thread support
482
enabled can be loaded by an interpreter that was compiled with disabled thread
485
The global interpreter lock is used to protect the pointer to the current thread
486
state. When releasing the lock and saving the thread state, the current thread
487
state pointer must be retrieved before the lock is released (since another
488
thread could immediately acquire the lock and store its own thread state in the
489
global variable). Conversely, when acquiring the lock and restoring the thread
490
state, the lock must be acquired before storing the thread state pointer.
492
Why am I going on with so much detail about this? Because when threads are
493
created from C, they don't have the global interpreter lock, nor is there a
494
thread state data structure for them. Such threads must bootstrap themselves
495
into existence, by first creating a thread state data structure, then acquiring
496
the lock, and finally storing their thread state pointer, before they can start
497
using the Python/C API. When they are done, they should reset the thread state
498
pointer, release the lock, and finally free their thread state data structure.
500
Threads can take advantage of the :cfunc:`PyGILState_\*` functions to do all of
501
the above automatically. The typical idiom for calling into Python from a C
504
PyGILState_STATE gstate;
505
gstate = PyGILState_Ensure();
507
/* Perform Python actions here. */
508
result = CallSomeFunction();
509
/* evaluate result */
511
/* Release the thread. No Python API allowed beyond this point. */
512
PyGILState_Release(gstate);
514
Note that the :cfunc:`PyGILState_\*` functions assume there is only one global
515
interpreter (created automatically by :cfunc:`Py_Initialize`). Python still
516
supports the creation of additional interpreters (using
517
:cfunc:`Py_NewInterpreter`), but mixing multiple interpreters and the
518
:cfunc:`PyGILState_\*` API is unsupported.
521
.. ctype:: PyInterpreterState
523
This data structure represents the state shared by a number of cooperating
524
threads. Threads belonging to the same interpreter share their module
525
administration and a few other internal items. There are no public members in
528
Threads belonging to different interpreters initially share nothing, except
529
process state like available memory, open file descriptors and such. The global
530
interpreter lock is also shared by all threads, regardless of to which
531
interpreter they belong.
534
.. ctype:: PyThreadState
536
This data structure represents the state of a single thread. The only public
537
data member is :ctype:`PyInterpreterState \*`:attr:`interp`, which points to
538
this thread's interpreter state.
541
.. cfunction:: void PyEval_InitThreads()
544
single: PyEval_ReleaseLock()
545
single: PyEval_ReleaseThread()
546
single: PyEval_SaveThread()
547
single: PyEval_RestoreThread()
549
Initialize and acquire the global interpreter lock. It should be called in the
550
main thread before creating a second thread or engaging in any other thread
551
operations such as :cfunc:`PyEval_ReleaseLock` or
552
``PyEval_ReleaseThread(tstate)``. It is not needed before calling
553
:cfunc:`PyEval_SaveThread` or :cfunc:`PyEval_RestoreThread`.
555
.. index:: single: Py_Initialize()
557
This is a no-op when called for a second time. It is safe to call this function
558
before calling :cfunc:`Py_Initialize`.
560
.. index:: module: _thread
562
When only the main thread exists, no lock operations are needed. This is a
563
common situation (most Python programs do not use threads), and the lock
564
operations slow the interpreter down a bit. Therefore, the lock is not created
565
initially. This situation is equivalent to having acquired the lock: when
566
there is only a single thread, all object accesses are safe. Therefore, when
567
this function initializes the lock, it also acquires it. Before the Python
568
:mod:`_thread` module creates a new thread, knowing that either it has the lock
569
or the lock hasn't been created yet, it calls :cfunc:`PyEval_InitThreads`. When
570
this call returns, it is guaranteed that the lock has been created and that the
571
calling thread has acquired it.
573
It is **not** safe to call this function when it is unknown which thread (if
574
any) currently has the global interpreter lock.
576
This function is not available when thread support is disabled at compile time.
579
.. cfunction:: int PyEval_ThreadsInitialized()
581
Returns a non-zero value if :cfunc:`PyEval_InitThreads` has been called. This
582
function can be called without holding the lock, and therefore can be used to
583
avoid calls to the locking API when running single-threaded. This function is
584
not available when thread support is disabled at compile time.
587
.. cfunction:: void PyEval_AcquireLock()
589
Acquire the global interpreter lock. The lock must have been created earlier.
590
If this thread already has the lock, a deadlock ensues. This function is not
591
available when thread support is disabled at compile time.
594
.. cfunction:: void PyEval_ReleaseLock()
596
Release the global interpreter lock. The lock must have been created earlier.
597
This function is not available when thread support is disabled at compile time.
600
.. cfunction:: void PyEval_AcquireThread(PyThreadState *tstate)
602
Acquire the global interpreter lock and set the current thread state to
603
*tstate*, which should not be *NULL*. The lock must have been created earlier.
604
If this thread already has the lock, deadlock ensues. This function is not
605
available when thread support is disabled at compile time.
608
.. cfunction:: void PyEval_ReleaseThread(PyThreadState *tstate)
610
Reset the current thread state to *NULL* and release the global interpreter
611
lock. The lock must have been created earlier and must be held by the current
612
thread. The *tstate* argument, which must not be *NULL*, is only used to check
613
that it represents the current thread state --- if it isn't, a fatal error is
614
reported. This function is not available when thread support is disabled at
618
.. cfunction:: PyThreadState* PyEval_SaveThread()
620
Release the interpreter lock (if it has been created and thread support is
621
enabled) and reset the thread state to *NULL*, returning the previous thread
622
state (which is not *NULL*). If the lock has been created, the current thread
623
must have acquired it. (This function is available even when thread support is
624
disabled at compile time.)
627
.. cfunction:: void PyEval_RestoreThread(PyThreadState *tstate)
629
Acquire the interpreter lock (if it has been created and thread support is
630
enabled) and set the thread state to *tstate*, which must not be *NULL*. If the
631
lock has been created, the current thread must not have acquired it, otherwise
632
deadlock ensues. (This function is available even when thread support is
633
disabled at compile time.)
636
.. cfunction:: void PyEval_ReInitThreads()
638
This function is called from :cfunc:`PyOS_AfterFork` to ensure that newly
639
created child processes don't hold locks referring to threads which
640
are not running in the child process.
643
The following macros are normally used without a trailing semicolon; look for
644
example usage in the Python source distribution.
647
.. cmacro:: Py_BEGIN_ALLOW_THREADS
649
This macro expands to ``{ PyThreadState *_save; _save = PyEval_SaveThread();``.
650
Note that it contains an opening brace; it must be matched with a following
651
:cmacro:`Py_END_ALLOW_THREADS` macro. See above for further discussion of this
652
macro. It is a no-op when thread support is disabled at compile time.
655
.. cmacro:: Py_END_ALLOW_THREADS
657
This macro expands to ``PyEval_RestoreThread(_save); }``. Note that it contains
658
a closing brace; it must be matched with an earlier
659
:cmacro:`Py_BEGIN_ALLOW_THREADS` macro. See above for further discussion of
660
this macro. It is a no-op when thread support is disabled at compile time.
663
.. cmacro:: Py_BLOCK_THREADS
665
This macro expands to ``PyEval_RestoreThread(_save);``: it is equivalent to
666
:cmacro:`Py_END_ALLOW_THREADS` without the closing brace. It is a no-op when
667
thread support is disabled at compile time.
670
.. cmacro:: Py_UNBLOCK_THREADS
672
This macro expands to ``_save = PyEval_SaveThread();``: it is equivalent to
673
:cmacro:`Py_BEGIN_ALLOW_THREADS` without the opening brace and variable
674
declaration. It is a no-op when thread support is disabled at compile time.
676
All of the following functions are only available when thread support is enabled
677
at compile time, and must be called only when the interpreter lock has been
681
.. cfunction:: PyInterpreterState* PyInterpreterState_New()
683
Create a new interpreter state object. The interpreter lock need not be held,
684
but may be held if it is necessary to serialize calls to this function.
687
.. cfunction:: void PyInterpreterState_Clear(PyInterpreterState *interp)
689
Reset all information in an interpreter state object. The interpreter lock must
693
.. cfunction:: void PyInterpreterState_Delete(PyInterpreterState *interp)
695
Destroy an interpreter state object. The interpreter lock need not be held.
696
The interpreter state must have been reset with a previous call to
697
:cfunc:`PyInterpreterState_Clear`.
700
.. cfunction:: PyThreadState* PyThreadState_New(PyInterpreterState *interp)
702
Create a new thread state object belonging to the given interpreter object. The
703
interpreter lock need not be held, but may be held if it is necessary to
704
serialize calls to this function.
707
.. cfunction:: void PyThreadState_Clear(PyThreadState *tstate)
709
Reset all information in a thread state object. The interpreter lock must be
713
.. cfunction:: void PyThreadState_Delete(PyThreadState *tstate)
715
Destroy a thread state object. The interpreter lock need not be held. The
716
thread state must have been reset with a previous call to
717
:cfunc:`PyThreadState_Clear`.
720
.. cfunction:: PyThreadState* PyThreadState_Get()
722
Return the current thread state. The interpreter lock must be held. When the
723
current thread state is *NULL*, this issues a fatal error (so that the caller
724
needn't check for *NULL*).
727
.. cfunction:: PyThreadState* PyThreadState_Swap(PyThreadState *tstate)
729
Swap the current thread state with the thread state given by the argument
730
*tstate*, which may be *NULL*. The interpreter lock must be held.
733
.. cfunction:: PyObject* PyThreadState_GetDict()
735
Return a dictionary in which extensions can store thread-specific state
736
information. Each extension should use a unique key to use to store state in
737
the dictionary. It is okay to call this function when no current thread state
738
is available. If this function returns *NULL*, no exception has been raised and
739
the caller should assume no current thread state is available.
742
.. cfunction:: int PyThreadState_SetAsyncExc(long id, PyObject *exc)
744
Asynchronously raise an exception in a thread. The *id* argument is the thread
745
id of the target thread; *exc* is the exception object to be raised. This
746
function does not steal any references to *exc*. To prevent naive misuse, you
747
must write your own C extension to call this. Must be called with the GIL held.
748
Returns the number of thread states modified; this is normally one, but will be
749
zero if the thread id isn't found. If *exc* is :const:`NULL`, the pending
750
exception (if any) for the thread is cleared. This raises no exceptions.
753
.. cfunction:: PyGILState_STATE PyGILState_Ensure()
755
Ensure that the current thread is ready to call the Python C API regardless of
756
the current state of Python, or of its thread lock. This may be called as many
757
times as desired by a thread as long as each call is matched with a call to
758
:cfunc:`PyGILState_Release`. In general, other thread-related APIs may be used
759
between :cfunc:`PyGILState_Ensure` and :cfunc:`PyGILState_Release` calls as long
760
as the thread state is restored to its previous state before the Release(). For
761
example, normal usage of the :cmacro:`Py_BEGIN_ALLOW_THREADS` and
762
:cmacro:`Py_END_ALLOW_THREADS` macros is acceptable.
764
The return value is an opaque "handle" to the thread state when
765
:cfunc:`PyGILState_Ensure` was called, and must be passed to
766
:cfunc:`PyGILState_Release` to ensure Python is left in the same state. Even
767
though recursive calls are allowed, these handles *cannot* be shared - each
768
unique call to :cfunc:`PyGILState_Ensure` must save the handle for its call
769
to :cfunc:`PyGILState_Release`.
771
When the function returns, the current thread will hold the GIL. Failure is a
775
.. cfunction:: void PyGILState_Release(PyGILState_STATE)
777
Release any resources previously acquired. After this call, Python's state will
778
be the same as it was prior to the corresponding :cfunc:`PyGILState_Ensure` call
779
(but generally this state will be unknown to the caller, hence the use of the
782
Every call to :cfunc:`PyGILState_Ensure` must be matched by a call to
783
:cfunc:`PyGILState_Release` on the same thread.
787
Asynchronous Notifications
788
==========================
790
A mechanism is provided to make asynchronous notifications to the the main
791
interpreter thread. These notifications take the form of a function
792
pointer and a void argument.
794
.. index:: single: setcheckinterval() (in module sys)
796
Every check interval, when the interpreter lock is released and reacquired,
797
python will also call any such provided functions. This can be used for
798
example by asynchronous IO handlers. The notification can be scheduled
799
from a worker thread and the actual call than made at the earliest
800
convenience by the main thread where it has possession of the global
801
interpreter lock and can perform any Python API calls.
803
.. cfunction:: void Py_AddPendingCall( int (*func)(void *, void *arg) )
805
.. index:: single: Py_AddPendingCall()
807
Post a notification to the Python main thread. If successful,
808
*func* will be called with the argument *arg* at the earliest
809
convenience. *func* will be called having the global interpreter
810
lock held and can thus use the full Python API and can take any
811
action such as setting object attributes to signal IO completion.
812
It must return 0 on success, or -1 signalling an exception.
813
The notification function won't be interrupted to perform another
814
asynchronous notification recursively,
815
but it can still be interrupted to switch threads if the interpreter
816
lock is released, for example, if it calls back into python code.
818
This function returns 0 on success in which case the notification has been
819
scheduled. Otherwise, for example if the notification buffer is full,
820
it returns -1 without setting any exception.
822
This function can be called on any thread, be it a Python thread or
823
some other system thread. If it is a Python thread, it doesn't matter if
824
it holds the global interpreter lock or not.
826
.. versionadded:: 2.7
832
Profiling and Tracing
833
=====================
835
.. sectionauthor:: Fred L. Drake, Jr. <fdrake@acm.org>
838
The Python interpreter provides some low-level support for attaching profiling
839
and execution tracing facilities. These are used for profiling, debugging, and
840
coverage analysis tools.
842
This C interface allows the profiling or tracing code to avoid the overhead of
843
calling through Python-level callable objects, making a direct C function call
844
instead. The essential attributes of the facility have not changed; the
845
interface allows trace functions to be installed per-thread, and the basic
846
events reported to the trace function are the same as had been reported to the
847
Python-level trace functions in previous versions.
850
.. ctype:: int (*Py_tracefunc)(PyObject *obj, PyFrameObject *frame, int what, PyObject *arg)
852
The type of the trace function registered using :cfunc:`PyEval_SetProfile` and
853
:cfunc:`PyEval_SetTrace`. The first parameter is the object passed to the
854
registration function as *obj*, *frame* is the frame object to which the event
855
pertains, *what* is one of the constants :const:`PyTrace_CALL`,
856
:const:`PyTrace_EXCEPTION`, :const:`PyTrace_LINE`, :const:`PyTrace_RETURN`,
857
:const:`PyTrace_C_CALL`, :const:`PyTrace_C_EXCEPTION`, or
858
:const:`PyTrace_C_RETURN`, and *arg* depends on the value of *what*:
860
+------------------------------+--------------------------------------+
861
| Value of *what* | Meaning of *arg* |
862
+==============================+======================================+
863
| :const:`PyTrace_CALL` | Always *NULL*. |
864
+------------------------------+--------------------------------------+
865
| :const:`PyTrace_EXCEPTION` | Exception information as returned by |
866
| | :func:`sys.exc_info`. |
867
+------------------------------+--------------------------------------+
868
| :const:`PyTrace_LINE` | Always *NULL*. |
869
+------------------------------+--------------------------------------+
870
| :const:`PyTrace_RETURN` | Value being returned to the caller. |
871
+------------------------------+--------------------------------------+
872
| :const:`PyTrace_C_CALL` | Name of function being called. |
873
+------------------------------+--------------------------------------+
874
| :const:`PyTrace_C_EXCEPTION` | Always *NULL*. |
875
+------------------------------+--------------------------------------+
876
| :const:`PyTrace_C_RETURN` | Always *NULL*. |
877
+------------------------------+--------------------------------------+
880
.. cvar:: int PyTrace_CALL
882
The value of the *what* parameter to a :ctype:`Py_tracefunc` function when a new
883
call to a function or method is being reported, or a new entry into a generator.
884
Note that the creation of the iterator for a generator function is not reported
885
as there is no control transfer to the Python bytecode in the corresponding
889
.. cvar:: int PyTrace_EXCEPTION
891
The value of the *what* parameter to a :ctype:`Py_tracefunc` function when an
892
exception has been raised. The callback function is called with this value for
893
*what* when after any bytecode is processed after which the exception becomes
894
set within the frame being executed. The effect of this is that as exception
895
propagation causes the Python stack to unwind, the callback is called upon
896
return to each frame as the exception propagates. Only trace functions receives
897
these events; they are not needed by the profiler.
900
.. cvar:: int PyTrace_LINE
902
The value passed as the *what* parameter to a trace function (but not a
903
profiling function) when a line-number event is being reported.
906
.. cvar:: int PyTrace_RETURN
908
The value for the *what* parameter to :ctype:`Py_tracefunc` functions when a
909
call is returning without propagating an exception.
912
.. cvar:: int PyTrace_C_CALL
914
The value for the *what* parameter to :ctype:`Py_tracefunc` functions when a C
915
function is about to be called.
918
.. cvar:: int PyTrace_C_EXCEPTION
920
The value for the *what* parameter to :ctype:`Py_tracefunc` functions when a C
921
function has thrown an exception.
924
.. cvar:: int PyTrace_C_RETURN
926
The value for the *what* parameter to :ctype:`Py_tracefunc` functions when a C
927
function has returned.
930
.. cfunction:: void PyEval_SetProfile(Py_tracefunc func, PyObject *obj)
932
Set the profiler function to *func*. The *obj* parameter is passed to the
933
function as its first parameter, and may be any Python object, or *NULL*. If
934
the profile function needs to maintain state, using a different value for *obj*
935
for each thread provides a convenient and thread-safe place to store it. The
936
profile function is called for all monitored events except the line-number
940
.. cfunction:: void PyEval_SetTrace(Py_tracefunc func, PyObject *obj)
942
Set the tracing function to *func*. This is similar to
943
:cfunc:`PyEval_SetProfile`, except the tracing function does receive line-number
946
.. cfunction:: PyObject* PyEval_GetCallStats(PyObject *self)
948
Return a tuple of function call counts. There are constants defined for the
949
positions within the tuple:
951
+-------------------------------+-------+
953
+===============================+=======+
954
| :const:`PCALL_ALL` | 0 |
955
+-------------------------------+-------+
956
| :const:`PCALL_FUNCTION` | 1 |
957
+-------------------------------+-------+
958
| :const:`PCALL_FAST_FUNCTION` | 2 |
959
+-------------------------------+-------+
960
| :const:`PCALL_FASTER_FUNCTION`| 3 |
961
+-------------------------------+-------+
962
| :const:`PCALL_METHOD` | 4 |
963
+-------------------------------+-------+
964
| :const:`PCALL_BOUND_METHOD` | 5 |
965
+-------------------------------+-------+
966
| :const:`PCALL_CFUNCTION` | 6 |
967
+-------------------------------+-------+
968
| :const:`PCALL_TYPE` | 7 |
969
+-------------------------------+-------+
970
| :const:`PCALL_GENERATOR` | 8 |
971
+-------------------------------+-------+
972
| :const:`PCALL_OTHER` | 9 |
973
+-------------------------------+-------+
974
| :const:`PCALL_POP` | 10 |
975
+-------------------------------+-------+
977
:const:`PCALL_FAST_FUNCTION` means no argument tuple needs to be created.
978
:const:`PCALL_FASTER_FUNCTION` means that the fast-path frame setup code is used.
980
If there is a method call where the call can be optimized by changing
981
the argument tuple and calling the function directly, it gets recorded
984
This function is only present if Python is compiled with :const:`CALL_PROFILE`
987
.. _advanced-debugging:
989
Advanced Debugger Support
990
=========================
992
.. sectionauthor:: Fred L. Drake, Jr. <fdrake@acm.org>
995
These functions are only intended to be used by advanced debugging tools.
998
.. cfunction:: PyInterpreterState* PyInterpreterState_Head()
1000
Return the interpreter state object at the head of the list of all such objects.
1003
.. cfunction:: PyInterpreterState* PyInterpreterState_Next(PyInterpreterState *interp)
1005
Return the next interpreter state object after *interp* from the list of all
1009
.. cfunction:: PyThreadState * PyInterpreterState_ThreadHead(PyInterpreterState *interp)
1011
Return the a pointer to the first :ctype:`PyThreadState` object in the list of
1012
threads associated with the interpreter *interp*.
1015
.. cfunction:: PyThreadState* PyThreadState_Next(PyThreadState *tstate)
1017
Return the next thread state object after *tstate* from the list of all such
1018
objects belonging to the same :ctype:`PyInterpreterState` object.