Fix a race in `PyMember_GetOne` and `PyMember_SetOne` for `Py_T_OBJECT_EX`.
These functions implement `__slots__` accesses for Python objects.
(cherry picked from commit 362cd2680b)
Co-authored-by: Daniele Parmeggiani <8658291+dpdani@users.noreply.github.com>
gh-120433: Mention ``chocolatey`` for installing llvm on Windows as an alternative option (GH-120434)
(cherry picked from commit 95737bbf18)
Co-authored-by: Kirill Podoprigora <kirill.bast9@mail.ru>
gh-117657: Add TSAN suppression for set_default_allocator_unlocked (GH-120500)
Add TSAN suppression for set_default_allocator_unlocked
(cherry picked from commit 2bacc2343c)
Co-authored-by: AN Long <aisk@users.noreply.github.com>
gh-117657: Make PyType_HasFeature atomic (GH-120210)
Make PyType_HasFeature atomic
(cherry picked from commit eebae2c460)
Co-authored-by: Ken Jin <kenjin@python.org>
The _strptime module object was cached in a static local variable (in the datetime.strptime() implementation). That's a problem when it crosses isolation boundaries, such as reinitializing the runtme or between interpreters. This change fixes the problem by dropping the static variable, instead always relying on the normal sys.modules cache (via PyImport_Import()).
(cherry picked from commit 127c1d2771, AKA gh-120224)
Co-authored-by: neonene <53406459+neonene@users.noreply.github.com>
gh-117657: Make Py_TYPE and Py_SET_TYPE thread safe (GH-120165)
(cherry picked from commit e16aed63f6)
Co-authored-by: Ken Jin <kenjin@python.org>
Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>
Co-authored-by: Nadeshiko Manju <me@manjusaka.me>
This adds a `_PyRecursiveMutex` type based on `PyMutex` and uses that
for the import lock. This fixes some data races in the free-threaded
build and generally simplifies the import lock code.
(cherry picked from commit e21057b999)
Co-authored-by: Sam Gross <colesbury@gmail.com>
The `_PyThreadState_Bind()` function is called before the first
`PyEval_AcquireThread()` so it's not synchronized with the stop the
world GC. We had a race where `gc_visit_heaps()` might visit a thread's
heap while it's being initialized.
Use a simple atomic int to avoid visiting heaps for threads that are not
yet fully initialized (i.e., before `tstate_mimalloc_bind()` is called).
The race was reproducible by running:
`python Lib/test/test_importlib/partial/pool_in_threads.py`.
(cherry picked from commit e69d068ad0)
Co-authored-by: Sam Gross <colesbury@gmail.com>
We make use of the same mechanism that we use for the static builtin types. This required a few tweaks.
This change is the final piece needed to make _datetime support multiple interpreters. I've updated the module slot accordingly.
(cherry picked from commit 105f22ea46, AKA gh-119929)
Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>
The free-threaded build currently immortalizes objects that use deferred
reference counting (see gh-117783). This typically happens once the
first non-main thread is created, but the behavior can be suppressed for
tests, in subinterpreters, or during a compile() call.
This fixes a race condition involving the tracking of whether the
behavior is suppressed.
(cherry picked from commit 47fb4327b5)
The `sem_clockwait` function is not currently instrumented, which leads
to false positives.
(cherry picked from commit 41c1cefbae)
Co-authored-by: Sam Gross <colesbury@gmail.com>
gh-117657: Fix data races report by TSAN unicode-hash (gh-119907)
(cherry picked from commit 0594a27e5f)
Co-authored-by: Donghee Na <donghee.na@python.org>
The GIL may be disabled concurrently with this call so we need to use a
relaxed atomic load.
(cherry picked from commit f3b89a63cb)
Co-authored-by: Sam Gross <colesbury@gmail.com>
Due to a limitation in TSAN, all reads from `PyThreadState.state` must be
atomic to avoid reported races.
(cherry picked from commit 90ec19fd33)
Co-authored-by: Sam Gross <colesbury@gmail.com>
Only call `gc_restore_tid()` from stop-the-world contexts.
`worklist_pop()` can be called while other threads are running, so use a
relaxed atomic to modify `ob_tid`.
(cherry picked from commit 60593b2052)
Co-authored-by: Sam Gross <colesbury@gmail.com>
gh-119661: Add _Py_SINGLETON() include in Argumenet Clinic (#119712)
When the _Py_SINGLETON() is used, Argument Clinic now adds an
explicit "pycore_runtime.h" include to get the macro. Previously, the
macro may or may not be included indirectly by another include.
(cherry picked from commit 7ca74a760a)
This is a backport of 3 commits that go together.
(cherry picked from commit a895756)
(cherry picked from commit b30d30c)
(cherry picked from commit a89fc26)
ensurepip forks a subprocess to run pip itself, but that subprocess only inherits a -I isolated mode flag (see _run_pip() in Lib/ensurepip/__init__.py), not the "-E -s" flags that the installer has been using. This means that parts of ensurepip don't actually run in an isolated environment and can make incorrect decisions based on packages installed in the user site-packages.
(cherry picked from commit c9073eb1a9)
Co-authored-by: Michael Vincent <377567+Vynce@users.noreply.github.com>
_PyArg_Parser holds static global data generated for modules by Argument Clinic. The _PyArg_Parser.kwtuple field is a tuple object, even though it's stored within a static global. In some cases the tuple is statically allocated and thus it's okay that it gets shared by multiple interpreters. However, in other cases the tuple is set lazily, allocated from the heap using the active interprepreter at the point the tuple is needed.
This is a problem once that interpreter is destroyed since _PyArg_Parser.kwtuple becomes at dangling pointer, leading to crashes. It isn't a problem if the tuple is allocated under the main interpreter, since its lifetime is bound to the lifetime of the runtime. The solution here is to temporarily switch to the main interpreter. The alternative would be to always statically allocate the tuple.
This change also fixes a bug where only the most recent parser was added to the global linked list.
(cherry picked from commit 81865002ae)
Co-authored-by: Eric Snow <ericsnowcurrently@gmail.com>
This ensures we don't lose races that occur in subprocesses or
interleave races from workers running in parallel.
Log files are collected and packaged into a zipfile that can be
downloaded from the "Artifacts" section of the workflow run.
(cherry picked from commit b88889e9ff)
Co-authored-by: mpage <mpage@meta.com>
`_Py_qsbr_unregister` is called when the PyThreadState is already
detached, so the access to `tstate->qsbr` isn't safe without locking the
shared mutex. Grab the `struct _qsbr_shared` from the interpreter
instead.
(cherry picked from commit 33d20199af)
Co-authored-by: Alex Turner <alexturner@meta.com>
Using `race:` filters out warnings if the function appears anywhere in
the stack trace. This can hide a lot of unrelated warnings, especially
for a function like `_PyEval_EvalFrameDefault`, which is somewhere on
the stack more often than not.
Change all free-threaded suppressions to `race_top:`, which only matches
the top frame, and add any new suppressions this exposes.
(cherry picked from commit 98ff3f65c0)
Co-authored-by: Brett Simmers <swtaarrs@users.noreply.github.com>
Use relaxed atomics when reading / writing to the field. There are still a
few places in the GC where we do not use atomics. Those should be safe as
the world is stopped.
(cherry picked from commit cb6f75a32c)
Co-authored-by: mpage <mpage@meta.com>
The module itself is a thin wrapper around calls to functions in
`Python/codecs.c`, so that's where the meaningful changes happened:
- Move codecs-related state that lives on `PyInterpreterState` to a
struct declared in `pycore_codecs.h`.
- In free-threaded builds, add a mutex to `codecs_state` to synchronize
operations on `search_path`. Because `search_path_mutex` is used as a
normal mutex and not a critical section, we must be extremely careful
with operations called while holding it.
- The codec registry is explicitly initialized as part of
`_PyUnicode_InitEncodings` to simplify thread-safety.
The code for Tier 2 is now only compiled when configured
with `--enable-experimental-jit[=yes|interpreter]`.
We drop support for `PYTHON_UOPS` and -`Xuops`,
but you can disable the interpreter or JIT
at runtime by setting `PYTHON_JIT=0`.
You can also build it without enabling it by default
using `--enable-experimental-jit=yes-off`;
enable with `PYTHON_JIT=1`.
On Windows, the `build.bat` script supports
`--experimental-jit`, `--experimental-jit-off`,
`--experimental-interpreter`.
In the C code, `_Py_JIT` is defined as before
when the JIT is enabled; the new variable
`_Py_TIER2` is defined when the JIT *or* the
interpreter is enabled. It is actually a bitmask:
1: JIT; 2: default-off; 4: interpreter.
Makes sys.settrace, sys.setprofile, and monitoring generally thread-safe.
Mostly uses a stop-the-world approach and synchronization around the code object's _co_instrumentation_version. There may be a little bit of extra synchronization around the monitoring data that's required to be TSAN clean.
Quiet erroneous TSAN reports of data races in `_PySeqLock`
TSAN reports a couple of data races between the compare/exchange in
`_PySeqLock_LockWrite` and the non-atomic loads in `_PySeqLock_{Abandon,Unlock}Write`.
This is another instance of TSAN incorrectly modeling failed compare/exchange
as a write instead of a load.
Fix data races in the method cache in free-threaded builds
These are technically data races, but I think they're benign (to
the extent that that is actually possible). We update cache entries
non-atomically but read them atomically from another thread, and there's
nothing that establishes a happens-before relationship between the
reads and writes that I can see.
Additionally, reduce the iterations for a few weakref tests that would
otherwise take a prohibitively long amount of time (> 1 hour) when TSAN
is enabled and the GIL is disabled.
* Move ifndef_symbols, includes and add_include() from Clinic to
Codegen. Add a 'codegen' (Codegen) attribute to Clinic.
* Remove libclinic.crenderdata module: move code to libclinic.codegen.
* BlockPrinter.print_block(): remove unused 'limited_capi' argument.
Remove also 'core_includes' parameter.
* Add get_includes() methods.
* Make Codegen.ifndef_symbols private.
* Make Codegen.includes private.
* Make CConverter.includes private.
Introduce a unified 16-bit backoff counter type (``_Py_BackoffCounter``),
shared between the Tier 1 adaptive specializer and the Tier 2 optimizer. The
API used for adaptive specialization counters is changed but the behavior is
(supposed to be) identical.
The behavior of the Tier 2 counters is changed:
- There are no longer dynamic thresholds (we never varied these).
- All counters now use the same exponential backoff.
- The counter for ``JUMP_BACKWARD`` starts counting down from 16.
- The ``temperature`` in side exits starts counting down from 64.
Add libclinic.clanguage module and move the following classes and
functions there:
* CLanguage
* declare_parser()
Add libclinic.codegen and move the following classes there:
* BlockPrinter
* BufferSeries
* Destination
Move the following functions to libclinic.function:
* permute_left_option_groups()
* permute_optional_groups()
* permute_right_option_groups()
* Add a new create_parser_namespace() function for
PythonParser to pass objects to executed code.
* In run_clinic(), list converters using 'converters' and
'return_converters' dictionarties.
* test_clinic: add 'object()' return converter.
* Use also create_parser_namespace() in eval_ast_expr().
Co-authored-by: Erlend E. Aasland <erlend@python.org>
Miss Islington (bot)