Eric Snow [Thu, 6 Apr 2023 00:42:02 +0000 (18:42 -0600)]
gh-101659: Add _Py_AtExit() (gh-103298)
The function is like Py_AtExit() but for a single interpreter. This is a companion to the atexit module's register() function, taking a C callback instead of a Python one.
We also update the _xxinterpchannels module to use _Py_AtExit(), which is the motivating case. (This is inspired by pain points felt while working on gh-101660.)
gh-99069: Consolidate checks for static_assert (#94766)
Several platforms don't define the static_assert macro despite having
compiler support for the _Static_assert keyword. The macro needs to be
defined since it is used unconditionally in the Python code. So it
should always be safe to define it if undefined and not in C++11 (or
later) mode.
Hence, remove the checks for particular platforms or libc versions,
and just define static_assert anytime it needs to be defined but isn't.
That way, all platforms that need the fix will get it, regardless of
whether someone specifically thought of them.
Also document that certain macOS versions are among the platforms that
need this.
The C2x draft (currently expected to become C23) makes static_assert
a keyword to match C++. So only define the macro for up to C17.
Co-authored-by: Victor Stinner <vstinner@python.org>
Michael Handler [Wed, 5 Apr 2023 04:55:24 +0000 (21:55 -0700)]
gh-66897: Upgrade HTTP CONNECT to protocol HTTP/1.1 (#8305)
* bpo-22708: Upgrade HTTP CONNECT to protocol HTTP/1.1 (GH-NNNN)
Use protocol HTTP/1.1 when sending HTTP CONNECT tunnelling requests;
generate Host: headers if one is not already provided (required by
HTTP/1.1), convert IDN domains to punycode in HTTP CONNECT requests.
* Refactor tests to pass under -bb (fix ByteWarnings); missed some lines >80.
* Use consistent 'tunnelling' spelling in Lib/http/client.py
* Lib/test/test_httplib: Remove remnant of obsoleted test.
* Use dict.copy() not copy.copy()
* fix version changed
* Update Lib/http/client.py
Co-authored-by: bgehman <bgehman@users.noreply.github.com>
* Switch to for/else: syntax, as suggested
Eric Snow [Tue, 4 Apr 2023 23:03:40 +0000 (17:03 -0600)]
gh-102660: Fix is_core_module() (gh-103257)
In gh-102744 we added is_core_module() (in Python/import.c), which relies on get_core_module_dict() (also added in that PR). The problem is that_PyImport_FixupBuiltin(), which ultimately calls is_core_module(), is called on the builtins module before interp->builtins_copyis set. Consequently, the builtins module isn't considered a "core" module while it is getting "fixed up" and its module def m_copy erroneously gets set. Under isolated interpreters this causes problems since sys and builtins are allowed even though they are still single-phase init modules. (This was discovered while working on gh-101660.)
The solution is to stop relying on get_core_module_dict() in is_core_module().
Ned Deily [Tue, 4 Apr 2023 21:06:00 +0000 (17:06 -0400)]
gh-103207: Add instructions to the macOS installer welcome display on how to workaround the macOS 13 Ventura “The installer encountered an error” failure. (GH-103251)
T. Wouters [Tue, 4 Apr 2023 14:51:30 +0000 (16:51 +0200)]
Fix a compiler warning in _xxsubinterpretermodule.c (#103245)
Fix a (correct) warning about potential uses of uninitialized memory in
_xxsubinterpreter. Unlike newly allocated PyObject structs or global
structs, stack-allocated structs are not initialised, and a few places in
the code expect the _sharedexception struct data to be either NULL or
initialised.
David Benjamin [Mon, 3 Apr 2023 22:44:00 +0000 (07:44 +0900)]
gh-100062: Remove error code tables from _ssl and err_names_to_codes (GH-100063)
Prior to https://github.com/python/cpython/pull/25300, the
make_ssl_data.py script used various tables, exposed in _ssl, to update
the error list.
After that PR, this is no longer used. Moreover, the err_names_to_codes
map isn't used at all. Clean those up. This gets them out of the way if,
in the future, OpenSSL provides an API to do what the code here is doing
directly. (https://github.com/openssl/openssl/issues/19848)
Follow on from https://github.com/python/cpython/pull/103116.
Expand list of clean docs files from 3 to 181. These files have no Sphinx warnings, and their presence in this list means that any new warnings introduced will fail the build.
The list was created by subtracting the list of files with warnings from a list of all files.
I tested with all of those, but found that `touch`ing two clean files (https://github.com/python/cpython/blob/main/Doc/includes/wasm-notavail.rst and https://github.com/python/cpython/blob/main/Doc/whatsnew/changelog.rst) caused a cascade effect and resulted in a number of dirty files being rebuilt too, and failing the build. So those two have been omitted.
Eric Snow [Wed, 29 Mar 2023 23:15:43 +0000 (17:15 -0600)]
gh-100227: Make the Global PyModuleDef Cache Safe for Isolated Interpreters (gh-103084)
Sharing mutable (or non-immortal) objects between interpreters is generally not safe. We can work around that but not easily.
There are two restrictions that are critical for objects that break interpreter isolation.
The first is that the object's state be guarded by a global lock. For now the GIL meets this requirement, but a granular global lock is needed once we have a per-interpreter GIL.
The second restriction is that the object (and, for a container, its items) be deallocated/resized only when the interpreter in which it was allocated is the current one. This is because every interpreter has (or will have, see gh-101660) its own object allocator. Deallocating an object with a different allocator can cause crashes.
The dict for the cache of module defs is completely internal, which simplifies what we have to do to meet those requirements. To do so, we do the following:
* add a mechanism for re-using a temporary thread state tied to the main interpreter in an arbitrary thread
* add _PyRuntime.imports.extensions.main_tstate`
* add _PyThreadState_InitDetached() and _PyThreadState_ClearDetached() (pystate.c)
* add _PyThreadState_BindDetached() and _PyThreadState_UnbindDetached() (pystate.c)
* make sure the cache dict (_PyRuntime.imports.extensions.dict) and its items are all owned by the main interpreter)
* add a placeholder using for a granular global lock
Note that the cache is only used for legacy extension modules and not for multi-phase init modules.
Brett Cannon [Wed, 29 Mar 2023 20:28:08 +0000 (13:28 -0700)]
GH-102973: add a dev container (GH-102975)
On content update, builds `python` and the docs. Also adds a Dockerfile that should include everything but autoconf 2.69 that's necessary to build CPython and the entire stdlib on Fedora.
Co-authored-by: Ronald Oussoren <ronaldoussoren@mac.com> Co-authored-by: Dusty Phillips <dusty@phillips.codes>
Steve Dower [Tue, 28 Mar 2023 23:47:13 +0000 (00:47 +0100)]
gh-103097: Add workaround for Windows ARM64 compiler bug (GH-103098)
See https://developercommunity.visualstudio.com/t/Regression-in-MSVC-1433-1434-ARM64-co/10224361 for details of the issue. It only applies to version 14.34.
Eric Snow [Tue, 28 Mar 2023 18:52:28 +0000 (12:52 -0600)]
gh-100227: Move the Dict of Interned Strings to PyInterpreterState (gh-102339)
We can revisit the options for keeping it global later, if desired. For now the approach seems quite complex, so we've gone with the simpler isolation solution in the meantime.
Chenxi Mao [Tue, 28 Mar 2023 08:52:22 +0000 (16:52 +0800)]
GH-102711: Fix warnings found by clang (#102712)
There are some warnings if build python via clang:
Parser/pegen.c:812:31: warning: a function declaration without a prototype is deprecated in all versions of C [-Wstrict-prototypes]
_PyPegen_clear_memo_statistics()
^
void
Parser/pegen.c:820:29: warning: a function declaration without a prototype is deprecated in all versions of C [-Wstrict-prototypes]
_PyPegen_get_memo_statistics()
^
void
This approach to keeping the interned strings safe is turning out to be too complex for my taste (due to obmalloc isolation). For now I'm going with the simpler solution, making the dict per-interpreter. We can revisit that later if we want a sharing solution.