Record success in `specialize`
This matches the existing behavior where we increment the success
stat for the generic opcode each time we successfully specialize
an instruction.
If Python fails to start newly created thread
due to failure of underlying PyThread_start_new_thread() call,
its state should be removed from interpreter' thread states list
to avoid its double cleanup.
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
This gets rid of the immortal check in `PyStackRef_FromPyObjectSteal()`.
Overall, this improves performance about 2% in the free threading
build.
This also renames `PyStackRef_Is()` to `PyStackRef_IsExactly()` because
the macro requires that the tag bits of the arguments match, which is
only true in certain special cases.
This change enables custom GHA runners for Ubuntu-24.04 that run on Arm hardware. It also prepares for Windows runners on Arm hardware, but doesn't enable that just yet, because the Arm GHA runner images for Windows need to be updated.
Add free-threaded specialization for `UNPACK_SEQUENCE` opcode.
`UNPACK_SEQUENCE_TUPLE/UNPACK_SEQUENCE_TWO_TUPLE` are already thread safe since tuples are immutable.
`UNPACK_SEQUENCE_LIST` is not thread safe because of nature of lists (there is nothing preventing another thread from adding items to or removing them the list while the instruction is executing). To achieve thread safety we add a critical section to the implementation of `UNPACK_SEQUENCE_LIST`, especially around the parts where we check the size of the list and push items onto the stack.
---------
Co-authored-by: Matt Page <mpage@meta.com>
Co-authored-by: mpage <mpage@cs.stanford.edu>
Threads are gone after fork, so clear the queues too. Otherwise the
child process (here created via multiprocessing.Process) crashes on
interpreter exit.
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
* Name without a PATHEXT extension is only searched if the mode does not
include X_OK.
* Support multi-component PATHEXT extensions (e.g. ".foo.bar").
* Support files without extensions in PATHEXT contains dot-only extension
(".", "..", etc).
* Support PATHEXT extensions that end with a dot (e.g. ".foo.").
The usage parameter of argparse.ArgumentParser no longer
affects the default value of the prog parameter in subparsers.
Previously the full custom usage of the main parser was used as
the prog prefix in subparsers.
`mmap`, `munmap`, and `mprotect` are used by CPython for memory
management, which may occur in the middle of the FileIO tests. The
system calls can also be used with files, so `strace` includes them
in its `%file` and `%desc` filters.
Filter out the `mmap` system calls related to memory allocation for the
file tests. Currently FileIO doesn't do `mmap` at all, so didn't add
code to track from `mmap` through `munmap` since it wouldn't be used.
For now if an `mmap` on a fd happens, the call will be included (which
may cause test to fail), and at that time support for tracking the
address throug `munmap` could be added.
The `methodcaller` C vectorcall implementation uses an arguments array
that is shared across calls. The first argument is modified on every
invocation. This isn't thread-safe in the free threading build. I think
it's also not safe in general, but for now just disable it in the free
threading build.
Decode a file URI like `file://///server/share` as a UNC path like
`\\server\share`. This form of file URI is created by software the simply
prepends `file:///` to any absolute Windows path.
Discard any 'localhost' authority from the beginning of a `file:` URI. As a
result, file URIs like `//localhost/etc/hosts` are correctly decoded as
`/etc/hosts`.
Adjust `pathname2url()` to encode embedded colon characters in Windows
paths, rather than bailing out with an `OSError`.
Co-authored-by: Steve Dower <steve.dower@microsoft.com>
Enable specialization of LOAD_GLOBAL in free-threaded builds.
Thread-safety of specialization in free-threaded builds is provided by the following:
A critical section is held on both the globals and builtins objects during specialization. This ensures we get an atomic view of both builtins and globals during specialization.
Generation of new keys versions is made atomic in free-threaded builds.
Existing helpers are used to atomically modify the opcode.
Thread-safety of specialized instructions in free-threaded builds is provided by the following:
Relaxed atomics are used when loading and storing dict keys versions. This avoids potential data races as the dict keys versions are read without holding the dictionary's per-object lock in version guards.
Dicts keys objects are passed from keys version guards to the downstream uops. This ensures that we are loading from the correct offset in the keys object. Once a unicode key has been stored in a keys object for a combined dictionary in free-threaded builds, the offset that it is stored in will never be reused for a different key. Once the version guard passes, we know that we are reading from the correct offset.
The dictionary read fast-path is used to read values from the dictionary once we know the correct offset.
This is a precursor to the actual fix for gh-114940, where we will change these macros to use the new lock. This change is almost entirely mechanical; the exceptions are the loops in codeobject.c and ceval.c, which now hold the "head" lock. Note that almost all of the uses of _Py_FOR_EACH_TSTATE_UNLOCKED() here will change to _Py_FOR_EACH_TSTATE_BEGIN() once we add the new per-interpreter lock.
* Fix support of STRING and GLOBAL opcodes with non-ASCII arguments.
* dis() now outputs non-ASCII bytes in STRING, BINSTRING and
SHORT_BINSTRING arguments as escaped (\xXX).
Distribution tooling (ex. sandbox on Gentoo and fakeroot on Debian) uses
LD_PRELOAD to intercept system calls and potentially modify them when
building. These tools can change the set of system calls, so disable
system call testing under these cases.
Co-authored-by: Michał Górny <mgorny@gentoo.org>
Don't take a reason in unspecialize
We only want to compute the reason if stats are enabled. Optimizing
compilers should optimize this away for us (gcc and clang do), but
it's better to be safe than sorry.
This adds authentication to the forkserver control socket. In the past only filesystem permissions protected this socket from code injection into the forkserver process by limiting access to the same UID, which didn't exist when Linux abstract namespace sockets were used (see issue) meaning that any process in the same system network namespace could inject code. We've since stopped using abstract namespace sockets by default, but protecting our control sockets regardless of type is a good idea.
This reuses the HMAC based shared key auth already used by `multiprocessing.connection` sockets for other purposes.
Doing this is useful so that filesystem permissions are not relied upon and trust isn't implied by default between all processes running as the same UID with access to the unix socket.
### pyperformance benchmarks
No significant changes. Including `concurrent_imap` which exercises `multiprocessing.Pool.imap` in that suite.
### Microbenchmarks
This does _slightly_ slow down forkserver use. How much so appears to depend on the platform. Modern platforms and simple platforms are less impacted. This PR adds additional IPC round trips to the control socket to tell forkserver to spawn a new process. Systems with potentially high latency IPC are naturally impacted more.
Typically a 1-4% slowdown on a very targeted process creation microbenchmark, with a worst case overloaded system slowdown of 20%. No evidence that these slowdowns appear in practical sense. See the PR for details.
Savannah Ostrowski
Rafael Fontenelle
mpage
Irit Katriel
Radislav Chugunov
Filipe Laíns 🇵🇸
Sam Gross
Łukasz Langa
Kirill Podoprigora
Eric Snow
Serhiy Storchaka
Andrei Bodrov
Jun Komoda
Victor Stinner
Sergey Muraviov
Cody Maloney
Tomas R.
Peter Bierma
Barney Gale
Jacob Bower
Donghee Na
Dino Viehland
Hugo van Kemenade
Nice Zombies
Hood Chatham
Zachary Ware
Mark Shannon
Gregory P. Smith
Brandt Bucher