gh-140544: store pointer to interpreter state as a thread local for fast access #140573
Conversation
kumaraditya303
requested review from
ZeroIntensity and
ericsnowcurrently
as code owners
kumaraditya303
force-pushed
the
interp-tls
branch
from
428d3e0 to
7115c75
Compare
Misc/NEWS.d/next/Core_and_Builtins/2025-10-25-07-25-52.gh-issue-140544.lwjtQe.rst
Show resolved
Hide resolved
| _Py_EnsureTstateNotNULL(tstate); | ||
| #endif | ||
| return tstate->interp; | ||
| #if !defined(Py_BUILD_CORE_MODULE) |
There was a problem hiding this comment.
Should this be
#if defined(HAVE_THREAD_LOCAL) && !defined(Py_BUILD_CORE_MODULE)
instead?
When testing the current main against mypyc, I'm seeing a new error message
In file included from /.../mypy/mypyc/lib-rt/misc_ops.c:1137:
In file included from /.../cpython/Include/internal/pycore_object.h:12:
In file included from /.../cpython/Include/internal/pycore_gc.h:12:
/.../cpython/Include/internal/pycore_pystate.h:213:12: error: use of undeclared identifier '_Py_tss_interp'
return _Py_tss_interp;
^
1 error generated.
Mypyc does use some of the internal headers. In particular
#define Py_BUILD_CORE
...
#include "internal/pycore_object.h"
There was a problem hiding this comment.
I guess it is because mypyc doesn't define HAVE_THREAD_LOCAL which is defined when python is built, does #140623 fixes it for you?
There was a problem hiding this comment.
I think we should just remove the check for HAVE_THREAD_LOCAL on line 92.
There was a problem hiding this comment.
I guess it is because mypyc doesn't define HAVE_THREAD_LOCAL which is defined when python is built, does #140623 fixes it for you?
Yes, that works.
I think we should just remove the check for
HAVE_THREAD_LOCALon line 92.
Guess, this would work too.
3 participants
Uh oh!
There was an error while loading. Please reload this page.