Bug report

Bug description:

Issue

ShareableList allocates string slots based on character count instead of UTF-8 byte count, causing corruption for multi-byte characters. Additionally,rstrip(b'\x00') strips legitimate trailing null bytes frombytes values.

Note: The trailing null bytes issue was originally reported in#106939 (July 2023) and documented as a known issue with a workaround. This fix attempts to resolve both that long-standing issue and the newly discovered UTF-8 corruption bug.

Reproducer

frommultiprocessing.shared_memoryimportShareableList# String corruptionsl=ShareableList(['0\U00010000\U00010000'])print(sl[0])# UnicodeDecodeErrorsl.shm.close();sl.shm.unlink()# Bytes corruptionsl=ShareableList([b'\x00'])print(repr(sl[0]))# b'' instead of b'\x00'sl.shm.close();sl.shm.unlink()

CPython versions tested on:

CPython main branch

Operating systems tested on:

No response

Linked PRs

• gh-145261: multiprocessing.shared_memory: fix ShareableList corruption for multi-byte strings and null bytes #145266