[ARROW-4582] [C++/Python] Memory corruption on Pandas->Arrow conversion

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.11.0, 0.11.1, 0.12.0
• Fix Version/s: 0.12.1, 0.13.0
• Component/s: C++, Python
• Labels: pull-request-available
• External issue URL: https://github.com/apache/arrow/issues/21126
• Assignee: Uwe Korn
• Reporter: Uwe Korn
• Created: 15/Feb/19 10:43
• Updated: 11/Jan/23 07:34
• Resolved: 15/Feb/19 15:45

Description

When converting DataFrames with numerical columns to Arrow tables we were seeing random segfaults in core Python code. This only happened in environments where we had a high level of parallelisation or slow code execution (e.g. in AddressSanitizer builds).

The reason for these segfaults was that we were incrementing the reference count of the underlying NumPy buffer but were not holding the GIL while changing the reference count.

Issue Links

• links to: GitHub Pull Request #3655