Message-ID: <f071e691-5930-4738-9dbd-43ed2da367fe@gmail.com>
Date: Tue, 28 Jul 2026 08:00:01 +0300
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Content-Language: en-US
From: Alexander Lakhin <exclusion@gmail.com>
Subject: plpython tests fail against python 3.14 under Valgrind
To: pgsql-hackers <pgsql-hackers@postgresql.org>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/f071e691-5930-4738-9dbd-43ed2da367fe%40gmail.com>
Precedence: bulk

Hello hackers,

As buildfarm animal skink shows, upgrade to Python 3.14 breaks plpython
tests under Valgrind [1]:
  61/400 plpython - postgresql:plpython/regress ERROR             49.88s   exit status 1

pgsql.build/testrun/plpython/regress/log/postmaster.log
Fatal Python error: pycore_init_builtins: failed to add exceptions to builtins
Python runtime state: preinitialized
object address  : 0x139da4a0
object refcount : 1
object type     : 0x1388d020
object type name: RecursionError
object repr     :
lost sys.stderr

Current thread 0x0000000008c62ec0 [memcheck-amd64-] (most recent call first):
   <no Python frame>

I've reproduced this locally and confirmed that it's not reproduced with
PLPYTHON="python3.13" ./configure ...
It's also not reproduced on master, probably just because of a bit lesser
stack usage. I've bisected the anomaly and found what makes master pass
the tests: aeb07c55f.

AI suggests the offending change in Python 3.14 is [2]:
Use Py_EnterRecursiveCall() to guard against runaway recursion in C code

[1] https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=skink&dt=2026-07-27%2013%3A09%3A55
[2] https://docs.python.org/3/whatsnew/3.14.html

Best regards,
Alexander