p2p_segwit.py fails with thread sanitizer #28801

Current behaviour

Using clang version 15.0.7 on Ubuntu and enabling the thread sanitizer, the p2p_segwit.py --descriptors test consistently fails.

It fails during the Subtest: test_wtxid_relay (Segwit active = True).

It's different from #28800 in that this isn't a timeout. The subtests fails after about 5 seconds.

Node returned unexpected exit code (66) vs (0) when stopping

Expected behaviour

Don't fail.

Steps to reproduce

export TSAN_OPTIONS="suppressions=$(pwd)/test/sanitizer_suppressions/tsan:halt_on_error=1:second_deadlock_stack=1"
./autogen
./configure CC=clang CXX=clang++ --enable-suppress-external-warnings --with-sanitizers=thread --without-gui
make -j33
test/functional/p2p_segwit.py

Relevant log output

stderr:
Traceback (most recent call last):
  File "/home/sjors/dev/bitcoin/test/functional/p2p_segwit.py", line 2070, in <module>
    SegWitTest().main()
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/test_framework.py", line 155, in main
    exit_code = self.shutdown()
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/test_framework.py", line 314, in shutdown
    self.stop_nodes()
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/test_framework.py", line 580, in stop_nodes
    node.wait_until_stopped()
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/test_node.py", line 409, in wait_until_stopped
    wait_until_helper_internal(lambda: self.is_node_stopped(expected_ret_code=expected_ret_code, **kwargs), timeout=timeout, timeout_factor=self.timeout_factor)
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/util.py", line 264, in wait_until_helper_internal
    if predicate():
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/test_node.py", line 409, in <lambda>
    wait_until_helper_internal(lambda: self.is_node_stopped(expected_ret_code=expected_ret_code, **kwargs), timeout=timeout, timeout_factor=self.timeout_factor)
  File "/home/sjors/dev/bitcoin/test/functional/test_framework/test_node.py", line 389, in is_node_stopped
    assert return_code == expected_ret_code, self._node_msg(
AssertionError: [node 0] Node returned unexpected exit code (66) vs (0) when stopping

How did you obtain Bitcoin Core

Compiled from source

What version of Bitcoin Core are you using?

master@953d302a242381ae13112ea42f87d57e6e796147

Operating system and version

Ubuntu 23.04

Machine specifications

AMD Ryzen 7950x

Some warnings from the sanitizer: https://gist.github.com/Sjors/607c3426136db1c91b89acea3c62affb

(running again with the supressions on)

Closing this. When I first saw this failure it was while running multiple tests in parallel. That was run by a script which does use supressions. But when I tried to reproduce I forgot the suppressions. Running the indidivual test with suppressions on doesn't fail. And when I ran the test suite in parallel again it also didn't fail. So I think it was just a random failure.

I'll try to get the logs from the parallel run next time; that way we can diagnose these more random failures.