So I poked around this bug today, I think that the kill signal is something valgrind is doing to dislodge a thread that it perceives as being blocked.
I cannot reproduce on my desktop running without valgrind, but can easily reproduce under valgrind running:
/usr/bin/valgrind --error-exitcode=1 --trace-children=yes --leak-check=full --show-leak-kinds=definite --errors-for-leak-kinds=definite bin/mir_acceptance_tests --gtest_filter="ServerDisconnect.causes_client_to_terminate_by_default" --gtest_repeat=10
So I poked around this bug today, I think that the kill signal is something valgrind is doing to dislodge a thread that it perceives as being blocked.
I cannot reproduce on my desktop running without valgrind, but can easily reproduce under valgrind running: children= yes --leak-check=full --show- leak-kinds= definite --errors- for-leak- kinds=definite bin/mir_ acceptance_ tests --gtest_ filter= "ServerDisconne ct.causes_ client_ to_terminate_ by_default" --gtest_repeat=10
/usr/bin/valgrind --error-exitcode=1 --trace-
(similar to what the CI infrastructure is doing)