https://bugs.kde.org/show_bug.cgi?id=458915
--- Comment #10 from Libor Peltan <libor.pel...@nic.cz> --- @Philippe Thanks much for your deep analysis! I wouldn't be able to see such things in conext. However, I think you are not entirely correct. I think the observed bug is not caused by other thread aborting. The crash on assert seems to be the consequence, not the cause of the mess. I prepared another scenario, where the program does not crash at all. It just ignores the wrong 0xe8 value returned from epoll and continues as if it's zero. In this case, it happened five times during the run and the program did not crash, and operated smoohtly up to the end of test case. By the way, writing to fd 2 (stderr) is done only by Valgrind, due to the setting `-d -d -d`. I'm going to attach those logs, this time also the stderr. I hope this helps to find the definitive cause :) -- You are receiving this mail because: You are watching all bug changes.