Hi! For testing rmind's latest tmpfs patch on NetBSD-6.99.25/amd64/clang/libc++, I ran my usual pbulk-in-a-tmpfs test and very repeatable got system hangs (not panics). I have since reverted to a tmpfs-bugfix version that was working stably for me for a long time and still see this behaviour.
They all have in common that the USB keyboard and mouse stop working. Sometimes I was sitting in front of the machine and could look at the widgets I have running (in X) to report network traffic and the current time. Here I saw the following behaviours: * Both network and time widget keep working (!) * (No network activity) time widget jumps between two seconds, e.g. 10:13:53 and 10:53:54, displaying 10:13:53 for ~3 seconds, switching to 10:13:54 for a very short time and then switching back. * Complete hang, no activity. A good way to cause this seems to be building lang/mono2, which is infamous for its not-quite-correct use of libpthread. For this reason my best guess is that it's something to do with libpthreads. I don't understand the behaviour I'm seeing though. Please note that the machine doesn't panic -- otherwise I'd have seen reboots instead. I'm regularly updating, so I think this problem is rather new (last month or so). I was usually running with clang; using libc++ is new, no idea if this is related or not (I guess not). Does anyone have any ideas about this? Thanks, Thomas