On Mon, Jun 13, 2016 at 08:42:35AM -0700, Philip Guenther wrote: > On Sun, 12 Jun 2016, Dimitris Papastamos wrote: > > I was building ports and at the same time started chromium and the > > kernel panic-ed. > > > > This has happened a few times in the last week. It started happening at > > some point towards the end of May. It is not easy to reproduce locally. > > Thank you for the report! When you say "not easy to reproduce locally", > do you mean you _can_ semireliably reproduce it, such that if we found a > suspicious commit to revert you would be able to be pretty sure whether it > was really fixed? If so, what's your best guess on how to tickle it?
Currently I can trigger this once every 4-5 days under normal usage. I think I will be able to confirm whether the fix works or not but that might take twice that period to be more confident (in case the revert only affects the race window). It seems to happen when there is heavy file I/O and I start chromium. >From a quick look it seems that unp_gc() is related to UNIX domain sockets. It might be interesting to ktrace chromium and see what it does in regards to UNIX domain sockets. A simple proof of concept that issues the same syscalls in more or less the same order can be written and executed repeatedly in a tight loop along with a ports build. Alternatively one can kill and start chromium every few seconds. I will try to make such a proof of concept program and see if I can reliably reproduce it. I mentioned heavy file I/O above but I have not confirmed this. It could also be networking related as the ports tree uses the network. I have seen the same crash by doing pkg_add -u and then starting chromium.
