The hs_err file shows the SIGILL in the compiler thread due to a handshake timeout. However the core dump shows that the compiler thread encountered a signal whilst executring mprotect form within malloc, and it then gets repeated signals until presumably we get the SIGILL due to the timeout. Meanwhile maloc is unusable and all other threads lock up and the WatcherThread then tries to kill the VM. All in all a bit of a train wreck and no way to be sure what happened first. See comments for stack details.