Does anyone know why m5 doesn't die when its stdout is closed? When tracediff gives up, instead of the two m5 jobs dying, they just keep going until you kill them explicitly. A typical program would die immediately on the SIGPIPE.
You can verify this behavior by running m5 with tracing on and piping the output through head. I tried a quick fix by adding 'signal(SIGPIPE, exitNowHandler);' to init.cc, but that didn't help, so I'm figuring that it's getting tangled up in python somewhere. However I don't see anywhere in our python code where we're catching IOError (which is what SIGPIPE shows up as). Steve
_______________________________________________ m5-dev mailing list m5-dev@m5sim.org http://m5sim.org/mailman/listinfo/m5-dev