> The code is broken anyways. If you free pages without flushingI've attached a more complicated patch that does a 2 stage effort to unmapping and freeing pages. My kernel no longer hangs with this patch... Jiri can you confirm? I'll look at the other issue separately.. Dave.
