large. Since we can't handle a rsize or wsize larger than that currently, cap those options at the number of kmap slots we have. We could consider capping it even lower, but we currently default to a max of 1M. Might as well allow those luddites on 32 bit arches enough rope to hang themselves. A more robust fix would be to teach the send and receive routines how to contend with an array of pages so we don't need to marshal up a kvec array at all. That's a fairly significant overhaul though, so we'll need this limit in place until that's ready. Cc: Reported-by: Jian Li Signed-off-by: Jeff Layton Signed-off-by: Steve French ‘‡ėT”8x