Re: [PATCH v4] netpoll: protect napi_poll and poll_controller during dev_[open|close]
From: David Miller <davem@davemloft.net>
Date: 2013-02-03 21:06:00
From: David Miller <davem@davemloft.net>
Date: 2013-02-03 21:06:00
From: Neil Horman <nhorman@tuxdriver.com> Date: Sat, 2 Feb 2013 13:51:16 -0500
I can't say Im a huge fan of returning -EBUSY in open/close, but I like it better than busy waiting while the poll path completes.
Sorry Neil, this is a show stopper for me. There is no reason why, just because of bad timing, a device open or close that would normally succeed, should fail. And this situation is only being created because the mutual exclusion and context handling is "hard", well that's too bad :-)