On Thu, Jan 31, 2019 at 04:45:35PM +0000, Michael Kelley wrote: > From: Sasha Levin <sas...@kernel.org> Sent: Thursday, January 31, 2019 7:20 AM > > > > I've queued this one for hyper-fixes, thanks all! > > > > Actually, please hold off on queuing this one. In a conversation I had > yesterday with Kim, they had identified a deadlock. Kim was going to be > looking at some revisions to avoid the deadlock. > > Kim -- please confirm. > > Michael
This is correct. I need to send a v2 of this patch which addresses two issues: 1) Check channel->state inside the mutex acquire/release in vmbus_chan_attr_show(). 2) Prevent a deadlock that can occur between the proposed mutex_lock() call in the vmbus_chan_attr_show() function and the sysfs/kernfs functions. I've identified two possible solutions to the deadlock: 1) Add a new mutex to the vmbus_channel struct which protects changes to "channel->state". Use this new mutex in vmbus_chan_attr_show() instead of "vmbus_connection.channel_mutex". 2) Use "vmbus_connection.channel_mutex" in vmbus_chan_attr_show() as originally proposed, and acquire it with mutex_trylock(). If the mutex cannot be acquired, return -EINVAL. I'm leaning towards (2), using mutex_trylock(). "vmbus_connection.channel_mutex" appears to be used only when a channel is being opened or closed, so vmbus_chan_attr_show() should be able to acquire the mutex when the channel is in use. If anyone has suggestions, please let me know. Thanks, Kim
