Re: [PATCH] HID: Separate struct hid_device's driver_lock into two locks.

On Mon, 26 Nov 2012, Benjamin Tissoires wrote:

It is clear that the goal is to make commit
4ea5454203d991ec85264f64f89ca8855fce69b0 less restrictive.
The problem is that this commit stops any communication with the
device, even configuration communication.

Logitech devices use their own protocol on top of the HID standard
protocol. For touch devices, this proprietary protocol requires to ask
the device for axis ranges, etc...

So here, the idea is not to open the can of worm for every hid devices
through hw_start() / hw_stop() calls, I think the idea of Andrew is
just to allow hid-logitech-dj to get rid of this restriction in some
particular circumstances.
Consider this as a controlled backdoor of the can of worms :)

I have been thinking for this for a while, and I wasn't able to come up 
with anything I'd personally consider substantially better aproach than 
splitting the lock. So I don't have strong objections at the very moment.

This will also allow us to get rid of what commit 596264082f introduced in 
unifying driver, right? (adding Nestor to CC).

-- 
Jiri Kosina
SUSE Labs