Re: CVS commit: src/sys/dev/bluetooth

[Radoslaw Kujawa <radoslaw.kujawa%c0ff33.net@localhost>]

Hi Iain.

Sorry, I should have consulted you before committing this.

On 1 sty 2012, at 20:44, Iain Hibbert wrote:

> On Sat, 31 Dec 2011, Radoslaw Kujawa wrote:
> 
>> Module Name: src
>> Committed By:        rkujawa
>> Date:                Sat Dec 31 01:16:09 UTC 2011
>> 
>> Modified Files:
>>      src/sys/dev/bluetooth: bthidev.c btkbd.c
>> 
>> Log Message:
>> Fix panic triggered by pressing the caps lock key:
>> http://c0ff33.net/drop/bt_caps_panic.jpg
> 
> this does not seem right - I've never had such a panic,

I can easily reproduce this problem with Apple Wireless Keyboard (version 
without numeric keypad) on NetBSD/amd64 HEAD. If this patch is not applied, 
pressing caps lock always results in panic, as in photo.

> and btkbd should
> really be Bluetooth agnostic since its the same as ukbd (except that
> ukbd is tied too closely to the USB stack)

So we shouldn't fiddle with bt_lock in btkbd.c ?

> btkbd_set_leds() may be called from wskbd directly (by pressing caps lock
> on your built-in keyboard for instance)

I've tested this patch by pressing the caps lock key on bt keyboard and issuing 
wsconsctl ledstate=1, and both methods work. 

> probably it should be that bt_lock is released in bthidev_input() before
> calling the hid_output function..

I'm not sure if I understand correctly. The bthidev_input() does not acquire 
bt_lock at all.

-- 
Best regards,
Radoslaw Kujawa