Re: [PATCH] [media] rc: call input_sync after scancode reports

[Jarod Wilson <jarod@redhat.com>]

Dmitry Torokhov wrote:
> Hi Jarod,
>
> On Thu, Jun 23, 2011 at 01:58:06PM -0400, Jarod Wilson wrote:
>> @@ -623,6 +624,7 @@ static void ir_do_keydown(struct rc_dev *dev, int scancode,
>>   			  u32 keycode, u8 toggle)
>>   {
>>   	input_event(dev->input_dev, EV_MSC, MSC_SCAN, scancode);
>> +	input_sync(dev->input_dev);
>>
>>   	/* Repeat event? */
>>   	if (dev->keypressed&&
>
> It looks like we would be issuing up to 3 input_sync() for a single
> keypress... Order of events is wrong too (we first issue MSC_SCAN for
> new key and then release old key). How about we change it a bit like in
> the patch below?

Yeah, your patch does result in a nicer overall flow of things (esp. the 
ordering of the release, which I hadn't noticed), and eliminates the 
extra unnecessary syncs. I've got one tiny tweak, where I just pass a 
true/false to ir_do_keyup to say whether or not it should do a sync to 
further reduce some code duplication. Building and testing here locally 
to make sure it does behave as expected, will then send it along.

-- 
Jarod Wilson
jarod@redhat.com