Keyboard drivers, polling vs. non-polling mode

[Maxim Ignatenko]

On 13 May 2014 22:26, Maxim Ignatenko <gelraen.ua at gmail.com> wrote:
> On 13 May 2014 11:17, Maxim Ignatenko <gelraen.ua at gmail.com> wrote:
>> On 13 May 2014 09:40, Ruslan Bukin <br at bsdpad.com> wrote:
>>> On Sun, May 11, 2014 at 11:33:42PM +0100, Maxim Ignatenko wrote:
>>>> Hello,
>>>>
>>>> I'm trying trying to get keyboard working in DDB on HP Chromebook 11 (ARM).
>>>> br@ said that it doesn't work there because polling mode is not implemented yet.
>>>> Where can I read about the difference between polling and non-polling
>>>> modes (and about keyboard drivers in general)?
>>>> sys/dev/kbd/kbdreg.h describes some structures and method signatures,
>>>> but I have no clue what is the expected behaviour of those methods.
>>>>
>>>> My current guess is that in polling mode keyboard driver just queues
>>>> up all the input coming from keyboard and then gives it to consumer
>>>> upon request, while in non-polling mode it invokes some callback
>>>> instead of queueing. But I cannot find any documentation to confirm or
>>>> disprove that.
>>>>
>>>
>>> Chrome Embedded Controller (EC) provides interrupt (KB_GPIO_INT pin, active low)
>>> reporting that there are pending data and you need to read the data using
>>> ec_command(..). After all data was read, pin comes to 1 (not active).
>>>
>>> We have no interrupts in KDB, so you have to check pin status manually and
>>> if status == 0 (active) then read new data.
>>>
>>> Probably you can start with patch attached (I did't tested).
>>
>> Thanks, I've tried something like that, except with invoking
>> ec_command immediately and comparing returned state to previous one
>> rather than reading status bit from GPIO pin.
>> I keep getting "fdb0: i2c transfer returned 6" and none of the printfs
>> I've added to iicbus_transfer_gen in sys/dev/iicbus/iiconf.c.
>
> Here's what I have now. After booting kernel from flash drive I get
> kdb prompt (because of panic() call in exynos5_ehci.c) and when I
> press any key - screen gets spammed with "fdb0: i2c transfer returned
> 6". printfs added to dev/iicbus/*.c are not triggered. Any ideas?

Sorry, it's "fbd0", not "fdb0".

-- 
Best regards,
Maxim