Re: [PATCH] Add support for M-5MOLS 8 Mega Pixel camera

["Kim, HeungJun" <riverful.kim@samsung.com>]

Hi Sylwester,

2011-03-20 오전 4:28, Sylwester Nawrocki 쓴 글:
> On 03/19/2011 04:11 PM, Kim HeungJun wrote:
>> Hi Sylwester,
>>
>> Thanks for the reviews. :)
>>
>> 2011. 3. 19., 오후 10:37, Sylwester Nawrocki 작성:
>>
>>> Hi HeungJun,
>>>
>>> On 03/16/2011 02:38 PM, Kim, Heungjun wrote:
>>>> Add I2C/V4L2 subdev driver for M-5MOLS camera sensor with integrated
>>>> image signal processor.
>>>>
[snip]
>>>>
>>>> 3. Speed-up whole I2C operation
>>>> 	: I've tested several times for decreasing the stabilization time
>>>> 	while I2C communication, and I have find proper time. Of course,
>>>> 	it's more faster than previous version.
>>>
>>> That sounds good. Do you think the delays before I2C read/write could
>>> be avoided in some (if not all) cases by using some status registers
>>> polling?
>> I don't understand literally. Could you explain more detailed with some examples?
>> My understanding is that it might be an issues or problem when getting some
>> status registers with polling it. is it right?
> 
> My concern is that we might not need an extra delay between consecutive 
> read or write operations in every case. Possibly it would be enough
> to read the status of some operations instead. But that just what I suspect.
Ah, I understand. The consecutive delay needs to read/write operations.
Because, in non-delay cases, I found the values is not changed to be read
when the driver is reading continuously at the different address. I means,
if there is not regular amount of delay in r/w operations, the value is kept
the previous value to be read by I2C operation. It's the same way in writing
case. It's wierd.

It might be different the kind of M-5MOLS sensor's FW and packaging type. But,
I expect that the internal ARM core of M-5MOLS takes time to proceed something
and to load values through I2C bus for the host process. 

Regards,
Heungjun Kim