Re: [RFCv2 PATCH 01/12] stk-webcam: the initial hflip and vflip setup was the wrong way around

[Hans de Goede <hdegoede@redhat.com>]

Hi,

On 02/11/2013 02:51 PM, Hans Verkuil wrote:
> On Mon February 11 2013 14:41:08 Hans de Goede wrote:
>> Hi,
>>
>> On 02/11/2013 02:21 PM, Hans Verkuil wrote:
>>> On Mon February 11 2013 14:08:44 Hans de Goede wrote:
>>>> Hi,
>>>>
>>>> Subject: stk-webcam: the initial hflip and vflip setup was the wrong way around
>>>>
>>>> No it is not.
>>>
>>> You are right, that patch makes no sense. It was a long day :-)
>>>
>>>> On 02/10/2013 06:52 PM, Hans Verkuil wrote:
>>>>> From: Hans Verkuil <hans.verkuil@cisco.com>
>>>>>
>>>>> This resulted in an upside-down picture.
>>>>
>>>> No it does not, the laptop having an upside down mounted camera and not being
>>>> in the dmi-table is what causes an upside down picture. For a non upside
>>>> down camera (so no dmi-match) hflip and vflip should be 0.
>>>>
>>>> The fix for the upside-down-ness Arvydas Sidorenko reported would be to
>>>> add his laptop to the upside down table.
>>>
>>> That doesn't make sense either. Arvydas, it worked fine for you before, right?
>>
>> Yes, it probably worked before, but not with...
>>
>>> That is, if you use e.g. v3.8-rc7 then your picture is the right side up.
>>
>> 3.8 will show it upside down for Arvydas
>>
>> The story goes likes this:
>>
>> 1) Once upon a time the stkwebcam driver was written
>> 2) The webcam in question was used mostly in Asus laptop models, including
>> the laptop of the original author of the driver, and in these models, in
>> typical Asus fashion (see the long long list for uvc cams inside v4l-utils),
>> they mounted the webcam-module the wrong way up. So the hflip and vflip
>> module options were given a default value of 1 (the correct value for
>> upside down mounted models)
>>
>> 3) Years later I got a bug report from a user with a laptop with stkwebcam,
>> where the module was actually mounted the right way up, and thus showed upside
>> down under Linux. So now I was facing the choice of 2 options:
>> a) Add a not-upside-down list to stkwebcam, which overrules the default
>> b) Do it like all the other drivers do, and make the default right for
>> cams mounted the proper way and add an upside-down model list, with models
>> where we need to flip-by-default.
>>
>> Despite knowing that going b) would cause a period of pain where we were
>> building the table (ie what we're discussing now) I opted to go for option
>> b), since a) is just too ugly, and worse different from how every other
>> driver does it leading to confusion in the long run.
>>
>> IOW this is entirely my fault, and I take full responsibility for it.
>
> Ah, OK. Now it makes sense. I wasn't aware of this history and it (clearly)
> confused me greatly.
>
> Can you perhaps provide me with a patch that adds some comments to the source
> explaining this. And in particular with which kernel this change took place?

Feel free to copy my 1) - 3) From above to a comment, step 3 landed in kernel 3.6
(you doing it seems better then me doing a patch conflicting with your patchset)

> The next time some poor sod (e.g. me) has to work on this the comments should
> explain this history.

Ack.

Regards,

Hans