On Tue, May 15, 2018 at 03:46:37AM +0000, Zheng, Jian Xu wrote:
> Hi Sakari,
> 
> > -----Original Message-----
> > From: [email protected] [mailto:linux-media-
> > [email protected]] On Behalf Of Sakari Ailus
> > Sent: Saturday, May 12, 2018 8:48 PM
> > To: Zheng, Jian Xu <[email protected]>
> > Cc: Chen, JasonX Z <[email protected]>; Tomasz Figa
> > <[email protected]>; Yeh, Andy <[email protected]>; Linux Media
> > Mailing List <[email protected]>; Chiang, AlanX
> > <[email protected]>; Qiu, Tian Shu <[email protected]>
> > Subject: Re: [PATCH v11] media: imx258: Add imx258 camera sensor driver
> > 
> > On Thu, May 10, 2018 at 11:08:43AM +0000, Zheng, Jian Xu wrote:
> > > Hi Sakari & Jason,
> > >
> > > > -----Original Message-----
> > > > From: [email protected] [mailto:linux-media-
> > > > [email protected]] On Behalf Of Sakari Ailus
> > > > Sent: Wednesday, May 9, 2018 5:43 PM
> > > > To: Chen, JasonX Z <[email protected]>
> > > > Cc: Tomasz Figa <[email protected]>; Yeh, Andy
> > > > <[email protected]>; Linux Media Mailing List
> > > > <[email protected]>; Chiang, AlanX
> > > > <[email protected]>
> > > > Subject: Re: [PATCH v11] media: imx258: Add imx258 camera sensor
> > > > driver
> > > >
> > > > Hi Jason,
> > > >
> > > > On Wed, May 09, 2018 at 09:28:30AM +0000, Chen, JasonX Z wrote:
> > > > > Hello Tomasz
> > > > >
> > > > > >> +/* Test Pattern Control */
> > > > > >> +#define IMX258_REG_TEST_PATTERN                0x0600
> > > > > >> +#define IMX258_TEST_PATTERN_DISABLE    0
> > > > > >> +#define IMX258_TEST_PATTERN_SOLID_COLOR        1
> > > > > >> +#define IMX258_TEST_PATTERN_COLOR_BARS 2 #define
> > > > > >> +IMX258_TEST_PATTERN_GREY_COLOR 3
> > > > > >> +#define IMX258_TEST_PATTERN_PN9                4
> > >
> > > I suppose we only use IMX258_TEST_PATTERN_COLOR_BARS. I heard that
> > > we'd better remove the functions/code no one would use. Is that true? e.g.
> > > remove all h_flip and v_flip ioctls because it's not used by anyone.
> > 
> > HFLIP and VFLIP support were AFAIR removed as they were not supported
> > correctly by the driver --- they do affect the pixel order, i.e. the format.

> So you mean that if the sensor is RAW format, we should not implement
> HFLIP/VFLIP. Because it changes the pixel order? I don't quite follow the
> logic here. In the meantime, if the logic works here, we need some other
> kind of HFLIP/VFLIP ioctl because it's supported by HW.

You can implement it but the implementation needs to be correctly reflect
the flipping in the pixel order reported by the driver.

-- 
Sakari Ailus
[email protected]

Reply via email to