Re: (EXT) Re: [PATCH 3/8] media: imx: Use dedicated format handler for i.MX7/8

[Alexander Stein]

Hi Laurent,

Am Samstag, 5. Februar 2022, 05:04:50 CET schrieb Laurent Pinchart:
> Hi Alexander and Dorota,
> 
> Thank you for the patch.
> 
> On Fri, Feb 04, 2022 at 01:15:09PM +0100, Alexander Stein wrote:
> > From: Dorota Czaplejewicz <dorota.czaplejewicz@xxxxxxx>
> > 
> > This splits out a format handler which takes into account
> > the capabilities of the i.MX7/8 video device,
> > as opposed to the default handler compatible with both i.MX5/6 and
> > i.MX7/8.
> > 
> > Signed-off-by: Dorota Czaplejewicz <dorota.czaplejewicz@xxxxxxx>
> > Signed-off-by: Alexander Stein <alexander.stein@xxxxxxxxxxxxxxx>
> > ---
> > Changes in comparison to original commit from Dorota:
> > * is_imx56 is used instead of enum
> > 
> >  drivers/staging/media/imx/imx-media-utils.c | 56 +++++++++++++++++++--
> >  1 file changed, 52 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/staging/media/imx/imx-media-utils.c
> > b/drivers/staging/media/imx/imx-media-utils.c index
> > 0daa6aad45f4..32aaa2e81bea 100644
> > --- a/drivers/staging/media/imx/imx-media-utils.c
> > +++ b/drivers/staging/media/imx/imx-media-utils.c
> > @@ -516,10 +516,9 @@ void imx_media_try_colorimetry(struct
> > v4l2_mbus_framefmt *tryfmt,> 
> >  }
> >  EXPORT_SYMBOL_GPL(imx_media_try_colorimetry);
> > 
> > -int imx_media_mbus_fmt_to_pix_fmt(struct v4l2_pix_format *pix,
> > -				  const struct v4l2_mbus_framefmt 
*mbus,
> > -				  const struct imx_media_pixfmt 
*cc,
> > -				  bool is_imx56)
> > +static int imx56_media_mbus_fmt_to_pix_fmt(struct v4l2_pix_format *pix,
> > +					   const struct 
v4l2_mbus_framefmt *mbus,
> > +					   const struct 
imx_media_pixfmt *cc)
> > 
> >  {
> >  
> >  	u32 width;
> >  	u32 stride;
> > 
> > @@ -568,6 +567,55 @@ int imx_media_mbus_fmt_to_pix_fmt(struct
> > v4l2_pix_format *pix,> 
> >  	return 0;
> >  
> >  }
> > 
> > +
> > +static int imx78_media_mbus_fmt_to_pix_fmt(struct v4l2_pix_format *pix,
> > +					   const struct 
v4l2_mbus_framefmt *mbus,
> > +					   const struct 
imx_media_pixfmt *cc)
> > +{
> > +	int ret;
> > +
> > +	if (!cc)
> > +		cc = imx_media_find_mbus_format(mbus->code, 
PIXFMT_SEL_ANY);
> > +
> > +	if (!cc)
> > +		return -EINVAL;
> > +	/*
> > +	 * The hardware can handle line lengths divisible by 4 pixels
> > +	 * as long as the whole buffer size ends up divisible by 8 bytes.
> > +	 * If not, use the value of 8 pixels recommended in the datasheet.
> > +	 */
> > +	ret = v4l2_fill_pixfmt(pix, cc->fourcc,
> > +			       round_up(mbus->width, 4), mbus-
>height);
> > +	if (ret)
> > +		return ret;
> > +
> > +	/* Only 8bits-per-pixel formats may need to get aligned to 8 
pixels,
> > +	 * because both 10-bit and 16-bit pixels occupy 2 bytes.
> > +	 * In those, 4-pixel aligmnent is equal to 8-byte alignment.
> > +	 */
> > +	if (pix->sizeimage % 8 != 0)
> > +		ret = v4l2_fill_pixfmt(pix, cc->fourcc,
> > +				       round_up(mbus->width, 8), 
mbus->height);
> 
> I think you could simplify this by using cc->bpp to figure out the
> alignment instead of calling v4l2_fill_pixfmt() twice.

Yeah, this should be possible. I can't test it (yet) though, I don't have a 8-
Bit sensor. Will do in the next version.

> > +	pix->colorspace = mbus->colorspace;
> > +	pix->xfer_func = mbus->xfer_func;
> > +	pix->ycbcr_enc = mbus->ycbcr_enc;
> > +	pix->quantization = mbus->quantization;
> > +	pix->field = mbus->field;
> 
> Should v4l2_fill_pixfmt() be updated to handle colorspace and field too
> instead of doing it manually here ?

I'm not so sure which parts of v4l2_pix_format shall be set into 
v4l2_fill_pixfmt(). Unfortunately there is no API description in the header, 
git log doesn't help much either. In the end this should be a separate patch.

Best regards,
Alexander

> > +
> > +	return ret;
> > +}
> > +
> > +int imx_media_mbus_fmt_to_pix_fmt(struct v4l2_pix_format *pix,
> > +				  const struct v4l2_mbus_framefmt 
*mbus,
> > +				  const struct imx_media_pixfmt 
*cc,
> > +				  bool is_imx56)
> > +{
> > +	if (is_imx56)
> > +		return imx56_media_mbus_fmt_to_pix_fmt(pix, mbus, cc);
> > +	else
> > +		return imx78_media_mbus_fmt_to_pix_fmt(pix, mbus, cc);
> > +}
> > 
> >  EXPORT_SYMBOL_GPL(imx_media_mbus_fmt_to_pix_fmt);
> >  
> >  void imx_media_free_dma_buf(struct device *dev,