Re: [PATCH v7 05/15] media: i2c: imx219: Add embedded data support

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Laurent,

On Wed, Mar 27, 2024 at 02:22:45AM +0200, Laurent Pinchart wrote:
> Hi Sakari,
> 
> On Tue, Mar 26, 2024 at 11:30:30PM +0000, Sakari Ailus wrote:
> > On Mon, Mar 25, 2024 at 12:08:41AM +0200, Laurent Pinchart wrote:
> > > The IMX219 generates embedded data unconditionally. Report it as an
> > > additional stream, with a new internal embedded data pad, and update
> > > subdev operations accordingly.
> > > 
> > > Signed-off-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
> > > ---
> > > Changes since v6:
> > > 
> > > - Get format from IMX219_STREAM_IMAGE in imx219_set_ctrl()
> > > - Fix mbus code for second stream in imx219_get_frame_desc()
> > > - Set V4L2_SUBDEV_ROUTE_FL_IMMUTABLE flag on route
> > > ---
> > >  drivers/media/i2c/imx219.c | 188 +++++++++++++++++++++++++++++++------
> > >  1 file changed, 160 insertions(+), 28 deletions(-)
> > > 
> > > diff --git a/drivers/media/i2c/imx219.c b/drivers/media/i2c/imx219.c
> > > index fa64bc402c9a..86a0ebf6d65f 100644
> > > --- a/drivers/media/i2c/imx219.c
> > > +++ b/drivers/media/i2c/imx219.c
> > > @@ -149,6 +149,9 @@
> > >  #define IMX219_PIXEL_ARRAY_WIDTH	3280U
> > >  #define IMX219_PIXEL_ARRAY_HEIGHT	2464U
> > >  
> > > +/* Embedded metadata stream height */
> > > +#define IMX219_EMBEDDED_DATA_HEIGHT	2U
> > > +
> > >  /* Mode : resolution and related config&values */
> > >  struct imx219_mode {
> > >  	/* Frame width */
> > > @@ -317,9 +320,15 @@ static const struct imx219_mode supported_modes[] = {
> > >  enum imx219_pad_ids {
> > >  	IMX219_PAD_SOURCE,
> > >  	IMX219_PAD_IMAGE,
> > > +	IMX219_PAD_EDATA,
> > >  	IMX219_NUM_PADS,
> > >  };
> > >  
> > > +enum imx219_stream_ids {
> > > +	IMX219_STREAM_IMAGE,
> > > +	IMX219_STREAM_EDATA,
> > > +};
> > > +
> > >  struct imx219 {
> > >  	struct v4l2_subdev sd;
> > >  	struct media_pad pads[IMX219_NUM_PADS];
> > > @@ -382,7 +391,8 @@ static int imx219_set_ctrl(struct v4l2_ctrl *ctrl)
> > >  	int ret = 0;
> > >  
> > >  	state = v4l2_subdev_get_locked_active_state(&imx219->sd);
> > > -	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE);
> > > +	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +					      IMX219_STREAM_IMAGE);
> > >  
> > >  	if (ctrl->id == V4L2_CID_VBLANK) {
> > >  		int exposure_max, exposure_def;
> > > @@ -610,6 +620,25 @@ static unsigned int imx219_format_bpp(u32 code)
> > >  	}
> > >  }
> > >  
> > > +/* Return the embedded data format corresponding to an image format. */
> > > +static u32 imx219_format_edata(u32 code)
> > > +{
> > > +	switch (code) {
> > > +	case MEDIA_BUS_FMT_SRGGB8_1X8:
> > > +	case MEDIA_BUS_FMT_SGRBG8_1X8:
> > > +	case MEDIA_BUS_FMT_SGBRG8_1X8:
> > > +	case MEDIA_BUS_FMT_SBGGR8_1X8:
> > > +		return MEDIA_BUS_FMT_META_8;
> > > +
> > > +	case MEDIA_BUS_FMT_SRGGB10_1X10:
> > > +	case MEDIA_BUS_FMT_SGRBG10_1X10:
> > > +	case MEDIA_BUS_FMT_SGBRG10_1X10:
> > > +	case MEDIA_BUS_FMT_SBGGR10_1X10:
> > > +	default:
> > > +		return MEDIA_BUS_FMT_META_10;
> > 
> > Something like this could be nice in the framework. But there are many
> > others that would be useful, too, so let's think of this separately.
> 
> Indeed. Let's see later, when we'll have multiple drivers sharing the
> same needs.
> 
> > > +	}
> > > +}
> > > +
> > >  static int imx219_set_framefmt(struct imx219 *imx219,
> > >  			       struct v4l2_subdev_state *state)
> > >  {
> > > @@ -619,7 +648,8 @@ static int imx219_set_framefmt(struct imx219 *imx219,
> > >  	u64 bin_h, bin_v;
> > >  	int ret = 0;
> > >  
> > > -	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE);
> > > +	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +					      IMX219_STREAM_IMAGE);
> > >  	crop = v4l2_subdev_state_get_crop(state, IMX219_PAD_IMAGE);
> > >  	bpp = imx219_format_bpp(format->code);
> > >  
> > > @@ -774,17 +804,33 @@ static int imx219_enum_mbus_code(struct v4l2_subdev *sd,
> > >  {
> > >  	struct imx219 *imx219 = to_imx219(sd);
> > >  
> > > -	if (code->pad == IMX219_PAD_IMAGE) {
> > > +	switch (code->pad) {
> > > +	case IMX219_PAD_IMAGE:
> > >  		/* The internal image pad is hardwired to the native format. */
> > > -		if (code->index)
> > > +		if (code->index > 0)
> > 
> > Why? It's unsigned so this has the same effect in a little bit more
> > complicated way.
> 
> I find it more explicit, as it reads as "index values larger than the
> limit are invalid" instead of "non-zero index values are invalid".
> Compare it with
> 
> 		if (code->index >= (ARRAY_SIZE(imx219_mbus_formats) / 4))
> 			return -EINVAL;

The redundant parentheses are extra btw.

> 
> As the value is unsigned, the compiler shouldn't generate less effective
> code.
> 
> > >  			return -EINVAL;
> > >  
> > >  		code->code = IMX219_NATIVE_FORMAT;
> > > -	} else {
> > > -		/*
> > > -		 * On the source pad, the sensor supports multiple raw formats
> > > -		 * with different bit depths.
> > > -		 */
> > > +		return 0;
> > > +
> > > +	case IMX219_PAD_EDATA:
> > > +		if (code->index > 0)
> > 
> > Same here actually, and elsewhere.
> > 
> > > +			return -EINVAL;
> > > +
> > > +		code->code = MEDIA_BUS_FMT_CCS_EMBEDDED;
> > > +		return 0;
> > > +
> > > +	case IMX219_PAD_SOURCE:
> > > +	default:
> > > +		break;
> > > +	}
> > > +
> > > +	/*
> > > +	 * On the source pad, the sensor supports multiple image raw formats
> > > +	 * with different bit depths. The embedded data format bit depth
> > > +	 * follows the image stream.
> > > +	 */
> > > +	if (code->stream == IMX219_STREAM_IMAGE) {
> > >  		u32 format;
> > >  
> > >  		if (code->index >= (ARRAY_SIZE(imx219_mbus_formats) / 4))
> > > @@ -792,6 +838,15 @@ static int imx219_enum_mbus_code(struct v4l2_subdev *sd,
> > >  
> > >  		format = imx219_mbus_formats[code->index * 4];
> > >  		code->code = imx219_get_format_code(imx219, format);
> > > +	} else {
> > > +		struct v4l2_mbus_framefmt *fmt;
> > > +
> > > +		if (code->index > 0)
> > > +			return -EINVAL;
> > > +
> > > +		fmt = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +						   IMX219_STREAM_EDATA);
> > > +		code->code = fmt->code;
> > >  	}
> > >  
> > >  	return 0;
> > > @@ -803,7 +858,8 @@ static int imx219_enum_frame_size(struct v4l2_subdev *sd,
> > >  {
> > >  	struct imx219 *imx219 = to_imx219(sd);
> > >  
> > > -	if (fse->pad == IMX219_PAD_IMAGE) {
> > > +	switch (fse->pad) {
> > > +	case IMX219_PAD_IMAGE:
> > >  		if (fse->code != IMX219_NATIVE_FORMAT || fse->index > 0)
> > >  			return -EINVAL;
> > >  
> > > @@ -811,7 +867,24 @@ static int imx219_enum_frame_size(struct v4l2_subdev *sd,
> > >  		fse->max_width = IMX219_NATIVE_WIDTH;
> > >  		fse->min_height = IMX219_NATIVE_HEIGHT;
> > >  		fse->max_height = IMX219_NATIVE_HEIGHT;
> > > -	} else {
> > > +		return 0;
> > > +
> > > +	case IMX219_PAD_EDATA:
> > > +		if (fse->code != MEDIA_BUS_FMT_CCS_EMBEDDED || fse->index > 0)
> > > +			return -EINVAL;
> > > +
> > > +		fse->min_width = IMX219_NATIVE_WIDTH;
> > > +		fse->max_width = IMX219_NATIVE_WIDTH;
> > > +		fse->min_height = IMX219_EMBEDDED_DATA_HEIGHT;
> > > +		fse->max_height = IMX219_EMBEDDED_DATA_HEIGHT;
> > > +		return 0;
> > > +
> > > +	case IMX219_PAD_SOURCE:
> > > +	default:
> > > +		break;
> > > +	}
> > > +
> > > +	if (fse->stream == IMX219_STREAM_IMAGE) {
> > >  		if (fse->code != imx219_get_format_code(imx219, fse->code) ||
> > >  		    fse->index >= ARRAY_SIZE(supported_modes))
> > >  			return -EINVAL;
> > > @@ -820,6 +893,21 @@ static int imx219_enum_frame_size(struct v4l2_subdev *sd,
> > >  		fse->max_width = fse->min_width;
> > >  		fse->min_height = supported_modes[fse->index].height;
> > >  		fse->max_height = fse->min_height;
> > > +	} else {
> > > +		struct v4l2_mbus_framefmt *fmt;
> > > +
> > > +		fmt = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +						   IMX219_STREAM_EDATA);
> > > +		if (fse->code != fmt->code)
> > > +			return -EINVAL;
> > > +
> > > +		if (fse->index)
> > > +			return -EINVAL;
> > 
> > But this seems nice. :-)
> 
> You're right that this lacks consistency, I can change this one :-)

:-o

> 
> > > +
> > > +		fse->min_width = fmt->width;
> > > +		fse->max_width = fmt->width;
> > > +		fse->min_height = IMX219_EMBEDDED_DATA_HEIGHT;
> > > +		fse->max_height = IMX219_EMBEDDED_DATA_HEIGHT;
> > >  	}
> > >  
> > >  	return 0;
> > > @@ -831,6 +919,7 @@ static int imx219_set_pad_format(struct v4l2_subdev *sd,
> > >  {
> > >  	struct imx219 *imx219 = to_imx219(sd);
> > >  	const struct imx219_mode *mode;
> > > +	struct v4l2_mbus_framefmt *ed_format;
> > >  	struct v4l2_mbus_framefmt *format;
> > >  	struct v4l2_rect *compose;
> > >  	struct v4l2_rect *crop;
> > > @@ -838,9 +927,9 @@ static int imx219_set_pad_format(struct v4l2_subdev *sd,
> > >  
> > >  	/*
> > >  	 * The driver is mode-based, the format can be set on the source pad
> > > -	 * only.
> > > +	 * only, and only for the image streeam.
> > >  	 */
> > > -	if (fmt->pad != IMX219_PAD_SOURCE)
> > > +	if (fmt->pad != IMX219_PAD_SOURCE || fmt->stream != IMX219_STREAM_IMAGE)
> > >  		return v4l2_subdev_get_fmt(sd, state, fmt);
> > >  
> > >  	/*
> > > @@ -897,15 +986,31 @@ static int imx219_set_pad_format(struct v4l2_subdev *sd,
> > >  	 * No mode use digital crop, the source pad crop rectangle size and
> > >  	 * format are thus identical to the image pad compose rectangle.
> > >  	 */
> > > -	crop = v4l2_subdev_state_get_crop(state, IMX219_PAD_SOURCE);
> > > +	crop = v4l2_subdev_state_get_crop(state, IMX219_PAD_SOURCE,
> > > +					  IMX219_STREAM_IMAGE);
> > >  	crop->left = 0;
> > >  	crop->top = 0;
> > >  	crop->width = fmt->format.width;
> > >  	crop->height = fmt->format.height;
> > >  
> > > -	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE);
> > > +	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +					      IMX219_STREAM_IMAGE);
> > >  	*format = fmt->format;
> > >  
> > > +	/*
> > > +	 * Finally, update the formats on the sink and source sides of the
> > > +	 * embedded data stream.
> > > +	 */
> > > +	ed_format = v4l2_subdev_state_get_format(state, IMX219_PAD_EDATA);
> > > +	ed_format->code = imx219_format_edata(format->code);
> > > +	ed_format->width = format->width;
> > > +	ed_format->height = IMX219_EMBEDDED_DATA_HEIGHT;
> > > +	ed_format->field = V4L2_FIELD_NONE;
> > > +
> > > +	format = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +					      IMX219_STREAM_EDATA);
> > > +	*format = *ed_format;
> > > +
> > >  	if (fmt->which == V4L2_SUBDEV_FORMAT_ACTIVE) {
> > >  		int exposure_max;
> > >  		int exposure_def;
> > > @@ -944,6 +1049,13 @@ static int imx219_get_selection(struct v4l2_subdev *sd,
> > >  {
> > >  	struct v4l2_rect *compose;
> > >  
> > > +	/*
> > > +	 * The embedded data stream doesn't support selection rectangles,
> > > +	 * neither on the embedded data pad nor on the source pad.
> > > +	 */
> > > +	if (sel->pad == IMX219_PAD_EDATA || sel->stream != 0)
> > 
> > I'd leave "!= 0" out. Up to you.
> 
> Same here, I find it better to write "only stream 0 is valid" than
> "streams are invalid". I actually went back and worth, and believe it or
> not, this was the result of careful consideration :-)

I've also considered this and disagree. :-) But overall, neither is faulty
obviously. Up to you.

> 
> > > +		return -EINVAL;
> > > +
> > >  	switch (sel->target) {
> > >  	case V4L2_SEL_TGT_NATIVE_SIZE:
> > >  		if (sel->pad != IMX219_PAD_IMAGE)
> > > @@ -996,12 +1108,19 @@ static int imx219_get_selection(struct v4l2_subdev *sd,
> > >  static int imx219_init_state(struct v4l2_subdev *sd,
> > >  			     struct v4l2_subdev_state *state)
> > >  {
> > > -	struct v4l2_subdev_route routes[1] = {
> > > +	struct v4l2_subdev_route routes[2] = {
> > 
> > Do you need to specify the number of the entries?
> 
> No, I'll drop it.
> 
> > >  		{
> > >  			.sink_pad = IMX219_PAD_IMAGE,
> > >  			.sink_stream = 0,
> > >  			.source_pad = IMX219_PAD_SOURCE,
> > > -			.source_stream = 0,
> > > +			.source_stream = IMX219_STREAM_IMAGE,
> > > +			.flags = V4L2_SUBDEV_ROUTE_FL_ACTIVE |
> > > +				 V4L2_SUBDEV_ROUTE_FL_IMMUTABLE,
> > > +		}, {
> > > +			.sink_pad = IMX219_PAD_EDATA,
> > > +			.sink_stream = 0,
> > > +			.source_pad = IMX219_PAD_SOURCE,
> > > +			.source_stream = IMX219_STREAM_EDATA,
> > >  			.flags = V4L2_SUBDEV_ROUTE_FL_ACTIVE |
> > >  				 V4L2_SUBDEV_ROUTE_FL_IMMUTABLE,
> > >  		},
> > > @@ -1014,7 +1133,7 @@ static int imx219_init_state(struct v4l2_subdev *sd,
> > >  	struct v4l2_subdev_format fmt = {
> > >  		.which = V4L2_SUBDEV_FORMAT_TRY,
> > >  		.pad = IMX219_PAD_SOURCE,
> > > -		.stream = 0,
> > > +		.stream = IMX219_STREAM_IMAGE,
> > >  		.format = {
> > >  			.code = MEDIA_BUS_FMT_SRGGB10_1X10,
> > >  			.width = supported_modes[0].width,
> > > @@ -1027,6 +1146,10 @@ static int imx219_init_state(struct v4l2_subdev *sd,
> > >  	if (ret)
> > >  		return ret;
> > >  
> > > +	/*
> > > +	 * Set the image stream format on the source pad. This will be
> > > +	 * propagated to all formats and selection rectangles internally.
> > > +	 */
> > >  	imx219_set_pad_format(sd, state, &fmt);
> > >  
> > >  	return 0;
> > > @@ -1035,29 +1158,36 @@ static int imx219_init_state(struct v4l2_subdev *sd,
> > >  static int imx219_get_frame_desc(struct v4l2_subdev *sd, unsigned int pad,
> > >  				 struct v4l2_mbus_frame_desc *fd)
> > >  {
> > > -	const struct v4l2_mbus_framefmt *fmt;
> > >  	struct v4l2_subdev_state *state;
> > > -	u32 code;
> > > +	u32 img_code;
> > > +	u32 ed_code;
> > >  
> > >  	if (pad != IMX219_PAD_SOURCE)
> > >  		return -EINVAL;
> > >  
> > >  	state = v4l2_subdev_lock_and_get_active_state(sd);
> > > -	fmt = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE, 0);
> > > -	code = fmt->code;
> > > +	img_code = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +						IMX219_STREAM_IMAGE)->code;
> > > +	ed_code = v4l2_subdev_state_get_format(state, IMX219_PAD_SOURCE,
> > > +					       IMX219_STREAM_EDATA)->code;
> > >  	v4l2_subdev_unlock_state(state);
> > >  
> > >  	fd->type = V4L2_MBUS_FRAME_DESC_TYPE_CSI2;
> > > -	fd->num_entries = 1;
> > > +	fd->num_entries = 2;
> > 
> > I'd keep an entry pointer and keep incrementing it, same for the number of
> > entries. Up to you. Two is fine.
> 
> I think I reviewed a patch of yours and proposed replacing the pointer
> incrementation by explicit indices :-) I find this more readable.

This approach does not scale. For a small number of entries such as two
it's fine but once you have ~ 5 I'd rather see the approach I suggested: no
explicit integers are used and you can easily validate the result at the
end by comparing it to the size of the array.

> 
> > >  
> > >  	memset(fd->entry, 0, sizeof(fd->entry));
> > >  
> > > -	fd->entry[0].pixelcode = code;
> > > -	fd->entry[0].stream = 0;
> > > +	fd->entry[0].pixelcode = img_code;
> > > +	fd->entry[0].stream = IMX219_STREAM_IMAGE;
> > >  	fd->entry[0].bus.csi2.vc = 0;
> > > -	fd->entry[0].bus.csi2.dt = imx219_format_bpp(code) == 8
> > > +	fd->entry[0].bus.csi2.dt = imx219_format_bpp(img_code) == 8
> > >  				 ? MIPI_CSI2_DT_RAW8 : MIPI_CSI2_DT_RAW10;
> > >  
> > > +	fd->entry[1].pixelcode = ed_code;
> > > +	fd->entry[1].stream = IMX219_STREAM_EDATA;
> > > +	fd->entry[1].bus.csi2.vc = 0;
> > > +	fd->entry[1].bus.csi2.dt = MIPI_CSI2_DT_EMBEDDED_8B;
> > > +
> > >  	return 0;
> > >  }
> > >  
> > > @@ -1321,12 +1451,14 @@ static int imx219_probe(struct i2c_client *client)
> > >  	/*
> > >  	 * Initialize the pads. To preserve backward compatibility with
> > >  	 * userspace that used the sensor before the introduction of the
> > > -	 * internal image pad, the external source pad is numbered 0 and the
> > > -	 * internal image pad numbered 1.
> > > +	 * internal pads, the external source pad is numbered 0 and the internal
> > > +	 * image and embedded data pads numbered 1 and 2 respectively.
> > >  	 */
> > >  	imx219->pads[IMX219_PAD_SOURCE].flags = MEDIA_PAD_FL_SOURCE;
> > >  	imx219->pads[IMX219_PAD_IMAGE].flags = MEDIA_PAD_FL_SINK
> > >  					     | MEDIA_PAD_FL_INTERNAL;
> > > +	imx219->pads[IMX219_PAD_EDATA].flags = MEDIA_PAD_FL_SINK
> > > +					     | MEDIA_PAD_FL_INTERNAL;
> > >  
> > >  	ret = media_entity_pads_init(&imx219->sd.entity,
> > >  				     ARRAY_SIZE(imx219->pads), imx219->pads);
> 

-- 
Regards,

Sakari Ailus




[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux