Re: [PATCH v3 08/13] davinci: vpif: add support for clipping on output data

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Hans,

On Monday 25 June 2012 15:08:10 Hans Verkuil wrote:
> On Mon 25 June 2012 14:54:39 Laurent Pinchart wrote:
> > On Monday 25 June 2012 16:37:30 Manjunath Hadli wrote:
> > > add hardware clipping support for VPIF output data. This
> > > is needed as it is possible that the external encoder
> > > might get confused between the FF or 00 which are a part
> > > of the data and that of the SAV or EAV codes.
> > > 
> > > Signed-off-by: Manjunath Hadli <manjunath.hadli@ti.com>
> > > Signed-off-by: Lad, Prabhakar <prabhakar.lad@ti.com>
> > > ---
> > > 
> > >  drivers/media/video/davinci/vpif.h         |   30 +++++++++++++++++++++
> > >  drivers/media/video/davinci/vpif_display.c |   10 +++++++++
> > >  include/media/davinci/vpif_types.h         |    2 +
> > >  3 files changed, 42 insertions(+), 0 deletions(-)
> > > 
> > > diff --git a/drivers/media/video/davinci/vpif.h
> > > b/drivers/media/video/davinci/vpif.h index a4d2141..c2ce4d9 100644
> > > --- a/drivers/media/video/davinci/vpif.h
> > > +++ b/drivers/media/video/davinci/vpif.h
> > > @@ -211,6 +211,12 @@ static inline void vpif_clr_bit(u32 reg, u32 bit)
> > > 
> > >  #define VPIF_CH3_INT_CTRL_SHIFT	(6)
> > >  #define VPIF_CH_INT_CTRL_SHIFT	(6)
> > > 
> > > +#define VPIF_CH2_CLIP_ANC_EN	14
> > > +#define VPIF_CH2_CLIP_ACTIVE_EN	13
> > > +
> > > +#define VPIF_CH3_CLIP_ANC_EN	14
> > > +#define VPIF_CH3_CLIP_ACTIVE_EN	13
> > > +
> > > 
> > >  /* enabled interrupt on both the fields on vpid_ch0_ctrl register */
> > >  #define channel0_intr_assert()	(regw((regr(VPIF_CH0_CTRL)|\
> > >  
> > >  	(VPIF_INT_BOTH << VPIF_CH0_INT_CTRL_SHIFT)), VPIF_CH0_CTRL))
> > > 
> > > @@ -515,6 +521,30 @@ static inline void channel3_raw_enable(int enable,
> > > u8
> > > index) vpif_clr_bit(VPIF_CH3_CTRL, mask);
> > > 
> > >  }
> > > 
> > > +/* function to enable clipping (for both active and blanking regions)
> > > on ch 2 */ +static inline void channel2_clipping_enable(int enable)
> > > +{
> > > +	if (enable) {
> > > +		vpif_set_bit(VPIF_CH2_CTRL, VPIF_CH2_CLIP_ANC_EN);
> > > +		vpif_set_bit(VPIF_CH2_CTRL, VPIF_CH2_CLIP_ACTIVE_EN);
> > > +	} else {
> > > +		vpif_clr_bit(VPIF_CH2_CTRL, VPIF_CH2_CLIP_ANC_EN);
> > > +		vpif_clr_bit(VPIF_CH2_CTRL, VPIF_CH2_CLIP_ACTIVE_EN);
> > > +	}
> > > +}
> > > +
> > > +/* function to enable clipping (for both active and blanking regions)
> > > on ch 2 */ +static inline void channel3_clipping_enable(int enable)
> > > +{
> > > +	if (enable) {
> > > +		vpif_set_bit(VPIF_CH3_CTRL, VPIF_CH3_CLIP_ANC_EN);
> > > +		vpif_set_bit(VPIF_CH3_CTRL, VPIF_CH3_CLIP_ACTIVE_EN);
> > > +	} else {
> > > +		vpif_clr_bit(VPIF_CH3_CTRL, VPIF_CH3_CLIP_ANC_EN);
> > > +		vpif_clr_bit(VPIF_CH3_CTRL, VPIF_CH3_CLIP_ACTIVE_EN);
> > > +	}
> > > +}
> > > +
> > > 
> > >  /* inline function to set buffer addresses in case of Y/C non mux mode
> > >  */
> > >  static inline void ch2_set_videobuf_addr_yc_nmux(unsigned long
> > > 
> > > top_strt_luma, unsigned long btm_strt_luma,
> > > diff --git a/drivers/media/video/davinci/vpif_display.c
> > > b/drivers/media/video/davinci/vpif_display.c index 61ea8bc..4436ef6
> > > 100644
> > > --- a/drivers/media/video/davinci/vpif_display.c
> > > +++ b/drivers/media/video/davinci/vpif_display.c
> > > @@ -1046,6 +1046,8 @@ static int vpif_streamon(struct file *file, void
> > > *priv, channel2_intr_assert();
> > > 
> > >  			channel2_intr_enable(1);
> > >  			enable_channel2(1);
> > > 
> > > +			if (vpif_config_data->ch2_clip_en)
> > > +				channel2_clipping_enable(1);
> > > 
> > >  		}
> > >  		
> > >  		if ((VPIF_CHANNEL3_VIDEO == ch->channel_id)
> > > 
> > > @@ -1053,6 +1055,8 @@ static int vpif_streamon(struct file *file, void
> > > *priv, channel3_intr_assert();
> > > 
> > >  			channel3_intr_enable(1);
> > >  			enable_channel3(1);
> > > 
> > > +			if (vpif_config_data->ch3_clip_en)
> > > +				channel3_clipping_enable(1);
> > > 
> > >  		}
> > >  		channel_first_int[VPIF_VIDEO_INDEX][ch->channel_id] = 1;
> > >  	
> > >  	}
> > > 
> > > @@ -1065,6 +1069,8 @@ static int vpif_streamoff(struct file *file, void
> > > *priv, struct vpif_fh *fh = priv;
> > > 
> > >  	struct channel_obj *ch = fh->channel;
> > >  	struct common_obj *common = &ch->common[VPIF_VIDEO_INDEX];
> > > 
> > > +	struct vpif_display_config *vpif_config_data =
> > > +					vpif_dev->platform_data;
> > > 
> > >  	if (buftype != V4L2_BUF_TYPE_VIDEO_OUTPUT) {
> > >  	
> > >  		vpif_err("buffer type not supported\n");
> > > 
> > > @@ -1084,11 +1090,15 @@ static int vpif_streamoff(struct file *file,
> > > void
> > > *priv, if (buftype == V4L2_BUF_TYPE_VIDEO_OUTPUT) {
> > > 
> > >  		/* disable channel */
> > >  		if (VPIF_CHANNEL2_VIDEO == ch->channel_id) {
> > > 
> > > +			if (vpif_config_data->ch2_clip_en)
> > > +				channel2_clipping_enable(0);
> > > 
> > >  			enable_channel2(0);
> > >  			channel2_intr_enable(0);
> > >  		
> > >  		}
> > >  		if ((VPIF_CHANNEL3_VIDEO == ch->channel_id) ||
> > >  		
> > >  					(2 == common->started)) {
> > > 
> > > +			if (vpif_config_data->ch3_clip_en)
> > > +				channel3_clipping_enable(0);
> > > 
> > >  			enable_channel3(0);
> > >  			channel3_intr_enable(0);
> > >  		
> > >  		}
> > > 
> > > diff --git a/include/media/davinci/vpif_types.h
> > > b/include/media/davinci/vpif_types.h index bd8217c..d8f6ab1 100644
> > > --- a/include/media/davinci/vpif_types.h
> > > +++ b/include/media/davinci/vpif_types.h
> > > @@ -50,6 +50,8 @@ struct vpif_display_config {
> > > 
> > >  	const char **output;
> > >  	int output_count;
> > >  	const char *card_name;
> > > 
> > > +	bool ch2_clip_en;
> > > +	bool ch3_clip_en;
> > 
> > Instead of hardcoding this in platform data, I think it would be better to
> > make this runtime-configurable. One option is to use the value of the
> > v4l2_pix_format::colorspace field configured by userspace. We already have
> > V4L2_COLORSPACE_JPEG which maps to the full 0-255 range, but we're missing
> > a colorspace for the clipped 1-254 range used by the VPIF and I'm not
> > sure whether it would really make sense to add one. Another option is to
> > use a V4L2 control.
> 
> This is something for a V4L2 control I think since clipping is colorspace
> independent (you have the same situation for e.g. YCbCr).

But some colorspaces define ranges smaller than 0-255. In this particular case 
I agree with you though, as the hardware clips to 1-255 without taking 
colorspaces into account.

-- 
Regards,

Laurent Pinchart