subject:"\[PATCH\] media\: i2c\: adv748x\: Fix pixel rate values"

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-05-05 Thread Niklas Söderlund

Hi Laurent,

On 2018-05-05 12:48:58 +0300, Laurent Pinchart wrote:
> Hi Niklas,
> 
> On Saturday, 5 May 2018 01:58:10 EEST Niklas Söderlund wrote:
> > On 2018-04-25 01:36:42 +0200, Niklas Söderlund wrote:
> > > On 2018-04-21 15:44:44 +0300, Laurent Pinchart wrote:
> > >> The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
> > >> include both horizontal and vertical blanking. Both the AFE and HDMI
> > >> receiver program it incorrectly:
> > >> 
> > >> - The HDMI receiver goes to the trouble of removing blanking to compute
> > >> the rate of active pixels. This is easy to fix by removing the
> > >> computation and returning the incoming pixel clock rate directly.
> > >> 
> > >> - The AFE performs similar calculation, while it should simply return
> > >> the fixed pixel rate for analog sources, mandated by the ADV748x to be
> > >> 28.63636 MHz.
> > >> 
> > >> Signed-off-by: Laurent Pinchart
> > >> 
> > > 
> > > Reviewed-by: Niklas Söderlund 
> > 
> > I'm afraid I would like to revoke this review tag, please see bellow.
> > 
> > > This patch uncovered a calculation error in rcar-csi2 which compensated
> > > for the removing of the blanking in the adv748x, thanks for that! Good
> > > thing that it's not merged yet, will include the fix in the next version
> > > of the CSI-2 driver.
> > > 
> > >> ---
> > >> 
> > >>  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
> > >>  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
> > >>  2 files changed, 6 insertions(+), 13 deletions(-)
> > >> 
> > >> diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c
> > >> b/drivers/media/i2c/adv748x/adv748x-afe.c index
> > >> 61514bae7e5c..3e18d5ae813b 100644
> > >> --- a/drivers/media/i2c/adv748x/adv748x-afe.c
> > >> +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
> > >> @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops
> > >> adv748x_afe_video_ops = {> > 
> > >>  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
> > >>  {
> > >>  struct v4l2_subdev *tx;
> > >> -unsigned int width, height, fps;
> > >> 
> > >>  tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
> > >>  if (!tx)
> > >>  return -ENOLINK;
> > >> 
> > >> -width = 720;
> > >> -height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
> > >> -fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
> > >> -
> > >> -return adv748x_csi2_set_pixelrate(tx, width * height * fps);
> > >> +/*
> > >> + * The ADV748x samples analog video signals using an externally
> > >> supplied
> > >> + * clock whose frequency is required to be 28.63636 MHz.
> > >> + */
> > >> +return adv748x_csi2_set_pixelrate(tx, 28636360);
> > 
> > I believe this is wrong. The sampling rate of the AFE is 28.63636 MHz
> > but the pixelrate output on the CSI-TXB might not be right?
> 
> There are only two ways that would allow the pixel rate output of the CSI-TXB 
> to be different than the sampling frequency. The SD core or TXB would need to 
> either resample or buffer the signal. Resampling would change the resolution 
> so that's out of question. Buffering would require a memory buffer large 
> enough to hold one line of data, which would be quite expensive in terms of 
> silicon, and is thus unlikely.

I'm the first to admit I have much to learn here :-) I tried reading up 
and with my limited understanding it seems possible that different 
components are sampled at different rates, and normal pixel clock for 
PAL and NTSC which kept popping up in my research was:

Digital video is a sampled form of analog video. The most common 
sampling schemes in use today are:

  Pixel Clock   HorizHorizVert
   Rate TotalActive
NTSC square pixel  12.27 MHz780  640  525
NTSC CCIR-601  13.5  MHz858  720  525
NTSC 4FSc  14.32 MHz910  768  525
PAL  square pixel  14.75 MHz944  768  625
PAL  CCIR-601  13.5  MHz864  720  625
PAL  4FSc  17.72 MHz   1135  948  625

For the CCIR-601 standards, the sampling is based on a static 
orthogonal sampling grid. The luminance component (Y) is sampled at 
13.5 MHz, while the two color difference signals, Cr and Cb are 
sampled at half that, or 6.75 MHz. The Cr and Cb samples are 
colocated with alternate Y samples, and they are taken at the same 
position on each line, such that one sample is coincident with the 
50% point of the falling edge of analog sync. The samples are coded 
to either 8 or 10 bits per component. 

ftp://ftp.sgi.com/sgi/video/rld/vidpage/sampling.html

Looking at timing diagrams it seems the VSYNC pulse is quiet long and 
that is AFIK represented by a short package on the CSI-2 bus so I feel 
the

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-05-05 Thread Laurent Pinchart

Hi Niklas,

On Saturday, 5 May 2018 01:58:10 EEST Niklas Söderlund wrote:
> On 2018-04-25 01:36:42 +0200, Niklas Söderlund wrote:
> > On 2018-04-21 15:44:44 +0300, Laurent Pinchart wrote:
> >> The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
> >> include both horizontal and vertical blanking. Both the AFE and HDMI
> >> receiver program it incorrectly:
> >> 
> >> - The HDMI receiver goes to the trouble of removing blanking to compute
> >> the rate of active pixels. This is easy to fix by removing the
> >> computation and returning the incoming pixel clock rate directly.
> >> 
> >> - The AFE performs similar calculation, while it should simply return
> >> the fixed pixel rate for analog sources, mandated by the ADV748x to be
> >> 28.63636 MHz.
> >> 
> >> Signed-off-by: Laurent Pinchart
> >> 
> > 
> > Reviewed-by: Niklas Söderlund 
> 
> I'm afraid I would like to revoke this review tag, please see bellow.
> 
> > This patch uncovered a calculation error in rcar-csi2 which compensated
> > for the removing of the blanking in the adv748x, thanks for that! Good
> > thing that it's not merged yet, will include the fix in the next version
> > of the CSI-2 driver.
> > 
> >> ---
> >> 
> >>  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
> >>  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
> >>  2 files changed, 6 insertions(+), 13 deletions(-)
> >> 
> >> diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c
> >> b/drivers/media/i2c/adv748x/adv748x-afe.c index
> >> 61514bae7e5c..3e18d5ae813b 100644
> >> --- a/drivers/media/i2c/adv748x/adv748x-afe.c
> >> +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
> >> @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops
> >> adv748x_afe_video_ops = {> > 
> >>  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
> >>  {
> >>struct v4l2_subdev *tx;
> >> -  unsigned int width, height, fps;
> >> 
> >>tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
> >>if (!tx)
> >>return -ENOLINK;
> >> 
> >> -  width = 720;
> >> -  height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
> >> -  fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
> >> -
> >> -  return adv748x_csi2_set_pixelrate(tx, width * height * fps);
> >> +  /*
> >> +   * The ADV748x samples analog video signals using an externally
> >> supplied
> >> +   * clock whose frequency is required to be 28.63636 MHz.
> >> +   */
> >> +  return adv748x_csi2_set_pixelrate(tx, 28636360);
> 
> I believe this is wrong. The sampling rate of the AFE is 28.63636 MHz
> but the pixelrate output on the CSI-TXB might not be right?

There are only two ways that would allow the pixel rate output of the CSI-TXB 
to be different than the sampling frequency. The SD core or TXB would need to 
either resample or buffer the signal. Resampling would change the resolution 
so that's out of question. Buffering would require a memory buffer large 
enough to hold one line of data, which would be quite expensive in terms of 
silicon, and is thus unlikely.

> The adv7482 is a complex and badly documented thing. But it's internal
> plumbing shows that the CVBS signal sampled by the AFE is passed to the SD
> core and then to the TXB CSI-2 transmitter.
> 
> Reading the documentation we have such registers as LTA[1:0] which
> controls the delay between the chroma and luma samples. I do believe the
> adv748x is running with the default of AUTO_PDC_EN which indicates that
> the timings are automatically controlled by the adv748x hardware. And
> there are other registers in the SD core which to me looks like it can
> be configured to make use of the samples so that it won't correlate 1:1
> with the pixel rate.
> 
> Looking at the CSI-TXA and CSI-TXB cores which are particularly badly
> documented and the driver contains a lot of undocumented register
> writes. One that is documented and I find interesting in this context is
> EN_AUTOCALC_DPHY_PARAMS which is set 'yes' for the adv748x driver. This
> leads me to believe that the pixel rate output on the CSI-2 bus is not
> correlated with the AFE sampling clock. There are lots of holes in the
> documentation here but some stand out, the undocumented hole around 0xda
> which contains the few documented bits in that area, MIPI_PLL_LOCK_FLAG,
> MIPI_PLL_CLK_DET and MIPI_PLL_EN. Maybe the true link_freq or pixel_rate
> could be read from a undocumented register in that darkness :-)
> 
> The real reason I started to dig into this is that after you corrected
> my assumption about how to setup the R-Car CSI-2 receiver link freq this
> change breaks capture of CVBS for me. Looking at how I on your
> suggestion calculate link speed.
> 
> link_freq = (pixel_rate * bits_per_sample) / (2 * nr_of_lanes)
> 
> pixel_rate = 28636360 (from the adv748x V4L2_CID_PIXEL_RATE)
> bits_per_sample = 16 (adv748x reports MEDIA_BUS_FMT_UYVY8_2X8)
> nr_of_lanes = 1 (Using TXB of the

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-05-04 Thread Niklas Söderlund

Hi Laurent,

On 2018-04-25 01:36:42 +0200, Niklas Söderlund wrote:
> Hi Laurent,
> 
> Thanks for your patch.
> 
> On 2018-04-21 15:44:44 +0300, Laurent Pinchart wrote:
> > The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
> > include both horizontal and vertical blanking. Both the AFE and HDMI
> > receiver program it incorrectly:
> > 
> > - The HDMI receiver goes to the trouble of removing blanking to compute
> > the rate of active pixels. This is easy to fix by removing the
> > computation and returning the incoming pixel clock rate directly.
> > 
> > - The AFE performs similar calculation, while it should simply return
> > the fixed pixel rate for analog sources, mandated by the ADV748x to be
> > 28.63636 MHz.
> > 
> > Signed-off-by: Laurent Pinchart 
> 
> Reviewed-by: Niklas Söderlund 

I'm afraid I would like to revoke this review tag, please see bellow.

> 
> This patch uncovered a calculation error in rcar-csi2 which compensated 
> for the removing of the blanking in the adv748x, thanks for that! Good 
> thing that it's not merged yet, will include the fix in the next version 
> of the CSI-2 driver.
> 
> > ---
> >  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
> >  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
> >  2 files changed, 6 insertions(+), 13 deletions(-)
> > 
> > diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c 
> > b/drivers/media/i2c/adv748x/adv748x-afe.c
> > index 61514bae7e5c..3e18d5ae813b 100644
> > --- a/drivers/media/i2c/adv748x/adv748x-afe.c
> > +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
> > @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops 
> > adv748x_afe_video_ops = {
> >  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
> >  {
> > struct v4l2_subdev *tx;
> > -   unsigned int width, height, fps;
> >  
> > tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
> > if (!tx)
> > return -ENOLINK;
> >  
> > -   width = 720;
> > -   height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
> > -   fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
> > -
> > -   return adv748x_csi2_set_pixelrate(tx, width * height * fps);
> > +   /*
> > +* The ADV748x samples analog video signals using an externally supplied
> > +* clock whose frequency is required to be 28.63636 MHz.
> > +*/
> > +   return adv748x_csi2_set_pixelrate(tx, 28636360);

I believe this is wrong. The sampling rate of the AFE is 28.63636 MHz 
but the pixelrate output on the CSI-TXB might not be right? The adv7482 
is a complex and badly documented thing. But it's internal plumbing 
shows that the CVBS signal sampled by the AFE is passed to the SD core 
and then to the TXB CSI-2 transmitter.  

Reading the documentation we have such registers as LTA[1:0] which 
controls the delay between the chroma and luma samples. I do believe the 
adv748x is running with the default of AUTO_PDC_EN which indicates that 
the timings are automatically controlled by the adv748x hardware. And 
there are other registers in the SD core which to me looks like it can 
be configured to make use of the samples so that it won't correlate 1:1 
with the pixel rate.

Looking at the CSI-TXA and CSI-TXB cores which are particularly badly 
documented and the driver contains a lot of undocumented register 
writes. One that is documented and I find interesting in this context is 
EN_AUTOCALC_DPHY_PARAMS which is set 'yes' for the adv748x driver. This 
leads me to believe that the pixel rate output on the CSI-2 bus is not 
correlated with the AFE sampling clock. There are lots of holes in the 
documentation here but some stand out, the undocumented hole around 0xda 
which contains the few documented bits in that area, MIPI_PLL_LOCK_FLAG, 
MIPI_PLL_CLK_DET and MIPI_PLL_EN. Maybe the true link_freq or pixel_rate 
could be read from a undocumented register in that darkness :-)

The real reason I started to dig into this is that after you corrected 
my assumption about how to setup the R-Car CSI-2 receiver link freq this 
change breaks capture of CVBS for me. Looking at how I on your 
suggestion calculate link speed.

link_freq = (pixel_rate * bits_per_sample) / (2 * nr_of_lanes)

pixel_rate = 28636360 (from the adv748x V4L2_CID_PIXEL_RATE)
bits_per_sample = 16 (adv748x reports MEDIA_BUS_FMT_UYVY8_2X8)
nr_of_lanes = 1 (Using TXB of the adv748x)

Gives a link_freq of 229Mhz or as the R-Car CSI-2 deals with in Mbps 
~460Mbps as CSI-2 is DDR. If I try to capture CVBS using such a high 
link speed capture fail. I tried using the lowest link speed R-Car CSI-2 
receiver supports of 80Mbps and then capture works perfectly. But 
settings as high as 235Mps still manage to capture.

I'm not sure how to proceed here and have the adv748x and rcar-csi2 to 
agree on the link speed. But I do think this is wrong.

The change for the adv748x HDMI pixel rate is tested and works for both

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-04-24 Thread Niklas Söderlund

Hi Laurent,

Thanks for your patch.

On 2018-04-21 15:44:44 +0300, Laurent Pinchart wrote:
> The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
> include both horizontal and vertical blanking. Both the AFE and HDMI
> receiver program it incorrectly:
> 
> - The HDMI receiver goes to the trouble of removing blanking to compute
> the rate of active pixels. This is easy to fix by removing the
> computation and returning the incoming pixel clock rate directly.
> 
> - The AFE performs similar calculation, while it should simply return
> the fixed pixel rate for analog sources, mandated by the ADV748x to be
> 28.63636 MHz.
> 
> Signed-off-by: Laurent Pinchart 

Reviewed-by: Niklas Söderlund 

This patch uncovered a calculation error in rcar-csi2 which compensated 
for the removing of the blanking in the adv748x, thanks for that! Good 
thing that it's not merged yet, will include the fix in the next version 
of the CSI-2 driver.

> ---
>  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
>  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
>  2 files changed, 6 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c 
> b/drivers/media/i2c/adv748x/adv748x-afe.c
> index 61514bae7e5c..3e18d5ae813b 100644
> --- a/drivers/media/i2c/adv748x/adv748x-afe.c
> +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
> @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops 
> adv748x_afe_video_ops = {
>  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
>  {
>   struct v4l2_subdev *tx;
> - unsigned int width, height, fps;
>  
>   tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
>   if (!tx)
>   return -ENOLINK;
>  
> - width = 720;
> - height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
> - fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
> -
> - return adv748x_csi2_set_pixelrate(tx, width * height * fps);
> + /*
> +  * The ADV748x samples analog video signals using an externally supplied
> +  * clock whose frequency is required to be 28.63636 MHz.
> +  */
> + return adv748x_csi2_set_pixelrate(tx, 28636360);
>  }
>  
>  static int adv748x_afe_enum_mbus_code(struct v4l2_subdev *sd,
> diff --git a/drivers/media/i2c/adv748x/adv748x-hdmi.c 
> b/drivers/media/i2c/adv748x/adv748x-hdmi.c
> index 10d229a4f088..aecc2a84dfec 100644
> --- a/drivers/media/i2c/adv748x/adv748x-hdmi.c
> +++ b/drivers/media/i2c/adv748x/adv748x-hdmi.c
> @@ -402,8 +402,6 @@ static int adv748x_hdmi_propagate_pixelrate(struct 
> adv748x_hdmi *hdmi)
>  {
>   struct v4l2_subdev *tx;
>   struct v4l2_dv_timings timings;
> - struct v4l2_bt_timings *bt = 
> - unsigned int fps;
>  
>   tx = adv748x_get_remote_sd(>pads[ADV748X_HDMI_SOURCE]);
>   if (!tx)
> @@ -411,11 +409,7 @@ static int adv748x_hdmi_propagate_pixelrate(struct 
> adv748x_hdmi *hdmi)
>  
>   adv748x_hdmi_query_dv_timings(>sd, );
>  
> - fps = DIV_ROUND_CLOSEST_ULL(bt->pixelclock,
> - V4L2_DV_BT_FRAME_WIDTH(bt) *
> - V4L2_DV_BT_FRAME_HEIGHT(bt));
> -
> - return adv748x_csi2_set_pixelrate(tx, bt->width * bt->height * fps);
> + return adv748x_csi2_set_pixelrate(tx, timings.bt.pixelclock);
>  }
>  
>  static int adv748x_hdmi_enum_mbus_code(struct v4l2_subdev *sd,
> -- 
> Regards,
> 
> Laurent Pinchart
> 

-- 
Regards,
Niklas Söderlund

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-04-23 Thread Kieran Bingham

Hi Laurent,

On 23/04/18 10:08, Laurent Pinchart wrote:
> Hi Kieran,
> 
> On Monday, 23 April 2018 11:59:04 EEST Kieran Bingham wrote:
>> On 21/04/18 13:44, Laurent Pinchart wrote:
>>> The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
>>> include both horizontal and vertical blanking. Both the AFE and HDMI
>>> receiver program it incorrectly:
>>>
>>> - The HDMI receiver goes to the trouble of removing blanking to compute
>>> the rate of active pixels. This is easy to fix by removing the
>>> computation and returning the incoming pixel clock rate directly.
>>>
>>> - The AFE performs similar calculation, while it should simply return
>>> the fixed pixel rate for analog sources, mandated by the ADV748x to be
>>> 28.63636 MHz.
>>>
>>> Signed-off-by: Laurent Pinchart
>>> 
>>
>> This looks quite reasonable, and simplifies the code. Win win.
>>
>> I presume this will have implications on the pixel receiver side (VIN in our
>> case)... are there changes required there, or was it 'just wrong' here.
> 
> No change should be required on the VIN side. This patch was prompted by a 
> BSP 
> patch for VIN that aimed at fixing the same issue, but in the wrong location. 
> See message 2461975.rTQ0tvdgNJ@avalon in the "[periperi] 2018-04-05 
> Multimedia 
> group chat report" thread (Date: Sat, 21 Apr 2018 15:49:49 +0300).
> 
>> Either way,
>>
>> Reviewed-by: Kieran Bingham 
> 
> As you maintain the adv748x driver, could you make sure this patch gets 
> upstream ? :-)

Yes, that's fine.
I'll collect the patch and test it the next time I spin up a Salvator-XS.

Regards

Kieran


>>> ---
>>>
>>>  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
>>>  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
>>>  2 files changed, 6 insertions(+), 13 deletions(-)
>>>
>>> diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c
>>> b/drivers/media/i2c/adv748x/adv748x-afe.c index
>>> 61514bae7e5c..3e18d5ae813b 100644
>>> --- a/drivers/media/i2c/adv748x/adv748x-afe.c
>>> +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
>>> @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops
>>> adv748x_afe_video_ops = {
>>>  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
>>>  {
>>> struct v4l2_subdev *tx;
>>> -   unsigned int width, height, fps;
>>>
>>> tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
>>> if (!tx)
>>> return -ENOLINK;
>>>
>>> -   width = 720;
>>> -   height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
>>> -   fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
>>> -
>>> -   return adv748x_csi2_set_pixelrate(tx, width * height * fps);
>>> +   /*
>>> +* The ADV748x samples analog video signals using an externally 
> supplied
>>> +* clock whose frequency is required to be 28.63636 MHz.
>>> +*/
>>> +   return adv748x_csi2_set_pixelrate(tx, 28636360);
>>>  }
>>>  
>>>  static int adv748x_afe_enum_mbus_code(struct v4l2_subdev *sd,
>>> diff --git a/drivers/media/i2c/adv748x/adv748x-hdmi.c
>>> b/drivers/media/i2c/adv748x/adv748x-hdmi.c index
>>> 10d229a4f088..aecc2a84dfec 100644
>>> --- a/drivers/media/i2c/adv748x/adv748x-hdmi.c
>>> +++ b/drivers/media/i2c/adv748x/adv748x-hdmi.c
>>> @@ -402,8 +402,6 @@ static int adv748x_hdmi_propagate_pixelrate(struct
>>> adv748x_hdmi *hdmi)> 
>>>  {
>>> struct v4l2_subdev *tx;
>>> struct v4l2_dv_timings timings;
>>> -   struct v4l2_bt_timings *bt = 
>>> -   unsigned int fps;
>>>
>>> tx = adv748x_get_remote_sd(>pads[ADV748X_HDMI_SOURCE]);
>>> if (!tx)
>>> @@ -411,11 +409,7 @@ static int adv748x_hdmi_propagate_pixelrate(struct
>>> adv748x_hdmi *hdmi)
>>> adv748x_hdmi_query_dv_timings(>sd, );
>>>
>>> -   fps = DIV_ROUND_CLOSEST_ULL(bt->pixelclock,
>>> -   V4L2_DV_BT_FRAME_WIDTH(bt) *
>>> -   V4L2_DV_BT_FRAME_HEIGHT(bt));
>>> -
>>> -   return adv748x_csi2_set_pixelrate(tx, bt->width * bt->height * fps);
>>> +   return adv748x_csi2_set_pixelrate(tx, timings.bt.pixelclock);
>>>  }
>>>  
>>>  static int adv748x_hdmi_enum_mbus_code(struct v4l2_subdev *sd,
>

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-04-23 Thread Laurent Pinchart

Hi Kieran,

On Monday, 23 April 2018 11:59:04 EEST Kieran Bingham wrote:
> On 21/04/18 13:44, Laurent Pinchart wrote:
> > The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
> > include both horizontal and vertical blanking. Both the AFE and HDMI
> > receiver program it incorrectly:
> > 
> > - The HDMI receiver goes to the trouble of removing blanking to compute
> > the rate of active pixels. This is easy to fix by removing the
> > computation and returning the incoming pixel clock rate directly.
> > 
> > - The AFE performs similar calculation, while it should simply return
> > the fixed pixel rate for analog sources, mandated by the ADV748x to be
> > 28.63636 MHz.
> > 
> > Signed-off-by: Laurent Pinchart
> > 
> 
> This looks quite reasonable, and simplifies the code. Win win.
> 
> I presume this will have implications on the pixel receiver side (VIN in our
> case)... are there changes required there, or was it 'just wrong' here.

No change should be required on the VIN side. This patch was prompted by a BSP 
patch for VIN that aimed at fixing the same issue, but in the wrong location. 
See message 2461975.rTQ0tvdgNJ@avalon in the "[periperi] 2018-04-05 Multimedia 
group chat report" thread (Date: Sat, 21 Apr 2018 15:49:49 +0300).

> Either way,
> 
> Reviewed-by: Kieran Bingham 

As you maintain the adv748x driver, could you make sure this patch gets 
upstream ? :-)

> > ---
> > 
> >  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
> >  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
> >  2 files changed, 6 insertions(+), 13 deletions(-)
> > 
> > diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c
> > b/drivers/media/i2c/adv748x/adv748x-afe.c index
> > 61514bae7e5c..3e18d5ae813b 100644
> > --- a/drivers/media/i2c/adv748x/adv748x-afe.c
> > +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
> > @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops
> > adv748x_afe_video_ops = {
> >  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
> >  {
> > struct v4l2_subdev *tx;
> > -   unsigned int width, height, fps;
> > 
> > tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
> > if (!tx)
> > return -ENOLINK;
> > 
> > -   width = 720;
> > -   height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
> > -   fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
> > -
> > -   return adv748x_csi2_set_pixelrate(tx, width * height * fps);
> > +   /*
> > +* The ADV748x samples analog video signals using an externally 
supplied
> > +* clock whose frequency is required to be 28.63636 MHz.
> > +*/
> > +   return adv748x_csi2_set_pixelrate(tx, 28636360);
> >  }
> >  
> >  static int adv748x_afe_enum_mbus_code(struct v4l2_subdev *sd,
> > diff --git a/drivers/media/i2c/adv748x/adv748x-hdmi.c
> > b/drivers/media/i2c/adv748x/adv748x-hdmi.c index
> > 10d229a4f088..aecc2a84dfec 100644
> > --- a/drivers/media/i2c/adv748x/adv748x-hdmi.c
> > +++ b/drivers/media/i2c/adv748x/adv748x-hdmi.c
> > @@ -402,8 +402,6 @@ static int adv748x_hdmi_propagate_pixelrate(struct
> > adv748x_hdmi *hdmi)> 
> >  {
> > struct v4l2_subdev *tx;
> > struct v4l2_dv_timings timings;
> > -   struct v4l2_bt_timings *bt = 
> > -   unsigned int fps;
> > 
> > tx = adv748x_get_remote_sd(>pads[ADV748X_HDMI_SOURCE]);
> > if (!tx)
> > @@ -411,11 +409,7 @@ static int adv748x_hdmi_propagate_pixelrate(struct
> > adv748x_hdmi *hdmi)
> > adv748x_hdmi_query_dv_timings(>sd, );
> > 
> > -   fps = DIV_ROUND_CLOSEST_ULL(bt->pixelclock,
> > -   V4L2_DV_BT_FRAME_WIDTH(bt) *
> > -   V4L2_DV_BT_FRAME_HEIGHT(bt));
> > -
> > -   return adv748x_csi2_set_pixelrate(tx, bt->width * bt->height * fps);
> > +   return adv748x_csi2_set_pixelrate(tx, timings.bt.pixelclock);
> >  }
> >  
> >  static int adv748x_hdmi_enum_mbus_code(struct v4l2_subdev *sd,

-- 
Regards,

Laurent Pinchart

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

2018-04-23 Thread Kieran Bingham

Hi Laurent,

On 21/04/18 13:44, Laurent Pinchart wrote:
> The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
> include both horizontal and vertical blanking. Both the AFE and HDMI
> receiver program it incorrectly:
> 
> - The HDMI receiver goes to the trouble of removing blanking to compute
> the rate of active pixels. This is easy to fix by removing the
> computation and returning the incoming pixel clock rate directly.
> 
> - The AFE performs similar calculation, while it should simply return
> the fixed pixel rate for analog sources, mandated by the ADV748x to be
> 28.63636 MHz.
> 
> Signed-off-by: Laurent Pinchart 

This looks quite reasonable, and simplifies the code. Win win.

I presume this will have implications on the pixel receiver side (VIN in our
case)... are there changes required there, or was it 'just wrong' here.


Either way,

Reviewed-by: Kieran Bingham 



> ---
>  drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
>  drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
>  2 files changed, 6 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c 
> b/drivers/media/i2c/adv748x/adv748x-afe.c
> index 61514bae7e5c..3e18d5ae813b 100644
> --- a/drivers/media/i2c/adv748x/adv748x-afe.c
> +++ b/drivers/media/i2c/adv748x/adv748x-afe.c
> @@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops 
> adv748x_afe_video_ops = {
>  static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
>  {
>   struct v4l2_subdev *tx;
> - unsigned int width, height, fps;
>  
>   tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
>   if (!tx)
>   return -ENOLINK;
>  
> - width = 720;
> - height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
> - fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
> -
> - return adv748x_csi2_set_pixelrate(tx, width * height * fps);
> + /*
> +  * The ADV748x samples analog video signals using an externally supplied
> +  * clock whose frequency is required to be 28.63636 MHz.
> +  */
> + return adv748x_csi2_set_pixelrate(tx, 28636360);
>  }
>  
>  static int adv748x_afe_enum_mbus_code(struct v4l2_subdev *sd,
> diff --git a/drivers/media/i2c/adv748x/adv748x-hdmi.c 
> b/drivers/media/i2c/adv748x/adv748x-hdmi.c
> index 10d229a4f088..aecc2a84dfec 100644
> --- a/drivers/media/i2c/adv748x/adv748x-hdmi.c
> +++ b/drivers/media/i2c/adv748x/adv748x-hdmi.c
> @@ -402,8 +402,6 @@ static int adv748x_hdmi_propagate_pixelrate(struct 
> adv748x_hdmi *hdmi)
>  {
>   struct v4l2_subdev *tx;
>   struct v4l2_dv_timings timings;
> - struct v4l2_bt_timings *bt = 
> - unsigned int fps;
>  
>   tx = adv748x_get_remote_sd(>pads[ADV748X_HDMI_SOURCE]);
>   if (!tx)
> @@ -411,11 +409,7 @@ static int adv748x_hdmi_propagate_pixelrate(struct 
> adv748x_hdmi *hdmi)
>  
>   adv748x_hdmi_query_dv_timings(>sd, );
>  
> - fps = DIV_ROUND_CLOSEST_ULL(bt->pixelclock,
> - V4L2_DV_BT_FRAME_WIDTH(bt) *
> - V4L2_DV_BT_FRAME_HEIGHT(bt));
> -
> - return adv748x_csi2_set_pixelrate(tx, bt->width * bt->height * fps);
> + return adv748x_csi2_set_pixelrate(tx, timings.bt.pixelclock);
>  }
>  
>  static int adv748x_hdmi_enum_mbus_code(struct v4l2_subdev *sd,
>

[PATCH] media: i2c: adv748x: Fix pixel rate values

2018-04-21 Thread Laurent Pinchart

The pixel rate, as reported by the V4L2_CID_PIXEL_RATE control, must
include both horizontal and vertical blanking. Both the AFE and HDMI
receiver program it incorrectly:

- The HDMI receiver goes to the trouble of removing blanking to compute
the rate of active pixels. This is easy to fix by removing the
computation and returning the incoming pixel clock rate directly.

- The AFE performs similar calculation, while it should simply return
the fixed pixel rate for analog sources, mandated by the ADV748x to be
28.63636 MHz.

Signed-off-by: Laurent Pinchart 
---
 drivers/media/i2c/adv748x/adv748x-afe.c  | 11 +--
 drivers/media/i2c/adv748x/adv748x-hdmi.c |  8 +---
 2 files changed, 6 insertions(+), 13 deletions(-)

diff --git a/drivers/media/i2c/adv748x/adv748x-afe.c 
b/drivers/media/i2c/adv748x/adv748x-afe.c
index 61514bae7e5c..3e18d5ae813b 100644
--- a/drivers/media/i2c/adv748x/adv748x-afe.c
+++ b/drivers/media/i2c/adv748x/adv748x-afe.c
@@ -321,17 +321,16 @@ static const struct v4l2_subdev_video_ops 
adv748x_afe_video_ops = {
 static int adv748x_afe_propagate_pixelrate(struct adv748x_afe *afe)
 {
struct v4l2_subdev *tx;
-   unsigned int width, height, fps;
 
tx = adv748x_get_remote_sd(>pads[ADV748X_AFE_SOURCE]);
if (!tx)
return -ENOLINK;
 
-   width = 720;
-   height = afe->curr_norm & V4L2_STD_525_60 ? 480 : 576;
-   fps = afe->curr_norm & V4L2_STD_525_60 ? 30 : 25;
-
-   return adv748x_csi2_set_pixelrate(tx, width * height * fps);
+   /*
+* The ADV748x samples analog video signals using an externally supplied
+* clock whose frequency is required to be 28.63636 MHz.
+*/
+   return adv748x_csi2_set_pixelrate(tx, 28636360);
 }
 
 static int adv748x_afe_enum_mbus_code(struct v4l2_subdev *sd,
diff --git a/drivers/media/i2c/adv748x/adv748x-hdmi.c 
b/drivers/media/i2c/adv748x/adv748x-hdmi.c
index 10d229a4f088..aecc2a84dfec 100644
--- a/drivers/media/i2c/adv748x/adv748x-hdmi.c
+++ b/drivers/media/i2c/adv748x/adv748x-hdmi.c
@@ -402,8 +402,6 @@ static int adv748x_hdmi_propagate_pixelrate(struct 
adv748x_hdmi *hdmi)
 {
struct v4l2_subdev *tx;
struct v4l2_dv_timings timings;
-   struct v4l2_bt_timings *bt = 
-   unsigned int fps;
 
tx = adv748x_get_remote_sd(>pads[ADV748X_HDMI_SOURCE]);
if (!tx)
@@ -411,11 +409,7 @@ static int adv748x_hdmi_propagate_pixelrate(struct 
adv748x_hdmi *hdmi)
 
adv748x_hdmi_query_dv_timings(>sd, );
 
-   fps = DIV_ROUND_CLOSEST_ULL(bt->pixelclock,
-   V4L2_DV_BT_FRAME_WIDTH(bt) *
-   V4L2_DV_BT_FRAME_HEIGHT(bt));
-
-   return adv748x_csi2_set_pixelrate(tx, bt->width * bt->height * fps);
+   return adv748x_csi2_set_pixelrate(tx, timings.bt.pixelclock);
 }
 
 static int adv748x_hdmi_enum_mbus_code(struct v4l2_subdev *sd,
-- 
Regards,

Laurent Pinchart

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

Re: [PATCH] media: i2c: adv748x: Fix pixel rate values

[PATCH] media: i2c: adv748x: Fix pixel rate values

8 matches

Site Navigation

Mail list logo

Footer information