[PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Nasser Afshin]

According to the datasheet, INTREQ/GPCL/VBLK should have a pull-up/down
resistor if it's been disabled. On hardware that does not have such
resistor, we should use the default output enable value.
This prevents the color burst lock instability problem.

Signed-off-by: Nasser Afshin <Afshin.Nasser@gmail.com>
---
 drivers/media/i2c/tvp5150.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/media/i2c/tvp5150.c b/drivers/media/i2c/tvp5150.c
index 2476d812f669..0e9713814816 100644
--- a/drivers/media/i2c/tvp5150.c
+++ b/drivers/media/i2c/tvp5150.c
@@ -328,7 +328,7 @@ static const struct i2c_reg_value tvp5150_init_default[] = {
 		TVP5150_OP_MODE_CTL,0x00
 	},
 	{ /* 0x03 */
-		TVP5150_MISC_CTL,0x01
+		TVP5150_MISC_CTL,0x21
 	},
 	{ /* 0x06 */
 		TVP5150_COLOR_KIL_THSH_CTL,0x10
@@ -1072,7 +1072,8 @@ static int tvp5150_s_stream(struct v4l2_subdev *sd, int enable)
 		 * Enable the YCbCr and clock outputs. In discrete sync mode
 		 * (non-BT.656) additionally enable the the sync outputs.
 		 */
-		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE;
+		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE |
+			TVP5150_MISC_CTL_INTREQ_OE;
 		if (decoder->mbus_type == V4L2_MBUS_PARALLEL)
 			val |= TVP5150_MISC_CTL_SYNC_OE;
 	}
-- 
2.15.0

Re: [PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Mauro Carvalho Chehab]

Hi Nasser,

Em Mon, 26 Mar 2018 03:26:33 +0430
Nasser Afshin <afshin.nasser@gmail.com> escreveu:

> According to the datasheet, INTREQ/GPCL/VBLK should have a pull-up/down
> resistor if it's been disabled. On hardware that does not have such
> resistor, we should use the default output enable value.
> This prevents the color burst lock instability problem.

If this is hardware-dependent, you should instead store it at
OF (for SoC) or pass via platform_data (for PCI/USB devices).

> 
> Signed-off-by: Nasser Afshin <Afshin.Nasser@gmail.com>
> ---
>  drivers/media/i2c/tvp5150.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/media/i2c/tvp5150.c b/drivers/media/i2c/tvp5150.c
> index 2476d812f669..0e9713814816 100644
> --- a/drivers/media/i2c/tvp5150.c
> +++ b/drivers/media/i2c/tvp5150.c
> @@ -328,7 +328,7 @@ static const struct i2c_reg_value tvp5150_init_default[] = {
>  		TVP5150_OP_MODE_CTL,0x00
>  	},
>  	{ /* 0x03 */
> -		TVP5150_MISC_CTL,0x01
> +		TVP5150_MISC_CTL,0x21
>  	},
>  	{ /* 0x06 */
>  		TVP5150_COLOR_KIL_THSH_CTL,0x10
> @@ -1072,7 +1072,8 @@ static int tvp5150_s_stream(struct v4l2_subdev *sd, int enable)
>  		 * Enable the YCbCr and clock outputs. In discrete sync mode
>  		 * (non-BT.656) additionally enable the the sync outputs.
>  		 */
> -		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE;
> +		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE |
> +			TVP5150_MISC_CTL_INTREQ_OE;
>  		if (decoder->mbus_type == V4L2_MBUS_PARALLEL)
>  			val |= TVP5150_MISC_CTL_SYNC_OE;
>  	}



Thanks,
Mauro

Re: [PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Nasser]

On Mon, Mar 26, 2018 at 06:43:53AM -0300, Mauro Carvalho Chehab wrote:
> Hi Nasser,
> 
> Em Mon, 26 Mar 2018 03:26:33 +0430
> Nasser Afshin <afshin.nasser@gmail.com> escreveu:
> 
> > According to the datasheet, INTREQ/GPCL/VBLK should have a pull-up/down
> > resistor if it's been disabled. On hardware that does not have such
> > resistor, we should use the default output enable value.
> > This prevents the color burst lock instability problem.
>

Color burst lock instability is just a side effect of not using the
recommended value for this bit. If we use the recommended setting, we
will support more hardware while not breaking anything.

> If this is hardware-dependent, you should instead store it at
> OF (for SoC) or pass via platform_data (for PCI/USB devices).
>

We have used the recommended value for this bit (as the datasheet
suggests) while we are in tvp5150_init_enable but in tvp5150_s_stream
we are using the wrong value.

Also we have this comment at line 319:
    /* Default values as sugested at TVP5150AM1 datasheet */
But as you see, TVP5150_MISC_CTL is not set to its suggested default
value.
 
> > 
> > Signed-off-by: Nasser Afshin <Afshin.Nasser@gmail.com>
> > ---
> >  drivers/media/i2c/tvp5150.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/media/i2c/tvp5150.c b/drivers/media/i2c/tvp5150.c
> > index 2476d812f669..0e9713814816 100644
> > --- a/drivers/media/i2c/tvp5150.c
> > +++ b/drivers/media/i2c/tvp5150.c
> > @@ -328,7 +328,7 @@ static const struct i2c_reg_value tvp5150_init_default[] = {
> >  		TVP5150_OP_MODE_CTL,0x00
> >  	},
> >  	{ /* 0x03 */
> > -		TVP5150_MISC_CTL,0x01
> > +		TVP5150_MISC_CTL,0x21
> >  	},
> >  	{ /* 0x06 */
> >  		TVP5150_COLOR_KIL_THSH_CTL,0x10
> > @@ -1072,7 +1072,8 @@ static int tvp5150_s_stream(struct v4l2_subdev *sd, int enable)
> >  		 * Enable the YCbCr and clock outputs. In discrete sync mode
> >  		 * (non-BT.656) additionally enable the the sync outputs.
> >  		 */
> > -		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE;
> > +		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE |
> > +			TVP5150_MISC_CTL_INTREQ_OE;
> >  		if (decoder->mbus_type == V4L2_MBUS_PARALLEL)
> >  			val |= TVP5150_MISC_CTL_SYNC_OE;
> >  	}
> 
> 
> 
> Thanks,
> Mauro

Re: [PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Nasser]

On Tue, Mar 27, 2018 at 02:59:21AM +0430, Nasser wrote:
Hi Mauro,

Thank you for taking time to review my patch.

May be I should rephrase the commit message to something like:
	Use the default register values as suggested in TVP5150AM1 datasheet

As this is not a hardware-dependent issue. Am I missing something?

> On Mon, Mar 26, 2018 at 06:43:53AM -0300, Mauro Carvalho Chehab wrote:
> > Hi Nasser,
> > 
> > Em Mon, 26 Mar 2018 03:26:33 +0430
> > Nasser Afshin <afshin.nasser@gmail.com> escreveu:
> > 
> > > According to the datasheet, INTREQ/GPCL/VBLK should have a pull-up/down
> > > resistor if it's been disabled. On hardware that does not have such
> > > resistor, we should use the default output enable value.
> > > This prevents the color burst lock instability problem.
> >
> 
> Color burst lock instability is just a side effect of not using the
> recommended value for this bit. If we use the recommended setting, we
> will support more hardware while not breaking anything.
> 
> > If this is hardware-dependent, you should instead store it at
> > OF (for SoC) or pass via platform_data (for PCI/USB devices).
> >
> 
> We have used the recommended value for this bit (as the datasheet
> suggests) while we are in tvp5150_init_enable but in tvp5150_s_stream
> we are using the wrong value.
> 
> Also we have this comment at line 319:
>     /* Default values as sugested at TVP5150AM1 datasheet */
> But as you see, TVP5150_MISC_CTL is not set to its suggested default
> value.
>  
> > > 
> > > Signed-off-by: Nasser Afshin <Afshin.Nasser@gmail.com>
> > > ---
> > >  drivers/media/i2c/tvp5150.c | 5 +++--
> > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/media/i2c/tvp5150.c b/drivers/media/i2c/tvp5150.c
> > > index 2476d812f669..0e9713814816 100644
> > > --- a/drivers/media/i2c/tvp5150.c
> > > +++ b/drivers/media/i2c/tvp5150.c
> > > @@ -328,7 +328,7 @@ static const struct i2c_reg_value tvp5150_init_default[] = {
> > >  		TVP5150_OP_MODE_CTL,0x00
> > >  	},
> > >  	{ /* 0x03 */
> > > -		TVP5150_MISC_CTL,0x01
> > > +		TVP5150_MISC_CTL,0x21
> > >  	},
> > >  	{ /* 0x06 */
> > >  		TVP5150_COLOR_KIL_THSH_CTL,0x10
> > > @@ -1072,7 +1072,8 @@ static int tvp5150_s_stream(struct v4l2_subdev *sd, int enable)
> > >  		 * Enable the YCbCr and clock outputs. In discrete sync mode
> > >  		 * (non-BT.656) additionally enable the the sync outputs.
> > >  		 */
> > > -		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE;
> > > +		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE |
> > > +			TVP5150_MISC_CTL_INTREQ_OE;
> > >  		if (decoder->mbus_type == V4L2_MBUS_PARALLEL)
> > >  			val |= TVP5150_MISC_CTL_SYNC_OE;
> > >  	}
> > 
> > 
> > 
> > Thanks,
> > Mauro

Thanks,
Nasser

Re: [PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Mauro Carvalho Chehab]

Em Thu, 29 Mar 2018 19:04:35 +0430
Nasser <afshin.nasser@gmail.com> escreveu:

> On Tue, Mar 27, 2018 at 02:59:21AM +0430, Nasser wrote:
> Hi Mauro,
> 
> Thank you for taking time to review my patch.
> 
> May be I should rephrase the commit message to something like:
> 	Use the default register values as suggested in TVP5150AM1 datasheet
> 
> As this is not a hardware-dependent issue. Am I missing something?

It is not a matter of rephasing, but, instead, to be sure that it won't
cause regressions on existing hardware.

Yet, it would worth if you could describe at the patch what hardware
did you test it, and if VBI was tested too.

Anyway, I'll try to find some time to run some tests on the hardware
I have with tvp5150 too.

Regards,
Mauro

> 
> > On Mon, Mar 26, 2018 at 06:43:53AM -0300, Mauro Carvalho Chehab wrote:  
> > > Hi Nasser,
> > > 
> > > Em Mon, 26 Mar 2018 03:26:33 +0430
> > > Nasser Afshin <afshin.nasser@gmail.com> escreveu:
> > >   
> > > > According to the datasheet, INTREQ/GPCL/VBLK should have a pull-up/down
> > > > resistor if it's been disabled. On hardware that does not have such
> > > > resistor, we should use the default output enable value.
> > > > This prevents the color burst lock instability problem.  
> > >  
> > 
> > Color burst lock instability is just a side effect of not using the
> > recommended value for this bit. If we use the recommended setting, we
> > will support more hardware while not breaking anything.
> >   
> > > If this is hardware-dependent, you should instead store it at
> > > OF (for SoC) or pass via platform_data (for PCI/USB devices).
> > >  
> > 
> > We have used the recommended value for this bit (as the datasheet
> > suggests) while we are in tvp5150_init_enable but in tvp5150_s_stream
> > we are using the wrong value.
> > 
> > Also we have this comment at line 319:
> >     /* Default values as sugested at TVP5150AM1 datasheet */
> > But as you see, TVP5150_MISC_CTL is not set to its suggested default
> > value.
> >    
> > > > 
> > > > Signed-off-by: Nasser Afshin <Afshin.Nasser@gmail.com>
> > > > ---
> > > >  drivers/media/i2c/tvp5150.c | 5 +++--
> > > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > > 
> > > > diff --git a/drivers/media/i2c/tvp5150.c b/drivers/media/i2c/tvp5150.c
> > > > index 2476d812f669..0e9713814816 100644
> > > > --- a/drivers/media/i2c/tvp5150.c
> > > > +++ b/drivers/media/i2c/tvp5150.c
> > > > @@ -328,7 +328,7 @@ static const struct i2c_reg_value tvp5150_init_default[] = {
> > > >  		TVP5150_OP_MODE_CTL,0x00
> > > >  	},
> > > >  	{ /* 0x03 */
> > > > -		TVP5150_MISC_CTL,0x01
> > > > +		TVP5150_MISC_CTL,0x21
> > > >  	},
> > > >  	{ /* 0x06 */
> > > >  		TVP5150_COLOR_KIL_THSH_CTL,0x10
> > > > @@ -1072,7 +1072,8 @@ static int tvp5150_s_stream(struct v4l2_subdev *sd, int enable)
> > > >  		 * Enable the YCbCr and clock outputs. In discrete sync mode
> > > >  		 * (non-BT.656) additionally enable the the sync outputs.
> > > >  		 */
> > > > -		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE;
> > > > +		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE |
> > > > +			TVP5150_MISC_CTL_INTREQ_OE;
> > > >  		if (decoder->mbus_type == V4L2_MBUS_PARALLEL)
> > > >  			val |= TVP5150_MISC_CTL_SYNC_OE;
> > > >  	}  
> > > 
> > > 
> > > 
> > > Thanks,
> > > Mauro  
> 
> Thanks,
> Nasser



Thanks,
Mauro

Re: [PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Nasser]

On Thu, Mar 29, 2018 at 12:02:40PM -0300, Mauro Carvalho Chehab wrote:
> Em Thu, 29 Mar 2018 19:04:35 +0430
> Nasser <afshin.nasser@gmail.com> escreveu:
> 
> > On Tue, Mar 27, 2018 at 02:59:21AM +0430, Nasser wrote:
> > Hi Mauro,
> > 
> > Thank you for taking time to review my patch.
> > 
> > May be I should rephrase the commit message to something like:
> > 	Use the default register values as suggested in TVP5150AM1 datasheet
> > 
> > As this is not a hardware-dependent issue. Am I missing something?
> 
> It is not a matter of rephasing, but, instead, to be sure that it won't
> cause regressions on existing hardware.
> 
> Yet, it would worth if you could describe at the patch what hardware
> did you test it, and if VBI was tested too.
> 

Does this means that I should resend the patch with this additional info?
Sorry for not being clear about that. This was a custom board based on
ARM. The VBI was not used.

> Anyway, I'll try to find some time to run some tests on the hardware
> I have with tvp5150 too.

It sounds great.

> 
> Regards,
> Mauro
> 
> > 
> > > On Mon, Mar 26, 2018 at 06:43:53AM -0300, Mauro Carvalho Chehab wrote:  
> > > > Hi Nasser,
> > > > 
> > > > Em Mon, 26 Mar 2018 03:26:33 +0430
> > > > Nasser Afshin <afshin.nasser@gmail.com> escreveu:
> > > >   
> > > > > According to the datasheet, INTREQ/GPCL/VBLK should have a pull-up/down
> > > > > resistor if it's been disabled. On hardware that does not have such
> > > > > resistor, we should use the default output enable value.
> > > > > This prevents the color burst lock instability problem.  
> > > >  
> > > 
> > > Color burst lock instability is just a side effect of not using the
> > > recommended value for this bit. If we use the recommended setting, we
> > > will support more hardware while not breaking anything.
> > >   
> > > > If this is hardware-dependent, you should instead store it at
> > > > OF (for SoC) or pass via platform_data (for PCI/USB devices).
> > > >  
> > > 
> > > We have used the recommended value for this bit (as the datasheet
> > > suggests) while we are in tvp5150_init_enable but in tvp5150_s_stream
> > > we are using the wrong value.
> > > 
> > > Also we have this comment at line 319:
> > >     /* Default values as sugested at TVP5150AM1 datasheet */
> > > But as you see, TVP5150_MISC_CTL is not set to its suggested default
> > > value.
> > >    

Any way the assignment to tvp5150_init_default after the above comment
seems not to be correct according to the "3.21.4 Miscellaneous Controls
Register" part in the datasheet. While following the same comment phrase
on line 455, we see the correct assignment as the default values to 
tvp5150_init_enable.

Sorry this is so lengthy.

Thank you,
Nasser

> > > > > 
> > > > > Signed-off-by: Nasser Afshin <Afshin.Nasser@gmail.com>
> > > > > ---
> > > > >  drivers/media/i2c/tvp5150.c | 5 +++--
> > > > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/media/i2c/tvp5150.c b/drivers/media/i2c/tvp5150.c
> > > > > index 2476d812f669..0e9713814816 100644
> > > > > --- a/drivers/media/i2c/tvp5150.c
> > > > > +++ b/drivers/media/i2c/tvp5150.c
> > > > > @@ -328,7 +328,7 @@ static const struct i2c_reg_value tvp5150_init_default[] = {
> > > > >  		TVP5150_OP_MODE_CTL,0x00
> > > > >  	},
> > > > >  	{ /* 0x03 */
> > > > > -		TVP5150_MISC_CTL,0x01
> > > > > +		TVP5150_MISC_CTL,0x21
> > > > >  	},
> > > > >  	{ /* 0x06 */
> > > > >  		TVP5150_COLOR_KIL_THSH_CTL,0x10
> > > > > @@ -1072,7 +1072,8 @@ static int tvp5150_s_stream(struct v4l2_subdev *sd, int enable)
> > > > >  		 * Enable the YCbCr and clock outputs. In discrete sync mode
> > > > >  		 * (non-BT.656) additionally enable the the sync outputs.
> > > > >  		 */
> > > > > -		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE;
> > > > > +		val |= TVP5150_MISC_CTL_YCBCR_OE | TVP5150_MISC_CTL_CLOCK_OE |
> > > > > +			TVP5150_MISC_CTL_INTREQ_OE;
> > > > >  		if (decoder->mbus_type == V4L2_MBUS_PARALLEL)
> > > > >  			val |= TVP5150_MISC_CTL_SYNC_OE;
> > > > >  	}  
> > > > 
> > > > 
> > > > 
> > > > Thanks,
> > > > Mauro  
> > 
> > Thanks,
> > Nasser
> 
> 
> 
> Thanks,
> Mauro

Re: [PATCH] media: i2c: tvp5150: fix color burst lock instability on some hardware

[Mauro Carvalho Chehab]

Em Thu, 29 Mar 2018 22:37:57 +0430
Nasser <afshin.nasser@gmail.com> escreveu:

> On Thu, Mar 29, 2018 at 12:02:40PM -0300, Mauro Carvalho Chehab wrote:
> > Em Thu, 29 Mar 2018 19:04:35 +0430
> > Nasser <afshin.nasser@gmail.com> escreveu:
> >   
> > > On Tue, Mar 27, 2018 at 02:59:21AM +0430, Nasser wrote:
> > > Hi Mauro,
> > > 
> > > Thank you for taking time to review my patch.
> > > 
> > > May be I should rephrase the commit message to something like:
> > > 	Use the default register values as suggested in TVP5150AM1 datasheet
> > > 
> > > As this is not a hardware-dependent issue. Am I missing something?  
> > 
> > It is not a matter of rephasing, but, instead, to be sure that it won't
> > cause regressions on existing hardware.
> > 
> > Yet, it would worth if you could describe at the patch what hardware
> > did you test it, and if VBI was tested too.
> >   
> 
> Does this means that I should resend the patch with this additional info?
> Sorry for not being clear about that. This was a custom board based on
> ARM. The VBI was not used.

That's what I suspected when you send your patch :-)

The tvp5150 is used by several USB and PCI devices for analog TV and
for video stream capture, with may have different requirements from what
you're doing on your ARM board.

Changing a register setting global wide without enough care will
very likely break support for existing boards.

That's why I said that the right thing to do is to pass a parameter to
the driver specifying what "extra" features are needed.

Without looking at tvp5150/tvp5151/tvp5150am datasheets nor testing,
I'd risk to say that, at the specific case of your patch, touching at
VBLK settings have a potential risk of affecting VBI handling.

> > Anyway, I'll try to find some time to run some tests on the hardware
> > I have with tvp5150 too.  
> 
> It sounds great.

Yes, but I won't be able to do it on the next couple of weeks. We're
approaching the merge window for Kernel 4.17.

Thanks,
Mauro