[Patch v2 0/3] media: ov5640: updates

[Patch v2 0/3] media: ov5640: updates

[Benoit Parrot]

This patch series is a collection of patches we have been carrying for a
while.

First, it adds support for PIXEL_RATE control which is used by some
CSI2 receiver driver to properly set-up their DPHY.

Then we fix an issue related to having extra sensor enable/disable in
the register array for the 1920x1080 mode.

Finally we restrict the largest resolution which should only be
available at the lowest FPS.

Changes since v1:
- Addressed comment from Sakari.
  added a function to calculate the pixel rate and remove the need to
  cache its value

Benoit Parrot (3):
  media: ov5640: add PIXEL_RATE control
  media: ov5640: Fix 1920x1080 mode to remove extra enable/disable
  media: ov5640: Make 2592x1944 mode only available at 15 fps

 drivers/media/i2c/ov5640.c | 33 +++++++++++++++++++++++++++++++--
 1 file changed, 31 insertions(+), 2 deletions(-)

-- 
2.17.1

[Patch v2 1/3] media: ov5640: add PIXEL_RATE control

[Benoit Parrot]

Add v4l2 controls to report the pixel rates of each mode. This is
needed by some CSI2 receiver in order to perform proper DPHY
configuration.

Signed-off-by: Benoit Parrot <bparrot@ti.com>
---
 drivers/media/i2c/ov5640.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
index 500d9bbff10b..5198dc887400 100644
--- a/drivers/media/i2c/ov5640.c
+++ b/drivers/media/i2c/ov5640.c
@@ -193,6 +193,9 @@ struct ov5640_mode_info {
 
 struct ov5640_ctrls {
 	struct v4l2_ctrl_handler handler;
+	struct {
+		struct v4l2_ctrl *pixel_rate;
+	};
 	struct {
 		struct v4l2_ctrl *auto_exp;
 		struct v4l2_ctrl *exposure;
@@ -2194,6 +2197,16 @@ static int ov5640_try_fmt_internal(struct v4l2_subdev *sd,
 	return 0;
 }
 
+static u64 ov5640_calc_pixel_rate(struct ov5640_dev *sensor)
+{
+	u64 rate;
+
+	rate = sensor->current_mode->vtot * sensor->current_mode->htot;
+	rate *= ov5640_framerates[sensor->current_fr];
+
+	return rate;
+}
+
 static int ov5640_set_fmt(struct v4l2_subdev *sd,
 			  struct v4l2_subdev_pad_config *cfg,
 			  struct v4l2_subdev_format *format)
@@ -2233,6 +2246,8 @@ static int ov5640_set_fmt(struct v4l2_subdev *sd,
 	if (mbus_fmt->code != sensor->fmt.code)
 		sensor->pending_fmt_change = true;
 
+	__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
+				 ov5640_calc_pixel_rate(sensor));
 out:
 	mutex_unlock(&sensor->lock);
 	return ret;
@@ -2657,6 +2672,13 @@ static int ov5640_init_controls(struct ov5640_dev *sensor)
 	/* we can use our own mutex for the ctrl lock */
 	hdl->lock = &sensor->lock;
 
+	/* Clock related controls */
+	ctrls->pixel_rate =
+		v4l2_ctrl_new_std(hdl, ops,
+				  V4L2_CID_PIXEL_RATE, 0, INT_MAX, 1,
+				  ov5640_calc_pixel_rate(sensor));
+	ctrls->pixel_rate->flags |= V4L2_CTRL_FLAG_READ_ONLY;
+
 	/* Auto/manual white balance */
 	ctrls->auto_wb = v4l2_ctrl_new_std(hdl, ops,
 					   V4L2_CID_AUTO_WHITE_BALANCE,
@@ -2816,6 +2838,9 @@ static int ov5640_s_frame_interval(struct v4l2_subdev *sd,
 		sensor->frame_interval = fi->interval;
 		sensor->current_mode = mode;
 		sensor->pending_mode_change = true;
+
+		__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
+					 ov5640_calc_pixel_rate(sensor));
 	}
 out:
 	mutex_unlock(&sensor->lock);
-- 
2.17.1

[Patch v2 2/3] media: ov5640: Fix 1920x1080 mode to remove extra enable/disable

[Benoit Parrot]

In the 1920x1080 register array an extra pair of reset ctrl disable
re-enable was causing unwanted init delays.

Signed-off-by: Benoit Parrot <bparrot@ti.com>
---
 drivers/media/i2c/ov5640.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
index 5198dc887400..103a4e8f88e1 100644
--- a/drivers/media/i2c/ov5640.c
+++ b/drivers/media/i2c/ov5640.c
@@ -492,7 +492,6 @@ static const struct reg_value ov5640_setting_720P_1280_720[] = {
 };
 
 static const struct reg_value ov5640_setting_1080P_1920_1080[] = {
-	{0x3008, 0x42, 0, 0},
 	{0x3c07, 0x08, 0, 0},
 	{0x3c09, 0x1c, 0, 0}, {0x3c0a, 0x9c, 0, 0}, {0x3c0b, 0x40, 0, 0},
 	{0x3814, 0x11, 0, 0},
@@ -520,7 +519,7 @@ static const struct reg_value ov5640_setting_1080P_1920_1080[] = {
 	{0x3a0e, 0x03, 0, 0}, {0x3a0d, 0x04, 0, 0}, {0x3a14, 0x04, 0, 0},
 	{0x3a15, 0x60, 0, 0}, {0x4407, 0x04, 0, 0},
 	{0x460b, 0x37, 0, 0}, {0x460c, 0x20, 0, 0}, {0x3824, 0x04, 0, 0},
-	{0x4005, 0x1a, 0, 0}, {0x3008, 0x02, 0, 0},
+	{0x4005, 0x1a, 0, 0},
 };
 
 static const struct reg_value ov5640_setting_QSXGA_2592_1944[] = {
-- 
2.17.1

[Patch v2 3/3] media: ov5640: Make 2592x1944 mode only available at 15 fps

[Benoit Parrot]

The sensor data sheet clearly state that 2592x1944 only works at 15 fps
make sure we don't try to miss configure the pll out of acceptable
range.

Signed-off-by: Benoit Parrot <bparrot@ti.com>
---
 drivers/media/i2c/ov5640.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
index 103a4e8f88e1..d5b0be2c7a0a 100644
--- a/drivers/media/i2c/ov5640.c
+++ b/drivers/media/i2c/ov5640.c
@@ -1613,6 +1613,11 @@ ov5640_find_mode(struct ov5640_dev *sensor, enum ov5640_frame_rate fr,
 	    !(mode->hact == 640 && mode->vact == 480))
 		return NULL;
 
+	/* 2592x1944 only works at 15fps */
+	if (fr != OV5640_15_FPS &&
+	    (mode->hact == 2592 && mode->vact == 1944))
+		return NULL;
+
 	return mode;
 }
 
-- 
2.17.1

Re: [Patch v2 1/3] media: ov5640: add PIXEL_RATE control

[Jacopo Mondi]

[-- Attachment #1: Type: text/plain, Size: 3218 bytes --]

Hi Benoit,

On Wed, Oct 02, 2019 at 08:51:32AM -0500, Benoit Parrot wrote:
> Add v4l2 controls to report the pixel rates of each mode. This is
> needed by some CSI2 receiver in order to perform proper DPHY
> configuration.
>
> Signed-off-by: Benoit Parrot <bparrot@ti.com>
> ---
>  drivers/media/i2c/ov5640.c | 25 +++++++++++++++++++++++++
>  1 file changed, 25 insertions(+)
>
> diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> index 500d9bbff10b..5198dc887400 100644
> --- a/drivers/media/i2c/ov5640.c
> +++ b/drivers/media/i2c/ov5640.c
> @@ -193,6 +193,9 @@ struct ov5640_mode_info {
>
>  struct ov5640_ctrls {
>  	struct v4l2_ctrl_handler handler;
> +	struct {
> +		struct v4l2_ctrl *pixel_rate;
> +	};

Do you need to wrap this v4l2_ctrl in it's own unnamed struct? Other
controls here declared in this way are clustered and, if I'm not
mistaken, using unnamed struct to wrap them is just a typographically
nice way to convey that. I think your new control could be declared
without a wrapping struct { }.

>  	struct {
>  		struct v4l2_ctrl *auto_exp;
>  		struct v4l2_ctrl *exposure;
> @@ -2194,6 +2197,16 @@ static int ov5640_try_fmt_internal(struct v4l2_subdev *sd,
>  	return 0;
>  }
>
> +static u64 ov5640_calc_pixel_rate(struct ov5640_dev *sensor)
> +{
> +	u64 rate;
> +
> +	rate = sensor->current_mode->vtot * sensor->current_mode->htot;
> +	rate *= ov5640_framerates[sensor->current_fr];
> +
> +	return rate;
> +}
> +

Just to point out this is the -theoretical- pixel rate, and might be
quite different from the one calculated by the clock tree tuning
procedure (which should be updated to match Hugues' latest findings).

>  static int ov5640_set_fmt(struct v4l2_subdev *sd,
>  			  struct v4l2_subdev_pad_config *cfg,
>  			  struct v4l2_subdev_format *format)
> @@ -2233,6 +2246,8 @@ static int ov5640_set_fmt(struct v4l2_subdev *sd,
>  	if (mbus_fmt->code != sensor->fmt.code)
>  		sensor->pending_fmt_change = true;
>
> +	__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> +				 ov5640_calc_pixel_rate(sensor));
>  out:
>  	mutex_unlock(&sensor->lock);
>  	return ret;
> @@ -2657,6 +2672,13 @@ static int ov5640_init_controls(struct ov5640_dev *sensor)
>  	/* we can use our own mutex for the ctrl lock */
>  	hdl->lock = &sensor->lock;
>
> +	/* Clock related controls */
> +	ctrls->pixel_rate =
> +		v4l2_ctrl_new_std(hdl, ops,

If you like it better, this could fit in 1 line

	ctrls->pixel_rate = v4l2_ctrl_new_std(hdl, ops, V4L2_CID_PIXEL_RATE,
					      0, INT_MAX, 1,
					      ov5640_calc_pixel_rate(sensor)

Thanks
   j

> +				  V4L2_CID_PIXEL_RATE, 0, INT_MAX, 1,
> +				  ov5640_calc_pixel_rate(sensor));


> +	ctrls->pixel_rate->flags |= V4L2_CTRL_FLAG_READ_ONLY;
> +
>  	/* Auto/manual white balance */
>  	ctrls->auto_wb = v4l2_ctrl_new_std(hdl, ops,
>  					   V4L2_CID_AUTO_WHITE_BALANCE,
> @@ -2816,6 +2838,9 @@ static int ov5640_s_frame_interval(struct v4l2_subdev *sd,
>  		sensor->frame_interval = fi->interval;
>  		sensor->current_mode = mode;
>  		sensor->pending_mode_change = true;
> +
> +		__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> +					 ov5640_calc_pixel_rate(sensor));
>  	}
>  out:
>  	mutex_unlock(&sensor->lock);
> --
> 2.17.1
>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [Patch v2 1/3] media: ov5640: add PIXEL_RATE control

[Sakari Ailus]

Hi Jacopo, Benoit,

On Thu, Oct 03, 2019 at 09:17:14AM +0200, Jacopo Mondi wrote:
> Hi Benoit,
> 
> On Wed, Oct 02, 2019 at 08:51:32AM -0500, Benoit Parrot wrote:
> > Add v4l2 controls to report the pixel rates of each mode. This is
> > needed by some CSI2 receiver in order to perform proper DPHY
> > configuration.
> >
> > Signed-off-by: Benoit Parrot <bparrot@ti.com>
> > ---
> >  drivers/media/i2c/ov5640.c | 25 +++++++++++++++++++++++++
> >  1 file changed, 25 insertions(+)
> >
> > diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> > index 500d9bbff10b..5198dc887400 100644
> > --- a/drivers/media/i2c/ov5640.c
> > +++ b/drivers/media/i2c/ov5640.c
> > @@ -193,6 +193,9 @@ struct ov5640_mode_info {
> >
> >  struct ov5640_ctrls {
> >  	struct v4l2_ctrl_handler handler;
> > +	struct {
> > +		struct v4l2_ctrl *pixel_rate;
> > +	};
> 
> Do you need to wrap this v4l2_ctrl in it's own unnamed struct? Other
> controls here declared in this way are clustered and, if I'm not
> mistaken, using unnamed struct to wrap them is just a typographically
> nice way to convey that. I think your new control could be declared
> without a wrapping struct { }.
> 
> >  	struct {
> >  		struct v4l2_ctrl *auto_exp;
> >  		struct v4l2_ctrl *exposure;
> > @@ -2194,6 +2197,16 @@ static int ov5640_try_fmt_internal(struct v4l2_subdev *sd,
> >  	return 0;
> >  }
> >
> > +static u64 ov5640_calc_pixel_rate(struct ov5640_dev *sensor)
> > +{
> > +	u64 rate;
> > +
> > +	rate = sensor->current_mode->vtot * sensor->current_mode->htot;
> > +	rate *= ov5640_framerates[sensor->current_fr];
> > +
> > +	return rate;
> > +}
> > +
> 
> Just to point out this is the -theoretical- pixel rate, and might be
> quite different from the one calculated by the clock tree tuning
> procedure (which should be updated to match Hugues' latest findings).

Hmm. Considering the xclk rate may be pretty much anything, I'd suppose
the value above would only be correct for a given xclk rate.

Could this be simply calculated from the clock tree configuration, to get
the right value in all cases?

> 
> >  static int ov5640_set_fmt(struct v4l2_subdev *sd,
> >  			  struct v4l2_subdev_pad_config *cfg,
> >  			  struct v4l2_subdev_format *format)
> > @@ -2233,6 +2246,8 @@ static int ov5640_set_fmt(struct v4l2_subdev *sd,
> >  	if (mbus_fmt->code != sensor->fmt.code)
> >  		sensor->pending_fmt_change = true;
> >
> > +	__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > +				 ov5640_calc_pixel_rate(sensor));
> >  out:
> >  	mutex_unlock(&sensor->lock);
> >  	return ret;
> > @@ -2657,6 +2672,13 @@ static int ov5640_init_controls(struct ov5640_dev *sensor)
> >  	/* we can use our own mutex for the ctrl lock */
> >  	hdl->lock = &sensor->lock;
> >
> > +	/* Clock related controls */
> > +	ctrls->pixel_rate =
> > +		v4l2_ctrl_new_std(hdl, ops,
> 
> If you like it better, this could fit in 1 line
> 
> 	ctrls->pixel_rate = v4l2_ctrl_new_std(hdl, ops, V4L2_CID_PIXEL_RATE,
> 					      0, INT_MAX, 1,
> 					      ov5640_calc_pixel_rate(sensor)
> 
> Thanks
>    j
> 
> > +				  V4L2_CID_PIXEL_RATE, 0, INT_MAX, 1,
> > +				  ov5640_calc_pixel_rate(sensor));
> 
> 
> > +	ctrls->pixel_rate->flags |= V4L2_CTRL_FLAG_READ_ONLY;

Note that ctrls->pixel_rate is NULL if e.g. memory allocation fails when
creating the control.

> > +
> >  	/* Auto/manual white balance */
> >  	ctrls->auto_wb = v4l2_ctrl_new_std(hdl, ops,
> >  					   V4L2_CID_AUTO_WHITE_BALANCE,
> > @@ -2816,6 +2838,9 @@ static int ov5640_s_frame_interval(struct v4l2_subdev *sd,
> >  		sensor->frame_interval = fi->interval;
> >  		sensor->current_mode = mode;
> >  		sensor->pending_mode_change = true;
> > +
> > +		__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > +					 ov5640_calc_pixel_rate(sensor));
> >  	}
> >  out:
> >  	mutex_unlock(&sensor->lock);

-- 
Regards,

Sakari Ailus
sakari.ailus@linux.intel.com

Re: [Patch v2 2/3] media: ov5640: Fix 1920x1080 mode to remove extra enable/disable

[Jacopo Mondi]

[-- Attachment #1: Type: text/plain, Size: 1696 bytes --]

Hi Benoit,
   good catch. I wonder why those power down/up sequences where only
set for this mode...

I also wonder which kind of power down mode do we enter, if the chip
is set in 'software power down mode' with 0x3008=0x42 at the beginning
of the register blob write sequence, and we still can successfully
program registers..

In any case, assuming 720p still works:
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>

Thanks
   j

On Wed, Oct 02, 2019 at 08:51:33AM -0500, Benoit Parrot wrote:
> In the 1920x1080 register array an extra pair of reset ctrl disable
> re-enable was causing unwanted init delays.
>
> Signed-off-by: Benoit Parrot <bparrot@ti.com>
> ---
>  drivers/media/i2c/ov5640.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> index 5198dc887400..103a4e8f88e1 100644
> --- a/drivers/media/i2c/ov5640.c
> +++ b/drivers/media/i2c/ov5640.c
> @@ -492,7 +492,6 @@ static const struct reg_value ov5640_setting_720P_1280_720[] = {
>  };
>
>  static const struct reg_value ov5640_setting_1080P_1920_1080[] = {
> -	{0x3008, 0x42, 0, 0},
>  	{0x3c07, 0x08, 0, 0},
>  	{0x3c09, 0x1c, 0, 0}, {0x3c0a, 0x9c, 0, 0}, {0x3c0b, 0x40, 0, 0},
>  	{0x3814, 0x11, 0, 0},
> @@ -520,7 +519,7 @@ static const struct reg_value ov5640_setting_1080P_1920_1080[] = {
>  	{0x3a0e, 0x03, 0, 0}, {0x3a0d, 0x04, 0, 0}, {0x3a14, 0x04, 0, 0},
>  	{0x3a15, 0x60, 0, 0}, {0x4407, 0x04, 0, 0},
>  	{0x460b, 0x37, 0, 0}, {0x460c, 0x20, 0, 0}, {0x3824, 0x04, 0, 0},
> -	{0x4005, 0x1a, 0, 0}, {0x3008, 0x02, 0, 0},
> +	{0x4005, 0x1a, 0, 0},
>  };
>
>  static const struct reg_value ov5640_setting_QSXGA_2592_1944[] = {
> --
> 2.17.1
>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [Patch v2 3/3] media: ov5640: Make 2592x1944 mode only available at 15 fps

[Jacopo Mondi]

[-- Attachment #1: Type: text/plain, Size: 1516 bytes --]

Hi Benoit,

On Wed, Oct 02, 2019 at 08:51:34AM -0500, Benoit Parrot wrote:
> The sensor data sheet clearly state that 2592x1944 only works at 15 fps
> make sure we don't try to miss configure the pll out of acceptable
> range.

The datasheet clearly indicates that 15 fps is the maximum achievable
rate with that resolution, so I guess preventing it from being set
to anything faster than that it's acceptable.
>
> Signed-off-by: Benoit Parrot <bparrot@ti.com>
> ---
>  drivers/media/i2c/ov5640.c | 5 +++++
>  1 file changed, 5 insertions(+)
>
> diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> index 103a4e8f88e1..d5b0be2c7a0a 100644
> --- a/drivers/media/i2c/ov5640.c
> +++ b/drivers/media/i2c/ov5640.c
> @@ -1613,6 +1613,11 @@ ov5640_find_mode(struct ov5640_dev *sensor, enum ov5640_frame_rate fr,
>  	    !(mode->hact == 640 && mode->vact == 480))
>  		return NULL;
>
> +	/* 2592x1944 only works at 15fps */
> +	if (fr != OV5640_15_FPS &&

As long as 15 fps is the lower framerate declared in
ov5640_framerates[] this is ok, but I would make this condition a
check for "fr  > OV5640_15_FPS" so that it's safe for future
extensions.

(And I would check for the resolution first then FPS, so you check
the most unlikely condition first, but that's really a minor
optimization).

With the above small details addressed
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>

Thanks
   j

> +	    (mode->hact == 2592 && mode->vact == 1944))
> +		return NULL;
> +
>  	return mode;
>  }
>
> --
> 2.17.1
>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [Patch v2 1/3] media: ov5640: add PIXEL_RATE control

[Benoit Parrot]

Jacopo Mondi <jacopo@jmondi.org> wrote on Thu [2019-Oct-03 09:17:14 +0200]:
> Hi Benoit,
> 
> On Wed, Oct 02, 2019 at 08:51:32AM -0500, Benoit Parrot wrote:
> > Add v4l2 controls to report the pixel rates of each mode. This is
> > needed by some CSI2 receiver in order to perform proper DPHY
> > configuration.
> >
> > Signed-off-by: Benoit Parrot <bparrot@ti.com>
> > ---
> >  drivers/media/i2c/ov5640.c | 25 +++++++++++++++++++++++++
> >  1 file changed, 25 insertions(+)
> >
> > diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> > index 500d9bbff10b..5198dc887400 100644
> > --- a/drivers/media/i2c/ov5640.c
> > +++ b/drivers/media/i2c/ov5640.c
> > @@ -193,6 +193,9 @@ struct ov5640_mode_info {
> >
> >  struct ov5640_ctrls {
> >  	struct v4l2_ctrl_handler handler;
> > +	struct {
> > +		struct v4l2_ctrl *pixel_rate;
> > +	};
> 
> Do you need to wrap this v4l2_ctrl in it's own unnamed struct? Other
> controls here declared in this way are clustered and, if I'm not
> mistaken, using unnamed struct to wrap them is just a typographically
> nice way to convey that. I think your new control could be declared
> without a wrapping struct { }.

Probably not, just tried to be consistent with the rest of code here.

> 
> >  	struct {
> >  		struct v4l2_ctrl *auto_exp;
> >  		struct v4l2_ctrl *exposure;
> > @@ -2194,6 +2197,16 @@ static int ov5640_try_fmt_internal(struct v4l2_subdev *sd,
> >  	return 0;
> >  }
> >
> > +static u64 ov5640_calc_pixel_rate(struct ov5640_dev *sensor)
> > +{
> > +	u64 rate;
> > +
> > +	rate = sensor->current_mode->vtot * sensor->current_mode->htot;
> > +	rate *= ov5640_framerates[sensor->current_fr];
> > +
> > +	return rate;
> > +}
> > +
> 
> Just to point out this is the -theoretical- pixel rate, and might be
> quite different from the one calculated by the clock tree tuning
> procedure (which should be updated to match Hugues' latest findings).

True, and to my surprise my receiver worked with all of those value even if
some actual value maybe off, I guess in my case they were close enough.

> 
> >  static int ov5640_set_fmt(struct v4l2_subdev *sd,
> >  			  struct v4l2_subdev_pad_config *cfg,
> >  			  struct v4l2_subdev_format *format)
> > @@ -2233,6 +2246,8 @@ static int ov5640_set_fmt(struct v4l2_subdev *sd,
> >  	if (mbus_fmt->code != sensor->fmt.code)
> >  		sensor->pending_fmt_change = true;
> >
> > +	__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > +				 ov5640_calc_pixel_rate(sensor));
> >  out:
> >  	mutex_unlock(&sensor->lock);
> >  	return ret;
> > @@ -2657,6 +2672,13 @@ static int ov5640_init_controls(struct ov5640_dev *sensor)
> >  	/* we can use our own mutex for the ctrl lock */
> >  	hdl->lock = &sensor->lock;
> >
> > +	/* Clock related controls */
> > +	ctrls->pixel_rate =
> > +		v4l2_ctrl_new_std(hdl, ops,
> 
> If you like it better, this could fit in 1 line
> 
> 	ctrls->pixel_rate = v4l2_ctrl_new_std(hdl, ops, V4L2_CID_PIXEL_RATE,
> 					      0, INT_MAX, 1,
> 					      ov5640_calc_pixel_rate(sensor)
>

Either way works for me.

Benoit
 
> Thanks
>    j
> 
> > +				  V4L2_CID_PIXEL_RATE, 0, INT_MAX, 1,
> > +				  ov5640_calc_pixel_rate(sensor));
> 
> 
> > +	ctrls->pixel_rate->flags |= V4L2_CTRL_FLAG_READ_ONLY;
> > +
> >  	/* Auto/manual white balance */
> >  	ctrls->auto_wb = v4l2_ctrl_new_std(hdl, ops,
> >  					   V4L2_CID_AUTO_WHITE_BALANCE,
> > @@ -2816,6 +2838,9 @@ static int ov5640_s_frame_interval(struct v4l2_subdev *sd,
> >  		sensor->frame_interval = fi->interval;
> >  		sensor->current_mode = mode;
> >  		sensor->pending_mode_change = true;
> > +
> > +		__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > +					 ov5640_calc_pixel_rate(sensor));
> >  	}
> >  out:
> >  	mutex_unlock(&sensor->lock);
> > --
> > 2.17.1
> >

Re: [Patch v2 1/3] media: ov5640: add PIXEL_RATE control

[Benoit Parrot]

Sakari Ailus <sakari.ailus@linux.intel.com> wrote on Thu [2019-Oct-03 10:22:51 +0300]:
> Hi Jacopo, Benoit,
> 
> On Thu, Oct 03, 2019 at 09:17:14AM +0200, Jacopo Mondi wrote:
> > Hi Benoit,
> > 
> > On Wed, Oct 02, 2019 at 08:51:32AM -0500, Benoit Parrot wrote:
> > > Add v4l2 controls to report the pixel rates of each mode. This is
> > > needed by some CSI2 receiver in order to perform proper DPHY
> > > configuration.
> > >
> > > Signed-off-by: Benoit Parrot <bparrot@ti.com>
> > > ---
> > >  drivers/media/i2c/ov5640.c | 25 +++++++++++++++++++++++++
> > >  1 file changed, 25 insertions(+)
> > >
> > > diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> > > index 500d9bbff10b..5198dc887400 100644
> > > --- a/drivers/media/i2c/ov5640.c
> > > +++ b/drivers/media/i2c/ov5640.c
> > > @@ -193,6 +193,9 @@ struct ov5640_mode_info {
> > >
> > >  struct ov5640_ctrls {
> > >  	struct v4l2_ctrl_handler handler;
> > > +	struct {
> > > +		struct v4l2_ctrl *pixel_rate;
> > > +	};
> > 
> > Do you need to wrap this v4l2_ctrl in it's own unnamed struct? Other
> > controls here declared in this way are clustered and, if I'm not
> > mistaken, using unnamed struct to wrap them is just a typographically
> > nice way to convey that. I think your new control could be declared
> > without a wrapping struct { }.
> > 
> > >  	struct {
> > >  		struct v4l2_ctrl *auto_exp;
> > >  		struct v4l2_ctrl *exposure;
> > > @@ -2194,6 +2197,16 @@ static int ov5640_try_fmt_internal(struct v4l2_subdev *sd,
> > >  	return 0;
> > >  }
> > >
> > > +static u64 ov5640_calc_pixel_rate(struct ov5640_dev *sensor)
> > > +{
> > > +	u64 rate;
> > > +
> > > +	rate = sensor->current_mode->vtot * sensor->current_mode->htot;
> > > +	rate *= ov5640_framerates[sensor->current_fr];
> > > +
> > > +	return rate;
> > > +}
> > > +
> > 
> > Just to point out this is the -theoretical- pixel rate, and might be
> > quite different from the one calculated by the clock tree tuning
> > procedure (which should be updated to match Hugues' latest findings).
> 
> Hmm. Considering the xclk rate may be pretty much anything, I'd suppose
> the value above would only be correct for a given xclk rate.

I am not sure about that, different xclk rate might yield slightly
different byte clock, but all in all the resolution and framerate pretty
much dictate the end result, no?

> 
> Could this be simply calculated from the clock tree configuration, to get
> the right value in all cases?

It probably could, and as I said earlier I gave it a try and failed, since
the theoretical value worked for me that's what I went with. Those are the
same values that Maxime's patch referred to. (dfbfb7aa832cdb media: ov5640:
Compute the clock rate at runtime).

Here I am just "publishing it".

Benoit

> 
> > 
> > >  static int ov5640_set_fmt(struct v4l2_subdev *sd,
> > >  			  struct v4l2_subdev_pad_config *cfg,
> > >  			  struct v4l2_subdev_format *format)
> > > @@ -2233,6 +2246,8 @@ static int ov5640_set_fmt(struct v4l2_subdev *sd,
> > >  	if (mbus_fmt->code != sensor->fmt.code)
> > >  		sensor->pending_fmt_change = true;
> > >
> > > +	__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > > +				 ov5640_calc_pixel_rate(sensor));
> > >  out:
> > >  	mutex_unlock(&sensor->lock);
> > >  	return ret;
> > > @@ -2657,6 +2672,13 @@ static int ov5640_init_controls(struct ov5640_dev *sensor)
> > >  	/* we can use our own mutex for the ctrl lock */
> > >  	hdl->lock = &sensor->lock;
> > >
> > > +	/* Clock related controls */
> > > +	ctrls->pixel_rate =
> > > +		v4l2_ctrl_new_std(hdl, ops,
> > 
> > If you like it better, this could fit in 1 line
> > 
> > 	ctrls->pixel_rate = v4l2_ctrl_new_std(hdl, ops, V4L2_CID_PIXEL_RATE,
> > 					      0, INT_MAX, 1,
> > 					      ov5640_calc_pixel_rate(sensor)
> > 
> > Thanks
> >    j
> > 
> > > +				  V4L2_CID_PIXEL_RATE, 0, INT_MAX, 1,
> > > +				  ov5640_calc_pixel_rate(sensor));
> > 
> > 
> > > +	ctrls->pixel_rate->flags |= V4L2_CTRL_FLAG_READ_ONLY;
> 
> Note that ctrls->pixel_rate is NULL if e.g. memory allocation fails when
> creating the control.
> 
> > > +
> > >  	/* Auto/manual white balance */
> > >  	ctrls->auto_wb = v4l2_ctrl_new_std(hdl, ops,
> > >  					   V4L2_CID_AUTO_WHITE_BALANCE,
> > > @@ -2816,6 +2838,9 @@ static int ov5640_s_frame_interval(struct v4l2_subdev *sd,
> > >  		sensor->frame_interval = fi->interval;
> > >  		sensor->current_mode = mode;
> > >  		sensor->pending_mode_change = true;
> > > +
> > > +		__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > > +					 ov5640_calc_pixel_rate(sensor));
> > >  	}
> > >  out:
> > >  	mutex_unlock(&sensor->lock);
> 
> -- 
> Regards,
> 
> Sakari Ailus
> sakari.ailus@linux.intel.com

Re: [Patch v2 3/3] media: ov5640: Make 2592x1944 mode only available at 15 fps

[Benoit Parrot]

Jacopo Mondi <jacopo@jmondi.org> wrote on Thu [2019-Oct-03 09:31:55 +0200]:
> Hi Benoit,
> 
> On Wed, Oct 02, 2019 at 08:51:34AM -0500, Benoit Parrot wrote:
> > The sensor data sheet clearly state that 2592x1944 only works at 15 fps
> > make sure we don't try to miss configure the pll out of acceptable
> > range.
> 
> The datasheet clearly indicates that 15 fps is the maximum achievable
> rate with that resolution, so I guess preventing it from being set
> to anything faster than that it's acceptable.
> >
> > Signed-off-by: Benoit Parrot <bparrot@ti.com>
> > ---
> >  drivers/media/i2c/ov5640.c | 5 +++++
> >  1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> > index 103a4e8f88e1..d5b0be2c7a0a 100644
> > --- a/drivers/media/i2c/ov5640.c
> > +++ b/drivers/media/i2c/ov5640.c
> > @@ -1613,6 +1613,11 @@ ov5640_find_mode(struct ov5640_dev *sensor, enum ov5640_frame_rate fr,
> >  	    !(mode->hact == 640 && mode->vact == 480))
> >  		return NULL;
> >
> > +	/* 2592x1944 only works at 15fps */
> > +	if (fr != OV5640_15_FPS &&
> 
> As long as 15 fps is the lower framerate declared in
> ov5640_framerates[] this is ok, but I would make this condition a
> check for "fr  > OV5640_15_FPS" so that it's safe for future
> extensions.
> 
> (And I would check for the resolution first then FPS, so you check
> the most unlikely condition first, but that's really a minor
> optimization).

Ah, very good I'll change that.

Benoit

> 
> With the above small details addressed
> Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
> 
> Thanks
>    j
> 
> > +	    (mode->hact == 2592 && mode->vact == 1944))
> > +		return NULL;
> > +
> >  	return mode;
> >  }
> >
> > --
> > 2.17.1
> >

Re: [Patch v2 1/3] media: ov5640: add PIXEL_RATE control

[Sakari Ailus]

Hi Benoit,

On Thu, Oct 03, 2019 at 07:07:41AM -0500, Benoit Parrot wrote:
> Sakari Ailus <sakari.ailus@linux.intel.com> wrote on Thu [2019-Oct-03 10:22:51 +0300]:
> > Hi Jacopo, Benoit,
> > 
> > On Thu, Oct 03, 2019 at 09:17:14AM +0200, Jacopo Mondi wrote:
> > > Hi Benoit,
> > > 
> > > On Wed, Oct 02, 2019 at 08:51:32AM -0500, Benoit Parrot wrote:
> > > > Add v4l2 controls to report the pixel rates of each mode. This is
> > > > needed by some CSI2 receiver in order to perform proper DPHY
> > > > configuration.
> > > >
> > > > Signed-off-by: Benoit Parrot <bparrot@ti.com>
> > > > ---
> > > >  drivers/media/i2c/ov5640.c | 25 +++++++++++++++++++++++++
> > > >  1 file changed, 25 insertions(+)
> > > >
> > > > diff --git a/drivers/media/i2c/ov5640.c b/drivers/media/i2c/ov5640.c
> > > > index 500d9bbff10b..5198dc887400 100644
> > > > --- a/drivers/media/i2c/ov5640.c
> > > > +++ b/drivers/media/i2c/ov5640.c
> > > > @@ -193,6 +193,9 @@ struct ov5640_mode_info {
> > > >
> > > >  struct ov5640_ctrls {
> > > >  	struct v4l2_ctrl_handler handler;
> > > > +	struct {
> > > > +		struct v4l2_ctrl *pixel_rate;
> > > > +	};
> > > 
> > > Do you need to wrap this v4l2_ctrl in it's own unnamed struct? Other
> > > controls here declared in this way are clustered and, if I'm not
> > > mistaken, using unnamed struct to wrap them is just a typographically
> > > nice way to convey that. I think your new control could be declared
> > > without a wrapping struct { }.
> > > 
> > > >  	struct {
> > > >  		struct v4l2_ctrl *auto_exp;
> > > >  		struct v4l2_ctrl *exposure;
> > > > @@ -2194,6 +2197,16 @@ static int ov5640_try_fmt_internal(struct v4l2_subdev *sd,
> > > >  	return 0;
> > > >  }
> > > >
> > > > +static u64 ov5640_calc_pixel_rate(struct ov5640_dev *sensor)
> > > > +{
> > > > +	u64 rate;
> > > > +
> > > > +	rate = sensor->current_mode->vtot * sensor->current_mode->htot;
> > > > +	rate *= ov5640_framerates[sensor->current_fr];
> > > > +
> > > > +	return rate;
> > > > +}
> > > > +
> > > 
> > > Just to point out this is the -theoretical- pixel rate, and might be
> > > quite different from the one calculated by the clock tree tuning
> > > procedure (which should be updated to match Hugues' latest findings).
> > 
> > Hmm. Considering the xclk rate may be pretty much anything, I'd suppose
> > the value above would only be correct for a given xclk rate.
> 
> I am not sure about that, different xclk rate might yield slightly
> different byte clock, but all in all the resolution and framerate pretty
> much dictate the end result, no?

Interestingly, the driver determines the PLL configuration based on the
pixels per line and lines per frame (including blanking) and the frames per
seconds. I guess it's always been like that in this driver.

So I agree the target frame rate can be used for this.

You could change ov5640_set_mode() to use this function as well to avoid
doing the same calculation twice in different places in the driver. Up to
you.

> 
> > 
> > Could this be simply calculated from the clock tree configuration, to get
> > the right value in all cases?
> 
> It probably could, and as I said earlier I gave it a try and failed, since
> the theoretical value worked for me that's what I went with. Those are the
> same values that Maxime's patch referred to. (dfbfb7aa832cdb media: ov5640:
> Compute the clock rate at runtime).
> 
> Here I am just "publishing it".
> 
> Benoit
> 
> > 
> > > 
> > > >  static int ov5640_set_fmt(struct v4l2_subdev *sd,
> > > >  			  struct v4l2_subdev_pad_config *cfg,
> > > >  			  struct v4l2_subdev_format *format)
> > > > @@ -2233,6 +2246,8 @@ static int ov5640_set_fmt(struct v4l2_subdev *sd,
> > > >  	if (mbus_fmt->code != sensor->fmt.code)
> > > >  		sensor->pending_fmt_change = true;
> > > >
> > > > +	__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > > > +				 ov5640_calc_pixel_rate(sensor));
> > > >  out:
> > > >  	mutex_unlock(&sensor->lock);
> > > >  	return ret;
> > > > @@ -2657,6 +2672,13 @@ static int ov5640_init_controls(struct ov5640_dev *sensor)
> > > >  	/* we can use our own mutex for the ctrl lock */
> > > >  	hdl->lock = &sensor->lock;
> > > >
> > > > +	/* Clock related controls */
> > > > +	ctrls->pixel_rate =
> > > > +		v4l2_ctrl_new_std(hdl, ops,
> > > 
> > > If you like it better, this could fit in 1 line
> > > 
> > > 	ctrls->pixel_rate = v4l2_ctrl_new_std(hdl, ops, V4L2_CID_PIXEL_RATE,
> > > 					      0, INT_MAX, 1,
> > > 					      ov5640_calc_pixel_rate(sensor)
> > > 
> > > Thanks
> > >    j
> > > 
> > > > +				  V4L2_CID_PIXEL_RATE, 0, INT_MAX, 1,
> > > > +				  ov5640_calc_pixel_rate(sensor));
> > > 
> > > 
> > > > +	ctrls->pixel_rate->flags |= V4L2_CTRL_FLAG_READ_ONLY;
> > 
> > Note that ctrls->pixel_rate is NULL if e.g. memory allocation fails when
> > creating the control.
> > 
> > > > +
> > > >  	/* Auto/manual white balance */
> > > >  	ctrls->auto_wb = v4l2_ctrl_new_std(hdl, ops,
> > > >  					   V4L2_CID_AUTO_WHITE_BALANCE,
> > > > @@ -2816,6 +2838,9 @@ static int ov5640_s_frame_interval(struct v4l2_subdev *sd,
> > > >  		sensor->frame_interval = fi->interval;
> > > >  		sensor->current_mode = mode;
> > > >  		sensor->pending_mode_change = true;
> > > > +
> > > > +		__v4l2_ctrl_s_ctrl_int64(sensor->ctrls.pixel_rate,
> > > > +					 ov5640_calc_pixel_rate(sensor));
> > > >  	}
> > > >  out:
> > > >  	mutex_unlock(&sensor->lock);
> > 
> > -- 
> > Regards,
> > 
> > Sakari Ailus
> > sakari.ailus@linux.intel.com

-- 
Sakari Ailus
sakari.ailus@linux.intel.com