mbox series

[v7,0/5] media: i2c: imx334: support lower bandwidth mode

Message ID 20221219061526.3169369-1-shravan.chippa@microchip.com
Headers show
Series media: i2c: imx334: support lower bandwidth mode | expand

Message

shravan chippa Dec. 19, 2022, 6:15 a.m. UTC
From: Shravan Chippa <shravan.chippa@microchip.com>

Hi

This patch series is for imx334 sensor driver support for lower bandwidth

Some platforms may not be capable of supporting the bandwidth
required for 12 bit or 3840x2160@60 resolutions.

Add support for dynamically selecting 10 bit and 1920x1080@30
resolutions while leaving the existing configuration as default

V6 -> V7
Reloved: kernel test robot warning
"drivers/media/i2c/imx334.c:767:15: warning: unused variable 'i' "

V5 -> V6
-Drop the dt-binding patch
-Optimize the code to avoid duplicating the lines
-Added proper mutex while imx334_mbus_codes array
-Modified Function __v4l2_ctrl_modify_range arguments as per the review commants
-Added hblank dummy set ctrl
-Removed Redundant comment
-corrected code alignment 
-All commit msgs are re-written

V4 -> V5
-Added 5 more patchs as per the review comments witch has below updates
-Updated 1782000000Mbps link frequency for 3840x2160@60 as per the mode
values
-Updated 1782000000Mbps link frequency in dt-bindings also
-Updated 3840x2160@60 mode array with default(reset) values

-Updated hblank __v4l2_ctrl_s_ctrl() to __v4l2_ctrl_modify_range()
Suggested-by: Jacopo Mondi <jacopo@jmondi.org>

-Current mode update only when we try to set V4L2_SUBDEV_FORMAT_ACTIVE
-Added link frequency (891000000Mbps) and pixel rate (74250000) to
1920x1080@30 mode
Suggested-by: Sakari Ailus <sakari.ailus@iki.fi>

-Updated commit message

V3 -> V4
- Make the 12 bit and 3840x2160 as default
- Set bus code SRGGB12 if set format fails

V2 -> V3
- Fixed the warning reported by kernel test robot

V1 -> V2
- Addressed the review comment given by Jacopo Mondi,
  Which has bug in imx334_enum_frame_size() loop function,
- Renamed array codes[] to imx334_mbus_codes[]


Shravan Chippa (5):
  media: i2c: imx334: modify link frequency as for the configureation
  media: i2c: imx334: replace __v4l2_ctrl_s_ctrl to
    __v4l2_ctrl_modify_range
  media: i2c: imx334: add missing reset values for mode 3840x2160_regs[]
  media: i2c: imx334: support lower bandwidth mode
  media: i2c: imx334: update pixel and link frequency

 drivers/media/i2c/imx334.c | 337 ++++++++++++++++++++++++++++++++++---
 1 file changed, 309 insertions(+), 28 deletions(-)

Comments

Jacopo Mondi Dec. 21, 2022, 8:43 a.m. UTC | #1
Hi Shravan

On Tue, Dec 20, 2022 at 11:11:15AM +0000, Shravan.Chippa@microchip.com wrote:
>
>
> > -----Original Message-----
> > From: Jacopo Mondi <jacopo@jmondi.org>
> > Sent: 19 December 2022 08:14 PM
> > To: shravan Chippa - I35088 <Shravan.Chippa@microchip.com>
> > Cc: paul.j.murphy@intel.com; daniele.alessandrelli@intel.com;
> > mchehab@kernel.org; linux-media@vger.kernel.org; linux-
> > kernel@vger.kernel.org
> > Subject: Re: [PATCH v7 1/5] media: i2c: imx334: modify link frequency as for
> > the configureation
> >
> > EXTERNAL EMAIL: Do not click links or open attachments unless you know the
> > content is safe
> >
> > Hi Shravan
> >
> > On Mon, Dec 19, 2022 at 11:45:22AM +0530, shravan kumar wrote:
> > > From: Shravan Chippa <shravan.chippa@microchip.com>
> > >
> > > Currently imx334 sensor driver is configured for 1782Mbps/lane for
> > > 3840x2160@60 resolution with reqired reg mode values but if we run the
> > > command "v4l2-ctl --all -d /dev/v4l-subdevX" it is showing incorrect
> > > link frequeny, This is because of the incorrect value of
> > > IMX334_LINK_FREQ witch is 891000000. it should be 1782000000.
> > >
> > > In general with the value of 891000000 link frequency it is not
> > > possible to configure 3840x2160@60 resolution.
> > >
> > > Fixes: 9746b11715c3 ("media: i2c: Add imx334 camera sensor driver")
> > >
> > > Signed-off-by: Shravan Chippa <shravan.chippa@microchip.com>
> > > ---
> > >  drivers/media/i2c/imx334.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/media/i2c/imx334.c b/drivers/media/i2c/imx334.c
> > > index 7b0a9086447d..acc9f9f15e47 100644
> > > --- a/drivers/media/i2c/imx334.c
> > > +++ b/drivers/media/i2c/imx334.c
> > > @@ -49,7 +49,7 @@
> > >  #define IMX334_INCLK_RATE    24000000
> > >
> > >  /* CSI2 HW configuration */
> > > -#define IMX334_LINK_FREQ     891000000
> > > +#define IMX334_LINK_FREQ     1782000000
> >
> > Is this your reasoning ?
> >
> > width: 3840
> > hblank: 560
> > height: 2160
> > vblank: 2340
> > bpp: 12
> > fps: 60
> > lanes: 4
> >
> > Total bandwidth: (3840 + 560) * (2160 + 2340) * 60 * 12 = 14.256.000.000
> > Bandwidth per lane = Total / 4 = 3.564.000.000 mipi clock =
> > Bandwidth_per_lane / 2 = 1.782.000.000
> >
> > Two questions:
> >
> > - Should you update the pixel clock as well ? It is currently set to
> >   594000000 while as per the above reasoning it should be doubled too.
> >
> > - Where is the sensor's clock tree programmed in the driver ?
> >   It's kind of weird that the pixel_clock and link_freq in the driver
> >   are half of what they theoretically should be...
> >
> >
> As per my understanding.
> the mode_3840x2160_regs[] array value which is written through the i2c bus is 4k resolution, 60fps,  link frequency 1782Mbps per lane
> but the vblank value is dynamic from user space.
> Min-90 to Max-130000, default value is 2340. With the default value, we will get 30fps.

Ah, it's 30, not 60. So my calculations above should be halved

>
> if we set vblank value from user space it will change FPS.
>

Sure, but the link frequency stays the same, and it should be computed
with the FPS resulting from the current blankings

> Total bandwidth: (3840 + 560) * (2160 + 2340) * 30 * 12 = 7.128.000.000

Correct

> Bandwidth per lane = Total / 4 = 1.782.000.000

Correct.

But with CID_LINK_FREQ you're reporting the bus link frequency, not
the lane bandwidth. As MIPI CSI-2 uses DDR read mode, two bits per
clock cycle are transmitted, hence the bus frequency is half of the lane
bandwidth.

TL;DR you don't need this patch, the current value is correct as it is in my
understanding.

Thanks
  j

>
> Thanks,
> Shravan
>
> >
> > >  #define IMX334_NUM_DATA_LANES        4
> > >
> > >  #define IMX334_REG_MIN               0x00
> > > --
> > > 2.34.1
> > >