[v4,0/2] media: imx415: Add more clock configurations

Message

Alexander Stein Jan. 17, 2024, 7:39 a.m. UTC

Hi,

this small series converts the driver to using CCI helpers.
On top of that more clock setups are added to the driver.
These are all taken from the datasheet. I suspect they all can be calculated
dynamically later on.
Tested on TQMa8MPxL + MBa8MPxL + ISP + Vision Components IMX415 camera.

While working on the driver I notived that imx415 sets mbus code to
MEDIA_BUS_FMT_SGBRG10_1X10, while imx290 uses MEDIA_BUS_FMT_SRGGB10_1X10.
But the datasheets show the same color coding pattern. But both settings seem
to be correct, e.g. the resulting image has correct colors.

I had to remove the identity check locally. I suspect that Vision Components
interjects reading some registers.

Changes in v4:
* Fixed reading chip ID
* Collected Laurent's R-b

Changes in v3:
* Lower case register definitions
* Add missing variable initialization
* Fix line lengths on several locations
* Simplify imx415_s_ctrl()'s cci_write usage
* Suffix lane_rate (u64) values by UL

Changes in v2:
* Removed explicit selection of REGMAP_I2C
* Dropped patch 3 for now. Mode selection shall be done by dynamic VMAX/HMAX
  calculations later on

Best regards,
Alexander

Alexander Stein (2):
  media: i2c: imx415: Convert to new CCI register access helpers
  media: i2c: imx415: Add more clock configurations

 drivers/media/i2c/Kconfig  |   1 +
 drivers/media/i2c/imx415.c | 672 +++++++++++++++++++++++--------------
 2 files changed, 430 insertions(+), 243 deletions(-)

Comments

Michael Riesch Jan. 17, 2024, 8:25 a.m. UTC | #1

Hi Alexander,

On 1/17/24 08:39, Alexander Stein wrote:
> Hi,
> 
> this small series converts the driver to using CCI helpers.
> On top of that more clock setups are added to the driver.
> These are all taken from the datasheet. I suspect they all can be calculated
> dynamically later on.
> Tested on TQMa8MPxL + MBa8MPxL + ISP + Vision Components IMX415 camera.
> 
> While working on the driver I notived that imx415 sets mbus code to
> MEDIA_BUS_FMT_SGBRG10_1X10, while imx290 uses MEDIA_BUS_FMT_SRGGB10_1X10.
> But the datasheets show the same color coding pattern. But both settings seem
> to be correct, e.g. the resulting image has correct colors.
> 
> I had to remove the identity check locally. I suspect that Vision Components
> interjects reading some registers.
> 
> Changes in v4:
> * Fixed reading chip ID
> * Collected Laurent's R-b
> 
> Changes in v3:
> * Lower case register definitions
> * Add missing variable initialization
> * Fix line lengths on several locations
> * Simplify imx415_s_ctrl()'s cci_write usage
> * Suffix lane_rate (u64) values by UL
> 
> Changes in v2:
> * Removed explicit selection of REGMAP_I2C
> * Dropped patch 3 for now. Mode selection shall be done by dynamic VMAX/HMAX
>   calculations later on
> 
> Best regards,
> Alexander
> 
> Alexander Stein (2):
>   media: i2c: imx415: Convert to new CCI register access helpers
>   media: i2c: imx415: Add more clock configurations
> 
>  drivers/media/i2c/Kconfig  |   1 +
>  drivers/media/i2c/imx415.c | 672 +++++++++++++++++++++++--------------
>  2 files changed, 430 insertions(+), 243 deletions(-)
> 

very nice! Haven't had a chance to test your series yet, but it LGTM.
For both patches

Reviewed-by: Michael Riesch <[e-mail]>

Thanks and best regards,
Michael