mbox series

[v6,0/9] media: i2c: ov5670: OF support, runtime_pm, regulators

Message ID 20230126165909.121302-1-jacopo.mondi@ideasonboard.com (mailing list archive)
Headers show
Series media: i2c: ov5670: OF support, runtime_pm, regulators | expand

Message

Jacopo Mondi Jan. 26, 2023, 4:59 p.m. UTC
Hello
  this small series introduces OF support for the ov5670 sensor and
adds support for regulators clocks and GPIOs.

Last year I dropped the ball as I didn't have access to HW anymore.
Luca (in cc) has reported he has a sensor and might give this new version
a spin, thanks!

Cheers
  j

v5->v6:
- Rework clock parsing as suggested by Sakari
- Move runtime_pm enablement after async subdev registration
- Use DIV_ROUND_UP to round clock freq

v4->v5:
- Enable clock in ov5670_runtime_resume() as suggested by Luca
- Add a patch to handle HBLANK, PIXEL_RATE and LINK_FREQ in .set_ctrl()
  to fix a warning again reported by Luca

v3->v4:
- Rework power enablement in power up sequence to support !CONFIG_OF
- Minor changes as per Sakari's review

v2->v3:
- bindings:
  - Drop assigned-clock properties from schema (moved to example)
  - s/pwdn-gpios/powerdown-gpios/

- driver
  - Use is_of_node() to decide how to parse clocks
  - Fix:
    drivers/media/i2c/ov5670.c:1787:18: error: initializer element is not a compile-time constant
                   .analog_crop = ov5670_analog_crop,
                                  ^~~~~~~~~~~~~~~~~~

    reported by kernel test robot and Nathan Chancellor with
    clang15 and gcc < 8

v1->v2:
- Address Krzysztof comments on bindings
- 2/8: new patch to use the common clock framework
- Address Lauren's comment on runtime_pm function names
- 7/8: new patch to implement init_cfg as suggested by Laurent
- Rework 8/8 which was incorrect as reported by Laurent

Thanks
   j

Jacopo Mondi (8):
  media: dt-bindings: Add OV5670
  media: i2c: ov5670: Allow probing with OF
  media: i2c: ov5670: Use common clock framework
  media: i2c: ov5670: Probe regulators
  media: i2c: ov5670: Probe GPIOs
  media: i2c: ov5670: Add runtime_pm operations
  media: i2c: ov5670: Implement init_cfg
  media: i2c: ov5670: Handle RO controls in set_ctrl

Jean-Michel Hautbois (1):
  media: i2c: ov5670: Add .get_selection() support

 .../bindings/media/i2c/ovti,ov5670.yaml       |  92 ++++++
 MAINTAINERS                                   |   1 +
 drivers/media/i2c/ov5670.c                    | 312 +++++++++++++++---
 3 files changed, 360 insertions(+), 45 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/media/i2c/ovti,ov5670.yaml

--
2.39.0

Comments

Luca Weiss Jan. 28, 2023, 9:27 p.m. UTC | #1
Hi Jacopo,

On Donnerstag, 26. Jänner 2023 17:59:00 CET Jacopo Mondi wrote:
> Hello
>   this small series introduces OF support for the ov5670 sensor and
> adds support for regulators clocks and GPIOs.
> 
> Last year I dropped the ball as I didn't have access to HW anymore.
> Luca (in cc) has reported he has a sensor and might give this new version
> a spin, thanks!

With this series the sensor initializes correctly for me on msm8974pro-
fairphone-fp2. I locally have essentially one more change that configures the 
sensor for single-lane operation (currently dual-lane is hardcoded) and with 
that I can get image that using libcamera.

Tested-by: Luca Weiss <luca@z3ntu.xyz>


Thanks for respinning this!

Regards
Luca

> 
> Cheers
>   j
> 
> v5->v6:
> - Rework clock parsing as suggested by Sakari
> - Move runtime_pm enablement after async subdev registration
> - Use DIV_ROUND_UP to round clock freq
> 
> v4->v5:
> - Enable clock in ov5670_runtime_resume() as suggested by Luca
> - Add a patch to handle HBLANK, PIXEL_RATE and LINK_FREQ in .set_ctrl()
>   to fix a warning again reported by Luca
> 
> v3->v4:
> - Rework power enablement in power up sequence to support !CONFIG_OF
> - Minor changes as per Sakari's review
> 
> v2->v3:
> - bindings:
>   - Drop assigned-clock properties from schema (moved to example)
>   - s/pwdn-gpios/powerdown-gpios/
> 
> - driver
>   - Use is_of_node() to decide how to parse clocks
>   - Fix:
>     drivers/media/i2c/ov5670.c:1787:18: error: initializer element is not a
> compile-time constant .analog_crop = ov5670_analog_crop,
>                                   ^~~~~~~~~~~~~~~~~~
> 
>     reported by kernel test robot and Nathan Chancellor with
>     clang15 and gcc < 8
> 
> v1->v2:
> - Address Krzysztof comments on bindings
> - 2/8: new patch to use the common clock framework
> - Address Lauren's comment on runtime_pm function names
> - 7/8: new patch to implement init_cfg as suggested by Laurent
> - Rework 8/8 which was incorrect as reported by Laurent
> 
> Thanks
>    j
> 
> Jacopo Mondi (8):
>   media: dt-bindings: Add OV5670
>   media: i2c: ov5670: Allow probing with OF
>   media: i2c: ov5670: Use common clock framework
>   media: i2c: ov5670: Probe regulators
>   media: i2c: ov5670: Probe GPIOs
>   media: i2c: ov5670: Add runtime_pm operations
>   media: i2c: ov5670: Implement init_cfg
>   media: i2c: ov5670: Handle RO controls in set_ctrl
> 
> Jean-Michel Hautbois (1):
>   media: i2c: ov5670: Add .get_selection() support
> 
>  .../bindings/media/i2c/ovti,ov5670.yaml       |  92 ++++++
>  MAINTAINERS                                   |   1 +
>  drivers/media/i2c/ov5670.c                    | 312 +++++++++++++++---
>  3 files changed, 360 insertions(+), 45 deletions(-)
>  create mode 100644
> Documentation/devicetree/bindings/media/i2c/ovti,ov5670.yaml
> 
> --
> 2.39.0