Date: Wed, 21 Feb 2024 18:02:15 +0300
From: Mikhail Rudenko <mike.rudenko@...il.com>
To: Mikhail Rudenko <mike.rudenko@...il.com>
Cc: linux-media@...r.kernel.org, linux-kernel@...r.kernel.org, Sakari Ailus
 <sakari.ailus@...ux.intel.com>, Laurent Pinchart
 <laurent.pinchart@...asonboard.com>, Jacopo Mondi <jacopo@...ndi.org>,
 Tommaso Merciai <tomm.merciai@...il.com>, Christophe JAILLET
 <christophe.jaillet@...adoo.fr>, Dave Stevenson
 <dave.stevenson@...pberrypi.com>, Mauro Carvalho Chehab
 <mchehab@...nel.org>
Subject: Re: [PATCH v2 00/20] Omnivision OV4689 refactoring and improvements


Hi,

On 2023-12-18 at 20:40 +03, Mikhail Rudenko <mike.rudenko@...il.com> wrote:

> Hi,
>
> This series contains refactoring and new features implementation for
> the Omnivision OV4689 sensor driver. Specifically, patches 1, 2, 3, 5,
> 6, 10, 15, 16, 18, and 19 are refactorings, and are not supposed to
> introduce any functional change. Patches 4 and 7 perform migration to
> CCI helpers and subdevice active state respectively, and should not
> introduce any hardware- and/or user-visible change either. Patch 8
> fixes a possible race condition due to v4l2_async_register_subdev_sensor
> being called too early in ov4689_probe, and patch 9 migrates power
> management to PM autosuspend.
>
> Patches 11-14 expose more sensor controls to the userspace, such as
> (read-write) HBLANK, VFLIP/HFLIP, digital gain, and color
> balance. Patch 17 implements configurable analogue crop rectangle via
> .set_selection callback. And finally, patch 20 enables 2x2 binning
> option. It should be noted that publicly available sensor
> documentation is lacking description of many registers and their value
> ranges, so a lot of values had to be found by experimentation.

Gentle ping on this series. Anything I can do to help getting it
reviewed and merged? Maybe split patches 15-20 which implement cropping
and binning and change the driver away from register list based model
into a separate series? Anyone?

> Changes in v2:
> - collect Laurent's r-b's
> - squash together "CCI conversion" and "Set gain in one 16 bit write"
> - use ctrl->val in ov4689_set_ctrl
> - rename try_fmt to fmt in ov4689_init_cfg and drop corresponding comment
> - rebase on top of media-stage and rename init_cfg->init_state
> - sort register definitions by address throughout the whole series
> - fix number of controls hint in v4l2_ctrl_handler_init
> - make all hexadecimal constants lowercase
> - disable runtime pm in probe error path
> - implement pm autosuspend
>
> Mikhail Rudenko (20):
>   media: i2c: ov4689: Clean up and annotate the register table
>   media: i2c: ov4689: Sort register definitions by address
>   media: i2c: ov4689: Fix typo in a comment
>   media: i2c: ov4689: CCI conversion
>   media: i2c: ov4689: Remove i2c_client from ov4689 struct
>   media: i2c: ov4689: Refactor ov4689_set_ctrl
>   media: i2c: ov4689: Use sub-device active state
>   media: i2c: ov4689: Enable runtime PM before registering sub-device
>   media: i2c: ov4689: Use runtime PM autosuspend
>   media: i2c: ov4689: Remove max_fps field from struct ov4689_mode
>   media: i2c: ov4689: Make horizontal blanking configurable
>   media: i2c: ov4689: Implement vflip/hflip controls
>   media: i2c: ov4689: Implement digital gain control
>   media: i2c: ov4689: Implement manual color balance controls
>   media: i2c: ov4689: Move pixel array size out of struct ov4689_mode
>   media: i2c: ov4689: Set timing registers programmatically
>   media: i2c: ov4689: Configurable analogue crop
>   media: i2c: ov4689: Eliminate struct ov4689_mode
>   media: i2c: ov4689: Refactor ov4689_s_stream
>   media: i2c: ov4689: Implement 2x2 binning
>
>  drivers/media/i2c/Kconfig  |   1 +
>  drivers/media/i2c/ov4689.c | 964 +++++++++++++++++++++++--------------
>  2 files changed, 592 insertions(+), 373 deletions(-)


--
Best regards,
Mikhail Rudenko

Powered by blists - more mailing lists