[v4,0/2] Add imx471 soft-ISP sensor support
mbox series

Message ID CALr1FG__gS_agbUnB5iWZ8wQypxdJch_UHziw2xuuG5Axp3vtQ@mail.gmail.com
Headers show
Series
  • Add imx471 soft-ISP sensor support
Related show

Message

John Cronin July 31, 2026, 3:15 p.m. UTC
Add CameraSensorHelper and static sensor properties for Sony IMX471
(SONY471A), used on Lenovo ThinkPad X9 / X1 Carbon Gen14 (IPU7).

Changes since v3:
- Document measured analogue gain model and black level on real
  hardware (raw Bayer sweep, fixed exposure).
- Confirm G = 1024/(1024-code) within ~1.5% over codes 0..800; max
  ~4.57x is the V4L2 code range under that model.
- Black level ~64 DN @ 10-bit from min exposure/gain raw frames.

v3 process notes (still apply): new thread, split patches, empty
sensorDelays, no self Tested-by.

Tested on Fedora 44 / ThinkPad X9-15 Gen 1 (kernel 7.1.5).

John Cronin (2):
  libipa: camera_sensor_helper: Add imx471
  libcamera: camera_sensor_properties: Add imx471

 src/ipa/libipa/camera_sensor_helper.cpp       | 29 +++++++++++++++++++
 .../sensor/camera_sensor_properties.cpp       | 14 +++++++++
 2 files changed, 43 insertions(+)

Comments

Jacopo Mondi July 31, 2026, 4:23 p.m. UTC | #1
Hi John

how did you formatted/sent the patches ?
I can't apply them with `git am` as I usually do:

error: git diff header lacks filename information when removing 1 leading pathname component at .git/rebase-apply/patch:6

Apparently the client you used to send emails broke lines which makes
the patch untreatable for git am.

Did you use git-send-email to send the patch ?
See Documentation/contributing.rst in particular
https://git-send-email.io/

Thanks
  j

On Fri, Jul 31, 2026 at 08:15:05AM -0700, John Cronin wrote:
> Add CameraSensorHelper and static sensor properties for Sony IMX471
> (SONY471A), used on Lenovo ThinkPad X9 / X1 Carbon Gen14 (IPU7).
>
> Changes since v3:
> - Document measured analogue gain model and black level on real
>   hardware (raw Bayer sweep, fixed exposure).
> - Confirm G = 1024/(1024-code) within ~1.5% over codes 0..800; max
>   ~4.57x is the V4L2 code range under that model.
> - Black level ~64 DN @ 10-bit from min exposure/gain raw frames.
>
> v3 process notes (still apply): new thread, split patches, empty
> sensorDelays, no self Tested-by.
>
> Tested on Fedora 44 / ThinkPad X9-15 Gen 1 (kernel 7.1.5).
>
> John Cronin (2):
>   libipa: camera_sensor_helper: Add imx471
>   libcamera: camera_sensor_properties: Add imx471
>
>  src/ipa/libipa/camera_sensor_helper.cpp       | 29 +++++++++++++++++++
>  .../sensor/camera_sensor_properties.cpp       | 14 +++++++++
>  2 files changed, 43 insertions(+)
>
> --
> 2.55.0
John Cronin July 31, 2026, 4:33 p.m. UTC | #2
Hi Jacopo,

On Fri, Jul 31, 2026 at 06:23:31PM +0200, Jacopo Mondi wrote:
> how did you formatted/sent the patches ?
> I can't apply them with `git am` as I usually do:
>
> error: git diff header lacks filename information when removing 1 leading
> pathname component ...
>
> Apparently the client you used to send emails broke lines which makes
> the patch untreatable for git am.
>
> Did you use git-send-email to send the patch ?

Sorry about that — no, I did not use git-send-email. The patches were
sent as plain-text bodies through Gmail's API/web path, which wrapped
long lines and mangled the format-patch headers for `git am`.

I do not have git-send-email/SMTP set up on this machine yet (will fix
that for future submissions; thanks for the contributing.rst pointer).

To unblock apply without relying on the MUA, here is a **git-am-ready**
format-patch series (raw files, tabs preserved, verified with `git am`
locally):

  https://gist.github.com/johncronin3/51dfc0bb2204b8ab13c678b9fad0bb13

  curl -fsSL https://gist.githubusercontent.com/johncronin3/51dfc0bb2204b8ab13c678b9fad0bb13/raw/0001-libipa-camera_sensor_helper-Add-imx471.patch
| git am
  curl -fsSL https://gist.githubusercontent.com/johncronin3/51dfc0bb2204b8ab13c678b9fad0bb13/raw/0002-libcamera-camera_sensor_properties-Add-imx471.patch
| git am

That series is **v5** vs the broken Gmail v4:
- Drop the long in-code helper comment (measurement detail stays in the
  commit message only) — please do drop it if you still prefer that.
- Drop the redundant test-pattern comment on properties (as on your R-b).
- Same functional code (blackLevel 4096, AnalogueGainLinear 1024 model).

Also fine if you apply from the earlier content and just drop the
comments yourself; the gist is only to give you a clean `git am` path.

Measurement write-up/script remains at:
  https://gist.github.com/johncronin3/15d9eacbf64f91ecc600fe583686325f

Thanks for your patience,
John Cronin