On 22/09/2022 12:19, Bryan O'Donoghue wrote:
On 22/09/2022 12:16, Dave Stevenson wrote:
It may*eventually* work for all three parts, but isn't the time to
add the compatible string at the point where it is actually compatible
with the driver?
Yes. I forgot about the 0x477 chip id on your part.
I'm happy enough to drop 477 from the compat string or indeed to allow
0x0477 as a valid chip identifier in imx412.
Sakari, what would you like to do ?
---
bod
Right.
So I got myself the official rpi imx477 sensor and ran the imx412/imx577
driver on the rpi 5.19.y tree
It looks like the rpi driver configures imx477 for two MIPI data-lanes,
whereas upstream imx412 wants four MIPI data-lanes.
So already that means the imx412 config as-is won't work.
But, we do know these sensors are very very close.
I think the right medium term thing to do is try take in the majority of
the imx477 code - including the various test modes, and resolutions and
support for different MIPI data-lane configurations.
Its not clear to me that the imx412/imx577 and imx378/imx477 can
genuinely live in the same codebase though.
Anyway I think adding imx477 to the imx412 driver should be considered
out of scope pending answering most of those questions and getting the
code to work.
Anyway that merging of rpi imx477 and upstream imx412/imx577 code feels
like an entirely different series.
So I'll resend this series minus the imx477 bits.
---
bod
