Today's level of yak shaving:
Wanted to get the cameras on my Dell Latitude 5290 working. Found out those are also IPU3 based ones. Then wondered why they wouldn't show up at all. Now I am compiling my own Debian kernel to at least see what's missing in terms of functionality.
On the upside I now know how to compile the kernel and will create a patch (soon™) for the IPU3 and CIO2 modules to be enabled by default in #Debian.
Wanted to get the cameras on my Dell Latitude 5290 working. Found out those are also IPU3 based ones. Then wondered why they wouldn't show up at all. Now I am compiling my own Debian kernel to at least see what's missing in terms of functionality.
On the upside I now know how to compile the kernel and will create a patch (soon™) for the IPU3 and CIO2 modules to be enabled by default in #Debian.
so I finished compiling the kernel. MR enabling the IPU3 and CIO2 modules by default is submitted and pending approval. This is what I like about free software: ones personal itch might just help another person.
Thanks to the helpful libcamera folks I now know that it needs some additional ov* drivers enabled. Also made a MR for that. With those enabled running the cameras should be as simple as installing the v4l2loopback kernel module and executing a launch script. Unfortunately the kernel module did not want to compile with my newer kernel, so I'll wait until this is mainlined. v4l2 --list returns the correct cameras now, so that's progress.
After some more digging it turns out that Intel was lazy. They used a similar camera setup in my device and the Surface device line. However, the voltage to the camera sensor is controlled by a tps68470 IC which can deliver variable voltages. Since the Linux kernel does not know the board (and the correct voltage levels), the camera is without power. Now I need to boot Windows (ugh) and get the correct values from reading the I2C traffic.
- replies
- 1
- announces
- 0
- likes
- 0