It's Becoming Possible To Use The Webcam On Newer Intel Laptops With Open-Source Linux

[pong]

IIRC UVC = USB video class.
I don't know what the chip this attaches to is, but if it isn't one connected to the CPU via USB (which seems modestly unlikely) then UVC wouldn't be applicable.
Maybe the MIPI signals are piped down a cable to the motherboard where a chipset IC ingests them directly and it shows up as a PCIE attached chip
with whatever image processing / acquisition aspect to it?
Just a guess.

Yeah. I thought Microsoft made non-UVC webcams disqualifying for Windows Logo Certification starting with Windows Vista.

[pong]

It's not just the webcam Intel has been doing a terrible job on for FOSS support.
On their own brand / manufactured Arc GPU boards there's no way I can see to do things like
read the temperatures, read / control the clock speeds, read / control the fan speeds,
set the power management properties, enable support for ASPM / enable low power operation.

All of those things are supported with a control & monitoring utility under ms windows and have been since ~ 1+ years ago launch.

And though they've published a fair amount of GPU interface "video driver" related documentation for their GPUs
these basic (and very simple) "what's the temperatures?", "read / set the fan speeds" things don't even have any API / register
documentation otherwise I'm sure they'd already have been supported by FOSS sensor & GPU settings utilities
even if Intel hadn't written the GUI / CLI for them.

I guess the temperature / clock / fan monitoring and control APIs / registers must be "very secret"
moreso than the actually graphics relevant parts of the GPU which they've generated hundreds of
pages of documentation on.

[chithanh]

On linux they have generally been pretty good.

This camera tech stuff is one of their rare opensource failures

No. Intel had a number of opensource failures. It was just until the camera mostly confined to their low-end Atom platforms.

Hans himself did a lot of work to rectify for example Cherry Trail support and implemented essential parts like battery monitoring for the MID/UMPC platform.

Before that there was Bay Trail which simply locked up when entering ACPI sleep states (kernel bugzilla report had hundreds of comments).

(the other major one being gma500 from around a decade ago -

Intel Poulsbo/GMA500 was launched in 2008, more than 15 years ago.
Since then, Intel launched several more Atom processors with PowerVR graphics and largely the same driver situation.
Oak Trail/GMA600 (2010)
Cedar Trail/GMA3600 (2011)
Clover Trail/GMA3650 (2012)
Moorefield/G6430 (2014)

After that they had Spreadtrum continue this business with SC98xx (2016),
while Intel briefly switched to Mali graphics for their Atom x3 SoFIA (2015) and then went back to inhouse GPU designs.

It is therefore not one failure, it is a string of failures spanning the better part of a decade.

[lkraav]

Is there any hope for IPU4 on Ice Lake hansdegoede

[hansdegoede]

Is there any hope for IPU4 on Ice Lake hansdegoede

ATM I do not know of anyone working on this. Are you sure that Ice Lake has an IPU4 ? I thought it was some early version of the IPU6 ?

[lkraav]

ATM I do not know of anyone working on this. Are you sure that Ice Lake has an IPU4 ? I thought it was some early version of the IPU6 ?

Out of the blue, this announcement landed today https://github.com/endeavour/DellXps...ent-1961035146 hansdegoede

> "Intel is working on upstreaming a driver for the IPU6 on the linux-media mailing list, and I've done some of the work to port this to IPU4 (PCI 5a88). It does currently not support IPU4P (PCI 8a19) which this thread is about."

Andreas Helbech Kleist (not sure if he's on Phoronix) seems to perhaps have the ability to pull this backport off?