Re: Need testers for ALPHA RPi-5 fan control, GE NIC, WiFi drivers for FreeBSD-CURRENT

On 6/5/26 19:46, Jeremy McMillan wrote:
> I remember some hair pulling around FDT supplied interfaces just not
> working, and resorting to copying the Linux code's behavior, which I
> suspect a close reading will show uses hardcoded addresses wherever the
> FDT stuff is missing or wrong.

I'll note that there is another kernel based on FDT (not ACPI) for RPi5
support these days: OpenBSD --not just mainline Linux and the RapsiOS
(my abbr.) kernels.

I've no clue if any of the OpenBSD code makes doing anything notably
easier. But it would be the case that if some code could be directly
used, the licensing should work out fine.

As I understand, OpenBSD too is based on the RPi* *.dtb* information,
not on mainline Linux *.dtb* information.

> 
> On Fri, 2026-06-05 at 12:46 -0700, Mark Millard wrote:
>> . . .
> 
> I'm game to refactor any hardcoded addresses and/or reliance on ACPI
> (which I suspect is just the UEFI dumping the same DTB data as ACPI
> enumeration) if you can help provide an overlay that exposes things
> properly.

The FreeBSD kernel has never tried to use all the RPi[234]B FDT
information via live extraction, instead hand coding much information
into the kernel.

I'd be really surprised if anyone actually objected to such, from what
I've seen of the prior RPi* related FreeBSD kernel code.

(I do get nervous about trying to mix mainline and RPi* *.dtb* related
sources of information without checking out any potential differences,
since my understanding is that the files, for example, are not the same
in various ways and FreeBSD historically targeted use of the RPI* *.dtb*
based context --not that I know the details of the differences.)

> The whole point of RPi as a platform is supposed to enable
> hardware modification by plugging modular stuff into i2c or spi
> interfaces, and it would be nice to keep that as modular as possible.

I also doubt anyone would object to more live extraction and use of FDT
information if degree of effort allows for such.

> 
> Thanks for digging into this! I'd be lost this deep.
> 
> 
> 


-- 
===
Mark Millard
marklmi at yahoo.com