NetBSD on HardKernel Odroid M1S

[Jeff Frasca <thatguy%jeff-frasca.name@localhost>]

I've been working on getting NetBSD running on the Odroid M1S:
https://wiki.odroid.com/odroid-m1s/hardware/start

It's an rk3566 board like the Quartz64, so I've been using the GENERIC64
kernel and using Jared McNeill's UEFI firmware.

I put my current WIP code here:
https://git.sr.ht/~phaedrus/quartz64_uefi/tree/odroid-m1s

Most of the heavy lifting is done in the board init driver:
https://git.sr.ht/~phaedrus/quartz64_uefi/tree/odroid-m1s/item/edk2-rockchip/Platform/HardKernel/Odroid-M1S/Drivers/BoardInitDxe

If anyone wants a build, let me know, I'll find a place to post them and drop
a link.  (Getting a working dev environment is a bit of an adventure right
now.)

I haven't posted builds because it's not fully functional.  The thing that is
not working is getting the ethernet to talk to the PHY.  I get this output in
dmesg on both 10.1 and 11_BETA:

[   1.0000040] eqos0 at acpi0 (MAC1, PRP0001-3): mem
0xfe010000-0xfe01ffff irq 64,61
[   1.0000040] eqos0: DesignWare EQOS ver 0x51 (0x30)
[   1.0000040] eqos0: Ethernet address 8e:0d:79:1a:a0:82
[   1.0000040] eqos0: reset timeout!

As best as I can tell, the software reset that's timing out is waiting for
the PHY clock to reset, and that's not happening.  If I comment out the reset
code, the driver fails when probing the PHY.  The current version of my board
init tries to probe the PHY as well, and I just get 0x0000 back for both PHY
Id registers and the BMSR (which should _always_ be non-zero).

The Odroid M1S has an RTL8211F, which should be the same as the Firefly
ROC-RK3566-PC.  There's at least one dmesg for that board up on nycbug's
dmesgd: https://dmesgd.nycbug.org/index.cgi?do=view&id=6572

Which is the output I'd expect to see.

I know the PHY works, I've got the board booting ubuntu and chimera.  The
chimera sdcard uses my UEFI firmware to boot, and I can break the ethernet by
not running the init gmac routine (so I know _I_ am the one setting up the
pin functions, and linux isn't re-doing it).

Does anyone still have one of those Firefly boards running (actually, any of
the rk3566 boards)?  Does your ethernet work?  Have you updated to 10.1 or
11_BETA?

(No need to comment on the TianoCore code, I posted it for transparency
purposes.  I'll go track down Jared on gh later.  OTOH, if anyone here has
worked on that firmware and has any debugging tips or things to try, I would
love to hear it.)