> Date: Thu, 21 Aug 2025 13:02:08 +0000 > From: Klemens Nanni <[email protected]> > > 21.08.2025 13:22, Mark Kettenis пишет: > > Hmm. We've seen that before on ther SoCs. Typically that happened > > when essential clocks were gated, the device was still in reset or the > > USB PHY wasn't initialized properly. > > > > Can you send me the eeprom -p output for this machine? > > See 'eeprom -p | gzip' attached.
Thanks, that confirms what I thought. The ehci(4) and its ohci(4) companion use the same clocks, and PHY. And ehci(4) works it seems. Can you try adding some printf's in dev/sys/ohci.c:ohci_fdt_attach() to narrow down where it hangs? > > FWIW, I also see all-zeo MAC addresses on rge0 and rge1, only dwqe0 has one. That is pretty normal for this type of machines. There is a way to make u-boot assign addresses based on the serial number of the board. That requires modifications to the device tree though, and I'm not sure upstream will accept those. Easiest thing to do is to put something like: lladdr 00:11:22:33:44:55 in your /etc/hotsname.rge0. I usually add the random MAC address that the kernel generates in there after the first boot.
