Re: rpi3b+ ue0 too late to be configured on boot?

On Sun, 1 Jan 2023, Mike Karels wrote:

> On 30 Dec 2022, at 16:16, Mike Karels wrote:
>
>> On 30 Dec 2022, at 16:03, Bjoern A. Zeeb wrote:
>>
>>> On Thu, 29 Dec 2022, Mark Millard wrote:
>>>
>>>> Bjoern did not report enough information for the configuration
>>>> of that EDK2 based context to replicate his boot context in a
>>>> test.
>>>
>>> Sorry, I applied the config.txt change from the open pull request there
>>> where I left a comment a year or so ago.  I was running with the
>>> previous version quite fine.
>>>
>>> I did disable ACPI for a change in the setup.  No idea how that would be
>>> persistent.
>>>
>>> I used to run with the previous version but loader now panics on that so I
>>> found the fairly recent new one.
>>>
>>> I also have my own kernel admittingly; same builds on years because I
>>> run them with a memory disk.
>>>
>>> The GE switch port are on auto but given this is USB device attaching
>>> time it seem that shouldn't make the huge difference (and I checked that
>>> it wa on an auto-port before and so is my other one).
>>>
>>> In either way, @mike, does your boot indiciate that ue0 is setup during
>>> the first netif run or is devd doing it later for you and it'd be there
>>> by the time you login)?  Just checking before I'll put it onto the TODO
>>> list to one day dig deeper.
>>
>> I think it was at the normal time (netif), according to the console
>> output.  I’ll double-check later.
>
> Yes, ue0 was detected before devd started.  However, this happened after
> root_hold_wait (in /etc/rc.subr, called from mountcritlocal) printed
> “Waiting 30s for the root mount holders: usbus1”, after which uhub1,
> ue0 and other things configured after a very short delay.

Thanks a lot Mike for checking again!  This may be the clue.

/bz

-- 
Bjoern A. Zeeb                                                     r15:7