Re: Unable to have network reliably start at boot time

From: Bjoern A. Zeeb <>
Date: Mon, 14 Mar 2022 19:36:51 UTC
On Mon, 14 Mar 2022, Kevin Oberman wrote:

> On Mon, Mar 14, 2022 at 10:47 AM Bjoern A. Zeeb <
>> wrote:
>> On Mon, 14 Mar 2022, Kevin Oberman wrote:
>> Hi,
>>> I have tried loading if_iwlwifi at boot by both /boot/loader.conf and
>>> /etc/rc.conf. In the /boot/loader case, the firmware fails to load with
>> the
>>> message "File size way too small!" for all available versions of
>> microcode
>>> including "wlwifi-cc-a0-68.ucode" which should be used for my AX200. For
>> Loading from loader is discuraged;  if you really want to see this
>> thread:
>> On FreeBSD Current, you don't even have to load the module at all anymore;
>> it'll automatically loaded for you (soon also in stable/13).
>>> /etc/rc.conf, there appears to be a race condition where the network
>> tries
>>> to start before the microcode load has completed.
>> That is a weird case and I thought that should not happen anymore as
>> the wlan interface should always only be created after the firmware
>> successfully loaded.
>> I see I have an email from you in my inbox as well; in case it's not
>> in there, can you send me the console (dmesg -a) output of that case
>> so I can have a look?
>> /bz
>> --
>> Bjoern A. Zeeb                                                     r15:7
> I'm only assuming that the firmware load is leading to the problem, but
> I've looked closer and that is likely not the case as I see:
> iwlwifi0: <iwlwifi> mem 0xd1300000-0xd1303fff at device 0.0 on pci2
> iwlwifi0: successfully loaded firmware image 'iwlwifi-cc-a0-67.ucode'
> iwlwifi0: api flags index 2 larger than supported by driver
> iwlwifi0: TLV_FW_FSEQ_VERSION: FSEQ Version:
> iwlwifi0: loaded firmware version 67.8f59b80b.0 cc-a0-67.ucode op_mode
> iwlmvm
> iwlwifi0: Detected Intel(R) Wi-Fi 6 AX200 160MHz, REV=0x340
> and, after the network fails to start, I see:
> iwlwifi0: Detected RF HR B3, rfid=0x10a100
> iwlwifi0: base HW address: 6c:6a:77:ed:ce:d9
> The failure generates no messages. Just shows git0 and lo0 starting
> normally. I assume that this is because wlan0 has not been created at this
> point. While there are many lines of messages in the log between these
> messages, they are all logged in the same second.

I had a look.  You are on stable/13, right?

If so, I just asked for e0ab0ff104cb0e817d47a1a2a235f915ec1a31f4 to be
MFCed as that likely prevents your wlan from starting if the wlan
interface is not there yet by the time we run (the "global") netif start.

If this was on HEAD please let me know and I'll have to dig deeper.


Bjoern A. Zeeb                                                     r15:7