Re: iwlwifi: i7260 does not load firmware properly/wifi not working

From: Bjoern A. Zeeb <>
Date: Mon, 21 Mar 2022 18:33:55 UTC
  This message is in MIME format.  The first part should be readable text,
  while the remaining parts are likely unreadable without MIME-aware tools.

Content-Type: TEXT/PLAIN; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8BIT

On Mon, 21 Mar 2022, Stefan Blachmann wrote:

> @"FreeBSDUser" maybe try with the iwm driver instead of the iwlwifi one?

I understood that iwm with lagg wasn't working well either for
"FreeBSD User" or did I misread this?

> It works fine for me. The only thing is the firmware loading thing:
> it does not work for me when I set in /boot/loader.conf
>    iwmfw_load="YES"
> which for some reason causes error messages about firmware not found.
> However, explicitly specifying
>    iwmXXXXfw_load="YES"
> works just fine.
> @Björn, is this normal (mistake on my side), or could it be considered a bug?

"iwmfw" is a catch-all to compile all iwm firmare into the kernel.

As both man iwm and man iwmfw state (ot my memory) you have to load the
individual firmware from loader directly if they are not compiled into the
kernel (either indidivdually or by catch-all);  there is no catch-all module.
We could add that but that would just load a lot of blobs into memory we
don't need.

In general I'd suggest to not load any WiFi drivers from loader
anymore these days (that's a relict of old times).  At least move to
kld_list=".." in rc.conf if the drivers do not properly support
auto-loading yet.  It's way faster to load the driver that way rather
than from loader.

I would hope that in the longer-term we get away from loading
non-essential drivers from loader (or compiling them into GENERIC) given we
have way better facilities these days (faster and more flexible).  One of
the reasons iwlwifi cannot even be compiled into the kernel anymore;  just
don't want to start it at all anymore unless someone knows a good reason it
is needed in the kernel before we go multi-user (*).


(*) One is securelevels but that hasn't come up yet...

Bjoern A. Zeeb                                                     r15:7