Re: HT and VHT testing for modern iwlwifi chipsets (notes about others)

Hello Bjoern,

My wifi card (AX201) doesn't seem to connect to a 802.11n network; it 
only uses 802.11a

Here's the card itself:

iwlwifi0@pci0:0:20:3:   class=0x028000 rev=0x00 hdr=0x00 vendor=0x8086 
device=0x02f0 subvendor=0x8086 subdevice=0x0070
     vendor     = 'Intel Corporation'
     device     = 'Comet Lake PCH-LP CNVi WiFi'
     class      = network

% grep AX201 /var/log/all.log
Mar  1 03:34:07 hamster kernel: iwlwifi0: Detected Intel(R) Wi-Fi 6 
AX201 160MHz

Here's my OS info:

% uname -a
FreeBSD hamster 15.0-CURRENT FreeBSD 15.0-CURRENT #16 
main-n275691-5b511473999a: Wed Feb 26 20:55:54 +05 2025 
root@hamster:/usr/obj/usr/src/amd64.amd64/sys/DEBOOG amd64

Here's what I chnged in loader.conf:

% grep linuxkpi /boot/loader.conf
compat.linuxkpi.80211.hw_crypto=1
compat.linuxkpi.iwlwifi_disable_11ac=0
compat.linuxkpi.iwlwifi_11n_disable=0

Another symptom would be that iwlwifi_disable_11ac doesn't seem to set to 0:

% sysctl compat.linuxkpi.iwlwifi_disable_11ac
compat.linuxkpi.iwlwifi_disable_11ac: 1

On 01.03.2025 02:35, Bjoern A. Zeeb wrote:
> Hi,
>
> upfront: rtw88/rtw89 and older iwlwifi chipsets all need more work which
> will continue. rtw88/rtw89 need LinuxKPI and net80211 changes, modern
> iwlwifi chipsets will need a net80211 change for a clean solution which
> is currently hardcoed in LinuxKPI, and older iwlwifi chipsets still need
> driver changes due to GPL-only code.
>
> If you have an iwlwifi AX2xx or BExxx and you have successfully tested
> HW_CYRPTO (see the other email thread here on wireless) then you have a
> chance to test a first cut of HT20/40 and VHT20/40/80 (no 160 yet)
> support by adding the following two tunables next to your hw_crypto one
> in loader.conf:
>
> compat.linuxkpi.iwlwifi_11n_disable=0
> compat.linuxkpi.iwlwifi_disable_11ac=0
>
> If you try this with any other (older) iwlwifi chipset (anything with
> just numbers 3xxx, 7xxx, 8xxx, 9xxx etc.), with rtw88 or rtw89 it will
> fail and I'll not provide support at this time unless you want to help
> to make it work.  Otherwise you'll have to wait a few days/weeks
> longer -- sorry.
>
> Also if you try this without the hw_crypto tunable on, it'll fail badly,
> so please test that (independently) *first* if you haven't as that will
> be my first question.
>
> I am sure there'll be plenty of rough edges to fix for me... but I
> wanted to get the code out at this point.
>
> The last bits of the code have been in stable/14 since yesterday [1] and
> main since [2] (beware that main has more wireless changes I haven't
> rebased to yet for extra testing but they should not interfere).
>
> Once we have sufficient confidence that things work well enough I will
> likely add the tunables to the firmware ports so it'll do the right
> thing out of the box like was done for the rtw88 skb tunable and
> eventually slip the policy once the majority works.
>
>
> If things work the thanks should go to The FreeBSD Foundation,
> if things don't work the problem reports go to me.
>
> Happy testing and a blessed weekend,
> Bjoern
>
> [1] stable/14 
> https://cgit.freebsd.org/src/commit/?h=stable/14&id=63c197298957c89fe3e7bd0c3392da74ed72a24b
> [2] main: 
> https://cgit.freebsd.org/src/commit/?id=2c44f1ff6935e93f7774e526f2f1ab0a2812eab5
>
-- 
wbr, Maxim Filimonov <che@bein.link>