Re: iwlwifi: HE not negotiated on Wi-Fi 7 BE201 (iwlmld, kernel 7.0.9)

Hi Emmanuel,

Just a mail to say thank you sir, closing the loop on this one, with good news.

After upgrading to kernel 7.1.7-arch1-1 and linux-firmware 20260622
(the card now loads firmware 102.07fca168.0 bz-b0-fm-c0-c102.ucode,
up from 101.6e695a70.0 c101 in the original report), the problem no
longer reproduces here:

- Zero cTDP_CONFIG_CMD firmware crashes across the last 10 boots
(2026-07-26 through 2026-08-11). It previously fired on roughly
40% of boots.
- HE now engages on every association. The current link runs
HE-MCS 11 HE-NSS 2 (tx 1200 Mbps) on an 80 MHz 5 GHz channel.
- Band 4 (6 GHz) is exposed in `iw phy phy0 info` in the ZA
regdomain on every boot, which also resolves the 6 GHz issue I
reported separately at
https://www.google.com/url?q=https://lore.kernel.org/linux-wireless/CAAecyrURg5npc%[email protected]/&source=gmail&ust=1786520475419000&sa=E

This matches the correlation from my previous mail: with the boot
crash gone, both downstream symptoms (no HE, no 6 GHz) went with it.
One root cause after all.

Since I no longer get crash boots, I can't capture the trace-cmd
data or file a meaningful bugzilla ticket. If it ever recurs I'll
capture the trace and file it as discussed. As far as I'm concerned
both of my reports are resolved on 7.1.7 + firmware c102.

Thank you again for all your help! Much appreciated.

Kernel: 7.1.7-arch1-1
Firmware: 102.07fca168.0 bz-b0-fm-c0-c102.ucode op_mode iwlmld
linux-firmware: 20260622-1

--
Michael Blake
[email protected]


On Sun, 24 May 2026 at 08:50, Michael Blake <[email protected]> wrote:
>
> Hi Emmanuel,
>
> Thanks for the quick response.
>
> Before answering, one new observation that may unify the two issues
> you split into separate bugs: I have data across several boots
> suggesting the cTDP_CONFIG_CMD firmware crash, HE non-engagement, and
> the 6 GHz band invisibility (which I separately reported at
> https://lore.kernel.org/linux-wireless/CAAecyrURg5npc+coGw7nG58EMUoM2wu1rao0EGL3cV8CqcVA-g@mail.gmail.com/)
> all appear to be the same single problem.
>
> Across the last 5 boots on this BE201:
>
>   Boot       | cTDP crash | HE on assoc | 6 GHz Band exposed
>   -----------+------------+-------------+-------------------
>   today A    | no         | yes         | yes
>   today B    | no         | yes         | yes
>   yesterday  | no         | (untested)  | (untested)
>   yesterday  | YES        | no (VHT/HT) | no (only Band 2)
>   yesterday  | YES        | no (HT)     | no
>
> When the firmware boots clean (~60% of boots), HE engages on
> association and Band 4 (6 GHz) appears in `iw phy phy0 info`. When
> the cTDP crash fires (~40% of boots) the driver recovers enough to
> associate but HE is never negotiated and 6 GHz channels are not
> exposed to userspace.
>
> So there is most likely one root cause -- the post-crash recovery
> state -- not three independent bugs. Let me know whether you'd still
> prefer separate bugzilla entries (firmware crash and HE non-assoc) or
> a single one that frames them together. Happy to file either way.
>
> > Can you attach the full message?
> > Or even better, open a bug on bugzilla.kernel.org.
> > Pick the drivers network wireless intel as the componen.
>
> Full dmesg from boot 2026-05-23 16:21 (a crash boot) is attached as
> `firmware-crash-dmesg-2026-05-23-boot-3.log` (212 lines). I will also
> file this on bugzilla.kernel.org under
> drivers / network-wireless / intel and link the bug here once it's
> filed.
>
> > > Is HE expected to engage on BE201 with the current iwlmld driver?
> > > Happy to capture more detailed logs (mac80211 trace, iwlwifi debug
> > > masks, association IE dumps) or test patches if useful.
> >
> > That would be another bug.
> > You can run
> > trace-cmd record -T -e iwlwifi -e iwlwifi_dbg -e mac80211 -e cfg80211
> > and provide the trace.dat file it'll create.
>
> Will do. trace-cmd isn't installed yet on this machine -- I'll
> install it and capture trace.dat the next time the firmware crashes
> (can't easily reproduce HE non-engagement on a clean boot since HE
> engages by default there). Given the ~40% crash rate, a few reboots
> after install should produce a captured trace.
>
> Kernel/firmware unchanged from the original report:
>   Kernel:          7.0.9-arch1-1
>   Firmware:        101.6e695a70.0 bz-b0-fm-c0-c101.ucode op_mode iwlmld
>   linux-firmware:  20260410
>
> Thanks,
> --
> Michael Blake
> [email protected]