Re: does not connect with intel 3168 past kernel 6.16

[James Prestwood <prestwoj@gmail.com>]

Hi Rosen,

On 2/19/26 4:26 PM, Rosen Penev wrote:
> On Thu, Feb 19, 2026 at 10:03 AM James Prestwood <prestwoj@gmail.com> wrote:
>> Hi Rosen,
>>
>> On 2/11/26 3:15 PM, Rosen Penev wrote:
>>> Output from 6.17 dmesg:
>>>
>>> iwlwifi 0000:08:00.0: enabling device (0000 -> 0002)
>>> iwlwifi 0000:08:00.0: Detected crf-id 0x0, cnv-id 0x0 wfpm id 0x0
>>> iwlwifi 0000:08:00.0: PCI dev 24fb/2110, rev=0x220, rfid=0xd55555d5
>>> iwlwifi 0000:08:00.0: Detected Intel(R) Dual Band Wireless-AC 3168
>>> iwlwifi 0000:08:00.0: loaded firmware version 29.0bd893f3.0
>>> 3168-29.ucode op_mode iwlmvm
>>> iwlwifi 0000:08:00.0: base HW address: xx:xx:xx:xx:xx:xx, OTP minor version: 0x0
>>>
>>> There's also iwd dmesg spam:
>>>
>>> invalid HE capabilities for xx:xx:xx:xx:xx:x1
>>> invalid HE capabilities for xx:xx:xx:xx:xx:x0
>>> event: connect-info, ssid: X, bss: xx:xx:xx:xx:xx:x1, signal: -76, load: 52/255
>>> event: state, old: autoconnect_quick, new: connecting (auto)
>>> event: authentication-timeout,
>>> event: connect-failed, status: 1
>>> event: connect-info, ssid: X, bss: xx:xx:xx:xx:xx:x0, signal: -78, load: 96/255
>>> event: authentication-timeout,
>>> event: connect-failed, status: 1
>>> event: state, old: connecting (auto), new: disconnected
>>> event: state, old: disconnected, new: autoconnect_full
>>> invalid HE capabilities for xx:xx:xx:xx:xx:x1\
>> Is there any kernel output during this time?
> It's basically this continuously. I use impala to manage iwd but even
> if remove it and use iwctl manally, same thing.
>>> 6.16 output:
>>>
>>> iwlwifi 0000:08:00.0: enabling device (0000 -> 0002)
>>> iwlwifi 0000:08:00.0: Detected crf-id 0x0, cnv-id 0x0 wfpm id 0x0
>>> iwlwifi 0000:08:00.0: PCI dev 24fb/2110, rev=0x220, rfid=0xd55555d5
>>> iwlwifi 0000:08:00.0: Detected Intel(R) Dual Band Wireless-AC 3168
>>> iwlwifi 0000:08:00.0: loaded firmware version 29.0bd893f3.0
>>> 3168-29.ucode op_mode iwlmvm
>>> iwlwifi 0000:08:00.0: base HW address: xx:xx:xx:xx:xx:xx, OTP minor version: 0x0
>>> iwlwifi 0000:08:00.0: Registered PHC clock: iwlwifi-PTP, with index: 1
>>>
>>> Note the extra PHC clock message.
>>>
>>> This is still an issue with kernel 6.18. I work around this by using a
>>> USB WiFi adapter, which is not ideal.
>> So the USB adapter does work with 6.18? Do you have any other adapters
>> to try? Does wpa_supplicant + iwlwifi adapter work on 6.18?
> Not sure, I can try later.
>> Trying to determine where the issue is. To me, if a USB adapter works
>> that tells me its not necessarily an IWD issue, but something with
>> iwlwifi and that kernel version. I'd be most interested if
>> wpa_supplicant + iwlwifi works, if so we can try getting some iwmon
>> logging of the two setups to see what differs.
> It's definitely a kernel issue introduced between 6.16 and 6.17. Still
> present on 6.18.
>
> Interestingly enough, If I boot to Windows and then reboot to Linux,
> it works. But booting directly to Linux fails. That just tells me
> firmware loading is broken in some way. Or something is not being
> initialized.
This all smells like a driver regression to me. Or at the very least 
something in iwlwifi/mac80211 changed, and IWD is passing or not passing 
some attribute that is now expected/unexpected. If you can successfully 
get wpa_supplicant to work with this setup I'd be very curious to see 
the iwmon logs with wpa_supplicant vs IWD.
>> Thanks,
>>
>> James
>>