Bug 215523

Created attachment 300306 [details]
complete dmesg output

I can confirm the problem. With Intel AX210 neither wifi nor bluetooth work.
This also affects the vanilla kernels 5.5.18 and 5.6.x!
with kernel 5.14.12 it works.

Sorry, I meant the problem also affects kernels 5.15.18 and 5.16.x!

In kernel 5.16.4 the problem persists.

Bluetooth seems to work, but generates the following errors:

[   49.898869] Bluetooth: RFCOMM TTY layer initialized
[   49.898879] Bluetooth: RFCOMM socket layer initialized
[   49.898884] Bluetooth: RFCOMM ver 1.11
[   51.610867] Bluetooth: hci0: Failed to read codec capabilities (-56)
[   51.611863] Bluetooth: hci0: Failed to read codec capabilities (-56)
[   51.612954] Bluetooth: hci0: Failed to read codec capabilities (-56)
[   51.613865] Bluetooth: hci0: Failed to read codec capabilities (-56)
[   51.614945] Bluetooth: hci0: Failed to read codec capabilities (-56)
[   51.615897] Bluetooth: hci0: Failed to read codec capabilities (-56)


For me Intel AX210 wifi works with the following temporary solution:

Replace the intel wifi driver with the version from kernel 5.14.21 - this will work.

Example:
1. rename linux-5.16.4/drivers/net/wireless/intel/iwlwifi, e.g. to iwlwifi.bak or delete it.
2. download the latest kernel 5.14.21. 
3. copy linux-5.14.21/drivers/net/wireless/intel/iwlwifi to linux-5.16.4/drivers/net/wireless/intel/iwlwifi
4. recompile the kernel.

Created attachment 300416 [details]
complete dmesg output

Problem also happens with 5.17.0-rc

Also happens with  firmware version 68.01d30b0c

Just noticed this bug, I wonder if this is a dupe of Bug 215488, which is to be fixed by https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=master&id=92883a524ae918736a7b8acef98698075507b8c1

It is probably a separate issue, because my kernels are compiled with

CONFIG_WLAN_VENDOR_INTEL=y
# CONFIG_IPW2100 is not set
# CONFIG_IPW2200 is not set
# CONFIG_IWL4965 is not set
# CONFIG_IWL3945 is not set
CONFIG_IWLWIFI=y
CONFIG_IWLWIFI_LEDS=y
# CONFIG_IWLDVM is not set
CONFIG_IWLMVM=y
# CONFIG_IWLWIFI_BCAST_FILTERING is not set

so broadcast filtering isn't compiled in.

(In reply to Udo Steinberg from comment #9)
> It is probably a separate issue, because my kernels are compiled with
> […]
> so broadcast filtering isn't compiled in.

But your dump mentioned the "ADVANCED_SYSASSERT" stuff mentioned in the compile, so I'd say it's worth testing the patch. If it doesn't help, I'll add this regression to the tracking

Asking the developers in the other bug for advice might be worth a shot as well.

I applied the patch from https://bugzilla.kernel.org/show_bug.cgi?id=215523#c8 to Linux-5.15.22, but the problem persists.

Created attachment 300422 [details]
Disable TWT from client

Hi,

I wonder why i missed this bug from my query. will check.
anyway, this is a TWT related issue, mainly indicates a interoperability issue 
with the AP.
as this feature is not yet well enough supported in the echo system, i'd like to ask you to check the attached patch, and if it works for you i'll upstream it till the feature is better tested in the echo-system.

(In reply to Golan Ben Ami from comment #13)

many thx for looking into this!

> I wonder why i missed this bug from my query. will check.

Side note: to me it looks like bugs submitted against "Drivers->network-wireless-intel" are not mailed to any developers(¹). Is this known or did I misdiagnose things? 

(¹) it at least looks like it to me, when I first commented here only the reporter and the person that wrote comment 2 got a mail according to the info bugzilla showed.

We have automation creating tickets in out internal Intel DB for better tracking.
we track this query:
https://bugzilla.kernel.org/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&bug_status=NEEDINFO&component=network-wireless-intel&f1=short_desc&f2=short_desc&f3=short_desc&f4=short_desc&f5=short_desc&f6=short_desc&f7=short_desc&list_id=1105640&o3=notsubstring&product=Drivers&product=Networking&query_format=advanced&short_desc=Intel%2Ciwlwifi&short_desc_type=anywordssubstr&v3=dvm

I assume we missed this bug because the summary doesn't consist iwlwifi/intel.
we should probably relax our query.

> I assume we missed this bug because the summary doesn't consist
> iwlwifi/intel.
> we should probably relax our query.

I think it would be good to filter on the component field being "network-wireless-intel".

(In reply to Golan Ben Ami from comment #13)
> Created attachment 300422 [details]
> Disable TWT from client
> anyway, this is a TWT related issue, mainly indicates a interoperability
> issue with the AP. as this feature is not yet well enough supported in the
> echo system, i'd like to ask you to check the attached patch, and if it works
> for you i'll upstream it till the feature is better tested in the
> echo-system.

With this patch I haven't seen the issue reappear for 8 hours, so I think your analysis is spot on. Please go ahead and upstream that patch. Thanks!

will do, marking this as resolved in the meanwhile.

(In reply to Golan Ben Ami from comment #18)
> marking this as resolved in the meanwhile.

I've seen you doing this in another bug as well. Is this really wise? That makes it even harder for users that run into the issue now to find it in this tracker, as only "open" bugs are searched by default. And there is always a risk that the patch gets forgotten for some reason. I'd say it should only be closed if the fix is mainlined.

(In reply to Golan Ben Ami from comment #18)
> marking this as resolved in the meanwhile.

What's the status of this? As the culprit apparently was backported it would be good to get this fixed rather sooner than later. On a quick search I couldn't find a patch on lore that looked similar to the one attached here about a week ago.

Tend to agree, i'm re-opening till Luca will upstream the patch.

The patch was sent out now:

https://lore.kernel.org/linux-wireless/20220301072926.153969-1-luca@coelho.fi

I'll close this bug.