Just in case anyone is wondering - this is a local privilege escalation issue, not a remotely exploitable wifi vulnerability.

https://access.redhat.com/security/cve/cve-2022-1679

https://ubuntu.com/security/notices/USN-5544-1 would be a better submission URL, IMO.

Yes, we should refrain linking to Packetstorm. Also, it isn't Ubuntu specific.

I looked into various sources and NIST's NVD had the most complete info [1]. In this specific case, better than Mitre. Though this could change.

Its also a bug from May 2022.

[1] https://nvd.nist.gov/vuln/detail/CVE-2022-1679

Given that this is the ath9k driver issue in particular, it should be https://ubuntu.com/security/CVE-2022-1679 dated May instead?

I had an ath9k driver on my dell xps 13 9370 with ubuntu. Got kernel hangs once a week and ath9k in syslog was always the culprit. Never was able to fix it and many others were affected by the same thing. Switched to windows and still have occasional soft hangs, but its quickly resolved by closing and opening the lid. Dell going with a qualcomm wireless SoC is a major reason I would avoid buying from them again.

Dell tweaks the hardware configs if you specify Linux when you purchase it, usually switching the wifi controller to Intel.

I ordered the Linux dev edition with Linux pre-installed. I still got the ath9k unfortunately.

If it's on a mini-pcie card you may be able to swap.

It's soldered in.

I am using an asus laptop with linux (5.10.0) and I have this problem. If I physically move the laptop the network manager crushes, the only way to resolve it is by rebooting. I have been told that if I downgrade my kernel it will resolve this issue.

For the record, WiFi chipset in Dell XPS 13 9380 (next gen than yours) used ath10k driver and the one in 7390 (two gens above) switched to iwlwifi. Dunno about newer ones.

It sounds like this machine is in the past for you, but these cards are often user-replaceable. (I am not sure how to obtain the card though.)

You can buy them off Amazon. I did that to replace a wonky Realtek with an Intel.

So, why don’t we have a separate kernel (a VM really) handling WiFi hardware and perhaps some other network stuff? For a laptop the overhead shouldn’t matter match.

Because that would take work; while you're at it why not switch to a decent microkernel and run all of these buggy-as-hell drivers in their own services. I have devices which I have 99% of the time on the kernel blacklist but enable when I need them as the system won't stay up reliably for more than a day or two with them enabled.

Thats what wifibox does on FreeBSD (though I believe the reason is more for drivers than for isolation): https://github.com/pgj/freebsd-wifibox

Literally what qubes os does

The only wifi devices that have not caused driver issuese for me on linux are the Intel ones. Realtek and Atheros drivers are so buggy it's crazy.

I believe that's because Atheros chipsets are intended more to operate in Master mode (e.g., as a wireless access point). For that, it's been pretty gd stable for me.

Ironically, I think, some purists choose ath9k because it does not depend on closed firmwares.

The ath9k purists will have their drivers fixed while the people relying on closed source drivers will be ignored as the company decided to stop maintaining the driver. How... ironic?

Yeah it is pretty crazy how no WiFi chipsets with open source firmware exist that operate on the 5 GHz band. And even the old WiFi 4 ath9k chips aren't really made anymore. If you want to use a fully open source WiFi stack, you max out at 150 Mbps!