Live data from Hacker News

Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

bugzilla.kernel.org

41–50 of 148 posts

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#42
post #35

It appears this is a non-US issue? "iw reg get" shows "country 00: DFS-UNSET" with 9 frequencies (incl. 755-928 MHz ... huh) and "country US: DFS-UNSET" with a long list going from 2402-7125 MHz on my Wi-Fi 6E AX211

I was confused by this too. 5 GHz channels with 40/80 MHz width definitely work for me, without having to set a regulatory domain. It appears that this problem is related to Intel's WiFi firmware automatic detection of the regulatory domain, which is LAR. This is done on the basis of nearby access points, although the specifics of how this works is beyond me. The problem is that if this gets detected incorrectly, the…

This indeed affects more than just the US.

A nearby AP is not always sufficient for LAR to work properly, making AP startup finicky and slow at best. While also causing the issue that scans later on might affect functionality (https://tildearrow.org/?p=post&month=7&year=2022&item=lar). I guess you could also end up with totally incorrect (and illegal) location detection, but that's less likely than it not working at all (leaving everything disabled).

With 6GHz in the mix, the likelihood of automatic detection failing to find any nearby relevant APs rises significantly. Breaking all IR (initiate-radiation) functionality, AP mode included.

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#43
post #35

Earlier quoted context omitted.

I was confused by this too. 5 GHz channels with 40/80 MHz width definitely work for me, without having to set a regulatory domain. It appears that this problem is related to Intel's WiFi firmware automatic detection of the regulatory domain, which is LAR. This is done on the basis of nearby access points, although the specifics of how this works is beyond me. The problem is that if this gets detected incorrectly, the…

> So I don't think this is a non-US issue, it's an issue for people who live in a place where nearby access points will teach the WiFi driver the wrong regulatory domain ... however that works. It's probably less common for this to happen in the US, but not impossible. Could it be as simple as somebody moving from Indonesia or wherever to America, and bringing their Indonesian/etc AP with them?

Possibly, but it could also be one of your neighbors, or a nearby AP enabling channels that are not supposed to be enabled in the US. The issue is that when Intel's drivers make the wrong decision about the regulatory domain, you don't usually have a lot of insight about why. Seems this issue affects Windows users as well.

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#44

A friend of mine had intermitent WiFi issues for a long time. One WiFi they could just never even find, until we figured out the Intel WiFi driver for their WiFi-Chipset contained a bug causing 2ghz WiFis on channel 13 to not be visible. It is a tiring process to identify issues like this in drivers. Really sucks when manufacturers just do not test their software.

Sounds like a misconfigured regulatory zone, TBH. Channels 12 and 13 are not legally usable in every country, including the USA: https://www.howtogeek.com/402142/why-wi-fi-channels-12-13-an...

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#45

I'm so relieved to find that its not just me going crazy all this time. Its gotten so bad that I've had to install a bloody MediaTek WiFi card in my ThinkPad just to make it work reliably. I remember when the 8000 series radios were the gold standard for mobile wireless, all the newer AX cards have been nothing but pain and suffering.

If it makes you feel better, I've had serious issues with AX cards on enterprise Windows ThinkPads.

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#47
post #6

I knew something was up. Goofy problems with wireless abound for a _long time_ with seemingly no solution.

"Goofy problems [...] abound for a _long time_ with seemingly no solution" is the norm in desktop Linux.

> "Goofy problems [...] abound for a _long time_ with seemingly no solution" is the norm in desktop Linux.

Hey, it's still better than Windows:) To quote a bit from upthread,

>>i should add that this happens on both windows and linux but at least on linux i had a way of disabling LAR so my regulatory domain gets detected properly and all the wifi channels/widths works as they should.

(https://news.ycombinator.com/item?id=35199660)

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#48

A friend of mine had intermitent WiFi issues for a long time. One WiFi they could just never even find, until we figured out the Intel WiFi driver for their WiFi-Chipset contained a bug causing 2ghz WiFis on channel 13 to not be visible. It is a tiring process to identify issues like this in drivers. Really sucks when manufacturers just do not test their software.

That’s a feature for outdoor radios so they don’t interfere with airplane radars. Not sure why it was in a laptop. As the other poster said it’s for some non-US countries.

My APs have an outdoor mode I was curious about and turns out it’s for plane radars. I was very confused when I turned it on. They were outside APs so I turned it on. I was looking at the channel interfaces and noticed missing channels.

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#49

I'm so relieved to find that its not just me going crazy all this time. Its gotten so bad that I've had to install a bloody MediaTek WiFi card in my ThinkPad just to make it work reliably. I remember when the 8000 series radios were the gold standard for mobile wireless, all the newer AX cards have been nothing but pain and suffering.

And Intel was the last reliable wifi chip maker I knew. What are good, well supported (on linux) wifi chipsets these days?

Mediatek and Qualcomm are pretty well supported afaict. Realtek is hit or miss. Broadcom should probably be avoided.

Re: Linux Intel WiFi driver broken with 5&6GHz bands for longer than three years

#50
post #19
post #2

>looks like lar_disable functionality was removed as of 5.5 kernel as a way to fix the firmware crash issue with iwlwifi, is that right? >[...] i've tried replacing the wifi adapter but this happens on intel wireless-ac 3165, 8265, and 9260 dual band 160Mhz (currently installed adapter). >[...] On Intel ax210ngw all 5ghz channels are disabled and cannot be used to transmit So it's iwlwifi specifically. >i should add…

Is there a patch for the kernel 6 line to add back the lar_disable functionality?

This was the patch that removed it:

https://github.com/torvalds/linux/commit/f06021a18fcf8d8a1e7...

But this commit message is burying the lede a bit. There are two problems here:

1) you can revert the change, but a large part of the regulatory logic lives in the firmware blob running on the chip and you can not change this.

2) the FCC doesn't look kindly upon functionality that ignores regulatory limits being available to end customers. It was already not legal for Intel to add this in the first place, and they certainly won't help you reinstate it.

The second part might be why these devices have been in a "broken" regulatory state forever now where they stay away from everything that touches upon e.g. DFS; someone at Intel might have realized that letting users pick their regulatory region is already not something they should be offering.

The reality is that these WiFi chips should probably have some fuse in them that decides what regulatory region they were made for and then run off that, but you can see the business people losing their mind over that: now you need to have a hundred different SKUs and worse you need to understand your distribution chain. So a software setting it is, and then you see why the FCC is upset.

Post reply on HN