Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

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

bugzilla.kernel.org

Post reply on HN
Print view View original

101–110 of 148 posts

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

#101

Post by csdvrx »

phoronixrly wrote: post #63
stefan_ wrote: post #50

Earlier quoted context omitted.

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 availa…

Can I just point out how insane this method of determining the regulatory rules is... Considering the average user would never bother to set their correct regdomain when they buy a crap Chinese wifi router, you are pretty much guaranteed to always be using the wrong (mostly CN) regdomain, because that's what your neighbours are using...

> you are pretty much guaranteed to always be using the wrong (mostly CN) regdomain

Exactly. The software should make it easier for the average user to do the right thing. Most people would select "US" in a dropdown menu if offered the chance. The windows driver could also pass the information to the firmware using geolocalization or the windows locale settings if what the firmware setting suggested by default seems awfully wrong (ex: CN regdomain while Windows locale = US, geo IP = US, GPS=US etc)

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

#102

Post by jrockway »

magila wrote: post #77
eikenberry wrote: post #21

Earlier quoted context omitted.

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

Qualcomm's Atheros descended chips are generally the best bet for Linux support, especially for use in an AP. In other words you want a chip supported by the ath10k or ath11k driver depending on which generation of wifi you want. 10k = 802.11ac, 11k = 802.11ax.

ath10k isn't current hardware, though. We used that for Google Fiber wifi back in 2015. The world now has "Wifi 6" and other such nonsense.

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

#103

Post by stametseater »

jeroenhd wrote: post #98
stametseater wrote: post #95

Earlier quoted context omitted.

In my experience, tearing only occurs with Intel GPUs if you're using X without a compositor. With a compositor it isn't an issue.

Compositors still have issues in some setups (mine, specifically: Intel + Nvidia + mux chip). On most Linux laptops with no dedicated GPU, Intel GPUs seem to work fine, though.

Ah I see, I have no experience with dual-GPU setups.

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

#104

Post by csdvrx »

Avamander wrote: post #68
stefan_ wrote: post #50

Earlier quoted context omitted.

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 availa…

Those problems are valid, but LAR is currently often unable to detect the correct location. Even though it seems to fail to a very restrictive default, it might not, potentially causing even worse problems than a (rare) manual override would. It also doesn't seem to be that much of an issue for other vendors to allow the user to configure their regulatory locale, it definitely raises the question how necessary this c…

> In the end, Intel's current LAR implementation hurts users that want to use their hardware within their locale's regulatory limits. It does not prevent intentionally misleading the LAR implementation or acquiring other equipment to be disruptive. To me it seems to achieve very little positive, and I've not even touched the freedom aspect of Linux being violated.

Fully agreed with all this. If there's no existing patch for the 6.2 kernel I'll make my own later today because for whatever reason my Intel wifi card thinks it's not in the US (!!!) and I want it to obey the FCC limits.

# iw reg get

global

country 00: DFS-UNSET

        (2402 - 2472 @ 40), (6, 20), (N/A)

        (2457 - 2482 @ 20), (6, 20), (N/A), AUTO-BW, PASSIVE-SCAN

        (2474 - 2494 @ 20), (6, 20), (N/A), NO-OFDM, PASSIVE-SCAN

        (5170 - 5250 @ 80), (6, 20), (N/A), AUTO-BW, PASSIVE-SCAN

        (5250 - 5330 @ 80), (6, 20), (0 ms), DFS, AUTO-BW, PASSIVE-SCAN

        (5490 - 5730 @ 160), (6, 20), (0 ms), DFS, PASSIVE-SCAN

        (5735 - 5835 @ 80), (6, 20), (N/A), PASSIVE-SCAN

        (57240 - 63720 @ 2160), (N/A, 0), (N/A)
I've tried but failed to add regulatory.db to my bzImage:

# dmesg | grep regula

  cfg80211: Loading compiled-in X.509 certificates for regulatory database

  platform regulatory.0: Direct firmware load for regulatory.db failed with error -2

  cfg80211: failed to load regulatory.db
In my .config:

CONFIG_EXTRA_FIRMWARE="iwlwifi-so-a0-gf-a0-72.ucode iwlwifi-so-a0-gf-a0.pnvm regulatory.db (...)

regulatory.db exists in the expected path and the md5sum 968432874991bf18d99a4f508fb1515d seems correct.

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

#105

Post by devanl »

dcow wrote: post #85
stefan_ wrote: post #76

Earlier quoted context omitted.

Keep in mind that for stations only (e.g. mobile devices most of the time), the rules are relaxed: if the station sees an AP operating on some channel, it is allowed to assume that it can operate on that channel. And of course passively scanning a channel to check for the presence of an AP is not transmitting, so no regulatory problem there. But don't worry, for 6 GHz the FCC has come up with even more insane stuff:…

Why does the FCC even care so much? What nefarious things can the average consumer possibly do that a motivated attacker can't already do? A motivated attacker who wants to change all the street lights to green will have hardware that lets them do whatever they want. The other 99.999% of people will just use their system as configured by their software vendor, legally. Why do we need hardware failsafes? Why do we nee…

Setting aside issues that can be safety related where consumer equipment overlaps with legacy equipment, there is a tragedy of the commons effect to not attempting to regulate radio power limits.

It's incredible how well WiFi works for the amount of power it uses.

Obviously turning up the transmit power makes it better... for that one person. For everyone else, that channel has a little bit more noise than before, making it harder to receive their signal.

So maybe their neighbors will also want to increase their transmit power to get better range / speed again. Having a limit across the board for all mass manufactured devices prevents an escalating spiral of vendors selling 30, then 40, then 50dbm etc routers.

As mentioned elsewhere, different countries have different power limits, but it's more economical to make a single radio for all markets with software power limits. One hypothetical way for a vendor to get a market advantage is to sell a radio that is software limited, but wink wink can be patched with easily googled instructions to increase the power to work better. Maybe by downloading a tool from some sketchy website that even actually works and can be spammed across the internet or social media.

So the FCC has to strongly discourage anything that could lead to lots of radios deliberately exceeding the regulatory limits and disproportionately making the spectrum worse compared to a compliant device.

I don't think that necessarily justifies rather invasive schemes like geolocated AFC, but preserving the use of the radio spectrum so that everyone can make efficient use of it is the FCC's mandate.

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

#106

Post by jeroenhd »

antisthenes wrote: post #82
sadlywebdev wrote: post #25

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.

> WiFi-Chipset contained a bug causing 2ghz WiFis on channel 13 to not be visible. Don't use non-standard channels. You're making it worse for everyone.

Channels 12+ are perfectly valid for WiFi use in some jurisdictions. In fact, they're useful exactly because they mostly transmit outside the range of normal WiFi channels.

802.11n works perfectly fine with channels 1+5+9+13, letting the guard bands intersect. The 1/6/11 set common to the USA leaves plenty of guard space to prevent overlaps, but wastes bandwidth in jurisdictions where channel 12 and 13 are disallowed.

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

#107

Post by mjevans »

jodrellblank wrote: post #73
mjevans wrote: post #56

Earlier quoted context omitted.

But what if someone moves or simply uses hardware in a different region that's sold? What if the laws change? The correct 'work around' is to build a kernel with a DEFAULT regulatory domain, abstract from any vendor's specific implementation, allow that domain to be updated at runtime (particularly for mobile devices), and follow the laws there in. I also advocate (in this post) for wifi_regulations=BREAKTHELAWANDDIS…

If you aren't breaking any laws transmitting like that in International Waters, why call it BREAKTHELAW and call out that use case? If you are breaking laws transmitting like that in International Waters, why would this setting be called out as specifically useful in International Waters? Neither seems to make sense. (And does this come from the Microsoft team who came up with setup.exe /IAcceptExchangeServerLicenseT…

It doesn't make sense to call the same option multiple things for such a small number of use cases; you really want the last thing, but someone might not know that in _many_ circumstances it could be against the law to use it, hence the prefix to document that.

Edit: ALSO, in case it wasn't clear... E.G. wifi_regulations=USA or wifi_regulations=CDN or wifi_regulations=GBR or some other set of country codes / match in the table of known regulations. A longer string that manually specifies allowed frequencies in some way might also work.

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

#108

Post by csdvrx »

jiggawatts wrote: post #91
dcow wrote: post #85

Earlier quoted context omitted.

Why does the FCC even care so much? What nefarious things can the average consumer possibly do that a motivated attacker can't already do? A motivated attacker who wants to change all the street lights to green will have hardware that lets them do whatever they want. The other 99.999% of people will just use their system as configured by their software vendor, legally. Why do we need hardware failsafes? Why do we nee…

The FAA raised a huge stink recently because they weren’t ready for 5G rollout and were worried that the altimeter radar would give false readings on approach to landing. Agencies — even when in the wrong — can throw their weight around and demand that the FCC ban things. Consumers like you and me have no real representation in government so we have no political clout and can’t throw our weight around to demand simpl…

> The FAA raised a huge stink recently because they weren’t ready for 5G rollout and were worried that the altimeter radar would give false readings on approach to landing.

If the FAA didn't do its job and certified altimeters that shouldn't have been, as the 5G rollout was announced and previsible, maybe they should start working instead of complaining that others help them do their job?

Also, if all it takes to mess with altimeters is a rogue 5G cell, the right approach is to harden the altimeters, instead of hoping no 5G cell ever will be misconfigured or pushed wrong configuation parameters by a enemy country or ransomware hacker group.

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

#109

Post by csdvrx »

devanl wrote: post #105
dcow wrote: post #85

Earlier quoted context omitted.

Why does the FCC even care so much? What nefarious things can the average consumer possibly do that a motivated attacker can't already do? A motivated attacker who wants to change all the street lights to green will have hardware that lets them do whatever they want. The other 99.999% of people will just use their system as configured by their software vendor, legally. Why do we need hardware failsafes? Why do we nee…

Setting aside issues that can be safety related where consumer equipment overlaps with legacy equipment, there is a tragedy of the commons effect to not attempting to regulate radio power limits. It's incredible how well WiFi works for the amount of power it uses. Obviously turning up the transmit power makes it better... for that one person. For everyone else, that channel has a little bit more noise than before, ma…

> It's incredible how well WiFi works for the amount of power it uses.

Actually, WiFi disproves that we'd have a tragedy of the commons.

> Having a limit across the board for all mass manufactured devices prevents an escalating spiral of vendors selling 30, then 40, then 50dbm etc routers.

No, the AP and the device both have a strong interest to limit the power used: not just to limit interference with other devices inside the home, but also to increase battery life!!

> but wink wink can be patched with easily googled instructions to increase the power to work better

I just don't understand why most posters here assume the worst by default. Most people are nice and want to obey the law.

WiFi proved that very little legal oversight was necessary to make it work.

In fact, it did the opposite: by comparing the efficiency of the use of the 2.4Ghz band (noisy with microwaves etc) to the rest of the spectrum managed by the heavy hand of the FCC, any reasonable person would argue for removing regulations or more parts of the spectrum (starting maybe with the huge chunks waste on HAM radio!)

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

#110

Post by jeroenhd »

stametseater wrote: post #103
jeroenhd wrote: post #98

Earlier quoted context omitted.

Compositors still have issues in some setups (mine, specifically: Intel + Nvidia + mux chip). On most Linux laptops with no dedicated GPU, Intel GPUs seem to work fine, though.

Ah I see, I have no experience with dual-GPU setups.

Before this laptop, I used an Intel+AMD laptop which worked fine (except for AMD's driver messing up in suspend, because the architecture was uncommon and left behind in the middle of a driver model switch).

"Avoid Nvidia; if you have to use Nvidia, avoid other brands if possible" seems to be the way to go.

I'm still mad about Nvidia setting an arbitrary limit to the amount of X11 displays. There is no good reason for it other than locking their customers into data center GPUs.

Post reply on HN