Linux Kernel Patches Posted For USB Support With Realtek RTW88 WiFi Driver

[phoronix]

Phoronix: Linux Kernel Patches Posted For USB Support With Realtek RTW88 WiFi Driver

Back in 2019 Realtek contributed the open-source "RTW88" WiFi driver to the Linux kernel for supporting their newer wireless chipsets on Linux. To date the mainline driver has just supported their PCI-based WiFi adapters while a patch series now out for review extends the driver to supporting their newer USB-specific chipsets...

https://www.phoronix.com/scan.php?page=news_item&px=Linux-Patches-RTW88-WiFi-USB

[kiffmet]

rtw8821cu - nice! I got copys of 4 different out-of-tree repos to have at least one working driver for that hardware for a given kernel version. The wireless sticks themselves are cheap, dual antenna, AC-capable and perform very well, it's just the software situation that has been messy.

[skeevy420]

Funny. That image has the same WiFi stick I put in my Mom's PC.

Sadly, we bought her a Skylake PC about a week or two before all the TPM bullshit was announced. Good to know all her hardware will be supported on Linux whenever Windows 10 gets depreciated.

[zboszor]

What about the 88xxau chips?

[Linuxhippy]

Was this sponsored by realtek, or didn't they care at all?

[kiffmet]

zboszor They seem to need a different driver. Check out whether aircrack-ng or someone else does offer one for your dongle's specific chipset on github/gitlab.

Linuxhippy This is based on open-sourced Realtek wifi drivers. The thing with those is that they all stem from a common, Realtek internal codebase that is being forked, adapted and stripped of support for other devices before publishing it. So there is a multitude of open-source wifi drivers out there that are difficult to consolidate into one big driver due to version mismatches (a driver for one device may have been forked from internal version 4.0, while one for a different chip stems from 5.3.4, etc.), bad practice like device specific changes in common code files, and so on.

The code drops happen because a hardware vendor (i.e. someone who buys realtek chips in bulk to mass-produce wifi dongles) pays for it and the drivers tend to stop recieving official updates, depending on the popularity of the chips and the state of their product life-cycle. This allows Realtek to save costs by stopping to care about maintenance at some point, leading to community developers forking the code again and changing it to keep the drivers working with more recent kernel versions, which as a side effect increases divergence even further.

The situation is quite messy to say the least and upstreaming also needs certain coding style and quality criteria being respected. I just hope that RTW88 can serve as the base for this, with the community patching in support for more devices as time passes, despite all of the additional effort needed.

When not buying wifi hardware with Intel chipsets, checking the state of driver support before buying is unfortunately still pretty much needed. Qualcomm may be good aswell, Broadcom and Mediatek are hit and miss.

[Franco Castillo]

How forgotten is rtl8812au 😒😒😒

[kiffmet]

How forgotten is rtl8812au 😒😒😒

It's not too bad - even community support for 5.17 is a thing.

https://github.com/morrownr/8812au

https://github.com/svpcom/rtl8812au

https://github.com/aircrack-ng/rtl8812au

[Franco Castillo]

It's not too bad - even community support for 5.17 is a thing.

https://github.com/morrownr/8812au

https://github.com/svpcom/rtl8812au

https://github.com/aircrack-ng/rtl8812au

Yes, thanks to the community, but not the Kernel.

[Linuxxx]

Yes, thanks to the community, but not the Kernel.

Are you actually using that out-of-tree driver with an USB WiFi dongle housing said chip by Realtek?
If so, could you please have a look whether that driver is causing absurdly high CPU load on your system?

I'm asking because I once tried that community driver, and it was causing around 25 % CPU load on one core with an Intel i5-6500, which is simply unacceptable, thus useless.