Driver MT7902 Fix Wifi

Hello everyone,

I would like to share a repository I created to fix the MediaTek MT7902 driver on Fedora 44 Plasma. I developed this solution after trying numerous codes from GitHub and community forums, none of which fully worked on my Asus Vivobook Go 14/15 (Model: E1404FA-FHD352).

I have tested this code on Linux kernel version 6.19.10-300.fc44.x86_64, and both the Wi-Fi and Bluetooth functionalities are working well.

However, there is a known issue in my code that I have not yet been able to resolve. When both Wi-Fi and Bluetooth are turned on simultaneously, the Wi-Fi network experiences interference. This causes the internet connection to slow down significantly, and in some cases, it drops to the point where even light browsing becomes impossible, although the Bluetooth continues to function normally.

If you are interested in using this workaround, or if you would like to contribute and help fix the co-existence issue via a Pull Request, please feel free to check out the repository here:

Repository: GitHub - risqinf/mt7902-driver: this a code firmware mt7902 unofficial to enable wifi and bluetooth. · GitHub

Note: I have not yet tested this driver on any operating systems other than Fedora.

Everyone, please know that this might be genuine, but it might also be a malware attempt. The driver git repository was created just 30 minutes before creating this Discussion topic. The github account has no public contributions history, and has only been active since the beginning of this year. There appears to be a lot of AI generated stuff on that account page and the linked project.

I advise everyone extreme caution. It’s not a good idea to run scripts from random sources that don’t have any community history and have been created today.

Of course I might be doing Riqsi injustice. But you should think about it before running any unknown code.

Hi Kamil Páral,

Thank you for pointing this out. You are absolutely right to warn the community, and I completely understand your skepticism. A brand new GitHub account posting a system driver with AI assistance—created just 30 minutes before posting—definitely looks like a major red flag. I don’t blame you at all for urging extreme caution; I would do the exact same thing.

To be completely transparent, I have absolutely no background in C programming. I actually put this together by looking for references from several existing repositories, and then I relied heavily on Claude to help me write and compile the final driver after spending all night desperately trying to get my Asus Vivobook Wi-Fi to work.

I only created this repository today because the existing solutions I found (like the GitHub - OnlineLearningTutorials/mt7902_temp: Linux driver development for mt7902 in the hope of getting running something · GitHub repo) failed to connect my specific device to the network. Once I finally got it working with Claude’s help, I just wanted to share it because maybe someone is experiencing the same problem as me on what might be the exact same device.

I have no bad intentions, but I 100% agree with your advice: nobody should blindly run scripts from a random new source. The repository is fully open, and I would actually be very grateful if experienced developers like you could audit the code or point out why the Wi-Fi/Bluetooth interference is happening.

Thank you again for looking out for the community’s safety!

Please don’t generate your responses with AI, use your own words. Here’s your original reply:

you can see and review the code directly, fully open and this code I made because it was my first time using fedora and previously I used alpine linux and used different devices, the repo I made purely because of a problem on my laptop because I used the mt7902_temp repo before the wifi managed to activate the wifi but failed to scan my wifi and when trying to scan it didn’t connect, I made this code yesterday and of course I used claude as my tool because I had never coded C before, if you don’t believe it it doesn’t matter I’m just sharing it because maybe someone has experienced the same problem.

In general the guidelines for using AI for posts on this forum is to explicitly state what is in your own words and what came from AI. Most of it should be your own.
Follow the guidelines.

Usually a persons own words are much more concise and compact.

@risqinf AI tools have made generating massive amounts of code quite easily. Using LLMs to write code can be valuable in a lot of contexts, but that requires good judgement. I won’t get into it here.

That said, I can’t stress this point enough: don’t expect others to review or “audit” your AI-generated code if you yourself have not spent time reviewing and testing it thoroughly. This applies for all communities, not just Fedora.

As an addendum to what @kparal has said, this is a forum for Fedora humans. Please just express your thoughts as they occur to you, grammar mistakes or typos are not an issue.

Using AI tools to generate code in a language you don’t understand and therefore can’t review seems like a strong antipattern.

Kernel 7.1 supports MT7902. Could you give rawhide kernel a try?

I haven’t tried the Rawhide kernel yet—I might give it a go if I have the time. Right now, I’m not using Fedora but Rocky Linux with kernel 6.12.0-211.26.1.el10_2.x86_64.