Added command to delete the driver folder in /var/lib/dkms/ when the driver is removed #145
Conversation
…driver is removed
|
Hi @gmrparcerao Thanks for the patch. I have not merged it yet because I am thinking about it. I think you have an issue that needs to be addressed but it is not clear to me if this is the optimal solution. If I am understanding you correctly, this is an issue that only comes up with a new kernel is installed causing the driver to be compiled for the new kernel automatically? Right? If so, it would be best if this is properly handled automatically instead of having to run remove-driver.sh . I think need to determine what the real problem is and why it does not happen when installing or reinstalling with install-driver.sh. The automatic compilation when a new kernel is installed uses dkms.conf and the only external code that it uses is dkms-make.sh .
I tested using remove-driver.sh and the /var/lib/dkms folder is deleted. So, that means I am not duplicating the problem. If you can take a deeper look at this and help, I would appreciate it. |
|
Some time ago, I have also encountered this problem. I don't remember the details, but it took an efforts and dancing to resolve it manually. And Yes, like You I don't know (remember) how to reproduce it. I am using Debian where kernel is updated automatically. If I am not wrong, the incident caused by two events that happened simultaneously: the kernel was updated with new version and I have pulled updates and installed new version of the driver. I am not sure if I am describing the scenario exactly, but it is something like that. |
I've been thinking about this and I can come up with an idea what the issue might be but it is a pain to reproduce.
That is what I am thinking. Something goes wrong when a new kernel comes in with a distro version upgrade. What I think I am going to do is use @gmrparcerao patch but slightly modified so as to use it for testing but ultimately we need to figure out a way to do this that does not include running remove-driver.sh. Any ideas are welcome. |
|
Why do You not submit Your driver to stock kernel? |
This is not my driver. it is Realtek's driver. I maintain the driver with help and I make it easier to install. There is an effort to improve the in-kernel driver for this chip. See: The in-kernel driver is improving. Five patches recently went into kernel 6.9. Those patches dramatically improve the experience with client mode. AP mode needs work. I hope we can get the in-kernel driver in good enough shape for me to shut this repo down. I need to be spending more time with other parts of the site. |
|
@morrownr wrote:
it comes back reproducing itself: see log: |
|
The log makes it look like an API change has failed to make it into this driver. Let me test. I see your kernel is 6.9. |
|
Your log makes it look like you are using a version of the driver that is not updated. Did you run the following before updating the kernel? $ git pull |
Yes I did. Though I think it may not be enough and reinstalling the driver with your script is necessary too. I will try, but before I want to test stock driver rtw88_8821cu with new kernel. |
|
Here is the result. I am using this driver with hostpad serving 2 clients. It works for a few minutes then lost connection: |
|
Also reinstalling your driver give no result too. So rollback to old kernel |
|
I will do some testing here with kernel 6.9. Please post results for rtw_8821cu in the following Issue: For what it is worth, AP mode in rtw88 is bad in all drivers accept for rtw_8821au and it is not even in the kernel yet... hopefully it will be soon. |
When the driver is removed using the remove-driver.sh script (mostly after updating the Linux kernel), the driver folder remains in the /var/lib/dkms/ folder, which causes issues to reinstall the driver after kernel updates and even blocks another DKMS software to update. I simply added two lines to remove the driver folder in the designed location, solving this problem.