Comment 99 for bug 404626

Would it be possible to devise a different process to deal with issues like these (namely bugs in drivers that are kernel modules) in the future?

Due to the need of guaranteeing the stability of the kernel even fixes for relatively simple bugs involving drivers like this one take months, with the result that people skip some distribution releases altogether.

Wouldn't it be possible and maybe easier in the future to deal with problems like this one by shipping out special packages containing only the source code of the fixed driver and using dkms to compile themselves and overriding the bugged modules?

In this way, only the people affected by the bug would install the fixing package, which could then be distributed with much less paranoia than a full kernel.