Description of problem: It has come to my attention that the kernels availble for upgrade with Fedora could be better -- they sometimes don't even work. While a bugzilla could be filed on each individual event that occurs for each kernel, I think it may be a better idea to try and do what it takes to get major kernel errors from updates down. Version-Release number of selected component (if applicable): I stopped using Fedora packaged kernels because updating the kernel was never a known of what the result would be. EX: will I get a kernel panic, will my ethernet not work? and while a user can boot to the old kernel easily, still i think its something to consider. How reproducible: When I used Fedora kernels, updates were not always a garantee of working. There were quite a number of Fedora kernel updates that gave me kernel panics and others caused my ethernet to be broken. for instance here's two forums on fedoraforum which imply kernel panics with latest kernel - 2.6.17-1.2187_FC5 http://forums.fedoraforum.org/showthread.php?t=124820 http://forums.fedoraforum.org/showthread.php?t=124822 Steps to Reproduce: 1. 2. 3. Actual results: Fedora Kernel updates can cause kernel panics and other unwanted errors Expected results: Kernel errors in general should try to be reduced especially if its a kernel panic or causes something like ethernet not to work. Additional info: While its just natural that there will be issues as code is changed in kernel.org's upstream and eg when Fedora adds patches to help do X or make kernel better in some way. Either more time needs to be spent testing the kernel or more time needs to be taken to look through say the kernel config and added patches to try and limit kernel panics and undesired results from upgrading to a new fedora kernel.
There are many reasons regressions can occur, and dealing with them on a case-by-case basis is the only way. If there was a quick one-size-fits all solution to everyones problems, we'd have implemented it already. Of the two reports you reference above: The first bug indicates the problem exists on an earlier kernel too. This may have been related to an update in another package that got updated at the same time as the kernel. The second bug is uninteresting as it has binary only modules loaded.