Comment 10 for bug 1482711

I know I took a long time again. My primary observation is that this bug does not appear when the nvidia driver is not installed.

I tested the upstream kernel version that matches the most recent repository kernel update(as well as the most recent upstream kernel version) before installing the most recent repository kernel update(3.13.11-031311ckt27 which the package manager calls 3.13.0-66.108). With both of the kernels from the upstream source, the nvidia module failed to install, and neither exhibited the bug, but when installed from the package manager the bug did occur in that same kernel version.

Unfortunately I don't know how to make the nvidia driver install for the upstream kernels, if it is even possible, and thus cannot test them properly. This is complicated by the fact that this bug appeared when the package manager forced an update to the nvidia module from 331 which supports my video card to 340 which makes no such claim.

Other than the nvidia driver not installing, the upstream kernels worked fine, though with a lot of warnings when I booted them, all about apparmor rules not being enforced. If you want, I ran a dmesg with each of them booted into a file, but I hesitate to post them, and thus pollute the associated data without your say so.

Given this information I fully expect that the answer here is either, "go report this to ___ instead of here.", "Sorry but your hardware is not supported by the newer kernel versions. Have a nice day." or possibly a "Well, see if it happens with the repository kernels if you uninstall the ndivia drivers." (I made a half hearted attempt to do that even before I posted this bug, but I didn't figure out how to tell the package manager to uninstall all the nvidia drivers without trying to install some random other version.)

Another thing that might provide some clues: some time ago I had installed the CUDA develpoment tools, and made a simple test
program. On the version of the kernel for which the desktop works, that CUDA program will not run (it says that there is no device), however, when switched over to a virtual terminal from the blank screen caused by the bug, the program runs just fine. This tells me that perhaps the bug isn't occurring for this kernel version because the nvidia driver isn't working for it.

So I hope this is somewhat useful even though technically the response is unable to test upstream. As per the instructions I will mark it as confirmed again. I hope that that is the right thing to do, even though I wasn't really able to test what you wanted well.