Salix Forums

I finally hit a machine with the oops issue with os-prober myself by chance. Therefor it was just a matter of minutes to determine modprobe jfs as the actual culprit. The problem is that Salix uses the huge kernel which has many filesystem modules already builtin like jfs in this case. But there's only one module package for huge and generic kernel, which also contains the jfs module, which is not actually needed for the huge kernel, just the generic kernel. Now loading the module, which os-prober does, at first works because it does exist, but usually just causes an error message and also used to for jfs in earlier kernel versions afaik. But with 13.37 loading the jfs module causes an oops. So I guess this is what one would call a kernel regression.
Anyway I was always able to regain control of my system by pressing Ctrl + Alt + F1-7 in sequence. This worked so I didn't bother testing whether less would work too.

I'll try to develop a patch for os-prober soon, which should prevent it from loading jfs with the huge kernel, by checking for jfs in the /proc/filesystems listing. If anybody knows of a better solution, please give me a yell (lsmod is none, because it doesn't list builtin modules).

Maybe I'm also going to give Pat and/or the kernel devs a yell. But while Pat tends to ignore mails, reporting to lkml would require me to test some new kernel development version for this issue and if it persist would probably require bisecting.

New build of the os-prober package uploaded to the repository which includes a workaround for the kernel in 13.37. This workaround prevents the kernel from crashing/oopsing here.

Shador wrote:New build of the os-prober package uploaded to the repository which includes a workaround for the kernel in 13.37. This workaround prevents the kernel from crashing/oopsing here.

Great!