> Don't you need to: (1) load the LKM and (2) reproduce the probmlem?
> If the full (raw) data is in your syslog then you should be able to
> run a user-level program to decode it (or pretty-print it) even if
> you can not easily reproduce the problem.

I was thinking along the lines of "some people like kernel clutter, some
people like readable messages"; the latter would load the LKM as a matter of
course (hoping not to need the decode during boot), and the former wouldn't.
Of course, some times you don't fully realize how much you need readable
messages until after you have a blob of binary data in your syslog...
It shouldn't be too hard to devise a user-level program that could be built
from the same sources as the LKM, which partially ameliorates this problem,
as well as the problem of people sending mail with raw data in their reports.
But as someone pointed out, not all the kernel ports HAVE the LKM mechanism,
so perhaps it's not worth the effort to make it an LKM.