Tim wrote:
This seems odd, as I've seen no evidence to suggest that the boot sector
gets altered by updating a kernel, when you're using grub as your
bootloader. After all, you can use the same boot record to boot either
kernel, and since the changes are in the grub.conf file, there's no need
to make changes elsewhere, as well. Lilo's another matter, though.
I can, quite easily, believe that Windows may have simply stuffed the
file up that you'd previously used through NTLDR.
I think the obvious test would be to take that boot record file from
before the kernel update, and compare it to the one after the update.
Use the "diff" command.
If they are the same, as I expect, that means you're chasing a red
herring. If they're not the same, you'd have to work out if that's just
co-incidental, or if there really are kernel-dependent changes in the
boot record.
Diff just stated that the files are different (after all, they are
binary files). But going through very quickly with hexdump -C show that
the first 80 bytes or so of the boot sector files are almost completely
different. I'm not inclined to believe in a coincidence in this case; I
have enough evidence convince me if no one else that this was related to
a kernel change.
As for others' suggestion that I'd have to do this for every kernel
update, we'll see. I didn't need to put the new boot sector file in
place after the new kernel came down from yum last night (after I got
Fedora booting again), and I didn't regularly have to do it before.
Maybe it's just on major kernel revisions that I'll have to do this.
Regardless, I at least now what to look for now, and can make the
"repair" in a couple minutes.
Thanks again!
Raymond
