DragonFly BSD
DragonFly kernel List (threaded) for 2004-12
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: /boot/loader question


From: Yury Tarasievich <grog@xxxxxxx>
Date: Wed, 01 Dec 2004 14:50:25 +0200

walt wrote:

On Tue, 30 Nov 2004, Simon 'corecode' Schubert wrote:



On Tuesday, 30. November 2004 22:40, walt wrote:

My question:  is there any inherent limitation in /boot/loader
which would prevent loading the kernel from a logical partition?


yes: loader doesn't contain the needed logic to find and use a non-primary
slice. can you try manually setting the boot device in loader to the right
device? if that doesn't work, somebody needs to hack loader to support it (or
not)


Heh, my guess would be "or not" ;o)  When I type "show" at the loader
prompt I see curdev and loaddev, which are both pointed at the correct
logical/extended partition, but I don't see bootdev.  (Picky, picky!)

This is a very old topic in the *BSD newsgroups.  The OpenBSD guys are
openly hostile about the subject of installing to extended partitions
but they refuse to admit their real objections (which are pretty clear,
IMHO).

The facts (as I see them) are these:  no serious developer wants to
admit that his OS (which is superior in every way to any other OS)
does not rate one of the four PRIMARY partitions!  (Delete Windows,
for f*ks sake!)

The real thinking big you get in Darwin and Solaris, demanding TWO primaries each.


Anyway, when I looked into bootloader code, there didn't seem to be serious technicalities not to boot from logical drive? Remove type 5 from unknown, another buffer, another load-recognize cycle? I didn't look in loader2, but I guess it uses standard intra-partition navigation anyway?

btw, there exists some kind of (interoperation?) ambiguity when booting two different bsd partitions from lilo. What gets installed last, takes over TWO selections in lilo menu somehow.




[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]