[TUHS] Uploaded 44bsd rootdump

Warner Losh imp at bsdimp.com
Fri Feb 17 03:59:09 AEST 2006

> On my FreeBSD/i386 host, I was unable to extract the "rootdump" 
> file of the 4.4BSD-Alpha distribution available on Minnie.
> Fortunately, I succeeded on my NetBSD/vax machine, even though
> its version of NetBSD is far older than FreeBSD 5.3 which is
> what I have on the i386.
> The VAX "restore" program mentioned something about quad-
> swapping, so perhaps FreeBSD's "restore" does not translate
> the byte order.  I think the 4.4BSD distribution is from a
> big-endian machine ("hp300").  It's funny, then, that I was
> sure that *at least* byte-order could *not* be the problem
> since both the i386 and VAX are little-endian and I was sure
> the 4.4BSD distribution was a VAX one!
> After extracting the rootdump, I created a .tar file instead
> and uploaded it to minnie.tuhs.org:/incoming, file name
> bsd44a-shoppa-rootdump.tar.  You may wish to compress it and
> move it to the same directory as the other 4.4BSD-Alpha files
> so others won't have to go through the trouble I experienced.

When Sam Leffler added ufs2 support to FreeBSD's dump/restore
programs, he eliminated support for reading very old dump files.
Supposedly, these were only 4.3BSD based dump files, but it appears
that some newer versions of dump still use this older format.
NetBSD's hasn't been updated to support ufs2, afaik, and even if it
has, the support for the old formats hasn't been removed.

FreeBSD's restore, like all the other restores, does have support for
different endians.  If you have a 4.x old older FreeBSD restore
binary, then you should still be able to read these files.

Just for future reference.  There was talk about extracting restore at
the tip of RELENG_4 and making it a port for dealing with really old
restore files, but nothing ever came of that talk.

Glad to see you were able to deal none the less.  This issue seems to
come up every few months.  Unfortunately, re-merging the old format
support back into restore appears to be kinda hard (where hard here
means that the old code won't work due to structural changes, although
one could imagine making it work w/o too much hassle).


More information about the TUHS mailing list