Boot loader doesn't see [root filesystem on] ATA disk after successful install

Joel rees at ddcom.co.jp
Tue May 10 23:05:37 PDT 2005


On Tue, 10 May 2005 22:23:28 -0700 (PDT)
Brian O'Shea <b_oshea at yahoo.com> wrote

> [...]
> There is a boot loader installed in the MBR of both disks, but I am
> trying to boot the IDE disk from the loader on its own MBR.  I didn't
> try booting the loader from the SCSI disk and then having it try to
> boot the IDE disk, but I don't think it would make a difference.  I'll
> give it a try tomorrow though, just for the sake completeness.

From what little I've seen, it could be worth a try if you have the time.

> [...]
> It can boot from the CD-ROM drive, which is on the same ATA controller.
> I have both the IDE disk and the CD-ROM drive set to cable select for
> their master/slave configuration.  Could this be the problem? 

I am told that cable select is more reliable with modern cables, less
reliable with older cables. (UDMA being modern if I recall correctly.)
You can search the web on "UDMA cable" to get quite a bit of information
on the cabling and positioning issues. There's a site called, I think,
pcguide, that is quite helpful.

> I'll
> try explicitly setting the IDE disk to master and CD-ROM to slave.

From what I've read and what I've experienced, putting hard disks and
CD-ROMs on the same channel is counterproductive. Boot problems and data
problems are said to be likely on many controller and drive combinations.

> > If you can boot from the SCSI, check the dmesg there to see whether the
> > ATA controller is recognized by the older system. That wouldn't give an
> > absolute answer, but might yield a clue.
> 
> The older system can see the CD-ROM drive, so it must be recognizing the
> ATA controller.  I'll post the relevant dmesg output tomorrow though.
> 
> > I hear that it's usually best to just let freeBSD's formatting utilities
> > do what they think they should and not try to meddle with that. 
> 
> Oh, well.  Too late for that!  Maybe I can set it back to the old
> values. 

When I was trying to set the geometry by hand, I found that fdisk and
disklabel would go back to the geometry they thought was best.

> However, when I changed the geometry to what the BIOS thought
> was correct, the reported disk size was closer to the advertised size.
> In both cases though the symptom was the same.

I thought I noticed something like that happen at one point while trying
to set the geometry by hand, but I dom't remember the details.

--
Joel Rees   <rees at ddcom.co.jp>
digitcom, inc.   株式会社デジコム
Kobe, Japan   +81-78-672-8800
** <http://www.ddcom.co.jp> **



More information about the freebsd-questions mailing list