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

Re: bootstrap: panic: ... invalid IO size

On Sat, Oct 14, 2000 at 12:53:32PM -0700, Marc Singer wrote:
> On Sat, Oct 14, 2000 at 11:55:26AM -0700, Steve Bowman wrote:
> [deletia]
> > > It took some fiddling, but I use the root= command to determine that
> > > this drive was called hd1.  It must think that hd0 is the SCSI drive
> > > because the BIOS is configured to boot to the SCSI controller.  I
> > > found the kernel using root=hd1s3.  Then, when the kernel booted, it
> > > didn't like that, so I tried hd2s3 and received the above message.
> > 
> > You got this error trying hd2s3 but you found the kernel on hd1s3?
> > What was the problem with hd1s3?
> That's what's strange.  Let me summarize.
> According to the documentation, the first hard drive, no matter what
> else is in the system, is named hd0.  The third partition (slice) on
> this device would be hd0s3.

Well, I don't have a SCSI device to check, but as I read in the Easy
Guide, grub would know the third partition on the first SCSI drive as
(hd0,2) (assuming you have no IDE drive on the system) but hurd would
know the same partition by sd0s3 (vice hd0s3).

> According to GRUB, there is no hd0s3.  It does find hd1s3 to have the
> correct signatures, partition type and fs type.  I suspect that the
> BIOS may be lying about the devices so that it can boot the SCSI
> disk.  

The grub 'find' command is useful (e.g., find /boot/gnumach.gz).  IIRC,
it'll return the grub name of the partition.  It'll probably return the
one you expect; it looks like the problem is getting the right hurd name
for the kernel command.

> When I use the GRUB command
>   kernel=/boot/gnumach.gz root=hd1s3 -s
> and boot, it reports that there is no hd1s3.  The partition check
> concurs.  It tells me that there is an sd0 and an hd2.  At the prompt,

This seems to confirm the above....

> [snip]
> Yet, I don't believe that that is what it means since I get a
> different error when I use a verifiably non-existent device spec such
> as sd5 or hd5. 

I'm not sure what to make of this.  Perhaps it's trying to read a real
device like the cdrom instead of a non-existent device so the error
_should_ be different.

> [snip]

Good luck,

Steve Bowman  <sbowman@frostwork.net> (preferred)
Buckeye, AZ   <sbowman@goodnet.com> <bowmanc@acm.org>

Powered by Debian GNU/Linux and GNU/Hurd <http://www.debian.org>

Reply to: