[ previous ] [ next ] [ threads ]
 From:  Fred Wright <fw at well dot com>
 To:  m0n0wall at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall] M0n0wall on a HDD info
 Date:  Wed, 4 Jun 2003 12:44:24 -0700 (PDT)
On Wed, 4 Jun 2003, bino-psn wrote:

> I renamed the file to *.gz ... and do this :
> indolab:/home/tenggo# gzip -d ./net45xx-pb11r401.img.gz
> indolab:/# dd if=/home/tenggo/net45xx-pb11r401.img of=/dev/hdc
> 10240+0 records in
> 10240+0 records out
> indolab:/# fdisk /dev/hdc
> Command (m for help): p
> Disk /dev/hdc: 8 heads, 16 sectors, 500 cylinders
> Units = cylinders of 128 * 512 bytes
> 8 partitions:
> #       start       end      size     fstype   [fsize bsize   cpg]
>   a:        1         5         5     4.2BSD     1024  8192    26
>   c:        1         5         5     unused        0     0
> But I still got "Read Error" message at boot time.
> The disk is work well in DOS (used to be a LRP disk).
> Is it about "dd" compatibility problem ?

First of all, you can verify the data by reading it back to another file
with "dd" and using "cmp".  Just remember that you need to specify "count"
in that direction (10240 in this case), since it can't infer it when the
input is a drive.

Secondly, you may be running afoul of BIOS geometry issues.  Different
BIOSes have different opinions about how software disk addresses map to
device disk addresses, and if the disklabel and bootloader blocks aren't
set up consistently with that, it won't boot.  That's a fundamental
weakness in the "prepared drive image" approach, although it generally
works for the 45xx BIOS, even when the "geometry" doesn't match the CF's

Thirdly, it's very unlikely that a 45xx image would work usefully on a PC,
anyway, due to the kernel configuration.  Even if it's a 486, it doesn't
have the expected Elan-specific hardware, and is also quite likely not to
use the one specific NIC type that the 45xx kernel is built for.

> ----- Original Message ----- 
> From: "Michael Iedema" <iedemam at pluto dot dsu dot edu>
> Sent: Monday, June 02, 2003 10:47 AM
> > The images (hdd & soekris), are gzipped but are named .img so internet
> > explorer doesn't uncompress them on the fly.  To unzip them I believe
> > you have to rename them to *.img.gz, then do a gzip -d name.  

Yes, unfortunately the new naming is more confusing, just to make WinIE
happy (Mac IE doesn't have that disease).  Perhaps it would be better to
use .img.gz and teach the updater to accept both compressed and
uncompressed images.

					Fred Wright