Subject: Re: Anyone care to post if restore cds work ?
To: David Brownlee <abs@netbsd.org>
From: Alex Pelts <alexp@broadcom.com>
List: port-cobalt
Date: 03/08/2005 05:55:13
Yep,
That is exactly what I figured after looking through init scripts and 
init binary itself. It calls mount_mfs and MAKEDEV when /dev/console is 
not present. There is no parameter to init that defines mfs size (at 
least documented parameter).

While it would be useful to make init create mfs filesystem of 
sufficient size, I would like to know why MAKDEV does not work for 2.0 
and does work for 1.6.1 on the original cd ? I will try to dig out hard 
drive from one of my old machines and run installation again. My problem 
is that qube is my only unix machine at home. My other computer is XP 
and my wife's laptop is XP as well. I have lots of old parts and 
probably can put anoter machine together for x86 NetBSD box. I am 
normally a linux person (don't throw too many stones :)).

Thanks,
Alex


David Brownlee wrote:
> On Mon, 7 Mar 2005, Alex Pelts wrote:
> 
>> Vadim,
>> You are second person to report this problem.
>> This is great. This means it mostly worked. I don't know what this 
>> /dev filesystem is, it is not present in my 1.6.2. It is also means 
>> that 1.6.2 should work just fine. Anyone can confirm this ?
>>
>> As soon as I find out how to fix the /dev problem I will build anoter 
>> 2.0 restore cd. If anyone can recommend a fix looking at the output I 
>> would be happy to incorporate it.
>> Partition table looks about right - 20% for var 5% for tmp, the rest 
>> is for root.
>>
>> /dev is new to 2.0 and the fact that it boots means we can make it work.
> 
> 
>     init checks for a populated /dev, and if not found will mount
>     a memoy filesystem on /dev/ and run /dev/MAKEDEV. It looks like
>     you have two issues:
>         a) init is creating too small a filesystem
>         b) it believes /dev is not populated :)
> 
>     If you have time it would be useful to play with changing the
>     mfs size in init so the mfs created is big enough - if you want
>     to let me know what values work I'll update the master source :)
> 
>