Re: SingleUser-Problem (2)

[Michael Boehnisch <billy%psycho.uni-paderborn.de@localhost>]

> Hi there!
> 
> I wanted to make a more precisely report of my problem:
> 
> 1) I install the miniroot.fs like written in the manual.
> 2) I start with loadbsd -a netbsd
>    @ the beginning of the screen of every boot come the two lines:
>     *[netbsd d.out symbole table not valid]*
> *    [no symbol tables format found]*
> 3) The booting goes ahead normal until after the line '10 views', there
>    stands:
>     *warning found rdb->secpercyl(209)!=rdb->nsectors(210)*rdb->nheads(1)*
>    but it continues normaly...

I get a similar message from my IBM DNES 39100 drive.  I think this indicates
that the drive has one sector per cylinder reserved for automatic bad block
remapping and thus, the logical geometry differs from the physical geometry.
I have successfully ignored the message for some time and then switched off
verbose SCSI diagnostics in the kernel without encountering problems.

> 4) I install the sets of your 1.4.1-FTP-Side over a AmigaDos-Partition:
>    The sets: base, comp, etc, man, misc, xserver. The kern I downloaded
>    too, but I don't know what happens with that.
> 5) The other sets I take from the Gateway-CD (NetBSD 1.3.2)
> 6) After Time-Code etc. it comes the message, that it was installed and
>    is ok.
> 7) So I reboot with loadbsd -a netbsd.
> 8) @ the line the fsck starts the failure comes like that:
>    */dev/rsd0a: bad super block: magic number wrong*
> *   /dev/rsd0a: unexpected inconsistency; run fsck_ffs manually*
> *   automatic file system check failed; help!*
> 9) So then comes (surely) the single-user mode. If a start fsck or fsck_ffs
>    mostely comes: 
>    * bad super block: magic number wrong*
> *    fsck: /dev/rsd0a: floating point exception.*
> 10) If I start fsck often, it could happen, that it really does something:
>     The normal phase 1.. checking.
>     *Phase 5: CG 0: bad magic number blk(s) missing in bit maps, salvage? y*
> *    summary information bad, salvage? y*
> *    mark filesystem clean? y*
> 11) So If I then reboot, the same thing happens.
> 

You may have given the information already in a previous mail, so excuse
me if I ask how much memory you have installed.  Your problem seems very
similar to mine when I upgraded to 128MB of RAM on my Cyberstorm PPC and
is related to the infamous "big machine crashes".  The generic kernel safely
used to boot with up to 32MB of memory only.  Ignatios recently increased
some internal values of the VM page handling and thus this border may have
moved upward (I didn't try for some time).  Still, I have to create my
own kernel to make netbsd boot with 128+MB RAM installed.  The file I
have to modify is usr/src/sys/arch/amiga/include/vmparam.h and I change the
value of VM_KERNEL_PT_PAGES from 4 to 8 (near the end of the file).

This cured the problem for me.

Try to limit the RAM used for netbsd to, say, 32MB (there is an command line
option for this in loadbsd, -m or something?) and see if the problems vanish.

Just my $.02,

   Michael B"ohnisch.