On Fri, 7 Feb 1997, Miles Sabin wrote:

> I've just sucessfully installed RiscBSD from the UniqueWay CD (I got
> fed up with waiting for the 'official' CD release), and although
> everything's gone surprisingly smoothly, I've got a few questions.
> 
> First some setup details:
> 
>   Kernel, SA110, 4740
>   base.set     10/01/96 (or is that 01/10/96 ?)
>   misc.set     10/01/96 (  "              "   )
>   etc.set      26/10/96
>   x11base.set  25/10/96
>   xarm27.set   25/10/96
> 
>   BtRiscBSD v.2.87
> 
>   RPC700, 24+2, AKF85, SA, RO3.70
>   850mb internal IDE
>   Standard issue Acorn internal double speed ATAPI CDROM drive
>   Cumana SCSI II, 1.2GB Quantum Fireball, SCSI ZIP drive.
> 
>   RiscBSD is installed on the Quantum,
>     400MB RiscOS partition.
>     800MB RiscBSD partition,
>       /root+/var 30MB
>       50MB swap
>       /usr remainder
> 
> Right, now for the problems (you can probably guess that I'm a Unix
> novice).
> 
> 1.  After installing the kernel as /netbsd and booting native, ps
>     gives sensible results, but vmstat gives me,
> 
>       vmstat: undefined symbols: _dk_ndrive _dk_xfer

You will need to install the patch sets, patch12-*.* - I think patch12-3
fixes this.

>     does this mean that the vmstat in the installed base set is
>     incompatible with the kernel I'm using?
> 
> 2.  Starting X applications gave me lots of 'Possible process deadlock
>     due to shortage of L1 page tables' errors. Upping max processes
>     from the default 32 to 64 seems to fix it, but is that a sensible
>     thing to do?

Yes.  That is the current 'fix' for the problem, (until L1 pages are
sorted out).

> 3.  After starting X, whenever I run xmh I get,
> 
>       Error: Cannot perform realloc
> 
>     Is this a genuine lack of resources, or some other problem?

I get this too - I have tried using unlimit as well to try and solve it,
but can't seem to get xmh to run.  I use pine anyway, so not really a
problem.

> 4.  Attempting to do a non-native with boot with the latest SA kernel
>     (4958, unzipped with SparkFS) results in a blank screen (no
>     startup text at all) and unnerving disc activity on the RiscOS
>     IDE, rather than the SCSI disc the RiscBSD partition is on.
> 
>     Could this be a kernel that doesn't have Cumana support? Is my
>     bootloader too old?

Pass.

> 5.  The only way I could figure out to create a user who could su to
>     root was to give it a uid of 'root', as well as a gid of 'wheel'.
>     Trouble is, I have a sneaking suspicion that logging in as this
>     user is just as dangerous as logging in as root. Is that right? If
>     so, what's the right way of creating a user that can su root?

You need to modify /etc/group, for example:

wheel:*:root,ben

allows me to su to root, and no one else.

> 6.  I can use 1280x1024x8 F60 without any trouble, but I'd prefer to
>     use F75, as I do under RiscOS. I think I need to set up something
>     like a RiscBSD equivalent of a mode file, but I've no idea where
>     to look ... can anyone give me any pointers?

I don't think you can do anything about this without kernel changes, but I
could be wrong.  The monitor description files are kept in
/usr/local/etc/monitors if that helps.

> 7.  Which of the entries in /etc/passwd *must* be left in there? root,
>     obviously, and any other users I create, but what about all the
>     others with the silly names?

Offhand I think you need root, daemon, bin.  You could lose others like
uucp, if you don't use uucp I suppose, most of the others could go I would
guess, though I could be wrong...

> 8.  How do I go about formatting ZIP discs for use with RiscBSD?

Pass.

> 9.  Can anyone give me advice on setting up xdm?

Yep.  You need to look in /usr/X11/lib/X11/xdm and read the manual page on
xdm (its quite good as manual pages go).  I edited Xsession to
basically run the users .xinitrc file and not use the X session manager.
Then in .xinitrc I load a few xterms etc. and run fvwm as the last thing.
Works a treat.

> 10. Can anyone give me advice on configuring the network components:
>     I've got a PPP, dynamic-ip dialup account, with POP3 mail?

The pppd should be on your system, you will need to get the popclient bits
from /pub/acorn/RiscBSD/1.2-contrib/sets/popclient.* on the ftp site.

The popclient bit is easy - just read the man page for it.

I can't really explain exactly how to set up a connection using ppp since
there are so many different ways it can be done, some of which depend on
the nature of your ppp account.

However, the following is what I use:

1. A script to start it all and connect (dialppp):

#!/bin/sh
#
pppd tty00 57600 connect 'chat -v -f /etc/ppp/provider.dial.chat' 
noipdefault defaultroute debug name <login>

...the pppd line is all one line, and <login> is my login.  I use ppp to
authenticate myself with the server.

2. A chat script to dial the server (provider.dial.chat)

ABORT BUSY ABORT 'NO CARRIER' '' ATZ OK ATDT01810000000 CONNECT

If you read the man pages whilst looking at these, then you should be able
to work it out, and it may work for you.  Briefly, the dialppp script is a
shell script which runs 'pppd' on tty00 (serial port) at 57600 baud.  To
connect it executes the string in quotes - basically it runs the chat
program with the second script as an argument to dial the modem.

Since I am using ppp to authenticate myself, I have a file pap-secrets
with my username and passwords in.  Hopefully this should get you going.

Ben.

-------------------------------------------------------------------------------
Ben Strawson                                            b.strawson@cs.ucl.ac.uk
Department of Computer Science
University College London           http://www.cs.ucl.ac.uk/students/b.strawson
-------------------------------------------------------------------------------