Installing NetBSD 2.1 on Quadra 950

Hi,

I have tried, as an exercise, to install NetBSD 2.1 on a
Quadra 950.
The installation I've done consists of booting the netbsd
install kernel
from MacOS with Booter 2.0 and then run the installation
totally within
NetBSD via the serial console. Everything went ok until
reboot.
After reboot I could not get to a login prompt. Instead of a
login
prompt came only the message that the ttyE0 device was not
present.
After that I have rebooted in single user mode and found out
that in
my case (Quadra 950) the proper device is ttyZ0 (corresponds
to the
zstty detected on QUADRA at boot time -- see /dev/MAKEDEV).

So I added the following line in /etc/ttys:

# Define the console that we actually run getty on.
ttyZ0   "/usr/libexec/getty Pc"         vt220  
on secure

and rebooted. This time I've got the login prompt on the
serial console.

NetBSD/mac68k (quadra950) (ttyZ0)

login:

after logging in as root:

quadra950# w
 9:53PM  up  2:02, 1 user, load averages: 0.67, 0.38, 0.85
USER    TTY FROM              LOGIN  IDLE WHAT
root     Z0 -                 9:53PM     0 w

And now the Quadra is usable either over serial console or
over network.
Maybe somebody could add this to /etc/ttys in repo?

Regards,

Gheorghe Ardelean.

P.S. Here is the dmesg form the Quadra:
Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002,
2003, 2004, 2005
    The NetBSD Foundation, Inc.  All rights reserved.
Copyright (c) 1982, 1986, 1989, 1991, 1993
    The Regents of the University of California.  All rights
reserved.

NetBSD 2.1 (GENERIC) #0: Mon Oct 24 22:13:16 UTC 2005

buildsb2.netbsd.org:/home/builds/ab/netbsd-2-1-RELEASE/mac6
8k/200510241747Z-obj/home/builds/ab/netbsd-2-1-RELEASE/src/s
ys/arch/mac68k/compile/GENERIC
Apple Macintosh Quadra 950  (68040)
cpu: delay factor 1067
total memory = 81920 KB
avail memory = 75656 KB
mrg: 'Quadra950 class ROMs' ROM glue, tracing off, debug
off, silent traps
mainbus0 (root)
obio0 at mainbus0
esp0 at obio0 (quick): address 0x897000: NCR53C96, 25MHz,
SCSI ID 7
scsibus0 at esp0: 8 targets, 8 luns per target
esp1 at obio0 (quick): address 0x897402: NCR53C96, 25MHz,
SCSI ID 7
scsibus1 at esp1: 8 targets, 8 luns per target
adb0 at obio0
asc0 at obio0: Apple Sound Chip
intvid0 at obio0  f9001000: DAFB video subsystem, monitor
sense 1
intvid0: 640 x 480, monochrome
macfb0 at intvid0
wsdisplay0 at macfb0 (kbdmux ignored)
sn0 at obio0: integrated Ethernet adapter
sn0: Ethernet address 08:00:07:d4:65:4e
iwm0 at obio0: Apple GCR floppy disk controller
fd0 at iwm0 drive 0: (HD disk -- not supported)
fd1 at iwm0 drive 1: (HD disk -- not supported)
zsc0 at obio0 chip type 0
zsc0 channel 0: d_speed   9600 DCD clk 0 CTS clk 0
zstty0 at zsc0 channel 0 (console i/o)
zsc0 channel 1: d_speed   9600 DCD clk 0 CTS clk 0
zstty1 at zsc0 channel 1
nubus0 at mainbus0
fpu0 at mainbus0 (mc68040)
adb0 (direct, IOP): 0 targets
aed0 at adb0 addr 0: ADB Event device
scsibus0: waiting 2 seconds for devices to settle...
scsibus1: waiting 2 seconds for devices to settle...
sd0 at scsibus0 target 0 lun 0: <SEAGATE, ST32430N,
0266> disk fixed
sd0: 2047 MB, 3992 cyl, 9 head, 116 sec, 512 bytes/sect x
4193415 sectors
sd0: sync (160.00ns offset 15), 8-bit (6.250MB/s) transfers
cd0 at scsibus0 target 2 lun 0: <MATSHITA, CD-ROM
CR-8004, 1.1f> cdrom removable
cd0: sync (480.00ns offset 8), 8-bit (2.083MB/s) transfers
boot device: sd0
root on sd0a dumps on sd0b
root file system type: ffs

At 0:03 Uhr +0200 4.5.2006, Gheorghe Ardelean wrote:
>After that I have rebooted in single user mode and found
out that in
>my case (Quadra 950) the proper device is ttyZ0
(corresponds to the
>zstty detected on QUADRA at boot time -- see
/dev/MAKEDEV).
>
>So I added the following line in /etc/ttys:
>
># Define the console that we actually run getty on.
>ttyZ0   "/usr/libexec/getty Pc"        
vt220   on secure
>
>and rebooted. This time I've got the login prompt on
the serial console.

Funny you should mention that, since I ran into the same
thing while
installing -current on a Quadra 650 yesterday. Actually it
was even
weirder, because I had

tty00   "/usr/libexec/getty std.9600"        
vt100   on secure

in /etc/ttys and could log in on serial console as vanilla
user, but not as
root, in spite of the 'secure'.

Could somebody please point me to the motivation of changing
the built-in
serial ports from tty0{0,1} to ttyZ{0,1}? Why are these new
devices not
used in /etc/ttys? Why is there a surplus of tty devices in
/etc/ttys,
anyway, which have never been and will never be used?

	hauke

--
"It's never straight up and down"     (DEVO)

On Thu, 4 May 2006, Hauke Fath wrote:

> Funny you should mention that, since I ran into the
same thing while
> installing -current on a Quadra 650 yesterday. Actually
it was even
> weirder, because I had
>
> tty00   "/usr/libexec/getty std.9600"      
  vt100   on secure
>
> in /etc/ttys and could log in on serial console as
vanilla user, but not as
> root, in spite of the 'secure'.

The real device node is ttyZ0 (as described in /dev/MAKEDEV)
but in this
case tty00 is just a link to it. See the manual page for
zstty(4) also!

The same problem is discussed here:
http://mail-index.netbsd.org/port-macppc/2005/01/21
/0002.html
http://mail-index.netbsd.org/port-macppc/2005/01/21
/0004.html

> Could somebody please point me to the motivation of
changing the built-in
> serial ports from tty0{0,1} to ttyZ{0,1}?

I think tty0? used to represent serial devices based on
NS16550 (and the
like) while ttyZ are device nodes for zstty which are based
on Zilog 8530.

> Why are these new devices not used in /etc/ttys?

Maybe, because the people are often using the normal console
(keyboard +
video/framebuffer) and not the serial ports on mac68k? In
this case it's hard to find out the missing parts because
by
activating getty on tty00 the serial ports just work (the
links are
correctly made in /dev).

> Why is there a surplus of tty devices in /etc/ttys,
> anyway, which have never been and will never be used?

I think they are there because one can use multiport
serial cards and needs more ttys (especially on other
archs).

Regards,

Johny.

At 22:17 Uhr +0200 4.5.2006, Gheorghe Ardelean wrote:

>> tty00   "/usr/libexec/getty std.9600"  
      vt100   on secure
>>
>> in /etc/ttys and could log in on serial console as
vanilla user, but not as
>> root, in spite of the 'secure'.
>
>The real device node is ttyZ0 (as described in
/dev/MAKEDEV) but in this
>case tty00 is just a link to it. See the manual page for
zstty(4) also!
>
>The same problem is discussed here:
>http://mail-index.netbsd.org/port-macppc/2005/01/21
/0002.html
>http://mail-index.netbsd.org/port-macppc/2005/01/21
/0004.html

Ah, okay. Thanks!

>> Could somebody please point me to the motivation of
changing the built-in
>> serial ports from tty0{0,1} to ttyZ{0,1}?
>
>I think tty0? used to represent serial devices based on
NS16550 (and the
>like) while ttyZ are device nodes for zstty which are
based on Zilog 8530.

NetBSD/mac68k had the Zilog 8530 and /dev/tty0 for a
looong time,
probably at least until 2.0.

>> Why are these new devices not used in /etc/ttys?
>
>Maybe, because the people are often using the normal
console (keyboard +
>video/framebuffer) and not the serial ports on mac68k?
In
>this case it's hard to find out the missing parts
because by
>activating getty on tty00 the serial ports just work
(the links are
>correctly made in /dev).

Yep:

[haukefattie] /<5>compile/GENERIC > l /dev/tty0?
lrwx------  1 root  wheel  5 May  3 22:35 /dev/tty00 ->
ttyZ0
lrwx------  1 root  wheel  5 May  3 22:35 /dev/tty01 ->
ttyZ1
[haukefattie] /<5>compile/GENERIC >

>> Why is there a surplus of tty devices in /etc/ttys,
>> anyway, which have never been and will never be
used?
>
>I think they are there because one can use multiport
>serial cards and needs more ttys (especially on other
archs).

There is no driver for a multiport card on mac68k. And since
/etc/ttys is
machine-dependent anyway, it should probably be changed.
I'll think about a
PR.

	hauke


--
"It's never straight up and down"     (DEVO)

On Thu, 4 May 2006, Hauke Fath wrote:

> NetBSD/mac68k had the Zilog 8530 and /dev/tty0 for
a looong time,
> probably at least until 2.0.

Well, I have just tried out. 
On Q700, LC III, LC475 I have never tried
to use the serial ports because of the supported ADB.

> There is no driver for a multiport card on mac68k.

Yes, now I recall I have seen this on a wish list on
macbsd.com.

> And since /etc/ttys is
> machine-dependent anyway, it should probably be
changed.

I'd like to see /etc/ttys on this plaftorm modified to
include at least ttyZ0 as console!

> I'll think about a PR.

This is a good idea! 

Regards,

Gheorghe Ardelean.