Sun-Spots Digest, v6n248
William LeFebvre
Sun-Spots-Request at Rice.edu
Wed Oct 5 14:24:24 AEST 1988
SUN-SPOTS DIGEST Tuesday, 4 October 1988 Volume 6 : Issue 248
Today's Topics:
Re: FootMouse
Re: Problem: SLOW Boot on diskless Suns
Re: Force periodic password changes
Re: Sharing roots under SunOS 4.0; one solution
Re: Problems with VME address spaces
Congrats on a billion-dollar year
More NULLs in files
cc after dbxtool
Lack of DNI (DECnet) documentation
Lightweight Processes in SunOS 4.0
Problem with TB+ dialing out at 2400 baud
Problem setting up Parallel Printer on Sun 4
TAAC board queries
Using DTR on a serial printer connection?
SIGFLUSH?
PCNFS problems?
videoing a sun screen?
porting an applications from SUNs to networked Macs?
Send contributions to: sun-spots at rice.edu
Send subscription add/delete requests to: sun-spots-request at rice.edu
Bitnet readers can subscribe directly with the CMS command:
TELL LISTSERV AT RICE SUBSCRIBE SUNSPOTS My Full Name
Recent backissues are available via anonymous FTP from "titan.rice.edu".
For volume X, issue Y, "get sun-spots/vXnY". They are also accessible
through the archive server: mail the request "send sun-spots vXnY" to
"archive-server at rice.edu" or mail the word "help" to the same address
for more information.
----------------------------------------------------------------------
Date: Thu, 29 Sep 88 11:58:37 PDT
From: weiser.pa at xerox.com
Subject: Re: FootMouse
My student Glenn Pearson and I had papers in each of the last two
Computer-Human Interaction (CHI '87 and CHI '88) conferences, describing
several different models of foot-mouse Glenn built for Sun and Xerox
workstations, and various experiments of their utility for different
tasks.
-mark
------------------------------
Date: Thu, 29 Sep 88 13:46:22 EDT
From: Jean-Francois Lamy <lamy at ai.utoronto.ca>
Subject: Re: Problem: SLOW Boot on diskless Suns
We have 3 3/60s in the same room, on the same DELNI. Oddly enough, one
machine is black and white and has never exhibited the slow TFTP problem.
The colour and the grey-scale ones do it all the time (1 fast boot every
full moon). 3/50s exhibit some variability, but nowhere as extreme. We
have the exact opposite problem with 2/50s booting off a Sun 4 under SunOS
4.0. TFTP is very quick, loading the kernel takes forever and a half.
Telling the server to route packets through slower machines does not help,
and since /etc/bootparams does not allow specifying wsize and rsize for
the initial root and swap mounts, there is little else we can do. As I
recall, Sun's answer for the Sun 2 problem was "reboot until it goes
quickly". Sure.
Jean-Francois Lamy lamy at ai.utoronto.ca, uunet!ai.utoronto.ca!lamy
AI Group, Department of Computer Science, University of Toronto, Canada M5S 1A4
------------------------------
Date: Thu, 29 Sep 88 15:01:53 PDT
From: pixar!r2d2!dv at sun.com (David W. Vezie)
Subject: Re: Force periodic password changes
I don't have any code that'll do that, but I can get you on the right track.
There's an undocumented feature in passwd (with support from at least
"finger", maybe others) that if a '*' appears as the first character in
the gecos (real name) field of the password file when the password is
changed, then it will be removed. "Finger" knows about this, and when it
does it's name lookup, removes any '*' from output.
I don't know if yppasswd{,d} honor the same convention.
At any case, it wouldn't be too hard to write a program that checks every
so often to see who still has a '*' in their gecos field, and reminds them
to change their password (afterwards, making sure everyone has a '*' in
their gecos field).
Reminder: It's undocumented, so if you start using it, and Sun sees that
it's useful, they might take it away :-).
Name: David Vezie
Address: pixar!lfl!dv
Company: Lucasfilm Ltd.
------------------------------
Date: Thu, 29 Sep 88 21:40:18 EDT
From: scs%lokkur.UUCP at umix.cc.umich.edu (Steve Simmons)
Subject: Re: Sharing roots under SunOS 4.0; one solution
Reference: v6n234
Jean-Francois Lamy <lamy at ai.utoronto.ca> writes
>The general idea is to have clients mount an appropriate root read-only,
>and give them devices and whatever files need be private as soon as
>possible. [[and goes on to give a nice description]]
This is a useful thing, but I believe we accomplished the same at
Schlumberger with considerably less "strain" on the 4.0 layout. Also a
lot less work! You'll have to forgive me if some of the details are
slightly wrong; I no longer work at Schlumberger and this is from memory.
All the root file systems are simply directorys in a master root
partition. Since the great bulk of these files are (a) identical and (b)
essentially read-only, we did the following.
Go thru the root file systems and delete all but one copy of any given
executable. Link the remaining one back to the file systems you have
deleted it from. Do a similar thing for any data file that is (in effect)
read-only.
If you have ten clients on a server, you'll find you've recovered about
80% of the disk space in that partition. Yes, you run the risk that one
user with his local root login can muck up *all* the workstations on that
server, but in practice this never happened. (Maybe we had nice users?
Nah... :-) ). You still look like a vanilla 4.0 system, but you're a hero
in providing a lot more room than the Sun default for /tmp.
>There are obvious disadvantages to this (you can't give a particular
>client a different kernel named "vmunix", because there is nothing private
>at boot time), but if you have several clients configured identically
>making them share their root may worthwhile.
Here customizing a user's kernal is easy. Just rm /vmunix from his
partition and copy the new one in. Similar customizations apply to any
other utility/setup.
------------------------------
Date: Thu, 29 Sep 88 04:49:29 EDT
From: attcan!utzoo!henry at uunet.uu.net
Subject: Re: Problems with VME address spaces
>We have a device which looks like a piece of memory in the vme24d16...
>On sun 3 machines running SunOS 3.4 or 3.5 we have had no problem in
>writing to it using bcopy() or dereferencing integer pointers. We
>understood that any 32 data bit accesses were being handled by the MMU and
>split into 2 * 16 bit accesses.
No, this is not correct. The MMU has nothing to do with it. It is the
68020 CPU itself that is splitting up the accesses.
>However, on a Sun 4/260 running SunOS 4.0 a bus error is given...
The SPARC CPU will not do the splitting for you. At all. No way, no how.
It has nothing to do with caches or MMUs. This is one of the slightly-
dubious features of the decidedly-complex 68020 that did not make it into
the much simpler SPARC. You are going to have to change your code to do
things the way your device wants them; the SPARC will not try to cover up
the differences for you.
>Also, using the read() system call to read from a file or a raw disk
>directly into the mmap'ed memory works fine on the Sun 3 but hangs the Sun
>4 (the disk controller being SCSI on the Sun 3 and Xylogics SMD on the Sun
>4). Presumably the DVMA system on the Sun 3 handles this ok...
I'm not sure about this one, but again I think you have a misdiagnosis:
it's the difference in controller that's biting you, would be my guess.
I'd suspect that the Sun 4's controller, or the software running it,
expects to be able to do 32-bit DMA (while the Sun 3's controller is only
doing 16-bit DMA, so it happens to work).
Aren't I/O buses fun? :-( :-(
Henry Spencer at U of Toronto Zoology
uunet!attcan!utzoo!henry henry at zoo.toronto.edu
------------------------------
Date: Thu, 29 Sep 88 18:40:01 -0700
From: gnu at toad.com
Subject: Congrats on a billion-dollar year
I'm surprised I haven't seen any comments in Sun-Spots on Sun's 1988
fiscal year results. They tried really hard to break a billion dollars in
sales, despite the RAM crunch, and went $51 million over! In six years
they have introduced four generations of products and sold us more than
100,000 systems.
>> Congratulations, folks! <<
John Gilmore
------------------------------
Date: Thu, 29 Sep 88 16:13:21 -0500
From: abe at mace.cc.purdue.edu (Vic Abell)
Subject: More NULLs in files
I previously reported an exotic server/client execution combination that
sometimes caused file characters to be changed to NULLs. I have now
experienced the same problem using vi on a file on the server, then on the
client. Here's a sample sequence (staff.cc is the 3/280 server and vic.cc
is my 3/50 client):
vic.cc % rlogin staff
staff.cc % vi <file> # add some stuff to <file>
staff.cc % ^Z
vic.cc % vi <file> # the added stuff is replaced with NULLs
The file being edited resides on the server's disk, a r/w partition
mounted by the client. The transformation to NULLs is a permanent
condition.
Vic Abell
[[ Could be that disk addresses for the new blocks are being dropped from
the inode. In this case those blocks would look like "holes" and upon
reading would turn into NULLs. Just a thought. --wnl ]]
------------------------------
Date: Wed, 28 Sep 88 12:27 CDT
From: CS_MRDF_10 at uta.edu
Subject: cc after dbxtool
Greetings fellow SunTanners:
This message is in regard to the problems with doing cc after using
dbxtool. I had the same problem when I quit dbxtool using say, CTRL C, or
the "quit" option on frame menu. But I found that this problem does not
occour if I type "quit" at the dbxtool prompt inside the window. I hope
this helps. P.S. I'm runing a Sun3/260 with Unix 4.2 Version 3.4
Murthy Suryam
Dept. of Computer Science Engineering
Univ. Texas, Arlington
e.mail : cs_mrdf_10 at uta.edu
------------------------------
Date: Thu, 29 Sep 88 13:23:32 ADT
From: dalcs!david at uunet.uu.net
Subject: Lack of DNI (DECnet) documentation
Further to the recent note on the lack of online graphics documentation:
this is also missing for Sun's DECnet product SunDNI. We reported this as
a bug but heard nothing more about it. It is certainly not in the Unix
tradition. If I had known about it ahead of time, I would not have bought
the product.
------------------------------
Date: Thu, 29 Sep 88 11:46:38 pdt
From: kcooley at teknowledge-vaxc.arpa (Keith Cooley)
Subject: Lightweight Processes in SunOS 4.0
Has anyone tried using the lightweight process library of SunOS 4.0 ? I
wrote some small programs that used the monitor facilities but was unable
to link on a 3/110 because:
ld: undefined symbol
REGOFFSET
Subsequently I tried linking the examples in Chapter 6 of the System
Services Overview but I obtained the same message from ld. Sun support
identified this as bug id:1012022 for which there currently is no known
bug or workaround. As far as I can see this makes the lightweight process
library unusable. I want to find out if anyone has had more success with
lightweight processes.
Keith Cooley
kcooley at teknowledge.arpa
{uunet!sun!ucbvax!decwrl!uw-beaver}!kcooley%teknowledge.arpa
------------------------------
Date: Thu, 29 Sep 88 20:20:45 PDT
From: versatc!tran at sun.com (Tony Tran 9824317)
Subject: Problem with TB+ dialing out at 2400 baud
Did anybody have any success with the Trailblazer+ dialing out at 2400
baud to a Hayes?
I can only do FAST (s50=255X3) or at 1200 baud (s50=2X0) on my
Trailblazer. I was able to connect at 2400 baud, but could not log in.
Did I do anything wrong ?
Tony Tran
Versatec, Inc
(408) 982-4317
UUCP: {sun|mips|pyramid}!versatc!tran
------------------------------
Date: 27 Sep 88 18:00:21 GMT
From: Paul Steele <@relay.cs.net:aucs!paul at dalcs.uucp>
Subject: Problem setting up Parallel Printer on Sun 4
I am trying to set up a parallel printer on the Sun-4 parallel port
(referred to as /dev/mcpp0). I set up a printcap entry as follows:
mcp|CI-600 printer in Carnegie:\
:lp=/dev/mcpp0:pw#80:sd=/usr/spool/mcp:\
:fs#06320:if=/usr/lib/lpf:lf=/usr/adm/lpd-errs:
I think this is the correct printcap entry, but the spooler reports that
the printer is not ready (maybe offline?). Anyone got any ideas? Please
mail responses directly to me since I don't read this newgroup very often.
Thanks.
--
Paul H. Steele USENET: {uunet|watmath|utai|garfield}!dalcs!aucs!Paul
Acadia University BITNET: Paul at Acadia
Wolfville, NS Internet: Paul%Acadia.BITNET at CUNYVM.CUNY.EDU
CANADA B0P 1X0 (902) 542-2201x587
------------------------------
Date: Fri, 30 Sep 88 9:03:48 BST
From: everson%COMPSCI.BRISTOL.AC.UK at cunyvm.cuny.edu
Subject: TAAC board queries
We're just about to take delivery of Sun's new TAAC board and are
interested in how other TAAC owners are using the board - for instance are
you concentrating on using C or are you mainly using TAAC assembler?
We'd also be interested in what fields people are using the board in,
although I realise that for some people this may be very sensitive
information.
Phill Everson
University of Bristol, UK
------------------------------
Date: 27 Sep 88 16:33:41 GMT
From: Paul Steele <@relay.cs.net:aucs!paul at dalcs.uucp>
Subject: Using DTR on a serial printer connection?
Is there any way to connect a printer to the Sun with DTR protocol. By
default, async lines on unix systems use xon/xoff and we need DTR since
the printer is connected to a printer sharing box. There must be a way to
make use of this common protocol.
Mail replies to me if possible since I don't read this news group very
often.
Paul H. Steele USENET: {uunet|watmath|utai|garfield}!dalcs!aucs!Paul
Acadia University BITNET: Paul at Acadia
Wolfville, NS Internet: Paul%Acadia.BITNET at CUNYVM.CUNY.EDU
CANADA B0P 1X0 (902) 542-2201x587
------------------------------
Date: Thu, 29 Sep 88 13:27:09 PDT
From: chaos%gojira.Berkeley.EDU at jade.berkeley.edu
Subject: SIGFLUSH?
Is there any way under SunOS to get an already executing background
process to flush its I/O buffers to open files, before one kills it?
Often I would like to check the progress of a long simulation by looking
at its output files without having to write in fflush into the code.
Jim Crutchfield
Physics, UCB
(415) 642-1287
------------------------------
Date: Thu, 29 Sep 88 17:23:43 EDT
From: ted at braggvax.arpa
Subject: PCNFS problems?
We have Sun PCNFS 3.0 with Micom-Interlan NI5210 ethernet cards connected
to Latticenet twisted pair ethernet. The ethernet NFS driver is from
Micom, not Sun. Our PCs are AT class with 640K, our servers are Sun
3/160s running 3.5.
Now, the question. Can the Sun NFS overrun the PC in this configuration?
To get the lifeline POP daemon to stop overrunning the PCs, I had to
insert a manual delay in it, otherwise, long messages got trashed. I
suspect that something similar is happening with NFS. Sometimes when I
try to transfer many files to a remote drive, the transfer hangs, and I
get the famous MSDOS abort retry ignore message. After this, even the
local hardisk is unresponsive and a reboot is necessary. I was able to
reproduce this today several times with the same transfer, then it cleared
up.
I suspect that if I call Sun, they'll tell me my ethernet board is
unsupported, so I wanted to find out if other people are having similar
problems with pop and nfs.
Also, does anyone know a way to set the rsize and wsize for PC mounts if
this is a factor? If there is a way, it doesn't seem to be in the manual.
Ted Nolan
ted at braggvax.arpa
------------------------------
Date: Fri, 30 Sep 88 11:24:28 BST
From: everson%COMPSCI.BRISTOL.AC.UK at cunyvm.cuny.edu
Subject: videoing a sun screen?
Has anyone successfully managed to video a sun screen? We have tried but
have had considerable flicker problems with the sun refresh rate of 66Hz
(?) and our video camera's rate of 25 Hz (?).
Is there a solution? I've seen Sun promo videos of sun screens so it must
be possivble somehow - I hope you don't have to get a genlock video
camera! [[ That's probably how they did it. --wnl ]]
Any pointers or suggestions would be appreciated.
Phill Everson
University of Bristol
------------------------------
Date: 27 Sep 88 00:21:46 GMT
From: werner at rascal.ics.utexas.edu (Werner Uhrig)
Subject: porting an applications from SUNs to networked Macs?
We are considering a port of an application implemented on a network of
SUNs to a network (including or exclusively) of Macs.
The application (in C) uses SunView, and the Simplify/Unify relational
dbms and the Sun3 ethernet.
If anyone has done any similar work or thoughts on the topic, I'd
appreciate hearing from you.
Cheers, ---Werner
(ARPA) werner at rascal.ics.utexas.edu (Internet: 128.83.144.1)
(INTERNET) werner%rascal.ics.utexas.edu at cs.utexas.edu
(UUCP) ..!utastro!werner or ..!uunet!rascal.ics.utexas.edu!werner
------------------------------
End of SUN-Spots Digest
***********************
More information about the Comp.sys.sun
mailing list