Sun-Spots-Request@RICE.EDU (William LeFebvre) (09/18/87)
SUN-SPOTS DIGEST Thursday, 17 September 1987 Volume 5 : Issue 43
Today's Topics:
Moderator's note: bug reports
Source for Sun to VAX/VMS mail using SUNLINK/DNA interface
Re: ZMAGIC format a.out files
> 8 processes on Sun-4
Re: tar tape troubles
Re: NeWS on ST
Bug in 3.2 setup
Undocumented messages from Sun-3/160C and FPA
Device driver question
New Symbolics Flavors?
lockscreen, power off, or screen power off?
Sun as a Personal Computer at Home?
Clearpoint memory for Sun 3/260?
Graphics Request.
Apollo & Sun for image processing?
----------------------------------------------------------------------
Date: Thu, 17 Sep 87 16:23:19 CDT
From: William LeFebvre <phil@Rice.edu>
Subject: Moderator's note: bug reports
In the past few issues there have been several reports of bugs in the
software distributed by Sun (known as SunOS). The reports sent to
Sun-Spots rarely make any mention of an attempt to contact Sun support for
help. This had me a little concerned. I was recently contacted by a
member of the Sun technical staff regarding, specifically, the report
about problems with the GKS support. No one had ever contacted Sun about
this problem. This matter now has me greatly concerned. Gentle readers,
Sun has sold you a product and they should stand behind their product.
But how can they even demonstrate their willingness to support their
customers when they are never given the opportunity to do so?
I would like to propose a few guidelines for handling bugs and bug reports
on Sun-Spots. When you find a bug, your first move should be to contact
Sun support. Do as much as you can to attempt to resolve the problem with
them. If the problem is resolved, send a brief summary of your experience
(with patch reference numbers, if appropriate) to Sun-Spots. If, however,
Sun support cannot or is not willing to help you with your problem in a
timely fashion, then send a message to Sun-Spots pleading for help. Make
sure to note in your message what steps you have already taken (such as
contacting Sun) to try to solve the problem.
In summary: I think it only fair to Sun and to their other customers that
you go to Sun *first* with your bugs. If they fail you, for whatever
reason, then it is perfectly reasonable to query the Sun-Spots readership.
William LeFebvre
Department of Computer Science
Rice University
<phil@Rice.edu>
------------------------------
Date: 10 Sep 87 17:41:25 GMT
From: allen@pyr.gatech.edu (P. Allen Jensen)
Subject: Source for Sun to VAX/VMS mail using SUNLINK/DNA interface
The following interfaces directly to VAX Mail over DECNET using
SUNLINK/DNA It may be used with SENDMAIL (see /usr/lib/sendmail.cf and the
tutorials in the back of the Sys. Admin. Manual). You will have to change
the variables server_???? to match your dec system or add them as
parameters.
The following programs are completely un-supported and guaranteed to have
some problems. They have, however, been of great use to myself and other
people on our network. There are no restrictions on the use of the
following code.
[[ The source is in the archives under the name "sun-source/dnamail.shar".
It can be retrieved via anonymous FTP to host "titan.rice.edu" or through
the archive server. For more information about the archive server, send a
message containing the word "help" to "archive-server@Rice.edu". The
makefile for this software requires an additional source file called
"dnaerror.c". It is my understanding that this file is provided with all
Sunlink/DNA software, so those interested in using Mr. Jensen's program
will already have this file. --wnl ]]
P. Allen Jensen
Department of Civil Engineering - Georgia Institute of Technology
Atlanta Georgia, 30332-0355 (gatech!gitpyr!allen)
USA, Earth
jensen@strudl.gatech.edu
------------------------------
Date: 14 Sep 1987 17:16 EDT
From: Anund.Lie@cs.cmu.edu
Subject: Re: ZMAGIC format a.out files
As far as I know, the swap area on the disk is *still* used for text
segments, even with demand load executables (ZMAGIC).
When an a.out file with ZMAGIC magic number is loaded, the page table for
the text segment are initialized with "executable fill" entries pointing
to the file which the contents are to be demand loaded from.
On each page fault, one page is paged in from the executable file and the
page table entry is changed to that for an in-core page. However, now
there is nowhere the system can record the fact that this page really came
from executable file F at offset O. Therefore, the best it can do is to
mark it "dirty" so it will be paged out to the designated swap area if the
page daemon decides that it needs that page frame again. So, on a loaded
system, most of the executables in use end up being copied to the swap
partition anyway.
If you start investigating the usage of the swap partition, you will find
that there is much more space allocated there than is accounted for by the
data segments alone. (Text segments are made read only/demand load by
"ld" by default, so as long as nobody uses weird loader flags -- or the
debugger, no text segment should ever need any backing store on the swap
partition.) This is really a nuisance, because you have to allocate a
swap partition for the maximum virtual memory space expected -- and count
the *huge* executables into this. (Swap partition size of 4x physical
memory space has been suggested as a rule of thumb.)
Anund Lie
Division of Information Engineering
Norwegian Inst. of Technology
A_Lie%vax.runit.unit.uninett@tor.nta.no
(Current address:
Department of Computer Science
Carnegie-Mellon University
Anund.Lie@gandalf.cs.cmu.edu)
------------------------------
Date: Tue, 15 Sep 87 07:14:43 EST
From: Steve D. Miller <steve@brillig.umd.edu>
Subject: > 8 processes on Sun-4
The spec sheet I have for the Sun-4 seems to indicate that one can have
sixteen (instead of the usual eight) contexts in hardware.
-Steve
Spoken: Steve Miller Domain: steve@mimsy.umd.edu UUCP: uunet!mimsy!steve
Phone: +1-301-454-1516 USPS: UMIACS, Univ. of Maryland, College Park, MD 20742
------------------------------
Date: Wed, 16 Sep 87 22:01:07 EDT
From: bzs@bu-cs.bu.edu (Barry Shein)
Subject: Re: tar tape troubles
Try the following:
mt -f /dev/rst0 fsf 1 # skip first "file"
tar tif /dev/rst0 # try to list the tape 'i'gnore errors
You will probably get an error for the first file (directory checksum
error) and then things will get back into sync. If that works then you
can:
mt -f /dev/rst0 rew # rewind it
mt -f /dev/rst0 fsf 1 # skip file again
tar xif /dev/rst0 # extract the tape
Just did this for a student here the other day, tho it was on 1/2".
-Barry Shein, Boston University
------------------------------
Date: 17 Sep 87 10:02:13 GMT
From: unido!pbinfo!mike@uunet.uu.net (Michael Utermoehle)
Subject: Re: NeWS on ST
> From: mckay@ea.ecn.purdue.edu (Dwight D Mckay)
> Organization: Purdue University Engineering Computer Network
> Do the folks who are suggesting a port of X for the ST know that NeWS has
> already been ported to the ST?
> That's right, Sun's NeWS window system has been ported to the ST by some
> folks at Sun. I've seen pictures of it. And my Sun salesman claims Sun is
> undecided about what to do with.
Stop !! Don't throw away your Sun !
Buy an Atari too !
My guess: (worst case):
The Atari ST act's as a window server, looking for PostScript
coming down the serial line. So you have to connect the ST
to your Sun (running the clients) via RS 232 (using Serial Line IP
for communication).
Bibliography:
(1) Sun User Group - README Vol.2, No.3 Page 7
... the system was demonstrated at various shows with NeWS
running on an Atari ST and communicating via RS-232 using SLIP.
(2) Sun - NeWS Preliminary Technical Overview
Porting the Operating System Interface
(my comment: we first need MINIX)
Let's ask the people reading comp.windows.news or comp.sys.sun
mike.
UUCP: ...!seismo!unido!pbinfo!mike | Post: Michael Utermoehle
or mike@pbinfo.UUCP | Universitaet-GH Paderborn
| FB 17 - Informatik
CSNET: mike%pbinfo.uucp@Germany.CSNET | Warburger Str. 100
| D-4790 Paderborn
ARPA: mike%pbinfo.uucp@seismo.css.gov | West Germany
| Tel.: +49 5251 603070
------------------------------
Date: 16 Sep 87 19:30:21 GMT
From: luth!al@uunet.uu.net (Arne Lundberg)
Subject: Bug in 3.2 setup
There exists a bug in setup version 3.2 that creates a configuration that
can prevent diskless workstations from booting. During boot of a diskless
client it does a RARP to get its IP address. This address is then
converted to a hexadecimal string and used as a name for the bootfile that
is requested by tftp from the server. The problem is that setup sometimes
creates bootfile names from hostnumbers in a different manner than the
monitor in the client does. In our case we had a client with IP address
2.10. When booting this client the monitor requests a bootfile with name
0200000A, and the link that setup created to the proper bootfile is named
200000A. The effect of this is that the bootfile is not found and the only
error message is "tftp: timeout".
I would think that this bug would show up every time you have an IP
address that would start with a 0 in the hexadecimal representation.
As I don't think we, in the north of Sweden, ever will be connected to the
Internet we selected a networknumber without consulting NIC. Our first net
was number 1 and then number 2 is an easy choice for the next net.
[[ No more arguments about the ethical implications! I promise! --wnl ]]
Arne Lundberg -- al@luth.UUCP
Computer Science Dept, University of Lulea, SWEDEN
------------------------------
Date: Tue, 15 Sep 87 15:46:00 EDT
From: sean@cadre.dsl.pittsburgh.edu (Sean McLinden)
Subject: Undocumented messages from Sun-3/160C and FPA
Something is occuring with programs compiled with the -ffpa option (and
linked, appropriately) on a Sun 3/160C with Rev 2 of the FPA. We
installed the FPA board, recompiled the kernel with device fpa, did a
"MAKEDEV fpa" in the /dev directory.
The following is run by /etc/rc.local:
if [ -f /dev/fpa ]; then
/usr/etc/fpa/fpa_download -q -u /usr/etc/fpa/fpa_micro_bin -m /usr/etc/fpa/fpa_micro_map -c /usr/etc/fpa/fpa_constants -r >/dev/console
fi
if [ -f /dev/fpa ]; then
/usr/etc/fpa/fparel >/dev/console
fi
I don't have a copy of what is printed out on the console because it is
not saved in /usr/adm/messages.
When programs are compiled with the -ffpa option (in this case, suntools),
I get the following:
Warning! Sun FPA works best with 68881 mask A93N
Any clues?
Sean McLinden
Decision Systems Laboratory
University of Pittsburgh
------------------------------
Date: Sun, 13 Sep 87 15:09:26 -0400
From: Steve Dyer <dyer@spdcc.com>
Subject: Device driver question
When writing a device driver for a multibus board attached to a Sun 3/160
via their VME-to-Multibus converter, how does the interrupt priority set
on the Multibus card relate (if at all) to the priority field specified
for the controller in the kernel config file and the choice of an
appropriate spl level in the driver to lock out interrupts? I've written
device drivers for other 4.2BSD machines, but am not particularly familiar
with the vagaries of the Sun architecture when it comes to handling
interrupts. Sun's manual for device driver writers is a little vague on
this point, and I have no other documentation available.
I am a little suspicious of the hardware I'm working with; it seems to
cause bus timeout problems with other VME peripherals like the SCSI
controller when activated, and I'm getting "spurious level 4 interrupts"
occasionally. This may all be due to faulty hardware in the controller,
adapter or backplane, but if it rings any bells with anyone more familiar
with writing drivers for the Sun, as being due to some malfeasance on my
part, I'd appreciate a note. Thanks.
Steve Dyer
dyer@harvard.harvard.edu
dyer@spdcc.COM aka {ihnp4,harvard,linus,ima,bbn,m2c}!spdcc!dyer
------------------------------
Date: Tue, 15 Sep 87 16:24:23 EDT
From: cognac!elias@pt.cs.cmu.edu (Glenn S. Elias)
Subject: New Symbolics Flavors?
Does anybody have a lisp environment running "New Symbolics Flavors" on a
Sun? I currently run Lucid Common Lisp, and have the sources to the New
Flavors written in Symbolics - Extended Common Lisp, and have found that I
have neither the time or the knowledge about Symbolics Lisp to do the
conversion (if it is even possible!).
Any help would be much appreciated.
Glenn Elias
cognac!elias@pt.cs.cmu.edu
Westinghouse Research and Developement Center
(412) 256-2706
------------------------------
Date: 16 Sep 87 10:27:18 GMT
From: rhett@seismo.css.gov (Rhett Butler)
Subject: lockscreen, power off, or screen power off?
On a single user Sun 3/50 with 70Mb disk which is a node on an
ethernetwork, which choice is best when you leave at night:
1) power off system
2) power off screen only
3) run lockscreen only
Can any knowledgeable users comment?
[[ It partly depends on how much you trust the 120 VAC power going in to
your machine. As a rule, it is best to leave a hard disk spinning. It is
much harder on a disk to spin it down and then back up than it is to just
leave it running. But if you don't trust the power coming out of the wall
to remain stable overnight and you don't have a regulator between it and
your machine, it might be safer to power it off. If you are going to
leave the machine on, I see no reason why you shouldn't leave the screen
on as well, provided you are running a screen saver program (such as
lockscreen). --wnl ]]
------------------------------
Date: Mon, 14 Sep 87 01:46:42 EDT
From: Prabhaker Mateti <pm%mandrill.cwru.edu@relay.cs.net>
Subject: Sun as a Personal Computer at Home?
This is a request for advice. I am thinking of buying a Sun 3/xx for use
at home. I am a computer science professor and a hacker. I can only
spend at most $7500. I am buying it through the university and will be
doing university work on it; so I expect to get a 20% discount. Here are
the questions.
- What would you Sun experts advice that I buy? What would you buy from
whom (Sun and 3rd party)?
- Is 70MB disk 'reasonable', or is it too small?
- Since I can get the disk 'loaded' at the university on their Sun network
and bring the disk home, is it worth spending some $2k for the 60MB tape
drive? I can live with archiving of my own files via the RS232 port to my
other personal computers and via modem to the University's various
systems.
- Is maintenace necessary, or is it worth gambling on sending the parts
for repair when they break?
--
prabhaker mateti, case western reserve university, cleveland, oh 44106
{decvax,cbosgd,sun}!mandrill!pm pm@cwru.edu (216) 368-2816
------------------------------
Date: Wed, 16 Sep 87 10:28:37 GMT
From: Paul Cooper <mcvax!stl.stc.co.uk!pac@uunet.uu.net>
Subject: Clearpoint memory for Sun 3/260?
We have had on order from a UK distributor for an 8MB Clearpoint memory
board for our 3/260 for about 3-4 months. We keep being told that it will
be "here soon". Are these boards available in the US yet? Does anyone
know when they will be available? Can anyone recommend an alternative?
Does anyone know about how Sun intend to "lock out the third party memory
suppliers"?
-- Paul Cooper (pac@stl.stc.co.uk <or> ...seismo!mcvax!ukc!stl!pac
<or> PSI%234237100122::PAC <or> (44) 279-29531 x2435)
------------------------------
Date: Tue, 15 Sep 87 22:08:26 cdt
From: rackow@anl-mcs.arpa (Gene Rackow)
Subject: Graphics Request.
I have a group that needs a "simple" graphics program. Their applications
run on CRAYs, BIG IBMs etc. and they get a list of numbers. Presently
they take that list and run it through a graphics program on another type
of system (RSX on an 11/70) and get paper output. They then take and
annotate this by hand. Drawing circles and arrows to point out the
highlights of the data. The plots are usually simple XY line or scatter
plots. These hand modified drawings are then used as "final" output. I
would like to give them better.
Does anyone have a package that can take this info and do it with the
features of a Sun3? I don't need all the fancy whiz-bang stuff of the
very expensive 3rd party packages, just a "simple" small package. If such
a package exists, or is under development, I would greatly appreciate
hearing about it and it's features (cost). Else if a 3rd party package is
available, and you have used it, a short description, or even a pointer
would help.
Many thanks in advance.
Gene Rackow ARPA: rackow@anl-mcs.arpa
Mathematics and Computer Science Voice: 312-972-7126
Argonne National Lab
Argonne Il. 60439
------------------------------
Date: 12 Sep 87 02:11:57 GMT
From: kato%etl.jp@relay.cs.net (Toshikazu Kato)
Subject: Apollo & Sun for image processing?
Dear everyone!
I am now trying to build a good software developing environment for my
research (computer vision, image processing and image database system).
Please suggest to me any ideas for making use of the following computer
facilities;
Apollo WSs (on Domain Net; AEGIS, 4.2, V):
DSP80 as file server, gateway to Ethernet, master node
DN300 with high resolution color image scanner camera
DN320
DN3000 (diskless node)
DN570 with image processor, monochrome camera (NTSC)
Sun WSs (on EtherNet; 4.2):
Sun3/260C (?) (maybe with color image scanner)
Sun3/60FC (?)
* (?) means "planning to buy".
I prefer Sun WSs, because there are many Suns and UNIX hackers in my
laboratory (ETL).
I personally am not familiar to OS, software tools, graphic softwares for
such WSs. Please advise me the smart usage of these resources.
Thanks in advance.
PS. This is my first post to CSNET. Thank you!
--
Toshikazu KATO
Information Systems Section, Electrotechincal Laboratory, Japan
CSNET: kato%etl.jp@relay.cs.net
------------------------------
End of SUN-Spots Digest
***********************