Sun-Spots-Request@RICE.EDU (William LeFebvre) (10/12/87)
SUN-SPOTS DIGEST Monday, 12 October 1987 Volume 5 : Issue 50
Today's Topics:
Re: DES hardware for Suns (2)
NFS over LANBridge
Re: install could be faster (tape position)
SMD cables
Re: Wanted: SCSI Disks for Sun3's
Re: Version 3.4 C-library routines (memory mgmt)
Patching rpc.mountd in SunOS 3.2L
Problems with scrolling in Curses
Using SunOS 3.4 compiler, make with SunOS 3.2
Information about RTI's Ingres?
'as'-syntax for 020-specific adressing modes?
Porting GKS ncar to Suns?
UUCP over Ethernet and X.25?
schematic capture/pc board layout?
Apollo & Sun for image processing, summary
----------------------------------------------------------------------
Date: 6 Oct 87 03:52:22 GMT
From: elroy!david@ames.arpa (David Robinson)
Subject: Re: DES hardware for Suns (1)
> From: weiser.pa@xerox.com
>
> There is a device driver, there is a command 'des(1)' which discusses
> hardware, but there is not on the sun price list any mention of DES
> hardware for suns, and my Sun salesperson says she never heard of such an
> option. Any one know how to order it?
The chip refered to is the AMZ8086 Data Ciphering Chip which can be
plugged directly into a socket on Sun-2 Multibus boards and there are
traces and holes for one on a Sun-3/200 CPU board. As far as I know this
is not a standard option but it appears that it is suported in software
and would only take a minor amount of work to put one it. (*I* have never
done this so take your own chances!).
Sun does have on its spare parts list (Oct 1986) a:
100-1032 IC, AMZ8086 Data Ciphering Chip $200 (No discount)
So it does exist. How to order? Try calling sales?
-David
David Robinson elroy!david@csvax.caltech.edu ARPA
david@elroy.jpl.nasa.gov (new)
seismo!cit-vax!elroy!david UUCP
Disclaimer: No one listens to me anyway!
------------------------------
Date: Wed, 7 Oct 87 09:28:41 PDT
From: weiser.pa@xerox.com
Subject: Re: DES hardware for Suns (2)
This is a partial answer only because this does not yet say which suns
accept this chip, etc. But I am hoping that if I point my salesperson at
this place in the price book, she can get the rest of the information for
me.
-mark
__________
From: david@Sun.COM (David DiGiacomo)
It's listed in the spares price list as:
100-1032 IC, AMZ8086 Data Ciphering Chip ND $200
Of course, the chip number should be Am8068.
David DiGiacomo, Sun Microsystems, Mt. View, CA sun!david david@sun.com
------------------------------
Date: Mon, 5 Oct 87 09:25:36 EDT
From: mark@cbpavo.mis.oh.att.com (Mark Horton)
Subject: NFS over LANBridge
Don't put a layer 3 or 4 bridge/gateway between a client and its server,
although low traffic NFS mounts, between servers, for example, may be OK.
We've tried an AT&T ISN EBIM bridge, with capacity 1 Mbit/sec. The
fileserver was on one side and 7 workstations on the other side. It works
fine for a while, but as soon as the traffic gets heavy, all hell breaks
loose and it remains unusable. NFS and ND retry 10 times per second, with
no backoff. The queues on the bridge quickly get backlogged and the
clients don't let up. Booting a workstation is one very effective way to
put a big load on the bridge, although if all is quiet sometimes you can
boot without overloading the bridge.
Mark
------------------------------
Date: Tue, 06 Oct 87 00:16:21 EDT
From: ted@braggvax.arpa
Subject: Re: install could be faster (tape position)
Maybe I've just been lucky, but whenever I have a tape with multiple tar
or dump files, which I don't want to rewind or fsf, I just do all my
operations in order. That is to say if I have a tape with 3 tar files, I
will do something like this:
mt -f /dev/rst0 rew
tar xvbpf 200 /dev/nrst0 /* gets first file */
tar xvbpf 200 /dev/nrst0 /* barfs - get an error message */
tar xvbpf 200 /dev/nrst0 /* gets second file */
tar xvbpf 200 /dev/nrst0 /* barfs - get an error message */
tar xvbpf 200 /dev/rst0 /* gets third file and rewinds */
Note that tar fails every other time; but this doesn't hurt anything, is
quick, and leaves the tape positioned correctly for the next extract. The
same is true for restore. It is not asthetically pleasing, but harmless
and much faster than a rewind and fsf before each operation. Hasn't
failed yet (after many such invocations).
Ted Nolan
ted@braggvax.arpa
------------------------------
Date: 6 Oct 1987 09:54-EDT
From: Anund.Lie@gandalf.cs.cmu.edu
Subject: SMD cables
In Roy Smith's message in Sun-Spots Digest v5n47
> I don't think that's correct. Don't the command (I forget which is A and
> which is B, but I know the skinny one is the command) cables run radially
> from the controller to each drive and the data cables daisy-chain from
> drive to drive? That's the way I have my twin 2351's on a XY-450 and it
> works fine.
I think you got it backwards:
The A cable is the command cable and may be daisy-chained from drive to
drive, with a resistor pack terminating it at the last drive. (However,
it can also be run in a star configuration with a separate cable from the
controller to each disk and terminated at each drive. This probably
depends on the driving electronics in the controller.)
The thinner B cable is always separate for each drive, presumably because
of the (much) higher data rate. Also, (at least in theory, with a high
enough DMA bandwidth and duplicated DMA hardware in the controller) I
believe it might be possible to do parallel read or write operations on
different drives. The command is issued on the A bus to start the drive
operation, afterwards the controller is free to use the A bus to talk to
another drive, as long as it polls the status of the first drive to tell
when it's ready. (Well, I'm not a SMD expert, but this is what I
understand from the description of the SMD interface in the Xylogics and
Fujitsu manuals we've got.)
Anund Lie
Dept. of EE&CS ! Current address:
Division of Information Engineering ! Dept. of Computer Science
Norwegian Institute of Technology ! Carnegie Mellon University
N-7034 Trondheim, Norway ! Pittsburgh, PA 15213
E-mail (ARPA):
A_Lie%vax.runit.unit.uninett@tor.nta.no
Anund.Lie@gandalf.cs.cmu.edu
------------------------------
Date: 6 Oct 1987 14:55-EDT
From: Anund.Lie@gandalf.cs.cmu.edu
Subject: Re: Wanted: SCSI Disks for Sun3's
We run a Wren III SCSI on a Sun-3/160 - old SCSI host adapter, Sun-UNIX
3.2 with essentially no problems. We bought the disk from a Norwegian
dealer who packages them with power supply and cabling for 3/50's and
installs Sun-UNIX and sets them up as boot disks.
We upgraded to Sun-UNIX 3.2 in parallel with installing the disk, and
didn't manage to access the disk from 3.0. We did not investigate whether
this was due to a fundamental inability of the 3.0 SCSI driver to support
this disk, or whether it simply did not support the format the 3.2 diag
had formatted the disk with. (I strongly suspect that pre-3.0 diag is not
able to format this disk, though.)
The most serious problem with this combination is the failure of the host
adapter to release the SCSI bus with tape in motion: If there is
filesystem activity (not to say swap activity) on the disk when someone
starts to use the tape, you'll get some nasty error messages when the disk
accesses time out..... (Probably, you risk a kernel panic, so better make
sure that nobody accesses the tape!)
Cabling was quite easy: (Some initial problems were due to sloppy
workmanship by us and resulting bad connections.) We had to get a 2x50
male ribbon connector (as far as I remember) to replace the D-type
connector on the cable for Sun-3/50. Then we just plugged this into the
connector at the end of the SCSI cable within the 3/160 enclosure and let
the ribbon cable out under the edge of one of the side panels. Not very
pretty, but good enough, as the machine is placed in a locked machine
room. (We could have mounted it inside the cabinet, but did not have the
right mounting brackets. We could probably have hade some made up, but
I'm afraid that it would be easy to impair the air flow within the cabinet
and risk overheating of cards.)
As for speed: The small, quiet, inexpensive Wren is nearly as as fast as
the big, noisy, heavy, expensive Eagle that is the main disk on this
machine....
Actually, I don't think anyone ever would mind having the Wren in their
office. The enclosure we got it in has enough soundproofing (approx.
1/3" of plastic foam on the walls) to make it virtually soundless. (Well,
you can hear a little chirping when the disk seeks, that's all. You have
to put your fingers over the vent to convince yourself that there's a fan
in the box.)
Does anybody have experience with other SCSI drives?
Anund Lie
Dept. of EE&CS ! Current address:
Division of Information Engineering ! Dept. of Computer Science
Norwegian Institute of Technology ! Carnegie Mellon University
N-7034 Trondheim, Norway ! Pittsburgh, PA 15213
E-mail (ARPA):
A_Lie%vax.runit.unit.uninett@tor.nta.no
Anund.Lie@gandalf.cs.cmu.edu
------------------------------
Date: Tue, 6 Oct 87 23:02:48 EST
From: Chris Torek <chris@mimsy.umd.edu>
Subject: Re: Version 3.4 C-library routines (memory mgmt)
In Sun-Spots Digest v5n48, sheffler@gateway.mitre.org asks:
>something in the standard library calls "f_prealloc" which seems to be
>defined along with "valloc". I think it has something to do with opening
>files. Anyway, since we don't have a source-code license (does anybody?)
>I would appreciate it if someone could tell me what this routine thinks it
>does.
The following is from the 4.3BSD `bugs' section under fopen(3s):
In order to support the same number of open files as does
the system, fopen must allocate additional memory for data
structures using calloc after 20 files have been opened.
This confuses some programs which use their own memory allo-
cators. An undocumented routine, f_prealloc, may be called
to force immediate allocation of all internal memory except
for buffers.
I would be surprised if similar wording is not hidden away in the
Sun manuals.
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7690)
Domain: chris@mimsy.umd.edu Path: uunet!mimsy!chris
------------------------------
Date: Tue, 6 Oct 87 13:11:10 PDT
From: allegra!mp@ucbvax.berkeley.edu
Subject: Patching rpc.mountd in SunOS 3.2L
People here would like to start using their Sun-4's as NFS servers, and I
don't want to do so until I can turn on the nfs_portmon and other
variables in rpc.mountd. rpc.mountd was, as usual, distributed stripped,
and Sun says an unstripped version won't be available until FCS. Has
anyone figured out what addresses to patch?
Mark Plotnick
allegra!mp
------------------------------
Date: 5 Oct 87 18:09:48 GMT
From: sun!wdl1!mas1!joej@decwrl.dec.com (Joseph Jou)
Subject: Problems with scrolling in Curses
I had a difficulty on using the 'SCROLL' or 'DELETELN()' function of the
Curses library. The problem happened right after the function was called.
According to the manual, these two functions will bump up each line on the
screen and leave the last line blank.
Apparently it only does the first part (scrolling) and leave the original
text of buttom line on the screen. This situtation happened for Sun-3
workstation (as a client) and Xterm under Ultrix.
Does anybody out there who have experiences using the same function out of
the Curses library under Unix environment? If so, can you point out any
solution? Your great advice is highly appreciated.
Joe Jou
Measurex Automation System
(408) 973-1800 x707
------------------------------
Date: Tue, 6 Oct 87 15:37:17 EDT
From: Matt Landau <mlandau@diamond.bbn.com>
Subject: Using SunOS 3.4 compiler, make with SunOS 3.2
I'm in a position where I'd like to use the enhanced make from SunOS 3.4
(which I assume also requires the 3.4 compiler and other programs from the
"SunPro" part of the 3.4 tape), but I cannot do the complete upgrade to
3.4 at the moment. Is it feasible to install just the SunPro stuff, and
run it under a system that's otherwise SunOS 3.2? Has anyone done this,
and is there anything special I need to be careful of?
Matt Landau
mlandau@bbn.com
------------------------------
Date: Sun, 04 Oct 87 11:55:17 EDT
From: gorman@braggvax.arpa
Subject: Information about RTI's Ingres?
I am posting this to several lists, some which I do not regularly read, so
I am requesting that any replies be sent directly to me at my ARPANET
address.
Several months back, I read a small note in either UNIX Review or UNIX
World stating that RTI had demonstrated its distributed Ingres product--at
either UniForum or USENIX during the winter conferences in
Washington--running on a network on Suns, MicroVAXen, and PCs. The
demonstration involved some level of remote queries and updates from the
different machines to a distributed database avialbale to all the
machines. I plan to call and speak to RTI about its product, but if
anyone, other than employees of RTI, can share any observations about this
product, I would be grateful. The sort of information I'm looking for is:
Were the VAXen were running VMS or Ultrix or UNIX 4.X? Were the devices
on the same network or separate networks gateway'ed together? Were the
systems all using the same network protocols, like TCP/IP?
Thanks in advance,
Bryan Gorman
gorman@braggvax.arpa
------------------------------
Date: Mon, 5 Oct 87 18:18:58 +0100
From: Knut Barra <barra%vax.nr.uninett@tor.nta.no>
Subject: 'as'-syntax for 020-specific adressing modes?
What is the 'as'-syntax for the 68020-specific adressing modes? These are
the Memory Indirect Post- or Pre-indexed and the PC-Memory Indirect Post-
or Pre-Indexed. The documentation that followed our Sun3/50 is not
properly updated at this point.
------------------------------
Date: 7 Oct 87 01:41:04 GMT
From: munnari!pjr@uunet.uu.net (Peter Rayner)
Subject: Porting GKS ncar to Suns?
We are contemplating porting the current implementation of the ncar
graphics package to the sun. This is the gks implementation. We have the
1981 version ported by Minesoft but we are interested in upgrading.
Before we sort out licensing and carry out the port: has anybody else done
it? If anyone has any information or is interested in what we're doing,
e-mail me directly.
Peter Rayner
------------------------------
Date: Tue, 6 Oct 87 15:21:21 EDT
From: umix!itivax!lokkur!scs@rutgers.edu
Subject: UUCP over Ethernet and X.25?
For various good reasons, we're interested in using uucp over ethernet and
X.25 links. From references to honeydanber UUCP it seem to be able to
handle X.25; we're hoping this capability is more widespread. In a
nutshell, here's what we've got:
Two sites, physically separated by 1000 miles. We have a dedicated 56Kb
line which is used for a virtual X.25 circuit. A portion of the bandwidth
is reserved for direct connections, and is essentially free between the
two sites (line rental costs only). Another portion is used by a BBN X.25
general network connection, and every packet costs $$$$. Sun 3/75s are on
both ends of the circuit, both running Sun X.25 software. We'd like to
batch up uucp jobs (news and software distributions both) to run late at
night over the "free" line. For the record:
1) rcp does *not* work over the X.25 links, and requires a degree of
openness we don't want on our systems.
2) ftp is possible for the software distributions, but doesn't seem
to deal well with news. In addition, we don't want to have user
have accounts going and coming on all the system. Uucps
spool-oriented "put file X on system Y with name Z" just exactly
fits the bill.
If anyone has any experience with this we'd be most interested.
We are sending this message both to the Sun hotline and to SunSpots
mailing list.
Steve Simmons
scs@lokkur.UUCP
------------------------------
Date: Wed, 7 Oct 87 09:49:02 EDT
From: Ned Danieley <ndd@sunbar.mc.duke.edu>
Subject: schematic capture/pc board layout?
We are interested in obtaining a program or programs to do schematic
capture and/or pc board layout on a Sun workstation. Does anyone know of
any such programs (preferably under $5000)?
Ned Danieley (ndd@sunbar.mc.duke.edu)
Basic Arrhythmia Laboratory
Duke University Medical Center
Durham, NC 27710
(919) 684-6807 or 684-6942
------------------------------
Date: 4 Oct 87 08:46:56 GMT
From: kddlab!kato%etlcom.etl.junet@uunet.uu.net (Toshikazu Kato)
Subject: Apollo & Sun for image processing, summary
Dear everyone!
I have received many responses on my previous article. Thank you so much!
This is the summary of these e-mails.
[0] My original question is;
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;
.....
[1] John W. Peterson suggests the software libraries "Utah Raster
Toolkit" developed at Utah Univ. (Raster Toolkit seems one of the PDS:
public domain software). It is as follows;
The Utah Raster toolkit is a collection of programs and C routines for
dealing with raster images commonly encountered in computer graphics. It
provides the following major functions:
* A device and system independent image format for storing images
and information about them. Called the RLE format, it uses run
length encoding to reduce storage space for most images.
* A library of C routines for reading, writing and manipulating
images stored in the RLE format.
* A collections of programs for manipulating and displaying RLE
images.
[2] Jim Helman told me of another image processing software;
As far as I know there aren't any standard libraries for image processing,
in part because of the specialized hardware required to do it rapidly,
e.g. the Vicom and IBM each has it's own software. There are very
complete packages for image transformation such as GIPSY, written in
Fortran which should run on anything with a f77 compiler, e.g. Sun, Apollo
or Vax, but it is much slower than using specialized hardware like the
Vicom or IBM. Vicom has a product out that is based around a Sun
workstation that could be interesting (C or f77 programmable, fast for
convolutions and filtering, but expensive).
.....
I don't know about the 2-D standards, such as Core and GKS, but for 3-D
work, PHIGS seems to be emerging as a standard. We plan on using
Template's version called Figaro, on our Suns and Irises. I think it also
runs on Apollos. There is also a nascent X/3D standard that is supposed to
integrate X window with PHIGS.
[3] Tait Cyrus also informed me the XVision project at New Mexico Univ.;
Here at the University of New Mexico, a research project (which I am NOT
affiliated with) has written a software package called XVision. This
package runs on Sun's & uvaxII's under the X window environment and allows
the user to numerous image processing tasks.
XVision is a research and development tool designed to:
* facilitate investigation into new computer vision/image
processing techniques and
* facilitate the development of turn key vision solutions for
various application areas (automation, medicine, biology,
astronomy, etc).
XVision utilizes the X Window System, which provides a network transparent
windowing environment, and will soon be interfaced to frame rate image
processing hardware to provide the vision programmer with a high-level,
interactive vision system.
[4] I have received other short comments on image processing environments
on WS.
Thank you so much, again!
Toshikazu KATO
Information Systems Section, Electrotechincal Laboratory, Japan
JUNET: kato@etl.junet
CSNET: kato%etl.jp@relay.cs.net
------------------------------
End of SUN-Spots Digest
***********************