Sun-Spots Digest, v6n217
William LeFebvre
Sun-Spots-Request at Rice.edu
Tue Sep 6 09:27:15 AEST 1988
SUN-SPOTS DIGEST Sunday, 4 September 1988 Volume 6 : Issue 217
Today's Topics:
Re: SunOS 4.0 -- initgroups error (2)
Re: SunOS 4.0 Source Code licensing problem
Re: Yellow Pages problem
Sun 4/110 surprises
Using automount
Feature (or bug?) in SunOS 3.5 Pascal Compiler (pc)
clock losing one month
automount questions (SunOS 4.0)
Is /Tty/Bold_style broken in 4.0?
SMD on 4/110?
how do i receive multicast packets?
Non-Sun memory for Suns?
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: Fri, 02 Sep 88 13:44:32 -0700
From: bien at aerospace.aero.org
Subject: Re: SunOS 4.0 -- initgroups error (1)
> If a user is in more than 8 groups, they cannot log into a SunOS 4.0
> system. Root also cannot su to that user id.
Change this to "If a user is in more than 8 groups *and* the users home
directory and/or login shell are on NFS mounted file systems, then the
user cannot log into a SunOS 4.0 system and you cannot su to that user
id."
A bug report has been filed at Sun.
--Cheryl Bien
The Aerospace Corporation
bien at aerospace.aero.org
------------------------------
Date: Sat, 03 Sep 88 10:13:40 EST
From: trinkle at purdue.edu
Subject: Re: SunOS 4.0 -- initgroups error (2)
Sun did us a favor by increasing the number of groups (NGROUP) to 16.
This works fine on the local machine, but NFS still restricts the number
of groups. Now this may be a restriction only when talking to non 4.0 NFS
servers, but I don't think so. Even though Sun would like to think the
whole world is Suns, it isn't. We have many other machines that run NFS
(supposedly a standard).
On our Mt. Xinu systems I fixed this so that only the first 8 groups are
used by NFS (RPC) authentication. This not the "best" solution, but it
worked better than Sun's method (breaking NFS completely). We later just
went back to a maximum of 8 groups everywhere. We found everybody could
live with that.
Daniel Trinkle trinkle at cs.purdue.edu ARPA
Department of Computer Sciences trinkle%purdue.edu at relay.cs.net CSNET
Purdue University {ucbvax,decvax}!purdue!trinkle UUCP
West Lafayette, IN 47907 (317) 494-7844 PHONE
------------------------------
Date: 2 Sep 88 19:59:52 GMT
From: ehrlich at blitz.cs.psu.edu (Dan Ehrlich)
Subject: Re: SunOS 4.0 Source Code licensing problem
Rererence: v6n204
weiser.pa at xerox.com writes:
>SunOS 4.0 source apparently requires an AT&T System V rel 3.1 source
>license. Notice that ".1". We have a Vr3 license (required for 3.5
>source), but not a Vr3.1 license, and so Sun will not ship us the 4.0
>source until we upgrade yet again. ...
*** FLAME ON ***
Actually we were told that we needed a Sys V rel 3.2 license. We should
have recieved the source tapes months ago but our sales rep messed up.
First, when the order was taken no license was requested. Sun's leagel
beagel in CA asked for a license. As the sale person did not specify that
we need release 3.2 we, also, sent along a copy of our 3.0 license. After
another two weeks we get a call that we don't have the right flavor System
V license. Blood pressure is starting to rise by now. As the UNIX
license is held by our University and not the CS deparment it took an act
of god to get the update pushed through in less than a month. Sun has now
had the correct System V Release 3.2 license for a week and we still don't
have any source tapes. Talking to them doesn't do any good. Maybe we
should start talking to MIPS and Encore instead (dare I include DEC???)
???
*** FLAME OFF ***
> ... We are doing this, but it would have
>been much more convenient and faster if Sun had announced this up front...
Sure would have, but then there would not have been the dramatic increase
in the sales on antiacids and assorted other stomach upset remedies.
>(I am disappointed that Sun once again, as they do every major new
>release, alienates its most influential customers (the outspoken, trusted,
>techies) by not taking source code availability seriously.)
*** FLAME ON ***
Here, here!! I agree with Mark completely on this. Sun is getting
entirely too arrogant towards their customers. They have lost site of
their customers needs (including source code availablity). Then again
they are also probably cowering a bit under that 20% stock holding that
AT&T has acquired. They seem to have lost a lot of the entrepenurial
spirit now that they have a big brother standing near by with a very large
stick watching their every move.
*** FLAME OFF ***
Dan Ehrlich <ehrlich at blitz.cs.psu.edu>
The Pennsylvania State University
Department of Computer Science
University Park, PA 16802
------------------------------
Date: Sat, 03 Sep 88 10:04:50 EST
From: trinkle at purdue.edu
Subject: Re: Yellow Pages problem
> Hi. When I do a make after changing Yellow pages password, the following
> message appears:
> updated passwd
> Can't get an address for server .
> Can't get an address for server .
> pushed passwd
Your ypserver map is not set up properly. It is the YP map of all slave
YP server to whom you are going to push copies of the YP maps. There are
instructions in the YP manual sections for updating this map. Another way
to do it is to rerun ypinit on the master.
Daniel Trinkle trinkle at cs.purdue.edu ARPA
Department of Computer Sciences trinkle%purdue.edu at relay.cs.net CSNET
Purdue University {ucbvax,decvax}!purdue!trinkle UUCP
West Lafayette, IN 47907 (317) 494-7844 PHONE
------------------------------
Date: Sat, 3 Sep 88 16:32:43 PDT
From: Paul O'Neill <pvo3366 at neptune.oce.orst.edu>
Subject: Sun 4/110 surprises
Reference: sunspots v6n201 & v6n215
jeff at kestrel.arpa (Jeff Kitson) writes:
>Subject: why is system time so high for TeX on a Sun 4/110?
>...
>Anyone know why the system time is so high on the 4/110, or what should be
>done to improve it?
*AND*
roberts%studguppy at lanl.gov(Doug Roberts @ Los Alamos National Laboratory) writes
>Subject: Non-Sun SCSI disks & Sun4/110
>....(see below)....
THAT DOES IT!! I've just got to post this.
You probably won't learn the following till you open your 4/110 box (or
read this):
a VISUAL INSPECTION:
The cpu board is double-height. There's only 1 empty slot in your
three-slot backplane.
It has thin ethernet (as well as thick).
exerpts from the HARDWARE README FIRST:
...
SCSI Disk Use
The Small Computer Systems Interface(SCSI) connector on the Sun 4100 CPU
board installed in your 4/110 system may not adhere completely to the SCSI
specification. Pin 26 on the SCSI connector may be grounded. Do NOT use
non-Sun SCSI disk drives with any Sun 4/110 or any 4100 CPU based system.
Use of a SCSI drive not purchased from Sun may result in the damage or
destruction of the disk drive/s. {not a problem--pin 26 is termination
power--configure your SCSI peripheral to supply it's own local termination
power--pvo}
VME Information
The CPU in the 4/110 uses a VME Master and Arbiter only. VME devices may
not execute Direct Memory Accesses into the CPU board's memory. (They may
however, DMA into other memory on the VME bus). Data is movable to and
from the VME bus by program I/O only.
The Sun 4100 CPU board's Memory Management Unit (MMU) does not have bits
28 through 31. Bit 27 is extended to encompass bits 27-31 on the VME
address lines. Therefore, /dev/vme32dxx is not a true 32-bit address bus,
instead you may use /dev/vme24d32.
...
FPU Option Deficiency
The following information applies only if you have an FPU equipped 4100
CPU board, and only if the board or FPU upgrade was ordered before October
1st 1988. There are several deficient performance factors in FPU chips
shipped prior to October 1988. These factors will cause the chip to
under-perform its design specification during heavy "number crunching" or
benchmarking processes. There should be no indication of deficient
performance during normal operation of the system.
Jeff--your 4/110 doesn't have an FPU, does it? (please e-mail me an answer)
(He did--this was the problem)
We delayed our FPU delivery until November after reading the above and
then discovered that the floating-point software emulation on the 4/110 is
absurdly slow.
Large FORTRAN applications and simple FORTRAN & C floating-point
benchmarks show the following speed for an FPU-less 4/110:
FORTRAN C
----------- -----------
4/110 vs. 3/50 6 x slower 14 x slower
4/110 vs. 3/50 w/ 68881 22 x slower 96 x slower
4/110 vs. 3/280 12 x slower 27 x slower
4/110 vs. 3/280 w/ 68881 30 x slower 131 x slower
4/110 vs. 3/280 w/ fpa 102 x slower 455 x slower
That's right, your screaming SPARC is 10x slower than a 3/50 when it does
FP in software. It's wonderfully fast for Suntools and system stuff, but
forget doing any math on it. This is what you'd expect for RISC--farm
that complicated stuff out to specialized hardware. I'm not bummed at Sun
that it's so slow, I'm annoyed that no one at Sun warned me it would be so
slow w/o an FPU.
Finally, to answer Jeff's system-time question. There are no compile time
floating-point options on sun4's. The FPU is assumed to be there. If
it's not, then the *KERNEL*, not linked library stuff, emulates the FPU.
This time gets logged as system time, not user time. You can see the
kernel stuff in SunOs 4.0 /usr/share/sys/sun4/OBJ. I've never used Tex.
I'm surprised a formatting package uses floating-point, but I'll bet
that's what's happening. Solution--order an FPU.
Our Technical Sales Engineer is bringing an FPU by in a couple days so we
can see how bad this "performance deficiency" is. Stay tuned for 4/110 w/
FPU vs. the above machines.
editorial comment--This SunStuff is the best computer gear I've used in 20
years. I'm afraid that most of the Sun-bashing that's done in this group
is by folks who don't realize that even simple systems have bugs. I'm
always amazed that anything this complex running anything as mind-boggling
as Unix runs at all. I'm glad I didn't have to design/write/debug/document
all the tools, calls, drivers, etc. that do work.
Paul O'Neill pvo at oce.orst.edu
Coastal Imaging Lab
OSU--Oceanography
Corvallis, OR 97331 503-754-3251
------------------------------
Date: 3 Sep 88 11:19:53 PDT (Saturday)
From: "Larry_Masinter.PARC"@xerox.com
Subject: Using automount
I've found automount quite useful just using
automount /net -hosts
I set up all the other links using symbolic links, e.g., if I want
/usr/local/lde to point to the directory on a server mas, then I just put
a symbolic link
ln -s /net/mas/usr/local/lde /usr/local/lde
What would be the advantage of other mount points?
------------------------------
Date: Fri, 02 Sep 88 15:46:00 CDT
From: AARON KONSTAM <79343382 at TRINITY.BITNET>
Subject: Feature (or bug?) in SunOS 3.5 Pascal Compiler (pc)
When one runs the following program through pc with the -L and -C option
you get the run time error: Subrange or array subscript is out of range.
__________
program test(input,output);
const
M=50;
var
L:0..M;
j:1..M;
begin
L:=0;
for j:=1 to L do write(' ',j);
end.
__________
It took a great deal of head scratching to realize that what was being
objected to is that j has a subrange that does not include 0 and L had a
value of 0. The fix is to change j to have the subrange: 0..M
I assume defining j to be just integer would also work. I believe one
could get an argument on any street corner among Pascal users and compiler
writers whether this a feature or a bug. By the way TurboPascal 4.0 acts
the same way. Earlier versions of TurboPascal did not treat this as an
error.
Aaron Konstam
Trinity University
San Antonio, Tx
------------------------------
Date: Sat, 3 Sep 88 16:36:14 CDT
From: keith at barnum.ksl.umn.edu
Subject: clock losing one month
I have a 3/60 that started mysteriously losing a month when it is
rebooted. Suddenly August 31 becomes July 31! The date command resets
the date and all is fine until the next time its booted.
The 3/60 and 3/50's that sit next to it do not seem to display this
backward behavior.
------------------------------
Date: Fri, 2 Sep 88 16:32:37 PDT
From: mcooper at oberon.usc.edu
Subject: automount questions (SunOS 4.0)
I've got a couple of questions (doesn't everybody?) regarding automount
under SunOS 4.0:
1) We are running 4.0 on several systems that are served by about 6 YP
servers all running SunOS 3.2. I can't seem to get automount to find
anything in a given YP map. It always says it "cannot find" a given
directory. I can ypcat and ypmatch on the given map on the 4.0 host
just fine. Giving it a filename instead of a YP map name works fine.
2) How do you forcibly unmount a filesystem mounted by automount?
(Sometimes I don't want to wait for the timeout). I tried umount'ing
the /tmp_mnt/auto* mount, but that leaves the link in /auto (the
directory name I pass to automount). Since /auto is mount read-only
by automount, I can't get rid of it. Hence, automount never tries to
remount the filesystem since the link is there. Of course, killing
the automount process results in the system waiting for the YP server
to come back and so you're basicly toast at that point.
Overall, I'd say automount is just what we need. It lacks only a little
detailed documentation.
mike
Michael A. Cooper, University Computing Services, U of Southern California
UUCP: ...!uunet!oberon!mcooper BITNET: mcooper at kylara
ARPA: mcooper at oberon.USC.EDU PHONE: (213) 743-2957
------------------------------
Date: Sat, 3 Sep 88 08:32:10 cdt
From: wsmith at m.cs.uiuc.edu (William Smith)
Subject: Is /Tty/Bold_style broken in 4.0?
I created this line with the defaults editor in my .defaults file. I
appears to have no effect in SunOs 4.0. In version 3.5 of the Sun
operating System, it would cause highlighting to be a bold style instead
of reverse video.
/Tty/Bold_style "Offset_X_and_Y"
Do I need to do anything more to make it work? Is this feature broken?
An acceptable alternative would be to turn the cursor into an underscore,
but I prefer the Offset_X_and_Y.
Bill Smith uiucdcs!wsmith wsmith at cs.uiuc.edu
------------------------------
Date: 2 Sep 88 16:05:51 GMT
From: felix!arcturus!dav at hplabs.hp.com (David L. Markowitz)
Subject: SMD on 4/110?
Does anyone out there use SMD disks on a 4/110? I am especially
interested in hearing about configurations based on the Xylogics 753/7052.
Sun doesn't sell a 4/110-SMD configuration. Anyone know why?
We need a cheap, fast NFS server for some 386-based machines. Does anyone
have a better solution than a 4/110-8 w/ SMD disks?
David L. Markowitz Rockwell International
dav at arcturus ...!sun!sunkist!arcturus!dav
------------------------------
Date: Fri, 2 Sep 88 13:43:54 MDT
From: Charles Shub <cdash at boulder.colorado.edu>
Subject: how do i receive multicast packets?
We have a sun 3-160 running sun 3.2 and want to receive multicast packets.
we have "enhanced" the driver (/sys/sunif/if_ie.c) to recognize the
special packets, but somehow or other, the multicast packets are not
making it up to the driver.
When we did a similar thing on a microvax, we had to change the chip reset
software to get the chip to pass the multicast packets. Do we have to do
this on the sun, or is there some other trick?
thanks...
cdash aka cdash at boulder.Colorado.EDU -or- ..!{ncar|nbires}!boulder!cdash
------------------------------
Date: Fri, 2 Sep 88 17:09:51 PDT
From: Joseph LaRocca <hqpyr1!oracle!jlarocca at hplabs.hp.com>
Subject: Non-Sun memory for Suns?
Since it seems nearly impossible to get memory boards for Sun Workstations
from Sun, could anyone tell me another source for them? I've heard
someone mention a company called Helios, but haven't been able to locate
them; anyone have an address?
Thanks in advance,
Joseph T. LaRocca
(415) 598-8191
------------------------------
End of SUN-Spots Digest
***********************
More information about the Comp.sys.sun
mailing list