Sun-Spots-Request@RICE.EDU (Vicky Riffle) (08/17/87)
SUN-SPOTS DIGEST Monday, 17 August 1987 Volume 5 : Issue 33
Today's Topics:
Administrivia
Open question on NFS, efficiency, etc.
Strange ftp problems
Answer to ftp problems
Re: Problems with yellow pages in SunOS 3.2
Re: rwhod not broadcasting on proper address
vadvise
multiple-machine executables
Re: Video recording from a Sun workstation
OCLI coating for Hires mono
Re: PIC tool or TeX tool?
Re: dsun on SunView
netstat -I missing in Sun O.S.
problems with ICMP routing on Sun 3.4
Root over NFS in SunOS 4.0?
Sun III/60M-4-P10 vs. VaxStation 2000?
Machine performance questions?
Sun OS 3.2 Assembler Problem?
----------------------------------------------------------------------
Date: Mon, 17 Aug 87 10:55:50 CDT
From: William LeFebvre <Sun-Spots-Request@Rice.edu>
Subject: Administrivia
Some of you may have received multiple copies of issues 32 and 31. My
apologies for this. It was caused by a combination of name resolver
problems and a smidgen of stupidity on my part. I will make sure that it
will not happen again. I also inadvertently left in a Berkeley From line
in issue 32. Again, I apologize and I hope that it didn't cause too many
problems down the line.
People have been submitting icon bitmaps to this list since issue 25. I
have collected up these icons and placed them in our public FTP area in
the directory sun-icons. The files are named as follows: wnl-name.icon,
where "wnl" are the initials of the submitter and "name" is some one-word
description of the icon's use. Here is an example: "dpz-rlogin.icon".
This directory, like its counterpart sun-source, contains "00index" and
"00directory" files.
I am still working on setting up an archive server. More on that when it
becomes available.
William LeFebvre
<phil@Rice.edu>
------------------------------
Date: Mon, 10 Aug 87 11:32:28 PDT
From: ultra!wayne@ames.arpa (Wayne Hathaway)
Subject: Open question on NFS, efficiency, etc.
While waiting at my diskless Sun workstation for a "cp" to complete, an
obvious question wandered into my otherwise empty head. When I do "cp a
b" on my workstation under NFS, what is in fact happening is that disk
blocks are being read on the server, forwarded over the Ethernet to my
workstation, processed briefly, sent back across the Ethernet to the
server, and finally written back to the same disk they started from; a
rather tortuous path at best. So I did a quick set of timings on a 2.3
megabyte file: the local "cp" gave 0.0u, 4.0s, 1:26 elapsed. To get
timings for running the "cp" directly on the server (via "rsh") I timed
both locally ("time rsh server cp a b") and remotely ("rsh server time cp
a b"); local was 0.0u, 0.3s, 0:21 elapsed and remote was 0.0u, 2.1s, 0:15
elapsed. I don't know how to time the server's part of a local "cp", but
looking at "perfmeter" didn't show any significant difference in CPU
usage. So in other words, the time perceived by the workstation user went
from 86 seconds down to 21 seconds, with no apparent increase in server
load! Given this, has anybody taken the obvious step of automating this
process, sort of like having "cp" say "hmm, both these files are on the
same server and the file is big enough to cover startup time; I'll just
have him do the copy for me"? (I realize that "rcp" already does
something like this for "remote-to-remote" situations, but then NFS is
supposed to make things more transparent, right? And by the way, "time
rcp server:a server:b" gave 0.1u, 0.8s, 0:36 elapsed; not as good as "rsh
cp" but still a lot better than "cp"!)
Anyway, I realize this is the tip of a rather large iceberg (the "when do
you move the data to the computation and when do you move the computation
to the data?" question), and I have in fact been chipping away at that
iceberg myself (for example, I already have procedures set up to do
lengthy "find"s on the server itself). But I was just curious if anybody
has done anything "scientific" about this rather limited version of the
general question? Or UNscientific, for that matter (like my proposed
hacking of "cp" above). Sure seems like something obvious to look at for
(e.g.) SunOS.
Well, my "cp" is finally done, so back to work ...
Wayne Hathaway ultra!wayne@Ames.ARPA
Ultra Network Technologies
2140 Bering drive with a domain server:
San Jose, CA 95131 wayne@Ultra.COM
408-922-0100
------------------------------
Date: Mon, 10 Aug 87 20:42:06 EST
From: bhc@l.cc.purdue.edu (Sam Kimery)
Subject: Strange ftp problems
I have been having real problems with ftp running under SUNos 3.3. I have
a 3/280 server and 9 3/50 clients and a single 3/110 client. I can make
connections from any of the machines, but trying to perform an operation
that requires a data connection (get, put, ls, etc) from any of the
clients, dies with the following message:
"ftp: bind: can't assign requested address"
commands (ls, get, put, etc) that open data connections initiated from any
host not in this group of SUNs works fine.
After several call to the USA4SUN hotline, they are VERY confused. They
claim that either /etc/services or /etc/hosts has been scrambled by YP for
the clients. They have not. A ypcat of either gives a clean listing.
For kicks I crufted a simple program to request the port address of both
FTP and FTP-DATA using the getservbyname() call. It also returns correct
information (0x20 and 0x21).
At this point I'm desperate. Any ideas?
Sam Kimery bhc@l.cc.purdue.edu
System Mangler
Purdue University
Department of Mathematics
------------------------------
Date: Wed, 12 Aug 87 09:46:24 EST
From: bhc@l.cc.purdue.edu (Sam Kimery)
Subject: Answer to ftp problems
I finally received an answer from the USA4SUN hotline concerning the ftp
problem I reported earlier. The conclusion is that even though 3.3 is
claimed to support subnets, it really doesn't. None of the network tools
(ftp, telnet, etc) had the code added to provide proper subnet support.
Thus my internet addresses of the form 128.210.3.X broke ftp.
This is supposedly fixed in 3.4.
Sam Kimery kimery@helicon.math.purdue.edu (or)
System Admin. bhc@l.cc.purdue.edu
Purdue University
Mathematics Department
------------------------------
Date: Wed, 12 Aug 87 12:40:03 -0700
From: Craig Rolandelli <craig@rome.uci.edu>
Subject: Re: Problems with yellow pages in SunOS 3.2
> On a couple of occations we've created a user with a name that is a prefix
> of another user name. In that case it sometimes happens that the new user
> is prohibited from logging on. (It seems as if the wrong entry is
> retrieved from the "dbm" file.) It happened with the names "steinar" and
> "stein", but I wasn't able to provoke the error with the names "ab", "abc"
> ... "abcde".
>
> Has this problem occurred to anybody else?
Yes, we had this problem about 6 months ago. But it only happened when
the person with the shorter name tried to change thier password. The
problem is that yppasswdd uses strncmp instead of strcmp. This is what
Sun told me (I do not have sources). The guy I talked to said it must
have been done for speed, but that it is a bug. He said it should be
fixed in 3.4.
As to why it happens when someone logs in I do not know as it uses ypserv
and dbm to retrieve that information.
There are two solutions to the problem:
1) Upgrade to 3.4. We have not done this, so I do not know if it was fixed.
2) Make sure that your master yellow pages password file is sorted by login
name length. With the shortest at the top of the file.
Craig Rolandelli (craig@ics.uci.edu, ucbvax!ucivax!craig)
Computer Science Dept.
University of California, Irvine
Irvine Ca. 92717
(714) 856-4222
------------------------------
Date: 13 Aug 87 00:27:41 GMT
From: elroy!david@csvax.caltech.edu (David Robinson)
Subject: Re: rwhod not broadcasting on proper address
From: hi!kurt@hc.dspo.gov (Kurt Zeilenga)
>
> We are configuring our hosts (Release 3.4) like:
> /etc/ifconfig ie0 sunhost netmask 255.255.248.0 broadcast 129.24.15.255 \
> -trailers up
> where the address of the hostname would be something like:
> 129.24.13.3 sunhosts
> We broadcast on the subnet address 129.24.8 on host all '1 (ie:
> 129.24.15.255) as configured above. This is to conform with Internet
> Standards. On our Ultrix and 4.3BSD systems this works fine.
> On our SUNs, in.routed seems to understand this, yet in.rwhod doesn't.
> Rwho seems to ignore the broadcast address ifconfiged in. Do you have a
> fix for this or have any suggests?
There are a number of subnet bugs in 3.3 and 3.4 that have to do with
broadcasts. In 3.3 the following programs did not use the broadcast
address: /etc/umount /etc/ypbind /usr/ucb/rup /usr/ucb/rusers
/usr/etc/in.rwhod. I had reported all of them to Sun and had gotten fixes
and promises that they would be in the 3.4 release with the exception of
in.rwhod which was not reported until 3.4 was out of engineering and the
whole thing is to be replaced with 4.3bsd in.rwhod in SunOS 4.0.
In the 3.4 documentation the bugs are noted but unfortunately someone at
Sun goofed and left the new binaries off of the 3.4 upgrade tape. I have
reported this and they now know about it. You may wish to contact
Software Support and ask for a patch release that contains the missing
binaries, with the exception of in.rwhod. If you have the 4.3bsd sources
it is trivial to port rwhod to SunOS 3.4 and will give you the correct
behaviour.
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: 12 Aug 87 11:49 EST
From: barnett%vdsvax.tcpip@ge-crd.arpa
Subject: vadvise
I remember a benchmark that someone wrote that accessed a large 2
dimensonal array along the diagonal. The peculiar thing was that when the
program went back down the same diagonal, the Sun OS (2.0) started to
thrash.
The execution time was ten (or more) times slower. I think the look-ahead
paging scheme was flailing in this case.
The solution was to include a
vadvise(VA_FLUSH)
before traversing down the sparse matrix.
See /usr/include/sys/vadvise.h
Maybe this will help?
Bruce G. Barnett <barnett@ge-crd.ARPA> <barnett@steinmetz.UUCP>
------------------------------
Date: Tue, 11 Aug 87 05:10 EDT
From: SIMON%M_SCRVX2%sdr.slb.com@relay.cs.net
Subject: multiple-machine executables
Bill Mitchell made a comment about the inconvenience of Sun's
object-incompatible hardware generations (Sun 2, Sun 3, Sun 4, ???)
While I don't agree with the suggestion that executables should contain
code for all the different architectures, the problems are real
nonetheless and can only become more severe with the advent of RISC
processors. It's not that one has to recompile programs for a new
generation, which one does, or store an additional copy of the operating
system, which one does, but that when a LAN-full of different workstations
is being supported, the ability to move easily from machine to machine is
lost. I'm sure that this will make many Sun users think twice about
installing the next generation.
It may be timely to recall one of the main reasons that microcode was
invented: to support the same instruction set on different hardware. In
adopting RISC, computer manufacturers are able to design faster hardware
sooner by sacrificing object code compatibility.
[usual disclaimers]
Simon Barnes
simon%m_scr%sdr.slb.com@relay.cs.net
Cambridge, U.K.
------------------------------
Date: 12 Aug 87 18:52:55 GMT
From: ralphw@ius2.cs.cmu.edu (Ralph Hyre)
Subject: Re: Video recording from a Sun workstation
BSD@PSUVM.BITNET (Scott Dickson) writes:
>We want to be able to get video output from our Sun workstations.
If you want to tape the signal coming from the Sun frame buffer, it is
66Hz refresh at some unusual scan rate. Flicker is objectionable, and
scan rate conversion is real expensive. We're currently trying to fool
SunView into using our pixrect-supporting frame buffers. (Matrox MIP-512
boards). [Does anyone know what's needed, other than the FBIOGETTYPE
ioctl?] One easy thing to do is setting the pixwin's backing pixrect to
the MIP-512 pixrect. This has some drawbacks, though. It's hard to get a
handle on the pixwin struct from outside the program. The other solution
is to recode the program to use pixrects only. -- - Ralph W. Hyre, Jr.
Internet: ralphw@ius2.cs.cmu.edu Phone:(412)268-{2847,3275} CMU-{BUGS,DARK}
Amateur Packet Radio: N3FGW@W2XO, or c/o W3VC, CMU Radio Club, Pittsburgh, PA
------------------------------
Date: 10 Aug 1987 18:34 EDT
From: Ralph.Hyre@ius2.cs.cmu.edu
Subject: OCLI coating for Hires mono
I've heard a rumor that the hires tube is no different than normal, just a
couple of boards inside the case are changed to allow the doubled scan
rates. This would imply that the no OCLI coating is a marketing decision
rather than a technical one.
Has anybody out there upgrade from a Sun-3/160 to Sun-3/200 machine? Does
Sun give you those boards which accept the higher scan rate with the
upgrade kit? Seems like nobody would want to waste their old monitor. We
certainly don't.
------------------------------
Date: Thu, 13 Aug 87 13:04:50 CDT
From: knutson@huey.cc.utexas.edu (Jim Knutson)
Subject: Re: PIC tool or TeX tool?
There is a tool called fig that was written by Supoj Sutanthavibul
(supoj@sally.UTEXAS.EDU) here at the University of Texas. It will allow
you to create pictures and output the appropriate PIC commands for
troffing.
I don't know if he is still around or not, but try there. Also, the Sun
Users Group may have this or other tools like it on their user contributed
software tape.
Jim Knutson
knutson@ngp.utexas.edu
------------------------------
Date: Wed, 12 Aug 87 18:24:15 pdt
From: Jeff Lo <sdcsvax!ames!elan!jlo@ucbvax.berkeley.edu>
Subject: Re: dsun on SunView
sdcrdcf!eggert@cam.unisys.com (Paul Eggert) said:
>
> In Sun-Spots v5n29 tamir@cs.ucla.edu (Yuval Tamir) asks whether anybody
> has gotten the old dsun program to work under SunView. The following
> fixes should get dsun.tool to work under Sun UNIX Release 3.4....
>
> Surely something better than this should be available these days!
Something better is available. Elan sells a product called Eview/Sun to
go along with EROFF (see below). It supports all of DWB 2.0 and Elan's
enhancments. It allows you to zoom the image to see the entire page at
once on the screen, see about half well enough to easily read, or zoom
in for a view at full (300dpi) resolution.
Elan Computer Group, Inc. sells a package called EROFF which is the AT&T
Documenter's WorkBench (DWB), including device independent troff, nroff,
tbl, eqn, pic, grap, macros, etc., plus bug fixes and enhancements
including bitmap graphics inclusion, and a driver for your choice of
printer (All HP LaserJet models, PostScript printers, or Imagen Impress
printers). Automatic HP Soft Font downloading is done for the LaserJets.
DWB and the printer drivers are also available separately.
Please write or call for more information.
Jeff Lo
ELAN Computer Group
410 Cambridge Avenue, Suite A
Palo Alto, CA 94306
(415) 322-2450
..!{ames,hplabs}!elan!jlo
------------------------------
Date: Tue, 11 Aug 87 17:44:18 EDT
From: Terry Slattery <tcs@cad.usna.mil>
Subject: netstat -I missing in Sun O.S.
In attempting to debug a network related problem here tonight, I wanted to
run netstat on a 3/50 with release 3.3 for a specific interface (le0 vice
lo0). To my dismay I find that the "-I interface" option to netstat has
been discarded in favor of the following (copied from the Sun netstat
manual page):
This display consists of a column summarizing information for all
interfaces, and a column for the interface with the most traffic
since the system was last rebooted.
Not being able to select the interface being displayed on the 3/50 isn't
that much of a big deal since there are only two. However, on the larger
models, it may present a problem. Would someone who knows (preferably
from Sun) care to comment on why it is desirable to not be able to select
the interface to be displayed?
-tcs
------------------------------
Date: Wed, 12 Aug 87 15:34:53 EDT
From: dms@wheaties.ai.mit.edu (David M. Siegel)
Subject: problems with ICMP routing on Sun 3.4
We have a subnetted class B network, and configure our Suns with a netmask
of 0xffffff00. Our client Suns are gatewayed to a Sun backbone, where
their server is connected to both the client subnet and the backbone
subnet.
When a client attempts to ping another host off its cable, the ping always
failed, however, telneting, etc, to that host worked. It seems that icmp
requests were not being routed correctly. I took a look in
netinet/raw_ip.c and sure enough, the socket option to do routing was
being ignored. However, when I changed the code to do the routing
correctly, something really wiered happend: I could ping hosts that didn't
exist on other server's client subnets!
In any case, does someone know how to correctly fix Suns network code to
make it route ICMP the same way it routes TCP/UDP?
-Dave
------------------------------
Date: 10 Aug 87 15:52 EST
From: montnaro%sprite.tcpip@ge-crd.arpa
Subject: Root over NFS in SunOS 4.0?
We are told that in SunOS 4.0 ND will be scrapped in favor of NFS-mounted
root and swap partitions. One thing that currently is a nagging problem is
that when a root user reads or writes files to an NFS-mounted partition,
the effective user id is "nobody". Does anybody know how this will be
gotten around in 4.0 without a major security breach?
Thanks,
Skip Montanaro (montanaro@ge-crd.arpa or uunet!steinmetz!desdemona!montanaro)
------------------------------
Date: 6 Aug 87 14:36:56 GMT
From: mimsy!mmlai!aplcen!jhunix!msc_wmhs@seismo.css.gov
Subject: Sun III/60M-4-P10 vs. VaxStation 2000?
Our department (math sciences, johns hopkins u) wants to buy a stand-alone
workstation. Our mail applications will be: numerical mathematics
(optimization, statistics), text processing (TeX), and general utility
computing (communications, uucp, file transfer). We are considering two
comparably loaded and priced machines---Sun III/60M and VS2000 (6+ Megs
memory, 150 + Disk, ethernet, FPA, and Tape backup system). We will want
to connect a laser printer (LaserWriter, I assume).
We are not system's programmer types, but we have had some previous
experience maintaining our own unix machines.
I would greatly appreciate some insight and comments to help make this
choice. Are there functional differences between the capabilities of
these machines? Is one easier to maintain?
I will summarize responses if it seems appropriate. I apologize if this
inquiry is not appropriate for these newsgroups.
Thanks.
michael schneider
msc_wmhs@jhunix.bitnet and msc_wmhs@jhunix.uucp
------------------------------
Date: Tue, 11 Aug 87 09:46:24 EDT
From: sunlakes!xox!bigbird!jeff@sun.com (jeff henslin)
Subject: Machine performance questions?
I am thinking of purchasing several Sun 3/60's, but am concerned about
possible performance problems.
My Sun sales rep believes I may purchase and comfortably use three 3/60's
each with 16Mbytes of memory, all served by a 3/160 with an eagle disk.
The server is currently serving only a 3/75. Each of these client machines
must be capable of running at least one Lisp session (preferably two on
some machines). Can a 3/160 with 16Mbytes of memory serve a 3/75, and 3
3/60's all of which are running Lisp? My sales rep says that with clients
running on 16M memory, swapping will pose little problem, but I am
wondering ...
If anyone has a similar set-up, or any information, I would appreciate
comments and advice. Thanks,
Jeff Henslin
XOX Corporation
UUCP: ...!sun!sunlakes!xox!jeff
PHONE: (607) 277-6661
------------------------------
Date: Wed, 12 Aug 87 16:58:39 EDT
From: "Timothy P. Donahue" <tdonahue@lf-server-2.bbn.com>
Subject: Sun OS 3.2 Assembler Problem?
I'm looking for some help in figuring out the correct Sun OS 3.2 assembler
syntax for the MC68020 bit field instructions, such as BFFFO. Page 59 of
the Assembly Language Reference Manual gives the following:
bfffo dn,ea{n,n}
We have tried this (and several dozen variations) and received the
"Invalid operand" error each time. Curiously, after a day of trying, we
were unable to get the Sun C compiler to generate any 68020 bit field
instructions...leading us to believe that they are not working in the
assembler.
Does anyone out there in netland know the correct syntax? We'd love to
know.
Thanks in advance
Tim
P.S. Please send flames about assembly to comp.arch, there's a nice
religious war going on there right now. :-)
------------------------------
End of SUN-Spots Digest
***********************