Sun-Spots-Request@RICE.EDU (William LeFebvre) (08/04/88)
SUN-SPOTS DIGEST Monday, 1 August 1988 Volume 6 : Issue 162
Today's Topics:
Re: login problem when NFS server crashes (2)
3.2 on the 3/60
Sparc Processor Info
Exabyte tape label generator (in perl)
SLIP & TrailBlazer
SunView problems
Graphing program
Sunlink/ibmftp problems
Spurious level 7 interupt?
How to get X10V4 for Sun?
Any Help in SUN VME and SCSI?
Send contributions to: sun-spots@rice.edu
Send subscription add/delete requests to: sun-spots-request@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@rice.edu" or mail the word "help" to the same address
for more information.
----------------------------------------------------------------------
Date: 20 Jul 88 22:25:37 GMT
From: etnibsd!vsh@uunet.uu.net (Steve Harris)
Subject: Re: login problem when NFS server crashes (1)
Reference: v6n143
Thanks, wnl, for the detailed discussion of pwd, getwd, and what happens
when one logs in. Here is a small detail which may be of intrest:
As reported, getwd stats every file in the directory "..", searching for
the entry with the same inode as the current directory "." Note that
getwd does a stat, not an lstat! This means that if, in the directory
"..", there is a symbolic link to the real entry (the one with the same
inode as "."), and the symbolic link precedes the real entry, getwd (and
pwd) will use the name of the symbolic link for that component of the
path. At least, that's the way it seems to work on my system (a 3/260
running 3.4). [[ Yes, this is true. In fact, it was during the course of
Sun-Spots discussion on this same behavior that someone pointed out the
cause of the "hanging login" problem. Once I thought about it, the
problem was obvious (even tho I had been puzzling over it for many
months). --wnl ]]
I use gnutar to do my backups, and was tired of the way it traversed the
/usr.MC68020 tree before /bin, /lib, /etc (note: /usr is a symbolic link
to /usr.MC68020). So, in single-user mode, I re-ordered the entries in /,
putting /usr.MC68020 after everything else. Lo and behold! Now whenever
I do a pwd, instead of "/usr.MC68020/foo/bar", it returns "/usr/foo/bar"!
Somehow, I don't think this is what pwd ought to do; it ought to return
the true path. It would seem to be an easy fix (Sun, are you listening?
:-)
BTW, if I have the C shell "hardpaths" variable set, the "cwd" variable
does not include any symbolic links in its components.
Steve Harris
Eaton Corporation, 108 Cherry Hill Drive, Beverly, MA 01915
1-617-921-0750 ext. 6542
uunet!etnibsd!vsh
------------------------------
Date: Thu, 21 Jul 88 09:40:12 EDT
From: Godfrey Lee <glee%cognos@waterloo.edu>
Subject: Re: login problem when NFS server crashes (2)
In Sun-Spots Digest, v6n143, William LeFebvre <phil@Rice.edu> writes:
>So, if you do what most people do and put all your NFS partitions in "/",
>then you *might* hang when one of those servers goes down.
>So how do you work around the problem? [.....] I was
>able to solve the problem for most of the users by making sure that the
>directory for their file system appeared before any other NFS file systems
>in "/".
Thanks for the explanation, I was wondering about that myself. But, isn't
it easier if one just puts the nfs mounts under a separate subdirectory
under the / directory? e.g. /n/xxxx. I know it is not as nice, and I must
admit that I would probably just live with the problem unless it becomes
unbearable.
[[ Yes, that is easier from an administrative position, but a little
harder from a user's standpoint. As a user, I like having all the
servers' main partitions available at the root (until one of them goes
down, anyway). --wnl ]]
Godfrey Lee P.O. Box 9707
Cognos Incorporated 3755 Riverside Dr.
VOICE: (613) 738-1440 FAX: (613) 738-0002 Ottawa, Ontario
UUCP: uunet!mnetor!dciem!nrcaer!cognos!glee CANADA K1G 3Z4
------------------------------
Date: Thu, 21 Jul 88 14:19:39 EDT
From: steve@umiacs.umd.edu (Steven D. Miller)
Subject: 3.2 on the 3/60
I just passed this along to Charles Hedrick, and I'll pass it along to any
others that might think the information useful: I've ported SunOS 3.2 to
the 3/60. (Hedrick's rule used to be "a 3/60 is a 3/160"; my rule is "a
3/60 is a 3/160, except for the onboard SCSI, where it's a 3/50, and the
hi-res monochrome, where it's a 3/260, and the cgfour, where it's like
nothing else in 3.2.) I have the ported version running on color, hi-res,
and normal mono 3/60s with local disk, though I think it will work on
other 3/60 variants, and it should work fine on other hardware supported
in SunOS 3.2.
I have instructions for those who might wish them. If you don't have
sources, though, my instructions won't help you a bit. (Sorry.) I know
that my 3.2, for which I have sources and to which I've added almost-
reasonable subnet support, works better on a subnetted network than what
comes right off the tape... and Sun didn't tell us that we needed a
System V Release 3.0 license to get SunOS 3.4 or later sources until many
months after we'd placed our order.
-Steve
Spoken: Steve Miller Domain: steve@mimsy.umd.edu UUCP: uunet!mimsy!steve
Phone: +1-301-454-1808 USPS: UMIACS, Univ. of Maryland, College Park, MD 20742
------------------------------
Date: Thu, 21 Jul 88 22:31:03 EDT
From: dff@morgan.com (Daniel F. Fisher)
Subject: Sparc Processor Info
Reference: v6n146
Craig Leres asks:
> If any one knows how to get a processor handbook, please let the rest
> of us in on the secret.
I don't know where to get a processor handbook per se. However, while
searching for the "SPARC Processor Architecture document" referred to in
the Sun-4 Assembly Language Reference, I managed to get 3 documents that
may be adaquate substitutes. One is "The SPARC Architecture Handbook"
Part No. 800-1399-07, Revision 50 of August 8, 1987, which gives a full
description of the instructions, registers, traps, exceptions, error
handling, etc. This was obtained after numerous calls to Sun Support and
my Sun salesperson, while they tried determine whether it existed, whether
it was released, whether it could be released and what to charge for it.
At one point, I was given the phone number of a Fujitsu representative
(Fujitsu is a manufacturer of SPARC chips). They were kind enough to send
me two specification documents, describing the Fujitsu MB86900 High
Performance 32-bit RISC Processor SPARC and the Fujitsu MB86910 RISC
Floating Point Controller SPARC. I will provide contact info for the
Fujitsu Representative to anyone who is interested, though I found the
Fujitsu Spec Sheets to hardware oriented for my needs.
Dan
------------------------------
Date: Thu, 21 Jul 88 15:53:58 PDT
From: mkp@hac2arpa.hac.com (Mike K. Peterson)
Subject: Exabyte tape label generator (in perl)
If you're using one of those new-fangled 8mm videotape backup units and
haven't yet whipped up a label maker for inclusion in the cartridge box,
here's one written in perl.
The PostScript part is basically right out of an audio cassette label
program written by Tom Smith (analog!smith) which was posted to
comp.sources.unix back in February, and tweaked for the dimensions of the
8mm tape box. The script makes labels of the form "X-mm/dd/yy," where X
is the first character of the host name. We keep a list of disk device /
mount point pairs in a file, and list those on the label.
No doubt this will not fit anyone's local customs, but at least provides a
jumping-off place in terms of boiler plate.
Mike Peterson mkp@hac2arpa.com
mkp@scgvaxd.scg.hac.com
---------------------------8><---------------------------------------
#! /usr/bin/perl
#
# This program writes PostScript to generate a label for a
# Exabyte 8mm tape.
#
{
# construct tape number
$initial = substr(`hostname`, 0, 1);
# capitalize as necessary
$initial =~ y/[a-z]/[A-Z]/;
$buf = $initial . '-' . `date +%D`;
chop($buf);
# draw the lines
printf("%!\n");
printf("/inches { 72 mul } def\n");
printf("1 inches 10 inches moveto\n");
printf("4.7 inches 10 inches lineto\n");
printf("4.7 inches 6.35 inches lineto\n");
printf("1 inches 6.35 inches lineto\n");
printf("1 inches 10 inches lineto\n");
printf("1 inches 9.45 inches moveto\n");
printf("4.7 inches 9.45 inches lineto\n");
printf("1 inches 8.85 inches moveto\n");
printf("4.7 inches 8.85 inches lineto\n");
# write tape label
printf("/Helvetica findfont 35 scalefont setfont\n");
printf("1.2 inches 9 inches moveto\n");
printf("(%s\)show\n", $buf);
printf("/Helvetica findfont 10 scalefont setfont\n");
printf("1.1 inches 8.6 inches moveto\n");
printf("(Tape File #\)show\n");
printf("2.25 inches 8.6 inches moveto\n");
printf("(File System\)show\n");
printf("3.25 inches 8.6 inches moveto\n");
printf("(Mount Point\)show\n");
(open(fp, "/usr/local/etc/dumplist")) ||
die("unable to open dumplist file.\n");
$i = 1;
$y = 8.3;
while(<fp>)
{
chop;
($fs, $mp) = split;
printf("1.3 inches %5.2f inches moveto\n", $y);
printf("(%d\)show\n", $i++);
printf("2.25 inches %5.2f inches moveto\n", $y);
printf("(%s\)show\n", $fs);
printf("3.25 inches %5.2f inches moveto\n", $y);
printf("(%s\)show\n", $mp);
$y = $y - .2;
}
close(fp);
printf("stroke\n");
printf("showpage\n");
}
------------------------------
Date: 21 Jul 88 15:47 -0700
From: honman <honman%cmpt.sfu.cdn@ean.ubc.ca>
Subject: SLIP & TrailBlazer
Have anybody got SLIP to work over telephone lines with TrailBlazerPlus
modems? Would you please tell me what I did wrong?
I was using SUN-OS3.4 on 2 SUN3/50, and the SLIP code retrieved from the
sun-spots archive. When I connected the SUN using a null modem cable (pin
2-3 crossed, 7 straight thru), everything worked. I added the dialing
code and put TrailBlazerPlus modems on both end. I used the factory
setting for the modems and cables with pin 1,2,3 and 7 straight thru, and
pin 4-5, 6-20 tied on the modem's end. After the connection had been
established and I started the 'slattach-dstaddr-ifconfig' sequence,
'netstat -i' indicated that everything was okay. However, the two
SUN-3/50 couldn't communicate. When I did a 'ping' on SUN #1, I could see
the 'sd' LED on the front of the modem blinked, but nothing seemed to be
coming back. On SUN #2, 'netstat -i' indicated that it had received some
packets, but no packet was sent out.
It was not a baud rate problem because the dialing code waited until it
saw "CONNECT" from the modem, then forked and 'exec' slattach.
Hon-Man Wong
honman@cmpt.sfu.cdn
..\!ubc-cs\!fornax\!honman
[[ I later received this additional note: --wnl ]]
I have found the solution to my problem. The factory setting for the
TrailBlazerPlus enables software handshake. After turning the software
handshake off, everything works!
Hon-Man Wong
------------------------------
Date: Thu, 21 Jul 88 14:28:57 CDT
From: sdj@flora.wustl.edu (Scott Johnson)
Subject: SunView problems
I have been using SunView for a few months now, and there are two problems
that are bothering me, but never enough to worry about them. Both seem to
occur randomly (sometimes they happen, sometimes they don't). Now I am
getting tired of ignoring the problems, since they are not pretty. If you
could help me with them, I'd appreciate it.
I'm having some problems interfacing the SunCGI and SunView graphics
systems on a Sun 3/50. I am running a simulation in a SunView canvas,
writing into the canvas with both pw_ routines and cgipw_ routines. I use
the procedures open_pw_cgi() and open_cgi_pw(pw,&vpw,&cginame) to
initialize the system, run the simultaion, then quit and
close_cgi_pw(&vpw) and close_pw_cgi(). Upon closure, the program usually
either hangs up with no response or throws out a huge number of "WIN ioctl
number xxxxxxxx: Bad file number" error messages. The occurence of the
error message seem to be random; sometimes there is no error, but usually
it occurs.
Another problem I'm having is with the SunView event system. In the
simulation there are a few actions (panel button initiated) which take a
large amount of time to complete. While they are running, it seems that
the event queue is suspended in that inputs are stored in the queue but
not dequeued. I.e., during the running of these hefty procedures, mouse
movements are stored in this queue. Thus, if I wish to register the mouse
buttons after the procedures are completed, I can't because the queue is
full of mouse movement events. If 100 move events occur during the
routine, I won't find a mouse down event until those 100 move events are
flushed from the queue with another 100 move events. Is there a simple
way to shut off the event system while these long routines are running? I
tried WIN_IGNORE_PICK_EVENTS without any amount of consistency.
If you could help me with either of these problems, I would be grateful.
Thanks,
S Johnson (Washington U St. Louis)
------------------------------
Date: Thu, 21 Jul 88 15:03:18 CDT
From: Peter Wisnovsky <pswisnov@shrike.bcm.tmc.edu>
Subject: Graphing program
Hi: I am in need of a good public-domain data plotting program capable of
rendering nice output under both SunView and on a LaserWriter. I have been
looking into GnuPlot but it seems more suited to function plotting than
data plotting, but this may just be because I lack the right documentation
(I am missing the documentation shar files of version 1.1: if anyone can
tell me were to get the complete package I would also be very grateful).
If you can point me to such an animal I would deeply appreciate it.
Thanks,
Peter Wisnovsky
pswisnov@shrike.bcm.tmc.edu
pswisnov@phoenix.princeton.edu
------------------------------
Date: Wed, 20 Jul 88 09:16:12 PDT
From: otoole@sun.com (Mary O'Toole)
Subject: Sunlink/ibmftp problems
I need to know how SUNLINK's imbftp causes to execute on the IBM side. I
am pretty familiar with the sun network, but I don't know what is being
executed on the mainframe side to get the files to and from the hosts.
When ibmftp is forked in te3279, someinfo is echoed on the TSO screen for
the user (IND$FILE GET) or (IND$FILE PUT). These appear to be macros, but
again, I have no info so far on what is being executed to get the file
transfered.
Any information at all is muchly appreciated!
Pls print in sun-spots for all to see.
Thanks!
------------------------------
Date: Thu, 21 Jul 88 14:48:51 CDT
From: AARON KONSTAM <79343382@TRINITY.BITNET>
Subject: Spurious level 7 interupt?
Periodicaly I recieve the message:
spurious level 7 interupt.
The message seems to have no effect on the continued operation of our system.
What does this mean?
We have a 3/160 server with 2 3/50 diskless clients. We recently upgraded to
SunOS 3.5.
Aaron Konstam
Trinity University
79343382@trinity.bitnet
[[ There has been sporadic conversation about level X interrupts over the
past few weeks. You might want to check recent backissues. --wnl ]]
------------------------------
Date: 22 Jul 88 03:11:41 GMT
From: nick%marvax.Berkeley.EDU@ucbvax.berkeley.edu (nicholas socci)
Subject: How to get X10V4 for Sun?
I have recently got a Sun 3/50 and a Sun 4/260, and would like to install
X windows, in particular X10V4.
I would appreciate any pointers to where I may get the necessary files,
and any tips or advice in installation.
Please reply to the above account:
nick@marvax.berkeley.edu
Thanks to all.
------------------------------
Date: 21 Jul 88 22:28:57 GMT
From: iyw@cos.com (Izumi Y. Wilson)
Subject: Any Help in SUN VME and SCSI?
I want to design a board which plugs into SUN VME bus and I want to talk
to it via SCSI interface.
I need information on SUN VME electrical and mechanical set up. Also I
don't know what kind of flaver SUN SCSI interface is.
Any help?
P.S. For curious people, I need to talk to my board via SCSI for future
'out of cabinet' operation. But initially the board needs to sit
in SUN 3/160.
------------------------------
End of SUN-Spots Digest
***********************