Sun-Spots-Request@RICE.EDU (William LeFebvre) (01/08/88)
SUN-SPOTS DIGEST Thursday, 7 January 1988 Volume 6 : Issue 1
Today's Topics:
Administrivia and Non-trivia
Sun TOD Clock bug Patches for all releases
Sendmail 'Oi' fix breaks SMTP mail
Re: rows,cols (window sizing)
Re: disk versus memory
Diffs to make gnuplot 1.1.0 run under SunCGI
PD indexing software
Low Cost SCSI drives for Sun's
Sun 3 keyboards drop characters when you type too fast
Strange failure of bind(2) on diskless clients
Attempt to compile dumpregion.c
ping bug?
Removable Disk Storage on Suns?
Maximum disk space limitation?
How to do hybrid slider/text panel items?
SI83 8" SMD disks and SMD controllers?
Ethernet controllers for Suns?
----------------------------------------------------------------------
Date: Wed, 6 Jan 88 17:46:46 CST
From: William LeFebvre <phil@Rice.edu>
Subject: Administrivia and Non-trivia
I just got all caught up with Sun-Spots and then Christmas break hit.
Oh well......
Two of the files in the SunRPC distribution (contained in the archives)
suffer from the infamous dot-on-a-line-by-itself disease. I hope to take
care of the problem in the very near future (I may just ask the author to
repack those files with a better mail packer). The afflicted files are
"rpcsrc.doc.11" and "rpcsrc.doc.13". Sorry for the inconvenience.
Finally, what new year would be complete without a widespread and serious
bug to plague everyone. No, I'm not talking about the flu. I'm talking
about a disturbance in the temporal plane: the now infamous drifting
clock bug. As most of you probably know by now, there is a bug in SunOS
that, as of the beginning of the year, makes the TOD clock drift or
otherwise behave oddly. Chuq at Sun has been kind enough to send a patch
to Sun-Spots for this bug. I recommend this patch over all others---it's
straight from Sun. It is all contained in the next message. I felt it
important enough to pull it out of the queue of waiting message and run it
immediately.
William LeFebvre
Department of Computer Science
Rice University
<phil@Rice.edu>
------------------------------
Date: Thu, 7 Jan 88 10:29:15 PST
From: chuq@sun.com (Chuq Von Rospach)
Subject: Sun TOD Clock bug Patches for all releases
[These are the official patches from Sun for all known affected releases
in the field. If you have any questions or problems, please call Sun
Tech Support]
chuq
Sun Tech Support
There exists a problem for all Sun3 (68020) machines running SunOS
Releases 3.0-3.5, and all Sun4 (SPARC) machines running SunOS
Release Sys4-3.2 FCS and Sys4-3.2L GAMMA. This problem does not
exist for Sun-2's.
As of Jan 1 00:00 1988, the clock routine in the kernel will put the
clock chip into an uncertain state if you attempt to set the date.
The visible effects of this is to 1) cause the message
WARNING: TOD clock not initialized -- CHECK AND RESET THE DATE!
to appear while booting vmunix, and to 2) cause the system date to start
to drift widely. Any attempts to actually *set* the date will have only a
temporary effect (i.e., the date you set will be good for about 30 seconds).
In order to solve this problem, you must patch both the kernel and system
object files.
[[ NOTE that there are three separate patches. Make sure you use the
right one. --wnl ]]
==============================================================================
Sun3 System Patch
Releases 3.2, 3.3, 3.4, 3.5
This is for Diskful and Server Machines only. Diskless machines need to
be fixed on the server.
As root, run the follwing command:
echo 'resettodr+c0?i' | adb /vmunix - | grep reset
You should see the following printed out:
_resettodr+c0: bnes _resettodr+0xca
If you see instead:
_resettodr+c0: bnes _resettodr+0xce
the patch has already been applied to this system.
Proceed with the rest of the patch procedure anyway!
If you do not see either of these messages, go no further with this patch,
and please contact Sun Microsystems Customer Service.
If you do see either of those messages, then run, as root,
the following commands:
echo 'resettodr+c0?w 660c' | adb -w /vmunix
Reboot and then *set* the date.
If you build kernels for your system, or are a server for diskless clients,
do, as root
cp /sys/OBJ/clock.o /sys/OBJ/clock.o-
echo 'resettodr+c0?w 660c' | adb -w /sys/OBJ/clock.o
and then rebuild your kernel and/or the kernels for your diskless clients.
==============================================================================
Sun3 System Patch
Release SunOS Release 3.0
This is for Diskful and Server Machines only. Diskless machines need to
be fixed on the server.
As root, run the following command:
echo 'todset+0xb4?i' | adb /vmunix -
You should see the following printed out:
_todset+0xb4: bnes _todset+0xbe
If you see instead:
_todset+0xb4: bnes _todset+0xc2
the patch has already been applied to this system.
Proceed with the rest of the patch procedure anyway!
If you do not see either of these messages, go no further with this patch,
and please contact Sun Microsystems Customer Service.
If you do see either of those messages, then run, as root,
the following command:
echo 'todset+0xb4?w 0x660c' | adb -w /vmunix
Reboot and then *set* the date.
If you build kernels for your system, or are a server for diskless clients,
do, as root
cp /sys/OBJ/clock.o /sys/OBJ/clock.o-
echo 'todset+0xb4?w 0x660c' | adb -w /sys/OBJ/clock.o
and then rebuild your kernel and/or the kernels for your diskless clients.
==============================================================================
Sun4 System Patch
Release Sys4-3.2 FCS, Sys4-3.2L GAMMA
This is for Diskful and Server Machines only. Diskless machines need to
be fixed on the server.
echo 'resettodr+0x110?i' | adb /vmunix -
You should see the following printed out:
_resettodr+0x110: sub %i5, 0x1, %i5
If you see instead:
_resettodr+0x110: sub %i5, 0x0, %i5
the patch has already been applied to this system.
Proceed with the rest of the patch procedure anyway!
If you do not see either of these messages, go no further with this patch,
and please contact Sun Microsystems Customer Service.
If you do see either of those messages, then run, as root,
the following command:
echo 'resettodr+0x110?W ba276000' | adb -w -k /vmunix /dev/mem
Reboot and then *set* the date.
If you build kernels for your system, or are a server for diskless clients,
do, as root
cp /sys/sun4/OBJ/clock.o /sys/sun4/OBJ/clock.o-
echo 'resettodr+0x110?W ba276000' | adb -w /sys/sun4/OBJ/clock.o
and then rebuild your kernel and/or the kernels for your diskless clients.
------------------------------
Date: Sat, 12 Dec 87 18:53:42 PST
From: hoptoad!gnu@cgl.ucsf.edu (John Gilmore)
Subject: Sendmail 'Oi' fix breaks SMTP mail
Remember the Sun-Spots issue that arrived truncated due to a dot by itself
on a line in the middle of the text?
Remember my suggested fix of setting the "Oi" option in sendmail.cf?
Well, it works for sites without networks, but the SMTP code also uses the
"i" option's value -- incorrectly in my opinion -- to control whether a
period on a line by itself terminates a message arriving by SMTP. Since
there is no other way to terminate a message in SMTP, if you set the Oi
option, incoming connections will hang and fail to deliver mail.
If you have a network, even two workstations, I recommend backing out the
change. I have a two-workstation network, and only noticed when mail sent
from "polliwog" was still queued there after a few days.
Bill Nowicki at Sun knows about the problem, and I presume it will be
"fixed in 4.0" or some later release. He says he knows of no way to
configure a SunOS 3.3 sendmail to handle "." properly for both rmail and
SMTP mail, so we are mostly stuck until it's fixed.
John Gilmore
------------------------------
Date: Wed, 16 Dec 87 09:41:58 EST
From: ted@braggvax.arpa
Subject: Re: rows,cols (window sizing)
Here's sunsize, a little program I find useful for sizing my windows on
remote suns and 4.3 vaxes. With no arguments, it asks your window how
big it is and sets the internal size accordingly. With 2 arguments, it
makes your window that big in rows and cols and sets the internal size. I
find it useful in my .login file on the vax and remote suns. It runs on
Sun0S 2.X and 3.X and Bsd 4.3. Yes, it's a hack.
Ted Nolan
ted@braggvax.arpa
[[ Available in the archives as "sun-source/sunsize.shar". --wnl ]]
------------------------------
Date: 16 Dec 87 14:21 EST
From: CONKLIN@ge-crd.arpa
Subject: Re: disk versus memory
> Given a development environment --- compiles, loads, debugs ---
> which would you rather have - a 3/60 with lots of memory (say in
> 16 - 24 MB range) or a local disk?
Running a 3/60 with only 4 MB (can't get it with less) is like driving
with the emergency brake on - it's always slowing you down. Adding a
local disk would certainly help, but that would probably decrease the load
on your network and server more than it would noticeably speed up your
3/60. After all, the local disk would probably be slower than your
server's big, fast disk, and accessing a disk over the network is
(supposedly) only 20% slower than accessing a local disk (or is that just
for NFS?). What really kills you is paging, and a 4 MB Sun, with or
without a local disk, is going to do a ton of paging. Bumping the
physical memory up to 16 MB (or 24 MB for really heavy applications) will
give you the performance improvement you're looking for.
We have several servers here supporting ten or more diskless machines.
The servers supporting 16 MB clients running mostly LISP type applications
(including KEE and ART) run like a charm. The servers supporting the 4 MB
3/50's doing not much more than editing, compiles, etc., are under
constant strain.
Remember also, the following negative aspects of a local disk:
(1) It's more expensive per megabyte than an Eagle-type disk
(2) It's one more thing for your system manager to deal with
(setup, upgrade, ...)
(3) Assuming you put more than just root and swap on the local
disk, it's one more thing you have to backup.
Bottom line: get the memory!
Joel Conklin
General Electric Corporate Research & Development
Schenectady, NY
(518) 387-5817
email: conklin@ge-crd.ARPA, or steinmetz!crd!conklin
------------------------------
Date: Sun, 13 Dec 87 18:18:23 CDT
From: Stan Barber <sob@tmc.edu>
Subject: Diffs to make gnuplot 1.1.0 run under SunCGI
Here are the diffs to make gnuplot 1.1.0 work on a Sun using SunCGI.
Enjoy.
STan
[[ The diffs are available in the archives as "sun-source/gnuplot.diff".
It can be retrieved via anonymous FTP from the host "titan.rice.edu" or
via the archive server. For more information about the archive server,
send a mail message containing the word "help" to the address
"archive-server@rice.edu". --wnl ]]
------------------------------
Date: 15 December 1987 0918-PST (Tuesday)
From: wallen%ics@sdcsvax.ucsd.edu (Mark Wallen)
Subject: PD indexing software
Actually there is such a beast; it's the programs that the "refer"
bibliography software is based on: /usr/lib/refer/{mkey,inv,hunt}.
Unfortunately, there is NO documentation on how to use them other than the
original paper describing the work "Some Applications of Inverted Indexes
on the UNIX System" (June 21, 1978) by M. E. Lesk. It's possible to
index, for instance, all the man pages. I've included a couple of scripts
that call these programs that I've used a bit so you can get a flavor.
Warning: this stuff is OLD and most UNIX vendors only exercise the stuff
directly related to refer (see /usr/bin/indxbib); there are bugs in many
of the versions!
Mark Wallen
UC San Diego
wallen@ucsd.edu
[[ The shar file is available in the archives as
"sun-source/indexutils.shar". -wnl ]]
------------------------------
Date: Mon, 14 Dec 87 14:27:36 EST
From: berger@datacube.com (Bob Berger)
Subject: Low Cost SCSI drives for Sun's
I had mentioned in an earlier note that CDC WREN IV 300 Mbyte drives could
be gotten for around $2k. It turns out that that was a discounted price
from Arrow Electronics, a large distributer of electronic components. It
might be hard to get that price for single quantities, but a friend of
mine did talk them down to almost that much for one. It shouldn't be hard
to get that price for small quantities (about 5).
PS. Our WREN's are still working without a problem!
Bob Berger
Datacube Inc. Systems / Software Group 4 Dearborn Rd. Peabody, Ma 01960
VOICE: 617-535-6644; FAX: (617) 535-5643; TWX: (710) 347-0125
UUCP: berger@datacube.COM, rutgers!datacube!berger, ihnp4!datacube!berger
{cbosgd,cuae2,mit-eddie}!mirror!datacube!berger
------------------------------
Date: Mon, 14 Dec 87 10:35:11 EST
From: mike@thumper.bellcore.com (Michael Caplinger)
Subject: Sun 3 keyboards drop characters when you type too fast
Occasionally when I switch into furious typing mode on my Sun 3, the
keyboard starts beeping and my keystrokes are dropped. Why is this
allowed to happen? Are there any plans to fix it? I can't say it bodes
well when a machine can't keep up with something on the human timescale...
Mike
------------------------------
Date: Mon, 14 Dec 87 17:03:29 EST
From: jenkins@purdue.edu
Subject: Strange failure of bind(2) on diskless clients
We have two diskless clients that do not function properly when user
programs use the bind() system call. The symptoms are that bind() returns
an EADDRNOTAVAIL when called in programs like ftp, uwm (an X Windows
window manager), etc. Other network software works with no noticeable
problem (eg rlogin, rcp, etc.). The problem appears to be isolated to the
ethernet interface, and does not involve the localhost interface. This is
indicated by the fact that "uwm unix:0" works while "uwm `hostname`:0"
does not work. According to the sources, EADDRNOTAVAIL should only be
returned if there are no interfaces, the protocol requested is not
available on the interface, or the machine interface address is not found.
Netstat shows both interfaces up, and general network traffic flows as
expected.
This happens on both a Sun 3/50 (le0) and a Sun 3/160 (ie0). Both are
clients on the same Sun 3/180 server. There are other clients on this
particular server that display no problems. Only clients on this server
have the problem. All clients are booting the same public kernel. If we
boot the hardware exhibiting the error symptoms from a different
partition, it works properly. We have checked everything we can think of
as far as the ND partition setup. They all look the same.
Any pointers as to where to look for the cause of this problem would be
greatly appreciated.
Colin Jenkins jenkins@cs.purdue.edu ARPA
Computer Science Department jenkins%purdue.edu@relay.cs.net CSNET
Purdue University {ucbvax,decvax,ihnp4}!purdue!jenkins UUCP
West Lafayette, IN 47907 (317) 494-7832 PHONE
------------------------------
Date: Mon, 14 Dec 87 22:29:16 EST
From: Comer Duncan <duncan%andy.bgsu.edu@relay.cs.net>
Subject: Attempt to compile dumpregion.c
I just received the dumpregion.c code mentioned in the most recent
sun.spots.digest. On attempting to compile the code, I got a failure with
the remark that an include file could not be found. I am writing to ask
whether you know if the files:
/usr/include/images/stretchNW.cursor
/usr/include/images/stretchSE.cursor
are included with the person's submission, whether they are supposedly
part and parcel of the Sun distribution, or none of the above. I do
not find the above files in my include directory. I am running
3.2 on a Sun3/110. Thanks for any clues. This looks to be a useful
piece of code. I hope it can be brought up without too much pain...
Comer Duncan
Department of Physics and Astronomy
Bowling Green State University
Bowling Green, OH 43403
duncan@bgsu.edu --- CSNET address on Vax785
ATT: (419) 372 8108
[[ The machine that I am currently on, which is running SunOS 3.2, has
those two files online. Perhaps the person who set up your machine did
not unload these files from the distribution tape. --wnl ]]
------------------------------
Date: 16 Dec 87 17:10:34 GMT
From: mkhaw@teknowledge-vaxc.arpa (Mike Khaw)
Subject: ping bug?
I have a shell script that essentially does:
for host in a b c d e f
do
ping $host > /tmp/$host$$ &
done && wait
cat /tmp/*$$
It turns out that if, for example, "a" and "c" are actually up, but "b"
isn't, I get "x is alive" for all three; i.e., if a ping to a down host is
sandwiched between pings to up hosts, I get a false up indication for the
down host.
Is this something inherent in the ping protocol (or whatever), or is it an
implementation bug?
Thanks,
Mike Khaw
internet: mkhaw@teknowledge-vaxc.arpa
usenet: {uunet|sun|ucbvax|decwrl|uw-beaver}!mkhaw%teknowledge-vaxc.arpa
USnail: Teknowledge Inc, 1850 Embarcadero Rd, POB 10119, Palo Alto, CA 94303
------------------------------
Date: 14 Dec 87 16:52:39 GMT
From: km@emory.UUCP (Ken Mandelberg)
Subject: Removable Disk Storage on Suns?
Sun doesn't sell removable media disk drives, but I'm sure there must be
third party vendors that do. Can anyone tell me what is available? I
susppose the old DEC RMO5 (CDC whatever) will run off the standard SMD
controller, but those are really uneconomic to maintain. Surely, there is
something better now a days.
An optimal solution for us would be a SMD drive that could be run either
off a Vax 780 (using an SI controller) or off the Sun's Xylogics.
Ken Mandelberg | {decvax,sun!sunatl,gatech}!emory!km UUCP
Emory University | km@emory BITNET
Dept of Math and CS | km@emory.ARPA ARPA,CSNET
Atlanta, GA 30322 | Phone: (404) 727-7963
------------------------------
Date: Tue, 15 Dec 87 15:48:47 PST
From: Mark Lottor <MKL@sri-nic.arpa>
Subject: Maximum disk space limitation?
Sun specifications list the 3/260 as only being able to handle 1 gigabyte
of disk space, and a 3/280 at 2 gigabytes.
What's the limitation here? Hardware/software/disk controllers?
Does anyone run with more than 2 gigabytes on a 280?
------------------------------
Date: 15 Dec 87 17:29:50 GMT
From: roy%phri@uunet.uu.net (Roy Smith)
Subject: How to do hybrid slider/text panel items?
I'm trying to make a hybrid slider/text panel item. What I want is
something which you can grab and pull like a slider, but I also want to be
able to click on the slider-supplied "[current-value]" and type in a new
value from the keyboard.
The best I have been able to come up with is a slider and a separate text
item next to it which update each other:
[a_text and a_slider are global Panel_item's]
a_text = panel_create_item (control_panel, PANEL_TEXT,
PANEL_NOTIFY_PROC, a_text_proc,
/* lots of other argument pairs not shown */
0);
a_slider = panel_create_item (control_panel, PANEL_SLIDER,
PANEL_SHOW_VALUE, FALSE,
PANEL_NOTIFY_LEVEL, PANEL_ALL,
PANEL_NOTIFY_PROC, a_slider_proc,
/* ditto */
0);
void a_slider_proc (item, value, event)
Panel_item item;
int value;
Event *event;
{
extern int a_value;
char buf[100];
a_value = value;
sprintf (buf, "[%d]", a_value);
panel_set_value (a_text, buf);
}
void a_text_proc (item, event)
Panel_item item;
Event *event;
{
extern int a_value;
a_value = atoi (panel_get_value (item));
panel_set_value (a_slider, a_value);
}
The result is workable, but ugly and slow. Is there a better way to do
what I want?
Roy Smith, {allegra,cmcl2,philabs}!phri!roy
System Administrator, Public Health Research Institute
455 First Avenue, New York, NY 10016
------------------------------
Date: 15 Dec 87 23:10:43 GMT
From: dave@rosesun.rosemount.com (Dave Marquardt)
Subject: SI83 8" SMD disks and SMD controllers?
We're considering purchasing an SI83 8" SMD disk in place of a Super
Eagle. The SI83 is actually a Fujitsu M2344K/KS, and the unformatted
capacity is 690.1 MB, just slightly more than a Super Eagle. Is anyone
out there using these drives? What is your impression of this drive? And
if you got it through SI, what's your impression of them as a service
organization.
Now, a second question. Does anyone have any information on when and what
Sun will be choosing as a SMD disk controller for the VME bus? Or did I
miss an announcement somewhere?
Dave
------------------------------
Date: Tue, 15 Dec 87 18:01:32 PST
From: John Lekashman <lekash@orville.nas.nasa.gov>
Subject: Ethernet controllers for Suns?
Hi.
I'm looking for additional ethernet controllers for a VME sun3/260. Price
is not a particular object, performance is. Does anyone make something
which runs faster than the standard 450A from Sun?
Also of interest is multiple such interfaces per board. Something with
four interfaces and Lance chipsets, for example, would be ideal.
john
------------------------------
End of SUN-Spots Digest
***********************