Sun-Spots Digest, v7n14

William LeFebvre Sun-Spots-Request at Rice.edu
Wed Nov 16 15:08:30 AEST 1988


SUN-SPOTS DIGEST         Monday, 14 November 1988      Volume 7 : Issue 14

Today's Topics:
                  Can't trap BUS signal under SunOS 4.0
                kernel reconfig and max processes problems
          Problems with Fujitsu-M2382K on Xylogic 735 on a 3/280
                              ALM-II lockups
                           Ungraceful shutdowns
                diskless suns booted from non-sun server?
                       Events during other events?
                           Sunview - X server?
                          SCSI vs E-net thruput?
            Documentation for Sun VME-Multibus adapter cards?
                             Online bug list?
                         optomizing C compilers?
                        document writing systems?
                            tn3270 for SunOS?
                       TIFF-to-Sunraster revisited?
                    GatePost (Sun/Appletalk) questions
                       PC-NFS application questions

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:    Tue, 8 Nov 88 12:49:30 EST
From:    @aplvax.jhuapl.edu:trn at warper.jhuapl.edu
Subject: Can't trap BUS signal under SunOS 4.0

Has anyone out there been able to successfully trap a BUS signal on SunOS
4.0?

I am using a Sun 4/110, and am mapping to a card on the VMEbus.
Intermittently I get a bus error [signal BUS (bus error) when using
dbxtool].  I've tried to screen the bus error, but to no avail.  Even the
code

	signal( SIGBUS, TrapError );	/* TrapError halts process */
	i = ( *reg & msk );		/* "reg" is mapped variable */
	signal( SIGBUS, SIG_DFL );
or
	signal( SIGBUS, SIG_IGN );
	i = ( *reg & msk );		/* "reg" is mapped variable */
	signal( SIGBUS, SIG_DFL );

fails to divert the bus error from trashing my process.

Incidentally, my code runs on a Sun 3/110 under SunOS 3.2.  On that
machine the signal I have to trap is SIGSEGV.  (Yes, I've tried trapping
both SIGBUS and SIGSEGV on the Sun 4.  No luck.)

Any suggestions?

	Tony Nardo
	Johns Hopkins University/APL
	INET: trn%warper at aplvax.jhuapl.edu
	UUCP: {backbone!}umd5!aplcen!aplcomm!warper!trn

------------------------------

Date:    8 Nov 88 18:53:35 GMT
From:    eric at eddie.mit.edu (Eric Van Tassell)
Subject: kernel reconfig and max processes problems

What am I doing wrong?? I did the following to allow more processes on my
sun-3.

		cd /sys/conf
		sed 's/maxusers 6/maxusers 8/' <PSCNH >PSCNH.new
		mv PSCNH PSCNH.old
		mv PSCNH.new PSCNH
		/etc/config PSCNH
		cd ../PSCNH
		make
		mv vmunix /vmunix

my question is why does size have the same output for the new and the old
kernel? Shouldn't something have gotten bigger? The checksums however were
different. Also the value of param.o:_nproc was 0x14 in both cases. Please
e-mail me if you can alleviate my confusion. Thanks.

	eric at eddie.mit.edu

------------------------------

Date:    Mon, 7 Nov 88 15:15:07 EST
From:    cxf at cs.rit.edu
Subject: Problems with Fujitsu-M2382K on Xylogic 735 on a 3/280

We recently received a Fujitsu-M2382K on Xylogic 735 and I was trying to
set it flying on a 3/280 when I ran into problems with insufficient inodes
in the client root partition. I tried to use to "-i" option in "newfs" to
increase the inodes allocation but nothing happened no matter how I
changed the "bpi" value. Then I tried to drop the # of cyl per group down
to 10 (default 16, according to the manual) but the "mkfs" responded that
cyl/grp has to be multiple of 16. Finally, I decided to use 4K logical
block size (default is 8K, according to the manual) and then I was able to
lower the cyl/grp to 8 (more groups, more inodes). The "mkfs" didn't
complain, and neither did "fsck", but when I tried to mount the new fs and
"cd" into it, the system crashed (didn't remember the exact error
message). Whatever I tried, fs with 4K block size simply doesn't work.
When I remake the fs with 8K blocks, it works fine but I am back with the
inodes problem again.

I know of at least one other site that was faced with the same inodes
deficiency problem on a M2372K. I wonder if any kind soul out there can
help us with this problem. Questions that I have are:

(1) how come the "-i" option in "newfs" didn't do what the manual says
    it should do (I know, I know, manauls lie all the time :-< )?
(2) is there something peculiar with the geometry of the M2382K that
    restricts the choice of # of cyl/grp as mentioned above?
(3) does anyone has a possible work around for this problem?

Thanks for any help that you can give us.

    Charles Fung		UUCP: {allegra,decvax}!rochester!rit!cxf
    Systems Analyst			MILNET: cxf at cs.rit.edu
    Rochester Institute of Technology	CSNET: cxf at cs.rit.edu
    Undergraduate Computer Science Dept	BITNET: CXFICS at RITVAX

------------------------------

Date:    Mon, 7 Nov 88 17:27:19 MST
From:    sitongia at hao.ucar.edu (Leonard Sitongia)
Subject: ALM-II lockups

System:	Sun-4/280S
OS: 4.0
Category: software/Operating System

Problem: a user logs off a terminal and the terminal immediately locks up.
This terminal is directly connected to an ALM-II serial port.  Two
different terminals have been tried and show the same problem.  No other
ports (we have 3 ALM-II's on this system) show this problem.

Looking at the process status gives:

      F UID   PID  PPID CP PRI NI  SZ  RSS WCHAN    STAT TT  TIME COMMAND
 408401 135   250     1  0   3  0  64    0 lbolt    S    j8  0:01 <exiting>

Logout turned the csh into a zombie and it looks like it was inherited by
init (as it should be), but init didnt finish up.  init is (right now):

      F UID   PID  PPID CP PRI NI  SZ  RSS WCHAN    STAT TT  TIME COMMAND
      3   0     0     0  0 -25  0   0    0 runout   D    ?   0:00 swapper
   8001   0     1     0  0   5  0  72   88 child    I    ?   0:07 /sbin/init -

What do you think is causing this?

------------------------------

Date:    Mon,  7-Nov-88 17:03:10 PST
From:    portal!cup.portal.com!donp at sun.com
Subject: Ungraceful shutdowns

Our site has a Sun 3/160M with two hard disks which is connected to a
recently installed 1500 watt backup power supply system.  If a power
outage is of sufficient duration, there will usually (but not always) be
personnel available that can "gracefully" shut down the system before the
backup power system runs out of battery power.

If a sysop person is NOT available, which of the following procedures
would do the least potential "damage":

A.  Spin down the disks, then turn off the CPU.
B.  Turn off the CPU, then spin down the disks.
C.  Turn off the disks and CPU simultaneously.

A (possibly) pertinent fact:  We have a 16 channel multiplexer so processes
may be running other than those active on the console.

Thanks, in advance, for any feedback or opinions...

Don Parmentier         email:  donp at cup.portal.com.UUCP  or  don at msunix.UUCP

[[ If you have enough time, run the command "sync" (it can be run by any
user) before turning the system off.  It will at least try to write the
disk caches out.  Of course, if there is heavy activity at the time,
you're still out of luck.  Back in the days of VAXes and DEC maintenance,
I got tired of having to rendezvous with the DEC guy once a month just to
shut the machine down for him.  So I wrote a program to be used in place
of a login shell.  It could only be run from the console and only as a
login process (child of init).  It would basically start a shutdown with
an explanatory message.  That way the DEC guy could shut the machine down
himself without having to know the root password.  Perhaps something
similar could be adopted here.  Security is maintained because the program
will exit immediately if it is not being run on the console and the
console is behind a locked door next to the machine.  --wnl ]]

------------------------------

Date:    Mon, 7 Nov 88 15:19:15 EST
From:    marc at aplpy.jhuapl.edu (Marcus Gates)
Subject: diskless suns booted from non-sun server?

Can diskless suns (3/50, 3/60) be booted from something other than a sun
server.  If yes, what are the requirements for the server? (NFS?) In
particular, can a hp835S support diskless suns sitting on an ethernet?

------------------------------

Date:    Mon, 7 Nov 88 19:37:43 EST
From:    rbw at williams.edu
Subject: Events during other events?

Does anyone know of a way to detect events during other events?  My
problem is that I have a sunwindow type tool to draw functions on a
canvas, and I would like some way of interrupting a draw (started by
clicking a panel button) if, for example, it is not progressing cor-
rectly.  Any info would be greatly appreciated.

-Richard Ward
rbw at cs.williams.edu		Williams College, Williamstown, MA

------------------------------

Date:    Fri, 4 Nov 88 14:14:26 PST
From:    ssc-vax!shuksan!scott at beaver.cs.washington.edu
Subject: Sunview - X server?

We were informed that Sun has a product (maybe in beta test) that provided
a Sunview tookit for X. One could use this toolkit, which all our code is
written in, but be running in the X environment. There is still some
rewrite work for conforming to the new X control flow but it is supposed
to be rather easy.

Anyone have more information on this phantom product? Will the toolkit
source be available for rehosting on other platforms using X? Are we
barking up the wrong tree such that we should be rewriting in an X
toolkit?

Thanks in advance,

	Scott Moody @ Boeing Mountain Network

------------------------------

Date:    Mon, 7 Nov 88 12:11:03 CST
From:    apctrc!drd!mark at uunet.uu.net (Mark Lawrence)
Subject: SCSI vs E-net thruput?

I'm thinking about putting a Wren IV on my 3/50 (which happens to host the
modem which provides our gateway to the Greater E-world) and I'm wondering
what the relative observed real-world thruput figures are.

In other words, the 3/50 is currently being served off of a 3/260.  When
netnews comes in, the network and the 3/260 get rather sluggish.  I assume
that this is due to all the nd and nfs I/O that is taking place due to
storing data coming in from the modem, serving paging for the programs
that are running on the 3/50 to do comm, uncompressing and unbatching and
all the virtual disk I/O that those entail.

If I had a Wren IV local to the 3/50 via SCSI, would it the I/O thruput be
better?  Enough to warrent placing the 3/50's swap/page partition on it?
(Of course, saving some CPU cycles on the 3/260 by not having to serve up
all that I/O and page traffic would be nice too, but I ignore that for the
sake of this discussion).

	Mark Lawrence			tulsun!drd!mark at Sun.COM
	DRD Corporation			mlawrence at jarsun1.ZONE1.COM
	 (918)743-3013			drd!mark at apctrc.UU.NET
					okstate!romed!drd!mark at rutgers.EDU

------------------------------

Date:    Mon, 7 Nov 88 23:50:01 EST
From:    eap at bu-it.bu.edu (Eric A. Pearce)
Subject: Documentation for Sun VME-Multibus adapter cards?

I'm looking for a complete description of all the dip-switch settings on
these cards.  Has anybody ever obtained one from Sun or some other source?

 Eric Pearce
 Boston University Info-Tech
 111 Cummington Street
 Boston, MA 02215
 617-353-2780
 eap at bu-it.bu.edu

------------------------------

Date:    Tue, 08 Nov 88 10:34:59 EST
From:    ted at braggvax.arpa
Subject: Online bug list?

In v7n2  Preston Mullen <mullen at itd.nrl.navy.mil> writes:
>There is a known bug whereby portmap crashes.  The following description
>was taken from Sun's on-line bug list a few months ago:

Where is this buglist?  How does one access it?

	Thanks,

	Ted Nolan
	ted at braggvax.arpa

------------------------------

Date:    8 Nov 88 08:12:00 CST
From:    "OACIS::NAVOCEANO" <navoceano%oacis.decnet at norda.arpa>
Subject: optomizing C compilers?

Has anyone else been underwhelmed by the optomization in the Sun-3 C
compiler? Does anyone have experience with or information about C
compilers that do a realy good job of optomizing for the 68020? Or perhaps
there are other optomizing methods that I don't know about (short of
manual) that will improve register usage, looping, concurrency and
instruction alignment.

Please respond directly to my address. I will provide a synopsis of
responses to Spots.

	Donald R. Newcomb
	U.S. Naval Oceanographic Office
	Stennis Space Center, MS 39522
	(601) 688-5998
	navoceano%oacis.decnet at norda.arpa
	(SPAN) OACIS::NAVOCEANO

------------------------------

Date:    Tue, 8 Nov 88 11:39:56 EST
From:    panda!jpn at talcott.harvard.edu (John P. Nelson)
Subject: document writing systems?

I am looking for information on software engineering-oriented
documentation systems.

We are currently using "troff" for all of our documents, occasionally
using "pic" to imbed pictures and diagrams (we have an internally written
tool that generates "pic" output).  While this fills our needs for the
short term, I'm sure that there must be better solutions.  We have over 50
diskless SUN workstations, and we would like to take advantage of the CPU
power at our disposal.

Most of the WYSIWYG documentation programs seem heavily oriented towards
producing end-user documentation.  As engineers, we like to use the
information in our documents in various ways:  not just as a end-product
hard-copy.  Ideally, we would like to use some kind of CASE system, but
most of these seem like vaporware at the moment.  Barring that, we would
like to use a document system with an open interface: that lets us get
information in and out of the system easily, (perhaps storing the
documents in some kind of database?)  We would like to be able to imbed
pictures into our documents, but our needs are simple: we mostly generate
block diagrams, flow-charts, that sort of thing.

Does anyone even make such a thing for SUN systems?  Also, because of our
1 user per cpu configuration, we would prefer to find software that is not
licensed on a per-cpu basis (or is discounted for network configurations
or large numbers of CPUs).

Thanks in advance.  I'll summarize any responses I receive.

     john nelson

UUCP:	{decvax,mit-eddie}!genrad!teddy!jpn
smail:	jpn at teddy.genrad.com

------------------------------

Date:    Tue, 08 Nov 88 11:41:20 PDT
From:    "M. Delparastaran" <MEHRAN at suwatson.stanford.edu>
Subject: tn3270 for SunOS?

I don't know if I am posting in the right place or not. I am interested in
tn3270 on SunOs. I know of tn3270 that is part of BSD. How would it port
to SunOs? any mod required?

Mehran

------------------------------

Date:    Sat, 05 Nov 88 13:25:16 EST
From:    Scott Thomas <13501SGT at MSU.BITNET>
Subject: TIFF-to-Sunraster revisited?

I am trying to find out if there was ever a re-release of the
TIFF-to-SunRaster filter that was mentioned during the summer.

------------------------------

Date:    Tue, 8 Nov 88 11:44:46 EST
From:    albert at mssun7.msi.cornell.edu (Jay Albert)
Subject: GatePost (Sun/Appletalk) questions

We have a copy of GatePost lying around, which is a program to run on a
Macintosh that allows one to connect a Sun to an Appletalk network, e.g.
to share a LaserWriter. The instructions are vague on how to make the
connection between the Sun and the Mac. How should the RS232 port on the
Sun be configured, and what about pin assignments? Is anyone out there
using this program? Thanks, Jay Albert (albert at mssun7.msi.cornell.edu)

------------------------------

Date:    Sat, 5 Nov 88 22:40:43 EST
From:    edsews!mancer!dharvey at uunet.uu.net (Doug Harvey)
Subject: PC-NFS application questions

I have a situation where I would like to transfer files to a PC from a Sun
workstation.  I know that Sun has a product called PC-NFS that allows you
to do this.  However, it doesn't suit my needs exactly, and besides, it is
rather expensive.

I was thinking of trying to write an MS-DOS program that would do the same
thing, except over a serial connection.   I know that PC-NFS can be
installed to run over a serial line.  In addition, in my case the
application does not need to be a device driver or a TSR.  All NFS
activity could be handled within the application itself, eg. execute the
pgm and perform all file transfers from within the pgm.  A couple of
questions: 

	o Does anything need to be written on the Sun side?  
	  We have pc-nfs installed and have a daemon on the sun called
	  'pcnfsd'.  Is this (or something similar) required in order 
	  to communicate with a PC?

	o Is it possible to write a PC application that uses the NFS 
	  protocol to communicate with the Sun in the same manner?  

	o How involved would something like this be? 

Feel free to tell me if this idea just won't work.  I don't want to waste
a lot of time for nothing.  Any pointers to documentation and/or example
source code would be appreciated. as well as any words of wisdom.

Thanks in advance,

Doug Harvey

Douglas Harvey                   UUCP: {uunet.uu.net}!edsews!mancer!dharvey
Electronic Data Systems          ARPA: edsews!mancer!dharvey at uunet.uu.net
(313) 556-0791

------------------------------

End of SUN-Spots Digest
***********************



More information about the Comp.sys.sun mailing list