Sun-Spots-Request@RICE.EDU (William LeFebvre) (12/02/87)
SUN-SPOTS DIGEST Tuesday, 1 December 1987 Volume 5 : Issue 65
Today's Topics:
Re: Use of 4.2 stuff on Sun OS
Re: Help with permission problems on server
Re: Porting timed to Suns
Re: "dbx" problem
Re: BUG in libmath.a on Sun 3.4
Re: process limit under 3.3
Re: Help printing large file on a laserwriter (2)
SUID uudecode
texsun announcement
Network Software Bug
Alternative Hardware Services?
Unconnected thin Ether hangs ESDI under SunOS 3.3?
Obtaining or developing a talk-like tool?
Chemistry package?
Tektronix 4107 emulator?
----------------------------------------------------------------------
Date: Tue, 24 Nov 87 11:31:42 PST
From: sklower@ucbarpa.berkeley.edu (Keith Sklower)
Subject: Re: Use of 4.2 stuff on Sun OS
I am not a lawyer, and ultimately such questions are answered by att's
licensing department. However I can offer the following opinion based on
what is done around here at berkeley: You should be able to extend your
unix 32/v, systemIII, or system V license that you had to acquire in order
to receive 4.2 on your vax to cover your sun as well. This should not be
very expensive (something like $400).
Having done so, you can copy sources and compile at will. Without
acquiring the 4.2 source license for the sun, ATT prohibits anybody except
the holder of the binary reseller's license for your Sun, to supply
executables derived from ATT source code to your machine. (Which means if
you had a 4.2 program, you could send the source to Sun, ask them to
compile it and give it back to you.).
------------------------------
Date: Tue, 24 Nov 87 08:32:02 EST
From: Dan Trinkle <trinkle@purdue.edu>
Subject: Re: Help with permission problems on server
I noticed this exact behavior when the underlying mount point directories
for NFS mounts were not readable/executable.
Daniel Trinkle trinkle@cs.purdue.edu ARPA
Computer Science Department trinkle%purdue.edu@relay.cs.net CSNET
Purdue University {ucbvax,decvax,ihnp4}!purdue!trinkle UUCP
West Lafayette, IN 47907 (317) 494-7844 PHONE
------------------------------
Date: Mon, 23 Nov 87 12:00:39 EST
From: Steve D. Miller <steve@brillig.umd.edu>
Subject: Re: Porting timed to Suns
One problem that I think a lot of people run across is the difference
between raw ICMP sockets under 4.2BSD/SunOS and 4.3BSD. The 4.3BSD raw
ICMP socket interface returns both the IP and ICMP headers on a
read/recvfrom when a packet is received; the 4.2BSD version of the
interface returned only the ICMP header. If you haven't fiddled the code
in measure.c to take care of this difference, you should.
I haven't ported it myself, but I helped someone else here who ported a
variant of the code, and I remember mentioning this to someone else who
had the same problem.
-Steve
------------------------------
Date: Mon, 23 Nov 87 12:04:58 -0800
From: Don Kark <wiley!don@csvax.caltech.edu>
Subject: Re: "dbx" problem
We just had this same problem. It turns out to be a bug in the Sun loader
"ld". Our problem showed up when we were doing partial, relocatable loads
"ld -r". Apparently the debugging information doesn't get correlated
correctly in the output file under certain conditions...which is why
turning the "-g" flag on and off on certain modules either solves or
provokes the problem. The bug still existed in release 3.4 (which we are
running).
The good news is that Sun has a patch for the bug. The patch tape we
received contained a new version of "/bin/ld" for 68010's and 68020's.
The patch "name" on the tape was "ld -r and gettype". I'm afraid I don't
actually have a patch number, but the Sun hotline knew what the problem
was when we described it. Hope this helps.
-- Don
=============================================================================
Don Kark | {cit-vax,trwrb,uunet}!wiley!don
TRW Inc. | wiley!don@csvax.caltech.edu
Redondo Beach, CA | don@wiley.uucp
------------------------------
Date: Thu, 12 Nov 87 09:40:38 EST
From: riedl@purdue.edu
Subject: Re: BUG in libmath.a on Sun 3.4
I'm running Sun Unix Release 3.4 on a 3/50 with no floating point chip,
and the log function correctly sets errno for me. It also prints a
message on stderr, as wnl mentioned. A work-around is to define the
function "matherr"(3M) to return a non-zero integer:
-------------
/*
* Stupid Sun 3.2 tan routine calls matherr and *prints* a message if
* it returns a 0.
*/
int matherr()
{
return(1);
}
-------------
In the man page for matherr there are two interesting passages that may
explain the problem:
If the user's matherr function returns non-zero, no error
message will be printed, and errno will not be set.
and
In the Sun environment, the facilities provided by matherr
are only available when a program is built with the software
floating point library, as there would be a substantial per-
formance penalty imposed by providing these facilities with
the libraries that support various Sun floating point
hardware options.
Hope this helps,
John Riedl
{ucbvax,decvax,hplabs}!purdue!riedl -or- riedl@mordred.cs.purdue.edu
------------------------------
Date: Tue, 24 Nov 87 00:56:13 EST
From: dan%wind@bellcore.com (Daniel R. Strick)
Subject: Re: process limit under 3.3
The line "#define MAXUPRC (NPROC - 5)" has been in param.c since
essentially the beginning of Sun Unix and remains unchanged through modern
times (release 3.5 beta). The justification is presumably that sun
workstations are single user machines, the window manager creates lots of
processes, and people with special requirements can always hack param.c.
NPROC-5 is probably too large for MAXUPRC even for a single user system
since a typical sun on a network using nfs has about 20 system processes
running in the background. The size of the text table, 24+MAXUSERS, is
also probably inappropriate.
Someone once asked me to find out what was wrong with Sun workstations
that caused them to crash whenever he ran his little "benchmark" program.
I reluctantly agreed to examine his code:
for (i=0; i = 5; ++i){
pid = fork();
printf("pid %d \n",pid);
}
I hope you enjoy this as much as I did. (Hint: there are two errors here.)
------------------------------
Date: Tue, 24 Nov 87 11:03 EST
From: PAUL <BESL%gmr.com@relay.cs.net>
Subject: Re: Help printing large file on a laserwriter
Disregard the previous mail message. Turns out that the PostScript file
was incorrectly truncated (so I'm told), which caused the problem. All ok
now.
Jim Marselle
GM Research Labs
marselle@gmr.com
------------------------------
Date: Mon, 23 Nov 87 23:36:12 PST
From: Brent Chapman <capmkt!brent@lll-tis.arpa>
Subject: Re: Help printing large file on a laserwriter
>I'm trying to print a PostScript file which is about 130K on a Sun
>laserwriter. When I do an "lpq" it says the file is in the queue and the
>lights on the printer blink, but nothing gets printed....
In general, the LaserWriters seem to print jobs a page at a time, as they
receive them (in other words, they don't try to receive the whole job
before beginning printing, just the next page). Does the job you're
trying to print include some especially complicated pages (graphics, or
fancy text, or bitmaps, or something)? The memory limitations on a
LaserWriter generally limit what you can do on a single page, not what you
can do in a complete job.
Error messages from a printer can be found in the printer log file whose
name is specified in the "lf" field of the /etc/printcap entry for that
printer.
There are other, more serious problems with the TranScript software that
Sun ships to run the LaserWriters. The system seems to wedge at the
slightest irregularity (like one of the printers runs out of paper and
doesn't get refilled immediately, or someone tries to remove a job from
the queue, or any of a number of other seemingly "normal" things). I've
heard from various sources that the version that Sun ships is not Adobe's
latest version, and that Sun's current version has several rather serious
bugs, such as the internal structures of the spooler or one of the filters
(I'm not sure which) can get walked on when the system gets an error
message back from the printer. Can anyone (preferably from Sun or Adobe,
though I'll listen to anybody who knows what they're talking about)
confirm or refute this?
-Brent
--
Brent Chapman Capital Market Technology, Inc.
Senior Programmer/Analyst 1995 University Ave., Suite 390
{lll-tis,ucbvax!cogsci}!capmkt!brent Berkeley, CA 94704
capmkt!brent@{lll-tis.arpa,cogsci.berkeley.edu} Phone: 415/540-6400
------------------------------
Date: Mon, 23 Nov 87 10:42:23 EST
From: im4u!rutgers!rochester!srs!matt@ut-sally.UUCP
Subject: SUID uudecode
The problem with uudecode being SUID to uucp has nothing to do with
uudecode itself, it's because of the alias that exists on many systems
(Sun in particular) for "decode":
decode: "|/usr/bin/uudecode"
The problem is that uudecode gets called as user "daemon". Since
uudecode, when it is run, creats a file that is specified within the file
itself, this can be a serious problem if uudecode is NOT SUID to some less
priveleged user. Now, as far as I can tell, having the `decode' alias is
still pretty risky (think about it) as far as uucp is concerned. Probably
the best way is either:
1) Don't support the `decode' alias and make uudecode mode 755.
2) Make a copy of uudecode mode 4755 to some useless user created
just for the purpose of sending to the decode alias. To wit:
% cd /usr/bin
% cp uudecode decode
% chown dumb decode
% chmod 4755 decode
% ...change the alias in /usr/lib/aliases to use the...
...new /usr/bin/decode
% newaliases
% ...possibly cd to /etc/yp and `make aliases'
% ...make an entry in /etc/passwd for user `dumb'
% ...make `dumb's home directory mode 777
% ...do anything else I've forgotten...
% ...give up and do #1...
Anyway, I did #1 several months ago. Who actually uses the `decode' alias
anyway?
Matt Goheen
S.R. Systems
{allegra,rutgers,ames}!rochester!srs!matt
------------------------------
Date: Sun, 22 Nov 87 21:19:36 CST
From: grunwald%guitar.cs.uiuc.edu@a.cs.uiuc.edu (Dirk Grunwald)
Subject: texsun announcement
Previously, I had announced the availability of a DVI previewer for use
with the X-11 window system. I've used the same driver code to implement a
previewer for the SunView window system.
Previously, the X-11 previewer was called 'xdvi'. It has been renamed to
'texx' (pronounced the way you're not supposed to pronounced 'TeX'), and
the SunView previewer is called 'texsun' (i.e. someone from TeXaS).
The package also includes a re-distribution of the dvi -> impress program
written by Chris Torek at the Univ. of Maryland.
The entire package is available via anonymous FTP from a.cs.uiuc.edu
(192.5.69.1) in the file pub/iptex.tar.Z. I have also made arrangements to
have this code distributed via the USENET newgroup 'comp.sources.unix'
There is an installation guide, a users guide and the programs.
Currently, the man pages refer you to the users guide. I welcome
improvements & fixes, particularly to the man pages.
Dirk Grunwald
Univ. of Illinois
grunwald@m.cs.uiuc.edu
------------------------------
From: telesoft!souza@sdcsvax.ucsd.edu (Steve Souza @eldest)
Date: Tue, 24 Nov 87 10:21:33 PST
Subject: Network Software Bug
We've seen a strange symptom in a network software package that, until
SunOS version 3.4 (on a Sun 160), functioned normally.
The basic scenario is a server process that accepts network socket
connection requests from a client process, executes a specified command,
and sends the output back to the client process, not unlike the r-shell
software.
Under 3.4, huge pieces of /etc/group randomly appear in the output of the
client process, often in a repeating pattern of usernames.
Here are excerpts from the code [pardon the crudity of the i/o -- I'm
still working on it!]:
----------------------------------------------------------------Server:
write_output(s, cmd)
int s; /* `s' is a socket of type AF_INET */
char *cmd; /* `cmd' is the command line */
{
FILE *pp;
char buf[BUFSIZ];
pp = popen(cmd, "r");
...
while (fgets(buf, BUFSIZ, pp) != NULL)
if (write(s, buf, BUFSIZ) < 0) {
pclose(pp);
return(1<<8);
}
return(pclose(pp));
}
----------------------------------------------------------------Client:
read_output()
{
/* `sock' is the other end of the AF_INET socket */
extern int sock;
char obuf[BUFSIZ];
while (1) {
if (read(sock, obuf, BUFSIZ) > 0)
write(1, obuf, strlen(obuf));
else
break;
}
}
-----------------------------------------------------------------------
Anybody know of changes in 3.4 that might cause this? We don't run YP,
BTW. Also, the server process calls initgroups(). Could that be related?
Any suggestions would be m u c h appreciated...
Thanks!
Steve Souza ...sdcsvax!telesoft!souza
TELESOFT Inc., San Diego, CA (619)457-2700 x277
------------------------------
Date: Mon, 23 Nov 87 20:33:21 EST
From: berger@datacube.com (Bob Berger)
Subject: Alternative Hardware Services?
Has anyone had any experience with alternatives to Sun's Hardware Service
Contracts? We are getting attractive quotes from a company called Sigma
Services. Their claims of support sound as good as Sun's both on-site
and repair center support. Their prices are about 20% lower than Sun's.
We have also found a compay called Apex Computer Co. that offers
independent board repair for Sun computers. This is a service for
getting Sun Computer components (boards, disks, monitors) repair or
swapped at a cost lower than Sun's Spare parts repair costs.
We now have around 15 Sun's, mostly 3/50's and 3/60's, some 3/110's and a
3/280S and 3/160 as servers. We are mostly concerned with keeping the
servers up as we can rearrange the 3/50's and 3/60's while things get
repaired. The yearly cost on a service contract for the whole shebang
would pay for more servers and workstations!
We are thinking of putting the servers on contract and just repair the
others on a per event basis. What do people who have a large installed
base of Sun's do about service?
Here are the contacts for the two servies I mentioned:
Sigma Services.
Larry Springer
61 Inverness Drive East
Englewood, Co. 80112
800-525-7050
Apex Computer Company
Chuck Hughes
4500 150th Ave. NE
Redmond, Wa. 98052
206-867-1900
------------------------------
Date: Tue, 24 Nov 87 08:01:43 EST
From: Steve D. Miller <steve@brillig.umd.edu>
Subject: Unconnected thin Ether hangs ESDI under SunOS 3.3?
I had an interesting problem recently. My colleagues and I were bringing
up a Sun-3/50 with a Sun-supplied 141MB disk in an area that was going to
have thin Ethernet installed, but where the ether wasn't yet in place. We
said, "gee, that's no problem -- just bring the beast up standalone, set
the switches inside so that the thin Ethernet tap gets used, ifconfig the
interface (a Sun/Lance -- le -- interface, by the way) as normal, and let
the prof plug the machine in by himself when the ether arrives." Sounds
OK to you, right?
Guess again. The machine worked fine when configured for thick (but still
unconnected) Ether, but soon after switching to thin, it started to get
SCSI failures a la "sd0: SCSI FAILURE". I tried about ten times to get
the beast to fsck and/or to dump it, and I would usually get that message
(and the disk would go to lunch) before I could achieve anything useful.
Curiously enough, I'd occasionally have dumps fail because of SCSI
failures associated with the local tape drive, and not the disk.
With a sigh, I brought the machine back to where I could get to a server
and started fiddling with it some more. At one point, I booted
single-user without running /etc/rc.boot (and the ifconfig), and guess
what? Everything suddenly started working. A disk that I couldn't keep
online for more than five minutes stayed online for much more than an hour
while I fsck'ed and dumped more than 100MB of data to the
(now-functioning) local tape. The machine was still configured for thin
Ether, and was still unconnected from the net. If I ifconfig'ed the
interface and exercised the disk for a while, everything broke again in
the same old way; a reboot without the ifconfig, and everything ran fine
again. This was completely repeatable.
I ended up taking the machine back to the prof's office, commenting out
the ifconfig, and letting him use it like that until we connected it up
more than a week later. He saw no further problems.
Has anyone heard of a hardware or software bug that might cause such
shenanigans? (If this is documented somewhere, a polite RTFM and a
reference would be appreciated, too.) I can, of course, provide more
information if necessary.
-Steve Miller
Spoken: Steve Miller Domain: steve@mimsy.umd.edu UUCP: uunet!mimsy!steve
Phone: +1-301-454-1516 USPS: UMIACS, Univ. of Maryland, College Park, MD 20742
------------------------------
Date: Mon, 9 Nov 87 16:40:27 EST
From: decvax!cvbnet!caramel!bandreas@decwrl.dec.com (Barry Andreasen)
Subject: Obtaining or developing a talk-like tool?
I am interested in obtaining (developing if necessary) a UNIX talk like
tool. This tool would provide a similar function as the importantly would
provide added functionality that would be, in my opinion, very valuable in
the development environment. Among some of the enhancements I see are:
* a .rc file to customize setup
* speed dialing buttons
* conference call capabilities (more than 2 parties all interacting)
* a graphics (white board) area with basic drawing capabilities.
* gfx pick and stuff capabilities
* call screening or privacy function
* call hold
* an answering service
.
.
.
.
The list could go on and on, but basically I believe it should have the
minimal functionality to allow such things as design to occur among
individuals without requiring their presence in order to graphically
display their thoughts. Any clues to the existence of such a tool or
helps in getting better ideas for implementation would be greatly
appreciated. Thanks
Since I am not sure of my mail posting route here's my conventional
address: B. Andreasen
Computervision
Crosby Rd. 5-2
Bedford, Mass. 01730
------------------------------
Date: Mon, 23 Nov 87 10:21:41 EST
From: bernhold@qtp.ufl.edu
Subject: Chemistry package?
Does anyone know of a package similar to ChemDraw for the Macintosh which
allows you to draw structures of molecules nicely and output them to a
file or printer? We have Sunview and would prefer postscript output, but
we'll take what we can get. Thanks in advance...
Dave Bernholdt bernhold@orange.qtp.ufl.edu
Quantum Theory Project bernhold@ufpine (BITnet)
Univ. of Florida
------------------------------
Date: 24 Nov 1987 0935-EST
From: franceus@tycho.arpa (Paul Franceus)
Subject: Tektronix 4107 emulator?
Does anyone out there know of a Tek4107 emulator for the Sun? I would
prefer a PD implementation of it if possible.
Also, we have an Interleaf laser printer. Is there anyway to get it to
talk Postscript? ie Postscript to Impress filter or something of that
sort.
Thanks for your help,
Paul Franceus
National Computer Security Center
------------------------------
End of SUN-Spots Digest
***********************