🚀 go-pugleaf

SUN-SPOTS DIGEST           Friday, 6 March 1987    	Volume 5 : Issue 4

Today's Topics:
			       Administrivia
			  Re: Is NeWS good news?
		    Re: mail on diskless workstations?
		     root pathname dependance in YP
	         Re: Improper cleanup when exiting Suntools?
		        Re: Sun 3.0 and tty ownership
		        solution to mutual NFS hang
			      Sun 3.2 pipe()
		  Updated NFS Change to merge filesystems
		        Re: nd and arp problems? (2)
			     Re: TU80 on Sun-3
		       Lisp Floating Point on a Sun
		serious error in 160/180 installation manual
			  Bugs in Sun-3 compilers
		  Sun OS 3.2 ypserv -i and nameservers?
			 Serial ports for the Sun3?
		         3.2 rlogin vs interrupts?
		      RCS for Sun-3: I need patches?
			    Prolog and Sun3.2?
		        touch screen info request?
			       MAXTOR XT8760?
			      Used sun brokers?
			Generic SMD disks on a SUN3?
		    Re: Disk speeds of Sun file servers?
		Problem with events using Canvas and Panel?
	     peaceful co-existence with a diskless Sun-3/160C?
		      Graphics packages a la Macpaint?
			    YP for /etc/services?
			     used Sun equipment?
		 gremlin and ditroff previewer for SUN 3/50?
			  Comments on the Sun 3/110?
			    RFS server for Suns?
			  SUN graphics libraries?
			Ethernet<-->SNA interconnect?
			       Sun keyboard?
			   System V layers support?
		       Clearpooint memory and Sun 3/110?
		      VT100 emulator for Sun 3 windows?
			    Tape drives on sun?

------------------------------------------------------------------------
Date: Fri, 03 Mar 87 14:01:02 cst 
From: Vicky Riffle <rif@rice.edu>
Subject: Administrivia

Apologies for the lateness of this issue.  I have been unable to
connect to many sites over the past 1 1/2 months due to some local net
problems and was trying to hold off sending Sun-Spots until these
were all resolved.  I have decided to go ahead with the distribution
regardless.  If you are having problems receiving Sun-Spots, please 
send e-mail to sun-spots-request@rice.edu.  If you do not receive a
response, it means that I am unable to reach you electronically.
In that case -  please call me at 713-527-8101 x3844.

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

Date: Thu, 15 Jan 87 16:59:27 est
From: Bob Sutterfield <bob@ohio-state.ARPA>
Subject: Re: Is NeWS good news?

In article <1987.01.06.12.45.03.615.00452@iapetus.rice> mike@bambi.bellcore.com (Mike Caplinger) writes:
>Date: Fri, 5 Dec 86 11:04:56 est
>From: mike@bambi.bellcore.com (Mike Caplinger)
>Subject: Is NeWS good news?
> ...
>I am wary of Sun establishing standards that their competitors might like
>to use -- note that NFS has appeared so far on superminis and Crayettes,
>not on competing workstations.

	In choosing workstations for evaluation, we have specified NFS
as a requirement,  unless  the manufacturer  can offer something  much
better.   We note that   Hewlett Packard, Data  General,  Apollo,  and
others already support NFS, with moves afoot  in several other realms.
And you can always buy VAX NFS from Mt  Xinu.  Many even plan (or have
announced or are  about to announce - I  can never remember what their
status is) true diskless nodes based upon NFS.

	These are planned to (or already do) compete with the Sun 3/50
and on up the line.    How well they  do  so  is,  of course, open  to
evaluation in your very own smoke-filled room, but there do seem to be
some `competing workstations' running NFS out there.

	Now, we can all still wonder whether they can do it again with
NeWS...

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

Date: Mon, 19 Jan 87 10:49:57 EST
From: butler@stsci.ARPA (Lee Butler)
Subject: Re: mail on diskless workstations?

We have a small collection of diskless Sun3/50's which are fileserved by a
3/180.  Like you, we mount all the accounts from the fileserver which also
shares its passwd via YP.  We also mount the file server's /usr/spool so that
all mail and news is centrally located (and not duplicated on other, diskful
workstations).

Lee A. Butler
Space Telescope Science Institute   3700 San Martin Dr.   Baltimore MD 21218
Arpanet: butler@stsci.arpa  |  butler@brl.arpa
 Usenet: seismo!stsci.arpa!butler | {noao,astrovax,cfa,charm,nrao1}!stsci!butler
  Phone: (301) 338-4531

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

Date: Mon, 19 Jan 87 11:31:41 -0500
From: lrj@lasspvax.tn.cornell.edu
Subject: root pathname dependance in YP

  On all my machines, i have the root path set so that "." is
not a part of it.  This apparently breaks /etc/yp/Makefile when
it calls "makedbm".  When i put an explicit path into the
makefile, it works.  This is something to watch out for, as
YP passwd files all of a sudden weren't working properly.

  Also, please check your /etc/exports file(s).  Sun cleverly
has the default from Setup such that all filesystems are
exported to EVERYONE.  I know of one machine on the Internet whose
filesystems i was able to mount from here.

  Lastly, although it was mentioned on the TCP-IP mailing list,
	i thought i was worth mentioning here: Watch out for the
	password-less entries in Sun's default /etc/passwd file.

				-- Lewis R. Jansen, LASSP Systems Grunt
					lrj@lasspvax.tn.cornell.edu

	PS: anyone have any interesting rasterfiles for use as a
	    Suntools background?

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

Date: 19 Jan 87 06:27:19 GMT
From: Mike Edmonds <uss%tektools.tek.com@RELAY.CS.NET>
Subject: Re: Improper cleanup when exiting Suntools?

> From: H. Eidnes <H_Eidnes%vax.runit.unit.uninett@NTA-VAX.ARPA>
> Subject: Improper cleanup when exiting Suntools?
> 
> Several times (always?) when I exit Suntools, it seems some status
> information is not properly written out. The symptoms are that
> when I run "w" after exiting, "w" shows a lot of processes logged
> on, but with no activity ("-") specified. The ttys are those that
> were used under Suntools. The sizes of the windows also seem not
> to be reset ...

There are two problems here: one is excess entries in utmp and the
other is badly sized terminals. The two problems are related only in
that suntools is involved in them both.

When suntools exits, or a window is "quit" (from the frame command)
then entries from utmp are correctly removed. However, it seems that
when the process inside a window (typically, for me, an rlogin) goes
away and the window subsequently collapses, then the entry for it in
utmp is left. It's as if suntools cannot tell that that window went
away, and so cannot clean up after it. (I don't have a source machine
so it may be just that simple.)

The second case occurs because of the way the ioctl-settable parameters
are controlled. Normally they are changed only by resizing instructions
from a mouse initiated action. (If you are on a serial port or coming
through the network, normally these do not get reset from 0,0 at all.)
When a window is closed, the pty it was using is resized to 0,0. Except ...
except when another process is still running on the pty. I don't know if
suntools won't or can't, but the terminal does not get resized to 0,0.
To duplicate, start a shelltool, put a sleep 30 in the background, and
kill the window. Now rlogin to that pty. Your session will behave very
oddly! Run this program to see what the problem is:

#include <sys/ioctl.h>
main()
{
    struct ttysize size;

    if (ioctl(0,TIOCGSIZE,&size) < 0) perror("No go");
    else printf("Size is %d, %d\n", size.ts_lines, size.ts_cols);
}

As long as I'm at it, here's the other brief program H. Eidnes spoke of:

#include <sys/ioctl.h>
main()
{
    struct ttysize size;

    size.ts_lines = 0;
    size.ts_cols = 0;
    if (ioctl(0,TIOCSSIZE,&size) < 0) perror("No go");
    else printf("Size is %d, %d\n", size.ts_lines, size.ts_cols);
}

Fixes for either? Can't tell without source. But at least you know the
cause now. It may in fact be fixed in 3.2; I just installed it two days
ago and haven't tested it for either of these conditions yet.

Rick

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

Date: 21 Jan 87 23:38:10 GMT
From: Matt Landau <mlandau@diamond.bbn.com>
Subject: Re: Sun 3.0 and tty ownership

In last month's exciting episode, I asked the following:
>
>Can anyone give me a good reason why the Sun 3.0 and 3.2 versions of
>/bin/login don't set the ownership of the tty or pty so the logged in user
>is the owner?  

The answer turns out to be that /dev must be owned by bin for the tty
ownership to be set correctly.  The 3.0 version of setup appears to make
root the owner of /dev on diskless clients, although on diskful machines
/dev is correctly owned by bin.  I've seen this on a number of clusters
so far, leading me to believe it's a bug in setup.  Just changing the
ownership solves the problem.
-- 

 Matt Landau				mlandau@diamond.bbn.com
 BBN Laboratories, Inc.		     ...seismo!diamond.bbn.com!mlandau

 "Yow!  Maybe we could paint GOLDIE HAWN a RICH PRUSSIAN BLUE..."

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

Date: Fri, 23 Jan 87 11:52:59 EST
From: (Chip Hitchcock): CJH@CCA.CCA.COM
Subject: solution to mutual NFS hang

A few months ago the question came up about how to prevent deadlock during boot
on two Suns each of which was NFS-mounting files from the other, e.g.

Hektor% df
Filesystem            kbytes    used   avail capacity  Mounted on
/dev/xy0a               7471    5451    1272    81%    /
/dev/xy0g              83983   50929   24655    67%    /u1
Seurat:/u2             40419   23316   13061    64%    /u2
Seurat:/usr            64871   54051    4332    93%    /usr

Seurat% df
Filesystem            kbytes    used   avail capacity  Mounted on
/dev/sd0a               8859    7379     594    93%    /
/dev/sd1g              64871   54051    4332    93%    /usr
/dev/sd0g              40419   23316   13061    64%    /u2
Hektor:/u1             83983   50929   24655    67%    /u1

I don't recall seeing a solution posted (and couldn't get one from Sun) so I
came up with a crufty solution involving replacing
	/etc/mount -vat nfs
in Seurat:/etc/rc.local with a call to a shell script. However, while digging
through the manual I found a reference to a cleaner working solution:

Seurat% cat /etc/fstab
/dev/sd0a / 4.2 rw 1 1
/dev/sd0g /u2 4.2 rw 1 2
/dev/sd1g /usr 4.2 rw 1 3
Hektor:/u1 /u1 nfs rw,bg,retry=10 0 0

The option 'bg', which is documented under fstab(5), lets boot put the mount
in the background if the first try fails (which it's apt to even if Hektor is
running, since Hektor hangs with the usual "NFS: Server not responding"
complaint while Seurat is rebooting).

	CHip (Chip Hitchcock)		ARPA: CJH@CCA.CCA.COM
uu: ...!{decvax!linus, seismo!harvard, cbosgd, caip!think}!cca!cjh

We are just the vessel for a lot of intestynal flora---
And if we own something they will own it too.			Dave Luckett

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

Date: Fri, 23 Jan 87 10:16:21 +0000
From: Julian Onions <jpo%computer-science.nottingham.ac.uk@Cs.Ucl.AC.UK>
Subject: Sun 3.2 pipe()

I was quite amazed today to discover that under sun 3.2 (and
maybe other releases of sun) that the pipe call has changed.

In 4.2/4.3 and V7 the pipe call gave two descriptors both open for
reading and writing. Under sun 3.2 the first descriptor is open for
reading & the second for writing.

On the average, this will break 50% of older programs, surely? Does
anyone know the reason for this change??

Julian.

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

Date: Fri, 5 Dec 86 22:12:03 EST
From: NFS Functionality Enhancement Committee <dpk@brl.arpa>
Subject: Updated NFS Change to merge filesystems

We are just beginning to use NFS around BRL and I have been amazed
at how little thought seems to have been put into using NFS in
a large collection of large hosts.  Many of our machine have 8 to 16
disk segments mounted, and almost as many physical disks, so there
is little that can be done to lower the number of mounted partitions.
We wish to make every file system available from every system (or a
close approximation of this).  If we were to use the normal SUN NFS
implementation, we would have mount tables with 100 to 200 mounted
filesystems.  This is a nightmare.  I like to sleep, so I have made
the following change to nfs/nfs_server.c and nfs/nfs_vnodeops.c,
both part of the NFS related kernel source.

The effect of this change is to make tree of mounted local file
systems appear as a single homogeneous file system to remote
system that mount the root of such a tree.

Mount points are invisibly followed as long as they go to a file
system of the same type (which in this case is local).  The restriction
on the same type of file system is necessary to prevent file system
loops.  When/If more local file system types are supported, the "if"
below would have to be made smarter.  The statfs operation is
somewhat meaningless with this change since it will only return
the stats for the file system you mounted and not any file systems
under it.

The change to nfs_vnodeops.c is to improve the information content
of the faked-up dev entry in a stat structure of a remote file.
The key problem is that the dev entry is still a short, making it
very hard to make useful dev entries for remote files.  My adhoc
scheme allows for up to 31 remote mounts (hosts) until things fall
apart.  st_dev should really be at least a long.  Ideally it would
be an object containing an fsid and a machineid.  Maybe on the next
version...

The end result of all this is that you can now make all the file
systems on a server system available by simply mounting the root file
system (actually directory, e.g. mount -t nfs -o bg,soft host:/ /n/host).

We have chosen to creat a directory /n and to make a directory in
it for each system we wish to make available.  We then mount the
root of each system as /n/hostA, /n/hostB, ...

It is quite possible some of you may be able to suggest some
improvments to this implementation, such as ways to make it conditional
or to better handle the statfs data.  For us, this change alone is
a big step forward in making NFS usable in a large cluster of
independent super-mini computers (Vaxen, Goulds, Alliants) as well
as workstations (Iris's, Suns).

Comments welcome.

	-Doug-

Encl.  Diff of /sys/nfs/nfs_server.c and nfs_vnodeops.c.   Line numbers are
	from the Gould version of the SUN 3.0 sources, your numbers may vary.

*** /tmp/,RCSt1000202	Mon Jan 26 23:30:03 1987
--- nfs_server.c	Mon Jan 26 23:03:44 1987
***************
*** 282,288 ****
--- 282,306 ----
  		return;
  	}
  
+ #ifdef BRL
  	/*
+ 	 * Handle ".." special case.
+ 	 *    If this vnode is the root of a mounted
+ 	 *    file system, then replace it with the
+ 	 *    vnode which was mounted on so we take the
+ 	 *    .. in the other file system.
+ 	 */
+ 	if (da->da_name[0]=='.' && da->da_name[1]=='.' && da->da_name[2]==0) {
+ 		while (dvp->v_flag & VROOT) {
+ 			vp = dvp->v_vfsp->vfs_vnodecovered;
+ 			VN_HOLD(vp);
+ 			VN_RELE(dvp);
+ 			dvp = vp;
+ 		}
+ 	}
+ #endif BRL
+ 
+ 	/*
  	 * do lookup.
  	 */
  	error = VOP_LOOKUP(dvp, da->da_name, &vp, u.u_cred);
***************
*** 289,294 ****
--- 307,345 ----
  	if (error) {
  		vp = (struct vnode *)0;
  	} else {
+ #ifdef BRL
+ 		register struct vfs *vfsp;
+ 		struct vnode *tvp;
+ 
+ 	        /*
+ 		 * The following allows the exporting of contiguous
+ 		 * collections of local file systems.  -DPK-
+ 		 *
+                  * If this vnode is mounted on, and the mounted VFS
+ 		 * is the same as the current one (local), then we
+ 		 * transparently indirect to the vnode which
+ 		 * is the root of the mounted file system.
+ 		 * Before we do this we must check that an unmount is not
+ 		 * in progress on this vnode. This maintains the fs status
+ 		 * quo while a possibly lengthy unmount is going on.
+ 		 */
+ mloop:
+ 		while ((vfsp = vp->v_vfsmountedhere) &&
+ 			vfsp->vfs_op == vp->v_vfsp->vfs_op) {
+ 			while (vfsp->vfs_flag & VFS_MLOCK) {
+ 				vfsp->vfs_flag |= VFS_MWAIT;
+ 				sleep((caddr_t)vfsp, PVFS);
+ 				goto mloop;
+ 			}
+ 			error = VFS_ROOT(vp->v_vfsmountedhere, &tvp);
+ 			VN_RELE(vp);
+ 			if (error) {
+ 				vp = (struct vnode *)0;
+ 				goto bad;
+ 			}
+ 			vp = tvp;
+ 		}
+ #endif BRL
  		error = VOP_GETATTR(vp, &va, u.u_cred);
  		if (!error) {
  			vattr_to_nattr(&va, &dr->dr_attr);
***************
*** 295,305 ****
  			error = makefh(&dr->dr_fhandle, vp);
  		}
  	}
  	dr->dr_status = puterrno(error);
! 	if (vp) {
  		VN_RELE(vp);
! 	}
! 	VN_RELE(dvp);
  #ifdef NFSDEBUG
  	dprint(nfsdebug, 5, "rfs_lookup: returning %d\n", error);
  #endif
--- 346,357 ----
  			error = makefh(&dr->dr_fhandle, vp);
  		}
  	}
+ bad:
  	dr->dr_status = puterrno(error);
! 	if (vp)
  		VN_RELE(vp);
! 	if (dvp)
! 		VN_RELE(dvp);
  #ifdef NFSDEBUG
  	dprint(nfsdebug, 5, "rfs_lookup: returning %d\n", error);
  #endif
*** /tmp/,RCSt1000210	Mon Jan 26 23:30:18 1987
--- nfs_vnodeops.c	Fri Jan  2 17:51:54 1987
***************
*** 579,585 ****
--- 578,590 ----
  		 */
  		rp = vtor(vp);
  		nattr_to_vattr(&rp->r_nfsattr, vap);
+ #ifdef BRL
+ 		/* a better better kludge ??? */
+ 		vap->va_fsid &= 0x7ff;
+ 		vap->va_fsid |= ((vtomi(vp)->mi_mntno+1)<<11);
+ #else
  		vap->va_fsid = 0xff00 | vtomi(vp)->mi_mntno;
+ #endif BRL
  		if (rp->r_size < vap->va_size) {
  			rp->r_size = vap->va_size;
  		} else if (vap->va_size < rp->r_size) {
***************
*** 600,606 ****
--- 605,617 ----
  			 * an dev from the mount number and an arbitrary major
  			 * number 255.
  			*/
+ #ifdef BRL
+ 			/* a better better kludge ??? */
+ 			vap->va_fsid &= 0x7ff;
+ 			vap->va_fsid |= ((vtomi(vp)->mi_mntno+1)<<11);
+ #else
  			vap->va_fsid = 0xff00 | vtomi(vp)->mi_mntno;
+ #endif BRL
  			if (rp->r_size < vap->va_size) {
  				rp->r_size = vap->va_size;
  			} else if (vap->va_size < rp->r_size) {

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

Date: Wed, 28 Jan 87 19:17:46 CST
From: wca@ngp.utexas.edu (William C. Anderson)
Subject: Re: nd and arp problems? (SUN-Spots Digest, v5n3) (1)

I had the "bogus hardware ethernet address" problem on my beloved 2/50 until
I inspected /etc/hosts one day.  The hostname of my 2/50 wasn't there.  I
suggest that you make sure that your hostname is in your ND root partition
host table.

William Anderson - CompCenter - University of Texas at Austin

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

Date: Wed, 28 Jan 87 19:17:46 CST
From: Craig Rolandelli <craig%craig.uci.edu@icsg.UCI.EDU>
Subject: nd and arp problems? (2)

 > I have a SUN-3/75 which boots via nd from a SUN-3/180.  While booting, the
 > ARP table entry for the diskless node changes from 8:0:20:1:3d:ee (good)
 > to da:0:0:0:0:0 (not good) on the diskless node and on the nd server.  For
 > some reason, this works fine, until the diskless node is rebooted.  Then
 > rarpd on the nd server does not recognize the diskless node until an
 > "arp -d disklessnode" is done on the server, so the diskless node repeatedly
 > gets "tftp timeout" messages.  I have a SUN-3/110C which experiences exactly
 > the same problem, only its ARP table entry gets changed to ca:0:0:0:0:0.
 > However, there are seven other SUN-3s booting from the same nd server which
 > do not experience this problem.
 > 
 > [stuff removed]

 We have this same problem.  We have two fileservers each with 10 3/50's, 
 but only one of the fileservers has the problem.  Not all the diskless
 nodes are effected, but the same ones are always effected.  One of the 
 diskless nodes number gets changed to de:0:0:0:0:0  and the other two get 
 changed to e:0:0:0:0:0.  If you find out what is happening and/or how to 
 fix it, please let me know ASAP.  Until a fix is found, I am going to 
 place the same line in our crontab file.  Do you get "spurious level 3 
 interupts" on your server by some chance?  I wonder if they have anything 
 to do with it.

 Craig Rolandelli
 UCI ICS Support Group
 Universiaty of California, Irvine 
 Irvine, California 92717
 (714) 856-4222

 Internet: craig@ics.uci.edu
 Bitnet:   craig@uci
 UUCP:     ...!ucbvax!ucivax!craig

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

Date: Fri, 30 Jan 87 01:18:38 EST
From: drilex!dricej@harvard.HARVARD.EDU
Subject: Re: TU80 on Sun-3

The TU80 is a CDC drive; I don't remember the actual model number, but
I was able to get some numbers off of mine which CDC could understand.
(Actually, I have a Zilog drive, but it is the same drive as a TU80.)

The Interphase controller which is described in the Catalyst book should
be able to drive it, I'm told.  I haven't actually done it yet.

Craig Jackson
UUCP: {harvard!axiom,linus!axiom,ll-xn}!drilex!dricej
BIX:  cjackson

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

Date: Thu, 29 Jan 87 16:19:19 PST
From: Malcolm Slaney <spar!malcolm@decwrl.DEC.COM>
Subject: Lisp Floating Point on a Sun

Lucid sent me a new Lisp Compiler that does floating point correctly.
Here are the times to execute 10 1024 point fft's.  Note that this
version of the Lucid compiler will optionally generate code for the
68881.  There is no support yet for the FPA.  The times below are
in seconds.

Machine->	160	260	Symbolics	160/FPA	260/FPA 
Language \/	------------------------------------------------
Native		 5.4	 4.6	 4.6		 3.1	 1.9
Franz(Current)	81.	47.			 NA	 NA
Lucid(New)	 6.2	 5.3			 NA	 NA
Lucid(Current)	74.0	43.1

The entries labeled Native mean C on the Sun's and Lisp on the
Symbolics.

There are two amazing pieces of news in these numbers.  First, 
Lucid Lisp on a Sun/3-260 with a 68881 runs within 15% of a 
Symbolics without an IFU.  I expect that when the IFU is factored 
in then the Symbolics will be about 50% faster.  All this changes 
when (if?) the Sun Lisps' support the FPA.  And then there are
Sun-4s....

The second amazing news is that Lucid is now within 15% of the raw
speed of the machine.  

I just talked to John Foderaro at Franz, Inc. and they are working
on similar enhancements.  If your Lisp of choice is Franz then you
shouldn't have to wait long.

This is a pre-pre-prerelease of the Lucid compiler.  Many thanks to
David Posner at Lucid and Eric Schoen at Schlumberger and Stanford
for helping with these benchmarks.  No, I don't know when the new
compiler will be released.

Cheers.

							Malcolm
P.S.  Like all benchmarks, your mileage may vary.  These times record
just CPU time on machines with enough memory so there was no paging.

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

Date: Thu, 29 Jan 87 17:04:01 est
From: <utzoo!henry@seismo.CSS.GOV>
Subject: serious error in 160/180 installation manual

Being as how we are rather paranoid folks, with little faith in the perfect
reliability of hardware manufacturers, when we get something new in the door
the first thing we do is read the hardware manuals and check all the switch
and jumper options to make sure they're set right.  This does pay off from
time to time.  When we got our 3/180S, we went through the manuals and the
boards checking things.  Everything was right except on the CPU board, where
two things were wrong.  One appears to have been a genuine error:  the FPP
was a 16MHz chip but was jumpered for the 12MHz clock.  (That's the sort of
thing that would be very likely to slip by Sun testing unless the tests were
specifically looking for it, and the local Sun hardware people say they've
seen that one before.)  The other one looked equally straightforward...

In the Hardware Installation Manual for the Sun-3/160 SunStation (apparently
there is no separate manual for the 180, or wasn't at the time we got ours),
part 800-1314-01, rev 50 of 15 Oct 1985, on page 49, Table 4-4, "CPU Board
Jumpers", contains a serious error.  About a third of the way down, we have
J3102 and J3101 labelled respectively "in for 4MB board" and "in for 2MB
board".  If you look at a normal 4MB 160 or 180, you will find the jumper
that is probably J3102 (the manual is unclear -- to put it mildly -- about
which jumpers are which) out and probably-J3101 in.  By the manual,
this would appear to give you a 2MB machine.  The manual is lying.  Leave
the jumpers as they are.  If you try making the obvious change to set it
up for 4MB, the power-up self-test will flunk the memory diagnostic.  We
got suspicious about this when a replacement CPU board from Sun had the
same "error" in the jumper settings; sure enough, with the jumpers set back
the way they were, our "defective" board ran fine.

Sun has been informed; our local Sun people say it's the first they've heard
of it.  I guess we're just unusually paranoid...

				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,decvax,pyramid}!utzoo!henry

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

Date: 4 Feb 1987 20:02:52 GMT
From: mcvax!bond!uk.ac.bath.maths!jhd@seismo.CSS.GOV
Subject: Bugs in Sun-3 compilers

The following piece of roff describes a bug I submitted five weeks ago,
but which has yet to appear in the bulletin board.  Since it describes
bugs in three different compilers, and affects TeX, I feel many people
would like to be aware of it.
     James Davenport

.ce
\fBBug in Sun's compilers\fP
.pp
We (Professor R. Sibson and myself) have discovered an amusing bug in
the Sun C and Fortran 77 compilers [and, later, in the Pascal
compilers], which appears to be present in both release 3.1 and 3.2,
and to exist whether or not the optimiser is used.  We recall the Sun
convention that registers D0 and D1 are scratch registers, whereas
D2-D7 are "permanent", and any routine using them has to save and
restore them appropriately. The bug is that remainder calculations
(MOD in Fortran, % in C) modulo an explicit power of two can generate
code which uses D2, but D2 is not saved or restored. The work-around
is to use a variable as the second argument.  We reported this bug (in
general terms) to Sun before Christmas, and will contact them tomorrow
with more details.
.br
J.H. Davenport - New Year's Day MCMLXXXVII
.pp
Sun's (very belated) reaction is that the bug is to be fixed in 3.4
(note that we are running 3.1 on one of our machines, since 3.2 is not
available as an upgrade yet), and that it can be bypassed by using the
-mc68010 flag on the compilers. Of course, this generates worse code,
and, in particular, means that the FPA is not available.
.pp
Here is the Fortran version:
.nf
      INTEGER FUNCTION PLTPEN(PEN)
      INTEGER PEN
      PLTPEN = (MOD(PEN,5)) + 8 * MOD((PEN / 5), 4)
      RETURN
      END
.fi
while a C program doing much the same (and generating the same bug) is
.nf
pltpen_(pen)
int *pen;
{
     int sty = ((*pen) % 5) + 8 * (((*pen) / 5) % 4);
     return sty;
}
.fi
The following main() demonstrates the nature of the C bug quite clearly
.nf
main()
{ register int a,b,c,d,e,f; /* This should mean that f is D2 */
  int z;
  a=b=c=d=e=f=0;
  z=6;
  pltpen_(&z);
  printf("%d\n",f); /* prints 3, which is somewhat odd */ }
.fi
The assembler produced from the Fortran is (my annotations start ***)
.nf
	.data
	.data1
	.bss
	.data
	.align 4
	.text
|#PROC# 07
	.text
	.globl	_pltpen_

_pltpen_:
|#PROLOGUE# 0
	link	a6,#0
	addl	#-LF1,sp
	moveml	#LS1,sp@	*** An attempt to save registers
				*** but LS1 is in fact 0, so no
				*** registers are saved
|#PROLOGUE# 1
L15:
	jra	L12		*** eccentric
L12:
	movl	a6@(0x8),a0
	movl	a0@,d0
	divsll	#0x5,d1:d0	*** the MOD(PEN,5) calculation
	movl	a6@(0x8),a0
	movl	a0@,d0
	divsl	#5,d0		*** PEN/5
				*** Now we are going to do the MOD(...,4)
				*** which we will do by ANDing with 3
				*** except that this won't work for negative
				*** numbers
	moveq	#3,d2		*** Load 3 into D2, which we have not saved
	tstl	d0
	jge	L2000000
	negl	d0		*** the negative case
	andl	d2,d0
	negl	d0
	jra	L2000001
L2000000:	andl	d2,d0	*** the positive case
L2000001:	asll	#0x3,d0
	addl	d0,d1
	movl	d1,a6@(-0x8)
	jra	L11
L11:
	movl	a6@(-0x8),d0
	jra LE1
LE1:
	unlk	a6		*** no register restore
	rts
	LF1 = 32
	LS1 = 0x0
	LFF1 = 32
	LSS1 = 0x0
	LP1 =	0x8
	.data1
.fi
.pp
It would appear that the presence of the bug in the Pascal compiler "pc"
explains why TeX on Suns fails Knuth's "trip" test - various parts of TeX
are incorrectly compiled.
In fact, re-compiling TeX with -mc68010 means that it does pass the
"trip" test, so that the case can be considered proven.

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

Date: Fri, 16 Jan 87 09:55:38 EST
From: Mark Mendell <mendell%utopus.toronto.edu@RELAY.CS.NET>
Subject: Sun OS 3.2 ypserv -i and nameservers?

   We are running Sun OS 3.2 (68020), and the Berkeley (not the Sun 
distributed) named (BIND), with the usual number of bug fixes.  The problem 
is that "ypserv -i" is supposed to query the name server, but just forks a
couple of times and waits.  We don't have the source to 3.2 yet (although we
do have to 3.0), and the executable is stripped, so I can't figure out how to
turn on debugging to figure out what is wrong.  In addition the resolver(3)
manual page seems to be missing.

    Does anyone know what the problem is, or how to track it down?  If there
is a local named, is /etc/resolv.conf necessary?  Any other clues?

Mark Mendell
	    Computer Systems Research Institute    University of Toronto
	    Usenet:	{linus, ihnp4, allegra, decvax, floyd}!utcsri!mendell
	    CSNET:	mendell@Toronto
	    ARPA:	mendell%Toronto@CSNet-Relay

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

Date: Mon, 19 Jan 87 10:08:55 EST
From: butler@stsci.ARPA (Lee Butler)
Subject: Serial ports for the Sun3?

Does Anyone have any experience with any non-Sun serial boards for the
Sun 3 systems?  I have heard of folks who are contemplating the DY210
and the Microprojects board, but nothing form anyone who has tried
anything.  Any comments, good/bad/indifferent on any sources would be
welcome.

Lee A. Butler
Space Telescope Science Institute   3700 San Martin Dr.   Baltimore MD 21218
Arpanet: butler@stsci.arpa  |  butler@brl.arpa
 Usenet: seismo!stsci.arpa!butler | {noao,astrovax,cfa,charm,nrao1}!stsci!butler
  Phone: (301) 338-4531

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

Date: 19 Jan 87 19:33:51 EST (Mon)
From: Jean-Francois Lamy <utai!lamy@seismo.CSS.GOV>
Subject: 3.2 rlogin vs interrupts?

When remote logged in to a Sun 3 running 3.2, interrupt characters
take a very long time to be processed.  Same behaviour when logged
from a 4.2 BSD Vax, an OS 3.0 Sun 2 or an OS 3.2 Sun 3.

This does not happen when the remote machine is running 3.0.

Jean-Francois Lamy 		       CSNet:  lamy@ai.toronto.edu
AI Group, Dept of Computer Science     ARPA:   lamy%ai.toronto.edu@relay.cs.net
University of Toronto		       EAN:    lamy@ai.toronto.cdn

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

Date: Tue, 20 Jan 87 16:02:15 -0100
From: ames!seismo!mcvax!inria!shapiro@cad.Berkeley.EDU (Marc Shapiro)
Subject: RCS for Sun-3: I need patches?

I know this has been answered before, but my site does not keep
news archives.

Can someone please send me a complete list of pathces to make RCS
work on a Sun-3 (i.e. diffs from the Vax 4.2 or 4.3 version).
Thanks.

        Marc Shapiro
        INRIA Bat 11
        BP 105
        78153 Le Chesnay Cedex
                France
        mcvax!inria!shapiro

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

From: Thomas Rodden <mcvax!comp.lancs.ac.uk!tam@seismo.CSS.GOV>
Date: 20 Jan 87 16:28:59 GMT
Subject: Prolog and Sun3.2?

I am currently looking for a Prolog that will allow me to interface
easily with the sun window system  on Sun UNIX 4.2 Release 3.2. Has 
anyone either interfaced  a prolog to sun windows, or can recommend one
with a working interface.
				Many Thanks
				   Tam Rodden

					(email:tam@uk.ac.lancs.cs)

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

Date: Fri, 23 Jan 87 10:02:07 pst
From: klee@ads.ARPA (Ken Lee)
Subject: touch screen info request?

One of our user interface research projects is looking into the use of touch
screens.  Has anyone had any experience with touch screens with Suns?  What
manufacturers make screens this size?  How well do they integrate with Sun
hardware?  Are they easy to integrate with the SunView VUID software?  Please
reply by mail, I'll post to the list if there is interest.  Thank in advance.

Ken Lee
Advanced Decision Systems, Mountain View, California
klee@ads.arpa

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

Date: Thu, 29 Jan 87 00:33:30 EST
From: Ken Mandelberg <km@EMORY.ARPA>
Subject: MAXTOR XT8760?

Maxtor is advertising a new drive, the XT-8760E. Its a 5.25 inch 
winchester with 765 Meg (601 Meg formatted), an 18ms seek time,
and a quantity one list price under $8K (which one would expect
to drop a couple of $K depending on discounts). It has an EDSI
interface.

It seems like if you put it in a shoebox with Sun's EDSI controller
that this would make a Sun 3/50 into quite a powerful server. Am
I missing something?

Maxtor also has a SCSI version. Presumably you could add it to
the Sun without any inteface, if there was a SCSI driver that
used the right SCSI command set. Can anyone comment on whether
the Sun SCSI driver is being changed to support the Common Command
set, or is it stuck on whatever dialect the old Adaptec board
wanted?

Ken Mandelberg      |  {akgua,sb1,gatech}!emory!km   USENET
Emory University    |  km@emory                      CSNET,BITNET
Dept of Math and CS |  km.emory@csnet-relay          ARPANET 
Atlanta, Ga 30322   |  Phone: (404) 727-7963

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

Date: Wed, 28 Jan 87 17:21:54 EST
From: Root Boy Jim <rbj@icst-cmr.arpa>
Subject: Used sun brokers?

   Date: Tue, 26 Aug 86 19:37:46 EDT
   From: Rick Adams <rick@seismo.CSS.GOV>

   I have a Sun 2/120 that isn't much use anymore (all of our other
   Suns are Sun3s). Is there any company that would accept a
   Sun 2 as a trade in on a Sun 3. We can't sell it outright, since
   it's government owned equipment.

How about an inter agency transfer? NBS is part of DOC. Drool.

   ---rick

(Root Boy) Jim "Just Say Yes" Cottrell	<rbj@icst-cmr.arpa>

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

Date: Wed, 28 Jan 87 17:16:01 EST
From: Root Boy Jim <rbj@icst-cmr.arpa>
Subject: Generic SMD disks on a SUN3?

   Has anyone got any experience putting non-eagle SMD drives on a SUN
   using a Xylogics 450 controller.  I've got a CDC drive just burning a
   hole in my machine room floor that I would love to connect to the SUN.

Yes. We are running a SUN 150U upgraded to a SUN-2 with two CDC 9762
(RM03 clones) on an XY450. It's a bit tricky, see below.

   So far, my luck with the drive has been all bad.  It seems that the
   SMD interface used by the eagle and that used by the CDC drive is
   not the same (everyone should have at least one standard, namely their own).

There are two types of SMD configurations: radial and daisy chained.
The skinny cable is called the B (or data) cable and there is one per
drive.  The wider cable is called the A (command) cable. In the radial
configuration, there is also one per drive. When daisy chained, there
is an A cable to the first drive, another A cable between the first
and second drive, one between the second and third, etc, and the last
one is terminated.

The first secret is just to try one disk on the controller. When only
one drive is used, the radial and daisy chained cases are identical.
Try running diag on your lone disk.

The problem is that the index and sector pulses are duplicated on both
the A & B cables, and can be gated with select, or continuous. Because
the 450 looks for these signals on the A cable, they must be gated.
Continuous signals might be good for optimizing rotation in radial mode.

There is a card known as the `_TVV' where `_' is `F' for continuous
and `G' for gated. There are other possible combinations. You didn't
say what model you had, but for the 9762 the only difference is
resoldering a jumper on the B01 card. There is another fix for 9766's
(RM05 clones) and probably others.

   ANY information, good or bad, is welcome.

I hope this is in the former category.

One more warning, and the reason I am posting to the entire newsgroup:

		BEWARE OF SUN's CRETINOUS `SETUP' PROGRAM!!!

The little bugger INSISTS on writing LABELS on your disks!!! At least
SUN2.0 supported a disk that had the same geometry as the RM03, but
SUN3.x does not. This means installing everything by hand. Boo Hiss!

		   John <scott@mitre-gateway.arpa>

(Root Boy) Jim "Just Say Yes" Cottrell	<rbj@icst-cmr.arpa>

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

Date: Thu, 29 Jan 87 11:30:44 +0100
From: H. Eidnes <H_Eidnes%vax.runit.unit.uninett@NTA-VAX.ARPA>
Subject: Re: Disk speeds of Sun file servers?

First I must thank Don Speck for his excellent (as far as I can judge)
survey of his measurements.

I was a little bit surprised by the fact/conclusion that the DMA on the
machines was unable to keep up with the disks. I would very much like to
get some comments on this from someone a little more technically competent
than me; specifically:

1) Can some of the effect described come from the fact that the 450 and
451 controllers both are 16-bit Multibus devices, and does the adapter
card add anything to slow down the transfers?

2) I would also like to know if using a "real" 32-bits disk controller
(eg. the Interphase V/SMD 3200) would give any benefit? Measuerements
directly comparable to Don Speck's figures would of course be the best.


3) Is the DMA-to-virtual-memory feature a possible bottleneck that can
explain why the DMA does not keep up with the disk?

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

Date: Thu, 29 Jan 87 12:39:43 gmt
From: David England <DE%UK.AC.LANCS.COMP@ac.uk>
Subject: Problem with events using Canvas and Panel?

I'm writing an interactive tool for drawing window-based tool layouts
(consisting of a Canvas drawing surface with a controlling Panel)
and I'm having problems with some events sent to the Canvas window not having
any effect until I cross over the window boundary into the Panel in the
same tool. In the panel subwindow I set the type of object I want to draw
(menu box, button etc) and then locate its position on the canvas. When
the drawing operation is complete I reset the object type menu to object
select mode. I suspect that the problem is trying to use panel_set from
within the canvas_select function i.e. the program is still dealing with
canvas events whilst updating the panel item image. Has any one had any
similar problems and have they found a way round them ??


     Dave    uucp: ...!mcvax!ukc!dcl-cs!de
      arpa: de%lancs.comp@ucl-cs

"... you know of course this means war !"

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

Date: 29 Jan 1987 1215-PST (Thursday)
From: Keith Lantz <lantz@gregorio.stanford.edu>
Subject: peaceful co-existence with a diskless Sun-3/160C?

Much to my surprise, diskless 3/160C's generate a LOT of noise.  (I
had expected the heat.)  Does anyone have any great ideas for reducing
that problem without moving the shoebox into another room?

Thanks in advance, Keith Lantz

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

Date: Fri, 30 Jan 87 14:42:38 -0500
From: boncelet@huey.udel.edu
Subject: Graphics packages a la Macpaint?

We have just gotten some Suns and are just learning...

Is there a graphics package similar to MacPaint so that we can
do pictures, transparencies, etc?  We have a pretty standard
configuration of Sun 3's and an Apple LaserWriter.  The ideal
solution would be a public domain package.  Otherwise, can
anyone recommend (or steer us away from) specific commercial
products?  Price is a consideration.

Thanks. 
		Charlie <boncelet@huey.udel.edu>

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

Date:  3 Feb 87 03:28 EST
From: montnaro%sprite.tcpip@ge-crd.arpa
Subject: YP for /etc/services?

We have four Sun-3/260s, one server and three clients, running Unix 3.2 on
our local ethernet (along with a scad of other things, ranging from PCs, IP
routers to other networks, other Suns, [micro and big]VAXes, and a 3081).
The server provides yp service for the three clients. When this
configuration is run straight "out-of-the-box", built using setup, the
clients are not given an /etc/services file, but get the services
database from the yp server, as confirmed by ypwhich and ypcat.  "ypcat
services" yields entries for both who and new-rwho, however "ypmatch who
services" and "ypmatch new-rwho services" give error messages like "Can't
match who.  Reason: no such key in map." In fact, even though we are running
/usr/etc/in.rwhod on the clients, they appear to neither gather nor send out
information. (They do accumulate CPU time). In order to make things work
properly, we have to grab the /etc/services file from another machine.
Then, regardless of whether or not we are running ypbind, rwho seems to work
properly. This leads me to believe that rwhod is not getting its port
information from yp.

What is the correct way to set up the clients? Are they or aren't they
supposed to have an /etc/services file? Will adding it harm anything else?

Thanks,

Skip Montanaro

ARPA: montanaro%desdemona.tcpip@ge-crd.arpa
UUCP: seismo!rochester!steinmetz!desdemona!montanaro
GE DECnet: csbvax::mrgate!montanaro@desdemona@smtp@tcpgateway

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

Date: 3 February 1987 0844-PST (Tuesday)
From: wallen%ics@sdcsvax.ucsd.edu (Mark Wallen)
Subject: used Sun equipment?

We've been considering upgrading our Sun 2/120 and
Sun 2/100U systems, and a question has come up:

Are used Sun 2/120 or Sun 2/100 systems worth anything?
Are there dealers that buy used Sun equipment like there
are with DEC systems?

Thanks

Mark R. Wallen

Institute for Cognitive Science
UC San Diego

wallen@ucsd.edu
wallen@nprdc.arpa

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

Date: 29 Jan 87 15:26:34 GMT
From: Kishore Ramachandran <rama%gt-stratus%gatech.csnet@RELAY.CS.NET>
Subject: gremlin and ditroff previewer for SUN 3/50?

I am looking for ditroff previewer and gremlin for SUN 3/50
running 'suntools' window manager.
If anyone has any info on how I can get these please mail
me at the following address:
	uucp: seismo!gatech!rama
	csnet: rama%gatech@csnet-relay.arpa

Kishore Ramachandran
(404) 894-3813

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

Date: Fri, 06 Feb 87 23:08:22 PST
From: Yuval Tamir <tamir@CS.UCLA.EDU>
Subject: Comments on the Sun 3/110 ?

The 3/110 is Sun's cheapest color workstation.
I have heard from several sources that there are some problems with it.
Specifically:
(1) The graphics is incompatible with other Suns (e.g. you cannot
    use the colors under X-Windows).
(2) Graphics is MUCH slower than on a 3/160.  Specifically,
    Filled rectangle drawing is half the speed of a 3/160C for solid
    rectangles, and something like 1/10th the speed for stipples.

I would like to hear from users of the 3/110:
Are you generally happy with it ?
If you had to do it over, would you get the 110 again or would you
choose between paying more for a 160C or sticking with monchrome ?
What are your major complaints regarding the 110 ?

Please respond by e-mail.

			   Yuval Tamir

Internet: tamir@cs.ucla.edu
    UUCP: ...!{ihnp4,ucbvax,sdcrdcf,trwspp}!ucla-cs!tamir

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

Date: Sat 7 Feb 87 18:46:24-EST
From: David F. Bacon <DFB@CS.COLUMBIA.EDU>
Subject: RFS server for Suns?

i would like to run an RFS server on my sun 3/180.  has anyone ported RFS
to the sun?  it seems to run on a tektronix 68000 workstation, but a naive
compile and run with debug simply exits with no information.

david

ps: this is the public domain RFS, not the AT+T version.

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

Date: Sun, 8 Feb 87 19:04:13 +0100
From: Anund Lie <a_lie%vax.runit.unit.uninett@NTA-VAX.ARPA>?
Subject: SUN graphics libraries?

1)  Is it possible to run suncore within a sunview window (i.e. in the
pixwin behind it)?  The intention was to use the scrollbars etc.
offered by sunview.  We tried to pass "init_view_surface" in suncore
the WINDOW_FD and the WINDOW_DEVICE_NAME, respectively; but it seemed
to ignore the hints and created a new window....

2) If I try to "window_get" both the WINDOW_FD and the
WINDOW_DEVICE_NAME attribute, strange things happen: When I quit the
window, the program core-dumps inside a "getenv" call deep within
"window_main_loop"! It doesn't matter if I "window_get" from two
distinct windows (a "frame" and a "canvas" within it) either. The
attribute values returned appears correct, and if I comment out either
window_get-call, everything is just fine.


Anund Lie
Division of Computer Science
Norwegian Institute of Technology
(A_Lie%vax.runit.unit.uninett@nta-vax.arpa)

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

Date: 8 Feb 87 18:14:35 GMT
From: mcvax!cwi.nl!uucp@seismo.CSS.GOV (The Communications Owner)
Subject: Ethernet<-->SNA interconnect?

We are in the process of connecting a large SNA network with many, many
desktop PC's and a number of Sun workstation ethernet rings.  The PC
connection is reasonably straightforward (as opposed to simple), but we
are in the dark about connecting the Sun and SNA networks.  We would
like to be able to have 3270 emulation and LU 6.2 services available
for applications running on the Suns.  The ability to do file transfers
between various IBM mainframes and the Suns is essential.

Does anyone know of products which would be helpful in this endeavor?

Please send reponses to me and I will summarize for the net.


			Owen Medd
--
UUCP:     {mit-eddie,seismo!umix,ihnp4}!umich!cosivax!osm

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

Date: Fri, 16 Jan 87 11:53:26 CST
From: carey@a.cs.uiuc.edu (John Carey)
Subject: Sun keyboard?

We have several Suns (sun-2).  We would like to place the console
about 50 feet from the CPU, but never have been able to because 
the keyboard can send only about 25 feet.  (These are the microswitch keyboards)
There are two signals from the keyboard to the CPU.  One is data,
the other is a sync signal.  Five volts is sent from the CPU to
the keyboard.  It seems you could put a little box in the keyboard cable. 
The box would contain a little power supply, and 
some kind of appropriate driver chip to boost the signal level
of the signals from the keyboard.

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

Date: Mon, 19 Jan 87 16:08:16 EST
From: strange!color!kurt@ucbvax.Berkeley.EDU (K. Gluck)
Subject: System V layers support?

Has anyone written the code for sun 3.2 that will allow one to use the sun
as a blit/5620 terminal into a vax ?

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

Date: Wed, 21 Jan 87 14:43:52 GMT
From: Paul Cooper <mcvax!stl.stc.co.uk!pac@seismo.CSS.GOV>
Subject: Clearpoint memory and Sun 3/110?

We have a 4MB Sun 3/110 to which we are trying to add a Clearpoint 8MB
memory expansion.  We've got the jumpers set correctly and have changed
the EEPROM configuration parameters but the diagnostics fail when trying
to address over the 8-12MB range.  Any hints as to what is wrong?
Please reply directly to.....
--

-- Paul Cooper	(...!mcvax!ukc!stc!stl!pac)

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

Date: Thu, 22 Jan 87 14:45:58 PST
From: Gary R. Thrapp <thrapp%cod@nosc.ARPA>
Subject: VT100 emulator for Sun 3 windows?

I need to send VT100 graphics and text output to several
SunTools windows.  Are there any vt100 emulators available
to use in Sun 3 windows?

Gary R. Thrapp
Naval Ocean Systems Center
San Diego, CA

DDN: thrapp@nosc.mil
UUCP: {ihnp4,akgua,decvax,dcdwest,ucbvax}!sdcsvax!noscvax!thrapp

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

Date: 28 Jan 87 18:16 +0500
From: naren <naren%systems.carleton.cdn%ubc.csnet@RELAY.CS.NET>
Subject: Tape drives on sun?

I have to buy a third party tape drive on a sun-160.  I have been quoted
a Thorn-EMI (model 9900), CDC (models 92181 and 92185) and cipher (models
M890, M891 and M990) with a Xylogics 472 controller.  CDC 92185 and 
Cipher M990 are 6250 bpi tape drive.  Is one better than the other?  If
you are using one of these, would you recommend it?  Thanks for any info.

Narendra Mehta, 
Systems & Computer Eng. Department, Carleton University, Ottawa, Canada 

UUCP:	{allegra,decvax,ihnp4,linus,utzoo}!watmath!clan!naren
CDN:	naren@systems.carleton.cdn
ARPA:	naren%systems.carleton.cdn%ubc.csnet@csnet-relay.arpa
CSNET:	naren%systems.carleton.cdn@ubc.csnet

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

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