Sun-Spots Digest, v6n267
William LeFebvre
Sun-Spots-Request at Rice.edu
Fri Oct 21 08:17:46 AEST 1988
SUN-SPOTS DIGEST Wednesday, 19 October 1988 Volume 6 : Issue 267
Today's Topics:
Re: ALM-2 and hardware flow control
Re: Question about Window display lock broken after time limit
Re: mounting remote tape drives
Re: Laserwriter hanging with framedevice error
Re: opinions on configuration wanted
win_* calls
Possible bug in c compiler under SunOS 3.3 ?
Process hangs if its open NFS file is deleted by the server?
Re: client making (includes source)
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: Thu, 13 Oct 88 15:27:53 +0200
From: mcvax!tut.fi!vjk at uunet.uu.net
Subject: Re: ALM-2 and hardware flow control
wbh at hoh-2.att.com (Wilson H. Bent, Jr):
> We though this was true, too, until Sun pointed us in the direction of "16
> Channel Asynchronous Line Multiplexer-2 Field Service Manual and
> Installation Notes" (catchy title), Part No: 813-1029-05 of Dec 7, 1987,
> Table 6-2: ports 0 through 3 have RTS, CTS, TXC, and RXC, ......
Well, I missed that too.
*BUT*
Hardware support is this case is not enough. You have to get mcp driver to
really use those signals. What you can do depends on what version of SunOs
you are running:
SunOs 3.5:
There's nothing you can do.
SunOs 4.0:
You can set flag CRTSCTS flag on some termio struct to get
hardware flow control on *outgoing* stream.
Now your only problem is how to set that flag. You *cannot* set it
as default parameter of line. You cannot set it in getty(tab). You
can set it with stty(1) once you have logged in (don't look
manual, it's not there. Just write 'stty crtscts').
If you want to use that on printer or call-out-modem lines, you
have to patch source of lpd, uucico, tip etc... to set it with
ioctl. (What, no sources. Too bad).
Looks bad, doesn't it.
Vesa
------------------------------
Date: Mon, 17 Oct 88 08:54:17 EDT
From: Chuck Musciano <chuck at trantor.harris-atd.com>
Subject: Re: Question about Window display lock broken after time limit
This error message occurs when a process locks the window system in order
to perform some atomic task, like a confirmer window. The time limit is
not wall clock, but CPU time used, and is rather small. You can ignore
the message. I had it occur when I wrote a menu which dynamically builds
submenus to match your file system. Since a menu display locks the window
system, as I burned CPU time looking at the file system, I would break the
lock. I believe you can dicker with the kernel to change the timeout, but
the value is intertwined with all sorts of other window parameters and is
best left alone.
Chuck Musciano
Advanced Technology Department
Harris Corporation
(407) 727-6131
ARPA: chuck at trantor.harris-atd.com
------------------------------
Date: Mon, 17 Oct 88 10:27:06 EDT
From: bob at allosaur.cis.ohio-state.edu (Bob Sutterfield)
Subject: Re: mounting remote tape drives
Reference: v6n259
edsews!mancer!wintermute!dharvey at uunet.uu.net (Doug Harvey) writes:
>Is it possible to access a remote tape drive without 'rlogin'ing to
>the remote machine?
You can't get to it in the same way as a local device, by MAKEDEV etc.
That works in ATT RFS-land but not on a Sun (yet?).
For backups, use rdump/rmt. If you want to write a tar tape, see the
tar(1) manual page for an example incantation like:
tar cvfb - 20 files | rsh krypton dd of=/dev/rmt0 obs=20b
or the similar spell to read a tape:
rsh krypton dd if=/dev/rmt0 bs=20b | tar xvBfb - 20 files
Neither of these solutions requires you to have an interactive shell
session on the machine with the tape drive (via rlogin or telnet or
whatever). Since dump is run as root, rdump will require an entry in the
tapeful machine's /.rhosts. The rsh/tar scheme will require an entry in
the tapeful machine's /etc/hosts.equiv.
--Bob
------------------------------
Date: Thu, 13 Oct 88 16:15:31 BST
From: Paul Hudson <mcvax!moncam!paul at uunet.uu.net>
Subject: Re: Laserwriter hanging with framedevice error
Sounds like the LaserWriter doesn't have enough room to set up it's
framebuffer. In normal use the LaserWriter reclaims memory used in each
job, but there is a way of downloading stuff that survives from job to
job. (Jargon is "outside the server loop"). I'd guess that one of your
PostScript jobs is filling up VM outside the server loop - commonest way
is to always download your stuff, rather than checking it's there already
- and eventually leaving too little VM for normal operation.
Paul Hudson
Snail mail: Monotype ADG Email: ...!ukc!acorn!moncam!paul
Science Park, paul at moncam.co.uk
Milton Road,
Cambridge,
CB4 4FQ
------------------------------
Date: Mon, 17 Oct 88 10:19:55 -0400
From: Henry B.J. Krempel <krempel at pacrat.npac.syr.edu>
Subject: Re: opinions on configuration wanted
Reference: v6n259
>From: Steve Elias <eli at spdcc.com>
>on the following setup: 1 sun 3/60 with 327M disk & tape drive,
> 4 diskless sun 3/50 clients.
I think you're going to have some trouble acquiring more memory
"after-market" for your 50's. 4M is the maximum for that model. If you
want to add inexpensive (but upgradeable) clients, your best bet is a
3/60, monochrome if you like, but color's only a couple of thou more
(what's a few thou between friends?)
8M for clients is really suggested. Also, be sure you have more than 8M
in the server (you don't want that to swap at all). If you must get 50's,
the local swap disk is a good idea.
As far as the disk performace goes, I'm skeptical about putting a lot of
systems (especially swapping 50's) on that 327; but I'd be curious to hear
if it works out. You might just make it with a 5-node network.
------------------------------
Date: Mon, 17 Oct 88 10:28:26 -0400
From: mesard at bbn.com
Subject: win_* calls
Phone: 617-873-1878
Anthony Worrall writes:
> ...In order to create new windows I have used the call
>
> win_insert(windowfd);
[and more about wmgr_* calls.]
>
> PS. If anybody else is using these functions I would be please to here any
> experiences and/or tips they might care to pass on.
In my experience, it is rarely necessary for an application to call the
system level routines (i.e., those documented in the SunView System
Programmer's Guide).
I've coded in win_* calls many times, only to take them out a week later
when I stumbled across a way to do the same thing at the application
level.
Of course there are exceptions. But by and large these are restricted to
non-application-like things such as vuid input (cf. CM's eyecon program)
or *intense* window-hacking (cf. DH's vine non-application).
There is one time that I used a win_* call and nothing else would have
sufficed (and it fits into my second category, above):
win_screendestroy((int) window_get(frame, WIN_FD));
But calls to this function should be used sparingly in an application:-).
------------------------------
Date: 12 Oct 88 20:56:12 GMT
From: Paul Harrington <mcvax!cs.tcd.ie!phrrngtn at uunet.uu.net>
Subject: Possible bug in c compiler under SunOS 3.3 ?
The following code was tested on a SUN 3/50 under SunOS 3.3. Depending on
wheter you give an initialiser to the enum elements in the declaration of
the type of p or not you get very different outputs from test_fn() ranging
from 1,3,5 to 1,2,4 to 2,4,6 with my old favourite 2,3,4 when you say
a=1,b=2,c=3 in the declaration of the type of p in test_fn()
An RPC package called Courier generates this unusual type of code with
parameter types being redeclared even for functions that have the same
effective types for their parameters.
why does the compiler (I don't know which compiler it is. I assume it is
the standard one supplied with a SUN 3/50) produce this result and is it a
bug of this compiler or a "feature" of C ?
typedef struct {
enum{a=1,b,c}choice;
union{
int a;
long b;
short c;
}u;
}new_struct;
typedef enum {e,f,g,h} e1;
void test_fn1(p)
struct {
enum{a=1,b=2,c=3}choice;
union{
int a;
long b;
short c;
}u;
}*p;
{
printf("the size of p is %d\n",sizeof(p));
printf("(test_fn1) a=%d,b=%d,c=%d\n",a,b,c);
}
main()
{
new_struct *v;
printf("(main) a=%d,b=%d,c=%d\n",a,b,c);
printf("the size of e1 is %d\n",sizeof(e1));
printf("the size of a is %d\n",sizeof(a));
test_fn1(v);
}
------------------------------
Date: Mon, 17 Oct 88 08:32:12 EDT
From: elsie!ado at ncifcrf.gov (Arthur David Olson)
Subject: Process hangs if its open NFS file is deleted by the server?
The scoop: "flower" is a diskless 3/60 that uses NFS-mounted disks
provided by a 3/280 named "elsie"; both are running SunOS 4.0. This
script shows what happens if flower opens a file and then elsie deletes it
while flower has it open:
Script started on Mon Oct 17 08:03:20 1988
flower$ tty
/dev/ttyp0
flower$ cat try.c
#include <stdio.h>
main(argc, argv)
int argc;
char * argv[];
{
register FILE * fp;
register int c;
fp = fopen("xyz", "w+");
(void) fprintf(fp, "This is only a test.\n");
(void) fflush(fp);
if (argc > 1)
(void) system("rsh elsie rm xyz");
(void) rewind(fp);
while((c = getc(fp)) != EOF)
(void) putchar(c);
return 0;
}
flower$ cc try.c -o try
flower$ try &
[1] 171
[1] + Stopped(tty output) try &
flower$ %1
try
This is only a test.
flower$ try + &
[1] 172
>>>> flower$ [1] + Stopped (tty input) try + &
ps -l#172
F UID PID PPID CP PRI NI SZ RSS WCHAN STAT TT TIME COMMAND
8021 102 172 162 7 5 0 48 208 T p0 0:00 try +
flower$ exit
You have stopped jobs
flower$ exit
script done on Mon Oct 17 08:04:49 1988
Heaven only knows where that "tty input" came from; in any event, if I
were to bring the "try" process back into foreground at the marked point,
it would be oblivious to any character I typed (specifically, in a "D"
type sleep on "kernelmap").
And so the question: am I doing something wrong here, or is this a
problem with SunOS 4.0?
ado at ncifcrf.gov ADO is a trademark of Ampex.
------------------------------
Date: Mon, 17 Oct 88 11:04:04 EDT
From: mcgrew at topaz.rutgers.edu (Charles)
Subject: Re: client making (includes source)
Reference: v6n260
Bennett Todd <bet at bent.mc.duke.edu> writes
> Note that I didn't take the Sun approach of specifying the nd
> locations out in the middle of partition C and hoping I get it
> right. I don't know why Sun suggests that.
... because in a certain configuration, there aren't any partitions left
(root, /usr.MC68010, /usr.MC68020, /pub.MC68010, /pub.MC68020, swap,
whole-disk, and one for user files is 8, the max under bsd).
Here's a program to generate nd.local files that you may find useful, it
writes a file called nd.local into the current directory. The client
names are just 'client-1', 'client-2', and so on. Your best bet is to
create it somewhere besides /etc, then edit the result to add the name...
/* this program is copyright 1988 by Charles McGrew, and is placed in
* the public domain. It may be changed, enhanced, redistributed and
* otherwised abused, as long as this copyright remains intact. If
* you come up with neat enhancements, let me know!
* mcgrew at topaz.rutgers.edu
*/
#include <stdio.h>
main()
{
int
both, /* 0 if its only 68010, 1 if only 68020, 2 if both */
partition,
root_size,
sector,
swap_size,
num_clients,
little_swap_size,
medium_swap_size,
large_swap_size,
little_swap_areas,
medium_swap_areas,
large_swap_areas,
num_partition_types,
pub_0_size,
pub_1_size,
offset;
char reply;
FILE *nd_file;
printf("You should put client roots and swaps on cylinder boundaries\n");
printf("\n if you don't know how big a cylinder is in sectors, ^C this\n");
printf("and run /etc/dkinfo. Multiply the number of heads by the number\n");
printf("of sectors/track. That's the number of sectors per cylinder.\n");
printf("This program will keep things on the boundaries as long as you\n");
printf("start on a cylinder boundary and give sizes in cylinder boundaries\n");
printf("\n\n");
printf("how many clients?>");
scanf("%d", &num_clients);
if ((num_clients < 1) || (num_clients > 15))
{ printf("bad number - %d. try again.\n",num_clients);
exit(0);
};
retry:
printf("are there 68010 AND 68020 clients here? (y/n)>");
scanf(" %c", &reply);
if ((reply != 'y') && (reply != 'n'))
{ printf("say what?\n");
goto retry;
};
if (reply == 'y') both = 2;
if (reply == 'n')
{
retry2:
printf("All 68020 (y/n)?>");
scanf(" %c", &reply);
if ((reply != 'y') && (reply != 'n'))
{ printf("say what?\n");
goto retry2;
};
if (reply == 'y') both = 1; else both = 0;
};
printf("how may big a root in each client? (in sectors)>");
scanf("%d", &root_size);
retry3:
printf("how many different swap areas do you want to use (1,2, or 3)?>");
scanf("%d", &num_partition_types);
if ((num_partition_types < 1) || (num_partition_types > 3))
{ printf("bad number, try again\n");
goto retry3;
};
if (num_partition_types > 1) printf("first ");
printf("swap space size (in sectors)?>");
scanf("%d", &little_swap_size);
if (num_partition_types > 1)
{
printf("second swap space size (in sectors)?>");
scanf("%d", &medium_swap_size);
};
if (num_partition_types > 2)
{
printf("third swap space size (in sectors)?>");
scanf("%d", &large_swap_size);
};
retry4:
little_swap_areas = 0;
medium_swap_areas = 0;
large_swap_areas = 0;
if (num_partition_types > 1)
{
printf("number of clients with first swap size?>");
scanf("%d",&little_swap_areas);
printf("number of clients with second swap size?>");
scanf("%d",&medium_swap_areas);
if (num_partition_types > 2)
{
printf("number of clients with third swap size?>");
scanf("%d",&large_swap_areas);
};
if ((little_swap_areas + medium_swap_areas + large_swap_areas) !=
num_clients)
{ printf("numbers don't match, lets try again\n");
goto retry4;
};
}
else little_swap_areas = num_clients;
pub_0_size = -1;
pub_1_size = -1;
if (both != 1)
{ printf("68010 pub partition size?>");
scanf("%d",&pub_0_size);
};
if (both != 0)
{ printf("68020 pub partition size?>");
scanf("%d",&pub_1_size);
};
printf("where does the partition begin (offset in sectors\n");
printf(" from 0 of /dev/xy0c)?>");
scanf("%d",&offset);
nd_file = fopen("nd.local","w");
/* some setup lines */
fprintf(nd_file,"clear\n");
fprintf(nd_file,"version 1\n");
/* first the pub1 (68020) area, now offset from xy0f */
if (pub_1_size != -1) fprintf(nd_file,"user 0 1 /dev/xy0f 0 %d -1\n",
pub_1_size);
/* the pub0 (68010) area, offset from xy0e */
if (pub_0_size != -1) fprintf(nd_file,"user 0 0 /dev/xy0e 0 %d -1\n",
pub_0_size);
sector = offset;
for(partition=0; partition < num_clients; partition++) {
fprintf(nd_file,"user client-%d 0 /dev/xy0c %d %d %d\n",
partition, sector, root_size, partition);
sector += root_size;
swap_size = little_swap_size;
if (partition >= (little_swap_areas+medium_swap_areas))
swap_size = large_swap_size;
else if (partition >= little_swap_areas)
swap_size = medium_swap_size;
fprintf(nd_file,"user client-%d 1 /dev/xy0c %d %d -1\n",
partition, sector, swap_size);
sector += swap_size;
}
fprintf(nd_file,"son\n");
printf("nd partition ends at sector %d\n",sector);
}
Commands of use in this sort of thing:
/etc/mkfs - what to do is described in the admin manual, under
'adding a client to your system'
/usr/etc/setup.files - copies one client partition to another
(except devices. for those, you'll need to
mount /dev/ndlX, where X is 0..numclients-1,
then:)
MAKEDEV std nd0 nd1 ndp0 ndp1 win0 win1 win2 pty0 pty1 pty2 bwtwo0
(to make the devices, in /mnt/dev)
... for the linking gag suggested by Bennett, I'd suggest also
linking rc.local, /stand/diag and copy (you can rebuild the client
from the server), and vmunix (to pub/vmunix).
Hope that helps,
Charles
------------------------------
End of SUN-Spots Digest
***********************
More information about the Comp.sys.sun
mailing list