netnews@wnuxb.UUCP (Heiby) (06/27/85)
Unix Technical Digest Thu, 27 Jun 85 Volume 2 : Issue 13
Today's Topics:
'tablet' line discipline in 4.2 BSD
/etc/passwd protocols
ARPA time in SYS III?
Callan (v7) losing blocks upon power-up
chroot(2)
Emulex CS32 multiplexer problems
requests to mod.unix
----------------------------------------------------------------------
Date: Tue, 25 Jun 85 09:41:11 MDT
From: utah-cs!utah-gr!thomas@utah-gr.UTAH-CS (Spencer W. Thomas)
Subject: 'tablet' line discipline in 4.2 BSD
The 4.2 "tablet" discipline supports a couple of tablet types -- yours
might be one of them. Basically, all you need to do is get your tablet
running, then select the appropriate line discipline. When you read,
you get a struct
struct tbpos {
int xpos;
int ypos;
short status;
short scount;
}
back. If there isn't support for your tablet type, you need to do a
little more work, but reading the code and modifying one of the supplied
types isn't too hard. I did it for a bitpad a couple of years ago.
--
=Spencer ({ihnp4,decvax}!utah-cs!thomas, thomas@utah-cs.ARPA)
------------------------------
Date: Mon, 24 Jun 85 14:47:25 cst
From: ihnp4!sask!zaphod!bobd (Bob Dalgleish)
Subject: /etc/passwd protocols
On Ultrix, (close enough to 4.2BSD for the moment), there are suite of
programs and protocols for manipulating the password file. These
include 'vipw', 'chfn', 'chsh', etc. We wish to provide our own chfn
and finger programs (less than .01% of the 4.2 community actually works
in either Evans or Cory halls), but updating the password file is still
unclear to me. The following is the procedure that we might follow:
Get READ-ONLY permission on /etc/passwd (using flock(2)).
Copy /etc/passwd into /etc/ptmp (put EXCLUSIVE-WRITE permission on
/etc/ptmp).
Modify /etc/ptmp to our hearts content.
Change the READ-ONLY permission on /etc/passwd to EXCLUSIVE-WRITE.
Copy /etc/ptmp back to /etc/passwd.
Release the lock on /etc/passwd, and delete /etc/ptmp.
This is obviously a correct and sufficient procedure, if
1) All of the other programs follow the same sequence. Do
they?
2) It prevents all simultaneous updates to /etc/passwd but can
cause an unreasonable(?) amount of lockout while vipw is
running.
3) It can obviously be circumvented by any program that does
not use the advisory locking mechanism.
Is there a more efficient way to do the lockout at the cost of
simultaneity? Suggestions will be greatly appreciated.
------------------------------
Date: Mon, 24 Jun 85 17:17:54 est
From: seismo!hadron!jsdy (Joseph S. D. Yao)
Subject: ARPA time in SYS III?
Two suggestions. (1) use localtime() [see ctime(3)] to get the local
time; extern long int timezone will contain the (numeric) time zone;
and you can format it with printf to your heart's content. Or you
can do a system("date +FORMAT"); where FORMAT is arpa format. Or
have I missed your r e a l question?
Joe Yao hadron!jsdy@seismo.{ARPA,UUCP}
------------------------------
Date: 25-Jun-1985 18:03-EDT (Tuesday)
From: masscomp!dobbs!leiby
Subject: Callan (v7) losing blocks upon power-up
Dieter Muller (djm@nmtvax.UUCP) writes:
> After a power-up, I have it perform a file system check. It
> consistently comes up with about 25-30 blocks misplaced.
> (I power-off right after de-buffering everything with 'sync').
Read the man page for the /etc/reboot command, I'm pretty sure v7 has it. :-)
Most computing systems provide a privileged command of some sort to take
the system down cleanly. Just syncing the disks is not enuf.
--
Rt. Rev. Mike Leibensperger, Archbishop of Chelmsford
Church of St. Clint the Righteous ("Feel lucky, Pink Boy?")
Masscomp; 1 Technology Park; Westford, MA 01886
{decvax,harpo,tektronix}!masscomp!leiby
------------------------------
Date: Tue, 25 Jun 85 09:41:11 MDT
From: utah-cs!utah-gr!thomas@utah-gr.UTAH-CS (Spencer W. Thomas)
Subject: chroot(2)
>main(argc,argv,envp)
>char **argv, **envp;
>int argc;
>{
> if (chroot("/ultrix") != 0) {
> perror("newroot");
> exit(1);
> }
> execve("/bin/csh", argv, envp);
If you don't have /ultrix/bin/csh, then the exec will fail because it
won't be able to find the csh. You need to make a copy of most of the
"system" files on your new root (including /etc/passwd, if you
want to use ~).
> printf("exec failed. ");
> perror("newroot");
>}
>
--
=Spencer ({ihnp4,decvax}!utah-cs!thomas, thomas@utah-cs.ARPA)
------------------------------
Date: Tue, 25 Jun 85 11:20:32 EST
From: seismo!munnari!basser.oz!john
Subject: Emulex CS32 multiplexer problems
We have recently purchased one of these. We are driving it with a
VMZ driver which we wrote ourseleves (and has been working fine
on VMZs for a long time). A careful perusal of the manual doesn't
reveal any significant differences from a VMZ, yet, with the thing
driving one 9600 baud terminal and the 780 in single-user mode,
it outputs characters in bursts. The bursts are maybe 40 or so
characters long, with pauses of about three or four seconds in
between.
Needless to say, this makes it useless. If we can't make it work
better than that we'll have to send it back. Any assistance would
be appreciated (indeed, I'd be glad just to hear from someone who
has one of the things working... I've begun to wonder if they are
ALL like that!). Also, I seem to recall someone mentioning
on net.unix-wizards sometime last year that they had had dealings
with Emulex Customer Service via USENET, and found them helpful:
I'd love their address.
Thanks in advance, and I'll be glad to summarize replies.
John Mackin, Basser Department of Computer Science,
University of Sydney, Sydney, Australia
------------------------------
Date: Mon, 24 Jun 85 14:22:34 edt
From: packard!harvard!cybvax0!fbp (Rick Peralta)
Subject: requests to mod.unix
I always have a lot of questions and would like to submit a few to you.
1) Where are there "GOOD" sources for internall system descriptions.
I have Kernagan & Ritchie, Kernagan & Pike as well as some other
UNIX texts. Associated with that What schools are worth looking
in to for "advanced" UNIX hacking? Is there a list somewhere of
all of the schools that have source (should I see net.college) ?
[Ed note: Without a source license, good internals info cannot legally be
acquired. With a source license, one good source is AT&T courses. I took
the Internals course from AT&T and found it to be quite good. There are
other good courses, as well. RWH.]
2) When compiling a kernel (my pc runs Xenix and at work 4.2 and ultrix)
what do the config variables mean ?
buffers disk blocks kept (cached) in memory ?
inodes inodes kept in memory ?
files # open files ?
clists ???
locks # system semaphoers
There are some more I think, but can't remember.
3) Is it more efficient to use multiple devices ? ie: one large disk
or a few smaller ones. Theoretically it shouldn't make a differance.
However if the device is very large, it should get fragmented more
severely and data would be spread all over the media (lost of seek
time). Also the trip from swap to the end of the data segment can
get expensive when running a disk intensive task (like compiling).
Rick
...!cybvax0!fbp
"A likely story. I don't believe a word of it."
------------------------------
End of Unix Technical Digest
******************************
--
Ronald W. Heiby / netnews@wnuxb.UUCP | unix-request@cbosgd.UUCP
AT&T Information Systems, Inc., Lisle, IL (CU-D21)