jr@amanue.UUCP (10/03/87)
A while back I posted a couple of articles about problems getting getty to
respond to BREAK on tty000. As I recall, the last news I mentioned was
that the hotline had verified that the driver was indeed eating a null,
which is what getty uses to tell it to switch baud rates. (A BREAK should
come through as a null with a framing error at any baud rate.) My respondent
at the hotline said he'd passed it along to the group that actually fixes bugs.
The other day I was actually *sent* a floppy by AT&T containing a new kernel
that does indeed seem to fix it! Three cheers for AT&T and for the hotline!!
I'm impressed. I was pleased that whoever made up the floppy just made it a
good ole fashioned cpio archive -- I didn't even have to mess with ua to
install it. The new kernel is shown as UNIX3.51.1. When it boots, though,
where the old kernel said something about 3.51, the new one simply says 000a.
One curious thing about this kernel is that it's nearly 200 bytes *smaller*
than the old one. That looks like more than a tiny change to the driver!
Now for everyone else out in UNIX-PC-land I'm not sure what the right procedure
is for you to get this fix. I've only had my machine a couple of months or
so -- I gather there are things called FIXDISKs that are official AT&T fixes.
I just read an article by someone claiming that AT&T had told them there was
*no* FIXDISK for 3.51. It's possible that whoever authored this fix just
hasn't had time for it to percolate through the normal channel whereby a
FIXDISK is created, and sent it to me "hot off the press". I'm assuming in
due course there will be a way for everyone else to get this. It was an AT&T
unit in Chicago, & they called me to tell me they would have a fix; good
thing because he'd built a 3.5 kernel. So that means you should in theory
be able to get the fix even if you're not at 3.51.
Maybe someone from AT&T on this group can clarify how everyone will be able
to get this fix.
--
Jim Rosenberg
CIS: 71515,124 decvax!idis! \
WELL: jer allegra! ---- pitt!amanue!jr
BIX: jrosenberg seismo!cmcl2!cadre! /ford@crash.CTS.COM (Michael Ditto) (10/04/87)
In article <241@amanue.UUCP> jr@amanue.UUCP (Jim Rosenberg) writes: >The other day I was actually *sent* a floppy by AT&T containing a new kernel >that does indeed seem to fix it! Three cheers for AT&T and for the hotline!! Three more cheers from here... I, too, called the hotline and reported this bug, and I, too, received a disk containing a cpio archive of UNIX3.51.1 about a week ago. The first person I talked to at the hotline (after the person who asks for serial #, that is) claimed that it was a bug in /etc/getty introduced when they changed it for the windowing system (sounded questionable to me). I later was called by a "real person" who sounded like he had the code in front of him, and he asked me about the problem. He called me back a few hours later and said he had found and fixed the problem (in the tty000 device driver) in the 3.5 release and was getting the 3.51 source on his system to make the same fix to it. I got the disk a week later. It seems that there are "real people" hidden in the depths of AT&T support, and that they can and do support this system and actually fix bugs. It also seems that when they fix the bug, everyone who reported it gets the fix. Apparrently they will only listen to you if your software service agreement is in effect (90 days after purchase); I may find out soon, as mine runs out in a few weeks. I wonder how many people reported the tty000 break bug, and whether the number of people had anything to do with how quickly it got fixed. I also was told a few days ago by AT&T that a fix for the ksh bug is on its way to me. (The bug is in emacs editing mode; typing more than 100 characters on a line causes a core dump). I have repeatedly been very impressed by AT&T service for this machine, especially considering the fire sale price. My machine did not work when it first arrived; after a toll-free call to AT&T service, a serviceman arrived within two hours and replaced the motherboard -- no charge. Maybe I'll even give three MORE cheerss... -- Michael "Ford" Ditto -=] Ford [=- P.O. Box 1721 ford@crash.CTS.COM Bonita, CA 92002 ford%oz@prep.mit.ai.edu