This is a further update to say that the problem has been resolved as far as
we are able to test. A patch is required (a replacement ace.mod file) and a
kernel rebuild. We have been successful with 4.0d patch level 3 + ace.mod
patch.
Thanks to:
Dr. Tom Blinn
Dave Bono (our Compaq on site engineering support)
The Unnamed (those in Compaq support and engineering who have obviously
contributed but are unknown to me)
Synopsis:
There is a bug in 4.0d which means that when it is run on 2000 boxes
(Jensens) the async ports do not work properly. If the console is on tty00
then tty01 is inaccessible, if the console is on the QVision display then
port tty00 can be used but tty01 cannot.
Compaq response:
We have now been officially (through our support channel) sent by Compaq a
test patch which can be applied after patch #3 to 4.0d to fix the problem
and make both ports accessible again.
Status (Changed since last SUMMARY message):
e have tested the patch as best we can and it works fine. Our testing has
been limited to systems with the console on tty00. we have not been able to
test on systems with the console on the QVision graphics device.
Other comments:
1. As far as we know this problem is only relevant to the Jensen hardware.
2. There is no agreed date for inclusion of the test patch in an official
patch yet.
3. I have no idea how anyone else would obtain the patch except through the
support channels.
------------------- Original problem statement -------------------------
We have a number of old 2000 boxes (Jensens) which we have been recommended
by our support engineers to upgrade to 4.0d for millennium compliance. They
were running 4.0.
Under 4.0 we used a default kernel configuration and had no problems running
our software which uses the inbuilt COM2 serial port (/dev/tty01). The COM1
port is used by the console since the systems have no graphics head.
After the upgrade we found that trying to open the device /dev/tty01 gives
an open failure. We have hunted around, and been recommended to try various
things including:
deleting the special files and recreating using MAKEDEV ace0
rebuilding the kernel
Neither have achieved the desired result. There has been no change in the
hardware so we have not rerun the EISA config (the site engineers did not
seem very keen to try this, I think they have had bad problems with ECU).
Anyone got any ideas? Does a simple upgrade require ECU to be rerun?
Maybe this would help: if you try to run something like cat or echo with
output redirected to the device you get a message back
cannot create /dev/tty01
The special files are set world read/write
---------------------------------------------------------------
Malcolm Melville email: malcolm_at_eml.co.uk
Enterprise Melville Limited Tel: +44 1435 869911
Network and Aviation Consultancy Fax: +44 1435 869912
Mobile: 07771 78-14-78
---------------------------------------------------------------
Received on Thu Feb 04 1999 - 16:15:40 NZDT