A short while ago I posted the following request for help:
I am having problem with a Sun 4/280 835E0598) server crashing due to
what appears to be an Ethernet problem. The Sun is running SunOS 4.0.3
and is a file and boot server for five 4/110's and four Sparcstation 1's.
Our Sun repair person has been out twice in the last week and has replaced
the CPU board on both occassions. The information in /var/adm/messages
from last night's crash is as follows:
vmunix: ie0: tbd: ff09c1fc, es->es_utbd: ff09c1ec
vmunix: es->es_tbdring: ff09c1ac, tfd: ff09c26c
vmunix: tfd->ietfd_tbd: fc01, es->es_base: ff09c000
vmunix: panic: iexmitdone: tbds out of sync
vmunix: ie0: tbd: ff09c20c, es->es_utbd: ff09c1ec
vmunix: es->es_tbdring: ff09c1ac, tfd: ff09c2ac
vmunix: tfd->ietfd_tbd: c02, es->es_base: ff09c000
vmunix: panic: iexmitdone: tbds out of sync
vmunix:
vmunix: dumping to vp ff005a98, offset 52097
Several people responded. My thanks to Chris Barry, Daniel Trinkle
and also the SunBugs people. Sun's hardware support people also tracked
the problem down. As for the solution, there is a patch for the IECHKCCA
problem (I also had iechkcca crashes). Call usa4sun to get it.
I installed the patch 4 days ago and haven't seen the problem again.
Following is most of the README file that comes with the IECHKCCA patch.
Once again, many thanks to the people on this mailing list.
Sam Cole
Chemistry Computer Center
University of Utah
cole@chemistry.chem.utah.edu
------------------------ from the README file ------------------------------
Bug Report Id:
kenel/network:1014717
Bug Fix Id:
unix/4.0.3/sun4_panic_iechkcca
Problem:
This patch may fix problems with "panic: iechkcca" occuring
on sun-4 machines.
In 4.0.3, a machine will panic with "iechkcca" if the
82586 (ie) chip does not clear it's command register within
an expected "time frame". In general, system panics of
this type have most often been found to be the result of
network problems (e.g., a bad transceiver, DELNI box, or
ethernet cable) or other hardware problems (e.g., a bad or badly
configured disc controller) on the system itself.
A bug has been identified in the 4.0.3 ie driver which
causes the "time frame" mentioned above to be shorter
on a sun-4 than on a sun-3. And, that is, of course, shorter
than had been intended. This patch correctly adjusts the
timer.
In addition, this patch attempts to reset the chip up to
three times, if the time limit expires before the command
register is cleared.
It is important to note that that this fix may not have
any affect on iechkcca panics which are a result of the
types of possible network and/or hardware problems indicated
above.
This patch was originally generated for the SunOS 4.0 release,
and has been recompiled for the 4.0.3 release.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:05:56 CDT