ACC Release Notes for the B.03.01 Release

25
Advanced Communications Controller Release Notes for the B.03.01 Release
Patches and Fixes for this Version
10. Reference: SR #4701391862
Reference: SR #4701391912
Reference: TPO0h02130
No current method to determine
hardware revision.
An enhancement to detect hardware
revisions of ACC cards was needed. A
standard interface has been defined to
identify hardware revisions of all ACC
cards. The mx command of zmntr has
been enhanced to include the display of
the hardware revision.
11. Reference: SR #5003380527
When kernel parameters msgssz &
msgseg are incorrectly configured, it
causes zmasterd to hang during ZCOM
subsystem start up.
The incorrect parameters caused the
kernel to run out of message buffer. A
timeout mechanism is added to zmasterd,
so it can drop out and display an error
when there is not enough message buffer
in the system.
12. Reference: SR #5003398362
Reference: TPO0h01774
A large number of short packets
received with the M-bit set can lead to a
firmware failure (ACC card restart).
A defect in the X.25 firmware code,
caused queue corruption on the ACC
whenever a short DATA packet was
received (one that is less than the full
packet size but with the M-bit set and
without the D-bit set). The error handling
causes the queue corruption. The ACC
firmware can survive for some time with
this queue corruption, but a lot of these
errors will eventually cause the card to
fail. This defect is fixed in this release.
13. Reference: SR #5003415463
Occasionally, zmlog displays ‘bad ZCOM
log header, and skips log messages
during system start up. This happens
more often in systems with large ACC
configurations.
This problem is extremely difficult to
reproduce in the lab. The relevant
software is enhanced to display more
information when this problem happens.
Refer to “Known Problems and
Workarounds” for more details about this
problem and the enhancement made.
14. Reference: SR #5003417972
Reference: TPO0h02086
System panics when zx25l2stat_rcvr() is
called rapidly from a remote node.
The panic was due to ZCOM memory
corruption. One of the Znode kernel
routines was calling a ZCOM routine
with a missing argument. This corrupted
the ZCOM memory. This is fixed in this
release.
Symptoms Defect / Fix