| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 5218.1 | AXP CIXCD Change history | PROXY::MOORE | Tom Moore PK1/B1 DTN-223-6309 | Fri Jan 24 1997 11:49 | 45 |
| AXP CIXCD Change history
Version 8
Port timeout Fix - Elements were incorrectly inserted onto the response queue.
It showed up as a "port timeout".
Misc. Error 8407 - This error caused an unnecessary port timeout.
Version 7
TTO problem refix - There appeared to be a problem in customer tests for the
TTO fix in REV V6. It appears now that customer needed the fix in V8.
Version 6
TTO problem - Errorlog shows TTO XMI error with the error address = FFFFFFFE00.
Sometimes corrupted message packets were found by a SYSAP where the
data after a pagelet crossing would be from another message.
Rewrite of BDLT routines to simplify and correct errors. No reports that these
bugs were excited by VMS.
Version 5
ARB_TIMEOUT - Reduced the arbitration timeout period to a more reasonable
value. The longer value resulted in REQ_ID requests accumulating on
the command queue until NON-PAGED POOL was consumed and the system and
cluster hung. The system would not crash and would hang the cluster.
Slot time - Since there is only one legal value the firmware forces it to 10
regardless of the jumper setting.
Version 4
2 gig memory fix - The system will crash with a NXM error if the following
conditions were present:
-system has more than 2 gbyte of memory
-NON-PAGED pool is forced to expand because the system is not
configured with enough at boot
-I/O greater than 16 kbyte is requested.
Version 3
4 gig memory fix - The system would crash with a NXM error if an I/O was
requested to s buffer higher than 4 gbyte.
Version 2
2 gig memory fix - The system would crash with a NXM error if an I/O was
requested to a buffer higher than 2 gbyte.
RBD tests - Fix tests so that they would run.
Queue Purging - Fixed command queue purging on VC and connection shutdown.
Version 1
initial release
|
| 5218.2 | VAX CIXCD CHANGE HISTORY | PROXY::MOORE | Tom Moore PK1/B1 DTN-223-6309 | Fri Jan 24 1997 11:50 | 31 |
| VAX CIXCD CHANGE HISTORY
REV 48:
ARB_TO - When the adapter is broken such that it cannot arbitrate on the bus,
REQ_ID commands from the driver accumilate on the command queue until all of
NON-PAGED poll is consumed and the system hangs but does not crash. This in
turn causes the cluster to hang. The fix is use a shorter more appropriate
value which will keep the command queue empty.
SLOT_TIME jumper - Since there is only one allowed value and this causes
configuration problems when incorrently set, the value is forced to 10 ticks
and the jumper is ignored.
REV 47:
RBD tests - Fixed several problems in RBD tests. RBD5 which is VAX7000 specific
now works. The "T/R" command can now be also typed "t/r".
MEI interrupt test - Because the IOP incorrectly reports this selftest it has
been removed to avoid confusing errorlogs.
REV 46
VAX7000 support added
REV 45 or REV 2.5
Address support for more than 512meg added.
Version numbers changed from major.minor form to just REV form. rev 2.5
translates to rev 46. The LFU program show this as a decimal 70.
|
| 5218.1 | AXP CIXCD Change history | PROXY::MOORE | Tom Moore PK1/B1 DTN-223-6309 | Mon Feb 10 1997 13:46 | 52 |
| AXP CIXCD Change history
Version 8 >>>>>>>>> DO NOT USE THIS VERION <<<<<<<<<<
>>This version seems to have a serious bug and should not be installed. The
>>foot print is:
>> "Bugcheck Type: INVEXCEPTN, Exception while above ASTDEL"
>>At this time we do not know for sure if the problem was introduced in V6, V7
>>or V8. At this time recommend V5 because there is a pndriver workaround for
>>the problems fixed in V6 and V7 for VMS versions V6.2 and V7.1
Port timeout Fix - Elements were incorrectly inserted onto the response queue.
It showed up as a "port timeout".
Misc. Error 8407 - This error caused an unnecessary port timeout.
Version 7
TTO problem refix - There appeared to be a problem in customer tests for the
TTO fix in REV V6. It appears now that customer needed the fix in V8.
Version 6
TTO problem - Errorlog shows TTO XMI error with the error address = FFFFFFFE00.
Sometimes corrupted message packets were found by a SYSAP where the
data after a pagelet crossing would be from another message.
Rewrite of BDLT routines to simplify and correct errors. No reports that these
bugs were excited by VMS.
Version 5
ARB_TIMEOUT - Reduced the arbitration timeout period to a more reasonable
value. The longer value resulted in REQ_ID requests accumulating on
the command queue until NON-PAGED POOL was consumed and the system and
cluster hung. The system would not crash and would hang the cluster.
Slot time - Since there is only one legal value the firmware forces it to 10
regardless of the jumper setting.
Version 4
2 gig memory fix - The system will crash with a NXM error if the following
conditions were present:
-system has more than 2 gbyte of memory
-NON-PAGED pool is forced to expand because the system is not
configured with enough at boot
-I/O greater than 16 kbyte is requested.
Version 3
4 gig memory fix - The system would crash with a NXM error if an I/O was
requested to s buffer higher than 4 gbyte.
Version 2
2 gig memory fix - The system would crash with a NXM error if an I/O was
requested to a buffer higher than 2 gbyte.
RBD tests - Fix tests so that they would run.
Queue Purging - Fixed command queue purging on VC and connection shutdown.
Version 1
initial release
|