DEFINITY Enterprise Communications Server Release 7
Maintenance for R7r
555-230-126 Issue 4
June 1999
Maintenance Object Repair Procedures
9-1348PROCR (RISC Processor Circuit Pack)
9
Processor Cache Parity Audit (#896)
This audit checks the state of the Cach e Parity bit maintained by the R3000 CPU.
If this bit is set, there may be p roblems with either the CPU, the Instruction, or
Data Cache. An indication of b ad cac he parity won’t itself mean that the system
won’t operate, but minor to severe perf ormance degrad ation may be present. For
example, a single bad bit in either cache c ould cause this bi t to be set. If the
faulty word is seldom acc essed, the impac t will be small. If the bit is ac cessed
frequently, there could be serious p roblems with much, or all, of the c ache (stuck
data bit) with the only direc t indication agai n being the cac he parity error or
cache test error.
Table 9-529. TEST #896 Processor Cache Parity Audit
Error
Code Test
Result Description/ Recommendation
100 ABORT The requested test did not complete within the allowable time period.
1. Retry the command.
1022 ABORT Internal system error
1. Retry the command.
1335 ABORT This test is only run on the active SPE since the standby processor does not
use the cache memory. Cache failures may cause cause cache parity errors
on the active SPE which can be log ged against a standby SPE is an SPE
interchange has taken p lace.
If a PROC 1026 error code is present in the hardware error log for a processor
on the standby SPE:
a. Enter the test memory long command to run the Processor Cache test
(#895) to verify that the cache has not failed.
b. Request an SPE interchange by entering reset system interc h ange
and run the short processor test sequence test processor while the
proc essor is on t he act ive SPE.
c. If all test p ass but a MAJOR alarm is present as a result of a 1026 error
in the hardware error log, rep lace the proc essor circuit pa ck since
multiple cache parity errors have occurred in the past and will probably
occur again which can cause call p rocessing failures.
2500 ABORT Internal system error
1. Retry the command.
1338 ABORT The test is not allowed to run since a planned SPE interchange is in progress.
This may be caused by a planned interchange initiated automatic ally during 24
hour scheduled testing.
1. Wait 3 minutes and retry the command.
Continued on next page