"XPNET 3.1 Interim Bug Fixes (IBFs) and Enhancements - X25NI" SCR# Change Description ----- ------------------ SCR #3260 X25NI - rel3^ver0^X2518^050831 31-AUG-05 X25NI - rel3^ver1^X2502^20050831 Reference: Case #399105 Symptom: When running X.25 on XPNET, a line logs the following log messages and then goes abnormal: Line write timed out Line trunk failure 12 (file in use) on , circuits suspended Problem: XPNET's write to the X.25 circuit timed out after 45 seconds and XPNET didn't disconnect. The line was configured to Make Initial Call (MICALL) so when XPNET tried to re-open the circuit it was already in use. Change: Added logic to disconnect the circuit after the write timeout. If the line is configured to MICALL, it will try and re-connect immediately. If configured to Wait for Call (WFCALL) or Make Call (MCALL), it will stay in the started state waiting for the appropriate time to connect. Implementation: Install the new X25NI code and re-build the network using NETBC. Dependencies: None. SCR #3357 X25NI - rel3^ver0^X2519^060905 05Sep2006 X25NI - rel3^ver1^X2504^20060905 Reference: Case #411267 Symptom: When running X.25 on XPNET, the node dumps with trap #8409. Problem: SCR 3260 was put in place to fix a disconnect problem. With SCR 3260 and the line configured to MICALL, XPNET attempts to disconnect twice when there is a write timeout. Change: Added logic to correct the disconnect problem when the line is configured to MICALL. Implementation: Install the new X25NI code and re-build the network using NETBC. Dependencies: None. SCR #3359 X25NI - rel3^ver0^X2519^060905 13Nov2006 X25NI - rel3^ver1^X2504^20060905 Reference: Case #421578 Symptom: When running X.25 on XPNET, the node dumps with trap #8092. Problem: XPNET issues a write on an X.25 station and starts the write timer and an activity timer . The activity timer expires before the write completes and the write is canceled without deleting the write timer. When the write timer expires the line has the wrong status so XPNET dumps. Change: Added logic to delete the write timer when the activity timer expires. Implementation: Install the new X25NI code and re-build the network using NETBC. Dependencies: None. SCR #3484 X25NI - rel3^ver0^X2520^090330 30-MAR-09 - rel3^ver1^x2505^20090330 Reference: Case 493096 Symptom: - XPNET DUMP (trap 8092) in X25 protocol. - SCR3208 is installed but doesn't fix this instance of trap #8092. - Messages similar to the following may log just before the dump. 09-03-26;18:22:56.532 ACI.XPNET.3100 6392 Line L1A^X25 write timed out 08-11-10 11:42:35.967 ACI.XPNET.3100 6022 Line L1A^X25 clear request or invitation to clear timed out - The stack trace will look something like the below: TRAP 8092 (XPNET 3.1 X2504) 0 TAL #PROGRAMMATIC^DUMP.#1117(DLIB06TS) + %HAI 1 TAL #NETDUMP.#18894(EXEC11TS) + %H3I 2 TAL #X25NI^IO.#7269(X2504TS) + %H4I 3 TAL #X25NI^DATACOM.#2997(X2504TS) 4 TAL #INTERUPT^LINE.#6081(DCOM06TS) + %H6I 5 TAL #REL3^VER1^EXEC11^20080622.EXECUTE.#4379 (EXEC11TS) 6 TAL #REL3^VER1^EXEC11^20080622.#4636(EXEC11TS) Problem: A write timed out while an i/o to audit input was pending, SCR3165 didn't allow for that combination. Change: Modified the X25NI protocol to handle the situation where a write timer expires when an audit input is in progress. Implementation: Replace the existing X25NI module and re-build the network. Dependencies: None. SCR #3551 X25NI - REL3^VER1^X2506^20110301 01-MAR-11 Reference: Case #01097905 Symptom: - XPNET traps with SIGNUM 8 (arithmetic overflow) in procedure X25NI^LINE^ERROR Problem: XPNET 3.1 was changed to increase many counters from an INT to an INT(32) including lin.input^fail. In X25NI there is logic to generate an EMS error message approximately every 5 minutes depending on the line timeout. In order to do this modulo division is performed (lin.input^fail '\' 300). The result of this modulo division must be an INT (0-65535) or an arithmetic overflow will occur. This was not an issue when the input failure counter was an INT, but when it was changed to an INT(32) it allowed lin.input^fail to become large enough (19,660,800 or greater) to cause the result of the modulo division to be greater than 65535 causing the XPNET node to abend. Change: Under PTAL modulo division was changed to allow the use of FIXED and INT(32) operands which allows the result to be an INT(32) [2,147,483,647] verses an INT [65535]. Modified the logic to take advantage of this which prevents the trap. NOTE: There is a work around for this until the fix can be installed. RESETSTATS LINE before the failure count reaches 19,660,800. This count can be seen on the STATISTICS LINE , DETAIL response. NCPCOM > STATISTICS LINE *, DETAIL LINE: \SYS1.P1A^NODE.L1A^X25^1 MSGS PENDING: 0 PROTOCOL: X.25 INPUT COUNT: 0 INPUT FAILURES: 0 <<< OUTPUT COUNT: 0 OUTPUT FAILURES: 0 LAST RESET: Implementation: Install the new files on XPNET subvol. Rebuild the NETWORK object using the NETB file. Dependencies: XPNET 3.1 only.