INVEXCEPTN, Exception while above ASTDEL

[Mohamed K Ahmed]

Hello all,

Suddenly my ES40 OpenVMS 7.2-1 system crashed, After it rebooted and the shadow disks were merging, I checked the disks and I found one was offline, it was bad and I replaced it.
I tried to find why the system crashed and looked in the Internet for any clues regarding this problem, what I found is it might be caused by a hardware problem or a software problem where some instruction failed while the IPL limit is above ASTDEL.
What does these abbreviations stand for?

Attached is the SDA CLUE CRASH output, may be someone can give me a clue !!

Thanks
Mohamed

[John Gillings]

Mohamed,

This forum is not really suitable for analysis of crash dumps. I've looked at your CLUE output, but not been able to find a match for it. To get a proper analysis, please log a case against your service contract with your local Customer Support Centre.

What the acronyms stand for:

IPL is "Interrupt Priority Level" - it's a mechanism used by the operating system to synchronise access to data structures by software interrupts. Put simply and interrupt has a "level" in the range 0 - 31. At any time a processor is running at a particular level. Only interrupts with a higher level than current are allowed.

Normal user mode code runs at IPL 0. ASTDEL is the IPL at which ASTs are delivered to processes. This is the highest level at which exceptions are tolerated, since anything running at a higher IPL is, by definition, part of the operating system, rather than user code.

INVEXCEPTN means an exception has occurred where it shouldn't have. The signal array gives the exception, which would translate to the message:

%SYSTEM-F-ACCVIO, access violation, reason mask=04, virtual address=0236A4EC, PC=800ECFD0, PS=00000804

Interpretation... reason mask 04 means an attempt to write to non-existent memory at the stated virtual address. The PC points to module PROCESS_MANAGEMENT, routine SCH$QAST offset 100. The PS indicates the CPU was in KERNEL mode at IPL 8, executing an interrupt.

QAST is attempting to deliver an AST to a process. That section of code is inserting the ACB (AST Control Block) into the appropriate queue. So, it appears the ACB has been corrupted somehow. That may have something to do with your bad disk, but it would require deeper analysis of the dump itself to say for sure.

Other than logging a case... consider upgrading to a current version of OpenVMS, or installing all rating 1 ECOs for V7.2-1, plus any ECOs relating to the I/O subsystem and shadowing.

[Mohamed K Ahmed]

Thanks for the info.
Mohamed

[Vinit Adya_1]

Last time I had a similar crash I had fauly backup cache on the CPU. It would be nice to run DIAGNOSE for the last couple of days and check if any hardware CPU/Memory errors are present.