1753335 Members
5418 Online
108792 Solutions
New Discussion юеВ

System Crash!!

 
Peter Clarke
Regular Advisor

System Crash!!

Hi all,

My system has just crashed ive managed to get it back up and running but would like to know what caused it.
I have attached the clue file that i found, anyone got any ideas??

Cheers

Peter
4 REPLIES 4
Kris Clippeleyr
Honored Contributor

Re: System Crash!!

Peter,

Being no expert in crash dump analysis, it seems to me that someone messed up register 22 (contains 0). The failing instruction tries to get the return address into R26 from the contents of R22 (plus 8).
And since this happens in kernel mode, with elevated IPL (2), in the routine that delivers an AST, that's fatal for VMS.

Maybe Volker can shed more light on this.

Greetz,

Kris
I'm gonna hit the highway like a battering ram on a silver-black phantom bike...
Ian Miller.
Honored Contributor

Re: System Crash!!

Its best to have hp have a look as they can check it against known crash footprints. I see tcp/ip related data in the crash so I guess the version of TCPIP will be of interest.
____________________
Purely Personal Opinion
Volker Halle
Honored Contributor

Re: System Crash!!

Peter,

the immediate reason for the crash, ACCVIO = access violation, when trying to access address 8(R22), is the invalid contents of R22 = 0.

Looking at the instruction stream preceeding the failing instruction:

SCH$ASTDEL_K_C+00130: LDL R22,#X0020(R5) <<< load R22 here
SCH$ASTDEL_K_C+00134: ADDQ SP,#X04,R18
SCH$ASTDEL_K_C+00138: ADDL R31,R14,R19
SCH$ASTDEL_K_C+0013C: ADDL R31,R5,R17
SCH$ASTDEL_K_C+00140: LDQ R26,#X0008(R22) <<< crash here (R22=0)

shows that R22 has been loaded from 20(R5). The CLUE REGISTER output shows R5 to point to an IRP for device BG5466, if you format the IRP, offset 0x20 is the offset for the IRP$L_KAST field, the Kernel AST address.

So something may have clobbered the IRP$L_KAST address or the IRP has ended up in the wrong queue or it has been re-used (double deallocation of pool). This is an SMP system.

The system has been up for 41 days, so this does seem to be a very 'intermittent' problem.

Please contact HP with the contents of the CLUE file and the system dump and have them run the CLUE file through the CCAT crash analysis tool.

Volker.
Volker Halle
Honored Contributor

Re: System Crash!!

Peter,

you may also want to consider to install a 'current' version of the ES40 firmware (your system is running V6.5-19, current FW version is V6.8).

Volker.