FATAL BUGCHECK, SSRVEXCEPT, Unexpected system service exception

[Navipa]

Hi,

our environment; Charon-VAX 4000 model 108 and VMS v6.1.

Today our one of the test server rebooted automatically while it was running one of the application program with SSRVEXCEPT BUGCHEK. Our SYSDUMP file is date =(c,m) very old. But Ana/Err shows a " process name  = BAT", I don't know what is this BAT process, we don't know what is this process "BAT". I searched the entire server for any program with that name, nothing.

I have pasted  the console $ANA/ERR output nelow

******************************* ENTRY 10797. *******************************
ERROR SEQUENCE 18027. LOGGED ON: SID 13002003
DATE/TIME 18-JUN-2015 12:01:28.17 SYS_TYPE 03210B01
SYSTEM UPTIME: 1 DAYS 06:31:16
SCS NODE: TELKOM1 VAX/VMS V6.1

FATAL BUGCHECK KA53 CPU Microcode Rev # 3. CONSOLE FW REV# 2.1
Standard Microcode Patch Patch Rev # 16.

SSRVEXCEPT, Unexpected system service exception

PROCESS NAME BAT

PROCESS ID 004A0042

ERROR PC 8BF849E3
ERROR PSL 00020000
INTERRUPT PRIORITY LEVEL = 02.
PREVIOUS MODE = KERNEL
CURRENT MODE = KERNEL
FIRST PART DONE CLEAR

STACK POINTERS

KSP 7FFE7678 ESP 0038952C SSP 7FFECA44 USP 7FE0A274 ISP 8D71F600

GENERAL REGISTERS

R0 00020009 R1 7FFE7780 R2 00039910 R3 8BF3A274 R4 8CD04A00
R5 00000000 R6 00000031 R7 8B054900 R8 8CDAA980 R9 003C0E00
R10 00039917 R11 00000006 AP 7FFE775C FP 7FFE7744 SP 7FFE773C

SYSTEM REGISTERS

P0BR 94E85800
P0 PTE BASE (VIRT ADDRS)
P0LR 00002167
TOTAL P0 PAGES
P1BR 9488D600
P1 PTE BASE (VIRT ADDRS)
P1LR 001FF049
TOTAL NON-EXISTENT P1 PAGES
SBR 0FBFFA00
SYSTEM PTE BASE (PHYS ADDRS)
SLR 000F9180
TOTAL PAGES "SYSTEM" VIRT MEM
PCBB 0A083E20
PCB BASE (PHYS ADDRS)
SCBB 0FBFA600
SCB BASE (PHYS ADDRS)
ASTLVL 00000001
EXECUTIVE MODE AST PENDING
SISR 00000000
INTERRUPT REQUEST ACTIVE = 0.
ICCS 000000C1
RUN
INTERRUPT ENABLE
INTERRUPT

KA53 REGISTER SUBPACKET

TODR 652FE782
BPCR FECA0000
PAMODE 00000000
30 bit physical address mode
MMEPTE 00000000
MMESTS 00000000
PCSCR 00000000
Patchable control store disabled
no microcode patch
CPU microcode Patch Rev # = 0.
ICSR 00000001
virtual instruction cache enabled
ECR 0000000A
SUBSET INTERVAL TIMER ENABLED
TBSTS 00000000
PCCTL FFFFFE13
pcache enabled for D-stream reference
pcache enabled for I-stream reference
pcache parity error detection enabled
pcache redundancy elements enabled
PCSTS 00000000
CCTL 00000021
backup cache enabled
BCACHE TAG RAM SPEED:
read = 3 cycles, write = 3 cycles
BCACHE DATA RAM SPEED:
read = 2 cycles, write = 3 cycles
512 kilobyte backup cache
BCEDSTS 00000000
BCETSTS 00000000
MESR 00000000
MMCDSR 08000000
2600 cycles before disown write tmeout
enable logging soft errors
CQBIC on CP_I02
CESR 00000000
CMCDSR 0000C108
DMA prefetching enabled
3200 Cycles Before NDAL Timeout
144 cycles before cp1 mt timeout
144000 cycles before cp2 mt timeout
cp1 interrupts pending:
none
cp2 interrupts pending:
none
CEFSTS 00019200
NESTS 00000000
NEOCMD 00000000
NEICMD 00000000
DSER 00000000

Any clue?

Thanks

Navipa

[Steven Schweda]

> Ana/Err shows a " process name = BAT", I don't know what is this BAT
> process, we don't know what is this process "BAT". I searched the
> entire server for any program with that name, nothing.

   A process name may have nothing to do with an executable name. For
example, around here:

alp $ show system /batch
OpenVMS V8.3  on node ALP 18-JUN-2015 15:32:15.61 Uptime 48 16:18:01
   Pid    Process Name    State  Pri      I/O      CPU      Page flts  Pages
20200430 BATCH_767      LEF      5  3522675   0 00:08:00.70   2512520    101  B
20200442 DTSS$PROVIDER   LEF      5     2141   0 00:00:00.49       512    424  B
20211B51 BATCH_596      LEF      5      199   0 00:00:00.14       179    135  B

   I don't have any executables named "BATCH_596" or "BATCH_767",
either.  See, for example:

      help set process /name
      help spawn /process
      help submit /name

And there are system services which can affect it, too.

   Possibly useful:

      help analyze /crash_dump

[abrsvc]

The problem occurred in system space, so I'm not sure the process name is important anyway.

Use the following to narrow down the area within OpenVMS where the exception occurred:

SDA> READ/EXEC

SDA> MAP <offending address>

[Volker Halle]

Navipa,

the bugcheck errlog entry is not enough to get an idea about the problem. You need to analyze the dump file. Note that the creation/modification date of the dump file does NOT matter, those dates refer to the creation of the file and NOT to the last dump written.

Has there a valid crash dump been written ?

$ ANA/CRASH SYS$SYSTEM:SYSDUMP

SDA> SHOW CRASH

SDA> EXIT

As you're running OpenVMS VAX V6.1, there should also be a CLUE file with a summary of the crash information. Is there a CLUE$OUTPUT:CLUE$LAST_*.LIS file ? If so, it might contain the crashdump information, if you have not yet shut down this system after the crash.

Otherwise try:

$ CLUE:==$CLUE

$ CLUE/DISPLAY                                                ! will give you the crash history of this system

CLUE>EXTRACT/OUT=filename.TXT n     ! where 'n' is the number of the crash from column 1

CLUE> EXIT

Then look at 'filename.TXT' and consider to attach that file to your next reply.

A SSRVEXCEPT crash happens in the context of the current process, but the current image is not necessarily the culprit. The image has called some system service and a problem has happened while executing the system service, thus causing the crash.

Volker.

[Navipa]

Thanks Volker,

I was expecting the response from you, because I see there are many OpenVMS CRASH related posting in the net with your name.

I have taken the output and attached here. I searched application source and sys$disk for any reference to LIB$EXTZV routine, but I don't find any except $3$DIA0:[SYSE.SYSCOMMON.SYSLIB]PASCAL$LIB_ROUTINES.PAS.

Thanks

[Volker Halle]

Navipa,

I cannot see any attached file, can you ?

Volker.

[Navipa]

Hi Volker, 

oh! I had attached .RTF.  Now I have reattached .doc here.

Thanks

Navipa

[Volker Halle]

Navipa,

this 'SSRVEXCEPT crash at VAXCLUSTER_CACHE+022D1 EXTZV #00,04(R4),R10,R11 in  VMS 6.1' seems to be a known problem. The reported solution is supposed to be included in VAXSHAD04_061 (or higher).

The crash happens due to a 'reserved operand fault'  executing the following instruction:

VAXCLUSTER_CACHE+22D1  PC>8BF3A2D1 EXTZV   #00,04(R4),R10,R11

This can only happen, if the length value stored in @R4+4 is GTRU 32. You can verify this in the dump with:

$ ANA/CRASH SYS$SYSTEM:SYSDUMP.DMP

SDA> EXA @R4+4

8CD04A04: ?

SDA> EXIT

Do you see previous crashes with this footprint, if you look at the crash history of this system using $ CLUE/DISPLAY ?

Is that patch installed on this system ?

Volker.

[Navipa]

Thanks Volker,

yes I see similar kind of crash earlier "SSRVEXCEPT crash at VAXCLUSTER_CACHE+022D1 EXTZV #00,04(R4),R10,R11 " recorded in the SYSDUMP. I will work with vendor. I have applied patches earlier to VAX, Alpha and Itanium, but not to CharonVAX server. Do I need to follow the same procedure/method to apply the patch to this CharonVAX too?

Btw, what is this VAXCLUSTER_CACHE, our server is not clustered.

Thanks

Navipa

[Volker Halle · Accepted solution]

Navipa,

this crash is a result of an OpenVMS VAX V6.1 software problem, it's irrelevant whether you're running this version on a real VAX or on a CHARON-VAX emulator. You will need to do some resaerch to find out, if that VAXSHAD04_061 patch or a higher version has been installed on this system.

VAXCLUSTER_CACHE is the 'virtual I/O cache'. It also is active also on non-clustered systems.

A worarkound to prevent this crash from happening is to disable this cache by setting the system parameter VBN_CACHE_S=0

Volker.