I64 V8.2-1 update DVD gets CPUSANITY Bugcheck

[James Amend]

I am trying to update by I64 HP rx2600 (1.40GHz/1.5MB)OpenVMS V8.2 to V8.2-1. I have updated the firmware from:
MP FW : E.02.29
BMC FW : 01.52
EFI FW : 01.22
System FW : 02.31
to:
MP FW : E.03.15
BMC FW : 01.53
EFI FW : 01.22
System FW : 02.31
When I boot up on the V8.2-1 CD, I get the following:

1) Upgrade, install or reconfigure OpenVMS I64 Version V8.2-1
2) Display products and patches that this procedure can install
3) Install or upgrade layered products and patches
4) Show installed products
5) Reconfigure installed products
6) Remove installed products
7) Execute DCL commands and procedures
8) Shut down this system
Enter CHOICE or ? for help: (1/2/3/4/5/6/7/8/?)

**** OpenVMS I64 Operating System V8.2-1 - BUGCHECK ****
** Bugcheck code = 00000774: CPUSANITY, CPU sanity timer expired
** Crash CPU: 01 Primary CPU: 00 Active CPUs: 00000003
** Current Process = NULL
** Current PSB ID = 00000001
** Image Name =
**** Error log buffers not dumped to SYS$ERRLOG.DMP
**** Canceled memory dump, no dump file available

I do not have to select anything from the menu. After about 30 seconds I get the bugcheck.

Has anybody else had this problem? If so what did you do to solve it.

thanks
Jim Amend

[John Gillings · Accepted solution]

Jim,

This has been reported to OpenVMS engineering.

Since CPUSANITY is only possible with multiple CPUs, perhaps if you entered DCL and STOP/CPU 1? (though that probably won't help for an upgrade). Another thing to try is boot conversationally and:

SYSBOOT> SET MULTIPROCESSING 0
SYSBOOT> CONTINUE

[Alex Daniels]

I had that on a rx7620 with V8.2 (which is of course an unsupported config, hence why I didn't progress it with my CSC).

John Gillings' suggestion will however mean you only use one CPU.

What I did and what worked for me (with V8.2, which let me use all of my eight CPU's) was at SYSBOOT add six zeros onto the end of SMP_SANITY_CNT and a zero on the end of SMP_TICK_CNT.

I had no CPUSANITY problems after that.

Of course you will want to progress this, with your CSC for a proper resolution, but that should get you running.

[James Amend]

I was able to get into DCL and stop/cpu 1. I was able to install V8.2-1. However, when it rebooted to do autogen, it failed with cpusanity again. I then went into EFI and disabled 1 cpu and rebooted. When it did the autogen, it got a machine check.

Attached is the crash output: mancheck.txt.

I tried the multiprocessing = 0, and changing smp* to no avail. Guenther suggested that I pull my FC controller which is not hooked up anyway because the FGE driver is on the stack. I am going to do that now.

jim

[Robert Brooks_1]

FGEDRIVER is *not* supported on I64, so yes, the card should be pulled.

[James Amend]

Pulling the FC card took care of the problem. I am up and running now. Thanks for all of the input.

jim

[Volker Halle]

Jim,

let me try to summarize this:

OpenVMS I64 only supports the A6826A (Dual port fibre channel PCI-X HBA), also called QLogic ISP2312, which is driven by SYS$PGQDRIVER (a port driver to the existing DKDRIVER). The devices are still called FGA0, FGB0 ...

For instructions on how to verify installation of the A6826A FC HBA see:

http://docs.hp.com/en/A6826-96007/ch02s07.html

If you've seen SYS$FGEDRIVER in one of the registers in the MACHINECHK crash, this seems to indicate, that you were using an unsupported FC HBA.

Volker.