Unaligned kernel access

[Prakash_16]

Hi all,

I have two Compw ES40 Servers running tru64 5.1B cluster. Oracle 9.2.0 is running on this.

Today one server rebooted with unexpectedly with the following erro:

Jul 29 11:13:38 atd1 vmunix: Unaligned kernel access va=0xfffffc00012da3dc pc=0xffffffff000b8ef8 ra=0xffffffff000bfcd0
Jul 29 11:13:38 atd1 vmunix: panic (cpu 1): Unaligned kernel space access from kernel mode

The other cluster member was intact.
It created some crash files, i can send it, if required

Can anybody give some input?

Thanks in adance

Prakash

[Johan Brusche]

Based on that info nobody can give any other input than "install patchkit #4". 2 cases of "Unaligned kerne; space access..." are corrected with it.

Maybe with the crash-data.y we could make some more educated guesses, but to really analyze such panic one needs the vmunix.y and vmzcore.y

__Johan__

[Ralf Puchner]

the given message is without any technical information. Please post crash-data next time.

the "unaligned kernel access" message is too common and adresses are useless without using the same environment for debugging/analysis.

[Prakash_16]

Hi

Thanks for the replies.
Patchkit 4 is already applied without any error. I am the crash-data file. Vmunix.N & vmzcore.N are too big to attach ( 29 & 75 MB)

Any update?

Kind regards

[Ralf Puchner]

this crash-data requires deeper analyze, please open a call within the HP support center.

Analyzing and working on complex problems here in the forum is not possible.

[Johan Brusche]

Panic seems to be triggered by LAT streams code. If you do not use LAT, you should set LAT_SETUP=0 into /etc/rc.config(.common) and unload lat from the running system.

Anyway you should log a service request and make sure this incident is "elevated" to Tru64 engineering.

__Johan__

[Ralf Puchner]

correct, it seems the streams module within lat will be involved, but without an escalation this will not be fixed and maybe some "risk" after disabling lat will exists.