We have Itanium based SD16. One of the npar configured with one cell. On this npar already installed HP-UX 11.23. HPUX boots via SAN from EVA5000 (active-passive). We use secure path software for multi-passing.
And now we add one more cell and get panic while boot: ------------------- THE SPINLOCK CODE HAS NOT FAILED! You are seeing this message because some kernel code has attempted to yield the processor holding spinlocks. Please look for the code that did not drop the spinlock!
System Panic:
panic: spinlock: locker forgot to unlock Stack Trace: IP Function Name 0xe000000001607060 spinlocks_held_leaving_processor+0xb0 0xe000000000740540 exit+0x120held_leaving_processor+0xb0 0xe00000000071fd90 execve+0x1210d_leaving_processor+0xb0 0xe00000000069de70 syscall+0x920d_leaving_processor+0xb0 End of Stack Trace ----------------------------- That means this panic? How to solve this problem?
You did this with the system up, or did you bring down the superdome complex?
Please provide more information on what you are doing.
With I/O cards, you need to run olrad before pulling them and replacing them.
My assumption is to add a cell to an npar the npar needs to be down. Your message does not make that clear.
SEP
Steven E Protter Owner of ISN Corporation http://isnamerica.com http://hpuxconsulting.com Sponsor: http://hpux.ws Twitter: http://twitter.com/hpuxlinux Founder http://newdatacloud.com
Assuming that you rebooted the nPar for reconfiguration -- you're going to have to engage support to solve the problem.
It would need to be determined:
+ If the nPartition correctly added the cell or is in an incorrect state (causing O/S issues).
or
+ If the additional cpus triggered an O/S issue instead (bug in the kernel) which would almost certainly entail finding out what lock was held on exit, and how it was acquired.
(Is that also really what the console showed on output? Because that stack trace shouldn't have picked up the "d_leaving_processor+0xb0" on the subsequent lines -- looks rather like the string buffer wasn't terminating with the '\0' character as it should... may be related).
Neither of which is likely something folks can walk you through/determine through forum posts.
I'd expect Support to start with full partstatus output of the pertinent partition, Firmware versions/levels and then to attempt to acquire the dump [if this is in a panic/boot loop as it sounds, that would likely involve removing the cell first -- but I wouldn't do that until they asked you to so they can understand the nPar state causing the issue].
