Has anyone else experienced a similar problem?

This abend is a result of the z/VM Control Program (CP) being unable to
acquire the IPTE Interlock in a reasonable time. The IPTE Interlock can be
acquired by both CP and by Guests (running under SIE). The approach CP uses
to acquire the lock is to try once every 50-100 microseconds for up to
10,000 attempts. If not acquired at the end of that, the LCK009 abend is
taken. We have validated that the abend was not a result of the Guest
holding the interlock indefinitely, but rather the rare timing that
different vCPUs of the guest in question would acquire it in between CP's
attempts to acquire it. The following are factors contributing to the
possibility of resulting in the abend:
1. The rate at which the Guest is doing IPTEs. The higher the rate, the
greater the possibility of the abend.
2. The number of virtual CPUs configured for a given virtual machine. Since
IPTE can be issued by any of the virtual CPUs, the greater the number the
greater the possibility.
3. The number of logical processors. Just having multiple virtual CPUs is
not enough, but being able to run them concurrently increases the
possibility.
4. Load on the system by other guests. The lighter the load overall, the
more likely multiple vCPUs of the same guest could be dispatched at once
and the greater the possibility.
5. Frequency at which CP attempts to acquire the lock. Currently 50-100
microseconds. The less frequently CP attempts to acquire the lock, the more
likely CP misses an opportunity where the lock was available.

-- 
George Henke
(C) 845 401 5614

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to