Monitoring for missing interrupts does not take place until the system is reini­
tialized, or until the class B user issues another SET MITIME command.
Detennining Time Interval Settings
Diagnostic Aids
System Messages
The class B user can determine the current missing interrupt handler time intervals
by issuing the following CP command: QUERY MITIME
The system issues:
The time interval setting for each device group in minutes and seconds • The response MITIME OFF An error message if the user specified an invalid parameter. • The response that the missing interrupt handler is not available if DMKDID is
not in the loadlist during system generation.
Missing interrupt handler support provides aids so that the system programmer can
determine the frequency and status of interrupts and also know when he has made
an error in using the support. Diagnostic aids available when using the missing
interrupt handler include: • System messages
Macro notes VM/SP system's error recording area
Trace table
Messages inform the system operator when a missing interrupt occurs and indicate
if the condition has been cleared or if the interrupt is still pending. Other messages
indicate that the module DMKDID is not in the loadlist or that the user specified an
invalid parameter on the QUERY or SET MITIME command. See VM / SP System
Messages and Codes for a complete discussion of messages that the missing inter­
rupt handler issues.
The system programmer can use message information to increase the availability of
the system. If a particular device class causes frequent interrupts even if the system
clears the condition, the system programmer may want to change the time interval.
Changing the time interval prevents the overhead of frequent timer interrupts, fre­
quent trips through the detector routine, and rescheduling of timer request queues. On the other hand, if the control program did not clear the condition, the messages
make the system programmer or system operator aware of the condition and one of
them can reset the hardware either physically or using CP commands.
Macro notes (MNOTES) inform the user that SYSMIH is not present in DMKSYS or that the user specified an invalid time value in the SYSMIH macro. The system
uses the default interval time values and informs the user.
Interruption Handling 21

System's Error Recording Area Trace. Table
Program Interrupt
Machine Check Interrupt
Whether or not CP succeeds in correcting a missing interrupt situation, it creates a
record of the event in the system's error recording area (LOGREC). CP also traces the simulated interrupt and records it as trace table entry X' 19'.
Refer to Figure 67 on page 507, "CP Trace Table Entries", for the format of the
entry. The system programmer uses the trace table to determine the events that
preceded a CP system failure.
Program interrupts can occur in two states. If the processor is in supervisor state,
the interrupt indicates a system failure in the CP nucleus and causes the system to
abnormally terminate. If the processor is in problem state, a virtual machine is exe­
cuting. CP takes control to perform any required paging operations to satisfy the
exception, or to simulate the instruction. The fault is transparent to the virtual
machine execution. Any other program interrupt is a result of the virtual machine
processing and is reflected to the machine for handling.
When a machine check occurs, the CP Recovery Management Support (RMS) gains control to save data associated with the failure for the Field Engineer. RMS analyzes the failure to determine the extent of damage.
Damage assessment results in one of the following actions being taken: • System termination (CP disabled wait state) • Attached processor disabled (system continues in uniprocessor mode) One processor of a multiprocessor configuration disabled (system continues in
uniprocessor mode) • One or more failing channels disabled (system continues in same mode as at
time of the error) • Selective virtual user termination
Selective virtual machine reset
Refreshing of damaged information with no effect on system configuration
Refreshing of damaged information with the defective storage page removed
from further system use
Error recording only for certain soft machine checks
The system operator is informed of all actions taken by the RMS routines. When a
machine check occurs during VM/SP startup (before the system is sufficiently ini­
tialized to permit RMS to operate successfully), the processor goes into a disabled
wait state and places a completion code of X'OOB' in the leftmost bytes of the cur­
rent PSW. 22 VM/SP System Programmer's Guide