Abstract: A file validator technique for identifying a test specification to be used for testing validity of a file element associated with a file is provided. The test specification is a member in one of a set of test specifications embedded in the file validator and a set of test specifications not embedded in the file validator. According to the technique, a file element is selected, a pointer associated with the selected file element is selected, and a pointer is used to select the test specification associated with the selected file element, the selected test specification being selectable from one of the set of test specifications embedded in the file validator and the set of test specifications not embedded in the file validator.

Abstract: A method for determining if a measurement of an elapsed time for an execution of a software routine in a computer system is valid. A clock skew is used between the clocks of two processors such that the size of the clock skew is greater than the maximum possible elapsed time for the execution of the software routine. The software routine is executed with a clock value recorded before (start time) and after (end time) execution. An elapsed time is calculated as a difference between the start time and the end time. Whether the elapsed time is valid is determined by checking for a positive value of the elapsed time and comparing the value of the elapsed time with the clock skew.

Abstract: The present invention provides a method, system, computer program product and service, by which a unit of work manager can reliably start a follow-on unit of work after a first unit of work, in the event the first unit of work backs out without requiring to interlock the execution of the first unit of work and follow on unit of work. If the first unit of work completes before the follow-on is started information written to non-volatile memory, written as part of the first unit of work, is used on recovery to start is follow-on unit of work if no other information exists in the volatile storage which indicates the follow-on unit of work has started.

Abstract: A sampling mechanism is disclosed in which software can specify a property or properties which characterize samples of interest. For example, if the software is interested in cache behavior, the software can specify that information for memory operations, or only information for memory instructions which miss in one or more caches, be reported. The sampling mechanism may specify many such properties and events (properties and events may vary from processor to processor, and may also depend on which properties or events are considered useful for performance analysis).