Abstract: A multithread auto test method is disclosed for the test process of computer hardware. According to the exclusion relation among the unique IDs of the test items, a multithread executable logic is automatically generated. An appropriate parallel method is employed to find procedures for test items that do not have conflictions. Therefore, multithreads of test procedures are performed to increase the test efficiency and quality. The method includes the steps of: determining a unique ID of a test item; automatically generating a test logic table according to the exclusion relation among the unique IDs; and performing multithread test procedure according to the test logic given in the test logic table.

Abstract: The specification may disclose a computer system that may operate a portion of available memory as backup to a primary memory, and the computer system may be adapted to test the backup memory for memory errors at times other than execution of power-on self-test procedures.

Abstract: The disclosed embodiments relate to a system and method for message suppression. The system comprises a memory that is adapted to store a data element corresponding to a failure notification and to store a corresponding timestamp. Also included in the system is an interface adapted to receive the failure notification, create the corresponding timestamp, store the data element and the timestamp corresponding to the failure notification in the memory, send a message corresponding to the failure notification, suppress the sending of subsequent messages corresponding to the failure notification until a predetermined condition occurs, receive a notification to cancel the suppression of subsequent messages corresponding to the failure notification, and send subsequent messages corresponding to the failure notification in response to failure notifications received after the notification to cancel the suppression of subsequent messages.

Abstract: The present invention is directed to detecting leaks in computer memory. Memory leaks are detected by analyzing computer component memory usage records. The first derivative of the memory usage data is computed. If the area under a curve created by a plot of the first derivative of the memory usage data is greater than zero, the second derivative of the memory usage data is computed. If the area under a curve created by a plot of the second derivative of the memory usage data also is greater than or equal to zero, the memory usage data is analyzed to determine how many minima points are present in the memory usage data. If less than four memory usage data minima points are present, a memory leak exists. If four or more minima points are present in the memory usage data, the minima points are further processed to determine if a memory leak exists.

Abstract: A recovery process for embedded processors monitors other processes in the system. Each process may specify a recovery policy residing in nonvolatile electronic memory that preferably includes a recovery count, a recovery time, and a recovery action. If a process terminates unexpectedly, the recovery process determines whether the process had a corresponding recovery policy. If not, the recovery process does not recover the process. If the process has a corresponding recovery policy, the recovery process determines whether it can recover the process by examining the recovery count and recovery time specified in the recovery policy. If the process can be recovered, the recovery process performs the recovery action specified in the corresponding recovery policy. If the process cannot be recovered, the recovery process resets the system.