Sun Microsystems, Inc.

Oracle America, Inc.

Arista Networks, Inc.

Bart Smaalders

Darrin Paul Johnson

Eric Christopher Saxe

Jonathan Chew

Andrew G Tucker

Kevin J Clarke

Bryan M Cantrill

Nicolai Kosche

Madhusudhan Talluri

Andrei Dorofeev

Bruce W Curtis

Roman Olegovich Chertov

Tim P Marsland

Sean McGrath

Damien Farnham

Miguel Isenberg

Joseph J Bonasera

Sundeep R Panicker

Krishna Kondaka

Dave Singleton

Robert L Berube

Techniques for persistent crash logging. The techniques may include making, by a monitor, a first determination that an event information file exists in a volatile storage device location in a volatile storage device. The first determination may be made after an agent event occurs. The techniques may also include performing, by the monitor, an action on a portion of the event information file to obtain a uniquely identifiable filename portion and writing, by the monitor, an event information file copy to a non-volatile storage location in a non-volatile storage device. The techniques may also include, after writing the event information file copy to the non-volatile storage location is complete, renaming, by the monitor, the event information file copy to a uniquely identifiable filename comprising the uniquely identifiable filename portion.

Persistent crash logging

In general the invention relates to a system. The system includes processors each having a processing state. The system further includes a dispatcher operatively connected to the plurality of processors and configured to: receive a first thread to dispatch, select one of the processors to dispatch the thread to based on the processing state the processors and a power management policy, and dispatch the thread to the selected one of the plurality of processors.

Method and system for power-management aware dispatcher

A device, system, and method are directed towards managing threads in a computer system with one or more processing units, each processing unit having a corresponding hardware resource. Threads are characterized based on their use or requirements for access to the hardware resource. The threads are distributed among the processing units in a configuration that leaves at least one processing unit with threads that have an aggregate zero or low usage of the hardware resource. Power may be reduced or turned off to the instances of the hardware resource that have zero or low usage. Distribution may be based on one or more of a number of specifications or factors, such as user power management specifications, power usage, performance, and other factors.

Hardware utilization-aware thread management in multithreaded computer systems

Techniques for isolating software regressions are provided. In one approach, QTrace is used to generate DTrace scripts to gather timing information of one or more portions of source code during a test of particular version of a program. Based on results of one or more tests of multiple versions of a program, a regression is identified with respect to a particular portion of source code. It is determined that the particular portion of source code was changed during an intermediate version of the program (i.e., between two tests). The intermediate version and/or the particular portion of source code are identified as being a potential source of the regression.

Determining causes of software regressions based on regression and delta information

A device, system, and method are directed towards managing power consumption in a computer system with one or more processing units, each processing unit executing one or more threads. Threads are characterized based on a cycles per instruction (CPI) characteristic of the thread. A clock frequency of each processing unit may be configured based on the CPI of each thread assigned to the processing unit. In a system wherein higher clock frequencies consume greater amounts of power, the CPI may be used to determine a desirable clock frequency. The CPI of each thread may also be used to assign threads to each processing unit, so that threads having similar characteristics are grouped together. Techniques for assigning threads and configuring processor frequency may be combined to affect performance and power consumption. Various specifications or factors may also be considered when scheduling threads or determining processor frequencies.

Frequency scaling of processing unit based on aggregate thread CPI metric

One embodiment of the present invention provides a system that facilitates buffering data at a kernel in a computer system, wherein the data is buffered based on the structure of a message contained in the data. The system operates by receiving data at a computer system from an external source. Next, the system buffers the data at a kernel on the computer system. As the system buffers the data, the system also determines if the buffered data constitutes a complete message as defined by a communication protocol. If so, the system forwards the buffered data to an application on the computer system.

Method and apparatus for performing syntax-directed application layer data buffering

An abstraction layer is comprised in the operating system that represents the particulars of the PPMs. The abstractions in the abstraction layer are differentiated from one another by parameters representing the characteristics of the PPMs. The dispatcher uses the abstraction to balance processing loads when assigning execution threads to the PPMs. The assigning of the execution threads and the balancing of the processing loads is performed while taking account of the characteristics of the PPMs, such as shared resources and clock speed.

Processor specific dispatching in a heterogeneous configuration

Methods and apparatus, including computer program products, implementing and using techniques for alerting one or more computer software application threads waiting to retrieve events from an event port. An alert event generated by a computer software application is received at the event port. A state of the event port is changed to an alert state, if the event port is not already in an alert state, in response to receiving the alert event. One or more of the computer software application threads is notified about the alert state of the event port.

Queue alerts

A method and system of managing resource allocation in a non-uniform resource access computer system is disclosed. A method comprises determining access costs between resources in a computer system having non-uniform access costs between the resources. The method also includes constructing a hierarchical data structure comprising the access costs. The hierarchical data structure is traversed to manage a set of the resources.

Method and system for managing resource allocation in non-uniform resource access computer systems

In a computer system, a descriptive memory allocation system is described having a memory policy allocation module for setting memory allocation policies by an operating system in response to descriptive resource use requirements provided by an application requesting access to a specified address range in memory. The descriptive memory allocation system includes a descriptive resource allocator that uses descriptive memory use advice provided by an application to decide how to allocate memory to the application. The descriptive resource allocator includes memory allocation policies that may be set by the operating system after the operating system has determined the appropriate allocation scheme to implement based on an allocation advice provided by a requesting after the application. The application in providing its descriptive memory use information does not specify a specific allocation policy the operating system should use to allocate memory to it.

System and method of descriptively specifying memory placement in a computer system

Method for enabling the dynamic modification of cluster configurations, and apparatus including software to perform the method. To enable this dynamic modification, cluster configuration data is stored as a table in a cluster configuration repository that is accessible from all nodes in the cluster. Accordingly, the present invention enables the modification of the cluster configuration from any node in the cluster dynamically. When a reconfiguration command is given, the configuration table is changed and all the nodes in the cluster are notified of the changed configuration in parallel. Following the notification by the nodes of the changed cluster configuration, the changes to the cluster are implemented dynamically as specified by the command.

Method to dynamically change cluster or distributed system configuration

In a multiprocessing computer system, a schedulable process entity (such as a UNIX process, a Solaris lightweight process, or a Windows NT thread) sets a memory flag (sc.sub.-- nopreempt) before acquiring a shared resource. This flag tells the operating system that the process entity should not be preempted. When it is time for the process entity to be preempted, but sc.sub.-- nopreempt is set, the operating system sets a flag (sc.sub.-- yield) to tell the process entity that the entity should surrender the CPU when the entity releases the shared resource. However, the entity is not preempted but continues to run. When the entity releases the shared resource, the entity checks the sc.sub.-- yield flag. If the flag is set, the entity makes an OS call to surrender the CPU.

Method and apparatus for execution and preemption control of computer

A method and system for reordering sections of a computer program are disclosed. The computer program is executed during an experimental execution period. During the execution period, an access record is made. The access record identifies when each section of the computer program is accessed. After the execution period, the sections are reordered based on when the sections were accessed during the execution period. The sections may be code sections, data sections, or a sections that contain a combinatin of code and data (e.g. objects). According to one disclosed technique, the execution period is divided into time intervals. Each time interval corresponds to the time period in which a discrete high level operation of the computer program was performed. A data matrix is created where each row corresponds to a section, and each column corresponds to a time interval. Values are stored in each cell of the data matrix based on whether the corresponding section was accessed during the corresponding time interval. The sections are then reordered based on the values in the rows to which they correspond.

Growing community of inventors

Menlo Park, CA, United States of America

Bart Smaalders