International Business Machines Corporation

Richard Karl Kirkman

Patrick Joseph McCarthy

Don Darrell Reed

Lynn Keat Chung

Henry Joseph May

Kenneth Charles Vossen

Wade Byron Ouren

Mark Robert Funk

Christopher Francois

Michael Joseph Corrigan

Paul LuVerne Godtland

Scott Alan Plaetzer

Kenneth Roger Allen

Angela Mae Newton

Chistopher Francois

Naresh Nayar

William Joseph Armstrong

Kevin Jack Erickson

Jay Paul Kurtz

Brian Eldridge Clark

Michael Brian Brutman

Chris Francois

Patrick James Christenson

Donald Arthur Morrison

George David Timms, Jr

Ryan Harvey Bishop

Dennis A Towne

Method for measuring the performance of a multi-nodal computer system. In one embodiment, a number of system related events may be instrumented. When an instrumented event occurs, instrumentation routines may be invoked to record the state of system variables related to the event. In particular, events such as thread creation, dispatch, and nodal events such as events related to verifying or changing workload distribution in a multi-nodal system, among others, may be instrumented to record the state of variables as these events occur in a running multi-nodal computer system.

Analysis of nodal affinity behavior

Disclosed is an apparatus, method, and program product for identifying and grouping threads that have interdependent data access needs. The preferred embodiment of the present invention utilizes two different constructs to accomplish this grouping. A Memory Affinity Group (MAG) is disclosed. The MAG construct enables multiple threads to be associated with the same node without any foreknowledge of which threads will be involved in the association, and without any control over the particular node with which they are associated. A Logical Node construct is also disclosed. The Logical Node construct enables multiple threads to be associated with the same specified node without any foreknowledge of which threads will be involved in the association. While logical nodes do not explicitly identify the underlying physical nodes comprising the system, they provide a means of associating particular threads with the same node and other threads with other node(s).

Reducing remote memory accesses to shared data in a multi-nodal computer system

An apparatus, program product and method in which application program-specified resource allocation and allocation strength preferences are used to allocate hardware resources from a computer for that application program. The resource allocation preference is used to allocate hardware resources for the application program, while the allocation strength preference is used to determine whether to allocate alternate hardware resources when the resource allocation preference specified by the application program cannot be met.

Resource allocation in a NUMA architecture based on application specified resource and strength preferences for processor and memory resources

Resource allocation in a NUMA architecture based on separate application specified resource and strength preferences for processor and memory resources

An operating system kernel includes an attach mechanism and a detach mechanism. In addition, processes are tagged with an access attribute identifying the process as either a client process or a server process. Based on the access attribute, the operating system kernel lays out the process local address space differently depending on whether the process is a client process or a server process. A server process can 'attach' to a client process and reference all of the client process' local storage as though it were its own. The server process continues to reference its own process local storage, but in addition, it can reference the other storage, using the client process' local addresses. When access to the other storage is no longer needed, the server process can 'detach' from the client process. Once detached, the other storage can no longer be referenced.

Method, apparatus, and computer program product for accessing process local storage of another process

Methods, systems, and media for managing dynamic memory are disclosed. Embodiments may disclose identifying nodes with having memory for dynamic storage, and reserving a portion of the memory from the identified nodes for a heap pool. After generating a heap pool, embodiments may allocate dynamic storage from the heap pool to tasks received that are associated with one of the identified nodes. More specifically, embodiments identify the node or home node associated with the task, the amount of dynamic storage requested by the task, and create a heap object in the node associated with the task to provide the requested dynamic storage. Some embodiments involve de-allocating the dynamic storage assigned to the task upon receipt of an indication that the task is complete and the dynamic storage is no longer needed for the task. Several of such embodiments return the de-allocated dynamic storage to the heap pool for reuse.

Methods, systems, and media for managing dynamic storage

Mechanism for reducing remote memory accesses to shared data in a multi-nodal computer system

Disclosed is an apparatus, method, and program product that enables distribution of operating system resources on a nodal basis in the same proportions as the expected system workload. The preferred embodiment of the present invention accomplishes this by assigning various types of weights to each node to represent their proportion of the overall balance within the system. Target Weights represent the desired distribution of the workload based on the existing proportions of processor and memory resources on each node. The actual workload balance on the system is represented by Current Weights, which the operating system strives to keep as close to the Target Weights as possible, on an ongoing basis. When the system is started, operating system services distribute their resources nodally in the same proportions as the Target Weights, and can request to be notified if the Target Weights ever change. If processors and/or memory are subsequently added or removed, new Target Weights are calculated at that time, and all services which requested notification are notified so they can redistribute their resources according to the new Target Weights or a stepwise refinement thereof.

Mechanism for enabling the distribution of operating system resources in a multi-node computer system

Disclosed is an apparatus, method, and program product for observing the nodal workload balance of the system on an ongoing basis, and for dynamically changing the preferred nodes of existing threads in order to improve nodal balance. Workload balance is ascertained on a nodal basis and then thread-based workload information is collected. If the detected imbalance persists, the thread-based information is used to change the assignment of preferred nodes to threads to improve nodal workload balance.

Mechanism for dynamic workload rebalancing in a multi-nodal computer system

A method, apparatus, system, and signal-bearing medium that in an embodiment map a pointer to a target indirectly via a pointer map. In an embodiment, the pointer map is stored in a reserved area and an entry for the pointer in the pointer map includes a back-pointer containing an address of the pointer and a target-address containing the address of the target. When the pointer is accessed, the pointer is checked to ensure its contents point at the entry in the pointer map in the reserved area and that the back-pointer in the pointer map entry points back at the pointer. In this way, pointers are protected from being modified illegitimately.

Secure pointers

Diagnostic data, such as a time increment corresponding to how long a thread waits to access a shared resource, is stored within a predetermined location in a data structure, such as a hash bucket in a hash table. The location is preferably correlated to the resource such that a display of the diagnostic data may be tailored to reflect a user-specified relationship between the data and resource.

Dynamic diagnostic program for determining thread wait time

A linked list has an associated auxiliary data structure, containing external references to the linked list which are used by updating tasks. The auxiliary data structure is used to block subsets of the linked list from being altered by other tasks. Preferably, the linked list is doubly-linked, and list maintenance is accomplished using object-oriented programming techniques. The auxiliary data structure is an object, which comprises iterator objects, including inspector objects and mutator objects, blocker objects and view objects. Iterator objects are used by clients to access the linked list externally. An inspector object traverses the list without changing it; while a mutator alters the list, and may traverse it as well. Both inspectors and mutators have associated view objects, which contain the pointers to the actual list elements. Mutators also have associated blocker objects, which block access to selected list elements. In operation, a mutator object blocks off a portion of the linked list by altering pointers so that they point to the blocker object, the blocker object containing additional pointers by-passing the blocked list portion. Other inspectors may thereafter traverse the list through the by-passed portion. By blocking only a relatively small portion of a large linked list, multiple tasks may concurrently access different list elements and perform separate list updates.

Method and apparatus for maintaining a linked list

An apparatus and method provide simultaneous local and global addressing capabilities in a computer system. A global address space is defined that may be accessed by all processes. In addition, each process has a local address space that is local (and therefore available) only to that process. An address space processor is implemented in software to perform system functions that distinguish between local addresses and global addresses. In the preferred embodiments, the local address space has a size that is a multiple of the size of a segment of global address space. When the hardware indicates a page fault, the address space processor determines whether the address being translated is a local address or a global address. If the address is a local address, the address space processor uses a local directory to process the page fault. If the address is a global address, the address space processor uses a global directory to process the page fault. When the hardware indicates an addressing error because a computed address crosses a global segment boundary, the address space processor determines whether the address is a local address or a global address. If the address is a global address, the address space processor indicates an addressing error. If the address is a local address, the address space processor determines whether the address is within the process' local address space, and indicates an addressing error if the address is outside the process' local address space. Instructions are allowed to operate on both local and global addresses because the address space processor handles either type of address whenever software assistance is required, such as for servicing a page fault or checking a segment boundary crossing. In addition, the address space processor dynamically checks the addressing compatibility of called code before passing control to the called code.

Growing community of inventors

Rochester, MN, United States of America

Richard Karl Kirkman