International Business Machines Corporation

Jeffrey Scott Peek

Luke Matthew Browning

Bruce Gerard Mealey

Randal Craig Swanberg

Mark D Rogers

Michael Stephen Williams

John Thomas O'Quinn, Ii

Olivier Alain Cremel

A method and system for establishing a priority order of threads executed within a multiprocessor data processing system are disclosed. The priority order of threads is maintained on a global queue within the data processing system and the queue is ordered by selecting a most favored runnable thread from among either a current thread, a favored thread, or an affinity thread. Once the thread is chosen, be it the current thread, the most favored thread, or the affinity thread, the chosen thread is removed from a run queue and then is run on the data processing system on a selected processor. An affinity thread pointer is updated to the thread being run to select the next affinity thread or the thread having the greatest affinity.

Method and system for scheduling threads within a multiprocessor data processing system using an affinity scheduler

A method and system are disclosed for handling exceptions generated by a particular processor among a number of processors within a multiprocessor data processing system. The data processing system includes a global queue from which threads are dispatched in a priority order for execution by the processors and a number of local dispatch flags, which are each associated with one of the processors. In response to an occurrence of an exception during execution of a particular thread by a particular processor, a state of the particular thread, which is at least partially defined by volatile data within the particular processor, is saved. A selected exception handler associated with the exception is then executed. Next, a determination is made whether resumption of the particular thread depends upon an occurrence of a specified event. In response to determining that resumption of the particular thread depends upon an occurrence of the specified event, the local dispatch flag associated with the particular processor is set. Thereafter, a determination is made whether the particular thread can be interrupted by another thread. In response to the local dispatch flag associated with the particular processor not being set or in response to determining that the particular thread cannot be interrupted by another thread, the particular thread is resumed. In response to the local dispatch flag associated with the particular processor being set and a determination that the particular thread can be interrupted by another thread, a thread is selected from the global queue for execution by the particular processor.

Method and system for scheduling threads and handling exceptions within

A method and apparatus for enabling AIX device driver (DD) created for a uniprocessor (UP) system to run unchanged on a symmetrical multiprocessor system (SMP). Device drivers are processed by a funnelling mechanism so that UP device drivers always runs on a 'Master' processor in a multi-processor system. New device drivers written for the SMP system are permitted to bypass the funneling mechanism and proceed directly to execution on any available processor in the SMP system. Device registration services are provided which examine DD flags looking for a new unique flag indicating that the device driver is SMP enabled. If the flag is not present, then emulating a uniprocessor environment for execution of that device driver.

Method and apparatus for device driver funnelling

A computer-implemented method, computer system, and memory for performing a fork operation of a parent process is provided. The parent process includes one or more threads, each having a kernel stack. The method includes the steps of locating the kernel stacks in a first memory location (e.g., segment 1) if the parent process is multi-threaded, in response to an initiation of a fork by a calling thread, switching the kernel stack of the calling thread to a second memory location if the parent process is multi-threaded, and copying the second memory location (e.g., segment 2) to create a child process.

Efficient forking of a process

A method and system for using a single lock data structure for executing either development or non-development lock primitives contained within a kernel. The kernel includes a mode indication flag, which can be set by the user, for indicating whether the kernel is to operate in a development or a non-development mode. During the execution of the kernel, the mode indication flag is examined and the appropriate set of lock primitives is overlayed. During execution of the kernel in development mode, the single lock data structure is received in the kernel, and a development lock data structure is allocated. The data from the single lock data structure is copied to the development lock data structure, and the lock data structure is overloaded (redefined) as a pointer to the physical address of the development lock data structure. Finally, the lock data structure is initialized to point to the physical address of the development lock data structure. Thereby allowing the user to have full access to the development lock primitives using the redefined lock data structure.

Method and system for lock instrumentation in a data processing system

Libraries for use in a multithreaded computer environment which must be thread-safe and cannot be recoded are identified. Such libraries, after identification, are repackaged. Export routines for such libraries will thereby acquire a front end which effects several steps. First a write-exclusive lock for a particular called function of one of the libraries is generically acquired. A call is then made to the underlying function whose name has been remapped. Upon return from the real routine, the write-exclusive lock is unlocked and a return to the user is executed. The system and method ensure correct functioning and integrity of the identified library functions accessible by multiple threads, as well as providing for creation of multithread-safe shared libraries without necessitating extensive library source modifications.

Growing community of inventors

Austin, TX, United States of America

Jeffrey Scott Peek