Abstract:

My research interests/efforts that relate to this conference are associated with issues of multi-threading and multi-processing. This both in OS implementations and in user applications.

In my previous research position, I was a lead designer and developer of the Mach-US multi-server operating system at CMU under Rick Rashid and Brian Bershad. In that system, multiple system servers supplied specific system services (filesys, network, ttyio,...) and a UNIX emulation library was loaded into each process to support syscall semantics using the system servers. This work was described at a USENIX conference [Stevenson&Julin]. (See: https://www.cs.cmu.edu/afs/cs/project/mach/public/www/projects/mach_us.html)

My current related research is to develop software support system for rapidly and safely installing fine grained locking into existing C++ code. The target code is currently single threaded or "Big Lock" PThread distributed applications. This fine grained locking would occur while limiting the impact on straight-line execution speed. The threading support system is to run in both UNIX and NT based threading environments.

With my UNIX OS background with an emphasis on distributed multi-threaded systems, and current work transition issues for optimal fine grain threading and distribution in the NT environment, I am interested in this workshop.

Reference:

[Stevenson&Julin] J. Mark Stevenson, Daniel P. Julin.
Mach-US: UNIX on Generic OS Object Servers.
USENIX 1995 Technical Conference Proceedings
January 1995, pp. 119-130
J. Mark Stevenson
Sr. Systems Designer
CLARITECH Corp.
5201 Fifth Ave.
Pittsburgh, PA, 15232-2124

Phone: 412.621.0570
email: jmark@claritech.com