Ford, Bryan, Mike Hibler, and Jay Lepreau, "Notes on Thread Models in Mach 3.0," Department of Computer Science, University of Utah, Technical Report No. UUCS-93-012, Salt Lake City, Utah, April 1993.  Home/Search   Document Not in Database   Summary   Related Articles

....them, making it costly to remove the client thread from the client task and insert it into the server task. Doing so also opens up the server task to accidental or intentional assault by entities holding capabilities to the thread. Short of completely redesigning Mach s thread scheduling, as in [8], thread migration is not currently a practical solution. Furthermore, establishing a shared region of memory between two distinct tasks is fairly involved in Mach 3 [12] requiring either a) a common ancestor task, b) a common memory mapped file, or c) privileged access. Additionally, sharing a ....

....RPC call latency by putting servers into the kernel s address space, limiting the facility to those with kernel access and defeating some of the benefits of the microkernel architecture. This work led to the exploration of the use of thread migration to speed up directed control transfer in [8]. 4.3. Performance LRPC Mach3 is implemented on a Sequent multiprocessor containing 20 Intel i386 microprocessors. A single processor is used for all timing comparisons. The fundamental performance measurement is call latency the time required for a client to invoke and return from a server ....

Ford, Bryan, Mike Hibler, and Jay Lepreau, "Notes on Thread Models in Mach 3.0," Department of Computer Science, University of Utah, Technical Report No. UUCS-93-012, Salt Lake City, Utah, April 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC