Messages in this thread | | | From | "Sven-Thorsten Dietrich" <> | Subject | [ANNOUNCE] Robust Mutex Special Interest Group | Date | Tue, 15 Mar 2005 11:03:26 -0800 |
| |
Announcing the OSDL Robust Mutex Special Interest Group
http://developer.osdl.org/dev/mutexes/
PURPOSE
The general mission of the mutexes SIG is to establish a forum centralizing discussion pertaining to mutex-specific planning and development in Linux.
The mutexes SIG is needed to promote forward-looking discussion between developers, users and maintainers.
The mutexes SIG can reduce redundant development efforts, and could provision a base for exploring successful integration of mutex-related technology into the Linux kernel.
BACKGROUND
Mutex implementations, like the applications that use them are plentiful. Mutex implementations are found in Linux libraries, in Linux patches, as well as the Linux Kernel.
For each mutex implementation, functionality is specialized. Almost all multi-threaded applications depend on mutexes. College textbooks cite examples that work on Linux, but use mutex libraries dating back to older operating systems.
Amazing applications, some as mobile as autonomous Robots, run Linux, but many of them augment the Linux kernel with mutex enhancements that comply with given design criteria.
Advanced mutex functionality is minimal in the Linux kernel. With the advent of the real-time preemption concept, the mutex conundrum does not relent.
The real-time enhancements exist only as a kernel patch, as do many of the mutex implementations. But the real-time mutex implementation is not compatible with any other advanced mutex implementation.
INITIAL OBJECTIVES
The mutexes SIG will initially research the integration of two partially overlapped mutex implementations:
1. http://people.redhat.com/~mingo/realtime-preempt/ 2. http://developer.osdl.org/dev/robustmutexes/
The realtime-preempt mutex in Ingo Molnar's patch implements a generic (portable) priority-inheriting kernel mutex with deadlock detection support.
The Robust Mutex project (aka "Fusyn") also provides a generic (portable) priority-inheriting kernel / userspace mutex with deadlock detection.
Both projects have specific, but also complementary, and overlapping objectives. Both projects implement several possibly common subsystems, but the implementations are functionally conflicting. Binary compatibility concerns emerge.
It is conceivable that system-designers using Linux for advanced applications may wish to utilize the Robust (User-space) Mutex patches at the same time as the real-time kernel enhancements. It is inefficient at best, to have two independent PI mechanisms operating in the kernel. If RT and Fusyn are to coexist (they are birds of a feather after all), at least the priority inhertance functionality, as well as bounded list operations would need to be reconciled between the two implementations.
Please refer to the OSDL site:
http://developer.osdl.org/dev/mutexes/
for more information.
First conference call is Wed (3/16) at 9AM (pacific)
Thank You
Sven
*********************************** Sven-Thorsten Dietrich Real-Time Software Architect MontaVista Software, Inc. 1237 East Arques Ave. Sunnyvale, CA 94085
Phone: 408.992.4515 Fax: 408.328.9204 Email: sdietrich@mvista.com
http://www.mvista.com Platform To Innovate ***********************************
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |