Quantitative Analysis
Parallel Processing
Numerical Analysis
C++ Multithreading
Python for Excel
Python Utilities

I. Installation.
II. Threading primitives.
III. NonBlockingQueue.
1. NonBlockingQueue design.
2. Simplest example with NonBlockingQueue.
3. NonBlockingQueue prototypes.
A. NonBlockingQueue data fields.
B. NonBlockingQueue::push member function.
C. NonBlockingQueue::pop member function.
D. NonBlockingQueue::Element.
E. NonBlockingQueue::Node.
4. Python-based acceptance test of NonBlockingQueue.
IV. ThreadPool.
V. ThreadMaster.
VI. OTS Scheduler.
VII. Bibliography
Downloads. Index. Contents.

NonBlockingQueue::push member function.

he following is the implementation of the NonBlockingQueue::push member function.

First, line 04, we get a pointer to an Element "el" via get() that reads and increments theSwitch. Then, line 10, we attempt to gain ownership of el's mutex. If successful, line 11, then we gain ownership of Node's mutex, line 13, and do the insertion, line 16. If gaining ownership at line 10 fails then we proceed to the next Element on the line 19. The line 21 is added to insure that any accidental call pattern that may lead to repeated failure would be broken. It does exactly what it says: delays for a small random amount of time. If everything fails, which should be extremely rare, an exception is thrown from line 23.

01\template <class Data>

02\ void NonBlockingQueue<Data>::push(

volatile typename NonBlockingQueue<Data>::Node& node

) volatile

03\ {

04\ volatile Element* el=get();

05\ for( int j=0; j<100; ++j )

06\ {

07\ int L=circleLength();

08\ for( int i=0; i<L; ++i )

09\ {

10\ ElementTryGuard el_(el);

11\ if( el_ )

12\ {

13\ NodeWriterGuard node_(node);

14\ if( node_->isInQueue() )

15\ return;

16\ Node::doInsert(el_,node_);

17\ return;

18\ }

19\ el=el->next();

20\ }

21\ SmallRandomDelay();

22\ }

23\ throw UnableToPush();

24\ }

Downloads. Index. Contents.

Copyright 2007