[Corelinux-develop] Questions, Questions, and Questions.

[Hans D. <dul...@eg...>]

Frank encouraged me to keep asking question and now I do :-).
I am trying to understand the SemaphoreCommon class and
would like to ask the following questions:

1. Why CSA is needed? Is it used for obtaining the key of semget()?
   However, I also find SemaphoreGroup::tmpDir ("/tmp") and 
   SemaphoreGroup::tmpName (/tmp/clsgtemp.\d+) used by ftok to generate
   the key, but I could not find the connection between CSA and ftok.

2. SemaphoreCommon defines members setLock and setUnLock which eventually
   calls the Dijkstra's P()/down() and V()/up() operation.
   To implement EventSemaphore, I am planning to use the "zero()"
operation,
   that is semop with sembuf.sem_op set to 0.
   Will it be apropriate? If, yes, should a new member "setZero()" be added
   to SemaphoreCommon?

I also have a couple of comments:

1. SemaphoreCommon::setLock and SemaphoreCommon::setUnLock call
   semop and pass 'ON' as the third actual parameter.
   Semantically, 'ON' does not make sense to the call because
   the third formal parameter of semop is 'number of semaphore operations'.

2. I think the constant ON and OFF in SemaphoreCommon are overloaded
   for too many different context, in some cases sacrificing its semantics.
   For instance, the semantic of "-ON" [minus ON] is not clear.

-- 
Hans Dulimarta, Ph.D.                  dul...@co...
P: 517-432-7589             http://www.egr.msu.edu/~dulimart   
F: 760-281-7691             http://corelinux.sourceforge.net
Elec. & Comp. Engg., Mich. State Univ., E. Lansing, MI 48824