Re: Ada delay

[cis.ohio-state.edu!news.sei.cmu.edu!jldh@ucbvax.Berkeley.EDU (Jorge Luis Diaz-Herrera)]

In article <1992Sep17.142643.21378@sei.cmu.edu>, firth@sei.cmu.edu (Robert Firt
h) writes:
|> In article <Buq4Mr.D7H@math.uwaterloo.ca> amichail@plg.uwaterloo.ca (Amir Mi
chail) writes:
|> 
|> >How is the Ada delay statement implemented?  I suppose it uses a hardware 
|> >timer coupled with a clock server, but this would mean that the time
|> >taken for a delay is unbounded (since we have to search an ordered list
|> >of tasks).
|> 
|> Either a list is ordered or you have to search it, but not both, surely,
|> else somebody has it in the wrong order.
|> ...

We implemented (see ref.) simple delays using an ARRAY [1..MaxTasks] of Time.
The task name serves as the array index; the array entry is the task release
time (release time = current time + delaytime.) The array is searched at every
context switch and all tasks whose release time is less than or equal to the
current time are inserted in the corresponding ready queue (priority).  This
is consistent with the LRM which requires delays to be "at least" the
duration specified. Linear search is used since the number of delayed tasks
will probably be small.  In addition to timing through tick counting, using
the timer interrupt trapped, a variation was developed using calls to the
system clock.  This technique while inherently more efficient, can be highly
inaccurate since in our RTK time is updated only upon entry to the dispatcher.

For more information refer to: "ARTK-M2:A Kernel for Ada Tasking Requirements:
an implementation and an Automatic Generator" J.L.Diaz-Herrera et al,
Software Practice and Experience, vol. 22(4) 317-348 (April 1992).