Re: [ECOS] clarification need for "tm_baic" test case

["Gary D. Thomas" <gary dot thomas at mind dot be>]

On Wed, 2003-05-21 at 22:51, radhakrishnan R wrote:
> Hello,
>     Iam trying to run the "tm_basic" test case
> available in eCos "kernel" package for my target.
> 
> problem summary:
>     "
>     During the kernel primitives timing measurement,
> if the timer interrupt occurs, the timing for ISR is
> also taken in to account. Is this is expected
> implementation or the ISR timer need to be excluded in
> measurement?.
>     
>     To measure the timing for the kernel primitives a
> the ticks maintained in timer counter is used.
>     The timer counter works like this.
>     "The timer is programmed to interrupt the
> Processor every 1 milli second. The timer is
> initialized to count and increment every 1 Micro
> second.
>     So every 1 microsecond my timer counter will
> increment and when it reaches 1 milisecond it will
> interrupt the Processor."
>     The question is "when the timer interrupt occurs
> because it has reached the limit, the timer ISR is
> executed". Because of the ISR execution the time taken
> for the ISR is also included in the Kernel primitives
> timing measurement. Is the time taken to execute ISR
> also be included in measurement??.
> 

What actually gets measured is the latency - the time from
the interrupt occurring until the ISR routine actually runs.
Look at the code - the measurement happens early in the ISR.
There is no measurement of the ISR per-se, but there can be
an additional measurement of the ISR -> DSR latency.

Again, the code is all out there for you to read.
  kernel/current/src/common/clock.cxx 

-- 
Gary D. Thomas <gary.thomas@mind.be>


-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss