Re: About task-safeness

[Georg Bauhaus <rm.dash-bauhaus@futureapps.de>]

On 03.02.11 09:53, Niklas Holsti wrote:
> Randy Brukardt wrote:
>> "mockturtle" <framefritti@gmail.com> wrote in message
>> news:06ecb5ab-a9e5-4a5d-9370-6bbe137d3693@glegroupsg2000goo.googlegroups.com...
>> ...
>>> (For example, a package defining some type of object could keep the number
>>> of allocated
>>> objects, so it can give to each object a unique ID.)  If such a package was
>>> used  by two
>>> different tasks, and the counter was not protected, obscure bugs can arise.
>>> This type of
>>> structure maybe is not very recommended, but it happens... :-(
>>
>> Note that "protection" may simply be declaring the object Atomic. Presuming
>> the compiler supports that, there isn't a problem with multiple tasks
>> accessing the same counter.
> 
> Randy, could you be more explicit about your suggested use of Atomic? As I
> understand it, even if a counter variable N is Atomic, two tasks concurrently
> executing an assignment of the form
> 
>    N := N + 1;
> 
> can interleave their actions so that N is increased by only 1, not by 2 as
> intended.

When I first read about atomic counting, I thought so, too.  But then,
I noticed a new(?) Implementation Advice in the LRM that says, IIRC,
that compilers should make atomic processor ops available, including
decrement.  A neighboring paragraph is about intrinsic operations.