Non-blocking message_queue_send

[Isaac Gutekunst]

Hi Joel,

That's exactly my use case. I never actually went and tested that they 
blocked, just anticipated this would be a problem later, as I was using 
the queue as a mechanism to limit memory allocations.

This is actually for some application data tracing/logging and I didn't 
want the log statements to hold up other tasks.

Thanks for the help!

Isaac

On 02/12/2016 11:10 AM, Joel Sherrill wrote:
>
>
> On Fri, Feb 12, 2016 at 9:40 AM, Isaac Gutekunst
> <isaac.gutekunst at vecna.com <mailto:isaac.gutekunst at vecna.com>> wrote:
>
>     Hi All,
>
>     I have an application where having a non-blocking
>     rtems_message_queue_send would be useful.
>
>
>     I will have multiple producers and a single consumer task wise, so
>     using rtems_message_queue_get_number_pending as method to determine
>     if the next send will succeed won't work.
>
>     Has anyone had a similar problem before?
>
>
>     I'm thinking I can wrap all calls to this message queue with a
>     counting semaphore and use that to arbitrate. Any other suggestions?
>
>
> rtems_message_queue_send() does not block at all. Either the message can
> be enqueued or you get a message queue full error.
>
> POSIX message queues can optionally block on the mq_send() but not
> Classic API message queues.
>
> Am I missing your use case? Or did you misinterpret something?
>
> --joel
>
>     Isaac
>     _______________________________________________
>     users mailing list
>     users at rtems.org <mailto:users at rtems.org>
>     http://lists.rtems.org/mailman/listinfo/users
>
>