On 11.02.23 17:27, Martin Erik Werner wrote:
Regenerate message queue directives from the rtems-central spec in
order
to include the detailed note about non-atomicity of broadcast.
This file is selectively added, other changes in generated files are
omitted.
Update #4804.
---
c-user/message/directives.rst | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/c-user/message/directives.rst b/c-
user/message/directives.rst
index a13e4c7..ac53ec4 100644
--- a/c-user/message/directives.rst
+++ b/c-user/message/directives.rst
@@ -758,6 +758,14 @@ The execution time of this directive is directly
related to the number of tasks
waiting on the message queue, although it is more efficient than the
equivalent
number of invocations of :ref:`InterfaceRtemsMessageQueueSend`.
+:ref:`InterfaceRtemsMessageQueueBroadcast` unblocks :term:`receivers
+<receiver>` in a non-atomic way. Meaning, it will not only
:term:`unblock`
+those :term:`receivers <receiver>` it finds waiting at the queue when
+:ref:`InterfaceRtemsMessageQueueBroadcast` is invoked but also any new
+:term:`receivers <receiver>` which start waiting for messages after
+:ref:`InterfaceRtemsMessageQueueBroadcast` is invoked and before it
returns.
+This may lead to infinite unblocking loops.
+
.. rubric:: CONSTRAINTS:
The following constraints apply to this directive:
Thanks for the updated documentation. It let me think about the current
implementation since the current behaviour is a bit scary.
In uniprocessor configurations, the current behaviour could be fixed by
disabling thread dispatching for the entire broadcast. This should be
acceptable. The disabled thread dispatching prevents that new waiting
threads show up in the queue.
In SMP configurations, this is not sufficient (new waiting threads can
show up on other processors). We could add a 32-bit generation number to
the message queue. When a waiting thread is enqueued, it increments and
gets the generation number. In the broadcast we dequeue only threads
with a generation number less than the one at broadcast start (modulo
operation).
--
embedded brains GmbH
Herr Sebastian HUBER
Dornierstr. 4
82178 Puchheim
Germany
email: sebastian.hu...@embedded-brains.de
phone: +49-89-18 94 741 - 16
fax: +49-89-18 94 741 - 08
Registergericht: Amtsgericht München
Registernummer: HRB 157899
Vertretungsberechtigte Geschäftsführer: Peter Rasmussen, Thomas Dörfler
Unsere Datenschutzerklärung finden Sie hier:
https://embedded-brains.de/datenschutzerklaerung/
_______________________________________________
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
