Hello we are using the pSOS interface on top of the Mercury core. Under heavy stress conditions we see sporadically that messages are getting lost.
Attached you find a test application (multi_queue.c) that helped me to find the issue and verify that the attached patch actually solves the issue as well. To reproduce the issue I start the test application several times in parallel. The cpuload of the board will reach a value of 100%. Maybe it is not needed to start several application in parallel, but if I do so, I can reproduce the issue in a matter of seconds. As soon as one message is lost, the test stops itself (exit). After some time I observe that all applications have stopped ... What the test basically does is sending messages between a manager and 2 worker threads and verify, by using sequence numbers in the messages, that no message is lost. The test might look too complex but it is basically a simulation of a scenario that we perform in our real application. The patch is ported from an older version of Xenomai. I did not actually run the test with the latest version but by looking to the code I can see that the issue is still present in the latest version as well. Best regards, Ronny -------------- next part -------------- A non-text attachment was scrubbed... Name: multi_queue.c Type: text/x-csrc Size: 4747 bytes Desc: not available URL: <http://xenomai.org/pipermail/xenomai/attachments/20160205/5a4434b5/attachment.c> -------------- next part -------------- A non-text attachment was scrubbed... Name: psos_queue_correction.patch Type: application/octet-stream Size: 2106 bytes Desc: not available URL: <http://xenomai.org/pipermail/xenomai/attachments/20160205/5a4434b5/attachment.obj> _______________________________________________ Xenomai mailing list Xenomai@xenomai.org http://xenomai.org/mailman/listinfo/xenomai
