-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 14.02.2012 22:15, schrieb George Broz: > Yes - the application cycle (as far as I can tell) should be > synchronized to the return of rt_timer_read(). > > I've tried two methods: (1) a regular Xenomai task using > rt_task_wait_period() (after setup with rt_task_set_periodic()), > and > > (2) a Xenomai task awoken on an interrupt/timeout using > rt_intr_wait() where the timeout period is calculated using > rt_timer_read() on entry/exit from rt_intr_wait(). > > The period of the real-time task is set for 250 micro-seconds. > Xenomai task latency is on average in the 1 to 4 micro-second > range, sometimes seeing a datapoint in the 10 to 20 micro-second > range.
This is how we do it: http://etherlab.hg.sourceforge.net/hgweb/etherlab/etherlab/file/f07b08426fc9/tools/rt_appcore/rt_main.c (see app_main_thread()) http://etherlab.hg.sourceforge.net/hgweb/etherlab/etherlab/file/f07b08426fc9/rtw/lib/ecrt_support.c (see ecs_send()) - -- Best regards, Florian -----BEGIN PGP SIGNATURE----- Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk9F9mYACgkQABFOIMygR8wEgwCfX34uPt4SnFYp59bhQsIv3WS+ UDsAoIXNrn4PEYIJ7Boz1N1XOn2bKiaR =nLYW -----END PGP SIGNATURE----- _______________________________________________ etherlab-users mailing list etherlab-users@etherlab.org http://lists.etherlab.org/mailman/listinfo/etherlab-users