OS task causing 50mS stall every 60 sec.
Posted: Fri Jan 09, 2015 8:40 pm
I have a Netburner 5270B that I'm using in an application that involves reading data from an external hardware FIFO and periodically sending it out in a UDP data-gram. I had an issue where periodically my FIFO would overflow. I created a simple program that toggled a pin in an endless loop and found expected jitter due to OS background tasks of a few uS when monitoring with a scope. I also found a periodic stall of 80uS and most disturbingly a stall of 50mS every 60 Sec. I used task monitor and found a task called TCPD with a priority of 40 being called every 1200 or so ticks. That comes out to about 60 sec. so I think it's the culprit.
Question is what is that task and is there a way to kill it or cause it to take less than 50mS away from user main at any one time. It would be acceptable to lose less time more often for example 5 mS every 6 sec. As it stands my FIFO is not large enough to ride through the 50mS stall.
Thanks in advance for any help.
Question is what is that task and is there a way to kill it or cause it to take less than 50mS away from user main at any one time. It would be acceptable to lose less time more often for example 5 mS every 6 sec. As it stands my FIFO is not large enough to ride through the 50mS stall.
Thanks in advance for any help.