On Fri, 2015-10-30 at 21:37 +0100, Richard Cochran wrote:
>
> NAK. We want to have the time stamp ASAP, because the delay between
> the transmission of the Sync and its application in the servo
> degrades or even spoils synchronization. Think of an ARM SoC with
> HZ=100. Two jiffies are 20 milliseconds.
The workqueue is only used for delivering packets where the timestamp
was lost - if a timestamp is received, the packet is delivered
immediately. This patch may introduce a slight delay of the packets
where the timestamp never arrives, by these packets are properly not
very useful to the application anyway.
Stefan
N�����r��y����b�X��ǧv�^�){.n�+���z�^�)����w* jg��� �����ݢj/���z�ޖ��2�ޙ����&�)ߡ�a���� �G���h��j:+v���w��٥
