Jan Ceuleers <janspam.ceule...@skynet.be> wrote: > On 16/01/11 09:11, Chris Albertson wrote: >> No, if it is not _processed" right at the UTC second it is pointless. >> The Motorola GPS allows you to adjust the timing of the pulse to >> account for delay in the antenna feed line and serial line. > > I was also thinking about avoiding interrupt collisions. In an ideal > world, if the PPS interrupt occurs exactly at the UTC second it is going > to coincide with the system's timer interrupt, is it not? That's even if > the system has only one PPS source.
You assume that system's timer interrupt is somehow being synchronized with the UTC second. I don't think that is what is really happening. The interrupt remains free-running (can occur at any moment, and will usually occur several times a second) and what is varied is the keeping of a local time variable in relation to this interrupt. The system remembers at what time the interrupt occurs and sets the system time to that (incremented) value whenever the interrupt comes in. When a program would read the system time at the moment of the interrupt, it would not read xxx.00000000 seconds. _______________________________________________ questions mailing list questions@lists.ntp.org http://lists.ntp.org/listinfo/questions