Funny you should mention this - I've been poking at my zaurus to see about keeping track of time across power loss. It seems that the RTC is powered from the battery/wall and if there is no power, we lose the RTC. Why oh why couldn't they have stuck in a little wristwatch timer, BACKED BY AN INDEPENDENT watch battery...? ya ya, cost and size...
Anyway, I made friends with the processor manual: www.intel.com/design/pca/applicationsprocessors/manuals/280000.htm On 6/8/05, Matthias Kilian <[EMAIL PROTECTED]> wrote: > Running ntpd -s on my zaurus, I sometimes (not allways) get log > messages like > > reply from ...: negative delay -0.000022 > > IMHO, potential candidates are (in decreasing order): Don't forget about a possibly off-kilter realtime clock. TFM says that there are two 1Hz clocks, one 100Hz clock and one 1kHz clock all accurate to one part in 32768. obRTFM: page 935-937 of the PXA27x processor family manual. > OpenBSD 3.7-current (GENERIC) #0: Sun Jun 5 13:30:53 CEST 2005 > ... > saost0 at pxaip0 addr 0x40a00000 > saost0: SA-11x0 OS Timer > ... > clock: hz=100 stathz=64 10ms clock, +/- 30.5usec. Not exactly the best thing in the world if you're looking for sub-millisecond timing. 1/32768 = .0000305 .0000305 > fabs(-0.000022) In short, the delay error is within the error limits of the onboard clocks. -- GDB has a 'break' feature; why doesn't it have 'fix' too?
