On Thu, 29 Apr 2021 09:40:00 +0200 Sean Nyekjaer <sean@xxxxxxxxxx> wrote: > On 28/04/2021 18.32, Jonathan Cameron wrote: > >> When buffered sampling is enabled, the accelerometer will dump data into > >> the internal fifo and interrupt at watermark. Then the driver flushes > >> all data to the iio buffer. > >> As the accelerometer doesn't have internal timestamps, they are aproximated > >> between to interrupts. > > two? > > > > This tends to be a noisy approach, so people often try to apply a filter. > > However, no need to do that for an initial version. > > > > There are some things in here referring to enabling triggered modes, but I'm > > not seeing code to actually do so. The fun question when dealing with fifos > > and triggered mode is what the interface is to switch between the two. > > One option I think we've used before is to just have 'no trigger' match > > up to fifo mode and if a trigger is set, don't use the fifo. > > > Thanks Jonathan. > > Fixed the above text to: > As the accelerometer doesn't have internal timestamps, > they are approximated between the current and last interrupt. Nice. > > I don't know the correct term here, the accelerometer via the watermark, > is doing interrupts. > Is that called no-trigger / trigger ? Triggers are one per scan (Set of samples take at approximately the same time) So anything involving a fifo is without trigger (we don't expose anything because it is of no use for doing synchronous capture across multiple devices). One day I'll add a terminology section to the docs! J > > /Sean
