RTLinux

If you've got the source, you've got all that you need.

Measuring interrupt latency is a matter if recording the time at the start of the interrupt and recording it again at the end. This may be in software, or it may just be a matter of toggling a GPIO line and watching it on an oscilloscope. But you're going to have to tweak the interrupt handlers to do this.

I would suggest asking FSM Labs directly if they already support this or not. They may already have these numbers for a variety of systems.

-- Darin Johnson

To be specific, it's a question of measuring the time between - the event that causes the interrupt - the interrupt being serviced (taking the required action)

It's important to note that the start of the interval is not within the interrupt: what's of most interest is the time taken for the service routine to start, and is a function of the longest time the interrupt can be disabled for any reason anywhere in the code.

It's not an easy thing to measure; it's usually calculated. It's very likely to be available as a metric, since this is one of the defining metrics of an RTOS.

Steve