signals on a power pc 500

Zaa wrote: : Hi, : : I have a device driver that generates a signal sgmin signo 37 I believe. : It is supposed to generate this unix signal at a certain rate. : I want to be able to measure the event happening and detect any drift or : delta in firings of the event.

You want to measure the timing of a software event, is that it? Unfortunately, as far as my U*IX knowledge goes, signals are to user land applications what (HW) interrupts or, in a wider sense, exceptions are to the CPU/kernel.

Thus, signals will not be detecteable as anything out of the ordinary per se, hardware wise. If OTOH the signal is not to frequent, you could just read the RT-clock at every received signal to gain your difference. Else, set up a fast, free-running timer by means of your device driver and read current count by your signal reciving application. You will then have your diff in the form of number of timer ticks.

: How can I do that with a scope. Is there a pin I can take this reading : off? Or am I better off going through the Jtag interface to get this info.

• To be able to use a scope, you need to toggle a GPIO pin.
• To toggle a pin, you ordinarily need a device driver. (It is theoretically possible to memory map IO devices into a userland reachable region and have all cache coherence issues staighten out. Can't help you there.)
• Calling a device driver (might) induce variable delay. This will also plague the free running timer approach.

Hence, you will be able to measure the _generation_ of your signal quite accurately by flipping the pin just after posting your signal. The reception of signals in userland is a different story. I might be wrong, but this is my understanding.