Hi,
I'm looking at code coverage and timing analysis tools and it would solve some headaches if I can accomplish those goals using program counter data streamed through the ETM, From ARM's documentation it seems like the number of TPIU pins needed depends on the rate of data consumed and that the number of pins used can be configured. However, the ETM-R5 documentation implies that all 32 pins must be used. Unfortunately ETMDATA[31-8] are shared with various EMIF control, address, and data pins most of which I'm already using.
Given the overlap, is it true to say that: real time streaming of the PC is not possible if a device is also using the EMIF?
If it is true, I'm wondering what the rationale was for the pin overlap, specifically if is there is another hardware assisted method to obtain high rate core trace data that's just less optimal. For example buffering the trace data and then streaming it out when the EMIF does not have an active transfer in progress.
Thanks for any information,
Matt
