I am busy putting together a set of traceXXXX macros covering the whole scope of functions. Essentially I am mapping all functions into 3 different categories being: vFreeRTOSTraceTASK(): all task related functions vFreeRTOSTraceQUEUE(): queue, mutex and semaphore vFreeRTOSTraceTIMER(): timer related functions Within each of the functions I am just streaming max 8 character string (built compact just identifying the object and event) as fast and direct as possible to the UART. What I have noticed is that with counting semaphores, no parameter is passed. When compared with queues and mutexes, I would have assumed passing the xHandle value would make sense. Am I missing something, a reason why nothing is passed? Rgds Andre
Some of the macros were updated to accommodate the specific needs of the FreeRTOS+Trace tool, so it might be related to that. Can you give a specific example of a queue trace macro that does include the paraemter, and a mutex macro that doesn’t include the parameter, so I may be able to give a more specific answer. Regards.
As an example, have a look at calling of: traceCREATEMUTEX(pxNewQueue) vs traceCREATECOUNTING_SEMAPHORE() Would it not make sense to provide the xHandle as parameter to identify the semaphore?
I would agree it is probably sub-optimal. In the case of a counting semaphore more than one trace macro is invoiced, first the macros in xQueueGenericCreate() [which is itself called by xQueueCreateCountingSemaphore()] and then traceCREATECOUNTINGSEMAPHORE() – so presumably the handle is already known from when xQueueGenericCreate() was called. You can still obtain the handle, but you would need to look at the code to see what the handle was called at the point the macro was called. In the case highlighted it is simply “xHandle”. Just looking at the macro definitions used by FreeRTOS+Trace (which are in /FreeRTOS-Plus/Source/FreeRTOS-Plus-Trace/Include/trcKernelPort.h) it appears traceCREATECOUNTINGSEMAPHORE() is not used – so maybe that is why it never had the parameter? Regards.
As an aside, I’m not sure a serial port would be sufficiently fast to stream data at the rate a running application would generate it. Regards.
Currently running UART at 961.2k, and limiting the specific macros enabled to limit output volume, works reasonably for th exceptions I need to track. Next step will be to provide support for GPIO pin status, mainly to track task switching on a logic analyser. I can then mix the 2 output mechanisms based on the nature of the trace..
Copyright (C) Amazon Web Services, Inc. or its affiliates. All rights reserved.