xQueueCreate Failing

Kernel
1

Hello,
I am relatively new to FreeRTOS and trying to port FreeRTOS to R7 core. I don’t understand why xQueueCreate( mainQUEUE_LENGTH, sizeof( uint32_t ) ); fails.

While trying to debug, I understood that the assertion " configASSERT( uxSchedulerSuspended );", inside xTaskResumeAll() function fails, as uxSchedulerSuspended becomes zero inside vTaskSuspendAll(). How can I solve this issue??

Thanks in advance.

2

We already have some Cortex-R ports. Which compiler are you using? Also, which chip?

Each call to vTaskSuspendAll() needs to be paired with a falling call to xTaskResumeAll(). xTaskResumeAll() should therefore never find uxSchedulerSuspended is set to zero as it should have been incremented by a previous call to vTaskSuspendAll(). If uxSchedulerSuspended is 0 then either you are calling xTaskResumeAll() without first calling vTaskSuspendAll(), or uxSchedulerSuspended has been corrupted.

3

Thank you @rtel for responding. I am using the cortex-R5 port in Zynq Ultrascale as a base. My intention is to port FreeRTOS into R7 core in RCAR V3H by compiling using arm-none-eabi- toolchain.

The function I am calling from my application is xQueueCreate(). Before incrementing the uxSchedulerSuspended value in vTaskSuspendAll(), it is coming to be -1. So after that when the call to the vTaskResumeAll() is invoked, uxSchedulerSuspended is zero and the assertion fails. I am not sure if I am doing something wrong.

4

So it sounds like something has gone wrong before vTaskSuspendAll() is called within xQueueCreate(). Can you add the variable to the debugger window so you can see when it gets set to -1? Is it initialised to its correct initial value before main() is called. Also could you attach the .map file so I can see where the offending variable is in the memory space relative to other things.

5

I don’t have a debugger to work with. I am trying to debug within the code itself using serial console. I have attached the objdump file for your refernce. objdump.zip (85.3 KB)

6

Hmm, I think you will be more efficient if you use can get to use a debugger. I don’t think the serial console can give you much visibility without itself interfering with the code’s operation.

7

Are you out of heap?

Call xPortGetFreeHeapSize(); before you try to create the queue and make sure you have enough memory.

Jacob

8

Thank you for the suggestions. Sorry for the late reply. I have solved that issue, the bss section initialization was getting skipped which caused this issue.