timer issue

williamjsell wrote on Tuesday, May 19, 2015:

Target: LCP177x_8x
RTOS version: 8.0.0
IDE: Rowley Crossworks for ARM

We have a debounce algorithm for a membrane switch, which relies on the setting and resetting of a timer in RTOS, based on button edge interrupts. This works great for a while, then the system appears to hang up after many repeated button presses. I traced the reason to this line in timers.c:

static void prvProcessReceivedCommands( void )
DaemonTaskMessage_t xMessage;
Timer_t *pxTimer;
BaseType_t xTimerListsWereSwitched, xResult;
TickType_t xTimeNow;

while( xQueueReceive( xTimerQueue, &xMessage, tmrNO_DELAY ) != pdFAIL )

where suddenly I get a return status from the xTimerStopFromISR or xTimerStartFromISR indicating that the queue is full. I put a printf debug after the xQueueReceive and recreated the problem, and indeed this no longer was processing the messages from the timers. So the queue fills, the kernel seems quite happy, but we cannot navigate from the current screen as the buttons are frozen…any ideas what is stopping this from working?

rtel wrote on Wednesday, May 20, 2015:

So if I understand correctly, the code you show is the loop which drains the timer command queue, the queue is full, but the timer task is no longer draining the queue.

If the timer task is not in the correct state then it could be an interrupt priority issue. Do you have configASSERT() defined? In later versions of FreeRTOS (not sure about V8.0.0) that will automatically catch interrupt priority mis-configurations.

I would recommend updating your FreeRTOS version anyway as V8 had an issue with the event groups “set bits from ISR” function http://www.freertos.org/History.txt.


williamjsell wrote on Wednesday, May 20, 2015:

Yes, the problem is as you describe. Let me upgrade to the latest RTOS version first and see if this helps. How do you turn on the assertions (#define configASSERT in the FreeRTOSConfig.h file)?

rtel wrote on Wednesday, May 20, 2015:

Yes - define it in FreeRTOSConfig.h, normal assert semantics. In its
most basic form you can:

#define configASSERT( ( x ) ) if( ( x ) == 0 ){ 
portDISABLE_INTERRUPTS(); for(;;); }

to just bring the system to a stop on the offending line.



williamjsell wrote on Wednesday, May 20, 2015:

OK, making some progress, the assert was very helpful. What I am seeing when the button is pressed is a trap in vPortValidateInterruptPriority(); the assert fails here:
configASSERT( ucCurrentPriority >= ucMaxSysCallPriority );

my confusion is the button is set to 0 priority, but the ucMaxSysCallPriority is set to 40. I thought (based on the doco at http://www.freertos.org/RTOS-Cortex-M3-M4.html and the LPC1788 UM, the max priority levels for an interrupt is 32. Still trying to understand how this all works…

williamjsell wrote on Wednesday, May 20, 2015:

ok, problem solved and embarrassingly I had set all other system interrupt priorities correctly, but forgot the GPIO irq. Just required this line:


where the configGPIO_INTERRUPT_PRIORITY is greater than the configMAX_SYSCALL_INTERRUPT_PRIORITY value…