ch.rlist initialization with CH_CFG_NO_IDLE_THREAD Topic is solved

[lungj]

It looks like if CH_CFG_NO_IDLE_THREAD is enabled, the thread ready list's next/prev values are set incorrectly. This causes problems when attempting to sleep: an invalid thread address is selected as the next element by queue_fifo_remove in chSchGoSleepS. Is this behaviour intentional?

Code: Select all

int main(void) {
   halInit();
   chSysInit();

   while(1) {
      chThdSleepMilliseconds(1000); // This fails if CH_CFG_NO_IDLE_THREAD is TRUE.
   }
   return 0;
}

[Giovanni]

Hi,

When CH_CFG_NO_IDLE_THREAD is enabled then the main() function is the idle thread, you must never try to sleep in there. Doing so causes the problems you are describing.

Giovanni

[faisal]

Hi,

When CH_CFG_NO_IDLE_THREAD is enabled then the main() function is the idle thread, you must never try to sleep in there. Doing so causes the problems you are describing.

Giovanni

Is this still the case? If so, there should be an assert somewhere to catch this. I think chSchGoSleepS is the right place.

[Giovanni]

Yes, an assertion is a good idea.

Giovanni

[Dzarda]

I got burned by this again, I think that an assertion should be very suitable.

[Giovanni]

Moving into "bug reports" so it is not lost in the noise.

Giovanni

[Giovanni]

Hi,

Assertion added in thunk code.

Giovanni