k00hacksPWM.ino

Go to the documentation of this file.

// 220216 -
#include <krnl.h>
 
struct k_t *pTask; // pointer to task descriptor
 
// stak to be used by task
#define STK 150
unsigned char taskStak[STK];
 
void t1() {
 
  // blink approx 1 Hz
  while (1) // your task loop - forever
  {
    digitalWrite(13, !digitalRead(13)); // togle led
    k_sleep(500);  // instead of delay
  }
}
 
void(* resetFunc) (void) = 0;//declare reset function at address 0
 
void setup() {
  int res;
 
  pinMode(13, OUTPUT); // for debug
 
  k_init(1, 0, 0); // init with space for one task
  //            ^------- number of message queues
  //         ^--------- number of semaphores
  //      ^------- number of tasks
  // RETURNS 0 is all went ok
 
  // priority low number higher priority than higher number
  pTask = k_crt_task(t1, 10, taskStak, STK); // t1 as task, priority 10, 100 B stak
  //                                    ^--- stak size in bytes
  //                                ^-- array used for stak
  //                      ^---------- task priority 1 is highest
  //                  ^------------- function to be used as body for task
  // return NULL ptr if something went wrong otherwise pointer to task descriptior
 
  res = k_start(); // 1 milli sec tick
 
  // you will never return from k_start
  // if you come here is it bq krnl has not started due to an error
  // ret vakue is negative and is a count of initalisation errors like
  // no slot for tasks etc
  // see krnl.c approx line ++1270
 
  // RESET
 
  (*resetFunc)();
  while (1) ;
}
 
void loop() {} // is NEVER used
 
/*
   README README
   In task t1 we are waiting 500 msec in each loop by calling k_eat_msec
   This is busy waiting so t1 is running on CPU all time
   The debug k_break function below turn led13 on when dummy is running (dummey
   has pRun->nr  == 0) But led 13 never light up - meaning 100% usage of cpu :-(
 
   Changing from k_eat_msec(500)  to k_sleep(500) in t1 we still are waiting 500
   msec but now in passive mode So try do remove comment from k_sleep and
   comment out k_eat_msec Now you will observe that led13 is ON all time. Why?
   Because printing an integer takes nearly no time compared to sleeping in 500
   msec It you observe by a logic analyser on pin13 you will see it goes low for
   a very sjort amount of time and then high for 500 msec. So we dont use the
   CPU for just waiting -
 
   2)
   Try to change k_eat_msec and k_sleep to (and no comment in from of no one ot
   them k_eat_msec(500); k_sleep(2000);
 
   Then will observe the led13 is off foo 500 msec (when printing and running
   k_eat_msec) and led13 is ON for 2000 msec(when we are sleeping in
   k_sleep(2000)
 
   You can see when your are printing becasue the led marked TX is ON when the
   uno is doing serial printing
 
   THIS EXAMPLE IS WITHOUT printf. using printf takes at least 200 bytes of ram
   Compare with same example  just named k01myfirsttask-w-printf so see diff i ram usage
   /JDN
*/
 
/***
 
  extern "C" {
 
  void k_breakout(void) // called every task shift from dispatcher
  {
 
  if (pRun->nr == 0) // 0 is dummy task - the eater of excessive CPU when all
                    // user tasks are idling
  {
   PORTB = PORTB | B00100000; // led13 (bit 5) on let the rest be untouched
  } else {
   PORTB = PORTB & B11011111; // led13 off uno
  }
  }
  }
 
  // on MEGA led13 is on PORTB bit 7.    so use B10000000 instead of B00100000 and
  // B011111111 instead of B11011111
 
*/