October 6, 2016

A Kyu tutorial - the race condition fixed.

Here we fix the race in the previous problem by adding a locking (or "mutual exclusion") semaphore. Commonly abbreviated by the slangy term "mutex". Here we wrap the lock around the incrementing of all the variables, but truth be known, it should really only be needed around "wally". 
#include "kyu.h"
#include "thread.h"

static volatile int wally = 0;
static volatile int count1 = 0;
static volatile int count2 = 0;
static struct sem *lock_sem;

#define HZ	10000

void
racer1 ( int xx )
{
	for ( ;; ) {
	    thr_delay ( 1 );

	    sem_block ( lock_sem );
	    wally++;
	    count1++;
	    sem_unblock ( lock_sem );
	}
}

void
racer2 ( int xx )
{
	int races = 0;
	int got_race;

	while ( races < 10 && count1 < 60*HZ ) {

	    sem_block ( lock_sem );
	    wally++;
	    count2++;
	    got_race = (wally != count1 + count2 );
	    sem_unblock ( lock_sem );

	    if ( got_race ) {
		printf ( "Race after %d / %d (%d)\n", count1, count2, wally );
		wally = count1 + count2;
		races++;
	    }
	}
	printf ( "Done\n" );
}

void
user_init ( int xx )
{
	timer_rate_set ( HZ );
	lock_sem = sem_mutex_new ( SEM_PRIO );
	(void) thr_new ( "racer1", racer1, (void *) 0, 20, 0 );
	(void) thr_new ( "racer2", racer2, (void *) 0, 30, 0 );
}

Have any comments? Questions? Drop me a line!

Kyu / tom@mmto.org