Critical-Section Problem. (Dekker)

[Peon666]

I read this code in my OS book but I'm having trouble understanding it. Would someone explain to me what's happening in this code? This code is the first-known correct software solution to the critical-section problem for two processes and was developed by Dekker.

boolean flag[2];
int turn;

// Above two variables are shared by the two processes.

do {
    flag[i]=TRUE;
    while (flag[j]){
        if (turn == j){
            while (turn == j)
               ;// do nothing
            flag[i]=FALSE;
        }
    }

// critical section

turn = j;
flag[i]=FALSE;

// remainder section.

}while (TRUE);

I'm totally clueless about what's going on in this code.

 

[rcgldr]

The example you show isn't quite right, wiki shows the proper version:

http://en.wikipedia.org/wiki/Dekker's_algorithm

This matches the wiki version:

volatile boolean flag[2] = {FALSE, FALSE}  // global shared variable
volatile int turn = 0;                     // global shared variable

static int i, j;

// initialize
// for process 0, i = 0, j = 1
// for process 1, i = 1, j = 0

do{

// non-critical section

    flag[i] = TRUE;              // set flag[i] to request ownership
    while(flag[j] == TRUE){      // while other process has set flag[j]
        if(turn != i){           //   if turn != this process
             flag[i] = FALSE;    //     clear flag[i] so other process can run
             while(turn != i);   //     wait for other process to change turn
             flag[i] = TRUE;     //     set flag[i] to request ownership
        }
    }                            // wait until other process clears flag[j]
 
// critical section

    turn = j;                    // set turn to other process
    flag[i] = FALSE;             // clear flag[i] so other process can run

// non-critical section

}while (TRUE);