#include <stdio.h>
#include "locks.hfa"
#include <stdlib.hfa>
#include <thread.hfa>
#include <containers/array.hfa>

static unsigned int taskCount = 4;
static unsigned int threadCount = 2;
static unsigned int lockCount = 1;
static unsigned int total_times = 320000;
static unsigned int num_times;
static const int workBufferSize = 16;
static unsigned int work_unlocked = 10000;
static unsigned int work_locked = 10000;

// taken from martin's thread_test
static inline void dowork(volatile int* buffer, unsigned int steps) {
  int value = 0;
  for (unsigned int i = 0; i < steps; i += 1) {
    // a little more work than just a single memory access helps with stability
    value += (buffer[i % workBufferSize] * 17) / 23 + 55;
  }
  buffer[0] += value;
}

thread worker {
    linear_backoff_then_block_lock * locks;
    bool improved;
};

void ?{}( worker & w, linear_backoff_then_block_lock * locks, bool improved ) {
	w.locks = locks;
    w.improved = improved;
}


void main( worker & this ) with(this) {
	int buffer[workBufferSize];
    for (int i = 0; i < workBufferSize; i += 1) buffer[i] = rand() % 1024;
    unsigned int lck = rand() % lockCount;
    linear_backoff_then_block_lock * curr_lock = &locks[lck];
    for (unsigned int i = 0; i < num_times; i++) {
        dowork(buffer, work_unlocked);
        if (improved) lock_improved(*curr_lock);
        else lock(*curr_lock);
        dowork(buffer, work_locked);
        unlock(*curr_lock);
        lck = rand() % lockCount;
        curr_lock = &locks[lck];
    }
}


int main(int argc, char* argv[]) {
    switch (argc) {
        case 7:
            work_unlocked = atoi(argv[5]);
        case 6:
            work_locked = atoi(argv[5]);
        case 5:
            num_times = atoi(argv[4]);
        case 4:
            lockCount = atoi(argv[3]);
        case 3:
            threadCount = atoi(argv[2]);
        case 2:
            taskCount = atoi(argv[1]);
        case 1:
            break;
        default:
            break;
    }
	processor p[threadCount];
    linear_backoff_then_block_lock locks[lockCount];
    worker * worker_arr[taskCount];
    num_times = total_times  / taskCount;

	//printf("Start Test: martin lock simple\n");
	clock_t begin = clock();
	for (unsigned int i = 0; i < taskCount; i++) {
        worker_arr[i] = new( locks, false );
    }
    for (unsigned int i = 0; i < taskCount; i++) {
        delete( worker_arr[i] );
    }
	clock_t end = clock();
	double time_spent = (double)(end - begin) / CLOCKS_PER_SEC;
	printf("norm: %f\n", time_spent);

    //printf("Start Test: martin lock improved\n");
	begin = clock();
	for (unsigned int i = 0; i < taskCount; i++) {
        worker_arr[i] = new( locks, true );
    }
    for (unsigned int i = 0; i < taskCount; i++) {
        delete( worker_arr[i] );
    }
	end = clock();
	time_spent = (double)(end - begin) / CLOCKS_PER_SEC;
	printf("improved: %f\n", time_spent);
}