#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <pthread.h>

static uint32_t *workerID;
static uint32_t numWorkers;
static uint32_t m_cost;

// Memory to be hashed
static uint32_t *mem;
static uint32_t *hash;

// Do L3-cache intensive work in a regular manner, not SSE enabled.
static void *doWork(void *workerIDPtr) {
    uint32_t workerID = *(uint32_t *)workerIDPtr;
    uint64_t len = (1024ull << m_cost)/(sizeof(uint32_t)*numWorkers);
    printf("Hashing %lu memory on thread %u\n", len*sizeof(uint32_t), workerID);
    uint32_t *p = mem + len*workerID;
    p[0] = 2*workerID + 1; // Always odd
    uint64_t randAddr = 0;
    for(uint64_t i = 1; i < len; i++) {
        p[i] = p[randAddr] * (randAddr | 1) + 1234; // Always odd
        randAddr ^= p[i] >> 8;
    }
    hash[workerID] = randAddr;
    pthread_exit(NULL);
}

// Start numWorker worker threads that do regular speed cache thrashing.
static void startWorkers(pthread_t *workerThreads, uint32_t numWorkers) {
    for(uint32_t i = 0; i < numWorkers; i++) {
        workerID[i] = i;
        pthread_create(&workerThreads[i], NULL, doWork, workerID + i);
    }
}

// Join numWorker worker threads.
static void joinWorkers(pthread_t *workerThreads, uint32_t numWorkers) {
    for(uint32_t i = 0; i < numWorkers; i++) {
        pthread_join(workerThreads[i], NULL);
    }
}

int main(int argc, char **argv) {
    if(argc != 3) {
        printf("Usage: %s numWorkers m_cost\n", argv[0]);
        return 1;
    }
    numWorkers = atoi(argv[1]);
    m_cost = atoi(argv[2]);
    bool hashMemory = atoi(argv[4]);
    workerID = calloc(numWorkers, sizeof(uint32_t));
    hash = calloc(numWorkers, sizeof(uint32_t));
    mem = malloc(1024llu << m_cost);
    printf("Allocated %llu memory\n", 1024ll << m_cost);
    pthread_t workerThreads[numWorkers];
    startWorkers(workerThreads, numWorkers);
    joinWorkers(workerThreads, numWorkers);
    uint32_t result = 0;
    for(uint32_t i = 0; i < numWorkers; i++) {
        result += hash[i];
    }
    printf("Result: %u\n", result);
    return 0;
}