/*
 * fractal-gen - Generate iteration-based fractals in PNM format
 * Copyright (c) 2016 David Phillips <dbphillipsnz@gmail.com>
 * All rights reserved
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#include "fractal-gen.h"
#include "generator.h"

#include <stdio.h>
#include <stdlib.h>
#include <libgen.h>
#include <unistd.h>
#include <signal.h>
#include <math.h>
#include <sys/mman.h>
#include <string.h>

pid_t child = -1;
struct frame f;

static struct section_generator generators[] = {
	{ "mandelbrot-gen" , &generate_mandelbrot_section },
	{ "burning-ship-gen" , &generate_burning_ship_section }
};

char *ram_units[] = {
	"B", "KiB", "MiB", "GiB", "TiB"
};


void
defaultsd(double *who, double def)
{
	if (isnan(*who))
		*who = def;
}

double
timespec_diff(struct timespec start, struct timespec end) {
	long weight = 1000000000;
	time_t s = (end.tv_sec - start.tv_sec);
	long ns = (end.tv_nsec - start.tv_nsec) % weight;

	return s + ((double)ns)/weight;
}

void
*generate(void *section) {
	data_section *s = (data_section*)section;
	clock_gettime(CLOCK_THREAD_CPUTIME_ID, &(s->time_start));
	(s->generator)(s);
	clock_gettime(CLOCK_THREAD_CPUTIME_ID, &(s->time_end));
	return NULL;
}

void
handle_signal(int sig) {
	switch (sig) {
	case SIGSEGV:
		/* kill the forked thread and abort */
		if (child > 0) {
			kill(child, SIGKILL);
			abort();
		}
		break;
	default:
		/* do nothing */
		break;
	}
}

void
install_sigaction(void handler(int)) {
	struct sigaction sa;

	memset(&sa, 0, sizeof(sa));
	sa.sa_handler = handler;
	sigaction(SIGSEGV, &sa, NULL);
}

void
destroy_sections(data_section **s, unsigned long count) {
	unsigned long i = 0;

	/* free malloced data memory */
	for (i = 0; i < count; i++) {
		free((*s)[i].data);
	}

	/* unmap mmapped sections */
	munmap(s, sizeof(data_section)*threads);
}

int
main(int argc, char **argv) {
	unsigned long x = 0;
	unsigned long width = 0;
	size_t toalloc = 0;
	unsigned long y = 0;
	unsigned long i = 0;
	double ram_nice = 0.f; /* Forecast RAM usage, divided down to < 1024 */
	char* ram_unit = NULL; /* Unit for ram_nice */
	char* bname = NULL;
	data_section* sections = NULL;
	data_section *s = NULL;
	generator_func generator = NULL;

	/* who are we? */
	argv0 = argv[0];

	/* set up signal handler */
	install_sigaction(handle_signal);

	/* Select correct generator for the fractal type */
	bname = basename(argv[0]);
	generator = select_generator(bname);

	if (generator == NULL) {
		fprintf(stderr, "Unknown fractal '%s' (perhaps try running symlink to me)\n",
			bname);
		return 1;
	}

	if (parse_args(argc, argv)) {
		show_help();
		return 1;
	}

	/* Allocate memory for sections */
	if ((sections = mmap(NULL, sizeof(data_section)*threads, PROT_READ|PROT_WRITE,
	MAP_SHARED|MAP_ANONYMOUS, -1, 0)) == (data_section*)MAP_FAILED) {
		perror("mmap");
		return 1;
	}

	ram_nice = (size*size)/clust_total;

	i = 1;
	ram_unit = ram_units[0];
	while (   ram_nice > 1024
	       && i < sizeof(ram_units) / sizeof(ram_units[0])) {
		ram_unit = ram_units[i++];
		ram_nice /= 1024;
	}

	/* FIXME clean up */
	struct timespec time_start, time_end;
	clock_gettime(CLOCK_REALTIME, &time_start);

	fprintf(stderr,
		"Forecast resource use:\n"
		" Threads: %lu\n"
		" RAM    : ~%.4f %s\n",
		threads,
		ram_nice,
		ram_unit);
	/* Spawn all the threads! Something something interlacing */
	for (i = 0; i < threads; i++) {
		/* A bit complex, icky, will document later */
		if (i < (size%threads))
			width = (size/threads)+1;
		else
			width = (size/threads);

		toalloc = width*size;
		toalloc = ceilf((double)toalloc/clust_total);

		if ((sections[i].data = malloc(toalloc)) == NULL) {
			fprintf(stderr, "\nmalloc of %zd bytes failed\n", toalloc);
			perror("malloc");

			/* free resources allocated to sections array */
			destroy_sections(&sections, i-1);
			return 1;
		}
		/* FIXME repetition */
		sections[i].core = i;
		sections[i].width = width;
		sections[i].parent_frame.y = f.y;
		sections[i].parent_frame.x = f.x;
		sections[i].generator = generator;
		sections[i].parent_frame.scale = f.scale;
		sections[i].datasize = toalloc;
		fprintf(stderr, " -> Thread %lu\r", i);
		if (pthread_create(&sections[i].thread, NULL, generate, &(sections[i])) != 0) {
			perror("pthread_create");
			destroy_sections(&sections, i);
			return 1;
		}
	}

	s = &(sections[threads-1]);

	switch (child = fork()) {
	case 0:
		while(1) {
			fprintf(stderr, "Thread %lu: %.4f%%\r",
					threads-1,
					100.f*(double)s->idx/s->datasize);
			sleep(1);
		}
		break;
	case -1:
		perror("progress thread fork");
	default:
		break;
	}
	/* Wait for each thread to complete */
	for (i = 0; i < threads; i++)
		pthread_join(sections[i].thread, NULL);

	kill(child, SIGKILL);

	clock_gettime(CLOCK_REALTIME, &time_end);

	fprintf(stderr, "\nDone\n");

	double time_wall = timespec_diff(time_start, time_end);
	double time_ch = 0;

	for (i = 0; i < threads; i++) {
		data_section *s = &(sections[i]);
		time_ch += (timespec_diff(s->time_start, s->time_end)) / sysconf(_SC_NPROCESSORS_ONLN);
	}

	fprintf(stderr,
		"Wall-clock time: %.2f seconds\n"
		"Average time per CPU thread: %.2f seconds\n"
		"Multi-core efficiency: %.2f%%\n"
		, time_wall, time_ch, 100*(time_ch)/time_wall);

	/* Output PGM Header */
	printf("P5\n%lu\n%lu\n255\n",size,size/clust_total);

	/* Vomit the data segments onto stdout, interlacing frames from threads
	 * FIXME: look at buffering if at all possible */
	for (y = 0; y < size/clust_total; y++) {
		for (x = 0; x < size; x++)
		{
			s = &(sections[x%threads]);
			putchar(s->data[y*(s->width) + x/threads]);
		}
	}

	destroy_sections(&sections, threads);
	return 0;
}


int
parse_args(int argc, char **argv)
{
	char opt = '\0';

	/* first things first: preload default or initial values */
	size = 0;
	iterat = 0;
	power = 2;
	threads = sysconf(_SC_NPROCESSORS_ONLN);
	thread_mult = 1;
	clust_id = 0;
	clust_total = 1;

	f.x = nan("");
	f.y = nan("");
	f.scale = nan("");

	/* bail out early if no arguments are supplied */
	if (argc <= 1)
		return 1;

	while ( (opt = getopt(argc, argv, "s:i:e:c:t:N:T:x:y:z:")) != -1 ) {
		switch (opt)
		{
			case 's': size = atoi(optarg); break;
			case 'i': iterat = atoi(optarg); break;
			case 'e': power = atof(optarg); break;

			case 'c': threads = atoi(optarg); break;
			case 't': thread_mult = atof(optarg); break;

			case 'N': clust_id = atoi(optarg); break;
			case 'T': clust_total = atoi(optarg); break;

			case 'x': f.x = atof(optarg); break;
			case 'y': f.y = atof(optarg); break;
			case 'z': f.scale = atof(optarg); break;

			/* redundant case for '?', but explicitness is best */
			case '?':
			default:
				return 1;
				break;
		}
	}

	/* Extend number of threads to multiplier value */
	threads *= thread_mult;

	if (size <= 0) {
		fprintf(stderr, "ERROR: size must be positive\n");
		return 1;
	}

	if (iterat <= 0) {
		fprintf(stderr, "ERROR: max iteration count must be positive\n");
		return 1;
	}

	if (clust_id >= clust_total) {
		fprintf(stderr, "WARN: Node ID cannot be >= node count\n");
		return 1;
	}

	/* Interlacing is row-based, can't have more workers than columns */
	if (threads > size) {
		threads = size;
		fprintf(stderr, "WARN: Capping number of threads to image size (%lu)\n", threads);
	}

	if (size % clust_total != 0) {
		fprintf(stderr, "ERROR: image size must be an exact multiple of clust_total\n");
		return 1;
	}

	if (threads <= 0) {
		fprintf(stderr, "ERROR: core counts should be positive\n");
		return 1;
	}
	return 0;
}


generator_func select_generator(const char* name)
{
	unsigned long i = 0;
	for (i = 0; i < sizeof(generators)/sizeof(struct section_generator); i++)
		if (strcmp(name, generators[i].executable_name) == 0)
			return generators[i].generator;

	return NULL;
}

void show_help()
{
	fprintf(stderr,
			"%s -s size -i iterat [-e exponent]\n"
			"        [-c threads] [-t thread_multiplier]\n"
			"        [-N cluster-id -T cluster-total]\n",
			argv0);
}