open-cas-linux/casadm/cas_lib.c

/*
* Copyright(c) 2012-2019 Intel Corporation
* SPDX-License-Identifier: BSD-3-Clause-Clear
*/

#include <stdio.h>
#include <errno.h>
#include <assert.h>
#include <fcntl.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <inttypes.h>
#include <limits.h>
#include <fstab.h>
#include <linux/fs.h>
#include <linux/types.h>
#include <linux/major.h>
#include <mntent.h>
#include <sys/mman.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
#include <sys/types.h>
#include <time.h>
#include <pthread.h>
#include <dirent.h>
#include <stdbool.h>
#include "cas_lib.h"
#include "extended_err_msg.h"
#include "cas_lib_utils.h"
#include "csvparse.h"
#include "statistics_view.h"
#include "safeclib/safe_mem_lib.h"
#include "safeclib/safe_str_lib.h"
#include "safeclib/safe_lib.h"
#include <cas_ioctl_codes.h>
#include <cas_version.h>
#include "psort.h"
#define PRINT_STAT(x) header->cmd_input.cache_stats.x

#define CORE_ADD_MAX_TIMEOUT 30

#define CHECK_IF_CACHE_IS_MOUNTED -1

/**
 * @brief Routine verifies if filesystem is currently mounted for given cache/core
 *
 *        If FAILURE is returned, reason for failure is printed onto
 *        standard error.
 * @param cache_id cache id of filesystem (to verify if it is mounted)
 * @param core_id core id of filesystem (to verify if it is mounted); if this
 *        parameter is set to negative value, it is only checked if any core belonging
 *        to given cache is mounted;
 * @return SUCCESS if is not mounted; FAILURE if filesystem is mounted
 */
int check_if_mounted(int cache_id, int core_id);

/* KCAS_IOCTL_CACHE_CHECK_DEVICE  wrapper */
int _check_cache_device(const char *device_path,
		struct kcas_cache_check_device *cmd_info);

static const char *cache_states_name[ocf_cache_state_max + 1] = {
		[ocf_cache_state_running] = "Running",
		[ocf_cache_state_stopping] = "Stopping",
		[ocf_cache_state_initializing] = "Initializing",
		[ocf_cache_state_incomplete] = "Incomplete",
		[ocf_cache_state_max] = "Unknown",
};

static const char *core_states_name[] = {
		[ocf_core_state_active] = "Active",
		[ocf_core_state_inactive] = "Inactive",
};

#define NOT_RUNNING_STATE "Not running"

#define CACHE_STATE_LENGHT 20

#define CAS_LOG_FILE "/var/log/opencas.log"
#define CAS_LOG_LEVEL LOG_INFO

int vcaslog(int log_level, const char *template, va_list args)
{
	FILE *log;
	time_t t;
	struct tm *tm;
	char *timestamp;
	int ret;

	if (log_level > CAS_LOG_LEVEL)
		return 0;

	log = fopen(CAS_LOG_FILE, "a");
	if (!log)
		return FAILURE;

	ret = lockf(fileno(log), F_LOCK, 0);
	if (ret < 0)
		goto out;

	t = time(NULL);
	tm = localtime(&t);
	if (!tm) {
		ret = FAILURE;
		goto out;
	}

	timestamp = asctime(tm);
	if (!timestamp) {
		ret = FAILURE;
		goto out;
	}

	timestamp[strnlen(timestamp, SIZE_MAX)-1] = 0;

	fseek(log, 0, SEEK_END);
	fprintf(log, "%s casadm: ", timestamp);
	vfprintf(log, template, args);
	fflush(log);

	lockf(fileno(log), F_ULOCK, 0);

out:
	fclose(log);
	return ret;
}

__attribute__((format(printf, 2, 3)))
int caslog(int log_level, const char *template, ...)
{
	va_list args;
	va_start(args, template);
	vcaslog(log_level, template, args);
	va_end(args);
	return 0;
}

__attribute__((format(printf, 2, 3)))
int std_printf(int log_level, const char *template, ...)
{
	va_list args;
	va_start(args, template);
	if (LOG_WARNING >= log_level) {
		va_list args_copy;
		va_copy(args_copy, args);
		vfprintf(stderr, template, args);
		vcaslog(log_level, template, args_copy);
		va_end(args_copy);
	} else {
		vfprintf(stdout, template, args);
	}
	va_end(args);
	return 0;
}

cas_printf_t cas_printf = std_printf;

int validate_dev(const char *dev_path)
{
	struct fstab *fstab_entry;
	fstab_entry = getfsspec(dev_path);
	if (fstab_entry != NULL) {
		return FAILURE;
	}
	return SUCCESS;
}

int validate_path(const char *path, int exist)
{
	if (NULL == path) {
		return FAILURE;
	}

	if (0 == path[0]) {
		cas_printf(LOG_ERR, "Empty path\n");
		return FAILURE;
	}

	if (strnlen(path, MAX_STR_LEN) >= MAX_STR_LEN) {
		cas_printf(LOG_ERR, "File path too long\n");
		return FAILURE;
	}

	if (exist) {
		struct stat _stat = { 0 };
		int result = stat(path, &_stat);
		if (result) {
			cas_printf(LOG_ERR, "File does not exist\n");
			return FAILURE;
		}
	}

	return SUCCESS;
}

int __validate_str_num(const char *source_str, const char *msg,
		long long int min, long long int max, bool validate_sbd)
{
	uint64_t ret;
	char *endptr = NULL;

	errno = 0;
	ret = strtoul(source_str, &endptr, 10);
	if (endptr == source_str || (endptr && *endptr != '\0') ||
			((ret == 0 || ret == ULONG_MAX) && errno)) {
		cas_printf(LOG_ERR, "Invalid %s, must be a correct unsigned decimal integer.\n",
			   msg);
		return FAILURE;
	} else if (ret < min || ret > max) {
		cas_printf(LOG_ERR, "Invalid %s, must be in the range %lld-%lld.\n",
			   msg, min, max);
		return FAILURE;
	} else if (validate_sbd && __builtin_popcount(ret) != 1) {
		cas_printf(LOG_ERR, "Invalid %s, must be a power of 2.\n", msg);
		return FAILURE;
	}

	return SUCCESS;
}

int validate_str_num(const char *source_str, const char *msg, long long int min, long long int max)
{
	return __validate_str_num(source_str, msg, min, max, false);
}

int validate_str_num_sbd(const char *source_str, const char *msg, int min, int max)
{
	return __validate_str_num(source_str, msg, min, max, true);
}

int validate_str_unum(const char *source_str, const char *msg, unsigned int min,
		unsigned int max)
{
	return __validate_str_num(source_str, msg, min, max, false);
}

struct name_to_val_mapping {
	const char* short_name;
	const char* long_name;
	int value;
};

static struct name_to_val_mapping eviction_policy_names[] = {
	{ .short_name = "lru", .value = ocf_eviction_lru },
	{ NULL }
};

static struct name_to_val_mapping cache_mode_names[] = {
	{ .short_name = "wt", .long_name = "Write-Through", .value = ocf_cache_mode_wt },
	{ .short_name = "wb", .long_name = "Write-Back", .value = ocf_cache_mode_wb },
	{ .short_name = "wa", .long_name = "Write-Around", .value = ocf_cache_mode_wa },
	{ .short_name = "pt", .long_name = "Pass-Through", .value = ocf_cache_mode_pt },
#ifdef WI_AVAILABLE
	{ .short_name = "wi", .long_name = "Write-Invalidate", .value = ocf_cache_mode_wi },
#endif
	{ NULL }
};

static struct name_to_val_mapping cleaning_policy_names[] = {
	{ .short_name = "nop", .value = ocf_cleaning_nop },
	{ .short_name = "alru", .value = ocf_cleaning_alru },
	{ .short_name = "acp", .value = ocf_cleaning_acp },
	{ NULL }
};

static struct name_to_val_mapping metadata_mode_names[] = {
	{ .short_name = "normal", .value = CAS_METADATA_MODE_NORMAL },
	{ .short_name = "atomic", .value = CAS_METADATA_MODE_ATOMIC },
	{ NULL }
};

static struct name_to_val_mapping seq_cutoff_policy_names[] = {
	{ .short_name = "always", .value = ocf_seq_cutoff_policy_always },
	{ .short_name = "full", .value = ocf_seq_cutoff_policy_full },
	{ .short_name = "never", .value = ocf_seq_cutoff_policy_never },
	{ NULL }
};

static struct name_to_val_mapping stats_filters_names[] = {
	{ .short_name = "conf", .value = STATS_FILTER_CONF },
	{ .short_name = "usage", .value = STATS_FILTER_USAGE },
	{ .short_name = "req", .value = STATS_FILTER_REQ },
	{ .short_name = "blk", .value = STATS_FILTER_BLK },
	{ .short_name = "err", .value = STATS_FILTER_ERR },
	{ .short_name = "all", .value = STATS_FILTER_ALL },
	{ NULL }
};

static struct name_to_val_mapping output_formats_names[] = {
	{ .short_name = "table", .value = OUTPUT_FORMAT_TABLE },
	{ .short_name = "csv", .value = OUTPUT_FORMAT_CSV },
	{ NULL }
};

static struct name_to_val_mapping metadata_modes_names[] = {
	{ .short_name = "normal", .value = METADATA_MODE_NORMAL },
	{ .short_name = "atomic", .value = METADATA_MODE_ATOMIC },
	{ NULL }
};

static int validate_str_val_mapping(const char* s,
				    const struct name_to_val_mapping* mappings,
				    int invalid_value)
{
	int i;

	if (strempty(s)) {
		return invalid_value;
	}

	for (i = 0; NULL != mappings[i].short_name; ++i) {
		if (0 == strncmp(mappings[i].short_name, s, MAX_STR_LEN)) {
			return mappings[i].value;
		}
	}

	return invalid_value;
}

static int validate_str_val_mapping_multi(const char* s,
					  const struct name_to_val_mapping* mappings,
					  int invalid_value)
{
	const char* p;
	char* token;
	char* delim;
	int value = 0;
	int token_val;

	if (strempty(s)) {
		return invalid_value;
	}

	p = s;
	while (p[0]) {
		delim = strchr(p, ',');
		if (delim == p) {
			/* Empty tokens not allowed */
			return invalid_value;
		}

		if (delim) {
			token = strndup(p, delim - p);
			/* Skip token and comma */
			p = delim + 1;
			if (!p[0]) {
				/* Trailing comma not allowed */
				free(token);
				return invalid_value;
			}
		} else {
			size_t len = strnlen(p, MAX_STR_LEN);
			if (len >= MAX_STR_LEN) {
				return invalid_value;
			}

			token = strdup(p);
			p += len;
		}

		token_val = validate_str_val_mapping(token, mappings, invalid_value);
		if (token_val == invalid_value) {
			free(token);
			return invalid_value;
		}

		value |= token_val;
		free(token);
	}
	return value;
}

static const char* val_to_long_name(int value, const struct name_to_val_mapping* mappings,
			    const char* other_name)
{
	int i;
	for (i = 0; NULL != mappings[i].long_name; ++i) {
		if (mappings[i].value == value) {
			return mappings[i].long_name;
		}
	}
	return other_name;
}

static const char* val_to_short_name(int value, const struct name_to_val_mapping* mappings,
			     const char* other_name)
{
	int i;
	for (i = 0; NULL != mappings[i].short_name; ++i) {
		if (mappings[i].value == value) {
			return mappings[i].short_name;
		}
	}
	return other_name;
}

/* Returns non-negative policy index or
 * negative number in case of error.
 */
inline int validate_str_ev_policy(const char *s)
{
	return validate_str_val_mapping(s, eviction_policy_names, -1);
}

inline const char *eviction_policy_to_name(uint8_t policy)
{
	return val_to_short_name(policy, eviction_policy_names, "Unknown");
}

inline const char *cache_mode_to_name(uint8_t cache_mode)
{
	return val_to_short_name(cache_mode, cache_mode_names, "Unknown");
}

static inline const char *cache_mode_to_name_long(uint8_t cache_mode)
{
	return val_to_long_name(cache_mode, cache_mode_names, "??");
}

inline int validate_str_cache_mode(const char *s)
{
	return validate_str_val_mapping(s, cache_mode_names, -1);
}

inline int validate_str_cln_policy(const char *s)
{
	return validate_str_val_mapping(s, cleaning_policy_names, -1);
}

inline const char *cleaning_policy_to_name(uint8_t policy)
{
	return val_to_short_name(policy, cleaning_policy_names, "Unknown");
}

const char *metadata_mode_to_name(uint8_t metadata_mode)
{
	return val_to_short_name(metadata_mode, metadata_mode_names, "Invalid");
}

const char *seq_cutoff_policy_to_name(uint8_t seq_cutoff_policy)
{
	return val_to_short_name(seq_cutoff_policy,
			seq_cutoff_policy_names, "Invalid");
}

inline void metadata_memory_footprint(uint64_t size, float *footprint,
	const char **units)
{
	float factor = 1;
	static const char *units_names[] = {"B", "KiB", "MiB", "GiB", "TiB"};
	uint32_t i;

	for (i = 0; i < sizeof(units_names) / sizeof(units_names[0]); i++) {
		*footprint = ((float) (size)) / factor;
		*units = units_names[i];

		if (*footprint < 1024.0) {
			break;
		}

		factor *= 1024;
	}
}

/* Returns one of or combination of STATS_FILTER values
 * or STATS_FILTER_INVALID in case of error.
 */
int validate_str_stats_filters(const char* s)
{
	return validate_str_val_mapping_multi(s, stats_filters_names,
					      STATS_FILTER_INVALID);
}

/* Returns one of OUTPUT_FORMAT values
 * or OUTPUT_FORMAT_INVALID in case of error.
 */
int validate_str_output_format(const char* s)
{
	return validate_str_val_mapping(s, output_formats_names,
					OUTPUT_FORMAT_INVALID);
}

/* Returns one of METADATA_MODE values
 * or METADATA_MODE_INVALID in case of error.
 */
int validate_str_metadata_mode(const char* s)
{
	return validate_str_val_mapping(s, metadata_modes_names,
					METADATA_MODE_INVALID);
}

void print_err(int error_code)
{
	const char *msg = cas_strerr(error_code);

	if (msg)
		cas_printf(LOG_ERR, "%s\n", msg);
}

const char *get_cache_state_name(int cache_state)
{
	int i;
	/* iterate over states in reverse order, so that combined states "running&stopping"
	 * would be described as "stopping" */
	for (i = ocf_cache_state_max - 1; i >= 0; --i) {
		if ((cache_state & (1 << i)) > 0) {
			return cache_states_name[i];
		}
	}
	return NOT_RUNNING_STATE;
}

const char *get_core_state_name(int core_state)
{
	if (core_state < 0 || core_state >= ocf_core_state_max)
		return "Invalid";

	return core_states_name[core_state];
}


/* check if device is atomic and print information about potential slow start */
void print_slow_atomic_cache_start_info(const char *device_path)
{
	struct kcas_cache_check_device cmd_info;
	int ret = _check_cache_device(device_path, &cmd_info);

	if (!ret && cmd_info.format_atomic) {
		cas_printf(LOG_INFO,
			"Starting new cache instance on a device with atomic metadata format may take \n"
			"several minutes depending on device model and size.\n");
	}
}

/**
  * @brief get special device file path (/dev/sdX) for disk.
  */
int get_dev_path(const char* disk, char* buf, size_t num)
{
	char *path;
	int err;

	path = realpath(disk, NULL);
	if (!path)
		return FAILURE;

	err = strncpy_s(buf, num, path, MAX_STR_LEN);

	free(path);
	return err;
}

int get_core_info(int fd, int cache_id, int core_id, struct kcas_core_info *info)
{
	memset(info, 0, sizeof(*info));
	info->cache_id = cache_id;
	info->core_id = core_id;

	if (ioctl(fd, KCAS_IOCTL_CORE_INFO, info) < 0) {
		return FAILURE;
	}

	/* internally use device special file path to describe core */
	if (get_dev_path(info->core_path_name,
			 info->core_path_name,
			 sizeof(info->core_path_name))) {
		cas_printf(LOG_WARNING, "WARNING: Can not resolve path to core "
			"%d from cache %d. By-id path will be shown for that core.\n",
			core_id, cache_id);
	}

	return SUCCESS;
}

static int get_core_device(int cache_id, int core_id, struct core_device *core)
{
	int fd;
	struct kcas_core_info cmd_info;

	if (!core)
		return FAILURE;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (get_core_info(fd, cache_id, core_id, &cmd_info)) {
		cas_printf(LOG_ERR, "Error while retrieving stats\n");
		print_err(cmd_info.ext_err_code);
		close(fd);
		return FAILURE;
	}

	close(fd);

	core->id = core_id;
	core->cache_id = cache_id;
	strncpy_s(core->path, sizeof(core->path), cmd_info.core_path_name,
		sizeof(cmd_info.core_path_name));
	memcpy_s(&core->info, sizeof(core->info),
		&cmd_info, sizeof(cmd_info));

	return SUCCESS;
}

int get_cache_count(int fd)
{
	struct kcas_cache_count cmd;

	if (ioctl(fd, KCAS_IOCTL_GET_CACHE_COUNT, &cmd) < 0)
		return 0;

	return cmd.cache_count;
}

int *get_cache_ids(int *caches_count)
{
	int i, fd, status;
	struct kcas_cache_list cache_list;
	int *cache_ids = NULL;
	int count, chunk_size;

	fd = open_ctrl_device();
	if (fd == -1)
		return NULL;

	count = get_cache_count(fd);

	if (count <= 0) {
		goto error_out;
	}

	cache_ids = malloc(count * sizeof(*cache_ids));
	if (cache_ids == NULL) {
		goto error_out;
	}

	memset(&cache_list, 0, sizeof(cache_list));

	*caches_count = 0;

	chunk_size = CACHE_LIST_ID_LIMIT;
	cache_list.id_position = 0;
	cache_list.in_out_num = chunk_size;
	do {
		if ((status = ioctl(fd, KCAS_IOCTL_LIST_CACHE, &cache_list)) < 0) {
			if (errno != EINVAL) {
				cas_printf(LOG_ERR, "Error while retrieving cache properties %d %d\n",
					errno, status);
				free(cache_ids);
				cache_ids = NULL;
				*caches_count = 0;
				break;
			}
		}

		/* iterate through id table and get status */
		for (i = 0; i < cache_list.in_out_num; i++) {
			cache_ids[(*caches_count)] = cache_list.cache_id_tab[i];
			(*caches_count)++;
			if (*caches_count >= count) {
				break;
			}
		}

		cache_list.id_position += chunk_size;
	} while (cache_list.in_out_num >= chunk_size); /* repeat until there is no more devices on the list */

error_out:
	close(fd);
	return cache_ids;
}

/**
 * @brief function returns pointer to cache device given cache_info structure.
 *
 * structure is mallocated, and therefore it is callers responsibility to free it.
 *
 * @return valid pointer to a structure or NULL if error happened
 */
struct cache_device *get_cache_device(const struct kcas_cache_info *info)
{
	int core_id, cache_id, ret;
	struct cache_device *cache;
	struct core_device core;
	cache_id = info->cache_id;
	size_t cache_size;

	cache_size = sizeof(*cache);
	cache_size += info->info.core_count * sizeof(cache->cores[0]);

	cache = (struct cache_device *) malloc(cache_size);
	if (NULL == cache) {
		return NULL;
	}

	cache->core_count = 0;
	cache->expected_core_count = info->info.core_count;
	cache->id = cache_id;
	cache->state = info->info.state;
	strncpy_s(cache->device, sizeof(cache->device), info->cache_path_name,
		  strnlen_s(info->cache_path_name, sizeof(info->cache_path_name)));
	cache->mode = info->info.cache_mode;
	cache->dirty = info->info.dirty;
	cache->flushed = info->info.flushed;
	cache->eviction_policy = info->info.eviction_policy;
	cache->cleaning_policy = info->info.cleaning_policy;
	cache->size = info->info.cache_line_size;

	if ((info->info.state & (1 << ocf_cache_state_running)) == 0) {
		return cache;
	}

	for (cache->core_count = 0; cache->core_count < info->info.core_count; ++cache->core_count) {
		core_id = info->core_id[cache->core_count];

		ret = get_core_device(cache_id, core_id, &core);
		if (0 != ret) {
			break;
		} else {
			memcpy_s(&cache->cores[cache->core_count],
				sizeof(cache->cores[cache->core_count]),
				&core, sizeof(core));
		}
	}

	return cache;
}

/**
 * @brief function returns pointer to cache device given cache id and fd of /dev/cas_ctrl
 *
 * structure is mallocated, and therefore it is callers responsibility to free it.
 *
 * @param fd valid file descriptor to /dev/cas_ctrl
 * @param cache_id cache id (1...)
 * @return valid pointer to a structure or NULL if error happened
 */
struct cache_device *get_cache_device_by_id_fd(int cache_id, int fd)
{
	struct kcas_cache_info cmd_info;

	memset(&cmd_info, 0, sizeof(cmd_info));
	cmd_info.cache_id = cache_id;

	if (ioctl(fd, KCAS_IOCTL_CACHE_INFO, &cmd_info) < 0) {
		if (errno != EINVAL)
			return NULL;
	}

	return get_cache_device(&cmd_info);
}

void free_cache_devices_list(struct cache_device **caches, int caches_count)
{
	int i;
	for (i = 0; i < caches_count; ++i) {
		free(caches[i]);
		caches[i] = NULL;
	}
	free(caches);
}

struct cache_device **get_cache_devices(int *caches_count)
{
	int i, fd, status, chunk_size, count;
	struct kcas_cache_list cache_list;
	struct cache_device **caches = NULL;
	struct cache_device *tmp_cache;

	*caches_count = -1;

	fd = open_ctrl_device();
	if (fd == -1)
		return NULL;

	*caches_count = count = get_cache_count(fd);
	if (count <= 0) {
		goto error_out;
	}

	memset(&cache_list, 0, sizeof(cache_list));
	caches = malloc(count * sizeof(*caches));

	if (NULL == caches) {
		*caches_count = -1;
		goto error_out;
	}

	(*caches_count) = 0;

	chunk_size = CACHE_LIST_ID_LIMIT;

	cache_list.id_position = 0;
	cache_list.in_out_num = chunk_size;
	do {
		if ((status = ioctl(fd, KCAS_IOCTL_LIST_CACHE, &cache_list)) < 0) {
			if (errno != EINVAL) {
				cas_printf(LOG_ERR, "Error while retrieving cache properties %d %d\n",
					errno, status);
				free_cache_devices_list(caches, *caches_count);
				*caches_count = -1;
				caches = NULL;
				goto error_out;
			}
		}

		/* iterate through id table and get status */
		for (i = 0; i < cache_list.in_out_num; i++) {
			if ((tmp_cache = get_cache_device_by_id_fd(cache_list.cache_id_tab[i], fd)) == NULL) {
				cas_printf(LOG_ERR, "Failed to retrieve cache information!\n");
				continue;
			}
			caches[(*caches_count)] = tmp_cache;
			(*caches_count)++;
			if (*caches_count >= count) {
				break;
			}
		}
		cache_list.id_position += chunk_size;
	} while (cache_list.in_out_num >= chunk_size); /* repeat until there is no more devices on the list */

error_out:
	close(fd);
	return caches;
}

int caches_compare(const void *a, const void *b)
{
	int a_id = (*(struct cache_device**)a)->id;
	int b_id = (*(struct cache_device**)b)->id;
	return a_id - b_id;
}

int check_cache_already_added(const char *cache_device) {
	struct cache_device **caches, *curr_cache;
	int caches_count, i;

	caches = get_cache_devices(&caches_count);

	if (NULL == caches) {
		return SUCCESS;
	}

	for (i = 0; i < caches_count; ++i) {
		curr_cache = caches[i];
		if (0 == strncmp(cache_device, curr_cache->device, MAX_STR_LEN)) {
			free_cache_devices_list(caches, caches_count);
			return FAILURE;
		}
	}

	free_cache_devices_list(caches, caches_count);

	return SUCCESS;
}

static void check_cache_scheduler(const char *cache_device, const char *elv_name)
{
	if (strnlen_s(elv_name, MAX_ELEVATOR_NAME) == 3 &&
	    !strncmp(elv_name, "cfq", 3)) {
		cas_printf(LOG_INFO,
			   "I/O scheduler for cache device %s is %s. This could cause performance drop.\n"
			   "Consider switching I/O scheduler to deadline or noop.\n",
			   cache_device, elv_name);
	}
}

int start_cache(ocf_cache_id_t cache_id, unsigned int cache_init,
		const char *cache_device, ocf_cache_mode_t cache_mode,
		ocf_eviction_t eviction_policy_type,
		ocf_cache_line_size_t line_size, int force)
{
	int fd = 0;
	struct kcas_start_cache cmd;
	struct cache_device **caches;
	struct cache_device *cache;
	int i, status, caches_count;
	double min_free_ram_gb;

	/* check if cache device given exists */
	fd = open(cache_device, 0);
	if (fd < 0) {
		cas_printf(LOG_ERR, "Device %s not found.\n", cache_device);
		return FAILURE;
	}
	close(fd);

	if (cache_init == CACHE_INIT_NEW)
		print_slow_atomic_cache_start_info(cache_device);

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (cache_id == 0) {
		cache_id = 1;
		caches = get_cache_devices(&caches_count);
		if (caches != NULL) {
			psort(caches, caches_count, sizeof(struct cache_device*), caches_compare);
			for (i = 0; i < caches_count; ++i) {
				if (caches[i]->id == cache_id) {
					cache_id += 1;
				}
			}

			free_cache_devices_list(caches, caches_count);
		}
	}

	memset(&cmd, 0, sizeof(cmd));

	cmd.cache_id = cache_id;
	cmd.init_cache = cache_init;
	strncpy_s(cmd.cache_path_name,
		  sizeof(cmd.cache_path_name),
		  cache_device,
		  strnlen_s(cache_device,
			    sizeof(cmd.cache_path_name)));
	cmd.caching_mode = cache_mode;
	cmd.eviction_policy = eviction_policy_type;
	cmd.line_size = line_size;
	cmd.force = (uint8_t)force;

	if (run_ioctl_interruptible(fd, KCAS_IOCTL_START_CACHE, &cmd,
			"Starting cache", cache_id, OCF_CORE_ID_INVALID) < 0) {
		close(fd);
		if (cmd.ext_err_code == OCF_ERR_NO_FREE_RAM) {
			min_free_ram_gb = cmd.min_free_ram;
			min_free_ram_gb /= GiB;

			cas_printf(LOG_ERR, "Not enough free RAM.\n"
					"You need at least %0.2gGB to start cache"
					" with cache line size equal %llukB.\n",
					min_free_ram_gb, line_size / KiB);

			if (64 * KiB > line_size)
				cas_printf(LOG_ERR, "Try with greater cache line size.\n");

			return FAILURE;
		} else {
			cas_printf(LOG_ERR, "Error inserting cache %d\n", cache_id);
			if (OCF_ERR_NOT_OPEN_EXC == cmd.ext_err_code &&
				FAILURE == check_cache_already_added(cache_device)) {
				cas_printf(LOG_ERR, "Cache device '%s' is already used as cache.\n",
				cache_device);
			} else {
				print_err(cmd.ext_err_code);
			}
			return FAILURE;
		}
	}

	if (!cmd.metadata_mode_optimal)
		cas_printf(LOG_NOTICE, "Selected metadata mode is not optimal for device %s.\n"
			"You can improve cache performance by formating your device\n"
			"to use optimal metadata mode with following command:\n"
			"casadm --nvme --format atomic --device %s\n",
			cache_device, cache_device);

	check_cache_scheduler(cache_device,
			      cmd.cache_elevator);

	status = SUCCESS;

	for (i = 0; i < CORE_ADD_MAX_TIMEOUT; ++i) {
		cache = get_cache_device_by_id_fd(cache_id, fd);
		status = FAILURE;

		if (cache == NULL) {
			break;
		}

		if (cache->core_count == cache->expected_core_count) {
			if (cache->state & (1 << ocf_cache_state_incomplete)) {
				cas_printf(LOG_WARNING, "WARNING: Cache is in incomplete state - at least one core is inactive\n");
			}
			status = SUCCESS;
			free(cache);
			cache = NULL;
			break;
		}

		free(cache);
		cache = NULL;

		sleep(1);
	}

	close(fd);

	if (status == SUCCESS) {
		cas_printf(LOG_INFO, "Successfully added cache instance %u\n", cache_id);
	} else {
		cas_printf(LOG_ERR, "Failed to start cache\n");
		return FAILURE;
	}

	return SUCCESS;
}

int stop_cache(ocf_cache_id_t cache_id, int flush)
{
	int fd = 0;
	struct kcas_stop_cache cmd;

	/* don't even attempt ioctl if filesystem is mounted */
	if (check_if_mounted(cache_id, CHECK_IF_CACHE_IS_MOUNTED) == FAILURE) {
		return FAILURE;
	}

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	memset(&cmd, 0, sizeof(cmd));
	cmd.cache_id = cache_id;
	cmd.flush_data = flush;

	if(run_ioctl_interruptible(fd, KCAS_IOCTL_STOP_CACHE, &cmd, "Stopping cache",
			cache_id, OCF_CORE_ID_INVALID) < 0) {
		close(fd);
		if (OCF_ERR_FLUSHING_INTERRUPTED == cmd.ext_err_code) {
			cas_printf(LOG_ERR, "You have interrupted stopping of cache. CAS continues\n"
				"to operate normally. If you want to stop cache without fully\n"
				"flushing dirty data, use '-n' option.\n");
			return INTERRUPTED;
		} else {
			cas_printf(LOG_ERR, "Error while removing cache %d\n", cache_id);
			print_err(cmd.ext_err_code);
			return FAILURE;
		}
	}
	close(fd);
	return SUCCESS;
}

/*
 * @brief check caching mode
 * @param[in] ctrl_fd file descriptor of opened control utility
 * @param[in] cache_id id of cache device
 * @param[out] mode mode identifier as integer
 * @return exit code of successful completion is 0; nonzero exit code means failure
 */
int get_cache_mode(int ctrl_fd, unsigned int cache_id, int *mode)
{
	struct kcas_cache_info cmd_info;

	memset(&cmd_info, 0, sizeof(cmd_info));
	cmd_info.cache_id = cache_id;

	if (ioctl(ctrl_fd, KCAS_IOCTL_CACHE_INFO, &cmd_info) < 0)
		return FAILURE;

	*mode = cmd_info.info.cache_mode;
	return SUCCESS;
}

int set_cache_mode(unsigned int cache_mode, unsigned int cache_id, int flush)
{
	int fd = 0;
	int orig_mode;
	struct kcas_set_cache_state cmd;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (get_cache_mode(fd, cache_id, &orig_mode)) {
		cas_printf(LOG_ERR, "Error while retrieving cache properties.\n");
		close(fd);
		return FAILURE;
	}

	/* if flushing mode is undefined, set it to default (but only if original mode is write back mode) */
	if (-1 == flush) {
		if (ocf_cache_mode_wb == orig_mode) {
			cas_printf(LOG_ERR, "Error: Required parameter (‘--flush-cache’) was not specified.\n");
			close(fd);
			return FAILURE;
		} else {
			flush=NO;
		}
	}

	if (ocf_cache_mode_wb == orig_mode) {
		if (1 == flush) {
			cas_printf(LOG_INFO, "CAS is currently flushing dirty data to primary storage devices.\n");
		} else {
			cas_printf(LOG_INFO, "CAS is currently migrating from Write-Back to %s mode.\n"
				"Dirty data are being flushed to primary storage device in background.\n"
				"Please find flushing progress via list caches command (‘casadm -L’) or\n"
				"via statistics command (‘casadm -P’).\n",
				cache_mode_to_name_long(cache_mode));
		}
	}
	memset(&cmd, 0, sizeof(cmd));
	cmd.cache_id = cache_id;
	cmd.caching_mode = cache_mode;
	cmd.flush_data = flush;

	if (run_ioctl_interruptible(fd, KCAS_IOCTL_SET_CACHE_STATE, &cmd, "Setting mode",
			cache_id, OCF_CORE_ID_INVALID) < 0) {
		close(fd);
		if (OCF_ERR_FLUSHING_INTERRUPTED == cmd.ext_err_code) {
			assert(flush);
			cas_printf(LOG_ERR,
				"Interrupted flushing of dirty data. Software prevented switching\n"
				"of cache mode. If you want to switch cache mode immediately, use\n"
				"'--flush-cache no' parameter.\n");
			return INTERRUPTED;
		} else {
			cas_printf(LOG_ERR, "Error while setting cache state for cache %d\n",
				cache_id);
			print_err(cmd.ext_err_code);
			return FAILURE;
		}
	}
	close(fd);

	return SUCCESS;
}

static void print_param(FILE *intermediate_file, struct cas_param *param)
{
	if (param->value_names) {
		fprintf(intermediate_file, "%s%s,%s\n", TAG(TABLE_ROW),
			param->name, param->value_names[param->value]);
	} else {
		char *unit = param->unit ?: "";
		fprintf(intermediate_file, "%s%s,%u %s\n", TAG(TABLE_ROW),
			param->name, param->value, unit);
	}
	fflush(intermediate_file);
}

int core_params_set(unsigned int cache_id, unsigned int core_id,
		struct cas_param *params)
{
	int cache_mode = ocf_cache_mode_none;
	struct kcas_set_core_param cmd = {0};
	int fd = 0;
	int i;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (get_cache_mode(fd, cache_id, &cache_mode)) {
		close(fd);
		return FAILURE;
	}

	if (ocf_cache_mode_pt == cache_mode) {
		cas_printf(LOG_WARNING, "Changing parameters for core in Pass-Through mode."
				"New values will be saved but will not be effective"
				"until switching to another cache mode.\n");
	}

	for (i = 0; params[i].name; ++i) {
		if (!params[i].select)
			continue;

		cmd.cache_id = cache_id;
		cmd.core_id = core_id;
		cmd.param_id = i;
		cmd.param_value = params[i].value;

		if (run_ioctl(fd, KCAS_IOCTL_SET_CORE_PARAM, &cmd) < 0) {
			close(fd);
			return FAILURE;
		}
	}

	close(fd);
	return SUCCESS;
}

int core_params_get(unsigned int cache_id, unsigned int core_id,
		struct cas_param *params, unsigned int output_format)
{
	struct kcas_get_core_param cmd = {0};
	FILE *intermediate_file[2];
	int fd = 0;
	int i;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (create_pipe_pair(intermediate_file)) {
		cas_printf(LOG_ERR,"Failed to create unidirectional pipe.\n");
		close(fd);
		return FAILURE;
	}

	fprintf(intermediate_file[1], TAG(TABLE_HEADER) "Parameter name,Value\n");
	fflush(intermediate_file[1]);

	for (i = 0; params[i].name; ++i) {
		if (!params[i].select)
			continue;

		cmd.cache_id = cache_id;
		cmd.core_id = core_id;
		cmd.param_id = i;

		if (run_ioctl(fd, KCAS_IOCTL_GET_CORE_PARAM, &cmd) < 0) {
			if (cmd.ext_err_code == OCF_ERR_CACHE_NOT_EXIST)
				cas_printf(LOG_ERR, "Cache id %d not running\n", cache_id);
			else if (cmd.ext_err_code == OCF_ERR_CORE_NOT_AVAIL)
				cas_printf(LOG_ERR, "Core id %d not available\n", core_id);
			else
				cas_printf(LOG_ERR, "Can't get parameters\n");
			fclose(intermediate_file[0]);
			fclose(intermediate_file[1]);
			close(fd);
			return FAILURE;
		}

		if (params[i].transform_value)
			params[i].value = params[i].transform_value(cmd.param_value);
		else
			params[i].value = cmd.param_value;

		print_param(intermediate_file[1], &params[i]);
	}

	close(fd);

	fclose(intermediate_file[1]);
	stat_format_output(intermediate_file[0], stdout, output_format);
	fclose(intermediate_file[0]);

	return SUCCESS;
}

int cache_params_set(unsigned int cache_id, struct cas_param *params)
{
	int cache_mode = ocf_cache_mode_none;
	struct kcas_set_cache_param cmd = {0};
	int fd = 0;
	int i;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (get_cache_mode(fd, cache_id, &cache_mode)) {
		close(fd);
		return FAILURE;
	}

	if (ocf_cache_mode_pt == cache_mode) {
		cas_printf(LOG_WARNING, "Changing parameters for core in Pass-Through mode."
				" New values will be saved but will not be effective"
				" until switching to another cache mode.\n");
	}

	for (i = 0; params[i].name; ++i) {
		if (!params[i].select)
			continue;

		cmd.cache_id = cache_id;
		cmd.param_id = i;
		cmd.param_value = params[i].value;

		if (run_ioctl(fd, KCAS_IOCTL_SET_CACHE_PARAM, &cmd) < 0) {
			close(fd);
			return FAILURE;
		}
	}

	close(fd);
	return SUCCESS;
}

int cache_get_param(unsigned int cache_id, unsigned int param_id,
		struct cas_param *param)
{
	struct kcas_get_cache_param cmd = { 0 };
	int fd = 0;

	if (param_id >= cache_param_id_max)
		return FAILURE;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	cmd.param_id = param_id;
	cmd.cache_id = cache_id;

	if (run_ioctl(fd, KCAS_IOCTL_GET_CACHE_PARAM, &cmd) < 0) {
		if (cmd.ext_err_code == OCF_ERR_CACHE_NOT_EXIST)
			cas_printf(LOG_ERR, "Cache id %d not running\n", cache_id);
		else
			cas_printf(LOG_ERR, "Can't get parameters\n");
		close(fd);
		return FAILURE;
	}

	if (param->transform_value)
		param->value = param->transform_value(cmd.param_value);
	else
		param->value = cmd.param_value;

	close(fd);

	return SUCCESS;
}

int cache_params_get(unsigned int cache_id, struct cas_param *params,
		unsigned int output_format)
{
	struct kcas_get_cache_param cmd = {0};
	FILE *intermediate_file[2];
	int fd = 0;
	int i;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (create_pipe_pair(intermediate_file)) {
		cas_printf(LOG_ERR,"Failed to create unidirectional pipe.\n");
		close(fd);
		return FAILURE;
	}

	fprintf(intermediate_file[1], TAG(TABLE_HEADER) "Parameter name,Value\n");
	fflush(intermediate_file[1]);

	for (i = 0; params[i].name; ++i) {
		if (!params[i].select)
			continue;

		cmd.cache_id = cache_id;
		cmd.param_id = i;

		if (run_ioctl(fd, KCAS_IOCTL_GET_CACHE_PARAM, &cmd) < 0) {
			if (cmd.ext_err_code == OCF_ERR_CACHE_NOT_EXIST)
				cas_printf(LOG_ERR, "Cache id %d not running\n", cache_id);
			else
				cas_printf(LOG_ERR, "Can't get parameters\n");
			fclose(intermediate_file[0]);
			fclose(intermediate_file[1]);
			close(fd);
			return FAILURE;
		}

		if (params[i].transform_value)
			params[i].value = params[i].transform_value(cmd.param_value);
		else
			params[i].value = cmd.param_value;

		print_param(intermediate_file[1], &params[i]);
	}

	close(fd);

	fclose(intermediate_file[1]);
	stat_format_output(intermediate_file[0], stdout, output_format);
	fclose(intermediate_file[0]);

	return SUCCESS;
}

int check_core_already_cached(const char *core_device) {
	struct cache_device **caches, *curr_cache;
	struct core_device *curr_core;
	int caches_count, i, j;
	char core_device_path[MAX_STR_LEN];

	if (get_dev_path(core_device, core_device_path, sizeof(core_device_path)))
		return SUCCESS;

	caches = get_cache_devices(&caches_count);

	if (NULL == caches) {
		return SUCCESS;
	}

	for (i = 0; i < caches_count; ++i) {
		curr_cache = caches[i];
		for (j = 0; j < curr_cache->core_count; ++j) {
			curr_core = &curr_cache->cores[j];
			if (0 ==
			strncmp(core_device_path, curr_core->path, MAX_STR_LEN)) {
				free_cache_devices_list(caches, caches_count);
				return FAILURE;
			}
		}
	}

	free_cache_devices_list(caches, caches_count);

	return SUCCESS;
}

/**
 * @brief convert string to int
 *
 * @param[in] start string beginning
 * @param[out] end optional pointer to character past the end of integer
 * @param[out] val integer value
 * @return true in case of success, false in case of failure
 */
bool str_to_int(const char* start, char** end, int *val)
{
	long int _val;
	char *_end = (char *)start;

	_val = strtol(start, &_end, 10);

	if (_end == start) {
		/* no integer found */
		return false;
	}

	if (_val < INT_MIN || _val > INT_MAX) {
		/* value out of int range */
		return false;
	}

	/* 0 might indicate strtol error, so try to check if the
	   input is really 0. This might not be  bullet-proof, but enough
	   for us. */
	if (_val == 0 && *(_end - 1) != '0') {
		/* doesn't look like 0, more likely a parsing error */
		return false;
	}

	*val = (int)_val;
	if (end)
		*end = _end;

	return true;
}


static bool get_core_cache_id_from_string(char *str,
	int *cache_id, int *core_id)
{
	char *end;

	if (!str_to_int(str, &end, cache_id))
		return false;

	if (*end != '-') {
		/* invalid separator */
		return false;
	}

	if (!str_to_int(end + 1, NULL, core_id))
		return false;

	return true;
}

int get_inactive_core_count(const struct kcas_cache_info *cache_info)
{
	struct cache_device *cache;
	int inactive_cores = 0;
	int i;

	cache = get_cache_device(cache_info);
	if (!cache)
		return -1;

	for (i = 0; i < cache->core_count; i++) {
		if (cache->cores[i].info.state == ocf_core_state_inactive)
			inactive_cores++;
	}

	free(cache);

	return inactive_cores;
}


/**
 * @brief check for illegal recursive core configuration
 *
 * Function returns 1 (FAILURE/true) if it detects that adding core_device to
 *   cache_id will  result in illegal multilevel configuration.
 * Function returns 0 (SUCCESS/false) if it detects that it is fine to add
 *   core_device to cache_id  and it will NOT result in illegal multilevel
 *   configuration.
 *
 * Here is example of such illegal configuration:
 *
 * type      id   disk             device
 * cache     1    /dev/sdc1        -
 * +core     1    /dev/sdd1        /dev/cas1-1
 * +core     2    /dev/cas1-1	   /dev/cas1-2
 *
 * Here is another example of illegal configuration (notice that it is indirect, and hence
 * whole multilevel caching hierarchy has to be parsed)
 *
 * type      id   disk              device
 * cache     1    /dev/sdc1         -
 * +core     1    /dev/sdd1         /dev/cas1-1
 * +core     2    /dev/cas2-1	    /dev/cas1-2
 * cache     2    /dev/sdc2         -
 * +core     1    /dev/cas1-1	    /dev/cas2-1
 *
 * (in above example adding core 2 to cache shouldn't be allowed as this is effectively adding same
 * disk device (/dev/sdd1) to the same cache (/dev/sdc1) twice).
 *
 * @param cache_id cache to which new core is being added
 * @param core_device path to a core device that is being added
 * @param fd valid file descriptor for /dev/cas_ctrl device
 * @return 0 if check is successful and on illegal recursion is detected.
 *		1 if illegal config detected.
 */
int illegal_recursive_core(unsigned int cache_id, const char *core_device, int core_path_size, int fd)
{
	char tmp_path[MAX_STR_LEN];
	char core_path[MAX_STR_LEN];	/* extracted actual path */
	int dev_core_id, dev_cache_id;  /* cache_id and core_id for currently
					 * analyzed device */
	struct stat st_buf;
	int i;
	static const char cas_pattern[] = "/dev/cas";
	struct cache_device *cache; /*structure containing data on cache device*/

	while (true) {
		/*
		 * if core_device is an cas device (or a symlink to
		 * cas device) check if its cache device is cache id. if
		 * it is, return an error, as this will lead to illegal
		 * multilevel configuration.
		 */
		if (lstat(core_device, &st_buf)) {
			cas_printf(LOG_ERR, "ERROR: lstat failed for %s.\n",
				   core_device);
			return FAILURE;
		}

		if (get_dev_path(core_device, core_path, sizeof(core_path)))
			return FAILURE;

		/* if core_path does NOT begin with /dev/cas, report success
		 * as it certainly is not case of */
		if (strncmp(cas_pattern, core_path, sizeof(cas_pattern) - 1)) {
			return SUCCESS;
		}

		if (!get_core_cache_id_from_string(
				core_path + sizeof(cas_pattern) - 1,
				&dev_cache_id,
				&dev_core_id)) {
			cas_printf(LOG_ERR, "Failed to extract core/cache "
				   "id from %s path\n", core_path);
			return FAILURE;
		}

		if (dev_cache_id == cache_id) {
			cas_printf(LOG_ERR, "Core device '%s' is already cached"
				   " on cache device %d. - "
				   "illegal multilevel caching configuration.\n",
				   core_device, cache_id);
			return FAILURE;
		}
		/* possibly legal multilevel caching configuration - do one more
		 * iteration of this loop*/

		/* get underlying core device of dev_cache_id-dev_core_id */
		cache = get_cache_device_by_id_fd(dev_cache_id, fd);

		if (!cache) {
			cas_printf(LOG_ERR, "Failed to extract statistics for "
				   "cache device %d\n", dev_cache_id);
			return FAILURE;
		}

		/* lookup for record for appropriate core */
		for (i = 0; i != cache->core_count ; ++i) {
			if (cache->cores[i].id == dev_core_id) {
				strncpy_s(tmp_path, sizeof(tmp_path),
					  cache->cores[i].path,
					  strnlen_s(cache->cores[i].path,
						    sizeof(cache->cores[i].path)));
				core_device = tmp_path;
				break;
			}
		}

		/* make sure that loop above resulted in correct assignment */
		if (i == cache->core_count) {
			cas_printf(LOG_ERR, "Failed to extract statistics for "
				   "core device %d-%d. Does it exist?\n",
				   dev_cache_id, dev_core_id);
			free(cache);
			return FAILURE;
		}

		free(cache);
	}
}

/* Indicate whether given entry in /dev/disk/by-id should be ignored -
   we ignore software created links like 'lvm-' since these can point to
   both CAS exported object and core device depending on initialization order.
*/
static bool dev_link_blacklisted(const char* entry)
{
	static const char* const prefix_blacklist[] = {"lvm"};
	static const unsigned count = ARRAY_SIZE(prefix_blacklist);
	const char* curr;
	unsigned i;

	for (i = 0; i < count; i++) {
		curr = prefix_blacklist[i];
		if (!strncmp(entry, curr, strnlen_s(curr, MAX_STR_LEN)))
			return true;
	}

	return false;
}

/* get device link starting with /dev/disk/by-id */
static int get_dev_link(const char* disk, char* buf, size_t num)
{
	static const char dev_by_id_dir[] = "/dev/disk/by-id";
	int err;
	struct dirent *entry;
	DIR* dir;
	char disk_dev[MAX_STR_LEN];  /* input disk device file */
	char dev_by_id[MAX_STR_LEN]; /* current device path by id */
	char curr_dev[MAX_STR_LEN];  /* current device file - compared against disk_dev[] */
	int n;

	dir = opendir(dev_by_id_dir);
	if (!dir) {
		/* no disk available by id? */
		cas_printf(LOG_WARNING, "Unable to open disk alias directory.\n");
		return FAILURE;
	}

	if (get_dev_path(disk, disk_dev, sizeof(disk_dev))) {
		err = FAILURE;
		goto close_dir;
	}

	err = FAILURE;
	while (err != SUCCESS && (entry = readdir(dir))) {
		/* check if link is blacklisted */
		if (dev_link_blacklisted(entry->d_name))
			continue;

		/* construct device-by-id path for current device */
		n = snprintf(dev_by_id, sizeof(dev_by_id), "%s/%s",
				dev_by_id_dir, entry->d_name);
		if (n < 0 || n >= sizeof(dev_by_id)) {
			cas_printf(LOG_WARNING,
				"Error constructing disk device by-link path.\n");
			continue;
		}
		/* get device path for current device */
		if (get_dev_path(dev_by_id, curr_dev, sizeof(curr_dev))) {
			/* it's normal to have stale links in /dev/ - no log */
			continue;
		}
		/* compare current device path against disk device path */
		if (!strncmp(disk_dev, curr_dev, sizeof(curr_dev))) {
			if (n >= num) {
				cas_printf(LOG_WARNING, "Buffer to short to store device link.\n");
			} else {
				strncpy_s(buf, num, dev_by_id, sizeof(dev_by_id));
				err = SUCCESS;
			}
		}
	}

close_dir:
	closedir(dir);

	return err;
}

static int set_core_path(char *path, const char *core_device, size_t len)
{
	/* attempt to get disk device path by id */
	if (get_dev_link(core_device, path, len) == SUCCESS)
		return SUCCESS;

	/* .. if this failed, try to get standard /dev/sd* path */
	if (get_dev_path(core_device, path, len) == SUCCESS)
		return SUCCESS;

	/* if everything else failed - fall back to user-provided path */
	if (!strncpy_s(path, len, core_device, strnlen_s(core_device, MAX_STR_LEN)))
		return SUCCESS;

	return FAILURE;
}

int add_core(unsigned int cache_id, unsigned int core_id, const char *core_device,
		int try_add, int update_path)
{
	int fd = 0, user_core_path_size;
	struct kcas_insert_core cmd;
	struct stat query_core;
	const char *core_path;      /* core path sent down to kernel  */
	const char *user_core_path; /* core path provided by user */

	/* Check if core device provided is valid */
	fd = open(core_device, 0);
	if (fd < 0) {
		cas_printf(LOG_ERR, "Device %s not found.\n", core_device);
		return FAILURE;
	}
	close(fd);

	/* Check if the core device is a block device or a file */
	if (stat(core_device, &query_core)) {
		cas_printf(LOG_ERR, "Could not stat target core device %s!\n", core_device);
		return FAILURE;
	}

	if (!S_ISBLK(query_core.st_mode)) {
		cas_printf(LOG_ERR, "Core object %s is not supported!\n", core_device);
		return FAILURE;
	}

	memset(&cmd, 0, sizeof(cmd));
	if (set_core_path(cmd.core_path_name, core_device, MAX_STR_LEN) != SUCCESS) {
		cas_printf(LOG_ERR, "Failed to copy core path\n");
		return FAILURE;
	}

	user_core_path = core_device;
	user_core_path_size = strnlen_s(core_device, MAX_STR_LEN);
	core_path = cmd.core_path_name;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	/* check for illegal rec ursive caching config. */
	if (illegal_recursive_core(cache_id, user_core_path,
		user_core_path_size, fd)) {
		close(fd);
		return FAILURE;
	}

	cmd.cache_id = cache_id;
	cmd.core_id = core_id;
	cmd.try_add = try_add;
	cmd.update_path = update_path;

	if (ioctl(fd, KCAS_IOCTL_INSERT_CORE, &cmd) < 0) {
		close(fd);
		cas_printf(LOG_ERR, "Error while adding core device to cache instance %d\n",
			cache_id);
		if (OCF_ERR_NOT_OPEN_EXC == cmd.ext_err_code) {
			if (FAILURE == check_core_already_cached(core_path)) {
				cas_printf(LOG_ERR, "Core device '%s' is already cached.\n",
					user_core_path);
			} else {
				cas_printf(LOG_ERR, "Failed to open '%s' device"
				  " exclusively. Please close all applications "
				  "accessing it or unmount the device.\n",
				  user_core_path);
			}
		} else {
			print_err(cmd.ext_err_code);
		}
		return FAILURE;
	}
	close(fd);

	if (try_add) {
		cas_printf(LOG_INFO, "Successfully added device in try add mode %s\n", user_core_path);
	} else {
		core_id = cmd.core_id;

		cas_printf(LOG_INFO, "Successfully added core %u to cache instance %u\n", core_id, cache_id);
	}

	return SUCCESS;
}

int check_if_mounted(int cache_id, int core_id)
{
	FILE *mtab;
	struct mntent *mstruct;
	char dev_buf[80];
	int dev_buf_len;
	if (0 <= core_id) {
		/* verify if specific core is mounted */
		snprintf(dev_buf, sizeof(dev_buf), "/dev/cas%d-%d", cache_id, core_id);
	} else {
		/* verify if any core from given cache is mounted */
		snprintf(dev_buf, sizeof(dev_buf), "/dev/cas%d-", cache_id);
	}
	dev_buf_len = strnlen(dev_buf, sizeof(dev_buf));

	mtab = setmntent("/etc/mtab", "r");
	if (!mtab)
	{
		cas_printf(LOG_ERR, "Error while accessing /etc/mtab\n");
		return FAILURE;
	}

	while ((mstruct = getmntent(mtab)) != NULL) {
		/* mstruct->mnt_fsname is /dev/... block device path, not a mountpoint */
		if ((NULL != mstruct->mnt_fsname)
		    && (strncmp(mstruct->mnt_fsname, dev_buf, dev_buf_len) == 0)) {
			if (core_id<0) {
				cas_printf(LOG_ERR,
					   "Can't stop cache instance %d. Device %s is mounted!\n",
					   cache_id, mstruct->mnt_fsname);
			} else {
				cas_printf(LOG_ERR,
					   "Can't remove core %d from cache %d."
					   " Device %s is mounted!\n",
					   core_id, cache_id, mstruct->mnt_fsname);
			}
			return FAILURE;
		}
	}
	return SUCCESS;

}

int remove_core(unsigned int cache_id, unsigned int core_id,
		bool detach, bool force_no_flush)
{
	int fd = 0;
	struct kcas_remove_core cmd;

	/* don't even attempt ioctl if filesystem is mounted */
	if (SUCCESS != check_if_mounted(cache_id, core_id)) {
		return FAILURE;
	}

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	memset(&cmd, 0, sizeof(cmd));
	cmd.cache_id = cache_id;
	cmd.core_id = core_id;
	cmd.force_no_flush = force_no_flush;
	cmd.detach = detach;

	if (run_ioctl_interruptible(fd, KCAS_IOCTL_REMOVE_CORE, &cmd,
			"Removing core", cache_id, core_id) < 0) {
		close(fd);
		if (OCF_ERR_FLUSHING_INTERRUPTED == cmd.ext_err_code) {
			cas_printf(LOG_ERR, "You have interrupted removal of core. CAS continues to operate normally.\n");
			return INTERRUPTED;
		} else {
			cas_printf(LOG_ERR, "Error while removing core device %d from cache instance %d\n",
				   core_id, cache_id);
			print_err(cmd.ext_err_code);
			return FAILURE;
		}
	}
	close(fd);

	return SUCCESS;
}

int core_pool_remove(const char *core_device)
{
	struct kcas_core_pool_remove cmd;
	int fd;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	if (set_core_path(cmd.core_path_name, core_device, MAX_STR_LEN) != SUCCESS) {
		cas_printf(LOG_ERR, "Failed to copy core path\n");
		close(fd);
		return FAILURE;
	}

	if (ioctl(fd, KCAS_IOCTL_CORE_POOL_REMOVE, &cmd) < 0) {
		cas_printf(LOG_ERR, "Error while removing device %s from core pool\n",
				core_device);
		print_err(cmd.ext_err_code);
		close(fd);
		return FAILURE;
	}

	close(fd);
	return SUCCESS;
}

#define DIRTY_FLUSHING_WARNING "You have interrupted flushing of cache dirty data. CAS continues to operate\nnormally and dirty data that remains on cache device will be flushed by cleaning thread.\n"
int flush_cache(unsigned int cache_id)
{
	int fd = 0;
	struct kcas_flush_cache cmd;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	memset(&cmd, 0, sizeof(cmd));
	cmd.cache_id = cache_id;
	/* synchronous flag */
	if (run_ioctl_interruptible(fd, KCAS_IOCTL_FLUSH_CACHE, &cmd, "Flushing cache",
			cache_id, OCF_CORE_ID_INVALID) < 0) {
		close(fd);
		if (OCF_ERR_FLUSHING_INTERRUPTED == cmd.ext_err_code) {
			cas_printf(LOG_ERR, DIRTY_FLUSHING_WARNING);
			return INTERRUPTED;
		} else {
			print_err(cmd.ext_err_code);
			return FAILURE;
		}
	}

	close(fd);
	return SUCCESS;
}

int flush_core(unsigned int cache_id, unsigned int core_id)
{
	int fd = 0;
	struct kcas_flush_core cmd;

	memset(&cmd, 0, sizeof(cmd));
	cmd.cache_id = cache_id;
	cmd.core_id = core_id;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	/* synchronous flag */
	if (run_ioctl_interruptible(fd, KCAS_IOCTL_FLUSH_CORE, &cmd, "Flushing core", cache_id, core_id) < 0) {
		close(fd);
		if (OCF_ERR_FLUSHING_INTERRUPTED == cmd.ext_err_code) {
			cas_printf(LOG_ERR, DIRTY_FLUSHING_WARNING);
			return INTERRUPTED;
		} else {
			print_err(cmd.ext_err_code);
			return FAILURE;
		}
	}
	close(fd);
	return SUCCESS;
}

struct partition_config_col {
	const char *name;
	int pos;
};

static struct partition_config_col partition_config_columns[] = {
	{ .name = "IO class id", .pos = -1 },
	{ .name = "IO class name", .pos = -1 },
	{ .name = "Eviction priority", .pos = -1 },
	{ .name = "Allocation", .pos = -1 },
	{ .name = NULL }
};

void partition_list_line(FILE *out, struct kcas_io_class *cls, bool csv)
{
	char buffer[128];
	const char *prio;
	const char *allocation;
	if (cls->info.cache_mode != ocf_cache_mode_pt)
		allocation = csv ? "1" : "YES";
	else
		allocation = csv ? "0" : "NO";

	if (OCF_IO_CLASS_PRIO_PINNED == cls->info.priority) {
		prio = csv ? "" : "Pinned";
	} else {
		snprintf(buffer, sizeof(buffer), "%d", cls->info.priority);
		prio = buffer;
	}

	fprintf(out, TAG(TABLE_ROW)"%u,%s,%s,%s\n",
		cls->class_id, cls->info.name, prio, allocation);

}

int partition_list(unsigned int cache_id, unsigned int output_format)
{
	struct kcas_io_class io_class = { .ext_err_code = 0 };
	int fd, i = 0, result = 0;
	/* 1 is writing end, 0 is reading end of a pipe */
	FILE *intermediate_file[2];
	bool use_csv, first_col;

	fd = open_ctrl_device();
	if (fd == -1 )
		return FAILURE;

	if (create_pipe_pair(intermediate_file)) {
		cas_printf(LOG_ERR,"Failed to create unidirectional pipe.\n");
		close(fd);
		return FAILURE;
	}

	use_csv = (output_format == OUTPUT_FORMAT_CSV);

	first_col = true;
	fprintf(intermediate_file[1], TAG(TABLE_HEADER));
	for (i = 0; partition_config_columns[i].name; i++) {
		if (!first_col) {
			fputc(',', intermediate_file[1]);
		}
		fprintf(intermediate_file[1], "%s",
			partition_config_columns[i].name);
		first_col = false;
	}
	fputc('\n', intermediate_file[1]);

	for (i = 0; i < OCF_IO_CLASS_MAX; i++, io_class.ext_err_code = 0) {
		io_class.cache_id = cache_id;
		io_class.class_id = i;

		result = run_ioctl(fd, KCAS_IOCTL_PARTITION_STATS, &io_class);
		if (result) {
			if (OCF_ERR_IO_CLASS_NOT_EXIST == io_class.ext_err_code) {
				result = SUCCESS;
				continue;
			} else {
				result = FAILURE;
				break;
			}
		}

		partition_list_line(intermediate_file[1],
			&io_class, use_csv);

	}

	if (io_class.ext_err_code) {
		print_err(io_class.ext_err_code);
	}

	fclose(intermediate_file[1]);
	if (!result && stat_format_output(intermediate_file[0], stdout,
					  use_csv?RAW_CSV:TEXT)) {
		cas_printf(LOG_ERR, "An error occured during statistics formatting.\n");
		result = FAILURE;
	}
	fclose(intermediate_file[0]);
	close(fd);

	return result;
}

enum {
	part_csv_coll_id = 0, part_csv_coll_name, part_csv_coll_prio,
	part_csv_coll_alloc, part_csv_coll_max
};

int partition_is_name_valid(const char *name)
{
	int i;
	int length = strnlen(name, OCF_IO_CLASS_NAME_MAX);
	if (0 == length || length >= OCF_IO_CLASS_NAME_MAX) {
		cas_printf(LOG_ERR, "Empty or too long IO class name\n");
		return FAILURE;
	}

	for (i = 0; i < length; i++) {
		if (name[i] == ',' || name[i] == '"' ||
		    name[i] < 32 || name[i] > 126) {
			cas_printf(LOG_ERR, "Only characters allowed in IO "
				   "class name are low ascii characters, "
				   "excluding control characters, comma and "
				   "quotation mark.\n");
			return FAILURE;
		}
	}

	return SUCCESS;
}

static inline const char *partition_get_csv_col(CSVFILE *csv, int col,
						int *error_col)
{
	const char *val;

	val = csv_get_col(csv, partition_config_columns[col].pos);
	if (!val) {
		*error_col = col;
	}
	return val;
}

static inline int partition_get_line(CSVFILE *csv,
				     struct kcas_io_classes *cnfg,
				     int *error_col)
{
	uint32_t part_id;
	uint32_t value;
	const char *id, *name, *prio, *alloc;

	id = partition_get_csv_col(csv, part_csv_coll_id, error_col);
	if (!id) {
		return FAILURE;
	}
	name = partition_get_csv_col(csv, part_csv_coll_name, error_col);
	if (!name) {
		return FAILURE;
	}
	prio = partition_get_csv_col(csv, part_csv_coll_prio, error_col);
	if (!prio) {
		return FAILURE;
	}
	alloc = partition_get_csv_col(csv, part_csv_coll_alloc, error_col);
	if (!alloc) {
		return FAILURE;
	}

	/* Validate ID */
	*error_col = part_csv_coll_id;
	if (strempty(id)) {
		return FAILURE;
	}
	if (validate_str_num(id, "id", 0, OCF_IO_CLASS_ID_MAX)) {
		return FAILURE;
	}
	part_id = strtoul(id, NULL, 10);
	if (part_id > OCF_IO_CLASS_ID_MAX) {
		cas_printf(LOG_ERR, "Invalid partition id\n");
		return FAILURE;
	}
	if (!strempty(cnfg->info[part_id].name)) {
		cas_printf(LOG_ERR, "Double configuration for IO class id %u\n",
				part_id);
		return FAILURE;
	}

	/* Validate name */
	*error_col = part_csv_coll_name;
	if (SUCCESS != partition_is_name_valid(name)) {
		return FAILURE;
	}
	strncpy_s(cnfg->info[part_id].name, sizeof(cnfg->info[part_id].name),
		  name, strnlen_s(name, sizeof(cnfg->info[part_id].name)));

	/* Validate Priority*/
	*error_col = part_csv_coll_prio;
	if (strempty(prio)) {
		value = OCF_IO_CLASS_PRIO_PINNED;
	} else {
		if (validate_str_num(prio, "prio", OCF_IO_CLASS_PRIO_HIGHEST,
				OCF_IO_CLASS_PRIO_LOWEST)) {
			return FAILURE;
		}
		value = strtoul(prio, NULL, 10);
	}
	cnfg->info[part_id].priority = value;

	/* Validate Allocation */
	*error_col = part_csv_coll_alloc;
	if (strempty(alloc)) {
		return FAILURE;
	}
	if (validate_str_num(alloc, "alloc", 0, 1)) {
		return FAILURE;
	}
	value = strtoul(alloc, NULL, 10);
	if (0 == value) {
		cnfg->info[part_id].cache_mode = ocf_cache_mode_pt;
	} else if (1 == value) {
		cnfg->info[part_id].cache_mode = ocf_cache_mode_max;
	} else {
		return FAILURE;
	}

	cnfg->info[part_id].min_size = 0;
	cnfg->info[part_id].max_size = UINT32_MAX;

	return 0;
}

static int partition_parse_header(CSVFILE *csv)
{
	int i, j, csv_cols;
	const char *col_name;

	csv_cols = csv_count_cols(csv);
	for (i = 0; i < csv_cols; i++) {
		col_name = csv_get_col(csv, i);

		if (!col_name) {
			cas_printf(LOG_ERR, "Cannot parse configuration file.\n");
			return FAILURE;
		}

		for (j = 0; partition_config_columns[j].name; j++) {
			if (!strncmp(col_name, partition_config_columns[j].name, MAX_STR_LEN)) {
				partition_config_columns[j].pos = i;
				break;
			}
		}
		if (!partition_config_columns[j].name) {
			cas_printf(LOG_ERR,
				   "Cannot parse configuration file - unknown column \"%s\".\n",
				   col_name);
			return FAILURE;
		}
	}

	for (i = 0; partition_config_columns[i].name; i++) {
		if (partition_config_columns[i].pos < 0) {
			cas_printf(LOG_ERR,
				   "Cannot parse configuration file - missing column \"%s\".\n",
				   partition_config_columns[i].name);
			return FAILURE;
		}
	}
	return SUCCESS;
}

int partition_get_config(CSVFILE *csv, struct kcas_io_classes *cnfg,
		int cache_id)
{
	int result = 0, count = 0;
	int line = 1;
	int error_col = -1;

	cnfg->cache_id = cache_id;

	/* before reading io class configuration check header */
	if (csv_read(csv)) {
		if (csv_feof(csv)) {
			cas_printf(LOG_ERR,
				   "Empty IO Classes configuration file"
				   " supplied.\n");
			return FAILURE;
		} else {
			cas_printf(LOG_ERR,
				   "I/O error occured while reading"
				   " IO Classes configuration file"
				   " supplied.\n");
			return FAILURE;
		}
	}

	if (partition_parse_header(csv)) {
		cas_printf(LOG_ERR, "Failed to parse I/O classes"
			   " configuration file header. It is either"
			   " malformed or missing.\n"
			   "Please consult Admin Guide to check how"
			   " columns in configuration file should"
			   " be named.\n");
		return FAILURE;
	}

	/* check all lines of input */
	while (!csv_feof(csv)) {
		line++;
		if (csv_read(csv)) {
			if (csv_feof(csv)) {
				break;
			} else {
				result = FAILURE;
				break;
			}
		}

		if (part_csv_coll_max != csv_count_cols(csv)) {
			if (csv_empty_line(csv)) {
				continue;
			} else {
				result = FAILURE;
				break;
			}
		}

		if (partition_get_line(csv, cnfg, &error_col)) {
			result = FAILURE;
			break;
		}

		count++;
	}

	if (result) {
		if (error_col >= 0) {
			cas_printf(LOG_ERR,
				   "Cannot parse configuration file - error in line %d in column %d (%s).\n",
				   line,
				   partition_config_columns[error_col].pos+1,
				   partition_config_columns[error_col].name);
		} else {
			cas_printf(LOG_ERR, "Cannot parse configuration file - error in line %d.\n", line);
		}
	} else if (0 == count) {
		result = FAILURE;
		cas_printf(LOG_ERR, "Empty configuration file\n");
	}

	return result;
}

int partition_set_config(struct kcas_io_classes *cnfg)
{
	int fd;
	int result = 0;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	result = run_ioctl(fd, KCAS_IOCTL_PARTITION_SET, cnfg);
	if (result) {
		if (OCF_ERR_IO_CLASS_NOT_EXIST == cnfg->ext_err_code) {
			result = SUCCESS;
		} else {
			print_err(cnfg->ext_err_code);
			result = FAILURE;
		}
	}

	close(fd);
	return result;
}

int partition_setup(unsigned int cache_id, const char *file)
{
	int result = 0;
	CSVFILE *in;
	struct kcas_io_classes *cnfg = calloc(1, KCAS_IO_CLASSES_SIZE);

	if (!cnfg)
		return FAILURE;

	if (strempty(file)) {
		cas_printf(LOG_ERR, "Invalid path of configuration file\n");
		result = FAILURE;
		goto exit;
	}

	if ('-'==file[0] && (!file[1])) {
		/* configuration is supposed to be read from stdin. Setup
		 * a csv parser treating standard input as input file instead
		 * of opening a regular file */
		in = csv_fopen(stdin);
	} else {
		/* read ioclass configuration from a regular file */
		in = csv_open(file, "r");
	}
	if (NULL == in) {
		cas_printf(LOG_ERR, "Cannot open configuration file %s\n",
				file);
		result = FAILURE;
		goto exit;
	}

	if (0 == partition_get_config(in, cnfg, cache_id)) {
		result = partition_set_config(cnfg);
	} else {
		result = FAILURE;
	}

	if ('-' == file[0] && (!file[1])) {
		/* free assets allocated by CSV parser without actually
		 * closing a file */
		csv_close_nu(in);
	} else {
		csv_close(in);
	}

exit:
	free(cnfg);
	return result;
}

int reset_counters(unsigned int cache_id, unsigned int core_id)
{
	struct kcas_reset_stats cmd;
	int fd = 0;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	memset(&cmd, 0, sizeof(cmd));
	cmd.cache_id = cache_id;
	cmd.core_id = core_id;

	if (ioctl(fd, KCAS_IOCTL_RESET_STATS, &cmd) < 0) {
		close(fd);
		cas_printf(LOG_ERR, "Error encountered while reseting counters\n");
		print_err(cmd.ext_err_code);
		return FAILURE;
	}

	close(fd);
	return SUCCESS;
}

int cas_module_version(char *buff, int size)
{
	FILE *fd;
	int n_read;

	if (size <= 0 || size > MAX_STR_LEN) {
		return FAILURE;
	}
	memset(buff, 0, size);

	fd = fopen("/sys/module/cas_cache/version", "r");
	if (!fd) {
		return FAILURE;
	}

	n_read = fread(buff, 1, size, fd);
	if (ferror(fd)) {
		n_read = 0;
	}
	fclose(fd);

	if (n_read > 0) {
		buff[n_read - 1] = '\0';
		return SUCCESS;
	} else {
		return FAILURE;
	}
}

int disk_module_version(char *buff, int size)
{
	FILE *fd;
	int n_read;

	if (size <= 0 || size > MAX_STR_LEN) {
		return FAILURE;
	}

	fd = fopen("/sys/module/cas_disk/version", "r");
	if (!fd) {
		return FAILURE;
	}

	n_read = fread(buff, 1, size, fd);
	if (ferror(fd)) {
		n_read = 0;
	}
	fclose(fd);

	if (n_read > 0) {
		buff[n_read - 1] = '\0';
		return SUCCESS;
	} else {
		return FAILURE;
	}
}

float calculate_flush_progress(unsigned dirty, unsigned flushed)
{
	unsigned total_dirty;

	if (!flushed)
		return 0;

	total_dirty = dirty + flushed;
	return total_dirty ? 100. * flushed / total_dirty : 100;
}

int get_flush_progress(int unsigned cache_id, float *progress)
{
	struct kcas_cache_info cmd_info;
	int fd = 0;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	memset(&cmd_info, 0, sizeof(cmd_info));

	cmd_info.cache_id = cache_id;
	if (ioctl(fd, KCAS_IOCTL_CACHE_INFO, &cmd_info) < 0) {
		close(fd);
		return FAILURE;
	}

	*progress = calculate_flush_progress(cmd_info.info.dirty,
			cmd_info.info.flushed);

	close(fd);
	return SUCCESS;
}

struct list_printout_ctx
{
	FILE *intermediate;
	FILE *out;
	int type;
	int result;
};

void *list_printout(void *ctx)
{
	struct list_printout_ctx *spc = ctx;
	if (stat_format_output(spc->intermediate,
			       spc->out, spc->type)) {
		cas_printf(LOG_ERR, "An error occured during statistics formatting.\n");
		spc->result = FAILURE;
	} else {
		spc->result = SUCCESS;
	}

	return NULL;
}

int get_core_pool_count(int fd)
{
	struct kcas_core_pool_count cmd;

	if (ioctl(fd, KCAS_IOCTL_GET_CORE_POOL_COUNT, &cmd) < 0)
		return 0;

	return cmd.core_pool_count;
}

int get_core_pool_devices(struct kcas_core_pool_path *cmd)
{
	int fd, status, result = SUCCESS;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	cmd->core_pool_count = get_core_pool_count(fd);
	if (cmd->core_pool_count <= 0) {
		goto error_out;
	}

	cmd->core_path_tab = malloc(cmd->core_pool_count * MAX_STR_LEN);
	if (NULL == cmd->core_path_tab) {
		cmd->core_pool_count = 0;
		goto error_out;
	}

	if ((status = ioctl(fd, KCAS_IOCTL_GET_CORE_POOL_PATHS, cmd)) < 0) {
		cas_printf(LOG_ERR, "Error while retrieving core pool list %d %d\n",
				errno, status);
		free(cmd->core_path_tab);
		result = FAILURE;
		goto error_out;
	}

error_out:
	close(fd);
	return result;
}

int list_caches(unsigned int list_format)
{
	struct cache_device **caches, *curr_cache;
	struct kcas_core_pool_path core_pool_path_cmd = {0};
	struct core_device *curr_core;
	int caches_count, i, j;
	/* 1 is writing end, 0 is reading end of a pipe */
	FILE *intermediate_file[2];
	int result = SUCCESS;
	pthread_t thread;
	struct list_printout_ctx printout_ctx;

	caches = get_cache_devices(&caches_count);
	if (caches_count < 0) {
		cas_printf(LOG_INFO, "Error getting caches list\n");
		return FAILURE;
	}

	if (get_core_pool_devices(&core_pool_path_cmd)) {
		free_cache_devices_list(caches, caches_count);
		cas_printf(LOG_INFO, "Error getting cores in pool list\n");
		return FAILURE;
	}

	if (caches == NULL && !core_pool_path_cmd.core_pool_count) {
		cas_printf(LOG_INFO, "No caches running\n");
		return SUCCESS;
	}

	if (create_pipe_pair(intermediate_file)) {
		cas_printf(LOG_ERR,"Failed to create unidirectional pipe.\n");
		free(core_pool_path_cmd.core_path_tab);
		free_cache_devices_list(caches, caches_count);
		return FAILURE;
	}

	printout_ctx.intermediate = intermediate_file[0];
	printout_ctx.out = stdout;
	printout_ctx.type = (OUTPUT_FORMAT_CSV == list_format ? RAW_CSV : TEXT);

	if (pthread_create(&thread, 0, list_printout, &printout_ctx)) {
		cas_printf(LOG_ERR,"Failed to create thread.\n");
		free(core_pool_path_cmd.core_path_tab);
		free_cache_devices_list(caches, caches_count);
		fclose(intermediate_file[0]);
		fclose(intermediate_file[1]);
		return FAILURE;
	}

	if (caches_count || core_pool_path_cmd.core_pool_count) {
		fprintf(intermediate_file[1],
			TAG(TREE_HEADER)"%s,%s,%s,%s,%s,%s\n",
			"type", "id", "disk", "status",
			"write policy", "device");
	}

	if (core_pool_path_cmd.core_pool_count) {
		fprintf(intermediate_file[1], TAG(TREE_BRANCH)
			"%s,%s,%s,%s,%s,%s\n",
			"core pool", /* type */
			"-", /* id */
			"-",
			"-",
			"-", /* write policy */
			"-" /* device */);
		for (i = 0; i < core_pool_path_cmd.core_pool_count; i++) {
			char *core_path = core_pool_path_cmd.core_path_tab + (MAX_STR_LEN * i);
			if (get_dev_path(core_path, core_path, MAX_STR_LEN)) {
				cas_printf(LOG_WARNING, "WARNING: Can not resolve path to core. "
						"By-id path will be shown for that core.\n");
			}
			fprintf(intermediate_file[1], TAG(TREE_LEAF)
			"%s,%s,%s,%s,%s,%s\n",
			"core", /* type */
			"-", /* id */
			core_path,
			"Detached",
			"-", /* write policy */
			"-" /* device */);
		}
	}

	for (i = 0; i < caches_count; ++i) {
		curr_cache = caches[i];

		char status_buf[CACHE_STATE_LENGHT];
		const char *tmp_status;
		char mode_string[12];
		float cache_flush_prog;
		float core_flush_prog;

		get_dev_path(curr_cache->device, curr_cache->device, sizeof(curr_cache->device));

		cache_flush_prog = calculate_flush_progress(curr_cache->dirty, curr_cache->flushed);
		if (cache_flush_prog) {
			snprintf(status_buf, sizeof(status_buf),
				"%s (%3.1f %%)", "Flushing", cache_flush_prog);
			tmp_status = status_buf;
			snprintf(mode_string, sizeof(mode_string), "wb->%s",
					cache_mode_to_name(curr_cache->mode));
		} else {
			tmp_status = get_cache_state_name(curr_cache->state);
			snprintf(mode_string, sizeof(mode_string), "%s",
					cache_mode_to_name(curr_cache->mode));
		}

		fprintf(intermediate_file[1], TAG(TREE_BRANCH)
			"%s,%u,%s,%s,%s,%s\n",
			"cache", /* type */
			curr_cache->id, /* id */
			curr_cache->device, /* device path */
			tmp_status, /* cache status */
			mode_string, /* write policy */
			"-" /* device */);

		for (j = 0; j < curr_cache->core_count; ++j) {
			char* core_path;

			curr_core = &curr_cache->cores[j];
			core_path = curr_core->path;

			core_flush_prog = calculate_flush_progress(curr_core->info.stats.dirty,
					curr_core->info.stats.flushed);

			if (!core_flush_prog && cache_flush_prog) {
				core_flush_prog = curr_core->info.stats.dirty ? 0 : 100;
			}

			if (core_flush_prog || cache_flush_prog) {
				snprintf(status_buf, CACHE_STATE_LENGHT,
						"%s (%3.1f %%)", "Flushing", core_flush_prog);
				tmp_status = status_buf;
			} else {
				tmp_status = get_core_state_name(curr_core->info.state);
			}

			fprintf(intermediate_file[1], TAG(TREE_LEAF)
					"%s,%u,%s,%s,%s,/dev/cas%d-%d\n",
					"core", /* type */
					curr_core->id, /* id */
					core_path, /* path to core*/
					tmp_status, /* core status */
					"-", /* write policy */
					curr_cache->id, /* core id (part of path)*/
					curr_core->id /* cache id (part of path)*/ );
		}
	}

	free_cache_devices_list(caches, caches_count);
	free(core_pool_path_cmd.core_path_tab);

	fclose(intermediate_file[1]);
	pthread_join(thread, 0);
	if (printout_ctx.result) {
		result = 1;
		cas_printf(LOG_ERR, "An error occured during list formatting.\n");

	}
	fclose(intermediate_file[0]);
	return result;
}

int _get_cas_capabilites(struct kcas_capabilites *caps, int quiet)
{
	static bool retrieved = false;
	static struct kcas_capabilites caps_buf;
	int status = SUCCESS;
	int ctrl_fd;
	if (!retrieved) {
		if (quiet) {
			ctrl_fd = open_ctrl_device_quiet();
		} else {
			ctrl_fd = open_ctrl_device();
		}

		if (ctrl_fd < 0) {
			if (!quiet)
				print_err(KCAS_ERR_SYSTEM);

			return FAILURE;
		}

		status = ioctl(ctrl_fd, KCAS_IOCTL_GET_CAPABILITIES, &caps_buf);
		close(ctrl_fd);

		if (status) {
			return FAILURE;
		}
		retrieved = true;
	}

	memcpy_s(caps, sizeof(*caps), &caps_buf, sizeof(caps_buf));
	return status;
}

int get_cas_capabilites_quiet(struct kcas_capabilites *caps)
{
	return _get_cas_capabilites(caps, true);
}

int get_cas_capabilites(struct kcas_capabilites *caps)
{
	return _get_cas_capabilites(caps, false);
}

int nvme_format(const char *device_path, int metadata_mode, int force)
{
	struct kcas_nvme_format cmd_info;
	int fd;
	int result = 0;

	strncpy_s(cmd_info.device_path_name,
		sizeof(cmd_info.device_path_name), device_path,
		strnlen_s(device_path, sizeof(cmd_info.device_path_name)));

	switch (metadata_mode) {
	case METADATA_MODE_NORMAL:
		cmd_info.metadata_mode = CAS_METADATA_MODE_NORMAL;
		break;
	case METADATA_MODE_ATOMIC:
		cmd_info.metadata_mode = CAS_METADATA_MODE_ATOMIC;
		break;
	default:
		return FAILURE;
	}
	cmd_info.force = force;

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	/* Format NVMe  */
	result = run_ioctl(fd, KCAS_IOCTL_NVME_FORMAT, &cmd_info);
	close(fd);

	if (result) {
		result = cmd_info.ext_err_code ? : KCAS_ERR_SYSTEM;
		cas_printf(LOG_INFO, "Changing NVMe format failed!\n");
		print_err(result);
		return FAILURE;
	}

	cas_printf(LOG_INFO, "Changing NVMe format succeeded.\n"
			"IMPORTANT: Reboot is required!\n");

	return SUCCESS;
}

int _check_cache_device(const char *device_path,
		struct kcas_cache_check_device *cmd_info)
{
	int result, fd;

	strncpy_s(cmd_info->path_name, sizeof(cmd_info->path_name), device_path,
		strnlen_s(device_path, sizeof(cmd_info->path_name)));

	fd = open_ctrl_device();
	if (fd == -1)
		return FAILURE;

	result = run_ioctl(fd, KCAS_IOCTL_CACHE_CHECK_DEVICE, cmd_info);

	close(fd);

	return result;
}

int check_cache_device(const char *device_path)
{
	struct kcas_cache_check_device cmd_info;
	FILE *intermediate_file[2];
	int result;

	result = _check_cache_device(device_path, &cmd_info);

	if (result) {
		result = cmd_info.ext_err_code ? : KCAS_ERR_SYSTEM;
		print_err(result);
		return FAILURE;
	}

	if (create_pipe_pair(intermediate_file)) {
		cas_printf(LOG_ERR,"Failed to create unidirectional pipe.\n");
		return FAILURE;
	}

	fprintf(intermediate_file[1], TAG(TABLE_HEADER) "Is cache,Clean Shutdown,Cache dirty\n");

	fprintf(intermediate_file[1], TAG(TABLE_ROW));
	if (cmd_info.is_cache_device) {
		fprintf(intermediate_file[1], "yes,%s,%s\n",
				cmd_info.clean_shutdown ? "yes" : "no",
				cmd_info.cache_dirty ? "yes" : "no");
	} else {
		fprintf(intermediate_file[1], "no,-,-\n");
	}

	fclose(intermediate_file[1]);
	stat_format_output(intermediate_file[0], stdout, RAW_CSV);
	fclose(intermediate_file[0]);

	return SUCCESS;
}