X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=init.c;h=44c74e315a4b758ccdd7bb8055094014d395b12e;hp=87052b9e5476aae581ffc6475293aa5529bf7590;hb=ad705bcb7e79a7cdb9891db17b4c40b13b6c30c3;hpb=142575e6579462656a6d0f7c50ec8c35b8a08802

diff --git a/init.c b/init.c
index 87052b9e..44c74e31 100644
--- a/init.c
+++ b/init.c
@@ -9,34 +9,42 @@
 #include <string.h>
 #include <errno.h>
 #include <sys/ipc.h>
-#include <sys/shm.h>
 #include <sys/types.h>
 #include <sys/stat.h>
 
 #include "fio.h"
+#ifndef FIO_NO_HAVE_SHM_H
+#include <sys/shm.h>
+#endif
+
 #include "parse.h"
 #include "smalloc.h"
 #include "filehash.h"
 #include "verify.h"
 #include "profile.h"
 #include "server.h"
+#include "idletime.h"
 
 #include "lib/getopt.h"
+#include "lib/strcasestr.h"
 
-static char fio_version_string[] = "fio 1.58";
+const char fio_version_string[] = FIO_VERSION;
 
 #define FIO_RANDSEED		(0xb1899bedUL)
 
 static char **ini_file;
 static int max_jobs = FIO_MAX_JOBS;
 static int dump_cmdline;
+static int def_timeout;
+static int parse_only;
 
 static struct thread_data def_thread;
 struct thread_data *threads = NULL;
 
 int exitall_on_terminate = 0;
-int terse_output = 0;
-int eta_print;
+int output_format = FIO_OUTPUT_NORMAL;
+int eta_print = FIO_ETA_AUTO;
+int eta_new_line = 0;
 unsigned long long mlock_size = 0;
 FILE *f_out = NULL;
 FILE *f_err = NULL;
@@ -44,10 +52,10 @@ char **job_sections = NULL;
 int nr_job_sections = 0;
 char *exec_profile = NULL;
 int warnings_fatal = 0;
-int terse_version = 2;
+int terse_version = 3;
 int is_backend = 0;
-int is_client = 0;
-char *client;
+int nr_clients = 0;
+int log_syslog = 0;
 
 int write_bw_log = 0;
 int read_only = 0;
@@ -61,6 +69,9 @@ unsigned int fio_debug_jobno = -1;
 unsigned int *fio_debug_jobp = NULL;
 
 static char cmd_optstr[256];
+static int did_arg;
+
+#define FIO_CLIENT_FLAG		(1 << 16)
 
 /*
  * Command line options. These will contain the above, plus a few
@@ -70,122 +81,228 @@ static struct option l_opts[FIO_NR_OPTIONS] = {
 	{
 		.name		= (char *) "output",
 		.has_arg	= required_argument,
-		.val		= 'o',
+		.val		= 'o' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "timeout",
 		.has_arg	= required_argument,
-		.val		= 't',
+		.val		= 't' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "latency-log",
 		.has_arg	= required_argument,
-		.val		= 'l',
+		.val		= 'l' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "bandwidth-log",
 		.has_arg	= required_argument,
-		.val		= 'b',
+		.val		= 'b' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "minimal",
 		.has_arg	= optional_argument,
-		.val		= 'm',
+		.val		= 'm' | FIO_CLIENT_FLAG,
+	},
+	{
+		.name		= (char *) "output-format",
+		.has_arg	= optional_argument,
+		.val		= 'F' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "version",
 		.has_arg	= no_argument,
-		.val		= 'v',
+		.val		= 'v' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "help",
 		.has_arg	= no_argument,
-		.val		= 'h',
+		.val		= 'h' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "cmdhelp",
 		.has_arg	= optional_argument,
-		.val		= 'c',
+		.val		= 'c' | FIO_CLIENT_FLAG,
+	},
+	{
+		.name		   = (char *) "enghelp",
+		.has_arg	= optional_argument,
+		.val		    = 'i' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "showcmd",
 		.has_arg	= no_argument,
-		.val		= 's',
+		.val		= 's' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "readonly",
 		.has_arg	= no_argument,
-		.val		= 'r',
+		.val		= 'r' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "eta",
 		.has_arg	= required_argument,
-		.val		= 'e',
+		.val		= 'e' | FIO_CLIENT_FLAG,
+	},
+	{
+		.name		= (char *) "eta-newline",
+		.has_arg	= required_argument,
+		.val		= 'E' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "debug",
 		.has_arg	= required_argument,
-		.val		= 'd',
+		.val		= 'd' | FIO_CLIENT_FLAG,
+	},
+	{
+		.name		= (char *) "parse-only",
+		.has_arg	= no_argument,
+		.val		= 'P' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "section",
 		.has_arg	= required_argument,
-		.val		= 'x',
+		.val		= 'x' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "alloc-size",
 		.has_arg	= required_argument,
-		.val		= 'a',
+		.val		= 'a' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "profile",
 		.has_arg	= required_argument,
-		.val		= 'p',
+		.val		= 'p' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "warnings-fatal",
 		.has_arg	= no_argument,
-		.val		= 'w',
+		.val		= 'w' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "max-jobs",
 		.has_arg	= required_argument,
-		.val		= 'j',
+		.val		= 'j' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "terse-version",
 		.has_arg	= required_argument,
-		.val		= 'V',
+		.val		= 'V' | FIO_CLIENT_FLAG,
 	},
 	{
 		.name		= (char *) "server",
-		.has_arg	= no_argument,
+		.has_arg	= optional_argument,
 		.val		= 'S',
 	},
-	{
-		.name		= (char *) "net-port",
+	{	.name		= (char *) "daemonize",
 		.has_arg	= required_argument,
-		.val		= 'P',
+		.val		= 'D',
 	},
 	{
 		.name		= (char *) "client",
 		.has_arg	= required_argument,
 		.val		= 'C',
 	},
+	{
+		.name		= (char *) "cpuclock-test",
+		.has_arg	= no_argument,
+		.val		= 'T',
+	},
+	{
+		.name		= (char *) "idle-prof",
+		.has_arg	= required_argument,
+		.val		= 'I',
+	},
 	{
 		.name		= NULL,
 	},
 };
 
+static void free_shm(void)
+{
+	struct shmid_ds sbuf;
+
+	if (threads) {
+		void *tp = threads;
+
+		threads = NULL;
+		file_hash_exit();
+		flow_exit();
+		fio_debug_jobp = NULL;
+		shmdt(tp);
+		shmctl(shm_id, IPC_RMID, &sbuf);
+	}
+
+	scleanup();
+}
+
+/*
+ * The thread area is shared between the main process and the job
+ * threads/processes. So setup a shared memory segment that will hold
+ * all the job info. We use the end of the region for keeping track of
+ * open files across jobs, for file sharing.
+ */
+static int setup_thread_area(void)
+{
+	void *hash;
+
+	if (threads)
+		return 0;
+
+	/*
+	 * 1024 is too much on some machines, scale max_jobs if
+	 * we get a failure that looks like too large a shm segment
+	 */
+	do {
+		size_t size = max_jobs * sizeof(struct thread_data);
+
+		size += file_hash_size;
+		size += sizeof(unsigned int);
+
+		shm_id = shmget(0, size, IPC_CREAT | 0600);
+		if (shm_id != -1)
+			break;
+		if (errno != EINVAL && errno != ENOMEM && errno != ENOSPC) {
+			perror("shmget");
+			break;
+		}
+
+		max_jobs >>= 1;
+	} while (max_jobs);
+
+	if (shm_id == -1)
+		return 1;
+
+	threads = shmat(shm_id, NULL, 0);
+	if (threads == (void *) -1) {
+		perror("shmat");
+		return 1;
+	}
+
+	memset(threads, 0, max_jobs * sizeof(struct thread_data));
+	hash = (void *) threads + max_jobs * sizeof(struct thread_data);
+	fio_debug_jobp = (void *) hash + file_hash_size;
+	*fio_debug_jobp = -1;
+	file_hash_init(hash);
+
+	flow_init();
+
+	return 0;
+}
+
 /*
  * Return a free job structure.
  */
-static struct thread_data *get_new_job(int global, struct thread_data *parent)
+static struct thread_data *get_new_job(int global, struct thread_data *parent,
+				       int preserve_eo)
 {
 	struct thread_data *td;
 
 	if (global)
 		return &def_thread;
+	if (setup_thread_area()) {
+		log_err("error: failed to setup shm segment\n");
+		return NULL;
+	}
 	if (thread_number >= max_jobs) {
 		log_err("error: maximum number of jobs (%d) reached.\n",
 				max_jobs);
@@ -195,14 +312,22 @@ static struct thread_data *get_new_job(int global, struct thread_data *parent)
 	td = &threads[thread_number++];
 	*td = *parent;
 
+	td->io_ops = NULL;
+	if (!preserve_eo)
+		td->eo = NULL;
+
 	td->o.uid = td->o.gid = -1U;
 
 	dup_files(td, parent);
-	options_mem_dupe(td);
+	fio_options_mem_dupe(td);
 
 	profile_add_hooks(td);
 
 	td->thread_number = thread_number;
+
+	if (!parent || !parent->o.group_reporting)
+		stat_number++;
+
 	return td;
 }
 
@@ -212,10 +337,15 @@ static void put_job(struct thread_data *td)
 		return;
 
 	profile_td_exit(td);
+	flow_exit_job(td);
 
 	if (td->error)
 		log_info("fio: %s\n", td->verror);
 
+	fio_options_free(td);
+	if (td->io_ops)
+		free_ioengine(td);
+
 	memset(&threads[td->thread_number - 1], 0, sizeof(*td));
 	thread_number--;
 }
@@ -223,21 +353,19 @@ static void put_job(struct thread_data *td)
 static int __setup_rate(struct thread_data *td, enum fio_ddir ddir)
 {
 	unsigned int bs = td->o.min_bs[ddir];
-	unsigned long long bytes_per_sec;
 
 	assert(ddir_rw(ddir));
 
 	if (td->o.rate[ddir])
-		bytes_per_sec = td->o.rate[ddir];
+		td->rate_bps[ddir] = td->o.rate[ddir];
 	else
-		bytes_per_sec = td->o.rate_iops[ddir] * bs;
+		td->rate_bps[ddir] = td->o.rate_iops[ddir] * bs;
 
-	if (!bytes_per_sec) {
+	if (!td->rate_bps[ddir]) {
 		log_err("rate lower than supported\n");
 		return -1;
 	}
 
-	td->rate_nsec_cycle[ddir] = 1000000000ULL / bytes_per_sec;
 	td->rate_pending_usleep[ddir] = 0;
 	return 0;
 }
@@ -250,6 +378,8 @@ static int setup_rate(struct thread_data *td)
 		ret = __setup_rate(td, DDIR_READ);
 	if (td->o.rate[DDIR_WRITE] || td->o.rate_iops[DDIR_WRITE])
 		ret |= __setup_rate(td, DDIR_WRITE);
+	if (td->o.rate[DDIR_TRIM] || td->o.rate_iops[DDIR_TRIM])
+		ret |= __setup_rate(td, DDIR_TRIM);
 
 	return ret;
 }
@@ -258,7 +388,9 @@ static int fixed_block_size(struct thread_options *o)
 {
 	return o->min_bs[DDIR_READ] == o->max_bs[DDIR_READ] &&
 		o->min_bs[DDIR_WRITE] == o->max_bs[DDIR_WRITE] &&
-		o->min_bs[DDIR_READ] == o->min_bs[DDIR_WRITE];
+		o->min_bs[DDIR_TRIM] == o->max_bs[DDIR_TRIM] &&
+		o->min_bs[DDIR_READ] == o->min_bs[DDIR_WRITE] &&
+		o->min_bs[DDIR_READ] == o->min_bs[DDIR_TRIM];
 }
 
 /*
@@ -288,11 +420,18 @@ static int fixup_options(struct thread_data *td)
 	}
 
 	/*
-	 * only really works for sequential io for now, and with 1 file
+	 * only really works with 1 file
 	 */
-	if (o->zone_size && td_random(td) && o->open_files == 1)
+	if (o->zone_size && o->open_files > 1)
 		o->zone_size = 0;
 
+	/*
+	 * If zone_range isn't specified, backward compatibility dictates it
+	 * should be made equal to zone_size.
+	 */
+	if (o->zone_size && !o->zone_range)
+		o->zone_range = o->zone_size;
+
 	/*
 	 * Reads can do overwrites, we always need to pre-create the file
 	 */
@@ -307,8 +446,14 @@ static int fixup_options(struct thread_data *td)
 		o->min_bs[DDIR_WRITE] = o->bs[DDIR_WRITE];
 	if (!o->max_bs[DDIR_WRITE])
 		o->max_bs[DDIR_WRITE] = o->bs[DDIR_WRITE];
+	if (!o->min_bs[DDIR_TRIM])
+		o->min_bs[DDIR_TRIM] = o->bs[DDIR_TRIM];
+	if (!o->max_bs[DDIR_TRIM])
+		o->max_bs[DDIR_TRIM] = o->bs[DDIR_TRIM];
+
 
 	o->rw_min_bs = min(o->min_bs[DDIR_READ], o->min_bs[DDIR_WRITE]);
+	o->rw_min_bs = min(o->min_bs[DDIR_TRIM], o->rw_min_bs);
 
 	/*
 	 * For random IO, allow blockalign offset other than min_bs.
@@ -317,9 +462,12 @@ static int fixup_options(struct thread_data *td)
 		o->ba[DDIR_READ] = o->min_bs[DDIR_READ];
 	if (!o->ba[DDIR_WRITE] || !td_random(td))
 		o->ba[DDIR_WRITE] = o->min_bs[DDIR_WRITE];
+	if (!o->ba[DDIR_TRIM] || !td_random(td))
+		o->ba[DDIR_TRIM] = o->min_bs[DDIR_TRIM];
 
 	if ((o->ba[DDIR_READ] != o->min_bs[DDIR_READ] ||
-	    o->ba[DDIR_WRITE] != o->min_bs[DDIR_WRITE]) &&
+	    o->ba[DDIR_WRITE] != o->min_bs[DDIR_WRITE] ||
+	    o->ba[DDIR_TRIM] != o->min_bs[DDIR_TRIM]) &&
 	    !o->norandommap) {
 		log_err("fio: Any use of blockalign= turns off randommap\n");
 		o->norandommap = 1;
@@ -348,17 +496,8 @@ static int fixup_options(struct thread_data *td)
 	/*
 	 * The low water mark cannot be bigger than the iodepth
 	 */
-	if (o->iodepth_low > o->iodepth || !o->iodepth_low) {
-		/*
-		 * syslet work around - if the workload is sequential,
-		 * we want to let the queue drain all the way down to
-		 * avoid seeking between async threads
-		 */
-		if (!strcmp(td->io_ops->name, "syslet-rw") && !td_random(td))
-			o->iodepth_low = 1;
-		else
-			o->iodepth_low = o->iodepth;
-	}
+	if (o->iodepth_low > o->iodepth || !o->iodepth_low)
+		o->iodepth_low = o->iodepth;
 
 	/*
 	 * If batch number isn't set, default to the same as iodepth
@@ -372,15 +511,19 @@ static int fixup_options(struct thread_data *td)
 	if (o->open_files > o->nr_files || !o->open_files)
 		o->open_files = o->nr_files;
 
-	if (((o->rate[0] + o->rate[1]) && (o->rate_iops[0] + o->rate_iops[1]))||
-	    ((o->ratemin[0] + o->ratemin[1]) && (o->rate_iops_min[0] +
-		o->rate_iops_min[1]))) {
+	if (((o->rate[DDIR_READ] + o->rate[DDIR_WRITE] + o->rate[DDIR_TRIM]) &&
+	    (o->rate_iops[DDIR_READ] + o->rate_iops[DDIR_WRITE] + o->rate_iops[DDIR_TRIM])) ||
+	    ((o->ratemin[DDIR_READ] + o->ratemin[DDIR_WRITE] + o->ratemin[DDIR_TRIM]) &&
+	    (o->rate_iops_min[DDIR_READ] + o->rate_iops_min[DDIR_WRITE] + o->rate_iops_min[DDIR_TRIM]))) {
 		log_err("fio: rate and rate_iops are mutually exclusive\n");
 		ret = 1;
 	}
-	if ((o->rate[0] < o->ratemin[0]) || (o->rate[1] < o->ratemin[1]) ||
-	    (o->rate_iops[0] < o->rate_iops_min[0]) ||
-	    (o->rate_iops[1] < o->rate_iops_min[1])) {
+	if ((o->rate[DDIR_READ] < o->ratemin[DDIR_READ]) ||
+	    (o->rate[DDIR_WRITE] < o->ratemin[DDIR_WRITE]) ||
+	    (o->rate[DDIR_TRIM] < o->ratemin[DDIR_TRIM]) ||
+	    (o->rate_iops[DDIR_READ] < o->rate_iops_min[DDIR_READ]) ||
+	    (o->rate_iops[DDIR_WRITE] < o->rate_iops_min[DDIR_WRITE]) ||
+	    (o->rate_iops[DDIR_TRIM] < o->rate_iops_min[DDIR_TRIM])) {
 		log_err("fio: minimum rate exceeds rate\n");
 		ret = 1;
 	}
@@ -417,7 +560,14 @@ static int fixup_options(struct thread_data *td)
 		}
 	}
 
-#ifndef FIO_HAVE_FDATASYNC
+	if (!o->unit_base) {
+		if (td->io_ops->flags & FIO_BIT_BASED)
+			o->unit_base = 1;
+		else
+			o->unit_base = 8;
+	}
+
+#ifndef CONFIG_FDATASYNC
 	if (o->fdatasync_blocks) {
 		log_info("fio: this platform does not support fdatasync()"
 			 " falling back to using fsync().  Use the 'fsync'"
@@ -429,6 +579,35 @@ static int fixup_options(struct thread_data *td)
 	}
 #endif
 
+#ifdef WIN32
+	/*
+	 * Windows doesn't support O_DIRECT or O_SYNC with the _open interface,
+	 * so fail if we're passed those flags
+	 */
+	if ((td->io_ops->flags & FIO_SYNCIO) && (td->o.odirect || td->o.sync_io)) {
+		log_err("fio: Windows does not support direct or non-buffered io with"
+				" the synchronous ioengines. Use the 'windowsaio' ioengine"
+				" with 'direct=1' and 'iodepth=1' instead.\n");
+		ret = 1;
+	}
+#endif
+
+	/*
+	 * For fully compressible data, just zero them at init time.
+	 * It's faster than repeatedly filling it.
+	 */
+	if (td->o.compress_percentage == 100) {
+		td->o.zero_buffers = 1;
+		td->o.compress_percentage = 0;
+	}
+
+	/*
+	 * Using a non-uniform random distribution excludes usage of
+	 * a random map
+	 */
+	if (td->o.random_distribution != FIO_RAND_DIST_RANDOM)
+		td->o.norandommap = 1;
+
 	return ret;
 }
 
@@ -449,7 +628,7 @@ static char *to_kmg(unsigned int val)
 		p++;
 	} while (*p);
 
-	snprintf(buf, 31, "%u%c", val, *p);
+	snprintf(buf, 32, "%u%c", val, *p);
 	return buf;
 }
 
@@ -484,44 +663,44 @@ static int exists_and_not_file(const char *filename)
 
 static void td_fill_rand_seeds_os(struct thread_data *td)
 {
-	os_random_seed(td->rand_seeds[0], &td->bsrange_state);
-	os_random_seed(td->rand_seeds[1], &td->verify_state);
-	os_random_seed(td->rand_seeds[2], &td->rwmix_state);
+	os_random_seed(td->rand_seeds[FIO_RAND_BS_OFF], &td->bsrange_state);
+	os_random_seed(td->rand_seeds[FIO_RAND_VER_OFF], &td->verify_state);
+	os_random_seed(td->rand_seeds[FIO_RAND_MIX_OFF], &td->rwmix_state);
 
 	if (td->o.file_service_type == FIO_FSERVICE_RANDOM)
-		os_random_seed(td->rand_seeds[3], &td->next_file_state);
+		os_random_seed(td->rand_seeds[FIO_RAND_FILE_OFF], &td->next_file_state);
 
-	os_random_seed(td->rand_seeds[5], &td->file_size_state);
-	os_random_seed(td->rand_seeds[6], &td->trim_state);
+	os_random_seed(td->rand_seeds[FIO_RAND_FILE_SIZE_OFF], &td->file_size_state);
+	os_random_seed(td->rand_seeds[FIO_RAND_TRIM_OFF], &td->trim_state);
 
 	if (!td_random(td))
 		return;
 
 	if (td->o.rand_repeatable)
-		td->rand_seeds[4] = FIO_RANDSEED * td->thread_number;
+		td->rand_seeds[FIO_RAND_BLOCK_OFF] = FIO_RANDSEED * td->thread_number;
 
-	os_random_seed(td->rand_seeds[4], &td->random_state);
+	os_random_seed(td->rand_seeds[FIO_RAND_BLOCK_OFF], &td->random_state);
 }
 
 static void td_fill_rand_seeds_internal(struct thread_data *td)
 {
-	init_rand_seed(&td->__bsrange_state, td->rand_seeds[0]);
-	init_rand_seed(&td->__verify_state, td->rand_seeds[1]);
-	init_rand_seed(&td->__rwmix_state, td->rand_seeds[2]);
+	init_rand_seed(&td->__bsrange_state, td->rand_seeds[FIO_RAND_BS_OFF]);
+	init_rand_seed(&td->__verify_state, td->rand_seeds[FIO_RAND_VER_OFF]);
+	init_rand_seed(&td->__rwmix_state, td->rand_seeds[FIO_RAND_MIX_OFF]);
 
 	if (td->o.file_service_type == FIO_FSERVICE_RANDOM)
-		init_rand_seed(&td->__next_file_state, td->rand_seeds[3]);
+		init_rand_seed(&td->__next_file_state, td->rand_seeds[FIO_RAND_FILE_OFF]);
 
-	init_rand_seed(&td->__file_size_state, td->rand_seeds[5]);
-	init_rand_seed(&td->__trim_state, td->rand_seeds[6]);
+	init_rand_seed(&td->__file_size_state, td->rand_seeds[FIO_RAND_FILE_SIZE_OFF]);
+	init_rand_seed(&td->__trim_state, td->rand_seeds[FIO_RAND_TRIM_OFF]);
 
 	if (!td_random(td))
 		return;
 
 	if (td->o.rand_repeatable)
-		td->rand_seeds[4] = FIO_RANDSEED * td->thread_number;
+		td->rand_seeds[FIO_RAND_BLOCK_OFF] = FIO_RANDSEED * td->thread_number;
 
-	init_rand_seed(&td->__random_state, td->rand_seeds[4]);
+	init_rand_seed(&td->__random_state, td->rand_seeds[FIO_RAND_BLOCK_OFF]);
 }
 
 void td_fill_rand_seeds(struct thread_data *td)
@@ -531,35 +710,179 @@ void td_fill_rand_seeds(struct thread_data *td)
 	else
 		td_fill_rand_seeds_internal(td);
 
-	init_rand_seed(&td->buf_state, td->rand_seeds[7]);
+	init_rand_seed(&td->buf_state, td->rand_seeds[FIO_RAND_BUF_OFF]);
 }
 
 /*
- * Initialize the various random states we need (random io, block size ranges,
- * read/write mix, etc).
+ * Initializes the ioengine configured for a job, if it has not been done so
+ * already.
  */
-static int init_random_state(struct thread_data *td)
+int ioengine_load(struct thread_data *td)
 {
-	int fd;
+	const char *engine;
 
-	fd = open("/dev/urandom", O_RDONLY);
-	if (fd == -1) {
-		td_verror(td, errno, "open");
+	/*
+	 * Engine has already been loaded.
+	 */
+	if (td->io_ops)
+		return 0;
+
+	engine = get_engine_name(td->o.ioengine);
+	td->io_ops = load_ioengine(td, engine);
+	if (!td->io_ops) {
+		log_err("fio: failed to load engine %s\n", engine);
 		return 1;
 	}
 
-	if (read(fd, td->rand_seeds, sizeof(td->rand_seeds)) <
-	    (int) sizeof(td->rand_seeds)) {
-		td_verror(td, EIO, "read");
-		close(fd);
-		return 1;
+	if (td->io_ops->option_struct_size && td->io_ops->options) {
+		/*
+		 * In cases where td->eo is set, clone it for a child thread.
+		 * This requires that the parent thread has the same ioengine,
+		 * but that requirement must be enforced by the code which
+		 * cloned the thread.
+		 */
+		void *origeo = td->eo;
+		/*
+		 * Otherwise use the default thread options.
+		 */
+		if (!origeo && td != &def_thread && def_thread.eo &&
+		    def_thread.io_ops->options == td->io_ops->options)
+			origeo = def_thread.eo;
+
+		options_init(td->io_ops->options);
+		td->eo = malloc(td->io_ops->option_struct_size);
+		/*
+		 * Use the default thread as an option template if this uses the
+		 * same options structure and there are non-default options
+		 * used.
+		 */
+		if (origeo) {
+			memcpy(td->eo, origeo, td->io_ops->option_struct_size);
+			options_mem_dupe(td->eo, td->io_ops->options);
+		} else {
+			memset(td->eo, 0, td->io_ops->option_struct_size);
+			fill_default_options(td->eo, td->io_ops->options);
+		}
+		*(struct thread_data **)td->eo = td;
+	}
+
+	return 0;
+}
+
+static void init_flags(struct thread_data *td)
+{
+	struct thread_options *o = &td->o;
+
+	if (o->verify_backlog)
+		td->flags |= TD_F_VER_BACKLOG;
+	if (o->trim_backlog)
+		td->flags |= TD_F_TRIM_BACKLOG;
+	if (o->read_iolog_file)
+		td->flags |= TD_F_READ_IOLOG;
+	if (o->refill_buffers)
+		td->flags |= TD_F_REFILL_BUFFERS;
+	if (o->scramble_buffers)
+		td->flags |= TD_F_SCRAMBLE_BUFFERS;
+	if (o->verify != VERIFY_NONE)
+		td->flags |= TD_F_VER_NONE;
+}
+
+static int setup_random_seeds(struct thread_data *td)
+{
+	unsigned long seed;
+	unsigned int i;
+
+	if (!td->o.rand_repeatable)
+		return init_random_state(td, td->rand_seeds, sizeof(td->rand_seeds));
+
+	for (seed = 0x89, i = 0; i < 4; i++)
+		seed *= 0x9e370001UL;
+
+	for (i = 0; i < FIO_RAND_NR_OFFS; i++) {
+		td->rand_seeds[i] = seed;
+		seed *= 0x9e370001UL;
 	}
 
-	close(fd);
 	td_fill_rand_seeds(td);
 	return 0;
 }
 
+enum {
+	FPRE_NONE = 0,
+	FPRE_JOBNAME,
+	FPRE_JOBNUM,
+	FPRE_FILENUM
+};
+
+static struct fpre_keyword {
+	const char *keyword;
+	size_t strlen;
+	int key;
+} fpre_keywords[] = {
+	{ .keyword = "$jobname",	.key = FPRE_JOBNAME, },
+	{ .keyword = "$jobnum",		.key = FPRE_JOBNUM, },
+	{ .keyword = "$filenum",	.key = FPRE_FILENUM, },
+	{ .keyword = NULL, },
+	};
+
+static char *make_filename(char *buf, struct thread_options *o,
+			   const char *jobname, int jobnum, int filenum)
+{
+	struct fpre_keyword *f;
+	char copy[PATH_MAX];
+
+	if (!o->filename_format || !strlen(o->filename_format)) {
+		sprintf(buf, "%s.%d.%d", jobname, jobnum, filenum);
+		return NULL;
+	}
+
+	for (f = &fpre_keywords[0]; f->keyword; f++)
+		f->strlen = strlen(f->keyword);
+
+	strcpy(buf, o->filename_format);
+	memset(copy, 0, sizeof(copy));
+	for (f = &fpre_keywords[0]; f->keyword; f++) {
+		do {
+			size_t pre_len, post_start = 0;
+			char *str, *dst = copy;
+
+			str = strcasestr(buf, f->keyword);
+			if (!str)
+				break;
+
+			pre_len = str - buf;
+			if (strlen(str) != f->strlen)
+				post_start = pre_len + f->strlen;
+
+			if (pre_len) {
+				strncpy(dst, buf, pre_len);
+				dst += pre_len;
+			}
+
+			switch (f->key) {
+			case FPRE_JOBNAME:
+				dst += sprintf(dst, "%s", jobname);
+				break;
+			case FPRE_JOBNUM:
+				dst += sprintf(dst, "%d", jobnum);
+				break;
+			case FPRE_FILENUM:
+				dst += sprintf(dst, "%d", filenum);
+				break;
+			default:
+				assert(0);
+				break;
+			}
+
+			if (post_start)
+				strcpy(dst, buf + post_start);
+
+			strcpy(buf, copy);
+		} while (1);
+	}
+
+	return buf;
+}
 /*
  * Adds a job to the list of things todo. Sanitizes the various options
  * to make sure we don't have conflicts, and initializes various
@@ -568,11 +891,12 @@ static int init_random_state(struct thread_data *td)
 static int add_job(struct thread_data *td, const char *jobname, int job_add_num)
 {
 	const char *ddir_str[] = { NULL, "read", "write", "rw", NULL,
-				   "randread", "randwrite", "randrw" };
+				   "randread", "randwrite", "randrw",
+				   "trim", NULL, NULL, NULL, "randtrim" };
 	unsigned int i;
-	const char *engine;
 	char fname[PATH_MAX];
 	int numjobs, file_alloced;
+	struct thread_options *o = &td->o;
 
 	/*
 	 * the def_thread is just for options, it's not a real job
@@ -580,10 +904,12 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num)
 	if (td == &def_thread)
 		return 0;
 
+	init_flags(td);
+
 	/*
 	 * if we are just dumping the output command line, don't add the job
 	 */
-	if (dump_cmdline) {
+	if (dump_cmdline || parse_only) {
 		put_job(td);
 		return 0;
 	}
@@ -591,39 +917,41 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num)
 	if (profile_td_init(td))
 		goto err;
 
-	engine = get_engine_name(td->o.ioengine);
-	td->io_ops = load_ioengine(td, engine);
-	if (!td->io_ops) {
-		log_err("fio: failed to load engine %s\n", engine);
+	if (ioengine_load(td))
 		goto err;
-	}
 
-	if (td->o.use_thread)
+	if (o->use_thread)
 		nr_thread++;
 	else
 		nr_process++;
 
-	if (td->o.odirect)
+	if (o->odirect)
 		td->io_ops->flags |= FIO_RAWIO;
 
 	file_alloced = 0;
-	if (!td->o.filename && !td->files_index && !td->o.read_iolog_file) {
+	if (!o->filename && !td->files_index && !o->read_iolog_file) {
 		file_alloced = 1;
 
-		if (td->o.nr_files == 1 && exists_and_not_file(jobname))
+		if (o->nr_files == 1 && exists_and_not_file(jobname))
 			add_file(td, jobname);
 		else {
-			for (i = 0; i < td->o.nr_files; i++) {
-				sprintf(fname, "%s.%d.%d", jobname,
-							td->thread_number, i);
-				add_file(td, fname);
-			}
+			for (i = 0; i < o->nr_files; i++)
+				add_file(td, make_filename(fname, o, jobname, td->thread_number, i));
 		}
 	}
 
 	if (fixup_options(td))
 		goto err;
 
+	flow_init_job(td);
+
+	/*
+	 * IO engines only need this for option callbacks, and the address may
+	 * change in subprocesses.
+	 */
+	if (td->eo)
+		*(struct thread_data **)td->eo = NULL;
+
 	if (td->io_ops->flags & FIO_DISKLESSIO) {
 		struct fio_file *f;
 
@@ -631,21 +959,21 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num)
 			f->real_file_size = -1ULL;
 	}
 
-	td->mutex = fio_mutex_init(0);
+	td->mutex = fio_mutex_init(FIO_MUTEX_LOCKED);
 
-	td->ts.clat_percentiles = td->o.clat_percentiles;
-	if (td->o.overwrite_plist)
-		td->ts.percentile_list = td->o.percentile_list;
-	else
-		td->ts.percentile_list = NULL;
+	td->ts.clat_percentiles = o->clat_percentiles;
+	td->ts.percentile_precision = o->percentile_precision;
+	memcpy(td->ts.percentile_list, o->percentile_list, sizeof(o->percentile_list));
 
-	td->ts.clat_stat[0].min_val = td->ts.clat_stat[1].min_val = ULONG_MAX;
-	td->ts.slat_stat[0].min_val = td->ts.slat_stat[1].min_val = ULONG_MAX;
-	td->ts.lat_stat[0].min_val = td->ts.lat_stat[1].min_val = ULONG_MAX;
-	td->ts.bw_stat[0].min_val = td->ts.bw_stat[1].min_val = ULONG_MAX;
-	td->ddir_seq_nr = td->o.ddir_seq_nr;
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
+		td->ts.clat_stat[i].min_val = ULONG_MAX;
+		td->ts.slat_stat[i].min_val = ULONG_MAX;
+		td->ts.lat_stat[i].min_val = ULONG_MAX;
+		td->ts.bw_stat[i].min_val = ULONG_MAX;
+	}
+	td->ddir_seq_nr = o->ddir_seq_nr;
 
-	if ((td->o.stonewall || td->o.new_group) && prev_group_jobs) {
+	if ((o->stonewall || o->new_group) && prev_group_jobs) {
 		prev_group_jobs = 0;
 		groupid++;
 	}
@@ -653,50 +981,56 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num)
 	td->groupid = groupid;
 	prev_group_jobs++;
 
-	if (init_random_state(td))
+	if (setup_random_seeds(td)) {
+		td_verror(td, errno, "init_random_state");
 		goto err;
+	}
 
 	if (setup_rate(td))
 		goto err;
 
-	if (td->o.write_lat_log) {
-		setup_log(&td->ts.lat_log);
-		setup_log(&td->ts.slat_log);
-		setup_log(&td->ts.clat_log);
+	if (o->write_lat_log) {
+		setup_log(&td->lat_log, o->log_avg_msec);
+		setup_log(&td->slat_log, o->log_avg_msec);
+		setup_log(&td->clat_log, o->log_avg_msec);
 	}
-	if (td->o.write_bw_log)
-		setup_log(&td->ts.bw_log);
+	if (o->write_bw_log)
+		setup_log(&td->bw_log, o->log_avg_msec);
+	if (o->write_iops_log)
+		setup_log(&td->iops_log, o->log_avg_msec);
 
-	if (!td->o.name)
-		td->o.name = strdup(jobname);
+	if (!o->name)
+		o->name = strdup(jobname);
 
-	if (!terse_output) {
+	if (output_format == FIO_OUTPUT_NORMAL) {
 		if (!job_add_num) {
 			if (!strcmp(td->io_ops->name, "cpuio")) {
 				log_info("%s: ioengine=cpu, cpuload=%u,"
-					 " cpucycle=%u\n", td->o.name,
-							td->o.cpuload,
-							td->o.cpucycle);
+					 " cpucycle=%u\n", o->name,
+						o->cpuload, o->cpucycle);
 			} else {
-				char *c1, *c2, *c3, *c4;
+				char *c1, *c2, *c3, *c4, *c5, *c6;
 
-				c1 = to_kmg(td->o.min_bs[DDIR_READ]);
-				c2 = to_kmg(td->o.max_bs[DDIR_READ]);
-				c3 = to_kmg(td->o.min_bs[DDIR_WRITE]);
-				c4 = to_kmg(td->o.max_bs[DDIR_WRITE]);
+				c1 = to_kmg(o->min_bs[DDIR_READ]);
+				c2 = to_kmg(o->max_bs[DDIR_READ]);
+				c3 = to_kmg(o->min_bs[DDIR_WRITE]);
+				c4 = to_kmg(o->max_bs[DDIR_WRITE]);
+				c5 = to_kmg(o->min_bs[DDIR_TRIM]);
+				c6 = to_kmg(o->max_bs[DDIR_TRIM]);
 
-				log_info("%s: (g=%d): rw=%s, bs=%s-%s/%s-%s,"
+				log_info("%s: (g=%d): rw=%s, bs=%s-%s/%s-%s/%s-%s,"
 					 " ioengine=%s, iodepth=%u\n",
-						td->o.name, td->groupid,
-						ddir_str[td->o.td_ddir],
-						c1, c2, c3, c4,
-						td->io_ops->name,
-						td->o.iodepth);
+						o->name, td->groupid,
+						ddir_str[o->td_ddir],
+						c1, c2, c3, c4, c5, c6,
+						td->io_ops->name, o->iodepth);
 
 				free(c1);
 				free(c2);
 				free(c3);
 				free(c4);
+				free(c5);
+				free(c6);
 			}
 		} else if (job_add_num == 1)
 			log_info("...\n");
@@ -706,9 +1040,9 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num)
 	 * recurse add identical jobs, clear numjobs and stonewall options
 	 * as they don't apply to sub-jobs
 	 */
-	numjobs = td->o.numjobs;
+	numjobs = o->numjobs;
 	while (--numjobs) {
-		struct thread_data *td_new = get_new_job(0, td);
+		struct thread_data *td_new = get_new_job(0, td, 1);
 
 		if (!td_new)
 			goto err;
@@ -756,11 +1090,11 @@ void add_job_opts(const char **o)
 			sprintf(jobname, "%s", o[i] + 5);
 		}
 		if (in_global && !td_parent)
-			td_parent = get_new_job(1, &def_thread);
+			td_parent = get_new_job(1, &def_thread, 0);
 		else if (!in_global && !td) {
 			if (!td_parent)
 				td_parent = &def_thread;
-			td = get_new_job(0, td_parent);
+			td = get_new_job(0, td_parent, 0);
 		}
 		if (in_global)
 			fio_options_parse(td_parent, (char **) &o[i], 1);
@@ -858,7 +1192,7 @@ int parse_jobs_ini(char *file, int is_buf, int stonewall_flag)
 			if (is_buf)
 				p = strsep(&file, "\n");
 			else
-				p = fgets(string, 4095, f);
+				p = fgets(string, 4096, f);
 			if (!p)
 				break;
 		}
@@ -895,7 +1229,7 @@ int parse_jobs_ini(char *file, int is_buf, int stonewall_flag)
 			first_sect = 0;
 		}
 
-		td = get_new_job(global, &def_thread);
+		td = get_new_job(global, &def_thread, 0);
 		if (!td) {
 			ret = 1;
 			break;
@@ -966,6 +1300,12 @@ int parse_jobs_ini(char *file, int is_buf, int stonewall_flag)
 	if (dump_cmdline)
 		log_info("\n");
 
+	i = 0;
+	while (i < nr_job_sections) {
+		free(job_sections[i]);
+		i++;
+	}
+
 	for (i = 0; i < num_opts; i++)
 		free(opts[i]);
 
@@ -982,7 +1322,8 @@ static int fill_def_thread(void)
 	memset(&def_thread, 0, sizeof(def_thread));
 
 	fio_getaffinity(getpid(), &def_thread.o.cpumask);
-
+	def_thread.o.timeout = def_timeout;
+	def_thread.o.error_dump = 1;
 	/*
 	 * fill default options
 	 */
@@ -990,100 +1331,48 @@ static int fill_def_thread(void)
 	return 0;
 }
 
-static void free_shm(void)
-{
-	struct shmid_ds sbuf;
-
-	if (threads) {
-		void *tp = threads;
-
-		threads = NULL;
-		file_hash_exit();
-		fio_debug_jobp = NULL;
-		shmdt(tp);
-		shmctl(shm_id, IPC_RMID, &sbuf);
-	}
-
-	scleanup();
-}
-
-/*
- * The thread area is shared between the main process and the job
- * threads/processes. So setup a shared memory segment that will hold
- * all the job info. We use the end of the region for keeping track of
- * open files across jobs, for file sharing.
- */
-static int setup_thread_area(void)
-{
-	void *hash;
-
-	/*
-	 * 1024 is too much on some machines, scale max_jobs if
-	 * we get a failure that looks like too large a shm segment
-	 */
-	do {
-		size_t size = max_jobs * sizeof(struct thread_data);
-
-		size += file_hash_size;
-		size += sizeof(unsigned int);
-
-		shm_id = shmget(0, size, IPC_CREAT | 0600);
-		if (shm_id != -1)
-			break;
-		if (errno != EINVAL) {
-			perror("shmget");
-			break;
-		}
-
-		max_jobs >>= 1;
-	} while (max_jobs);
-
-	if (shm_id == -1)
-		return 1;
-
-	threads = shmat(shm_id, NULL, 0);
-	if (threads == (void *) -1) {
-		perror("shmat");
-		return 1;
-	}
-
-	memset(threads, 0, max_jobs * sizeof(struct thread_data));
-	hash = (void *) threads + max_jobs * sizeof(struct thread_data);
-	fio_debug_jobp = (void *) hash + file_hash_size;
-	*fio_debug_jobp = -1;
-	file_hash_init(hash);
-	atexit(free_shm);
-	return 0;
-}
-
 static void usage(const char *name)
 {
 	printf("%s\n", fio_version_string);
 	printf("%s [options] [job options] <job file(s)>\n", name);
-	printf("\t--debug=options\tEnable debug logging\n");
-	printf("\t--output\tWrite output to file\n");
-	printf("\t--timeout\tRuntime in seconds\n");
-	printf("\t--latency-log\tGenerate per-job latency logs\n");
-	printf("\t--bandwidth-log\tGenerate per-job bandwidth logs\n");
-	printf("\t--minimal\tMinimal (terse) output\n");
-	printf("\t--version\tPrint version info and exit\n");
-	printf("\t--terse-version=x Terse version output format\n");
-	printf("\t--help\t\tPrint this page\n");
-	printf("\t--cmdhelp=cmd\tPrint command help, \"all\" for all of"
+	printf("  --debug=options\tEnable debug logging. May be one/more of:\n"
+		"\t\t\tprocess,file,io,mem,blktrace,verify,random,parse,\n"
+		"\t\t\tdiskutil,job,mutex,profile,time,net\n");
+	printf("  --parse-only\t\tParse options only, don't start any IO\n");
+	printf("  --output\t\tWrite output to file\n");
+	printf("  --runtime\t\tRuntime in seconds\n");
+	printf("  --latency-log\t\tGenerate per-job latency logs\n");
+	printf("  --bandwidth-log\tGenerate per-job bandwidth logs\n");
+	printf("  --minimal\t\tMinimal (terse) output\n");
+	printf("  --output-format=x\tOutput format (terse,json,normal)\n");
+	printf("  --terse-version=x\tSet terse version output format to 'x'\n");
+	printf("  --version\t\tPrint version info and exit\n");
+	printf("  --help\t\tPrint this page\n");
+	printf("  --cpuclock-test\tPerform test/validation of CPU clock\n");
+	printf("  --cmdhelp=cmd\t\tPrint command help, \"all\" for all of"
 		" them\n");
-	printf("\t--showcmd\tTurn a job file into command line options\n");
-	printf("\t--eta=when\tWhen ETA estimate should be printed\n");
-	printf("\t          \tMay be \"always\", \"never\" or \"auto\"\n");
-	printf("\t--readonly\tTurn on safety read-only checks, preventing"
+	printf("  --enghelp=engine\tPrint ioengine help, or list"
+		" available ioengines\n");
+	printf("  --enghelp=engine,cmd\tPrint help for an ioengine"
+		" cmd\n");
+	printf("  --showcmd\t\tTurn a job file into command line options\n");
+	printf("  --eta=when\t\tWhen ETA estimate should be printed\n");
+	printf("            \t\tMay be \"always\", \"never\" or \"auto\"\n");
+	printf("  --eta-newline=time\tForce a new line for every 'time'");
+	printf(" period passed\n");
+	printf("  --readonly\t\tTurn on safety read-only checks, preventing"
 		" writes\n");
-	printf("\t--section=name\tOnly run specified section in job file\n");
-	printf("\t--alloc-size=kb\tSet smalloc pool to this size in kb"
+	printf("  --section=name\tOnly run specified section in job file\n");
+	printf("  --alloc-size=kb\tSet smalloc pool to this size in kb"
 		" (def 1024)\n");
-	printf("\t--warnings-fatal Fio parser warnings are fatal\n");
-	printf("\t--max-jobs\tMaximum number of threads/processes to support\n");
-	printf("\t--server\tStart a backend fio server\n");
-	printf("\t--client=hostname Talk to remove backend fio server at hostname\n");
-	printf("\t--net-port=port\tUse specified port for client/server connection\n");
+	printf("  --warnings-fatal\tFio parser warnings are fatal\n");
+	printf("  --max-jobs=nr\t\tMaximum number of threads/processes to support\n");
+	printf("  --server=args\t\tStart a backend fio server\n");
+	printf("  --daemonize=pidfile\tBackground fio server, write pid to file\n");
+	printf("  --client=hostname\tTalk to remote backend fio server at hostname\n");
+	printf("  --idle-prof=option\tReport cpu idleness on a system or percpu basis\n"
+		"\t\t\t(option=system,percpu) or run unit work\n"
+		"\t\t\tcalibration only (option=calibrate)\n");
 	printf("\nFio was written by Jens Axboe <jens.axboe@oracle.com>");
 	printf("\n                   Jens Axboe <jaxboe@fusionio.com>\n");
 }
@@ -1103,6 +1392,7 @@ struct debug_level debug_levels[] = {
 	{ .name = "mutex",	.shift = FD_MUTEX },
 	{ .name	= "profile",	.shift = FD_PROFILE },
 	{ .name = "time",	.shift = FD_TIME },
+	{ .name = "net",	.shift = FD_NET },
 	{ .name = NULL, },
 };
 
@@ -1187,19 +1477,57 @@ static void fio_options_fill_optstring(void)
 	ostr[c] = '\0';
 }
 
-static int parse_cmd_line(int argc, char *argv[])
+static int client_flag_set(char c)
+{
+	int i;
+
+	i = 0;
+	while (l_opts[i].name) {
+		int val = l_opts[i].val;
+
+		if (c == (val & 0xff))
+			return (val & FIO_CLIENT_FLAG);
+
+		i++;
+	}
+
+	return 0;
+}
+
+void parse_cmd_client(void *client, char *opt)
+{
+	fio_client_add_cmd_option(client, opt);
+}
+
+int parse_cmd_line(int argc, char *argv[])
 {
 	struct thread_data *td = NULL;
 	int c, ini_idx = 0, lidx, ret = 0, do_exit = 0, exit_val = 0;
 	char *ostr = cmd_optstr;
+	void *pid_file = NULL;
+	void *cur_client = NULL;
+	int backend = 0;
+
+	/*
+	 * Reset optind handling, since we may call this multiple times
+	 * for the backend.
+	 */
+	optind = 1;
 
 	while ((c = getopt_long_only(argc, argv, ostr, l_opts, &lidx)) != -1) {
+		did_arg = 1;
+
+		if ((c & FIO_CLIENT_FLAG) || client_flag_set(c)) {
+			parse_cmd_client(cur_client, argv[optind - 1]);
+			c &= ~FIO_CLIENT_FLAG;
+		}
+
 		switch (c) {
 		case 'a':
 			smalloc_pool_size = atoi(optarg);
 			break;
 		case 't':
-			def_thread.o.timeout = atoi(optarg);
+			def_timeout = atoi(optarg);
 			break;
 		case 'l':
 			write_lat_log = 1;
@@ -1216,13 +1544,36 @@ static int parse_cmd_line(int argc, char *argv[])
 			f_err = f_out;
 			break;
 		case 'm':
-			terse_output = 1;
+			output_format = FIO_OUTPUT_TERSE;
+			break;
+		case 'F':
+			if (!strcmp(optarg, "minimal") ||
+			    !strcmp(optarg, "terse") ||
+			    !strcmp(optarg, "csv"))
+				output_format = FIO_OUTPUT_TERSE;
+			else if (!strcmp(optarg, "json"))
+				output_format = FIO_OUTPUT_JSON;
+			else
+				output_format = FIO_OUTPUT_NORMAL;
 			break;
 		case 'h':
-			usage(argv[0]);
-			exit(0);
+			if (!cur_client) {
+				usage(argv[0]);
+				do_exit++;
+			}
+			break;
 		case 'c':
-			exit(fio_show_option_help(optarg));
+			if (!cur_client) {
+				fio_show_option_help(optarg);
+				do_exit++;
+			}
+			break;
+		case 'i':
+			if (!cur_client) {
+				fio_show_ioengine_help(optarg);
+				do_exit++;
+			}
+			break;
 		case 's':
 			dump_cmdline = 1;
 			break;
@@ -1230,11 +1581,15 @@ static int parse_cmd_line(int argc, char *argv[])
 			read_only = 1;
 			break;
 		case 'v':
-			log_info("%s\n", fio_version_string);
-			exit(0);
+			if (!cur_client) {
+				log_info("%s\n", fio_version_string);
+				do_exit++;
+			}
+			break;
 		case 'V':
 			terse_version = atoi(optarg);
-			if (terse_version != 2) {
+			if (!(terse_version == 2 || terse_version == 3 ||
+			     terse_version == 4)) {
 				log_err("fio: bad terse version format\n");
 				exit_val = 1;
 				do_exit++;
@@ -1246,10 +1601,24 @@ static int parse_cmd_line(int argc, char *argv[])
 			else if (!strcmp("never", optarg))
 				eta_print = FIO_ETA_NEVER;
 			break;
+		case 'E': {
+			long long t = 0;
+
+			if (str_to_decimal(optarg, &t, 0, NULL)) {
+				log_err("fio: failed parsing eta time %s\n", optarg);
+				exit_val = 1;
+				do_exit++;
+			}
+			eta_new_line = t;
+			break;
+			}
 		case 'd':
 			if (set_debug(optarg))
 				do_exit++;
 			break;
+		case 'P':
+			parse_only = 1;
+			break;
 		case 'x': {
 			size_t new_size;
 
@@ -1289,12 +1658,28 @@ static int parse_cmd_line(int argc, char *argv[])
 				if (is_section && skip_this_section(val))
 					continue;
 
-				td = get_new_job(global, &def_thread);
-				if (!td)
+				td = get_new_job(global, &def_thread, 1);
+				if (!td || ioengine_load(td))
 					return 0;
+				fio_options_set_ioengine_opts(l_opts, td);
 			}
 
 			ret = fio_cmd_option_parse(td, opt, val);
+
+			if (!ret && !strcmp(opt, "ioengine")) {
+				free_ioengine(td);
+				if (ioengine_load(td))
+					return 0;
+				fio_options_set_ioengine_opts(l_opts, td);
+			}
+			break;
+		}
+		case FIO_GETOPT_IOENGINE: {
+			const char *opt = l_opts[lidx].name;
+			char *val = optarg;
+			opt = l_opts[lidx].name;
+			val = optarg;
+			ret = fio_cmd_ioengine_option_parse(td, opt, val);
 			break;
 		}
 		case 'w':
@@ -1309,16 +1694,27 @@ static int parse_cmd_line(int argc, char *argv[])
 			}
 			break;
 		case 'S':
-			if (is_client) {
+			if (nr_clients) {
 				log_err("fio: can't be both client and server\n");
 				do_exit++;
 				exit_val = 1;
 				break;
 			}
+			if (optarg)
+				fio_server_set_arg(optarg);
 			is_backend = 1;
+			backend = 1;
 			break;
-		case 'P':
-			fio_net_port = atoi(optarg);
+		case 'D':
+			pid_file = strdup(optarg);
+			break;
+		case 'I':
+			if ((ret = fio_idle_prof_parse_opt(optarg))) {
+				/* exit on error and calibration only */
+				do_exit++;
+				if (ret == -1) 
+					exit_val = 1;
+			}
 			break;
 		case 'C':
 			if (is_backend) {
@@ -1327,34 +1723,61 @@ static int parse_cmd_line(int argc, char *argv[])
 				exit_val = 1;
 				break;
 			}
-			is_client = 1;
-			client = strdup(optarg);
+			if (fio_client_add(optarg, &cur_client)) {
+				log_err("fio: failed adding client %s\n", optarg);
+				do_exit++;
+				exit_val = 1;
+				break;
+			}
+			/*
+			 * If the next argument exists and isn't an option,
+			 * assume it's a job file for this client only.
+			 */
+			while (optind < argc) {
+				if (!strncmp(argv[optind], "--", 2) ||
+				    !strncmp(argv[optind], "-", 1))
+					break;
+
+				fio_client_add_ini_file(cur_client, argv[optind]);
+				optind++;
+			}
+			break;
+		case 'T':
+			do_exit++;
+			exit_val = fio_monotonic_clocktest();
 			break;
+		case '?':
+			log_err("%s: unrecognized option '%s'\n", argv[0],
+							argv[optind - 1]);
 		default:
 			do_exit++;
 			exit_val = 1;
 			break;
 		}
+		if (do_exit)
+			break;
 	}
 
-	if (do_exit)
-		exit(exit_val);
+	if (do_exit) {
+		if (exit_val && !(is_backend || nr_clients))
+			exit(exit_val);
+	}
 
-	if (is_client && fio_client_connect(client)) {
+	if (nr_clients && fio_clients_connect()) {
 		do_exit++;
 		exit_val = 1;
-		return 1;
+		return -1;
 	}
 
-	if (is_backend)
-		return fio_server();
+	if (is_backend && backend)
+		return fio_start_server(pid_file);
 
 	if (td) {
 		if (!ret)
 			ret = add_job(td, td->o.name ?: "fio", 0);
 	}
 
-	while (optind < argc) {
+	while (!ret && optind < argc) {
 		ini_idx++;
 		ini_file = realloc(ini_file, ini_idx * sizeof(char *));
 		ini_file[ini_idx - 1] = strdup(argv[optind]);
@@ -1374,40 +1797,55 @@ int parse_options(int argc, char *argv[])
 	fio_options_fill_optstring();
 	fio_options_dup_and_init(l_opts);
 
-	if (setup_thread_area())
-		return 1;
+	atexit(free_shm);
+
 	if (fill_def_thread())
 		return 1;
 
 	job_files = parse_cmd_line(argc, argv);
 
-	for (i = 0; i < job_files; i++) {
-		if (fill_def_thread())
-			return 1;
-		if (is_client) {
-			if (fio_client_send_ini(ini_file[i]))
-				return 1;
-		} else {
-			if (parse_jobs_ini(ini_file[i], 0, i))
+	if (job_files > 0) {
+		for (i = 0; i < job_files; i++) {
+			if (fill_def_thread())
 				return 1;
+			if (nr_clients) {
+				if (fio_clients_send_ini(ini_file[i]))
+					return 1;
+				free(ini_file[i]);
+			} else if (!is_backend) {
+				if (parse_jobs_ini(ini_file[i], 0, i))
+					return 1;
+				free(ini_file[i]);
+			}
 		}
-		free(ini_file[i]);
+	} else if (nr_clients) {
+		if (fill_def_thread())
+			return 1;
+		if (fio_clients_send_ini(NULL))
+			return 1;
 	}
 
 	free(ini_file);
-	options_mem_free(&def_thread);
+	fio_options_free(&def_thread);
 
 	if (!thread_number) {
-		if (dump_cmdline)
+		if (dump_cmdline || parse_only)
 			return 0;
 		if (exec_profile)
 			return 0;
-		if (is_backend || is_client)
+		if (is_backend || nr_clients)
+			return 0;
+		if (did_arg)
 			return 0;
 
 		log_err("No jobs(s) defined\n\n");
-		usage(argv[0]);
-		return 1;
+
+		if (!did_arg) {
+			usage(argv[0]);
+			return 1;
+		}
+
+		return 0;
 	}
 
 	if (def_thread.o.gtod_offload) {
@@ -1416,6 +1854,8 @@ int parse_options(int argc, char *argv[])
 		fio_gtod_cpu = def_thread.o.gtod_cpu;
 	}
 
-	log_info("%s\n", fio_version_string);
+	if (output_format == FIO_OUTPUT_NORMAL)
+		log_info("%s\n", fio_version_string);
+
 	return 0;
 }