X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=init.c;h=e80aec300770036346cfcc72d56f53d3c9623c65;hp=82b21c78568867d5fc901bc7e35eabcd7d38a714;hb=2ae7a90df2d2340c4503be8a91526f80b3b96789;hpb=d5b78f5a8fd73d21c1b84d71993c5690dbe098b5

diff --git a/init.c b/init.c
index 82b21c78..e80aec30 100644
--- a/init.c
+++ b/init.c
@@ -25,11 +25,13 @@
 #include "server.h"
 #include "idletime.h"
 #include "filelock.h"
+#include "steadystate.h"
 
 #include "oslib/getopt.h"
 #include "oslib/strcasestr.h"
 
 #include "crc/test.h"
+#include "lib/pow2.h"
 
 const char fio_version_string[] = FIO_VERSION;
 
@@ -38,7 +40,6 @@ const char fio_version_string[] = FIO_VERSION;
 static char **ini_file;
 static int max_jobs = FIO_MAX_JOBS;
 static int dump_cmdline;
-static long long def_timeout;
 static int parse_only;
 
 static struct thread_data def_thread;
@@ -47,7 +48,6 @@ static char **job_sections;
 static int nr_job_sections;
 
 int exitall_on_terminate = 0;
-int exitall_on_terminate_error = 0;
 int output_format = FIO_OUTPUT_NORMAL;
 int eta_print = FIO_ETA_AUTO;
 int eta_new_line = 0;
@@ -92,11 +92,6 @@ static struct option l_opts[FIO_NR_OPTIONS] = {
 		.has_arg	= required_argument,
 		.val		= 'o' | FIO_CLIENT_FLAG,
 	},
-	{
-		.name		= (char *) "timeout",
-		.has_arg	= required_argument,
-		.val		= 't' | FIO_CLIENT_FLAG,
-	},
 	{
 		.name		= (char *) "latency-log",
 		.has_arg	= required_argument,
@@ -104,7 +99,7 @@ static struct option l_opts[FIO_NR_OPTIONS] = {
 	},
 	{
 		.name		= (char *) "bandwidth-log",
-		.has_arg	= required_argument,
+		.has_arg	= no_argument,
 		.val		= 'b' | FIO_CLIENT_FLAG,
 	},
 	{
@@ -114,7 +109,7 @@ static struct option l_opts[FIO_NR_OPTIONS] = {
 	},
 	{
 		.name		= (char *) "output-format",
-		.has_arg	= optional_argument,
+		.has_arg	= required_argument,
 		.val		= 'F' | FIO_CLIENT_FLAG,
 	},
 	{
@@ -299,7 +294,6 @@ void free_threads_shm(void)
 static void free_shm(void)
 {
 	if (threads) {
-		file_hash_exit();
 		flow_exit();
 		fio_debug_jobp = NULL;
 		free_threads_shm();
@@ -310,8 +304,9 @@ static void free_shm(void)
 	free(trigger_remote_cmd);
 	trigger_file = trigger_cmd = trigger_remote_cmd = NULL;
 
-	options_free(fio_options, &def_thread);
+	options_free(fio_options, &def_thread.o);
 	fio_filelock_exit();
+	file_hash_exit();
 	scleanup();
 }
 
@@ -323,8 +318,6 @@ static void free_shm(void)
  */
 static int setup_thread_area(void)
 {
-	void *hash;
-
 	if (threads)
 		return 0;
 
@@ -335,7 +328,6 @@ static int setup_thread_area(void)
 	do {
 		size_t size = max_jobs * sizeof(struct thread_data);
 
-		size += file_hash_size;
 		size += sizeof(unsigned int);
 
 #ifndef CONFIG_NO_SHM
@@ -364,27 +356,19 @@ static int setup_thread_area(void)
 		perror("shmat");
 		return 1;
 	}
+	if (shm_attach_to_open_removed())
+		shmctl(shm_id, IPC_RMID, NULL);
 #endif
 
 	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 = (unsigned int *)(threads + max_jobs);
 	*fio_debug_jobp = -1;
-	file_hash_init(hash);
 
 	flow_init();
 
 	return 0;
 }
 
-static void set_cmd_options(struct thread_data *td)
-{
-	struct thread_options *o = &td->o;
-
-	if (!o->timeout)
-		o->timeout = def_timeout;
-}
-
 static void dump_print_option(struct print_option *p)
 {
 	const char *delim;
@@ -450,15 +434,13 @@ static void copy_opt_list(struct thread_data *dst, struct thread_data *src)
 /*
  * Return a free job structure.
  */
-static struct thread_data *get_new_job(int global, struct thread_data *parent,
-				       int preserve_eo, const char *jobname)
+static struct thread_data *get_new_job(bool global, struct thread_data *parent,
+				       bool preserve_eo, const char *jobname)
 {
 	struct thread_data *td;
 
-	if (global) {
-		set_cmd_options(&def_thread);
+	if (global)
 		return &def_thread;
-	}
 	if (setup_thread_area()) {
 		log_err("error: failed to setup shm segment\n");
 		return NULL;
@@ -477,6 +459,7 @@ static struct thread_data *get_new_job(int global, struct thread_data *parent,
 		copy_opt_list(td, parent);
 
 	td->io_ops = NULL;
+	td->io_ops_init = 0;
 	if (!preserve_eo)
 		td->eo = NULL;
 
@@ -496,7 +479,6 @@ static struct thread_data *get_new_job(int global, struct thread_data *parent,
 	if (!parent->o.group_reporting || parent == &def_thread)
 		stat_number++;
 
-	set_cmd_options(td);
 	return td;
 }
 
@@ -541,7 +523,7 @@ static int __setup_rate(struct thread_data *td, enum fio_ddir ddir)
 
 	td->rate_next_io_time[ddir] = 0;
 	td->rate_io_issue_bytes[ddir] = 0;
-	td->last_usec = 0;
+	td->last_usec[ddir] = 0;
 	return 0;
 }
 
@@ -585,6 +567,17 @@ static unsigned long long get_rand_start_delay(struct thread_data *td)
 	return delayrange;
 }
 
+/*
+ * <3 Johannes
+ */
+static unsigned int gcd(unsigned int m, unsigned int n)
+{
+	if (!n)
+		return m;
+
+	return gcd(n, m % n);
+}
+
 /*
  * Lazy way of fixing up options that depend on each other. We could also
  * define option callback handlers, but this is easier.
@@ -594,7 +587,7 @@ static int fixup_options(struct thread_data *td)
 	struct thread_options *o = &td->o;
 	int ret = 0;
 
-#ifndef FIO_HAVE_PSHARED_MUTEX
+#ifndef CONFIG_PSHARED
 	if (!o->use_thread) {
 		log_info("fio: this platform does not support process shared"
 			 " mutexes, forcing use of threads. Use the 'thread'"
@@ -627,7 +620,7 @@ static int fixup_options(struct thread_data *td)
 	/*
 	 * Reads can do overwrites, we always need to pre-create the file
 	 */
-	if (td_read(td) || td_rw(td))
+	if (td_read(td))
 		o->overwrite = 1;
 
 	if (!o->min_bs[DDIR_READ])
@@ -677,7 +670,7 @@ static int fixup_options(struct thread_data *td)
 			"verify limited\n");
 		ret = warnings_fatal;
 	}
-	if (o->bs_unaligned && (o->odirect || td->io_ops->flags & FIO_RAWIO))
+	if (o->bs_unaligned && (o->odirect || td_ioengine_flagged(td, FIO_RAWIO)))
 		log_err("fio: bs_unaligned may not work with raw io\n");
 
 	/*
@@ -705,6 +698,23 @@ static int fixup_options(struct thread_data *td)
 	if (o->iodepth_batch_complete_min > o->iodepth_batch_complete_max)
 		o->iodepth_batch_complete_max = o->iodepth_batch_complete_min;
 
+	/*
+	 * There's no need to check for in-flight overlapping IOs if the job
+	 * isn't changing data or the maximum iodepth is guaranteed to be 1
+	 */
+	if (o->serialize_overlap && !(td->flags & TD_F_READ_IOLOG) &&
+	    (!(td_write(td) || td_trim(td)) || o->iodepth == 1))
+		o->serialize_overlap = 0;
+	/*
+	 * Currently can't check for overlaps in offload mode
+	 */
+	if (o->serialize_overlap && o->io_submit_mode == IO_MODE_OFFLOAD) {
+		log_err("fio: checking for in-flight overlaps when the "
+			"io_submit_mode is offload is not supported\n");
+		o->serialize_overlap = 0;
+		ret = warnings_fatal;
+	}
+
 	if (o->nr_files > td->files_index)
 		o->nr_files = td->files_index;
 
@@ -738,13 +748,30 @@ static int fixup_options(struct thread_data *td)
 		o->size = -1ULL;
 
 	if (o->verify != VERIFY_NONE) {
-		if (td_write(td) && o->do_verify && o->numjobs > 1) {
-			log_info("Multiple writers may overwrite blocks that "
-				"belong to other jobs. This can cause "
+		if (td_write(td) && o->do_verify && o->numjobs > 1 &&
+		    (o->filename ||
+		     !(o->unique_filename &&
+		       strstr(o->filename_format, "$jobname") &&
+		       strstr(o->filename_format, "$jobnum") &&
+		       strstr(o->filename_format, "$filenum")))) {
+			log_info("fio: multiple writers may overwrite blocks "
+				"that belong to other jobs. This can cause "
 				"verification failures.\n");
 			ret = warnings_fatal;
 		}
 
+		/*
+		 * Warn if verification is requested but no verification of any
+		 * kind can be started due to time constraints
+		 */
+		if (td_write(td) && o->do_verify && o->timeout &&
+		    o->time_based && !td_read(td) && !o->verify_backlog) {
+			log_info("fio: verification read phase will never "
+				 "start because write phase uses all of "
+				 "runtime\n");
+			ret = warnings_fatal;
+		}
+
 		if (!fio_option_is_set(o, refill_buffers))
 			o->refill_buffers = 1;
 
@@ -760,11 +787,21 @@ static int fixup_options(struct thread_data *td)
 			o->verify_interval = o->min_bs[DDIR_WRITE];
 		else if (td_read(td) && o->verify_interval > o->min_bs[DDIR_READ])
 			o->verify_interval = o->min_bs[DDIR_READ];
+
+		/*
+		 * Verify interval must be a factor or both min and max
+		 * write size
+		 */
+		if (o->verify_interval % o->min_bs[DDIR_WRITE] ||
+		    o->verify_interval % o->max_bs[DDIR_WRITE])
+			o->verify_interval = gcd(o->min_bs[DDIR_WRITE],
+							o->max_bs[DDIR_WRITE]);
 	}
 
 	if (o->pre_read) {
-		o->invalidate_cache = 0;
-		if (td->io_ops->flags & FIO_PIPEIO) {
+		if (o->invalidate_cache)
+			o->invalidate_cache = 0;
+		if (td_ioengine_flagged(td, FIO_PIPEIO)) {
 			log_info("fio: cannot pre-read files with an IO engine"
 				 " that isn't seekable. Pre-read disabled.\n");
 			ret = warnings_fatal;
@@ -772,12 +809,17 @@ static int fixup_options(struct thread_data *td)
 	}
 
 	if (!o->unit_base) {
-		if (td->io_ops->flags & FIO_BIT_BASED)
+		if (td_ioengine_flagged(td, FIO_BIT_BASED))
 			o->unit_base = 1;
 		else
 			o->unit_base = 8;
 	}
 
+#ifndef FIO_HAVE_ANY_FALLOCATE
+	/* Platform doesn't support any fallocate so force it to none */
+	o->fallocate_mode = FIO_FALLOCATE_NONE;
+#endif
+
 #ifndef CONFIG_FDATASYNC
 	if (o->fdatasync_blocks) {
 		log_info("fio: this platform does not support fdatasync()"
@@ -795,7 +837,7 @@ static int fixup_options(struct thread_data *td)
 	 * 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)) {
+	if (td_ioengine_flagged(td, FIO_SYNCIO) && (o->odirect || 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");
@@ -821,30 +863,31 @@ static int fixup_options(struct thread_data *td)
 	 * 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;
+	if (o->random_distribution != FIO_RAND_DIST_RANDOM)
+		o->norandommap = 1;
 
 	/*
 	 * If size is set but less than the min block size, complain
 	 */
 	if (o->size && o->size < td_min_bs(td)) {
-		log_err("fio: size too small, must be larger than the IO size: %llu\n", (unsigned long long) o->size);
+		log_err("fio: size too small, must not be less than minimum block size: %llu < %u\n",
+			(unsigned long long) o->size, td_min_bs(td));
 		ret = 1;
 	}
 
 	/*
 	 * O_ATOMIC implies O_DIRECT
 	 */
-	if (td->o.oatomic)
-		td->o.odirect = 1;
+	if (o->oatomic)
+		o->odirect = 1;
 
 	/*
 	 * If randseed is set, that overrides randrepeat
 	 */
-	if (fio_option_is_set(&td->o, rand_seed))
-		td->o.rand_repeatable = 0;
+	if (fio_option_is_set(o, rand_seed))
+		o->rand_repeatable = 0;
 
-	if ((td->io_ops->flags & FIO_NOEXTEND) && td->o.file_append) {
+	if (td_ioengine_flagged(td, FIO_NOEXTEND) && o->file_append) {
 		log_err("fio: can't append/extent with IO engine %s\n", td->io_ops->name);
 		ret = 1;
 	}
@@ -859,69 +902,28 @@ static int fixup_options(struct thread_data *td)
 	if (!td->loops)
 		td->loops = 1;
 
-	if (td->o.block_error_hist && td->o.nr_files != 1) {
+	if (o->block_error_hist && o->nr_files != 1) {
 		log_err("fio: block error histogram only available "
 			"with a single file per job, but %d files "
-			"provided\n", td->o.nr_files);
+			"provided\n", o->nr_files);
 		ret = 1;
 	}
 
-	return ret;
-}
-
-/*
- * This function leaks the buffer
- */
-char *fio_uint_to_kmg(unsigned int val)
-{
-	char *buf = malloc(32);
-	char post[] = { 0, 'K', 'M', 'G', 'P', 'E', 0 };
-	char *p = post;
-
-	do {
-		if (val & 1023)
-			break;
-
-		val >>= 10;
-		p++;
-	} while (*p);
-
-	snprintf(buf, 32, "%u%c", val, *p);
-	return buf;
-}
-
-/* External engines are specified by "external:name.o") */
-static const char *get_engine_name(const char *str)
-{
-	char *p = strstr(str, ":");
-
-	if (!p)
-		return str;
-
-	p++;
-	strip_blank_front(&p);
-	strip_blank_end(p);
-	return p;
-}
-
-static int exists_and_not_file(const char *filename)
-{
-	struct stat sb;
-
-	if (lstat(filename, &sb) == -1)
-		return 0;
-
-#ifndef WIN32 /* NOT Windows */
-	if (S_ISREG(sb.st_mode))
-		return 0;
-#else
-	/* \\.\ is the device namespace in Windows, where every file
-	 * is a device node */
-	if (S_ISREG(sb.st_mode) && strncmp(filename, "\\\\.\\", 4) != 0)
-		return 0;
-#endif
+	if (fio_option_is_set(o, clat_percentiles) &&
+	    !fio_option_is_set(o, lat_percentiles)) {
+		o->lat_percentiles = !o->clat_percentiles;
+	} else if (fio_option_is_set(o, lat_percentiles) &&
+		   !fio_option_is_set(o, clat_percentiles)) {
+		o->clat_percentiles = !o->lat_percentiles;
+	} else if (fio_option_is_set(o, lat_percentiles) &&
+		   fio_option_is_set(o, clat_percentiles) &&
+		   o->lat_percentiles && o->clat_percentiles) {
+		log_err("fio: lat_percentiles and clat_percentiles are "
+			"mutually exclusive\n");
+		ret = 1;
+	}
 
-	return 1;
+	return ret;
 }
 
 static void init_rand_file_service(struct thread_data *td)
@@ -946,9 +948,9 @@ void td_fill_verify_state_seed(struct thread_data *td)
 	bool use64;
 
 	if (td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE64)
-		use64 = 1;
+		use64 = true;
 	else
-		use64 = 0;
+		use64 = false;
 
 	init_rand_seed(&td->verify_state, td->rand_seeds[FIO_RAND_VER_OFF],
 		use64);
@@ -958,7 +960,22 @@ static void td_fill_rand_seeds_internal(struct thread_data *td, bool use64)
 {
 	int i;
 
-	init_rand_seed(&td->bsrange_state, td->rand_seeds[FIO_RAND_BS_OFF], use64);
+	/*
+	 * trimwrite is special in that we need to generate the same
+	 * offsets to get the "write after trim" effect. If we are
+	 * using bssplit to set buffer length distributions, ensure that
+	 * we seed the trim and write generators identically.
+	 */
+	if (td_trimwrite(td)) {
+		init_rand_seed(&td->bsrange_state[DDIR_READ], td->rand_seeds[FIO_RAND_BS_OFF], use64);
+		init_rand_seed(&td->bsrange_state[DDIR_WRITE], td->rand_seeds[FIO_RAND_BS1_OFF], use64);
+		init_rand_seed(&td->bsrange_state[DDIR_TRIM], td->rand_seeds[FIO_RAND_BS1_OFF], use64);
+	} else {
+		init_rand_seed(&td->bsrange_state[DDIR_READ], td->rand_seeds[FIO_RAND_BS_OFF], use64);
+		init_rand_seed(&td->bsrange_state[DDIR_WRITE], td->rand_seeds[FIO_RAND_BS1_OFF], use64);
+		init_rand_seed(&td->bsrange_state[DDIR_TRIM], td->rand_seeds[FIO_RAND_BS2_OFF], use64);
+	}
+
 	td_fill_verify_state_seed(td);
 	init_rand_seed(&td->rwmix_state, td->rand_seeds[FIO_RAND_MIX_OFF], false);
 
@@ -970,7 +987,9 @@ static void td_fill_rand_seeds_internal(struct thread_data *td, bool use64)
 	init_rand_seed(&td->file_size_state, td->rand_seeds[FIO_RAND_FILE_SIZE_OFF], use64);
 	init_rand_seed(&td->trim_state, td->rand_seeds[FIO_RAND_TRIM_OFF], use64);
 	init_rand_seed(&td->delay_state, td->rand_seeds[FIO_RAND_START_DELAY], use64);
-	init_rand_seed(&td->poisson_state, td->rand_seeds[FIO_RAND_POISSON_OFF], 0);
+	init_rand_seed(&td->poisson_state[0], td->rand_seeds[FIO_RAND_POISSON_OFF], 0);
+	init_rand_seed(&td->poisson_state[1], td->rand_seeds[FIO_RAND_POISSON2_OFF], 0);
+	init_rand_seed(&td->poisson_state[2], td->rand_seeds[FIO_RAND_POISSON3_OFF], 0);
 	init_rand_seed(&td->dedupe_state, td->rand_seeds[FIO_DEDUPE_OFF], false);
 	init_rand_seed(&td->zone_state, td->rand_seeds[FIO_RAND_ZONE_OFF], false);
 
@@ -1002,9 +1021,9 @@ void td_fill_rand_seeds(struct thread_data *td)
 	}
 
 	if (td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE64)
-		use64 = 1;
+		use64 = true;
 	else
-		use64 = 0;
+		use64 = false;
 
 	td_fill_rand_seeds_internal(td, use64);
 
@@ -1018,22 +1037,27 @@ void td_fill_rand_seeds(struct thread_data *td)
  */
 int ioengine_load(struct thread_data *td)
 {
-	const char *engine;
-
-	/*
-	 * Engine has already been loaded.
-	 */
-	if (td->io_ops)
-		return 0;
 	if (!td->o.ioengine) {
 		log_err("fio: internal fault, no IO engine specified\n");
 		return 1;
 	}
 
-	engine = get_engine_name(td->o.ioengine);
-	td->io_ops = load_ioengine(td, engine);
+	if (td->io_ops) {
+		/* An engine is loaded, but the requested ioengine
+		 * may have changed.
+		 */
+		if (!strcmp(td->io_ops->name, td->o.ioengine)) {
+			/* The right engine is already loaded */
+			return 0;
+		}
+
+		/* Unload the old engine. */
+		free_ioengine(td);
+	}
+
+	td->io_ops = load_ioengine(td);
 	if (!td->io_ops) {
-		log_err("fio: failed to load engine %s\n", engine);
+		log_err("fio: failed to load engine\n");
 		return 1;
 	}
 
@@ -1061,7 +1085,7 @@ int ioengine_load(struct thread_data *td)
 		 */
 		if (origeo) {
 			memcpy(td->eo, origeo, td->io_ops->option_struct_size);
-			options_mem_dupe(td->eo, td->io_ops->options);
+			options_mem_dupe(td->io_ops->options, td->eo);
 		} else {
 			memset(td->eo, 0, td->io_ops->option_struct_size);
 			fill_default_options(td->eo, td->io_ops->options);
@@ -1069,6 +1093,10 @@ int ioengine_load(struct thread_data *td)
 		*(struct thread_data **)td->eo = td;
 	}
 
+	if (td->o.odirect)
+		td->io_ops->flags |= FIO_RAWIO;
+
+	td_set_ioengine_flags(td);
 	return 0;
 }
 
@@ -1101,6 +1129,9 @@ static void init_flags(struct thread_data *td)
 
 	if (o->verify_async || o->io_submit_mode == IO_MODE_OFFLOAD)
 		td->flags |= TD_F_NEED_LOCK;
+
+	if (o->mem_type == MEM_CUDA_MALLOC)
+		td->flags &= ~TD_F_SCRAMBLE_BUFFERS;
 }
 
 static int setup_random_seeds(struct thread_data *td)
@@ -1108,8 +1139,12 @@ static int setup_random_seeds(struct thread_data *td)
 	unsigned long seed;
 	unsigned int i;
 
-	if (!td->o.rand_repeatable && !fio_option_is_set(&td->o, rand_seed))
-		return init_random_state(td, td->rand_seeds, sizeof(td->rand_seeds));
+	if (!td->o.rand_repeatable && !fio_option_is_set(&td->o, rand_seed)) {
+		int ret = init_random_seeds(td->rand_seeds, sizeof(td->rand_seeds));
+		if (!ret)
+			td_fill_rand_seeds(td);
+		return ret;
+	}
 
 	seed = td->o.rand_seed;
 	for (i = 0; i < 4; i++)
@@ -1151,7 +1186,7 @@ static char *make_filename(char *buf, size_t buf_size,struct thread_options *o,
 
 	if (!o->filename_format || !strlen(o->filename_format)) {
 		sprintf(buf, "%s.%d.%d", jobname, jobnum, filenum);
-		return NULL;
+		return buf;
 	}
 
 	for (f = &fpre_keywords[0]; f->keyword; f++)
@@ -1244,7 +1279,7 @@ static char *make_filename(char *buf, size_t buf_size,struct thread_options *o,
 	return buf;
 }
 
-int parse_dryrun(void)
+bool parse_dryrun(void)
 {
 	return dump_cmdline || parse_only;
 }
@@ -1340,14 +1375,11 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 	if (ioengine_load(td))
 		goto err;
 
-	if (o->odirect)
-		td->io_ops->flags |= FIO_RAWIO;
-
 	file_alloced = 0;
 	if (!o->filename && !td->files_index && !o->read_iolog_file) {
 		file_alloced = 1;
 
-		if (o->nr_files == 1 && exists_and_not_file(jobname))
+		if (o->nr_files == 1 && exists_and_not_regfile(jobname))
 			add_file(td, jobname, job_add_num, 0);
 		else {
 			for (i = 0; i < o->nr_files; i++)
@@ -1373,7 +1405,7 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 	if (td->eo)
 		*(struct thread_data **)td->eo = NULL;
 
-	if (td->io_ops->flags & FIO_DISKLESSIO) {
+	if (td_ioengine_flagged(td, FIO_DISKLESSIO)) {
 		struct fio_file *f;
 
 		for_each_file(td, f, i)
@@ -1383,6 +1415,7 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 	td->mutex = fio_mutex_init(FIO_MUTEX_LOCKED);
 
 	td->ts.clat_percentiles = o->clat_percentiles;
+	td->ts.lat_percentiles = o->lat_percentiles;
 	td->ts.percentile_precision = o->percentile_precision;
 	memcpy(td->ts.percentile_list, o->percentile_list, sizeof(o->percentile_list));
 
@@ -1391,6 +1424,7 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 		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->ts.iops_stat[i].min_val = ULONG_MAX;
 	}
 	td->ddir_seq_nr = o->ddir_seq_nr;
 
@@ -1407,22 +1441,25 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 	prev_group_jobs++;
 
 	if (setup_random_seeds(td)) {
-		td_verror(td, errno, "init_random_state");
+		td_verror(td, errno, "setup_random_seeds");
 		goto err;
 	}
 
 	if (setup_rate(td))
 		goto err;
 
-	if (o->lat_log_file) {
+	if (o->write_lat_log) {
 		struct log_params p = {
 			.td = td,
 			.avg_msec = o->log_avg_msec,
+			.hist_msec = o->log_hist_msec,
+			.hist_coarseness = o->log_hist_coarseness,
 			.log_type = IO_LOG_TYPE_LAT,
 			.log_offset = o->log_offset,
 			.log_gz = o->log_gz,
 			.log_gz_store = o->log_gz_store,
 		};
+		const char *pre = o->lat_log_file ? o->lat_log_file : o->name;
 		const char *suf;
 
 		if (p.log_gz_store)
@@ -1430,65 +1467,109 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 		else
 			suf = "log";
 
-		gen_log_name(logname, sizeof(logname), "lat", o->lat_log_file,
+		gen_log_name(logname, sizeof(logname), "lat", pre,
 				td->thread_number, suf, o->per_job_logs);
 		setup_log(&td->lat_log, &p, logname);
 
-		gen_log_name(logname, sizeof(logname), "slat", o->lat_log_file,
+		gen_log_name(logname, sizeof(logname), "slat", pre,
 				td->thread_number, suf, o->per_job_logs);
 		setup_log(&td->slat_log, &p, logname);
 
-		gen_log_name(logname, sizeof(logname), "clat", o->lat_log_file,
+		gen_log_name(logname, sizeof(logname), "clat", pre,
 				td->thread_number, suf, o->per_job_logs);
 		setup_log(&td->clat_log, &p, logname);
 	}
-	if (o->bw_log_file) {
+
+	if (o->write_hist_log) {
 		struct log_params p = {
 			.td = td,
 			.avg_msec = o->log_avg_msec,
+			.hist_msec = o->log_hist_msec,
+			.hist_coarseness = o->log_hist_coarseness,
+			.log_type = IO_LOG_TYPE_HIST,
+			.log_offset = o->log_offset,
+			.log_gz = o->log_gz,
+			.log_gz_store = o->log_gz_store,
+		};
+		const char *pre = o->hist_log_file ? o->hist_log_file : o->name;
+		const char *suf;
+
+#ifndef CONFIG_ZLIB
+		if (td->client_type) {
+			log_err("fio: --write_hist_log requires zlib in client/server mode\n");
+			goto err;
+		}
+#endif
+
+		if (p.log_gz_store)
+			suf = "log.fz";
+		else
+			suf = "log";
+
+		gen_log_name(logname, sizeof(logname), "clat_hist", pre,
+				td->thread_number, suf, o->per_job_logs);
+		setup_log(&td->clat_hist_log, &p, logname);
+	}
+
+	if (o->write_bw_log) {
+		struct log_params p = {
+			.td = td,
+			.avg_msec = o->log_avg_msec,
+			.hist_msec = o->log_hist_msec,
+			.hist_coarseness = o->log_hist_coarseness,
 			.log_type = IO_LOG_TYPE_BW,
 			.log_offset = o->log_offset,
 			.log_gz = o->log_gz,
 			.log_gz_store = o->log_gz_store,
 		};
+		const char *pre = o->bw_log_file ? o->bw_log_file : o->name;
 		const char *suf;
 
 		if (fio_option_is_set(o, bw_avg_time))
 			p.avg_msec = min(o->log_avg_msec, o->bw_avg_time);
 		else
 			o->bw_avg_time = p.avg_msec;
+	
+		p.hist_msec = o->log_hist_msec;
+		p.hist_coarseness = o->log_hist_coarseness;
 
 		if (p.log_gz_store)
 			suf = "log.fz";
 		else
 			suf = "log";
 
-		gen_log_name(logname, sizeof(logname), "bw", o->bw_log_file,
+		gen_log_name(logname, sizeof(logname), "bw", pre,
 				td->thread_number, suf, o->per_job_logs);
 		setup_log(&td->bw_log, &p, logname);
 	}
-	if (o->iops_log_file) {
+	if (o->write_iops_log) {
 		struct log_params p = {
 			.td = td,
 			.avg_msec = o->log_avg_msec,
+			.hist_msec = o->log_hist_msec,
+			.hist_coarseness = o->log_hist_coarseness,
 			.log_type = IO_LOG_TYPE_IOPS,
 			.log_offset = o->log_offset,
 			.log_gz = o->log_gz,
 			.log_gz_store = o->log_gz_store,
 		};
+		const char *pre = o->iops_log_file ? o->iops_log_file : o->name;
 		const char *suf;
 
 		if (fio_option_is_set(o, iops_avg_time))
 			p.avg_msec = min(o->log_avg_msec, o->iops_avg_time);
 		else
 			o->iops_avg_time = p.avg_msec;
+	
+		p.hist_msec = o->log_hist_msec;
+		p.hist_coarseness = o->log_hist_coarseness;
 
 		if (p.log_gz_store)
 			suf = "log.fz";
 		else
 			suf = "log";
 
-		gen_log_name(logname, sizeof(logname), "iops", o->iops_log_file,
+		gen_log_name(logname, sizeof(logname), "iops", pre,
 				td->thread_number, suf, o->per_job_logs);
 		setup_log(&td->iops_log, &p, logname);
 	}
@@ -1501,18 +1582,19 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 			if (is_backend && !recursed)
 				fio_server_send_add_job(td);
 
-			if (!(td->io_ops->flags & FIO_NOIO)) {
+			if (!td_ioengine_flagged(td, FIO_NOIO)) {
 				char *c1, *c2, *c3, *c4;
 				char *c5 = NULL, *c6 = NULL;
+				int i2p = is_power_of_2(o->kb_base);
 
-				c1 = fio_uint_to_kmg(o->min_bs[DDIR_READ]);
-				c2 = fio_uint_to_kmg(o->max_bs[DDIR_READ]);
-				c3 = fio_uint_to_kmg(o->min_bs[DDIR_WRITE]);
-				c4 = fio_uint_to_kmg(o->max_bs[DDIR_WRITE]);
+				c1 = num2str(o->min_bs[DDIR_READ], 4, 1, i2p, N2S_BYTE);
+				c2 = num2str(o->max_bs[DDIR_READ], 4, 1, i2p, N2S_BYTE);
+				c3 = num2str(o->min_bs[DDIR_WRITE], 4, 1, i2p, N2S_BYTE);
+				c4 = num2str(o->max_bs[DDIR_WRITE], 4, 1, i2p, N2S_BYTE);
 
 				if (!o->bs_is_seq_rand) {
-					c5 = fio_uint_to_kmg(o->min_bs[DDIR_TRIM]);
-					c6 = fio_uint_to_kmg(o->max_bs[DDIR_TRIM]);
+					c5 = num2str(o->min_bs[DDIR_TRIM], 4, 1, i2p, N2S_BYTE);
+					c6 = num2str(o->max_bs[DDIR_TRIM], 4, 1, i2p, N2S_BYTE);
 				}
 
 				log_info("%s: (g=%d): rw=%s, ", td->o.name,
@@ -1520,10 +1602,10 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 							ddir_str(o->td_ddir));
 
 				if (o->bs_is_seq_rand)
-					log_info("bs(seq/rand)=%s-%s/%s-%s, ",
+					log_info("bs=(R) %s-%s, (W) %s-%s, bs_is_seq_rand, ",
 							c1, c2, c3, c4);
 				else
-					log_info("bs=%s-%s/%s-%s/%s-%s, ",
+					log_info("bs=(R) %s-%s, (W) %s-%s, (T) %s-%s, ",
 							c1, c2, c3, c4, c5, c6);
 
 				log_info("ioengine=%s, iodepth=%u\n",
@@ -1540,13 +1622,16 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 			log_info("...\n");
 	}
 
+	if (td_steadystate_init(td))
+		goto err;
+
 	/*
 	 * recurse add identical jobs, clear numjobs and stonewall options
 	 * as they don't apply to sub-jobs
 	 */
 	numjobs = o->numjobs;
 	while (--numjobs) {
-		struct thread_data *td_new = get_new_job(0, td, 1, jobname);
+		struct thread_data *td_new = get_new_job(false, td, true, jobname);
 
 		if (!td_new)
 			goto err;
@@ -1555,6 +1640,8 @@ static int add_job(struct thread_data *td, const char *jobname, int job_add_num,
 		td_new->o.stonewall = 0;
 		td_new->o.new_group = 0;
 		td_new->subjob_number = numjobs;
+		td_new->o.ss_dur = o->ss_dur * 1000000l;
+		td_new->o.ss_limit = o->ss_limit;
 
 		if (file_alloced) {
 			if (td_new->files) {
@@ -1605,11 +1692,11 @@ void add_job_opts(const char **o, int client_type)
 			sprintf(jobname, "%s", o[i] + 5);
 		}
 		if (in_global && !td_parent)
-			td_parent = get_new_job(1, &def_thread, 0, jobname);
+			td_parent = get_new_job(true, &def_thread, false, jobname);
 		else if (!in_global && !td) {
 			if (!td_parent)
 				td_parent = &def_thread;
-			td = get_new_job(0, td_parent, 0, jobname);
+			td = get_new_job(false, td_parent, false, jobname);
 		}
 		if (in_global)
 			fio_options_parse(td_parent, (char **) &o[i], 1);
@@ -1657,11 +1744,11 @@ static int is_empty_or_comment(char *line)
 /*
  * This is our [ini] type file parser.
  */
-int __parse_jobs_ini(struct thread_data *td,
+static int __parse_jobs_ini(struct thread_data *td,
 		char *file, int is_buf, int stonewall_flag, int type,
 		int nested, char *name, char ***popts, int *aopts, int *nopts)
 {
-	unsigned int global = 0;
+	bool global = false;
 	char *string;
 	FILE *f;
 	char *p;
@@ -1770,7 +1857,7 @@ int __parse_jobs_ini(struct thread_data *td,
 				first_sect = 0;
 			}
 
-			td = get_new_job(global, &def_thread, 0, name);
+			td = get_new_job(global, &def_thread, false, name);
 			if (!td) {
 				ret = 1;
 				break;
@@ -1974,6 +2061,11 @@ static void show_debug_categories(void)
 #endif
 }
 
+/*
+ * Following options aren't printed by usage().
+ * --append-terse - Equivalent to --output-format=terse, see f6a7df53.
+ * --latency-log - Deprecated option.
+ */
 static void usage(const char *name)
 {
 	printf("%s\n", fio_version_string);
@@ -1982,15 +2074,15 @@ static void usage(const char *name)
 	show_debug_categories();
 	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("  --bandwidth-log\tGenerate per-job bandwidth logs\n");
+	printf("  --bandwidth-log\tGenerate aggregate bandwidth logs\n");
 	printf("  --minimal\t\tMinimal (terse) output\n");
-	printf("  --output-format=x\tOutput format (terse,json,json+,normal)\n");
-	printf("  --terse-version=x\tSet terse version output format to 'x'\n");
+	printf("  --output-format=type\tOutput format (terse,json,json+,normal)\n");
+	printf("  --terse-version=type\tSet terse version output format"
+		" (default 3, or 2 or 4)\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("  --crctest\t\tTest speed of checksum functions\n");
+	printf("  --crctest=[type]\tTest speed of checksum functions\n");
 	printf("  --cmdhelp=cmd\t\tPrint command help, \"all\" for all of"
 		" them\n");
 	printf("  --enghelp=engine\tPrint ioengine help, or list"
@@ -2006,14 +2098,15 @@ static void usage(const char *name)
 	printf(" 't' period passed\n");
 	printf("  --readonly\t\tTurn on safety read-only checks, preventing"
 		" writes\n");
-	printf("  --section=name\tOnly run specified section in job file\n");
+	printf("  --section=name\tOnly run specified section in job file,"
+		" multiple sections can be specified\n");
 	printf("  --alloc-size=kb\tSet smalloc pool to this size in kb"
-		" (def 1024)\n");
+		" (def 16384)\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("  --client=hostname\tTalk to remote backend(s) fio server at hostname\n");
 	printf("  --remote-config=file\tTell fio server to load this local job file\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"
@@ -2022,7 +2115,7 @@ static void usage(const char *name)
 	printf("  --inflate-log=log\tInflate and output compressed log\n");
 #endif
 	printf("  --trigger-file=file\tExecute trigger cmd when file exists\n");
-	printf("  --trigger-timeout=t\tExecute trigger af this time\n");
+	printf("  --trigger-timeout=t\tExecute trigger at this time\n");
 	printf("  --trigger=cmd\t\tSet this command as local trigger\n");
 	printf("  --trigger-remote=cmd\tSet this command as remote trigger\n");
 	printf("  --aux-path=path\tUse this path for fio state generated files\n");
@@ -2097,6 +2190,14 @@ struct debug_level debug_levels[] = {
 	  .help = "Log compression logging",
 	  .shift = FD_COMPRESS,
 	},
+	{ .name = "steadystate",
+	  .help = "Steady state detection logging",
+	  .shift = FD_STEADYSTATE,
+	},
+	{ .name = "helperthread",
+	  .help = "Helper thread logging",
+	  .shift = FD_HELPERTHREAD,
+	},
 	{ .name = NULL, },
 };
 
@@ -2266,7 +2367,7 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 	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;
+	char *pid_file = NULL;
 	void *cur_client = NULL;
 	int backend = 0;
 
@@ -2285,13 +2386,8 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 		switch (c) {
 		case 'a':
 			smalloc_pool_size = atoi(optarg);
-			break;
-		case 't':
-			if (check_str_time(optarg, &def_timeout, 1)) {
-				log_err("fio: failed parsing time %s\n", optarg);
-				do_exit++;
-				exit_val = 1;
-			}
+			smalloc_pool_size <<= 10;
+			sinit();
 			break;
 		case 'l':
 			log_err("fio: --latency-log is deprecated. Use per-job latency log options.\n");
@@ -2301,27 +2397,26 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 		case 'b':
 			write_bw_log = 1;
 			break;
-		case 'o':
+		case 'o': {
+			FILE *tmp;
+
 			if (f_out && f_out != stdout)
 				fclose(f_out);
 
-			f_out = fopen(optarg, "w+");
-			if (!f_out) {
-				perror("fopen output");
-				exit(1);
+			tmp = fopen(optarg, "w+");
+			if (!tmp) {
+				log_err("fio: output file open error: %s\n", strerror(errno));
+				exit_val = 1;
+				do_exit++;
+				break;
 			}
-			f_err = f_out;
+			f_err = f_out = tmp;
 			break;
+			}
 		case 'm':
 			output_format = FIO_OUTPUT_TERSE;
 			break;
 		case 'F':
-			if (!optarg) {
-				log_err("fio: missing --output-format argument\n");
-				exit_val = 1;
-				do_exit++;
-				break;
-			}
 			if (parse_output_format(optarg)) {
 				log_err("fio: failed parsing output-format\n");
 				exit_val = 1;
@@ -2369,8 +2464,7 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 			break;
 		case 'V':
 			terse_version = atoi(optarg);
-			if (!(terse_version == 2 || terse_version == 3 ||
-			     terse_version == 4)) {
+			if (!(terse_version >= 2 && terse_version <= 5)) {
 				log_err("fio: bad terse version format\n");
 				exit_val = 1;
 				do_exit++;
@@ -2451,7 +2545,7 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 				if (is_section && skip_this_section(val))
 					continue;
 
-				td = get_new_job(global, &def_thread, 1, NULL);
+				td = get_new_job(global, &def_thread, true, NULL);
 				if (!td || ioengine_load(td)) {
 					if (td) {
 						put_job(td);
@@ -2481,7 +2575,6 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 			}
 
 			if (!ret && !strcmp(opt, "ioengine")) {
-				free_ioengine(td);
 				if (ioengine_load(td)) {
 					put_job(td);
 					td = NULL;
@@ -2689,7 +2782,7 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 		if (!ret) {
 			ret = add_job(td, td->o.name ?: "fio", 0, 0, client_type);
 			if (ret)
-				did_arg = 1;
+				exit(1);
 		}
 	}
 
@@ -2701,9 +2794,6 @@ int parse_cmd_line(int argc, char *argv[], int client_type)
 	}
 
 out_free:
-	if (pid_file)
-		free(pid_file);
-
 	return ini_idx;
 }
 
@@ -2772,7 +2862,7 @@ int parse_options(int argc, char *argv[])
 		if (did_arg)
 			return 0;
 
-		log_err("No jobs(s) defined\n\n");
+		log_err("No job(s) defined\n\n");
 
 		if (!did_arg) {
 			usage(argv[0]);