X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=stat.c;h=8835f7f960ad93b223ffd0ccfe29d16ae0661f04;hp=8b032b881d93d8fe1dc692a1c38898be42848bcd;hb=9db01ef976006c002b05fa4e4ec589eb029aac5b;hpb=2e33101f218a1603eeecca969f5b7a0e98696a01

diff --git a/stat.c b/stat.c
index 8b032b88..8835f7f9 100644
--- a/stat.c
+++ b/stat.c
@@ -10,13 +10,15 @@
 #include "fio.h"
 #include "diskutil.h"
 #include "lib/ieee754.h"
+#include "json.h"
+#include "lib/getrusage.h"
+#include "idletime.h"
 
 void update_rusage_stat(struct thread_data *td)
 {
 	struct thread_stat *ts = &td->ts;
 
-	getrusage(RUSAGE_SELF, &td->ru_end);
-
+	fio_getrusage(&td->ru_end);
 	ts->usr_time += mtime_since(&td->ru_start.ru_utime,
 					&td->ru_end.ru_utime);
 	ts->sys_time += mtime_since(&td->ru_start.ru_stime,
@@ -63,12 +65,12 @@ static unsigned int plat_val_to_idx(unsigned int val)
 
 	/*
 	 * Discard the error bits and apply the mask to find the
-         * index for the buckets in the group
+	 * index for the buckets in the group
 	 */
 	offset = (FIO_IO_U_PLAT_VAL - 1) & (val >> error_bits);
 
 	/* Make sure the index does not exceed (array size - 1) */
-	idx = (base + offset) < (FIO_IO_U_PLAT_NR - 1)?
+	idx = (base + offset) < (FIO_IO_U_PLAT_NR - 1) ?
 		(base + offset) : (FIO_IO_U_PLAT_NR - 1);
 
 	return idx;
@@ -86,11 +88,11 @@ static unsigned int plat_idx_to_val(unsigned int idx)
 
 	/* MSB <= (FIO_IO_U_PLAT_BITS-1), cannot be rounded off. Use
 	 * all bits of the sample as index */
-	if (idx < (FIO_IO_U_PLAT_VAL << 1) )
+	if (idx < (FIO_IO_U_PLAT_VAL << 1))
 		return idx;
 
 	/* Find the group and compute the minimum value of that group */
-	error_bits = (idx >> FIO_IO_U_PLAT_BITS) -1;
+	error_bits = (idx >> FIO_IO_U_PLAT_BITS) - 1;
 	base = 1 << (error_bits + FIO_IO_U_PLAT_BITS);
 
 	/* Find its bucket number of the group */
@@ -140,7 +142,7 @@ unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long nr,
 	 * isn't a worry. Also note that this does not work for NaN values.
 	 */
 	if (len > 1)
-		qsort((void*)plist, len, sizeof(plist[0]), double_cmp);
+		qsort((void *)plist, len, sizeof(plist[0]), double_cmp);
 
 	/*
 	 * Calculate bucket values, note down max and min values
@@ -178,11 +180,12 @@ unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long nr,
  * Find and display the p-th percentile of clat
  */
 static void show_clat_percentiles(unsigned int *io_u_plat, unsigned long nr,
-				  fio_fp64_t *plist)
+				  fio_fp64_t *plist, unsigned int precision)
 {
 	unsigned int len, j = 0, minv, maxv;
 	unsigned int *ovals;
-	int is_last, scale_down;
+	int is_last, per_line, scale_down;
+	char fmt[32];
 
 	len = calc_clat_percentiles(io_u_plat, nr, plist, &ovals, &maxv, &minv);
 	if (!len)
@@ -200,20 +203,23 @@ static void show_clat_percentiles(unsigned int *io_u_plat, unsigned long nr,
 		log_info("    clat percentiles (usec):\n     |");
 	}
 
+	snprintf(fmt, sizeof(fmt), "%%1.%uf", precision);
+	per_line = (80 - 7) / (precision + 14);
+
 	for (j = 0; j < len; j++) {
-		char fbuf[8];
+		char fbuf[16], *ptr = fbuf;
 
 		/* for formatting */
-		if (j != 0 && (j % 4) == 0)
+		if (j != 0 && (j % per_line) == 0)
 			log_info("     |");
 
 		/* end of the list */
 		is_last = (j == len - 1);
 
 		if (plist[j].u.f < 10.0)
-			sprintf(fbuf, " %2.2f", plist[j].u.f);
-		else
-			sprintf(fbuf, "%2.2f", plist[j].u.f);
+			ptr += sprintf(fbuf, " ");
+
+		snprintf(ptr, sizeof(fbuf), fmt, plist[j].u.f);
 
 		if (scale_down)
 			ovals[j] = (ovals[j] + 999) / 1000;
@@ -223,7 +229,7 @@ static void show_clat_percentiles(unsigned int *io_u_plat, unsigned long nr,
 		if (is_last)
 			break;
 
-		if (j % 4 == 3)	/* for formatting */
+		if ((j % per_line) == per_line - 1)	/* for formatting */
 			log_info("\n");
 	}
 
@@ -257,12 +263,12 @@ int calc_lat(struct io_stat *is, unsigned long *min, unsigned long *max,
 void show_group_stats(struct group_run_stats *rs)
 {
 	char *p1, *p2, *p3, *p4;
-	const char *ddir_str[] = { "   READ", "  WRITE" };
+	const char *ddir_str[] = { "   READ", "  WRITE" , "   TRIM"};
 	int i;
 
 	log_info("\nRun status group %d (all jobs):\n", rs->groupid);
 
-	for (i = 0; i <= DDIR_WRITE; i++) {
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
 		const int i2p = is_power_of_2(rs->kb_base);
 
 		if (!rs->max_run[i])
@@ -274,9 +280,9 @@ void show_group_stats(struct group_run_stats *rs)
 		p4 = num2str(rs->max_bw[i], 6, rs->kb_base, i2p);
 
 		log_info("%s: io=%sB, aggrb=%sB/s, minb=%sB/s, maxb=%sB/s,"
-			 " mint=%llumsec, maxt=%llumsec\n", ddir_str[i], p1, p2,
-						p3, p4, rs->min_run[i],
-						rs->max_run[i]);
+			 " mint=%llumsec, maxt=%llumsec\n",
+				rs->unified_rw_rep ? "  MIXED" : ddir_str[i],
+				p1, p2, p3, p4, rs->min_run[i], rs->max_run[i]);
 
 		free(p1);
 		free(p2);
@@ -306,7 +312,7 @@ void stat_calc_dist(unsigned int *map, unsigned long total, double *io_u_dist)
 static void stat_calc_lat(struct thread_stat *ts, double *dst,
 			  unsigned int *src, int nr)
 {
-	unsigned long total = ts_total_io_u(ts);
+	unsigned long total = ddir_rw_sum(ts->total_io_u);
 	int i;
 
 	/*
@@ -355,7 +361,7 @@ static void display_lat(const char *name, unsigned long min, unsigned long max,
 static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 			     int ddir)
 {
-	const char *ddir_str[] = { "read ", "write" };
+	const char *ddir_str[] = { "read ", "write", "trim" };
 	unsigned long min, max, runt;
 	unsigned long long bw, iops;
 	double mean, dev;
@@ -378,8 +384,8 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 	iops_p = num2str(iops, 6, 1, 0);
 
 	log_info("  %s: io=%sB, bw=%sB/s, iops=%s, runt=%6llumsec\n",
-					ddir_str[ddir], io_p, bw_p, iops_p,
-					ts->runtime[ddir]);
+				rs->unified_rw_rep ? "mixed" : ddir_str[ddir],
+				io_p, bw_p, iops_p, ts->runtime[ddir]);
 
 	free(io_p);
 	free(bw_p);
@@ -395,7 +401,8 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 	if (ts->clat_percentiles) {
 		show_clat_percentiles(ts->io_u_plat[ddir],
 					ts->clat_stat[ddir].samples,
-					ts->percentile_list);
+					ts->percentile_list,
+					ts->percentile_precision);
 	}
 	if (calc_lat(&ts->bw_stat[ddir], &min, &max, &mean, &dev)) {
 		double p_of_agg = 100.0;
@@ -485,19 +492,26 @@ void show_thread_status(struct thread_stat *ts, struct group_run_stats *rs)
 	double usr_cpu, sys_cpu;
 	unsigned long runtime;
 	double io_u_dist[FIO_IO_U_MAP_NR];
+	time_t time_p;
+	char time_buf[64];
 
-	if (!(ts->io_bytes[0] + ts->io_bytes[1]) &&
-	    !(ts->total_io_u[0] + ts->total_io_u[1]))
+	if (!(ts->io_bytes[DDIR_READ] + ts->io_bytes[DDIR_WRITE] +
+	    ts->io_bytes[DDIR_TRIM]) && !(ts->total_io_u[DDIR_READ] +
+	    ts->total_io_u[DDIR_WRITE] + ts->total_io_u[DDIR_TRIM]))
 		return;
 
+	time(&time_p);
+	os_ctime_r((const time_t *) &time_p, time_buf, sizeof(time_buf));
+
 	if (!ts->error) {
-		log_info("%s: (groupid=%d, jobs=%d): err=%2d: pid=%d\n",
+		log_info("%s: (groupid=%d, jobs=%d): err=%2d: pid=%d: %s",
 					ts->name, ts->groupid, ts->members,
-					ts->error, (int) ts->pid);
+					ts->error, (int) ts->pid, time_buf);
 	} else {
-		log_info("%s: (groupid=%d, jobs=%d): err=%2d (%s): pid=%d\n",
+		log_info("%s: (groupid=%d, jobs=%d): err=%2d (%s): pid=%d: %s",
 					ts->name, ts->groupid, ts->members,
-					ts->error, ts->verror, (int) ts->pid);
+					ts->error, ts->verror, (int) ts->pid,
+					time_buf);
 	}
 
 	if (strlen(ts->description))
@@ -507,6 +521,8 @@ void show_thread_status(struct thread_stat *ts, struct group_run_stats *rs)
 		show_ddir_status(rs, ts, DDIR_READ);
 	if (ts->io_bytes[DDIR_WRITE])
 		show_ddir_status(rs, ts, DDIR_WRITE);
+	if (ts->io_bytes[DDIR_TRIM])
+		show_ddir_status(rs, ts, DDIR_TRIM);
 
 	show_latencies(ts);
 
@@ -524,7 +540,7 @@ void show_thread_status(struct thread_stat *ts, struct group_run_stats *rs)
 	log_info("  cpu          : usr=%3.2f%%, sys=%3.2f%%, ctx=%lu, majf=%lu,"
 		 " minf=%lu\n", usr_cpu, sys_cpu, ts->ctx, ts->majf, ts->minf);
 
-	stat_calc_dist(ts->io_u_map, ts_total_io_u(ts), io_u_dist);
+	stat_calc_dist(ts->io_u_map, ddir_rw_sum(ts->total_io_u), io_u_dist);
 	log_info("  IO depths    : 1=%3.1f%%, 2=%3.1f%%, 4=%3.1f%%, 8=%3.1f%%,"
 		 " 16=%3.1f%%, 32=%3.1f%%, >=64=%3.1f%%\n", io_u_dist[0],
 					io_u_dist[1], io_u_dist[2],
@@ -573,7 +589,7 @@ static void show_ddir_status_terse(struct thread_stat *ts,
 	if (ts->runtime[ddir]) {
 		uint64_t runt = ts->runtime[ddir];
 
-		bw = ts->io_bytes[ddir] / runt;
+		bw = ((1000 * ts->io_bytes[ddir]) / runt) / 1024;
 		iops = (1000 * (uint64_t) ts->total_io_u[ddir]) / runt;
 	}
 
@@ -603,7 +619,7 @@ static void show_ddir_status_terse(struct thread_stat *ts,
 			log_info(";0%%=0");
 			continue;
 		}
-		log_info(";%2.2f%%=%u", ts->percentile_list[i].u.f, ovals[i]);
+		log_info(";%f%%=%u", ts->percentile_list[i].u.f, ovals[i]);
 	}
 
 	if (calc_lat(&ts->lat_stat[ddir], &min, &max, &mean, &dev))
@@ -628,8 +644,115 @@ static void show_ddir_status_terse(struct thread_stat *ts,
 		log_info(";%lu;%lu;%f%%;%f;%f", 0UL, 0UL, 0.0, 0.0, 0.0);
 }
 
+static void add_ddir_status_json(struct thread_stat *ts,
+		struct group_run_stats *rs, int ddir, struct json_object *parent)
+{
+	unsigned long min, max;
+	unsigned long long bw, iops;
+	unsigned int *ovals = NULL;
+	double mean, dev;
+	unsigned int len, minv, maxv;
+	int i;
+	const char *ddirname[] = {"read", "write", "trim"};
+	struct json_object *dir_object, *tmp_object, *percentile_object;
+	char buf[120];
+	double p_of_agg = 100.0;
+
+	assert(ddir_rw(ddir));
+
+	if (ts->unified_rw_rep && ddir != DDIR_READ)
+		return;
+
+	dir_object = json_create_object();
+	json_object_add_value_object(parent,
+		ts->unified_rw_rep ? "mixed" : ddirname[ddir], dir_object);
+
+	iops = bw = 0;
+	if (ts->runtime[ddir]) {
+		uint64_t runt = ts->runtime[ddir];
+
+		bw = ((1000 * ts->io_bytes[ddir]) / runt) / 1024;
+		iops = (1000 * (uint64_t) ts->total_io_u[ddir]) / runt;
+	}
+
+	json_object_add_value_int(dir_object, "io_bytes", ts->io_bytes[ddir] >> 10);
+	json_object_add_value_int(dir_object, "bw", bw);
+	json_object_add_value_int(dir_object, "iops", iops);
+	json_object_add_value_int(dir_object, "runtime", ts->runtime[ddir]);
+
+	if (!calc_lat(&ts->slat_stat[ddir], &min, &max, &mean, &dev)) {
+		min = max = 0;
+		mean = dev = 0.0;
+	}
+	tmp_object = json_create_object();
+	json_object_add_value_object(dir_object, "slat", tmp_object);
+	json_object_add_value_int(tmp_object, "min", min);
+	json_object_add_value_int(tmp_object, "max", max);
+	json_object_add_value_float(tmp_object, "mean", mean);
+	json_object_add_value_float(tmp_object, "stddev", dev);
+
+	if (!calc_lat(&ts->clat_stat[ddir], &min, &max, &mean, &dev)) {
+		min = max = 0;
+		mean = dev = 0.0;
+	}
+	tmp_object = json_create_object();
+	json_object_add_value_object(dir_object, "clat", tmp_object);
+	json_object_add_value_int(tmp_object, "min", min);
+	json_object_add_value_int(tmp_object, "max", max);
+	json_object_add_value_float(tmp_object, "mean", mean);
+	json_object_add_value_float(tmp_object, "stddev", dev);
+
+	if (ts->clat_percentiles) {
+		len = calc_clat_percentiles(ts->io_u_plat[ddir],
+					ts->clat_stat[ddir].samples,
+					ts->percentile_list, &ovals, &maxv,
+					&minv);
+	} else
+		len = 0;
+
+	percentile_object = json_create_object();
+	json_object_add_value_object(tmp_object, "percentile", percentile_object);
+	for (i = 0; i < FIO_IO_U_LIST_MAX_LEN; i++) {
+		if (i >= len) {
+			json_object_add_value_int(percentile_object, "0.00", 0);
+			continue;
+		}
+		snprintf(buf, sizeof(buf), "%f", ts->percentile_list[i].u.f);
+		json_object_add_value_int(percentile_object, (const char *)buf, ovals[i]);
+	}
+
+	if (!calc_lat(&ts->lat_stat[ddir], &min, &max, &mean, &dev)) {
+		min = max = 0;
+		mean = dev = 0.0;
+	}
+	tmp_object = json_create_object();
+	json_object_add_value_object(dir_object, "lat", tmp_object);
+	json_object_add_value_int(tmp_object, "min", min);
+	json_object_add_value_int(tmp_object, "max", max);
+	json_object_add_value_float(tmp_object, "mean", mean);
+	json_object_add_value_float(tmp_object, "stddev", dev);
+	if (ovals)
+		free(ovals);
+
+	if (!calc_lat(&ts->bw_stat[ddir], &min, &max, &mean, &dev)) {
+		if (rs->agg[ddir]) {
+			p_of_agg = mean * 100 / (double) rs->agg[ddir];
+			if (p_of_agg > 100.0)
+				p_of_agg = 100.0;
+		}
+	} else {
+		min = max = 0;
+		p_of_agg = mean = dev = 0.0;
+	}
+	json_object_add_value_int(dir_object, "bw_min", min);
+	json_object_add_value_int(dir_object, "bw_max", max);
+	json_object_add_value_float(dir_object, "bw_agg", mean);
+	json_object_add_value_float(dir_object, "bw_mean", mean);
+	json_object_add_value_float(dir_object, "bw_dev", dev);
+}
+
 static void show_thread_status_terse_v2(struct thread_stat *ts,
-				        struct group_run_stats *rs)
+					struct group_run_stats *rs)
 {
 	double io_u_dist[FIO_IO_U_MAP_NR];
 	double io_u_lat_u[FIO_IO_U_LAT_U_NR];
@@ -640,9 +763,11 @@ static void show_thread_status_terse_v2(struct thread_stat *ts,
 	/* General Info */
 	log_info("2;%s;%d;%d", ts->name, ts->groupid, ts->error);
 	/* Log Read Status */
-	show_ddir_status_terse(ts, rs, 0);
+	show_ddir_status_terse(ts, rs, DDIR_READ);
 	/* Log Write Status */
-	show_ddir_status_terse(ts, rs, 1);
+	show_ddir_status_terse(ts, rs, DDIR_WRITE);
+	/* Log Trim Status */
+	show_ddir_status_terse(ts, rs, DDIR_TRIM);
 
 	/* CPU Usage */
 	if (ts->total_run_time) {
@@ -659,7 +784,7 @@ static void show_thread_status_terse_v2(struct thread_stat *ts,
 								ts->minf);
 
 	/* Calc % distribution of IO depths, usecond, msecond latency */
-	stat_calc_dist(ts->io_u_map, ts_total_io_u(ts), io_u_dist);
+	stat_calc_dist(ts->io_u_map, ddir_rw_sum(ts->total_io_u), io_u_dist);
 	stat_calc_lat_u(ts, io_u_lat_u);
 	stat_calc_lat_m(ts, io_u_lat_m);
 
@@ -686,10 +811,8 @@ static void show_thread_status_terse_v2(struct thread_stat *ts,
 	log_info("\n");
 }
 
-#define FIO_TERSE_VERSION	"3"
-
-static void show_thread_status_terse_v3(struct thread_stat *ts,
-					struct group_run_stats *rs)
+static void show_thread_status_terse_v3_v4(struct thread_stat *ts,
+					   struct group_run_stats *rs, int ver)
 {
 	double io_u_dist[FIO_IO_U_MAP_NR];
 	double io_u_lat_u[FIO_IO_U_LAT_U_NR];
@@ -698,12 +821,15 @@ static void show_thread_status_terse_v3(struct thread_stat *ts,
 	int i;
 
 	/* General Info */
-	log_info("%s;%s;%s;%d;%d", FIO_TERSE_VERSION, fio_version_string,
+	log_info("%d;%s;%s;%d;%d", ver, fio_version_string,
 					ts->name, ts->groupid, ts->error);
 	/* Log Read Status */
-	show_ddir_status_terse(ts, rs, 0);
+	show_ddir_status_terse(ts, rs, DDIR_READ);
 	/* Log Write Status */
-	show_ddir_status_terse(ts, rs, 1);
+	show_ddir_status_terse(ts, rs, DDIR_WRITE);
+	/* Log Trim Status */
+	if (ver == 4)
+		show_ddir_status_terse(ts, rs, DDIR_TRIM);
 
 	/* CPU Usage */
 	if (ts->total_run_time) {
@@ -720,7 +846,7 @@ static void show_thread_status_terse_v3(struct thread_stat *ts,
 								ts->minf);
 
 	/* Calc % distribution of IO depths, usecond, msecond latency */
-	stat_calc_dist(ts->io_u_map, ts_total_io_u(ts), io_u_dist);
+	stat_calc_dist(ts->io_u_map, ddir_rw_sum(ts->total_io_u), io_u_dist);
 	stat_calc_lat_u(ts, io_u_lat_u);
 	stat_calc_lat_m(ts, io_u_lat_m);
 
@@ -737,16 +863,101 @@ static void show_thread_status_terse_v3(struct thread_stat *ts,
 		log_info(";%3.2f%%", io_u_lat_m[i]);
 
 	/* disk util stats, if any */
-	show_disk_util(1);
+	show_disk_util(1, NULL);
 
 	/* Additional output if continue_on_error set - default off*/
 	if (ts->continue_on_error)
 		log_info(";%lu;%d", ts->total_err_count, ts->first_error);
-	log_info("\n");
 
 	/* Additional output if description is set */
 	if (strlen(ts->description))
 		log_info(";%s", ts->description);
+
+	log_info("\n");
+}
+
+static struct json_object *show_thread_status_json(struct thread_stat *ts,
+				    struct group_run_stats *rs)
+{
+	struct json_object *root, *tmp;
+	double io_u_dist[FIO_IO_U_MAP_NR];
+	double io_u_lat_u[FIO_IO_U_LAT_U_NR];
+	double io_u_lat_m[FIO_IO_U_LAT_M_NR];
+	double usr_cpu, sys_cpu;
+	int i;
+
+	root = json_create_object();
+	json_object_add_value_string(root, "jobname", ts->name);
+	json_object_add_value_int(root, "groupid", ts->groupid);
+	json_object_add_value_int(root, "error", ts->error);
+
+	add_ddir_status_json(ts, rs, DDIR_READ, root);
+	add_ddir_status_json(ts, rs, DDIR_WRITE, root);
+	add_ddir_status_json(ts, rs, DDIR_TRIM, root);
+
+	/* CPU Usage */
+	if (ts->total_run_time) {
+		double runt = (double) ts->total_run_time;
+
+		usr_cpu = (double) ts->usr_time * 100 / runt;
+		sys_cpu = (double) ts->sys_time * 100 / runt;
+	} else {
+		usr_cpu = 0;
+		sys_cpu = 0;
+	}
+	json_object_add_value_float(root, "usr_cpu", usr_cpu);
+	json_object_add_value_float(root, "sys_cpu", sys_cpu);
+	json_object_add_value_int(root, "ctx", ts->ctx);
+	json_object_add_value_int(root, "majf", ts->majf);
+	json_object_add_value_int(root, "minf", ts->minf);
+
+
+	/* Calc % distribution of IO depths, usecond, msecond latency */
+	stat_calc_dist(ts->io_u_map, ddir_rw_sum(ts->total_io_u), io_u_dist);
+	stat_calc_lat_u(ts, io_u_lat_u);
+	stat_calc_lat_m(ts, io_u_lat_m);
+
+	tmp = json_create_object();
+	json_object_add_value_object(root, "iodepth_level", tmp);
+	/* Only show fixed 7 I/O depth levels*/
+	for (i = 0; i < 7; i++) {
+		char name[20];
+		if (i < 6)
+			snprintf(name, 20, "%d", 1 << i);
+		else
+			snprintf(name, 20, ">=%d", 1 << i);
+		json_object_add_value_float(tmp, (const char *)name, io_u_dist[i]);
+	}
+
+	tmp = json_create_object();
+	json_object_add_value_object(root, "latency_us", tmp);
+	/* Microsecond latency */
+	for (i = 0; i < FIO_IO_U_LAT_U_NR; i++) {
+		const char *ranges[] = { "2", "4", "10", "20", "50", "100",
+				 "250", "500", "750", "1000", };
+		json_object_add_value_float(tmp, ranges[i], io_u_lat_u[i]);
+	}
+	/* Millisecond latency */
+	tmp = json_create_object();
+	json_object_add_value_object(root, "latency_ms", tmp);
+	for (i = 0; i < FIO_IO_U_LAT_M_NR; i++) {
+		const char *ranges[] = { "2", "4", "10", "20", "50", "100",
+				 "250", "500", "750", "1000", "2000",
+				 ">=2000", };
+		json_object_add_value_float(tmp, ranges[i], io_u_lat_m[i]);
+	}
+
+	/* Additional output if continue_on_error set - default off*/
+	if (ts->continue_on_error) {
+		json_object_add_value_int(root, "total_err", ts->total_err_count);
+		json_object_add_value_int(root, "total_err", ts->first_error);
+	}
+
+	/* Additional output if description is set */
+	if (strlen(ts->description))
+		json_object_add_value_string(root, "desc", ts->description);
+
+	return root;
 }
 
 static void show_thread_status_terse(struct thread_stat *ts,
@@ -754,8 +965,8 @@ static void show_thread_status_terse(struct thread_stat *ts,
 {
 	if (terse_version == 2)
 		show_thread_status_terse_v2(ts, rs);
-	else if (terse_version == 3)
-		show_thread_status_terse_v3(ts, rs);
+	else if (terse_version == 3 || terse_version == 4)
+		show_thread_status_terse_v3_v4(ts, rs, terse_version);
 	else
 		log_err("fio: bad terse version!? %d\n", terse_version);
 }
@@ -799,7 +1010,7 @@ void sum_group_stats(struct group_run_stats *dst, struct group_run_stats *src)
 {
 	int i;
 
-	for (i = 0; i < 2; i++) {
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
 		if (dst->max_run[i] < src->max_run[i])
 			dst->max_run[i] = src->max_run[i];
 		if (dst->min_run[i] && dst->min_run[i] > src->min_run[i])
@@ -819,16 +1030,28 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, int nr)
 {
 	int l, k;
 
-	for (l = 0; l <= DDIR_WRITE; l++) {
-		sum_stat(&dst->clat_stat[l], &src->clat_stat[l], nr);
-		sum_stat(&dst->slat_stat[l], &src->slat_stat[l], nr);
-		sum_stat(&dst->lat_stat[l], &src->lat_stat[l], nr);
-		sum_stat(&dst->bw_stat[l], &src->bw_stat[l], nr);
+	for (l = 0; l < DDIR_RWDIR_CNT; l++) {
+		if (!dst->unified_rw_rep) {
+			sum_stat(&dst->clat_stat[l], &src->clat_stat[l], nr);
+			sum_stat(&dst->slat_stat[l], &src->slat_stat[l], nr);
+			sum_stat(&dst->lat_stat[l], &src->lat_stat[l], nr);
+			sum_stat(&dst->bw_stat[l], &src->bw_stat[l], nr);
+
+			dst->io_bytes[l] += src->io_bytes[l];
 
-		dst->io_bytes[l] += src->io_bytes[l];
+			if (dst->runtime[l] < src->runtime[l])
+				dst->runtime[l] = src->runtime[l];
+		} else {
+			sum_stat(&dst->clat_stat[0], &src->clat_stat[l], nr);
+			sum_stat(&dst->slat_stat[0], &src->slat_stat[l], nr);
+			sum_stat(&dst->lat_stat[0], &src->lat_stat[l], nr);
+			sum_stat(&dst->bw_stat[0], &src->bw_stat[l], nr);
 
-		if (dst->runtime[l] < src->runtime[l])
-			dst->runtime[l] = src->runtime[l];
+			dst->io_bytes[0] += src->io_bytes[l];
+
+			if (dst->runtime[0] < src->runtime[l])
+				dst->runtime[0] = src->runtime[l];
+		}
 	}
 
 	dst->usr_time += src->usr_time;
@@ -848,15 +1071,25 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, int nr)
 	for (k = 0; k < FIO_IO_U_LAT_M_NR; k++)
 		dst->io_u_lat_m[k] += src->io_u_lat_m[k];
 
-	for (k = 0; k <= 2; k++) {
-		dst->total_io_u[k] += src->total_io_u[k];
-		dst->short_io_u[k] += src->short_io_u[k];
+	for (k = 0; k < DDIR_RWDIR_CNT; k++) {
+		if (!dst->unified_rw_rep) {
+			dst->total_io_u[k] += src->total_io_u[k];
+			dst->short_io_u[k] += src->short_io_u[k];
+		} else {
+			dst->total_io_u[0] += src->total_io_u[k];
+			dst->short_io_u[0] += src->short_io_u[k];
+		}
 	}
 
-	for (k = 0; k <= DDIR_WRITE; k++) {
+	for (k = 0; k < DDIR_RWDIR_CNT; k++) {
 		int m;
-		for (m = 0; m < FIO_IO_U_PLAT_NR; m++)
-			dst->io_u_plat[k][m] += src->io_u_plat[k][m];
+
+		for (m = 0; m < FIO_IO_U_PLAT_NR; m++) {
+			if (!dst->unified_rw_rep)
+				dst->io_u_plat[k][m] += src->io_u_plat[k][m];
+			else
+				dst->io_u_plat[0][m] += src->io_u_plat[k][m];
+		}
 	}
 
 	dst->total_run_time += src->total_run_time;
@@ -866,9 +1099,11 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, int nr)
 
 void init_group_run_stat(struct group_run_stats *gs)
 {
+	int i;
 	memset(gs, 0, sizeof(*gs));
-	gs->min_bw[0] = gs->min_run[0] = ~0UL;
-	gs->min_bw[1] = gs->min_run[1] = ~0UL;
+
+	for (i = 0; i < DDIR_RWDIR_CNT; i++)
+		gs->min_bw[i] = gs->min_run[i] = ~0UL;
 }
 
 void init_thread_stat(struct thread_stat *ts)
@@ -877,7 +1112,7 @@ void init_thread_stat(struct thread_stat *ts)
 
 	memset(ts, 0, sizeof(*ts));
 
-	for (j = 0; j <= DDIR_WRITE; j++) {
+	for (j = 0; j < DDIR_RWDIR_CNT; j++) {
 		ts->lat_stat[j].min_val = -1UL;
 		ts->clat_stat[j].min_val = -1UL;
 		ts->slat_stat[j].min_val = -1UL;
@@ -893,6 +1128,8 @@ void show_run_stats(void)
 	struct thread_stat *threadstats, *ts;
 	int i, j, nr_ts, last_ts, idx;
 	int kb_base_warned = 0;
+	struct json_object *root = NULL;
+	struct json_array *array = NULL;
 
 	runstats = malloc(sizeof(struct group_run_stats) * (groupid + 1));
 
@@ -937,10 +1174,8 @@ void show_run_stats(void)
 		ts = &threadstats[j];
 
 		ts->clat_percentiles = td->o.clat_percentiles;
-		if (td->o.overwrite_plist)
-			memcpy(ts->percentile_list, td->o.percentile_list, sizeof(td->o.percentile_list));
-		else
-			memcpy(ts->percentile_list, def_percentile_list, sizeof(def_percentile_list));
+		ts->percentile_precision = td->o.percentile_precision;
+		memcpy(ts->percentile_list, td->o.percentile_list, sizeof(td->o.percentile_list));
 
 		idx++;
 		ts->members++;
@@ -956,6 +1191,11 @@ void show_run_stats(void)
 			else
 				memset(ts->description, 0, FIO_JOBNAME_SIZE);
 
+			/*
+			 * If multiple entries in this group, this is
+			 * the first member.
+			 */
+			ts->thread_number = td->thread_number;
 			ts->groupid = td->groupid;
 
 			/*
@@ -964,6 +1204,7 @@ void show_run_stats(void)
 			ts->pid = td->pid;
 
 			ts->kb_base = td->o.kb_base;
+			ts->unified_rw_rep = td->o.unified_rw_rep;
 		} else if (ts->kb_base != td->o.kb_base && !kb_base_warned) {
 			log_info("fio: kb_base differs for jobs in group, using"
 				 " %u as the base\n", ts->kb_base);
@@ -993,8 +1234,9 @@ void show_run_stats(void)
 		ts = &threadstats[i];
 		rs = &runstats[ts->groupid];
 		rs->kb_base = ts->kb_base;
+		rs->unified_rw_rep += ts->unified_rw_rep;
 
-		for (j = 0; j <= DDIR_WRITE; j++) {
+		for (j = 0; j < DDIR_RWDIR_CNT; j++) {
 			if (!ts->runtime[j])
 				continue;
 			if (ts->runtime[j] < rs->min_run[j] || !rs->min_run[j])
@@ -1004,10 +1246,11 @@ void show_run_stats(void)
 
 			bw = 0;
 			if (ts->runtime[j]) {
-				unsigned long runt;
+				unsigned long runt = ts->runtime[j];
+				unsigned long long kb;
 
-				runt = ts->runtime[j];
-				bw = ts->io_bytes[j] / runt;
+				kb = ts->io_bytes[j] / rs->kb_base;
+				bw = kb * 1000 / runt;
 			}
 			if (bw < rs->min_bw[j])
 				rs->min_bw[j] = bw;
@@ -1019,23 +1262,28 @@ void show_run_stats(void)
 	}
 
 	for (i = 0; i < groupid + 1; i++) {
-		unsigned long max_run[2];
+		int ddir;
 
 		rs = &runstats[i];
-		max_run[0] = rs->max_run[0];
-		max_run[1] = rs->max_run[1];
 
-		if (rs->max_run[0])
-			rs->agg[0] = (rs->io_kb[0] * 1000) / max_run[0];
-		if (rs->max_run[1])
-			rs->agg[1] = (rs->io_kb[1] * 1000) / max_run[1];
+		for (ddir = 0; ddir < DDIR_RWDIR_CNT; ddir++) {
+			if (rs->max_run[ddir])
+				rs->agg[ddir] = (rs->io_kb[ddir] * 1000) /
+						rs->max_run[ddir];
+		}
 	}
 
 	/*
 	 * don't overwrite last signal output
 	 */
-	if (!terse_output)
+	if (output_format == FIO_OUTPUT_NORMAL)
 		log_info("\n");
+	else if (output_format == FIO_OUTPUT_JSON) {
+		root = json_create_object();
+		json_object_add_value_string(root, "fio version", fio_version_string);
+		array = json_create_array();
+		json_object_add_value_array(root, "jobs", array);
+	}
 
 	for (i = 0; i < nr_ts; i++) {
 		ts = &threadstats[i];
@@ -1043,11 +1291,24 @@ void show_run_stats(void)
 
 		if (is_backend)
 			fio_server_send_ts(ts, rs);
-		else if (terse_output)
+		else if (output_format == FIO_OUTPUT_TERSE)
 			show_thread_status_terse(ts, rs);
-		else
+		else if (output_format == FIO_OUTPUT_JSON) {
+			struct json_object *tmp = show_thread_status_json(ts, rs);
+			json_array_add_value_object(array, tmp);
+		} else
 			show_thread_status(ts, rs);
 	}
+	if (output_format == FIO_OUTPUT_JSON) {
+		/* disk util stats, if any */
+		show_disk_util(1, root);
+
+		show_idle_prof_stats(FIO_OUTPUT_JSON, root);
+
+		json_print_object(root);
+		log_info("\n");
+		json_free_object(root);
+	}
 
 	for (i = 0; i < groupid + 1; i++) {
 		rs = &runstats[i];
@@ -1055,21 +1316,75 @@ void show_run_stats(void)
 		rs->groupid = i;
 		if (is_backend)
 			fio_server_send_gs(rs);
-		else if (!terse_output)
+		else if (output_format == FIO_OUTPUT_NORMAL)
 			show_group_stats(rs);
 	}
 
 	if (is_backend)
 		fio_server_send_du();
-	else if (!terse_output)
-		show_disk_util(0);
+	else if (output_format == FIO_OUTPUT_NORMAL)
+		show_disk_util(0, NULL);
 
-	free_disk_util();
+	show_idle_prof_stats(FIO_OUTPUT_NORMAL, NULL);
 
 	free(runstats);
 	free(threadstats);
 }
 
+static void *__show_running_run_stats(void *arg)
+{
+	struct thread_data *td;
+	unsigned long long *rt;
+	struct timeval tv;
+	int i;
+
+	rt = malloc(thread_number * sizeof(unsigned long long));
+	fio_gettime(&tv, NULL);
+
+	for_each_td(td, i) {
+		rt[i] = mtime_since(&td->start, &tv);
+		if (td_read(td) && td->io_bytes[DDIR_READ])
+			td->ts.runtime[DDIR_READ] += rt[i];
+		if (td_write(td) && td->io_bytes[DDIR_WRITE])
+			td->ts.runtime[DDIR_WRITE] += rt[i];
+		if (td_trim(td) && td->io_bytes[DDIR_TRIM])
+			td->ts.runtime[DDIR_TRIM] += rt[i];
+
+		update_rusage_stat(td);
+		td->ts.io_bytes[DDIR_READ] = td->io_bytes[DDIR_READ];
+		td->ts.io_bytes[DDIR_WRITE] = td->io_bytes[DDIR_WRITE];
+		td->ts.io_bytes[DDIR_TRIM] = td->io_bytes[DDIR_TRIM];
+		td->ts.total_run_time = mtime_since(&td->epoch, &tv);
+	}
+
+	show_run_stats();
+
+	for_each_td(td, i) {
+		if (td_read(td) && td->io_bytes[DDIR_READ])
+			td->ts.runtime[DDIR_READ] -= rt[i];
+		if (td_write(td) && td->io_bytes[DDIR_WRITE])
+			td->ts.runtime[DDIR_WRITE] -= rt[i];
+		if (td_trim(td) && td->io_bytes[DDIR_TRIM])
+			td->ts.runtime[DDIR_TRIM] -= rt[i];
+	}
+
+	free(rt);
+	return NULL;
+}
+
+/*
+ * Called from signal handler. It _should_ be safe to just run this inline
+ * in the sig handler, but we should be disturbing the system less by just
+ * creating a thread to do it.
+ */
+void show_running_run_stats(void)
+{
+	pthread_t thread;
+
+	pthread_create(&thread, NULL, __show_running_run_stats, NULL);
+	pthread_detach(thread);
+}
+
 static inline void add_stat_sample(struct io_stat *is, unsigned long data)
 {
 	double val = data;
@@ -1168,9 +1483,17 @@ static void add_log_sample(struct thread_data *td, struct io_log *iolog,
 		mw = iolog->avg_window[DDIR_WRITE].mean.u.f + 0.50;
 		__add_log_sample(iolog, mw, DDIR_WRITE, 0, elapsed);
 	}
+	if (iolog->avg_window[DDIR_TRIM].samples) {
+		unsigned long mw;
+
+		mw = iolog->avg_window[DDIR_TRIM].mean.u.f + 0.50;
+		__add_log_sample(iolog, mw, DDIR_TRIM, 0, elapsed);
+	}
+
 
 	reset_io_stat(&iolog->avg_window[DDIR_READ]);
 	reset_io_stat(&iolog->avg_window[DDIR_WRITE]);
+	reset_io_stat(&iolog->avg_window[DDIR_TRIM]);
 	iolog->avg_last = elapsed;
 }
 
@@ -1255,7 +1578,7 @@ void add_bw_sample(struct thread_data *td, enum fio_ddir ddir, unsigned int bs,
 	/*
 	 * Compute both read and write rates for the interval.
 	 */
-	for (ddir = DDIR_READ; ddir <= DDIR_WRITE; ddir++) {
+	for (ddir = DDIR_READ; ddir < DDIR_RWDIR_CNT; ddir++) {
 		uint64_t delta;
 
 		delta = td->this_io_bytes[ddir] - td->stat_io_bytes[ddir];
@@ -1290,7 +1613,7 @@ void add_iops_sample(struct thread_data *td, enum fio_ddir ddir,
 	/*
 	 * Compute both read and write rates for the interval.
 	 */
-	for (ddir = DDIR_READ; ddir <= DDIR_WRITE; ddir++) {
+	for (ddir = DDIR_READ; ddir < DDIR_RWDIR_CNT; ddir++) {
 		uint64_t delta;
 
 		delta = td->this_io_blocks[ddir] - td->stat_io_blocks[ddir];