X-Git-Url: https://git.kernel.dk/?a=blobdiff_plain;f=stat.c;h=a8a96c85a4120b0d70dee939c9102fe340565aae;hb=864314464e2772a9885da34ea041f130073affe9;hp=9d93dcd17040d6326a1c8ea9c0226d7dfd363bb6;hpb=b2a432bfbb6d10e93f2c8f8092d6db672d45af0d;p=fio.git

diff --git a/stat.c b/stat.c
index 9d93dcd1..a8a96c85 100644
--- a/stat.c
+++ b/stat.c
@@ -282,6 +282,46 @@ bool calc_lat(struct io_stat *is, unsigned long long *min,
 	return true;
 }
 
+void show_mixed_group_stats(struct group_run_stats *rs, struct buf_output *out) 
+{
+	char *io, *agg, *min, *max;
+	char *ioalt, *aggalt, *minalt, *maxalt;
+	uint64_t io_mix = 0, agg_mix = 0, min_mix = -1, max_mix = 0, min_run = -1, max_run = 0;
+	int i;
+	const int i2p = is_power_of_2(rs->kb_base);
+
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
+		if (!rs->max_run[i])
+			continue;
+		io_mix += rs->iobytes[i];
+		agg_mix += rs->agg[i];
+		min_mix = min_mix < rs->min_bw[i] ? min_mix : rs->min_bw[i];
+		max_mix = max_mix > rs->max_bw[i] ? max_mix : rs->max_bw[i];
+		min_run = min_run < rs->min_run[i] ? min_run : rs->min_run[i];
+		max_run = max_run > rs->max_run[i] ? max_run : rs->max_run[i];
+	}
+	io = num2str(io_mix, rs->sig_figs, 1, i2p, N2S_BYTE);
+	ioalt = num2str(io_mix, rs->sig_figs, 1, !i2p, N2S_BYTE);
+	agg = num2str(agg_mix, rs->sig_figs, 1, i2p, rs->unit_base);
+	aggalt = num2str(agg_mix, rs->sig_figs, 1, !i2p, rs->unit_base);
+	min = num2str(min_mix, rs->sig_figs, 1, i2p, rs->unit_base);
+	minalt = num2str(min_mix, rs->sig_figs, 1, !i2p, rs->unit_base);
+	max = num2str(max_mix, rs->sig_figs, 1, i2p, rs->unit_base);
+	maxalt = num2str(max_mix, rs->sig_figs, 1, !i2p, rs->unit_base);
+	log_buf(out, "  MIXED: bw=%s (%s), %s-%s (%s-%s), io=%s (%s), run=%llu-%llumsec\n",
+			agg, aggalt, min, max, minalt, maxalt, io, ioalt,
+			(unsigned long long) min_run,
+			(unsigned long long) max_run);
+	free(io);
+	free(agg);
+	free(min);
+	free(max);
+	free(ioalt);
+	free(aggalt);
+	free(minalt);
+	free(maxalt);
+}
+
 void show_group_stats(struct group_run_stats *rs, struct buf_output *out)
 {
 	char *io, *agg, *min, *max;
@@ -306,7 +346,7 @@ void show_group_stats(struct group_run_stats *rs, struct buf_output *out)
 		max = num2str(rs->max_bw[i], rs->sig_figs, 1, i2p, rs->unit_base);
 		maxalt = num2str(rs->max_bw[i], rs->sig_figs, 1, !i2p, rs->unit_base);
 		log_buf(out, "%s: bw=%s (%s), %s-%s (%s-%s), io=%s (%s), run=%llu-%llumsec\n",
-				rs->unified_rw_rep ? "  MIXED" : str[i],
+				(rs->unified_rw_rep == UNIFIED_MIXED) ? "  MIXED" : str[i],
 				agg, aggalt, min, max, minalt, maxalt, io, ioalt,
 				(unsigned long long) rs->min_run[i],
 				(unsigned long long) rs->max_run[i]);
@@ -320,6 +360,10 @@ void show_group_stats(struct group_run_stats *rs, struct buf_output *out)
 		free(minalt);
 		free(maxalt);
 	}
+	
+	/* Need to aggregate statisitics to show mixed values */
+	if (rs->unified_rw_rep == UNIFIED_BOTH) 
+		show_mixed_group_stats(rs, out);
 }
 
 void stat_calc_dist(uint64_t *map, unsigned long total, double *io_u_dist)
@@ -414,6 +458,180 @@ static void display_lat(const char *name, unsigned long long min,
 	free(maxp);
 }
 
+static double convert_agg_kbytes_percent(struct group_run_stats *rs, int ddir, int mean)
+{
+	double p_of_agg = 100.0;
+	if (rs && rs->agg[ddir] > 1024) {
+		p_of_agg = mean * 100.0 / (double) (rs->agg[ddir] / 1024.0);
+
+		if (p_of_agg > 100.0)
+			p_of_agg = 100.0;
+	}
+	return p_of_agg;
+}
+
+static void show_mixed_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
+			     struct buf_output *out)
+{
+	unsigned long runt;
+	unsigned long long min, max, bw, iops;
+	double mean, dev;
+	char *io_p, *bw_p, *bw_p_alt, *iops_p, *post_st = NULL;
+	struct thread_stat *ts_lcl;
+
+	int i2p;
+	int ddir = 0, i;
+
+	/* Handle aggregation of Reads (ddir = 0), Writes (ddir = 1), and Trims (ddir = 2) */
+	ts_lcl = malloc(sizeof(struct thread_stat));
+	memset((void *)ts_lcl, 0, sizeof(struct thread_stat));
+	ts_lcl->unified_rw_rep = UNIFIED_MIXED;               /* calculate mixed stats  */
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
+		ts_lcl->clat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->slat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->lat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->bw_stat[i].min_val = ULONG_MAX;
+		ts_lcl->iops_stat[i].min_val = ULONG_MAX;
+		ts_lcl->clat_high_prio_stat[i].min_val = ULONG_MAX;
+		ts_lcl->clat_low_prio_stat[i].min_val = ULONG_MAX;
+	}
+	ts_lcl->sync_stat.min_val = ULONG_MAX;
+
+	sum_thread_stats(ts_lcl, ts, 1);
+
+	assert(ddir_rw(ddir));
+
+	if (!ts_lcl->runtime[ddir])
+		return;
+
+	i2p = is_power_of_2(rs->kb_base);
+	runt = ts_lcl->runtime[ddir];
+
+	bw = (1000 * ts_lcl->io_bytes[ddir]) / runt;
+	io_p = num2str(ts_lcl->io_bytes[ddir], ts->sig_figs, 1, i2p, N2S_BYTE);
+	bw_p = num2str(bw, ts->sig_figs, 1, i2p, ts->unit_base);
+	bw_p_alt = num2str(bw, ts->sig_figs, 1, !i2p, ts->unit_base);
+
+	iops = (1000 * ts_lcl->total_io_u[ddir]) / runt;
+	iops_p = num2str(iops, ts->sig_figs, 1, 0, N2S_NONE);
+
+	log_buf(out, "  mixed: IOPS=%s, BW=%s (%s)(%s/%llumsec)%s\n",
+			iops_p, bw_p, bw_p_alt, io_p,
+			(unsigned long long) ts_lcl->runtime[ddir],
+			post_st ? : "");
+
+	free(post_st);
+	free(io_p);
+	free(bw_p);
+	free(bw_p_alt);
+	free(iops_p);
+
+	if (calc_lat(&ts_lcl->slat_stat[ddir], &min, &max, &mean, &dev))
+		display_lat("slat", min, max, mean, dev, out);
+	if (calc_lat(&ts_lcl->clat_stat[ddir], &min, &max, &mean, &dev))
+		display_lat("clat", min, max, mean, dev, out);
+	if (calc_lat(&ts_lcl->lat_stat[ddir], &min, &max, &mean, &dev))
+		display_lat(" lat", min, max, mean, dev, out);
+	if (calc_lat(&ts_lcl->clat_high_prio_stat[ddir], &min, &max, &mean, &dev)) {
+		display_lat(ts_lcl->lat_percentiles ? "high prio_lat" : "high prio_clat",
+				min, max, mean, dev, out);
+		if (calc_lat(&ts_lcl->clat_low_prio_stat[ddir], &min, &max, &mean, &dev))
+			display_lat(ts_lcl->lat_percentiles ? "low prio_lat" : "low prio_clat",
+					min, max, mean, dev, out);
+	}
+
+	if (ts->slat_percentiles && ts_lcl->slat_stat[ddir].samples > 0)
+		show_clat_percentiles(ts_lcl->io_u_plat[FIO_SLAT][ddir],
+				ts_lcl->slat_stat[ddir].samples,
+				ts->percentile_list,
+				ts->percentile_precision, "slat", out);
+	if (ts->clat_percentiles && ts_lcl->clat_stat[ddir].samples > 0)
+		show_clat_percentiles(ts_lcl->io_u_plat[FIO_CLAT][ddir],
+				ts_lcl->clat_stat[ddir].samples,
+				ts->percentile_list,
+				ts->percentile_precision, "clat", out);
+	if (ts->lat_percentiles && ts_lcl->lat_stat[ddir].samples > 0)
+		show_clat_percentiles(ts_lcl->io_u_plat[FIO_LAT][ddir],
+				ts_lcl->lat_stat[ddir].samples,
+				ts->percentile_list,
+				ts->percentile_precision, "lat", out);
+
+	if (ts->clat_percentiles || ts->lat_percentiles) {
+		const char *name = ts->lat_percentiles ? "lat" : "clat";
+		char prio_name[32];
+		uint64_t samples;
+
+		if (ts->lat_percentiles)
+			samples = ts_lcl->lat_stat[ddir].samples;
+		else
+			samples = ts_lcl->clat_stat[ddir].samples;
+
+		/* Only print this if some high and low priority stats were collected */
+		if (ts_lcl->clat_high_prio_stat[ddir].samples > 0 &&
+				ts_lcl->clat_low_prio_stat[ddir].samples > 0)
+		{
+			sprintf(prio_name, "high prio (%.2f%%) %s",
+					100. * (double) ts_lcl->clat_high_prio_stat[ddir].samples / (double) samples,
+					name);
+			show_clat_percentiles(ts_lcl->io_u_plat_high_prio[ddir],
+					ts_lcl->clat_high_prio_stat[ddir].samples,
+					ts->percentile_list,
+					ts->percentile_precision, prio_name, out);
+
+			sprintf(prio_name, "low prio (%.2f%%) %s",
+					100. * (double) ts_lcl->clat_low_prio_stat[ddir].samples / (double) samples,
+					name);
+			show_clat_percentiles(ts_lcl->io_u_plat_low_prio[ddir],
+					ts_lcl->clat_low_prio_stat[ddir].samples,
+					ts->percentile_list,
+					ts->percentile_precision, prio_name, out);
+		}
+	}
+
+	if (calc_lat(&ts_lcl->bw_stat[ddir], &min, &max, &mean, &dev)) {
+		double p_of_agg = 100.0, fkb_base = (double)rs->kb_base;
+		const char *bw_str;
+
+		if ((rs->unit_base == 1) && i2p)
+			bw_str = "Kibit";
+		else if (rs->unit_base == 1)
+			bw_str = "kbit";
+		else if (i2p)
+			bw_str = "KiB";
+		else
+			bw_str = "kB";
+
+		p_of_agg = convert_agg_kbytes_percent(rs, ddir, mean);
+
+		if (rs->unit_base == 1) {
+			min *= 8.0;
+			max *= 8.0;
+			mean *= 8.0;
+			dev *= 8.0;
+		}
+
+		if (mean > fkb_base * fkb_base) {
+			min /= fkb_base;
+			max /= fkb_base;
+			mean /= fkb_base;
+			dev /= fkb_base;
+			bw_str = (rs->unit_base == 1 ? "Mibit" : "MiB");
+		}
+
+		log_buf(out, "   bw (%5s/s): min=%5llu, max=%5llu, per=%3.2f%%, "
+			"avg=%5.02f, stdev=%5.02f, samples=%" PRIu64 "\n",
+			bw_str, min, max, p_of_agg, mean, dev,
+			(&ts_lcl->bw_stat[ddir])->samples);
+	}
+	if (calc_lat(&ts_lcl->iops_stat[ddir], &min, &max, &mean, &dev)) {
+		log_buf(out, "   iops        : min=%5llu, max=%5llu, "
+			"avg=%5.02f, stdev=%5.02f, samples=%" PRIu64 "\n",
+			min, max, mean, dev, (&ts_lcl->iops_stat[ddir])->samples);
+	}
+
+	free(ts_lcl);
+}
+
 static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 			     int ddir, struct buf_output *out)
 {
@@ -465,7 +683,7 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 	}
 
 	log_buf(out, "  %s: IOPS=%s, BW=%s (%s)(%s/%llumsec)%s\n",
-			rs->unified_rw_rep ? "mixed" : io_ddir_name(ddir),
+			(ts->unified_rw_rep == UNIFIED_MIXED) ? "mixed" : io_ddir_name(ddir),
 			iops_p, bw_p, bw_p_alt, io_p,
 			(unsigned long long) ts->runtime[ddir],
 			post_st ? : "");
@@ -482,27 +700,43 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 		display_lat("clat", min, max, mean, dev, out);
 	if (calc_lat(&ts->lat_stat[ddir], &min, &max, &mean, &dev))
 		display_lat(" lat", min, max, mean, dev, out);
-	if (calc_lat(&ts->clat_high_prio_stat[ddir], &min, &max, &mean, &dev))
-		display_lat("prio_clat", min, max, mean, dev, out);
+	if (calc_lat(&ts->clat_high_prio_stat[ddir], &min, &max, &mean, &dev)) {
+		display_lat(ts->lat_percentiles ? "high prio_lat" : "high prio_clat",
+				min, max, mean, dev, out);
+		if (calc_lat(&ts->clat_low_prio_stat[ddir], &min, &max, &mean, &dev))
+			display_lat(ts->lat_percentiles ? "low prio_lat" : "low prio_clat",
+					min, max, mean, dev, out);
+	}
+
+	if (ts->slat_percentiles && ts->slat_stat[ddir].samples > 0)
+		show_clat_percentiles(ts->io_u_plat[FIO_SLAT][ddir],
+					ts->slat_stat[ddir].samples,
+					ts->percentile_list,
+					ts->percentile_precision, "slat", out);
+	if (ts->clat_percentiles && ts->clat_stat[ddir].samples > 0)
+		show_clat_percentiles(ts->io_u_plat[FIO_CLAT][ddir],
+					ts->clat_stat[ddir].samples,
+					ts->percentile_list,
+					ts->percentile_precision, "clat", out);
+	if (ts->lat_percentiles && ts->lat_stat[ddir].samples > 0)
+		show_clat_percentiles(ts->io_u_plat[FIO_LAT][ddir],
+					ts->lat_stat[ddir].samples,
+					ts->percentile_list,
+					ts->percentile_precision, "lat", out);
 
 	if (ts->clat_percentiles || ts->lat_percentiles) {
-		const char *name = ts->clat_percentiles ? "clat" : " lat";
+		const char *name = ts->lat_percentiles ? "lat" : "clat";
 		char prio_name[32];
 		uint64_t samples;
 
-		if (ts->clat_percentiles)
-			samples = ts->clat_stat[ddir].samples;
-		else
+		if (ts->lat_percentiles)
 			samples = ts->lat_stat[ddir].samples;
-
-		show_clat_percentiles(ts->io_u_plat[ddir],
-					samples,
-					ts->percentile_list,
-					ts->percentile_precision, name, out);
+		else
+			samples = ts->clat_stat[ddir].samples;
 
 		/* Only print this if some high and low priority stats were collected */
 		if (ts->clat_high_prio_stat[ddir].samples > 0 &&
-			ts->clat_prio_stat[ddir].samples > 0)
+			ts->clat_low_prio_stat[ddir].samples > 0)
 		{
 			sprintf(prio_name, "high prio (%.2f%%) %s",
 					100. * (double) ts->clat_high_prio_stat[ddir].samples / (double) samples,
@@ -513,14 +747,15 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 						ts->percentile_precision, prio_name, out);
 
 			sprintf(prio_name, "low prio (%.2f%%) %s",
-					100. * (double) ts->clat_prio_stat[ddir].samples / (double) samples,
+					100. * (double) ts->clat_low_prio_stat[ddir].samples / (double) samples,
 					name);
-			show_clat_percentiles(ts->io_u_plat_prio[ddir],
-						ts->clat_prio_stat[ddir].samples,
+			show_clat_percentiles(ts->io_u_plat_low_prio[ddir],
+						ts->clat_low_prio_stat[ddir].samples,
 						ts->percentile_list,
 						ts->percentile_precision, prio_name, out);
 		}
 	}
+
 	if (calc_lat(&ts->bw_stat[ddir], &min, &max, &mean, &dev)) {
 		double p_of_agg = 100.0, fkb_base = (double)rs->kb_base;
 		const char *bw_str;
@@ -534,11 +769,7 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts,
 		else
 			bw_str = "kB";
 
-		if (rs->agg[ddir]) {
-			p_of_agg = mean * 100 / (double) (rs->agg[ddir] / 1024);
-			if (p_of_agg > 100.0)
-				p_of_agg = 100.0;
-		}
+		p_of_agg = convert_agg_kbytes_percent(rs, ddir, mean);
 
 		if (rs->unit_base == 1) {
 			min *= 8.0;
@@ -937,7 +1168,7 @@ void json_array_add_disk_util(struct disk_util_stat *dus,
 	obj = json_create_object();
 	json_array_add_value_object(array, obj);
 
-	json_object_add_value_string(obj, "name", dus->name);
+	json_object_add_value_string(obj, "name", (const char *)dus->name);
 	json_object_add_value_int(obj, "read_ios", dus->s.ios[0]);
 	json_object_add_value_int(obj, "write_ios", dus->s.ios[1]);
 	json_object_add_value_int(obj, "read_merges", dus->s.merges[0]);
@@ -1053,12 +1284,13 @@ static void show_thread_status_normal(struct thread_stat *ts,
 	if (strlen(ts->description))
 		log_buf(out, "  Description  : [%s]\n", ts->description);
 
-	if (ts->io_bytes[DDIR_READ])
-		show_ddir_status(rs, ts, DDIR_READ, out);
-	if (ts->io_bytes[DDIR_WRITE])
-		show_ddir_status(rs, ts, DDIR_WRITE, out);
-	if (ts->io_bytes[DDIR_TRIM])
-		show_ddir_status(rs, ts, DDIR_TRIM, out);
+	for_each_rw_ddir(ddir) {
+		if (ts->io_bytes[ddir])
+			show_ddir_status(rs, ts, ddir, out);
+	}
+
+	if (ts->unified_rw_rep == UNIFIED_BOTH)
+		show_mixed_ddir_status(rs, ts, out);
 
 	show_latencies(ts, out);
 
@@ -1170,14 +1402,19 @@ static void show_ddir_status_terse(struct thread_stat *ts,
 	else
 		log_buf(out, ";%llu;%llu;%f;%f", 0ULL, 0ULL, 0.0, 0.0);
 
-	if (ts->clat_percentiles || ts->lat_percentiles) {
-		len = calc_clat_percentiles(ts->io_u_plat[ddir],
+	if (ts->lat_percentiles)
+		len = calc_clat_percentiles(ts->io_u_plat[FIO_LAT][ddir],
+					ts->lat_stat[ddir].samples,
+					ts->percentile_list, &ovals, &maxv,
+					&minv);
+	else if (ts->clat_percentiles)
+		len = calc_clat_percentiles(ts->io_u_plat[FIO_CLAT][ddir],
 					ts->clat_stat[ddir].samples,
 					ts->percentile_list, &ovals, &maxv,
 					&minv);
-	} else
+	else
 		len = 0;
-
+	
 	for (i = 0; i < FIO_IO_U_LIST_MAX_LEN; i++) {
 		if (i >= len) {
 			log_buf(out, ";0%%=0");
@@ -1221,28 +1458,107 @@ static void show_ddir_status_terse(struct thread_stat *ts,
 	}
 }
 
+static void show_mixed_ddir_status_terse(struct thread_stat *ts,
+				   struct group_run_stats *rs,
+				   int ver, struct buf_output *out)
+{
+	struct thread_stat *ts_lcl;
+	int i;
+
+	/* Handle aggregation of Reads (ddir = 0), Writes (ddir = 1), and Trims (ddir = 2) */
+	ts_lcl = malloc(sizeof(struct thread_stat));
+	memset((void *)ts_lcl, 0, sizeof(struct thread_stat));
+	ts_lcl->unified_rw_rep = UNIFIED_MIXED;               /* calculate mixed stats  */
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
+		ts_lcl->clat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->slat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->lat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->bw_stat[i].min_val = ULONG_MAX;
+		ts_lcl->iops_stat[i].min_val = ULONG_MAX;
+		ts_lcl->clat_high_prio_stat[i].min_val = ULONG_MAX;
+		ts_lcl->clat_low_prio_stat[i].min_val = ULONG_MAX;
+	}
+	ts_lcl->sync_stat.min_val = ULONG_MAX;
+	ts_lcl->lat_percentiles = ts->lat_percentiles;
+	ts_lcl->clat_percentiles = ts->clat_percentiles;
+	ts_lcl->slat_percentiles = ts->slat_percentiles;
+	ts_lcl->percentile_precision = ts->percentile_precision;		
+	memcpy(ts_lcl->percentile_list, ts->percentile_list, sizeof(ts->percentile_list));
+	
+	sum_thread_stats(ts_lcl, ts, 1);
+
+	/* add the aggregated stats to json parent */
+	show_ddir_status_terse(ts_lcl, rs, DDIR_READ, ver, out);
+	free(ts_lcl);
+}
+
+static struct json_object *add_ddir_lat_json(struct thread_stat *ts, uint32_t percentiles,
+		struct io_stat *lat_stat, uint64_t *io_u_plat)
+{
+	char buf[120];
+	double mean, dev;
+	unsigned int i, len;
+	struct json_object *lat_object, *percentile_object, *clat_bins_object;
+	unsigned long long min, max, maxv, minv, *ovals = NULL;
+
+	if (!calc_lat(lat_stat, &min, &max, &mean, &dev)) {
+		min = max = 0;
+		mean = dev = 0.0;
+	}
+	lat_object = json_create_object();
+	json_object_add_value_int(lat_object, "min", min);
+	json_object_add_value_int(lat_object, "max", max);
+	json_object_add_value_float(lat_object, "mean", mean);
+	json_object_add_value_float(lat_object, "stddev", dev);
+	json_object_add_value_int(lat_object, "N", lat_stat->samples);
+
+	if (percentiles && lat_stat->samples) {
+		len = calc_clat_percentiles(io_u_plat, lat_stat->samples,
+				ts->percentile_list, &ovals, &maxv, &minv);
+
+		if (len > FIO_IO_U_LIST_MAX_LEN)
+			len = FIO_IO_U_LIST_MAX_LEN;
+
+		percentile_object = json_create_object();
+		json_object_add_value_object(lat_object, "percentile", percentile_object);
+		for (i = 0; i < len; i++) {
+			snprintf(buf, sizeof(buf), "%f", ts->percentile_list[i].u.f);
+			json_object_add_value_int(percentile_object, buf, ovals[i]);
+		}
+		free(ovals);
+
+		if (output_format & FIO_OUTPUT_JSON_PLUS) {
+			clat_bins_object = json_create_object();
+			json_object_add_value_object(lat_object, "bins", clat_bins_object);
+
+			for(i = 0; i < FIO_IO_U_PLAT_NR; i++)
+				if (io_u_plat[i]) {
+					snprintf(buf, sizeof(buf), "%llu", plat_idx_to_val(i));
+					json_object_add_value_int(clat_bins_object, buf, io_u_plat[i]);
+				}
+		}
+	}
+
+	return lat_object;
+}
+
 static void add_ddir_status_json(struct thread_stat *ts,
 		struct group_run_stats *rs, int ddir, struct json_object *parent)
 {
-	unsigned long long min, max, minv, maxv;
+	unsigned long long min, max;
 	unsigned long long bw_bytes, bw;
-	unsigned long long *ovals = NULL;
 	double mean, dev, iops;
-	unsigned int len;
-	int i;
-	struct json_object *dir_object, *tmp_object, *percentile_object = NULL,
-		*clat_bins_object = NULL;
-	char buf[120];
+	struct json_object *dir_object, *tmp_object;
 	double p_of_agg = 100.0;
 
 	assert(ddir_rw(ddir) || ddir_sync(ddir));
 
-	if (ts->unified_rw_rep && ddir != DDIR_READ)
+	if ((ts->unified_rw_rep == UNIFIED_MIXED) && ddir != DDIR_READ)
 		return;
 
 	dir_object = json_create_object();
 	json_object_add_value_object(parent,
-		ts->unified_rw_rep ? "mixed" : io_ddir_name(ddir), dir_object);
+		(ts->unified_rw_rep == UNIFIED_MIXED) ? "mixed" : io_ddir_name(ddir), dir_object);
 
 	if (ddir_rw(ddir)) {
 		bw_bytes = 0;
@@ -1266,233 +1582,55 @@ static void add_ddir_status_json(struct thread_stat *ts,
 		json_object_add_value_int(dir_object, "short_ios", ts->short_io_u[ddir]);
 		json_object_add_value_int(dir_object, "drop_ios", ts->drop_io_u[ddir]);
 
-		if (!calc_lat(&ts->slat_stat[ddir], &min, &max, &mean, &dev)) {
-			min = max = 0;
-			mean = dev = 0.0;
-		}
-		tmp_object = json_create_object();
+		tmp_object = add_ddir_lat_json(ts, ts->slat_percentiles,
+				&ts->slat_stat[ddir], ts->io_u_plat[FIO_SLAT][ddir]);
 		json_object_add_value_object(dir_object, "slat_ns", 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();
+
+		tmp_object = add_ddir_lat_json(ts, ts->clat_percentiles,
+				&ts->clat_stat[ddir], ts->io_u_plat[FIO_CLAT][ddir]);
 		json_object_add_value_object(dir_object, "clat_ns", 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);
-	} else {
-		if (!calc_lat(&ts->sync_stat, &min, &max, &mean, &dev)) {
-			min = max = 0;
-			mean = dev = 0.0;
-		}
 
-		tmp_object = json_create_object();
+		tmp_object = add_ddir_lat_json(ts, ts->lat_percentiles,
+				&ts->lat_stat[ddir], ts->io_u_plat[FIO_LAT][ddir]);
 		json_object_add_value_object(dir_object, "lat_ns", tmp_object);
+	} else {
 		json_object_add_value_int(dir_object, "total_ios", ts->total_io_u[DDIR_SYNC]);
-		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 || ts->lat_percentiles) {
-		if (ddir_rw(ddir)) {
-			uint64_t samples;
-
-			if (ts->clat_percentiles)
-				samples = ts->clat_stat[ddir].samples;
-			else
-				samples = ts->lat_stat[ddir].samples;
-
-			len = calc_clat_percentiles(ts->io_u_plat[ddir],
-					samples, ts->percentile_list, &ovals,
-					&maxv, &minv);
-		} else {
-			len = calc_clat_percentiles(ts->io_u_sync_plat,
-					ts->sync_stat.samples,
-					ts->percentile_list, &ovals, &maxv,
-					&minv);
-		}
-
-		if (len > FIO_IO_U_LIST_MAX_LEN)
-			len = FIO_IO_U_LIST_MAX_LEN;
-	} else
-		len = 0;
-
-	if (ts->clat_percentiles) {
-		percentile_object = json_create_object();
-		json_object_add_value_object(tmp_object, "percentile", percentile_object);
-		for (i = 0; i < len; i++) {
-			snprintf(buf, sizeof(buf), "%f",
-				 ts->percentile_list[i].u.f);
-			json_object_add_value_int(percentile_object, buf,
-						  ovals[i]);
-		}
-	}
-
-	free(ovals);
-
-	if (output_format & FIO_OUTPUT_JSON_PLUS && ts->clat_percentiles) {
-		clat_bins_object = json_create_object();
-		json_object_add_value_object(tmp_object, "bins",
-					     clat_bins_object);
-
-		for(i = 0; i < FIO_IO_U_PLAT_NR; i++) {
-			if (ddir_rw(ddir)) {
-				if (ts->io_u_plat[ddir][i]) {
-					snprintf(buf, sizeof(buf), "%llu", plat_idx_to_val(i));
-					json_object_add_value_int(clat_bins_object, buf, ts->io_u_plat[ddir][i]);
-				}
-			} else {
-				if (ts->io_u_sync_plat[i]) {
-					snprintf(buf, sizeof(buf), "%llu", plat_idx_to_val(i));
-					json_object_add_value_int(clat_bins_object, buf, ts->io_u_sync_plat[i]);
-				}
-			}
-		}
+		tmp_object = add_ddir_lat_json(ts, ts->lat_percentiles | ts->clat_percentiles,
+				&ts->sync_stat, ts->io_u_sync_plat);
+		json_object_add_value_object(dir_object, "lat_ns", tmp_object);
 	}
 
+	if (!ddir_rw(ddir))
+		return;
 
 	/* Only print PRIO latencies if some high priority samples were gathered */
 	if (ts->clat_high_prio_stat[ddir].samples > 0) {
-		/* START OF HIGH PRIO CLAT */
-	    if (!calc_lat(&ts->clat_high_prio_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_prio",
-				tmp_object);
-		json_object_add_value_int(tmp_object, "samples",
-				ts->clat_high_prio_stat[ddir].samples);
-		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_high_prio[ddir],
-						ts->clat_high_prio_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 (output_format & FIO_OUTPUT_JSON_PLUS) {
-			clat_bins_object = json_create_object();
-			json_object_add_value_object(tmp_object, "bins", clat_bins_object);
-			for(i = 0; i < FIO_IO_U_PLAT_NR; i++) {
-				snprintf(buf, sizeof(buf), "%d", i);
-				json_object_add_value_int(clat_bins_object, (const char *)buf,
-						ts->io_u_plat_high_prio[ddir][i]);
-			}
-			json_object_add_value_int(clat_bins_object, "FIO_IO_U_PLAT_BITS",
-					FIO_IO_U_PLAT_BITS);
-			json_object_add_value_int(clat_bins_object, "FIO_IO_U_PLAT_VAL",
-					FIO_IO_U_PLAT_VAL);
-			json_object_add_value_int(clat_bins_object, "FIO_IO_U_PLAT_NR",
-					FIO_IO_U_PLAT_NR);
-		}
-		/* END OF HIGH PRIO CLAT */
-
-		/* START OF PRIO CLAT */
-	    if (!calc_lat(&ts->clat_prio_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_low_prio",
-				tmp_object);
-		json_object_add_value_int(tmp_object, "samples",
-				ts->clat_prio_stat[ddir].samples);
-		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_prio[ddir],
-						ts->clat_prio_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]);
-		}
+		const char *high, *low;
 
-		if (output_format & FIO_OUTPUT_JSON_PLUS) {
-			clat_bins_object = json_create_object();
-			json_object_add_value_object(tmp_object, "bins", clat_bins_object);
-			for(i = 0; i < FIO_IO_U_PLAT_NR; i++) {
-				snprintf(buf, sizeof(buf), "%d", i);
-				json_object_add_value_int(clat_bins_object, (const char *)buf,
-						ts->io_u_plat_prio[ddir][i]);
-			}
-			json_object_add_value_int(clat_bins_object, "FIO_IO_U_PLAT_BITS",
-					FIO_IO_U_PLAT_BITS);
-			json_object_add_value_int(clat_bins_object, "FIO_IO_U_PLAT_VAL",
-					FIO_IO_U_PLAT_VAL);
-			json_object_add_value_int(clat_bins_object, "FIO_IO_U_PLAT_NR",
-					FIO_IO_U_PLAT_NR);
+		if (ts->lat_percentiles) {
+			high = "lat_high_prio";
+			low = "lat_low_prio";
+		} else {
+			high = "clat_high_prio";
+			low = "clat_low_prio";
 		}
-		/* END OF PRIO CLAT */
-	}
 
-	if (!ddir_rw(ddir))
-		return;
+		tmp_object = add_ddir_lat_json(ts, ts->clat_percentiles | ts->lat_percentiles,
+				&ts->clat_high_prio_stat[ddir], ts->io_u_plat_high_prio[ddir]);
+		json_object_add_value_object(dir_object, high, tmp_object);
 
-	if (!calc_lat(&ts->lat_stat[ddir], &min, &max, &mean, &dev)) {
-		min = max = 0;
-		mean = dev = 0.0;
+		tmp_object = add_ddir_lat_json(ts, ts->clat_percentiles | ts->lat_percentiles,
+				&ts->clat_low_prio_stat[ddir], ts->io_u_plat_low_prio[ddir]);
+		json_object_add_value_object(dir_object, low, tmp_object);
 	}
-	tmp_object = json_create_object();
-	json_object_add_value_object(dir_object, "lat_ns", 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->lat_percentiles)
-		json_object_add_value_object(tmp_object, "percentile", percentile_object);
-	if (output_format & FIO_OUTPUT_JSON_PLUS && ts->lat_percentiles)
-		json_object_add_value_object(tmp_object, "bins", clat_bins_object);
 
 	if (calc_lat(&ts->bw_stat[ddir], &min, &max, &mean, &dev)) {
-		if (rs->agg[ddir]) {
-			p_of_agg = mean * 100 / (double) (rs->agg[ddir] / 1024);
-			if (p_of_agg > 100.0)
-				p_of_agg = 100.0;
-		}
+		p_of_agg = convert_agg_kbytes_percent(rs, ddir, mean);
 	} 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", p_of_agg);
@@ -1523,6 +1661,39 @@ static void add_ddir_status_json(struct thread_stat *ts,
 	}
 }
 
+static void add_mixed_ddir_status_json(struct thread_stat *ts,
+		struct group_run_stats *rs, struct json_object *parent)
+{
+	struct thread_stat *ts_lcl;
+	int i;
+
+	/* Handle aggregation of Reads (ddir = 0), Writes (ddir = 1), and Trims (ddir = 2) */
+	ts_lcl = malloc(sizeof(struct thread_stat));
+	memset((void *)ts_lcl, 0, sizeof(struct thread_stat));
+	ts_lcl->unified_rw_rep = UNIFIED_MIXED;               /* calculate mixed stats  */
+	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
+		ts_lcl->clat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->slat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->lat_stat[i].min_val = ULONG_MAX;
+		ts_lcl->bw_stat[i].min_val = ULONG_MAX;
+		ts_lcl->iops_stat[i].min_val = ULONG_MAX;
+		ts_lcl->clat_high_prio_stat[i].min_val = ULONG_MAX;
+		ts_lcl->clat_low_prio_stat[i].min_val = ULONG_MAX;
+	}
+	ts_lcl->sync_stat.min_val = ULONG_MAX;
+	ts_lcl->lat_percentiles = ts->lat_percentiles;
+	ts_lcl->clat_percentiles = ts->clat_percentiles;
+	ts_lcl->slat_percentiles = ts->slat_percentiles;
+	ts_lcl->percentile_precision = ts->percentile_precision;		
+	memcpy(ts_lcl->percentile_list, ts->percentile_list, sizeof(ts->percentile_list));
+
+	sum_thread_stats(ts_lcl, ts, 1);
+
+	/* add the aggregated stats to json parent */
+	add_ddir_status_json(ts_lcl, rs, DDIR_READ, parent);
+	free(ts_lcl);
+}
+
 static void show_thread_status_terse_all(struct thread_stat *ts,
 					 struct group_run_stats *rs, int ver,
 					 struct buf_output *out)
@@ -1540,14 +1711,17 @@ static void show_thread_status_terse_all(struct thread_stat *ts,
 		log_buf(out, "%d;%s;%s;%d;%d", ver, fio_version_string,
 			ts->name, ts->groupid, ts->error);
 
-	/* Log Read Status */
+	/* Log Read Status, or mixed if unified_rw_rep = 1 */
 	show_ddir_status_terse(ts, rs, DDIR_READ, ver, out);
-	/* Log Write Status */
-	show_ddir_status_terse(ts, rs, DDIR_WRITE, ver, out);
-	/* Log Trim Status */
-	if (ver == 2 || ver == 4 || ver == 5)
-		show_ddir_status_terse(ts, rs, DDIR_TRIM, ver, out);
-
+	if (ts->unified_rw_rep != UNIFIED_MIXED) {
+		/* Log Write Status */
+		show_ddir_status_terse(ts, rs, DDIR_WRITE, ver, out);
+		/* Log Trim Status */
+		if (ver == 2 || ver == 4 || ver == 5)
+			show_ddir_status_terse(ts, rs, DDIR_TRIM, ver, out);
+	}
+	if (ts->unified_rw_rep == UNIFIED_BOTH)
+		show_mixed_ddir_status_terse(ts, rs, ver, out);
 	/* CPU Usage */
 	if (ts->total_run_time) {
 		double runt = (double) ts->total_run_time;
@@ -1613,12 +1787,8 @@ static void json_add_job_opts(struct json_object *root, const char *name,
 	json_object_add_value_object(root, name, dir_object);
 
 	flist_for_each(entry, opt_list) {
-		const char *pos = "";
-
 		p = flist_entry(entry, struct print_option, list);
-		if (p->value)
-			pos = p->value;
-		json_object_add_value_string(dir_object, p->name, pos);
+		json_object_add_value_string(dir_object, p->name, p->value);
 	}
 }
 
@@ -1656,6 +1826,9 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts,
 	add_ddir_status_json(ts, rs, DDIR_TRIM, root);
 	add_ddir_status_json(ts, rs, DDIR_SYNC, root);
 
+	if (ts->unified_rw_rep == UNIFIED_BOTH)
+		add_mixed_ddir_status_json(ts, rs, root);
+
 	/* CPU Usage */
 	if (ts->total_run_time) {
 		double runt = (double) ts->total_run_time;
@@ -1981,13 +2154,13 @@ void sum_group_stats(struct group_run_stats *dst, struct group_run_stats *src)
 void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src,
 		      bool first)
 {
-	int l, k;
+	int k, l, m;
 
 	for (l = 0; l < DDIR_RWDIR_CNT; l++) {
-		if (!dst->unified_rw_rep) {
+		if (!(dst->unified_rw_rep == UNIFIED_MIXED)) {
 			sum_stat(&dst->clat_stat[l], &src->clat_stat[l], first, false);
 			sum_stat(&dst->clat_high_prio_stat[l], &src->clat_high_prio_stat[l], first, false);
-			sum_stat(&dst->clat_prio_stat[l], &src->clat_prio_stat[l], first, false);
+			sum_stat(&dst->clat_low_prio_stat[l], &src->clat_low_prio_stat[l], first, false);
 			sum_stat(&dst->slat_stat[l], &src->slat_stat[l], first, false);
 			sum_stat(&dst->lat_stat[l], &src->lat_stat[l], first, false);
 			sum_stat(&dst->bw_stat[l], &src->bw_stat[l], first, true);
@@ -1999,8 +2172,8 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src,
 				dst->runtime[l] = src->runtime[l];
 		} else {
 			sum_stat(&dst->clat_stat[0], &src->clat_stat[l], first, false);
-			sum_stat(&dst->clat_high_prio_stat[l], &src->clat_high_prio_stat[l], first, false);
-			sum_stat(&dst->clat_prio_stat[l], &src->clat_prio_stat[l], first, false);
+			sum_stat(&dst->clat_high_prio_stat[0], &src->clat_high_prio_stat[l], first, false);
+			sum_stat(&dst->clat_low_prio_stat[0], &src->clat_low_prio_stat[l], first, false);
 			sum_stat(&dst->slat_stat[0], &src->slat_stat[l], first, false);
 			sum_stat(&dst->lat_stat[0], &src->lat_stat[l], first, false);
 			sum_stat(&dst->bw_stat[0], &src->bw_stat[l], first, true);
@@ -2039,11 +2212,8 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src,
 	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 < FIO_IO_U_PLAT_NR; k++)
-		dst->io_u_sync_plat[k] += src->io_u_sync_plat[k];
-
 	for (k = 0; k < DDIR_RWDIR_CNT; k++) {
-		if (!dst->unified_rw_rep) {
+		if (!(dst->unified_rw_rep == UNIFIED_MIXED)) {
 			dst->total_io_u[k] += src->total_io_u[k];
 			dst->short_io_u[k] += src->short_io_u[k];
 			dst->drop_io_u[k] += src->drop_io_u[k];
@@ -2056,18 +2226,25 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src,
 
 	dst->total_io_u[DDIR_SYNC] += src->total_io_u[DDIR_SYNC];
 
-	for (k = 0; k < DDIR_RWDIR_CNT; k++) {
-		int m;
+	for (k = 0; k < FIO_LAT_CNT; k++)
+		for (l = 0; l < DDIR_RWDIR_CNT; l++)
+			for (m = 0; m < FIO_IO_U_PLAT_NR; m++)
+				if (!(dst->unified_rw_rep == UNIFIED_MIXED))
+					dst->io_u_plat[k][l][m] += src->io_u_plat[k][l][m];
+				else
+					dst->io_u_plat[k][0][m] += src->io_u_plat[k][l][m];
+
+	for (k = 0; k < FIO_IO_U_PLAT_NR; k++)
+		dst->io_u_sync_plat[k] += src->io_u_sync_plat[k];
 
+	for (k = 0; k < DDIR_RWDIR_CNT; k++) {
 		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];
+			if (!(dst->unified_rw_rep == UNIFIED_MIXED)) {
 				dst->io_u_plat_high_prio[k][m] += src->io_u_plat_high_prio[k][m];
-				dst->io_u_plat_prio[k][m] += src->io_u_plat_prio[k][m];
+				dst->io_u_plat_low_prio[k][m] += src->io_u_plat_low_prio[k][m];
 			} else {
-				dst->io_u_plat[0][m] += src->io_u_plat[k][m];
 				dst->io_u_plat_high_prio[0][m] += src->io_u_plat_high_prio[k][m];
-				dst->io_u_plat_prio[0][m] += src->io_u_plat_prio[k][m];
+				dst->io_u_plat_low_prio[0][m] += src->io_u_plat_low_prio[k][m];
 			}
 
 		}
@@ -2103,7 +2280,7 @@ void init_thread_stat(struct thread_stat *ts)
 		ts->bw_stat[j].min_val = -1UL;
 		ts->iops_stat[j].min_val = -1UL;
 		ts->clat_high_prio_stat[j].min_val = -1UL;
-		ts->clat_prio_stat[j].min_val = -1UL;
+		ts->clat_low_prio_stat[j].min_val = -1UL;
 	}
 	ts->sync_stat.min_val = -1UL;
 	ts->groupid = -1;
@@ -2173,6 +2350,7 @@ void __show_run_stats(void)
 
 		ts->clat_percentiles = td->o.clat_percentiles;
 		ts->lat_percentiles = td->o.lat_percentiles;
+		ts->slat_percentiles = td->o.slat_percentiles;
 		ts->percentile_precision = td->o.percentile_precision;
 		memcpy(ts->percentile_list, td->o.percentile_list, sizeof(td->o.percentile_list));
 		opt_lists[j] = &td->opt_list;
@@ -2270,7 +2448,7 @@ void __show_run_stats(void)
 		rs->kb_base = ts->kb_base;
 		rs->unit_base = ts->unit_base;
 		rs->sig_figs = ts->sig_figs;
-		rs->unified_rw_rep += ts->unified_rw_rep;
+		rs->unified_rw_rep |= ts->unified_rw_rep;
 
 		for (j = 0; j < DDIR_RWDIR_CNT; j++) {
 			if (!ts->runtime[j])
@@ -2403,7 +2581,7 @@ void __show_run_stats(void)
 	free(opt_lists);
 }
 
-void __show_running_run_stats(void)
+int __show_running_run_stats(void)
 {
 	struct thread_data *td;
 	unsigned long long *rt;
@@ -2417,9 +2595,9 @@ void __show_running_run_stats(void)
 
 	for_each_td(td, i) {
 		td->update_rusage = 1;
-		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];
+		for_each_rw_ddir(ddir) {
+			td->ts.io_bytes[ddir] = td->io_bytes[ddir];
+		}
 		td->ts.total_run_time = mtime_since(&td->epoch, &ts);
 
 		rt[i] = mtime_since(&td->start, &ts);
@@ -2454,10 +2632,10 @@ void __show_running_run_stats(void)
 
 	free(rt);
 	fio_sem_up(stat_sem);
+
+	return 0;
 }
 
-static bool status_interval_init;
-static struct timespec status_time;
 static bool status_file_disabled;
 
 #define FIO_STATUS_FILE		"fio-dump-status"
@@ -2500,16 +2678,6 @@ static int check_status_file(void)
 
 void check_for_running_stats(void)
 {
-	if (status_interval) {
-		if (!status_interval_init) {
-			fio_gettime(&status_time, NULL);
-			status_interval_init = true;
-		} else if (mtime_since_now(&status_time) >= status_interval) {
-			show_running_run_stats();
-			fio_gettime(&status_time, NULL);
-			return;
-		}
-	}
 	if (check_status_file()) {
 		show_running_run_stats();
 		return;
@@ -2650,6 +2818,14 @@ void regrow_logs(struct thread_data *td)
 	td->flags &= ~TD_F_REGROW_LOGS;
 }
 
+void regrow_agg_logs(void)
+{
+	enum fio_ddir ddir;
+
+	for (ddir = 0; ddir < DDIR_RWDIR_CNT; ddir++)
+		regrow_log(agg_io_log[ddir]);
+}
+
 static struct io_logs *get_cur_log(struct io_log *iolog)
 {
 	struct io_logs *cur_log;
@@ -2728,11 +2904,11 @@ static inline void reset_io_stat(struct io_stat *ios)
 void reset_io_stats(struct thread_data *td)
 {
 	struct thread_stat *ts = &td->ts;
-	int i, j;
+	int i, j, k;
 
 	for (i = 0; i < DDIR_RWDIR_CNT; i++) {
 		reset_io_stat(&ts->clat_high_prio_stat[i]);
-		reset_io_stat(&ts->clat_prio_stat[i]);
+		reset_io_stat(&ts->clat_low_prio_stat[i]);
 		reset_io_stat(&ts->clat_stat[i]);
 		reset_io_stat(&ts->slat_stat[i]);
 		reset_io_stat(&ts->lat_stat[i]);
@@ -2746,14 +2922,18 @@ void reset_io_stats(struct thread_data *td)
 		ts->drop_io_u[i] = 0;
 
 		for (j = 0; j < FIO_IO_U_PLAT_NR; j++) {
-			ts->io_u_plat[i][j] = 0;
 			ts->io_u_plat_high_prio[i][j] = 0;
-			ts->io_u_plat_prio[i][j] = 0;
+			ts->io_u_plat_low_prio[i][j] = 0;
 			if (!i)
 				ts->io_u_sync_plat[j] = 0;
 		}
 	}
 
+	for (i = 0; i < FIO_LAT_CNT; i++)
+		for (j = 0; j < DDIR_RWDIR_CNT; j++)
+			for (k = 0; k < FIO_IO_U_PLAT_NR; k++)
+				ts->io_u_plat[i][j][k] = 0;
+
 	ts->total_io_u[DDIR_SYNC] = 0;
 
 	for (i = 0; i < FIO_IO_U_MAP_NR; i++) {
@@ -2847,9 +3027,10 @@ static unsigned long add_log_sample(struct thread_data *td,
 			return diff;
 	}
 
-	_add_stat_to_log(iolog, elapsed, td->o.log_max != 0, priority_bit);
+	__add_stat_to_log(iolog, ddir, elapsed, td->o.log_max != 0, priority_bit);
+
+	iolog->avg_last[ddir] = elapsed - (elapsed % iolog->avg_msec);
 
-	iolog->avg_last[ddir] = elapsed - (this_window - iolog->avg_msec);
 	return iolog->avg_msec;
 }
 
@@ -2892,19 +3073,27 @@ void add_sync_clat_sample(struct thread_stat *ts, unsigned long long nsec)
 	add_stat_sample(&ts->sync_stat, nsec);
 }
 
-static void add_clat_percentile_sample(struct thread_stat *ts,
-				unsigned long long nsec, enum fio_ddir ddir, uint8_t priority_bit)
+static void add_lat_percentile_sample_noprio(struct thread_stat *ts,
+				unsigned long long nsec, enum fio_ddir ddir, enum fio_lat lat)
 {
 	unsigned int idx = plat_val_to_idx(nsec);
 	assert(idx < FIO_IO_U_PLAT_NR);
 
-	ts->io_u_plat[ddir][idx]++;
+	ts->io_u_plat[lat][ddir][idx]++;
+}
 
-	if (!priority_bit) {
-		ts->io_u_plat_prio[ddir][idx]++;
-	} else {
+static void add_lat_percentile_sample(struct thread_stat *ts,
+				unsigned long long nsec, enum fio_ddir ddir, uint8_t priority_bit,
+				enum fio_lat lat)
+{
+	unsigned int idx = plat_val_to_idx(nsec);
+
+	add_lat_percentile_sample_noprio(ts, nsec, ddir, lat);
+
+	if (!priority_bit)
+		ts->io_u_plat_low_prio[ddir][idx]++;
+	else
 		ts->io_u_plat_high_prio[ddir][idx]++;
-	}
 }
 
 void add_clat_sample(struct thread_data *td, enum fio_ddir ddir,
@@ -2921,10 +3110,11 @@ void add_clat_sample(struct thread_data *td, enum fio_ddir ddir,
 
 	add_stat_sample(&ts->clat_stat[ddir], nsec);
 
-	if (priority_bit) {
-		add_stat_sample(&ts->clat_high_prio_stat[ddir], nsec);
-	} else {
-		add_stat_sample(&ts->clat_prio_stat[ddir], nsec);
+	if (!ts->lat_percentiles) {
+		if (priority_bit)
+			add_stat_sample(&ts->clat_high_prio_stat[ddir], nsec);
+		else
+			add_stat_sample(&ts->clat_low_prio_stat[ddir], nsec);
 	}
 
 	if (td->clat_log)
@@ -2932,7 +3122,10 @@ void add_clat_sample(struct thread_data *td, enum fio_ddir ddir,
 			       offset, priority_bit);
 
 	if (ts->clat_percentiles) {
-		add_clat_percentile_sample(ts, nsec, ddir, priority_bit);
+		if (ts->lat_percentiles)
+			add_lat_percentile_sample_noprio(ts, nsec, ddir, FIO_CLAT);
+		else
+			add_lat_percentile_sample(ts, nsec, ddir, priority_bit, FIO_CLAT);
 	}
 
 	if (iolog && iolog->hist_msec) {
@@ -2955,7 +3148,7 @@ void add_clat_sample(struct thread_data *td, enum fio_ddir ddir,
 			 * located in iolog.c after printing this sample to the
 			 * log file.
 			 */
-			io_u_plat = (uint64_t *) td->ts.io_u_plat[ddir];
+			io_u_plat = (uint64_t *) td->ts.io_u_plat[FIO_CLAT][ddir];
 			dst = malloc(sizeof(struct io_u_plat_entry));
 			memcpy(&(dst->io_u_plat), io_u_plat,
 				FIO_IO_U_PLAT_NR * sizeof(uint64_t));
@@ -2978,7 +3171,7 @@ void add_clat_sample(struct thread_data *td, enum fio_ddir ddir,
 }
 
 void add_slat_sample(struct thread_data *td, enum fio_ddir ddir,
-			unsigned long usec, unsigned long long bs, uint64_t offset,
+			unsigned long long nsec, unsigned long long bs, uint64_t offset,
 			uint8_t priority_bit)
 {
 	const bool needs_lock = td_async_processing(td);
@@ -2990,12 +3183,15 @@ void add_slat_sample(struct thread_data *td, enum fio_ddir ddir,
 	if (needs_lock)
 		__td_io_u_lock(td);
 
-	add_stat_sample(&ts->slat_stat[ddir], usec);
+	add_stat_sample(&ts->slat_stat[ddir], nsec);
 
 	if (td->slat_log)
-		add_log_sample(td, td->slat_log, sample_val(usec), ddir, bs, offset,
+		add_log_sample(td, td->slat_log, sample_val(nsec), ddir, bs, offset,
 			priority_bit);
 
+	if (ts->slat_percentiles)
+		add_lat_percentile_sample_noprio(ts, nsec, ddir, FIO_SLAT);
+
 	if (needs_lock)
 		__td_io_u_unlock(td);
 }
@@ -3019,9 +3215,14 @@ void add_lat_sample(struct thread_data *td, enum fio_ddir ddir,
 		add_log_sample(td, td->lat_log, sample_val(nsec), ddir, bs,
 			       offset, priority_bit);
 
-	if (ts->lat_percentiles)
-		add_clat_percentile_sample(ts, nsec, ddir, priority_bit);
+	if (ts->lat_percentiles) {
+		add_lat_percentile_sample(ts, nsec, ddir, priority_bit, FIO_LAT);
+		if (priority_bit)
+			add_stat_sample(&ts->clat_high_prio_stat[ddir], nsec);
+		else
+			add_stat_sample(&ts->clat_low_prio_stat[ddir], nsec);
 
+	}
 	if (needs_lock)
 		__td_io_u_unlock(td);
 }
@@ -3067,7 +3268,7 @@ static int __add_samples(struct thread_data *td, struct timespec *parent_tv,
 	next_log = avg_time;
 
 	spent = mtime_since(parent_tv, t);
-	if (spent < avg_time && avg_time - spent >= LOG_MSEC_SLACK)
+	if (spent < avg_time && avg_time - spent > LOG_MSEC_SLACK)
 		return avg_time - spent;
 
 	if (needs_lock)
@@ -3106,7 +3307,7 @@ static int __add_samples(struct thread_data *td, struct timespec *parent_tv,
 		stat_io_bytes[ddir] = this_io_bytes[ddir];
 	}
 
-	timespec_add_msec(parent_tv, avg_time);
+	*parent_tv = *t;
 
 	if (needs_lock)
 		__td_io_u_unlock(td);
@@ -3160,13 +3361,16 @@ static int add_iops_samples(struct thread_data *td, struct timespec *t)
 int calc_log_samples(void)
 {
 	struct thread_data *td;
-	unsigned int next = ~0U, tmp;
+	unsigned int next = ~0U, tmp = 0, next_mod = 0, log_avg_msec_min = -1U;
 	struct timespec now;
 	int i;
+	long elapsed_time = 0;
 
 	fio_gettime(&now, NULL);
 
 	for_each_td(td, i) {
+		elapsed_time = mtime_since_now(&td->epoch);
+
 		if (!td->o.stats)
 			continue;
 		if (in_ramp_time(td) ||
@@ -3177,17 +3381,34 @@ int calc_log_samples(void)
 		if (!td->bw_log ||
 			(td->bw_log && !per_unit_log(td->bw_log))) {
 			tmp = add_bw_samples(td, &now);
-			if (tmp < next)
-				next = tmp;
+
+			if (td->bw_log)
+				log_avg_msec_min = min(log_avg_msec_min, (unsigned int)td->bw_log->avg_msec);
 		}
 		if (!td->iops_log ||
 			(td->iops_log && !per_unit_log(td->iops_log))) {
 			tmp = add_iops_samples(td, &now);
-			if (tmp < next)
-				next = tmp;
+
+			if (td->iops_log)
+				log_avg_msec_min = min(log_avg_msec_min, (unsigned int)td->iops_log->avg_msec);
 		}
+
+		if (tmp < next)
+			next = tmp;
 	}
 
+	/* if log_avg_msec_min has not been changed, set it to 0 */
+	if (log_avg_msec_min == -1U)
+		log_avg_msec_min = 0;
+
+	if (log_avg_msec_min == 0)
+		next_mod = elapsed_time;
+	else
+		next_mod = elapsed_time % log_avg_msec_min;
+
+	/* correction to keep the time on the log avg msec boundary */
+	next = min(next, (log_avg_msec_min - next_mod));
+
 	return next == ~0U ? 0 : next;
 }
 
@@ -3224,3 +3445,4 @@ uint32_t *io_u_block_info(struct thread_data *td, struct io_u *io_u)
 	assert(idx < td->ts.nr_block_infos);
 	return info;
 }
+