X-Git-Url: https://git.kernel.dk/?p=fio.git;a=blobdiff_plain;f=stat.c;h=9d93dcd17040d6326a1c8ea9c0226d7dfd363bb6;hp=cc171a4d52ff69dab7df15037a7b89996e05ecaa;hb=HEAD;hpb=c619c0fdb28fbe043d7a7f75bba2ea82b4eca298 diff --git a/stat.c b/stat.c index cc171a4d..b98e8b27 100644 --- a/stat.c +++ b/stat.c @@ -1,10 +1,8 @@ #include #include +#include #include -#include #include -#include -#include #include #include "fio.h" @@ -17,10 +15,16 @@ #include "lib/output_buffer.h" #include "helper_thread.h" #include "smalloc.h" +#include "zbd.h" +#include "oslib/asprintf.h" -#define LOG_MSEC_SLACK 10 +#ifdef WIN32 +#define LOG_MSEC_SLACK 2 +#else +#define LOG_MSEC_SLACK 1 +#endif -struct fio_mutex *stat_mutex; +struct fio_sem *stat_sem; void clear_rusage_stat(struct thread_data *td) { @@ -135,13 +139,12 @@ static int double_cmp(const void *a, const void *b) return cmp; } -unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long long nr, +unsigned int calc_clat_percentiles(uint64_t *io_u_plat, unsigned long long nr, fio_fp64_t *plist, unsigned long long **output, unsigned long long *maxv, unsigned long long *minv) { unsigned long long sum = 0; unsigned int len, i, j = 0; - unsigned int oval_len = 0; unsigned long long *ovals = NULL; bool is_last; @@ -161,7 +164,11 @@ unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long long n * 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(plist, len, sizeof(plist[0]), double_cmp); + + ovals = malloc(len * sizeof(*ovals)); + if (!ovals) + return 0; /* * Calculate bucket values, note down max and min values @@ -169,14 +176,9 @@ unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long long n is_last = false; for (i = 0; i < FIO_IO_U_PLAT_NR && !is_last; i++) { sum += io_u_plat[i]; - while (sum >= (plist[j].u.f / 100.0 * nr)) { + while (sum >= ((long double) plist[j].u.f / 100.0 * nr)) { assert(plist[j].u.f <= 100.0); - if (j == oval_len) { - oval_len += 100; - ovals = realloc(ovals, oval_len * sizeof(*ovals)); - } - ovals[j] = plat_idx_to_val(i); if (ovals[j] < *minv) *minv = ovals[j]; @@ -191,6 +193,9 @@ unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long long n } } + if (!is_last) + log_err("fio: error calculating latency percentiles\n"); + *output = ovals; return len; } @@ -198,21 +203,20 @@ unsigned int calc_clat_percentiles(unsigned int *io_u_plat, unsigned long long n /* * Find and display the p-th percentile of clat */ -static void show_clat_percentiles(unsigned int *io_u_plat, unsigned long long nr, +static void show_clat_percentiles(uint64_t *io_u_plat, unsigned long long nr, fio_fp64_t *plist, unsigned int precision, - bool is_clat, struct buf_output *out) + const char *pre, struct buf_output *out) { unsigned int divisor, len, i, j = 0; unsigned long long minv, maxv; unsigned long long *ovals; int per_line, scale_down, time_width; - const char *pre = is_clat ? "clat" : " lat"; bool is_last; char fmt[32]; len = calc_clat_percentiles(io_u_plat, nr, plist, &ovals, &maxv, &minv); - if (!len) - goto out; + if (!len || !ovals) + return; /* * We default to nsecs, but if the value range is such that we @@ -259,9 +263,19 @@ static void show_clat_percentiles(unsigned int *io_u_plat, unsigned long long nr log_buf(out, "\n"); } -out: - if (ovals) - free(ovals); + free(ovals); +} + +static int get_nr_prios_with_samples(struct thread_stat *ts, enum fio_ddir ddir) +{ + int i, nr_prios_with_samples = 0; + + for (i = 0; i < ts->nr_clat_prio[ddir]; i++) { + if (ts->clat_prio[ddir][i].clat_stat.samples) + nr_prios_with_samples++; + } + + return nr_prios_with_samples; } bool calc_lat(struct io_stat *is, unsigned long long *min, @@ -284,6 +298,47 @@ 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; + uint64_t min_run = -1, max_run = 0; + const int i2p = is_power_of_2(rs->kb_base); + int i; + + 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; @@ -308,7 +363,7 @@ 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]); @@ -322,9 +377,13 @@ void show_group_stats(struct group_run_stats *rs, struct buf_output *out) free(minalt); free(maxalt); } + + /* Need to aggregate statistics to show mixed values */ + if (rs->unified_rw_rep == UNIFIED_BOTH) + show_mixed_group_stats(rs, out); } -void stat_calc_dist(unsigned int *map, unsigned long total, double *io_u_dist) +void stat_calc_dist(uint64_t *map, unsigned long total, double *io_u_dist) { int i; @@ -343,7 +402,7 @@ 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) + uint64_t *src, int nr) { unsigned long total = ddir_rw_sum(ts->total_io_u); int i; @@ -366,7 +425,7 @@ static void stat_calc_lat(struct thread_stat *ts, double *dst, * To keep the terse format unaltered, add all of the ns latency * buckets to the first us latency bucket */ -void stat_calc_lat_nu(struct thread_stat *ts, double *io_u_lat_u) +static void stat_calc_lat_nu(struct thread_stat *ts, double *io_u_lat_u) { unsigned long ntotal = 0, total = ddir_rw_sum(ts->total_io_u); int i; @@ -416,15 +475,70 @@ static void display_lat(const char *name, unsigned long long min, free(maxp); } +static struct thread_stat *gen_mixed_ddir_stats_from_ts(struct thread_stat *ts) +{ + struct thread_stat *ts_lcl; + + /* + * Handle aggregation of Reads (ddir = 0), Writes (ddir = 1), and + * Trims (ddir = 2) + */ + ts_lcl = malloc(sizeof(struct thread_stat)); + if (!ts_lcl) { + log_err("fio: failed to allocate local thread stat\n"); + return NULL; + } + + init_thread_stat(ts_lcl); + + /* calculate mixed stats */ + ts_lcl->unified_rw_rep = UNIFIED_MIXED; + 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); + + return ts_lcl; +} + +static double convert_agg_kbytes_percent(struct group_run_stats *rs, + enum fio_ddir 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_ddir_status(struct group_run_stats *rs, struct thread_stat *ts, - int ddir, struct buf_output *out) + enum fio_ddir ddir, struct buf_output *out) { - const char *str[] = { " read", "write", " trim" }; unsigned long runt; unsigned long long min, max, bw, iops; double mean, dev; - char *io_p, *bw_p, *bw_p_alt, *iops_p; - int i2p; + char *io_p, *bw_p, *bw_p_alt, *iops_p, *post_st = NULL; + int i2p, i; + const char *clat_type = ts->lat_percentiles ? "lat" : "clat"; + + if (ddir_sync(ddir)) { + if (calc_lat(&ts->sync_stat, &min, &max, &mean, &dev)) { + log_buf(out, " %s:\n", "fsync/fdatasync/sync_file_range"); + display_lat(io_ddir_name(ddir), min, max, mean, dev, out); + show_clat_percentiles(ts->io_u_sync_plat, + ts->sync_stat.samples, + ts->percentile_list, + ts->percentile_precision, + io_ddir_name(ddir), out); + } + return; + } assert(ddir_rw(ddir)); @@ -441,12 +555,26 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts, iops = (1000 * (uint64_t)ts->total_io_u[ddir]) / runt; iops_p = num2str(iops, ts->sig_figs, 1, 0, N2S_NONE); + if (ddir == DDIR_WRITE || ddir == DDIR_TRIM) + post_st = zbd_write_status(ts); + else if (ddir == DDIR_READ && ts->cachehit && ts->cachemiss) { + uint64_t total; + double hit; + + total = ts->cachehit + ts->cachemiss; + hit = (double) ts->cachehit / (double) total; + hit *= 100.0; + if (asprintf(&post_st, "; Cachehit=%0.2f%%", hit) < 0) + post_st = NULL; + } - log_buf(out, " %s: IOPS=%s, BW=%s (%s)(%s/%llumsec)\n", - rs->unified_rw_rep ? "mixed" : str[ddir], + log_buf(out, " %s: IOPS=%s, BW=%s (%s)(%s/%llumsec)%s\n", + (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]); + (unsigned long long) ts->runtime[ddir], + post_st ? : ""); + free(post_st); free(io_p); free(bw_p); free(bw_p_alt); @@ -459,20 +587,74 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts, if (calc_lat(&ts->lat_stat[ddir], &min, &max, &mean, &dev)) display_lat(" lat", min, max, mean, dev, out); + /* Only print per prio stats if there are >= 2 prios with samples */ + if (get_nr_prios_with_samples(ts, ddir) >= 2) { + for (i = 0; i < ts->nr_clat_prio[ddir]; i++) { + char buf[64]; + + if (!calc_lat(&ts->clat_prio[ddir][i].clat_stat, &min, + &max, &mean, &dev)) + continue; + + snprintf(buf, sizeof(buf), + "%s prio %u/%u/%u", + clat_type, + ioprio_class(ts->clat_prio[ddir][i].ioprio), + ioprio(ts->clat_prio[ddir][i].ioprio), + ioprio_hint(ts->clat_prio[ddir][i].ioprio)); + display_lat(buf, 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) { + char prio_name[64]; uint64_t samples; - if (ts->clat_percentiles) - samples = ts->clat_stat[ddir].samples; - else + if (ts->lat_percentiles) samples = ts->lat_stat[ddir].samples; + else + samples = ts->clat_stat[ddir].samples; - show_clat_percentiles(ts->io_u_plat[ddir], - samples, - ts->percentile_list, - ts->percentile_precision, - ts->clat_percentiles, out); + /* Only print per prio stats if there are >= 2 prios with samples */ + if (get_nr_prios_with_samples(ts, ddir) >= 2) { + for (i = 0; i < ts->nr_clat_prio[ddir]; i++) { + uint64_t prio_samples = + ts->clat_prio[ddir][i].clat_stat.samples; + + if (!prio_samples) + continue; + + snprintf(prio_name, sizeof(prio_name), + "%s prio %u/%u/%u (%.2f%% of IOs)", + clat_type, + ioprio_class(ts->clat_prio[ddir][i].ioprio), + ioprio(ts->clat_prio[ddir][i].ioprio), + ioprio_hint(ts->clat_prio[ddir][i].ioprio), + 100. * (double) prio_samples / (double) samples); + show_clat_percentiles(ts->clat_prio[ddir][i].io_u_plat, + prio_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; @@ -486,11 +668,7 @@ 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; @@ -519,6 +697,19 @@ static void show_ddir_status(struct group_run_stats *rs, struct thread_stat *ts, } } +static void show_mixed_ddir_status(struct group_run_stats *rs, + struct thread_stat *ts, + struct buf_output *out) +{ + struct thread_stat *ts_lcl = gen_mixed_ddir_stats_from_ts(ts); + + if (ts_lcl) + show_ddir_status(rs, ts_lcl, DDIR_READ, out); + + free_clat_prio_stats(ts_lcl); + free(ts_lcl); +} + static bool show_lat(double *io_u_lat, int nr, const char **ranges, const char *msg, struct buf_output *out) { @@ -610,8 +801,8 @@ static int block_state_category(int block_state) static int compare_block_infos(const void *bs1, const void *bs2) { - uint32_t block1 = *(uint32_t *)bs1; - uint32_t block2 = *(uint32_t *)bs2; + uint64_t block1 = *(uint64_t *)bs1; + uint64_t block2 = *(uint64_t *)bs2; int state1 = BLOCK_INFO_STATE(block1); int state2 = BLOCK_INFO_STATE(block2); int bscat1 = block_state_category(state1); @@ -659,9 +850,8 @@ static int calc_block_percentiles(int nr_block_infos, uint32_t *block_infos, * 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(plist, len, sizeof(plist[0]), double_cmp); - nr_uninit = 0; /* Start only after the uninit entries end */ for (nr_uninit = 0; nr_uninit < nr_block_infos @@ -762,6 +952,228 @@ static void show_ss_normal(struct thread_stat *ts, struct buf_output *out) free(p2); } +static void show_agg_stats(struct disk_util_agg *agg, int terse, + struct buf_output *out) +{ + if (!agg->slavecount) + return; + + if (!terse) { + log_buf(out, ", aggrios=%llu/%llu, aggsectors=%llu/%llu, " + "aggrmerge=%llu/%llu, aggrticks=%llu/%llu, " + "aggrin_queue=%llu, aggrutil=%3.2f%%", + (unsigned long long) agg->ios[0] / agg->slavecount, + (unsigned long long) agg->ios[1] / agg->slavecount, + (unsigned long long) agg->sectors[0] / agg->slavecount, + (unsigned long long) agg->sectors[1] / agg->slavecount, + (unsigned long long) agg->merges[0] / agg->slavecount, + (unsigned long long) agg->merges[1] / agg->slavecount, + (unsigned long long) agg->ticks[0] / agg->slavecount, + (unsigned long long) agg->ticks[1] / agg->slavecount, + (unsigned long long) agg->time_in_queue / agg->slavecount, + agg->max_util.u.f); + } else { + log_buf(out, ";slaves;%llu;%llu;%llu;%llu;%llu;%llu;%llu;%3.2f%%", + (unsigned long long) agg->ios[0] / agg->slavecount, + (unsigned long long) agg->ios[1] / agg->slavecount, + (unsigned long long) agg->merges[0] / agg->slavecount, + (unsigned long long) agg->merges[1] / agg->slavecount, + (unsigned long long) agg->ticks[0] / agg->slavecount, + (unsigned long long) agg->ticks[1] / agg->slavecount, + (unsigned long long) agg->time_in_queue / agg->slavecount, + agg->max_util.u.f); + } +} + +static void aggregate_slaves_stats(struct disk_util *masterdu) +{ + struct disk_util_agg *agg = &masterdu->agg; + struct disk_util_stat *dus; + struct flist_head *entry; + struct disk_util *slavedu; + double util; + + flist_for_each(entry, &masterdu->slaves) { + slavedu = flist_entry(entry, struct disk_util, slavelist); + dus = &slavedu->dus; + agg->ios[0] += dus->s.ios[0]; + agg->ios[1] += dus->s.ios[1]; + agg->merges[0] += dus->s.merges[0]; + agg->merges[1] += dus->s.merges[1]; + agg->sectors[0] += dus->s.sectors[0]; + agg->sectors[1] += dus->s.sectors[1]; + agg->ticks[0] += dus->s.ticks[0]; + agg->ticks[1] += dus->s.ticks[1]; + agg->time_in_queue += dus->s.time_in_queue; + agg->slavecount++; + + util = (double) (100 * dus->s.io_ticks / (double) slavedu->dus.s.msec); + /* System utilization is the utilization of the + * component with the highest utilization. + */ + if (util > agg->max_util.u.f) + agg->max_util.u.f = util; + + } + + if (agg->max_util.u.f > 100.0) + agg->max_util.u.f = 100.0; +} + +void print_disk_util(struct disk_util_stat *dus, struct disk_util_agg *agg, + int terse, struct buf_output *out) +{ + double util = 0; + + if (dus->s.msec) + util = (double) 100 * dus->s.io_ticks / (double) dus->s.msec; + if (util > 100.0) + util = 100.0; + + if (!terse) { + if (agg->slavecount) + log_buf(out, " "); + + log_buf(out, " %s: ios=%llu/%llu, sectors=%llu/%llu, " + "merge=%llu/%llu, ticks=%llu/%llu, in_queue=%llu, " + "util=%3.2f%%", + dus->name, + (unsigned long long) dus->s.ios[0], + (unsigned long long) dus->s.ios[1], + (unsigned long long) dus->s.sectors[0], + (unsigned long long) dus->s.sectors[1], + (unsigned long long) dus->s.merges[0], + (unsigned long long) dus->s.merges[1], + (unsigned long long) dus->s.ticks[0], + (unsigned long long) dus->s.ticks[1], + (unsigned long long) dus->s.time_in_queue, + util); + } else { + log_buf(out, ";%s;%llu;%llu;%llu;%llu;%llu;%llu;%llu;%3.2f%%", + dus->name, + (unsigned long long) dus->s.ios[0], + (unsigned long long) dus->s.ios[1], + (unsigned long long) dus->s.merges[0], + (unsigned long long) dus->s.merges[1], + (unsigned long long) dus->s.ticks[0], + (unsigned long long) dus->s.ticks[1], + (unsigned long long) dus->s.time_in_queue, + util); + } + + /* + * If the device has slaves, aggregate the stats for + * those slave devices also. + */ + show_agg_stats(agg, terse, out); + + if (!terse) + log_buf(out, "\n"); +} + +void json_array_add_disk_util(struct disk_util_stat *dus, + struct disk_util_agg *agg, struct json_array *array) +{ + struct json_object *obj; + double util = 0; + + if (dus->s.msec) + util = (double) 100 * dus->s.io_ticks / (double) dus->s.msec; + if (util > 100.0) + util = 100.0; + + obj = json_create_object(); + json_array_add_value_object(array, obj); + + 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_sectors", dus->s.sectors[0]); + json_object_add_value_int(obj, "write_sectors", dus->s.sectors[1]); + json_object_add_value_int(obj, "read_merges", dus->s.merges[0]); + json_object_add_value_int(obj, "write_merges", dus->s.merges[1]); + json_object_add_value_int(obj, "read_ticks", dus->s.ticks[0]); + json_object_add_value_int(obj, "write_ticks", dus->s.ticks[1]); + json_object_add_value_int(obj, "in_queue", dus->s.time_in_queue); + json_object_add_value_float(obj, "util", util); + + /* + * If the device has slaves, aggregate the stats for + * those slave devices also. + */ + if (!agg->slavecount) + return; + json_object_add_value_int(obj, "aggr_read_ios", + agg->ios[0] / agg->slavecount); + json_object_add_value_int(obj, "aggr_write_ios", + agg->ios[1] / agg->slavecount); + json_object_add_value_int(obj, "aggr_read_sectors", + agg->sectors[0] / agg->slavecount); + json_object_add_value_int(obj, "aggr_write_sectors", + agg->sectors[1] / agg->slavecount); + json_object_add_value_int(obj, "aggr_read_merges", + agg->merges[0] / agg->slavecount); + json_object_add_value_int(obj, "aggr_write_merge", + agg->merges[1] / agg->slavecount); + json_object_add_value_int(obj, "aggr_read_ticks", + agg->ticks[0] / agg->slavecount); + json_object_add_value_int(obj, "aggr_write_ticks", + agg->ticks[1] / agg->slavecount); + json_object_add_value_int(obj, "aggr_in_queue", + agg->time_in_queue / agg->slavecount); + json_object_add_value_float(obj, "aggr_util", agg->max_util.u.f); +} + +static void json_object_add_disk_utils(struct json_object *obj, + struct flist_head *head) +{ + struct json_array *array = json_create_array(); + struct flist_head *entry; + struct disk_util *du; + + json_object_add_value_array(obj, "disk_util", array); + + flist_for_each(entry, head) { + du = flist_entry(entry, struct disk_util, list); + + aggregate_slaves_stats(du); + json_array_add_disk_util(&du->dus, &du->agg, array); + } +} + +void show_disk_util(int terse, struct json_object *parent, + struct buf_output *out) +{ + struct flist_head *entry; + struct disk_util *du; + bool do_json; + + if (!is_running_backend()) + return; + + if (flist_empty(&disk_list)) + return; + + if ((output_format & FIO_OUTPUT_JSON) && parent) + do_json = true; + else + do_json = false; + + if (!terse && !do_json) + log_buf(out, "\nDisk stats (read/write):\n"); + + if (do_json) { + json_object_add_disk_utils(parent, &disk_list); + } else if (output_format & ~(FIO_OUTPUT_JSON | FIO_OUTPUT_JSON_PLUS)) { + flist_for_each(entry, &disk_list) { + du = flist_entry(entry, struct disk_util, list); + + aggregate_slaves_stats(du); + print_disk_util(&du->dus, &du->agg, terse, out); + } + } +} + static void show_thread_status_normal(struct thread_stat *ts, struct group_run_stats *rs, struct buf_output *out) @@ -774,7 +1186,7 @@ static void show_thread_status_normal(struct thread_stat *ts, if (!ddir_rw_sum(ts->io_bytes) && !ddir_rw_sum(ts->total_io_u)) return; - + memset(time_buf, 0, sizeof(time_buf)); time(&time_p); @@ -794,15 +1206,19 @@ 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); + if (ts->sync_stat.samples) + show_ddir_status(rs, ts, DDIR_SYNC, out); + runtime = ts->total_run_time; if (runtime) { double runt = (double) runtime; @@ -839,12 +1255,13 @@ static void show_thread_status_normal(struct thread_stat *ts, io_u_dist[1], io_u_dist[2], io_u_dist[3], io_u_dist[4], io_u_dist[5], io_u_dist[6]); - log_buf(out, " issued rwt: total=%llu,%llu,%llu," - " short=%llu,%llu,%llu," - " dropped=%llu,%llu,%llu\n", + log_buf(out, " issued rwts: total=%llu,%llu,%llu,%llu" + " short=%llu,%llu,%llu,0" + " dropped=%llu,%llu,%llu,0\n", (unsigned long long) ts->total_io_u[0], (unsigned long long) ts->total_io_u[1], (unsigned long long) ts->total_io_u[2], + (unsigned long long) ts->total_io_u[3], (unsigned long long) ts->short_io_u[0], (unsigned long long) ts->short_io_u[1], (unsigned long long) ts->short_io_u[2], @@ -874,8 +1291,9 @@ static void show_thread_status_normal(struct thread_stat *ts, } static void show_ddir_status_terse(struct thread_stat *ts, - struct group_run_stats *rs, int ddir, - int ver, struct buf_output *out) + struct group_run_stats *rs, + enum fio_ddir ddir, int ver, + struct buf_output *out) { unsigned long long min, max, minv, maxv, bw, iops; unsigned long long *ovals = NULL; @@ -907,13 +1325,19 @@ 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) { @@ -928,8 +1352,7 @@ 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 (ovals) - free(ovals); + free(ovals); bw_stat = calc_lat(&ts->bw_stat[ddir], &min, &max, &mean, &dev); if (bw_stat) { @@ -942,8 +1365,9 @@ static void show_ddir_status_terse(struct thread_stat *ts, } log_buf(out, ";%llu;%llu;%f%%;%f;%f", min, max, p_of_agg, mean, dev); - } else + } else { log_buf(out, ";%llu;%llu;%f%%;%f;%f", 0ULL, 0ULL, 0.0, 0.0, 0.0); + } if (ver == 5) { if (bw_stat) @@ -959,130 +1383,177 @@ 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 = gen_mixed_ddir_stats_from_ts(ts); + + if (ts_lcl) + show_ddir_status_terse(ts_lcl, rs, DDIR_READ, ver, out); + + free_clat_prio_stats(ts_lcl); + 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) + struct group_run_stats *rs, enum fio_ddir 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; - const char *ddirname[] = {"read", "write", "trim"}; - struct json_object *dir_object, *tmp_object, *percentile_object, *clat_bins_object = NULL; - char buf[120]; + struct json_object *dir_object, *tmp_object; double p_of_agg = 100.0; - assert(ddir_rw(ddir)); + 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" : ddirname[ddir], dir_object); - - bw_bytes = 0; - bw = 0; - iops = 0.0; - if (ts->runtime[ddir]) { - uint64_t runt = ts->runtime[ddir]; + (ts->unified_rw_rep == UNIFIED_MIXED) ? "mixed" : io_ddir_name(ddir), dir_object); + + if (ddir_rw(ddir)) { + bw_bytes = 0; + bw = 0; + iops = 0.0; + if (ts->runtime[ddir]) { + uint64_t runt = ts->runtime[ddir]; + + bw_bytes = ((1000 * ts->io_bytes[ddir]) / runt); /* Bytes/s */ + bw = bw_bytes / 1024; /* KiB/s */ + iops = (1000.0 * (uint64_t) ts->total_io_u[ddir]) / runt; + } - bw_bytes = ((1000 * ts->io_bytes[ddir]) / runt); /* Bytes/s */ - bw = bw_bytes / 1024; /* KiB/s */ - iops = (1000.0 * (uint64_t) ts->total_io_u[ddir]) / runt; + json_object_add_value_int(dir_object, "io_bytes", ts->io_bytes[ddir]); + json_object_add_value_int(dir_object, "io_kbytes", ts->io_bytes[ddir] >> 10); + json_object_add_value_int(dir_object, "bw_bytes", bw_bytes); + json_object_add_value_int(dir_object, "bw", bw); + json_object_add_value_float(dir_object, "iops", iops); + json_object_add_value_int(dir_object, "runtime", ts->runtime[ddir]); + json_object_add_value_int(dir_object, "total_ios", ts->total_io_u[ddir]); + 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]); + + 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); + + 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); + + 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]); + 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); } - json_object_add_value_int(dir_object, "io_bytes", ts->io_bytes[ddir]); - json_object_add_value_int(dir_object, "io_kbytes", ts->io_bytes[ddir] >> 10); - json_object_add_value_int(dir_object, "bw_bytes", bw_bytes); - json_object_add_value_int(dir_object, "bw", bw); - json_object_add_value_float(dir_object, "iops", iops); - json_object_add_value_int(dir_object, "runtime", ts->runtime[ddir]); - json_object_add_value_int(dir_object, "total_ios", ts->total_io_u[ddir]); - 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 (!ddir_rw(ddir)) + return; - 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_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); + /* Only include per prio stats if there are >= 2 prios with samples */ + if (get_nr_prios_with_samples(ts, ddir) >= 2) { + struct json_array *array = json_create_array(); + const char *obj_name; + int i; - 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_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) + obj_name = "lat_ns"; + else + obj_name = "clat_ns"; - if (ts->clat_percentiles || ts->lat_percentiles) { - len = calc_clat_percentiles(ts->io_u_plat[ddir], - ts->clat_stat[ddir].samples, - ts->percentile_list, &ovals, &maxv, - &minv); - } else - len = 0; + json_object_add_value_array(dir_object, "prios", array); - 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]); - } + for (i = 0; i < ts->nr_clat_prio[ddir]; i++) { + struct json_object *obj; - if (output_format & FIO_OUTPUT_JSON_PLUS) { - clat_bins_object = json_create_object(); - if (ts->clat_percentiles) - json_object_add_value_object(tmp_object, "bins", clat_bins_object); + if (!ts->clat_prio[ddir][i].clat_stat.samples) + continue; - for(i = 0; i < FIO_IO_U_PLAT_NR; i++) { - 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, (const char *)buf, ts->io_u_plat[ddir][i]); - } + obj = json_create_object(); + + json_object_add_value_int(obj, "prioclass", + ioprio_class(ts->clat_prio[ddir][i].ioprio)); + json_object_add_value_int(obj, "prio", + ioprio(ts->clat_prio[ddir][i].ioprio)); + json_object_add_value_int(obj, "priohint", + ioprio_hint(ts->clat_prio[ddir][i].ioprio)); + + tmp_object = add_ddir_lat_json(ts, + ts->clat_percentiles | ts->lat_percentiles, + &ts->clat_prio[ddir][i].clat_stat, + ts->clat_prio[ddir][i].io_u_plat); + json_object_add_value_object(obj, obj_name, tmp_object); + json_array_add_value_object(array, obj); } } - 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_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 (output_format & FIO_OUTPUT_JSON_PLUS && ts->lat_percentiles) - json_object_add_value_object(tmp_object, "bins", clat_bins_object); - - 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] / 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); @@ -1101,6 +1572,29 @@ static void add_ddir_status_json(struct thread_stat *ts, json_object_add_value_float(dir_object, "iops_stddev", dev); json_object_add_value_int(dir_object, "iops_samples", (&ts->iops_stat[ddir])->samples); + + if (ts->cachehit + ts->cachemiss) { + uint64_t total; + double hit; + + total = ts->cachehit + ts->cachemiss; + hit = (double) ts->cachehit / (double) total; + hit *= 100.0; + json_object_add_value_float(dir_object, "cachehit", hit); + } +} + +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 = gen_mixed_ddir_stats_from_ts(ts); + + /* add the aggregated stats to json parent */ + if (ts_lcl) + add_ddir_status_json(ts_lcl, rs, DDIR_READ, parent); + + free_clat_prio_stats(ts_lcl); + free(ts_lcl); } static void show_thread_status_terse_all(struct thread_stat *ts, @@ -1120,14 +1614,17 @@ 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; @@ -1162,24 +1659,25 @@ static void show_thread_status_terse_all(struct thread_stat *ts, log_buf(out, ";%3.2f%%", io_u_lat_m[i]); /* disk util stats, if any */ - if (ver >= 3) + if (ver >= 3 && is_running_backend()) show_disk_util(1, NULL, out); /* Additional output if continue_on_error set - default off*/ if (ts->continue_on_error) log_buf(out, ";%llu;%d", (unsigned long long) ts->total_err_count, ts->first_error); - if (ver == 2) - log_buf(out, "\n"); /* Additional output if description is set */ - if (strlen(ts->description)) + if (strlen(ts->description)) { + if (ver == 2) + log_buf(out, "\n"); log_buf(out, ";%s", ts->description); + } log_buf(out, "\n"); } static void json_add_job_opts(struct json_object *root, const char *name, - struct flist_head *opt_list, bool num_jobs) + struct flist_head *opt_list) { struct json_object *dir_object; struct flist_head *entry; @@ -1192,14 +1690,8 @@ 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 (!num_jobs && !strcmp(p->name, "numjobs")) - continue; - 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); } } @@ -1220,6 +1712,7 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, 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, "job_start", ts->job_start); json_object_add_value_int(root, "error", ts->error); /* ETA Info */ @@ -1227,14 +1720,19 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, if (je) { json_object_add_value_int(root, "eta", je->eta_sec); json_object_add_value_int(root, "elapsed", je->elapsed_sec); + free(je); } if (opt_list) - json_add_job_opts(root, "job options", opt_list, true); + json_add_job_opts(root, "job options", opt_list); 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); + 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) { @@ -1246,19 +1744,15 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, usr_cpu = 0; sys_cpu = 0; } + json_object_add_value_int(root, "job_runtime", ts->total_run_time); 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 */ + /* Calc % distribution of IO depths */ stat_calc_dist(ts->io_u_map, ddir_rw_sum(ts->total_io_u), io_u_dist); - stat_calc_lat_n(ts, io_u_lat_n); - 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*/ @@ -1271,6 +1765,44 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, json_object_add_value_float(tmp, (const char *)name, io_u_dist[i]); } + /* Calc % distribution of submit IO depths */ + stat_calc_dist(ts->io_u_submit, ts->total_submit, io_u_dist); + tmp = json_create_object(); + json_object_add_value_object(root, "iodepth_submit", tmp); + /* Only show fixed 7 I/O depth levels*/ + for (i = 0; i < 7; i++) { + char name[20]; + if (i == 0) + snprintf(name, 20, "0"); + else if (i < 6) + snprintf(name, 20, "%d", 1 << (i+1)); + else + snprintf(name, 20, ">=%d", 1 << i); + json_object_add_value_float(tmp, (const char *)name, io_u_dist[i]); + } + + /* Calc % distribution of completion IO depths */ + stat_calc_dist(ts->io_u_complete, ts->total_complete, io_u_dist); + tmp = json_create_object(); + json_object_add_value_object(root, "iodepth_complete", tmp); + /* Only show fixed 7 I/O depth levels*/ + for (i = 0; i < 7; i++) { + char name[20]; + if (i == 0) + snprintf(name, 20, "0"); + else if (i < 6) + snprintf(name, 20, "%d", 1 << (i+1)); + else + snprintf(name, 20, ">=%d", 1 << i); + json_object_add_value_float(tmp, (const char *)name, io_u_dist[i]); + } + + /* Calc % distribution of nsecond, usecond, msecond latency */ + stat_calc_dist(ts->io_u_map, ddir_rw_sum(ts->total_io_u), io_u_dist); + stat_calc_lat_n(ts, io_u_lat_n); + stat_calc_lat_u(ts, io_u_lat_u); + stat_calc_lat_m(ts, io_u_lat_m); + /* Nanosecond latency */ tmp = json_create_object(); json_object_add_value_object(root, "latency_ns", tmp); @@ -1338,7 +1870,7 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, snprintf(buf, sizeof(buf), "%f", ts->percentile_list[i].u.f); json_object_add_value_int(percentile_object, - (const char *)buf, + buf, percentiles[i]); } @@ -1356,8 +1888,9 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, if (ts->ss_dur) { struct json_object *data; struct json_array *iops, *bw; - int i, j, k; + int j, k, l; char ss_buf[64]; + int intervals = ts->ss_dur / (ss_check_interval / 1000L); snprintf(ss_buf, sizeof(ss_buf), "%s%s:%f%s", ts->ss_state & FIO_SS_IOPS ? "iops" : "bw", @@ -1391,9 +1924,9 @@ static struct json_object *show_thread_status_json(struct thread_stat *ts, if ((ts->ss_state & FIO_SS_ATTAINED) || !(ts->ss_state & FIO_SS_BUFFER_FULL)) j = ts->ss_head; else - j = ts->ss_head == 0 ? ts->ss_dur - 1 : ts->ss_head - 1; - for (i = 0; i < ts->ss_dur; i++) { - k = (j + i) % ts->ss_dur; + j = ts->ss_head == 0 ? intervals - 1 : ts->ss_head - 1; + for (l = 0; l < intervals; l++) { + k = (j + l) % intervals; json_array_add_value_int(bw, ts->ss_bw_data[k]); json_array_add_value_int(iops, ts->ss_iops_data[k]); } @@ -1433,13 +1966,10 @@ struct json_object *show_thread_status(struct thread_stat *ts, return ret; } -static void sum_stat(struct io_stat *dst, struct io_stat *src, bool first) +static void __sum_stat(struct io_stat *dst, struct io_stat *src, bool first) { double mean, S; - if (src->samples == 0) - return; - dst->min_val = min(dst->min_val, src->min_val); dst->max_val = max(dst->max_val, src->max_val); @@ -1466,9 +1996,43 @@ static void sum_stat(struct io_stat *dst, struct io_stat *src, bool first) dst->samples += src->samples; dst->mean.u.f = mean; dst->S.u.f = S; + } -void sum_group_stats(struct group_run_stats *dst, struct group_run_stats *src) +/* + * We sum two kinds of stats - one that is time based, in which case we + * apply the proper summing technique, and then one that is iops/bw + * numbers. For group_reporting, we should just add those up, not make + * them the mean of everything. + */ +static void sum_stat(struct io_stat *dst, struct io_stat *src, bool pure_sum) +{ + bool first = dst->samples == 0; + + if (src->samples == 0) + return; + + if (!pure_sum) { + __sum_stat(dst, src, first); + return; + } + + if (first) { + dst->min_val = src->min_val; + dst->max_val = src->max_val; + dst->samples = src->samples; + dst->mean.u.f = src->mean.u.f; + dst->S.u.f = src->S.u.f; + } else { + dst->min_val += src->min_val; + dst->max_val += src->max_val; + dst->samples += src->samples; + dst->mean.u.f += src->mean.u.f; + dst->S.u.f += src->S.u.f; + } +} + +void sum_group_stats(struct group_run_stats *dst, struct group_run_stats *src) { int i; @@ -1490,57 +2054,267 @@ void sum_group_stats(struct group_run_stats *dst, struct group_run_stats *src) dst->kb_base = src->kb_base; if (!dst->unit_base) dst->unit_base = src->unit_base; + if (!dst->sig_figs) + dst->sig_figs = src->sig_figs; } -void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, - bool first) +/* + * Free the clat_prio_stat arrays allocated by alloc_clat_prio_stat_ddir(). + */ +void free_clat_prio_stats(struct thread_stat *ts) { - int l, k; + enum fio_ddir ddir; + + if (!ts) + return; + + for (ddir = 0; ddir < DDIR_RWDIR_CNT; ddir++) { + sfree(ts->clat_prio[ddir]); + ts->clat_prio[ddir] = NULL; + ts->nr_clat_prio[ddir] = 0; + } +} + +/* + * Allocate a clat_prio_stat array. The array has to be allocated/freed using + * smalloc/sfree, so that it is accessible by the process/thread summing the + * thread_stats. + */ +int alloc_clat_prio_stat_ddir(struct thread_stat *ts, enum fio_ddir ddir, + int nr_prios) +{ + struct clat_prio_stat *clat_prio; + int i; + + clat_prio = scalloc(nr_prios, sizeof(*ts->clat_prio[ddir])); + if (!clat_prio) { + log_err("fio: failed to allocate ts clat data\n"); + return 1; + } + + for (i = 0; i < nr_prios; i++) + clat_prio[i].clat_stat.min_val = ULONG_MAX; + + ts->clat_prio[ddir] = clat_prio; + ts->nr_clat_prio[ddir] = nr_prios; + + return 0; +} + +static int grow_clat_prio_stat(struct thread_stat *dst, enum fio_ddir ddir) +{ + int curr_len = dst->nr_clat_prio[ddir]; + void *new_arr; + + new_arr = scalloc(curr_len + 1, sizeof(*dst->clat_prio[ddir])); + if (!new_arr) { + log_err("fio: failed to grow clat prio array\n"); + return 1; + } + + memcpy(new_arr, dst->clat_prio[ddir], + curr_len * sizeof(*dst->clat_prio[ddir])); + sfree(dst->clat_prio[ddir]); + + dst->clat_prio[ddir] = new_arr; + dst->clat_prio[ddir][curr_len].clat_stat.min_val = ULONG_MAX; + dst->nr_clat_prio[ddir]++; + + return 0; +} + +static int find_clat_prio_index(struct thread_stat *dst, enum fio_ddir ddir, + uint32_t ioprio) +{ + int i, nr_prios = dst->nr_clat_prio[ddir]; + + for (i = 0; i < nr_prios; i++) { + if (dst->clat_prio[ddir][i].ioprio == ioprio) + return i; + } + + return -1; +} + +static int alloc_or_get_clat_prio_index(struct thread_stat *dst, + enum fio_ddir ddir, uint32_t ioprio, + int *idx) +{ + int index = find_clat_prio_index(dst, ddir, ioprio); + + if (index == -1) { + index = dst->nr_clat_prio[ddir]; + + if (grow_clat_prio_stat(dst, ddir)) + return 1; + + dst->clat_prio[ddir][index].ioprio = ioprio; + } + + *idx = index; + + return 0; +} + +static int clat_prio_stats_copy(struct thread_stat *dst, struct thread_stat *src, + enum fio_ddir dst_ddir, enum fio_ddir src_ddir) +{ + size_t sz = sizeof(*src->clat_prio[src_ddir]) * + src->nr_clat_prio[src_ddir]; + + dst->clat_prio[dst_ddir] = smalloc(sz); + if (!dst->clat_prio[dst_ddir]) { + log_err("fio: failed to alloc clat prio array\n"); + return 1; + } + + memcpy(dst->clat_prio[dst_ddir], src->clat_prio[src_ddir], sz); + dst->nr_clat_prio[dst_ddir] = src->nr_clat_prio[src_ddir]; + + return 0; +} + +static int clat_prio_stat_add_samples(struct thread_stat *dst, + enum fio_ddir dst_ddir, uint32_t ioprio, + struct io_stat *io_stat, + uint64_t *io_u_plat) +{ + int i, dst_index; + + if (!io_stat->samples) + return 0; + + if (alloc_or_get_clat_prio_index(dst, dst_ddir, ioprio, &dst_index)) + return 1; + + sum_stat(&dst->clat_prio[dst_ddir][dst_index].clat_stat, io_stat, + false); + + for (i = 0; i < FIO_IO_U_PLAT_NR; i++) + dst->clat_prio[dst_ddir][dst_index].io_u_plat[i] += io_u_plat[i]; + + return 0; +} + +static int sum_clat_prio_stats_src_single_prio(struct thread_stat *dst, + struct thread_stat *src, + enum fio_ddir dst_ddir, + enum fio_ddir src_ddir) +{ + struct io_stat *io_stat; + uint64_t *io_u_plat; + + /* + * If src ts has no clat_prio_stat array, then all I/Os were submitted + * using src->ioprio. Thus, the global samples in src->clat_stat (or + * src->lat_stat) can be used as the 'per prio' samples for src->ioprio. + */ + assert(!src->clat_prio[src_ddir]); + assert(src->nr_clat_prio[src_ddir] == 0); + + if (src->lat_percentiles) { + io_u_plat = src->io_u_plat[FIO_LAT][src_ddir]; + io_stat = &src->lat_stat[src_ddir]; + } else { + io_u_plat = src->io_u_plat[FIO_CLAT][src_ddir]; + io_stat = &src->clat_stat[src_ddir]; + } + + return clat_prio_stat_add_samples(dst, dst_ddir, src->ioprio, io_stat, + io_u_plat); +} + +static int sum_clat_prio_stats_src_multi_prio(struct thread_stat *dst, + struct thread_stat *src, + enum fio_ddir dst_ddir, + enum fio_ddir src_ddir) +{ + int i; + + /* + * If src ts has a clat_prio_stat array, then there are multiple prios + * in use (i.e. src ts had cmdprio_percentage or cmdprio_bssplit set). + * The samples for the default prio will exist in the src->clat_prio + * array, just like the samples for any other prio. + */ + assert(src->clat_prio[src_ddir]); + assert(src->nr_clat_prio[src_ddir]); + + /* If the dst ts doesn't yet have a clat_prio array, simply memcpy. */ + if (!dst->clat_prio[dst_ddir]) + return clat_prio_stats_copy(dst, src, dst_ddir, src_ddir); + + /* The dst ts already has a clat_prio_array, add src stats into it. */ + for (i = 0; i < src->nr_clat_prio[src_ddir]; i++) { + struct io_stat *io_stat = &src->clat_prio[src_ddir][i].clat_stat; + uint64_t *io_u_plat = src->clat_prio[src_ddir][i].io_u_plat; + uint32_t ioprio = src->clat_prio[src_ddir][i].ioprio; + + if (clat_prio_stat_add_samples(dst, dst_ddir, ioprio, io_stat, io_u_plat)) + return 1; + } + + return 0; +} + +static int sum_clat_prio_stats(struct thread_stat *dst, struct thread_stat *src, + enum fio_ddir dst_ddir, enum fio_ddir src_ddir) +{ + if (dst->disable_prio_stat) + return 0; + + if (!src->clat_prio[src_ddir]) + return sum_clat_prio_stats_src_single_prio(dst, src, dst_ddir, + src_ddir); + + return sum_clat_prio_stats_src_multi_prio(dst, src, dst_ddir, src_ddir); +} + +void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src) +{ + int k, l, m; for (l = 0; l < DDIR_RWDIR_CNT; l++) { - if (!dst->unified_rw_rep) { - sum_stat(&dst->clat_stat[l], &src->clat_stat[l], first); - sum_stat(&dst->slat_stat[l], &src->slat_stat[l], first); - sum_stat(&dst->lat_stat[l], &src->lat_stat[l], first); - sum_stat(&dst->bw_stat[l], &src->bw_stat[l], first); - sum_stat(&dst->iops_stat[l], &src->iops_stat[l], first); + if (dst->unified_rw_rep != UNIFIED_MIXED) { + sum_stat(&dst->clat_stat[l], &src->clat_stat[l], false); + sum_stat(&dst->slat_stat[l], &src->slat_stat[l], false); + sum_stat(&dst->lat_stat[l], &src->lat_stat[l], false); + sum_stat(&dst->bw_stat[l], &src->bw_stat[l], true); + sum_stat(&dst->iops_stat[l], &src->iops_stat[l], true); + sum_clat_prio_stats(dst, src, l, 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], first); - sum_stat(&dst->slat_stat[0], &src->slat_stat[l], first); - sum_stat(&dst->lat_stat[0], &src->lat_stat[l], first); - sum_stat(&dst->bw_stat[0], &src->bw_stat[l], first); - sum_stat(&dst->iops_stat[0], &src->iops_stat[l], first); + sum_stat(&dst->clat_stat[0], &src->clat_stat[l], false); + sum_stat(&dst->slat_stat[0], &src->slat_stat[l], false); + sum_stat(&dst->lat_stat[0], &src->lat_stat[l], false); + sum_stat(&dst->bw_stat[0], &src->bw_stat[l], true); + sum_stat(&dst->iops_stat[0], &src->iops_stat[l], true); + sum_clat_prio_stats(dst, src, 0, l); dst->io_bytes[0] += src->io_bytes[l]; if (dst->runtime[0] < src->runtime[l]) dst->runtime[0] = src->runtime[l]; - - /* - * We're summing to the same destination, so override - * 'first' after the first iteration of the loop - */ - first = false; } } + sum_stat(&dst->sync_stat, &src->sync_stat, false); dst->usr_time += src->usr_time; dst->sys_time += src->sys_time; dst->ctx += src->ctx; dst->majf += src->majf; dst->minf += src->minf; - for (k = 0; k < FIO_IO_U_MAP_NR; k++) + for (k = 0; k < FIO_IO_U_MAP_NR; k++) { dst->io_u_map[k] += src->io_u_map[k]; - for (k = 0; k < FIO_IO_U_MAP_NR; k++) dst->io_u_submit[k] += src->io_u_submit[k]; - for (k = 0; k < FIO_IO_U_MAP_NR; k++) dst->io_u_complete[k] += src->io_u_complete[k]; + } + for (k = 0; k < FIO_IO_U_LAT_N_NR; k++) dst->io_u_lat_n[k] += src->io_u_lat_n[k]; for (k = 0; k < FIO_IO_U_LAT_U_NR; k++) @@ -1549,7 +2323,7 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, dst->io_u_lat_m[k] += src->io_u_lat_m[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]; @@ -1560,20 +2334,25 @@ void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, } } - for (k = 0; k < DDIR_RWDIR_CNT; k++) { - int m; + dst->total_io_u[DDIR_SYNC] += src->total_io_u[DDIR_SYNC]; - 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]; - } - } + 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]; dst->total_run_time += src->total_run_time; dst->total_submit += src->total_submit; dst->total_complete += src->total_complete; + dst->nr_zone_resets += src->nr_zone_resets; + dst->cachehit += src->cachehit; + dst->cachemiss += src->cachemiss; } void init_group_run_stat(struct group_run_stats *gs) @@ -1585,26 +2364,82 @@ void init_group_run_stat(struct group_run_stats *gs) gs->min_bw[i] = gs->min_run[i] = ~0UL; } -void init_thread_stat(struct thread_stat *ts) +void init_thread_stat_min_vals(struct thread_stat *ts) { - int j; + int i; + for (i = 0; i < DDIR_RWDIR_CNT; i++) { + ts->clat_stat[i].min_val = ULONG_MAX; + ts->slat_stat[i].min_val = ULONG_MAX; + ts->lat_stat[i].min_val = ULONG_MAX; + ts->bw_stat[i].min_val = ULONG_MAX; + ts->iops_stat[i].min_val = ULONG_MAX; + } + ts->sync_stat.min_val = ULONG_MAX; +} + +void init_thread_stat(struct thread_stat *ts) +{ memset(ts, 0, sizeof(*ts)); - 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; - ts->bw_stat[j].min_val = -1UL; - ts->iops_stat[j].min_val = -1UL; - } + init_thread_stat_min_vals(ts); ts->groupid = -1; } +static void init_per_prio_stats(struct thread_stat *threadstats, int nr_ts) +{ + struct thread_stat *ts; + int i, j, last_ts, idx; + enum fio_ddir ddir; + + j = 0; + last_ts = -1; + idx = 0; + + /* + * Loop through all tds, if a td requires per prio stats, temporarily + * store a 1 in ts->disable_prio_stat, and then do an additional + * loop at the end where we invert the ts->disable_prio_stat values. + */ + for_each_td(td) { + if (!td->o.stats) + continue; + if (idx && + (!td->o.group_reporting || + (td->o.group_reporting && last_ts != td->groupid))) { + idx = 0; + j++; + } + + last_ts = td->groupid; + ts = &threadstats[j]; + + /* idx == 0 means first td in group, or td is not in a group. */ + if (idx == 0) + ts->ioprio = td->ioprio; + else if (td->ioprio != ts->ioprio) + ts->disable_prio_stat = 1; + + for (ddir = 0; ddir < DDIR_RWDIR_CNT; ddir++) { + if (td->ts.clat_prio[ddir]) { + ts->disable_prio_stat = 1; + break; + } + } + + idx++; + } end_for_each(); + + /* Loop through all dst threadstats and fixup the values. */ + for (i = 0; i < nr_ts; i++) { + ts = &threadstats[i]; + ts->disable_prio_stat = !ts->disable_prio_stat; + } +} + void __show_run_stats(void) { struct group_run_stats *runstats, *rs; - struct thread_data *td; struct thread_stat *threadstats, *ts; int i, j, k, nr_ts, last_ts, idx; bool kb_base_warned = false; @@ -1625,7 +2460,7 @@ void __show_run_stats(void) */ nr_ts = 0; last_ts = -1; - for_each_td(td, i) { + for_each_td(td) { if (!td->o.group_reporting) { nr_ts++; continue; @@ -1637,7 +2472,7 @@ void __show_run_stats(void) last_ts = td->groupid; nr_ts++; - } + } end_for_each(); threadstats = malloc(nr_ts * sizeof(struct thread_stat)); opt_lists = malloc(nr_ts * sizeof(struct flist_head *)); @@ -1647,10 +2482,12 @@ void __show_run_stats(void) opt_lists[i] = NULL; } + init_per_prio_stats(threadstats, nr_ts); + j = 0; last_ts = -1; idx = 0; - for_each_td(td, i) { + for_each_td(td) { if (!td->o.stats) continue; if (idx && (!td->o.group_reporting || @@ -1665,21 +2502,22 @@ 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; idx++; - ts->members++; if (ts->groupid == -1) { /* * These are per-group shared already */ - strncpy(ts->name, td->o.name, FIO_JOBNAME_SIZE - 1); + snprintf(ts->name, sizeof(ts->name), "%s", td->o.name); if (td->o.description) - strncpy(ts->description, td->o.description, - FIO_JOBDESC_SIZE - 1); + snprintf(ts->description, + sizeof(ts->description), "%s", + td->o.description); else memset(ts->description, 0, FIO_JOBDESC_SIZE); @@ -1689,6 +2527,7 @@ void __show_run_stats(void) */ ts->thread_number = td->thread_number; ts->groupid = td->groupid; + ts->job_start = td->job_start; /* * first pid in group, not very useful... @@ -1716,12 +2555,12 @@ void __show_run_stats(void) if (!td->error && td->o.continue_on_error && td->first_error) { ts->error = td->first_error; - ts->verror[sizeof(ts->verror) - 1] = '\0'; - strncpy(ts->verror, td->verror, sizeof(ts->verror) - 1); + snprintf(ts->verror, sizeof(ts->verror), "%s", + td->verror); } else if (td->error) { ts->error = td->error; - ts->verror[sizeof(ts->verror) - 1] = '\0'; - strncpy(ts->verror, td->verror, sizeof(ts->verror) - 1); + snprintf(ts->verror, sizeof(ts->verror), "%s", + td->verror); } } @@ -1734,7 +2573,9 @@ void __show_run_stats(void) for (k = 0; k < ts->nr_block_infos; k++) ts->block_infos[k] = td->ts.block_infos[k]; - sum_thread_stats(ts, &td->ts, idx == 1); + sum_thread_stats(ts, &td->ts); + + ts->members++; if (td->o.ss_dur) { ts->ss_state = td->ss.state; @@ -1749,7 +2590,7 @@ void __show_run_stats(void) } else ts->ss_dur = ts->ss_state = 0; - } + } end_for_each(); for (i = 0; i < nr_ts; i++) { unsigned long long bw; @@ -1761,7 +2602,7 @@ 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]) @@ -1784,7 +2625,7 @@ void __show_run_stats(void) } for (i = 0; i < groupid + 1; i++) { - int ddir; + enum fio_ddir ddir; rs = &runstats[i]; @@ -1808,13 +2649,14 @@ void __show_run_stats(void) char time_buf[32]; struct timeval now; unsigned long long ms_since_epoch; + time_t tv_sec; gettimeofday(&now, NULL); ms_since_epoch = (unsigned long long)(now.tv_sec) * 1000 + (unsigned long long)(now.tv_usec) / 1000; - os_ctime_r((const time_t *) &now.tv_sec, time_buf, - sizeof(time_buf)); + tv_sec = now.tv_sec; + os_ctime_r(&tv_sec, time_buf, sizeof(time_buf)); if (time_buf[strlen(time_buf) - 1] == '\n') time_buf[strlen(time_buf) - 1] = '\0'; @@ -1824,7 +2666,7 @@ void __show_run_stats(void) json_object_add_value_int(root, "timestamp_ms", ms_since_epoch); json_object_add_value_string(root, "time", time_buf); global = get_global_options(); - json_add_job_opts(root, "global options", &global->opt_list, false); + json_add_job_opts(root, "global options", &global->opt_list); array = json_create_array(); json_object_add_value_array(root, "jobs", array); } @@ -1885,74 +2727,79 @@ void __show_run_stats(void) buf_output_free(out); } + fio_idle_prof_cleanup(); + log_info_flush(); free(runstats); + + /* free arrays allocated by sum_thread_stats(), if any */ + for (i = 0; i < nr_ts; i++) { + ts = &threadstats[i]; + free_clat_prio_stats(ts); + } free(threadstats); free(opt_lists); } -void show_run_stats(void) +int __show_running_run_stats(void) { - fio_mutex_down(stat_mutex); - __show_run_stats(); - fio_mutex_up(stat_mutex); -} - -void __show_running_run_stats(void) -{ - struct thread_data *td; unsigned long long *rt; struct timespec ts; - int i; - fio_mutex_down(stat_mutex); + fio_sem_down(stat_sem); rt = malloc(thread_number * sizeof(unsigned long long)); fio_gettime(&ts, NULL); - for_each_td(td, i) { + for_each_td(td) { + if (td->runstate >= TD_EXITED) + continue; + 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); + rt[__td_index] = mtime_since(&td->start, &ts); if (td_read(td) && td->ts.io_bytes[DDIR_READ]) - td->ts.runtime[DDIR_READ] += rt[i]; + td->ts.runtime[DDIR_READ] += rt[__td_index]; if (td_write(td) && td->ts.io_bytes[DDIR_WRITE]) - td->ts.runtime[DDIR_WRITE] += rt[i]; + td->ts.runtime[DDIR_WRITE] += rt[__td_index]; if (td_trim(td) && td->ts.io_bytes[DDIR_TRIM]) - td->ts.runtime[DDIR_TRIM] += rt[i]; - } + td->ts.runtime[DDIR_TRIM] += rt[__td_index]; + } end_for_each(); - for_each_td(td, i) { + for_each_td(td) { if (td->runstate >= TD_EXITED) continue; if (td->rusage_sem) { td->update_rusage = 1; - fio_mutex_down(td->rusage_sem); + fio_sem_down(td->rusage_sem); } td->update_rusage = 0; - } + } end_for_each(); __show_run_stats(); - for_each_td(td, i) { + for_each_td(td) { + if (td->runstate >= TD_EXITED) + continue; + if (td_read(td) && td->ts.io_bytes[DDIR_READ]) - td->ts.runtime[DDIR_READ] -= rt[i]; + td->ts.runtime[DDIR_READ] -= rt[__td_index]; if (td_write(td) && td->ts.io_bytes[DDIR_WRITE]) - td->ts.runtime[DDIR_WRITE] -= rt[i]; + td->ts.runtime[DDIR_WRITE] -= rt[__td_index]; if (td_trim(td) && td->ts.io_bytes[DDIR_TRIM]) - td->ts.runtime[DDIR_TRIM] -= rt[i]; - } + td->ts.runtime[DDIR_TRIM] -= rt[__td_index]; + } end_for_each(); free(rt); - fio_mutex_up(stat_mutex); + 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" @@ -1974,6 +2821,9 @@ static int check_status_file(void) } if (temp_dir == NULL) temp_dir = "/tmp"; +#ifdef __COVERITY__ + __coverity_tainted_data_sanitize__(temp_dir); +#endif snprintf(fio_status_file_path, sizeof(fio_status_file_path), "%s/%s", temp_dir, FIO_STATUS_FILE); @@ -1992,16 +2842,6 @@ 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; @@ -2027,33 +2867,42 @@ static inline void add_stat_sample(struct io_stat *is, unsigned long long data) is->samples++; } +static inline void add_stat_prio_sample(struct clat_prio_stat *clat_prio, + unsigned short clat_prio_index, + unsigned long long nsec) +{ + if (clat_prio) + add_stat_sample(&clat_prio[clat_prio_index].clat_stat, nsec); +} + /* * Return a struct io_logs, which is added to the tail of the log * list for 'iolog'. */ static struct io_logs *get_new_log(struct io_log *iolog) { - size_t new_size, new_samples; + size_t new_samples; struct io_logs *cur_log; /* * Cap the size at MAX_LOG_ENTRIES, so we don't keep doubling * forever */ - if (!iolog->cur_log_max) - new_samples = DEF_LOG_ENTRIES; - else { + if (!iolog->cur_log_max) { + if (iolog->td) + new_samples = iolog->td->o.log_entries; + else + new_samples = DEF_LOG_ENTRIES; + } else { new_samples = iolog->cur_log_max * 2; if (new_samples > MAX_LOG_ENTRIES) new_samples = MAX_LOG_ENTRIES; } - new_size = new_samples * log_entry_sz(iolog); - cur_log = smalloc(sizeof(*cur_log)); if (cur_log) { INIT_FLIST_HEAD(&cur_log->list); - cur_log->log = malloc(new_size); + cur_log->log = calloc(new_samples, log_entry_sz(iolog)); if (cur_log->log) { cur_log->nr_samples = 0; cur_log->max_samples = new_samples; @@ -2142,6 +2991,14 @@ 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; @@ -2163,18 +3020,21 @@ static struct io_logs *get_cur_log(struct io_log *iolog) * submissions, flag 'td' as needing a log regrow and we'll take * care of it on the submission side. */ - if (iolog->td->o.io_submit_mode == IO_MODE_OFFLOAD || + if ((iolog->td && iolog->td->o.io_submit_mode == IO_MODE_OFFLOAD) || !per_unit_log(iolog)) return regrow_log(iolog); - iolog->td->flags |= TD_F_REGROW_LOGS; - assert(iolog->pending->nr_samples < iolog->pending->max_samples); + if (iolog->td) + iolog->td->flags |= TD_F_REGROW_LOGS; + if (iolog->pending) + assert(iolog->pending->nr_samples < iolog->pending->max_samples); return iolog->pending; } static void __add_log_sample(struct io_log *iolog, union io_sample_data data, - enum fio_ddir ddir, unsigned int bs, - unsigned long t, uint64_t offset) + enum fio_ddir ddir, unsigned long long bs, + unsigned long t, uint64_t offset, + unsigned int priority) { struct io_logs *cur_log; @@ -2190,9 +3050,12 @@ static void __add_log_sample(struct io_log *iolog, union io_sample_data data, s = get_sample(iolog, cur_log, cur_log->nr_samples); s->data = data; - s->time = t + (iolog->td ? iolog->td->unix_epoch : 0); + s->time = t; + if (iolog->td && iolog->td->o.log_alternate_epoch) + s->time += iolog->td->alternate_epoch; io_sample_set_ddir(iolog, s, ddir); s->bs = bs; + s->priority = priority; if (iolog->log_offset) { struct io_sample_offset *so = (void *) s; @@ -2209,10 +3072,35 @@ static void __add_log_sample(struct io_log *iolog, union io_sample_data data, static inline void reset_io_stat(struct io_stat *ios) { - ios->max_val = ios->min_val = ios->samples = 0; + ios->min_val = -1ULL; + ios->max_val = ios->samples = 0; ios->mean.u.f = ios->S.u.f = 0; } +static inline void reset_io_u_plat(uint64_t *io_u_plat) +{ + int i; + + for (i = 0; i < FIO_IO_U_PLAT_NR; i++) + io_u_plat[i] = 0; +} + +static inline void reset_clat_prio_stats(struct thread_stat *ts) +{ + enum fio_ddir ddir; + int i; + + for (ddir = 0; ddir < DDIR_RWDIR_CNT; ddir++) { + if (!ts->clat_prio[ddir]) + continue; + + for (i = 0; i < ts->nr_clat_prio[ddir]; i++) { + reset_io_stat(&ts->clat_prio[ddir][i].clat_stat); + reset_io_u_plat(ts->clat_prio[ddir][i].io_u_plat); + } + } +} + void reset_io_stats(struct thread_data *td) { struct thread_stat *ts = &td->ts; @@ -2230,11 +3118,17 @@ void reset_io_stats(struct thread_data *td) ts->total_io_u[i] = 0; ts->short_io_u[i] = 0; ts->drop_io_u[i] = 0; - - for (j = 0; j < FIO_IO_U_PLAT_NR; j++) - ts->io_u_plat[i][j] = 0; } + for (i = 0; i < FIO_LAT_CNT; i++) + for (j = 0; j < DDIR_RWDIR_CNT; j++) + reset_io_u_plat(ts->io_u_plat[i][j]); + + reset_clat_prio_stats(ts); + + ts->total_io_u[DDIR_SYNC] = 0; + reset_io_u_plat(ts->io_u_sync_plat); + for (i = 0; i < FIO_IO_U_MAP_NR; i++) { ts->io_u_map[i] = 0; ts->io_u_submit[i] = 0; @@ -2250,10 +3144,12 @@ void reset_io_stats(struct thread_data *td) ts->total_submit = 0; ts->total_complete = 0; + ts->nr_zone_resets = 0; + ts->cachehit = ts->cachemiss = 0; } static void __add_stat_to_log(struct io_log *iolog, enum fio_ddir ddir, - unsigned long elapsed, bool log_max) + unsigned long elapsed, int log_max) { /* * Note an entry in the log. Use the mean from the logged samples, @@ -2263,29 +3159,37 @@ static void __add_stat_to_log(struct io_log *iolog, enum fio_ddir ddir, if (iolog->avg_window[ddir].samples) { union io_sample_data data; - if (log_max) - data.val = iolog->avg_window[ddir].max_val; - else - data.val = iolog->avg_window[ddir].mean.u.f + 0.50; + if (log_max == IO_LOG_SAMPLE_AVG) { + data.val.val0 = iolog->avg_window[ddir].mean.u.f + 0.50; + data.val.val1 = 0; + } else if (log_max == IO_LOG_SAMPLE_MAX) { + data.val.val0 = iolog->avg_window[ddir].max_val; + data.val.val1 = 0; + } else { + data.val.val0 = iolog->avg_window[ddir].mean.u.f + 0.50; + data.val.val1 = iolog->avg_window[ddir].max_val; + } - __add_log_sample(iolog, data, ddir, 0, elapsed, 0); + __add_log_sample(iolog, data, ddir, 0, elapsed, 0, 0); } reset_io_stat(&iolog->avg_window[ddir]); } static void _add_stat_to_log(struct io_log *iolog, unsigned long elapsed, - bool log_max) + int log_max) { - int ddir; + enum fio_ddir ddir; for (ddir = 0; ddir < DDIR_RWDIR_CNT; ddir++) __add_stat_to_log(iolog, ddir, elapsed, log_max); } -static long add_log_sample(struct thread_data *td, struct io_log *iolog, - union io_sample_data data, enum fio_ddir ddir, - unsigned int bs, uint64_t offset) +static unsigned long add_log_sample(struct thread_data *td, + struct io_log *iolog, + union io_sample_data data, + enum fio_ddir ddir, unsigned long long bs, + uint64_t offset, unsigned int ioprio) { unsigned long elapsed, this_window; @@ -2298,7 +3202,8 @@ static long add_log_sample(struct thread_data *td, struct io_log *iolog, * If no time averaging, just add the log sample. */ if (!iolog->avg_msec) { - __add_log_sample(iolog, data, ddir, bs, elapsed, offset); + __add_log_sample(iolog, data, ddir, bs, elapsed, offset, + ioprio); return 0; } @@ -2306,7 +3211,7 @@ static long add_log_sample(struct thread_data *td, struct io_log *iolog, * Add the sample. If the time period has passed, then * add that entry to the log and clear. */ - add_stat_sample(&iolog->avg_window[ddir], data.val); + add_stat_sample(&iolog->avg_window[ddir], data.val.val0); /* * If period hasn't passed, adding the above sample is all we @@ -2316,15 +3221,16 @@ static long add_log_sample(struct thread_data *td, struct io_log *iolog, if (elapsed < iolog->avg_last[ddir]) return iolog->avg_last[ddir] - elapsed; else if (this_window < iolog->avg_msec) { - int diff = iolog->avg_msec - this_window; + unsigned long diff = iolog->avg_msec - this_window; if (inline_log(iolog) || diff > LOG_MSEC_SLACK) return diff; } - __add_stat_to_log(iolog, ddir, elapsed, td->o.log_max != 0); + __add_stat_to_log(iolog, ddir, elapsed, td->o.log_max); + + iolog->avg_last[ddir] = elapsed - (elapsed % iolog->avg_msec); - iolog->avg_last[ddir] = elapsed - (this_window - iolog->avg_msec); return iolog->avg_msec; } @@ -2335,18 +3241,19 @@ void finalize_logs(struct thread_data *td, bool unit_logs) elapsed = mtime_since_now(&td->epoch); if (td->clat_log && unit_logs) - _add_stat_to_log(td->clat_log, elapsed, td->o.log_max != 0); + _add_stat_to_log(td->clat_log, elapsed, td->o.log_max); if (td->slat_log && unit_logs) - _add_stat_to_log(td->slat_log, elapsed, td->o.log_max != 0); + _add_stat_to_log(td->slat_log, elapsed, td->o.log_max); if (td->lat_log && unit_logs) - _add_stat_to_log(td->lat_log, elapsed, td->o.log_max != 0); + _add_stat_to_log(td->lat_log, elapsed, td->o.log_max); if (td->bw_log && (unit_logs == per_unit_log(td->bw_log))) - _add_stat_to_log(td->bw_log, elapsed, td->o.log_max != 0); + _add_stat_to_log(td->bw_log, elapsed, td->o.log_max); if (td->iops_log && (unit_logs == per_unit_log(td->iops_log))) - _add_stat_to_log(td->iops_log, elapsed, td->o.log_max != 0); + _add_stat_to_log(td->iops_log, elapsed, td->o.log_max); } -void add_agg_sample(union io_sample_data data, enum fio_ddir ddir, unsigned int bs) +void add_agg_sample(union io_sample_data data, enum fio_ddir ddir, + unsigned long long bs) { struct io_log *iolog; @@ -2354,35 +3261,83 @@ void add_agg_sample(union io_sample_data data, enum fio_ddir ddir, unsigned int return; iolog = agg_io_log[ddir]; - __add_log_sample(iolog, data, ddir, bs, mtime_since_genesis(), 0); + __add_log_sample(iolog, data, ddir, bs, mtime_since_genesis(), 0, 0); } -static void add_clat_percentile_sample(struct thread_stat *ts, - unsigned long long nsec, enum fio_ddir ddir) +void add_sync_clat_sample(struct thread_stat *ts, unsigned long long nsec) { unsigned int idx = plat_val_to_idx(nsec); assert(idx < FIO_IO_U_PLAT_NR); - ts->io_u_plat[ddir][idx]++; + ts->io_u_sync_plat[idx]++; + add_stat_sample(&ts->sync_stat, nsec); +} + +static inline void add_lat_percentile_sample(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[lat][ddir][idx]++; +} + +static inline void +add_lat_percentile_prio_sample(struct thread_stat *ts, unsigned long long nsec, + enum fio_ddir ddir, + unsigned short clat_prio_index) +{ + unsigned int idx = plat_val_to_idx(nsec); + + if (ts->clat_prio[ddir]) + ts->clat_prio[ddir][clat_prio_index].io_u_plat[idx]++; } void add_clat_sample(struct thread_data *td, enum fio_ddir ddir, - unsigned long long nsec, unsigned int bs, uint64_t offset) + unsigned long long nsec, unsigned long long bs, + uint64_t offset, unsigned int ioprio, + unsigned short clat_prio_index) { + const bool needs_lock = td_async_processing(td); unsigned long elapsed, this_window; struct thread_stat *ts = &td->ts; struct io_log *iolog = td->clat_hist_log; - td_io_u_lock(td); + if (needs_lock) + __td_io_u_lock(td); add_stat_sample(&ts->clat_stat[ddir], nsec); + /* + * When lat_percentiles=1 (default 0), the reported per priority + * percentiles and stats are used for describing total latency values, + * even though the variable names themselves start with clat_. + * + * Because of the above definition, add a prio stat sample only when + * lat_percentiles=0. add_lat_sample() will add the prio stat sample + * when lat_percentiles=1. + */ + if (!ts->lat_percentiles) + add_stat_prio_sample(ts->clat_prio[ddir], clat_prio_index, + nsec); + if (td->clat_log) add_log_sample(td, td->clat_log, sample_val(nsec), ddir, bs, - offset); + offset, ioprio); - if (ts->clat_percentiles) - add_clat_percentile_sample(ts, nsec, ddir); + if (ts->clat_percentiles) { + /* + * Because of the above definition, add a prio lat percentile + * sample only when lat_percentiles=0. add_lat_sample() will add + * the prio lat percentile sample when lat_percentiles=1. + */ + add_lat_percentile_sample(ts, nsec, ddir, FIO_CLAT); + if (!ts->lat_percentiles) + add_lat_percentile_prio_sample(ts, nsec, ddir, + clat_prio_index); + } if (iolog && iolog->hist_msec) { struct io_hist *hw = &iolog->hist_window[ddir]; @@ -2392,9 +3347,9 @@ void add_clat_sample(struct thread_data *td, enum fio_ddir ddir, if (!hw->hist_last) hw->hist_last = elapsed; this_window = elapsed - hw->hist_last; - + if (this_window >= iolog->hist_msec) { - unsigned int *io_u_plat; + uint64_t *io_u_plat; struct io_u_plat_entry *dst; /* @@ -2404,13 +3359,13 @@ 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 = (unsigned int *) 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(unsigned int)); + FIO_IO_U_PLAT_NR * sizeof(uint64_t)); flist_add(&dst->list, &hw->list); __add_log_sample(iolog, sample_plat(dst), ddir, bs, - elapsed, offset); + elapsed, offset, ioprio); /* * Update the last time we recorded as being now, minus @@ -2422,52 +3377,80 @@ void add_clat_sample(struct thread_data *td, enum fio_ddir ddir, } } - td_io_u_unlock(td); + if (needs_lock) + __td_io_u_unlock(td); } void add_slat_sample(struct thread_data *td, enum fio_ddir ddir, - unsigned long usec, unsigned int bs, uint64_t offset) + unsigned long long nsec, unsigned long long bs, + uint64_t offset, unsigned int ioprio) { + const bool needs_lock = td_async_processing(td); struct thread_stat *ts = &td->ts; if (!ddir_rw(ddir)) return; - td_io_u_lock(td); + 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, ioprio); - td_io_u_unlock(td); + if (ts->slat_percentiles) + add_lat_percentile_sample(ts, nsec, ddir, FIO_SLAT); + + if (needs_lock) + __td_io_u_unlock(td); } void add_lat_sample(struct thread_data *td, enum fio_ddir ddir, - unsigned long long nsec, unsigned int bs, uint64_t offset) + unsigned long long nsec, unsigned long long bs, + uint64_t offset, unsigned int ioprio, + unsigned short clat_prio_index) { + const bool needs_lock = td_async_processing(td); struct thread_stat *ts = &td->ts; if (!ddir_rw(ddir)) return; - td_io_u_lock(td); + if (needs_lock) + __td_io_u_lock(td); add_stat_sample(&ts->lat_stat[ddir], nsec); if (td->lat_log) add_log_sample(td, td->lat_log, sample_val(nsec), ddir, bs, - offset); - - if (ts->lat_percentiles) - add_clat_percentile_sample(ts, nsec, ddir); + offset, ioprio); - td_io_u_unlock(td); + /* + * When lat_percentiles=1 (default 0), the reported per priority + * percentiles and stats are used for describing total latency values, + * even though the variable names themselves start with clat_. + * + * Because of the above definition, add a prio stat and prio lat + * percentile sample only when lat_percentiles=1. add_clat_sample() will + * add the prio stat and prio lat percentile sample when + * lat_percentiles=0. + */ + if (ts->lat_percentiles) { + add_lat_percentile_sample(ts, nsec, ddir, FIO_LAT); + add_lat_percentile_prio_sample(ts, nsec, ddir, clat_prio_index); + add_stat_prio_sample(ts->clat_prio[ddir], clat_prio_index, + nsec); + } + if (needs_lock) + __td_io_u_unlock(td); } void add_bw_sample(struct thread_data *td, struct io_u *io_u, unsigned int bytes, unsigned long long spent) { + const bool needs_lock = td_async_processing(td); struct thread_stat *ts = &td->ts; unsigned long rate; @@ -2476,16 +3459,19 @@ void add_bw_sample(struct thread_data *td, struct io_u *io_u, else rate = 0; - td_io_u_lock(td); + if (needs_lock) + __td_io_u_lock(td); add_stat_sample(&ts->bw_stat[io_u->ddir], rate); if (td->bw_log) add_log_sample(td, td->bw_log, sample_val(rate), io_u->ddir, - bytes, io_u->offset); + bytes, io_u->offset, io_u->ioprio); td->stat_io_bytes[io_u->ddir] = td->this_io_bytes[io_u->ddir]; - td_io_u_unlock(td); + + if (needs_lock) + __td_io_u_unlock(td); } static int __add_samples(struct thread_data *td, struct timespec *parent_tv, @@ -2494,17 +3480,19 @@ static int __add_samples(struct thread_data *td, struct timespec *parent_tv, struct io_stat *stat, struct io_log *log, bool is_kb) { + const bool needs_lock = td_async_processing(td); unsigned long spent, rate; enum fio_ddir ddir; - unsigned int next, next_log; + unsigned long next, next_log; 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; - td_io_u_lock(td); + if (needs_lock) + __td_io_u_lock(td); /* * Compute both read and write rates for the interval. @@ -2527,21 +3515,23 @@ static int __add_samples(struct thread_data *td, struct timespec *parent_tv, add_stat_sample(&stat[ddir], rate); if (log) { - unsigned int bs = 0; + unsigned long long bs = 0; if (td->o.min_bs[ddir] == td->o.max_bs[ddir]) bs = td->o.min_bs[ddir]; - next = add_log_sample(td, log, sample_val(rate), ddir, bs, 0); + next = add_log_sample(td, log, sample_val(rate), ddir, + bs, 0, 0); next_log = min(next_log, next); } stat_io_bytes[ddir] = this_io_bytes[ddir]; } - timespec_add_msec(parent_tv, avg_time); + *parent_tv = *t; - td_io_u_unlock(td); + if (needs_lock) + __td_io_u_unlock(td); if (spent <= avg_time) next = avg_time; @@ -2561,18 +3551,22 @@ static int add_bw_samples(struct thread_data *td, struct timespec *t) void add_iops_sample(struct thread_data *td, struct io_u *io_u, unsigned int bytes) { + const bool needs_lock = td_async_processing(td); struct thread_stat *ts = &td->ts; - td_io_u_lock(td); + if (needs_lock) + __td_io_u_lock(td); add_stat_sample(&ts->iops_stat[io_u->ddir], 1); if (td->iops_log) add_log_sample(td, td->iops_log, sample_val(1), io_u->ddir, - bytes, io_u->offset); + bytes, io_u->offset, io_u->ioprio); td->stat_io_blocks[io_u->ddir] = td->this_io_blocks[io_u->ddir]; - td_io_u_unlock(td); + + if (needs_lock) + __td_io_u_unlock(td); } static int add_iops_samples(struct thread_data *td, struct timespec *t) @@ -2582,46 +3576,78 @@ static int add_iops_samples(struct thread_data *td, struct timespec *t) td->ts.iops_stat, td->iops_log, false); } +static bool td_in_logging_state(struct thread_data *td) +{ + if (in_ramp_time(td)) + return false; + + switch(td->runstate) { + case TD_RUNNING: + case TD_VERIFYING: + case TD_FINISHING: + case TD_EXITED: + return true; + default: + return false; + } +} + /* * Returns msecs to next event */ 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) { + fio_gettime(&now, NULL); + elapsed_time = mtime_since(&td->epoch, &now); - for_each_td(td, i) { if (!td->o.stats) continue; - if (in_ramp_time(td) || - !(td->runstate == TD_RUNNING || td->runstate == TD_VERIFYING)) { + if (!td_in_logging_state(td)) { next = min(td->o.iops_avg_time, td->o.bw_avg_time); continue; } 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; + } end_for_each(); + + /* 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; } void stat_init(void) { - stat_mutex = fio_mutex_init(FIO_MUTEX_UNLOCKED); + stat_sem = fio_sem_init(FIO_SEM_UNLOCKED); } void stat_exit(void) @@ -2630,8 +3656,8 @@ void stat_exit(void) * When we have the mutex, we know out-of-band access to it * have ended. */ - fio_mutex_down(stat_mutex); - fio_mutex_remove(stat_mutex); + fio_sem_down(stat_sem); + fio_sem_remove(stat_sem); } /* @@ -2652,3 +3678,4 @@ uint32_t *io_u_block_info(struct thread_data *td, struct io_u *io_u) assert(idx < td->ts.nr_block_infos); return info; } +