> if/when we up the size I can just see someone raising an issue about how using fio with 16GByte blocks at an iodepth of 8 didn't work... Agreed. But we already see folks today using bs=1MB, numjobs=512, and iodepth=1024 as a "test." Folks just need to be aware of what IO they are really trying to do. > Post the patch and we can take a look at it. Here's my partial patch. I didn't (yet) change anything in blktrace.c, and the hack to parse.c is rough. Also I have a few make warnings to resolve. Regards, Jeff diff --git a/backend.c b/backend.c index a7e9184..a1e83ff 100644 --- a/backend.c +++ b/backend.c @@ -1199,7 +1199,7 @@ static void cleanup_io_u(struct thread_data *td) static int init_io_u(struct thread_data *td) { struct io_u *io_u; - unsigned int max_bs, min_write; + unsigned long long max_bs, min_write; int cl_align, i, max_units; int data_xfer = 1, err; char *p; @@ -1234,7 +1234,7 @@ static int init_io_u(struct thread_data *td) td->orig_buffer_size += page_mask + td->o.mem_align; if (td->o.mem_type == MEM_SHMHUGE || td->o.mem_type == MEM_MMAPHUGE) { - unsigned long bs; + unsigned long long bs; bs = td->orig_buffer_size + td->o.hugepage_size - 1; td->orig_buffer_size = bs & ~(td->o.hugepage_size - 1); diff --git a/cconv.c b/cconv.c index bfd699d..9180bac 100644 --- a/cconv.c +++ b/cconv.c @@ -110,16 +110,16 @@ void convert_thread_options_to_cpu(struct thread_options *o, o->start_offset_percent = le32_to_cpu(top->start_offset_percent); for (i = 0; i < DDIR_RWDIR_CNT; i++) { - o->bs[i] = le32_to_cpu(top->bs[i]); - o->ba[i] = le32_to_cpu(top->ba[i]); - o->min_bs[i] = le32_to_cpu(top->min_bs[i]); - o->max_bs[i] = le32_to_cpu(top->max_bs[i]); + o->bs[i] = le64_to_cpu(top->bs[i]); + o->ba[i] = le64_to_cpu(top->ba[i]); + o->min_bs[i] = le64_to_cpu(top->min_bs[i]); + o->max_bs[i] = le64_to_cpu(top->max_bs[i]); o->bssplit_nr[i] = le32_to_cpu(top->bssplit_nr[i]); if (o->bssplit_nr[i]) { o->bssplit[i] = malloc(o->bssplit_nr[i] * sizeof(struct bssplit)); for (j = 0; j < o->bssplit_nr[i]; j++) { - o->bssplit[i][j].bs = le32_to_cpu(top->bssplit[i][j].bs); + o->bssplit[i][j].bs = le64_to_cpu(top->bssplit[i][j].bs); o->bssplit[i][j].perc = le32_to_cpu(top->bssplit[i][j].perc); } } @@ -203,7 +203,7 @@ void convert_thread_options_to_cpu(struct thread_options *o, o->gauss_dev.u.f = fio_uint64_to_double(le64_to_cpu(top->gauss_dev.u.i)); o->random_generator = le32_to_cpu(top->random_generator); o->hugepage_size = le32_to_cpu(top->hugepage_size); - o->rw_min_bs = le32_to_cpu(top->rw_min_bs); + o->rw_min_bs = le64_to_cpu(top->rw_min_bs); o->thinktime = le32_to_cpu(top->thinktime); o->thinktime_spin = le32_to_cpu(top->thinktime_spin); o->thinktime_blocks = le32_to_cpu(top->thinktime_blocks); @@ -410,7 +410,7 @@ void convert_thread_options_to_net(struct thread_options_pack *top, top->gauss_dev.u.i = __cpu_to_le64(fio_double_to_uint64(o->gauss_dev.u.f)); top->random_generator = cpu_to_le32(o->random_generator); top->hugepage_size = cpu_to_le32(o->hugepage_size); - top->rw_min_bs = cpu_to_le32(o->rw_min_bs); + top->rw_min_bs = cpu_to_le64(o->rw_min_bs); top->thinktime = cpu_to_le32(o->thinktime); top->thinktime_spin = cpu_to_le32(o->thinktime_spin); top->thinktime_blocks = cpu_to_le32(o->thinktime_blocks); @@ -488,10 +488,10 @@ void convert_thread_options_to_net(struct thread_options_pack *top, top->write_hist_log = cpu_to_le32(o->write_hist_log); for (i = 0; i < DDIR_RWDIR_CNT; i++) { - top->bs[i] = cpu_to_le32(o->bs[i]); - top->ba[i] = cpu_to_le32(o->ba[i]); - top->min_bs[i] = cpu_to_le32(o->min_bs[i]); - top->max_bs[i] = cpu_to_le32(o->max_bs[i]); + top->bs[i] = cpu_to_le64(o->bs[i]); + top->ba[i] = cpu_to_le64(o->ba[i]); + top->min_bs[i] = cpu_to_le64(o->min_bs[i]); + top->max_bs[i] = cpu_to_le64(o->max_bs[i]); top->bssplit_nr[i] = cpu_to_le32(o->bssplit_nr[i]); if (o->bssplit_nr[i]) { @@ -502,7 +502,7 @@ void convert_thread_options_to_net(struct thread_options_pack *top, bssplit_nr = BSSPLIT_MAX; } for (j = 0; j < bssplit_nr; j++) { - top->bssplit[i][j].bs = cpu_to_le32(o->bssplit[i][j].bs); + top->bssplit[i][j].bs = cpu_to_le64(o->bssplit[i][j].bs); top->bssplit[i][j].perc = cpu_to_le32(o->bssplit[i][j].perc); } } diff --git a/client.c b/client.c index 2a86ea9..e2525c8 100644 --- a/client.c +++ b/client.c @@ -1357,8 +1357,8 @@ static void client_flush_hist_samples(FILE *f, int hist_coarseness, void *sample entry = s->data.plat_entry; io_u_plat = entry->io_u_plat; - fprintf(f, "%lu, %u, %u, ", (unsigned long) s->time, - io_sample_ddir(s), s->bs); + fprintf(f, "%lu, %u, %llu, ", (unsigned long) s->time, + io_sample_ddir(s), (unsigned long long) s->bs); for (j = 0; j < FIO_IO_U_PLAT_NR - stride; j += stride) { fprintf(f, "%llu, ", (unsigned long long)hist_sum(j, stride, io_u_plat, NULL)); } @@ -1647,7 +1647,7 @@ static struct cmd_iolog_pdu *convert_iolog(struct fio_net_cmd *cmd, s->time = le64_to_cpu(s->time); s->data.val = le64_to_cpu(s->data.val); s->__ddir = le32_to_cpu(s->__ddir); - s->bs = le32_to_cpu(s->bs); + s->bs = le64_to_cpu(s->bs); if (ret->log_offset) { struct io_sample_offset *so = (void *) s; diff --git a/file.h b/file.h index 8fd34b1..c0a547e 100644 --- a/file.h +++ b/file.h @@ -86,7 +86,7 @@ struct fio_file { */ unsigned int major, minor; int fileno; - int bs; + unsigned long long bs; char *file_name; /* diff --git a/filesetup.c b/filesetup.c index a2427a1..accb67a 100644 --- a/filesetup.c +++ b/filesetup.c @@ -107,7 +107,7 @@ static int extend_file(struct thread_data *td, struct fio_file *f) { int new_layout = 0, unlink_file = 0, flags; unsigned long long left; - unsigned int bs; + unsigned long long bs; char *b = NULL; if (read_only) { @@ -260,7 +260,7 @@ static bool pre_read_file(struct thread_data *td, struct fio_file *f) { int r, did_open = 0, old_runstate; unsigned long long left; - unsigned int bs; + unsigned long long bs; bool ret = true; char *b; @@ -900,7 +900,7 @@ int setup_files(struct thread_data *td) unsigned int i, nr_fs_extra = 0; int err = 0, need_extend; int old_state; - const unsigned int bs = td_min_bs(td); + const unsigned long long bs = td_min_bs(td); uint64_t fs = 0; dprint(FD_FILE, "setup files\n"); diff --git a/fio.h b/fio.h index 51b8fdc..ab44347 100644 --- a/fio.h +++ b/fio.h @@ -736,17 +736,17 @@ static inline bool should_check_rate(struct thread_data *td) return ddir_rw_sum(td->bytes_done) != 0; } -static inline unsigned int td_max_bs(struct thread_data *td) +static inline unsigned long long td_max_bs(struct thread_data *td) { - unsigned int max_bs; + unsigned long long max_bs; max_bs = max(td->o.max_bs[DDIR_READ], td->o.max_bs[DDIR_WRITE]); return max(td->o.max_bs[DDIR_TRIM], max_bs); } -static inline unsigned int td_min_bs(struct thread_data *td) +static inline unsigned long long td_min_bs(struct thread_data *td) { - unsigned int min_bs; + unsigned long long min_bs; min_bs = min(td->o.min_bs[DDIR_READ], td->o.min_bs[DDIR_WRITE]); return min(td->o.min_bs[DDIR_TRIM], min_bs); diff --git a/init.c b/init.c index af4cc6b..8cb8117 100644 --- a/init.c +++ b/init.c @@ -531,7 +531,7 @@ static void put_job(struct thread_data *td) static int __setup_rate(struct thread_data *td, enum fio_ddir ddir) { - unsigned int bs = td->o.min_bs[ddir]; + unsigned long long bs = td->o.min_bs[ddir]; assert(ddir_rw(ddir)); @@ -891,7 +891,7 @@ static int fixup_options(struct thread_data *td) * If size is set but less than the min block size, complain */ if (o->size && o->size < td_min_bs(td)) { - log_err("fio: size too small, must not be less than minimum block size: %llu < %u\n", + log_err("fio: size too small, must not be less than minimum block size: %llu < %llu\n", (unsigned long long) o->size, td_min_bs(td)); ret |= 1; } diff --git a/io_u.c b/io_u.c index 580c414..bfbce39 100644 --- a/io_u.c +++ b/io_u.c @@ -33,9 +33,9 @@ static bool random_map_free(struct fio_file *f, const uint64_t block) */ static void mark_random_map(struct thread_data *td, struct io_u *io_u) { - unsigned int min_bs = td->o.min_bs[io_u->ddir]; + unsigned long long min_bs = td->o.min_bs[io_u->ddir]; struct fio_file *f = io_u->file; - unsigned int nr_blocks; + unsigned long long nr_blocks; uint64_t block; block = (io_u->offset - f->file_offset) / (uint64_t) min_bs; @@ -515,7 +515,7 @@ static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u, { int ddir = io_u->ddir; unsigned int buflen = 0; - unsigned int minbs, maxbs; + unsigned long long minbs, maxbs; uint64_t frand_max, r; bool power_2; @@ -2085,8 +2085,8 @@ static void save_buf_state(struct thread_data *td, struct frand_state *rs) frand_copy(&td->buf_state_prev, rs); } -void fill_io_buffer(struct thread_data *td, void *buf, unsigned int min_write, - unsigned int max_bs) +void fill_io_buffer(struct thread_data *td, void *buf, unsigned long long min_write, + unsigned long long max_bs) { struct thread_options *o = &td->o; @@ -2096,8 +2096,8 @@ void fill_io_buffer(struct thread_data *td, void *buf, unsigned int min_write, if (o->compress_percentage || o->dedupe_percentage) { unsigned int perc = td->o.compress_percentage; struct frand_state *rs; - unsigned int left = max_bs; - unsigned int this_write; + unsigned long long left = max_bs; + unsigned long long this_write; do { rs = get_buf_state(td); @@ -2106,7 +2106,7 @@ void fill_io_buffer(struct thread_data *td, void *buf, unsigned int min_write, if (perc) { this_write = min_not_zero(min_write, - td->o.compress_chunk); + (unsigned long long) td->o.compress_chunk); fill_random_buf_percentage(rs, buf, perc, this_write, this_write, @@ -2133,7 +2133,7 @@ void fill_io_buffer(struct thread_data *td, void *buf, unsigned int min_write, * "randomly" fill the buffer contents */ void io_u_fill_buffer(struct thread_data *td, struct io_u *io_u, - unsigned int min_write, unsigned int max_bs) + unsigned long long min_write, unsigned long long max_bs) { io_u->buf_filled_len = 0; fill_io_buffer(td, io_u->buf, min_write, max_bs); diff --git a/io_u.h b/io_u.h index 4f433c3..b068e5b 100644 --- a/io_u.h +++ b/io_u.h @@ -134,8 +134,8 @@ extern void io_u_queued(struct thread_data *, struct io_u *); extern int io_u_quiesce(struct thread_data *); extern void io_u_log_error(struct thread_data *, struct io_u *); extern void io_u_mark_depth(struct thread_data *, unsigned int); -extern void fill_io_buffer(struct thread_data *, void *, unsigned int, unsigned int); -extern void io_u_fill_buffer(struct thread_data *td, struct io_u *, unsigned int, unsigned int); +extern void fill_io_buffer(struct thread_data *, void *, unsigned long long, unsigned long long); +extern void io_u_fill_buffer(struct thread_data *td, struct io_u *, unsigned long long, unsigned long long); void io_u_mark_complete(struct thread_data *, unsigned int); void io_u_mark_submit(struct thread_data *, unsigned int); bool queue_full(const struct thread_data *); diff --git a/iolog.c b/iolog.c index 5be3e84..3a9c1ad 100644 --- a/iolog.c +++ b/iolog.c @@ -737,8 +737,8 @@ static void flush_hist_samples(FILE *f, int hist_coarseness, void *samples, entry_before = flist_first_entry(&entry->list, struct io_u_plat_entry, list); io_u_plat_before = entry_before->io_u_plat; - fprintf(f, "%lu, %u, %u, ", (unsigned long) s->time, - io_sample_ddir(s), s->bs); + fprintf(f, "%lu, %u, %llu, ", (unsigned long) s->time, + io_sample_ddir(s), (unsigned long long) s->bs); for (j = 0; j < FIO_IO_U_PLAT_NR - stride; j += stride) { fprintf(f, "%llu, ", (unsigned long long) hist_sum(j, stride, io_u_plat, io_u_plat_before)); @@ -770,17 +770,17 @@ void flush_samples(FILE *f, void *samples, uint64_t sample_size) s = __get_sample(samples, log_offset, i); if (!log_offset) { - fprintf(f, "%lu, %" PRId64 ", %u, %u\n", + fprintf(f, "%lu, %" PRId64 ", %u, %llu\n", (unsigned long) s->time, s->data.val, - io_sample_ddir(s), s->bs); + io_sample_ddir(s), (unsigned long long) s->bs); } else { struct io_sample_offset *so = (void *) s; - fprintf(f, "%lu, %" PRId64 ", %u, %u, %llu\n", + fprintf(f, "%lu, %" PRId64 ", %u, %llu, %llu\n", (unsigned long) s->time, s->data.val, - io_sample_ddir(s), s->bs, + io_sample_ddir(s), (unsigned long long) s->bs, (unsigned long long) so->offset); } } diff --git a/iolog.h b/iolog.h index a4e335a..3b8c901 100644 --- a/iolog.h +++ b/iolog.h @@ -42,7 +42,7 @@ struct io_sample { uint64_t time; union io_sample_data data; uint32_t __ddir; - uint32_t bs; + uint64_t bs; }; struct io_sample_offset { diff --git a/options.c b/options.c index a174e2c..439bd9b 100644 --- a/options.c +++ b/options.c @@ -119,7 +119,7 @@ static int bssplit_ddir(struct thread_options *o, enum fio_ddir ddir, char *str, bool data) { unsigned int i, perc, perc_missing; - unsigned int max_bs, min_bs; + unsigned long long max_bs, min_bs; struct split split; memset(&split, 0, sizeof(split)); diff --git a/parse.c b/parse.c index 6261fca..fd08a60 100644 --- a/parse.c +++ b/parse.c @@ -582,7 +582,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr, return 1; } - if (o->maxval && ull > o->maxval) { + if (o->maxval && ull > o->maxval && ((o->name!="bs") && (o->name!="ba"))) { log_err("max value out of range: %llu" " (%u max)\n", ull, o->maxval); return 1; diff --git a/server.c b/server.c index 7e7ffed..b966c66 100644 --- a/server.c +++ b/server.c @@ -1985,7 +1985,7 @@ int fio_send_iolog(struct thread_data *td, struct io_log *log, const char *name) s->time = cpu_to_le64(s->time); s->data.val = cpu_to_le64(s->data.val); s->__ddir = cpu_to_le32(s->__ddir); - s->bs = cpu_to_le32(s->bs); + s->bs = cpu_to_le64(s->bs); if (log->log_offset) { struct io_sample_offset *so = (void *) s; diff --git a/stat.c b/stat.c index a308eb8..4aa8d09 100644 --- a/stat.c +++ b/stat.c @@ -619,8 +619,8 @@ 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); @@ -2220,7 +2220,7 @@ static struct io_logs *get_cur_log(struct io_log *iolog) } static void __add_log_sample(struct io_log *iolog, union io_sample_data data, - enum fio_ddir ddir, unsigned int bs, + enum fio_ddir ddir, unsigned long long bs, unsigned long t, uint64_t offset) { struct io_logs *cur_log; @@ -2338,7 +2338,7 @@ static void _add_stat_to_log(struct io_log *iolog, unsigned long elapsed, static unsigned long add_log_sample(struct thread_data *td, struct io_log *iolog, union io_sample_data data, - enum fio_ddir ddir, unsigned int bs, + enum fio_ddir ddir, unsigned long long bs, uint64_t offset) { unsigned long elapsed, this_window; @@ -2400,7 +2400,7 @@ void finalize_logs(struct thread_data *td, bool unit_logs) _add_stat_to_log(td->iops_log, elapsed, td->o.log_max != 0); } -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; @@ -2430,7 +2430,7 @@ static void add_clat_percentile_sample(struct thread_stat *ts, } 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 long elapsed, this_window; struct thread_stat *ts = &td->ts; @@ -2489,7 +2489,7 @@ 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 int bs, uint64_t offset) + unsigned long usec, unsigned long long bs, uint64_t offset) { struct thread_stat *ts = &td->ts; @@ -2507,7 +2507,7 @@ void add_slat_sample(struct thread_data *td, enum fio_ddir ddir, } 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) { struct thread_stat *ts = &td->ts; @@ -2590,7 +2590,7 @@ 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]; diff --git a/stat.h b/stat.h index c5b8185..5dcaae0 100644 --- a/stat.h +++ b/stat.h @@ -308,12 +308,12 @@ extern void update_rusage_stat(struct thread_data *); extern void clear_rusage_stat(struct thread_data *); extern void add_lat_sample(struct thread_data *, enum fio_ddir, unsigned long long, - unsigned int, uint64_t); + unsigned long long, uint64_t); extern void add_clat_sample(struct thread_data *, enum fio_ddir, unsigned long long, - unsigned int, uint64_t); + unsigned long long, uint64_t); extern void add_slat_sample(struct thread_data *, enum fio_ddir, unsigned long, - unsigned int, uint64_t); -extern void add_agg_sample(union io_sample_data, enum fio_ddir, unsigned int); + unsigned long long, uint64_t); +extern void add_agg_sample(union io_sample_data, enum fio_ddir, unsigned long long); extern void add_iops_sample(struct thread_data *, struct io_u *, unsigned int); extern void add_bw_sample(struct thread_data *, struct io_u *, diff --git a/thread_options.h b/thread_options.h index 8d13b79..a250d2c 100644 --- a/thread_options.h +++ b/thread_options.h @@ -29,7 +29,7 @@ enum fio_memtype { #define ZONESPLIT_MAX 256 struct bssplit { - uint32_t bs; + uint64_t bs; uint32_t perc; }; @@ -82,10 +82,10 @@ struct thread_options { unsigned long long start_offset; unsigned long long start_offset_align; - unsigned int bs[DDIR_RWDIR_CNT]; - unsigned int ba[DDIR_RWDIR_CNT]; - unsigned int min_bs[DDIR_RWDIR_CNT]; - unsigned int max_bs[DDIR_RWDIR_CNT]; + unsigned long long bs[DDIR_RWDIR_CNT]; + unsigned long long ba[DDIR_RWDIR_CNT]; + unsigned long long min_bs[DDIR_RWDIR_CNT]; + unsigned long long max_bs[DDIR_RWDIR_CNT]; struct bssplit *bssplit[DDIR_RWDIR_CNT]; unsigned int bssplit_nr[DDIR_RWDIR_CNT]; @@ -164,7 +164,8 @@ struct thread_options { unsigned int perc_rand[DDIR_RWDIR_CNT]; unsigned int hugepage_size; - unsigned int rw_min_bs; + unsigned long long rw_min_bs; + unsigned int pad2; unsigned int thinktime; unsigned int thinktime_spin; unsigned int thinktime_blocks; @@ -363,10 +364,10 @@ struct thread_options_pack { uint64_t start_offset; uint64_t start_offset_align; - uint32_t bs[DDIR_RWDIR_CNT]; - uint32_t ba[DDIR_RWDIR_CNT]; - uint32_t min_bs[DDIR_RWDIR_CNT]; - uint32_t max_bs[DDIR_RWDIR_CNT]; + uint64_t bs[DDIR_RWDIR_CNT]; + uint64_t ba[DDIR_RWDIR_CNT]; + uint64_t min_bs[DDIR_RWDIR_CNT]; + uint64_t max_bs[DDIR_RWDIR_CNT]; struct bssplit bssplit[DDIR_RWDIR_CNT][BSSPLIT_MAX]; uint32_t bssplit_nr[DDIR_RWDIR_CNT]; @@ -443,7 +444,8 @@ struct thread_options_pack { uint32_t perc_rand[DDIR_RWDIR_CNT]; uint32_t hugepage_size; - uint32_t rw_min_bs; + uint64_t rw_min_bs; + uint32_t pad2; uint32_t thinktime; uint32_t thinktime_spin; uint32_t thinktime_blocks; ��.n��������+%������w��{.n�������^n�r������&��z�ޗ�zf���h���~����������_��+v���)ߣ�