fio max blocksize

fio max blocksize

[Jeff Furlong]

Hi All,
Currently it looks like fio has a max blocksize of uint32_t.  So commands such as 

# fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=4g --filename=/dev/nvme1n1 --number_ios=1 max value out of range: 4294967296 (4294967295 max)
fio: failed parsing bs=4g

will fail.  While I wouldn't normally do 4GB reads/writes, I would like to issue 4GB trims.  Otherwise I need to invoke a tool such as blkdiscard, but then summarizing all job IO is very difficult.  I went through the process of unsuccessfully changing the blocksize variables to uint64_t (failed getting buflen), but not sure if I missed a minor detail or a major roadblock that would prevent it.  How could >4GB blocksizes be supported?  Thanks.

Regards,
Jeff

Re: fio max blocksize

[Sitsofe Wheeler]

Hi,

On 6 July 2018 at 22:15, Jeff Furlong <jeff.furlong@wdc.com> wrote:
> Hi All,
> Currently it looks like fio has a max blocksize of uint32_t.  So commands such as
>
> # fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=4g --filename=/dev/nvme1n1 --number_ios=1 max value out of range: 4294967296 (4294967295 max)
> fio: failed parsing bs=4g
>
> will fail.  While I wouldn't normally do 4GB reads/writes, I would like to issue 4GB trims.  Otherwise I need to invoke a tool such as blkdiscard, but then summarizing all job IO is very difficult.  I went through the process of unsuccessfully changing the blocksize variables to uint64_t (failed getting buflen), but not sure if I missed a minor detail or a major roadblock that would prevent it.  How could >4GB blocksizes be supported?  Thanks.

Wow! I was going to say "surely discards can't be that huge?" but I've
just checked and even a cheap SSD has a queue/discard_max_hw_bytes of
2GBytes and I see some NVMe drives have 4GBytes for that value. It's
not really fair on fio because in hardware the disk's blocksize
remains the same it's just that asking for a discard sends down an
offset+range of blocks so it is able to reach huge sizes...

As to your question I'm not sure if fio has assumptions in that expect
blocks not to be giant... I suspect it just seemed a sane limit and
tradeoff at the time - 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... :-p

-- 
Sitsofe | http://sucs.org/~sits/

Re: fio max blocksize

[Jens Axboe]

On 7/6/18 3:15 PM, Jeff Furlong wrote:
> Hi All,
> Currently it looks like fio has a max blocksize of uint32_t.  So commands such as 
> 
> # fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=4g --filename=/dev/nvme1n1 --number_ios=1 max value out of range: 4294967296 (4294967295 max)
> fio: failed parsing bs=4g
> 
> will fail.  While I wouldn't normally do 4GB reads/writes, I would like to issue 4GB trims.  Otherwise I need to invoke a tool such as blkdiscard, but then summarizing all job IO is very difficult.  I went through the process of unsuccessfully changing the blocksize variables to uint64_t (failed getting buflen), but not sure if I missed a minor detail or a major roadblock that would prevent it.  How could >4GB blocksizes be supported?  Thanks.

You'd just need to change the type of the bs related options, and all
of the structs that hold it. Not impossible at all, an pretty
straight forward. So I'd say that what you did would likely work,
though it's hard to ensure that you didn't miss a conversion
somewhere. Post the patch and we can take a look at it.

-- 
Jens Axboe

RE: fio max blocksize

[Jeff Furlong]

> 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;

Re: fio max blocksize

[Jens Axboe]

On 7/12/18 12:08 PM, Jeff Furlong wrote:
>> 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.

"rough" is one way to express it, "not doing at all what you think" would be
another :-)

> 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"))) {

I'm guessing you do more python than C :-)

In any case, the real fix would be to make FIO_OPT_RANGE work on ull,
not unsignd ints. That one is trivial, the other block size related
options are worse since they are used by other options as well. Probably
the easiest to just introduce ULL versions of the ones used, and reuse
most of the code for that. Basically the unsigned int version should
just be a subset of the ull, with the min/max settings things would just
work out nicely.

-- 
Jens Axboe

RE: fio max blocksize

[Jeff Furlong]

>"rough" is one way to express it, "not doing at all what you think" would be another :-)
OK, fair enough.  It was a short way to describe what I wanted but even the compiler warned me it wasn't doing what I wanted.  The hack was just to force that line to evaluate to false and try the rest of the patch (which failed, so that seems to be my bigger issue).  So I thought to solve that parsing issue another day :)

> Basically the unsigned int version should just be a subset of the ull, with the min/max settings things would just work out nicely.
So the idea is to share the same case statement for both, not introduce unique case statements for UL and ULL, right?

Regards,
Jeff

Re: fio max blocksize

[Jens Axboe]

On 7/12/18 1:11 PM, Jeff Furlong wrote:
>> Basically the unsigned int version should just be a subset of the
>> ull, with the min/max settings things would just work out nicely.
>
> So the idea is to share the same case statement for both, not
> introduce unique case statements for UL and ULL, right?

Right, it's just converting (if needed) the general case to be ull
based. See how FIO_OPT_INT is basically the same as FIO_OPT_STR_VAL.
You might need to add an implicit knowledge that the uint based options
have an implied maxval of UINT_MAX, if maxval isn't set, to avoid cases
where we can parse much larger values than a particular option may be
able to store.

-- 
Jens Axboe

RE: fio max blocksize

[Jeff Furlong]

> Right, it's just converting (if needed) the general case to be ull based
I added a FIO_OPT_ULL case and made the maxval appropriate there.

# fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=4g --filename=/dev/nvme1n1 --number_ios=1 --debug=io
fio: set debug option io
test: (g=0): rw=trim, bs=(R) 4096MiB-4096MiB, (W) 4096MiB-4096MiB, (T) 4096MiB-4096MiB, ioengine=libaio, iodepth=1
fio-3.7-47-g19a8-dirty
Starting 1 process
io       81681 drop page cache /dev/nvme1n1
io       81681 io_u 0x2457380, failed getting buflen
io       81681 io_u 0x2457380, setting file failed
io       81681 get_io_u failed
test: No I/O performed by libaio, perhaps try --debug=io option for details?
io       81681 close ioengine libaio
io       81681 free ioengine libaio

I didn't address bsrange or blktrace, but they seem workable after I fix the above issue.

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..248cc86 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));
@@ -2112,7 +2112,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 		.name	= "bs",
 		.lname	= "Block size",
 		.alias	= "blocksize",
-		.type	= FIO_OPT_INT,
+		.type	= FIO_OPT_ULL,
 		.off1	= offsetof(struct thread_options, bs[DDIR_READ]),
 		.off2	= offsetof(struct thread_options, bs[DDIR_WRITE]),
 		.off3	= offsetof(struct thread_options, bs[DDIR_TRIM]),
@@ -2129,7 +2129,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 		.name	= "ba",
 		.lname	= "Block size align",
 		.alias	= "blockalign",
-		.type	= FIO_OPT_INT,
+		.type	= FIO_OPT_ULL,
 		.off1	= offsetof(struct thread_options, ba[DDIR_READ]),
 		.off2	= offsetof(struct thread_options, ba[DDIR_WRITE]),
 		.off3	= offsetof(struct thread_options, ba[DDIR_TRIM]),
diff --git a/parse.c b/parse.c
index 6261fca..d125ffc 100644
--- a/parse.c
+++ b/parse.c
@@ -32,6 +32,7 @@ static const char *opt_type_names[] = {
 	"OPT_STR_STORE",
 	"OPT_RANGE",
 	"OPT_INT",
+    "OPT_ULL",
 	"OPT_BOOL",
 	"OPT_FLOAT_LIST",
 	"OPT_STR_SET",
@@ -554,6 +555,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 	}
 	case FIO_OPT_STR_VAL_TIME:
 		is_time = 1;
+    case FIO_OPT_ULL:
 	case FIO_OPT_INT:
 	case FIO_OPT_STR_VAL: {
 		fio_opt_str_val_fn *fn = o->cb;
@@ -584,7 +586,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 
 		if (o->maxval && ull > o->maxval) {
 			log_err("max value out of range: %llu"
-					" (%u max)\n", ull, o->maxval);
+					" (%llu max)\n", ull, o->maxval);
 			return 1;
 		}
 		if (o->minval && ull < o->minval) {
@@ -615,25 +617,25 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 		if (fn)
 			ret = fn(data, &ull);
 		else {
-			if (o->type == FIO_OPT_INT) {
+			if ((o->type == FIO_OPT_INT) || (o->type == FIO_OPT_ULL)) {
 				if (first)
-					val_store(ilp, ull, o->off1, 0, data, o);
+					val_store(ullp, ull, o->off1, 0, data, o);
 				if (curr == 1) {
 					if (o->off2)
-						val_store(ilp, ull, o->off2, 0, data, o);
+						val_store(ullp, ull, o->off2, 0, data, o);
 				}
 				if (curr == 2) {
 					if (o->off3)
-						val_store(ilp, ull, o->off3, 0, data, o);
+						val_store(ullp, ull, o->off3, 0, data, o);
 				}
 				if (!more) {
 					if (curr < 1) {
 						if (o->off2)
-							val_store(ilp, ull, o->off2, 0, data, o);
+							val_store(ullp, ull, o->off2, 0, data, o);
 					}
 					if (curr < 2) {
 						if (o->off3)
-							val_store(ilp, ull, o->off3, 0, data, o);
+							val_store(ullp, ull, o->off3, 0, data, o);
 					}
 				}
 			} else {
@@ -851,7 +853,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 			break;
 
 		if (o->maxval && il > (int) o->maxval) {
-			log_err("max value out of range: %d (%d max)\n",
+			log_err("max value out of range: %d (%llu max)\n",
 								il, o->maxval);
 			return 1;
 		}
@@ -1325,6 +1327,10 @@ static void option_init(struct fio_option *o)
 		if (!o->maxval)
 			o->maxval = UINT_MAX;
 	}
+    if (o->type == FIO_OPT_ULL) {
+		if (!o->maxval)
+			o->maxval = ULLONG_MAX;
+	}
 	if (o->type == FIO_OPT_STR_SET && o->def && !o->no_warn_def) {
 		log_err("Option %s: string set option with"
 				" default will always be true\n", o->name);
diff --git a/parse.h b/parse.h
index 4de5e77..56f5218 100644
--- a/parse.h
+++ b/parse.h
@@ -16,6 +16,7 @@ enum fio_opt_type {
 	FIO_OPT_STR_STORE,
 	FIO_OPT_RANGE,
 	FIO_OPT_INT,
+    FIO_OPT_ULL,
 	FIO_OPT_BOOL,
 	FIO_OPT_FLOAT_LIST,
 	FIO_OPT_STR_SET,
@@ -52,7 +53,7 @@ struct fio_option {
 	unsigned int off4;
 	unsigned int off5;
 	unsigned int off6;
-	unsigned int maxval;		/* max and min value */
+	unsigned long long maxval;		/* max and min value */
 	int minval;
 	double maxfp;			/* max and min floating value */
 	double minfp;
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;

Re: fio max blocksize

[Jens Axboe]

On 7/12/18 4:13 PM, Jeff Furlong wrote:
>> Right, it's just converting (if needed) the general case to be ull based
> I added a FIO_OPT_ULL case and made the maxval appropriate there.

> diff --git a/parse.c b/parse.c
> index 6261fca..d125ffc 100644
> --- a/parse.c
> +++ b/parse.c
> @@ -615,25 +617,25 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
>  		if (fn)
>  			ret = fn(data, &ull);
>  		else {
> -			if (o->type == FIO_OPT_INT) {
> +			if ((o->type == FIO_OPT_INT) || (o->type == FIO_OPT_ULL)) {
>  				if (first)
> -					val_store(ilp, ull, o->off1, 0, data, o);
> +					val_store(ullp, ull, o->off1, 0, data, o);

This isn't going to work, as FIO_OPT_INT will be an integer pointer, not
a pointer to an unsigned long long. The above will overwrite nearby
members.

-- 
Jens Axboe

RE: fio max blocksize

[Jeff Furlong]

> The above will overwrite nearby members.
Was thinking the member was allocated as 64 bits but may only use 32 bits, but agreed that's not true for other options that use FIO_OPT_INT (e.g. nr_files).  How about just breaking the ULL if/else out separately:

diff --git a/parse.c b/parse.c
index 6261fca..35ec362 100644
--- a/parse.c
+++ b/parse.c
@@ -32,6 +32,7 @@ static const char *opt_type_names[] = {
 	"OPT_STR_STORE",
 	"OPT_RANGE",
 	"OPT_INT",
+    "OPT_ULL",
 	"OPT_BOOL",
 	"OPT_FLOAT_LIST",
 	"OPT_STR_SET",
@@ -554,6 +555,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 	}
 	case FIO_OPT_STR_VAL_TIME:
 		is_time = 1;
+    case FIO_OPT_ULL:
 	case FIO_OPT_INT:
 	case FIO_OPT_STR_VAL: {
 		fio_opt_str_val_fn *fn = o->cb;
@@ -584,7 +586,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 
 		if (o->maxval && ull > o->maxval) {
 			log_err("max value out of range: %llu"
-					" (%u max)\n", ull, o->maxval);
+					" (%llu max)\n", ull, o->maxval);
 			return 1;
 		}
 		if (o->minval && ull < o->minval) {
@@ -636,6 +638,28 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 							val_store(ilp, ull, o->off3, 0, data, o);
 					}
 				}
+            }
+            else if (o->type == FIO_OPT_ULL) {
+				if (first)
+					val_store(ullp, ull, o->off1, 0, data, o);
+				if (curr == 1) {
+					if (o->off2)
+						val_store(ullp, ull, o->off2, 0, data, o);
+				}
+				if (curr == 2) {
+					if (o->off3)
+						val_store(ullp, ull, o->off3, 0, data, o);
+				}
+				if (!more) {
+					if (curr < 1) {
+						if (o->off2)
+							val_store(ullp, ull, o->off2, 0, data, o);
+					}
+					if (curr < 2) {
+						if (o->off3)
+							val_store(ullp, ull, o->off3, 0, data, o);
+					}
+				}
 			} else {
 				if (first)
 					val_store(ullp, ull, o->off1, 0, data, o);
@@ -851,7 +875,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 			break;
 
 		if (o->maxval && il > (int) o->maxval) {
-			log_err("max value out of range: %d (%d max)\n",
+			log_err("max value out of range: %d (%llu max)\n",
 								il, o->maxval);
 			return 1;
 		}
@@ -1325,6 +1349,10 @@ static void option_init(struct fio_option *o)
 		if (!o->maxval)
 			o->maxval = UINT_MAX;
 	}
+    if (o->type == FIO_OPT_ULL) {
+		if (!o->maxval)
+			o->maxval = ULLONG_MAX;
+	}
 	if (o->type == FIO_OPT_STR_SET && o->def && !o->no_warn_def) {
 		log_err("Option %s: string set option with"
 				" default will always be true\n", o->name);


Regards,
Jeff

Re: fio max blocksize

[Jens Axboe]

On 7/12/18 4:47 PM, Jeff Furlong wrote:
>> The above will overwrite nearby members.
> Was thinking the member was allocated as 64 bits but may only use 32
> bits, but agreed that's not true for other options that use
> FIO_OPT_INT (e.g. nr_files).  How about just breaking the ULL if/else
> out separately:

That is much better, that'll work fine.

BTW, not on when you do a more final version, be wary of tabs and
spaces. You seem to mix and match a bit. It's always tabs.

-- 
Jens Axboe

RE: fio max blocksize

[Elliott, Robert (Persistent Memory)]

> -----Original Message-----
> From: fio-owner@vger.kernel.org [mailto:fio-owner@vger.kernel.org] On Behalf Of Jeff Furlong
> Sent: Thursday, July 12, 2018 5:14 PM
> To: Jens Axboe <axboe@kernel.dk>; fio@vger.kernel.org
> Subject: RE: fio max blocksize
...
> 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]);

That implies an interoperability issue in client/server mode - a new client could
generate values that an old server couldn't parse.  Update FIO_SERVER_VER to
require them to be compatible versions.


---
Robert Elliott, HPE Persistent Memory

RE: fio max blocksize

[Jeff Furlong]

> Update FIO_SERVER_VER to require them to be compatible versions.
Thanks, good catch.  I updated on my local working copy.

Regards,
Jeff

RE: fio max blocksize

[Jeff Furlong]

I fixed the buflen issue (just keep peeling the onion back and updating all associated vars), fixed bsrange, fixed "auto" tabs, and did some simple tests:

# fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=4g --filename=/dev/nvme1n1 --number_ios=1
# fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=1t --filename=/dev/nvme1n1 --number_ios=1
# fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bsrange=4g-8g --filename=/dev/nvme1n1 --number_ios=2

The open issue I have is parsing bssplit.  I plumbed out bssplit.bs as a 64 bit value, but using the existing value_pair helpers (32 bit) seem to conflict.  Don’t really want to duplicate chunks of code, so open to ideas there.

Regards,
Jeff


diff --git a/backend.c b/backend.c
index a7e9184..6b6efe1 100644
--- a/backend.c
+++ b/backend.c
@@ -454,7 +454,7 @@ int io_queue_event(struct thread_data *td, struct io_u *io_u, int *ret,
 			*ret = -io_u->error;
 			clear_io_u(td, io_u);
 		} else if (io_u->resid) {
-			int bytes = io_u->xfer_buflen - io_u->resid;
+			long long bytes = io_u->xfer_buflen - io_u->resid;
 			struct fio_file *f = io_u->file;
 
 			if (bytes_issued)
@@ -583,7 +583,7 @@ static bool in_flight_overlap(struct io_u_queue *q, struct io_u *io_u)
 
 			if (x1 < y2 && y1 < x2) {
 				overlap = true;
-				dprint(FD_IO, "in-flight overlap: %llu/%lu, %llu/%lu\n",
+				dprint(FD_IO, "in-flight overlap: %llu/%llu, %llu/%llu\n",
 						x1, io_u->buflen,
 						y1, check_io_u->buflen);
 				break;
@@ -1033,7 +1033,7 @@ static void do_io(struct thread_data *td, uint64_t *bytes_done)
 			log_io_piece(td, io_u);
 
 		if (td->o.io_submit_mode == IO_MODE_OFFLOAD) {
-			const unsigned long blen = io_u->xfer_buflen;
+			const unsigned long long blen = io_u->xfer_buflen;
 			const enum fio_ddir __ddir = acct_ddir(io_u);
 
 			if (td->error)
@@ -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..534bfb0 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..40375c3 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..bb73748 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..767ed00 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..efeb4cd 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..0d6cc64 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;
@@ -503,19 +503,19 @@ static int get_next_offset(struct thread_data *td, struct io_u *io_u,
 }
 
 static inline bool io_u_fits(struct thread_data *td, struct io_u *io_u,
-			     unsigned int buflen)
+				 unsigned long long buflen)
 {
 	struct fio_file *f = io_u->file;
 
 	return io_u->offset + buflen <= f->io_size + get_start_offset(td, f);
 }
 
-static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u,
+static unsigned long long get_next_buflen(struct thread_data *td, struct io_u *io_u,
 					bool is_random)
 {
 	int ddir = io_u->ddir;
-	unsigned int buflen = 0;
-	unsigned int minbs, maxbs;
+	unsigned long long buflen = 0;
+	unsigned long long minbs, maxbs;
 	uint64_t frand_max, r;
 	bool power_2;
 
@@ -541,10 +541,8 @@ static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u,
 		r = __rand(&td->bsrange_state[ddir]);
 
 		if (!td->o.bssplit_nr[ddir]) {
-			buflen = 1 + (unsigned int) ((double) maxbs *
+			buflen = minbs + (unsigned long long) ((double) maxbs *
 					(r / (frand_max + 1.0)));
-			if (buflen < minbs)
-				buflen = minbs;
 		} else {
 			long long perc = 0;
 			unsigned int i;
@@ -894,7 +892,7 @@ static int fill_io_u(struct thread_data *td, struct io_u *io_u)
 	}
 
 	if (io_u->offset + io_u->buflen > io_u->file->real_file_size) {
-		dprint(FD_IO, "io_u %p, off=0x%llx + len=0x%lx exceeds file size=0x%llx\n",
+		dprint(FD_IO, "io_u %p, off=0x%llx + len=0x%llx exceeds file size=0x%llx\n",
 			io_u,
 			(unsigned long long) io_u->offset, io_u->buflen,
 			(unsigned long long) io_u->file->real_file_size);
@@ -1585,7 +1583,7 @@ static bool check_get_verify(struct thread_data *td, struct io_u *io_u)
  */
 static void small_content_scramble(struct io_u *io_u)
 {
-	unsigned int i, nr_blocks = io_u->buflen >> 9;
+	unsigned long long i, nr_blocks = io_u->buflen >> 9;
 	unsigned int offset;
 	uint64_t boffset, *iptr;
 	char *p;
@@ -1729,7 +1727,7 @@ static void __io_u_log_error(struct thread_data *td, struct io_u *io_u)
 	if (td_non_fatal_error(td, eb, io_u->error) && !td->o.error_dump)
 		return;
 
-	log_err("fio: io_u error%s%s: %s: %s offset=%llu, buflen=%lu\n",
+	log_err("fio: io_u error%s%s: %s: %s offset=%llu, buflen=%llu\n",
 		io_u->file ? " on file " : "",
 		io_u->file ? io_u->file->file_name : "",
 		strerror(io_u->error),
@@ -1895,7 +1893,7 @@ static void io_completed(struct thread_data *td, struct io_u **io_u_ptr,
 	td->last_ddir = ddir;
 
 	if (!io_u->error && ddir_rw(ddir)) {
-		unsigned int bytes = io_u->buflen - io_u->resid;
+		unsigned long long bytes = io_u->buflen - io_u->resid;
 		int ret;
 
 		td->io_blocks[ddir]++;
@@ -2085,8 +2083,8 @@ 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 +2094,8 @@ 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 +2104,7 @@ 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 +2131,7 @@ 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..9a423b2 100644
--- a/io_u.h
+++ b/io_u.h
@@ -51,7 +51,7 @@ struct io_u {
 	/*
 	 * Allocated/set buffer and length
 	 */
-	unsigned long buflen;
+	unsigned long long buflen;
 	unsigned long long offset;
 	void *buf;
 
@@ -65,13 +65,13 @@ struct io_u {
 	 * partial transfers / residual data counts
 	 */
 	void *xfer_buf;
-	unsigned long xfer_buflen;
+	unsigned long long xfer_buflen;
 
 	/*
 	 * Parameter related to pre-filled buffers and
 	 * their size to handle variable block sizes.
 	 */
-	unsigned long buf_filled_len;
+	unsigned long long buf_filled_len;
 
 	struct io_piece *ipo;
 
@@ -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 *);
@@ -149,13 +149,13 @@ static inline void dprint_io_u(struct io_u *io_u, const char *p)
 	struct fio_file *f = io_u->file;
 
 	if (f)
-		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%lx,ddir=%d,file=%s\n",
+		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%llx,ddir=%d,file=%s\n",
 				p, io_u,
 				(unsigned long long) io_u->offset,
 				io_u->buflen, io_u->ddir,
 				f->file_name);
 	else
-		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%lx,ddir=%d\n",
+		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%llx,ddir=%d\n",
 				p, io_u,
 				(unsigned long long) io_u->offset,
 				io_u->buflen, io_u->ddir);
diff --git a/ioengines.c b/ioengines.c
index d579682..b7f9d40 100644
--- a/ioengines.c
+++ b/ioengines.c
@@ -279,7 +279,7 @@ out:
 enum fio_q_status td_io_queue(struct thread_data *td, struct io_u *io_u)
 {
 	const enum fio_ddir ddir = acct_ddir(io_u);
-	unsigned long buflen = io_u->xfer_buflen;
+	unsigned long long buflen = io_u->xfer_buflen;
 	enum fio_q_status ret;
 
 	dprint_io_u(io_u, "queue");
diff --git a/iolog.c b/iolog.c
index 5be3e84..7653175 100644
--- a/iolog.c
+++ b/iolog.c
@@ -35,7 +35,7 @@ void log_io_u(const struct thread_data *td, const struct io_u *io_u)
 	if (!td->o.write_iolog_file)
 		return;
 
-	fprintf(td->iolog_f, "%s %s %llu %lu\n", io_u->file->file_name,
+	fprintf(td->iolog_f, "%s %s %llu %llu\n", io_u->file->file_name,
 						io_ddir_name(io_u->ddir),
 						io_u->offset, io_u->buflen);
 }
@@ -161,7 +161,7 @@ int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 			io_u->buflen = ipo->len;
 			io_u->file = td->files[ipo->fileno];
 			get_file(io_u->file);
-			dprint(FD_IO, "iolog: get %llu/%lu/%s\n", io_u->offset,
+			dprint(FD_IO, "iolog: get %llu/%llu/%s\n", io_u->offset,
 						io_u->buflen, io_u->file->file_name);
 			if (ipo->delay)
 				iolog_delay(td, ipo->delay);
@@ -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..8780327 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..0007348 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));
@@ -2112,7 +2112,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 		.name	= "bs",
 		.lname	= "Block size",
 		.alias	= "blocksize",
-		.type	= FIO_OPT_INT,
+		.type	= FIO_OPT_ULL,
 		.off1	= offsetof(struct thread_options, bs[DDIR_READ]),
 		.off2	= offsetof(struct thread_options, bs[DDIR_WRITE]),
 		.off3	= offsetof(struct thread_options, bs[DDIR_TRIM]),
@@ -2129,7 +2129,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 		.name	= "ba",
 		.lname	= "Block size align",
 		.alias	= "blockalign",
-		.type	= FIO_OPT_INT,
+		.type	= FIO_OPT_ULL,
 		.off1	= offsetof(struct thread_options, ba[DDIR_READ]),
 		.off2	= offsetof(struct thread_options, ba[DDIR_WRITE]),
 		.off3	= offsetof(struct thread_options, ba[DDIR_TRIM]),
@@ -2163,7 +2163,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 	{
 		.name	= "bssplit",
 		.lname	= "Block size split",
-		.type	= FIO_OPT_STR,
+		.type	= FIO_OPT_STR_ULL,
 		.cb	= str_bssplit_cb,
 		.off1	= offsetof(struct thread_options, bssplit),
 		.help	= "Set a specific mix of block sizes",
diff --git a/os/windows/posix.c b/os/windows/posix.c
old mode 100755
new mode 100644
index d33250d..1a04473
diff --git a/parse.c b/parse.c
index 6261fca..27d7d04 100644
--- a/parse.c
+++ b/parse.c
@@ -26,12 +26,14 @@
 static const char *opt_type_names[] = {
 	"OPT_INVALID",
 	"OPT_STR",
+	"OPT_STR_ULL",
 	"OPT_STR_MULTI",
 	"OPT_STR_VAL",
 	"OPT_STR_VAL_TIME",
 	"OPT_STR_STORE",
 	"OPT_RANGE",
 	"OPT_INT",
+	"OPT_ULL",
 	"OPT_BOOL",
 	"OPT_FLOAT_LIST",
 	"OPT_STR_SET",
@@ -438,7 +440,7 @@ void strip_blank_end(char *p)
 	*(s + 1) = '\0';
 }
 
-static int check_range_bytes(const char *str, long *val, void *data)
+static int check_range_bytes(const char *str, long long *val, void *data)
 {
 	long long __val;
 
@@ -507,7 +509,8 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 	int il=0, *ilp;
 	fio_fp64_t *flp;
 	long long ull, *ullp;
-	long ul1, ul2;
+	long ul2;
+	long long ull1, ull2;
 	double uf;
 	char **cp = NULL;
 	int ret = 0, is_time = 0;
@@ -525,6 +528,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 
 	switch (o->type) {
 	case FIO_OPT_STR:
+	case FIO_OPT_STR_ULL:
 	case FIO_OPT_STR_MULTI: {
 		fio_opt_str_fn *fn = o->cb;
 
@@ -554,6 +558,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 	}
 	case FIO_OPT_STR_VAL_TIME:
 		is_time = 1;
+	case FIO_OPT_ULL:
 	case FIO_OPT_INT:
 	case FIO_OPT_STR_VAL: {
 		fio_opt_str_val_fn *fn = o->cb;
@@ -584,7 +589,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 
 		if (o->maxval && ull > o->maxval) {
 			log_err("max value out of range: %llu"
-					" (%u max)\n", ull, o->maxval);
+					" (%llu max)\n", ull, o->maxval);
 			return 1;
 		}
 		if (o->minval && ull < o->minval) {
@@ -636,6 +641,28 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 							val_store(ilp, ull, o->off3, 0, data, o);
 					}
 				}
+			}
+			else if (o->type == FIO_OPT_ULL) {
+				if (first)
+					val_store(ullp, ull, o->off1, 0, data, o);
+				if (curr == 1) {
+					if (o->off2)
+						val_store(ullp, ull, o->off2, 0, data, o);
+				}
+				if (curr == 2) {
+					if (o->off3)
+						val_store(ullp, ull, o->off3, 0, data, o);
+				}
+				if (!more) {
+					if (curr < 1) {
+						if (o->off2)
+							val_store(ullp, ull, o->off2, 0, data, o);
+					}
+					if (curr < 2) {
+						if (o->off3)
+							val_store(ullp, ull, o->off3, 0, data, o);
+					}
+				}
 			} else {
 				if (first)
 					val_store(ullp, ull, o->off1, 0, data, o);
@@ -790,43 +817,43 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 		p1 = tmp;
 
 		ret = 1;
-		if (!check_range_bytes(p1, &ul1, data) &&
-		    !check_range_bytes(p2, &ul2, data)) {
+		if (!check_range_bytes(p1, &ull1, data) &&
+			!check_range_bytes(p2, &ull2, data)) {
 			ret = 0;
-			if (ul1 > ul2) {
-				unsigned long foo = ul1;
+			if (ull1 > ull2) {
+				unsigned long long foo = ull1;
 
-				ul1 = ul2;
-				ul2 = foo;
+				ull1 = ull2;
+				ull2 = foo;
 			}
 
 			if (first) {
-				val_store(ilp, ul1, o->off1, 0, data, o);
-				val_store(ilp, ul2, o->off2, 0, data, o);
+				val_store(ullp, ull1, o->off1, 0, data, o);
+				val_store(ullp, ull2, o->off2, 0, data, o);
 			}
 			if (curr == 1) {
 				if (o->off3 && o->off4) {
-					val_store(ilp, ul1, o->off3, 0, data, o);
-					val_store(ilp, ul2, o->off4, 0, data, o);
+					val_store(ullp, ull1, o->off3, 0, data, o);
+					val_store(ullp, ull2, o->off4, 0, data, o);
 				}
 			}
 			if (curr == 2) {
 				if (o->off5 && o->off6) {
-					val_store(ilp, ul1, o->off5, 0, data, o);
-					val_store(ilp, ul2, o->off6, 0, data, o);
+					val_store(ullp, ull1, o->off5, 0, data, o);
+					val_store(ullp, ull2, o->off6, 0, data, o);
 				}
 			}
 			if (!more) {
 				if (curr < 1) {
 					if (o->off3 && o->off4) {
-						val_store(ilp, ul1, o->off3, 0, data, o);
-						val_store(ilp, ul2, o->off4, 0, data, o);
+						val_store(ullp, ull1, o->off3, 0, data, o);
+						val_store(ullp, ull2, o->off4, 0, data, o);
 					}
 				}
 				if (curr < 2) {
 					if (o->off5 && o->off6) {
-						val_store(ilp, ul1, o->off5, 0, data, o);
-						val_store(ilp, ul2, o->off6, 0, data, o);
+						val_store(ullp, ull1, o->off5, 0, data, o);
+						val_store(ullp, ull2, o->off6, 0, data, o);
 					}
 				}
 			}
@@ -851,7 +878,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 			break;
 
 		if (o->maxval && il > (int) o->maxval) {
-			log_err("max value out of range: %d (%d max)\n",
+			log_err("max value out of range: %d (%llu max)\n",
 								il, o->maxval);
 			return 1;
 		}
@@ -1325,6 +1352,10 @@ static void option_init(struct fio_option *o)
 		if (!o->maxval)
 			o->maxval = UINT_MAX;
 	}
+	if (o->type == FIO_OPT_ULL) {
+		if (!o->maxval)
+			o->maxval = ULLONG_MAX;
+	}
 	if (o->type == FIO_OPT_STR_SET && o->def && !o->no_warn_def) {
 		log_err("Option %s: string set option with"
 				" default will always be true\n", o->name);
diff --git a/parse.h b/parse.h
index 4de5e77..083f283 100644
--- a/parse.h
+++ b/parse.h
@@ -10,12 +10,14 @@
 enum fio_opt_type {
 	FIO_OPT_INVALID = 0,
 	FIO_OPT_STR,
+	FIO_OPT_STR_ULL,
 	FIO_OPT_STR_MULTI,
 	FIO_OPT_STR_VAL,
 	FIO_OPT_STR_VAL_TIME,
 	FIO_OPT_STR_STORE,
 	FIO_OPT_RANGE,
 	FIO_OPT_INT,
+	FIO_OPT_ULL,
 	FIO_OPT_BOOL,
 	FIO_OPT_FLOAT_LIST,
 	FIO_OPT_STR_SET,
@@ -52,7 +54,7 @@ struct fio_option {
 	unsigned int off4;
 	unsigned int off5;
 	unsigned int off6;
-	unsigned int maxval;		/* max and min value */
+	unsigned long long maxval;		/* max and min value */
 	int minval;
 	double maxfp;			/* max and min floating value */
 	double minfp;
diff --git a/server.c b/server.c
index 7e7ffed..defc4d3 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/server.h b/server.h
index b48bbe1..37d2f76 100644
--- a/server.h
+++ b/server.h
@@ -48,7 +48,7 @@ struct fio_net_cmd_reply {
 };
 
 enum {
-	FIO_SERVER_VER			= 73,
+	FIO_SERVER_VER			= 74,
 
 	FIO_SERVER_MAX_FRAGMENT_PDU	= 1024,
 	FIO_SERVER_MAX_CMD_MB		= 2048,
diff --git a/stat.c b/stat.c
index a308eb8..8863349 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..306c93e 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;
diff --git a/verify.c b/verify.c
index 40d484b..3f6405a 100644
--- a/verify.c
+++ b/verify.c
@@ -801,7 +801,7 @@ static int verify_trimmed_io_u(struct thread_data *td, struct io_u *io_u)
 
 	mem_is_zero_slow(io_u->buf, io_u->buflen, &offset);
 
-	log_err("trim: verify failed at file %s offset %llu, length %lu"
+	log_err("trim: verify failed at file %s offset %llu, length %llu"
 		", block offset %lu\n",
 			io_u->file->file_name, io_u->offset, io_u->buflen,
 			(unsigned long) offset);

Re: fio max blocksize

[Jens Axboe]

On 7/13/18 5:59 PM, Jeff Furlong wrote:
> I fixed the buflen issue (just keep peeling the onion back and
> updating all associated vars), fixed bsrange, fixed "auto" tabs, and
> did some simple tests:
> # fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=4g --filename=/dev/nvme1n1 --number_ios=1
> # fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bs=1t --filename=/dev/nvme1n1 --number_ios=1
> # fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bsrange=4g-8g --filename=/dev/nvme1n1 --number_ios=2
> 
> The open issue I have is parsing bssplit.  I plumbed out bssplit.bs as
> a 64 bit value, but using the existing value_pair helpers (32 bit)
> seem to conflict.  Don’t really want to duplicate chunks of code, so
> open to ideas there.

I'd just convert those helpers to deal with unsigned long long always,
that should work fine for the uint case.

Just a note that I'll be out all of next week, so you'll have plenty of
time to make this perfect before I get back :-)

-- 
Jens Axboe

RE: fio max blocksize

[Jeff Furlong]

> I'd just convert those helpers to deal with unsigned long long always, that should work fine for the uint case.
Done.  I tested with:

# fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bssplit=4g/100 --filename=/dev/nvme1n1 --number_ios=2

Here's the final patch unless someone finds an issue.  I also updated it so it applies nicely to the latest on git.  Tested with bs, ba, bsrange, bssplit options that have a >32 bit value.

Regards,
Jeff


diff --git a/backend.c b/backend.c
index a7e9184..3c45e78 100644
--- a/backend.c
+++ b/backend.c
@@ -454,7 +454,7 @@ int io_queue_event(struct thread_data *td, struct io_u *io_u, int *ret,
 			*ret = -io_u->error;
 			clear_io_u(td, io_u);
 		} else if (io_u->resid) {
-			int bytes = io_u->xfer_buflen - io_u->resid;
+			long long bytes = io_u->xfer_buflen - io_u->resid;
 			struct fio_file *f = io_u->file;
 
 			if (bytes_issued)
@@ -583,7 +583,7 @@ static bool in_flight_overlap(struct io_u_queue *q, struct io_u *io_u)
 
 			if (x1 < y2 && y1 < x2) {
 				overlap = true;
-				dprint(FD_IO, "in-flight overlap: %llu/%lu, %llu/%lu\n",
+				dprint(FD_IO, "in-flight overlap: %llu/%llu, %llu/%llu\n",
 						x1, io_u->buflen,
 						y1, check_io_u->buflen);
 				break;
@@ -1033,7 +1033,7 @@ static void do_io(struct thread_data *td, uint64_t *bytes_done)
 			log_io_piece(td, io_u);
 
 		if (td->o.io_submit_mode == IO_MODE_OFFLOAD) {
-			const unsigned long blen = io_u->xfer_buflen;
+			const unsigned long long blen = io_u->xfer_buflen;
 			const enum fio_ddir __ddir = acct_ddir(io_u);
 
 			if (td->error)
@@ -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..534bfb0 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 3ac552b..685aab1 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 5221a78..c58dcf0 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;
@@ -503,19 +503,19 @@ static int get_next_offset(struct thread_data *td, struct io_u *io_u,
 }
 
 static inline bool io_u_fits(struct thread_data *td, struct io_u *io_u,
-			     unsigned int buflen)
+			     unsigned long long buflen)
 {
 	struct fio_file *f = io_u->file;
 
 	return io_u->offset + buflen <= f->io_size + get_start_offset(td, f);
 }
 
-static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u,
+static unsigned long long get_next_buflen(struct thread_data *td, struct io_u *io_u,
 				    bool is_random)
 {
 	int ddir = io_u->ddir;
-	unsigned int buflen = 0;
-	unsigned int minbs, maxbs;
+	unsigned long long buflen = 0;
+	unsigned long long minbs, maxbs;
 	uint64_t frand_max, r;
 	bool power_2;
 
@@ -541,7 +541,7 @@ static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u,
 		r = __rand(&td->bsrange_state[ddir]);
 
 		if (!td->o.bssplit_nr[ddir]) {
-			buflen = minbs + (unsigned int) ((double) maxbs *
+			buflen = minbs + (unsigned long long) ((double) maxbs *
 					(r / (frand_max + 1.0)));
 		} else {
 			long long perc = 0;
@@ -891,7 +891,7 @@ static int fill_io_u(struct thread_data *td, struct io_u *io_u)
 	}
 
 	if (io_u->offset + io_u->buflen > io_u->file->real_file_size) {
-		dprint(FD_IO, "io_u %p, off=0x%llx + len=0x%lx exceeds file size=0x%llx\n",
+		dprint(FD_IO, "io_u %p, off=0x%llx + len=0x%llx exceeds file size=0x%llx\n",
 			io_u,
 			(unsigned long long) io_u->offset, io_u->buflen,
 			(unsigned long long) io_u->file->real_file_size);
@@ -1582,7 +1582,7 @@ static bool check_get_verify(struct thread_data *td, struct io_u *io_u)
  */
 static void small_content_scramble(struct io_u *io_u)
 {
-	unsigned int i, nr_blocks = io_u->buflen >> 9;
+	unsigned long long i, nr_blocks = io_u->buflen >> 9;
 	unsigned int offset;
 	uint64_t boffset, *iptr;
 	char *p;
@@ -1726,7 +1726,7 @@ static void __io_u_log_error(struct thread_data *td, struct io_u *io_u)
 	if (td_non_fatal_error(td, eb, io_u->error) && !td->o.error_dump)
 		return;
 
-	log_err("fio: io_u error%s%s: %s: %s offset=%llu, buflen=%lu\n",
+	log_err("fio: io_u error%s%s: %s: %s offset=%llu, buflen=%llu\n",
 		io_u->file ? " on file " : "",
 		io_u->file ? io_u->file->file_name : "",
 		strerror(io_u->error),
@@ -1892,7 +1892,7 @@ static void io_completed(struct thread_data *td, struct io_u **io_u_ptr,
 	td->last_ddir = ddir;
 
 	if (!io_u->error && ddir_rw(ddir)) {
-		unsigned int bytes = io_u->buflen - io_u->resid;
+		unsigned long long bytes = io_u->buflen - io_u->resid;
 		int ret;
 
 		td->io_blocks[ddir]++;
@@ -2082,8 +2082,8 @@ 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;
 
@@ -2093,8 +2093,8 @@ 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);
@@ -2103,7 +2103,7 @@ 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,
@@ -2130,7 +2130,7 @@ 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..9a423b2 100644
--- a/io_u.h
+++ b/io_u.h
@@ -51,7 +51,7 @@ struct io_u {
 	/*
 	 * Allocated/set buffer and length
 	 */
-	unsigned long buflen;
+	unsigned long long buflen;
 	unsigned long long offset;
 	void *buf;
 
@@ -65,13 +65,13 @@ struct io_u {
 	 * partial transfers / residual data counts
 	 */
 	void *xfer_buf;
-	unsigned long xfer_buflen;
+	unsigned long long xfer_buflen;
 
 	/*
 	 * Parameter related to pre-filled buffers and
 	 * their size to handle variable block sizes.
 	 */
-	unsigned long buf_filled_len;
+	unsigned long long buf_filled_len;
 
 	struct io_piece *ipo;
 
@@ -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 *);
@@ -149,13 +149,13 @@ static inline void dprint_io_u(struct io_u *io_u, const char *p)
 	struct fio_file *f = io_u->file;
 
 	if (f)
-		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%lx,ddir=%d,file=%s\n",
+		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%llx,ddir=%d,file=%s\n",
 				p, io_u,
 				(unsigned long long) io_u->offset,
 				io_u->buflen, io_u->ddir,
 				f->file_name);
 	else
-		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%lx,ddir=%d\n",
+		dprint(FD_IO, "%s: io_u %p: off=0x%llx,len=0x%llx,ddir=%d\n",
 				p, io_u,
 				(unsigned long long) io_u->offset,
 				io_u->buflen, io_u->ddir);
diff --git a/ioengines.c b/ioengines.c
index d579682..bce65ea 100644
--- a/ioengines.c
+++ b/ioengines.c
@@ -279,7 +279,7 @@ out:
 enum fio_q_status td_io_queue(struct thread_data *td, struct io_u *io_u)
 {
 	const enum fio_ddir ddir = acct_ddir(io_u);
-	unsigned long buflen = io_u->xfer_buflen;
+	unsigned long long buflen = io_u->xfer_buflen;
 	enum fio_q_status ret;
 
 	dprint_io_u(io_u, "queue");
diff --git a/iolog.c b/iolog.c
index 5be3e84..d51e49c 100644
--- a/iolog.c
+++ b/iolog.c
@@ -35,7 +35,7 @@ void log_io_u(const struct thread_data *td, const struct io_u *io_u)
 	if (!td->o.write_iolog_file)
 		return;
 
-	fprintf(td->iolog_f, "%s %s %llu %lu\n", io_u->file->file_name,
+	fprintf(td->iolog_f, "%s %s %llu %llu\n", io_u->file->file_name,
 						io_ddir_name(io_u->ddir),
 						io_u->offset, io_u->buflen);
 }
@@ -161,7 +161,7 @@ int read_iolog_get(struct thread_data *td, struct io_u *io_u)
 			io_u->buflen = ipo->len;
 			io_u->file = td->files[ipo->fileno];
 			get_file(io_u->file);
-			dprint(FD_IO, "iolog: get %llu/%lu/%s\n", io_u->offset,
+			dprint(FD_IO, "iolog: get %llu/%llu/%s\n", io_u->offset,
 						io_u->buflen, io_u->file->file_name);
 			if (ipo->delay)
 				iolog_delay(td, ipo->delay);
@@ -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..4b46402 100644
--- a/options.c
+++ b/options.c
@@ -52,7 +52,7 @@ static int bs_cmp(const void *p1, const void *p2)
 
 struct split {
 	unsigned int nr;
-	unsigned int val1[ZONESPLIT_MAX];
+	unsigned long long val1[ZONESPLIT_MAX];
 	unsigned long long val2[ZONESPLIT_MAX];
 };
 
@@ -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));
@@ -2112,7 +2112,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 		.name	= "bs",
 		.lname	= "Block size",
 		.alias	= "blocksize",
-		.type	= FIO_OPT_INT,
+		.type	= FIO_OPT_ULL,
 		.off1	= offsetof(struct thread_options, bs[DDIR_READ]),
 		.off2	= offsetof(struct thread_options, bs[DDIR_WRITE]),
 		.off3	= offsetof(struct thread_options, bs[DDIR_TRIM]),
@@ -2129,7 +2129,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 		.name	= "ba",
 		.lname	= "Block size align",
 		.alias	= "blockalign",
-		.type	= FIO_OPT_INT,
+		.type	= FIO_OPT_ULL,
 		.off1	= offsetof(struct thread_options, ba[DDIR_READ]),
 		.off2	= offsetof(struct thread_options, ba[DDIR_WRITE]),
 		.off3	= offsetof(struct thread_options, ba[DDIR_TRIM]),
@@ -2163,7 +2163,7 @@ struct fio_option fio_options[FIO_MAX_OPTS] = {
 	{
 		.name	= "bssplit",
 		.lname	= "Block size split",
-		.type	= FIO_OPT_STR,
+		.type	= FIO_OPT_STR_ULL,
 		.cb	= str_bssplit_cb,
 		.off1	= offsetof(struct thread_options, bssplit),
 		.help	= "Set a specific mix of block sizes",
diff --git a/os/windows/posix.c b/os/windows/posix.c
old mode 100755
new mode 100644
diff --git a/parse.c b/parse.c
index 6261fca..200f530 100644
--- a/parse.c
+++ b/parse.c
@@ -26,12 +26,14 @@
 static const char *opt_type_names[] = {
 	"OPT_INVALID",
 	"OPT_STR",
+	"OPT_STR_ULL",
 	"OPT_STR_MULTI",
 	"OPT_STR_VAL",
 	"OPT_STR_VAL_TIME",
 	"OPT_STR_STORE",
 	"OPT_RANGE",
 	"OPT_INT",
+	"OPT_ULL",
 	"OPT_BOOL",
 	"OPT_FLOAT_LIST",
 	"OPT_STR_SET",
@@ -438,7 +440,7 @@ void strip_blank_end(char *p)
 	*(s + 1) = '\0';
 }
 
-static int check_range_bytes(const char *str, long *val, void *data)
+static int check_range_bytes(const char *str, long long *val, void *data)
 {
 	long long __val;
 
@@ -507,7 +509,8 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 	int il=0, *ilp;
 	fio_fp64_t *flp;
 	long long ull, *ullp;
-	long ul1, ul2;
+	long ul2;
+	long long ull1, ull2;
 	double uf;
 	char **cp = NULL;
 	int ret = 0, is_time = 0;
@@ -525,6 +528,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 
 	switch (o->type) {
 	case FIO_OPT_STR:
+	case FIO_OPT_STR_ULL:
 	case FIO_OPT_STR_MULTI: {
 		fio_opt_str_fn *fn = o->cb;
 
@@ -540,8 +544,14 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 				break;
 			if (!strncmp(vp->ival, ptr, str_match_len(vp, ptr))) {
 				ret = 0;
+				if (o->type == FIO_OPT_STR_ULL) {
+					if (o->off1)
+						val_store(ullp, vp->oval, o->off1, vp->orval, data, o);
+				}
+				else {
 				if (o->off1)
 					val_store(ilp, vp->oval, o->off1, vp->orval, data, o);
+				}
 				continue;
 			}
 		}
@@ -554,6 +564,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 	}
 	case FIO_OPT_STR_VAL_TIME:
 		is_time = 1;
+	case FIO_OPT_ULL:
 	case FIO_OPT_INT:
 	case FIO_OPT_STR_VAL: {
 		fio_opt_str_val_fn *fn = o->cb;
@@ -584,7 +595,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 
 		if (o->maxval && ull > o->maxval) {
 			log_err("max value out of range: %llu"
-					" (%u max)\n", ull, o->maxval);
+					" (%llu max)\n", ull, o->maxval);
 			return 1;
 		}
 		if (o->minval && ull < o->minval) {
@@ -636,6 +647,28 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 							val_store(ilp, ull, o->off3, 0, data, o);
 					}
 				}
+			}
+			else if (o->type == FIO_OPT_ULL) {
+				if (first)
+					val_store(ullp, ull, o->off1, 0, data, o);
+				if (curr == 1) {
+					if (o->off2)
+						val_store(ullp, ull, o->off2, 0, data, o);
+				}
+				if (curr == 2) {
+					if (o->off3)
+						val_store(ullp, ull, o->off3, 0, data, o);
+				}
+				if (!more) {
+					if (curr < 1) {
+						if (o->off2)
+							val_store(ullp, ull, o->off2, 0, data, o);
+					}
+					if (curr < 2) {
+						if (o->off3)
+							val_store(ullp, ull, o->off3, 0, data, o);
+					}
+				}
 			} else {
 				if (first)
 					val_store(ullp, ull, o->off1, 0, data, o);
@@ -790,43 +823,43 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 		p1 = tmp;
 
 		ret = 1;
-		if (!check_range_bytes(p1, &ul1, data) &&
-		    !check_range_bytes(p2, &ul2, data)) {
+		if (!check_range_bytes(p1, &ull1, data) &&
+			!check_range_bytes(p2, &ull2, data)) {
 			ret = 0;
-			if (ul1 > ul2) {
-				unsigned long foo = ul1;
+			if (ull1 > ull2) {
+				unsigned long long foo = ull1;
 
-				ul1 = ul2;
-				ul2 = foo;
+				ull1 = ull2;
+				ull2 = foo;
 			}
 
 			if (first) {
-				val_store(ilp, ul1, o->off1, 0, data, o);
-				val_store(ilp, ul2, o->off2, 0, data, o);
+				val_store(ullp, ull1, o->off1, 0, data, o);
+				val_store(ullp, ull2, o->off2, 0, data, o);
 			}
 			if (curr == 1) {
 				if (o->off3 && o->off4) {
-					val_store(ilp, ul1, o->off3, 0, data, o);
-					val_store(ilp, ul2, o->off4, 0, data, o);
+					val_store(ullp, ull1, o->off3, 0, data, o);
+					val_store(ullp, ull2, o->off4, 0, data, o);
 				}
 			}
 			if (curr == 2) {
 				if (o->off5 && o->off6) {
-					val_store(ilp, ul1, o->off5, 0, data, o);
-					val_store(ilp, ul2, o->off6, 0, data, o);
+					val_store(ullp, ull1, o->off5, 0, data, o);
+					val_store(ullp, ull2, o->off6, 0, data, o);
 				}
 			}
 			if (!more) {
 				if (curr < 1) {
 					if (o->off3 && o->off4) {
-						val_store(ilp, ul1, o->off3, 0, data, o);
-						val_store(ilp, ul2, o->off4, 0, data, o);
+						val_store(ullp, ull1, o->off3, 0, data, o);
+						val_store(ullp, ull2, o->off4, 0, data, o);
 					}
 				}
 				if (curr < 2) {
 					if (o->off5 && o->off6) {
-						val_store(ilp, ul1, o->off5, 0, data, o);
-						val_store(ilp, ul2, o->off6, 0, data, o);
+						val_store(ullp, ull1, o->off5, 0, data, o);
+						val_store(ullp, ull2, o->off6, 0, data, o);
 					}
 				}
 			}
@@ -851,7 +884,7 @@ static int __handle_option(const struct fio_option *o, const char *ptr,
 			break;
 
 		if (o->maxval && il > (int) o->maxval) {
-			log_err("max value out of range: %d (%d max)\n",
+			log_err("max value out of range: %d (%llu max)\n",
 								il, o->maxval);
 			return 1;
 		}
@@ -1325,6 +1358,10 @@ static void option_init(struct fio_option *o)
 		if (!o->maxval)
 			o->maxval = UINT_MAX;
 	}
+	if (o->type == FIO_OPT_ULL) {
+		if (!o->maxval)
+			o->maxval = ULLONG_MAX;
+	}
 	if (o->type == FIO_OPT_STR_SET && o->def && !o->no_warn_def) {
 		log_err("Option %s: string set option with"
 				" default will always be true\n", o->name);
diff --git a/parse.h b/parse.h
index 4de5e77..b47a02c 100644
--- a/parse.h
+++ b/parse.h
@@ -10,12 +10,14 @@
 enum fio_opt_type {
 	FIO_OPT_INVALID = 0,
 	FIO_OPT_STR,
+	FIO_OPT_STR_ULL,
 	FIO_OPT_STR_MULTI,
 	FIO_OPT_STR_VAL,
 	FIO_OPT_STR_VAL_TIME,
 	FIO_OPT_STR_STORE,
 	FIO_OPT_RANGE,
 	FIO_OPT_INT,
+	FIO_OPT_ULL,
 	FIO_OPT_BOOL,
 	FIO_OPT_FLOAT_LIST,
 	FIO_OPT_STR_SET,
@@ -29,7 +31,7 @@ enum fio_opt_type {
  */
 struct value_pair {
 	const char *ival;		/* string option */
-	unsigned int oval;		/* output value */
+	unsigned long long oval;/* output value */
 	const char *help;		/* help text for sub option */
 	int orval;			/* OR value */
 	void *cb;			/* sub-option callback */
@@ -52,7 +54,7 @@ struct fio_option {
 	unsigned int off4;
 	unsigned int off5;
 	unsigned int off6;
-	unsigned int maxval;		/* max and min value */
+	unsigned long long maxval;		/* max and min value */
 	int minval;
 	double maxfp;			/* max and min floating value */
 	double minfp;
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/server.h b/server.h
index b48bbe1..37d2f76 100644
--- a/server.h
+++ b/server.h
@@ -48,7 +48,7 @@ struct fio_net_cmd_reply {
 };
 
 enum {
-	FIO_SERVER_VER			= 73,
+	FIO_SERVER_VER			= 74,
 
 	FIO_SERVER_MAX_FRAGMENT_PDU	= 1024,
 	FIO_SERVER_MAX_CMD_MB		= 2048,
diff --git a/stat.c b/stat.c
index a308eb8..1677a9d 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..8adba48 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;
diff --git a/verify.c b/verify.c
index 40d484b..0f2c118 100644
--- a/verify.c
+++ b/verify.c
@@ -801,7 +801,7 @@ static int verify_trimmed_io_u(struct thread_data *td, struct io_u *io_u)
 
 	mem_is_zero_slow(io_u->buf, io_u->buflen, &offset);
 
-	log_err("trim: verify failed at file %s offset %llu, length %lu"
+	log_err("trim: verify failed at file %s offset %llu, length %llu"
 		", block offset %lu\n",
 			io_u->file->file_name, io_u->offset, io_u->buflen,
 			(unsigned long) offset);

Re: fio max blocksize

[Jens Axboe]

On 7/17/18 3:38 PM, Jeff Furlong wrote:
>> I'd just convert those helpers to deal with unsigned long long always, that should work fine for the uint case.
> Done.  I tested with:
> 
> # fio --name=test --ioengine=libaio --direct=1 --rw=trim --iodepth=1 --bssplit=4g/100 --filename=/dev/nvme1n1 --number_ios=2
> 
> Here's the final patch unless someone finds an issue.  I also updated
> it so it applies nicely to the latest on git.  Tested with bs, ba,
> bsrange, bssplit options that have a >32 bit value.

Looks pretty good to me, I applied it with various little fixups in
terms of style. Please test current -git and ensure that it works
properly for you. Ran some basic testing here, and it still seems good.

-- 
Jens Axboe

RE: fio max blocksize

[Jeff Furlong]

> Please test current -git and ensure that it works properly for you.
Ran latest git (fio-3.8-5-g464b) and tests pass.  Thanks.

Regards,
Jeff