X-Git-Url: https://git.kernel.dk/?a=blobdiff_plain;f=io_u.c;h=7df57aa804316a01efe1422e0d924dfa77b82f65;hb=f16b7405fff7c3fc1da421b6bdf8552cc99c3156;hp=f020cac38af38e5d2ebb150c059564f023b14b18;hpb=44cbc6daaae3674ea5d0a113b66596ca24f372e0;p=fio.git

diff --git a/io_u.c b/io_u.c
index f020cac3..7df57aa8 100644
--- a/io_u.c
+++ b/io_u.c
@@ -16,7 +16,7 @@ struct io_completion_data {
 	int nr;				/* input */
 
 	int error;			/* output */
-	unsigned long bytes_done[DDIR_RWDIR_CNT];	/* output */
+	uint64_t bytes_done[DDIR_RWDIR_CNT];	/* output */
 	struct timeval time;		/* output */
 };
 
@@ -101,7 +101,7 @@ static int __get_next_rand_offset(struct thread_data *td, struct fio_file *f,
 			r = __rand(&td->__random_state);
 		}
 
-		dprint(FD_RANDOM, "off rand %llu\n", r);
+		dprint(FD_RANDOM, "off rand %llu\n", (unsigned long long) r);
 
 		*b = (lastb - 1) * (r / ((uint64_t) rmax + 1.0));
 	} else {
@@ -125,7 +125,8 @@ static int __get_next_rand_offset(struct thread_data *td, struct fio_file *f,
 	if (random_map_free(f, *b))
 		goto ret;
 
-	dprint(FD_RANDOM, "get_next_rand_offset: offset %llu busy\n", *b);
+	dprint(FD_RANDOM, "get_next_rand_offset: offset %llu busy\n",
+						(unsigned long long) *b);
 
 	*b = axmap_next_free(f->io_axmap, *b);
 	if (*b == (uint64_t) -1ULL)
@@ -172,19 +173,50 @@ static int get_off_from_method(struct thread_data *td, struct fio_file *f,
 	return 1;
 }
 
+/*
+ * Sort the reads for a verify phase in batches of verifysort_nr, if
+ * specified.
+ */
+static inline int should_sort_io(struct thread_data *td)
+{
+	if (!td->o.verifysort_nr || !td->o.do_verify)
+		return 0;
+	if (!td_random(td))
+		return 0;
+	if (td->runstate != TD_VERIFYING)
+		return 0;
+	if (td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE)
+		return 0;
+
+	return 1;
+}
+
+static int should_do_random(struct thread_data *td, enum fio_ddir ddir)
+{
+	unsigned int v;
+	unsigned long r;
+
+	if (td->o.perc_rand[ddir] == 100)
+		return 1;
+
+	if (td->o.use_os_rand) {
+		r = os_random_long(&td->seq_rand_state[ddir]);
+		v = 1 + (int) (100.0 * (r / (OS_RAND_MAX + 1.0)));
+	} else {
+		r = __rand(&td->__seq_rand_state[ddir]);
+		v = 1 + (int) (100.0 * (r / (FRAND_MAX + 1.0)));
+	}
+
+	return v <= td->o.perc_rand[ddir];
+}
+
 static int get_next_rand_offset(struct thread_data *td, struct fio_file *f,
 				enum fio_ddir ddir, uint64_t *b)
 {
 	struct rand_off *r;
 	int i, ret = 1;
 
-	/*
-	 * If sort not enabled, or not a pure random read workload without
-	 * any stored write metadata, just return a random offset
-	 */
-	if (!td->o.verifysort_nr || !(ddir == DDIR_READ && td->o.do_verify &&
-	    td->o.verify != VERIFY_NONE && td_random(td)) ||
-	    td->o.random_generator == FIO_RAND_GEN_TAUSWORTHE)
+	if (!should_sort_io(td))
 		return get_off_from_method(td, f, ddir, b);
 
 	if (!flist_empty(&td->next_rand_list)) {
@@ -230,7 +262,8 @@ static int get_next_rand_block(struct thread_data *td, struct fio_file *f,
 	}
 
 	dprint(FD_IO, "%s: rand offset failed, last=%llu, size=%llu\n",
-			f->file_name, f->last_pos, f->real_file_size);
+			f->file_name, (unsigned long long) f->last_pos,
+			(unsigned long long) f->real_file_size);
 	return 1;
 }
 
@@ -260,7 +293,8 @@ static int get_next_seq_offset(struct thread_data *td, struct fio_file *f,
 }
 
 static int get_next_block(struct thread_data *td, struct io_u *io_u,
-			  enum fio_ddir ddir, int rw_seq)
+			  enum fio_ddir ddir, int rw_seq,
+			  unsigned int *is_random)
 {
 	struct fio_file *f = io_u->file;
 	uint64_t b, offset;
@@ -271,17 +305,31 @@ static int get_next_block(struct thread_data *td, struct io_u *io_u,
 	b = offset = -1ULL;
 
 	if (rw_seq) {
-		if (td_random(td))
-			ret = get_next_rand_block(td, f, ddir, &b);
-		else
+		if (td_random(td)) {
+			if (should_do_random(td, ddir)) {
+				ret = get_next_rand_block(td, f, ddir, &b);
+				*is_random = 1;
+			} else {
+				*is_random = 0;
+				io_u->flags |= IO_U_F_BUSY_OK;
+				ret = get_next_seq_offset(td, f, ddir, &offset);
+				if (ret)
+					ret = get_next_rand_block(td, f, ddir, &b);
+			}
+		} else {
+			*is_random = 0;
 			ret = get_next_seq_offset(td, f, ddir, &offset);
+		}
 	} else {
 		io_u->flags |= IO_U_F_BUSY_OK;
+		*is_random = 0;
 
 		if (td->o.rw_seq == RW_SEQ_SEQ) {
 			ret = get_next_seq_offset(td, f, ddir, &offset);
-			if (ret)
+			if (ret) {
 				ret = get_next_rand_block(td, f, ddir, &b);
+				*is_random = 0;
+			}
 		} else if (td->o.rw_seq == RW_SEQ_IDENT) {
 			if (f->last_start != -1ULL)
 				offset = f->last_start - f->file_offset;
@@ -300,8 +348,7 @@ static int get_next_block(struct thread_data *td, struct io_u *io_u,
 		else if (b != -1ULL)
 			io_u->offset = b * td->o.ba[ddir];
 		else {
-			log_err("fio: bug in offset generation: offset=%llu, b=%llu\n",
-								offset, b);
+			log_err("fio: bug in offset generation: offset=%llu, b=%llu\n", (unsigned long long) offset, (unsigned long long) b);
 			ret = 1;
 		}
 	}
@@ -314,7 +361,8 @@ static int get_next_block(struct thread_data *td, struct io_u *io_u,
  * until we find a free one. For sequential io, just return the end of
  * the last io issued.
  */
-static int __get_next_offset(struct thread_data *td, struct io_u *io_u)
+static int __get_next_offset(struct thread_data *td, struct io_u *io_u,
+			     unsigned int *is_random)
 {
 	struct fio_file *f = io_u->file;
 	enum fio_ddir ddir = io_u->ddir;
@@ -327,35 +375,38 @@ static int __get_next_offset(struct thread_data *td, struct io_u *io_u)
 		td->ddir_seq_nr = td->o.ddir_seq_nr;
 	}
 
-	if (get_next_block(td, io_u, ddir, rw_seq_hit))
+	if (get_next_block(td, io_u, ddir, rw_seq_hit, is_random))
 		return 1;
 
 	if (io_u->offset >= f->io_size) {
 		dprint(FD_IO, "get_next_offset: offset %llu >= io_size %llu\n",
-					io_u->offset, f->io_size);
+					(unsigned long long) io_u->offset,
+					(unsigned long long) f->io_size);
 		return 1;
 	}
 
 	io_u->offset += f->file_offset;
 	if (io_u->offset >= f->real_file_size) {
 		dprint(FD_IO, "get_next_offset: offset %llu >= size %llu\n",
-					io_u->offset, f->real_file_size);
+					(unsigned long long) io_u->offset,
+					(unsigned long long) f->real_file_size);
 		return 1;
 	}
 
 	return 0;
 }
 
-static int get_next_offset(struct thread_data *td, struct io_u *io_u)
+static int get_next_offset(struct thread_data *td, struct io_u *io_u,
+			   unsigned int *is_random)
 {
 	if (td->flags & TD_F_PROFILE_OPS) {
 		struct prof_io_ops *ops = &td->prof_io_ops;
 
 		if (ops->fill_io_u_off)
-			return ops->fill_io_u_off(td, io_u);
+			return ops->fill_io_u_off(td, io_u, is_random);
 	}
 
-	return __get_next_offset(td, io_u);
+	return __get_next_offset(td, io_u, is_random);
 }
 
 static inline int io_u_fits(struct thread_data *td, struct io_u *io_u,
@@ -366,14 +417,20 @@ static inline int io_u_fits(struct thread_data *td, struct io_u *io_u,
 	return io_u->offset + buflen <= f->io_size + get_start_offset(td);
 }
 
-static unsigned int __get_next_buflen(struct thread_data *td, struct io_u *io_u)
+static unsigned int __get_next_buflen(struct thread_data *td, struct io_u *io_u,
+				      unsigned int is_random)
 {
-	const int ddir = io_u->ddir;
+	int ddir = io_u->ddir;
 	unsigned int buflen = 0;
 	unsigned int minbs, maxbs;
 	unsigned long r, rand_max;
 
-	assert(ddir_rw(ddir));
+	assert(ddir_rw(io_u->ddir));
+
+	if (td->o.bs_is_seq_rand)
+		ddir = is_random ? DDIR_WRITE: DDIR_READ;
+	else
+		ddir = io_u->ddir;
 
 	minbs = td->o.min_bs[ddir];
 	maxbs = td->o.max_bs[ddir];
@@ -418,6 +475,10 @@ static unsigned int __get_next_buflen(struct thread_data *td, struct io_u *io_u)
 			}
 		}
 
+		if (td->o.do_verify && td->o.verify != VERIFY_NONE)
+			buflen = (buflen + td->o.verify_interval - 1) &
+				~(td->o.verify_interval - 1);
+
 		if (!td->o.bs_unaligned && is_power_of_2(minbs))
 			buflen = (buflen + minbs - 1) & ~(minbs - 1);
 
@@ -426,16 +487,17 @@ static unsigned int __get_next_buflen(struct thread_data *td, struct io_u *io_u)
 	return buflen;
 }
 
-static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u)
+static unsigned int get_next_buflen(struct thread_data *td, struct io_u *io_u,
+				    unsigned int is_random)
 {
 	if (td->flags & TD_F_PROFILE_OPS) {
 		struct prof_io_ops *ops = &td->prof_io_ops;
 
 		if (ops->fill_io_u_size)
-			return ops->fill_io_u_size(td, io_u);
+			return ops->fill_io_u_size(td, io_u, is_random);
 	}
 
-	return __get_next_buflen(td, io_u);
+	return __get_next_buflen(td, io_u, is_random);
 }
 
 static void set_rwmix_bytes(struct thread_data *td)
@@ -470,6 +532,24 @@ static inline enum fio_ddir get_rand_ddir(struct thread_data *td)
 	return DDIR_WRITE;
 }
 
+void io_u_quiesce(struct thread_data *td)
+{
+	/*
+	 * We are going to sleep, ensure that we flush anything pending as
+	 * not to skew our latency numbers.
+	 *
+	 * Changed to only monitor 'in flight' requests here instead of the
+	 * td->cur_depth, b/c td->cur_depth does not accurately represent
+	 * io's that have been actually submitted to an async engine,
+	 * and cur_depth is meaningless for sync engines.
+	 */
+	while (td->io_u_in_flight) {
+		int fio_unused ret;
+
+		ret = io_u_queued_complete(td, 1, NULL);
+	}
+}
+
 static enum fio_ddir rate_ddir(struct thread_data *td, enum fio_ddir ddir)
 {
 	enum fio_ddir odir = ddir ^ 1;
@@ -485,7 +565,7 @@ static enum fio_ddir rate_ddir(struct thread_data *td, enum fio_ddir ddir)
 	 * We have too much pending sleep in this direction. See if we
 	 * should switch.
 	 */
-	if (td_rw(td)) {
+	if (td_rw(td) && td->o.rwmix[odir]) {
 		/*
 		 * Other direction does not have too much pending, switch
 		 */
@@ -506,20 +586,7 @@ static enum fio_ddir rate_ddir(struct thread_data *td, enum fio_ddir ddir)
 	} else
 		usec = td->rate_pending_usleep[ddir];
 
-	/*
-	 * We are going to sleep, ensure that we flush anything pending as
-	 * not to skew our latency numbers.
-	 *
-	 * Changed to only monitor 'in flight' requests here instead of the
-	 * td->cur_depth, b/c td->cur_depth does not accurately represent
-	 * io's that have been actually submitted to an async engine,
-	 * and cur_depth is meaningless for sync engines.
-	 */
-	if (td->io_u_in_flight) {
-		int fio_unused ret;
-
-		ret = io_u_queued_complete(td, td->io_u_in_flight, NULL);
-	}
+	io_u_quiesce(td);
 
 	fio_gettime(&t, NULL);
 	usec_sleep(td, usec);
@@ -533,6 +600,7 @@ static enum fio_ddir rate_ddir(struct thread_data *td, enum fio_ddir ddir)
 
 	if (ddir_trim(ddir))
 		return ddir;
+
 	return ddir;
 }
 
@@ -545,12 +613,6 @@ static enum fio_ddir get_rw_ddir(struct thread_data *td)
 {
 	enum fio_ddir ddir;
 
-	/*
-	 * If verify phase started, it's always a READ
-	 */
-	if (td->runstate == TD_VERIFYING)
-		return DDIR_READ;
-
 	/*
 	 * see if it's time to fsync
 	 */
@@ -606,7 +668,7 @@ static enum fio_ddir get_rw_ddir(struct thread_data *td)
 
 static void set_rw_ddir(struct thread_data *td, struct io_u *io_u)
 {
-	io_u->ddir = get_rw_ddir(td);
+	io_u->ddir = io_u->acct_ddir = get_rw_ddir(td);
 
 	if (io_u->ddir == DDIR_WRITE && (td->io_ops->flags & FIO_BARRIER) &&
 	    td->o.barrier_blocks &&
@@ -635,8 +697,7 @@ void put_io_u(struct thread_data *td, struct io_u *io_u)
 
 	if (io_u->flags & IO_U_F_IN_CUR_DEPTH)
 		td->cur_depth--;
-	flist_del_init(&io_u->list);
-	flist_add(&io_u->list, &td->io_u_freelist);
+	io_u_qpush(&td->io_u_freelist, io_u);
 	td_io_u_unlock(td);
 	td_io_u_free_notify(td);
 }
@@ -650,26 +711,29 @@ void clear_io_u(struct thread_data *td, struct io_u *io_u)
 void requeue_io_u(struct thread_data *td, struct io_u **io_u)
 {
 	struct io_u *__io_u = *io_u;
+	enum fio_ddir ddir = acct_ddir(__io_u);
 
 	dprint(FD_IO, "requeue %p\n", __io_u);
 
 	td_io_u_lock(td);
 
 	__io_u->flags |= IO_U_F_FREE;
-	if ((__io_u->flags & IO_U_F_FLIGHT) && ddir_rw(__io_u->ddir))
-		td->io_issues[__io_u->ddir]--;
+	if ((__io_u->flags & IO_U_F_FLIGHT) && ddir_rw(ddir))
+		td->io_issues[ddir]--;
 
 	__io_u->flags &= ~IO_U_F_FLIGHT;
 	if (__io_u->flags & IO_U_F_IN_CUR_DEPTH)
 		td->cur_depth--;
-	flist_del(&__io_u->list);
-	flist_add_tail(&__io_u->list, &td->io_u_requeues);
+
+	io_u_rpush(&td->io_u_requeues, __io_u);
 	td_io_u_unlock(td);
 	*io_u = NULL;
 }
 
 static int fill_io_u(struct thread_data *td, struct io_u *io_u)
 {
+	unsigned int is_random;
+
 	if (td->io_ops->flags & FIO_NOIO)
 		goto out;
 
@@ -695,12 +759,12 @@ static int fill_io_u(struct thread_data *td, struct io_u *io_u)
 	 * No log, let the seq/rand engine retrieve the next buflen and
 	 * position.
 	 */
-	if (get_next_offset(td, io_u)) {
+	if (get_next_offset(td, io_u, &is_random)) {
 		dprint(FD_IO, "io_u %p, failed getting offset\n", io_u);
 		return 1;
 	}
 
-	io_u->buflen = get_next_buflen(td, io_u);
+	io_u->buflen = get_next_buflen(td, io_u, is_random);
 	if (!io_u->buflen) {
 		dprint(FD_IO, "io_u %p, failed getting buflen\n", io_u);
 		return 1;
@@ -708,8 +772,9 @@ 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, offset too large\n", io_u);
-		dprint(FD_IO, "  off=%llu/%lu > %llu\n", io_u->offset,
-				io_u->buflen, io_u->file->real_file_size);
+		dprint(FD_IO, "  off=%llu/%lu > %llu\n",
+			(unsigned long long) io_u->offset, io_u->buflen,
+			(unsigned long long) io_u->file->real_file_size);
 		return 1;
 	}
 
@@ -719,13 +784,9 @@ static int fill_io_u(struct thread_data *td, struct io_u *io_u)
 	if (td_random(td) && file_randommap(td, io_u->file))
 		mark_random_map(td, io_u);
 
-	/*
-	 * If using a write iolog, store this entry.
-	 */
 out:
 	dprint_io_u(io_u, "fill_io_u");
 	td->zone_bytes += io_u->buflen;
-	log_io_u(td, io_u);
 	return 0;
 }
 
@@ -1068,15 +1129,15 @@ static int set_io_u_file(struct thread_data *td, struct io_u *io_u)
 
 struct io_u *__get_io_u(struct thread_data *td)
 {
-	struct io_u *io_u = NULL;
+	struct io_u *io_u;
 
 	td_io_u_lock(td);
 
 again:
-	if (!flist_empty(&td->io_u_requeues))
-		io_u = flist_entry(td->io_u_requeues.next, struct io_u, list);
-	else if (!queue_full(td)) {
-		io_u = flist_entry(td->io_u_freelist.next, struct io_u, list);
+	if (!io_u_rempty(&td->io_u_requeues))
+		io_u = io_u_rpop(&td->io_u_requeues);
+	else if (!io_u_qempty(&td->io_u_freelist)) {
+		io_u = io_u_qpop(&td->io_u_freelist);
 
 		io_u->buflen = 0;
 		io_u->resid = 0;
@@ -1091,8 +1152,7 @@ again:
 		io_u->flags &= ~IO_U_F_VER_LIST;
 
 		io_u->error = 0;
-		flist_del(&io_u->list);
-		flist_add_tail(&io_u->list, &td->io_u_busylist);
+		io_u->acct_ddir = -1;
 		td->cur_depth++;
 		io_u->flags |= IO_U_F_IN_CUR_DEPTH;
 	} else if (td->o.verify_async) {
@@ -1328,6 +1388,13 @@ static void account_io_completion(struct thread_data *td, struct io_u *io_u,
 		tusec = utime_since(&io_u->start_time, &icd->time);
 		add_lat_sample(td, idx, tusec, bytes);
 
+		if (td->flags & TD_F_PROFILE_OPS) {
+			struct prof_io_ops *ops = &td->prof_io_ops;
+
+			if (ops->io_u_lat)
+				icd->error = ops->io_u_lat(td, tusec);
+		}
+
 		if (td->o.max_latency && tusec > td->o.max_latency) {
 			if (!td->error)
 				log_err("fio: latency of %lu usec exceeds specified max (%u usec)\n", tusec, td->o.max_latency);
@@ -1344,7 +1411,10 @@ static void account_io_completion(struct thread_data *td, struct io_u *io_u,
 	if (!td->o.disable_bw)
 		add_bw_sample(td, idx, bytes, &icd->time);
 
-	add_iops_sample(td, idx, &icd->time);
+	add_iops_sample(td, idx, bytes, &icd->time);
+
+	if (td->o.number_ios && !--td->o.number_ios)
+		td->done = 1;
 }
 
 static long long usec_for_io(struct thread_data *td, enum fio_ddir ddir)
@@ -1489,7 +1559,7 @@ static void ios_completed(struct thread_data *td,
  * Complete a single io_u for the sync engines.
  */
 int io_u_sync_complete(struct thread_data *td, struct io_u *io_u,
-		       unsigned long *bytes)
+		       uint64_t *bytes)
 {
 	struct io_completion_data icd;
 
@@ -1518,7 +1588,7 @@ int io_u_sync_complete(struct thread_data *td, struct io_u *io_u,
  * Called to complete min_events number of io for the async engines.
  */
 int io_u_queued_complete(struct thread_data *td, int min_evts,
-			 unsigned long *bytes)
+			 uint64_t *bytes)
 {
 	struct io_completion_data icd;
 	struct timespec *tvp = NULL;
@@ -1567,14 +1637,9 @@ void io_u_queued(struct thread_data *td, struct io_u *io_u)
 	}
 }
 
-/*
- * "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)
+void fill_io_buffer(struct thread_data *td, void *buf, unsigned int min_write,
+		    unsigned int max_bs)
 {
-	io_u->buf_filled_len = 0;
-
 	if (!td->o.zero_buffers) {
 		unsigned int perc = td->o.compress_percentage;
 
@@ -1582,10 +1647,23 @@ void io_u_fill_buffer(struct thread_data *td, struct io_u *io_u,
 			unsigned int seg = min_write;
 
 			seg = min(min_write, td->o.compress_chunk);
-			fill_random_buf_percentage(&td->buf_state, io_u->buf,
+			if (!seg)
+				seg = min_write;
+
+			fill_random_buf_percentage(&td->buf_state, buf,
 						perc, seg, max_bs);
 		} else
-			fill_random_buf(&td->buf_state, io_u->buf, max_bs);
+			fill_random_buf(&td->buf_state, buf, max_bs);
 	} else
-		memset(io_u->buf, 0, max_bs);
+		memset(buf, 0, max_bs);
+}
+
+/*
+ * "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)
+{
+	io_u->buf_filled_len = 0;
+	fill_io_buffer(td, io_u->buf, min_write, max_bs);
 }