return odir;
/*
- * Both directions are ahead of rate. sleep the min
- * switch if necissary
+ * Both directions are ahead of rate. sleep the min,
+ * switch if necessary
*/
if (td->rate_next_io_time[ddir] <=
- td->rate_next_io_time[odir]) {
+ td->rate_next_io_time[odir]) {
usec = td->rate_next_io_time[ddir] - now;
} else {
usec = td->rate_next_io_time[odir] - now;
if (td->o.io_submit_mode == IO_MODE_INLINE)
io_u_quiesce(td);
- usec = usec_sleep(td, usec);
-
+ usec_sleep(td, usec);
return ddir;
}
*io_u = NULL;
}
+static void __fill_io_u_zone(struct thread_data *td, struct io_u *io_u)
+{
+ struct fio_file *f = io_u->file;
+
+ /*
+ * See if it's time to switch to a new zone
+ */
+ if (td->zone_bytes >= td->o.zone_size && td->o.zone_skip) {
+ td->zone_bytes = 0;
+ f->file_offset += td->o.zone_range + td->o.zone_skip;
+
+ /*
+ * Wrap from the beginning, if we exceed the file size
+ */
+ if (f->file_offset >= f->real_file_size)
+ f->file_offset = f->real_file_size - f->file_offset;
+ f->last_pos[io_u->ddir] = f->file_offset;
+ td->io_skip_bytes += td->o.zone_skip;
+ }
+
+ /*
+ * If zone_size > zone_range, then maintain the same zone until
+ * zone_bytes >= zone_size.
+ */
+ if (f->last_pos[io_u->ddir] >= (f->file_offset + td->o.zone_range)) {
+ dprint(FD_IO, "io_u maintain zone offset=%" PRIu64 "/last_pos=%" PRIu64 "\n",
+ f->file_offset, f->last_pos[io_u->ddir]);
+ f->last_pos[io_u->ddir] = f->file_offset;
+ }
+
+ /*
+ * For random: if 'norandommap' is not set and zone_size > zone_range,
+ * map needs to be reset as it's done with zone_range everytime.
+ */
+ if ((td->zone_bytes % td->o.zone_range) == 0) {
+ fio_file_reset(td, f);
+ }
+}
+
static int fill_io_u(struct thread_data *td, struct io_u *io_u)
{
unsigned int is_random;
goto out;
/*
- * See if it's time to switch to a new zone
+ * When file is zoned zone_range is always positive
*/
- if (td->zone_bytes >= td->o.zone_size && td->o.zone_skip) {
- struct fio_file *f = io_u->file;
-
- td->zone_bytes = 0;
- f->file_offset += td->o.zone_range + td->o.zone_skip;
-
- /*
- * Wrap from the beginning, if we exceed the file size
- */
- if (f->file_offset >= f->real_file_size)
- f->file_offset = f->real_file_size - f->file_offset;
- f->last_pos[io_u->ddir] = f->file_offset;
- td->io_skip_bytes += td->o.zone_skip;
+ if (td->o.zone_range) {
+ __fill_io_u_zone(td, io_u);
}
/*
}
if (io_u->offset + io_u->buflen > io_u->file->real_file_size) {
- dprint(FD_IO, "io_u %p, offset + buflen exceeds file size\n",
- io_u);
- dprint(FD_IO, " offset=%llu/buflen=%lu > %llu\n",
+ dprint(FD_IO, "io_u %p, off=0x%llx + len=0x%lx 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);
return 1;
mark_random_map(td, io_u);
out:
- dprint_io_u(io_u, "fill_io_u");
+ dprint_io_u(io_u, "fill");
td->zone_bytes += io_u->buflen;
return 0;
}
}
static void lat_fatal(struct thread_data *td, struct io_completion_data *icd,
- unsigned long tusec, unsigned long max_usec)
+ unsigned long long tnsec, unsigned long long max_nsec)
{
if (!td->error)
- log_err("fio: latency of %lu usec exceeds specified max (%lu usec)\n", tusec, max_usec);
+ log_err("fio: latency of %llu nsec exceeds specified max (%llu nsec)\n", tnsec, max_nsec);
td_verror(td, ETIMEDOUT, "max latency exceeded");
icd->error = ETIMEDOUT;
}
{
if (!(td->flags & TD_F_TRIM_BACKLOG))
return false;
+ if (!td->trim_entries)
+ return false;
- if (td->trim_entries) {
- int get_trim = 0;
-
- if (td->trim_batch) {
- td->trim_batch--;
- get_trim = 1;
- } else if (!(td->io_hist_len % td->o.trim_backlog) &&
- td->last_ddir != DDIR_READ) {
- td->trim_batch = td->o.trim_batch;
- if (!td->trim_batch)
- td->trim_batch = td->o.trim_backlog;
- get_trim = 1;
- }
-
- if (get_trim && get_next_trim(td, io_u))
+ if (td->trim_batch) {
+ td->trim_batch--;
+ if (get_next_trim(td, io_u))
+ return true;
+ } else if (!(td->io_hist_len % td->o.trim_backlog) &&
+ td->last_ddir != DDIR_READ) {
+ td->trim_batch = td->o.trim_batch;
+ if (!td->trim_batch)
+ td->trim_batch = td->o.trim_backlog;
+ if (get_next_trim(td, io_u))
return true;
}
*/
static void small_content_scramble(struct io_u *io_u)
{
- unsigned int i, nr_blocks = io_u->buflen / 512;
- uint64_t boffset;
+ unsigned int i, nr_blocks = io_u->buflen >> 9;
unsigned int offset;
- char *p, *end;
+ uint64_t boffset, *iptr;
+ char *p;
if (!nr_blocks)
return;
p = io_u->xfer_buf;
boffset = io_u->offset;
- io_u->buf_filled_len = 0;
+
+ if (io_u->buf_filled_len)
+ io_u->buf_filled_len = 0;
+
+ /*
+ * Generate random index between 0..7. We do chunks of 512b, if
+ * we assume a cacheline is 64 bytes, then we have 8 of those.
+ * Scramble content within the blocks in the same cacheline to
+ * speed things up.
+ */
+ offset = (io_u->start_time.tv_nsec ^ boffset) & 7;
for (i = 0; i < nr_blocks; i++) {
/*
- * Fill the byte offset into a "random" start offset of
- * the buffer, given by the product of the usec time
- * and the actual offset.
+ * Fill offset into start of cacheline, time into end
+ * of cacheline
*/
- offset = ((io_u->start_time.tv_nsec/1000) ^ boffset) & 511;
- offset &= ~(sizeof(uint64_t) - 1);
- if (offset >= 512 - sizeof(uint64_t))
- offset -= sizeof(uint64_t);
- memcpy(p + offset, &boffset, sizeof(boffset));
-
- end = p + 512 - sizeof(io_u->start_time);
- memcpy(end, &io_u->start_time, sizeof(io_u->start_time));
+ iptr = (void *) p + (offset << 6);
+ *iptr = boffset;
+
+ iptr = (void *) p + 64 - 2 * sizeof(uint64_t);
+ iptr[0] = io_u->start_time.tv_sec;
+ iptr[1] = io_u->start_time.tv_nsec;
+
p += 512;
boffset += 512;
}
struct prof_io_ops *ops = &td->prof_io_ops;
if (ops->io_u_lat)
- icd->error = ops->io_u_lat(td, tnsec/1000);
+ icd->error = ops->io_u_lat(td, tnsec);
}
- if (td->o.max_latency && tnsec/1000 > td->o.max_latency)
- lat_fatal(td, icd, tnsec/1000, td->o.max_latency);
- if (td->o.latency_target && tnsec/1000 > td->o.latency_target) {
+ if (td->o.max_latency && tnsec > td->o.max_latency)
+ lat_fatal(td, icd, tnsec, td->o.max_latency);
+ if (td->o.latency_target && tnsec > td->o.latency_target) {
if (lat_target_failed(td))
- lat_fatal(td, icd, tnsec/1000, td->o.latency_target);
+ lat_fatal(td, icd, tnsec, td->o.latency_target);
}
}
if (no_reduce && per_unit_log(td->iops_log))
add_iops_sample(td, io_u, bytes);
- }
+ } else if (ddir_sync(idx) && !td->o.disable_clat)
+ add_sync_clat_sample(&td->ts, llnsec);
if (td->ts.nr_block_infos && io_u->ddir == DDIR_TRIM) {
uint32_t *info = io_u_block_info(td, io_u);
f->last_write_idx = 0;
}
+static bool should_account(struct thread_data *td)
+{
+ return ramp_time_over(td) && (td->runstate == TD_RUNNING ||
+ td->runstate == TD_VERIFYING);
+}
+
static void io_completed(struct thread_data *td, struct io_u **io_u_ptr,
struct io_completion_data *icd)
{
enum fio_ddir ddir = io_u->ddir;
struct fio_file *f = io_u->file;
- dprint_io_u(io_u, "io complete");
+ dprint_io_u(io_u, "complete");
assert(io_u->flags & IO_U_F_FLIGHT);
io_u_clear(td, io_u, IO_U_F_FLIGHT | IO_U_F_BUSY_OK);
}
if (ddir_sync(ddir)) {
- td->last_was_sync = 1;
+ td->last_was_sync = true;
if (f) {
f->first_write = -1ULL;
f->last_write = -1ULL;
}
+ if (should_account(td))
+ account_io_completion(td, io_u, icd, ddir, io_u->buflen);
return;
}
- td->last_was_sync = 0;
+ td->last_was_sync = false;
td->last_ddir = ddir;
if (!io_u->error && ddir_rw(ddir)) {
int ret;
td->io_blocks[ddir]++;
- td->this_io_blocks[ddir]++;
td->io_bytes[ddir] += bytes;
- if (!(io_u->flags & IO_U_F_VER_LIST))
+ if (!(io_u->flags & IO_U_F_VER_LIST)) {
+ td->this_io_blocks[ddir]++;
td->this_io_bytes[ddir] += bytes;
+ }
if (ddir == DDIR_WRITE)
file_log_write_comp(td, f, io_u->offset, bytes);
- if (ramp_time_over(td) && (td->runstate == TD_RUNNING ||
- td->runstate == TD_VERIFYING))
+ if (should_account(td))
account_io_completion(td, io_u, icd, ddir, bytes);
icd->bytes_done[ddir] += bytes;
projects / fio.git / blobdiff