On Thu, 2020-08-13 at 13:57 +0900, Shin'ichiro Kawasaki wrote:
> When fio starts write workloads to zones with open status, fio does not
> reflect the zone status to open zone list. This results in inconsistent
> open zone accounting.
>
> To avoid this inconsistency, initialize the open zone list referring the
> zone status by calling zbd_open_zone() function before workload start.
> Since io_u is not available at that timing, make the function
> independent from io_u.
>
> Of note is that fio counts open zones within the write target range.
> Open zones out of the range are not counted or checked.
>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@xxxxxxx>
> ---
> zbd.c | 21 ++++++++++++++++-----
> 1 file changed, 16 insertions(+), 5 deletions(-)
>
> diff --git a/zbd.c b/zbd.c
> index 20f64b58..123d5530 100644
> --- a/zbd.c
> +++ b/zbd.c
> @@ -627,6 +627,9 @@ static int zbd_init_zone_info(struct thread_data *td, struct fio_file *file)
> return ret;
> }
>
> +static bool zbd_open_zone(struct thread_data *td, const struct fio_file *f,
> + uint32_t zone_idx);
> +
> int zbd_setup_files(struct thread_data *td)
> {
> struct fio_file *f;
> @@ -650,6 +653,8 @@ int zbd_setup_files(struct thread_data *td)
>
> for_each_file(td, f, i) {
> struct zoned_block_device_info *zbd = f->zbd_info;
> + struct fio_zone_info *z;
> + int zi;
>
> if (!zbd)
> continue;
> @@ -665,6 +670,13 @@ int zbd_setup_files(struct thread_data *td)
> log_err("'max_open_zones' value is limited by %u\n", ZBD_MAX_OPEN_ZONES);
> return 1;
> }
> +
> + for (zi = f->min_zone; zi < f->max_zone; zi++) {
> + z = &zbd->zone_info[zi];
> + if (z->cond == ZBD_ZONE_COND_IMP_OPEN ||
> + z->cond == ZBD_ZONE_COND_EXP_OPEN)
> + zbd_open_zone(td, f, zi);
> + }
What if zbd->max_open_zones is set less than the number of the currently
open zones in [max_zone:min_zone] range? In this case, the loop above will
open more zones than needed. Maybe this part should be changed to something
like the code below?
+ int zi, noz;
...
+ for (zi = f->min_zone, noz = 0; zi < f->max_zone; zi++) {
+ z = &zbd->zone_info[zi];
+ if (z->cond == ZBD_ZONE_COND_IMP_OPEN ||
+ z->cond == ZBD_ZONE_COND_EXP_OPEN) {
+ if (noz < zbd->max_open_zones)
+ zbd_open_zone(td, f, zi);
+ else
+ zbd_reset_zone(td, f, zi);
+ noz++;
+ }
+ }
> }
>
> return 0;
> @@ -934,11 +946,10 @@ static bool is_zone_open(const struct thread_data *td, const struct fio_file *f,
> * was not yet open and opening a new zone would cause the zone limit to be
> * exceeded.
> */
> -static bool zbd_open_zone(struct thread_data *td, const struct io_u *io_u,
> +static bool zbd_open_zone(struct thread_data *td, const struct fio_file *f,
> uint32_t zone_idx)
> {
> const uint32_t min_bs = td->o.min_bs[DDIR_WRITE];
> - const struct fio_file *f = io_u->file;
> struct fio_zone_info *z = &f->zbd_info->zone_info[zone_idx];
> bool res = true;
>
> @@ -1126,7 +1137,7 @@ open_other_zone:
> zone_lock(td, f, z);
> if (z->open)
> continue;
> - if (zbd_open_zone(td, io_u, zone_idx))
> + if (zbd_open_zone(td, f, zone_idx))
> goto out;
> }
>
> @@ -1169,7 +1180,7 @@ static struct fio_zone_info *zbd_replay_write_order(struct thread_data *td,
> const struct fio_file *f = io_u->file;
> const uint32_t min_bs = td->o.min_bs[DDIR_WRITE];
>
> - if (!zbd_open_zone(td, io_u, z - f->zbd_info->zone_info)) {
> + if (!zbd_open_zone(td, f, z - f->zbd_info->zone_info)) {
> pthread_mutex_unlock(&z->mutex);
> z = zbd_convert_to_open_zone(td, io_u);
> assert(z);
> @@ -1581,7 +1592,7 @@ enum io_u_action zbd_adjust_block(struct thread_data *td, struct io_u *io_u)
> case DDIR_WRITE:
> if (io_u->buflen > f->zbd_info->zone_size)
> goto eof;
> - if (!zbd_open_zone(td, io_u, zone_idx_b)) {
> + if (!zbd_open_zone(td, f, zone_idx_b)) {
> pthread_mutex_unlock(&zb->mutex);
> zb = zbd_convert_to_open_zone(td, io_u);
> if (!zb)
