On 2020/05/23 0:39, Hannes Reinecke wrote:
> Random and sequential zones should be part of the respective
> device structure to make arbitration between devices possible.
>
> Signed-off-by: Hannes Reinecke <hare@xxxxxxx>
> ---
> drivers/md/dm-zoned-metadata.c | 143 +++++++++++++++++++++++++----------------
> drivers/md/dm-zoned.h | 10 +++
> 2 files changed, 99 insertions(+), 54 deletions(-)
>
> diff --git a/drivers/md/dm-zoned-metadata.c b/drivers/md/dm-zoned-metadata.c
> index 1b9da698a812..5f44970a6187 100644
> --- a/drivers/md/dm-zoned-metadata.c
> +++ b/drivers/md/dm-zoned-metadata.c
> @@ -192,21 +192,12 @@ struct dmz_metadata {
> /* Zone allocation management */
> struct mutex map_lock;
> struct dmz_mblock **map_mblk;
> - unsigned int nr_rnd;
> - atomic_t unmap_nr_rnd;
> - struct list_head unmap_rnd_list;
> - struct list_head map_rnd_list;
>
> unsigned int nr_cache;
> atomic_t unmap_nr_cache;
> struct list_head unmap_cache_list;
> struct list_head map_cache_list;
>
> - unsigned int nr_seq;
> - atomic_t unmap_nr_seq;
> - struct list_head unmap_seq_list;
> - struct list_head map_seq_list;
> -
> atomic_t nr_reserved_seq_zones;
> struct list_head reserved_seq_zones_list;
>
> @@ -281,12 +272,22 @@ unsigned int dmz_nr_chunks(struct dmz_metadata *zmd)
>
> unsigned int dmz_nr_rnd_zones(struct dmz_metadata *zmd)
> {
> - return zmd->nr_rnd;
> + unsigned int nr_rnd_zones = 0;
> + int i;
> +
> + for (i = 0; i < zmd->nr_devs; i++)
> + nr_rnd_zones += zmd->dev[i].nr_rnd;
We could keep the total nr_rnd_zones in dmz_metadata to avoid this one since the
value will never change at run time.
> + return nr_rnd_zones;
> }
>
> unsigned int dmz_nr_unmap_rnd_zones(struct dmz_metadata *zmd)
> {
> - return atomic_read(&zmd->unmap_nr_rnd);
> + unsigned int nr_unmap_rnd_zones = 0;
> + int i;
> +
> + for (i = 0; i < zmd->nr_devs; i++)
> + nr_unmap_rnd_zones += atomic_read(&zmd->dev[i].unmap_nr_rnd);
> + return nr_unmap_rnd_zones;
> }
>
> unsigned int dmz_nr_cache_zones(struct dmz_metadata *zmd)
> @@ -301,12 +302,22 @@ unsigned int dmz_nr_unmap_cache_zones(struct dmz_metadata *zmd)
>
> unsigned int dmz_nr_seq_zones(struct dmz_metadata *zmd)
> {
> - return zmd->nr_seq;
> + unsigned int nr_seq_zones = 0;
> + int i;
> +
> + for (i = 0; i < zmd->nr_devs; i++)
> + nr_seq_zones += zmd->dev[i].nr_seq;
Same here. This value does not change at runtime.
> + return nr_seq_zones;
> }
>
> unsigned int dmz_nr_unmap_seq_zones(struct dmz_metadata *zmd)
> {
> - return atomic_read(&zmd->unmap_nr_seq);
> + unsigned int nr_unmap_seq_zones = 0;
> + int i;
> +
> + for (i = 0; i < zmd->nr_devs; i++)
> + nr_unmap_seq_zones += atomic_read(&zmd->dev[i].unmap_nr_seq);
> + return nr_unmap_seq_zones;
> }
>
> static struct dm_zone *dmz_get(struct dmz_metadata *zmd, unsigned int zone_id)
> @@ -1485,6 +1496,14 @@ static int dmz_init_zones(struct dmz_metadata *zmd)
>
> dev->metadata = zmd;
> zmd->nr_zones += dev->nr_zones;
> +
> + atomic_set(&dev->unmap_nr_rnd, 0);
> + INIT_LIST_HEAD(&dev->unmap_rnd_list);
> + INIT_LIST_HEAD(&dev->map_rnd_list);
> +
> + atomic_set(&dev->unmap_nr_seq, 0);
> + INIT_LIST_HEAD(&dev->unmap_seq_list);
> + INIT_LIST_HEAD(&dev->map_seq_list);
> }
>
> if (!zmd->nr_zones) {
> @@ -1702,9 +1721,9 @@ static int dmz_load_mapping(struct dmz_metadata *zmd)
> if (dmz_is_cache(dzone))
> list_add_tail(&dzone->link, &zmd->map_cache_list);
> else if (dmz_is_rnd(dzone))
> - list_add_tail(&dzone->link, &zmd->map_rnd_list);
> + list_add_tail(&dzone->link, &dzone->dev->map_rnd_list);
> else
> - list_add_tail(&dzone->link, &zmd->map_seq_list);
> + list_add_tail(&dzone->link, &dzone->dev->map_seq_list);
>
> /* Check buffer zone */
> bzone_id = le32_to_cpu(dmap[e].bzone_id);
> @@ -1738,7 +1757,7 @@ static int dmz_load_mapping(struct dmz_metadata *zmd)
> if (dmz_is_cache(bzone))
> list_add_tail(&bzone->link, &zmd->map_cache_list);
> else
> - list_add_tail(&bzone->link, &zmd->map_rnd_list);
> + list_add_tail(&bzone->link, &bzone->dev->map_rnd_list);
> next:
> chunk++;
> e++;
> @@ -1763,9 +1782,9 @@ static int dmz_load_mapping(struct dmz_metadata *zmd)
> if (dmz_is_cache(dzone))
> zmd->nr_cache++;
> else if (dmz_is_rnd(dzone))
> - zmd->nr_rnd++;
> + dzone->dev->nr_rnd++;
> else
> - zmd->nr_seq++;
> + dzone->dev->nr_seq++;
>
> if (dmz_is_data(dzone)) {
> /* Already initialized */
> @@ -1779,16 +1798,18 @@ static int dmz_load_mapping(struct dmz_metadata *zmd)
> list_add_tail(&dzone->link, &zmd->unmap_cache_list);
> atomic_inc(&zmd->unmap_nr_cache);
> } else if (dmz_is_rnd(dzone)) {
> - list_add_tail(&dzone->link, &zmd->unmap_rnd_list);
> - atomic_inc(&zmd->unmap_nr_rnd);
> + list_add_tail(&dzone->link,
> + &dzone->dev->unmap_rnd_list);
> + atomic_inc(&dzone->dev->unmap_nr_rnd);
> } else if (atomic_read(&zmd->nr_reserved_seq_zones) < zmd->nr_reserved_seq) {
> list_add_tail(&dzone->link, &zmd->reserved_seq_zones_list);
> set_bit(DMZ_RESERVED, &dzone->flags);
> atomic_inc(&zmd->nr_reserved_seq_zones);
> - zmd->nr_seq--;
> + dzone->dev->nr_seq--;
> } else {
> - list_add_tail(&dzone->link, &zmd->unmap_seq_list);
> - atomic_inc(&zmd->unmap_nr_seq);
> + list_add_tail(&dzone->link,
> + &dzone->dev->unmap_seq_list);
> + atomic_inc(&dzone->dev->unmap_nr_seq);
> }
> }
>
> @@ -1822,13 +1843,13 @@ static void __dmz_lru_zone(struct dmz_metadata *zmd, struct dm_zone *zone)
> list_del_init(&zone->link);
> if (dmz_is_seq(zone)) {
> /* LRU rotate sequential zone */
> - list_add_tail(&zone->link, &zmd->map_seq_list);
> + list_add_tail(&zone->link, &zone->dev->map_seq_list);
> } else if (dmz_is_cache(zone)) {
> /* LRU rotate cache zone */
> list_add_tail(&zone->link, &zmd->map_cache_list);
> } else {
> /* LRU rotate random zone */
> - list_add_tail(&zone->link, &zmd->map_rnd_list);
> + list_add_tail(&zone->link, &zone->dev->map_rnd_list);
> }
> }
>
> @@ -1910,14 +1931,24 @@ static struct dm_zone *dmz_get_rnd_zone_for_reclaim(struct dmz_metadata *zmd,
> {
> struct dm_zone *dzone = NULL;
> struct dm_zone *zone;
> - struct list_head *zone_list = &zmd->map_rnd_list;
> + struct list_head *zone_list;
>
> /* If we have cache zones select from the cache zone list */
> if (zmd->nr_cache) {
> zone_list = &zmd->map_cache_list;
> /* Try to relaim random zones, too, when idle */
> - if (idle && list_empty(zone_list))
> - zone_list = &zmd->map_rnd_list;
> + if (idle && list_empty(zone_list)) {
> + int i;
> +
> + for (i = 1; i < zmd->nr_devs; i++) {
> + zone_list = &zmd->dev[i].map_rnd_list;
> + if (!list_empty(zone_list))
> + break;
> + }
This is going to use the first zoned dev until it has no more random zones, then
switch to the next zoned dev. What about going round-robin on the devices to
increase parallelism between the drives ?
> + }
> + } else {
> + /* Otherwise the random zones are on the first disk */
> + zone_list = &zmd->dev[0].map_rnd_list;
> }
>
> list_for_each_entry(zone, zone_list, link) {
> @@ -1938,12 +1969,17 @@ static struct dm_zone *dmz_get_rnd_zone_for_reclaim(struct dmz_metadata *zmd,
> static struct dm_zone *dmz_get_seq_zone_for_reclaim(struct dmz_metadata *zmd)
> {
> struct dm_zone *zone;
> + int i;
>
> - list_for_each_entry(zone, &zmd->map_seq_list, link) {
> - if (!zone->bzone)
> - continue;
> - if (dmz_lock_zone_reclaim(zone))
> - return zone;
> + for (i = 0; i < zmd->nr_devs; i++) {
> + struct dmz_dev *dev = &zmd->dev[i];
> +
> + list_for_each_entry(zone, &dev->map_seq_list, link) {
> + if (!zone->bzone)
> + continue;
> + if (dmz_lock_zone_reclaim(zone))
> + return zone;
> + }
Same comment here.
> }
>
> return NULL;
> @@ -2129,7 +2165,7 @@ struct dm_zone *dmz_get_chunk_buffer(struct dmz_metadata *zmd,
> if (dmz_is_cache(bzone))
> list_add_tail(&bzone->link, &zmd->map_cache_list);
> else
> - list_add_tail(&bzone->link, &zmd->map_rnd_list);
> + list_add_tail(&bzone->link, &bzone->dev->map_rnd_list);
> out:
> dmz_unlock_map(zmd);
>
> @@ -2144,21 +2180,27 @@ struct dm_zone *dmz_alloc_zone(struct dmz_metadata *zmd, unsigned long flags)
> {
> struct list_head *list;
> struct dm_zone *zone;
> + unsigned int dev_idx = 0;
>
> +again:
> if (flags & DMZ_ALLOC_CACHE)
> list = &zmd->unmap_cache_list;
> else if (flags & DMZ_ALLOC_RND)
> - list = &zmd->unmap_rnd_list;
> + list = &zmd->dev[dev_idx].unmap_rnd_list;
> else
> - list = &zmd->unmap_seq_list;
> + list = &zmd->dev[dev_idx].unmap_seq_list;
>
> -again:
> if (list_empty(list)) {
> /*
> * No free zone: return NULL if this is for not reclaim.
> */
> if (!(flags & DMZ_ALLOC_RECLAIM))
> return NULL;
> + if (dev_idx < zmd->nr_devs) {
> + dev_idx++;
> + goto again;
> + }
> +
> /*
> * Fallback to the reserved sequential zones
> */
> @@ -2177,9 +2219,9 @@ struct dm_zone *dmz_alloc_zone(struct dmz_metadata *zmd, unsigned long flags)
> if (dmz_is_cache(zone))
> atomic_dec(&zmd->unmap_nr_cache);
> else if (dmz_is_rnd(zone))
> - atomic_dec(&zmd->unmap_nr_rnd);
> + atomic_dec(&zone->dev->unmap_nr_rnd);
> else
> - atomic_dec(&zmd->unmap_nr_seq);
> + atomic_dec(&zone->dev->unmap_nr_seq);
>
> if (dmz_is_offline(zone)) {
> dmz_zmd_warn(zmd, "Zone %u is offline", zone->id);
> @@ -2209,14 +2251,14 @@ void dmz_free_zone(struct dmz_metadata *zmd, struct dm_zone *zone)
> list_add_tail(&zone->link, &zmd->unmap_cache_list);
> atomic_inc(&zmd->unmap_nr_cache);
> } else if (dmz_is_rnd(zone)) {
> - list_add_tail(&zone->link, &zmd->unmap_rnd_list);
> - atomic_inc(&zmd->unmap_nr_rnd);
> + list_add_tail(&zone->link, &zone->dev->unmap_rnd_list);
> + atomic_inc(&zone->dev->unmap_nr_rnd);
> } else if (dmz_is_reserved(zone)) {
> list_add_tail(&zone->link, &zmd->reserved_seq_zones_list);
> atomic_inc(&zmd->nr_reserved_seq_zones);
> } else {
> - list_add_tail(&zone->link, &zmd->unmap_seq_list);
> - atomic_inc(&zmd->unmap_nr_seq);
> + list_add_tail(&zone->link, &zone->dev->unmap_seq_list);
> + atomic_inc(&zone->dev->unmap_nr_seq);
> }
>
> wake_up_all(&zmd->free_wq);
> @@ -2236,9 +2278,9 @@ void dmz_map_zone(struct dmz_metadata *zmd, struct dm_zone *dzone,
> if (dmz_is_cache(dzone))
> list_add_tail(&dzone->link, &zmd->map_cache_list);
> else if (dmz_is_rnd(dzone))
> - list_add_tail(&dzone->link, &zmd->map_rnd_list);
> + list_add_tail(&dzone->link, &dzone->dev->map_rnd_list);
> else
> - list_add_tail(&dzone->link, &zmd->map_seq_list);
> + list_add_tail(&dzone->link, &dzone->dev->map_seq_list);
> }
>
> /*
> @@ -2806,18 +2848,11 @@ int dmz_ctr_metadata(struct dmz_dev *dev, int num_dev,
> INIT_LIST_HEAD(&zmd->mblk_dirty_list);
>
> mutex_init(&zmd->map_lock);
> - atomic_set(&zmd->unmap_nr_rnd, 0);
> - INIT_LIST_HEAD(&zmd->unmap_rnd_list);
> - INIT_LIST_HEAD(&zmd->map_rnd_list);
>
> atomic_set(&zmd->unmap_nr_cache, 0);
> INIT_LIST_HEAD(&zmd->unmap_cache_list);
> INIT_LIST_HEAD(&zmd->map_cache_list);
>
> - atomic_set(&zmd->unmap_nr_seq, 0);
> - INIT_LIST_HEAD(&zmd->unmap_seq_list);
> - INIT_LIST_HEAD(&zmd->map_seq_list);
> -
> atomic_set(&zmd->nr_reserved_seq_zones, 0);
> INIT_LIST_HEAD(&zmd->reserved_seq_zones_list);
>
> @@ -2887,9 +2922,9 @@ int dmz_ctr_metadata(struct dmz_dev *dev, int num_dev,
> dmz_zmd_debug(zmd, " %u cache zones (%u unmapped)",
> zmd->nr_cache, atomic_read(&zmd->unmap_nr_cache));
> dmz_zmd_debug(zmd, " %u random zones (%u unmapped)",
> - zmd->nr_rnd, atomic_read(&zmd->unmap_nr_rnd));
> + dmz_nr_rnd_zones(zmd), dmz_nr_unmap_rnd_zones(zmd));
> dmz_zmd_debug(zmd, " %u sequential zones (%u unmapped)",
> - zmd->nr_seq, atomic_read(&zmd->unmap_nr_seq));
> + dmz_nr_seq_zones(zmd), dmz_nr_unmap_seq_zones(zmd));
> dmz_zmd_debug(zmd, " %u reserved sequential data zones",
> zmd->nr_reserved_seq);
> dmz_zmd_debug(zmd, "Format:");
> diff --git a/drivers/md/dm-zoned.h b/drivers/md/dm-zoned.h
> index 983f5b5e9fa0..56e138586d9b 100644
> --- a/drivers/md/dm-zoned.h
> +++ b/drivers/md/dm-zoned.h
> @@ -66,6 +66,16 @@ struct dmz_dev {
> unsigned int flags;
>
> sector_t zone_nr_sectors;
> +
> + unsigned int nr_rnd;
> + atomic_t unmap_nr_rnd;
> + struct list_head unmap_rnd_list;
> + struct list_head map_rnd_list;
> +
> + unsigned int nr_seq;
> + atomic_t unmap_nr_seq;
> + struct list_head unmap_seq_list;
> + struct list_head map_seq_list;
> };
>
> #define dmz_bio_chunk(zmd, bio) ((bio)->bi_iter.bi_sector >> \
>
--
Damien Le Moal
Western Digital Research
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel
