| Message ID | 20230731221458.437440-3-bvanassche@acm.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Improve the performance for zoned UFS devices | expand |
On 8/1/23 07:14, Bart Van Assche wrote: > Measurements have shown that limiting the queue depth to one per zone for > zoned writes has a significant negative performance impact on zoned UFS > devices. Hence this patch that disables zone locking by the mq-deadline > scheduler if the storage controller preserves the command order. This > patch is based on the following assumptions: > - It happens infrequently that zoned write requests are reordered by the > block layer. > - The I/O priority of all write requests is the same per zone. > - Either no I/O scheduler is used or an I/O scheduler is used that > submits write requests per zone in LBA order. > > Cc: Christoph Hellwig <hch@lst.de> > Cc: Damien Le Moal <dlemoal@kernel.org> > Cc: Ming Lei <ming.lei@redhat.com> > Signed-off-by: Bart Van Assche <bvanassche@acm.org> Looks OK. Reviewed-by: Damien Le Moal <dlemoal@kernel.org> > --- > block/mq-deadline.c | 24 ++++++++++++++++++------ > 1 file changed, 18 insertions(+), 6 deletions(-) > > diff --git a/block/mq-deadline.c b/block/mq-deadline.c > index 02a916ba62ee..1f4124dd4a0b 100644 > --- a/block/mq-deadline.c > +++ b/block/mq-deadline.c > @@ -338,6 +338,16 @@ static struct request *deadline_skip_seq_writes(struct deadline_data *dd, > return rq; > } > > +/* > + * Use write locking if either QUEUE_FLAG_NO_ZONE_WRITE_LOCK has not been set. > + * Not using zone write locking is only safe if the block driver preserves the > + * request order. > + */ > +static bool dd_use_zone_write_locking(struct request_queue *q) > +{ > + return blk_queue_is_zoned(q) && !blk_queue_no_zone_write_lock(q); > +} > + > /* > * For the specified data direction, return the next request to > * dispatch using arrival ordered lists. > @@ -353,7 +363,7 @@ deadline_fifo_request(struct deadline_data *dd, struct dd_per_prio *per_prio, > return NULL; > > rq = rq_entry_fifo(per_prio->fifo_list[data_dir].next); > - if (data_dir == DD_READ || !blk_queue_is_zoned(rq->q)) > + if (data_dir == DD_READ || !dd_use_zone_write_locking(rq->q)) > return rq; > > /* > @@ -398,7 +408,7 @@ deadline_next_request(struct deadline_data *dd, struct dd_per_prio *per_prio, > if (!rq) > return NULL; > > - if (data_dir == DD_READ || !blk_queue_is_zoned(rq->q)) > + if (data_dir == DD_READ || !dd_use_zone_write_locking(rq->q)) > return rq; > > /* > @@ -526,8 +536,9 @@ static struct request *__dd_dispatch_request(struct deadline_data *dd, > } > > /* > - * For a zoned block device, if we only have writes queued and none of > - * them can be dispatched, rq will be NULL. > + * For a zoned block device that requires write serialization, if we > + * only have writes queued and none of them can be dispatched, rq will > + * be NULL. > */ > if (!rq) > return NULL; > @@ -552,7 +563,8 @@ static struct request *__dd_dispatch_request(struct deadline_data *dd, > /* > * If the request needs its target zone locked, do it. > */ > - blk_req_zone_write_lock(rq); > + if (dd_use_zone_write_locking(rq->q)) > + blk_req_zone_write_lock(rq); > rq->rq_flags |= RQF_STARTED; > return rq; > } > @@ -933,7 +945,7 @@ static void dd_finish_request(struct request *rq) > > atomic_inc(&per_prio->stats.completed); > > - if (blk_queue_is_zoned(q)) { > + if (dd_use_zone_write_locking(rq->q)) { > unsigned long flags; > > spin_lock_irqsave(&dd->zone_lock, flags);
diff --git a/block/mq-deadline.c b/block/mq-deadline.c index 02a916ba62ee..1f4124dd4a0b 100644 --- a/block/mq-deadline.c +++ b/block/mq-deadline.c @@ -338,6 +338,16 @@ static struct request *deadline_skip_seq_writes(struct deadline_data *dd, return rq; } +/* + * Use write locking if either QUEUE_FLAG_NO_ZONE_WRITE_LOCK has not been set. + * Not using zone write locking is only safe if the block driver preserves the + * request order. + */ +static bool dd_use_zone_write_locking(struct request_queue *q) +{ + return blk_queue_is_zoned(q) && !blk_queue_no_zone_write_lock(q); +} + /* * For the specified data direction, return the next request to * dispatch using arrival ordered lists. @@ -353,7 +363,7 @@ deadline_fifo_request(struct deadline_data *dd, struct dd_per_prio *per_prio, return NULL; rq = rq_entry_fifo(per_prio->fifo_list[data_dir].next); - if (data_dir == DD_READ || !blk_queue_is_zoned(rq->q)) + if (data_dir == DD_READ || !dd_use_zone_write_locking(rq->q)) return rq; /* @@ -398,7 +408,7 @@ deadline_next_request(struct deadline_data *dd, struct dd_per_prio *per_prio, if (!rq) return NULL; - if (data_dir == DD_READ || !blk_queue_is_zoned(rq->q)) + if (data_dir == DD_READ || !dd_use_zone_write_locking(rq->q)) return rq; /* @@ -526,8 +536,9 @@ static struct request *__dd_dispatch_request(struct deadline_data *dd, } /* - * For a zoned block device, if we only have writes queued and none of - * them can be dispatched, rq will be NULL. + * For a zoned block device that requires write serialization, if we + * only have writes queued and none of them can be dispatched, rq will + * be NULL. */ if (!rq) return NULL; @@ -552,7 +563,8 @@ static struct request *__dd_dispatch_request(struct deadline_data *dd, /* * If the request needs its target zone locked, do it. */ - blk_req_zone_write_lock(rq); + if (dd_use_zone_write_locking(rq->q)) + blk_req_zone_write_lock(rq); rq->rq_flags |= RQF_STARTED; return rq; } @@ -933,7 +945,7 @@ static void dd_finish_request(struct request *rq) atomic_inc(&per_prio->stats.completed); - if (blk_queue_is_zoned(q)) { + if (dd_use_zone_write_locking(rq->q)) { unsigned long flags; spin_lock_irqsave(&dd->zone_lock, flags);
Measurements have shown that limiting the queue depth to one per zone for zoned writes has a significant negative performance impact on zoned UFS devices. Hence this patch that disables zone locking by the mq-deadline scheduler if the storage controller preserves the command order. This patch is based on the following assumptions: - It happens infrequently that zoned write requests are reordered by the block layer. - The I/O priority of all write requests is the same per zone. - Either no I/O scheduler is used or an I/O scheduler is used that submits write requests per zone in LBA order. Cc: Christoph Hellwig <hch@lst.de> Cc: Damien Le Moal <dlemoal@kernel.org> Cc: Ming Lei <ming.lei@redhat.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> --- block/mq-deadline.c | 24 ++++++++++++++++++------ 1 file changed, 18 insertions(+), 6 deletions(-)