Introduce zone write locking to avoid write request reordering with
zoned block devices. This is achieved using a finer selection of the
next request to dispatch:
1) Any non-write request is always allowed to proceed.
2) Any write to a conventional zone is always allowed to proceed.
3) For a write to a sequential zone, the zone lock is first checked.
a) If the zone is not locked, the write is allowed to proceed after
its target zone is locked.
b) If the zone is locked, the write request is skipped and the next
request in the dispatch queue tested (back to step 1).
For a write request that has locked its target zone, the zone is
unlocked either when the request completes with a call to the method
deadline_request_completed() or when the request is requeued using
dd_insert_request().
Requests targeting a locked zone are always left in the scheduler queue
to preserve the lba ordering for write requests. If no write request
can be dispatched, allow reads to be dispatched even if the write batch
is not done.
If the device used is not a zoned block device, or if zoned block device
support is disabled, this patch does not modify mq-deadline behavior.
Signed-off-by: Damien Le Moal <damien.lemoal@xxxxxxx>
Reviewed-by: Christoph Hellwig <hch@xxxxxx>
Reviewed-by: Martin K. Petersen <martin.petersen@xxxxxxxxxx>
---
block/mq-deadline.c | 89 +++++++++++++++++++++++++++++++++++++++++++++++++++--
1 file changed, 86 insertions(+), 3 deletions(-)
diff --git a/block/mq-deadline.c b/block/mq-deadline.c
index 8bd6db9e69c7..d56972e8ebda 100644
--- a/block/mq-deadline.c
+++ b/block/mq-deadline.c
@@ -59,6 +59,7 @@ struct deadline_data {
int front_merges;
spinlock_t lock;
+ spinlock_t zone_lock;
struct list_head dispatch;
};
@@ -198,13 +199,33 @@ static inline int deadline_check_fifo(struct deadline_data *dd, int ddir)
static struct request *
deadline_fifo_request(struct deadline_data *dd, int data_dir)
{
+ struct request *rq;
+ unsigned long flags;
+
if (WARN_ON_ONCE(data_dir != READ && data_dir != WRITE))
return NULL;
if (list_empty(&dd->fifo_list[data_dir]))
return NULL;
- return rq_entry_fifo(dd->fifo_list[data_dir].next);
+ rq = rq_entry_fifo(dd->fifo_list[data_dir].next);
+ if (data_dir == READ || !blk_queue_is_zoned(rq->q))
+ return rq;
+
+ /*
+ * Look for a write request that can be dispatched, that is one with
+ * an unlocked target zone.
+ */
+ spin_lock_irqsave(&dd->zone_lock, flags);
+ list_for_each_entry(rq, &dd->fifo_list[WRITE], queuelist) {
+ if (blk_req_can_dispatch_to_zone(rq))
+ goto out;
+ }
+ rq = NULL;
+out:
+ spin_unlock_irqrestore(&dd->zone_lock, flags);
+
+ return rq;
}
/*
@@ -214,10 +235,32 @@ deadline_fifo_request(struct deadline_data *dd, int data_dir)
static struct request *
deadline_next_request(struct deadline_data *dd, int data_dir)
{
+ struct request *rq;
+ unsigned long flags;
+
if (WARN_ON_ONCE(data_dir != READ && data_dir != WRITE))
return NULL;
- return dd->next_rq[data_dir];
+ rq = dd->next_rq[data_dir];
+ if (!rq)
+ return NULL;
+
+ if (data_dir == READ || !blk_queue_is_zoned(rq->q))
+ return rq;
+
+ /*
+ * Look for a write request that can be dispatched, that is one with
+ * an unlocked target zone.
+ */
+ spin_lock_irqsave(&dd->zone_lock, flags);
+ while (rq) {
+ if (blk_req_can_dispatch_to_zone(rq))
+ break;
+ rq = deadline_latter_request(rq);
+ }
+ spin_unlock_irqrestore(&dd->zone_lock, flags);
+
+ return rq;
}
/*
@@ -259,7 +302,8 @@ static struct request *__dd_dispatch_request(struct blk_mq_hw_ctx *hctx)
if (reads) {
BUG_ON(RB_EMPTY_ROOT(&dd->sort_list[READ]));
- if (writes && (dd->starved++ >= dd->writes_starved))
+ if (deadline_fifo_request(dd, WRITE) &&
+ (dd->starved++ >= dd->writes_starved))
goto dispatch_writes;
data_dir = READ;
@@ -304,6 +348,13 @@ static struct request *__dd_dispatch_request(struct blk_mq_hw_ctx *hctx)
rq = next_rq;
}
+ /*
+ * For a zoned block device, if we only have writes queued and none of
+ * them can be dispatched, rq will be NULL.
+ */
+ if (!rq)
+ return NULL;
+
dd->batching = 0;
dispatch_request:
@@ -313,6 +364,10 @@ static struct request *__dd_dispatch_request(struct blk_mq_hw_ctx *hctx)
dd->batching++;
deadline_move_request(dd, rq);
done:
+ /*
+ * If the request needs its target zone locked, do it.
+ */
+ blk_req_zone_write_lock(rq);
rq->rq_flags |= RQF_STARTED;
return rq;
}
@@ -368,6 +423,7 @@ static int dd_init_queue(struct request_queue *q, struct elevator_type *e)
dd->front_merges = 1;
dd->fifo_batch = fifo_batch;
spin_lock_init(&dd->lock);
+ spin_lock_init(&dd->zone_lock);
INIT_LIST_HEAD(&dd->dispatch);
q->elevator = eq;
@@ -424,6 +480,12 @@ static void dd_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
struct deadline_data *dd = q->elevator->elevator_data;
const int data_dir = rq_data_dir(rq);
+ /*
+ * This may be a requeue of a write request that has locked its
+ * target zone. If it is the case, this releases the zone lock.
+ */
+ blk_req_zone_write_unlock(rq);
+
if (blk_mq_sched_try_insert_merge(q, rq))
return;
@@ -468,6 +530,26 @@ static void dd_insert_requests(struct blk_mq_hw_ctx *hctx,
spin_unlock(&dd->lock);
}
+/*
+ * For zoned block devices, write unlock the target zone of
+ * completed write requests. Do this while holding the zone lock
+ * spinlock so that the zone is never unlocked while deadline_fifo_request()
+ * while deadline_next_request() are executing.
+ */
+static void dd_completed_request(struct request *rq)
+{
+ struct request_queue *q = rq->q;
+
+ if (blk_queue_is_zoned(q)) {
+ struct deadline_data *dd = q->elevator->elevator_data;
+ unsigned long flags;
+
+ spin_lock_irqsave(&dd->zone_lock, flags);
+ blk_req_zone_write_unlock(rq);
+ spin_unlock_irqrestore(&dd->zone_lock, flags);
+ }
+}
+
static bool dd_has_work(struct blk_mq_hw_ctx *hctx)
{
struct deadline_data *dd = hctx->queue->elevator->elevator_data;
@@ -669,6 +751,7 @@ static struct elevator_type mq_deadline = {
.ops.mq = {
.insert_requests = dd_insert_requests,
.dispatch_request = dd_dispatch_request,
+ .completed_request = dd_completed_request,
.next_request = elv_rb_latter_request,
.former_request = elv_rb_former_request,
.bio_merge = dd_bio_merge,
--
2.14.3
