block, bfq: add requeue-request hook
[sfrench/cifs-2.6.git] / block / bfq-iosched.c
index 47e6ec7427c44bfefeba20af5842f88711b1c3d9..aeca22d911010dc1a11ae9377a46f4ac99b3f629 100644 (file)
@@ -3823,24 +3823,26 @@ static struct request *__bfq_dispatch_request(struct blk_mq_hw_ctx *hctx)
                }
 
                /*
-                * We exploit the bfq_finish_request hook to decrement
-                * rq_in_driver, but bfq_finish_request will not be
-                * invoked on this request. So, to avoid unbalance,
-                * just start this request, without incrementing
-                * rq_in_driver. As a negative consequence,
-                * rq_in_driver is deceptively lower than it should be
-                * while this request is in service. This may cause
-                * bfq_schedule_dispatch to be invoked uselessly.
+                * We exploit the bfq_finish_requeue_request hook to
+                * decrement rq_in_driver, but
+                * bfq_finish_requeue_request will not be invoked on
+                * this request. So, to avoid unbalance, just start
+                * this request, without incrementing rq_in_driver. As
+                * a negative consequence, rq_in_driver is deceptively
+                * lower than it should be while this request is in
+                * service. This may cause bfq_schedule_dispatch to be
+                * invoked uselessly.
                 *
                 * As for implementing an exact solution, the
-                * bfq_finish_request hook, if defined, is probably
-                * invoked also on this request. So, by exploiting
-                * this hook, we could 1) increment rq_in_driver here,
-                * and 2) decrement it in bfq_finish_request. Such a
-                * solution would let the value of the counter be
-                * always accurate, but it would entail using an extra
-                * interface function. This cost seems higher than the
-                * benefit, being the frequency of non-elevator-private
+                * bfq_finish_requeue_request hook, if defined, is
+                * probably invoked also on this request. So, by
+                * exploiting this hook, we could 1) increment
+                * rq_in_driver here, and 2) decrement it in
+                * bfq_finish_requeue_request. Such a solution would
+                * let the value of the counter be always accurate,
+                * but it would entail using an extra interface
+                * function. This cost seems higher than the benefit,
+                * being the frequency of non-elevator-private
                 * requests very low.
                 */
                goto start_rq;
@@ -4515,6 +4517,8 @@ static inline void bfq_update_insert_stats(struct request_queue *q,
                                           unsigned int cmd_flags) {}
 #endif
 
+static void bfq_prepare_request(struct request *rq, struct bio *bio);
+
 static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
                               bool at_head)
 {
@@ -4541,6 +4545,18 @@ static void bfq_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
                else
                        list_add_tail(&rq->queuelist, &bfqd->dispatch);
        } else {
+               if (WARN_ON_ONCE(!bfqq)) {
+                       /*
+                        * This should never happen. Most likely rq is
+                        * a requeued regular request, being
+                        * re-inserted without being first
+                        * re-prepared. Do a prepare, to avoid
+                        * failure.
+                        */
+                       bfq_prepare_request(rq, rq->bio);
+                       bfqq = RQ_BFQQ(rq);
+               }
+
                idle_timer_disabled = __bfq_insert_request(bfqd, rq);
                /*
                 * Update bfqq, because, if a queue merge has occurred
@@ -4697,22 +4713,44 @@ static void bfq_completed_request(struct bfq_queue *bfqq, struct bfq_data *bfqd)
                bfq_schedule_dispatch(bfqd);
 }
 
-static void bfq_finish_request_body(struct bfq_queue *bfqq)
+static void bfq_finish_requeue_request_body(struct bfq_queue *bfqq)
 {
        bfqq->allocated--;
 
        bfq_put_queue(bfqq);
 }
 
-static void bfq_finish_request(struct request *rq)
+/*
+ * Handle either a requeue or a finish for rq. The things to do are
+ * the same in both cases: all references to rq are to be dropped. In
+ * particular, rq is considered completed from the point of view of
+ * the scheduler.
+ */
+static void bfq_finish_requeue_request(struct request *rq)
 {
-       struct bfq_queue *bfqq;
+       struct bfq_queue *bfqq = RQ_BFQQ(rq);
        struct bfq_data *bfqd;
 
-       if (!rq->elv.icq)
+       /*
+        * Requeue and finish hooks are invoked in blk-mq without
+        * checking whether the involved request is actually still
+        * referenced in the scheduler. To handle this fact, the
+        * following two checks make this function exit in case of
+        * spurious invocations, for which there is nothing to do.
+        *
+        * First, check whether rq has nothing to do with an elevator.
+        */
+       if (unlikely(!(rq->rq_flags & RQF_ELVPRIV)))
+               return;
+
+       /*
+        * rq either is not associated with any icq, or is an already
+        * requeued request that has not (yet) been re-inserted into
+        * a bfq_queue.
+        */
+       if (!rq->elv.icq || !bfqq)
                return;
 
-       bfqq = RQ_BFQQ(rq);
        bfqd = bfqq->bfqd;
 
        if (rq->rq_flags & RQF_STARTED)
@@ -4727,13 +4765,14 @@ static void bfq_finish_request(struct request *rq)
                spin_lock_irqsave(&bfqd->lock, flags);
 
                bfq_completed_request(bfqq, bfqd);
-               bfq_finish_request_body(bfqq);
+               bfq_finish_requeue_request_body(bfqq);
 
                spin_unlock_irqrestore(&bfqd->lock, flags);
        } else {
                /*
                 * Request rq may be still/already in the scheduler,
-                * in which case we need to remove it. And we cannot
+                * in which case we need to remove it (this should
+                * never happen in case of requeue). And we cannot
                 * defer such a check and removal, to avoid
                 * inconsistencies in the time interval from the end
                 * of this function to the start of the deferred work.
@@ -4748,9 +4787,26 @@ static void bfq_finish_request(struct request *rq)
                        bfqg_stats_update_io_remove(bfqq_group(bfqq),
                                                    rq->cmd_flags);
                }
-               bfq_finish_request_body(bfqq);
+               bfq_finish_requeue_request_body(bfqq);
        }
 
+       /*
+        * Reset private fields. In case of a requeue, this allows
+        * this function to correctly do nothing if it is spuriously
+        * invoked again on this same request (see the check at the
+        * beginning of the function). Probably, a better general
+        * design would be to prevent blk-mq from invoking the requeue
+        * or finish hooks of an elevator, for a request that is not
+        * referred by that elevator.
+        *
+        * Resetting the following fields would break the
+        * request-insertion logic if rq is re-inserted into a bfq
+        * internal queue, without a re-preparation. Here we assume
+        * that re-insertions of requeued requests, without
+        * re-preparation, can happen only for pass_through or at_head
+        * requests (which are not re-inserted into bfq internal
+        * queues).
+        */
        rq->elv.priv[0] = NULL;
        rq->elv.priv[1] = NULL;
 }
@@ -5426,7 +5482,8 @@ static struct elevator_type iosched_bfq_mq = {
        .ops.mq = {
                .limit_depth            = bfq_limit_depth,
                .prepare_request        = bfq_prepare_request,
-               .finish_request         = bfq_finish_request,
+               .requeue_request        = bfq_finish_requeue_request,
+               .finish_request         = bfq_finish_requeue_request,
                .exit_icq               = bfq_exit_icq,
                .insert_requests        = bfq_insert_requests,
                .dispatch_request       = bfq_dispatch_request,