Merge branch 'for-linus' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:
 "A set of fixes in the area of block IO, that should go into the next
  -rc release. This contains:

   - An OOPS fix from Dmitry, fixing a regression with the bio integrity
     code in this series.

   - Fix truncation of elevator io context cache name, from Eric
     Biggers.

   - NVMe pull from Christoph includes FC fixes from James, APST
     fixes/tweaks from Kai-Heng, removal fix from Rakesh, and an RDMA
     fix from Sagi.

   - Two tweaks for the block throttling code. One from Joseph Qi,
     fixing an oops from the timer code, and one from Shaohua, improving
     the behavior on rotatonal storage.

   - Two blk-mq fixes from Ming, fixing corner cases with the direct
     issue code.

   - Locking fix for bfq cgroups from Paolo"

* 'for-linus' of git://git.kernel.dk/linux-block:
  block, bfq: access and cache blkg data only when safe
  Fix loop device flush before configure v3
  blk-throttle: set default latency baseline for harddisk
  blk-throttle: fix NULL pointer dereference in throtl_schedule_pending_timer
  nvme: relax APST default max latency to 100ms
  nvme: only consider exit latency when choosing useful non-op power states
  nvme-fc: fix missing put reference on controller create failure
  nvme-fc: on lldd/transport io error, terminate association
  nvme-rdma: fast fail incoming requests while we reconnect
  nvme-pci: fix multiple ctrl removal scheduling
  nvme: fix hang in remove path
  elevator: fix truncation of icq_cache_name
  blk-mq: fix direct issue
  blk-mq: pass correct hctx to blk_mq_try_issue_directly
  bio-integrity: Do not allocate integrity context for bio w/o data

block/bfq-cgroup.c

@@ -52,7 +52,7 @@ BFQG_FLAG_FNS(idling)
 BFQG_FLAG_FNS(empty)
 #undef BFQG_FLAG_FNS
 
-/* This should be called with the queue_lock held. */
+/* This should be called with the scheduler lock held. */
 static void bfqg_stats_update_group_wait_time(struct bfqg_stats *stats)
 {
 	unsigned long long now;
@@ -67,7 +67,7 @@ static void bfqg_stats_update_group_wait_time(struct bfqg_stats *stats)
 	bfqg_stats_clear_waiting(stats);
 }
 
-/* This should be called with the queue_lock held. */
+/* This should be called with the scheduler lock held. */
 static void bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg,
 						 struct bfq_group *curr_bfqg)
 {
@@ -81,7 +81,7 @@ static void bfqg_stats_set_start_group_wait_time(struct bfq_group *bfqg,
 	bfqg_stats_mark_waiting(stats);
 }
 
-/* This should be called with the queue_lock held. */
+/* This should be called with the scheduler lock held. */
 static void bfqg_stats_end_empty_time(struct bfqg_stats *stats)
 {
 	unsigned long long now;
@@ -203,12 +203,30 @@ struct bfq_group *bfqq_group(struct bfq_queue *bfqq)
 
 static void bfqg_get(struct bfq_group *bfqg)
 {
-	return blkg_get(bfqg_to_blkg(bfqg));
+	bfqg->ref++;
 }
 
 void bfqg_put(struct bfq_group *bfqg)
 {
-	return blkg_put(bfqg_to_blkg(bfqg));
+	bfqg->ref--;
+
+	if (bfqg->ref == 0)
+		kfree(bfqg);
+}
+
+static void bfqg_and_blkg_get(struct bfq_group *bfqg)
+{
+	/* see comments in bfq_bic_update_cgroup for why refcounting bfqg */
+	bfqg_get(bfqg);
+
+	blkg_get(bfqg_to_blkg(bfqg));
+}
+
+void bfqg_and_blkg_put(struct bfq_group *bfqg)
+{
+	bfqg_put(bfqg);
+
+	blkg_put(bfqg_to_blkg(bfqg));
 }
 
 void bfqg_stats_update_io_add(struct bfq_group *bfqg, struct bfq_queue *bfqq,
@@ -312,7 +330,11 @@ void bfq_init_entity(struct bfq_entity *entity, struct bfq_group *bfqg)
 	if (bfqq) {
 		bfqq->ioprio = bfqq->new_ioprio;
 		bfqq->ioprio_class = bfqq->new_ioprio_class;
-		bfqg_get(bfqg);
+		/*
+		 * Make sure that bfqg and its associated blkg do not
+		 * disappear before entity.
+		 */
+		bfqg_and_blkg_get(bfqg);
 	}
 	entity->parent = bfqg->my_entity; /* NULL for root group */
 	entity->sched_data = &bfqg->sched_data;
@@ -399,6 +421,8 @@ struct blkg_policy_data *bfq_pd_alloc(gfp_t gfp, int node)
 		return NULL;
 	}
 
+	/* see comments in bfq_bic_update_cgroup for why refcounting */
+	bfqg_get(bfqg);
 	return &bfqg->pd;
 }
 
@@ -426,7 +450,7 @@ void bfq_pd_free(struct blkg_policy_data *pd)
 	struct bfq_group *bfqg = pd_to_bfqg(pd);
 
 	bfqg_stats_exit(&bfqg->stats);
-	return kfree(bfqg);
+	bfqg_put(bfqg);
 }
 
 void bfq_pd_reset_stats(struct blkg_policy_data *pd)
@@ -496,9 +520,10 @@ struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd,
  * Move @bfqq to @bfqg, deactivating it from its old group and reactivating
  * it on the new one.  Avoid putting the entity on the old group idle tree.
  *
- * Must be called under the queue lock; the cgroup owning @bfqg must
- * not disappear (by now this just means that we are called under
- * rcu_read_lock()).
+ * Must be called under the scheduler lock, to make sure that the blkg
+ * owning @bfqg does not disappear (see comments in
+ * bfq_bic_update_cgroup on guaranteeing the consistency of blkg
+ * objects).
  */
 void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq,
 		   struct bfq_group *bfqg)
@@ -519,16 +544,12 @@ void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq,
 		bfq_deactivate_bfqq(bfqd, bfqq, false, false);
 	else if (entity->on_st)
 		bfq_put_idle_entity(bfq_entity_service_tree(entity), entity);
-	bfqg_put(bfqq_group(bfqq));
+	bfqg_and_blkg_put(bfqq_group(bfqq));
 
-	/*
-	 * Here we use a reference to bfqg.  We don't need a refcounter
-	 * as the cgroup reference will not be dropped, so that its
-	 * destroy() callback will not be invoked.
-	 */
 	entity->parent = bfqg->my_entity;
 	entity->sched_data = &bfqg->sched_data;
-	bfqg_get(bfqg);
+	/* pin down bfqg and its associated blkg  */
+	bfqg_and_blkg_get(bfqg);
 
 	if (bfq_bfqq_busy(bfqq)) {
 		bfq_pos_tree_add_move(bfqd, bfqq);
@@ -545,8 +566,9 @@ void bfq_bfqq_move(struct bfq_data *bfqd, struct bfq_queue *bfqq,
  * @bic: the bic to move.
  * @blkcg: the blk-cgroup to move to.
  *
- * Move bic to blkcg, assuming that bfqd->queue is locked; the caller
- * has to make sure that the reference to cgroup is valid across the call.
+ * Move bic to blkcg, assuming that bfqd->lock is held; which makes
+ * sure that the reference to cgroup is valid across the call (see
+ * comments in bfq_bic_update_cgroup on this issue)
  *
  * NOTE: an alternative approach might have been to store the current
  * cgroup in bfqq and getting a reference to it, reducing the lookup
@@ -604,6 +626,57 @@ void bfq_bic_update_cgroup(struct bfq_io_cq *bic, struct bio *bio)
 		goto out;
 
 	bfqg = __bfq_bic_change_cgroup(bfqd, bic, bio_blkcg(bio));
+	/*
+	 * Update blkg_path for bfq_log_* functions. We cache this
+	 * path, and update it here, for the following
+	 * reasons. Operations on blkg objects in blk-cgroup are
+	 * protected with the request_queue lock, and not with the
+	 * lock that protects the instances of this scheduler
+	 * (bfqd->lock). This exposes BFQ to the following sort of
+	 * race.
+	 *
+	 * The blkg_lookup performed in bfq_get_queue, protected
+	 * through rcu, may happen to return the address of a copy of
+	 * the original blkg. If this is the case, then the
+	 * bfqg_and_blkg_get performed in bfq_get_queue, to pin down
+	 * the blkg, is useless: it does not prevent blk-cgroup code
+	 * from destroying both the original blkg and all objects
+	 * directly or indirectly referred by the copy of the
+	 * blkg.
+	 *
+	 * On the bright side, destroy operations on a blkg invoke, as
+	 * a first step, hooks of the scheduler associated with the
+	 * blkg. And these hooks are executed with bfqd->lock held for
+	 * BFQ. As a consequence, for any blkg associated with the
+	 * request queue this instance of the scheduler is attached
+	 * to, we are guaranteed that such a blkg is not destroyed, and
+	 * that all the pointers it contains are consistent, while we
+	 * are holding bfqd->lock. A blkg_lookup performed with
+	 * bfqd->lock held then returns a fully consistent blkg, which
+	 * remains consistent until this lock is held.
+	 *
+	 * Thanks to the last fact, and to the fact that: (1) bfqg has
+	 * been obtained through a blkg_lookup in the above
+	 * assignment, and (2) bfqd->lock is being held, here we can
+	 * safely use the policy data for the involved blkg (i.e., the
+	 * field bfqg->pd) to get to the blkg associated with bfqg,
+	 * and then we can safely use any field of blkg. After we
+	 * release bfqd->lock, even just getting blkg through this
+	 * bfqg may cause dangling references to be traversed, as
+	 * bfqg->pd may not exist any more.
+	 *
+	 * In view of the above facts, here we cache, in the bfqg, any
+	 * blkg data we may need for this bic, and for its associated
+	 * bfq_queue. As of now, we need to cache only the path of the
+	 * blkg, which is used in the bfq_log_* functions.
+	 *
+	 * Finally, note that bfqg itself needs to be protected from
+	 * destruction on the blkg_free of the original blkg (which
+	 * invokes bfq_pd_free). We use an additional private
+	 * refcounter for bfqg, to let it disappear only after no
+	 * bfq_queue refers to it any longer.
+	 */
+	blkg_path(bfqg_to_blkg(bfqg), bfqg->blkg_path, sizeof(bfqg->blkg_path));
 	bic->blkcg_serial_nr = serial_nr;
 out:
 	rcu_read_unlock();
@@ -640,8 +713,6 @@ static void bfq_reparent_leaf_entity(struct bfq_data *bfqd,
  * @bfqd: the device data structure with the root group.
  * @bfqg: the group to move from.
  * @st: the service tree with the entities.
- *
- * Needs queue_lock to be taken and reference to be valid over the call.
  */
 static void bfq_reparent_active_entities(struct bfq_data *bfqd,
 					 struct bfq_group *bfqg,
@@ -692,8 +763,7 @@ void bfq_pd_offline(struct blkg_policy_data *pd)
 		/*
 		 * The idle tree may still contain bfq_queues belonging
 		 * to exited task because they never migrated to a different
-		 * cgroup from the one being destroyed now.  No one else
-		 * can access them so it's safe to act without any lock.
+		 * cgroup from the one being destroyed now.
 		 */
 		bfq_flush_idle_tree(st);
 

block/bfq-iosched.c

@@ -3665,7 +3665,7 @@ void bfq_put_queue(struct bfq_queue *bfqq)
 
 	kmem_cache_free(bfq_pool, bfqq);
 #ifdef CONFIG_BFQ_GROUP_IOSCHED
-	bfqg_put(bfqg);
+	bfqg_and_blkg_put(bfqg);
 #endif
 }
 

block/bfq-iosched.h

@@ -759,6 +759,12 @@ struct bfq_group {
 	/* must be the first member */
 	struct blkg_policy_data pd;
 
+	/* cached path for this blkg (see comments in bfq_bic_update_cgroup) */
+	char blkg_path[128];
+
+	/* reference counter (see comments in bfq_bic_update_cgroup) */
+	int ref;
+
 	struct bfq_entity entity;
 	struct bfq_sched_data sched_data;
 
@@ -838,7 +844,7 @@ struct bfq_group *bfq_find_set_group(struct bfq_data *bfqd,
 struct blkcg_gq *bfqg_to_blkg(struct bfq_group *bfqg);
 struct bfq_group *bfqq_group(struct bfq_queue *bfqq);
 struct bfq_group *bfq_create_group_hierarchy(struct bfq_data *bfqd, int node);
-void bfqg_put(struct bfq_group *bfqg);
+void bfqg_and_blkg_put(struct bfq_group *bfqg);
 
 #ifdef CONFIG_BFQ_GROUP_IOSCHED
 extern struct cftype bfq_blkcg_legacy_files[];
@@ -910,20 +916,13 @@ void bfq_add_bfqq_busy(struct bfq_data *bfqd, struct bfq_queue *bfqq);
 struct bfq_group *bfqq_group(struct bfq_queue *bfqq);
 
 #define bfq_log_bfqq(bfqd, bfqq, fmt, args...)	do {			\
-	char __pbuf[128];						\
-									\
-	blkg_path(bfqg_to_blkg(bfqq_group(bfqq)), __pbuf, sizeof(__pbuf)); \
-	blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, (bfqq)->pid, \
+	blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, (bfqq)->pid,\
 			bfq_bfqq_sync((bfqq)) ? 'S' : 'A',		\
-			  __pbuf, ##args);				\
+			bfqq_group(bfqq)->blkg_path, ##args);		\
 } while (0)
 
-#define bfq_log_bfqg(bfqd, bfqg, fmt, args...)	do {			\
-	char __pbuf[128];						\
-									\
-	blkg_path(bfqg_to_blkg(bfqg), __pbuf, sizeof(__pbuf));		\
-	blk_add_trace_msg((bfqd)->queue, "%s " fmt, __pbuf, ##args);	\
-} while (0)
+#define bfq_log_bfqg(bfqd, bfqg, fmt, args...)	\
+	blk_add_trace_msg((bfqd)->queue, "%s " fmt, (bfqg)->blkg_path, ##args)
 
 #else /* CONFIG_BFQ_GROUP_IOSCHED */
 

block/bio-integrity.c

@@ -175,6 +175,9 @@ bool bio_integrity_enabled(struct bio *bio)
 	if (bio_op(bio) != REQ_OP_READ && bio_op(bio) != REQ_OP_WRITE)
 		return false;
 
+	if (!bio_sectors(bio))
+		return false;
+
 	/* Already protected? */
 	if (bio_integrity(bio))
 		return false;

block/blk-mq.c

@@ -1461,22 +1461,28 @@ static blk_qc_t request_to_qc_t(struct blk_mq_hw_ctx *hctx, struct request *rq)
 	return blk_tag_to_qc_t(rq->internal_tag, hctx->queue_num, true);
 }
 
-static void __blk_mq_try_issue_directly(struct request *rq, blk_qc_t *cookie,
-				      bool may_sleep)
+static void __blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
+					struct request *rq,
+					blk_qc_t *cookie, bool may_sleep)
 {
 	struct request_queue *q = rq->q;
 	struct blk_mq_queue_data bd = {
 		.rq = rq,
 		.last = true,
 	};
-	struct blk_mq_hw_ctx *hctx;
 	blk_qc_t new_cookie;
 	int ret;
+	bool run_queue = true;
+
+	if (blk_mq_hctx_stopped(hctx)) {
+		run_queue = false;
+		goto insert;
+	}
 
 	if (q->elevator)
 		goto insert;
 
-	if (!blk_mq_get_driver_tag(rq, &hctx, false))
+	if (!blk_mq_get_driver_tag(rq, NULL, false))
 		goto insert;
 
 	new_cookie = request_to_qc_t(hctx, rq);
@@ -1500,7 +1506,7 @@ static void __blk_mq_try_issue_directly(struct request *rq, blk_qc_t *cookie,
 
 	__blk_mq_requeue_request(rq);
 insert:
-	blk_mq_sched_insert_request(rq, false, true, false, may_sleep);
+	blk_mq_sched_insert_request(rq, false, run_queue, false, may_sleep);
 }
 
 static void blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
@@ -1508,7 +1514,7 @@ static void blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
 {
 	if (!(hctx->flags & BLK_MQ_F_BLOCKING)) {
 		rcu_read_lock();
-		__blk_mq_try_issue_directly(rq, cookie, false);
+		__blk_mq_try_issue_directly(hctx, rq, cookie, false);
 		rcu_read_unlock();
 	} else {
 		unsigned int srcu_idx;
@@ -1516,7 +1522,7 @@ static void blk_mq_try_issue_directly(struct blk_mq_hw_ctx *hctx,
 		might_sleep();
 
 		srcu_idx = srcu_read_lock(&hctx->queue_rq_srcu);
-		__blk_mq_try_issue_directly(rq, cookie, true);
+		__blk_mq_try_issue_directly(hctx, rq, cookie, true);
 		srcu_read_unlock(&hctx->queue_rq_srcu, srcu_idx);
 	}
 }
@@ -1619,9 +1625,12 @@ static blk_qc_t blk_mq_make_request(struct request_queue *q, struct bio *bio)
 
 		blk_mq_put_ctx(data.ctx);
 
-		if (same_queue_rq)
+		if (same_queue_rq) {
+			data.hctx = blk_mq_map_queue(q,
+					same_queue_rq->mq_ctx->cpu);
 			blk_mq_try_issue_directly(data.hctx, same_queue_rq,
 					&cookie);
+		}
 	} else if (q->nr_hw_queues > 1 && is_sync) {
 		blk_mq_put_ctx(data.ctx);
 		blk_mq_bio_to_request(rq, bio);

block/blk-throttle.c

@@ -27,6 +27,13 @@ static int throtl_quantum = 32;
 #define MIN_THROTL_IOPS (10)
 #define DFL_LATENCY_TARGET (-1L)
 #define DFL_IDLE_THRESHOLD (0)
+#define DFL_HD_BASELINE_LATENCY (4000L) /* 4ms */
+#define LATENCY_FILTERED_SSD (0)
+/*
+ * For HD, very small latency comes from sequential IO. Such IO is helpless to
+ * help determine if its IO is impacted by others, hence we ignore the IO
+ */
+#define LATENCY_FILTERED_HD (1000L) /* 1ms */
 
 #define SKIP_LATENCY (((u64)1) << BLK_STAT_RES_SHIFT)
 
@@ -212,6 +219,7 @@ struct throtl_data
 	struct avg_latency_bucket avg_buckets[LATENCY_BUCKET_SIZE];
 	struct latency_bucket __percpu *latency_buckets;
 	unsigned long last_calculate_time;
+	unsigned long filtered_latency;
 
 	bool track_bio_latency;
 };
@@ -698,7 +706,7 @@ static void throtl_dequeue_tg(struct throtl_grp *tg)
 static void throtl_schedule_pending_timer(struct throtl_service_queue *sq,
 					  unsigned long expires)
 {
-	unsigned long max_expire = jiffies + 8 * sq_to_tg(sq)->td->throtl_slice;
+	unsigned long max_expire = jiffies + 8 * sq_to_td(sq)->throtl_slice;
 
 	/*
 	 * Since we are adjusting the throttle limit dynamically, the sleep
@@ -2281,7 +2289,7 @@ void blk_throtl_bio_endio(struct bio *bio)
 		throtl_track_latency(tg->td, blk_stat_size(&bio->bi_issue_stat),
 			bio_op(bio), lat);
 
-	if (tg->latency_target) {
+	if (tg->latency_target && lat >= tg->td->filtered_latency) {
 		int bucket;
 		unsigned int threshold;
 
@@ -2417,14 +2425,20 @@ void blk_throtl_exit(struct request_queue *q)
 void blk_throtl_register_queue(struct request_queue *q)
 {
 	struct throtl_data *td;
+	int i;
 
 	td = q->td;
 	BUG_ON(!td);
 
-	if (blk_queue_nonrot(q))
+	if (blk_queue_nonrot(q)) {
 		td->throtl_slice = DFL_THROTL_SLICE_SSD;
-	else
+		td->filtered_latency = LATENCY_FILTERED_SSD;
+	} else {
 		td->throtl_slice = DFL_THROTL_SLICE_HD;
+		td->filtered_latency = LATENCY_FILTERED_HD;
+		for (i = 0; i < LATENCY_BUCKET_SIZE; i++)
+			td->avg_buckets[i].latency = DFL_HD_BASELINE_LATENCY;
+	}
 #ifndef CONFIG_BLK_DEV_THROTTLING_LOW
 	/* if no low limit, use previous default */
 	td->throtl_slice = DFL_THROTL_SLICE_HD;

drivers/block/loop.c

@@ -608,6 +608,9 @@ static int loop_switch(struct loop_device *lo, struct file *file)
  */
 static int loop_flush(struct loop_device *lo)
 {
+	/* loop not yet configured, no running thread, nothing to flush */
+	if (lo->lo_state != Lo_bound)
+		return 0;
 	return loop_switch(lo, NULL);
 }
 

drivers/nvme/host/core.c

@@ -56,7 +56,7 @@ MODULE_PARM_DESC(max_retries, "max number of retries a command may have");
 static int nvme_char_major;
 module_param(nvme_char_major, int, 0);
 
-static unsigned long default_ps_max_latency_us = 25000;
+static unsigned long default_ps_max_latency_us = 100000;
 module_param(default_ps_max_latency_us, ulong, 0644);
 MODULE_PARM_DESC(default_ps_max_latency_us,
 		 "max power saving latency for new devices; use PM QOS to change per device");
@@ -1342,7 +1342,7 @@ static void nvme_configure_apst(struct nvme_ctrl *ctrl)
 	 * transitioning between power states.  Therefore, when running
 	 * in any given state, we will enter the next lower-power
 	 * non-operational state after waiting 50 * (enlat + exlat)
-	 * microseconds, as long as that state's total latency is under
+	 * microseconds, as long as that state's exit latency is under
 	 * the requested maximum latency.
 	 *
 	 * We will not autonomously enter any non-operational state for
@@ -1387,7 +1387,7 @@ static void nvme_configure_apst(struct nvme_ctrl *ctrl)
 		 * lowest-power state, not the number of states.
 		 */
 		for (state = (int)ctrl->npss; state >= 0; state--) {
-			u64 total_latency_us, transition_ms;
+			u64 total_latency_us, exit_latency_us, transition_ms;
 
 			if (target)
 				table->entries[state] = target;
@@ -1408,12 +1408,15 @@ static void nvme_configure_apst(struct nvme_ctrl *ctrl)
 			      NVME_PS_FLAGS_NON_OP_STATE))
 				continue;
 
-			total_latency_us =
-				(u64)le32_to_cpu(ctrl->psd[state].entry_lat) +
-				+ le32_to_cpu(ctrl->psd[state].exit_lat);
-			if (total_latency_us > ctrl->ps_max_latency_us)
+			exit_latency_us =
+				(u64)le32_to_cpu(ctrl->psd[state].exit_lat);
+			if (exit_latency_us > ctrl->ps_max_latency_us)
 				continue;
 
+			total_latency_us =
+				exit_latency_us +
+				le32_to_cpu(ctrl->psd[state].entry_lat);
+
 			/*
 			 * This state is good.  Use it as the APST idle
 			 * target for higher power states.
@@ -2438,6 +2441,10 @@ void nvme_kill_queues(struct nvme_ctrl *ctrl)
 	struct nvme_ns *ns;
 
 	mutex_lock(&ctrl->namespaces_mutex);
+
+	/* Forcibly start all queues to avoid having stuck requests */
+	blk_mq_start_hw_queues(ctrl->admin_q);
+
 	list_for_each_entry(ns, &ctrl->namespaces, list) {
 		/*
 		 * Revalidating a dead namespace sets capacity to 0. This will

drivers/nvme/host/fc.c

@@ -1139,6 +1139,7 @@ nvme_fc_xmt_disconnect_assoc(struct nvme_fc_ctrl *ctrl)
 /* *********************** NVME Ctrl Routines **************************** */
 
 static void __nvme_fc_final_op_cleanup(struct request *rq);
+static void nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg);
 
 static int
 nvme_fc_reinit_request(void *data, struct request *rq)
@@ -1265,7 +1266,7 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 	struct nvme_command *sqe = &op->cmd_iu.sqe;
 	__le16 status = cpu_to_le16(NVME_SC_SUCCESS << 1);
 	union nvme_result result;
-	bool complete_rq;
+	bool complete_rq, terminate_assoc = true;
 
 	/*
 	 * WARNING:
@@ -1294,6 +1295,14 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 	 * fabricate a CQE, the following fields will not be set as they
 	 * are not referenced:
 	 *      cqe.sqid,  cqe.sqhd,  cqe.command_id
+	 *
+	 * Failure or error of an individual i/o, in a transport
+	 * detected fashion unrelated to the nvme completion status,
+	 * potentially cause the initiator and target sides to get out
+	 * of sync on SQ head/tail (aka outstanding io count allowed).
+	 * Per FC-NVME spec, failure of an individual command requires
+	 * the connection to be terminated, which in turn requires the
+	 * association to be terminated.
 	 */
 
 	fc_dma_sync_single_for_cpu(ctrl->lport->dev, op->fcp_req.rspdma,
@@ -1359,6 +1368,8 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 		goto done;
 	}
 
+	terminate_assoc = false;
+
 done:
 	if (op->flags & FCOP_FLAGS_AEN) {
 		nvme_complete_async_event(&queue->ctrl->ctrl, status, &result);
@@ -1366,7 +1377,7 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 		atomic_set(&op->state, FCPOP_STATE_IDLE);
 		op->flags = FCOP_FLAGS_AEN;	/* clear other flags */
 		nvme_fc_ctrl_put(ctrl);
-		return;
+		goto check_error;
 	}
 
 	complete_rq = __nvme_fc_fcpop_chk_teardowns(ctrl, op);
@@ -1379,6 +1390,10 @@ nvme_fc_fcpio_done(struct nvmefc_fcp_req *req)
 		nvme_end_request(rq, status, result);
 	} else
 		__nvme_fc_final_op_cleanup(rq);
+
+check_error:
+	if (terminate_assoc)
+		nvme_fc_error_recovery(ctrl, "transport detected io error");
 }
 
 static int
@@ -2791,6 +2806,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 		ctrl->ctrl.opts = NULL;
 		/* initiate nvme ctrl ref counting teardown */
 		nvme_uninit_ctrl(&ctrl->ctrl);
+		nvme_put_ctrl(&ctrl->ctrl);
 
 		/* as we're past the point where we transition to the ref
 		 * counting teardown path, if we return a bad pointer here,

drivers/nvme/host/pci.c

@@ -1367,7 +1367,7 @@ static bool nvme_should_reset(struct nvme_dev *dev, u32 csts)
 	bool nssro = dev->subsystem && (csts & NVME_CSTS_NSSRO);
 
 	/* If there is a reset ongoing, we shouldn't reset again. */
-	if (work_busy(&dev->reset_work))
+	if (dev->ctrl.state == NVME_CTRL_RESETTING)
 		return false;
 
 	/* We shouldn't reset unless the controller is on fatal error state
@@ -1903,7 +1903,7 @@ static void nvme_reset_work(struct work_struct *work)
 	bool was_suspend = !!(dev->ctrl.ctrl_config & NVME_CC_SHN_NORMAL);
 	int result = -ENODEV;
 
-	if (WARN_ON(dev->ctrl.state == NVME_CTRL_RESETTING))
+	if (WARN_ON(dev->ctrl.state != NVME_CTRL_RESETTING))
 		goto out;
 
 	/*
@@ -1913,9 +1913,6 @@ static void nvme_reset_work(struct work_struct *work)
 	if (dev->ctrl.ctrl_config & NVME_CC_ENABLE)
 		nvme_dev_disable(dev, false);
 
-	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING))
-		goto out;
-
 	result = nvme_pci_enable(dev);
 	if (result)
 		goto out;
@@ -2009,8 +2006,8 @@ static int nvme_reset(struct nvme_dev *dev)
 {
 	if (!dev->ctrl.admin_q || blk_queue_dying(dev->ctrl.admin_q))
 		return -ENODEV;
-	if (work_busy(&dev->reset_work))
-		return -ENODEV;
+	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING))
+		return -EBUSY;
 	if (!queue_work(nvme_workq, &dev->reset_work))
 		return -EBUSY;
 	return 0;
@@ -2136,6 +2133,7 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	if (result)
 		goto release_pools;
 
+	nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING);
 	dev_info(dev->ctrl.device, "pci function %s\n", dev_name(&pdev->dev));
 
 	queue_work(nvme_workq, &dev->reset_work);
@@ -2179,6 +2177,7 @@ static void nvme_remove(struct pci_dev *pdev)
 
 	nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_DELETING);
 
+	cancel_work_sync(&dev->reset_work);
 	pci_set_drvdata(pdev, NULL);
 
 	if (!pci_device_is_present(pdev)) {

drivers/nvme/host/rdma.c

@@ -753,28 +753,26 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	if (ret)
 		goto requeue;
 
-	blk_mq_start_stopped_hw_queues(ctrl->ctrl.admin_q, true);
-
 	ret = nvmf_connect_admin_queue(&ctrl->ctrl);
 	if (ret)
-		goto stop_admin_q;
+		goto requeue;
 
 	set_bit(NVME_RDMA_Q_LIVE, &ctrl->queues[0].flags);
 
 	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->cap);
 	if (ret)
-		goto stop_admin_q;
+		goto requeue;
 
 	nvme_start_keep_alive(&ctrl->ctrl);
 
 	if (ctrl->queue_count > 1) {
 		ret = nvme_rdma_init_io_queues(ctrl);
 		if (ret)
-			goto stop_admin_q;
+			goto requeue;
 
 		ret = nvme_rdma_connect_io_queues(ctrl);
 		if (ret)
-			goto stop_admin_q;
+			goto requeue;
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
@@ -782,7 +780,6 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	ctrl->ctrl.opts->nr_reconnects = 0;
 
 	if (ctrl->queue_count > 1) {
-		nvme_start_queues(&ctrl->ctrl);
 		nvme_queue_scan(&ctrl->ctrl);
 		nvme_queue_async_events(&ctrl->ctrl);
 	}
@@ -791,8 +788,6 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 
 	return;
 
-stop_admin_q:
-	blk_mq_stop_hw_queues(ctrl->ctrl.admin_q);
 requeue:
 	dev_info(ctrl->ctrl.device, "Failed reconnect attempt %d\n",
 			ctrl->ctrl.opts->nr_reconnects);
@@ -823,6 +818,13 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
 				nvme_cancel_request, &ctrl->ctrl);
 
+	/*
+	 * queues are not a live anymore, so restart the queues to fail fast
+	 * new IO
+	 */
+	blk_mq_start_stopped_hw_queues(ctrl->ctrl.admin_q, true);
+	nvme_start_queues(&ctrl->ctrl);
+
 	nvme_rdma_reconnect_or_remove(ctrl);
 }
 
@@ -1433,7 +1435,7 @@ nvme_rdma_timeout(struct request *rq, bool reserved)
 /*
  * We cannot accept any other command until the Connect command has completed.
  */
-static inline bool nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue,
+static inline int nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue,
 		struct request *rq)
 {
 	if (unlikely(!test_bit(NVME_RDMA_Q_LIVE, &queue->flags))) {
@@ -1441,11 +1443,22 @@ static inline bool nvme_rdma_queue_is_ready(struct nvme_rdma_queue *queue,
 
 		if (!blk_rq_is_passthrough(rq) ||
 		    cmd->common.opcode != nvme_fabrics_command ||
-		    cmd->fabrics.fctype != nvme_fabrics_type_connect)
-			return false;
+		    cmd->fabrics.fctype != nvme_fabrics_type_connect) {
+			/*
+			 * reconnecting state means transport disruption, which
+			 * can take a long time and even might fail permanently,
+			 * so we can't let incoming I/O be requeued forever.
+			 * fail it fast to allow upper layers a chance to
+			 * failover.
+			 */
+			if (queue->ctrl->ctrl.state == NVME_CTRL_RECONNECTING)
+				return -EIO;
+			else
+				return -EAGAIN;
+		}
 	}
 
-	return true;
+	return 0;
 }
 
 static int nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx,
@@ -1463,8 +1476,9 @@ static int nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx,
 
 	WARN_ON_ONCE(rq->tag < 0);
 
-	if (!nvme_rdma_queue_is_ready(queue, rq))
-		return BLK_MQ_RQ_QUEUE_BUSY;
+	ret = nvme_rdma_queue_is_ready(queue, rq);
+	if (unlikely(ret))
+		goto err;
 
 	dev = queue->device->dev;
 	ib_dma_sync_single_for_cpu(dev, sqe->dma,

include/linux/elevator.h

@@ -153,7 +153,7 @@ struct elevator_type
 #endif
 
 	/* managed by elevator core */
-	char icq_cache_name[ELV_NAME_MAX + 5];	/* elvname + "_io_cq" */
+	char icq_cache_name[ELV_NAME_MAX + 6];	/* elvname + "_io_cq" */
 	struct list_head list;
 };