Merge branch 'for-4.14/block' of git://git.kernel.dk/linux-block

Pull block layer updates from Jens Axboe:
 "This is the first pull request for 4.14, containing most of the code
  changes. It's a quiet series this round, which I think we needed after
  the churn of the last few series. This contains:

   - Fix for a registration race in loop, from Anton Volkov.

   - Overflow complaint fix from Arnd for DAC960.

   - Series of drbd changes from the usual suspects.

   - Conversion of the stec/skd driver to blk-mq. From Bart.

   - A few BFQ improvements/fixes from Paolo.

   - CFQ improvement from Ritesh, allowing idling for group idle.

   - A few fixes found by Dan's smatch, courtesy of Dan.

   - A warning fixup for a race between changing the IO scheduler and
     device remova. From David Jeffery.

   - A few nbd fixes from Josef.

   - Support for cgroup info in blktrace, from Shaohua.

   - Also from Shaohua, new features in the null_blk driver to allow it
     to actually hold data, among other things.

   - Various corner cases and error handling fixes from Weiping Zhang.

   - Improvements to the IO stats tracking for blk-mq from me. Can
     drastically improve performance for fast devices and/or big
     machines.

   - Series from Christoph removing bi_bdev as being needed for IO
     submission, in preparation for nvme multipathing code.

   - Series from Bart, including various cleanups and fixes for switch
     fall through case complaints"

* 'for-4.14/block' of git://git.kernel.dk/linux-block: (162 commits)
  kernfs: checking for IS_ERR() instead of NULL
  drbd: remove BIOSET_NEED_RESCUER flag from drbd_{md_,}io_bio_set
  drbd: Fix allyesconfig build, fix recent commit
  drbd: switch from kmalloc() to kmalloc_array()
  drbd: abort drbd_start_resync if there is no connection
  drbd: move global variables to drbd namespace and make some static
  drbd: rename "usermode_helper" to "drbd_usermode_helper"
  drbd: fix race between handshake and admin disconnect/down
  drbd: fix potential deadlock when trying to detach during handshake
  drbd: A single dot should be put into a sequence.
  drbd: fix rmmod cleanup, remove _all_ debugfs entries
  drbd: Use setup_timer() instead of init_timer() to simplify the code.
  drbd: fix potential get_ldev/put_ldev refcount imbalance during attach
  drbd: new disk-option disable-write-same
  drbd: Fix resource role for newly created resources in events2
  drbd: mark symbols static where possible
  drbd: Send P_NEG_ACK upon write error in protocol != C
  drbd: add explicit plugging when submitting batches
  drbd: change list_for_each_safe to while(list_first_entry_or_null)
  drbd: introduce drbd_recv_header_maybe_unplug
  ...

MAINTAINERS

@@ -12561,6 +12561,12 @@ M:	Ion Badulescu <ionut@badula.org>
 S:	Odd Fixes
 F:	drivers/net/ethernet/adaptec/starfire*
 
+STEC S1220 SKD DRIVER
+M:	Bart Van Assche <bart.vanassche@wdc.com>
+L:	linux-block@vger.kernel.org
+S:	Maintained
+F:	drivers/block/skd*[ch]
+
 STI CEC DRIVER
 M:	Benjamin Gaignard <benjamin.gaignard@linaro.org>
 S:	Maintained

arch/powerpc/sysdev/axonram.c

@@ -110,7 +110,7 @@ axon_ram_irq_handler(int irq, void *dev)
 static blk_qc_t
 axon_ram_make_request(struct request_queue *queue, struct bio *bio)
 {
-	struct axon_ram_bank *bank = bio->bi_bdev->bd_disk->private_data;
+	struct axon_ram_bank *bank = bio->bi_disk->private_data;
 	unsigned long phys_mem, phys_end;
 	void *user_mem;
 	struct bio_vec vec;

block/bfq-iosched.c

@@ -128,7 +128,7 @@ BFQ_BFQQ_FNS(busy);
 BFQ_BFQQ_FNS(wait_request);
 BFQ_BFQQ_FNS(non_blocking_wait_rq);
 BFQ_BFQQ_FNS(fifo_expire);
-BFQ_BFQQ_FNS(idle_window);
+BFQ_BFQQ_FNS(has_short_ttime);
 BFQ_BFQQ_FNS(sync);
 BFQ_BFQQ_FNS(IO_bound);
 BFQ_BFQQ_FNS(in_large_burst);
@@ -731,10 +731,10 @@ bfq_bfqq_resume_state(struct bfq_queue *bfqq, struct bfq_data *bfqd,
 	unsigned int old_wr_coeff = bfqq->wr_coeff;
 	bool busy = bfq_already_existing && bfq_bfqq_busy(bfqq);
 
-	if (bic->saved_idle_window)
-		bfq_mark_bfqq_idle_window(bfqq);
+	if (bic->saved_has_short_ttime)
+		bfq_mark_bfqq_has_short_ttime(bfqq);
 	else
-		bfq_clear_bfqq_idle_window(bfqq);
+		bfq_clear_bfqq_has_short_ttime(bfqq);
 
 	if (bic->saved_IO_bound)
 		bfq_mark_bfqq_IO_bound(bfqq);
@@ -2012,7 +2012,7 @@ static void bfq_bfqq_save_state(struct bfq_queue *bfqq)
 		return;
 
 	bic->saved_ttime = bfqq->ttime;
-	bic->saved_idle_window = bfq_bfqq_idle_window(bfqq);
+	bic->saved_has_short_ttime = bfq_bfqq_has_short_ttime(bfqq);
 	bic->saved_IO_bound = bfq_bfqq_IO_bound(bfqq);
 	bic->saved_in_large_burst = bfq_bfqq_in_large_burst(bfqq);
 	bic->was_in_burst_list = !hlist_unhashed(&bfqq->burst_list_node);
@@ -3038,8 +3038,8 @@ void bfq_bfqq_expire(struct bfq_data *bfqd,
 	}
 
 	bfq_log_bfqq(bfqd, bfqq,
-		"expire (%d, slow %d, num_disp %d, idle_win %d)", reason,
-		slow, bfqq->dispatched, bfq_bfqq_idle_window(bfqq));
+		"expire (%d, slow %d, num_disp %d, short_ttime %d)", reason,
+		slow, bfqq->dispatched, bfq_bfqq_has_short_ttime(bfqq));
 
 	/*
 	 * Increase, decrease or leave budget unchanged according to
@@ -3114,35 +3114,56 @@ static bool bfq_may_expire_for_budg_timeout(struct bfq_queue *bfqq)
 static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq)
 {
 	struct bfq_data *bfqd = bfqq->bfqd;
-	bool idling_boosts_thr, idling_boosts_thr_without_issues,
+	bool rot_without_queueing =
+		!blk_queue_nonrot(bfqd->queue) && !bfqd->hw_tag,
+		bfqq_sequential_and_IO_bound,
+		idling_boosts_thr, idling_boosts_thr_without_issues,
 		idling_needed_for_service_guarantees,
 		asymmetric_scenario;
 
 	if (bfqd->strict_guarantees)
 		return true;
 
+	/*
+	 * Idling is performed only if slice_idle > 0. In addition, we
+	 * do not idle if
+	 * (a) bfqq is async
+	 * (b) bfqq is in the idle io prio class: in this case we do
+	 * not idle because we want to minimize the bandwidth that
+	 * queues in this class can steal to higher-priority queues
+	 */
+	if (bfqd->bfq_slice_idle == 0 || !bfq_bfqq_sync(bfqq) ||
+	    bfq_class_idle(bfqq))
+		return false;
+
+	bfqq_sequential_and_IO_bound = !BFQQ_SEEKY(bfqq) &&
+		bfq_bfqq_IO_bound(bfqq) && bfq_bfqq_has_short_ttime(bfqq);
+
 	/*
 	 * The next variable takes into account the cases where idling
 	 * boosts the throughput.
 	 *
 	 * The value of the variable is computed considering, first, that
 	 * idling is virtually always beneficial for the throughput if:
-	 * (a) the device is not NCQ-capable, or
-	 * (b) regardless of the presence of NCQ, the device is rotational
-	 *     and the request pattern for bfqq is I/O-bound and sequential.
+	 * (a) the device is not NCQ-capable and rotational, or
+	 * (b) regardless of the presence of NCQ, the device is rotational and
+	 *     the request pattern for bfqq is I/O-bound and sequential, or
+	 * (c) regardless of whether it is rotational, the device is
+	 *     not NCQ-capable and the request pattern for bfqq is
+	 *     I/O-bound and sequential.
 	 *
 	 * Secondly, and in contrast to the above item (b), idling an
 	 * NCQ-capable flash-based device would not boost the
 	 * throughput even with sequential I/O; rather it would lower
 	 * the throughput in proportion to how fast the device
 	 * is. Accordingly, the next variable is true if any of the
-	 * above conditions (a) and (b) is true, and, in particular,
-	 * happens to be false if bfqd is an NCQ-capable flash-based
-	 * device.
+	 * above conditions (a), (b) or (c) is true, and, in
+	 * particular, happens to be false if bfqd is an NCQ-capable
+	 * flash-based device.
 	 */
-	idling_boosts_thr = !bfqd->hw_tag ||
-		(!blk_queue_nonrot(bfqd->queue) && bfq_bfqq_IO_bound(bfqq) &&
-		 bfq_bfqq_idle_window(bfqq));
+	idling_boosts_thr = rot_without_queueing ||
+		((!blk_queue_nonrot(bfqd->queue) || !bfqd->hw_tag) &&
+		 bfqq_sequential_and_IO_bound);
 
 	/*
 	 * The value of the next variable,
@@ -3313,16 +3334,13 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq)
 		asymmetric_scenario && !bfq_bfqq_in_large_burst(bfqq);
 
 	/*
-	 * We have now all the components we need to compute the return
-	 * value of the function, which is true only if both the following
-	 * conditions hold:
-	 * 1) bfqq is sync, because idling make sense only for sync queues;
-	 * 2) idling either boosts the throughput (without issues), or
-	 *    is necessary to preserve service guarantees.
+	 * We have now all the components we need to compute the
+	 * return value of the function, which is true only if idling
+	 * either boosts the throughput (without issues), or is
+	 * necessary to preserve service guarantees.
 	 */
-	return bfq_bfqq_sync(bfqq) &&
-		(idling_boosts_thr_without_issues ||
-		 idling_needed_for_service_guarantees);
+	return idling_boosts_thr_without_issues ||
+		idling_needed_for_service_guarantees;
 }
 
 /*
@@ -3338,10 +3356,7 @@ static bool bfq_bfqq_may_idle(struct bfq_queue *bfqq)
  */
 static bool bfq_bfqq_must_idle(struct bfq_queue *bfqq)
 {
-	struct bfq_data *bfqd = bfqq->bfqd;
-
-	return RB_EMPTY_ROOT(&bfqq->sort_list) && bfqd->bfq_slice_idle != 0 &&
-	       bfq_bfqq_may_idle(bfqq);
+	return RB_EMPTY_ROOT(&bfqq->sort_list) && bfq_bfqq_may_idle(bfqq);
 }
 
 /*
@@ -3783,7 +3798,6 @@ bfq_set_next_ioprio_data(struct bfq_queue *bfqq, struct bfq_io_cq *bic)
 	case IOPRIO_CLASS_IDLE:
 		bfqq->new_ioprio_class = IOPRIO_CLASS_IDLE;
 		bfqq->new_ioprio = 7;
-		bfq_clear_bfqq_idle_window(bfqq);
 		break;
 	}
 
@@ -3843,8 +3857,14 @@ static void bfq_init_bfqq(struct bfq_data *bfqd, struct bfq_queue *bfqq,
 		bfq_set_next_ioprio_data(bfqq, bic);
 
 	if (is_sync) {
+		/*
+		 * No need to mark as has_short_ttime if in
+		 * idle_class, because no device idling is performed
+		 * for queues in idle class
+		 */
 		if (!bfq_class_idle(bfqq))
-			bfq_mark_bfqq_idle_window(bfqq);
+			/* tentatively mark as has_short_ttime */
+			bfq_mark_bfqq_has_short_ttime(bfqq);
 		bfq_mark_bfqq_sync(bfqq);
 		bfq_mark_bfqq_just_created(bfqq);
 	} else
@@ -3985,18 +4005,19 @@ bfq_update_io_seektime(struct bfq_data *bfqd, struct bfq_queue *bfqq,
 		 blk_rq_sectors(rq) < BFQQ_SECT_THR_NONROT);
 }
 
-/*
- * Disable idle window if the process thinks too long or seeks so much that
- * it doesn't matter.
- */
-static void bfq_update_idle_window(struct bfq_data *bfqd,
-				   struct bfq_queue *bfqq,
-				   struct bfq_io_cq *bic)
+static void bfq_update_has_short_ttime(struct bfq_data *bfqd,
+				       struct bfq_queue *bfqq,
+				       struct bfq_io_cq *bic)
 {
-	int enable_idle;
+	bool has_short_ttime = true;
 
-	/* Don't idle for async or idle io prio class. */
-	if (!bfq_bfqq_sync(bfqq) || bfq_class_idle(bfqq))
+	/*
+	 * No need to update has_short_ttime if bfqq is async or in
+	 * idle io prio class, or if bfq_slice_idle is zero, because
+	 * no device idling is performed for bfqq in this case.
+	 */
+	if (!bfq_bfqq_sync(bfqq) || bfq_class_idle(bfqq) ||
+	    bfqd->bfq_slice_idle == 0)
 		return;
 
 	/* Idle window just restored, statistics are meaningless. */
@@ -4004,27 +4025,22 @@ static void bfq_update_idle_window(struct bfq_data *bfqd,
 				     bfqd->bfq_wr_min_idle_time))
 		return;
 
-	enable_idle = bfq_bfqq_idle_window(bfqq);
-
+	/* Think time is infinite if no process is linked to
+	 * bfqq. Otherwise check average think time to
+	 * decide whether to mark as has_short_ttime
+	 */
 	if (atomic_read(&bic->icq.ioc->active_ref) == 0 ||
-	    bfqd->bfq_slice_idle == 0 ||
-		(bfqd->hw_tag && BFQQ_SEEKY(bfqq) &&
-			bfqq->wr_coeff == 1))
-		enable_idle = 0;
-	else if (bfq_sample_valid(bfqq->ttime.ttime_samples)) {
-		if (bfqq->ttime.ttime_mean > bfqd->bfq_slice_idle &&
-			bfqq->wr_coeff == 1)
-			enable_idle = 0;
-		else
-			enable_idle = 1;
-	}
-	bfq_log_bfqq(bfqd, bfqq, "update_idle_window: enable_idle %d",
-		enable_idle);
+	    (bfq_sample_valid(bfqq->ttime.ttime_samples) &&
+	     bfqq->ttime.ttime_mean > bfqd->bfq_slice_idle))
+		has_short_ttime = false;
 
-	if (enable_idle)
-		bfq_mark_bfqq_idle_window(bfqq);
+	bfq_log_bfqq(bfqd, bfqq, "update_has_short_ttime: has_short_ttime %d",
+		     has_short_ttime);
+
+	if (has_short_ttime)
+		bfq_mark_bfqq_has_short_ttime(bfqq);
 	else
-		bfq_clear_bfqq_idle_window(bfqq);
+		bfq_clear_bfqq_has_short_ttime(bfqq);
 }
 
 /*
@@ -4040,14 +4056,12 @@ static void bfq_rq_enqueued(struct bfq_data *bfqd, struct bfq_queue *bfqq,
 		bfqq->meta_pending++;
 
 	bfq_update_io_thinktime(bfqd, bfqq);
+	bfq_update_has_short_ttime(bfqd, bfqq, bic);
 	bfq_update_io_seektime(bfqd, bfqq, rq);
-	if (bfqq->entity.service > bfq_max_budget(bfqd) / 8 ||
-	    !BFQQ_SEEKY(bfqq))
-		bfq_update_idle_window(bfqd, bfqq, bic);
 
 	bfq_log_bfqq(bfqd, bfqq,
-		     "rq_enqueued: idle_window=%d (seeky %d)",
-		     bfq_bfqq_idle_window(bfqq), BFQQ_SEEKY(bfqq));
+		     "rq_enqueued: has_short_ttime=%d (seeky %d)",
+		     bfq_bfqq_has_short_ttime(bfqq), BFQQ_SEEKY(bfqq));
 
 	bfqq->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq);
 
@@ -4787,16 +4801,13 @@ static ssize_t bfq_var_show(unsigned int var, char *page)
 	return sprintf(page, "%u\n", var);
 }
 
-static ssize_t bfq_var_store(unsigned long *var, const char *page,
-			     size_t count)
+static void bfq_var_store(unsigned long *var, const char *page)
 {
 	unsigned long new_val;
 	int ret = kstrtoul(page, 10, &new_val);
 
 	if (ret == 0)
 		*var = new_val;
-
-	return count;
 }
 
 #define SHOW_FUNCTION(__FUNC, __VAR, __CONV)				\
@@ -4838,7 +4849,7 @@ __FUNC(struct elevator_queue *e, const char *page, size_t count)	\
 {									\
 	struct bfq_data *bfqd = e->elevator_data;			\
 	unsigned long uninitialized_var(__data);			\
-	int ret = bfq_var_store(&__data, (page), count);		\
+	bfq_var_store(&__data, (page));					\
 	if (__data < (MIN))						\
 		__data = (MIN);						\
 	else if (__data > (MAX))					\
@@ -4849,7 +4860,7 @@ __FUNC(struct elevator_queue *e, const char *page, size_t count)	\
 		*(__PTR) = (u64)__data * NSEC_PER_MSEC;			\
 	else								\
 		*(__PTR) = __data;					\
-	return ret;							\
+	return count;							\
 }
 STORE_FUNCTION(bfq_fifo_expire_sync_store, &bfqd->bfq_fifo_expire[1], 1,
 		INT_MAX, 2);
@@ -4866,13 +4877,13 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)\
 {									\
 	struct bfq_data *bfqd = e->elevator_data;			\
 	unsigned long uninitialized_var(__data);			\
-	int ret = bfq_var_store(&__data, (page), count);		\
+	bfq_var_store(&__data, (page));					\
 	if (__data < (MIN))						\
 		__data = (MIN);						\
 	else if (__data > (MAX))					\
 		__data = (MAX);						\
 	*(__PTR) = (u64)__data * NSEC_PER_USEC;				\
-	return ret;							\
+	return count;							\
 }
 USEC_STORE_FUNCTION(bfq_slice_idle_us_store, &bfqd->bfq_slice_idle, 0,
 		    UINT_MAX);
@@ -4883,7 +4894,8 @@ static ssize_t bfq_max_budget_store(struct elevator_queue *e,
 {
 	struct bfq_data *bfqd = e->elevator_data;
 	unsigned long uninitialized_var(__data);
-	int ret = bfq_var_store(&__data, (page), count);
+
+	bfq_var_store(&__data, (page));
 
 	if (__data == 0)
 		bfqd->bfq_max_budget = bfq_calc_max_budget(bfqd);
@@ -4895,7 +4907,7 @@ static ssize_t bfq_max_budget_store(struct elevator_queue *e,
 
 	bfqd->bfq_user_max_budget = __data;
 
-	return ret;
+	return count;
 }
 
 /*
@@ -4907,7 +4919,8 @@ static ssize_t bfq_timeout_sync_store(struct elevator_queue *e,
 {
 	struct bfq_data *bfqd = e->elevator_data;
 	unsigned long uninitialized_var(__data);
-	int ret = bfq_var_store(&__data, (page), count);
+
+	bfq_var_store(&__data, (page));
 
 	if (__data < 1)
 		__data = 1;
@@ -4918,7 +4931,7 @@ static ssize_t bfq_timeout_sync_store(struct elevator_queue *e,
 	if (bfqd->bfq_user_max_budget == 0)
 		bfqd->bfq_max_budget = bfq_calc_max_budget(bfqd);
 
-	return ret;
+	return count;
 }
 
 static ssize_t bfq_strict_guarantees_store(struct elevator_queue *e,
@@ -4926,7 +4939,8 @@ static ssize_t bfq_strict_guarantees_store(struct elevator_queue *e,
 {
 	struct bfq_data *bfqd = e->elevator_data;
 	unsigned long uninitialized_var(__data);
-	int ret = bfq_var_store(&__data, (page), count);
+
+	bfq_var_store(&__data, (page));
 
 	if (__data > 1)
 		__data = 1;
@@ -4936,7 +4950,7 @@ static ssize_t bfq_strict_guarantees_store(struct elevator_queue *e,
 
 	bfqd->strict_guarantees = __data;
 
-	return ret;
+	return count;
 }
 
 static ssize_t bfq_low_latency_store(struct elevator_queue *e,
@@ -4944,7 +4958,8 @@ static ssize_t bfq_low_latency_store(struct elevator_queue *e,
 {
 	struct bfq_data *bfqd = e->elevator_data;
 	unsigned long uninitialized_var(__data);
-	int ret = bfq_var_store(&__data, (page), count);
+
+	bfq_var_store(&__data, (page));
 
 	if (__data > 1)
 		__data = 1;
@@ -4952,7 +4967,7 @@ static ssize_t bfq_low_latency_store(struct elevator_queue *e,
 		bfq_end_wr(bfqd);
 	bfqd->low_latency = __data;
 
-	return ret;
+	return count;
 }
 
 #define BFQ_ATTR(name) \
@@ -4998,6 +5013,7 @@ static struct elevator_type iosched_bfq_mq = {
 	.elevator_name =	"bfq",
 	.elevator_owner =	THIS_MODULE,
 };
+MODULE_ALIAS("bfq-iosched");
 
 static int __init bfq_init(void)
 {
@@ -5048,10 +5064,12 @@ static int __init bfq_init(void)
 
 	ret = elv_register(&iosched_bfq_mq);
 	if (ret)
-		goto err_pol_unreg;
+		goto slab_kill;
 
 	return 0;
 
+slab_kill:
+	bfq_slab_kill();
 err_pol_unreg:
 #ifdef CONFIG_BFQ_GROUP_IOSCHED
 	blkcg_policy_unregister(&blkcg_policy_bfq);

block/bfq-iosched.h

@@ -360,11 +360,11 @@ struct bfq_io_cq {
 	uint64_t blkcg_serial_nr; /* the current blkcg serial */
 #endif
 	/*
-	 * Snapshot of the idle window before merging; taken to
-	 * remember this value while the queue is merged, so as to be
-	 * able to restore it in case of split.
+	 * Snapshot of the has_short_time flag before merging; taken
+	 * to remember its value while the queue is merged, so as to
+	 * be able to restore it in case of split.
 	 */
-	bool saved_idle_window;
+	bool saved_has_short_ttime;
 	/*
 	 * Same purpose as the previous two fields for the I/O bound
 	 * classification of a queue.
@@ -638,7 +638,7 @@ enum bfqq_state_flags {
 				     * without idling the device
 				     */
 	BFQQF_fifo_expire,	/* FIFO checked in this slice */
-	BFQQF_idle_window,	/* slice idling enabled */
+	BFQQF_has_short_ttime,	/* queue has a short think time */
 	BFQQF_sync,		/* synchronous queue */
 	BFQQF_IO_bound,		/*
 				 * bfqq has timed-out at least once
@@ -667,7 +667,7 @@ BFQ_BFQQ_FNS(busy);
 BFQ_BFQQ_FNS(wait_request);
 BFQ_BFQQ_FNS(non_blocking_wait_rq);
 BFQ_BFQQ_FNS(fifo_expire);
-BFQ_BFQQ_FNS(idle_window);
+BFQ_BFQQ_FNS(has_short_ttime);
 BFQ_BFQQ_FNS(sync);
 BFQ_BFQQ_FNS(IO_bound);
 BFQ_BFQQ_FNS(in_large_burst);
@@ -929,13 +929,16 @@ 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 {			\
-	blk_add_trace_msg((bfqd)->queue, "bfq%d%c %s " fmt, (bfqq)->pid,\
-			bfq_bfqq_sync((bfqq)) ? 'S' : 'A',		\
-			bfqq_group(bfqq)->blkg_path, ##args);		\
+	blk_add_cgroup_trace_msg((bfqd)->queue,				\
+			bfqg_to_blkg(bfqq_group(bfqq))->blkcg,		\
+			"bfq%d%c " fmt, (bfqq)->pid,			\
+			bfq_bfqq_sync((bfqq)) ? 'S' : 'A', ##args);	\
 } while (0)
 
-#define bfq_log_bfqg(bfqd, bfqg, fmt, args...)	\
-	blk_add_trace_msg((bfqd)->queue, "%s " fmt, (bfqg)->blkg_path, ##args)
+#define bfq_log_bfqg(bfqd, bfqg, fmt, args...)	do {			\
+	blk_add_cgroup_trace_msg((bfqd)->queue,				\
+		bfqg_to_blkg(bfqg)->blkcg, fmt, ##args);		\
+} while (0)
 
 #else /* CONFIG_BFQ_GROUP_IOSCHED */
 

block/bio-integrity.c

@@ -146,7 +146,7 @@ int bio_integrity_add_page(struct bio *bio, struct page *page,
 	iv = bip->bip_vec + bip->bip_vcnt;
 
 	if (bip->bip_vcnt &&
-	    bvec_gap_to_prev(bdev_get_queue(bio->bi_bdev),
+	    bvec_gap_to_prev(bio->bi_disk->queue,
 			     &bip->bip_vec[bip->bip_vcnt - 1], offset))
 		return 0;
 
@@ -190,7 +190,7 @@ static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi,
 static blk_status_t bio_integrity_process(struct bio *bio,
 		struct bvec_iter *proc_iter, integrity_processing_fn *proc_fn)
 {
-	struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev);
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
 	struct blk_integrity_iter iter;
 	struct bvec_iter bviter;
 	struct bio_vec bv;
@@ -199,7 +199,7 @@ static blk_status_t bio_integrity_process(struct bio *bio,
 	void *prot_buf = page_address(bip->bip_vec->bv_page) +
 		bip->bip_vec->bv_offset;
 
-	iter.disk_name = bio->bi_bdev->bd_disk->disk_name;
+	iter.disk_name = bio->bi_disk->disk_name;
 	iter.interval = 1 << bi->interval_exp;
 	iter.seed = proc_iter->bi_sector;
 	iter.prot_buf = prot_buf;
@@ -236,8 +236,8 @@ static blk_status_t bio_integrity_process(struct bio *bio,
 bool bio_integrity_prep(struct bio *bio)
 {
 	struct bio_integrity_payload *bip;
-	struct blk_integrity *bi;
-	struct request_queue *q;
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
+	struct request_queue *q = bio->bi_disk->queue;
 	void *buf;
 	unsigned long start, end;
 	unsigned int len, nr_pages;
@@ -245,8 +245,9 @@ bool bio_integrity_prep(struct bio *bio)
 	unsigned int intervals;
 	blk_status_t status;
 
-	bi = bdev_get_integrity(bio->bi_bdev);
-	q = bdev_get_queue(bio->bi_bdev);
+	if (!bi)
+		return true;
+
 	if (bio_op(bio) != REQ_OP_READ && bio_op(bio) != REQ_OP_WRITE)
 		return true;
 
@@ -257,9 +258,6 @@ bool bio_integrity_prep(struct bio *bio)
 	if (bio_integrity(bio))
 		return true;
 
-	if (bi == NULL)
-		return true;
-
 	if (bio_data_dir(bio) == READ) {
 		if (!bi->profile->verify_fn ||
 		    !(bi->flags & BLK_INTEGRITY_VERIFY))
@@ -354,7 +352,7 @@ static void bio_integrity_verify_fn(struct work_struct *work)
 	struct bio_integrity_payload *bip =
 		container_of(work, struct bio_integrity_payload, bip_work);
 	struct bio *bio = bip->bip_bio;
-	struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev);
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
 	struct bvec_iter iter = bio->bi_iter;
 
 	/*
@@ -387,7 +385,7 @@ static void bio_integrity_verify_fn(struct work_struct *work)
  */
 bool __bio_integrity_endio(struct bio *bio)
 {
-	struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev);
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
 	struct bio_integrity_payload *bip = bio_integrity(bio);
 
 	if (bio_op(bio) == REQ_OP_READ && !bio->bi_status &&
@@ -413,7 +411,7 @@ bool __bio_integrity_endio(struct bio *bio)
 void bio_integrity_advance(struct bio *bio, unsigned int bytes_done)
 {
 	struct bio_integrity_payload *bip = bio_integrity(bio);
-	struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev);
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
 	unsigned bytes = bio_integrity_bytes(bi, bytes_done >> 9);
 
 	bip->bip_iter.bi_sector += bytes_done >> 9;
@@ -430,7 +428,7 @@ EXPORT_SYMBOL(bio_integrity_advance);
 void bio_integrity_trim(struct bio *bio)
 {
 	struct bio_integrity_payload *bip = bio_integrity(bio);
-	struct blk_integrity *bi = bdev_get_integrity(bio->bi_bdev);
+	struct blk_integrity *bi = blk_get_integrity(bio->bi_disk);
 
 	bip->bip_iter.bi_size = bio_integrity_bytes(bi, bio_sectors(bio));
 }

block/bio.c

@@ -593,10 +593,10 @@ void __bio_clone_fast(struct bio *bio, struct bio *bio_src)
 	BUG_ON(bio->bi_pool && BVEC_POOL_IDX(bio));
 
 	/*
-	 * most users will be overriding ->bi_bdev with a new target,
+	 * most users will be overriding ->bi_disk with a new target,
 	 * so we don't set nor calculate new physical/hw segment counts here
 	 */
-	bio->bi_bdev = bio_src->bi_bdev;
+	bio->bi_disk = bio_src->bi_disk;
 	bio_set_flag(bio, BIO_CLONED);
 	bio->bi_opf = bio_src->bi_opf;
 	bio->bi_write_hint = bio_src->bi_write_hint;
@@ -681,7 +681,7 @@ struct bio *bio_clone_bioset(struct bio *bio_src, gfp_t gfp_mask,
 	bio = bio_alloc_bioset(gfp_mask, bio_segments(bio_src), bs);
 	if (!bio)
 		return NULL;
-	bio->bi_bdev		= bio_src->bi_bdev;
+	bio->bi_disk		= bio_src->bi_disk;
 	bio->bi_opf		= bio_src->bi_opf;
 	bio->bi_write_hint	= bio_src->bi_write_hint;
 	bio->bi_iter.bi_sector	= bio_src->bi_iter.bi_sector;
@@ -936,6 +936,10 @@ static void submit_bio_wait_endio(struct bio *bio)
  *
  * Simple wrapper around submit_bio(). Returns 0 on success, or the error from
  * bio_endio() on failure.
+ *
+ * WARNING: Unlike to how submit_bio() is usually used, this function does not
+ * result in bio reference to be consumed. The caller must drop the reference
+ * on his own.
  */
 int submit_bio_wait(struct bio *bio)
 {
@@ -1732,29 +1736,29 @@ void bio_check_pages_dirty(struct bio *bio)
 	}
 }
 
-void generic_start_io_acct(int rw, unsigned long sectors,
-			   struct hd_struct *part)
+void generic_start_io_acct(struct request_queue *q, int rw,
+			   unsigned long sectors, struct hd_struct *part)
 {
 	int cpu = part_stat_lock();
 
-	part_round_stats(cpu, part);
+	part_round_stats(q, cpu, part);
 	part_stat_inc(cpu, part, ios[rw]);
 	part_stat_add(cpu, part, sectors[rw], sectors);
-	part_inc_in_flight(part, rw);
+	part_inc_in_flight(q, part, rw);
 
 	part_stat_unlock();
 }
 EXPORT_SYMBOL(generic_start_io_acct);
 
-void generic_end_io_acct(int rw, struct hd_struct *part,
-			 unsigned long start_time)
+void generic_end_io_acct(struct request_queue *q, int rw,
+			 struct hd_struct *part, unsigned long start_time)
 {
 	unsigned long duration = jiffies - start_time;
 	int cpu = part_stat_lock();
 
 	part_stat_add(cpu, part, ticks[rw], duration);
-	part_round_stats(cpu, part);
-	part_dec_in_flight(part, rw);
+	part_round_stats(q, cpu, part);
+	part_dec_in_flight(q, part, rw);
 
 	part_stat_unlock();
 }
@@ -1826,8 +1830,8 @@ again:
 		goto again;
 	}
 
-	if (bio->bi_bdev && bio_flagged(bio, BIO_TRACE_COMPLETION)) {
-		trace_block_bio_complete(bdev_get_queue(bio->bi_bdev), bio,
+	if (bio->bi_disk && bio_flagged(bio, BIO_TRACE_COMPLETION)) {
+		trace_block_bio_complete(bio->bi_disk->queue, bio,
 					 blk_status_to_errno(bio->bi_status));
 		bio_clear_flag(bio, BIO_TRACE_COMPLETION);
 	}

block/blk-cgroup.c

@@ -1067,7 +1067,7 @@ blkcg_css_alloc(struct cgroup_subsys_state *parent_css)
 		blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
 		if (!blkcg) {
 			ret = ERR_PTR(-ENOMEM);
-			goto free_blkcg;
+			goto unlock;
 		}
 	}
 
@@ -1111,8 +1111,10 @@ free_pd_blkcg:
 	for (i--; i >= 0; i--)
 		if (blkcg->cpd[i])
 			blkcg_policy[i]->cpd_free_fn(blkcg->cpd[i]);
-free_blkcg:
-	kfree(blkcg);
+
+	if (blkcg != &blkcg_root)
+		kfree(blkcg);
+unlock:
 	mutex_unlock(&blkcg_pol_mutex);
 	return ret;
 }

block/blk-core.c

@@ -280,7 +280,7 @@ EXPORT_SYMBOL(blk_start_queue_async);
 void blk_start_queue(struct request_queue *q)
 {
 	lockdep_assert_held(q->queue_lock);
-	WARN_ON(!irqs_disabled());
+	WARN_ON(!in_interrupt() && !irqs_disabled());
 	WARN_ON_ONCE(q->mq_ops);
 
 	queue_flag_clear(QUEUE_FLAG_STOPPED, q);
@@ -1469,15 +1469,10 @@ static void add_acct_request(struct request_queue *q, struct request *rq,
 	__elv_add_request(q, rq, where);
 }
 
-static void part_round_stats_single(int cpu, struct hd_struct *part,
-				    unsigned long now)
+static void part_round_stats_single(struct request_queue *q, int cpu,
+				    struct hd_struct *part, unsigned long now,
+				    unsigned int inflight)
 {
-	int inflight;
-
-	if (now == part->stamp)
-		return;
-
-	inflight = part_in_flight(part);
 	if (inflight) {
 		__part_stat_add(cpu, part, time_in_queue,
 				inflight * (now - part->stamp));
@@ -1488,6 +1483,7 @@ static void part_round_stats_single(int cpu, struct hd_struct *part,
 
 /**
  * part_round_stats() - Round off the performance stats on a struct disk_stats.
+ * @q: target block queue
  * @cpu: cpu number for stats access
  * @part: target partition
  *
@@ -1502,13 +1498,31 @@ static void part_round_stats_single(int cpu, struct hd_struct *part,
  * /proc/diskstats.  This accounts immediately for all queue usage up to
  * the current jiffies and restarts the counters again.
  */
-void part_round_stats(int cpu, struct hd_struct *part)
+void part_round_stats(struct request_queue *q, int cpu, struct hd_struct *part)
 {
+	struct hd_struct *part2 = NULL;
 	unsigned long now = jiffies;
+	unsigned int inflight[2];
+	int stats = 0;
+
+	if (part->stamp != now)
+		stats |= 1;
+
+	if (part->partno) {
+		part2 = &part_to_disk(part)->part0;
+		if (part2->stamp != now)
+			stats |= 2;
+	}
 
-	if (part->partno)
-		part_round_stats_single(cpu, &part_to_disk(part)->part0, now);
-	part_round_stats_single(cpu, part, now);
+	if (!stats)
+		return;
+
+	part_in_flight(q, part, inflight);
+
+	if (stats & 2)
+		part_round_stats_single(q, cpu, part2, now, inflight[1]);
+	if (stats & 1)
+		part_round_stats_single(q, cpu, part, now, inflight[0]);
 }
 EXPORT_SYMBOL_GPL(part_round_stats);
 
@@ -1896,40 +1910,15 @@ out_unlock:
 	return BLK_QC_T_NONE;
 }
 
-/*
- * If bio->bi_dev is a partition, remap the location
- */
-static inline void blk_partition_remap(struct bio *bio)
-{
-	struct block_device *bdev = bio->bi_bdev;
-
-	/*
-	 * Zone reset does not include bi_size so bio_sectors() is always 0.
-	 * Include a test for the reset op code and perform the remap if needed.
-	 */
-	if (bdev != bdev->bd_contains &&
-	    (bio_sectors(bio) || bio_op(bio) == REQ_OP_ZONE_RESET)) {
-		struct hd_struct *p = bdev->bd_part;
-
-		bio->bi_iter.bi_sector += p->start_sect;
-		bio->bi_bdev = bdev->bd_contains;
-
-		trace_block_bio_remap(bdev_get_queue(bio->bi_bdev), bio,
-				      bdev->bd_dev,
-				      bio->bi_iter.bi_sector - p->start_sect);
-	}
-}
-
 static void handle_bad_sector(struct bio *bio)
 {
 	char b[BDEVNAME_SIZE];
 
 	printk(KERN_INFO "attempt to access beyond end of device\n");
 	printk(KERN_INFO "%s: rw=%d, want=%Lu, limit=%Lu\n",
-			bdevname(bio->bi_bdev, b),
-			bio->bi_opf,
+			bio_devname(bio, b), bio->bi_opf,
 			(unsigned long long)bio_end_sector(bio),
-			(long long)(i_size_read(bio->bi_bdev->bd_inode) >> 9));
+			(long long)get_capacity(bio->bi_disk));
 }
 
 #ifdef CONFIG_FAIL_MAKE_REQUEST
@@ -1967,6 +1956,38 @@ static inline bool should_fail_request(struct hd_struct *part,
 
 #endif /* CONFIG_FAIL_MAKE_REQUEST */
 
+/*
+ * Remap block n of partition p to block n+start(p) of the disk.
+ */
+static inline int blk_partition_remap(struct bio *bio)
+{
+	struct hd_struct *p;
+	int ret = 0;
+
+	/*
+	 * Zone reset does not include bi_size so bio_sectors() is always 0.
+	 * Include a test for the reset op code and perform the remap if needed.
+	 */
+	if (!bio->bi_partno ||
+	    (!bio_sectors(bio) && bio_op(bio) != REQ_OP_ZONE_RESET))
+		return 0;
+
+	rcu_read_lock();
+	p = __disk_get_part(bio->bi_disk, bio->bi_partno);
+	if (likely(p && !should_fail_request(p, bio->bi_iter.bi_size))) {
+		bio->bi_iter.bi_sector += p->start_sect;
+		bio->bi_partno = 0;
+		trace_block_bio_remap(bio->bi_disk->queue, bio, part_devt(p),
+				bio->bi_iter.bi_sector - p->start_sect);
+	} else {
+		printk("%s: fail for partition %d\n", __func__, bio->bi_partno);
+		ret = -EIO;
+	}
+	rcu_read_unlock();
+
+	return ret;
+}
+
 /*
  * Check whether this bio extends beyond the end of the device.
  */
@@ -1978,7 +1999,7 @@ static inline int bio_check_eod(struct bio *bio, unsigned int nr_sectors)
 		return 0;
 
 	/* Test device or partition size, when known. */
-	maxsector = i_size_read(bio->bi_bdev->bd_inode) >> 9;
+	maxsector = get_capacity(bio->bi_disk);
 	if (maxsector) {
 		sector_t sector = bio->bi_iter.bi_sector;
 
@@ -2003,20 +2024,18 @@ generic_make_request_checks(struct bio *bio)
 	int nr_sectors = bio_sectors(bio);
 	blk_status_t status = BLK_STS_IOERR;
 	char b[BDEVNAME_SIZE];
-	struct hd_struct *part;
 
 	might_sleep();
 
 	if (bio_check_eod(bio, nr_sectors))
 		goto end_io;
 
-	q = bdev_get_queue(bio->bi_bdev);
+	q = bio->bi_disk->queue;
 	if (unlikely(!q)) {
 		printk(KERN_ERR
 		       "generic_make_request: Trying to access "
 			"nonexistent block-device %s (%Lu)\n",
-			bdevname(bio->bi_bdev, b),
-			(long long) bio->bi_iter.bi_sector);
+			bio_devname(bio, b), (long long)bio->bi_iter.bi_sector);
 		goto end_io;
 	}
 
@@ -2028,17 +2047,11 @@ generic_make_request_checks(struct bio *bio)
 	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_rq_based(q))
 		goto not_supported;
 
-	part = bio->bi_bdev->bd_part;
-	if (should_fail_request(part, bio->bi_iter.bi_size) ||
-	    should_fail_request(&part_to_disk(part)->part0,
-				bio->bi_iter.bi_size))
+	if (should_fail_request(&bio->bi_disk->part0, bio->bi_iter.bi_size))
 		goto end_io;
 
-	/*
-	 * If this device has partitions, remap block n
-	 * of partition p to block n+start(p) of the disk.
-	 */
-	blk_partition_remap(bio);
+	if (blk_partition_remap(bio))
+		goto end_io;
 
 	if (bio_check_eod(bio, nr_sectors))
 		goto end_io;
@@ -2067,16 +2080,16 @@ generic_make_request_checks(struct bio *bio)
 			goto not_supported;
 		break;
 	case REQ_OP_WRITE_SAME:
-		if (!bdev_write_same(bio->bi_bdev))
+		if (!q->limits.max_write_same_sectors)
 			goto not_supported;
 		break;
 	case REQ_OP_ZONE_REPORT:
 	case REQ_OP_ZONE_RESET:
-		if (!bdev_is_zoned(bio->bi_bdev))
+		if (!blk_queue_is_zoned(q))
 			goto not_supported;
 		break;
 	case REQ_OP_WRITE_ZEROES:
-		if (!bdev_write_zeroes_sectors(bio->bi_bdev))
+		if (!q->limits.max_write_zeroes_sectors)
 			goto not_supported;
 		break;
 	default:
@@ -2183,7 +2196,7 @@ blk_qc_t generic_make_request(struct bio *bio)
 	bio_list_init(&bio_list_on_stack[0]);
 	current->bio_list = bio_list_on_stack;
 	do {
-		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
+		struct request_queue *q = bio->bi_disk->queue;
 
 		if (likely(blk_queue_enter(q, bio->bi_opf & REQ_NOWAIT) == 0)) {
 			struct bio_list lower, same;
@@ -2201,7 +2214,7 @@ blk_qc_t generic_make_request(struct bio *bio)
 			bio_list_init(&lower);
 			bio_list_init(&same);
 			while ((bio = bio_list_pop(&bio_list_on_stack[0])) != NULL)
-				if (q == bdev_get_queue(bio->bi_bdev))
+				if (q == bio->bi_disk->queue)
 					bio_list_add(&same, bio);
 				else
 					bio_list_add(&lower, bio);
@@ -2244,7 +2257,7 @@ blk_qc_t submit_bio(struct bio *bio)
 		unsigned int count;
 
 		if (unlikely(bio_op(bio) == REQ_OP_WRITE_SAME))
-			count = bdev_logical_block_size(bio->bi_bdev) >> 9;
+			count = queue_logical_block_size(bio->bi_disk->queue);
 		else
 			count = bio_sectors(bio);
 
@@ -2261,8 +2274,7 @@ blk_qc_t submit_bio(struct bio *bio)
 			current->comm, task_pid_nr(current),
 				op_is_write(bio_op(bio)) ? "WRITE" : "READ",
 				(unsigned long long)bio->bi_iter.bi_sector,
-				bdevname(bio->bi_bdev, b),
-				count);
+				bio_devname(bio, b), count);
 		}
 	}
 
@@ -2431,8 +2443,8 @@ void blk_account_io_done(struct request *req)
 
 		part_stat_inc(cpu, part, ios[rw]);
 		part_stat_add(cpu, part, ticks[rw], duration);
-		part_round_stats(cpu, part);
-		part_dec_in_flight(part, rw);
+		part_round_stats(req->q, cpu, part);
+		part_dec_in_flight(req->q, part, rw);
 
 		hd_struct_put(part);
 		part_stat_unlock();
@@ -2489,8 +2501,8 @@ void blk_account_io_start(struct request *rq, bool new_io)
 			part = &rq->rq_disk->part0;
 			hd_struct_get(part);
 		}
-		part_round_stats(cpu, part);
-		part_inc_in_flight(part, rw);
+		part_round_stats(rq->q, cpu, part);
+		part_inc_in_flight(rq->q, part, rw);
 		rq->part = part;
 	}
 
@@ -2603,7 +2615,7 @@ struct request *blk_peek_request(struct request_queue *q)
 }
 EXPORT_SYMBOL(blk_peek_request);
 
-void blk_dequeue_request(struct request *rq)
+static void blk_dequeue_request(struct request *rq)
 {
 	struct request_queue *q = rq->q;
 
@@ -2630,9 +2642,6 @@ void blk_dequeue_request(struct request *rq)
  * Description:
  *     Dequeue @req and start timeout timer on it.  This hands off the
  *     request to the driver.
- *
- *     Block internal functions which don't want to start timer should
- *     call blk_dequeue_request().
  */
 void blk_start_request(struct request *req)
 {
@@ -3035,8 +3044,8 @@ void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
 	rq->__data_len = bio->bi_iter.bi_size;
 	rq->bio = rq->biotail = bio;
 
-	if (bio->bi_bdev)
-		rq->rq_disk = bio->bi_bdev->bd_disk;
+	if (bio->bi_disk)
+		rq->rq_disk = bio->bi_disk;
 }
 
 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE

block/blk-flush.c

@@ -1,12 +1,12 @@
 /*
- * Functions to sequence FLUSH and FUA writes.
+ * Functions to sequence PREFLUSH and FUA writes.
  *
  * Copyright (C) 2011		Max Planck Institute for Gravitational Physics
  * Copyright (C) 2011		Tejun Heo <tj@kernel.org>
  *
  * This file is released under the GPLv2.
  *
- * REQ_{FLUSH|FUA} requests are decomposed to sequences consisted of three
+ * REQ_{PREFLUSH|FUA} requests are decomposed to sequences consisted of three
  * optional steps - PREFLUSH, DATA and POSTFLUSH - according to the request
  * properties and hardware capability.
  *
@@ -16,9 +16,9 @@
  * REQ_FUA means that the data must be on non-volatile media on request
  * completion.
  *
- * If the device doesn't have writeback cache, FLUSH and FUA don't make any
- * difference.  The requests are either completed immediately if there's no
- * data or executed as normal requests otherwise.
+ * If the device doesn't have writeback cache, PREFLUSH and FUA don't make any
+ * difference.  The requests are either completed immediately if there's no data
+ * or executed as normal requests otherwise.
  *
  * If the device has writeback cache and supports FUA, REQ_PREFLUSH is
  * translated to PREFLUSH but REQ_FUA is passed down directly with DATA.
@@ -31,7 +31,7 @@
  * fq->flush_queue[fq->flush_pending_idx].  Once certain criteria are met, a
  * REQ_OP_FLUSH is issued and the pending_idx is toggled.  When the flush
  * completes, all the requests which were pending are proceeded to the next
- * step.  This allows arbitrary merging of different types of FLUSH/FUA
+ * step.  This allows arbitrary merging of different types of PREFLUSH/FUA
  * requests.
  *
  * Currently, the following conditions are used to determine when to issue
@@ -47,19 +47,19 @@
  * C3. The second condition is ignored if there is a request which has
  *     waited longer than FLUSH_PENDING_TIMEOUT.  This is to avoid
  *     starvation in the unlikely case where there are continuous stream of
- *     FUA (without FLUSH) requests.
+ *     FUA (without PREFLUSH) requests.
  *
  * For devices which support FUA, it isn't clear whether C2 (and thus C3)
  * is beneficial.
  *
- * Note that a sequenced FLUSH/FUA request with DATA is completed twice.
+ * Note that a sequenced PREFLUSH/FUA request with DATA is completed twice.
  * Once while executing DATA and again after the whole sequence is
  * complete.  The first completion updates the contained bio but doesn't
  * finish it so that the bio submitter is notified only after the whole
  * sequence is complete.  This is implemented by testing RQF_FLUSH_SEQ in
  * req_bio_endio().
  *
- * The above peculiarity requires that each FLUSH/FUA request has only one
+ * The above peculiarity requires that each PREFLUSH/FUA request has only one
  * bio attached to it, which is guaranteed as they aren't allowed to be
  * merged in the usual way.
  */
@@ -76,7 +76,7 @@
 #include "blk-mq-tag.h"
 #include "blk-mq-sched.h"
 
-/* FLUSH/FUA sequences */
+/* PREFLUSH/FUA sequences */
 enum {
 	REQ_FSEQ_PREFLUSH	= (1 << 0), /* pre-flushing in progress */
 	REQ_FSEQ_DATA		= (1 << 1), /* data write in progress */
@@ -148,7 +148,7 @@ static bool blk_flush_queue_rq(struct request *rq, bool add_front)
 
 /**
  * blk_flush_complete_seq - complete flush sequence
- * @rq: FLUSH/FUA request being sequenced
+ * @rq: PREFLUSH/FUA request being sequenced
  * @fq: flush queue
  * @seq: sequences to complete (mask of %REQ_FSEQ_*, can be zero)
  * @error: whether an error occurred
@@ -406,7 +406,7 @@ static void mq_flush_data_end_io(struct request *rq, blk_status_t error)
 }
 
 /**
- * blk_insert_flush - insert a new FLUSH/FUA request
+ * blk_insert_flush - insert a new PREFLUSH/FUA request
  * @rq: request to insert
  *
  * To be called from __elv_add_request() for %ELEVATOR_INSERT_FLUSH insertions.
@@ -525,7 +525,7 @@ int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask,
 		return -ENXIO;
 
 	bio = bio_alloc(gfp_mask, 0);
-	bio->bi_bdev = bdev;
+	bio_set_dev(bio, bdev);
 	bio->bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
 
 	ret = submit_bio_wait(bio);

block/blk-lib.c

@@ -77,7 +77,7 @@ int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
 
 		bio = next_bio(bio, 0, gfp_mask);
 		bio->bi_iter.bi_sector = sector;
-		bio->bi_bdev = bdev;
+		bio_set_dev(bio, bdev);
 		bio_set_op_attrs(bio, op, 0);
 
 		bio->bi_iter.bi_size = req_sects << 9;
@@ -168,7 +168,7 @@ static int __blkdev_issue_write_same(struct block_device *bdev, sector_t sector,
 	while (nr_sects) {
 		bio = next_bio(bio, 1, gfp_mask);
 		bio->bi_iter.bi_sector = sector;
-		bio->bi_bdev = bdev;
+		bio_set_dev(bio, bdev);
 		bio->bi_vcnt = 1;
 		bio->bi_io_vec->bv_page = page;
 		bio->bi_io_vec->bv_offset = 0;
@@ -241,7 +241,7 @@ static int __blkdev_issue_write_zeroes(struct block_device *bdev,
 	while (nr_sects) {
 		bio = next_bio(bio, 0, gfp_mask);
 		bio->bi_iter.bi_sector = sector;
-		bio->bi_bdev = bdev;
+		bio_set_dev(bio, bdev);
 		bio->bi_opf = REQ_OP_WRITE_ZEROES;
 		if (flags & BLKDEV_ZERO_NOUNMAP)
 			bio->bi_opf |= REQ_NOUNMAP;
@@ -323,7 +323,7 @@ int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
 		bio = next_bio(bio, __blkdev_sectors_to_bio_pages(nr_sects),
 			       gfp_mask);
 		bio->bi_iter.bi_sector = sector;
-		bio->bi_bdev   = bdev;
+		bio_set_dev(bio, bdev);
 		bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
 
 		while (nr_sects != 0) {

block/blk-merge.c

@@ -633,8 +633,8 @@ static void blk_account_io_merge(struct request *req)
 		cpu = part_stat_lock();
 		part = req->part;
 
-		part_round_stats(cpu, part);
-		part_dec_in_flight(part, rq_data_dir(req));
+		part_round_stats(req->q, cpu, part);
+		part_dec_in_flight(req->q, part, rq_data_dir(req));
 
 		hd_struct_put(part);
 		part_stat_unlock();
@@ -786,7 +786,7 @@ bool blk_rq_merge_ok(struct request *rq, struct bio *bio)
 		return false;
 
 	/* must be same device and not a special request */
-	if (rq->rq_disk != bio->bi_bdev->bd_disk || req_no_special_merge(rq))
+	if (rq->rq_disk != bio->bi_disk || req_no_special_merge(rq))
 		return false;
 
 	/* only merge integrity protected bio into ditto rq */

block/blk-mq-debugfs.c

@@ -48,8 +48,6 @@ static int blk_flags_show(struct seq_file *m, const unsigned long flags,
 static const char *const blk_queue_flag_name[] = {
 	QUEUE_FLAG_NAME(QUEUED),
 	QUEUE_FLAG_NAME(STOPPED),
-	QUEUE_FLAG_NAME(SYNCFULL),
-	QUEUE_FLAG_NAME(ASYNCFULL),
 	QUEUE_FLAG_NAME(DYING),
 	QUEUE_FLAG_NAME(BYPASS),
 	QUEUE_FLAG_NAME(BIDI),
@@ -744,7 +742,7 @@ static int blk_mq_debugfs_release(struct inode *inode, struct file *file)
 		return seq_release(inode, file);
 }
 
-const struct file_operations blk_mq_debugfs_fops = {
+static const struct file_operations blk_mq_debugfs_fops = {
 	.open		= blk_mq_debugfs_open,
 	.read		= seq_read,
 	.write		= blk_mq_debugfs_write,

block/blk-mq-tag.c

@@ -214,7 +214,11 @@ static bool bt_iter(struct sbitmap *bitmap, unsigned int bitnr, void *data)
 		bitnr += tags->nr_reserved_tags;
 	rq = tags->rqs[bitnr];
 
-	if (rq->q == hctx->queue)
+	/*
+	 * We can hit rq == NULL here, because the tagging functions
+	 * test and set the bit before assining ->rqs[].
+	 */
+	if (rq && rq->q == hctx->queue)
 		iter_data->fn(hctx, rq, iter_data->data, reserved);
 	return true;
 }
@@ -248,9 +252,15 @@ static bool bt_tags_iter(struct sbitmap *bitmap, unsigned int bitnr, void *data)
 
 	if (!reserved)
 		bitnr += tags->nr_reserved_tags;
+
+	/*
+	 * We can hit rq == NULL here, because the tagging functions
+	 * test and set the bit before assining ->rqs[].
+	 */
 	rq = tags->rqs[bitnr];
+	if (rq)
+		iter_data->fn(rq, iter_data->data, reserved);
 
-	iter_data->fn(rq, iter_data->data, reserved);
 	return true;
 }
 
@@ -288,11 +298,12 @@ void blk_mq_tagset_busy_iter(struct blk_mq_tag_set *tagset,
 }
 EXPORT_SYMBOL(blk_mq_tagset_busy_iter);
 
-int blk_mq_reinit_tagset(struct blk_mq_tag_set *set)
+int blk_mq_reinit_tagset(struct blk_mq_tag_set *set,
+			 int (reinit_request)(void *, struct request *))
 {
 	int i, j, ret = 0;
 
-	if (!set->ops->reinit_request)
+	if (WARN_ON_ONCE(!reinit_request))
 		goto out;
 
 	for (i = 0; i < set->nr_hw_queues; i++) {
@@ -305,8 +316,8 @@ int blk_mq_reinit_tagset(struct blk_mq_tag_set *set)
 			if (!tags->static_rqs[j])
 				continue;
 
-			ret = set->ops->reinit_request(set->driver_data,
-						tags->static_rqs[j]);
+			ret = reinit_request(set->driver_data,
+					     tags->static_rqs[j]);
 			if (ret)
 				goto out;
 		}

block/blk-mq.c

@@ -83,6 +83,41 @@ static void blk_mq_hctx_clear_pending(struct blk_mq_hw_ctx *hctx,
 	sbitmap_clear_bit(&hctx->ctx_map, ctx->index_hw);
 }
 
+struct mq_inflight {
+	struct hd_struct *part;
+	unsigned int *inflight;
+};
+
+static void blk_mq_check_inflight(struct blk_mq_hw_ctx *hctx,
+				  struct request *rq, void *priv,
+				  bool reserved)
+{
+	struct mq_inflight *mi = priv;
+
+	if (test_bit(REQ_ATOM_STARTED, &rq->atomic_flags) &&
+	    !test_bit(REQ_ATOM_COMPLETE, &rq->atomic_flags)) {
+		/*
+		 * index[0] counts the specific partition that was asked
+		 * for. index[1] counts the ones that are active on the
+		 * whole device, so increment that if mi->part is indeed
+		 * a partition, and not a whole device.
+		 */
+		if (rq->part == mi->part)
+			mi->inflight[0]++;
+		if (mi->part->partno)
+			mi->inflight[1]++;
+	}
+}
+
+void blk_mq_in_flight(struct request_queue *q, struct hd_struct *part,
+		      unsigned int inflight[2])
+{
+	struct mq_inflight mi = { .part = part, .inflight = inflight, };
+
+	inflight[0] = inflight[1] = 0;
+	blk_mq_queue_tag_busy_iter(q, blk_mq_check_inflight, &mi);
+}
+
 void blk_freeze_queue_start(struct request_queue *q)
 {
 	int freeze_depth;
@@ -624,11 +659,10 @@ static void blk_mq_requeue_work(struct work_struct *work)
 		container_of(work, struct request_queue, requeue_work.work);
 	LIST_HEAD(rq_list);
 	struct request *rq, *next;
-	unsigned long flags;
 
-	spin_lock_irqsave(&q->requeue_lock, flags);
+	spin_lock_irq(&q->requeue_lock);
 	list_splice_init(&q->requeue_list, &rq_list);
-	spin_unlock_irqrestore(&q->requeue_lock, flags);
+	spin_unlock_irq(&q->requeue_lock);
 
 	list_for_each_entry_safe(rq, next, &rq_list, queuelist) {
 		if (!(rq->rq_flags & RQF_SOFTBARRIER))
@@ -1102,9 +1136,19 @@ static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
 {
 	int srcu_idx;
 
+	/*
+	 * We should be running this queue from one of the CPUs that
+	 * are mapped to it.
+	 */
 	WARN_ON(!cpumask_test_cpu(raw_smp_processor_id(), hctx->cpumask) &&
 		cpu_online(hctx->next_cpu));
 
+	/*
+	 * We can't run the queue inline with ints disabled. Ensure that
+	 * we catch bad users of this early.
+	 */
+	WARN_ON_ONCE(in_interrupt());
+
 	if (!(hctx->flags & BLK_MQ_F_BLOCKING)) {
 		rcu_read_lock();
 		blk_mq_sched_dispatch_requests(hctx);
@@ -1218,7 +1262,7 @@ EXPORT_SYMBOL(blk_mq_queue_stopped);
 /*
  * This function is often used for pausing .queue_rq() by driver when
  * there isn't enough resource or some conditions aren't satisfied, and
- * BLK_MQ_RQ_QUEUE_BUSY is usually returned.
+ * BLK_STS_RESOURCE is usually returned.
  *
  * We do not guarantee that dispatch can be drained or blocked
  * after blk_mq_stop_hw_queue() returns. Please use
@@ -1235,7 +1279,7 @@ EXPORT_SYMBOL(blk_mq_stop_hw_queue);
 /*
  * This function is often used for pausing .queue_rq() by driver when
  * there isn't enough resource or some conditions aren't satisfied, and
- * BLK_MQ_RQ_QUEUE_BUSY is usually returned.
+ * BLK_STS_RESOURCE is usually returned.
  *
  * We do not guarantee that dispatch can be drained or blocked
  * after blk_mq_stop_hw_queues() returns. Please use

block/blk-mq.h

@@ -133,4 +133,7 @@ static inline bool blk_mq_hw_queue_mapped(struct blk_mq_hw_ctx *hctx)
 	return hctx->nr_ctx && hctx->tags;
 }
 
+void blk_mq_in_flight(struct request_queue *q, struct hd_struct *part,
+			unsigned int inflight[2]);
+
 #endif

block/blk-settings.c

@@ -68,6 +68,7 @@ EXPORT_SYMBOL_GPL(blk_queue_rq_timeout);
 
 void blk_queue_rq_timed_out(struct request_queue *q, rq_timed_out_fn *fn)
 {
+	WARN_ON_ONCE(q->mq_ops);
 	q->rq_timed_out_fn = fn;
 }
 EXPORT_SYMBOL_GPL(blk_queue_rq_timed_out);

block/blk-sysfs.c

@@ -931,7 +931,9 @@ void blk_unregister_queue(struct gendisk *disk)
 	if (WARN_ON(!q))
 		return;
 
+	mutex_lock(&q->sysfs_lock);
 	queue_flag_clear_unlocked(QUEUE_FLAG_REGISTERED, q);
+	mutex_unlock(&q->sysfs_lock);
 
 	wbt_exit(q);
 

block/blk-tag.c

@@ -290,7 +290,6 @@ void blk_queue_end_tag(struct request_queue *q, struct request *rq)
 	 */
 	clear_bit_unlock(tag, bqt->tag_map);
 }
-EXPORT_SYMBOL(blk_queue_end_tag);
 
 /**
  * blk_queue_start_tag - find a free tag and assign it

block/blk-throttle.c

@@ -373,10 +373,8 @@ static unsigned int tg_iops_limit(struct throtl_grp *tg, int rw)
 	if (likely(!blk_trace_note_message_enabled(__td->queue)))	\
 		break;							\
 	if ((__tg)) {							\
-		char __pbuf[128];					\
-									\
-		blkg_path(tg_to_blkg(__tg), __pbuf, sizeof(__pbuf));	\
-		blk_add_trace_msg(__td->queue, "throtl %s " fmt, __pbuf, ##args); \
+		blk_add_cgroup_trace_msg(__td->queue,			\
+			tg_to_blkg(__tg)->blkcg, "throtl " fmt, ##args);\
 	} else {							\
 		blk_add_trace_msg(__td->queue, "throtl " fmt, ##args);	\
 	}								\
@@ -2114,14 +2112,9 @@ static inline void throtl_update_latency_buckets(struct throtl_data *td)
 static void blk_throtl_assoc_bio(struct throtl_grp *tg, struct bio *bio)
 {
 #ifdef CONFIG_BLK_DEV_THROTTLING_LOW
-	int ret;
-
-	ret = bio_associate_current(bio);
-	if (ret == 0 || ret == -EBUSY)
+	if (bio->bi_css)
 		bio->bi_cg_private = tg;
 	blk_stat_set_issue(&bio->bi_issue_stat, bio_sectors(bio));
-#else
-	bio_associate_current(bio);
 #endif
 }
 

block/blk-zoned.c

@@ -116,7 +116,7 @@ int blkdev_report_zones(struct block_device *bdev,
 	if (!bio)
 		return -ENOMEM;
 
-	bio->bi_bdev = bdev;
+	bio_set_dev(bio, bdev);
 	bio->bi_iter.bi_sector = blk_zone_start(q, sector);
 	bio_set_op_attrs(bio, REQ_OP_ZONE_REPORT, 0);
 
@@ -234,7 +234,7 @@ int blkdev_reset_zones(struct block_device *bdev,
 
 		bio = bio_alloc(gfp_mask, 0);
 		bio->bi_iter.bi_sector = sector;
-		bio->bi_bdev = bdev;
+		bio_set_dev(bio, bdev);
 		bio_set_op_attrs(bio, REQ_OP_ZONE_RESET, 0);
 
 		ret = submit_bio_wait(bio);

block/blk.h

@@ -64,7 +64,6 @@ void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
 			struct bio *bio);
 void blk_queue_bypass_start(struct request_queue *q);
 void blk_queue_bypass_end(struct request_queue *q);
-void blk_dequeue_request(struct request *rq);
 void __blk_queue_free_tags(struct request_queue *q);
 void blk_freeze_queue(struct request_queue *q);
 
@@ -204,6 +203,8 @@ static inline void elv_deactivate_rq(struct request_queue *q, struct request *rq
 		e->type->ops.sq.elevator_deactivate_req_fn(q, rq);
 }
 
+struct hd_struct *__disk_get_part(struct gendisk *disk, int partno);
+
 #ifdef CONFIG_FAIL_IO_TIMEOUT
 int blk_should_fake_timeout(struct request_queue *);
 ssize_t part_timeout_show(struct device *, struct device_attribute *, char *);

block/bsg.c

@@ -932,15 +932,8 @@ static long bsg_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 
 		return ret;
 	}
-	/*
-	 * block device ioctls
-	 */
 	default:
-#if 0
-		return ioctl_by_bdev(bd->bdev, cmd, arg);
-#else
 		return -ENOTTY;
-#endif
 	}
 }
 

block/cfq-iosched.c

@@ -656,20 +656,17 @@ static inline void cfqg_put(struct cfq_group *cfqg)
 }
 
 #define cfq_log_cfqq(cfqd, cfqq, fmt, args...)	do {			\
-	char __pbuf[128];						\
-									\
-	blkg_path(cfqg_to_blkg((cfqq)->cfqg), __pbuf, sizeof(__pbuf));	\
-	blk_add_trace_msg((cfqd)->queue, "cfq%d%c%c %s " fmt, (cfqq)->pid, \
+	blk_add_cgroup_trace_msg((cfqd)->queue,				\
+			cfqg_to_blkg((cfqq)->cfqg)->blkcg,		\
+			"cfq%d%c%c " fmt, (cfqq)->pid,			\
 			cfq_cfqq_sync((cfqq)) ? 'S' : 'A',		\
 			cfqq_type((cfqq)) == SYNC_NOIDLE_WORKLOAD ? 'N' : ' ',\
-			  __pbuf, ##args);				\
+			  ##args);					\
 } while (0)
 
 #define cfq_log_cfqg(cfqd, cfqg, fmt, args...)	do {			\
-	char __pbuf[128];						\
-									\
-	blkg_path(cfqg_to_blkg(cfqg), __pbuf, sizeof(__pbuf));		\
-	blk_add_trace_msg((cfqd)->queue, "%s " fmt, __pbuf, ##args);	\
+	blk_add_cgroup_trace_msg((cfqd)->queue,				\
+			cfqg_to_blkg(cfqg)->blkcg, fmt, ##args);	\
 } while (0)
 
 static inline void cfqg_stats_update_io_add(struct cfq_group *cfqg,
@@ -2937,7 +2934,8 @@ static void cfq_arm_slice_timer(struct cfq_data *cfqd)
 	 * for devices that support queuing, otherwise we still have a problem
 	 * with sync vs async workloads.
 	 */
-	if (blk_queue_nonrot(cfqd->queue) && cfqd->hw_tag)
+	if (blk_queue_nonrot(cfqd->queue) && cfqd->hw_tag &&
+		!cfqd->cfq_group_idle)
 		return;
 
 	WARN_ON(!RB_EMPTY_ROOT(&cfqq->sort_list));
@@ -4714,13 +4712,12 @@ cfq_var_show(unsigned int var, char *page)
 	return sprintf(page, "%u\n", var);
 }
 
-static ssize_t
-cfq_var_store(unsigned int *var, const char *page, size_t count)
+static void
+cfq_var_store(unsigned int *var, const char *page)
 {
 	char *p = (char *) page;
 
 	*var = simple_strtoul(p, &p, 10);
-	return count;
 }
 
 #define SHOW_FUNCTION(__FUNC, __VAR, __CONV)				\
@@ -4766,7 +4763,7 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 {									\
 	struct cfq_data *cfqd = e->elevator_data;			\
 	unsigned int __data;						\
-	int ret = cfq_var_store(&__data, (page), count);		\
+	cfq_var_store(&__data, (page));					\
 	if (__data < (MIN))						\
 		__data = (MIN);						\
 	else if (__data > (MAX))					\
@@ -4775,7 +4772,7 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 		*(__PTR) = (u64)__data * NSEC_PER_MSEC;			\
 	else								\
 		*(__PTR) = __data;					\
-	return ret;							\
+	return count;							\
 }
 STORE_FUNCTION(cfq_quantum_store, &cfqd->cfq_quantum, 1, UINT_MAX, 0);
 STORE_FUNCTION(cfq_fifo_expire_sync_store, &cfqd->cfq_fifo_expire[1], 1,
@@ -4800,13 +4797,13 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 {									\
 	struct cfq_data *cfqd = e->elevator_data;			\
 	unsigned int __data;						\
-	int ret = cfq_var_store(&__data, (page), count);		\
+	cfq_var_store(&__data, (page));					\
 	if (__data < (MIN))						\
 		__data = (MIN);						\
 	else if (__data > (MAX))					\
 		__data = (MAX);						\
 	*(__PTR) = (u64)__data * NSEC_PER_USEC;				\
-	return ret;							\
+	return count;							\
 }
 USEC_STORE_FUNCTION(cfq_slice_idle_us_store, &cfqd->cfq_slice_idle, 0, UINT_MAX);
 USEC_STORE_FUNCTION(cfq_group_idle_us_store, &cfqd->cfq_group_idle, 0, UINT_MAX);

block/deadline-iosched.c

@@ -373,13 +373,12 @@ deadline_var_show(int var, char *page)
 	return sprintf(page, "%d\n", var);
 }
 
-static ssize_t
-deadline_var_store(int *var, const char *page, size_t count)
+static void
+deadline_var_store(int *var, const char *page)
 {
 	char *p = (char *) page;
 
 	*var = simple_strtol(p, &p, 10);
-	return count;
 }
 
 #define SHOW_FUNCTION(__FUNC, __VAR, __CONV)				\
@@ -403,7 +402,7 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 {									\
 	struct deadline_data *dd = e->elevator_data;			\
 	int __data;							\
-	int ret = deadline_var_store(&__data, (page), count);		\
+	deadline_var_store(&__data, (page));				\
 	if (__data < (MIN))						\
 		__data = (MIN);						\
 	else if (__data > (MAX))					\
@@ -412,7 +411,7 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 		*(__PTR) = msecs_to_jiffies(__data);			\
 	else								\
 		*(__PTR) = __data;					\
-	return ret;							\
+	return count;							\
 }
 STORE_FUNCTION(deadline_read_expire_store, &dd->fifo_expire[READ], 0, INT_MAX, 1);
 STORE_FUNCTION(deadline_write_expire_store, &dd->fifo_expire[WRITE], 0, INT_MAX, 1);

block/elevator.c

@@ -1055,6 +1055,10 @@ static int __elevator_change(struct request_queue *q, const char *name)
 	char elevator_name[ELV_NAME_MAX];
 	struct elevator_type *e;
 
+	/* Make sure queue is not in the middle of being removed */
+	if (!test_bit(QUEUE_FLAG_REGISTERED, &q->queue_flags))
+		return -ENOENT;
+
 	/*
 	 * Special case for mq, turn off scheduling
 	 */

block/genhd.c

@@ -45,6 +45,52 @@ static void disk_add_events(struct gendisk *disk);
 static void disk_del_events(struct gendisk *disk);
 static void disk_release_events(struct gendisk *disk);
 
+void part_inc_in_flight(struct request_queue *q, struct hd_struct *part, int rw)
+{
+	if (q->mq_ops)
+		return;
+
+	atomic_inc(&part->in_flight[rw]);
+	if (part->partno)
+		atomic_inc(&part_to_disk(part)->part0.in_flight[rw]);
+}
+
+void part_dec_in_flight(struct request_queue *q, struct hd_struct *part, int rw)
+{
+	if (q->mq_ops)
+		return;
+
+	atomic_dec(&part->in_flight[rw]);
+	if (part->partno)
+		atomic_dec(&part_to_disk(part)->part0.in_flight[rw]);
+}
+
+void part_in_flight(struct request_queue *q, struct hd_struct *part,
+		    unsigned int inflight[2])
+{
+	if (q->mq_ops) {
+		blk_mq_in_flight(q, part, inflight);
+		return;
+	}
+
+	inflight[0] = atomic_read(&part->in_flight[0]) +
+			atomic_read(&part->in_flight[1]);
+	if (part->partno) {
+		part = &part_to_disk(part)->part0;
+		inflight[1] = atomic_read(&part->in_flight[0]) +
+				atomic_read(&part->in_flight[1]);
+	}
+}
+
+struct hd_struct *__disk_get_part(struct gendisk *disk, int partno)
+{
+	struct disk_part_tbl *ptbl = rcu_dereference(disk->part_tbl);
+
+	if (unlikely(partno < 0 || partno >= ptbl->len))
+		return NULL;
+	return rcu_dereference(ptbl->part[partno]);
+}
+
 /**
  * disk_get_part - get partition
  * @disk: disk to look partition from
@@ -61,21 +107,12 @@ static void disk_release_events(struct gendisk *disk);
  */
 struct hd_struct *disk_get_part(struct gendisk *disk, int partno)
 {
-	struct hd_struct *part = NULL;
-	struct disk_part_tbl *ptbl;
-
-	if (unlikely(partno < 0))
-		return NULL;
+	struct hd_struct *part;
 
 	rcu_read_lock();
-
-	ptbl = rcu_dereference(disk->part_tbl);
-	if (likely(partno < ptbl->len)) {
-		part = rcu_dereference(ptbl->part[partno]);
-		if (part)
-			get_device(part_to_dev(part));
-	}
-
+	part = __disk_get_part(disk, partno);
+	if (part)
+		get_device(part_to_dev(part));
 	rcu_read_unlock();
 
 	return part;
@@ -1098,12 +1135,13 @@ static const struct attribute_group *disk_attr_groups[] = {
  * original ptbl is freed using RCU callback.
  *
  * LOCKING:
- * Matching bd_mutx locked.
+ * Matching bd_mutex locked or the caller is the only user of @disk.
  */
 static void disk_replace_part_tbl(struct gendisk *disk,
 				  struct disk_part_tbl *new_ptbl)
 {
-	struct disk_part_tbl *old_ptbl = disk->part_tbl;
+	struct disk_part_tbl *old_ptbl =
+		rcu_dereference_protected(disk->part_tbl, 1);
 
 	rcu_assign_pointer(disk->part_tbl, new_ptbl);
 
@@ -1122,14 +1160,16 @@ static void disk_replace_part_tbl(struct gendisk *disk,
  * uses RCU to allow unlocked dereferencing for stats and other stuff.
  *
  * LOCKING:
- * Matching bd_mutex locked, might sleep.
+ * Matching bd_mutex locked or the caller is the only user of @disk.
+ * Might sleep.
  *
  * RETURNS:
  * 0 on success, -errno on failure.
  */
 int disk_expand_part_tbl(struct gendisk *disk, int partno)
 {
-	struct disk_part_tbl *old_ptbl = disk->part_tbl;
+	struct disk_part_tbl *old_ptbl =
+		rcu_dereference_protected(disk->part_tbl, 1);
 	struct disk_part_tbl *new_ptbl;
 	int len = old_ptbl ? old_ptbl->len : 0;
 	int i, target;
@@ -1212,6 +1252,7 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 	struct disk_part_iter piter;
 	struct hd_struct *hd;
 	char buf[BDEVNAME_SIZE];
+	unsigned int inflight[2];
 	int cpu;
 
 	/*
@@ -1225,8 +1266,9 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 	disk_part_iter_init(&piter, gp, DISK_PITER_INCL_EMPTY_PART0);
 	while ((hd = disk_part_iter_next(&piter))) {
 		cpu = part_stat_lock();
-		part_round_stats(cpu, hd);
+		part_round_stats(gp->queue, cpu, hd);
 		part_stat_unlock();
+		part_in_flight(gp->queue, hd, inflight);
 		seq_printf(seqf, "%4d %7d %s %lu %lu %lu "
 			   "%u %lu %lu %lu %u %u %u %u\n",
 			   MAJOR(part_devt(hd)), MINOR(part_devt(hd)),
@@ -1239,7 +1281,7 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 			   part_stat_read(hd, merges[WRITE]),
 			   part_stat_read(hd, sectors[WRITE]),
 			   jiffies_to_msecs(part_stat_read(hd, ticks[WRITE])),
-			   part_in_flight(hd),
+			   inflight[0],
 			   jiffies_to_msecs(part_stat_read(hd, io_ticks)),
 			   jiffies_to_msecs(part_stat_read(hd, time_in_queue))
 			);
@@ -1321,6 +1363,14 @@ EXPORT_SYMBOL(alloc_disk);
 struct gendisk *alloc_disk_node(int minors, int node_id)
 {
 	struct gendisk *disk;
+	struct disk_part_tbl *ptbl;
+
+	if (minors > DISK_MAX_PARTS) {
+		printk(KERN_ERR
+			"block: can't allocated more than %d partitions\n",
+			DISK_MAX_PARTS);
+		minors = DISK_MAX_PARTS;
+	}
 
 	disk = kzalloc_node(sizeof(struct gendisk), GFP_KERNEL, node_id);
 	if (disk) {
@@ -1334,7 +1384,8 @@ struct gendisk *alloc_disk_node(int minors, int node_id)
 			kfree(disk);
 			return NULL;
 		}
-		disk->part_tbl->part[0] = &disk->part0;
+		ptbl = rcu_dereference_protected(disk->part_tbl, 1);
+		rcu_assign_pointer(ptbl->part[0], &disk->part0);
 
 		/*
 		 * set_capacity() and get_capacity() currently don't use

block/mq-deadline.c

@@ -457,13 +457,12 @@ deadline_var_show(int var, char *page)
 	return sprintf(page, "%d\n", var);
 }
 
-static ssize_t
-deadline_var_store(int *var, const char *page, size_t count)
+static void
+deadline_var_store(int *var, const char *page)
 {
 	char *p = (char *) page;
 
 	*var = simple_strtol(p, &p, 10);
-	return count;
 }
 
 #define SHOW_FUNCTION(__FUNC, __VAR, __CONV)				\
@@ -487,7 +486,7 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 {									\
 	struct deadline_data *dd = e->elevator_data;			\
 	int __data;							\
-	int ret = deadline_var_store(&__data, (page), count);		\
+	deadline_var_store(&__data, (page));				\
 	if (__data < (MIN))						\
 		__data = (MIN);						\
 	else if (__data > (MAX))					\
@@ -496,7 +495,7 @@ static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count)
 		*(__PTR) = msecs_to_jiffies(__data);			\
 	else								\
 		*(__PTR) = __data;					\
-	return ret;							\
+	return count;							\
 }
 STORE_FUNCTION(deadline_read_expire_store, &dd->fifo_expire[READ], 0, INT_MAX, 1);
 STORE_FUNCTION(deadline_write_expire_store, &dd->fifo_expire[WRITE], 0, INT_MAX, 1);
@@ -660,6 +659,7 @@ static struct elevator_type mq_deadline = {
 	.elevator_name = "mq-deadline",
 	.elevator_owner = THIS_MODULE,
 };
+MODULE_ALIAS("mq-deadline-iosched");
 
 static int __init deadline_init(void)
 {

block/partition-generic.c

@@ -112,11 +112,14 @@ ssize_t part_stat_show(struct device *dev,
 		       struct device_attribute *attr, char *buf)
 {
 	struct hd_struct *p = dev_to_part(dev);
+	struct request_queue *q = dev_to_disk(dev)->queue;
+	unsigned int inflight[2];
 	int cpu;
 
 	cpu = part_stat_lock();
-	part_round_stats(cpu, p);
+	part_round_stats(q, cpu, p);
 	part_stat_unlock();
+	part_in_flight(q, p, inflight);
 	return sprintf(buf,
 		"%8lu %8lu %8llu %8u "
 		"%8lu %8lu %8llu %8u "
@@ -130,7 +133,7 @@ ssize_t part_stat_show(struct device *dev,
 		part_stat_read(p, merges[WRITE]),
 		(unsigned long long)part_stat_read(p, sectors[WRITE]),
 		jiffies_to_msecs(part_stat_read(p, ticks[WRITE])),
-		part_in_flight(p),
+		inflight[0],
 		jiffies_to_msecs(part_stat_read(p, io_ticks)),
 		jiffies_to_msecs(part_stat_read(p, time_in_queue)));
 }
@@ -249,15 +252,20 @@ void __delete_partition(struct percpu_ref *ref)
 	call_rcu(&part->rcu_head, delete_partition_rcu_cb);
 }
 
+/*
+ * Must be called either with bd_mutex held, before a disk can be opened or
+ * after all disk users are gone.
+ */
 void delete_partition(struct gendisk *disk, int partno)
 {
-	struct disk_part_tbl *ptbl = disk->part_tbl;
+	struct disk_part_tbl *ptbl =
+		rcu_dereference_protected(disk->part_tbl, 1);
 	struct hd_struct *part;
 
 	if (partno >= ptbl->len)
 		return;
 
-	part = ptbl->part[partno];
+	part = rcu_dereference_protected(ptbl->part[partno], 1);
 	if (!part)
 		return;
 
@@ -277,6 +285,10 @@ static ssize_t whole_disk_show(struct device *dev,
 static DEVICE_ATTR(whole_disk, S_IRUSR | S_IRGRP | S_IROTH,
 		   whole_disk_show, NULL);
 
+/*
+ * Must be called either with bd_mutex held, before a disk can be opened or
+ * after all disk users are gone.
+ */
 struct hd_struct *add_partition(struct gendisk *disk, int partno,
 				sector_t start, sector_t len, int flags,
 				struct partition_meta_info *info)
@@ -292,7 +304,7 @@ struct hd_struct *add_partition(struct gendisk *disk, int partno,
 	err = disk_expand_part_tbl(disk, partno);
 	if (err)
 		return ERR_PTR(err);
-	ptbl = disk->part_tbl;
+	ptbl = rcu_dereference_protected(disk->part_tbl, 1);
 
 	if (ptbl->part[partno])
 		return ERR_PTR(-EBUSY);
@@ -391,7 +403,6 @@ out_del:
 	device_del(pdev);
 out_put:
 	put_device(pdev);
-	blk_free_devt(devt);
 	return ERR_PTR(err);
 }
 

drivers/block/DAC960.c

@@ -1678,9 +1678,12 @@ static bool DAC960_V1_ReadControllerConfiguration(DAC960_Controller_T
       Enquiry2->FirmwareID.FirmwareType = '0';
       Enquiry2->FirmwareID.TurnID = 0;
     }
-  sprintf(Controller->FirmwareVersion, "%d.%02d-%c-%02d",
-	  Enquiry2->FirmwareID.MajorVersion, Enquiry2->FirmwareID.MinorVersion,
-	  Enquiry2->FirmwareID.FirmwareType, Enquiry2->FirmwareID.TurnID);
+  snprintf(Controller->FirmwareVersion, sizeof(Controller->FirmwareVersion),
+	   "%d.%02d-%c-%02d",
+	   Enquiry2->FirmwareID.MajorVersion,
+	   Enquiry2->FirmwareID.MinorVersion,
+	   Enquiry2->FirmwareID.FirmwareType,
+	   Enquiry2->FirmwareID.TurnID);
   if (!((Controller->FirmwareVersion[0] == '5' &&
 	 strcmp(Controller->FirmwareVersion, "5.06") >= 0) ||
 	(Controller->FirmwareVersion[0] == '4' &&
@@ -6588,7 +6591,8 @@ static void DAC960_CreateProcEntries(DAC960_Controller_T *Controller)
 			    &dac960_proc_fops);
 	}
 
-	sprintf(Controller->ControllerName, "c%d", Controller->ControllerNumber);
+	snprintf(Controller->ControllerName, sizeof(Controller->ControllerName),
+		 "c%d", Controller->ControllerNumber);
 	ControllerProcEntry = proc_mkdir(Controller->ControllerName,
 					 DAC960_ProcDirectoryEntry);
 	proc_create_data("initial_status", 0, ControllerProcEntry, &dac960_initial_status_proc_fops, Controller);

drivers/block/Kconfig

@@ -17,6 +17,7 @@ if BLK_DEV
 
 config BLK_DEV_NULL_BLK
 	tristate "Null test block driver"
+	depends on CONFIGFS_FS
 
 config BLK_DEV_FD
 	tristate "Normal floppy disk support"

drivers/block/brd.c

@@ -294,14 +294,13 @@ out:
 
 static blk_qc_t brd_make_request(struct request_queue *q, struct bio *bio)
 {
-	struct block_device *bdev = bio->bi_bdev;
-	struct brd_device *brd = bdev->bd_disk->private_data;
+	struct brd_device *brd = bio->bi_disk->private_data;
 	struct bio_vec bvec;
 	sector_t sector;
 	struct bvec_iter iter;
 
 	sector = bio->bi_iter.bi_sector;
-	if (bio_end_sector(bio) > get_capacity(bdev->bd_disk))
+	if (bio_end_sector(bio) > get_capacity(bio->bi_disk))
 		goto io_error;
 
 	bio_for_each_segment(bvec, bio, iter) {

drivers/block/drbd/drbd_actlog.c

@@ -151,7 +151,7 @@ static int _drbd_md_sync_page_io(struct drbd_device *device,
 	op_flags |= REQ_SYNC;
 
 	bio = bio_alloc_drbd(GFP_NOIO);
-	bio->bi_bdev = bdev->md_bdev;
+	bio_set_dev(bio, bdev->md_bdev);
 	bio->bi_iter.bi_sector = sector;
 	err = -EIO;
 	if (bio_add_page(bio, device->md_io.page, size, 0) != size)

drivers/block/drbd/drbd_bitmap.c

@@ -1019,7 +1019,7 @@ static void bm_page_io_async(struct drbd_bm_aio_ctx *ctx, int page_nr) __must_ho
 		bm_store_page_idx(page, page_nr);
 	} else
 		page = b->bm_pages[page_nr];
-	bio->bi_bdev = device->ldev->md_bdev;
+	bio_set_dev(bio, device->ldev->md_bdev);
 	bio->bi_iter.bi_sector = on_disk_sector;
 	/* bio_add_page of a single page to an empty bio will always succeed,
 	 * according to api.  Do we want to assert that? */

drivers/block/drbd/drbd_int.h

@@ -63,19 +63,15 @@
 # define __must_hold(x)
 #endif
 
-/* module parameter, defined in drbd_main.c */
-extern unsigned int minor_count;
-extern bool disable_sendpage;
-extern bool allow_oos;
-void tl_abort_disk_io(struct drbd_device *device);
-
+/* shared module parameters, defined in drbd_main.c */
 #ifdef CONFIG_DRBD_FAULT_INJECTION
-extern int enable_faults;
-extern int fault_rate;
-extern int fault_devs;
+extern int drbd_enable_faults;
+extern int drbd_fault_rate;
 #endif
 
-extern char usermode_helper[];
+extern unsigned int drbd_minor_count;
+extern char drbd_usermode_helper[];
+extern int drbd_proc_details;
 
 
 /* This is used to stop/restart our threads.
@@ -181,8 +177,8 @@ _drbd_insert_fault(struct drbd_device *device, unsigned int type);
 static inline int
 drbd_insert_fault(struct drbd_device *device, unsigned int type) {
 #ifdef CONFIG_DRBD_FAULT_INJECTION
-	return fault_rate &&
-		(enable_faults & (1<<type)) &&
+	return drbd_fault_rate &&
+		(drbd_enable_faults & (1<<type)) &&
 		_drbd_insert_fault(device, type);
 #else
 	return 0;
@@ -745,6 +741,8 @@ struct drbd_connection {
 	unsigned current_tle_writes;	/* writes seen within this tl epoch */
 
 	unsigned long last_reconnect_jif;
+	/* empty member on older kernels without blk_start_plug() */
+	struct blk_plug receiver_plug;
 	struct drbd_thread receiver;
 	struct drbd_thread worker;
 	struct drbd_thread ack_receiver;
@@ -1131,7 +1129,8 @@ extern void conn_send_sr_reply(struct drbd_connection *connection, enum drbd_sta
 extern int drbd_send_rs_deallocated(struct drbd_peer_device *, struct drbd_peer_request *);
 extern void drbd_backing_dev_free(struct drbd_device *device, struct drbd_backing_dev *ldev);
 extern void drbd_device_cleanup(struct drbd_device *device);
-void drbd_print_uuids(struct drbd_device *device, const char *text);
+extern void drbd_print_uuids(struct drbd_device *device, const char *text);
+extern void drbd_queue_unplug(struct drbd_device *device);
 
 extern void conn_md_sync(struct drbd_connection *connection);
 extern void drbd_md_write(struct drbd_device *device, void *buffer);
@@ -1463,8 +1462,6 @@ extern struct drbd_resource *drbd_find_resource(const char *name);
 extern void drbd_destroy_resource(struct kref *kref);
 extern void conn_free_crypto(struct drbd_connection *connection);
 
-extern int proc_details;
-
 /* drbd_req */
 extern void do_submit(struct work_struct *ws);
 extern void __drbd_make_request(struct drbd_device *, struct bio *, unsigned long);
@@ -1628,8 +1625,8 @@ static inline void drbd_generic_make_request(struct drbd_device *device,
 					     int fault_type, struct bio *bio)
 {
 	__release(local);
-	if (!bio->bi_bdev) {
-		drbd_err(device, "drbd_generic_make_request: bio->bi_bdev == NULL\n");
+	if (!bio->bi_disk) {
+		drbd_err(device, "drbd_generic_make_request: bio->bi_disk == NULL\n");
 		bio->bi_status = BLK_STS_IOERR;
 		bio_endio(bio);
 		return;

drivers/block/drbd/drbd_main.c

@@ -77,41 +77,41 @@ MODULE_PARM_DESC(minor_count, "Approximate number of drbd devices ("
 MODULE_ALIAS_BLOCKDEV_MAJOR(DRBD_MAJOR);
 
 #include <linux/moduleparam.h>
-/* allow_open_on_secondary */
-MODULE_PARM_DESC(allow_oos, "DONT USE!");
 /* thanks to these macros, if compiled into the kernel (not-module),
- * this becomes the boot parameter drbd.minor_count */
-module_param(minor_count, uint, 0444);
-module_param(disable_sendpage, bool, 0644);
-module_param(allow_oos, bool, 0);
-module_param(proc_details, int, 0644);
+ * these become boot parameters (e.g., drbd.minor_count) */
 
 #ifdef CONFIG_DRBD_FAULT_INJECTION
-int enable_faults;
-int fault_rate;
-static int fault_count;
-int fault_devs;
+int drbd_enable_faults;
+int drbd_fault_rate;
+static int drbd_fault_count;
+static int drbd_fault_devs;
 /* bitmap of enabled faults */
-module_param(enable_faults, int, 0664);
+module_param_named(enable_faults, drbd_enable_faults, int, 0664);
 /* fault rate % value - applies to all enabled faults */
-module_param(fault_rate, int, 0664);
+module_param_named(fault_rate, drbd_fault_rate, int, 0664);
 /* count of faults inserted */
-module_param(fault_count, int, 0664);
+module_param_named(fault_count, drbd_fault_count, int, 0664);
 /* bitmap of devices to insert faults on */
-module_param(fault_devs, int, 0644);
+module_param_named(fault_devs, drbd_fault_devs, int, 0644);
 #endif
 
-/* module parameter, defined */
-unsigned int minor_count = DRBD_MINOR_COUNT_DEF;
-bool disable_sendpage;
-bool allow_oos;
-int proc_details;       /* Detail level in proc drbd*/
-
+/* module parameters we can keep static */
+static bool drbd_allow_oos; /* allow_open_on_secondary */
+static bool drbd_disable_sendpage;
+MODULE_PARM_DESC(allow_oos, "DONT USE!");
+module_param_named(allow_oos, drbd_allow_oos, bool, 0);
+module_param_named(disable_sendpage, drbd_disable_sendpage, bool, 0644);
+
+/* module parameters we share */
+int drbd_proc_details; /* Detail level in proc drbd*/
+module_param_named(proc_details, drbd_proc_details, int, 0644);
+/* module parameters shared with defaults */
+unsigned int drbd_minor_count = DRBD_MINOR_COUNT_DEF;
 /* Module parameter for setting the user mode helper program
  * to run. Default is /sbin/drbdadm */
-char usermode_helper[80] = "/sbin/drbdadm";
-
-module_param_string(usermode_helper, usermode_helper, sizeof(usermode_helper), 0644);
+char drbd_usermode_helper[80] = "/sbin/drbdadm";
+module_param_named(minor_count, drbd_minor_count, uint, 0444);
+module_param_string(usermode_helper, drbd_usermode_helper, sizeof(drbd_usermode_helper), 0644);
 
 /* in 2.6.x, our device mapping and config info contains our virtual gendisks
  * as member "struct gendisk *vdisk;"
@@ -923,7 +923,9 @@ void drbd_gen_and_send_sync_uuid(struct drbd_peer_device *peer_device)
 }
 
 /* communicated if (agreed_features & DRBD_FF_WSAME) */
-void assign_p_sizes_qlim(struct drbd_device *device, struct p_sizes *p, struct request_queue *q)
+static void
+assign_p_sizes_qlim(struct drbd_device *device, struct p_sizes *p,
+					struct request_queue *q)
 {
 	if (q) {
 		p->qlim->physical_block_size = cpu_to_be32(queue_physical_block_size(q));
@@ -1560,7 +1562,7 @@ static int _drbd_send_page(struct drbd_peer_device *peer_device, struct page *pa
 	 * put_page(); and would cause either a VM_BUG directly, or
 	 * __page_cache_release a page that would actually still be referenced
 	 * by someone, leading to some obscure delayed Oops somewhere else. */
-	if (disable_sendpage || (page_count(page) < 1) || PageSlab(page))
+	if (drbd_disable_sendpage || (page_count(page) < 1) || PageSlab(page))
 		return _drbd_no_send_page(peer_device, page, offset, size, msg_flags);
 
 	msg_flags |= MSG_NOSIGNAL;
@@ -1932,7 +1934,7 @@ static int drbd_open(struct block_device *bdev, fmode_t mode)
 	if (device->state.role != R_PRIMARY) {
 		if (mode & FMODE_WRITE)
 			rv = -EROFS;
-		else if (!allow_oos)
+		else if (!drbd_allow_oos)
 			rv = -EMEDIUMTYPE;
 	}
 
@@ -1952,6 +1954,19 @@ static void drbd_release(struct gendisk *gd, fmode_t mode)
 	mutex_unlock(&drbd_main_mutex);
 }
 
+/* need to hold resource->req_lock */
+void drbd_queue_unplug(struct drbd_device *device)
+{
+	if (device->state.pdsk >= D_INCONSISTENT && device->state.conn >= C_CONNECTED) {
+		D_ASSERT(device, device->state.role == R_PRIMARY);
+		if (test_and_clear_bit(UNPLUG_REMOTE, &device->flags)) {
+			drbd_queue_work_if_unqueued(
+				&first_peer_device(device)->connection->sender_work,
+				&device->unplug_work);
+		}
+	}
+}
+
 static void drbd_set_defaults(struct drbd_device *device)
 {
 	/* Beware! The actual layout differs
@@ -2008,18 +2023,14 @@ void drbd_init_set_defaults(struct drbd_device *device)
 	device->unplug_work.cb  = w_send_write_hint;
 	device->bm_io_work.w.cb = w_bitmap_io;
 
-	init_timer(&device->resync_timer);
-	init_timer(&device->md_sync_timer);
-	init_timer(&device->start_resync_timer);
-	init_timer(&device->request_timer);
-	device->resync_timer.function = resync_timer_fn;
-	device->resync_timer.data = (unsigned long) device;
-	device->md_sync_timer.function = md_sync_timer_fn;
-	device->md_sync_timer.data = (unsigned long) device;
-	device->start_resync_timer.function = start_resync_timer_fn;
-	device->start_resync_timer.data = (unsigned long) device;
-	device->request_timer.function = request_timer_fn;
-	device->request_timer.data = (unsigned long) device;
+	setup_timer(&device->resync_timer, resync_timer_fn,
+			(unsigned long)device);
+	setup_timer(&device->md_sync_timer, md_sync_timer_fn,
+			(unsigned long)device);
+	setup_timer(&device->start_resync_timer, start_resync_timer_fn,
+			(unsigned long)device);
+	setup_timer(&device->request_timer, request_timer_fn,
+			(unsigned long)device);
 
 	init_waitqueue_head(&device->misc_wait);
 	init_waitqueue_head(&device->state_wait);
@@ -2131,7 +2142,7 @@ static void drbd_destroy_mempools(void)
 static int drbd_create_mempools(void)
 {
 	struct page *page;
-	const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count;
+	const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * drbd_minor_count;
 	int i;
 
 	/* prepare our caches and mempools */
@@ -2167,13 +2178,12 @@ static int drbd_create_mempools(void)
 		goto Enomem;
 
 	/* mempools */
-	drbd_io_bio_set = bioset_create(BIO_POOL_SIZE, 0, BIOSET_NEED_RESCUER);
+	drbd_io_bio_set = bioset_create(BIO_POOL_SIZE, 0, 0);
 	if (drbd_io_bio_set == NULL)
 		goto Enomem;
 
 	drbd_md_io_bio_set = bioset_create(DRBD_MIN_POOL_PAGES, 0,
-					   BIOSET_NEED_BVECS |
-					   BIOSET_NEED_RESCUER);
+					   BIOSET_NEED_BVECS);
 	if (drbd_md_io_bio_set == NULL)
 		goto Enomem;
 
@@ -2409,7 +2419,6 @@ static void drbd_cleanup(void)
 		destroy_workqueue(retry.wq);
 
 	drbd_genl_unregister();
-	drbd_debugfs_cleanup();
 
 	idr_for_each_entry(&drbd_devices, device, i)
 		drbd_delete_device(device);
@@ -2420,6 +2429,8 @@ static void drbd_cleanup(void)
 		drbd_free_resource(resource);
 	}
 
+	drbd_debugfs_cleanup();
+
 	drbd_destroy_mempools();
 	unregister_blkdev(DRBD_MAJOR, "drbd");
 
@@ -2972,12 +2983,12 @@ static int __init drbd_init(void)
 {
 	int err;
 
-	if (minor_count < DRBD_MINOR_COUNT_MIN || minor_count > DRBD_MINOR_COUNT_MAX) {
-		pr_err("invalid minor_count (%d)\n", minor_count);
+	if (drbd_minor_count < DRBD_MINOR_COUNT_MIN || drbd_minor_count > DRBD_MINOR_COUNT_MAX) {
+		pr_err("invalid minor_count (%d)\n", drbd_minor_count);
 #ifdef MODULE
 		return -EINVAL;
 #else
-		minor_count = DRBD_MINOR_COUNT_DEF;
+		drbd_minor_count = DRBD_MINOR_COUNT_DEF;
 #endif
 	}
 
@@ -3900,12 +3911,12 @@ _drbd_insert_fault(struct drbd_device *device, unsigned int type)
 	static struct fault_random_state rrs = {0, 0};
 
 	unsigned int ret = (
-		(fault_devs == 0 ||
-			((1 << device_to_minor(device)) & fault_devs) != 0) &&
-		(((_drbd_fault_random(&rrs) % 100) + 1) <= fault_rate));
+		(drbd_fault_devs == 0 ||
+			((1 << device_to_minor(device)) & drbd_fault_devs) != 0) &&
+		(((_drbd_fault_random(&rrs) % 100) + 1) <= drbd_fault_rate));
 
 	if (ret) {
-		fault_count++;
+		drbd_fault_count++;
 
 		if (__ratelimit(&drbd_ratelimit_state))
 			drbd_warn(device, "***Simulating %s failure\n",

drivers/block/drbd/drbd_nl.c

@@ -344,7 +344,7 @@ int drbd_khelper(struct drbd_device *device, char *cmd)
 			 (char[60]) { }, /* address */
 			NULL };
 	char mb[14];
-	char *argv[] = {usermode_helper, cmd, mb, NULL };
+	char *argv[] = {drbd_usermode_helper, cmd, mb, NULL };
 	struct drbd_connection *connection = first_peer_device(device)->connection;
 	struct sib_info sib;
 	int ret;
@@ -359,19 +359,19 @@ int drbd_khelper(struct drbd_device *device, char *cmd)
 	 * write out any unsynced meta data changes now */
 	drbd_md_sync(device);
 
-	drbd_info(device, "helper command: %s %s %s\n", usermode_helper, cmd, mb);
+	drbd_info(device, "helper command: %s %s %s\n", drbd_usermode_helper, cmd, mb);
 	sib.sib_reason = SIB_HELPER_PRE;
 	sib.helper_name = cmd;
 	drbd_bcast_event(device, &sib);
 	notify_helper(NOTIFY_CALL, device, connection, cmd, 0);
-	ret = call_usermodehelper(usermode_helper, argv, envp, UMH_WAIT_PROC);
+	ret = call_usermodehelper(drbd_usermode_helper, argv, envp, UMH_WAIT_PROC);
 	if (ret)
 		drbd_warn(device, "helper command: %s %s %s exit code %u (0x%x)\n",
-				usermode_helper, cmd, mb,
+				drbd_usermode_helper, cmd, mb,
 				(ret >> 8) & 0xff, ret);
 	else
 		drbd_info(device, "helper command: %s %s %s exit code %u (0x%x)\n",
-				usermode_helper, cmd, mb,
+				drbd_usermode_helper, cmd, mb,
 				(ret >> 8) & 0xff, ret);
 	sib.sib_reason = SIB_HELPER_POST;
 	sib.helper_exit_code = ret;
@@ -396,24 +396,24 @@ enum drbd_peer_state conn_khelper(struct drbd_connection *connection, char *cmd)
 			 (char[60]) { }, /* address */
 			NULL };
 	char *resource_name = connection->resource->name;
-	char *argv[] = {usermode_helper, cmd, resource_name, NULL };
+	char *argv[] = {drbd_usermode_helper, cmd, resource_name, NULL };
 	int ret;
 
 	setup_khelper_env(connection, envp);
 	conn_md_sync(connection);
 
-	drbd_info(connection, "helper command: %s %s %s\n", usermode_helper, cmd, resource_name);
+	drbd_info(connection, "helper command: %s %s %s\n", drbd_usermode_helper, cmd, resource_name);
 	/* TODO: conn_bcast_event() ?? */
 	notify_helper(NOTIFY_CALL, NULL, connection, cmd, 0);
 
-	ret = call_usermodehelper(usermode_helper, argv, envp, UMH_WAIT_PROC);
+	ret = call_usermodehelper(drbd_usermode_helper, argv, envp, UMH_WAIT_PROC);
 	if (ret)
 		drbd_warn(connection, "helper command: %s %s %s exit code %u (0x%x)\n",
-			  usermode_helper, cmd, resource_name,
+			  drbd_usermode_helper, cmd, resource_name,
 			  (ret >> 8) & 0xff, ret);
 	else
 		drbd_info(connection, "helper command: %s %s %s exit code %u (0x%x)\n",
-			  usermode_helper, cmd, resource_name,
+			  drbd_usermode_helper, cmd, resource_name,
 			  (ret >> 8) & 0xff, ret);
 	/* TODO: conn_bcast_event() ?? */
 	notify_helper(NOTIFY_RESPONSE, NULL, connection, cmd, ret);
@@ -1236,12 +1236,18 @@ static void fixup_discard_if_not_supported(struct request_queue *q)
 
 static void decide_on_write_same_support(struct drbd_device *device,
 			struct request_queue *q,
-			struct request_queue *b, struct o_qlim *o)
+			struct request_queue *b, struct o_qlim *o,
+			bool disable_write_same)
 {
 	struct drbd_peer_device *peer_device = first_peer_device(device);
 	struct drbd_connection *connection = peer_device->connection;
 	bool can_do = b ? b->limits.max_write_same_sectors : true;
 
+	if (can_do && disable_write_same) {
+		can_do = false;
+		drbd_info(peer_device, "WRITE_SAME disabled by config\n");
+	}
+
 	if (can_do && connection->cstate >= C_CONNECTED && !(connection->agreed_features & DRBD_FF_WSAME)) {
 		can_do = false;
 		drbd_info(peer_device, "peer does not support WRITE_SAME\n");
@@ -1302,6 +1308,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi
 	struct request_queue *b = NULL;
 	struct disk_conf *dc;
 	bool discard_zeroes_if_aligned = true;
+	bool disable_write_same = false;
 
 	if (bdev) {
 		b = bdev->backing_bdev->bd_disk->queue;
@@ -1311,6 +1318,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi
 		dc = rcu_dereference(device->ldev->disk_conf);
 		max_segments = dc->max_bio_bvecs;
 		discard_zeroes_if_aligned = dc->discard_zeroes_if_aligned;
+		disable_write_same = dc->disable_write_same;
 		rcu_read_unlock();
 
 		blk_set_stacking_limits(&q->limits);
@@ -1321,7 +1329,7 @@ static void drbd_setup_queue_param(struct drbd_device *device, struct drbd_backi
 	blk_queue_max_segments(q, max_segments ? max_segments : BLK_MAX_SEGMENTS);
 	blk_queue_segment_boundary(q, PAGE_SIZE-1);
 	decide_on_discard_support(device, q, b, discard_zeroes_if_aligned);
-	decide_on_write_same_support(device, q, b, o);
+	decide_on_write_same_support(device, q, b, o, disable_write_same);
 
 	if (b) {
 		blk_queue_stack_limits(q, b);
@@ -1612,7 +1620,8 @@ int drbd_adm_disk_opts(struct sk_buff *skb, struct genl_info *info)
 	if (write_ordering_changed(old_disk_conf, new_disk_conf))
 		drbd_bump_write_ordering(device->resource, NULL, WO_BDEV_FLUSH);
 
-	if (old_disk_conf->discard_zeroes_if_aligned != new_disk_conf->discard_zeroes_if_aligned)
+	if (old_disk_conf->discard_zeroes_if_aligned != new_disk_conf->discard_zeroes_if_aligned
+	||  old_disk_conf->disable_write_same != new_disk_conf->disable_write_same)
 		drbd_reconsider_queue_parameters(device, device->ldev, NULL);
 
 	drbd_md_sync(device);
@@ -2140,34 +2149,13 @@ int drbd_adm_attach(struct sk_buff *skb, struct genl_info *info)
 
 static int adm_detach(struct drbd_device *device, int force)
 {
-	enum drbd_state_rv retcode;
-	void *buffer;
-	int ret;
-
 	if (force) {
 		set_bit(FORCE_DETACH, &device->flags);
 		drbd_force_state(device, NS(disk, D_FAILED));
-		retcode = SS_SUCCESS;
-		goto out;
+		return SS_SUCCESS;
 	}
 
-	drbd_suspend_io(device); /* so no-one is stuck in drbd_al_begin_io */
-	buffer = drbd_md_get_buffer(device, __func__); /* make sure there is no in-flight meta-data IO */
-	if (buffer) {
-		retcode = drbd_request_state(device, NS(disk, D_FAILED));
-		drbd_md_put_buffer(device);
-	} else /* already <= D_FAILED */
-		retcode = SS_NOTHING_TO_DO;
-	/* D_FAILED will transition to DISKLESS. */
-	drbd_resume_io(device);
-	ret = wait_event_interruptible(device->misc_wait,
-			device->state.disk != D_FAILED);
-	if ((int)retcode == (int)SS_IS_DISKLESS)
-		retcode = SS_NOTHING_TO_DO;
-	if (ret)
-		retcode = ERR_INTR;
-out:
-	return retcode;
+	return drbd_request_detach_interruptible(device);
 }
 
 /* Detaching the disk is a process in multiple stages.  First we need to lock

drivers/block/drbd/drbd_proc.c

@@ -127,7 +127,7 @@ static void drbd_syncer_progress(struct drbd_device *device, struct seq_file *se
 		seq_putc(seq, '=');
 	seq_putc(seq, '>');
 	for (i = 0; i < y; i++)
-		seq_printf(seq, ".");
+		seq_putc(seq, '.');
 	seq_puts(seq, "] ");
 
 	if (state.conn == C_VERIFY_S || state.conn == C_VERIFY_T)
@@ -179,7 +179,7 @@ static void drbd_syncer_progress(struct drbd_device *device, struct seq_file *se
 	seq_printf_with_thousands_grouping(seq, dbdt);
 	seq_puts(seq, " (");
 	/* ------------------------- ~3s average ------------------------ */
-	if (proc_details >= 1) {
+	if (drbd_proc_details >= 1) {
 		/* this is what drbd_rs_should_slow_down() uses */
 		i = (device->rs_last_mark + DRBD_SYNC_MARKS-1) % DRBD_SYNC_MARKS;
 		dt = (jiffies - device->rs_mark_time[i]) / HZ;
@@ -209,7 +209,7 @@ static void drbd_syncer_progress(struct drbd_device *device, struct seq_file *se
 	}
 	seq_printf(seq, " K/sec%s\n", stalled ? " (stalled)" : "");
 
-	if (proc_details >= 1) {
+	if (drbd_proc_details >= 1) {
 		/* 64 bit:
 		 * we convert to sectors in the display below. */
 		unsigned long bm_bits = drbd_bm_bits(device);
@@ -332,13 +332,13 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
 		    state.conn == C_VERIFY_T)
 			drbd_syncer_progress(device, seq, state);
 
-		if (proc_details >= 1 && get_ldev_if_state(device, D_FAILED)) {
+		if (drbd_proc_details >= 1 && get_ldev_if_state(device, D_FAILED)) {
 			lc_seq_printf_stats(seq, device->resync);
 			lc_seq_printf_stats(seq, device->act_log);
 			put_ldev(device);
 		}
 
-		if (proc_details >= 2)
+		if (drbd_proc_details >= 2)
 			seq_printf(seq, "\tblocked on activity log: %d\n", atomic_read(&device->ap_actlog_cnt));
 	}
 	rcu_read_unlock();

drivers/block/drbd/drbd_receiver.c

@@ -332,7 +332,7 @@ static void drbd_free_pages(struct drbd_device *device, struct page *page, int i
 	if (page == NULL)
 		return;
 
-	if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count)
+	if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * drbd_minor_count)
 		i = page_chain_free(page);
 	else {
 		struct page *tmp;
@@ -1100,7 +1100,10 @@ randomize:
 	idr_for_each_entry(&connection->peer_devices, peer_device, vnr)
 		mutex_lock(peer_device->device->state_mutex);
 
+	/* avoid a race with conn_request_state( C_DISCONNECTING ) */
+	spin_lock_irq(&connection->resource->req_lock);
 	set_bit(STATE_SENT, &connection->flags);
+	spin_unlock_irq(&connection->resource->req_lock);
 
 	idr_for_each_entry(&connection->peer_devices, peer_device, vnr)
 		mutex_unlock(peer_device->device->state_mutex);
@@ -1194,6 +1197,14 @@ static int decode_header(struct drbd_connection *connection, void *header, struc
 	return 0;
 }
 
+static void drbd_unplug_all_devices(struct drbd_connection *connection)
+{
+	if (current->plug == &connection->receiver_plug) {
+		blk_finish_plug(&connection->receiver_plug);
+		blk_start_plug(&connection->receiver_plug);
+	} /* else: maybe just schedule() ?? */
+}
+
 static int drbd_recv_header(struct drbd_connection *connection, struct packet_info *pi)
 {
 	void *buffer = connection->data.rbuf;
@@ -1209,6 +1220,36 @@ static int drbd_recv_header(struct drbd_connection *connection, struct packet_in
 	return err;
 }
 
+static int drbd_recv_header_maybe_unplug(struct drbd_connection *connection, struct packet_info *pi)
+{
+	void *buffer = connection->data.rbuf;
+	unsigned int size = drbd_header_size(connection);
+	int err;
+
+	err = drbd_recv_short(connection->data.socket, buffer, size, MSG_NOSIGNAL|MSG_DONTWAIT);
+	if (err != size) {
+		/* If we have nothing in the receive buffer now, to reduce
+		 * application latency, try to drain the backend queues as
+		 * quickly as possible, and let remote TCP know what we have
+		 * received so far. */
+		if (err == -EAGAIN) {
+			drbd_tcp_quickack(connection->data.socket);
+			drbd_unplug_all_devices(connection);
+		}
+		if (err > 0) {
+			buffer += err;
+			size -= err;
+		}
+		err = drbd_recv_all_warn(connection, buffer, size);
+		if (err)
+			return err;
+	}
+
+	err = decode_header(connection, connection->data.rbuf, pi);
+	connection->last_received = jiffies;
+
+	return err;
+}
 /* This is blkdev_issue_flush, but asynchronous.
  * We want to submit to all component volumes in parallel,
  * then wait for all completions.
@@ -1223,7 +1264,7 @@ struct one_flush_context {
 	struct issue_flush_context *ctx;
 };
 
-void one_flush_endio(struct bio *bio)
+static void one_flush_endio(struct bio *bio)
 {
 	struct one_flush_context *octx = bio->bi_private;
 	struct drbd_device *device = octx->device;
@@ -1265,7 +1306,7 @@ static void submit_one_flush(struct drbd_device *device, struct issue_flush_cont
 
 	octx->device = device;
 	octx->ctx = ctx;
-	bio->bi_bdev = device->ldev->backing_bdev;
+	bio_set_dev(bio, device->ldev->backing_bdev);
 	bio->bi_private = octx;
 	bio->bi_end_io = one_flush_endio;
 	bio->bi_opf = REQ_OP_FLUSH | REQ_PREFLUSH;
@@ -1548,7 +1589,7 @@ next_bio:
 	}
 	/* > peer_req->i.sector, unless this is the first bio */
 	bio->bi_iter.bi_sector = sector;
-	bio->bi_bdev = device->ldev->backing_bdev;
+	bio_set_dev(bio, device->ldev->backing_bdev);
 	bio_set_op_attrs(bio, op, op_flags);
 	bio->bi_private = peer_req;
 	bio->bi_end_io = drbd_peer_request_endio;
@@ -4085,7 +4126,7 @@ static int receive_uuids(struct drbd_connection *connection, struct packet_info
 		return config_unknown_volume(connection, pi);
 	device = peer_device->device;
 
-	p_uuid = kmalloc(sizeof(u64)*UI_EXTENDED_SIZE, GFP_NOIO);
+	p_uuid = kmalloc_array(UI_EXTENDED_SIZE, sizeof(*p_uuid), GFP_NOIO);
 	if (!p_uuid) {
 		drbd_err(device, "kmalloc of p_uuid failed\n");
 		return false;
@@ -4882,8 +4923,8 @@ static void drbdd(struct drbd_connection *connection)
 		struct data_cmd const *cmd;
 
 		drbd_thread_current_set_cpu(&connection->receiver);
-		update_receiver_timing_details(connection, drbd_recv_header);
-		if (drbd_recv_header(connection, &pi))
+		update_receiver_timing_details(connection, drbd_recv_header_maybe_unplug);
+		if (drbd_recv_header_maybe_unplug(connection, &pi))
 			goto err_out;
 
 		cmd = &drbd_cmd_handler[pi.cmd];
@@ -5375,8 +5416,11 @@ int drbd_receiver(struct drbd_thread *thi)
 		}
 	} while (h == 0);
 
-	if (h > 0)
+	if (h > 0) {
+		blk_start_plug(&connection->receiver_plug);
 		drbdd(connection);
+		blk_finish_plug(&connection->receiver_plug);
+	}
 
 	conn_disconnect(connection);
 

drivers/block/drbd/drbd_req.c

@@ -36,14 +36,18 @@ static bool drbd_may_do_local_read(struct drbd_device *device, sector_t sector,
 /* Update disk stats at start of I/O request */
 static void _drbd_start_io_acct(struct drbd_device *device, struct drbd_request *req)
 {
-	generic_start_io_acct(bio_data_dir(req->master_bio), req->i.size >> 9,
-			      &device->vdisk->part0);
+	struct request_queue *q = device->rq_queue;
+
+	generic_start_io_acct(q, bio_data_dir(req->master_bio),
+				req->i.size >> 9, &device->vdisk->part0);
 }
 
 /* Update disk stats when completing request upwards */
 static void _drbd_end_io_acct(struct drbd_device *device, struct drbd_request *req)
 {
-	generic_end_io_acct(bio_data_dir(req->master_bio),
+	struct request_queue *q = device->rq_queue;
+
+	generic_end_io_acct(q, bio_data_dir(req->master_bio),
 			    &device->vdisk->part0, req->start_jif);
 }
 
@@ -1175,7 +1179,7 @@ drbd_submit_req_private_bio(struct drbd_request *req)
 	else
 		type = DRBD_FAULT_DT_RD;
 
-	bio->bi_bdev = device->ldev->backing_bdev;
+	bio_set_dev(bio, device->ldev->backing_bdev);
 
 	/* State may have changed since we grabbed our reference on the
 	 * ->ldev member. Double check, and short-circuit to endio.
@@ -1275,6 +1279,57 @@ static bool may_do_writes(struct drbd_device *device)
 	return s.disk == D_UP_TO_DATE || s.pdsk == D_UP_TO_DATE;
 }
 
+struct drbd_plug_cb {
+	struct blk_plug_cb cb;
+	struct drbd_request *most_recent_req;
+	/* do we need more? */
+};
+
+static void drbd_unplug(struct blk_plug_cb *cb, bool from_schedule)
+{
+	struct drbd_plug_cb *plug = container_of(cb, struct drbd_plug_cb, cb);
+	struct drbd_resource *resource = plug->cb.data;
+	struct drbd_request *req = plug->most_recent_req;
+
+	kfree(cb);
+	if (!req)
+		return;
+
+	spin_lock_irq(&resource->req_lock);
+	/* In case the sender did not process it yet, raise the flag to
+	 * have it followed with P_UNPLUG_REMOTE just after. */
+	req->rq_state |= RQ_UNPLUG;
+	/* but also queue a generic unplug */
+	drbd_queue_unplug(req->device);
+	kref_put(&req->kref, drbd_req_destroy);
+	spin_unlock_irq(&resource->req_lock);
+}
+
+static struct drbd_plug_cb* drbd_check_plugged(struct drbd_resource *resource)
+{
+	/* A lot of text to say
+	 * return (struct drbd_plug_cb*)blk_check_plugged(); */
+	struct drbd_plug_cb *plug;
+	struct blk_plug_cb *cb = blk_check_plugged(drbd_unplug, resource, sizeof(*plug));
+
+	if (cb)
+		plug = container_of(cb, struct drbd_plug_cb, cb);
+	else
+		plug = NULL;
+	return plug;
+}
+
+static void drbd_update_plug(struct drbd_plug_cb *plug, struct drbd_request *req)
+{
+	struct drbd_request *tmp = plug->most_recent_req;
+	/* Will be sent to some peer.
+	 * Remember to tag it with UNPLUG_REMOTE on unplug */
+	kref_get(&req->kref);
+	plug->most_recent_req = req;
+	if (tmp)
+		kref_put(&tmp->kref, drbd_req_destroy);
+}
+
 static void drbd_send_and_submit(struct drbd_device *device, struct drbd_request *req)
 {
 	struct drbd_resource *resource = device->resource;
@@ -1347,6 +1402,12 @@ static void drbd_send_and_submit(struct drbd_device *device, struct drbd_request
 			no_remote = true;
 	}
 
+	if (no_remote == false) {
+		struct drbd_plug_cb *plug = drbd_check_plugged(resource);
+		if (plug)
+			drbd_update_plug(plug, req);
+	}
+
 	/* If it took the fast path in drbd_request_prepare, add it here.
 	 * The slow path has added it already. */
 	if (list_empty(&req->req_pending_master_completion))
@@ -1395,7 +1456,10 @@ void __drbd_make_request(struct drbd_device *device, struct bio *bio, unsigned l
 
 static void submit_fast_path(struct drbd_device *device, struct list_head *incoming)
 {
+	struct blk_plug plug;
 	struct drbd_request *req, *tmp;
+
+	blk_start_plug(&plug);
 	list_for_each_entry_safe(req, tmp, incoming, tl_requests) {
 		const int rw = bio_data_dir(req->master_bio);
 
@@ -1413,6 +1477,7 @@ static void submit_fast_path(struct drbd_device *device, struct list_head *incom
 		list_del_init(&req->tl_requests);
 		drbd_send_and_submit(device, req);
 	}
+	blk_finish_plug(&plug);
 }
 
 static bool prepare_al_transaction_nonblock(struct drbd_device *device,
@@ -1420,12 +1485,12 @@ static bool prepare_al_transaction_nonblock(struct drbd_device *device,
 					    struct list_head *pending,
 					    struct list_head *later)
 {
-	struct drbd_request *req, *tmp;
+	struct drbd_request *req;
 	int wake = 0;
 	int err;
 
 	spin_lock_irq(&device->al_lock);
-	list_for_each_entry_safe(req, tmp, incoming, tl_requests) {
+	while ((req = list_first_entry_or_null(incoming, struct drbd_request, tl_requests))) {
 		err = drbd_al_begin_io_nonblock(device, &req->i);
 		if (err == -ENOBUFS)
 			break;
@@ -1442,17 +1507,20 @@ static bool prepare_al_transaction_nonblock(struct drbd_device *device,
 	return !list_empty(pending);
 }
 
-void send_and_submit_pending(struct drbd_device *device, struct list_head *pending)
+static void send_and_submit_pending(struct drbd_device *device, struct list_head *pending)
 {
-	struct drbd_request *req, *tmp;
+	struct blk_plug plug;
+	struct drbd_request *req;
 
-	list_for_each_entry_safe(req, tmp, pending, tl_requests) {
+	blk_start_plug(&plug);
+	while ((req = list_first_entry_or_null(pending, struct drbd_request, tl_requests))) {
 		req->rq_state |= RQ_IN_ACT_LOG;
 		req->in_actlog_jif = jiffies;
 		atomic_dec(&device->ap_actlog_cnt);
 		list_del_init(&req->tl_requests);
 		drbd_send_and_submit(device, req);
 	}
+	blk_finish_plug(&plug);
 }
 
 void do_submit(struct work_struct *ws)

drivers/block/drbd/drbd_req.h

@@ -212,6 +212,11 @@ enum drbd_req_state_bits {
 	/* Should call drbd_al_complete_io() for this request... */
 	__RQ_IN_ACT_LOG,
 
+	/* This was the most recent request during some blk_finish_plug()
+	 * or its implicit from-schedule equivalent.
+	 * We may use it as hint to send a P_UNPLUG_REMOTE */
+	__RQ_UNPLUG,
+
 	/* The peer has sent a retry ACK */
 	__RQ_POSTPONED,
 
@@ -249,6 +254,7 @@ enum drbd_req_state_bits {
 #define RQ_WSAME           (1UL << __RQ_WSAME)
 #define RQ_UNMAP           (1UL << __RQ_UNMAP)
 #define RQ_IN_ACT_LOG      (1UL << __RQ_IN_ACT_LOG)
+#define RQ_UNPLUG          (1UL << __RQ_UNPLUG)
 #define RQ_POSTPONED	   (1UL << __RQ_POSTPONED)
 #define RQ_COMPLETION_SUSP (1UL << __RQ_COMPLETION_SUSP)
 #define RQ_EXP_RECEIVE_ACK (1UL << __RQ_EXP_RECEIVE_ACK)

drivers/block/drbd/drbd_state.c

@@ -346,7 +346,7 @@ static enum drbd_role min_role(enum drbd_role role1, enum drbd_role role2)
 
 enum drbd_role conn_highest_role(struct drbd_connection *connection)
 {
-	enum drbd_role role = R_UNKNOWN;
+	enum drbd_role role = R_SECONDARY;
 	struct drbd_peer_device *peer_device;
 	int vnr;
 
@@ -579,11 +579,14 @@ drbd_req_state(struct drbd_device *device, union drbd_state mask,
 	unsigned long flags;
 	union drbd_state os, ns;
 	enum drbd_state_rv rv;
+	void *buffer = NULL;
 
 	init_completion(&done);
 
 	if (f & CS_SERIALIZE)
 		mutex_lock(device->state_mutex);
+	if (f & CS_INHIBIT_MD_IO)
+		buffer = drbd_md_get_buffer(device, __func__);
 
 	spin_lock_irqsave(&device->resource->req_lock, flags);
 	os = drbd_read_state(device);
@@ -636,6 +639,8 @@ drbd_req_state(struct drbd_device *device, union drbd_state mask,
 	}
 
 abort:
+	if (buffer)
+		drbd_md_put_buffer(device);
 	if (f & CS_SERIALIZE)
 		mutex_unlock(device->state_mutex);
 
@@ -664,6 +669,47 @@ _drbd_request_state(struct drbd_device *device, union drbd_state mask,
 	return rv;
 }
 
+/*
+ * We grab drbd_md_get_buffer(), because we don't want to "fail" the disk while
+ * there is IO in-flight: the transition into D_FAILED for detach purposes
+ * may get misinterpreted as actual IO error in a confused endio function.
+ *
+ * We wrap it all into wait_event(), to retry in case the drbd_req_state()
+ * returns SS_IN_TRANSIENT_STATE.
+ *
+ * To avoid potential deadlock with e.g. the receiver thread trying to grab
+ * drbd_md_get_buffer() while trying to get out of the "transient state", we
+ * need to grab and release the meta data buffer inside of that wait_event loop.
+ */
+static enum drbd_state_rv
+request_detach(struct drbd_device *device)
+{
+	return drbd_req_state(device, NS(disk, D_FAILED),
+			CS_VERBOSE | CS_ORDERED | CS_INHIBIT_MD_IO);
+}
+
+enum drbd_state_rv
+drbd_request_detach_interruptible(struct drbd_device *device)
+{
+	enum drbd_state_rv rv;
+	int ret;
+
+	drbd_suspend_io(device); /* so no-one is stuck in drbd_al_begin_io */
+	wait_event_interruptible(device->state_wait,
+		(rv = request_detach(device)) != SS_IN_TRANSIENT_STATE);
+	drbd_resume_io(device);
+
+	ret = wait_event_interruptible(device->misc_wait,
+			device->state.disk != D_FAILED);
+
+	if (rv == SS_IS_DISKLESS)
+		rv = SS_NOTHING_TO_DO;
+	if (ret)
+		rv = ERR_INTR;
+
+	return rv;
+}
+
 enum drbd_state_rv
 _drbd_request_state_holding_state_mutex(struct drbd_device *device, union drbd_state mask,
 		    union drbd_state val, enum chg_state_flags f)

drivers/block/drbd/drbd_state.h

@@ -71,6 +71,10 @@ enum chg_state_flags {
 	CS_DC_SUSP       = 1 << 10,
 	CS_DC_MASK       = CS_DC_ROLE + CS_DC_PEER + CS_DC_CONN + CS_DC_DISK + CS_DC_PDSK,
 	CS_IGN_OUTD_FAIL = 1 << 11,
+
+	/* Make sure no meta data IO is in flight, by calling
+	 * drbd_md_get_buffer().  Used for graceful detach. */
+	CS_INHIBIT_MD_IO = 1 << 12,
 };
 
 /* drbd_dev_state and drbd_state are different types. This is to stress the
@@ -156,6 +160,10 @@ static inline int drbd_request_state(struct drbd_device *device,
 	return _drbd_request_state(device, mask, val, CS_VERBOSE + CS_ORDERED);
 }
 
+/* for use in adm_detach() (drbd_adm_detach(), drbd_adm_down()) */
+enum drbd_state_rv
+drbd_request_detach_interruptible(struct drbd_device *device);
+
 enum drbd_role conn_highest_role(struct drbd_connection *connection);
 enum drbd_role conn_highest_peer(struct drbd_connection *connection);
 enum drbd_disk_state conn_highest_disk(struct drbd_connection *connection);

drivers/block/drbd/drbd_worker.c

@@ -65,6 +65,11 @@ void drbd_md_endio(struct bio *bio)
 	device = bio->bi_private;
 	device->md_io.error = blk_status_to_errno(bio->bi_status);
 
+	/* special case: drbd_md_read() during drbd_adm_attach() */
+	if (device->ldev)
+		put_ldev(device);
+	bio_put(bio);
+
 	/* We grabbed an extra reference in _drbd_md_sync_page_io() to be able
 	 * to timeout on the lower level device, and eventually detach from it.
 	 * If this io completion runs after that timeout expired, this
@@ -79,9 +84,6 @@ void drbd_md_endio(struct bio *bio)
 	drbd_md_put_buffer(device);
 	device->md_io.done = 1;
 	wake_up(&device->misc_wait);
-	bio_put(bio);
-	if (device->ldev) /* special case: drbd_md_read() during drbd_adm_attach() */
-		put_ldev(device);
 }
 
 /* reads on behalf of the partner,
@@ -128,6 +130,14 @@ void drbd_endio_write_sec_final(struct drbd_peer_request *peer_req) __releases(l
 	block_id = peer_req->block_id;
 	peer_req->flags &= ~EE_CALL_AL_COMPLETE_IO;
 
+	if (peer_req->flags & EE_WAS_ERROR) {
+		/* In protocol != C, we usually do not send write acks.
+		 * In case of a write error, send the neg ack anyways. */
+		if (!__test_and_set_bit(__EE_SEND_WRITE_ACK, &peer_req->flags))
+			inc_unacked(device);
+		drbd_set_out_of_sync(device, peer_req->i.sector, peer_req->i.size);
+	}
+
 	spin_lock_irqsave(&device->resource->req_lock, flags);
 	device->writ_cnt += peer_req->i.size >> 9;
 	list_move_tail(&peer_req->w.list, &device->done_ee);
@@ -195,7 +205,8 @@ void drbd_peer_request_endio(struct bio *bio)
 	}
 }
 
-void drbd_panic_after_delayed_completion_of_aborted_request(struct drbd_device *device)
+static void
+drbd_panic_after_delayed_completion_of_aborted_request(struct drbd_device *device)
 {
 	panic("drbd%u %s/%u potential random memory corruption caused by delayed completion of aborted local request\n",
 		device->minor, device->resource->name, device->vnr);
@@ -1382,18 +1393,22 @@ static int drbd_send_barrier(struct drbd_connection *connection)
 	return conn_send_command(connection, sock, P_BARRIER, sizeof(*p), NULL, 0);
 }
 
+static int pd_send_unplug_remote(struct drbd_peer_device *pd)
+{
+	struct drbd_socket *sock = &pd->connection->data;
+	if (!drbd_prepare_command(pd, sock))
+		return -EIO;
+	return drbd_send_command(pd, sock, P_UNPLUG_REMOTE, 0, NULL, 0);
+}
+
 int w_send_write_hint(struct drbd_work *w, int cancel)
 {
 	struct drbd_device *device =
 		container_of(w, struct drbd_device, unplug_work);
-	struct drbd_socket *sock;
 
 	if (cancel)
 		return 0;
-	sock = &first_peer_device(device)->connection->data;
-	if (!drbd_prepare_command(first_peer_device(device), sock))
-		return -EIO;
-	return drbd_send_command(first_peer_device(device), sock, P_UNPLUG_REMOTE, 0, NULL, 0);
+	return pd_send_unplug_remote(first_peer_device(device));
 }
 
 static void re_init_if_first_write(struct drbd_connection *connection, unsigned int epoch)
@@ -1455,6 +1470,7 @@ int w_send_dblock(struct drbd_work *w, int cancel)
 	struct drbd_device *device = req->device;
 	struct drbd_peer_device *const peer_device = first_peer_device(device);
 	struct drbd_connection *connection = peer_device->connection;
+	bool do_send_unplug = req->rq_state & RQ_UNPLUG;
 	int err;
 
 	if (unlikely(cancel)) {
@@ -1470,6 +1486,9 @@ int w_send_dblock(struct drbd_work *w, int cancel)
 	err = drbd_send_dblock(peer_device, req);
 	req_mod(req, err ? SEND_FAILED : HANDED_OVER_TO_NETWORK);
 
+	if (do_send_unplug && !err)
+		pd_send_unplug_remote(peer_device);
+
 	return err;
 }
 
@@ -1484,6 +1503,7 @@ int w_send_read_req(struct drbd_work *w, int cancel)
 	struct drbd_device *device = req->device;
 	struct drbd_peer_device *const peer_device = first_peer_device(device);
 	struct drbd_connection *connection = peer_device->connection;
+	bool do_send_unplug = req->rq_state & RQ_UNPLUG;
 	int err;
 
 	if (unlikely(cancel)) {
@@ -1501,6 +1521,9 @@ int w_send_read_req(struct drbd_work *w, int cancel)
 
 	req_mod(req, err ? SEND_FAILED : HANDED_OVER_TO_NETWORK);
 
+	if (do_send_unplug && !err)
+		pd_send_unplug_remote(peer_device);
+
 	return err;
 }
 
@@ -1513,7 +1536,7 @@ int w_restart_disk_io(struct drbd_work *w, int cancel)
 		drbd_al_begin_io(device, &req->i);
 
 	drbd_req_make_private_bio(req, req->master_bio);
-	req->private_bio->bi_bdev = device->ldev->backing_bdev;
+	bio_set_dev(req->private_bio, device->ldev->backing_bdev);
 	generic_make_request(req->private_bio);
 
 	return 0;
@@ -1733,6 +1756,11 @@ void drbd_start_resync(struct drbd_device *device, enum drbd_conns side)
 		return;
 	}
 
+	if (!connection) {
+		drbd_err(device, "No connection to peer, aborting!\n");
+		return;
+	}
+
 	if (!test_bit(B_RS_H_DONE, &device->flags)) {
 		if (side == C_SYNC_TARGET) {
 			/* Since application IO was locked out during C_WF_BITMAP_T and

drivers/block/floppy.c

@@ -4134,7 +4134,7 @@ static int __floppy_read_block_0(struct block_device *bdev, int drive)
 	cbdata.drive = drive;
 
 	bio_init(&bio, &bio_vec, 1);
-	bio.bi_bdev = bdev;
+	bio_set_dev(&bio, bdev);
 	bio_add_page(&bio, page, size, 0);
 
 	bio.bi_iter.bi_sector = 0;

drivers/block/loop.c

@@ -1966,10 +1966,6 @@ static int __init loop_init(void)
 	struct loop_device *lo;
 	int err;
 
-	err = misc_register(&loop_misc);
-	if (err < 0)
-		return err;
-
 	part_shift = 0;
 	if (max_part > 0) {
 		part_shift = fls(max_part);
@@ -1987,12 +1983,12 @@ static int __init loop_init(void)
 
 	if ((1UL << part_shift) > DISK_MAX_PARTS) {
 		err = -EINVAL;
-		goto misc_out;
+		goto err_out;
 	}
 
 	if (max_loop > 1UL << (MINORBITS - part_shift)) {
 		err = -EINVAL;
-		goto misc_out;
+		goto err_out;
 	}
 
 	/*
@@ -2011,6 +2007,11 @@ static int __init loop_init(void)
 		range = 1UL << MINORBITS;
 	}
 
+	err = misc_register(&loop_misc);
+	if (err < 0)
+		goto err_out;
+
+
 	if (register_blkdev(LOOP_MAJOR, "loop")) {
 		err = -EIO;
 		goto misc_out;
@@ -2030,6 +2031,7 @@ static int __init loop_init(void)
 
 misc_out:
 	misc_deregister(&loop_misc);
+err_out:
 	return err;
 }
 

drivers/block/nbd.c

@@ -128,7 +128,7 @@ static struct dentry *nbd_dbg_dir;
 #define NBD_MAGIC 0x68797548
 
 static unsigned int nbds_max = 16;
-static int max_part;
+static int max_part = 16;
 static struct workqueue_struct *recv_workqueue;
 static int part_shift;
 
@@ -165,7 +165,7 @@ static ssize_t pid_show(struct device *dev,
 	return sprintf(buf, "%d\n", task_pid_nr(nbd->task_recv));
 }
 
-static struct device_attribute pid_attr = {
+static const struct device_attribute pid_attr = {
 	.attr = { .name = "pid", .mode = S_IRUGO},
 	.show = pid_show,
 };
@@ -1584,6 +1584,15 @@ again:
 		}
 	} else {
 		nbd = idr_find(&nbd_index_idr, index);
+		if (!nbd) {
+			ret = nbd_dev_add(index);
+			if (ret < 0) {
+				mutex_unlock(&nbd_index_mutex);
+				printk(KERN_ERR "nbd: failed to add new device\n");
+				return ret;
+			}
+			nbd = idr_find(&nbd_index_idr, index);
+		}
 	}
 	if (!nbd) {
 		printk(KERN_ERR "nbd: couldn't find device at index %d\n",
@@ -2137,4 +2146,4 @@ MODULE_LICENSE("GPL");
 module_param(nbds_max, int, 0444);
 MODULE_PARM_DESC(nbds_max, "number of network block devices to initialize (default: 16)");
 module_param(max_part, int, 0444);
-MODULE_PARM_DESC(max_part, "number of partitions per device (default: 0)");
+MODULE_PARM_DESC(max_part, "number of partitions per device (default: 16)");

drivers/block/null_blk.c

@@ -1,3 +1,7 @@
+/*
+ * Add configfs and memory store: Kyungchan Koh <kkc6196@fb.com> and
+ * Shaohua Li <shli@fb.com>
+ */
 #include <linux/module.h>
 
 #include <linux/moduleparam.h>
@@ -9,6 +13,24 @@
 #include <linux/blk-mq.h>
 #include <linux/hrtimer.h>
 #include <linux/lightnvm.h>
+#include <linux/configfs.h>
+#include <linux/badblocks.h>
+
+#define SECTOR_SHIFT		9
+#define PAGE_SECTORS_SHIFT	(PAGE_SHIFT - SECTOR_SHIFT)
+#define PAGE_SECTORS		(1 << PAGE_SECTORS_SHIFT)
+#define SECTOR_SIZE		(1 << SECTOR_SHIFT)
+#define SECTOR_MASK		(PAGE_SECTORS - 1)
+
+#define FREE_BATCH		16
+
+#define TICKS_PER_SEC		50ULL
+#define TIMER_INTERVAL		(NSEC_PER_SEC / TICKS_PER_SEC)
+
+static inline u64 mb_per_tick(int mbps)
+{
+	return (1 << 20) / TICKS_PER_SEC * ((u64) mbps);
+}
 
 struct nullb_cmd {
 	struct list_head list;
@@ -19,17 +41,82 @@ struct nullb_cmd {
 	unsigned int tag;
 	struct nullb_queue *nq;
 	struct hrtimer timer;
+	blk_status_t error;
 };
 
 struct nullb_queue {
 	unsigned long *tag_map;
 	wait_queue_head_t wait;
 	unsigned int queue_depth;
+	struct nullb_device *dev;
 
 	struct nullb_cmd *cmds;
 };
 
+/*
+ * Status flags for nullb_device.
+ *
+ * CONFIGURED:	Device has been configured and turned on. Cannot reconfigure.
+ * UP:		Device is currently on and visible in userspace.
+ * THROTTLED:	Device is being throttled.
+ * CACHE:	Device is using a write-back cache.
+ */
+enum nullb_device_flags {
+	NULLB_DEV_FL_CONFIGURED	= 0,
+	NULLB_DEV_FL_UP		= 1,
+	NULLB_DEV_FL_THROTTLED	= 2,
+	NULLB_DEV_FL_CACHE	= 3,
+};
+
+/*
+ * nullb_page is a page in memory for nullb devices.
+ *
+ * @page:	The page holding the data.
+ * @bitmap:	The bitmap represents which sector in the page has data.
+ *		Each bit represents one block size. For example, sector 8
+ *		will use the 7th bit
+ * The highest 2 bits of bitmap are for special purpose. LOCK means the cache
+ * page is being flushing to storage. FREE means the cache page is freed and
+ * should be skipped from flushing to storage. Please see
+ * null_make_cache_space
+ */
+struct nullb_page {
+	struct page *page;
+	unsigned long bitmap;
+};
+#define NULLB_PAGE_LOCK (sizeof(unsigned long) * 8 - 1)
+#define NULLB_PAGE_FREE (sizeof(unsigned long) * 8 - 2)
+
+struct nullb_device {
+	struct nullb *nullb;
+	struct config_item item;
+	struct radix_tree_root data; /* data stored in the disk */
+	struct radix_tree_root cache; /* disk cache data */
+	unsigned long flags; /* device flags */
+	unsigned int curr_cache;
+	struct badblocks badblocks;
+
+	unsigned long size; /* device size in MB */
+	unsigned long completion_nsec; /* time in ns to complete a request */
+	unsigned long cache_size; /* disk cache size in MB */
+	unsigned int submit_queues; /* number of submission queues */
+	unsigned int home_node; /* home node for the device */
+	unsigned int queue_mode; /* block interface */
+	unsigned int blocksize; /* block size */
+	unsigned int irqmode; /* IRQ completion handler */
+	unsigned int hw_queue_depth; /* queue depth */
+	unsigned int index; /* index of the disk, only valid with a disk */
+	unsigned int mbps; /* Bandwidth throttle cap (in MB/s) */
+	bool use_lightnvm; /* register as a LightNVM device */
+	bool blocking; /* blocking blk-mq device */
+	bool use_per_node_hctx; /* use per-node allocation for hardware context */
+	bool power; /* power on/off the device */
+	bool memory_backed; /* if data is stored in memory */
+	bool discard; /* if support discard */
+};
+
 struct nullb {
+	struct nullb_device *dev;
 	struct list_head list;
 	unsigned int index;
 	struct request_queue *q;
@@ -37,8 +124,10 @@ struct nullb {
 	struct nvm_dev *ndev;
 	struct blk_mq_tag_set *tag_set;
 	struct blk_mq_tag_set __tag_set;
-	struct hrtimer timer;
 	unsigned int queue_depth;
+	atomic_long_t cur_bytes;
+	struct hrtimer bw_timer;
+	unsigned long cache_flush_pos;
 	spinlock_t lock;
 
 	struct nullb_queue *queues;
@@ -49,7 +138,7 @@ struct nullb {
 static LIST_HEAD(nullb_list);
 static struct mutex lock;
 static int null_major;
-static int nullb_indexes;
+static DEFINE_IDA(nullb_indexes);
 static struct kmem_cache *ppa_cache;
 static struct blk_mq_tag_set tag_set;
 
@@ -65,15 +154,15 @@ enum {
 	NULL_Q_MQ		= 2,
 };
 
-static int submit_queues;
-module_param(submit_queues, int, S_IRUGO);
+static int g_submit_queues = 1;
+module_param_named(submit_queues, g_submit_queues, int, S_IRUGO);
 MODULE_PARM_DESC(submit_queues, "Number of submission queues");
 
-static int home_node = NUMA_NO_NODE;
-module_param(home_node, int, S_IRUGO);
+static int g_home_node = NUMA_NO_NODE;
+module_param_named(home_node, g_home_node, int, S_IRUGO);
 MODULE_PARM_DESC(home_node, "Home node for the device");
 
-static int queue_mode = NULL_Q_MQ;
+static int g_queue_mode = NULL_Q_MQ;
 
 static int null_param_store_val(const char *str, int *val, int min, int max)
 {
@@ -92,7 +181,7 @@ static int null_param_store_val(const char *str, int *val, int min, int max)
 
 static int null_set_queue_mode(const char *str, const struct kernel_param *kp)
 {
-	return null_param_store_val(str, &queue_mode, NULL_Q_BIO, NULL_Q_MQ);
+	return null_param_store_val(str, &g_queue_mode, NULL_Q_BIO, NULL_Q_MQ);
 }
 
 static const struct kernel_param_ops null_queue_mode_param_ops = {
@@ -100,38 +189,38 @@ static const struct kernel_param_ops null_queue_mode_param_ops = {
 	.get	= param_get_int,
 };
 
-device_param_cb(queue_mode, &null_queue_mode_param_ops, &queue_mode, S_IRUGO);
+device_param_cb(queue_mode, &null_queue_mode_param_ops, &g_queue_mode, S_IRUGO);
 MODULE_PARM_DESC(queue_mode, "Block interface to use (0=bio,1=rq,2=multiqueue)");
 
-static int gb = 250;
-module_param(gb, int, S_IRUGO);
+static int g_gb = 250;
+module_param_named(gb, g_gb, int, S_IRUGO);
 MODULE_PARM_DESC(gb, "Size in GB");
 
-static int bs = 512;
-module_param(bs, int, S_IRUGO);
+static int g_bs = 512;
+module_param_named(bs, g_bs, int, S_IRUGO);
 MODULE_PARM_DESC(bs, "Block size (in bytes)");
 
 static int nr_devices = 1;
 module_param(nr_devices, int, S_IRUGO);
 MODULE_PARM_DESC(nr_devices, "Number of devices to register");
 
-static bool use_lightnvm;
-module_param(use_lightnvm, bool, S_IRUGO);
+static bool g_use_lightnvm;
+module_param_named(use_lightnvm, g_use_lightnvm, bool, S_IRUGO);
 MODULE_PARM_DESC(use_lightnvm, "Register as a LightNVM device");
 
-static bool blocking;
-module_param(blocking, bool, S_IRUGO);
+static bool g_blocking;
+module_param_named(blocking, g_blocking, bool, S_IRUGO);
 MODULE_PARM_DESC(blocking, "Register as a blocking blk-mq driver device");
 
 static bool shared_tags;
 module_param(shared_tags, bool, S_IRUGO);
 MODULE_PARM_DESC(shared_tags, "Share tag set between devices for blk-mq");
 
-static int irqmode = NULL_IRQ_SOFTIRQ;
+static int g_irqmode = NULL_IRQ_SOFTIRQ;
 
 static int null_set_irqmode(const char *str, const struct kernel_param *kp)
 {
-	return null_param_store_val(str, &irqmode, NULL_IRQ_NONE,
+	return null_param_store_val(str, &g_irqmode, NULL_IRQ_NONE,
 					NULL_IRQ_TIMER);
 }
 
@@ -140,21 +229,358 @@ static const struct kernel_param_ops null_irqmode_param_ops = {
 	.get	= param_get_int,
 };
 
-device_param_cb(irqmode, &null_irqmode_param_ops, &irqmode, S_IRUGO);
+device_param_cb(irqmode, &null_irqmode_param_ops, &g_irqmode, S_IRUGO);
 MODULE_PARM_DESC(irqmode, "IRQ completion handler. 0-none, 1-softirq, 2-timer");
 
-static unsigned long completion_nsec = 10000;
-module_param(completion_nsec, ulong, S_IRUGO);
+static unsigned long g_completion_nsec = 10000;
+module_param_named(completion_nsec, g_completion_nsec, ulong, S_IRUGO);
 MODULE_PARM_DESC(completion_nsec, "Time in ns to complete a request in hardware. Default: 10,000ns");
 
-static int hw_queue_depth = 64;
-module_param(hw_queue_depth, int, S_IRUGO);
+static int g_hw_queue_depth = 64;
+module_param_named(hw_queue_depth, g_hw_queue_depth, int, S_IRUGO);
 MODULE_PARM_DESC(hw_queue_depth, "Queue depth for each hardware queue. Default: 64");
 
-static bool use_per_node_hctx = false;
-module_param(use_per_node_hctx, bool, S_IRUGO);
+static bool g_use_per_node_hctx;
+module_param_named(use_per_node_hctx, g_use_per_node_hctx, bool, S_IRUGO);
 MODULE_PARM_DESC(use_per_node_hctx, "Use per-node allocation for hardware context queues. Default: false");
 
+static struct nullb_device *null_alloc_dev(void);
+static void null_free_dev(struct nullb_device *dev);
+static void null_del_dev(struct nullb *nullb);
+static int null_add_dev(struct nullb_device *dev);
+static void null_free_device_storage(struct nullb_device *dev, bool is_cache);
+
+static inline struct nullb_device *to_nullb_device(struct config_item *item)
+{
+	return item ? container_of(item, struct nullb_device, item) : NULL;
+}
+
+static inline ssize_t nullb_device_uint_attr_show(unsigned int val, char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%u\n", val);
+}
+
+static inline ssize_t nullb_device_ulong_attr_show(unsigned long val,
+	char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%lu\n", val);
+}
+
+static inline ssize_t nullb_device_bool_attr_show(bool val, char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%u\n", val);
+}
+
+static ssize_t nullb_device_uint_attr_store(unsigned int *val,
+	const char *page, size_t count)
+{
+	unsigned int tmp;
+	int result;
+
+	result = kstrtouint(page, 0, &tmp);
+	if (result)
+		return result;
+
+	*val = tmp;
+	return count;
+}
+
+static ssize_t nullb_device_ulong_attr_store(unsigned long *val,
+	const char *page, size_t count)
+{
+	int result;
+	unsigned long tmp;
+
+	result = kstrtoul(page, 0, &tmp);
+	if (result)
+		return result;
+
+	*val = tmp;
+	return count;
+}
+
+static ssize_t nullb_device_bool_attr_store(bool *val, const char *page,
+	size_t count)
+{
+	bool tmp;
+	int result;
+
+	result = kstrtobool(page,  &tmp);
+	if (result)
+		return result;
+
+	*val = tmp;
+	return count;
+}
+
+/* The following macro should only be used with TYPE = {uint, ulong, bool}. */
+#define NULLB_DEVICE_ATTR(NAME, TYPE)						\
+static ssize_t									\
+nullb_device_##NAME##_show(struct config_item *item, char *page)		\
+{										\
+	return nullb_device_##TYPE##_attr_show(					\
+				to_nullb_device(item)->NAME, page);		\
+}										\
+static ssize_t									\
+nullb_device_##NAME##_store(struct config_item *item, const char *page,		\
+			    size_t count)					\
+{										\
+	if (test_bit(NULLB_DEV_FL_CONFIGURED, &to_nullb_device(item)->flags))	\
+		return -EBUSY;							\
+	return nullb_device_##TYPE##_attr_store(				\
+			&to_nullb_device(item)->NAME, page, count);		\
+}										\
+CONFIGFS_ATTR(nullb_device_, NAME);
+
+NULLB_DEVICE_ATTR(size, ulong);
+NULLB_DEVICE_ATTR(completion_nsec, ulong);
+NULLB_DEVICE_ATTR(submit_queues, uint);
+NULLB_DEVICE_ATTR(home_node, uint);
+NULLB_DEVICE_ATTR(queue_mode, uint);
+NULLB_DEVICE_ATTR(blocksize, uint);
+NULLB_DEVICE_ATTR(irqmode, uint);
+NULLB_DEVICE_ATTR(hw_queue_depth, uint);
+NULLB_DEVICE_ATTR(index, uint);
+NULLB_DEVICE_ATTR(use_lightnvm, bool);
+NULLB_DEVICE_ATTR(blocking, bool);
+NULLB_DEVICE_ATTR(use_per_node_hctx, bool);
+NULLB_DEVICE_ATTR(memory_backed, bool);
+NULLB_DEVICE_ATTR(discard, bool);
+NULLB_DEVICE_ATTR(mbps, uint);
+NULLB_DEVICE_ATTR(cache_size, ulong);
+
+static ssize_t nullb_device_power_show(struct config_item *item, char *page)
+{
+	return nullb_device_bool_attr_show(to_nullb_device(item)->power, page);
+}
+
+static ssize_t nullb_device_power_store(struct config_item *item,
+				     const char *page, size_t count)
+{
+	struct nullb_device *dev = to_nullb_device(item);
+	bool newp = false;
+	ssize_t ret;
+
+	ret = nullb_device_bool_attr_store(&newp, page, count);
+	if (ret < 0)
+		return ret;
+
+	if (!dev->power && newp) {
+		if (test_and_set_bit(NULLB_DEV_FL_UP, &dev->flags))
+			return count;
+		if (null_add_dev(dev)) {
+			clear_bit(NULLB_DEV_FL_UP, &dev->flags);
+			return -ENOMEM;
+		}
+
+		set_bit(NULLB_DEV_FL_CONFIGURED, &dev->flags);
+		dev->power = newp;
+	} else if (dev->power && !newp) {
+		mutex_lock(&lock);
+		dev->power = newp;
+		null_del_dev(dev->nullb);
+		mutex_unlock(&lock);
+		clear_bit(NULLB_DEV_FL_UP, &dev->flags);
+	}
+
+	return count;
+}
+
+CONFIGFS_ATTR(nullb_device_, power);
+
+static ssize_t nullb_device_badblocks_show(struct config_item *item, char *page)
+{
+	struct nullb_device *t_dev = to_nullb_device(item);
+
+	return badblocks_show(&t_dev->badblocks, page, 0);
+}
+
+static ssize_t nullb_device_badblocks_store(struct config_item *item,
+				     const char *page, size_t count)
+{
+	struct nullb_device *t_dev = to_nullb_device(item);
+	char *orig, *buf, *tmp;
+	u64 start, end;
+	int ret;
+
+	orig = kstrndup(page, count, GFP_KERNEL);
+	if (!orig)
+		return -ENOMEM;
+
+	buf = strstrip(orig);
+
+	ret = -EINVAL;
+	if (buf[0] != '+' && buf[0] != '-')
+		goto out;
+	tmp = strchr(&buf[1], '-');
+	if (!tmp)
+		goto out;
+	*tmp = '\0';
+	ret = kstrtoull(buf + 1, 0, &start);
+	if (ret)
+		goto out;
+	ret = kstrtoull(tmp + 1, 0, &end);
+	if (ret)
+		goto out;
+	ret = -EINVAL;
+	if (start > end)
+		goto out;
+	/* enable badblocks */
+	cmpxchg(&t_dev->badblocks.shift, -1, 0);
+	if (buf[0] == '+')
+		ret = badblocks_set(&t_dev->badblocks, start,
+			end - start + 1, 1);
+	else
+		ret = badblocks_clear(&t_dev->badblocks, start,
+			end - start + 1);
+	if (ret == 0)
+		ret = count;
+out:
+	kfree(orig);
+	return ret;
+}
+CONFIGFS_ATTR(nullb_device_, badblocks);
+
+static struct configfs_attribute *nullb_device_attrs[] = {
+	&nullb_device_attr_size,
+	&nullb_device_attr_completion_nsec,
+	&nullb_device_attr_submit_queues,
+	&nullb_device_attr_home_node,
+	&nullb_device_attr_queue_mode,
+	&nullb_device_attr_blocksize,
+	&nullb_device_attr_irqmode,
+	&nullb_device_attr_hw_queue_depth,
+	&nullb_device_attr_index,
+	&nullb_device_attr_use_lightnvm,
+	&nullb_device_attr_blocking,
+	&nullb_device_attr_use_per_node_hctx,
+	&nullb_device_attr_power,
+	&nullb_device_attr_memory_backed,
+	&nullb_device_attr_discard,
+	&nullb_device_attr_mbps,
+	&nullb_device_attr_cache_size,
+	&nullb_device_attr_badblocks,
+	NULL,
+};
+
+static void nullb_device_release(struct config_item *item)
+{
+	struct nullb_device *dev = to_nullb_device(item);
+
+	badblocks_exit(&dev->badblocks);
+	null_free_device_storage(dev, false);
+	null_free_dev(dev);
+}
+
+static struct configfs_item_operations nullb_device_ops = {
+	.release	= nullb_device_release,
+};
+
+static struct config_item_type nullb_device_type = {
+	.ct_item_ops	= &nullb_device_ops,
+	.ct_attrs	= nullb_device_attrs,
+	.ct_owner	= THIS_MODULE,
+};
+
+static struct
+config_item *nullb_group_make_item(struct config_group *group, const char *name)
+{
+	struct nullb_device *dev;
+
+	dev = null_alloc_dev();
+	if (!dev)
+		return ERR_PTR(-ENOMEM);
+
+	config_item_init_type_name(&dev->item, name, &nullb_device_type);
+
+	return &dev->item;
+}
+
+static void
+nullb_group_drop_item(struct config_group *group, struct config_item *item)
+{
+	struct nullb_device *dev = to_nullb_device(item);
+
+	if (test_and_clear_bit(NULLB_DEV_FL_UP, &dev->flags)) {
+		mutex_lock(&lock);
+		dev->power = false;
+		null_del_dev(dev->nullb);
+		mutex_unlock(&lock);
+	}
+
+	config_item_put(item);
+}
+
+static ssize_t memb_group_features_show(struct config_item *item, char *page)
+{
+	return snprintf(page, PAGE_SIZE, "memory_backed,discard,bandwidth,cache,badblocks\n");
+}
+
+CONFIGFS_ATTR_RO(memb_group_, features);
+
+static struct configfs_attribute *nullb_group_attrs[] = {
+	&memb_group_attr_features,
+	NULL,
+};
+
+static struct configfs_group_operations nullb_group_ops = {
+	.make_item	= nullb_group_make_item,
+	.drop_item	= nullb_group_drop_item,
+};
+
+static struct config_item_type nullb_group_type = {
+	.ct_group_ops	= &nullb_group_ops,
+	.ct_attrs	= nullb_group_attrs,
+	.ct_owner	= THIS_MODULE,
+};
+
+static struct configfs_subsystem nullb_subsys = {
+	.su_group = {
+		.cg_item = {
+			.ci_namebuf = "nullb",
+			.ci_type = &nullb_group_type,
+		},
+	},
+};
+
+static inline int null_cache_active(struct nullb *nullb)
+{
+	return test_bit(NULLB_DEV_FL_CACHE, &nullb->dev->flags);
+}
+
+static struct nullb_device *null_alloc_dev(void)
+{
+	struct nullb_device *dev;
+
+	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
+	if (!dev)
+		return NULL;
+	INIT_RADIX_TREE(&dev->data, GFP_ATOMIC);
+	INIT_RADIX_TREE(&dev->cache, GFP_ATOMIC);
+	if (badblocks_init(&dev->badblocks, 0)) {
+		kfree(dev);
+		return NULL;
+	}
+
+	dev->size = g_gb * 1024;
+	dev->completion_nsec = g_completion_nsec;
+	dev->submit_queues = g_submit_queues;
+	dev->home_node = g_home_node;
+	dev->queue_mode = g_queue_mode;
+	dev->blocksize = g_bs;
+	dev->irqmode = g_irqmode;
+	dev->hw_queue_depth = g_hw_queue_depth;
+	dev->use_lightnvm = g_use_lightnvm;
+	dev->blocking = g_blocking;
+	dev->use_per_node_hctx = g_use_per_node_hctx;
+	return dev;
+}
+
+static void null_free_dev(struct nullb_device *dev)
+{
+	kfree(dev);
+}
+
 static void put_tag(struct nullb_queue *nq, unsigned int tag)
 {
 	clear_bit_unlock(tag, nq->tag_map);
@@ -193,7 +619,7 @@ static struct nullb_cmd *__alloc_cmd(struct nullb_queue *nq)
 		cmd = &nq->cmds[tag];
 		cmd->tag = tag;
 		cmd->nq = nq;
-		if (irqmode == NULL_IRQ_TIMER) {
+		if (nq->dev->irqmode == NULL_IRQ_TIMER) {
 			hrtimer_init(&cmd->timer, CLOCK_MONOTONIC,
 				     HRTIMER_MODE_REL);
 			cmd->timer.function = null_cmd_timer_expired;
@@ -229,19 +655,21 @@ static struct nullb_cmd *alloc_cmd(struct nullb_queue *nq, int can_wait)
 static void end_cmd(struct nullb_cmd *cmd)
 {
 	struct request_queue *q = NULL;
+	int queue_mode = cmd->nq->dev->queue_mode;
 
 	if (cmd->rq)
 		q = cmd->rq->q;
 
 	switch (queue_mode)  {
 	case NULL_Q_MQ:
-		blk_mq_end_request(cmd->rq, BLK_STS_OK);
+		blk_mq_end_request(cmd->rq, cmd->error);
 		return;
 	case NULL_Q_RQ:
 		INIT_LIST_HEAD(&cmd->rq->queuelist);
-		blk_end_request_all(cmd->rq, BLK_STS_OK);
+		blk_end_request_all(cmd->rq, cmd->error);
 		break;
 	case NULL_Q_BIO:
+		cmd->bio->bi_status = cmd->error;
 		bio_endio(cmd->bio);
 		break;
 	}
@@ -267,25 +695,582 @@ static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer)
 
 static void null_cmd_end_timer(struct nullb_cmd *cmd)
 {
-	ktime_t kt = completion_nsec;
+	ktime_t kt = cmd->nq->dev->completion_nsec;
 
 	hrtimer_start(&cmd->timer, kt, HRTIMER_MODE_REL);
 }
 
 static void null_softirq_done_fn(struct request *rq)
 {
-	if (queue_mode == NULL_Q_MQ)
+	struct nullb *nullb = rq->q->queuedata;
+
+	if (nullb->dev->queue_mode == NULL_Q_MQ)
 		end_cmd(blk_mq_rq_to_pdu(rq));
 	else
 		end_cmd(rq->special);
 }
 
-static inline void null_handle_cmd(struct nullb_cmd *cmd)
+static struct nullb_page *null_alloc_page(gfp_t gfp_flags)
+{
+	struct nullb_page *t_page;
+
+	t_page = kmalloc(sizeof(struct nullb_page), gfp_flags);
+	if (!t_page)
+		goto out;
+
+	t_page->page = alloc_pages(gfp_flags, 0);
+	if (!t_page->page)
+		goto out_freepage;
+
+	t_page->bitmap = 0;
+	return t_page;
+out_freepage:
+	kfree(t_page);
+out:
+	return NULL;
+}
+
+static void null_free_page(struct nullb_page *t_page)
+{
+	__set_bit(NULLB_PAGE_FREE, &t_page->bitmap);
+	if (test_bit(NULLB_PAGE_LOCK, &t_page->bitmap))
+		return;
+	__free_page(t_page->page);
+	kfree(t_page);
+}
+
+static void null_free_sector(struct nullb *nullb, sector_t sector,
+	bool is_cache)
+{
+	unsigned int sector_bit;
+	u64 idx;
+	struct nullb_page *t_page, *ret;
+	struct radix_tree_root *root;
+
+	root = is_cache ? &nullb->dev->cache : &nullb->dev->data;
+	idx = sector >> PAGE_SECTORS_SHIFT;
+	sector_bit = (sector & SECTOR_MASK);
+
+	t_page = radix_tree_lookup(root, idx);
+	if (t_page) {
+		__clear_bit(sector_bit, &t_page->bitmap);
+
+		if (!t_page->bitmap) {
+			ret = radix_tree_delete_item(root, idx, t_page);
+			WARN_ON(ret != t_page);
+			null_free_page(ret);
+			if (is_cache)
+				nullb->dev->curr_cache -= PAGE_SIZE;
+		}
+	}
+}
+
+static struct nullb_page *null_radix_tree_insert(struct nullb *nullb, u64 idx,
+	struct nullb_page *t_page, bool is_cache)
+{
+	struct radix_tree_root *root;
+
+	root = is_cache ? &nullb->dev->cache : &nullb->dev->data;
+
+	if (radix_tree_insert(root, idx, t_page)) {
+		null_free_page(t_page);
+		t_page = radix_tree_lookup(root, idx);
+		WARN_ON(!t_page || t_page->page->index != idx);
+	} else if (is_cache)
+		nullb->dev->curr_cache += PAGE_SIZE;
+
+	return t_page;
+}
+
+static void null_free_device_storage(struct nullb_device *dev, bool is_cache)
+{
+	unsigned long pos = 0;
+	int nr_pages;
+	struct nullb_page *ret, *t_pages[FREE_BATCH];
+	struct radix_tree_root *root;
+
+	root = is_cache ? &dev->cache : &dev->data;
+
+	do {
+		int i;
+
+		nr_pages = radix_tree_gang_lookup(root,
+				(void **)t_pages, pos, FREE_BATCH);
+
+		for (i = 0; i < nr_pages; i++) {
+			pos = t_pages[i]->page->index;
+			ret = radix_tree_delete_item(root, pos, t_pages[i]);
+			WARN_ON(ret != t_pages[i]);
+			null_free_page(ret);
+		}
+
+		pos++;
+	} while (nr_pages == FREE_BATCH);
+
+	if (is_cache)
+		dev->curr_cache = 0;
+}
+
+static struct nullb_page *__null_lookup_page(struct nullb *nullb,
+	sector_t sector, bool for_write, bool is_cache)
+{
+	unsigned int sector_bit;
+	u64 idx;
+	struct nullb_page *t_page;
+	struct radix_tree_root *root;
+
+	idx = sector >> PAGE_SECTORS_SHIFT;
+	sector_bit = (sector & SECTOR_MASK);
+
+	root = is_cache ? &nullb->dev->cache : &nullb->dev->data;
+	t_page = radix_tree_lookup(root, idx);
+	WARN_ON(t_page && t_page->page->index != idx);
+
+	if (t_page && (for_write || test_bit(sector_bit, &t_page->bitmap)))
+		return t_page;
+
+	return NULL;
+}
+
+static struct nullb_page *null_lookup_page(struct nullb *nullb,
+	sector_t sector, bool for_write, bool ignore_cache)
+{
+	struct nullb_page *page = NULL;
+
+	if (!ignore_cache)
+		page = __null_lookup_page(nullb, sector, for_write, true);
+	if (page)
+		return page;
+	return __null_lookup_page(nullb, sector, for_write, false);
+}
+
+static struct nullb_page *null_insert_page(struct nullb *nullb,
+	sector_t sector, bool ignore_cache)
+{
+	u64 idx;
+	struct nullb_page *t_page;
+
+	t_page = null_lookup_page(nullb, sector, true, ignore_cache);
+	if (t_page)
+		return t_page;
+
+	spin_unlock_irq(&nullb->lock);
+
+	t_page = null_alloc_page(GFP_NOIO);
+	if (!t_page)
+		goto out_lock;
+
+	if (radix_tree_preload(GFP_NOIO))
+		goto out_freepage;
+
+	spin_lock_irq(&nullb->lock);
+	idx = sector >> PAGE_SECTORS_SHIFT;
+	t_page->page->index = idx;
+	t_page = null_radix_tree_insert(nullb, idx, t_page, !ignore_cache);
+	radix_tree_preload_end();
+
+	return t_page;
+out_freepage:
+	null_free_page(t_page);
+out_lock:
+	spin_lock_irq(&nullb->lock);
+	return null_lookup_page(nullb, sector, true, ignore_cache);
+}
+
+static int null_flush_cache_page(struct nullb *nullb, struct nullb_page *c_page)
+{
+	int i;
+	unsigned int offset;
+	u64 idx;
+	struct nullb_page *t_page, *ret;
+	void *dst, *src;
+
+	idx = c_page->page->index;
+
+	t_page = null_insert_page(nullb, idx << PAGE_SECTORS_SHIFT, true);
+
+	__clear_bit(NULLB_PAGE_LOCK, &c_page->bitmap);
+	if (test_bit(NULLB_PAGE_FREE, &c_page->bitmap)) {
+		null_free_page(c_page);
+		if (t_page && t_page->bitmap == 0) {
+			ret = radix_tree_delete_item(&nullb->dev->data,
+				idx, t_page);
+			null_free_page(t_page);
+		}
+		return 0;
+	}
+
+	if (!t_page)
+		return -ENOMEM;
+
+	src = kmap_atomic(c_page->page);
+	dst = kmap_atomic(t_page->page);
+
+	for (i = 0; i < PAGE_SECTORS;
+			i += (nullb->dev->blocksize >> SECTOR_SHIFT)) {
+		if (test_bit(i, &c_page->bitmap)) {
+			offset = (i << SECTOR_SHIFT);
+			memcpy(dst + offset, src + offset,
+				nullb->dev->blocksize);
+			__set_bit(i, &t_page->bitmap);
+		}
+	}
+
+	kunmap_atomic(dst);
+	kunmap_atomic(src);
+
+	ret = radix_tree_delete_item(&nullb->dev->cache, idx, c_page);
+	null_free_page(ret);
+	nullb->dev->curr_cache -= PAGE_SIZE;
+
+	return 0;
+}
+
+static int null_make_cache_space(struct nullb *nullb, unsigned long n)
 {
+	int i, err, nr_pages;
+	struct nullb_page *c_pages[FREE_BATCH];
+	unsigned long flushed = 0, one_round;
+
+again:
+	if ((nullb->dev->cache_size * 1024 * 1024) >
+	     nullb->dev->curr_cache + n || nullb->dev->curr_cache == 0)
+		return 0;
+
+	nr_pages = radix_tree_gang_lookup(&nullb->dev->cache,
+			(void **)c_pages, nullb->cache_flush_pos, FREE_BATCH);
+	/*
+	 * nullb_flush_cache_page could unlock before using the c_pages. To
+	 * avoid race, we don't allow page free
+	 */
+	for (i = 0; i < nr_pages; i++) {
+		nullb->cache_flush_pos = c_pages[i]->page->index;
+		/*
+		 * We found the page which is being flushed to disk by other
+		 * threads
+		 */
+		if (test_bit(NULLB_PAGE_LOCK, &c_pages[i]->bitmap))
+			c_pages[i] = NULL;
+		else
+			__set_bit(NULLB_PAGE_LOCK, &c_pages[i]->bitmap);
+	}
+
+	one_round = 0;
+	for (i = 0; i < nr_pages; i++) {
+		if (c_pages[i] == NULL)
+			continue;
+		err = null_flush_cache_page(nullb, c_pages[i]);
+		if (err)
+			return err;
+		one_round++;
+	}
+	flushed += one_round << PAGE_SHIFT;
+
+	if (n > flushed) {
+		if (nr_pages == 0)
+			nullb->cache_flush_pos = 0;
+		if (one_round == 0) {
+			/* give other threads a chance */
+			spin_unlock_irq(&nullb->lock);
+			spin_lock_irq(&nullb->lock);
+		}
+		goto again;
+	}
+	return 0;
+}
+
+static int copy_to_nullb(struct nullb *nullb, struct page *source,
+	unsigned int off, sector_t sector, size_t n, bool is_fua)
+{
+	size_t temp, count = 0;
+	unsigned int offset;
+	struct nullb_page *t_page;
+	void *dst, *src;
+
+	while (count < n) {
+		temp = min_t(size_t, nullb->dev->blocksize, n - count);
+
+		if (null_cache_active(nullb) && !is_fua)
+			null_make_cache_space(nullb, PAGE_SIZE);
+
+		offset = (sector & SECTOR_MASK) << SECTOR_SHIFT;
+		t_page = null_insert_page(nullb, sector,
+			!null_cache_active(nullb) || is_fua);
+		if (!t_page)
+			return -ENOSPC;
+
+		src = kmap_atomic(source);
+		dst = kmap_atomic(t_page->page);
+		memcpy(dst + offset, src + off + count, temp);
+		kunmap_atomic(dst);
+		kunmap_atomic(src);
+
+		__set_bit(sector & SECTOR_MASK, &t_page->bitmap);
+
+		if (is_fua)
+			null_free_sector(nullb, sector, true);
+
+		count += temp;
+		sector += temp >> SECTOR_SHIFT;
+	}
+	return 0;
+}
+
+static int copy_from_nullb(struct nullb *nullb, struct page *dest,
+	unsigned int off, sector_t sector, size_t n)
+{
+	size_t temp, count = 0;
+	unsigned int offset;
+	struct nullb_page *t_page;
+	void *dst, *src;
+
+	while (count < n) {
+		temp = min_t(size_t, nullb->dev->blocksize, n - count);
+
+		offset = (sector & SECTOR_MASK) << SECTOR_SHIFT;
+		t_page = null_lookup_page(nullb, sector, false,
+			!null_cache_active(nullb));
+
+		dst = kmap_atomic(dest);
+		if (!t_page) {
+			memset(dst + off + count, 0, temp);
+			goto next;
+		}
+		src = kmap_atomic(t_page->page);
+		memcpy(dst + off + count, src + offset, temp);
+		kunmap_atomic(src);
+next:
+		kunmap_atomic(dst);
+
+		count += temp;
+		sector += temp >> SECTOR_SHIFT;
+	}
+	return 0;
+}
+
+static void null_handle_discard(struct nullb *nullb, sector_t sector, size_t n)
+{
+	size_t temp;
+
+	spin_lock_irq(&nullb->lock);
+	while (n > 0) {
+		temp = min_t(size_t, n, nullb->dev->blocksize);
+		null_free_sector(nullb, sector, false);
+		if (null_cache_active(nullb))
+			null_free_sector(nullb, sector, true);
+		sector += temp >> SECTOR_SHIFT;
+		n -= temp;
+	}
+	spin_unlock_irq(&nullb->lock);
+}
+
+static int null_handle_flush(struct nullb *nullb)
+{
+	int err;
+
+	if (!null_cache_active(nullb))
+		return 0;
+
+	spin_lock_irq(&nullb->lock);
+	while (true) {
+		err = null_make_cache_space(nullb,
+			nullb->dev->cache_size * 1024 * 1024);
+		if (err || nullb->dev->curr_cache == 0)
+			break;
+	}
+
+	WARN_ON(!radix_tree_empty(&nullb->dev->cache));
+	spin_unlock_irq(&nullb->lock);
+	return err;
+}
+
+static int null_transfer(struct nullb *nullb, struct page *page,
+	unsigned int len, unsigned int off, bool is_write, sector_t sector,
+	bool is_fua)
+{
+	int err = 0;
+
+	if (!is_write) {
+		err = copy_from_nullb(nullb, page, off, sector, len);
+		flush_dcache_page(page);
+	} else {
+		flush_dcache_page(page);
+		err = copy_to_nullb(nullb, page, off, sector, len, is_fua);
+	}
+
+	return err;
+}
+
+static int null_handle_rq(struct nullb_cmd *cmd)
+{
+	struct request *rq = cmd->rq;
+	struct nullb *nullb = cmd->nq->dev->nullb;
+	int err;
+	unsigned int len;
+	sector_t sector;
+	struct req_iterator iter;
+	struct bio_vec bvec;
+
+	sector = blk_rq_pos(rq);
+
+	if (req_op(rq) == REQ_OP_DISCARD) {
+		null_handle_discard(nullb, sector, blk_rq_bytes(rq));
+		return 0;
+	}
+
+	spin_lock_irq(&nullb->lock);
+	rq_for_each_segment(bvec, rq, iter) {
+		len = bvec.bv_len;
+		err = null_transfer(nullb, bvec.bv_page, len, bvec.bv_offset,
+				     op_is_write(req_op(rq)), sector,
+				     req_op(rq) & REQ_FUA);
+		if (err) {
+			spin_unlock_irq(&nullb->lock);
+			return err;
+		}
+		sector += len >> SECTOR_SHIFT;
+	}
+	spin_unlock_irq(&nullb->lock);
+
+	return 0;
+}
+
+static int null_handle_bio(struct nullb_cmd *cmd)
+{
+	struct bio *bio = cmd->bio;
+	struct nullb *nullb = cmd->nq->dev->nullb;
+	int err;
+	unsigned int len;
+	sector_t sector;
+	struct bio_vec bvec;
+	struct bvec_iter iter;
+
+	sector = bio->bi_iter.bi_sector;
+
+	if (bio_op(bio) == REQ_OP_DISCARD) {
+		null_handle_discard(nullb, sector,
+			bio_sectors(bio) << SECTOR_SHIFT);
+		return 0;
+	}
+
+	spin_lock_irq(&nullb->lock);
+	bio_for_each_segment(bvec, bio, iter) {
+		len = bvec.bv_len;
+		err = null_transfer(nullb, bvec.bv_page, len, bvec.bv_offset,
+				     op_is_write(bio_op(bio)), sector,
+				     bio_op(bio) & REQ_FUA);
+		if (err) {
+			spin_unlock_irq(&nullb->lock);
+			return err;
+		}
+		sector += len >> SECTOR_SHIFT;
+	}
+	spin_unlock_irq(&nullb->lock);
+	return 0;
+}
+
+static void null_stop_queue(struct nullb *nullb)
+{
+	struct request_queue *q = nullb->q;
+
+	if (nullb->dev->queue_mode == NULL_Q_MQ)
+		blk_mq_stop_hw_queues(q);
+	else {
+		spin_lock_irq(q->queue_lock);
+		blk_stop_queue(q);
+		spin_unlock_irq(q->queue_lock);
+	}
+}
+
+static void null_restart_queue_async(struct nullb *nullb)
+{
+	struct request_queue *q = nullb->q;
+	unsigned long flags;
+
+	if (nullb->dev->queue_mode == NULL_Q_MQ)
+		blk_mq_start_stopped_hw_queues(q, true);
+	else {
+		spin_lock_irqsave(q->queue_lock, flags);
+		blk_start_queue_async(q);
+		spin_unlock_irqrestore(q->queue_lock, flags);
+	}
+}
+
+static blk_status_t null_handle_cmd(struct nullb_cmd *cmd)
+{
+	struct nullb_device *dev = cmd->nq->dev;
+	struct nullb *nullb = dev->nullb;
+	int err = 0;
+
+	if (test_bit(NULLB_DEV_FL_THROTTLED, &dev->flags)) {
+		struct request *rq = cmd->rq;
+
+		if (!hrtimer_active(&nullb->bw_timer))
+			hrtimer_restart(&nullb->bw_timer);
+
+		if (atomic_long_sub_return(blk_rq_bytes(rq),
+				&nullb->cur_bytes) < 0) {
+			null_stop_queue(nullb);
+			/* race with timer */
+			if (atomic_long_read(&nullb->cur_bytes) > 0)
+				null_restart_queue_async(nullb);
+			if (dev->queue_mode == NULL_Q_RQ) {
+				struct request_queue *q = nullb->q;
+
+				spin_lock_irq(q->queue_lock);
+				rq->rq_flags |= RQF_DONTPREP;
+				blk_requeue_request(q, rq);
+				spin_unlock_irq(q->queue_lock);
+				return BLK_STS_OK;
+			} else
+				/* requeue request */
+				return BLK_STS_RESOURCE;
+		}
+	}
+
+	if (nullb->dev->badblocks.shift != -1) {
+		int bad_sectors;
+		sector_t sector, size, first_bad;
+		bool is_flush = true;
+
+		if (dev->queue_mode == NULL_Q_BIO &&
+				bio_op(cmd->bio) != REQ_OP_FLUSH) {
+			is_flush = false;
+			sector = cmd->bio->bi_iter.bi_sector;
+			size = bio_sectors(cmd->bio);
+		}
+		if (dev->queue_mode != NULL_Q_BIO &&
+				req_op(cmd->rq) != REQ_OP_FLUSH) {
+			is_flush = false;
+			sector = blk_rq_pos(cmd->rq);
+			size = blk_rq_sectors(cmd->rq);
+		}
+		if (!is_flush && badblocks_check(&nullb->dev->badblocks, sector,
+				size, &first_bad, &bad_sectors)) {
+			cmd->error = BLK_STS_IOERR;
+			goto out;
+		}
+	}
+
+	if (dev->memory_backed) {
+		if (dev->queue_mode == NULL_Q_BIO) {
+			if (bio_op(cmd->bio) == REQ_OP_FLUSH)
+				err = null_handle_flush(nullb);
+			else
+				err = null_handle_bio(cmd);
+		} else {
+			if (req_op(cmd->rq) == REQ_OP_FLUSH)
+				err = null_handle_flush(nullb);
+			else
+				err = null_handle_rq(cmd);
+		}
+	}
+	cmd->error = errno_to_blk_status(err);
+out:
 	/* Complete IO by inline, softirq or timer */
-	switch (irqmode) {
+	switch (dev->irqmode) {
 	case NULL_IRQ_SOFTIRQ:
-		switch (queue_mode)  {
+		switch (dev->queue_mode)  {
 		case NULL_Q_MQ:
 			blk_mq_complete_request(cmd->rq);
 			break;
@@ -307,6 +1292,34 @@ static inline void null_handle_cmd(struct nullb_cmd *cmd)
 		null_cmd_end_timer(cmd);
 		break;
 	}
+	return BLK_STS_OK;
+}
+
+static enum hrtimer_restart nullb_bwtimer_fn(struct hrtimer *timer)
+{
+	struct nullb *nullb = container_of(timer, struct nullb, bw_timer);
+	ktime_t timer_interval = ktime_set(0, TIMER_INTERVAL);
+	unsigned int mbps = nullb->dev->mbps;
+
+	if (atomic_long_read(&nullb->cur_bytes) == mb_per_tick(mbps))
+		return HRTIMER_NORESTART;
+
+	atomic_long_set(&nullb->cur_bytes, mb_per_tick(mbps));
+	null_restart_queue_async(nullb);
+
+	hrtimer_forward_now(&nullb->bw_timer, timer_interval);
+
+	return HRTIMER_RESTART;
+}
+
+static void nullb_setup_bwtimer(struct nullb *nullb)
+{
+	ktime_t timer_interval = ktime_set(0, TIMER_INTERVAL);
+
+	hrtimer_init(&nullb->bw_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	nullb->bw_timer.function = nullb_bwtimer_fn;
+	atomic_long_set(&nullb->cur_bytes, mb_per_tick(nullb->dev->mbps));
+	hrtimer_start(&nullb->bw_timer, timer_interval, HRTIMER_MODE_REL);
 }
 
 static struct nullb_queue *nullb_to_queue(struct nullb *nullb)
@@ -366,20 +1379,20 @@ static blk_status_t null_queue_rq(struct blk_mq_hw_ctx *hctx,
 			 const struct blk_mq_queue_data *bd)
 {
 	struct nullb_cmd *cmd = blk_mq_rq_to_pdu(bd->rq);
+	struct nullb_queue *nq = hctx->driver_data;
 
 	might_sleep_if(hctx->flags & BLK_MQ_F_BLOCKING);
 
-	if (irqmode == NULL_IRQ_TIMER) {
+	if (nq->dev->irqmode == NULL_IRQ_TIMER) {
 		hrtimer_init(&cmd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 		cmd->timer.function = null_cmd_timer_expired;
 	}
 	cmd->rq = bd->rq;
-	cmd->nq = hctx->driver_data;
+	cmd->nq = nq;
 
 	blk_mq_start_request(bd->rq);
 
-	null_handle_cmd(cmd);
-	return BLK_STS_OK;
+	return null_handle_cmd(cmd);
 }
 
 static const struct blk_mq_ops null_mq_ops = {
@@ -438,7 +1451,8 @@ static int null_lnvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd)
 
 static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id)
 {
-	sector_t size = gb * 1024 * 1024 * 1024ULL;
+	struct nullb *nullb = dev->q->queuedata;
+	sector_t size = (sector_t)nullb->dev->size * 1024 * 1024ULL;
 	sector_t blksize;
 	struct nvm_id_group *grp;
 
@@ -460,7 +1474,7 @@ static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id)
 	id->ppaf.ch_offset = 56;
 	id->ppaf.ch_len = 8;
 
-	sector_div(size, bs); /* convert size to pages */
+	sector_div(size, nullb->dev->blocksize); /* convert size to pages */
 	size >>= 8; /* concert size to pgs pr blk */
 	grp = &id->grp;
 	grp->mtype = 0;
@@ -474,8 +1488,8 @@ static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id)
 	grp->num_blk = blksize;
 	grp->num_pln = 1;
 
-	grp->fpg_sz = bs;
-	grp->csecs = bs;
+	grp->fpg_sz = nullb->dev->blocksize;
+	grp->csecs = nullb->dev->blocksize;
 	grp->trdt = 25000;
 	grp->trdm = 25000;
 	grp->tprt = 500000;
@@ -483,7 +1497,7 @@ static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id)
 	grp->tbet = 1500000;
 	grp->tbem = 1500000;
 	grp->mpos = 0x010101; /* single plane rwe */
-	grp->cpar = hw_queue_depth;
+	grp->cpar = nullb->dev->hw_queue_depth;
 
 	return 0;
 }
@@ -568,19 +1582,44 @@ static void null_nvm_unregister(struct nullb *nullb) {}
 
 static void null_del_dev(struct nullb *nullb)
 {
+	struct nullb_device *dev = nullb->dev;
+
+	ida_simple_remove(&nullb_indexes, nullb->index);
+
 	list_del_init(&nullb->list);
 
-	if (use_lightnvm)
+	if (dev->use_lightnvm)
 		null_nvm_unregister(nullb);
 	else
 		del_gendisk(nullb->disk);
+
+	if (test_bit(NULLB_DEV_FL_THROTTLED, &nullb->dev->flags)) {
+		hrtimer_cancel(&nullb->bw_timer);
+		atomic_long_set(&nullb->cur_bytes, LONG_MAX);
+		null_restart_queue_async(nullb);
+	}
+
 	blk_cleanup_queue(nullb->q);
-	if (queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set)
+	if (dev->queue_mode == NULL_Q_MQ &&
+	    nullb->tag_set == &nullb->__tag_set)
 		blk_mq_free_tag_set(nullb->tag_set);
-	if (!use_lightnvm)
+	if (!dev->use_lightnvm)
 		put_disk(nullb->disk);
 	cleanup_queues(nullb);
+	if (null_cache_active(nullb))
+		null_free_device_storage(nullb->dev, true);
 	kfree(nullb);
+	dev->nullb = NULL;
+}
+
+static void null_config_discard(struct nullb *nullb)
+{
+	if (nullb->dev->discard == false)
+		return;
+	nullb->q->limits.discard_granularity = nullb->dev->blocksize;
+	nullb->q->limits.discard_alignment = nullb->dev->blocksize;
+	blk_queue_max_discard_sectors(nullb->q, UINT_MAX >> 9);
+	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, nullb->q);
 }
 
 static int null_open(struct block_device *bdev, fmode_t mode)
@@ -605,6 +1644,7 @@ static void null_init_queue(struct nullb *nullb, struct nullb_queue *nq)
 
 	init_waitqueue_head(&nq->wait);
 	nq->queue_depth = nullb->queue_depth;
+	nq->dev = nullb->dev;
 }
 
 static void null_init_queues(struct nullb *nullb)
@@ -652,13 +1692,13 @@ static int setup_commands(struct nullb_queue *nq)
 
 static int setup_queues(struct nullb *nullb)
 {
-	nullb->queues = kzalloc(submit_queues * sizeof(struct nullb_queue),
-								GFP_KERNEL);
+	nullb->queues = kzalloc(nullb->dev->submit_queues *
+		sizeof(struct nullb_queue), GFP_KERNEL);
 	if (!nullb->queues)
 		return -ENOMEM;
 
 	nullb->nr_queues = 0;
-	nullb->queue_depth = hw_queue_depth;
+	nullb->queue_depth = nullb->dev->hw_queue_depth;
 
 	return 0;
 }
@@ -668,7 +1708,7 @@ static int init_driver_queues(struct nullb *nullb)
 	struct nullb_queue *nq;
 	int i, ret = 0;
 
-	for (i = 0; i < submit_queues; i++) {
+	for (i = 0; i < nullb->dev->submit_queues; i++) {
 		nq = &nullb->queues[i];
 
 		null_init_queue(nullb, nq);
@@ -686,10 +1726,10 @@ static int null_gendisk_register(struct nullb *nullb)
 	struct gendisk *disk;
 	sector_t size;
 
-	disk = nullb->disk = alloc_disk_node(1, home_node);
+	disk = nullb->disk = alloc_disk_node(1, nullb->dev->home_node);
 	if (!disk)
 		return -ENOMEM;
-	size = gb * 1024 * 1024 * 1024ULL;
+	size = (sector_t)nullb->dev->size * 1024 * 1024ULL;
 	set_capacity(disk, size >> 9);
 
 	disk->flags |= GENHD_FL_EXT_DEVT | GENHD_FL_SUPPRESS_PARTITION_INFO;
@@ -704,49 +1744,86 @@ static int null_gendisk_register(struct nullb *nullb)
 	return 0;
 }
 
-static int null_init_tag_set(struct blk_mq_tag_set *set)
+static int null_init_tag_set(struct nullb *nullb, struct blk_mq_tag_set *set)
 {
 	set->ops = &null_mq_ops;
-	set->nr_hw_queues = submit_queues;
-	set->queue_depth = hw_queue_depth;
-	set->numa_node = home_node;
+	set->nr_hw_queues = nullb ? nullb->dev->submit_queues :
+						g_submit_queues;
+	set->queue_depth = nullb ? nullb->dev->hw_queue_depth :
+						g_hw_queue_depth;
+	set->numa_node = nullb ? nullb->dev->home_node : g_home_node;
 	set->cmd_size	= sizeof(struct nullb_cmd);
 	set->flags = BLK_MQ_F_SHOULD_MERGE;
 	set->driver_data = NULL;
 
-	if (blocking)
+	if ((nullb && nullb->dev->blocking) || g_blocking)
 		set->flags |= BLK_MQ_F_BLOCKING;
 
 	return blk_mq_alloc_tag_set(set);
 }
 
-static int null_add_dev(void)
+static void null_validate_conf(struct nullb_device *dev)
+{
+	dev->blocksize = round_down(dev->blocksize, 512);
+	dev->blocksize = clamp_t(unsigned int, dev->blocksize, 512, 4096);
+	if (dev->use_lightnvm && dev->blocksize != 4096)
+		dev->blocksize = 4096;
+
+	if (dev->use_lightnvm && dev->queue_mode != NULL_Q_MQ)
+		dev->queue_mode = NULL_Q_MQ;
+
+	if (dev->queue_mode == NULL_Q_MQ && dev->use_per_node_hctx) {
+		if (dev->submit_queues != nr_online_nodes)
+			dev->submit_queues = nr_online_nodes;
+	} else if (dev->submit_queues > nr_cpu_ids)
+		dev->submit_queues = nr_cpu_ids;
+	else if (dev->submit_queues == 0)
+		dev->submit_queues = 1;
+
+	dev->queue_mode = min_t(unsigned int, dev->queue_mode, NULL_Q_MQ);
+	dev->irqmode = min_t(unsigned int, dev->irqmode, NULL_IRQ_TIMER);
+
+	/* Do memory allocation, so set blocking */
+	if (dev->memory_backed)
+		dev->blocking = true;
+	else /* cache is meaningless */
+		dev->cache_size = 0;
+	dev->cache_size = min_t(unsigned long, ULONG_MAX / 1024 / 1024,
+						dev->cache_size);
+	dev->mbps = min_t(unsigned int, 1024 * 40, dev->mbps);
+	/* can not stop a queue */
+	if (dev->queue_mode == NULL_Q_BIO)
+		dev->mbps = 0;
+}
+
+static int null_add_dev(struct nullb_device *dev)
 {
 	struct nullb *nullb;
 	int rv;
 
-	nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, home_node);
+	null_validate_conf(dev);
+
+	nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, dev->home_node);
 	if (!nullb) {
 		rv = -ENOMEM;
 		goto out;
 	}
+	nullb->dev = dev;
+	dev->nullb = nullb;
 
 	spin_lock_init(&nullb->lock);
 
-	if (queue_mode == NULL_Q_MQ && use_per_node_hctx)
-		submit_queues = nr_online_nodes;
-
 	rv = setup_queues(nullb);
 	if (rv)
 		goto out_free_nullb;
 
-	if (queue_mode == NULL_Q_MQ) {
+	if (dev->queue_mode == NULL_Q_MQ) {
 		if (shared_tags) {
 			nullb->tag_set = &tag_set;
 			rv = 0;
 		} else {
 			nullb->tag_set = &nullb->__tag_set;
-			rv = null_init_tag_set(nullb->tag_set);
+			rv = null_init_tag_set(nullb, nullb->tag_set);
 		}
 
 		if (rv)
@@ -758,8 +1835,8 @@ static int null_add_dev(void)
 			goto out_cleanup_tags;
 		}
 		null_init_queues(nullb);
-	} else if (queue_mode == NULL_Q_BIO) {
-		nullb->q = blk_alloc_queue_node(GFP_KERNEL, home_node);
+	} else if (dev->queue_mode == NULL_Q_BIO) {
+		nullb->q = blk_alloc_queue_node(GFP_KERNEL, dev->home_node);
 		if (!nullb->q) {
 			rv = -ENOMEM;
 			goto out_cleanup_queues;
@@ -769,7 +1846,8 @@ static int null_add_dev(void)
 		if (rv)
 			goto out_cleanup_blk_queue;
 	} else {
-		nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock, home_node);
+		nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock,
+						dev->home_node);
 		if (!nullb->q) {
 			rv = -ENOMEM;
 			goto out_cleanup_queues;
@@ -781,20 +1859,34 @@ static int null_add_dev(void)
 			goto out_cleanup_blk_queue;
 	}
 
+	if (dev->mbps) {
+		set_bit(NULLB_DEV_FL_THROTTLED, &dev->flags);
+		nullb_setup_bwtimer(nullb);
+	}
+
+	if (dev->cache_size > 0) {
+		set_bit(NULLB_DEV_FL_CACHE, &nullb->dev->flags);
+		blk_queue_write_cache(nullb->q, true, true);
+		blk_queue_flush_queueable(nullb->q, true);
+	}
+
 	nullb->q->queuedata = nullb;
 	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, nullb->q);
 	queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, nullb->q);
 
 	mutex_lock(&lock);
-	nullb->index = nullb_indexes++;
+	nullb->index = ida_simple_get(&nullb_indexes, 0, 0, GFP_KERNEL);
+	dev->index = nullb->index;
 	mutex_unlock(&lock);
 
-	blk_queue_logical_block_size(nullb->q, bs);
-	blk_queue_physical_block_size(nullb->q, bs);
+	blk_queue_logical_block_size(nullb->q, dev->blocksize);
+	blk_queue_physical_block_size(nullb->q, dev->blocksize);
+
+	null_config_discard(nullb);
 
 	sprintf(nullb->disk_name, "nullb%d", nullb->index);
 
-	if (use_lightnvm)
+	if (dev->use_lightnvm)
 		rv = null_nvm_register(nullb);
 	else
 		rv = null_gendisk_register(nullb);
@@ -810,7 +1902,7 @@ static int null_add_dev(void)
 out_cleanup_blk_queue:
 	blk_cleanup_queue(nullb->q);
 out_cleanup_tags:
-	if (queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set)
+	if (dev->queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set)
 		blk_mq_free_tag_set(nullb->tag_set);
 out_cleanup_queues:
 	cleanup_queues(nullb);
@@ -825,51 +1917,63 @@ static int __init null_init(void)
 	int ret = 0;
 	unsigned int i;
 	struct nullb *nullb;
+	struct nullb_device *dev;
+
+	/* check for nullb_page.bitmap */
+	if (sizeof(unsigned long) * 8 - 2 < (PAGE_SIZE >> SECTOR_SHIFT))
+		return -EINVAL;
 
-	if (bs > PAGE_SIZE) {
+	if (g_bs > PAGE_SIZE) {
 		pr_warn("null_blk: invalid block size\n");
 		pr_warn("null_blk: defaults block size to %lu\n", PAGE_SIZE);
-		bs = PAGE_SIZE;
+		g_bs = PAGE_SIZE;
 	}
 
-	if (use_lightnvm && bs != 4096) {
+	if (g_use_lightnvm && g_bs != 4096) {
 		pr_warn("null_blk: LightNVM only supports 4k block size\n");
 		pr_warn("null_blk: defaults block size to 4k\n");
-		bs = 4096;
+		g_bs = 4096;
 	}
 
-	if (use_lightnvm && queue_mode != NULL_Q_MQ) {
+	if (g_use_lightnvm && g_queue_mode != NULL_Q_MQ) {
 		pr_warn("null_blk: LightNVM only supported for blk-mq\n");
 		pr_warn("null_blk: defaults queue mode to blk-mq\n");
-		queue_mode = NULL_Q_MQ;
+		g_queue_mode = NULL_Q_MQ;
 	}
 
-	if (queue_mode == NULL_Q_MQ && use_per_node_hctx) {
-		if (submit_queues < nr_online_nodes) {
-			pr_warn("null_blk: submit_queues param is set to %u.",
+	if (g_queue_mode == NULL_Q_MQ && g_use_per_node_hctx) {
+		if (g_submit_queues != nr_online_nodes) {
+			pr_warn("null_blk: submit_queues param is set to %u.\n",
 							nr_online_nodes);
-			submit_queues = nr_online_nodes;
+			g_submit_queues = nr_online_nodes;
 		}
-	} else if (submit_queues > nr_cpu_ids)
-		submit_queues = nr_cpu_ids;
-	else if (!submit_queues)
-		submit_queues = 1;
+	} else if (g_submit_queues > nr_cpu_ids)
+		g_submit_queues = nr_cpu_ids;
+	else if (g_submit_queues <= 0)
+		g_submit_queues = 1;
 
-	if (queue_mode == NULL_Q_MQ && shared_tags) {
-		ret = null_init_tag_set(&tag_set);
+	if (g_queue_mode == NULL_Q_MQ && shared_tags) {
+		ret = null_init_tag_set(NULL, &tag_set);
 		if (ret)
 			return ret;
 	}
 
+	config_group_init(&nullb_subsys.su_group);
+	mutex_init(&nullb_subsys.su_mutex);
+
+	ret = configfs_register_subsystem(&nullb_subsys);
+	if (ret)
+		goto err_tagset;
+
 	mutex_init(&lock);
 
 	null_major = register_blkdev(0, "nullb");
 	if (null_major < 0) {
 		ret = null_major;
-		goto err_tagset;
+		goto err_conf;
 	}
 
-	if (use_lightnvm) {
+	if (g_use_lightnvm) {
 		ppa_cache = kmem_cache_create("ppa_cache", 64 * sizeof(u64),
 								0, 0, NULL);
 		if (!ppa_cache) {
@@ -880,9 +1984,14 @@ static int __init null_init(void)
 	}
 
 	for (i = 0; i < nr_devices; i++) {
-		ret = null_add_dev();
-		if (ret)
+		dev = null_alloc_dev();
+		if (!dev)
+			goto err_dev;
+		ret = null_add_dev(dev);
+		if (ret) {
+			null_free_dev(dev);
 			goto err_dev;
+		}
 	}
 
 	pr_info("null: module loaded\n");
@@ -891,13 +2000,17 @@ static int __init null_init(void)
 err_dev:
 	while (!list_empty(&nullb_list)) {
 		nullb = list_entry(nullb_list.next, struct nullb, list);
+		dev = nullb->dev;
 		null_del_dev(nullb);
+		null_free_dev(dev);
 	}
 	kmem_cache_destroy(ppa_cache);
 err_ppa:
 	unregister_blkdev(null_major, "nullb");
+err_conf:
+	configfs_unregister_subsystem(&nullb_subsys);
 err_tagset:
-	if (queue_mode == NULL_Q_MQ && shared_tags)
+	if (g_queue_mode == NULL_Q_MQ && shared_tags)
 		blk_mq_free_tag_set(&tag_set);
 	return ret;
 }
@@ -906,16 +2019,22 @@ static void __exit null_exit(void)
 {
 	struct nullb *nullb;
 
+	configfs_unregister_subsystem(&nullb_subsys);
+
 	unregister_blkdev(null_major, "nullb");
 
 	mutex_lock(&lock);
 	while (!list_empty(&nullb_list)) {
+		struct nullb_device *dev;
+
 		nullb = list_entry(nullb_list.next, struct nullb, list);
+		dev = nullb->dev;
 		null_del_dev(nullb);
+		null_free_dev(dev);
 	}
 	mutex_unlock(&lock);
 
-	if (queue_mode == NULL_Q_MQ && shared_tags)
+	if (g_queue_mode == NULL_Q_MQ && shared_tags)
 		blk_mq_free_tag_set(&tag_set);
 
 	kmem_cache_destroy(ppa_cache);
@@ -924,5 +2043,5 @@ static void __exit null_exit(void)
 module_init(null_init);
 module_exit(null_exit);
 
-MODULE_AUTHOR("Jens Axboe <jaxboe@fusionio.com>");
+MODULE_AUTHOR("Jens Axboe <axboe@kernel.dk>");
 MODULE_LICENSE("GPL");

drivers/block/pktcdvd.c

@@ -1028,7 +1028,7 @@ static void pkt_gather_data(struct pktcdvd_device *pd, struct packet_data *pkt)
 		bio = pkt->r_bios[f];
 		bio_reset(bio);
 		bio->bi_iter.bi_sector = pkt->sector + f * (CD_FRAMESIZE >> 9);
-		bio->bi_bdev = pd->bdev;
+		bio_set_dev(bio, pd->bdev);
 		bio->bi_end_io = pkt_end_io_read;
 		bio->bi_private = pkt;
 
@@ -1122,7 +1122,7 @@ static int pkt_start_recovery(struct packet_data *pkt)
 	pkt->sector = new_sector;
 
 	bio_reset(pkt->bio);
-	pkt->bio->bi_bdev = pd->bdev;
+	bio_set_set(pkt->bio, pd->bdev);
 	bio_set_op_attrs(pkt->bio, REQ_OP_WRITE, 0);
 	pkt->bio->bi_iter.bi_sector = new_sector;
 	pkt->bio->bi_iter.bi_size = pkt->frames * CD_FRAMESIZE;
@@ -1267,7 +1267,7 @@ static void pkt_start_write(struct pktcdvd_device *pd, struct packet_data *pkt)
 
 	bio_reset(pkt->w_bio);
 	pkt->w_bio->bi_iter.bi_sector = pkt->sector;
-	pkt->w_bio->bi_bdev = pd->bdev;
+	bio_set_dev(pkt->w_bio, pd->bdev);
 	pkt->w_bio->bi_end_io = pkt_end_io_packet_write;
 	pkt->w_bio->bi_private = pkt;
 
@@ -2314,7 +2314,7 @@ static void pkt_make_request_read(struct pktcdvd_device *pd, struct bio *bio)
 
 	psd->pd = pd;
 	psd->bio = bio;
-	cloned_bio->bi_bdev = pd->bdev;
+	bio_set_dev(cloned_bio, pd->bdev);
 	cloned_bio->bi_private = psd;
 	cloned_bio->bi_end_io = pkt_end_io_read_cloned;
 	pd->stats.secs_r += bio_sectors(bio);
@@ -2415,8 +2415,7 @@ static blk_qc_t pkt_make_request(struct request_queue *q, struct bio *bio)
 
 	pd = q->queuedata;
 	if (!pd) {
-		pr_err("%s incorrect request queue\n",
-		       bdevname(bio->bi_bdev, b));
+		pr_err("%s incorrect request queue\n", bio_devname(bio, b));
 		goto end_io;
 	}
 

drivers/block/rsxx/dev.c

@@ -112,7 +112,7 @@ static const struct block_device_operations rsxx_fops = {
 
 static void disk_stats_start(struct rsxx_cardinfo *card, struct bio *bio)
 {
-	generic_start_io_acct(bio_data_dir(bio), bio_sectors(bio),
+	generic_start_io_acct(card->queue, bio_data_dir(bio), bio_sectors(bio),
 			     &card->gendisk->part0);
 }
 
@@ -120,8 +120,8 @@ static void disk_stats_complete(struct rsxx_cardinfo *card,
 				struct bio *bio,
 				unsigned long start_time)
 {
-	generic_end_io_acct(bio_data_dir(bio), &card->gendisk->part0,
-			   start_time);
+	generic_end_io_acct(card->queue, bio_data_dir(bio),
+				&card->gendisk->part0, start_time);
 }
 
 static void bio_dma_done_cb(struct rsxx_cardinfo *card,

drivers/block/skd_main.c

@@ -1,19 +1,12 @@
-/* Copyright 2012 STEC, Inc.
+/*
+ * Driver for sTec s1120 PCIe SSDs. sTec was acquired in 2013 by HGST and HGST
+ * was acquired by Western Digital in 2012.
+ *
+ * Copyright 2012 sTec, Inc.
+ * Copyright (c) 2017 Western Digital Corporation or its affiliates.
  *
- * This file is licensed under the terms of the 3-clause
- * BSD License (http://opensource.org/licenses/BSD-3-Clause)
- * or the GNU GPL-2.0 (http://www.gnu.org/licenses/gpl-2.0.html),
- * at your option. Both licenses are also available in the LICENSE file
- * distributed with this project. This file may not be copied, modified,
- * or distributed except in accordance with those terms.
- * Gordoni Waidhofer <gwaidhofer@stec-inc.com>
- * Initial Driver Design!
- * Thomas Swann <tswann@stec-inc.com>
- * Interrupt handling.
- * Ramprasad Chinthekindi <rchinthekindi@stec-inc.com>
- * biomode implementation.
- * Akhil Bhansali <abhansali@stec-inc.com>
- * Added support for DISCARD / FLUSH and FUA.
+ * This file is part of the Linux kernel, and is made available under
+ * the terms of the GNU General Public License version 2.
  */
 
 #include <linux/kernel.h>
@@ -23,11 +16,11 @@
 #include <linux/slab.h>
 #include <linux/spinlock.h>
 #include <linux/blkdev.h>
+#include <linux/blk-mq.h>
 #include <linux/sched.h>
 #include <linux/interrupt.h>
 #include <linux/compiler.h>
 #include <linux/workqueue.h>
-#include <linux/bitops.h>
 #include <linux/delay.h>
 #include <linux/time.h>
 #include <linux/hdreg.h>
@@ -37,9 +30,9 @@
 #include <linux/version.h>
 #include <linux/err.h>
 #include <linux/aer.h>
-#include <linux/ctype.h>
 #include <linux/wait.h>
-#include <linux/uio.h>
+#include <linux/stringify.h>
+#include <linux/slab_def.h>
 #include <scsi/scsi.h>
 #include <scsi/sg.h>
 #include <linux/io.h>
@@ -51,19 +44,6 @@
 static int skd_dbg_level;
 static int skd_isr_comp_limit = 4;
 
-enum {
-	STEC_LINK_2_5GTS = 0,
-	STEC_LINK_5GTS = 1,
-	STEC_LINK_8GTS = 2,
-	STEC_LINK_UNKNOWN = 0xFF
-};
-
-enum {
-	SKD_FLUSH_INITIALIZER,
-	SKD_FLUSH_ZERO_SIZE_FIRST,
-	SKD_FLUSH_DATA_SECOND,
-};
-
 #define SKD_ASSERT(expr) \
 	do { \
 		if (unlikely(!(expr))) { \
@@ -73,17 +53,11 @@ enum {
 	} while (0)
 
 #define DRV_NAME "skd"
-#define DRV_VERSION "2.2.1"
-#define DRV_BUILD_ID "0260"
 #define PFX DRV_NAME ": "
-#define DRV_BIN_VERSION 0x100
-#define DRV_VER_COMPL   "2.2.1." DRV_BUILD_ID
 
-MODULE_AUTHOR("bug-reports: support@stec-inc.com");
-MODULE_LICENSE("Dual BSD/GPL");
+MODULE_LICENSE("GPL");
 
-MODULE_DESCRIPTION("STEC s1120 PCIe SSD block driver (b" DRV_BUILD_ID ")");
-MODULE_VERSION(DRV_VERSION "-" DRV_BUILD_ID);
+MODULE_DESCRIPTION("STEC s1120 PCIe SSD block driver");
 
 #define PCI_VENDOR_ID_STEC      0x1B39
 #define PCI_DEVICE_ID_S1120     0x0001
@@ -96,34 +70,32 @@ MODULE_VERSION(DRV_VERSION "-" DRV_BUILD_ID);
 #define SKD_PAUSE_TIMEOUT       (5 * 1000)
 
 #define SKD_N_FITMSG_BYTES      (512u)
+#define SKD_MAX_REQ_PER_MSG	14
 
-#define SKD_N_SPECIAL_CONTEXT   32u
 #define SKD_N_SPECIAL_FITMSG_BYTES      (128u)
 
 /* SG elements are 32 bytes, so we can make this 4096 and still be under the
  * 128KB limit.  That allows 4096*4K = 16M xfer size
  */
 #define SKD_N_SG_PER_REQ_DEFAULT 256u
-#define SKD_N_SG_PER_SPECIAL    256u
 
 #define SKD_N_COMPLETION_ENTRY  256u
 #define SKD_N_READ_CAP_BYTES    (8u)
 
 #define SKD_N_INTERNAL_BYTES    (512u)
 
+#define SKD_SKCOMP_SIZE							\
+	((sizeof(struct fit_completion_entry_v1) +			\
+	  sizeof(struct fit_comp_error_info)) * SKD_N_COMPLETION_ENTRY)
+
 /* 5 bits of uniqifier, 0xF800 */
-#define SKD_ID_INCR             (0x400)
 #define SKD_ID_TABLE_MASK       (3u << 8u)
 #define  SKD_ID_RW_REQUEST      (0u << 8u)
 #define  SKD_ID_INTERNAL        (1u << 8u)
-#define  SKD_ID_SPECIAL_REQUEST (2u << 8u)
 #define  SKD_ID_FIT_MSG         (3u << 8u)
 #define SKD_ID_SLOT_MASK        0x00FFu
 #define SKD_ID_SLOT_AND_TABLE_MASK 0x03FFu
 
-#define SKD_N_TIMEOUT_SLOT      4u
-#define SKD_TIMEOUT_SLOT_MASK   3u
-
 #define SKD_N_MAX_SECTORS 2048u
 
 #define SKD_MAX_RETRIES 2u
@@ -141,7 +113,6 @@ enum skd_drvr_state {
 	SKD_DRVR_STATE_ONLINE,
 	SKD_DRVR_STATE_PAUSING,
 	SKD_DRVR_STATE_PAUSED,
-	SKD_DRVR_STATE_DRAINING_TIMEOUT,
 	SKD_DRVR_STATE_RESTARTING,
 	SKD_DRVR_STATE_RESUMING,
 	SKD_DRVR_STATE_STOPPING,
@@ -158,7 +129,6 @@ enum skd_drvr_state {
 #define SKD_WAIT_BOOT_TIMO      SKD_TIMER_SECONDS(90u)
 #define SKD_STARTING_TIMO       SKD_TIMER_SECONDS(8u)
 #define SKD_RESTARTING_TIMO     SKD_TIMER_MINUTES(4u)
-#define SKD_DRAINING_TIMO       SKD_TIMER_SECONDS(6u)
 #define SKD_BUSY_TIMO           SKD_TIMER_MINUTES(20u)
 #define SKD_STARTED_BUSY_TIMO   SKD_TIMER_SECONDS(60u)
 #define SKD_START_WAIT_SECONDS  90u
@@ -169,12 +139,6 @@ enum skd_req_state {
 	SKD_REQ_STATE_BUSY,
 	SKD_REQ_STATE_COMPLETED,
 	SKD_REQ_STATE_TIMEOUT,
-	SKD_REQ_STATE_ABORTED,
-};
-
-enum skd_fit_msg_state {
-	SKD_MSG_STATE_IDLE,
-	SKD_MSG_STATE_BUSY,
 };
 
 enum skd_check_status_action {
@@ -185,34 +149,29 @@ enum skd_check_status_action {
 	SKD_CHECK_STATUS_BUSY_IMMINENT,
 };
 
-struct skd_fitmsg_context {
-	enum skd_fit_msg_state state;
-
-	struct skd_fitmsg_context *next;
+struct skd_msg_buf {
+	struct fit_msg_hdr	fmh;
+	struct skd_scsi_request	scsi[SKD_MAX_REQ_PER_MSG];
+};
 
+struct skd_fitmsg_context {
 	u32 id;
-	u16 outstanding;
 
 	u32 length;
-	u32 offset;
 
-	u8 *msg_buf;
+	struct skd_msg_buf *msg_buf;
 	dma_addr_t mb_dma_address;
 };
 
 struct skd_request_context {
 	enum skd_req_state state;
 
-	struct skd_request_context *next;
-
 	u16 id;
 	u32 fitmsg_id;
 
-	struct request *req;
 	u8 flush_cmd;
 
-	u32 timeout_stamp;
-	u8 sg_data_dir;
+	enum dma_data_direction data_dir;
 	struct scatterlist *sg;
 	u32 n_sg;
 	u32 sg_byte_count;
@@ -224,38 +183,19 @@ struct skd_request_context {
 
 	struct fit_comp_error_info err_info;
 
+	blk_status_t status;
 };
-#define SKD_DATA_DIR_HOST_TO_CARD       1
-#define SKD_DATA_DIR_CARD_TO_HOST       2
 
 struct skd_special_context {
 	struct skd_request_context req;
 
-	u8 orphaned;
-
 	void *data_buf;
 	dma_addr_t db_dma_address;
 
-	u8 *msg_buf;
+	struct skd_msg_buf *msg_buf;
 	dma_addr_t mb_dma_address;
 };
 
-struct skd_sg_io {
-	fmode_t mode;
-	void __user *argp;
-
-	struct sg_io_hdr sg;
-
-	u8 cdb[16];
-
-	u32 dxfer_len;
-	u32 iovcnt;
-	struct sg_iovec *iov;
-	struct sg_iovec no_iov_iov;
-
-	struct skd_special_context *skspcl;
-};
-
 typedef enum skd_irq_type {
 	SKD_IRQ_LEGACY,
 	SKD_IRQ_MSI,
@@ -265,7 +205,7 @@ typedef enum skd_irq_type {
 #define SKD_MAX_BARS                    2
 
 struct skd_device {
-	volatile void __iomem *mem_map[SKD_MAX_BARS];
+	void __iomem *mem_map[SKD_MAX_BARS];
 	resource_size_t mem_phys[SKD_MAX_BARS];
 	u32 mem_size[SKD_MAX_BARS];
 
@@ -276,21 +216,20 @@ struct skd_device {
 
 	spinlock_t lock;
 	struct gendisk *disk;
+	struct blk_mq_tag_set tag_set;
 	struct request_queue *queue;
+	struct skd_fitmsg_context *skmsg;
 	struct device *class_dev;
 	int gendisk_on;
 	int sync_done;
 
-	atomic_t device_count;
 	u32 devno;
 	u32 major;
-	char name[32];
 	char isr_name[30];
 
 	enum skd_drvr_state state;
 	u32 drive_state;
 
-	u32 in_flight;
 	u32 cur_max_queue_depth;
 	u32 queue_low_water_mark;
 	u32 dev_max_queue_depth;
@@ -298,27 +237,20 @@ struct skd_device {
 	u32 num_fitmsg_context;
 	u32 num_req_context;
 
-	u32 timeout_slot[SKD_N_TIMEOUT_SLOT];
-	u32 timeout_stamp;
-	struct skd_fitmsg_context *skmsg_free_list;
 	struct skd_fitmsg_context *skmsg_table;
 
-	struct skd_request_context *skreq_free_list;
-	struct skd_request_context *skreq_table;
-
-	struct skd_special_context *skspcl_free_list;
-	struct skd_special_context *skspcl_table;
-
 	struct skd_special_context internal_skspcl;
 	u32 read_cap_blocksize;
 	u32 read_cap_last_lba;
 	int read_cap_is_valid;
 	int inquiry_is_valid;
 	u8 inq_serial_num[13];  /*12 chars plus null term */
-	u8 id_str[80];          /* holds a composite name (pci + sernum) */
 
 	u8 skcomp_cycle;
 	u32 skcomp_ix;
+	struct kmem_cache *msgbuf_cache;
+	struct kmem_cache *sglist_cache;
+	struct kmem_cache *databuf_cache;
 	struct fit_completion_entry_v1 *skcomp_table;
 	struct fit_comp_error_info *skerr_table;
 	dma_addr_t cq_dma_address;
@@ -329,7 +261,6 @@ struct skd_device {
 	u32 timer_countdown;
 	u32 timer_substate;
 
-	int n_special;
 	int sgs_per_request;
 	u32 last_mtd;
 
@@ -343,7 +274,7 @@ struct skd_device {
 
 	u32 timo_slot;
 
-
+	struct work_struct start_queue;
 	struct work_struct completion_worker;
 };
 
@@ -353,53 +284,32 @@ struct skd_device {
 
 static inline u32 skd_reg_read32(struct skd_device *skdev, u32 offset)
 {
-	u32 val;
-
-	if (likely(skdev->dbg_level < 2))
-		return readl(skdev->mem_map[1] + offset);
-	else {
-		barrier();
-		val = readl(skdev->mem_map[1] + offset);
-		barrier();
-		pr_debug("%s:%s:%d offset %x = %x\n",
-			 skdev->name, __func__, __LINE__, offset, val);
-		return val;
-	}
+	u32 val = readl(skdev->mem_map[1] + offset);
 
+	if (unlikely(skdev->dbg_level >= 2))
+		dev_dbg(&skdev->pdev->dev, "offset %x = %x\n", offset, val);
+	return val;
 }
 
 static inline void skd_reg_write32(struct skd_device *skdev, u32 val,
 				   u32 offset)
 {
-	if (likely(skdev->dbg_level < 2)) {
-		writel(val, skdev->mem_map[1] + offset);
-		barrier();
-	} else {
-		barrier();
-		writel(val, skdev->mem_map[1] + offset);
-		barrier();
-		pr_debug("%s:%s:%d offset %x = %x\n",
-			 skdev->name, __func__, __LINE__, offset, val);
-	}
+	writel(val, skdev->mem_map[1] + offset);
+	if (unlikely(skdev->dbg_level >= 2))
+		dev_dbg(&skdev->pdev->dev, "offset %x = %x\n", offset, val);
 }
 
 static inline void skd_reg_write64(struct skd_device *skdev, u64 val,
 				   u32 offset)
 {
-	if (likely(skdev->dbg_level < 2)) {
-		writeq(val, skdev->mem_map[1] + offset);
-		barrier();
-	} else {
-		barrier();
-		writeq(val, skdev->mem_map[1] + offset);
-		barrier();
-		pr_debug("%s:%s:%d offset %x = %016llx\n",
-			 skdev->name, __func__, __LINE__, offset, val);
-	}
+	writeq(val, skdev->mem_map[1] + offset);
+	if (unlikely(skdev->dbg_level >= 2))
+		dev_dbg(&skdev->pdev->dev, "offset %x = %016llx\n", offset,
+			val);
 }
 
 
-#define SKD_IRQ_DEFAULT SKD_IRQ_MSI
+#define SKD_IRQ_DEFAULT SKD_IRQ_MSIX
 static int skd_isr_type = SKD_IRQ_DEFAULT;
 
 module_param(skd_isr_type, int, 0444);
@@ -412,7 +322,7 @@ static int skd_max_req_per_msg = SKD_MAX_REQ_PER_MSG_DEFAULT;
 module_param(skd_max_req_per_msg, int, 0444);
 MODULE_PARM_DESC(skd_max_req_per_msg,
 		 "Maximum SCSI requests packed in a single message."
-		 " (1-14, default==1)");
+		 " (1-" __stringify(SKD_MAX_REQ_PER_MSG) ", default==1)");
 
 #define SKD_MAX_QUEUE_DEPTH_DEFAULT 64
 #define SKD_MAX_QUEUE_DEPTH_DEFAULT_STR "64"
@@ -429,10 +339,10 @@ MODULE_PARM_DESC(skd_sgs_per_request,
 		 "Maximum SG elements per block request."
 		 " (1-4096, default==256)");
 
-static int skd_max_pass_thru = SKD_N_SPECIAL_CONTEXT;
+static int skd_max_pass_thru = 1;
 module_param(skd_max_pass_thru, int, 0444);
 MODULE_PARM_DESC(skd_max_pass_thru,
-		 "Maximum SCSI pass-thru at a time." " (1-50, default==32)");
+		 "Maximum SCSI pass-thru at a time. IGNORED");
 
 module_param(skd_dbg_level, int, 0444);
 MODULE_PARM_DESC(skd_dbg_level, "s1120 debug level (0,1,2)");
@@ -449,9 +359,6 @@ static void skd_send_fitmsg(struct skd_device *skdev,
 			    struct skd_fitmsg_context *skmsg);
 static void skd_send_special_fitmsg(struct skd_device *skdev,
 				    struct skd_special_context *skspcl);
-static void skd_request_fn(struct request_queue *rq);
-static void skd_end_request(struct skd_device *skdev,
-		struct skd_request_context *skreq, blk_status_t status);
 static bool skd_preop_sg_list(struct skd_device *skdev,
 			     struct skd_request_context *skreq);
 static void skd_postop_sg_list(struct skd_device *skdev,
@@ -460,19 +367,14 @@ static void skd_postop_sg_list(struct skd_device *skdev,
 static void skd_restart_device(struct skd_device *skdev);
 static int skd_quiesce_dev(struct skd_device *skdev);
 static int skd_unquiesce_dev(struct skd_device *skdev);
-static void skd_release_special(struct skd_device *skdev,
-				struct skd_special_context *skspcl);
 static void skd_disable_interrupts(struct skd_device *skdev);
 static void skd_isr_fwstate(struct skd_device *skdev);
-static void skd_recover_requests(struct skd_device *skdev, int requeue);
+static void skd_recover_requests(struct skd_device *skdev);
 static void skd_soft_reset(struct skd_device *skdev);
 
-static const char *skd_name(struct skd_device *skdev);
 const char *skd_drive_state_to_str(int state);
 const char *skd_skdev_state_to_str(enum skd_drvr_state state);
 static void skd_log_skdev(struct skd_device *skdev, const char *event);
-static void skd_log_skmsg(struct skd_device *skdev,
-			  struct skd_fitmsg_context *skmsg, const char *event);
 static void skd_log_skreq(struct skd_device *skdev,
 			  struct skd_request_context *skreq, const char *event);
 
@@ -481,18 +383,20 @@ static void skd_log_skreq(struct skd_device *skdev,
  * READ/WRITE REQUESTS
  *****************************************************************************
  */
-static void skd_fail_all_pending(struct skd_device *skdev)
+static void skd_inc_in_flight(struct request *rq, void *data, bool reserved)
 {
-	struct request_queue *q = skdev->queue;
-	struct request *req;
+	int *count = data;
 
-	for (;; ) {
-		req = blk_peek_request(q);
-		if (req == NULL)
-			break;
-		blk_start_request(req);
-		__blk_end_request_all(req, BLK_STS_IOERR);
-	}
+	count++;
+}
+
+static int skd_in_flight(struct skd_device *skdev)
+{
+	int count = 0;
+
+	blk_mq_tagset_busy_iter(&skdev->tag_set, skd_inc_in_flight, &count);
+
+	return count;
 }
 
 static void
@@ -501,9 +405,9 @@ skd_prep_rw_cdb(struct skd_scsi_request *scsi_req,
 		unsigned count)
 {
 	if (data_dir == READ)
-		scsi_req->cdb[0] = 0x28;
+		scsi_req->cdb[0] = READ_10;
 	else
-		scsi_req->cdb[0] = 0x2a;
+		scsi_req->cdb[0] = WRITE_10;
 
 	scsi_req->cdb[1] = 0;
 	scsi_req->cdb[2] = (lba & 0xff000000) >> 24;
@@ -522,7 +426,7 @@ skd_prep_zerosize_flush_cdb(struct skd_scsi_request *scsi_req,
 {
 	skreq->flush_cmd = 1;
 
-	scsi_req->cdb[0] = 0x35;
+	scsi_req->cdb[0] = SYNCHRONIZE_CACHE;
 	scsi_req->cdb[1] = 0;
 	scsi_req->cdb[2] = 0;
 	scsi_req->cdb[3] = 0;
@@ -534,307 +438,194 @@ skd_prep_zerosize_flush_cdb(struct skd_scsi_request *scsi_req,
 	scsi_req->cdb[9] = 0;
 }
 
-static void skd_request_fn_not_online(struct request_queue *q);
-
-static void skd_request_fn(struct request_queue *q)
+/*
+ * Return true if and only if all pending requests should be failed.
+ */
+static bool skd_fail_all(struct request_queue *q)
 {
 	struct skd_device *skdev = q->queuedata;
-	struct skd_fitmsg_context *skmsg = NULL;
-	struct fit_msg_hdr *fmh = NULL;
-	struct skd_request_context *skreq;
-	struct request *req = NULL;
-	struct skd_scsi_request *scsi_req;
-	unsigned long io_flags;
-	u32 lba;
-	u32 count;
-	int data_dir;
-	u32 be_lba;
-	u32 be_count;
-	u64 be_dmaa;
-	u64 cmdctxt;
-	u32 timo_slot;
-	void *cmd_ptr;
-	int flush, fua;
 
-	if (skdev->state != SKD_DRVR_STATE_ONLINE) {
-		skd_request_fn_not_online(q);
-		return;
-	}
-
-	if (blk_queue_stopped(skdev->queue)) {
-		if (skdev->skmsg_free_list == NULL ||
-		    skdev->skreq_free_list == NULL ||
-		    skdev->in_flight >= skdev->queue_low_water_mark)
-			/* There is still some kind of shortage */
-			return;
-
-		queue_flag_clear(QUEUE_FLAG_STOPPED, skdev->queue);
-	}
+	SKD_ASSERT(skdev->state != SKD_DRVR_STATE_ONLINE);
 
-	/*
-	 * Stop conditions:
-	 *  - There are no more native requests
-	 *  - There are already the maximum number of requests in progress
-	 *  - There are no more skd_request_context entries
-	 *  - There are no more FIT msg buffers
+	skd_log_skdev(skdev, "req_not_online");
+	switch (skdev->state) {
+	case SKD_DRVR_STATE_PAUSING:
+	case SKD_DRVR_STATE_PAUSED:
+	case SKD_DRVR_STATE_STARTING:
+	case SKD_DRVR_STATE_RESTARTING:
+	case SKD_DRVR_STATE_WAIT_BOOT:
+	/* In case of starting, we haven't started the queue,
+	 * so we can't get here... but requests are
+	 * possibly hanging out waiting for us because we
+	 * reported the dev/skd0 already.  They'll wait
+	 * forever if connect doesn't complete.
+	 * What to do??? delay dev/skd0 ??
 	 */
-	for (;; ) {
-
-		flush = fua = 0;
-
-		req = blk_peek_request(q);
-
-		/* Are there any native requests to start? */
-		if (req == NULL)
-			break;
-
-		lba = (u32)blk_rq_pos(req);
-		count = blk_rq_sectors(req);
-		data_dir = rq_data_dir(req);
-		io_flags = req->cmd_flags;
-
-		if (req_op(req) == REQ_OP_FLUSH)
-			flush++;
-
-		if (io_flags & REQ_FUA)
-			fua++;
-
-		pr_debug("%s:%s:%d new req=%p lba=%u(0x%x) "
-			 "count=%u(0x%x) dir=%d\n",
-			 skdev->name, __func__, __LINE__,
-			 req, lba, lba, count, count, data_dir);
-
-		/* At this point we know there is a request */
+	case SKD_DRVR_STATE_BUSY:
+	case SKD_DRVR_STATE_BUSY_IMMINENT:
+	case SKD_DRVR_STATE_BUSY_ERASE:
+		return false;
 
-		/* Are too many requets already in progress? */
-		if (skdev->in_flight >= skdev->cur_max_queue_depth) {
-			pr_debug("%s:%s:%d qdepth %d, limit %d\n",
-				 skdev->name, __func__, __LINE__,
-				 skdev->in_flight, skdev->cur_max_queue_depth);
-			break;
-		}
+	case SKD_DRVR_STATE_BUSY_SANITIZE:
+	case SKD_DRVR_STATE_STOPPING:
+	case SKD_DRVR_STATE_SYNCING:
+	case SKD_DRVR_STATE_FAULT:
+	case SKD_DRVR_STATE_DISAPPEARED:
+	default:
+		return true;
+	}
+}
 
-		/* Is a skd_request_context available? */
-		skreq = skdev->skreq_free_list;
-		if (skreq == NULL) {
-			pr_debug("%s:%s:%d Out of req=%p\n",
-				 skdev->name, __func__, __LINE__, q);
-			break;
-		}
-		SKD_ASSERT(skreq->state == SKD_REQ_STATE_IDLE);
-		SKD_ASSERT((skreq->id & SKD_ID_INCR) == 0);
-
-		/* Now we check to see if we can get a fit msg */
-		if (skmsg == NULL) {
-			if (skdev->skmsg_free_list == NULL) {
-				pr_debug("%s:%s:%d Out of msg\n",
-					 skdev->name, __func__, __LINE__);
-				break;
-			}
-		}
+static blk_status_t skd_mq_queue_rq(struct blk_mq_hw_ctx *hctx,
+				    const struct blk_mq_queue_data *mqd)
+{
+	struct request *const req = mqd->rq;
+	struct request_queue *const q = req->q;
+	struct skd_device *skdev = q->queuedata;
+	struct skd_fitmsg_context *skmsg;
+	struct fit_msg_hdr *fmh;
+	const u32 tag = blk_mq_unique_tag(req);
+	struct skd_request_context *const skreq = blk_mq_rq_to_pdu(req);
+	struct skd_scsi_request *scsi_req;
+	unsigned long flags = 0;
+	const u32 lba = blk_rq_pos(req);
+	const u32 count = blk_rq_sectors(req);
+	const int data_dir = rq_data_dir(req);
 
-		skreq->flush_cmd = 0;
-		skreq->n_sg = 0;
-		skreq->sg_byte_count = 0;
+	if (unlikely(skdev->state != SKD_DRVR_STATE_ONLINE))
+		return skd_fail_all(q) ? BLK_STS_IOERR : BLK_STS_RESOURCE;
 
-		/*
-		 * OK to now dequeue request from q.
-		 *
-		 * At this point we are comitted to either start or reject
-		 * the native request. Note that skd_request_context is
-		 * available but is still at the head of the free list.
-		 */
-		blk_start_request(req);
-		skreq->req = req;
-		skreq->fitmsg_id = 0;
-
-		/* Either a FIT msg is in progress or we have to start one. */
-		if (skmsg == NULL) {
-			/* Are there any FIT msg buffers available? */
-			skmsg = skdev->skmsg_free_list;
-			if (skmsg == NULL) {
-				pr_debug("%s:%s:%d Out of msg skdev=%p\n",
-					 skdev->name, __func__, __LINE__,
-					 skdev);
-				break;
-			}
-			SKD_ASSERT(skmsg->state == SKD_MSG_STATE_IDLE);
-			SKD_ASSERT((skmsg->id & SKD_ID_INCR) == 0);
+	blk_mq_start_request(req);
 
-			skdev->skmsg_free_list = skmsg->next;
+	WARN_ONCE(tag >= skd_max_queue_depth, "%#x > %#x (nr_requests = %lu)\n",
+		  tag, skd_max_queue_depth, q->nr_requests);
 
-			skmsg->state = SKD_MSG_STATE_BUSY;
-			skmsg->id += SKD_ID_INCR;
+	SKD_ASSERT(skreq->state == SKD_REQ_STATE_IDLE);
 
-			/* Initialize the FIT msg header */
-			fmh = (struct fit_msg_hdr *)skmsg->msg_buf;
-			memset(fmh, 0, sizeof(*fmh));
-			fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT;
-			skmsg->length = sizeof(*fmh);
-		}
+	dev_dbg(&skdev->pdev->dev,
+		"new req=%p lba=%u(0x%x) count=%u(0x%x) dir=%d\n", req, lba,
+		lba, count, count, data_dir);
 
-		skreq->fitmsg_id = skmsg->id;
+	skreq->id = tag + SKD_ID_RW_REQUEST;
+	skreq->flush_cmd = 0;
+	skreq->n_sg = 0;
+	skreq->sg_byte_count = 0;
 
-		/*
-		 * Note that a FIT msg may have just been started
-		 * but contains no SoFIT requests yet.
-		 */
+	skreq->fitmsg_id = 0;
 
-		/*
-		 * Transcode the request, checking as we go. The outcome of
-		 * the transcoding is represented by the error variable.
-		 */
-		cmd_ptr = &skmsg->msg_buf[skmsg->length];
-		memset(cmd_ptr, 0, 32);
+	skreq->data_dir = data_dir == READ ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
 
-		be_lba = cpu_to_be32(lba);
-		be_count = cpu_to_be32(count);
-		be_dmaa = cpu_to_be64((u64)skreq->sksg_dma_address);
-		cmdctxt = skreq->id + SKD_ID_INCR;
+	if (req->bio && !skd_preop_sg_list(skdev, skreq)) {
+		dev_dbg(&skdev->pdev->dev, "error Out\n");
+		skreq->status = BLK_STS_RESOURCE;
+		blk_mq_complete_request(req);
+		return BLK_STS_OK;
+	}
 
-		scsi_req = cmd_ptr;
-		scsi_req->hdr.tag = cmdctxt;
-		scsi_req->hdr.sg_list_dma_address = be_dmaa;
+	dma_sync_single_for_device(&skdev->pdev->dev, skreq->sksg_dma_address,
+				   skreq->n_sg *
+				   sizeof(struct fit_sg_descriptor),
+				   DMA_TO_DEVICE);
 
-		if (data_dir == READ)
-			skreq->sg_data_dir = SKD_DATA_DIR_CARD_TO_HOST;
-		else
-			skreq->sg_data_dir = SKD_DATA_DIR_HOST_TO_CARD;
+	/* Either a FIT msg is in progress or we have to start one. */
+	if (skd_max_req_per_msg == 1) {
+		skmsg = NULL;
+	} else {
+		spin_lock_irqsave(&skdev->lock, flags);
+		skmsg = skdev->skmsg;
+	}
+	if (!skmsg) {
+		skmsg = &skdev->skmsg_table[tag];
+		skdev->skmsg = skmsg;
 
-		if (flush == SKD_FLUSH_ZERO_SIZE_FIRST) {
-			skd_prep_zerosize_flush_cdb(scsi_req, skreq);
-			SKD_ASSERT(skreq->flush_cmd == 1);
+		/* Initialize the FIT msg header */
+		fmh = &skmsg->msg_buf->fmh;
+		memset(fmh, 0, sizeof(*fmh));
+		fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT;
+		skmsg->length = sizeof(*fmh);
+	} else {
+		fmh = &skmsg->msg_buf->fmh;
+	}
 
-		} else {
-			skd_prep_rw_cdb(scsi_req, data_dir, lba, count);
-		}
+	skreq->fitmsg_id = skmsg->id;
 
-		if (fua)
-			scsi_req->cdb[1] |= SKD_FUA_NV;
+	scsi_req = &skmsg->msg_buf->scsi[fmh->num_protocol_cmds_coalesced];
+	memset(scsi_req, 0, sizeof(*scsi_req));
 
-		if (!req->bio)
-			goto skip_sg;
+	scsi_req->hdr.tag = skreq->id;
+	scsi_req->hdr.sg_list_dma_address =
+		cpu_to_be64(skreq->sksg_dma_address);
 
-		if (!skd_preop_sg_list(skdev, skreq)) {
-			/*
-			 * Complete the native request with error.
-			 * Note that the request context is still at the
-			 * head of the free list, and that the SoFIT request
-			 * was encoded into the FIT msg buffer but the FIT
-			 * msg length has not been updated. In short, the
-			 * only resource that has been allocated but might
-			 * not be used is that the FIT msg could be empty.
-			 */
-			pr_debug("%s:%s:%d error Out\n",
-				 skdev->name, __func__, __LINE__);
-			skd_end_request(skdev, skreq, BLK_STS_RESOURCE);
-			continue;
-		}
+	if (req_op(req) == REQ_OP_FLUSH) {
+		skd_prep_zerosize_flush_cdb(scsi_req, skreq);
+		SKD_ASSERT(skreq->flush_cmd == 1);
+	} else {
+		skd_prep_rw_cdb(scsi_req, data_dir, lba, count);
+	}
 
-skip_sg:
-		scsi_req->hdr.sg_list_len_bytes =
-			cpu_to_be32(skreq->sg_byte_count);
+	if (req->cmd_flags & REQ_FUA)
+		scsi_req->cdb[1] |= SKD_FUA_NV;
 
-		/* Complete resource allocations. */
-		skdev->skreq_free_list = skreq->next;
-		skreq->state = SKD_REQ_STATE_BUSY;
-		skreq->id += SKD_ID_INCR;
+	scsi_req->hdr.sg_list_len_bytes = cpu_to_be32(skreq->sg_byte_count);
 
-		skmsg->length += sizeof(struct skd_scsi_request);
-		fmh->num_protocol_cmds_coalesced++;
+	/* Complete resource allocations. */
+	skreq->state = SKD_REQ_STATE_BUSY;
 
-		/*
-		 * Update the active request counts.
-		 * Capture the timeout timestamp.
-		 */
-		skreq->timeout_stamp = skdev->timeout_stamp;
-		timo_slot = skreq->timeout_stamp & SKD_TIMEOUT_SLOT_MASK;
-		skdev->timeout_slot[timo_slot]++;
-		skdev->in_flight++;
-		pr_debug("%s:%s:%d req=0x%x busy=%d\n",
-			 skdev->name, __func__, __LINE__,
-			 skreq->id, skdev->in_flight);
+	skmsg->length += sizeof(struct skd_scsi_request);
+	fmh->num_protocol_cmds_coalesced++;
 
-		/*
-		 * If the FIT msg buffer is full send it.
-		 */
-		if (skmsg->length >= SKD_N_FITMSG_BYTES ||
-		    fmh->num_protocol_cmds_coalesced >= skd_max_req_per_msg) {
-			skd_send_fitmsg(skdev, skmsg);
-			skmsg = NULL;
-			fmh = NULL;
-		}
-	}
+	dev_dbg(&skdev->pdev->dev, "req=0x%x busy=%d\n", skreq->id,
+		skd_in_flight(skdev));
 
 	/*
-	 * Is a FIT msg in progress? If it is empty put the buffer back
-	 * on the free list. If it is non-empty send what we got.
-	 * This minimizes latency when there are fewer requests than
-	 * what fits in a FIT msg.
+	 * If the FIT msg buffer is full send it.
 	 */
-	if (skmsg != NULL) {
-		/* Bigger than just a FIT msg header? */
-		if (skmsg->length > sizeof(struct fit_msg_hdr)) {
-			pr_debug("%s:%s:%d sending msg=%p, len %d\n",
-				 skdev->name, __func__, __LINE__,
-				 skmsg, skmsg->length);
+	if (skd_max_req_per_msg == 1) {
+		skd_send_fitmsg(skdev, skmsg);
+	} else {
+		if (mqd->last ||
+		    fmh->num_protocol_cmds_coalesced >= skd_max_req_per_msg) {
 			skd_send_fitmsg(skdev, skmsg);
-		} else {
-			/*
-			 * The FIT msg is empty. It means we got started
-			 * on the msg, but the requests were rejected.
-			 */
-			skmsg->state = SKD_MSG_STATE_IDLE;
-			skmsg->id += SKD_ID_INCR;
-			skmsg->next = skdev->skmsg_free_list;
-			skdev->skmsg_free_list = skmsg;
+			skdev->skmsg = NULL;
 		}
-		skmsg = NULL;
-		fmh = NULL;
+		spin_unlock_irqrestore(&skdev->lock, flags);
 	}
 
-	/*
-	 * If req is non-NULL it means there is something to do but
-	 * we are out of a resource.
-	 */
-	if (req)
-		blk_stop_queue(skdev->queue);
+	return BLK_STS_OK;
 }
 
-static void skd_end_request(struct skd_device *skdev,
-		struct skd_request_context *skreq, blk_status_t error)
+static enum blk_eh_timer_return skd_timed_out(struct request *req,
+					      bool reserved)
 {
-	if (unlikely(error)) {
-		struct request *req = skreq->req;
-		char *cmd = (rq_data_dir(req) == READ) ? "read" : "write";
-		u32 lba = (u32)blk_rq_pos(req);
-		u32 count = blk_rq_sectors(req);
-
-		pr_err("(%s): Error cmd=%s sect=%u count=%u id=0x%x\n",
-		       skd_name(skdev), cmd, lba, count, skreq->id);
-	} else
-		pr_debug("%s:%s:%d id=0x%x error=%d\n",
-			 skdev->name, __func__, __LINE__, skreq->id, error);
+	struct skd_device *skdev = req->q->queuedata;
+
+	dev_err(&skdev->pdev->dev, "request with tag %#x timed out\n",
+		blk_mq_unique_tag(req));
+
+	return BLK_EH_RESET_TIMER;
+}
 
-	__blk_end_request_all(skreq->req, error);
+static void skd_complete_rq(struct request *req)
+{
+	struct skd_request_context *skreq = blk_mq_rq_to_pdu(req);
+
+	blk_mq_end_request(req, skreq->status);
 }
 
 static bool skd_preop_sg_list(struct skd_device *skdev,
 			     struct skd_request_context *skreq)
 {
-	struct request *req = skreq->req;
-	int writing = skreq->sg_data_dir == SKD_DATA_DIR_HOST_TO_CARD;
-	int pci_dir = writing ? PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE;
-	struct scatterlist *sg = &skreq->sg[0];
+	struct request *req = blk_mq_rq_from_pdu(skreq);
+	struct scatterlist *sgl = &skreq->sg[0], *sg;
 	int n_sg;
 	int i;
 
 	skreq->sg_byte_count = 0;
 
-	/* SKD_ASSERT(skreq->sg_data_dir == SKD_DATA_DIR_HOST_TO_CARD ||
-		   skreq->sg_data_dir == SKD_DATA_DIR_CARD_TO_HOST); */
+	WARN_ON_ONCE(skreq->data_dir != DMA_TO_DEVICE &&
+		     skreq->data_dir != DMA_FROM_DEVICE);
 
-	n_sg = blk_rq_map_sg(skdev->queue, req, sg);
+	n_sg = blk_rq_map_sg(skdev->queue, req, sgl);
 	if (n_sg <= 0)
 		return false;
 
@@ -842,7 +633,7 @@ static bool skd_preop_sg_list(struct skd_device *skdev,
 	 * Map scatterlist to PCI bus addresses.
 	 * Note PCI might change the number of entries.
 	 */
-	n_sg = pci_map_sg(skdev->pdev, sg, n_sg, pci_dir);
+	n_sg = pci_map_sg(skdev->pdev, sgl, n_sg, skreq->data_dir);
 	if (n_sg <= 0)
 		return false;
 
@@ -850,10 +641,10 @@ static bool skd_preop_sg_list(struct skd_device *skdev,
 
 	skreq->n_sg = n_sg;
 
-	for (i = 0; i < n_sg; i++) {
+	for_each_sg(sgl, sg, n_sg, i) {
 		struct fit_sg_descriptor *sgd = &skreq->sksg_list[i];
-		u32 cnt = sg_dma_len(&sg[i]);
-		uint64_t dma_addr = sg_dma_address(&sg[i]);
+		u32 cnt = sg_dma_len(sg);
+		uint64_t dma_addr = sg_dma_address(sg);
 
 		sgd->control = FIT_SGD_CONTROL_NOT_LAST;
 		sgd->byte_count = cnt;
@@ -866,16 +657,16 @@ static bool skd_preop_sg_list(struct skd_device *skdev,
 	skreq->sksg_list[n_sg - 1].control = FIT_SGD_CONTROL_LAST;
 
 	if (unlikely(skdev->dbg_level > 1)) {
-		pr_debug("%s:%s:%d skreq=%x sksg_list=%p sksg_dma=%llx\n",
-			 skdev->name, __func__, __LINE__,
-			 skreq->id, skreq->sksg_list, skreq->sksg_dma_address);
+		dev_dbg(&skdev->pdev->dev,
+			"skreq=%x sksg_list=%p sksg_dma=%llx\n",
+			skreq->id, skreq->sksg_list, skreq->sksg_dma_address);
 		for (i = 0; i < n_sg; i++) {
 			struct fit_sg_descriptor *sgd = &skreq->sksg_list[i];
-			pr_debug("%s:%s:%d   sg[%d] count=%u ctrl=0x%x "
-				 "addr=0x%llx next=0x%llx\n",
-				 skdev->name, __func__, __LINE__,
-				 i, sgd->byte_count, sgd->control,
-				 sgd->host_side_addr, sgd->next_desc_ptr);
+
+			dev_dbg(&skdev->pdev->dev,
+				"  sg[%d] count=%u ctrl=0x%x addr=0x%llx next=0x%llx\n",
+				i, sgd->byte_count, sgd->control,
+				sgd->host_side_addr, sgd->next_desc_ptr);
 		}
 	}
 
@@ -885,9 +676,6 @@ static bool skd_preop_sg_list(struct skd_device *skdev,
 static void skd_postop_sg_list(struct skd_device *skdev,
 			       struct skd_request_context *skreq)
 {
-	int writing = skreq->sg_data_dir == SKD_DATA_DIR_HOST_TO_CARD;
-	int pci_dir = writing ? PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE;
-
 	/*
 	 * restore the next ptr for next IO request so we
 	 * don't have to set it every time.
@@ -895,51 +683,7 @@ static void skd_postop_sg_list(struct skd_device *skdev,
 	skreq->sksg_list[skreq->n_sg - 1].next_desc_ptr =
 		skreq->sksg_dma_address +
 		((skreq->n_sg) * sizeof(struct fit_sg_descriptor));
-	pci_unmap_sg(skdev->pdev, &skreq->sg[0], skreq->n_sg, pci_dir);
-}
-
-static void skd_request_fn_not_online(struct request_queue *q)
-{
-	struct skd_device *skdev = q->queuedata;
-	int error;
-
-	SKD_ASSERT(skdev->state != SKD_DRVR_STATE_ONLINE);
-
-	skd_log_skdev(skdev, "req_not_online");
-	switch (skdev->state) {
-	case SKD_DRVR_STATE_PAUSING:
-	case SKD_DRVR_STATE_PAUSED:
-	case SKD_DRVR_STATE_STARTING:
-	case SKD_DRVR_STATE_RESTARTING:
-	case SKD_DRVR_STATE_WAIT_BOOT:
-	/* In case of starting, we haven't started the queue,
-	 * so we can't get here... but requests are
-	 * possibly hanging out waiting for us because we
-	 * reported the dev/skd0 already.  They'll wait
-	 * forever if connect doesn't complete.
-	 * What to do??? delay dev/skd0 ??
-	 */
-	case SKD_DRVR_STATE_BUSY:
-	case SKD_DRVR_STATE_BUSY_IMMINENT:
-	case SKD_DRVR_STATE_BUSY_ERASE:
-	case SKD_DRVR_STATE_DRAINING_TIMEOUT:
-		return;
-
-	case SKD_DRVR_STATE_BUSY_SANITIZE:
-	case SKD_DRVR_STATE_STOPPING:
-	case SKD_DRVR_STATE_SYNCING:
-	case SKD_DRVR_STATE_FAULT:
-	case SKD_DRVR_STATE_DISAPPEARED:
-	default:
-		error = -EIO;
-		break;
-	}
-
-	/* If we get here, terminate all pending block requeusts
-	 * with EIO and any scsi pass thru with appropriate sense
-	 */
-
-	skd_fail_all_pending(skdev);
+	pci_unmap_sg(skdev->pdev, &skreq->sg[0], skreq->n_sg, skreq->data_dir);
 }
 
 /*
@@ -950,12 +694,22 @@ static void skd_request_fn_not_online(struct request_queue *q)
 
 static void skd_timer_tick_not_online(struct skd_device *skdev);
 
+static void skd_start_queue(struct work_struct *work)
+{
+	struct skd_device *skdev = container_of(work, typeof(*skdev),
+						start_queue);
+
+	/*
+	 * Although it is safe to call blk_start_queue() from interrupt
+	 * context, blk_mq_start_hw_queues() must not be called from
+	 * interrupt context.
+	 */
+	blk_mq_start_hw_queues(skdev->queue);
+}
+
 static void skd_timer_tick(ulong arg)
 {
 	struct skd_device *skdev = (struct skd_device *)arg;
-
-	u32 timo_slot;
-	u32 overdue_timestamp;
 	unsigned long reqflags;
 	u32 state;
 
@@ -972,37 +726,9 @@ static void skd_timer_tick(ulong arg)
 	if (state != skdev->drive_state)
 		skd_isr_fwstate(skdev);
 
-	if (skdev->state != SKD_DRVR_STATE_ONLINE) {
+	if (skdev->state != SKD_DRVR_STATE_ONLINE)
 		skd_timer_tick_not_online(skdev);
-		goto timer_func_out;
-	}
-	skdev->timeout_stamp++;
-	timo_slot = skdev->timeout_stamp & SKD_TIMEOUT_SLOT_MASK;
-
-	/*
-	 * All requests that happened during the previous use of
-	 * this slot should be done by now. The previous use was
-	 * over 7 seconds ago.
-	 */
-	if (skdev->timeout_slot[timo_slot] == 0)
-		goto timer_func_out;
-
-	/* Something is overdue */
-	overdue_timestamp = skdev->timeout_stamp - SKD_N_TIMEOUT_SLOT;
-
-	pr_debug("%s:%s:%d found %d timeouts, draining busy=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 skdev->timeout_slot[timo_slot], skdev->in_flight);
-	pr_err("(%s): Overdue IOs (%d), busy %d\n",
-	       skd_name(skdev), skdev->timeout_slot[timo_slot],
-	       skdev->in_flight);
 
-	skdev->timer_countdown = SKD_DRAINING_TIMO;
-	skdev->state = SKD_DRVR_STATE_DRAINING_TIMEOUT;
-	skdev->timo_slot = timo_slot;
-	blk_stop_queue(skdev->queue);
-
-timer_func_out:
 	mod_timer(&skdev->timer, (jiffies + HZ));
 
 	spin_unlock_irqrestore(&skdev->lock, reqflags);
@@ -1015,9 +741,9 @@ static void skd_timer_tick_not_online(struct skd_device *skdev)
 	case SKD_DRVR_STATE_LOAD:
 		break;
 	case SKD_DRVR_STATE_BUSY_SANITIZE:
-		pr_debug("%s:%s:%d drive busy sanitize[%x], driver[%x]\n",
-			 skdev->name, __func__, __LINE__,
-			 skdev->drive_state, skdev->state);
+		dev_dbg(&skdev->pdev->dev,
+			"drive busy sanitize[%x], driver[%x]\n",
+			skdev->drive_state, skdev->state);
 		/* If we've been in sanitize for 3 seconds, we figure we're not
 		 * going to get anymore completions, so recover requests now
 		 */
@@ -1025,22 +751,21 @@ static void skd_timer_tick_not_online(struct skd_device *skdev)
 			skdev->timer_countdown--;
 			return;
 		}
-		skd_recover_requests(skdev, 0);
+		skd_recover_requests(skdev);
 		break;
 
 	case SKD_DRVR_STATE_BUSY:
 	case SKD_DRVR_STATE_BUSY_IMMINENT:
 	case SKD_DRVR_STATE_BUSY_ERASE:
-		pr_debug("%s:%s:%d busy[%x], countdown=%d\n",
-			 skdev->name, __func__, __LINE__,
-			 skdev->state, skdev->timer_countdown);
+		dev_dbg(&skdev->pdev->dev, "busy[%x], countdown=%d\n",
+			skdev->state, skdev->timer_countdown);
 		if (skdev->timer_countdown > 0) {
 			skdev->timer_countdown--;
 			return;
 		}
-		pr_debug("%s:%s:%d busy[%x], timedout=%d, restarting device.",
-			 skdev->name, __func__, __LINE__,
-			 skdev->state, skdev->timer_countdown);
+		dev_dbg(&skdev->pdev->dev,
+			"busy[%x], timedout=%d, restarting device.",
+			skdev->state, skdev->timer_countdown);
 		skd_restart_device(skdev);
 		break;
 
@@ -1054,12 +779,12 @@ static void skd_timer_tick_not_online(struct skd_device *skdev)
 		 * revcover at some point. */
 		skdev->state = SKD_DRVR_STATE_FAULT;
 
-		pr_err("(%s): DriveFault Connect Timeout (%x)\n",
-		       skd_name(skdev), skdev->drive_state);
+		dev_err(&skdev->pdev->dev, "DriveFault Connect Timeout (%x)\n",
+			skdev->drive_state);
 
 		/*start the queue so we can respond with error to requests */
 		/* wakeup anyone waiting for startup complete */
-		blk_start_queue(skdev->queue);
+		schedule_work(&skdev->start_queue);
 		skdev->gendisk_on = -1;
 		wake_up_interruptible(&skdev->waitq);
 		break;
@@ -1072,29 +797,6 @@ static void skd_timer_tick_not_online(struct skd_device *skdev)
 	case SKD_DRVR_STATE_PAUSED:
 		break;
 
-	case SKD_DRVR_STATE_DRAINING_TIMEOUT:
-		pr_debug("%s:%s:%d "
-			 "draining busy [%d] tick[%d] qdb[%d] tmls[%d]\n",
-			 skdev->name, __func__, __LINE__,
-			 skdev->timo_slot,
-			 skdev->timer_countdown,
-			 skdev->in_flight,
-			 skdev->timeout_slot[skdev->timo_slot]);
-		/* if the slot has cleared we can let the I/O continue */
-		if (skdev->timeout_slot[skdev->timo_slot] == 0) {
-			pr_debug("%s:%s:%d Slot drained, starting queue.\n",
-				 skdev->name, __func__, __LINE__);
-			skdev->state = SKD_DRVR_STATE_ONLINE;
-			blk_start_queue(skdev->queue);
-			return;
-		}
-		if (skdev->timer_countdown > 0) {
-			skdev->timer_countdown--;
-			return;
-		}
-		skd_restart_device(skdev);
-		break;
-
 	case SKD_DRVR_STATE_RESTARTING:
 		if (skdev->timer_countdown > 0) {
 			skdev->timer_countdown--;
@@ -1103,8 +805,9 @@ static void skd_timer_tick_not_online(struct skd_device *skdev)
 		/* For now, we fault the drive. Could attempt resets to
 		 * revcover at some point. */
 		skdev->state = SKD_DRVR_STATE_FAULT;
-		pr_err("(%s): DriveFault Reconnect Timeout (%x)\n",
-		       skd_name(skdev), skdev->drive_state);
+		dev_err(&skdev->pdev->dev,
+			"DriveFault Reconnect Timeout (%x)\n",
+			skdev->drive_state);
 
 		/*
 		 * Recovering does two things:
@@ -1124,18 +827,18 @@ static void skd_timer_tick_not_online(struct skd_device *skdev)
 			/* It never came out of soft reset. Try to
 			 * recover the requests and then let them
 			 * fail. This is to mitigate hung processes. */
-			skd_recover_requests(skdev, 0);
+			skd_recover_requests(skdev);
 		else {
-			pr_err("(%s): Disable BusMaster (%x)\n",
-			       skd_name(skdev), skdev->drive_state);
+			dev_err(&skdev->pdev->dev, "Disable BusMaster (%x)\n",
+				skdev->drive_state);
 			pci_disable_device(skdev->pdev);
 			skd_disable_interrupts(skdev);
-			skd_recover_requests(skdev, 0);
+			skd_recover_requests(skdev);
 		}
 
 		/*start the queue so we can respond with error to requests */
 		/* wakeup anyone waiting for startup complete */
-		blk_start_queue(skdev->queue);
+		schedule_work(&skdev->start_queue);
 		skdev->gendisk_on = -1;
 		wake_up_interruptible(&skdev->waitq);
 		break;
@@ -1154,13 +857,11 @@ static int skd_start_timer(struct skd_device *skdev)
 {
 	int rc;
 
-	init_timer(&skdev->timer);
 	setup_timer(&skdev->timer, skd_timer_tick, (ulong)skdev);
 
 	rc = mod_timer(&skdev->timer, (jiffies + HZ));
 	if (rc)
-		pr_err("%s: failed to start timer %d\n",
-		       __func__, rc);
+		dev_err(&skdev->pdev->dev, "failed to start timer %d\n", rc);
 	return rc;
 }
 
@@ -1171,664 +872,37 @@ static void skd_kill_timer(struct skd_device *skdev)
 
 /*
  *****************************************************************************
- * IOCTL
+ * INTERNAL REQUESTS -- generated by driver itself
  *****************************************************************************
  */
-static int skd_ioctl_sg_io(struct skd_device *skdev,
-			   fmode_t mode, void __user *argp);
-static int skd_sg_io_get_and_check_args(struct skd_device *skdev,
-					struct skd_sg_io *sksgio);
-static int skd_sg_io_obtain_skspcl(struct skd_device *skdev,
-				   struct skd_sg_io *sksgio);
-static int skd_sg_io_prep_buffering(struct skd_device *skdev,
-				    struct skd_sg_io *sksgio);
-static int skd_sg_io_copy_buffer(struct skd_device *skdev,
-				 struct skd_sg_io *sksgio, int dxfer_dir);
-static int skd_sg_io_send_fitmsg(struct skd_device *skdev,
-				 struct skd_sg_io *sksgio);
-static int skd_sg_io_await(struct skd_device *skdev, struct skd_sg_io *sksgio);
-static int skd_sg_io_release_skspcl(struct skd_device *skdev,
-				    struct skd_sg_io *sksgio);
-static int skd_sg_io_put_status(struct skd_device *skdev,
-				struct skd_sg_io *sksgio);
-
-static void skd_complete_special(struct skd_device *skdev,
-				 volatile struct fit_completion_entry_v1
-				 *skcomp,
-				 volatile struct fit_comp_error_info *skerr,
-				 struct skd_special_context *skspcl);
-
-static int skd_bdev_ioctl(struct block_device *bdev, fmode_t mode,
-			  uint cmd_in, ulong arg)
-{
-	static const int sg_version_num = 30527;
-	int rc = 0, timeout;
-	struct gendisk *disk = bdev->bd_disk;
-	struct skd_device *skdev = disk->private_data;
-	int __user *p = (int __user *)arg;
-
-	pr_debug("%s:%s:%d %s: CMD[%s] ioctl  mode 0x%x, cmd 0x%x arg %0lx\n",
-		 skdev->name, __func__, __LINE__,
-		 disk->disk_name, current->comm, mode, cmd_in, arg);
-
-	if (!capable(CAP_SYS_ADMIN))
-		return -EPERM;
-
-	switch (cmd_in) {
-	case SG_SET_TIMEOUT:
-		rc = get_user(timeout, p);
-		if (!rc)
-			disk->queue->sg_timeout = clock_t_to_jiffies(timeout);
-		break;
-	case SG_GET_TIMEOUT:
-		rc = jiffies_to_clock_t(disk->queue->sg_timeout);
-		break;
-	case SG_GET_VERSION_NUM:
-		rc = put_user(sg_version_num, p);
-		break;
-	case SG_IO:
-		rc = skd_ioctl_sg_io(skdev, mode, (void __user *)arg);
-		break;
 
-	default:
-		rc = -ENOTTY;
-		break;
-	}
+static int skd_format_internal_skspcl(struct skd_device *skdev)
+{
+	struct skd_special_context *skspcl = &skdev->internal_skspcl;
+	struct fit_sg_descriptor *sgd = &skspcl->req.sksg_list[0];
+	struct fit_msg_hdr *fmh;
+	uint64_t dma_address;
+	struct skd_scsi_request *scsi;
 
-	pr_debug("%s:%s:%d %s:  completion rc %d\n",
-		 skdev->name, __func__, __LINE__, disk->disk_name, rc);
-	return rc;
-}
+	fmh = &skspcl->msg_buf->fmh;
+	fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT;
+	fmh->num_protocol_cmds_coalesced = 1;
 
-static int skd_ioctl_sg_io(struct skd_device *skdev, fmode_t mode,
-			   void __user *argp)
-{
-	int rc;
-	struct skd_sg_io sksgio;
+	scsi = &skspcl->msg_buf->scsi[0];
+	memset(scsi, 0, sizeof(*scsi));
+	dma_address = skspcl->req.sksg_dma_address;
+	scsi->hdr.sg_list_dma_address = cpu_to_be64(dma_address);
+	skspcl->req.n_sg = 1;
+	sgd->control = FIT_SGD_CONTROL_LAST;
+	sgd->byte_count = 0;
+	sgd->host_side_addr = skspcl->db_dma_address;
+	sgd->dev_side_addr = 0;
+	sgd->next_desc_ptr = 0LL;
 
-	memset(&sksgio, 0, sizeof(sksgio));
-	sksgio.mode = mode;
-	sksgio.argp = argp;
-	sksgio.iov = &sksgio.no_iov_iov;
+	return 1;
+}
 
-	switch (skdev->state) {
-	case SKD_DRVR_STATE_ONLINE:
-	case SKD_DRVR_STATE_BUSY_IMMINENT:
-		break;
-
-	default:
-		pr_debug("%s:%s:%d drive not online\n",
-			 skdev->name, __func__, __LINE__);
-		rc = -ENXIO;
-		goto out;
-	}
-
-	rc = skd_sg_io_get_and_check_args(skdev, &sksgio);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_obtain_skspcl(skdev, &sksgio);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_prep_buffering(skdev, &sksgio);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_copy_buffer(skdev, &sksgio, SG_DXFER_TO_DEV);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_send_fitmsg(skdev, &sksgio);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_await(skdev, &sksgio);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_copy_buffer(skdev, &sksgio, SG_DXFER_FROM_DEV);
-	if (rc)
-		goto out;
-
-	rc = skd_sg_io_put_status(skdev, &sksgio);
-	if (rc)
-		goto out;
-
-	rc = 0;
-
-out:
-	skd_sg_io_release_skspcl(skdev, &sksgio);
-
-	if (sksgio.iov != NULL && sksgio.iov != &sksgio.no_iov_iov)
-		kfree(sksgio.iov);
-	return rc;
-}
-
-static int skd_sg_io_get_and_check_args(struct skd_device *skdev,
-					struct skd_sg_io *sksgio)
-{
-	struct sg_io_hdr *sgp = &sksgio->sg;
-	int i, acc;
-
-	if (!access_ok(VERIFY_WRITE, sksgio->argp, sizeof(sg_io_hdr_t))) {
-		pr_debug("%s:%s:%d access sg failed %p\n",
-			 skdev->name, __func__, __LINE__, sksgio->argp);
-		return -EFAULT;
-	}
-
-	if (__copy_from_user(sgp, sksgio->argp, sizeof(sg_io_hdr_t))) {
-		pr_debug("%s:%s:%d copy_from_user sg failed %p\n",
-			 skdev->name, __func__, __LINE__, sksgio->argp);
-		return -EFAULT;
-	}
-
-	if (sgp->interface_id != SG_INTERFACE_ID_ORIG) {
-		pr_debug("%s:%s:%d interface_id invalid 0x%x\n",
-			 skdev->name, __func__, __LINE__, sgp->interface_id);
-		return -EINVAL;
-	}
-
-	if (sgp->cmd_len > sizeof(sksgio->cdb)) {
-		pr_debug("%s:%s:%d cmd_len invalid %d\n",
-			 skdev->name, __func__, __LINE__, sgp->cmd_len);
-		return -EINVAL;
-	}
-
-	if (sgp->iovec_count > 256) {
-		pr_debug("%s:%s:%d iovec_count invalid %d\n",
-			 skdev->name, __func__, __LINE__, sgp->iovec_count);
-		return -EINVAL;
-	}
-
-	if (sgp->dxfer_len > (PAGE_SIZE * SKD_N_SG_PER_SPECIAL)) {
-		pr_debug("%s:%s:%d dxfer_len invalid %d\n",
-			 skdev->name, __func__, __LINE__, sgp->dxfer_len);
-		return -EINVAL;
-	}
-
-	switch (sgp->dxfer_direction) {
-	case SG_DXFER_NONE:
-		acc = -1;
-		break;
-
-	case SG_DXFER_TO_DEV:
-		acc = VERIFY_READ;
-		break;
-
-	case SG_DXFER_FROM_DEV:
-	case SG_DXFER_TO_FROM_DEV:
-		acc = VERIFY_WRITE;
-		break;
-
-	default:
-		pr_debug("%s:%s:%d dxfer_dir invalid %d\n",
-			 skdev->name, __func__, __LINE__, sgp->dxfer_direction);
-		return -EINVAL;
-	}
-
-	if (copy_from_user(sksgio->cdb, sgp->cmdp, sgp->cmd_len)) {
-		pr_debug("%s:%s:%d copy_from_user cmdp failed %p\n",
-			 skdev->name, __func__, __LINE__, sgp->cmdp);
-		return -EFAULT;
-	}
-
-	if (sgp->mx_sb_len != 0) {
-		if (!access_ok(VERIFY_WRITE, sgp->sbp, sgp->mx_sb_len)) {
-			pr_debug("%s:%s:%d access sbp failed %p\n",
-				 skdev->name, __func__, __LINE__, sgp->sbp);
-			return -EFAULT;
-		}
-	}
-
-	if (sgp->iovec_count == 0) {
-		sksgio->iov[0].iov_base = sgp->dxferp;
-		sksgio->iov[0].iov_len = sgp->dxfer_len;
-		sksgio->iovcnt = 1;
-		sksgio->dxfer_len = sgp->dxfer_len;
-	} else {
-		struct sg_iovec *iov;
-		uint nbytes = sizeof(*iov) * sgp->iovec_count;
-		size_t iov_data_len;
-
-		iov = kmalloc(nbytes, GFP_KERNEL);
-		if (iov == NULL) {
-			pr_debug("%s:%s:%d alloc iovec failed %d\n",
-				 skdev->name, __func__, __LINE__,
-				 sgp->iovec_count);
-			return -ENOMEM;
-		}
-		sksgio->iov = iov;
-		sksgio->iovcnt = sgp->iovec_count;
-
-		if (copy_from_user(iov, sgp->dxferp, nbytes)) {
-			pr_debug("%s:%s:%d copy_from_user iovec failed %p\n",
-				 skdev->name, __func__, __LINE__, sgp->dxferp);
-			return -EFAULT;
-		}
-
-		/*
-		 * Sum up the vecs, making sure they don't overflow
-		 */
-		iov_data_len = 0;
-		for (i = 0; i < sgp->iovec_count; i++) {
-			if (iov_data_len + iov[i].iov_len < iov_data_len)
-				return -EINVAL;
-			iov_data_len += iov[i].iov_len;
-		}
-
-		/* SG_IO howto says that the shorter of the two wins */
-		if (sgp->dxfer_len < iov_data_len) {
-			sksgio->iovcnt = iov_shorten((struct iovec *)iov,
-						     sgp->iovec_count,
-						     sgp->dxfer_len);
-			sksgio->dxfer_len = sgp->dxfer_len;
-		} else
-			sksgio->dxfer_len = iov_data_len;
-	}
-
-	if (sgp->dxfer_direction != SG_DXFER_NONE) {
-		struct sg_iovec *iov = sksgio->iov;
-		for (i = 0; i < sksgio->iovcnt; i++, iov++) {
-			if (!access_ok(acc, iov->iov_base, iov->iov_len)) {
-				pr_debug("%s:%s:%d access data failed %p/%d\n",
-					 skdev->name, __func__, __LINE__,
-					 iov->iov_base, (int)iov->iov_len);
-				return -EFAULT;
-			}
-		}
-	}
-
-	return 0;
-}
-
-static int skd_sg_io_obtain_skspcl(struct skd_device *skdev,
-				   struct skd_sg_io *sksgio)
-{
-	struct skd_special_context *skspcl = NULL;
-	int rc;
-
-	for (;;) {
-		ulong flags;
-
-		spin_lock_irqsave(&skdev->lock, flags);
-		skspcl = skdev->skspcl_free_list;
-		if (skspcl != NULL) {
-			skdev->skspcl_free_list =
-				(struct skd_special_context *)skspcl->req.next;
-			skspcl->req.id += SKD_ID_INCR;
-			skspcl->req.state = SKD_REQ_STATE_SETUP;
-			skspcl->orphaned = 0;
-			skspcl->req.n_sg = 0;
-		}
-		spin_unlock_irqrestore(&skdev->lock, flags);
-
-		if (skspcl != NULL) {
-			rc = 0;
-			break;
-		}
-
-		pr_debug("%s:%s:%d blocking\n",
-			 skdev->name, __func__, __LINE__);
-
-		rc = wait_event_interruptible_timeout(
-				skdev->waitq,
-				(skdev->skspcl_free_list != NULL),
-				msecs_to_jiffies(sksgio->sg.timeout));
-
-		pr_debug("%s:%s:%d unblocking, rc=%d\n",
-			 skdev->name, __func__, __LINE__, rc);
-
-		if (rc <= 0) {
-			if (rc == 0)
-				rc = -ETIMEDOUT;
-			else
-				rc = -EINTR;
-			break;
-		}
-		/*
-		 * If we get here rc > 0 meaning the timeout to
-		 * wait_event_interruptible_timeout() had time left, hence the
-		 * sought event -- non-empty free list -- happened.
-		 * Retry the allocation.
-		 */
-	}
-	sksgio->skspcl = skspcl;
-
-	return rc;
-}
-
-static int skd_skreq_prep_buffering(struct skd_device *skdev,
-				    struct skd_request_context *skreq,
-				    u32 dxfer_len)
-{
-	u32 resid = dxfer_len;
-
-	/*
-	 * The DMA engine must have aligned addresses and byte counts.
-	 */
-	resid += (-resid) & 3;
-	skreq->sg_byte_count = resid;
-
-	skreq->n_sg = 0;
-
-	while (resid > 0) {
-		u32 nbytes = PAGE_SIZE;
-		u32 ix = skreq->n_sg;
-		struct scatterlist *sg = &skreq->sg[ix];
-		struct fit_sg_descriptor *sksg = &skreq->sksg_list[ix];
-		struct page *page;
-
-		if (nbytes > resid)
-			nbytes = resid;
-
-		page = alloc_page(GFP_KERNEL);
-		if (page == NULL)
-			return -ENOMEM;
-
-		sg_set_page(sg, page, nbytes, 0);
-
-		/* TODO: This should be going through a pci_???()
-		 * routine to do proper mapping. */
-		sksg->control = FIT_SGD_CONTROL_NOT_LAST;
-		sksg->byte_count = nbytes;
-
-		sksg->host_side_addr = sg_phys(sg);
-
-		sksg->dev_side_addr = 0;
-		sksg->next_desc_ptr = skreq->sksg_dma_address +
-				      (ix + 1) * sizeof(*sksg);
-
-		skreq->n_sg++;
-		resid -= nbytes;
-	}
-
-	if (skreq->n_sg > 0) {
-		u32 ix = skreq->n_sg - 1;
-		struct fit_sg_descriptor *sksg = &skreq->sksg_list[ix];
-
-		sksg->control = FIT_SGD_CONTROL_LAST;
-		sksg->next_desc_ptr = 0;
-	}
-
-	if (unlikely(skdev->dbg_level > 1)) {
-		u32 i;
-
-		pr_debug("%s:%s:%d skreq=%x sksg_list=%p sksg_dma=%llx\n",
-			 skdev->name, __func__, __LINE__,
-			 skreq->id, skreq->sksg_list, skreq->sksg_dma_address);
-		for (i = 0; i < skreq->n_sg; i++) {
-			struct fit_sg_descriptor *sgd = &skreq->sksg_list[i];
-
-			pr_debug("%s:%s:%d   sg[%d] count=%u ctrl=0x%x "
-				 "addr=0x%llx next=0x%llx\n",
-				 skdev->name, __func__, __LINE__,
-				 i, sgd->byte_count, sgd->control,
-				 sgd->host_side_addr, sgd->next_desc_ptr);
-		}
-	}
-
-	return 0;
-}
-
-static int skd_sg_io_prep_buffering(struct skd_device *skdev,
-				    struct skd_sg_io *sksgio)
-{
-	struct skd_special_context *skspcl = sksgio->skspcl;
-	struct skd_request_context *skreq = &skspcl->req;
-	u32 dxfer_len = sksgio->dxfer_len;
-	int rc;
-
-	rc = skd_skreq_prep_buffering(skdev, skreq, dxfer_len);
-	/*
-	 * Eventually, errors or not, skd_release_special() is called
-	 * to recover allocations including partial allocations.
-	 */
-	return rc;
-}
-
-static int skd_sg_io_copy_buffer(struct skd_device *skdev,
-				 struct skd_sg_io *sksgio, int dxfer_dir)
-{
-	struct skd_special_context *skspcl = sksgio->skspcl;
-	u32 iov_ix = 0;
-	struct sg_iovec curiov;
-	u32 sksg_ix = 0;
-	u8 *bufp = NULL;
-	u32 buf_len = 0;
-	u32 resid = sksgio->dxfer_len;
-	int rc;
-
-	curiov.iov_len = 0;
-	curiov.iov_base = NULL;
-
-	if (dxfer_dir != sksgio->sg.dxfer_direction) {
-		if (dxfer_dir != SG_DXFER_TO_DEV ||
-		    sksgio->sg.dxfer_direction != SG_DXFER_TO_FROM_DEV)
-			return 0;
-	}
-
-	while (resid > 0) {
-		u32 nbytes = PAGE_SIZE;
-
-		if (curiov.iov_len == 0) {
-			curiov = sksgio->iov[iov_ix++];
-			continue;
-		}
-
-		if (buf_len == 0) {
-			struct page *page;
-			page = sg_page(&skspcl->req.sg[sksg_ix++]);
-			bufp = page_address(page);
-			buf_len = PAGE_SIZE;
-		}
-
-		nbytes = min_t(u32, nbytes, resid);
-		nbytes = min_t(u32, nbytes, curiov.iov_len);
-		nbytes = min_t(u32, nbytes, buf_len);
-
-		if (dxfer_dir == SG_DXFER_TO_DEV)
-			rc = __copy_from_user(bufp, curiov.iov_base, nbytes);
-		else
-			rc = __copy_to_user(curiov.iov_base, bufp, nbytes);
-
-		if (rc)
-			return -EFAULT;
-
-		resid -= nbytes;
-		curiov.iov_len -= nbytes;
-		curiov.iov_base += nbytes;
-		buf_len -= nbytes;
-	}
-
-	return 0;
-}
-
-static int skd_sg_io_send_fitmsg(struct skd_device *skdev,
-				 struct skd_sg_io *sksgio)
-{
-	struct skd_special_context *skspcl = sksgio->skspcl;
-	struct fit_msg_hdr *fmh = (struct fit_msg_hdr *)skspcl->msg_buf;
-	struct skd_scsi_request *scsi_req = (struct skd_scsi_request *)&fmh[1];
-
-	memset(skspcl->msg_buf, 0, SKD_N_SPECIAL_FITMSG_BYTES);
-
-	/* Initialize the FIT msg header */
-	fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT;
-	fmh->num_protocol_cmds_coalesced = 1;
-
-	/* Initialize the SCSI request */
-	if (sksgio->sg.dxfer_direction != SG_DXFER_NONE)
-		scsi_req->hdr.sg_list_dma_address =
-			cpu_to_be64(skspcl->req.sksg_dma_address);
-	scsi_req->hdr.tag = skspcl->req.id;
-	scsi_req->hdr.sg_list_len_bytes =
-		cpu_to_be32(skspcl->req.sg_byte_count);
-	memcpy(scsi_req->cdb, sksgio->cdb, sizeof(scsi_req->cdb));
-
-	skspcl->req.state = SKD_REQ_STATE_BUSY;
-	skd_send_special_fitmsg(skdev, skspcl);
-
-	return 0;
-}
-
-static int skd_sg_io_await(struct skd_device *skdev, struct skd_sg_io *sksgio)
-{
-	unsigned long flags;
-	int rc;
-
-	rc = wait_event_interruptible_timeout(skdev->waitq,
-					      (sksgio->skspcl->req.state !=
-					       SKD_REQ_STATE_BUSY),
-					      msecs_to_jiffies(sksgio->sg.
-							       timeout));
-
-	spin_lock_irqsave(&skdev->lock, flags);
-
-	if (sksgio->skspcl->req.state == SKD_REQ_STATE_ABORTED) {
-		pr_debug("%s:%s:%d skspcl %p aborted\n",
-			 skdev->name, __func__, __LINE__, sksgio->skspcl);
-
-		/* Build check cond, sense and let command finish. */
-		/* For a timeout, we must fabricate completion and sense
-		 * data to complete the command */
-		sksgio->skspcl->req.completion.status =
-			SAM_STAT_CHECK_CONDITION;
-
-		memset(&sksgio->skspcl->req.err_info, 0,
-		       sizeof(sksgio->skspcl->req.err_info));
-		sksgio->skspcl->req.err_info.type = 0x70;
-		sksgio->skspcl->req.err_info.key = ABORTED_COMMAND;
-		sksgio->skspcl->req.err_info.code = 0x44;
-		sksgio->skspcl->req.err_info.qual = 0;
-		rc = 0;
-	} else if (sksgio->skspcl->req.state != SKD_REQ_STATE_BUSY)
-		/* No longer on the adapter. We finish. */
-		rc = 0;
-	else {
-		/* Something's gone wrong. Still busy. Timeout or
-		 * user interrupted (control-C). Mark as an orphan
-		 * so it will be disposed when completed. */
-		sksgio->skspcl->orphaned = 1;
-		sksgio->skspcl = NULL;
-		if (rc == 0) {
-			pr_debug("%s:%s:%d timed out %p (%u ms)\n",
-				 skdev->name, __func__, __LINE__,
-				 sksgio, sksgio->sg.timeout);
-			rc = -ETIMEDOUT;
-		} else {
-			pr_debug("%s:%s:%d cntlc %p\n",
-				 skdev->name, __func__, __LINE__, sksgio);
-			rc = -EINTR;
-		}
-	}
-
-	spin_unlock_irqrestore(&skdev->lock, flags);
-
-	return rc;
-}
-
-static int skd_sg_io_put_status(struct skd_device *skdev,
-				struct skd_sg_io *sksgio)
-{
-	struct sg_io_hdr *sgp = &sksgio->sg;
-	struct skd_special_context *skspcl = sksgio->skspcl;
-	int resid = 0;
-
-	u32 nb = be32_to_cpu(skspcl->req.completion.num_returned_bytes);
-
-	sgp->status = skspcl->req.completion.status;
-	resid = sksgio->dxfer_len - nb;
-
-	sgp->masked_status = sgp->status & STATUS_MASK;
-	sgp->msg_status = 0;
-	sgp->host_status = 0;
-	sgp->driver_status = 0;
-	sgp->resid = resid;
-	if (sgp->masked_status || sgp->host_status || sgp->driver_status)
-		sgp->info |= SG_INFO_CHECK;
-
-	pr_debug("%s:%s:%d status %x masked %x resid 0x%x\n",
-		 skdev->name, __func__, __LINE__,
-		 sgp->status, sgp->masked_status, sgp->resid);
-
-	if (sgp->masked_status == SAM_STAT_CHECK_CONDITION) {
-		if (sgp->mx_sb_len > 0) {
-			struct fit_comp_error_info *ei = &skspcl->req.err_info;
-			u32 nbytes = sizeof(*ei);
-
-			nbytes = min_t(u32, nbytes, sgp->mx_sb_len);
-
-			sgp->sb_len_wr = nbytes;
-
-			if (__copy_to_user(sgp->sbp, ei, nbytes)) {
-				pr_debug("%s:%s:%d copy_to_user sense failed %p\n",
-					 skdev->name, __func__, __LINE__,
-					 sgp->sbp);
-				return -EFAULT;
-			}
-		}
-	}
-
-	if (__copy_to_user(sksgio->argp, sgp, sizeof(sg_io_hdr_t))) {
-		pr_debug("%s:%s:%d copy_to_user sg failed %p\n",
-			 skdev->name, __func__, __LINE__, sksgio->argp);
-		return -EFAULT;
-	}
-
-	return 0;
-}
-
-static int skd_sg_io_release_skspcl(struct skd_device *skdev,
-				    struct skd_sg_io *sksgio)
-{
-	struct skd_special_context *skspcl = sksgio->skspcl;
-
-	if (skspcl != NULL) {
-		ulong flags;
-
-		sksgio->skspcl = NULL;
-
-		spin_lock_irqsave(&skdev->lock, flags);
-		skd_release_special(skdev, skspcl);
-		spin_unlock_irqrestore(&skdev->lock, flags);
-	}
-
-	return 0;
-}
-
-/*
- *****************************************************************************
- * INTERNAL REQUESTS -- generated by driver itself
- *****************************************************************************
- */
-
-static int skd_format_internal_skspcl(struct skd_device *skdev)
-{
-	struct skd_special_context *skspcl = &skdev->internal_skspcl;
-	struct fit_sg_descriptor *sgd = &skspcl->req.sksg_list[0];
-	struct fit_msg_hdr *fmh;
-	uint64_t dma_address;
-	struct skd_scsi_request *scsi;
-
-	fmh = (struct fit_msg_hdr *)&skspcl->msg_buf[0];
-	fmh->protocol_id = FIT_PROTOCOL_ID_SOFIT;
-	fmh->num_protocol_cmds_coalesced = 1;
-
-	scsi = (struct skd_scsi_request *)&skspcl->msg_buf[64];
-	memset(scsi, 0, sizeof(*scsi));
-	dma_address = skspcl->req.sksg_dma_address;
-	scsi->hdr.sg_list_dma_address = cpu_to_be64(dma_address);
-	sgd->control = FIT_SGD_CONTROL_LAST;
-	sgd->byte_count = 0;
-	sgd->host_side_addr = skspcl->db_dma_address;
-	sgd->dev_side_addr = 0;
-	sgd->next_desc_ptr = 0LL;
-
-	return 1;
-}
-
-#define WR_BUF_SIZE SKD_N_INTERNAL_BYTES
+#define WR_BUF_SIZE SKD_N_INTERNAL_BYTES
 
 static void skd_send_internal_skspcl(struct skd_device *skdev,
 				     struct skd_special_context *skspcl,
@@ -1846,11 +920,9 @@ static void skd_send_internal_skspcl(struct skd_device *skdev,
 		 */
 		return;
 
-	SKD_ASSERT((skspcl->req.id & SKD_ID_INCR) == 0);
 	skspcl->req.state = SKD_REQ_STATE_BUSY;
-	skspcl->req.id += SKD_ID_INCR;
 
-	scsi = (struct skd_scsi_request *)&skspcl->msg_buf[64];
+	scsi = &skspcl->msg_buf->scsi[0];
 	scsi->hdr.tag = skspcl->req.id;
 
 	memset(scsi->cdb, 0, sizeof(scsi->cdb));
@@ -1940,32 +1012,35 @@ static void skd_log_check_status(struct skd_device *skdev, u8 status, u8 key,
 	/* If the check condition is of special interest, log a message */
 	if ((status == SAM_STAT_CHECK_CONDITION) && (key == 0x02)
 	    && (code == 0x04) && (qual == 0x06)) {
-		pr_err("(%s): *** LOST_WRITE_DATA ERROR *** key/asc/"
-		       "ascq/fruc %02x/%02x/%02x/%02x\n",
-		       skd_name(skdev), key, code, qual, fruc);
+		dev_err(&skdev->pdev->dev,
+			"*** LOST_WRITE_DATA ERROR *** key/asc/ascq/fruc %02x/%02x/%02x/%02x\n",
+			key, code, qual, fruc);
 	}
 }
 
 static void skd_complete_internal(struct skd_device *skdev,
-				  volatile struct fit_completion_entry_v1
-				  *skcomp,
-				  volatile struct fit_comp_error_info *skerr,
+				  struct fit_completion_entry_v1 *skcomp,
+				  struct fit_comp_error_info *skerr,
 				  struct skd_special_context *skspcl)
 {
 	u8 *buf = skspcl->data_buf;
 	u8 status;
 	int i;
-	struct skd_scsi_request *scsi =
-		(struct skd_scsi_request *)&skspcl->msg_buf[64];
+	struct skd_scsi_request *scsi = &skspcl->msg_buf->scsi[0];
+
+	lockdep_assert_held(&skdev->lock);
 
 	SKD_ASSERT(skspcl == &skdev->internal_skspcl);
 
-	pr_debug("%s:%s:%d complete internal %x\n",
-		 skdev->name, __func__, __LINE__, scsi->cdb[0]);
+	dev_dbg(&skdev->pdev->dev, "complete internal %x\n", scsi->cdb[0]);
+
+	dma_sync_single_for_cpu(&skdev->pdev->dev,
+				skspcl->db_dma_address,
+				skspcl->req.sksg_list[0].byte_count,
+				DMA_BIDIRECTIONAL);
 
 	skspcl->req.completion = *skcomp;
 	skspcl->req.state = SKD_REQ_STATE_IDLE;
-	skspcl->req.id += SKD_ID_INCR;
 
 	status = skspcl->req.completion.status;
 
@@ -1981,14 +1056,15 @@ static void skd_complete_internal(struct skd_device *skdev,
 			skd_send_internal_skspcl(skdev, skspcl, WRITE_BUFFER);
 		else {
 			if (skdev->state == SKD_DRVR_STATE_STOPPING) {
-				pr_debug("%s:%s:%d TUR failed, don't send anymore state 0x%x\n",
-					 skdev->name, __func__, __LINE__,
-					 skdev->state);
+				dev_dbg(&skdev->pdev->dev,
+					"TUR failed, don't send anymore state 0x%x\n",
+					skdev->state);
 				return;
 			}
-			pr_debug("%s:%s:%d **** TUR failed, retry skerr\n",
-				 skdev->name, __func__, __LINE__);
-			skd_send_internal_skspcl(skdev, skspcl, 0x00);
+			dev_dbg(&skdev->pdev->dev,
+				"**** TUR failed, retry skerr\n");
+			skd_send_internal_skspcl(skdev, skspcl,
+						 TEST_UNIT_READY);
 		}
 		break;
 
@@ -1997,14 +1073,15 @@ static void skd_complete_internal(struct skd_device *skdev,
 			skd_send_internal_skspcl(skdev, skspcl, READ_BUFFER);
 		else {
 			if (skdev->state == SKD_DRVR_STATE_STOPPING) {
-				pr_debug("%s:%s:%d write buffer failed, don't send anymore state 0x%x\n",
-					 skdev->name, __func__, __LINE__,
-					 skdev->state);
+				dev_dbg(&skdev->pdev->dev,
+					"write buffer failed, don't send anymore state 0x%x\n",
+					skdev->state);
 				return;
 			}
-			pr_debug("%s:%s:%d **** write buffer failed, retry skerr\n",
-				 skdev->name, __func__, __LINE__);
-			skd_send_internal_skspcl(skdev, skspcl, 0x00);
+			dev_dbg(&skdev->pdev->dev,
+				"**** write buffer failed, retry skerr\n");
+			skd_send_internal_skspcl(skdev, skspcl,
+						 TEST_UNIT_READY);
 		}
 		break;
 
@@ -2014,33 +1091,31 @@ static void skd_complete_internal(struct skd_device *skdev,
 				skd_send_internal_skspcl(skdev, skspcl,
 							 READ_CAPACITY);
 			else {
-				pr_err(
-				       "(%s):*** W/R Buffer mismatch %d ***\n",
-				       skd_name(skdev), skdev->connect_retries);
+				dev_err(&skdev->pdev->dev,
+					"*** W/R Buffer mismatch %d ***\n",
+					skdev->connect_retries);
 				if (skdev->connect_retries <
 				    SKD_MAX_CONNECT_RETRIES) {
 					skdev->connect_retries++;
 					skd_soft_reset(skdev);
 				} else {
-					pr_err(
-					       "(%s): W/R Buffer Connect Error\n",
-					       skd_name(skdev));
+					dev_err(&skdev->pdev->dev,
+						"W/R Buffer Connect Error\n");
 					return;
 				}
 			}
 
 		} else {
 			if (skdev->state == SKD_DRVR_STATE_STOPPING) {
-				pr_debug("%s:%s:%d "
-					 "read buffer failed, don't send anymore state 0x%x\n",
-					 skdev->name, __func__, __LINE__,
-					 skdev->state);
+				dev_dbg(&skdev->pdev->dev,
+					"read buffer failed, don't send anymore state 0x%x\n",
+					skdev->state);
 				return;
 			}
-			pr_debug("%s:%s:%d "
-				 "**** read buffer failed, retry skerr\n",
-				 skdev->name, __func__, __LINE__);
-			skd_send_internal_skspcl(skdev, skspcl, 0x00);
+			dev_dbg(&skdev->pdev->dev,
+				"**** read buffer failed, retry skerr\n");
+			skd_send_internal_skspcl(skdev, skspcl,
+						 TEST_UNIT_READY);
 		}
 		break;
 
@@ -2054,10 +1129,9 @@ static void skd_complete_internal(struct skd_device *skdev,
 				(buf[4] << 24) | (buf[5] << 16) |
 				(buf[6] << 8) | buf[7];
 
-			pr_debug("%s:%s:%d last lba %d, bs %d\n",
-				 skdev->name, __func__, __LINE__,
-				 skdev->read_cap_last_lba,
-				 skdev->read_cap_blocksize);
+			dev_dbg(&skdev->pdev->dev, "last lba %d, bs %d\n",
+				skdev->read_cap_last_lba,
+				skdev->read_cap_blocksize);
 
 			set_capacity(skdev->disk, skdev->read_cap_last_lba + 1);
 
@@ -2068,13 +1142,10 @@ static void skd_complete_internal(struct skd_device *skdev,
 			   (skerr->key == MEDIUM_ERROR)) {
 			skdev->read_cap_last_lba = ~0;
 			set_capacity(skdev->disk, skdev->read_cap_last_lba + 1);
-			pr_debug("%s:%s:%d "
-				 "**** MEDIUM ERROR caused READCAP to fail, ignore failure and continue to inquiry\n",
-				 skdev->name, __func__, __LINE__);
+			dev_dbg(&skdev->pdev->dev, "**** MEDIUM ERROR caused READCAP to fail, ignore failure and continue to inquiry\n");
 			skd_send_internal_skspcl(skdev, skspcl, INQUIRY);
 		} else {
-			pr_debug("%s:%s:%d **** READCAP failed, retry TUR\n",
-				 skdev->name, __func__, __LINE__);
+			dev_dbg(&skdev->pdev->dev, "**** READCAP failed, retry TUR\n");
 			skd_send_internal_skspcl(skdev, skspcl,
 						 TEST_UNIT_READY);
 		}
@@ -2091,8 +1162,7 @@ static void skd_complete_internal(struct skd_device *skdev,
 		}
 
 		if (skd_unquiesce_dev(skdev) < 0)
-			pr_debug("%s:%s:%d **** failed, to ONLINE device\n",
-				 skdev->name, __func__, __LINE__);
+			dev_dbg(&skdev->pdev->dev, "**** failed, to ONLINE device\n");
 		 /* connection is complete */
 		skdev->connect_retries = 0;
 		break;
@@ -2120,27 +1190,20 @@ static void skd_send_fitmsg(struct skd_device *skdev,
 			    struct skd_fitmsg_context *skmsg)
 {
 	u64 qcmd;
-	struct fit_msg_hdr *fmh;
 
-	pr_debug("%s:%s:%d dma address 0x%llx, busy=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 skmsg->mb_dma_address, skdev->in_flight);
-	pr_debug("%s:%s:%d msg_buf 0x%p, offset %x\n",
-		 skdev->name, __func__, __LINE__,
-		 skmsg->msg_buf, skmsg->offset);
+	dev_dbg(&skdev->pdev->dev, "dma address 0x%llx, busy=%d\n",
+		skmsg->mb_dma_address, skd_in_flight(skdev));
+	dev_dbg(&skdev->pdev->dev, "msg_buf %p\n", skmsg->msg_buf);
 
 	qcmd = skmsg->mb_dma_address;
 	qcmd |= FIT_QCMD_QID_NORMAL;
 
-	fmh = (struct fit_msg_hdr *)skmsg->msg_buf;
-	skmsg->outstanding = fmh->num_protocol_cmds_coalesced;
-
 	if (unlikely(skdev->dbg_level > 1)) {
 		u8 *bp = (u8 *)skmsg->msg_buf;
 		int i;
 		for (i = 0; i < skmsg->length; i += 8) {
-			pr_debug("%s:%s:%d msg[%2d] %8ph\n",
-				 skdev->name, __func__, __LINE__, i, &bp[i]);
+			dev_dbg(&skdev->pdev->dev, "msg[%2d] %8ph\n", i,
+				&bp[i]);
 			if (i == 0)
 				i = 64 - 8;
 		}
@@ -2160,6 +1223,12 @@ static void skd_send_fitmsg(struct skd_device *skdev,
 		 */
 		qcmd |= FIT_QCMD_MSGSIZE_64;
 
+	dma_sync_single_for_device(&skdev->pdev->dev, skmsg->mb_dma_address,
+				   skmsg->length, DMA_TO_DEVICE);
+
+	/* Make sure skd_msg_buf is written before the doorbell is triggered. */
+	smp_wmb();
+
 	SKD_WRITEQ(skdev, qcmd, FIT_Q_COMMAND);
 }
 
@@ -2168,30 +1237,31 @@ static void skd_send_special_fitmsg(struct skd_device *skdev,
 {
 	u64 qcmd;
 
+	WARN_ON_ONCE(skspcl->req.n_sg != 1);
+
 	if (unlikely(skdev->dbg_level > 1)) {
 		u8 *bp = (u8 *)skspcl->msg_buf;
 		int i;
 
 		for (i = 0; i < SKD_N_SPECIAL_FITMSG_BYTES; i += 8) {
-			pr_debug("%s:%s:%d  spcl[%2d] %8ph\n",
-				 skdev->name, __func__, __LINE__, i, &bp[i]);
+			dev_dbg(&skdev->pdev->dev, " spcl[%2d] %8ph\n", i,
+				&bp[i]);
 			if (i == 0)
 				i = 64 - 8;
 		}
 
-		pr_debug("%s:%s:%d skspcl=%p id=%04x sksg_list=%p sksg_dma=%llx\n",
-			 skdev->name, __func__, __LINE__,
-			 skspcl, skspcl->req.id, skspcl->req.sksg_list,
-			 skspcl->req.sksg_dma_address);
+		dev_dbg(&skdev->pdev->dev,
+			"skspcl=%p id=%04x sksg_list=%p sksg_dma=%llx\n",
+			skspcl, skspcl->req.id, skspcl->req.sksg_list,
+			skspcl->req.sksg_dma_address);
 		for (i = 0; i < skspcl->req.n_sg; i++) {
 			struct fit_sg_descriptor *sgd =
 				&skspcl->req.sksg_list[i];
 
-			pr_debug("%s:%s:%d   sg[%d] count=%u ctrl=0x%x "
-				 "addr=0x%llx next=0x%llx\n",
-				 skdev->name, __func__, __LINE__,
-				 i, sgd->byte_count, sgd->control,
-				 sgd->host_side_addr, sgd->next_desc_ptr);
+			dev_dbg(&skdev->pdev->dev,
+				"  sg[%d] count=%u ctrl=0x%x addr=0x%llx next=0x%llx\n",
+				i, sgd->byte_count, sgd->control,
+				sgd->host_side_addr, sgd->next_desc_ptr);
 		}
 	}
 
@@ -2202,6 +1272,20 @@ static void skd_send_special_fitmsg(struct skd_device *skdev,
 	qcmd = skspcl->mb_dma_address;
 	qcmd |= FIT_QCMD_QID_NORMAL + FIT_QCMD_MSGSIZE_128;
 
+	dma_sync_single_for_device(&skdev->pdev->dev, skspcl->mb_dma_address,
+				   SKD_N_SPECIAL_FITMSG_BYTES, DMA_TO_DEVICE);
+	dma_sync_single_for_device(&skdev->pdev->dev,
+				   skspcl->req.sksg_dma_address,
+				   1 * sizeof(struct fit_sg_descriptor),
+				   DMA_TO_DEVICE);
+	dma_sync_single_for_device(&skdev->pdev->dev,
+				   skspcl->db_dma_address,
+				   skspcl->req.sksg_list[0].byte_count,
+				   DMA_BIDIRECTIONAL);
+
+	/* Make sure skd_msg_buf is written before the doorbell is triggered. */
+	smp_wmb();
+
 	SKD_WRITEQ(skdev, qcmd, FIT_Q_COMMAND);
 }
 
@@ -2212,8 +1296,8 @@ static void skd_send_special_fitmsg(struct skd_device *skdev,
  */
 
 static void skd_complete_other(struct skd_device *skdev,
-			       volatile struct fit_completion_entry_v1 *skcomp,
-			       volatile struct fit_comp_error_info *skerr);
+			       struct fit_completion_entry_v1 *skcomp,
+			       struct fit_comp_error_info *skerr);
 
 struct sns_info {
 	u8 type;
@@ -2262,21 +1346,20 @@ static struct sns_info skd_chkstat_table[] = {
 
 static enum skd_check_status_action
 skd_check_status(struct skd_device *skdev,
-		 u8 cmp_status, volatile struct fit_comp_error_info *skerr)
+		 u8 cmp_status, struct fit_comp_error_info *skerr)
 {
-	int i, n;
+	int i;
 
-	pr_err("(%s): key/asc/ascq/fruc %02x/%02x/%02x/%02x\n",
-	       skd_name(skdev), skerr->key, skerr->code, skerr->qual,
-	       skerr->fruc);
+	dev_err(&skdev->pdev->dev, "key/asc/ascq/fruc %02x/%02x/%02x/%02x\n",
+		skerr->key, skerr->code, skerr->qual, skerr->fruc);
 
-	pr_debug("%s:%s:%d stat: t=%02x stat=%02x k=%02x c=%02x q=%02x fruc=%02x\n",
-		 skdev->name, __func__, __LINE__, skerr->type, cmp_status,
-		 skerr->key, skerr->code, skerr->qual, skerr->fruc);
+	dev_dbg(&skdev->pdev->dev,
+		"stat: t=%02x stat=%02x k=%02x c=%02x q=%02x fruc=%02x\n",
+		skerr->type, cmp_status, skerr->key, skerr->code, skerr->qual,
+		skerr->fruc);
 
 	/* Does the info match an entry in the good category? */
-	n = sizeof(skd_chkstat_table) / sizeof(skd_chkstat_table[0]);
-	for (i = 0; i < n; i++) {
+	for (i = 0; i < ARRAY_SIZE(skd_chkstat_table); i++) {
 		struct sns_info *sns = &skd_chkstat_table[i];
 
 		if (sns->mask & 0x10)
@@ -2300,10 +1383,9 @@ skd_check_status(struct skd_device *skdev,
 				continue;
 
 		if (sns->action == SKD_CHECK_STATUS_REPORT_SMART_ALERT) {
-			pr_err("(%s): SMART Alert: sense key/asc/ascq "
-			       "%02x/%02x/%02x\n",
-			       skd_name(skdev), skerr->key,
-			       skerr->code, skerr->qual);
+			dev_err(&skdev->pdev->dev,
+				"SMART Alert: sense key/asc/ascq %02x/%02x/%02x\n",
+				skerr->key, skerr->code, skerr->qual);
 		}
 		return sns->action;
 	}
@@ -2312,335 +1394,80 @@ skd_check_status(struct skd_device *skdev,
 	 * zero status means good
 	 */
 	if (cmp_status) {
-		pr_debug("%s:%s:%d status check: error\n",
-			 skdev->name, __func__, __LINE__);
+		dev_dbg(&skdev->pdev->dev, "status check: error\n");
 		return SKD_CHECK_STATUS_REPORT_ERROR;
 	}
 
-	pr_debug("%s:%s:%d status check good default\n",
-		 skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "status check good default\n");
 	return SKD_CHECK_STATUS_REPORT_GOOD;
 }
 
 static void skd_resolve_req_exception(struct skd_device *skdev,
-				      struct skd_request_context *skreq)
+				      struct skd_request_context *skreq,
+				      struct request *req)
 {
 	u8 cmp_status = skreq->completion.status;
 
 	switch (skd_check_status(skdev, cmp_status, &skreq->err_info)) {
 	case SKD_CHECK_STATUS_REPORT_GOOD:
 	case SKD_CHECK_STATUS_REPORT_SMART_ALERT:
-		skd_end_request(skdev, skreq, BLK_STS_OK);
+		skreq->status = BLK_STS_OK;
+		blk_mq_complete_request(req);
 		break;
 
 	case SKD_CHECK_STATUS_BUSY_IMMINENT:
 		skd_log_skreq(skdev, skreq, "retry(busy)");
-		blk_requeue_request(skdev->queue, skreq->req);
-		pr_info("(%s) drive BUSY imminent\n", skd_name(skdev));
+		blk_requeue_request(skdev->queue, req);
+		dev_info(&skdev->pdev->dev, "drive BUSY imminent\n");
 		skdev->state = SKD_DRVR_STATE_BUSY_IMMINENT;
 		skdev->timer_countdown = SKD_TIMER_MINUTES(20);
 		skd_quiesce_dev(skdev);
 		break;
 
 	case SKD_CHECK_STATUS_REQUEUE_REQUEST:
-		if ((unsigned long) ++skreq->req->special < SKD_MAX_RETRIES) {
+		if ((unsigned long) ++req->special < SKD_MAX_RETRIES) {
 			skd_log_skreq(skdev, skreq, "retry");
-			blk_requeue_request(skdev->queue, skreq->req);
+			blk_requeue_request(skdev->queue, req);
 			break;
 		}
-	/* fall through to report error */
+		/* fall through */
 
 	case SKD_CHECK_STATUS_REPORT_ERROR:
 	default:
-		skd_end_request(skdev, skreq, BLK_STS_IOERR);
+		skreq->status = BLK_STS_IOERR;
+		blk_mq_complete_request(req);
 		break;
 	}
 }
 
-/* assume spinlock is already held */
 static void skd_release_skreq(struct skd_device *skdev,
 			      struct skd_request_context *skreq)
 {
-	u32 msg_slot;
-	struct skd_fitmsg_context *skmsg;
-
-	u32 timo_slot;
-
-	/*
-	 * Reclaim the FIT msg buffer if this is
-	 * the first of the requests it carried to
-	 * be completed. The FIT msg buffer used to
-	 * send this request cannot be reused until
-	 * we are sure the s1120 card has copied
-	 * it to its memory. The FIT msg might have
-	 * contained several requests. As soon as
-	 * any of them are completed we know that
-	 * the entire FIT msg was transferred.
-	 * Only the first completed request will
-	 * match the FIT msg buffer id. The FIT
-	 * msg buffer id is immediately updated.
-	 * When subsequent requests complete the FIT
-	 * msg buffer id won't match, so we know
-	 * quite cheaply that it is already done.
-	 */
-	msg_slot = skreq->fitmsg_id & SKD_ID_SLOT_MASK;
-	SKD_ASSERT(msg_slot < skdev->num_fitmsg_context);
-
-	skmsg = &skdev->skmsg_table[msg_slot];
-	if (skmsg->id == skreq->fitmsg_id) {
-		SKD_ASSERT(skmsg->state == SKD_MSG_STATE_BUSY);
-		SKD_ASSERT(skmsg->outstanding > 0);
-		skmsg->outstanding--;
-		if (skmsg->outstanding == 0) {
-			skmsg->state = SKD_MSG_STATE_IDLE;
-			skmsg->id += SKD_ID_INCR;
-			skmsg->next = skdev->skmsg_free_list;
-			skdev->skmsg_free_list = skmsg;
-		}
-	}
-
-	/*
-	 * Decrease the number of active requests.
-	 * Also decrements the count in the timeout slot.
-	 */
-	SKD_ASSERT(skdev->in_flight > 0);
-	skdev->in_flight -= 1;
-
-	timo_slot = skreq->timeout_stamp & SKD_TIMEOUT_SLOT_MASK;
-	SKD_ASSERT(skdev->timeout_slot[timo_slot] > 0);
-	skdev->timeout_slot[timo_slot] -= 1;
-
-	/*
-	 * Reset backpointer
-	 */
-	skreq->req = NULL;
-
 	/*
 	 * Reclaim the skd_request_context
-	 */
-	skreq->state = SKD_REQ_STATE_IDLE;
-	skreq->id += SKD_ID_INCR;
-	skreq->next = skdev->skreq_free_list;
-	skdev->skreq_free_list = skreq;
-}
-
-#define DRIVER_INQ_EVPD_PAGE_CODE   0xDA
-
-static void skd_do_inq_page_00(struct skd_device *skdev,
-			       volatile struct fit_completion_entry_v1 *skcomp,
-			       volatile struct fit_comp_error_info *skerr,
-			       uint8_t *cdb, uint8_t *buf)
-{
-	uint16_t insert_pt, max_bytes, drive_pages, drive_bytes, new_size;
-
-	/* Caller requested "supported pages".  The driver needs to insert
-	 * its page.
-	 */
-	pr_debug("%s:%s:%d skd_do_driver_inquiry: modify supported pages.\n",
-		 skdev->name, __func__, __LINE__);
-
-	/* If the device rejected the request because the CDB was
-	 * improperly formed, then just leave.
-	 */
-	if (skcomp->status == SAM_STAT_CHECK_CONDITION &&
-	    skerr->key == ILLEGAL_REQUEST && skerr->code == 0x24)
-		return;
-
-	/* Get the amount of space the caller allocated */
-	max_bytes = (cdb[3] << 8) | cdb[4];
-
-	/* Get the number of pages actually returned by the device */
-	drive_pages = (buf[2] << 8) | buf[3];
-	drive_bytes = drive_pages + 4;
-	new_size = drive_pages + 1;
-
-	/* Supported pages must be in numerical order, so find where
-	 * the driver page needs to be inserted into the list of
-	 * pages returned by the device.
-	 */
-	for (insert_pt = 4; insert_pt < drive_bytes; insert_pt++) {
-		if (buf[insert_pt] == DRIVER_INQ_EVPD_PAGE_CODE)
-			return; /* Device using this page code. abort */
-		else if (buf[insert_pt] > DRIVER_INQ_EVPD_PAGE_CODE)
-			break;
-	}
-
-	if (insert_pt < max_bytes) {
-		uint16_t u;
-
-		/* Shift everything up one byte to make room. */
-		for (u = new_size + 3; u > insert_pt; u--)
-			buf[u] = buf[u - 1];
-		buf[insert_pt] = DRIVER_INQ_EVPD_PAGE_CODE;
-
-		/* SCSI byte order increment of num_returned_bytes by 1 */
-		skcomp->num_returned_bytes =
-			be32_to_cpu(skcomp->num_returned_bytes) + 1;
-		skcomp->num_returned_bytes =
-			be32_to_cpu(skcomp->num_returned_bytes);
-	}
-
-	/* update page length field to reflect the driver's page too */
-	buf[2] = (uint8_t)((new_size >> 8) & 0xFF);
-	buf[3] = (uint8_t)((new_size >> 0) & 0xFF);
-}
-
-static void skd_get_link_info(struct pci_dev *pdev, u8 *speed, u8 *width)
-{
-	int pcie_reg;
-	u16 pci_bus_speed;
-	u8 pci_lanes;
-
-	pcie_reg = pci_find_capability(pdev, PCI_CAP_ID_EXP);
-	if (pcie_reg) {
-		u16 linksta;
-		pci_read_config_word(pdev, pcie_reg + PCI_EXP_LNKSTA, &linksta);
-
-		pci_bus_speed = linksta & 0xF;
-		pci_lanes = (linksta & 0x3F0) >> 4;
-	} else {
-		*speed = STEC_LINK_UNKNOWN;
-		*width = 0xFF;
-		return;
-	}
-
-	switch (pci_bus_speed) {
-	case 1:
-		*speed = STEC_LINK_2_5GTS;
-		break;
-	case 2:
-		*speed = STEC_LINK_5GTS;
-		break;
-	case 3:
-		*speed = STEC_LINK_8GTS;
-		break;
-	default:
-		*speed = STEC_LINK_UNKNOWN;
-		break;
-	}
-
-	if (pci_lanes <= 0x20)
-		*width = pci_lanes;
-	else
-		*width = 0xFF;
-}
-
-static void skd_do_inq_page_da(struct skd_device *skdev,
-			       volatile struct fit_completion_entry_v1 *skcomp,
-			       volatile struct fit_comp_error_info *skerr,
-			       uint8_t *cdb, uint8_t *buf)
-{
-	struct pci_dev *pdev = skdev->pdev;
-	unsigned max_bytes;
-	struct driver_inquiry_data inq;
-	u16 val;
-
-	pr_debug("%s:%s:%d skd_do_driver_inquiry: return driver page\n",
-		 skdev->name, __func__, __LINE__);
-
-	memset(&inq, 0, sizeof(inq));
-
-	inq.page_code = DRIVER_INQ_EVPD_PAGE_CODE;
-
-	skd_get_link_info(pdev, &inq.pcie_link_speed, &inq.pcie_link_lanes);
-	inq.pcie_bus_number = cpu_to_be16(pdev->bus->number);
-	inq.pcie_device_number = PCI_SLOT(pdev->devfn);
-	inq.pcie_function_number = PCI_FUNC(pdev->devfn);
-
-	pci_read_config_word(pdev, PCI_VENDOR_ID, &val);
-	inq.pcie_vendor_id = cpu_to_be16(val);
-
-	pci_read_config_word(pdev, PCI_DEVICE_ID, &val);
-	inq.pcie_device_id = cpu_to_be16(val);
-
-	pci_read_config_word(pdev, PCI_SUBSYSTEM_VENDOR_ID, &val);
-	inq.pcie_subsystem_vendor_id = cpu_to_be16(val);
-
-	pci_read_config_word(pdev, PCI_SUBSYSTEM_ID, &val);
-	inq.pcie_subsystem_device_id = cpu_to_be16(val);
-
-	/* Driver version, fixed lenth, padded with spaces on the right */
-	inq.driver_version_length = sizeof(inq.driver_version);
-	memset(&inq.driver_version, ' ', sizeof(inq.driver_version));
-	memcpy(inq.driver_version, DRV_VER_COMPL,
-	       min(sizeof(inq.driver_version), strlen(DRV_VER_COMPL)));
-
-	inq.page_length = cpu_to_be16((sizeof(inq) - 4));
-
-	/* Clear the error set by the device */
-	skcomp->status = SAM_STAT_GOOD;
-	memset((void *)skerr, 0, sizeof(*skerr));
-
-	/* copy response into output buffer */
-	max_bytes = (cdb[3] << 8) | cdb[4];
-	memcpy(buf, &inq, min_t(unsigned, max_bytes, sizeof(inq)));
-
-	skcomp->num_returned_bytes =
-		be32_to_cpu(min_t(uint16_t, max_bytes, sizeof(inq)));
-}
-
-static void skd_do_driver_inq(struct skd_device *skdev,
-			      volatile struct fit_completion_entry_v1 *skcomp,
-			      volatile struct fit_comp_error_info *skerr,
-			      uint8_t *cdb, uint8_t *buf)
-{
-	if (!buf)
-		return;
-	else if (cdb[0] != INQUIRY)
-		return;         /* Not an INQUIRY */
-	else if ((cdb[1] & 1) == 0)
-		return;         /* EVPD not set */
-	else if (cdb[2] == 0)
-		/* Need to add driver's page to supported pages list */
-		skd_do_inq_page_00(skdev, skcomp, skerr, cdb, buf);
-	else if (cdb[2] == DRIVER_INQ_EVPD_PAGE_CODE)
-		/* Caller requested driver's page */
-		skd_do_inq_page_da(skdev, skcomp, skerr, cdb, buf);
-}
-
-static unsigned char *skd_sg_1st_page_ptr(struct scatterlist *sg)
-{
-	if (!sg)
-		return NULL;
-	if (!sg_page(sg))
-		return NULL;
-	return sg_virt(sg);
-}
-
-static void skd_process_scsi_inq(struct skd_device *skdev,
-				 volatile struct fit_completion_entry_v1
-				 *skcomp,
-				 volatile struct fit_comp_error_info *skerr,
-				 struct skd_special_context *skspcl)
-{
-	uint8_t *buf;
-	struct fit_msg_hdr *fmh = (struct fit_msg_hdr *)skspcl->msg_buf;
-	struct skd_scsi_request *scsi_req = (struct skd_scsi_request *)&fmh[1];
-
-	dma_sync_sg_for_cpu(skdev->class_dev, skspcl->req.sg, skspcl->req.n_sg,
-			    skspcl->req.sg_data_dir);
-	buf = skd_sg_1st_page_ptr(skspcl->req.sg);
-
-	if (buf)
-		skd_do_driver_inq(skdev, skcomp, skerr, scsi_req->cdb, buf);
+	 */
+	skreq->state = SKD_REQ_STATE_IDLE;
 }
 
-
 static int skd_isr_completion_posted(struct skd_device *skdev,
 					int limit, int *enqueued)
 {
-	volatile struct fit_completion_entry_v1 *skcmp = NULL;
-	volatile struct fit_comp_error_info *skerr;
+	struct fit_completion_entry_v1 *skcmp;
+	struct fit_comp_error_info *skerr;
 	u16 req_id;
-	u32 req_slot;
+	u32 tag;
+	u16 hwq = 0;
+	struct request *rq;
 	struct skd_request_context *skreq;
-	u16 cmp_cntxt = 0;
-	u8 cmp_status = 0;
-	u8 cmp_cycle = 0;
-	u32 cmp_bytes = 0;
+	u16 cmp_cntxt;
+	u8 cmp_status;
+	u8 cmp_cycle;
+	u32 cmp_bytes;
 	int rc = 0;
 	int processed = 0;
 
+	lockdep_assert_held(&skdev->lock);
+
 	for (;; ) {
 		SKD_ASSERT(skdev->skcomp_ix < SKD_N_COMPLETION_ENTRY);
 
@@ -2652,16 +1479,14 @@ static int skd_isr_completion_posted(struct skd_device *skdev,
 
 		skerr = &skdev->skerr_table[skdev->skcomp_ix];
 
-		pr_debug("%s:%s:%d "
-			 "cycle=%d ix=%d got cycle=%d cmdctxt=0x%x stat=%d "
-			 "busy=%d rbytes=0x%x proto=%d\n",
-			 skdev->name, __func__, __LINE__, skdev->skcomp_cycle,
-			 skdev->skcomp_ix, cmp_cycle, cmp_cntxt, cmp_status,
-			 skdev->in_flight, cmp_bytes, skdev->proto_ver);
+		dev_dbg(&skdev->pdev->dev,
+			"cycle=%d ix=%d got cycle=%d cmdctxt=0x%x stat=%d busy=%d rbytes=0x%x proto=%d\n",
+			skdev->skcomp_cycle, skdev->skcomp_ix, cmp_cycle,
+			cmp_cntxt, cmp_status, skd_in_flight(skdev),
+			cmp_bytes, skdev->proto_ver);
 
 		if (cmp_cycle != skdev->skcomp_cycle) {
-			pr_debug("%s:%s:%d end of completions\n",
-				 skdev->name, __func__, __LINE__);
+			dev_dbg(&skdev->pdev->dev, "end of completions\n");
 			break;
 		}
 		/*
@@ -2680,49 +1505,38 @@ static int skd_isr_completion_posted(struct skd_device *skdev,
 		 * r/w request (see skd_start() above) or a special request.
 		 */
 		req_id = cmp_cntxt;
-		req_slot = req_id & SKD_ID_SLOT_AND_TABLE_MASK;
+		tag = req_id & SKD_ID_SLOT_AND_TABLE_MASK;
 
 		/* Is this other than a r/w request? */
-		if (req_slot >= skdev->num_req_context) {
+		if (tag >= skdev->num_req_context) {
 			/*
 			 * This is not a completion for a r/w request.
 			 */
+			WARN_ON_ONCE(blk_mq_tag_to_rq(skdev->tag_set.tags[hwq],
+						      tag));
 			skd_complete_other(skdev, skcmp, skerr);
 			continue;
 		}
 
-		skreq = &skdev->skreq_table[req_slot];
+		rq = blk_mq_tag_to_rq(skdev->tag_set.tags[hwq], tag);
+		if (WARN(!rq, "No request for tag %#x -> %#x\n", cmp_cntxt,
+			 tag))
+			continue;
+		skreq = blk_mq_rq_to_pdu(rq);
 
 		/*
 		 * Make sure the request ID for the slot matches.
 		 */
 		if (skreq->id != req_id) {
-			pr_debug("%s:%s:%d mismatch comp_id=0x%x req_id=0x%x\n",
-				 skdev->name, __func__, __LINE__,
-				 req_id, skreq->id);
-			{
-				u16 new_id = cmp_cntxt;
-				pr_err("(%s): Completion mismatch "
-				       "comp_id=0x%04x skreq=0x%04x new=0x%04x\n",
-				       skd_name(skdev), req_id,
-				       skreq->id, new_id);
+			dev_err(&skdev->pdev->dev,
+				"Completion mismatch comp_id=0x%04x skreq=0x%04x new=0x%04x\n",
+				req_id, skreq->id, cmp_cntxt);
 
-				continue;
-			}
+			continue;
 		}
 
 		SKD_ASSERT(skreq->state == SKD_REQ_STATE_BUSY);
 
-		if (skreq->state == SKD_REQ_STATE_ABORTED) {
-			pr_debug("%s:%s:%d reclaim req %p id=%04x\n",
-				 skdev->name, __func__, __LINE__,
-				 skreq, skreq->id);
-			/* a previously timed out command can
-			 * now be cleaned up */
-			skd_release_skreq(skdev, skreq);
-			continue;
-		}
-
 		skreq->completion = *skcmp;
 		if (unlikely(cmp_status == SAM_STAT_CHECK_CONDITION)) {
 			skreq->err_info = *skerr;
@@ -2734,27 +1548,17 @@ static int skd_isr_completion_posted(struct skd_device *skdev,
 		if (skreq->n_sg > 0)
 			skd_postop_sg_list(skdev, skreq);
 
-		if (!skreq->req) {
-			pr_debug("%s:%s:%d NULL backptr skdreq %p, "
-				 "req=0x%x req_id=0x%x\n",
-				 skdev->name, __func__, __LINE__,
-				 skreq, skreq->id, req_id);
-		} else {
-			/*
-			 * Capture the outcome and post it back to the
-			 * native request.
-			 */
-			if (likely(cmp_status == SAM_STAT_GOOD))
-				skd_end_request(skdev, skreq, BLK_STS_OK);
-			else
-				skd_resolve_req_exception(skdev, skreq);
-		}
+		skd_release_skreq(skdev, skreq);
 
 		/*
-		 * Release the skreq, its FIT msg (if one), timeout slot,
-		 * and queue depth.
+		 * Capture the outcome and post it back to the native request.
 		 */
-		skd_release_skreq(skdev, skreq);
+		if (likely(cmp_status == SAM_STAT_GOOD)) {
+			skreq->status = BLK_STS_OK;
+			blk_mq_complete_request(rq);
+		} else {
+			skd_resolve_req_exception(skdev, skreq, rq);
+		}
 
 		/* skd_isr_comp_limit equal zero means no limit */
 		if (limit) {
@@ -2765,8 +1569,8 @@ static int skd_isr_completion_posted(struct skd_device *skdev,
 		}
 	}
 
-	if ((skdev->state == SKD_DRVR_STATE_PAUSING)
-		&& (skdev->in_flight) == 0) {
+	if (skdev->state == SKD_DRVR_STATE_PAUSING &&
+	    skd_in_flight(skdev) == 0) {
 		skdev->state = SKD_DRVR_STATE_PAUSED;
 		wake_up_interruptible(&skdev->waitq);
 	}
@@ -2775,21 +1579,22 @@ static int skd_isr_completion_posted(struct skd_device *skdev,
 }
 
 static void skd_complete_other(struct skd_device *skdev,
-			       volatile struct fit_completion_entry_v1 *skcomp,
-			       volatile struct fit_comp_error_info *skerr)
+			       struct fit_completion_entry_v1 *skcomp,
+			       struct fit_comp_error_info *skerr)
 {
 	u32 req_id = 0;
 	u32 req_table;
 	u32 req_slot;
 	struct skd_special_context *skspcl;
 
+	lockdep_assert_held(&skdev->lock);
+
 	req_id = skcomp->tag;
 	req_table = req_id & SKD_ID_TABLE_MASK;
 	req_slot = req_id & SKD_ID_SLOT_MASK;
 
-	pr_debug("%s:%s:%d table=0x%x id=0x%x slot=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 req_table, req_id, req_slot);
+	dev_dbg(&skdev->pdev->dev, "table=0x%x id=0x%x slot=%d\n", req_table,
+		req_id, req_slot);
 
 	/*
 	 * Based on the request id, determine how to dispatch this completion.
@@ -2799,28 +1604,12 @@ static void skd_complete_other(struct skd_device *skdev,
 	switch (req_table) {
 	case SKD_ID_RW_REQUEST:
 		/*
-		 * The caller, skd_completion_posted_isr() above,
+		 * The caller, skd_isr_completion_posted() above,
 		 * handles r/w requests. The only way we get here
 		 * is if the req_slot is out of bounds.
 		 */
 		break;
 
-	case SKD_ID_SPECIAL_REQUEST:
-		/*
-		 * Make sure the req_slot is in bounds and that the id
-		 * matches.
-		 */
-		if (req_slot < skdev->n_special) {
-			skspcl = &skdev->skspcl_table[req_slot];
-			if (skspcl->req.id == req_id &&
-			    skspcl->req.state == SKD_REQ_STATE_BUSY) {
-				skd_complete_special(skdev,
-						     skcomp, skerr, skspcl);
-				return;
-			}
-		}
-		break;
-
 	case SKD_ID_INTERNAL:
 		if (req_slot == 0) {
 			skspcl = &skdev->internal_skspcl;
@@ -2851,72 +1640,9 @@ static void skd_complete_other(struct skd_device *skdev,
 	 */
 }
 
-static void skd_complete_special(struct skd_device *skdev,
-				 volatile struct fit_completion_entry_v1
-				 *skcomp,
-				 volatile struct fit_comp_error_info *skerr,
-				 struct skd_special_context *skspcl)
-{
-	pr_debug("%s:%s:%d  completing special request %p\n",
-		 skdev->name, __func__, __LINE__, skspcl);
-	if (skspcl->orphaned) {
-		/* Discard orphaned request */
-		/* ?: Can this release directly or does it need
-		 * to use a worker? */
-		pr_debug("%s:%s:%d release orphaned %p\n",
-			 skdev->name, __func__, __LINE__, skspcl);
-		skd_release_special(skdev, skspcl);
-		return;
-	}
-
-	skd_process_scsi_inq(skdev, skcomp, skerr, skspcl);
-
-	skspcl->req.state = SKD_REQ_STATE_COMPLETED;
-	skspcl->req.completion = *skcomp;
-	skspcl->req.err_info = *skerr;
-
-	skd_log_check_status(skdev, skspcl->req.completion.status, skerr->key,
-			     skerr->code, skerr->qual, skerr->fruc);
-
-	wake_up_interruptible(&skdev->waitq);
-}
-
-/* assume spinlock is already held */
-static void skd_release_special(struct skd_device *skdev,
-				struct skd_special_context *skspcl)
-{
-	int i, was_depleted;
-
-	for (i = 0; i < skspcl->req.n_sg; i++) {
-		struct page *page = sg_page(&skspcl->req.sg[i]);
-		__free_page(page);
-	}
-
-	was_depleted = (skdev->skspcl_free_list == NULL);
-
-	skspcl->req.state = SKD_REQ_STATE_IDLE;
-	skspcl->req.id += SKD_ID_INCR;
-	skspcl->req.next =
-		(struct skd_request_context *)skdev->skspcl_free_list;
-	skdev->skspcl_free_list = (struct skd_special_context *)skspcl;
-
-	if (was_depleted) {
-		pr_debug("%s:%s:%d skspcl was depleted\n",
-			 skdev->name, __func__, __LINE__);
-		/* Free list was depleted. Their might be waiters. */
-		wake_up_interruptible(&skdev->waitq);
-	}
-}
-
 static void skd_reset_skcomp(struct skd_device *skdev)
 {
-	u32 nbytes;
-	struct fit_completion_entry_v1 *skcomp;
-
-	nbytes = sizeof(*skcomp) * SKD_N_COMPLETION_ENTRY;
-	nbytes += sizeof(struct fit_comp_error_info) * SKD_N_COMPLETION_ENTRY;
-
-	memset(skdev->skcomp_table, 0, nbytes);
+	memset(skdev->skcomp_table, 0, SKD_SKCOMP_SIZE);
 
 	skdev->skcomp_ix = 0;
 	skdev->skcomp_cycle = 1;
@@ -2941,7 +1667,7 @@ static void skd_completion_worker(struct work_struct *work)
 	 * process everything in compq
 	 */
 	skd_isr_completion_posted(skdev, 0, &flush_enqueued);
-	skd_request_fn(skdev->queue);
+	schedule_work(&skdev->start_queue);
 
 	spin_unlock_irqrestore(&skdev->lock, flags);
 }
@@ -2951,14 +1677,13 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev);
 static irqreturn_t
 skd_isr(int irq, void *ptr)
 {
-	struct skd_device *skdev;
+	struct skd_device *skdev = ptr;
 	u32 intstat;
 	u32 ack;
 	int rc = 0;
 	int deferred = 0;
 	int flush_enqueued = 0;
 
-	skdev = (struct skd_device *)ptr;
 	spin_lock(&skdev->lock);
 
 	for (;; ) {
@@ -2967,8 +1692,8 @@ skd_isr(int irq, void *ptr)
 		ack = FIT_INT_DEF_MASK;
 		ack &= intstat;
 
-		pr_debug("%s:%s:%d intstat=0x%x ack=0x%x\n",
-			 skdev->name, __func__, __LINE__, intstat, ack);
+		dev_dbg(&skdev->pdev->dev, "intstat=0x%x ack=0x%x\n", intstat,
+			ack);
 
 		/* As long as there is an int pending on device, keep
 		 * running loop.  When none, get out, but if we've never
@@ -3018,12 +1743,12 @@ skd_isr(int irq, void *ptr)
 	}
 
 	if (unlikely(flush_enqueued))
-		skd_request_fn(skdev->queue);
+		schedule_work(&skdev->start_queue);
 
 	if (deferred)
 		schedule_work(&skdev->completion_worker);
 	else if (!flush_enqueued)
-		skd_request_fn(skdev->queue);
+		schedule_work(&skdev->start_queue);
 
 	spin_unlock(&skdev->lock);
 
@@ -3033,13 +1758,13 @@ skd_isr(int irq, void *ptr)
 static void skd_drive_fault(struct skd_device *skdev)
 {
 	skdev->state = SKD_DRVR_STATE_FAULT;
-	pr_err("(%s): Drive FAULT\n", skd_name(skdev));
+	dev_err(&skdev->pdev->dev, "Drive FAULT\n");
 }
 
 static void skd_drive_disappeared(struct skd_device *skdev)
 {
 	skdev->state = SKD_DRVR_STATE_DISAPPEARED;
-	pr_err("(%s): Drive DISAPPEARED\n", skd_name(skdev));
+	dev_err(&skdev->pdev->dev, "Drive DISAPPEARED\n");
 }
 
 static void skd_isr_fwstate(struct skd_device *skdev)
@@ -3052,10 +1777,9 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 	sense = SKD_READL(skdev, FIT_STATUS);
 	state = sense & FIT_SR_DRIVE_STATE_MASK;
 
-	pr_err("(%s): s1120 state %s(%d)=>%s(%d)\n",
-	       skd_name(skdev),
-	       skd_drive_state_to_str(skdev->drive_state), skdev->drive_state,
-	       skd_drive_state_to_str(state), state);
+	dev_err(&skdev->pdev->dev, "s1120 state %s(%d)=>%s(%d)\n",
+		skd_drive_state_to_str(skdev->drive_state), skdev->drive_state,
+		skd_drive_state_to_str(state), state);
 
 	skdev->drive_state = state;
 
@@ -3066,7 +1790,7 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 			break;
 		}
 		if (skdev->state == SKD_DRVR_STATE_RESTARTING)
-			skd_recover_requests(skdev, 0);
+			skd_recover_requests(skdev);
 		if (skdev->state == SKD_DRVR_STATE_WAIT_BOOT) {
 			skdev->timer_countdown = SKD_STARTING_TIMO;
 			skdev->state = SKD_DRVR_STATE_STARTING;
@@ -3087,11 +1811,11 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 			skdev->cur_max_queue_depth * 2 / 3 + 1;
 		if (skdev->queue_low_water_mark < 1)
 			skdev->queue_low_water_mark = 1;
-		pr_info(
-		       "(%s): Queue depth limit=%d dev=%d lowat=%d\n",
-		       skd_name(skdev),
-		       skdev->cur_max_queue_depth,
-		       skdev->dev_max_queue_depth, skdev->queue_low_water_mark);
+		dev_info(&skdev->pdev->dev,
+			 "Queue depth limit=%d dev=%d lowat=%d\n",
+			 skdev->cur_max_queue_depth,
+			 skdev->dev_max_queue_depth,
+			 skdev->queue_low_water_mark);
 
 		skd_refresh_device_data(skdev);
 		break;
@@ -3107,7 +1831,7 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 		 */
 		skdev->state = SKD_DRVR_STATE_BUSY_SANITIZE;
 		skdev->timer_countdown = SKD_TIMER_SECONDS(3);
-		blk_start_queue(skdev->queue);
+		schedule_work(&skdev->start_queue);
 		break;
 	case FIT_SR_DRIVE_BUSY_ERASE:
 		skdev->state = SKD_DRVR_STATE_BUSY_ERASE;
@@ -3128,8 +1852,7 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 		}
 		break;
 	case FIT_SR_DRIVE_FW_BOOTING:
-		pr_debug("%s:%s:%d ISR FIT_SR_DRIVE_FW_BOOTING %s\n",
-			 skdev->name, __func__, __LINE__, skdev->name);
+		dev_dbg(&skdev->pdev->dev, "ISR FIT_SR_DRIVE_FW_BOOTING\n");
 		skdev->state = SKD_DRVR_STATE_WAIT_BOOT;
 		skdev->timer_countdown = SKD_WAIT_BOOT_TIMO;
 		break;
@@ -3141,17 +1864,17 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 
 	case FIT_SR_DRIVE_FAULT:
 		skd_drive_fault(skdev);
-		skd_recover_requests(skdev, 0);
-		blk_start_queue(skdev->queue);
+		skd_recover_requests(skdev);
+		schedule_work(&skdev->start_queue);
 		break;
 
 	/* PCIe bus returned all Fs? */
 	case 0xFF:
-		pr_info("(%s): state=0x%x sense=0x%x\n",
-		       skd_name(skdev), state, sense);
+		dev_info(&skdev->pdev->dev, "state=0x%x sense=0x%x\n", state,
+			 sense);
 		skd_drive_disappeared(skdev);
-		skd_recover_requests(skdev, 0);
-		blk_start_queue(skdev->queue);
+		skd_recover_requests(skdev);
+		schedule_work(&skdev->start_queue);
 		break;
 	default:
 		/*
@@ -3159,92 +1882,33 @@ static void skd_isr_fwstate(struct skd_device *skdev)
 		 */
 		break;
 	}
-	pr_err("(%s): Driver state %s(%d)=>%s(%d)\n",
-	       skd_name(skdev),
-	       skd_skdev_state_to_str(prev_driver_state), prev_driver_state,
-	       skd_skdev_state_to_str(skdev->state), skdev->state);
+	dev_err(&skdev->pdev->dev, "Driver state %s(%d)=>%s(%d)\n",
+		skd_skdev_state_to_str(prev_driver_state), prev_driver_state,
+		skd_skdev_state_to_str(skdev->state), skdev->state);
 }
 
-static void skd_recover_requests(struct skd_device *skdev, int requeue)
+static void skd_recover_request(struct request *req, void *data, bool reserved)
 {
-	int i;
-
-	for (i = 0; i < skdev->num_req_context; i++) {
-		struct skd_request_context *skreq = &skdev->skreq_table[i];
-
-		if (skreq->state == SKD_REQ_STATE_BUSY) {
-			skd_log_skreq(skdev, skreq, "recover");
-
-			SKD_ASSERT((skreq->id & SKD_ID_INCR) != 0);
-			SKD_ASSERT(skreq->req != NULL);
-
-			/* Release DMA resources for the request. */
-			if (skreq->n_sg > 0)
-				skd_postop_sg_list(skdev, skreq);
-
-			if (requeue &&
-			    (unsigned long) ++skreq->req->special <
-			    SKD_MAX_RETRIES)
-				blk_requeue_request(skdev->queue, skreq->req);
-			else
-				skd_end_request(skdev, skreq, BLK_STS_IOERR);
-
-			skreq->req = NULL;
-
-			skreq->state = SKD_REQ_STATE_IDLE;
-			skreq->id += SKD_ID_INCR;
-		}
-		if (i > 0)
-			skreq[-1].next = skreq;
-		skreq->next = NULL;
-	}
-	skdev->skreq_free_list = skdev->skreq_table;
+	struct skd_device *const skdev = data;
+	struct skd_request_context *skreq = blk_mq_rq_to_pdu(req);
 
-	for (i = 0; i < skdev->num_fitmsg_context; i++) {
-		struct skd_fitmsg_context *skmsg = &skdev->skmsg_table[i];
-
-		if (skmsg->state == SKD_MSG_STATE_BUSY) {
-			skd_log_skmsg(skdev, skmsg, "salvaged");
-			SKD_ASSERT((skmsg->id & SKD_ID_INCR) != 0);
-			skmsg->state = SKD_MSG_STATE_IDLE;
-			skmsg->id += SKD_ID_INCR;
-		}
-		if (i > 0)
-			skmsg[-1].next = skmsg;
-		skmsg->next = NULL;
-	}
-	skdev->skmsg_free_list = skdev->skmsg_table;
+	if (skreq->state != SKD_REQ_STATE_BUSY)
+		return;
 
-	for (i = 0; i < skdev->n_special; i++) {
-		struct skd_special_context *skspcl = &skdev->skspcl_table[i];
+	skd_log_skreq(skdev, skreq, "recover");
 
-		/* If orphaned, reclaim it because it has already been reported
-		 * to the process as an error (it was just waiting for
-		 * a completion that didn't come, and now it will never come)
-		 * If busy, change to a state that will cause it to error
-		 * out in the wait routine and let it do the normal
-		 * reporting and reclaiming
-		 */
-		if (skspcl->req.state == SKD_REQ_STATE_BUSY) {
-			if (skspcl->orphaned) {
-				pr_debug("%s:%s:%d orphaned %p\n",
-					 skdev->name, __func__, __LINE__,
-					 skspcl);
-				skd_release_special(skdev, skspcl);
-			} else {
-				pr_debug("%s:%s:%d not orphaned %p\n",
-					 skdev->name, __func__, __LINE__,
-					 skspcl);
-				skspcl->req.state = SKD_REQ_STATE_ABORTED;
-			}
-		}
-	}
-	skdev->skspcl_free_list = skdev->skspcl_table;
+	/* Release DMA resources for the request. */
+	if (skreq->n_sg > 0)
+		skd_postop_sg_list(skdev, skreq);
 
-	for (i = 0; i < SKD_N_TIMEOUT_SLOT; i++)
-		skdev->timeout_slot[i] = 0;
+	skreq->state = SKD_REQ_STATE_IDLE;
+	skreq->status = BLK_STS_IOERR;
+	blk_mq_complete_request(req);
+}
 
-	skdev->in_flight = 0;
+static void skd_recover_requests(struct skd_device *skdev)
+{
+	blk_mq_tagset_busy_iter(&skdev->tag_set, skd_recover_request, skdev);
 }
 
 static void skd_isr_msg_from_dev(struct skd_device *skdev)
@@ -3255,8 +1919,8 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev)
 
 	mfd = SKD_READL(skdev, FIT_MSG_FROM_DEVICE);
 
-	pr_debug("%s:%s:%d mfd=0x%x last_mtd=0x%x\n",
-		 skdev->name, __func__, __LINE__, mfd, skdev->last_mtd);
+	dev_dbg(&skdev->pdev->dev, "mfd=0x%x last_mtd=0x%x\n", mfd,
+		skdev->last_mtd);
 
 	/* ignore any mtd that is an ack for something we didn't send */
 	if (FIT_MXD_TYPE(mfd) != FIT_MXD_TYPE(skdev->last_mtd))
@@ -3267,13 +1931,10 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev)
 		skdev->proto_ver = FIT_PROTOCOL_MAJOR_VER(mfd);
 
 		if (skdev->proto_ver != FIT_PROTOCOL_VERSION_1) {
-			pr_err("(%s): protocol mismatch\n",
-			       skdev->name);
-			pr_err("(%s):   got=%d support=%d\n",
-			       skdev->name, skdev->proto_ver,
-			       FIT_PROTOCOL_VERSION_1);
-			pr_err("(%s):   please upgrade driver\n",
-			       skdev->name);
+			dev_err(&skdev->pdev->dev, "protocol mismatch\n");
+			dev_err(&skdev->pdev->dev, "  got=%d support=%d\n",
+				skdev->proto_ver, FIT_PROTOCOL_VERSION_1);
+			dev_err(&skdev->pdev->dev, "  please upgrade driver\n");
 			skdev->state = SKD_DRVR_STATE_PROTOCOL_MISMATCH;
 			skd_soft_reset(skdev);
 			break;
@@ -3327,9 +1988,8 @@ static void skd_isr_msg_from_dev(struct skd_device *skdev)
 		SKD_WRITEL(skdev, mtd, FIT_MSG_TO_DEVICE);
 		skdev->last_mtd = mtd;
 
-		pr_err("(%s): Time sync driver=0x%x device=0x%x\n",
-		       skd_name(skdev),
-		       skdev->connect_time_stamp, skdev->drive_jiffies);
+		dev_err(&skdev->pdev->dev, "Time sync driver=0x%x device=0x%x\n",
+			skdev->connect_time_stamp, skdev->drive_jiffies);
 		break;
 
 	case FIT_MTD_ARM_QUEUE:
@@ -3351,8 +2011,7 @@ static void skd_disable_interrupts(struct skd_device *skdev)
 	sense = SKD_READL(skdev, FIT_CONTROL);
 	sense &= ~FIT_CR_ENABLE_INTERRUPTS;
 	SKD_WRITEL(skdev, sense, FIT_CONTROL);
-	pr_debug("%s:%s:%d sense 0x%x\n",
-		 skdev->name, __func__, __LINE__, sense);
+	dev_dbg(&skdev->pdev->dev, "sense 0x%x\n", sense);
 
 	/* Note that the 1s is written. A 1-bit means
 	 * disable, a 0 means enable.
@@ -3371,13 +2030,11 @@ static void skd_enable_interrupts(struct skd_device *skdev)
 	/* Note that the compliment of mask is written. A 1-bit means
 	 * disable, a 0 means enable. */
 	SKD_WRITEL(skdev, ~val, FIT_INT_MASK_HOST);
-	pr_debug("%s:%s:%d interrupt mask=0x%x\n",
-		 skdev->name, __func__, __LINE__, ~val);
+	dev_dbg(&skdev->pdev->dev, "interrupt mask=0x%x\n", ~val);
 
 	val = SKD_READL(skdev, FIT_CONTROL);
 	val |= FIT_CR_ENABLE_INTERRUPTS;
-	pr_debug("%s:%s:%d control=0x%x\n",
-		 skdev->name, __func__, __LINE__, val);
+	dev_dbg(&skdev->pdev->dev, "control=0x%x\n", val);
 	SKD_WRITEL(skdev, val, FIT_CONTROL);
 }
 
@@ -3393,8 +2050,7 @@ static void skd_soft_reset(struct skd_device *skdev)
 
 	val = SKD_READL(skdev, FIT_CONTROL);
 	val |= (FIT_CR_SOFT_RESET);
-	pr_debug("%s:%s:%d control=0x%x\n",
-		 skdev->name, __func__, __LINE__, val);
+	dev_dbg(&skdev->pdev->dev, "control=0x%x\n", val);
 	SKD_WRITEL(skdev, val, FIT_CONTROL);
 }
 
@@ -3411,8 +2067,7 @@ static void skd_start_device(struct skd_device *skdev)
 
 	sense = SKD_READL(skdev, FIT_STATUS);
 
-	pr_debug("%s:%s:%d initial status=0x%x\n",
-		 skdev->name, __func__, __LINE__, sense);
+	dev_dbg(&skdev->pdev->dev, "initial status=0x%x\n", sense);
 
 	state = sense & FIT_SR_DRIVE_STATE_MASK;
 	skdev->drive_state = state;
@@ -3425,25 +2080,23 @@ static void skd_start_device(struct skd_device *skdev)
 
 	switch (skdev->drive_state) {
 	case FIT_SR_DRIVE_OFFLINE:
-		pr_err("(%s): Drive offline...\n", skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "Drive offline...\n");
 		break;
 
 	case FIT_SR_DRIVE_FW_BOOTING:
-		pr_debug("%s:%s:%d FIT_SR_DRIVE_FW_BOOTING %s\n",
-			 skdev->name, __func__, __LINE__, skdev->name);
+		dev_dbg(&skdev->pdev->dev, "FIT_SR_DRIVE_FW_BOOTING\n");
 		skdev->state = SKD_DRVR_STATE_WAIT_BOOT;
 		skdev->timer_countdown = SKD_WAIT_BOOT_TIMO;
 		break;
 
 	case FIT_SR_DRIVE_BUSY_SANITIZE:
-		pr_info("(%s): Start: BUSY_SANITIZE\n",
-		       skd_name(skdev));
+		dev_info(&skdev->pdev->dev, "Start: BUSY_SANITIZE\n");
 		skdev->state = SKD_DRVR_STATE_BUSY_SANITIZE;
 		skdev->timer_countdown = SKD_STARTED_BUSY_TIMO;
 		break;
 
 	case FIT_SR_DRIVE_BUSY_ERASE:
-		pr_info("(%s): Start: BUSY_ERASE\n", skd_name(skdev));
+		dev_info(&skdev->pdev->dev, "Start: BUSY_ERASE\n");
 		skdev->state = SKD_DRVR_STATE_BUSY_ERASE;
 		skdev->timer_countdown = SKD_STARTED_BUSY_TIMO;
 		break;
@@ -3454,14 +2107,13 @@ static void skd_start_device(struct skd_device *skdev)
 		break;
 
 	case FIT_SR_DRIVE_BUSY:
-		pr_err("(%s): Drive Busy...\n", skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "Drive Busy...\n");
 		skdev->state = SKD_DRVR_STATE_BUSY;
 		skdev->timer_countdown = SKD_STARTED_BUSY_TIMO;
 		break;
 
 	case FIT_SR_DRIVE_SOFT_RESET:
-		pr_err("(%s) drive soft reset in prog\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "drive soft reset in prog\n");
 		break;
 
 	case FIT_SR_DRIVE_FAULT:
@@ -3471,9 +2123,8 @@ static void skd_start_device(struct skd_device *skdev)
 		 */
 		skd_drive_fault(skdev);
 		/*start the queue so we can respond with error to requests */
-		pr_debug("%s:%s:%d starting %s queue\n",
-			 skdev->name, __func__, __LINE__, skdev->name);
-		blk_start_queue(skdev->queue);
+		dev_dbg(&skdev->pdev->dev, "starting queue\n");
+		schedule_work(&skdev->start_queue);
 		skdev->gendisk_on = -1;
 		wake_up_interruptible(&skdev->waitq);
 		break;
@@ -3483,38 +2134,33 @@ static void skd_start_device(struct skd_device *skdev)
 		 * to the BAR1 addresses. */
 		skd_drive_disappeared(skdev);
 		/*start the queue so we can respond with error to requests */
-		pr_debug("%s:%s:%d starting %s queue to error-out reqs\n",
-			 skdev->name, __func__, __LINE__, skdev->name);
-		blk_start_queue(skdev->queue);
+		dev_dbg(&skdev->pdev->dev,
+			"starting queue to error-out reqs\n");
+		schedule_work(&skdev->start_queue);
 		skdev->gendisk_on = -1;
 		wake_up_interruptible(&skdev->waitq);
 		break;
 
 	default:
-		pr_err("(%s) Start: unknown state %x\n",
-		       skd_name(skdev), skdev->drive_state);
+		dev_err(&skdev->pdev->dev, "Start: unknown state %x\n",
+			skdev->drive_state);
 		break;
 	}
 
 	state = SKD_READL(skdev, FIT_CONTROL);
-	pr_debug("%s:%s:%d FIT Control Status=0x%x\n",
-		 skdev->name, __func__, __LINE__, state);
+	dev_dbg(&skdev->pdev->dev, "FIT Control Status=0x%x\n", state);
 
 	state = SKD_READL(skdev, FIT_INT_STATUS_HOST);
-	pr_debug("%s:%s:%d Intr Status=0x%x\n",
-		 skdev->name, __func__, __LINE__, state);
+	dev_dbg(&skdev->pdev->dev, "Intr Status=0x%x\n", state);
 
 	state = SKD_READL(skdev, FIT_INT_MASK_HOST);
-	pr_debug("%s:%s:%d Intr Mask=0x%x\n",
-		 skdev->name, __func__, __LINE__, state);
+	dev_dbg(&skdev->pdev->dev, "Intr Mask=0x%x\n", state);
 
 	state = SKD_READL(skdev, FIT_MSG_FROM_DEVICE);
-	pr_debug("%s:%s:%d Msg from Dev=0x%x\n",
-		 skdev->name, __func__, __LINE__, state);
+	dev_dbg(&skdev->pdev->dev, "Msg from Dev=0x%x\n", state);
 
 	state = SKD_READL(skdev, FIT_HW_VERSION);
-	pr_debug("%s:%s:%d HW version=0x%x\n",
-		 skdev->name, __func__, __LINE__, state);
+	dev_dbg(&skdev->pdev->dev, "HW version=0x%x\n", state);
 
 	spin_unlock_irqrestore(&skdev->lock, flags);
 }
@@ -3529,14 +2175,12 @@ static void skd_stop_device(struct skd_device *skdev)
 	spin_lock_irqsave(&skdev->lock, flags);
 
 	if (skdev->state != SKD_DRVR_STATE_ONLINE) {
-		pr_err("(%s): skd_stop_device not online no sync\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "%s not online no sync\n", __func__);
 		goto stop_out;
 	}
 
 	if (skspcl->req.state != SKD_REQ_STATE_IDLE) {
-		pr_err("(%s): skd_stop_device no special\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "%s no special\n", __func__);
 		goto stop_out;
 	}
 
@@ -3554,16 +2198,13 @@ static void skd_stop_device(struct skd_device *skdev)
 
 	switch (skdev->sync_done) {
 	case 0:
-		pr_err("(%s): skd_stop_device no sync\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "%s no sync\n", __func__);
 		break;
 	case 1:
-		pr_err("(%s): skd_stop_device sync done\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "%s sync done\n", __func__);
 		break;
 	default:
-		pr_err("(%s): skd_stop_device sync error\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "%s sync error\n", __func__);
 	}
 
 stop_out:
@@ -3593,8 +2234,8 @@ stop_out:
 	}
 
 	if (dev_state != FIT_SR_DRIVE_INIT)
-		pr_err("(%s): skd_stop_device state error 0x%02x\n",
-		       skd_name(skdev), dev_state);
+		dev_err(&skdev->pdev->dev, "%s state error 0x%02x\n", __func__,
+			dev_state);
 }
 
 /* assume spinlock is held */
@@ -3607,8 +2248,7 @@ static void skd_restart_device(struct skd_device *skdev)
 
 	state = SKD_READL(skdev, FIT_STATUS);
 
-	pr_debug("%s:%s:%d drive status=0x%x\n",
-		 skdev->name, __func__, __LINE__, state);
+	dev_dbg(&skdev->pdev->dev, "drive status=0x%x\n", state);
 
 	state &= FIT_SR_DRIVE_STATE_MASK;
 	skdev->drive_state = state;
@@ -3628,9 +2268,8 @@ static int skd_quiesce_dev(struct skd_device *skdev)
 	switch (skdev->state) {
 	case SKD_DRVR_STATE_BUSY:
 	case SKD_DRVR_STATE_BUSY_IMMINENT:
-		pr_debug("%s:%s:%d stopping %s queue\n",
-			 skdev->name, __func__, __LINE__, skdev->name);
-		blk_stop_queue(skdev->queue);
+		dev_dbg(&skdev->pdev->dev, "stopping queue\n");
+		blk_mq_stop_hw_queues(skdev->queue);
 		break;
 	case SKD_DRVR_STATE_ONLINE:
 	case SKD_DRVR_STATE_STOPPING:
@@ -3642,8 +2281,8 @@ static int skd_quiesce_dev(struct skd_device *skdev)
 	case SKD_DRVR_STATE_RESUMING:
 	default:
 		rc = -EINVAL;
-		pr_debug("%s:%s:%d state [%d] not implemented\n",
-			 skdev->name, __func__, __LINE__, skdev->state);
+		dev_dbg(&skdev->pdev->dev, "state [%d] not implemented\n",
+			skdev->state);
 	}
 	return rc;
 }
@@ -3655,8 +2294,7 @@ static int skd_unquiesce_dev(struct skd_device *skdev)
 
 	skd_log_skdev(skdev, "unquiesce");
 	if (skdev->state == SKD_DRVR_STATE_ONLINE) {
-		pr_debug("%s:%s:%d **** device already ONLINE\n",
-			 skdev->name, __func__, __LINE__);
+		dev_dbg(&skdev->pdev->dev, "**** device already ONLINE\n");
 		return 0;
 	}
 	if (skdev->drive_state != FIT_SR_DRIVE_ONLINE) {
@@ -3669,8 +2307,7 @@ static int skd_unquiesce_dev(struct skd_device *skdev)
 		 * to become available.
 		 */
 		skdev->state = SKD_DRVR_STATE_BUSY;
-		pr_debug("%s:%s:%d drive BUSY state\n",
-			 skdev->name, __func__, __LINE__);
+		dev_dbg(&skdev->pdev->dev, "drive BUSY state\n");
 		return 0;
 	}
 
@@ -3689,26 +2326,24 @@ static int skd_unquiesce_dev(struct skd_device *skdev)
 	case SKD_DRVR_STATE_IDLE:
 	case SKD_DRVR_STATE_LOAD:
 		skdev->state = SKD_DRVR_STATE_ONLINE;
-		pr_err("(%s): Driver state %s(%d)=>%s(%d)\n",
-		       skd_name(skdev),
-		       skd_skdev_state_to_str(prev_driver_state),
-		       prev_driver_state, skd_skdev_state_to_str(skdev->state),
-		       skdev->state);
-		pr_debug("%s:%s:%d **** device ONLINE...starting block queue\n",
-			 skdev->name, __func__, __LINE__);
-		pr_debug("%s:%s:%d starting %s queue\n",
-			 skdev->name, __func__, __LINE__, skdev->name);
-		pr_info("(%s): STEC s1120 ONLINE\n", skd_name(skdev));
-		blk_start_queue(skdev->queue);
+		dev_err(&skdev->pdev->dev, "Driver state %s(%d)=>%s(%d)\n",
+			skd_skdev_state_to_str(prev_driver_state),
+			prev_driver_state, skd_skdev_state_to_str(skdev->state),
+			skdev->state);
+		dev_dbg(&skdev->pdev->dev,
+			"**** device ONLINE...starting block queue\n");
+		dev_dbg(&skdev->pdev->dev, "starting queue\n");
+		dev_info(&skdev->pdev->dev, "STEC s1120 ONLINE\n");
+		schedule_work(&skdev->start_queue);
 		skdev->gendisk_on = 1;
 		wake_up_interruptible(&skdev->waitq);
 		break;
 
 	case SKD_DRVR_STATE_DISAPPEARED:
 	default:
-		pr_debug("%s:%s:%d **** driver state %d, not implemented \n",
-			 skdev->name, __func__, __LINE__,
-			 skdev->state);
+		dev_dbg(&skdev->pdev->dev,
+			"**** driver state %d, not implemented\n",
+			skdev->state);
 		return -EBUSY;
 	}
 	return 0;
@@ -3726,11 +2361,10 @@ static irqreturn_t skd_reserved_isr(int irq, void *skd_host_data)
 	unsigned long flags;
 
 	spin_lock_irqsave(&skdev->lock, flags);
-	pr_debug("%s:%s:%d MSIX = 0x%x\n",
-		 skdev->name, __func__, __LINE__,
-		 SKD_READL(skdev, FIT_INT_STATUS_HOST));
-	pr_err("(%s): MSIX reserved irq %d = 0x%x\n", skd_name(skdev),
-	       irq, SKD_READL(skdev, FIT_INT_STATUS_HOST));
+	dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n",
+		SKD_READL(skdev, FIT_INT_STATUS_HOST));
+	dev_err(&skdev->pdev->dev, "MSIX reserved irq %d = 0x%x\n", irq,
+		SKD_READL(skdev, FIT_INT_STATUS_HOST));
 	SKD_WRITEL(skdev, FIT_INT_RESERVED_MASK, FIT_INT_STATUS_HOST);
 	spin_unlock_irqrestore(&skdev->lock, flags);
 	return IRQ_HANDLED;
@@ -3742,9 +2376,8 @@ static irqreturn_t skd_statec_isr(int irq, void *skd_host_data)
 	unsigned long flags;
 
 	spin_lock_irqsave(&skdev->lock, flags);
-	pr_debug("%s:%s:%d MSIX = 0x%x\n",
-		 skdev->name, __func__, __LINE__,
-		 SKD_READL(skdev, FIT_INT_STATUS_HOST));
+	dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n",
+		SKD_READL(skdev, FIT_INT_STATUS_HOST));
 	SKD_WRITEL(skdev, FIT_ISH_FW_STATE_CHANGE, FIT_INT_STATUS_HOST);
 	skd_isr_fwstate(skdev);
 	spin_unlock_irqrestore(&skdev->lock, flags);
@@ -3759,19 +2392,18 @@ static irqreturn_t skd_comp_q(int irq, void *skd_host_data)
 	int deferred;
 
 	spin_lock_irqsave(&skdev->lock, flags);
-	pr_debug("%s:%s:%d MSIX = 0x%x\n",
-		 skdev->name, __func__, __LINE__,
-		 SKD_READL(skdev, FIT_INT_STATUS_HOST));
+	dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n",
+		SKD_READL(skdev, FIT_INT_STATUS_HOST));
 	SKD_WRITEL(skdev, FIT_ISH_COMPLETION_POSTED, FIT_INT_STATUS_HOST);
 	deferred = skd_isr_completion_posted(skdev, skd_isr_comp_limit,
 						&flush_enqueued);
 	if (flush_enqueued)
-		skd_request_fn(skdev->queue);
+		schedule_work(&skdev->start_queue);
 
 	if (deferred)
 		schedule_work(&skdev->completion_worker);
 	else if (!flush_enqueued)
-		skd_request_fn(skdev->queue);
+		schedule_work(&skdev->start_queue);
 
 	spin_unlock_irqrestore(&skdev->lock, flags);
 
@@ -3784,9 +2416,8 @@ static irqreturn_t skd_msg_isr(int irq, void *skd_host_data)
 	unsigned long flags;
 
 	spin_lock_irqsave(&skdev->lock, flags);
-	pr_debug("%s:%s:%d MSIX = 0x%x\n",
-		 skdev->name, __func__, __LINE__,
-		 SKD_READL(skdev, FIT_INT_STATUS_HOST));
+	dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n",
+		SKD_READL(skdev, FIT_INT_STATUS_HOST));
 	SKD_WRITEL(skdev, FIT_ISH_MSG_FROM_DEV, FIT_INT_STATUS_HOST);
 	skd_isr_msg_from_dev(skdev);
 	spin_unlock_irqrestore(&skdev->lock, flags);
@@ -3799,9 +2430,8 @@ static irqreturn_t skd_qfull_isr(int irq, void *skd_host_data)
 	unsigned long flags;
 
 	spin_lock_irqsave(&skdev->lock, flags);
-	pr_debug("%s:%s:%d MSIX = 0x%x\n",
-		 skdev->name, __func__, __LINE__,
-		 SKD_READL(skdev, FIT_INT_STATUS_HOST));
+	dev_dbg(&skdev->pdev->dev, "MSIX = 0x%x\n",
+		SKD_READL(skdev, FIT_INT_STATUS_HOST));
 	SKD_WRITEL(skdev, FIT_INT_QUEUE_FULL, FIT_INT_STATUS_HOST);
 	spin_unlock_irqrestore(&skdev->lock, flags);
 	return IRQ_HANDLED;
@@ -3850,8 +2480,7 @@ static int skd_acquire_msix(struct skd_device *skdev)
 	rc = pci_alloc_irq_vectors(pdev, SKD_MAX_MSIX_COUNT, SKD_MAX_MSIX_COUNT,
 			PCI_IRQ_MSIX);
 	if (rc < 0) {
-		pr_err("(%s): failed to enable MSI-X %d\n",
-		       skd_name(skdev), rc);
+		dev_err(&skdev->pdev->dev, "failed to enable MSI-X %d\n", rc);
 		goto out;
 	}
 
@@ -3859,8 +2488,7 @@ static int skd_acquire_msix(struct skd_device *skdev)
 			sizeof(struct skd_msix_entry), GFP_KERNEL);
 	if (!skdev->msix_entries) {
 		rc = -ENOMEM;
-		pr_err("(%s): msix table allocation error\n",
-		       skd_name(skdev));
+		dev_err(&skdev->pdev->dev, "msix table allocation error\n");
 		goto out;
 	}
 
@@ -3877,16 +2505,15 @@ static int skd_acquire_msix(struct skd_device *skdev)
 				msix_entries[i].handler, 0,
 				qentry->isr_name, skdev);
 		if (rc) {
-			pr_err("(%s): Unable to register(%d) MSI-X "
-			       "handler %d: %s\n",
-			       skd_name(skdev), rc, i, qentry->isr_name);
+			dev_err(&skdev->pdev->dev,
+				"Unable to register(%d) MSI-X handler %d: %s\n",
+				rc, i, qentry->isr_name);
 			goto msix_out;
 		}
 	}
 
-	pr_debug("%s:%s:%d %s: <%s> msix %d irq(s) enabled\n",
-		 skdev->name, __func__, __LINE__,
-		 pci_name(pdev), skdev->name, SKD_MAX_MSIX_COUNT);
+	dev_dbg(&skdev->pdev->dev, "%d msix irq(s) enabled\n",
+		SKD_MAX_MSIX_COUNT);
 	return 0;
 
 msix_out:
@@ -3909,8 +2536,8 @@ static int skd_acquire_irq(struct skd_device *skdev)
 		if (!rc)
 			return 0;
 
-		pr_err("(%s): failed to enable MSI-X, re-trying with MSI %d\n",
-		       skd_name(skdev), rc);
+		dev_err(&skdev->pdev->dev,
+			"failed to enable MSI-X, re-trying with MSI %d\n", rc);
 	}
 
 	snprintf(skdev->isr_name, sizeof(skdev->isr_name), "%s%d", DRV_NAME,
@@ -3920,8 +2547,8 @@ static int skd_acquire_irq(struct skd_device *skdev)
 		irq_flag |= PCI_IRQ_MSI;
 	rc = pci_alloc_irq_vectors(pdev, 1, 1, irq_flag);
 	if (rc < 0) {
-		pr_err("(%s): failed to allocate the MSI interrupt %d\n",
-			skd_name(skdev), rc);
+		dev_err(&skdev->pdev->dev,
+			"failed to allocate the MSI interrupt %d\n", rc);
 		return rc;
 	}
 
@@ -3930,8 +2557,8 @@ static int skd_acquire_irq(struct skd_device *skdev)
 			skdev->isr_name, skdev);
 	if (rc) {
 		pci_free_irq_vectors(pdev);
-		pr_err("(%s): failed to allocate interrupt %d\n",
-			skd_name(skdev), rc);
+		dev_err(&skdev->pdev->dev, "failed to allocate interrupt %d\n",
+			rc);
 		return rc;
 	}
 
@@ -3965,20 +2592,45 @@ static void skd_release_irq(struct skd_device *skdev)
  *****************************************************************************
  */
 
+static void *skd_alloc_dma(struct skd_device *skdev, struct kmem_cache *s,
+			   dma_addr_t *dma_handle, gfp_t gfp,
+			   enum dma_data_direction dir)
+{
+	struct device *dev = &skdev->pdev->dev;
+	void *buf;
+
+	buf = kmem_cache_alloc(s, gfp);
+	if (!buf)
+		return NULL;
+	*dma_handle = dma_map_single(dev, buf, s->size, dir);
+	if (dma_mapping_error(dev, *dma_handle)) {
+		kfree(buf);
+		buf = NULL;
+	}
+	return buf;
+}
+
+static void skd_free_dma(struct skd_device *skdev, struct kmem_cache *s,
+			 void *vaddr, dma_addr_t dma_handle,
+			 enum dma_data_direction dir)
+{
+	if (!vaddr)
+		return;
+
+	dma_unmap_single(&skdev->pdev->dev, dma_handle, s->size, dir);
+	kmem_cache_free(s, vaddr);
+}
+
 static int skd_cons_skcomp(struct skd_device *skdev)
 {
 	int rc = 0;
 	struct fit_completion_entry_v1 *skcomp;
-	u32 nbytes;
 
-	nbytes = sizeof(*skcomp) * SKD_N_COMPLETION_ENTRY;
-	nbytes += sizeof(struct fit_comp_error_info) * SKD_N_COMPLETION_ENTRY;
+	dev_dbg(&skdev->pdev->dev,
+		"comp pci_alloc, total bytes %zd entries %d\n",
+		SKD_SKCOMP_SIZE, SKD_N_COMPLETION_ENTRY);
 
-	pr_debug("%s:%s:%d comp pci_alloc, total bytes %d entries %d\n",
-		 skdev->name, __func__, __LINE__,
-		 nbytes, SKD_N_COMPLETION_ENTRY);
-
-	skcomp = pci_zalloc_consistent(skdev->pdev, nbytes,
+	skcomp = pci_zalloc_consistent(skdev->pdev, SKD_SKCOMP_SIZE,
 				       &skdev->cq_dma_address);
 
 	if (skcomp == NULL) {
@@ -4000,14 +2652,14 @@ static int skd_cons_skmsg(struct skd_device *skdev)
 	int rc = 0;
 	u32 i;
 
-	pr_debug("%s:%s:%d skmsg_table kzalloc, struct %lu, count %u total %lu\n",
-		 skdev->name, __func__, __LINE__,
-		 sizeof(struct skd_fitmsg_context),
-		 skdev->num_fitmsg_context,
-		 sizeof(struct skd_fitmsg_context) * skdev->num_fitmsg_context);
+	dev_dbg(&skdev->pdev->dev,
+		"skmsg_table kcalloc, struct %lu, count %u total %lu\n",
+		sizeof(struct skd_fitmsg_context), skdev->num_fitmsg_context,
+		sizeof(struct skd_fitmsg_context) * skdev->num_fitmsg_context);
 
-	skdev->skmsg_table = kzalloc(sizeof(struct skd_fitmsg_context)
-				     *skdev->num_fitmsg_context, GFP_KERNEL);
+	skdev->skmsg_table = kcalloc(skdev->num_fitmsg_context,
+				     sizeof(struct skd_fitmsg_context),
+				     GFP_KERNEL);
 	if (skdev->skmsg_table == NULL) {
 		rc = -ENOMEM;
 		goto err_out;
@@ -4020,9 +2672,8 @@ static int skd_cons_skmsg(struct skd_device *skdev)
 
 		skmsg->id = i + SKD_ID_FIT_MSG;
 
-		skmsg->state = SKD_MSG_STATE_IDLE;
 		skmsg->msg_buf = pci_alloc_consistent(skdev->pdev,
-						      SKD_N_FITMSG_BYTES + 64,
+						      SKD_N_FITMSG_BYTES,
 						      &skmsg->mb_dma_address);
 
 		if (skmsg->msg_buf == NULL) {
@@ -4030,22 +2681,13 @@ static int skd_cons_skmsg(struct skd_device *skdev)
 			goto err_out;
 		}
 
-		skmsg->offset = (u32)((u64)skmsg->msg_buf &
-				      (~FIT_QCMD_BASE_ADDRESS_MASK));
-		skmsg->msg_buf += ~FIT_QCMD_BASE_ADDRESS_MASK;
-		skmsg->msg_buf = (u8 *)((u64)skmsg->msg_buf &
-				       FIT_QCMD_BASE_ADDRESS_MASK);
-		skmsg->mb_dma_address += ~FIT_QCMD_BASE_ADDRESS_MASK;
-		skmsg->mb_dma_address &= FIT_QCMD_BASE_ADDRESS_MASK;
+		WARN(((uintptr_t)skmsg->msg_buf | skmsg->mb_dma_address) &
+		     (FIT_QCMD_ALIGN - 1),
+		     "not aligned: msg_buf %p mb_dma_address %#llx\n",
+		     skmsg->msg_buf, skmsg->mb_dma_address);
 		memset(skmsg->msg_buf, 0, SKD_N_FITMSG_BYTES);
-
-		skmsg->next = &skmsg[1];
 	}
 
-	/* Free list is in order starting with the 0th entry. */
-	skdev->skmsg_table[i - 1].next = NULL;
-	skdev->skmsg_free_list = skdev->skmsg_table;
-
 err_out:
 	return rc;
 }
@@ -4055,18 +2697,14 @@ static struct fit_sg_descriptor *skd_cons_sg_list(struct skd_device *skdev,
 						  dma_addr_t *ret_dma_addr)
 {
 	struct fit_sg_descriptor *sg_list;
-	u32 nbytes;
-
-	nbytes = sizeof(*sg_list) * n_sg;
 
-	sg_list = pci_alloc_consistent(skdev->pdev, nbytes, ret_dma_addr);
+	sg_list = skd_alloc_dma(skdev, skdev->sglist_cache, ret_dma_addr,
+				GFP_DMA | __GFP_ZERO, DMA_TO_DEVICE);
 
 	if (sg_list != NULL) {
 		uint64_t dma_address = *ret_dma_addr;
 		u32 i;
 
-		memset(sg_list, 0, nbytes);
-
 		for (i = 0; i < n_sg - 1; i++) {
 			uint64_t ndp_off;
 			ndp_off = (i + 1) * sizeof(struct fit_sg_descriptor);
@@ -4079,153 +2717,63 @@ static struct fit_sg_descriptor *skd_cons_sg_list(struct skd_device *skdev,
 	return sg_list;
 }
 
-static int skd_cons_skreq(struct skd_device *skdev)
+static void skd_free_sg_list(struct skd_device *skdev,
+			     struct fit_sg_descriptor *sg_list,
+			     dma_addr_t dma_addr)
 {
-	int rc = 0;
-	u32 i;
-
-	pr_debug("%s:%s:%d skreq_table kzalloc, struct %lu, count %u total %lu\n",
-		 skdev->name, __func__, __LINE__,
-		 sizeof(struct skd_request_context),
-		 skdev->num_req_context,
-		 sizeof(struct skd_request_context) * skdev->num_req_context);
-
-	skdev->skreq_table = kzalloc(sizeof(struct skd_request_context)
-				     * skdev->num_req_context, GFP_KERNEL);
-	if (skdev->skreq_table == NULL) {
-		rc = -ENOMEM;
-		goto err_out;
-	}
-
-	pr_debug("%s:%s:%d alloc sg_table sg_per_req %u scatlist %lu total %lu\n",
-		 skdev->name, __func__, __LINE__,
-		 skdev->sgs_per_request, sizeof(struct scatterlist),
-		 skdev->sgs_per_request * sizeof(struct scatterlist));
-
-	for (i = 0; i < skdev->num_req_context; i++) {
-		struct skd_request_context *skreq;
-
-		skreq = &skdev->skreq_table[i];
-
-		skreq->id = i + SKD_ID_RW_REQUEST;
-		skreq->state = SKD_REQ_STATE_IDLE;
-
-		skreq->sg = kzalloc(sizeof(struct scatterlist) *
-				    skdev->sgs_per_request, GFP_KERNEL);
-		if (skreq->sg == NULL) {
-			rc = -ENOMEM;
-			goto err_out;
-		}
-		sg_init_table(skreq->sg, skdev->sgs_per_request);
-
-		skreq->sksg_list = skd_cons_sg_list(skdev,
-						    skdev->sgs_per_request,
-						    &skreq->sksg_dma_address);
-
-		if (skreq->sksg_list == NULL) {
-			rc = -ENOMEM;
-			goto err_out;
-		}
-
-		skreq->next = &skreq[1];
-	}
-
-	/* Free list is in order starting with the 0th entry. */
-	skdev->skreq_table[i - 1].next = NULL;
-	skdev->skreq_free_list = skdev->skreq_table;
+	if (WARN_ON_ONCE(!sg_list))
+		return;
 
-err_out:
-	return rc;
+	skd_free_dma(skdev, skdev->sglist_cache, sg_list, dma_addr,
+		     DMA_TO_DEVICE);
 }
 
-static int skd_cons_skspcl(struct skd_device *skdev)
+static int skd_init_request(struct blk_mq_tag_set *set, struct request *rq,
+			    unsigned int hctx_idx, unsigned int numa_node)
 {
-	int rc = 0;
-	u32 i, nbytes;
-
-	pr_debug("%s:%s:%d skspcl_table kzalloc, struct %lu, count %u total %lu\n",
-		 skdev->name, __func__, __LINE__,
-		 sizeof(struct skd_special_context),
-		 skdev->n_special,
-		 sizeof(struct skd_special_context) * skdev->n_special);
-
-	skdev->skspcl_table = kzalloc(sizeof(struct skd_special_context)
-				      * skdev->n_special, GFP_KERNEL);
-	if (skdev->skspcl_table == NULL) {
-		rc = -ENOMEM;
-		goto err_out;
-	}
-
-	for (i = 0; i < skdev->n_special; i++) {
-		struct skd_special_context *skspcl;
+	struct skd_device *skdev = set->driver_data;
+	struct skd_request_context *skreq = blk_mq_rq_to_pdu(rq);
 
-		skspcl = &skdev->skspcl_table[i];
-
-		skspcl->req.id = i + SKD_ID_SPECIAL_REQUEST;
-		skspcl->req.state = SKD_REQ_STATE_IDLE;
-
-		skspcl->req.next = &skspcl[1].req;
-
-		nbytes = SKD_N_SPECIAL_FITMSG_BYTES;
-
-		skspcl->msg_buf =
-			pci_zalloc_consistent(skdev->pdev, nbytes,
-					      &skspcl->mb_dma_address);
-		if (skspcl->msg_buf == NULL) {
-			rc = -ENOMEM;
-			goto err_out;
-		}
-
-		skspcl->req.sg = kzalloc(sizeof(struct scatterlist) *
-					 SKD_N_SG_PER_SPECIAL, GFP_KERNEL);
-		if (skspcl->req.sg == NULL) {
-			rc = -ENOMEM;
-			goto err_out;
-		}
-
-		skspcl->req.sksg_list = skd_cons_sg_list(skdev,
-							 SKD_N_SG_PER_SPECIAL,
-							 &skspcl->req.
-							 sksg_dma_address);
-		if (skspcl->req.sksg_list == NULL) {
-			rc = -ENOMEM;
-			goto err_out;
-		}
-	}
+	skreq->state = SKD_REQ_STATE_IDLE;
+	skreq->sg = (void *)(skreq + 1);
+	sg_init_table(skreq->sg, skd_sgs_per_request);
+	skreq->sksg_list = skd_cons_sg_list(skdev, skd_sgs_per_request,
+					    &skreq->sksg_dma_address);
 
-	/* Free list is in order starting with the 0th entry. */
-	skdev->skspcl_table[i - 1].req.next = NULL;
-	skdev->skspcl_free_list = skdev->skspcl_table;
+	return skreq->sksg_list ? 0 : -ENOMEM;
+}
 
-	return rc;
+static void skd_exit_request(struct blk_mq_tag_set *set, struct request *rq,
+			     unsigned int hctx_idx)
+{
+	struct skd_device *skdev = set->driver_data;
+	struct skd_request_context *skreq = blk_mq_rq_to_pdu(rq);
 
-err_out:
-	return rc;
+	skd_free_sg_list(skdev, skreq->sksg_list, skreq->sksg_dma_address);
 }
 
 static int skd_cons_sksb(struct skd_device *skdev)
 {
 	int rc = 0;
 	struct skd_special_context *skspcl;
-	u32 nbytes;
 
 	skspcl = &skdev->internal_skspcl;
 
 	skspcl->req.id = 0 + SKD_ID_INTERNAL;
 	skspcl->req.state = SKD_REQ_STATE_IDLE;
 
-	nbytes = SKD_N_INTERNAL_BYTES;
-
-	skspcl->data_buf = pci_zalloc_consistent(skdev->pdev, nbytes,
-						 &skspcl->db_dma_address);
+	skspcl->data_buf = skd_alloc_dma(skdev, skdev->databuf_cache,
+					 &skspcl->db_dma_address,
+					 GFP_DMA | __GFP_ZERO,
+					 DMA_BIDIRECTIONAL);
 	if (skspcl->data_buf == NULL) {
 		rc = -ENOMEM;
 		goto err_out;
 	}
 
-	nbytes = SKD_N_SPECIAL_FITMSG_BYTES;
-	skspcl->msg_buf = pci_zalloc_consistent(skdev->pdev, nbytes,
-						&skspcl->mb_dma_address);
+	skspcl->msg_buf = skd_alloc_dma(skdev, skdev->msgbuf_cache,
+					&skspcl->mb_dma_address,
+					GFP_DMA | __GFP_ZERO, DMA_TO_DEVICE);
 	if (skspcl->msg_buf == NULL) {
 		rc = -ENOMEM;
 		goto err_out;
@@ -4247,6 +2795,14 @@ err_out:
 	return rc;
 }
 
+static const struct blk_mq_ops skd_mq_ops = {
+	.queue_rq	= skd_mq_queue_rq,
+	.complete	= skd_complete_rq,
+	.timeout	= skd_timed_out,
+	.init_request	= skd_init_request,
+	.exit_request	= skd_exit_request,
+};
+
 static int skd_cons_disk(struct skd_device *skdev)
 {
 	int rc = 0;
@@ -4268,31 +2824,46 @@ static int skd_cons_disk(struct skd_device *skdev)
 	disk->fops = &skd_blockdev_ops;
 	disk->private_data = skdev;
 
-	q = blk_init_queue(skd_request_fn, &skdev->lock);
-	if (!q) {
-		rc = -ENOMEM;
+	memset(&skdev->tag_set, 0, sizeof(skdev->tag_set));
+	skdev->tag_set.ops = &skd_mq_ops;
+	skdev->tag_set.nr_hw_queues = 1;
+	skdev->tag_set.queue_depth = skd_max_queue_depth;
+	skdev->tag_set.cmd_size = sizeof(struct skd_request_context) +
+		skdev->sgs_per_request * sizeof(struct scatterlist);
+	skdev->tag_set.numa_node = NUMA_NO_NODE;
+	skdev->tag_set.flags = BLK_MQ_F_SHOULD_MERGE |
+		BLK_MQ_F_SG_MERGE |
+		BLK_ALLOC_POLICY_TO_MQ_FLAG(BLK_TAG_ALLOC_FIFO);
+	skdev->tag_set.driver_data = skdev;
+	rc = blk_mq_alloc_tag_set(&skdev->tag_set);
+	if (rc)
+		goto err_out;
+	q = blk_mq_init_queue(&skdev->tag_set);
+	if (IS_ERR(q)) {
+		blk_mq_free_tag_set(&skdev->tag_set);
+		rc = PTR_ERR(q);
 		goto err_out;
 	}
-	blk_queue_bounce_limit(q, BLK_BOUNCE_HIGH);
+	q->queuedata = skdev;
 
 	skdev->queue = q;
 	disk->queue = q;
-	q->queuedata = skdev;
 
 	blk_queue_write_cache(q, true, true);
 	blk_queue_max_segments(q, skdev->sgs_per_request);
 	blk_queue_max_hw_sectors(q, SKD_N_MAX_SECTORS);
 
-	/* set sysfs ptimal_io_size to 8K */
+	/* set optimal I/O size to 8KB */
 	blk_queue_io_opt(q, 8192);
 
 	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
 	queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, q);
 
+	blk_queue_rq_timeout(q, 8 * HZ);
+
 	spin_lock_irqsave(&skdev->lock, flags);
-	pr_debug("%s:%s:%d stopping %s queue\n",
-		 skdev->name, __func__, __LINE__, skdev->name);
-	blk_stop_queue(skdev->queue);
+	dev_dbg(&skdev->pdev->dev, "stopping queue\n");
+	blk_mq_stop_hw_queues(skdev->queue);
 	spin_unlock_irqrestore(&skdev->lock, flags);
 
 err_out:
@@ -4306,13 +2877,13 @@ static struct skd_device *skd_construct(struct pci_dev *pdev)
 {
 	struct skd_device *skdev;
 	int blk_major = skd_major;
+	size_t size;
 	int rc;
 
 	skdev = kzalloc(sizeof(*skdev), GFP_KERNEL);
 
 	if (!skdev) {
-		pr_err(PFX "(%s): memory alloc failure\n",
-		       pci_name(pdev));
+		dev_err(&pdev->dev, "memory alloc failure\n");
 		return NULL;
 	}
 
@@ -4320,60 +2891,71 @@ static struct skd_device *skd_construct(struct pci_dev *pdev)
 	skdev->pdev = pdev;
 	skdev->devno = skd_next_devno++;
 	skdev->major = blk_major;
-	sprintf(skdev->name, DRV_NAME "%d", skdev->devno);
 	skdev->dev_max_queue_depth = 0;
 
 	skdev->num_req_context = skd_max_queue_depth;
 	skdev->num_fitmsg_context = skd_max_queue_depth;
-	skdev->n_special = skd_max_pass_thru;
 	skdev->cur_max_queue_depth = 1;
 	skdev->queue_low_water_mark = 1;
 	skdev->proto_ver = 99;
 	skdev->sgs_per_request = skd_sgs_per_request;
 	skdev->dbg_level = skd_dbg_level;
 
-	atomic_set(&skdev->device_count, 0);
-
 	spin_lock_init(&skdev->lock);
 
+	INIT_WORK(&skdev->start_queue, skd_start_queue);
 	INIT_WORK(&skdev->completion_worker, skd_completion_worker);
 
-	pr_debug("%s:%s:%d skcomp\n", skdev->name, __func__, __LINE__);
-	rc = skd_cons_skcomp(skdev);
-	if (rc < 0)
+	size = max(SKD_N_FITMSG_BYTES, SKD_N_SPECIAL_FITMSG_BYTES);
+	skdev->msgbuf_cache = kmem_cache_create("skd-msgbuf", size, 0,
+						SLAB_HWCACHE_ALIGN, NULL);
+	if (!skdev->msgbuf_cache)
 		goto err_out;
-
-	pr_debug("%s:%s:%d skmsg\n", skdev->name, __func__, __LINE__);
-	rc = skd_cons_skmsg(skdev);
-	if (rc < 0)
+	WARN_ONCE(kmem_cache_size(skdev->msgbuf_cache) < size,
+		  "skd-msgbuf: %d < %zd\n",
+		  kmem_cache_size(skdev->msgbuf_cache), size);
+	size = skd_sgs_per_request * sizeof(struct fit_sg_descriptor);
+	skdev->sglist_cache = kmem_cache_create("skd-sglist", size, 0,
+						SLAB_HWCACHE_ALIGN, NULL);
+	if (!skdev->sglist_cache)
+		goto err_out;
+	WARN_ONCE(kmem_cache_size(skdev->sglist_cache) < size,
+		  "skd-sglist: %d < %zd\n",
+		  kmem_cache_size(skdev->sglist_cache), size);
+	size = SKD_N_INTERNAL_BYTES;
+	skdev->databuf_cache = kmem_cache_create("skd-databuf", size, 0,
+						 SLAB_HWCACHE_ALIGN, NULL);
+	if (!skdev->databuf_cache)
 		goto err_out;
+	WARN_ONCE(kmem_cache_size(skdev->databuf_cache) < size,
+		  "skd-databuf: %d < %zd\n",
+		  kmem_cache_size(skdev->databuf_cache), size);
 
-	pr_debug("%s:%s:%d skreq\n", skdev->name, __func__, __LINE__);
-	rc = skd_cons_skreq(skdev);
+	dev_dbg(&skdev->pdev->dev, "skcomp\n");
+	rc = skd_cons_skcomp(skdev);
 	if (rc < 0)
 		goto err_out;
 
-	pr_debug("%s:%s:%d skspcl\n", skdev->name, __func__, __LINE__);
-	rc = skd_cons_skspcl(skdev);
+	dev_dbg(&skdev->pdev->dev, "skmsg\n");
+	rc = skd_cons_skmsg(skdev);
 	if (rc < 0)
 		goto err_out;
 
-	pr_debug("%s:%s:%d sksb\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "sksb\n");
 	rc = skd_cons_sksb(skdev);
 	if (rc < 0)
 		goto err_out;
 
-	pr_debug("%s:%s:%d disk\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "disk\n");
 	rc = skd_cons_disk(skdev);
 	if (rc < 0)
 		goto err_out;
 
-	pr_debug("%s:%s:%d VICTORY\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "VICTORY\n");
 	return skdev;
 
 err_out:
-	pr_debug("%s:%s:%d construct failed\n",
-		 skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "construct failed\n");
 	skd_destruct(skdev);
 	return NULL;
 }
@@ -4386,14 +2968,9 @@ err_out:
 
 static void skd_free_skcomp(struct skd_device *skdev)
 {
-	if (skdev->skcomp_table != NULL) {
-		u32 nbytes;
-
-		nbytes = sizeof(skdev->skcomp_table[0]) *
-			 SKD_N_COMPLETION_ENTRY;
-		pci_free_consistent(skdev->pdev, nbytes,
+	if (skdev->skcomp_table)
+		pci_free_consistent(skdev->pdev, SKD_SKCOMP_SIZE,
 				    skdev->skcomp_table, skdev->cq_dma_address);
-	}
 
 	skdev->skcomp_table = NULL;
 	skdev->cq_dma_address = 0;
@@ -4412,8 +2989,6 @@ static void skd_free_skmsg(struct skd_device *skdev)
 		skmsg = &skdev->skmsg_table[i];
 
 		if (skmsg->msg_buf != NULL) {
-			skmsg->msg_buf += skmsg->offset;
-			skmsg->mb_dma_address += skmsg->offset;
 			pci_free_consistent(skdev->pdev, SKD_N_FITMSG_BYTES,
 					    skmsg->msg_buf,
 					    skmsg->mb_dma_address);
@@ -4426,109 +3001,23 @@ static void skd_free_skmsg(struct skd_device *skdev)
 	skdev->skmsg_table = NULL;
 }
 
-static void skd_free_sg_list(struct skd_device *skdev,
-			     struct fit_sg_descriptor *sg_list,
-			     u32 n_sg, dma_addr_t dma_addr)
-{
-	if (sg_list != NULL) {
-		u32 nbytes;
-
-		nbytes = sizeof(*sg_list) * n_sg;
-
-		pci_free_consistent(skdev->pdev, nbytes, sg_list, dma_addr);
-	}
-}
-
-static void skd_free_skreq(struct skd_device *skdev)
-{
-	u32 i;
-
-	if (skdev->skreq_table == NULL)
-		return;
-
-	for (i = 0; i < skdev->num_req_context; i++) {
-		struct skd_request_context *skreq;
-
-		skreq = &skdev->skreq_table[i];
-
-		skd_free_sg_list(skdev, skreq->sksg_list,
-				 skdev->sgs_per_request,
-				 skreq->sksg_dma_address);
-
-		skreq->sksg_list = NULL;
-		skreq->sksg_dma_address = 0;
-
-		kfree(skreq->sg);
-	}
-
-	kfree(skdev->skreq_table);
-	skdev->skreq_table = NULL;
-}
-
-static void skd_free_skspcl(struct skd_device *skdev)
-{
-	u32 i;
-	u32 nbytes;
-
-	if (skdev->skspcl_table == NULL)
-		return;
-
-	for (i = 0; i < skdev->n_special; i++) {
-		struct skd_special_context *skspcl;
-
-		skspcl = &skdev->skspcl_table[i];
-
-		if (skspcl->msg_buf != NULL) {
-			nbytes = SKD_N_SPECIAL_FITMSG_BYTES;
-			pci_free_consistent(skdev->pdev, nbytes,
-					    skspcl->msg_buf,
-					    skspcl->mb_dma_address);
-		}
-
-		skspcl->msg_buf = NULL;
-		skspcl->mb_dma_address = 0;
-
-		skd_free_sg_list(skdev, skspcl->req.sksg_list,
-				 SKD_N_SG_PER_SPECIAL,
-				 skspcl->req.sksg_dma_address);
-
-		skspcl->req.sksg_list = NULL;
-		skspcl->req.sksg_dma_address = 0;
-
-		kfree(skspcl->req.sg);
-	}
-
-	kfree(skdev->skspcl_table);
-	skdev->skspcl_table = NULL;
-}
-
 static void skd_free_sksb(struct skd_device *skdev)
 {
-	struct skd_special_context *skspcl;
-	u32 nbytes;
-
-	skspcl = &skdev->internal_skspcl;
-
-	if (skspcl->data_buf != NULL) {
-		nbytes = SKD_N_INTERNAL_BYTES;
+	struct skd_special_context *skspcl = &skdev->internal_skspcl;
 
-		pci_free_consistent(skdev->pdev, nbytes,
-				    skspcl->data_buf, skspcl->db_dma_address);
-	}
+	skd_free_dma(skdev, skdev->databuf_cache, skspcl->data_buf,
+		     skspcl->db_dma_address, DMA_BIDIRECTIONAL);
 
 	skspcl->data_buf = NULL;
 	skspcl->db_dma_address = 0;
 
-	if (skspcl->msg_buf != NULL) {
-		nbytes = SKD_N_SPECIAL_FITMSG_BYTES;
-		pci_free_consistent(skdev->pdev, nbytes,
-				    skspcl->msg_buf, skspcl->mb_dma_address);
-	}
+	skd_free_dma(skdev, skdev->msgbuf_cache, skspcl->msg_buf,
+		     skspcl->mb_dma_address, DMA_TO_DEVICE);
 
 	skspcl->msg_buf = NULL;
 	skspcl->mb_dma_address = 0;
 
-	skd_free_sg_list(skdev, skspcl->req.sksg_list, 1,
+	skd_free_sg_list(skdev, skspcl->req.sksg_list,
 			 skspcl->req.sksg_dma_address);
 
 	skspcl->req.sksg_list = NULL;
@@ -4539,15 +3028,20 @@ static void skd_free_disk(struct skd_device *skdev)
 {
 	struct gendisk *disk = skdev->disk;
 
-	if (disk != NULL) {
-		struct request_queue *q = disk->queue;
+	if (disk && (disk->flags & GENHD_FL_UP))
+		del_gendisk(disk);
 
-		if (disk->flags & GENHD_FL_UP)
-			del_gendisk(disk);
-		if (q)
-			blk_cleanup_queue(q);
-		put_disk(disk);
+	if (skdev->queue) {
+		blk_cleanup_queue(skdev->queue);
+		skdev->queue = NULL;
+		if (disk)
+			disk->queue = NULL;
 	}
+
+	if (skdev->tag_set.tags)
+		blk_mq_free_tag_set(&skdev->tag_set);
+
+	put_disk(disk);
 	skdev->disk = NULL;
 }
 
@@ -4556,26 +3050,25 @@ static void skd_destruct(struct skd_device *skdev)
 	if (skdev == NULL)
 		return;
 
+	cancel_work_sync(&skdev->start_queue);
 
-	pr_debug("%s:%s:%d disk\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "disk\n");
 	skd_free_disk(skdev);
 
-	pr_debug("%s:%s:%d sksb\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "sksb\n");
 	skd_free_sksb(skdev);
 
-	pr_debug("%s:%s:%d skspcl\n", skdev->name, __func__, __LINE__);
-	skd_free_skspcl(skdev);
-
-	pr_debug("%s:%s:%d skreq\n", skdev->name, __func__, __LINE__);
-	skd_free_skreq(skdev);
-
-	pr_debug("%s:%s:%d skmsg\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "skmsg\n");
 	skd_free_skmsg(skdev);
 
-	pr_debug("%s:%s:%d skcomp\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "skcomp\n");
 	skd_free_skcomp(skdev);
 
-	pr_debug("%s:%s:%d skdev\n", skdev->name, __func__, __LINE__);
+	kmem_cache_destroy(skdev->databuf_cache);
+	kmem_cache_destroy(skdev->sglist_cache);
+	kmem_cache_destroy(skdev->msgbuf_cache);
+
+	dev_dbg(&skdev->pdev->dev, "skdev\n");
 	kfree(skdev);
 }
 
@@ -4592,9 +3085,8 @@ static int skd_bdev_getgeo(struct block_device *bdev, struct hd_geometry *geo)
 
 	skdev = bdev->bd_disk->private_data;
 
-	pr_debug("%s:%s:%d %s: CMD[%s] getgeo device\n",
-		 skdev->name, __func__, __LINE__,
-		 bdev->bd_disk->disk_name, current->comm);
+	dev_dbg(&skdev->pdev->dev, "%s: CMD[%s] getgeo device\n",
+		bdev->bd_disk->disk_name, current->comm);
 
 	if (skdev->read_cap_is_valid) {
 		capacity = get_capacity(skdev->disk);
@@ -4609,18 +3101,16 @@ static int skd_bdev_getgeo(struct block_device *bdev, struct hd_geometry *geo)
 
 static int skd_bdev_attach(struct device *parent, struct skd_device *skdev)
 {
-	pr_debug("%s:%s:%d add_disk\n", skdev->name, __func__, __LINE__);
+	dev_dbg(&skdev->pdev->dev, "add_disk\n");
 	device_add_disk(parent, skdev->disk);
 	return 0;
 }
 
 static const struct block_device_operations skd_blockdev_ops = {
 	.owner		= THIS_MODULE,
-	.ioctl		= skd_bdev_ioctl,
 	.getgeo		= skd_bdev_getgeo,
 };
 
-
 /*
  *****************************************************************************
  * PCIe DRIVER GLUE
@@ -4671,10 +3161,8 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	char pci_str[32];
 	struct skd_device *skdev;
 
-	pr_info("STEC s1120 Driver(%s) version %s-b%s\n",
-	       DRV_NAME, DRV_VERSION, DRV_BUILD_ID);
-	pr_info("(skd?:??:[%s]): vendor=%04X device=%04x\n",
-	       pci_name(pdev), pdev->vendor, pdev->device);
+	dev_dbg(&pdev->dev, "vendor=%04X device=%04x\n", pdev->vendor,
+		pdev->device);
 
 	rc = pci_enable_device(pdev);
 	if (rc)
@@ -4685,16 +3173,13 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(64));
 	if (!rc) {
 		if (pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64))) {
-
-			pr_err("(%s): consistent DMA mask error %d\n",
-			       pci_name(pdev), rc);
+			dev_err(&pdev->dev, "consistent DMA mask error %d\n",
+				rc);
 		}
 	} else {
-		(rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)));
+		rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32));
 		if (rc) {
-
-			pr_err("(%s): DMA mask error %d\n",
-			       pci_name(pdev), rc);
+			dev_err(&pdev->dev, "DMA mask error %d\n", rc);
 			goto err_out_regions;
 		}
 	}
@@ -4714,19 +3199,17 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	}
 
 	skd_pci_info(skdev, pci_str);
-	pr_info("(%s): %s 64bit\n", skd_name(skdev), pci_str);
+	dev_info(&pdev->dev, "%s 64bit\n", pci_str);
 
 	pci_set_master(pdev);
 	rc = pci_enable_pcie_error_reporting(pdev);
 	if (rc) {
-		pr_err(
-		       "(%s): bad enable of PCIe error reporting rc=%d\n",
-		       skd_name(skdev), rc);
+		dev_err(&pdev->dev,
+			"bad enable of PCIe error reporting rc=%d\n", rc);
 		skdev->pcie_error_reporting_is_enabled = 0;
 	} else
 		skdev->pcie_error_reporting_is_enabled = 1;
 
-
 	pci_set_drvdata(pdev, skdev);
 
 	for (i = 0; i < SKD_MAX_BARS; i++) {
@@ -4735,21 +3218,19 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		skdev->mem_map[i] = ioremap(skdev->mem_phys[i],
 					    skdev->mem_size[i]);
 		if (!skdev->mem_map[i]) {
-			pr_err("(%s): Unable to map adapter memory!\n",
-			       skd_name(skdev));
+			dev_err(&pdev->dev,
+				"Unable to map adapter memory!\n");
 			rc = -ENODEV;
 			goto err_out_iounmap;
 		}
-		pr_debug("%s:%s:%d mem_map=%p, phyd=%016llx, size=%d\n",
-			 skdev->name, __func__, __LINE__,
-			 skdev->mem_map[i],
-			 (uint64_t)skdev->mem_phys[i], skdev->mem_size[i]);
+		dev_dbg(&pdev->dev, "mem_map=%p, phyd=%016llx, size=%d\n",
+			skdev->mem_map[i], (uint64_t)skdev->mem_phys[i],
+			skdev->mem_size[i]);
 	}
 
 	rc = skd_acquire_irq(skdev);
 	if (rc) {
-		pr_err("(%s): interrupt resource error %d\n",
-		       skd_name(skdev), rc);
+		dev_err(&pdev->dev, "interrupt resource error %d\n", rc);
 		goto err_out_iounmap;
 	}
 
@@ -4771,29 +3252,14 @@ static int skd_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	} else {
 		/* we timed out, something is wrong with the device,
 		   don't add the disk structure */
-		pr_err(
-		       "(%s): error: waiting for s1120 timed out %d!\n",
-		       skd_name(skdev), rc);
+		dev_err(&pdev->dev, "error: waiting for s1120 timed out %d!\n",
+			rc);
 		/* in case of no error; we timeout with ENXIO */
 		if (!rc)
 			rc = -ENXIO;
 		goto err_out_timer;
 	}
 
-
-#ifdef SKD_VMK_POLL_HANDLER
-	if (skdev->irq_type == SKD_IRQ_MSIX) {
-		/* MSIX completion handler is being used for coredump */
-		vmklnx_scsi_register_poll_handler(skdev->scsi_host,
-						  skdev->msix_entries[5].vector,
-						  skd_comp_q, skdev);
-	} else {
-		vmklnx_scsi_register_poll_handler(skdev->scsi_host,
-						  skdev->pdev->irq, skd_isr,
-						  skdev);
-	}
-#endif  /* SKD_VMK_POLL_HANDLER */
-
 	return rc;
 
 err_out_timer:
@@ -4826,7 +3292,7 @@ static void skd_pci_remove(struct pci_dev *pdev)
 
 	skdev = pci_get_drvdata(pdev);
 	if (!skdev) {
-		pr_err("%s: no device data for PCI\n", pci_name(pdev));
+		dev_err(&pdev->dev, "no device data for PCI\n");
 		return;
 	}
 	skd_stop_device(skdev);
@@ -4834,7 +3300,7 @@ static void skd_pci_remove(struct pci_dev *pdev)
 
 	for (i = 0; i < SKD_MAX_BARS; i++)
 		if (skdev->mem_map[i])
-			iounmap((u32 *)skdev->mem_map[i]);
+			iounmap(skdev->mem_map[i]);
 
 	if (skdev->pcie_error_reporting_is_enabled)
 		pci_disable_pcie_error_reporting(pdev);
@@ -4855,7 +3321,7 @@ static int skd_pci_suspend(struct pci_dev *pdev, pm_message_t state)
 
 	skdev = pci_get_drvdata(pdev);
 	if (!skdev) {
-		pr_err("%s: no device data for PCI\n", pci_name(pdev));
+		dev_err(&pdev->dev, "no device data for PCI\n");
 		return -EIO;
 	}
 
@@ -4865,7 +3331,7 @@ static int skd_pci_suspend(struct pci_dev *pdev, pm_message_t state)
 
 	for (i = 0; i < SKD_MAX_BARS; i++)
 		if (skdev->mem_map[i])
-			iounmap((u32 *)skdev->mem_map[i]);
+			iounmap(skdev->mem_map[i]);
 
 	if (skdev->pcie_error_reporting_is_enabled)
 		pci_disable_pcie_error_reporting(pdev);
@@ -4885,7 +3351,7 @@ static int skd_pci_resume(struct pci_dev *pdev)
 
 	skdev = pci_get_drvdata(pdev);
 	if (!skdev) {
-		pr_err("%s: no device data for PCI\n", pci_name(pdev));
+		dev_err(&pdev->dev, "no device data for PCI\n");
 		return -1;
 	}
 
@@ -4903,15 +3369,14 @@ static int skd_pci_resume(struct pci_dev *pdev)
 	if (!rc) {
 		if (pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64))) {
 
-			pr_err("(%s): consistent DMA mask error %d\n",
-			       pci_name(pdev), rc);
+			dev_err(&pdev->dev, "consistent DMA mask error %d\n",
+				rc);
 		}
 	} else {
 		rc = pci_set_dma_mask(pdev, DMA_BIT_MASK(32));
 		if (rc) {
 
-			pr_err("(%s): DMA mask error %d\n",
-			       pci_name(pdev), rc);
+			dev_err(&pdev->dev, "DMA mask error %d\n", rc);
 			goto err_out_regions;
 		}
 	}
@@ -4919,8 +3384,8 @@ static int skd_pci_resume(struct pci_dev *pdev)
 	pci_set_master(pdev);
 	rc = pci_enable_pcie_error_reporting(pdev);
 	if (rc) {
-		pr_err("(%s): bad enable of PCIe error reporting rc=%d\n",
-		       skdev->name, rc);
+		dev_err(&pdev->dev,
+			"bad enable of PCIe error reporting rc=%d\n", rc);
 		skdev->pcie_error_reporting_is_enabled = 0;
 	} else
 		skdev->pcie_error_reporting_is_enabled = 1;
@@ -4932,21 +3397,17 @@ static int skd_pci_resume(struct pci_dev *pdev)
 		skdev->mem_map[i] = ioremap(skdev->mem_phys[i],
 					    skdev->mem_size[i]);
 		if (!skdev->mem_map[i]) {
-			pr_err("(%s): Unable to map adapter memory!\n",
-			       skd_name(skdev));
+			dev_err(&pdev->dev, "Unable to map adapter memory!\n");
 			rc = -ENODEV;
 			goto err_out_iounmap;
 		}
-		pr_debug("%s:%s:%d mem_map=%p, phyd=%016llx, size=%d\n",
-			 skdev->name, __func__, __LINE__,
-			 skdev->mem_map[i],
-			 (uint64_t)skdev->mem_phys[i], skdev->mem_size[i]);
+		dev_dbg(&pdev->dev, "mem_map=%p, phyd=%016llx, size=%d\n",
+			skdev->mem_map[i], (uint64_t)skdev->mem_phys[i],
+			skdev->mem_size[i]);
 	}
 	rc = skd_acquire_irq(skdev);
 	if (rc) {
-
-		pr_err("(%s): interrupt resource error %d\n",
-		       pci_name(pdev), rc);
+		dev_err(&pdev->dev, "interrupt resource error %d\n", rc);
 		goto err_out_iounmap;
 	}
 
@@ -4984,15 +3445,15 @@ static void skd_pci_shutdown(struct pci_dev *pdev)
 {
 	struct skd_device *skdev;
 
-	pr_err("skd_pci_shutdown called\n");
+	dev_err(&pdev->dev, "%s called\n", __func__);
 
 	skdev = pci_get_drvdata(pdev);
 	if (!skdev) {
-		pr_err("%s: no device data for PCI\n", pci_name(pdev));
+		dev_err(&pdev->dev, "no device data for PCI\n");
 		return;
 	}
 
-	pr_err("%s: calling stop\n", skd_name(skdev));
+	dev_err(&pdev->dev, "calling stop\n");
 	skd_stop_device(skdev);
 }
 
@@ -5012,21 +3473,6 @@ static struct pci_driver skd_driver = {
  *****************************************************************************
  */
 
-static const char *skd_name(struct skd_device *skdev)
-{
-	memset(skdev->id_str, 0, sizeof(skdev->id_str));
-
-	if (skdev->inquiry_is_valid)
-		snprintf(skdev->id_str, sizeof(skdev->id_str), "%s:%s:[%s]",
-			 skdev->name, skdev->inq_serial_num,
-			 pci_name(skdev->pdev));
-	else
-		snprintf(skdev->id_str, sizeof(skdev->id_str), "%s:??:[%s]",
-			 skdev->name, pci_name(skdev->pdev));
-
-	return skdev->id_str;
-}
-
 const char *skd_drive_state_to_str(int state)
 {
 	switch (state) {
@@ -5078,8 +3524,6 @@ const char *skd_skdev_state_to_str(enum skd_drvr_state state)
 		return "PAUSING";
 	case SKD_DRVR_STATE_PAUSED:
 		return "PAUSED";
-	case SKD_DRVR_STATE_DRAINING_TIMEOUT:
-		return "DRAINING_TIMEOUT";
 	case SKD_DRVR_STATE_RESTARTING:
 		return "RESTARTING";
 	case SKD_DRVR_STATE_RESUMING:
@@ -5106,18 +3550,6 @@ const char *skd_skdev_state_to_str(enum skd_drvr_state state)
 	}
 }
 
-static const char *skd_skmsg_state_to_str(enum skd_fit_msg_state state)
-{
-	switch (state) {
-	case SKD_MSG_STATE_IDLE:
-		return "IDLE";
-	case SKD_MSG_STATE_BUSY:
-		return "BUSY";
-	default:
-		return "???";
-	}
-}
-
 static const char *skd_skreq_state_to_str(enum skd_req_state state)
 {
 	switch (state) {
@@ -5131,8 +3563,6 @@ static const char *skd_skreq_state_to_str(enum skd_req_state state)
 		return "COMPLETED";
 	case SKD_REQ_STATE_TIMEOUT:
 		return "TIMEOUT";
-	case SKD_REQ_STATE_ABORTED:
-		return "ABORTED";
 	default:
 		return "???";
 	}
@@ -5140,58 +3570,34 @@ static const char *skd_skreq_state_to_str(enum skd_req_state state)
 
 static void skd_log_skdev(struct skd_device *skdev, const char *event)
 {
-	pr_debug("%s:%s:%d (%s) skdev=%p event='%s'\n",
-		 skdev->name, __func__, __LINE__, skdev->name, skdev, event);
-	pr_debug("%s:%s:%d   drive_state=%s(%d) driver_state=%s(%d)\n",
-		 skdev->name, __func__, __LINE__,
-		 skd_drive_state_to_str(skdev->drive_state), skdev->drive_state,
-		 skd_skdev_state_to_str(skdev->state), skdev->state);
-	pr_debug("%s:%s:%d   busy=%d limit=%d dev=%d lowat=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 skdev->in_flight, skdev->cur_max_queue_depth,
-		 skdev->dev_max_queue_depth, skdev->queue_low_water_mark);
-	pr_debug("%s:%s:%d   timestamp=0x%x cycle=%d cycle_ix=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 skdev->timeout_stamp, skdev->skcomp_cycle, skdev->skcomp_ix);
-}
-
-static void skd_log_skmsg(struct skd_device *skdev,
-			  struct skd_fitmsg_context *skmsg, const char *event)
-{
-	pr_debug("%s:%s:%d (%s) skmsg=%p event='%s'\n",
-		 skdev->name, __func__, __LINE__, skdev->name, skmsg, event);
-	pr_debug("%s:%s:%d   state=%s(%d) id=0x%04x length=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 skd_skmsg_state_to_str(skmsg->state), skmsg->state,
-		 skmsg->id, skmsg->length);
+	dev_dbg(&skdev->pdev->dev, "skdev=%p event='%s'\n", skdev, event);
+	dev_dbg(&skdev->pdev->dev, "  drive_state=%s(%d) driver_state=%s(%d)\n",
+		skd_drive_state_to_str(skdev->drive_state), skdev->drive_state,
+		skd_skdev_state_to_str(skdev->state), skdev->state);
+	dev_dbg(&skdev->pdev->dev, "  busy=%d limit=%d dev=%d lowat=%d\n",
+		skd_in_flight(skdev), skdev->cur_max_queue_depth,
+		skdev->dev_max_queue_depth, skdev->queue_low_water_mark);
+	dev_dbg(&skdev->pdev->dev, "  cycle=%d cycle_ix=%d\n",
+		skdev->skcomp_cycle, skdev->skcomp_ix);
 }
 
 static void skd_log_skreq(struct skd_device *skdev,
 			  struct skd_request_context *skreq, const char *event)
 {
-	pr_debug("%s:%s:%d (%s) skreq=%p event='%s'\n",
-		 skdev->name, __func__, __LINE__, skdev->name, skreq, event);
-	pr_debug("%s:%s:%d   state=%s(%d) id=0x%04x fitmsg=0x%04x\n",
-		 skdev->name, __func__, __LINE__,
-		 skd_skreq_state_to_str(skreq->state), skreq->state,
-		 skreq->id, skreq->fitmsg_id);
-	pr_debug("%s:%s:%d   timo=0x%x sg_dir=%d n_sg=%d\n",
-		 skdev->name, __func__, __LINE__,
-		 skreq->timeout_stamp, skreq->sg_data_dir, skreq->n_sg);
-
-	if (skreq->req != NULL) {
-		struct request *req = skreq->req;
-		u32 lba = (u32)blk_rq_pos(req);
-		u32 count = blk_rq_sectors(req);
-
-		pr_debug("%s:%s:%d "
-			 "req=%p lba=%u(0x%x) count=%u(0x%x) dir=%d\n",
-			 skdev->name, __func__, __LINE__,
-			 req, lba, lba, count, count,
-			 (int)rq_data_dir(req));
-	} else
-		pr_debug("%s:%s:%d req=NULL\n",
-			 skdev->name, __func__, __LINE__);
+	struct request *req = blk_mq_rq_from_pdu(skreq);
+	u32 lba = blk_rq_pos(req);
+	u32 count = blk_rq_sectors(req);
+
+	dev_dbg(&skdev->pdev->dev, "skreq=%p event='%s'\n", skreq, event);
+	dev_dbg(&skdev->pdev->dev, "  state=%s(%d) id=0x%04x fitmsg=0x%04x\n",
+		skd_skreq_state_to_str(skreq->state), skreq->state, skreq->id,
+		skreq->fitmsg_id);
+	dev_dbg(&skdev->pdev->dev, "  sg_dir=%d n_sg=%d\n",
+		skreq->data_dir, skreq->n_sg);
+
+	dev_dbg(&skdev->pdev->dev,
+		"req=%p lba=%u(0x%x) count=%u(0x%x) dir=%d\n", req, lba, lba,
+		count, count, (int)rq_data_dir(req));
 }
 
 /*
@@ -5202,7 +3608,14 @@ static void skd_log_skreq(struct skd_device *skdev,
 
 static int __init skd_init(void)
 {
-	pr_info(PFX " v%s-b%s loaded\n", DRV_VERSION, DRV_BUILD_ID);
+	BUILD_BUG_ON(sizeof(struct fit_completion_entry_v1) != 8);
+	BUILD_BUG_ON(sizeof(struct fit_comp_error_info) != 32);
+	BUILD_BUG_ON(sizeof(struct skd_command_header) != 16);
+	BUILD_BUG_ON(sizeof(struct skd_scsi_request) != 32);
+	BUILD_BUG_ON(sizeof(struct driver_inquiry_data) != 44);
+	BUILD_BUG_ON(offsetof(struct skd_msg_buf, fmh) != 0);
+	BUILD_BUG_ON(offsetof(struct skd_msg_buf, scsi) != 64);
+	BUILD_BUG_ON(sizeof(struct skd_msg_buf) != SKD_N_FITMSG_BYTES);
 
 	switch (skd_isr_type) {
 	case SKD_IRQ_LEGACY:
@@ -5222,7 +3635,8 @@ static int __init skd_init(void)
 		skd_max_queue_depth = SKD_MAX_QUEUE_DEPTH_DEFAULT;
 	}
 
-	if (skd_max_req_per_msg < 1 || skd_max_req_per_msg > 14) {
+	if (skd_max_req_per_msg < 1 ||
+	    skd_max_req_per_msg > SKD_MAX_REQ_PER_MSG) {
 		pr_err(PFX "skd_max_req_per_msg %d invalid, re-set to %d\n",
 		       skd_max_req_per_msg, SKD_MAX_REQ_PER_MSG_DEFAULT);
 		skd_max_req_per_msg = SKD_MAX_REQ_PER_MSG_DEFAULT;
@@ -5246,19 +3660,11 @@ static int __init skd_init(void)
 		skd_isr_comp_limit = 0;
 	}
 
-	if (skd_max_pass_thru < 1 || skd_max_pass_thru > 50) {
-		pr_err(PFX "skd_max_pass_thru %d invalid, re-set to %d\n",
-		       skd_max_pass_thru, SKD_N_SPECIAL_CONTEXT);
-		skd_max_pass_thru = SKD_N_SPECIAL_CONTEXT;
-	}
-
 	return pci_register_driver(&skd_driver);
 }
 
 static void __exit skd_exit(void)
 {
-	pr_info(PFX " v%s-b%s unloading\n", DRV_VERSION, DRV_BUILD_ID);
-
 	pci_unregister_driver(&skd_driver);
 
 	if (skd_major)

drivers/block/skd_s1120.h

@@ -1,19 +1,15 @@
-/* Copyright 2012 STEC, Inc.
+/*
+ * Copyright 2012 STEC, Inc.
+ * Copyright (c) 2017 Western Digital Corporation or its affiliates.
  *
- * This file is licensed under the terms of the 3-clause
- * BSD License (http://opensource.org/licenses/BSD-3-Clause)
- * or the GNU GPL-2.0 (http://www.gnu.org/licenses/gpl-2.0.html),
- * at your option. Both licenses are also available in the LICENSE file
- * distributed with this project. This file may not be copied, modified,
- * or distributed except in accordance with those terms.
+ * This file is part of the Linux kernel, and is made available under
+ * the terms of the GNU General Public License version 2.
  */
 
 
 #ifndef SKD_S1120_H
 #define SKD_S1120_H
 
-#pragma pack(push, s1120_h, 1)
-
 /*
  * Q-channel, 64-bit r/w
  */
@@ -30,7 +26,7 @@
 #define  FIT_QCMD_MSGSIZE_128		(0x1 << 4)
 #define  FIT_QCMD_MSGSIZE_256		(0x2 << 4)
 #define  FIT_QCMD_MSGSIZE_512		(0x3 << 4)
-#define  FIT_QCMD_BASE_ADDRESS_MASK	(0xFFFFFFFFFFFFFFC0ull)
+#define  FIT_QCMD_ALIGN			L1_CACHE_BYTES
 
 /*
  * Control, 32-bit r/w
@@ -250,7 +246,7 @@ struct fit_msg_hdr {
  *  20-23 of the FIT_MTD_FITFW_INIT response.
  */
 struct fit_completion_entry_v1 {
-	uint32_t	num_returned_bytes;
+	__be32		num_returned_bytes;
 	uint16_t	tag;
 	uint8_t		status;  /* SCSI status */
 	uint8_t		cycle;
@@ -278,7 +274,7 @@ struct fit_comp_error_info {
 	uint16_t	sks_low; /* 10: Sense Key Specific (LSW) */
 	uint16_t	reserved3; /* 12: Part of additional sense bytes (unused) */
 	uint16_t	uec; /* 14: Additional Sense Bytes */
-	uint64_t	per; /* 16: Additional Sense Bytes */
+	uint64_t	per __packed; /* 16: Additional Sense Bytes */
 	uint8_t		reserved4[2]; /* 1E: Additional Sense Bytes (unused) */
 };
 
@@ -292,11 +288,11 @@ struct fit_comp_error_info {
  * Version one has the last 32 bits sg_list_len_bytes;
  */
 struct skd_command_header {
-	uint64_t	sg_list_dma_address;
+	__be64		sg_list_dma_address;
 	uint16_t	tag;
 	uint8_t		attribute;
 	uint8_t		add_cdb_len;     /* In 32 bit words */
-	uint32_t	sg_list_len_bytes;
+	__be32		sg_list_len_bytes;
 };
 
 struct skd_scsi_request {
@@ -309,22 +305,20 @@ struct driver_inquiry_data {
 	uint8_t		peripheral_device_type:5;
 	uint8_t		qualifier:3;
 	uint8_t		page_code;
-	uint16_t	page_length;
-	uint16_t	pcie_bus_number;
+	__be16		page_length;
+	__be16		pcie_bus_number;
 	uint8_t		pcie_device_number;
 	uint8_t		pcie_function_number;
 	uint8_t		pcie_link_speed;
 	uint8_t		pcie_link_lanes;
-	uint16_t	pcie_vendor_id;
-	uint16_t	pcie_device_id;
-	uint16_t	pcie_subsystem_vendor_id;
-	uint16_t	pcie_subsystem_device_id;
+	__be16		pcie_vendor_id;
+	__be16		pcie_device_id;
+	__be16		pcie_subsystem_vendor_id;
+	__be16		pcie_subsystem_device_id;
 	uint8_t		reserved1[2];
 	uint8_t		reserved2[3];
 	uint8_t		driver_version_length;
 	uint8_t		driver_version[0x14];
 };
 
-#pragma pack(pop, s1120_h)
-
 #endif /* SKD_S1120_H */

drivers/block/virtio_blk.c

@@ -265,7 +265,7 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 	}
 
 	spin_lock_irqsave(&vblk->vqs[qid].lock, flags);
-	if (req_op(req) == REQ_OP_SCSI_IN || req_op(req) == REQ_OP_SCSI_OUT)
+	if (blk_rq_is_scsi(req))
 		err = virtblk_add_req_scsi(vblk->vqs[qid].vq, vbr, vbr->sg, num);
 	else
 		err = virtblk_add_req(vblk->vqs[qid].vq, vbr, vbr->sg, num);

drivers/block/xen-blkback/blkback.c

@@ -705,9 +705,9 @@ static unsigned int xen_blkbk_unmap_prepare(
 				    GNTMAP_host_map, pages[i]->handle);
 		pages[i]->handle = BLKBACK_INVALID_HANDLE;
 		invcount++;
-       }
+	}
 
-       return invcount;
+	return invcount;
 }
 
 static void xen_blkbk_unmap_and_respond_callback(int result, struct gntab_unmap_queue_data *data)
@@ -1251,6 +1251,7 @@ static int dispatch_rw_block_io(struct xen_blkif_ring *ring,
 		break;
 	case BLKIF_OP_WRITE_BARRIER:
 		drain = true;
+		/* fall through */
 	case BLKIF_OP_FLUSH_DISKCACHE:
 		ring->st_f_req++;
 		operation = REQ_OP_WRITE;
@@ -1362,7 +1363,7 @@ static int dispatch_rw_block_io(struct xen_blkif_ring *ring,
 				goto fail_put_bio;
 
 			biolist[nbio++] = bio;
-			bio->bi_bdev    = preq.bdev;
+			bio_set_dev(bio, preq.bdev);
 			bio->bi_private = pending_req;
 			bio->bi_end_io  = end_block_io_op;
 			bio->bi_iter.bi_sector  = preq.sector_number;
@@ -1381,7 +1382,7 @@ static int dispatch_rw_block_io(struct xen_blkif_ring *ring,
 			goto fail_put_bio;
 
 		biolist[nbio++] = bio;
-		bio->bi_bdev    = preq.bdev;
+		bio_set_dev(bio, preq.bdev);
 		bio->bi_private = pending_req;
 		bio->bi_end_io  = end_block_io_op;
 		bio_set_op_attrs(bio, operation, operation_flags);

drivers/block/xen-blkback/xenbus.c

@@ -816,7 +816,8 @@ static void frontend_changed(struct xenbus_device *dev,
 		xenbus_switch_state(dev, XenbusStateClosed);
 		if (xenbus_dev_is_online(dev))
 			break;
-		/* fall through if not online */
+		/* fall through */
+		/* if not online */
 	case XenbusStateUnknown:
 		/* implies xen_blkif_disconnect() via xen_blkbk_remove() */
 		device_unregister(&dev->dev);

drivers/block/xen-blkfront.c

@@ -2456,7 +2456,7 @@ static void blkback_changed(struct xenbus_device *dev,
 	case XenbusStateClosed:
 		if (dev->state == XenbusStateClosed)
 			break;
-		/* Missed the backend's Closing state -- fallthrough */
+		/* fall through */
 	case XenbusStateClosing:
 		if (info)
 			blkfront_closing(info);

drivers/block/zram/zram_drv.c

@@ -467,7 +467,7 @@ static int read_from_bdev_async(struct zram *zram, struct bio_vec *bvec,
 		return -ENOMEM;
 
 	bio->bi_iter.bi_sector = entry * (PAGE_SIZE >> 9);
-	bio->bi_bdev = zram->bdev;
+	bio_set_dev(bio, zram->bdev);
 	if (!bio_add_page(bio, bvec->bv_page, bvec->bv_len, bvec->bv_offset)) {
 		bio_put(bio);
 		return -EIO;
@@ -561,7 +561,7 @@ static int write_to_bdev(struct zram *zram, struct bio_vec *bvec,
 	}
 
 	bio->bi_iter.bi_sector = entry * (PAGE_SIZE >> 9);
-	bio->bi_bdev = zram->bdev;
+	bio_set_dev(bio, zram->bdev);
 	if (!bio_add_page(bio, bvec->bv_page, bvec->bv_len,
 					bvec->bv_offset)) {
 		bio_put(bio);
@@ -1171,9 +1171,10 @@ static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
 {
 	unsigned long start_time = jiffies;
 	int rw_acct = is_write ? REQ_OP_WRITE : REQ_OP_READ;
+	struct request_queue *q = zram->disk->queue;
 	int ret;
 
-	generic_start_io_acct(rw_acct, bvec->bv_len >> SECTOR_SHIFT,
+	generic_start_io_acct(q, rw_acct, bvec->bv_len >> SECTOR_SHIFT,
 			&zram->disk->part0);
 
 	if (!is_write) {
@@ -1185,7 +1186,7 @@ static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
 		ret = zram_bvec_write(zram, bvec, index, offset, bio);
 	}
 
-	generic_end_io_acct(rw_acct, &zram->disk->part0, start_time);
+	generic_end_io_acct(q, rw_acct, &zram->disk->part0, start_time);
 
 	if (unlikely(ret < 0)) {
 		if (!is_write)

drivers/ide/ide-floppy.c

@@ -72,7 +72,7 @@ static int ide_floppy_callback(ide_drive_t *drive, int dsc)
 		drive->failed_pc = NULL;
 
 	if (pc->c[0] == GPCMD_READ_10 || pc->c[0] == GPCMD_WRITE_10 ||
-	    (req_op(rq) == REQ_OP_SCSI_IN || req_op(rq) == REQ_OP_SCSI_OUT))
+	    blk_rq_is_scsi(rq))
 		uptodate = 1; /* FIXME */
 	else if (pc->c[0] == GPCMD_REQUEST_SENSE) {
 

drivers/md/bcache/debug.c

@@ -49,7 +49,7 @@ void bch_btree_verify(struct btree *b)
 	v->keys.ops = b->keys.ops;
 
 	bio = bch_bbio_alloc(b->c);
-	bio->bi_bdev		= PTR_CACHE(b->c, &b->key, 0)->bdev;
+	bio_set_dev(bio, PTR_CACHE(b->c, &b->key, 0)->bdev);
 	bio->bi_iter.bi_sector	= PTR_OFFSET(&b->key, 0);
 	bio->bi_iter.bi_size	= KEY_SIZE(&v->key) << 9;
 	bio->bi_opf		= REQ_OP_READ | REQ_META;

drivers/md/bcache/io.c

@@ -34,7 +34,7 @@ void __bch_submit_bbio(struct bio *bio, struct cache_set *c)
 	struct bbio *b = container_of(bio, struct bbio, bio);
 
 	bio->bi_iter.bi_sector	= PTR_OFFSET(&b->key, 0);
-	bio->bi_bdev		= PTR_CACHE(c, &b->key, 0)->bdev;
+	bio_set_dev(bio, PTR_CACHE(c, &b->key, 0)->bdev);
 
 	b->submit_time_us = local_clock_us();
 	closure_bio_submit(bio, bio->bi_private);

drivers/md/bcache/journal.c

@@ -53,7 +53,7 @@ reread:		left = ca->sb.bucket_size - offset;
 
 		bio_reset(bio);
 		bio->bi_iter.bi_sector	= bucket + offset;
-		bio->bi_bdev	= ca->bdev;
+		bio_set_dev(bio, ca->bdev);
 		bio->bi_iter.bi_size	= len << 9;
 
 		bio->bi_end_io	= journal_read_endio;
@@ -452,7 +452,7 @@ static void do_journal_discard(struct cache *ca)
 		bio_set_op_attrs(bio, REQ_OP_DISCARD, 0);
 		bio->bi_iter.bi_sector	= bucket_to_sector(ca->set,
 						ca->sb.d[ja->discard_idx]);
-		bio->bi_bdev		= ca->bdev;
+		bio_set_dev(bio, ca->bdev);
 		bio->bi_iter.bi_size	= bucket_bytes(ca);
 		bio->bi_end_io		= journal_discard_endio;
 
@@ -623,7 +623,7 @@ static void journal_write_unlocked(struct closure *cl)
 
 		bio_reset(bio);
 		bio->bi_iter.bi_sector	= PTR_OFFSET(k, i);
-		bio->bi_bdev	= ca->bdev;
+		bio_set_dev(bio, ca->bdev);
 		bio->bi_iter.bi_size = sectors << 9;
 
 		bio->bi_end_io	= journal_write_endio;

drivers/md/bcache/request.c

@@ -607,7 +607,8 @@ static void request_endio(struct bio *bio)
 static void bio_complete(struct search *s)
 {
 	if (s->orig_bio) {
-		generic_end_io_acct(bio_data_dir(s->orig_bio),
+		struct request_queue *q = s->orig_bio->bi_disk->queue;
+		generic_end_io_acct(q, bio_data_dir(s->orig_bio),
 				    &s->d->disk->part0, s->start_time);
 
 		trace_bcache_request_end(s->d, s->orig_bio);
@@ -734,7 +735,7 @@ static void cached_dev_read_done(struct closure *cl)
 	if (s->iop.bio) {
 		bio_reset(s->iop.bio);
 		s->iop.bio->bi_iter.bi_sector = s->cache_miss->bi_iter.bi_sector;
-		s->iop.bio->bi_bdev = s->cache_miss->bi_bdev;
+		bio_copy_dev(s->iop.bio, s->cache_miss);
 		s->iop.bio->bi_iter.bi_size = s->insert_bio_sectors << 9;
 		bch_bio_map(s->iop.bio, NULL);
 
@@ -793,7 +794,7 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 	    !(bio->bi_opf & REQ_META) &&
 	    s->iop.c->gc_stats.in_use < CUTOFF_CACHE_READA)
 		reada = min_t(sector_t, dc->readahead >> 9,
-			      bdev_sectors(bio->bi_bdev) - bio_end_sector(bio));
+			      get_capacity(bio->bi_disk) - bio_end_sector(bio));
 
 	s->insert_bio_sectors = min(sectors, bio_sectors(bio) + reada);
 
@@ -819,7 +820,7 @@ static int cached_dev_cache_miss(struct btree *b, struct search *s,
 		goto out_submit;
 
 	cache_bio->bi_iter.bi_sector	= miss->bi_iter.bi_sector;
-	cache_bio->bi_bdev		= miss->bi_bdev;
+	bio_copy_dev(cache_bio, miss);
 	cache_bio->bi_iter.bi_size	= s->insert_bio_sectors << 9;
 
 	cache_bio->bi_end_io	= request_endio;
@@ -918,7 +919,7 @@ static void cached_dev_write(struct cached_dev *dc, struct search *s)
 			struct bio *flush = bio_alloc_bioset(GFP_NOIO, 0,
 							     dc->disk.bio_split);
 
-			flush->bi_bdev	= bio->bi_bdev;
+			bio_copy_dev(flush, bio);
 			flush->bi_end_io = request_endio;
 			flush->bi_private = cl;
 			flush->bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
@@ -955,13 +956,13 @@ static blk_qc_t cached_dev_make_request(struct request_queue *q,
 					struct bio *bio)
 {
 	struct search *s;
-	struct bcache_device *d = bio->bi_bdev->bd_disk->private_data;
+	struct bcache_device *d = bio->bi_disk->private_data;
 	struct cached_dev *dc = container_of(d, struct cached_dev, disk);
 	int rw = bio_data_dir(bio);
 
-	generic_start_io_acct(rw, bio_sectors(bio), &d->disk->part0);
+	generic_start_io_acct(q, rw, bio_sectors(bio), &d->disk->part0);
 
-	bio->bi_bdev = dc->bdev;
+	bio_set_dev(bio, dc->bdev);
 	bio->bi_iter.bi_sector += dc->sb.data_offset;
 
 	if (cached_dev_get(dc)) {
@@ -1071,10 +1072,10 @@ static blk_qc_t flash_dev_make_request(struct request_queue *q,
 {
 	struct search *s;
 	struct closure *cl;
-	struct bcache_device *d = bio->bi_bdev->bd_disk->private_data;
+	struct bcache_device *d = bio->bi_disk->private_data;
 	int rw = bio_data_dir(bio);
 
-	generic_start_io_acct(rw, bio_sectors(bio), &d->disk->part0);
+	generic_start_io_acct(q, rw, bio_sectors(bio), &d->disk->part0);
 
 	s = search_alloc(bio, d);
 	cl = &s->cl;

drivers/md/bcache/super.c

@@ -257,7 +257,7 @@ void bch_write_bdev_super(struct cached_dev *dc, struct closure *parent)
 	closure_init(cl, parent);
 
 	bio_reset(bio);
-	bio->bi_bdev	= dc->bdev;
+	bio_set_dev(bio, dc->bdev);
 	bio->bi_end_io	= write_bdev_super_endio;
 	bio->bi_private = dc;
 
@@ -303,7 +303,7 @@ void bcache_write_super(struct cache_set *c)
 		SET_CACHE_SYNC(&ca->sb, CACHE_SYNC(&c->sb));
 
 		bio_reset(bio);
-		bio->bi_bdev	= ca->bdev;
+		bio_set_dev(bio, ca->bdev);
 		bio->bi_end_io	= write_super_endio;
 		bio->bi_private = ca;
 
@@ -508,7 +508,7 @@ static void prio_io(struct cache *ca, uint64_t bucket, int op,
 	closure_init_stack(cl);
 
 	bio->bi_iter.bi_sector	= bucket * ca->sb.bucket_size;
-	bio->bi_bdev		= ca->bdev;
+	bio_set_dev(bio, ca->bdev);
 	bio->bi_iter.bi_size	= bucket_bytes(ca);
 
 	bio->bi_end_io	= prio_endio;

drivers/md/bcache/writeback.c

@@ -181,7 +181,7 @@ static void write_dirty(struct closure *cl)
 	dirty_init(w);
 	bio_set_op_attrs(&io->bio, REQ_OP_WRITE, 0);
 	io->bio.bi_iter.bi_sector = KEY_START(&w->key);
-	io->bio.bi_bdev		= io->dc->bdev;
+	bio_set_dev(&io->bio, io->dc->bdev);
 	io->bio.bi_end_io	= dirty_endio;
 
 	closure_bio_submit(&io->bio, cl);
@@ -250,8 +250,7 @@ static void read_dirty(struct cached_dev *dc)
 		dirty_init(w);
 		bio_set_op_attrs(&io->bio, REQ_OP_READ, 0);
 		io->bio.bi_iter.bi_sector = PTR_OFFSET(&w->key, 0);
-		io->bio.bi_bdev		= PTR_CACHE(dc->disk.c,
-						    &w->key, 0)->bdev;
+		bio_set_dev(&io->bio, PTR_CACHE(dc->disk.c, &w->key, 0)->bdev);
 		io->bio.bi_end_io	= read_dirty_endio;
 
 		if (bio_alloc_pages(&io->bio, GFP_KERNEL))

drivers/md/dm-bio-record.h

@@ -18,21 +18,24 @@
  */
 
 struct dm_bio_details {
-	struct block_device *bi_bdev;
+	struct gendisk *bi_disk;
+	u8 bi_partno;
 	unsigned long bi_flags;
 	struct bvec_iter bi_iter;
 };
 
 static inline void dm_bio_record(struct dm_bio_details *bd, struct bio *bio)
 {
-	bd->bi_bdev = bio->bi_bdev;
+	bd->bi_disk = bio->bi_disk;
+	bd->bi_partno = bio->bi_partno;
 	bd->bi_flags = bio->bi_flags;
 	bd->bi_iter = bio->bi_iter;
 }
 
 static inline void dm_bio_restore(struct dm_bio_details *bd, struct bio *bio)
 {
-	bio->bi_bdev = bd->bi_bdev;
+	bio->bi_disk = bd->bi_disk;
+	bio->bi_partno = bd->bi_partno;
 	bio->bi_flags = bd->bi_flags;
 	bio->bi_iter = bd->bi_iter;
 }

drivers/md/dm-bufio.c

@@ -616,7 +616,7 @@ static void use_inline_bio(struct dm_buffer *b, int rw, sector_t sector,
 
 	bio_init(&b->bio, b->bio_vec, DM_BUFIO_INLINE_VECS);
 	b->bio.bi_iter.bi_sector = sector;
-	b->bio.bi_bdev = b->c->bdev;
+	bio_set_dev(&b->bio, b->c->bdev);
 	b->bio.bi_end_io = inline_endio;
 	/*
 	 * Use of .bi_private isn't a problem here because

drivers/md/dm-cache-target.c

@@ -833,7 +833,7 @@ static bool is_discarded_oblock(struct cache *cache, dm_oblock_t b)
  *--------------------------------------------------------------*/
 static void remap_to_origin(struct cache *cache, struct bio *bio)
 {
-	bio->bi_bdev = cache->origin_dev->bdev;
+	bio_set_dev(bio, cache->origin_dev->bdev);
 }
 
 static void remap_to_cache(struct cache *cache, struct bio *bio,
@@ -842,7 +842,7 @@ static void remap_to_cache(struct cache *cache, struct bio *bio,
 	sector_t bi_sector = bio->bi_iter.bi_sector;
 	sector_t block = from_cblock(cblock);
 
-	bio->bi_bdev = cache->cache_dev->bdev;
+	bio_set_dev(bio, cache->cache_dev->bdev);
 	if (!block_size_is_power_of_two(cache))
 		bio->bi_iter.bi_sector =
 			(block * cache->sectors_per_block) +

drivers/md/dm-crypt.c

@@ -932,9 +932,6 @@ static int dm_crypt_integrity_io_alloc(struct dm_crypt_io *io, struct bio *bio)
 	bip->bip_iter.bi_size = tag_len;
 	bip->bip_iter.bi_sector = io->cc->start + io->sector;
 
-	/* We own the metadata, do not let bio_free to release it */
-	bip->bip_flags &= ~BIP_BLOCK_INTEGRITY;
-
 	ret = bio_integrity_add_page(bio, virt_to_page(io->integrity_metadata),
 				     tag_len, offset_in_page(io->integrity_metadata));
 	if (unlikely(ret != tag_len))
@@ -1546,7 +1543,7 @@ static void clone_init(struct dm_crypt_io *io, struct bio *clone)
 
 	clone->bi_private = io;
 	clone->bi_end_io  = crypt_endio;
-	clone->bi_bdev    = cc->dev->bdev;
+	bio_set_dev(clone, cc->dev->bdev);
 	clone->bi_opf	  = io->base_bio->bi_opf;
 }
 
@@ -2795,7 +2792,7 @@ static int crypt_map(struct dm_target *ti, struct bio *bio)
 	 */
 	if (unlikely(bio->bi_opf & REQ_PREFLUSH ||
 	    bio_op(bio) == REQ_OP_DISCARD)) {
-		bio->bi_bdev = cc->dev->bdev;
+		bio_set_dev(bio, cc->dev->bdev);
 		if (bio_sectors(bio))
 			bio->bi_iter.bi_sector = cc->start +
 				dm_target_offset(ti, bio->bi_iter.bi_sector);

drivers/md/dm-delay.c

@@ -282,7 +282,7 @@ static int delay_map(struct dm_target *ti, struct bio *bio)
 	struct delay_c *dc = ti->private;
 
 	if ((bio_data_dir(bio) == WRITE) && (dc->dev_write)) {
-		bio->bi_bdev = dc->dev_write->bdev;
+		bio_set_dev(bio, dc->dev_write->bdev);
 		if (bio_sectors(bio))
 			bio->bi_iter.bi_sector = dc->start_write +
 				dm_target_offset(ti, bio->bi_iter.bi_sector);
@@ -290,7 +290,7 @@ static int delay_map(struct dm_target *ti, struct bio *bio)
 		return delay_bio(dc, dc->write_delay, bio);
 	}
 
-	bio->bi_bdev = dc->dev_read->bdev;
+	bio_set_dev(bio, dc->dev_read->bdev);
 	bio->bi_iter.bi_sector = dc->start_read +
 		dm_target_offset(ti, bio->bi_iter.bi_sector);
 

drivers/md/dm-era-target.c

@@ -1192,7 +1192,7 @@ static dm_block_t get_block(struct era *era, struct bio *bio)
 
 static void remap_to_origin(struct era *era, struct bio *bio)
 {
-	bio->bi_bdev = era->origin_dev->bdev;
+	bio_set_dev(bio, era->origin_dev->bdev);
 }
 
 /*----------------------------------------------------------------

drivers/md/dm-flakey.c

@@ -274,7 +274,7 @@ static void flakey_map_bio(struct dm_target *ti, struct bio *bio)
 {
 	struct flakey_c *fc = ti->private;
 
-	bio->bi_bdev = fc->dev->bdev;
+	bio_set_dev(bio, fc->dev->bdev);
 	if (bio_sectors(bio) || bio_op(bio) == REQ_OP_ZONE_RESET)
 		bio->bi_iter.bi_sector =
 			flakey_map_sector(ti, bio->bi_iter.bi_sector);

drivers/md/dm-integrity.c

@@ -250,7 +250,8 @@ struct dm_integrity_io {
 
 	struct completion *completion;
 
-	struct block_device *orig_bi_bdev;
+	struct gendisk *orig_bi_disk;
+	u8 orig_bi_partno;
 	bio_end_io_t *orig_bi_end_io;
 	struct bio_integrity_payload *orig_bi_integrity;
 	struct bvec_iter orig_bi_iter;
@@ -1164,7 +1165,8 @@ static void integrity_end_io(struct bio *bio)
 	struct dm_integrity_io *dio = dm_per_bio_data(bio, sizeof(struct dm_integrity_io));
 
 	bio->bi_iter = dio->orig_bi_iter;
-	bio->bi_bdev = dio->orig_bi_bdev;
+	bio->bi_disk = dio->orig_bi_disk;
+	bio->bi_partno = dio->orig_bi_partno;
 	if (dio->orig_bi_integrity) {
 		bio->bi_integrity = dio->orig_bi_integrity;
 		bio->bi_opf |= REQ_INTEGRITY;
@@ -1681,8 +1683,9 @@ sleep:
 
 	dio->orig_bi_iter = bio->bi_iter;
 
-	dio->orig_bi_bdev = bio->bi_bdev;
-	bio->bi_bdev = ic->dev->bdev;
+	dio->orig_bi_disk = bio->bi_disk;
+	dio->orig_bi_partno = bio->bi_partno;
+	bio_set_dev(bio, ic->dev->bdev);
 
 	dio->orig_bi_integrity = bio_integrity(bio);
 	bio->bi_integrity = NULL;

drivers/md/dm-io.c

@@ -347,7 +347,7 @@ static void do_region(int op, int op_flags, unsigned region,
 
 		bio = bio_alloc_bioset(GFP_NOIO, num_bvecs, io->client->bios);
 		bio->bi_iter.bi_sector = where->sector + (where->count - remaining);
-		bio->bi_bdev = where->bdev;
+		bio_set_dev(bio, where->bdev);
 		bio->bi_end_io = endio;
 		bio_set_op_attrs(bio, op, op_flags);
 		store_io_and_region_in_bio(bio, io, region);

drivers/md/dm-linear.c

@@ -88,7 +88,7 @@ static void linear_map_bio(struct dm_target *ti, struct bio *bio)
 {
 	struct linear_c *lc = ti->private;
 
-	bio->bi_bdev = lc->dev->bdev;
+	bio_set_dev(bio, lc->dev->bdev);
 	if (bio_sectors(bio) || bio_op(bio) == REQ_OP_ZONE_RESET)
 		bio->bi_iter.bi_sector =
 			linear_map_sector(ti, bio->bi_iter.bi_sector);

drivers/md/dm-log-writes.c

@@ -198,7 +198,7 @@ static int write_metadata(struct log_writes_c *lc, void *entry,
 	}
 	bio->bi_iter.bi_size = 0;
 	bio->bi_iter.bi_sector = sector;
-	bio->bi_bdev = lc->logdev->bdev;
+	bio_set_dev(bio, lc->logdev->bdev);
 	bio->bi_end_io = log_end_io;
 	bio->bi_private = lc;
 	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
@@ -263,7 +263,7 @@ static int log_one_block(struct log_writes_c *lc,
 	}
 	bio->bi_iter.bi_size = 0;
 	bio->bi_iter.bi_sector = sector;
-	bio->bi_bdev = lc->logdev->bdev;
+	bio_set_dev(bio, lc->logdev->bdev);
 	bio->bi_end_io = log_end_io;
 	bio->bi_private = lc;
 	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
@@ -285,7 +285,7 @@ static int log_one_block(struct log_writes_c *lc,
 			}
 			bio->bi_iter.bi_size = 0;
 			bio->bi_iter.bi_sector = sector;
-			bio->bi_bdev = lc->logdev->bdev;
+			bio_set_dev(bio, lc->logdev->bdev);
 			bio->bi_end_io = log_end_io;
 			bio->bi_private = lc;
 			bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
@@ -539,7 +539,7 @@ static void normal_map_bio(struct dm_target *ti, struct bio *bio)
 {
 	struct log_writes_c *lc = ti->private;
 
-	bio->bi_bdev = lc->dev->bdev;
+	bio_set_dev(bio, lc->dev->bdev);
 }
 
 static int log_writes_map(struct dm_target *ti, struct bio *bio)

drivers/md/dm-mpath.c

@@ -565,7 +565,7 @@ static int __multipath_map_bio(struct multipath *m, struct bio *bio, struct dm_m
 	mpio->nr_bytes = nr_bytes;
 
 	bio->bi_status = 0;
-	bio->bi_bdev = pgpath->path.dev->bdev;
+	bio_set_dev(bio, pgpath->path.dev->bdev);
 	bio->bi_opf |= REQ_FAILFAST_TRANSPORT;
 
 	if (pgpath->pg->ps.type->start_io)

drivers/md/dm-raid1.c

@@ -145,7 +145,7 @@ static void dispatch_bios(void *context, struct bio_list *bio_list)
 
 struct dm_raid1_bio_record {
 	struct mirror *m;
-	/* if details->bi_bdev == NULL, details were not saved */
+	/* if details->bi_disk == NULL, details were not saved */
 	struct dm_bio_details details;
 	region_t write_region;
 };
@@ -464,7 +464,7 @@ static sector_t map_sector(struct mirror *m, struct bio *bio)
 
 static void map_bio(struct mirror *m, struct bio *bio)
 {
-	bio->bi_bdev = m->dev->bdev;
+	bio_set_dev(bio, m->dev->bdev);
 	bio->bi_iter.bi_sector = map_sector(m, bio);
 }
 
@@ -1199,7 +1199,7 @@ static int mirror_map(struct dm_target *ti, struct bio *bio)
 	struct dm_raid1_bio_record *bio_record =
 	  dm_per_bio_data(bio, sizeof(struct dm_raid1_bio_record));
 
-	bio_record->details.bi_bdev = NULL;
+	bio_record->details.bi_disk = NULL;
 
 	if (rw == WRITE) {
 		/* Save region for mirror_end_io() handler */
@@ -1266,7 +1266,7 @@ static int mirror_end_io(struct dm_target *ti, struct bio *bio,
 		goto out;
 
 	if (unlikely(*error)) {
-		if (!bio_record->details.bi_bdev) {
+		if (!bio_record->details.bi_disk) {
 			/*
 			 * There wasn't enough memory to record necessary
 			 * information for a retry or there was no other
@@ -1291,7 +1291,7 @@ static int mirror_end_io(struct dm_target *ti, struct bio *bio,
 			bd = &bio_record->details;
 
 			dm_bio_restore(bd, bio);
-			bio_record->details.bi_bdev = NULL;
+			bio_record->details.bi_disk = NULL;
 			bio->bi_status = 0;
 
 			queue_bio(ms, bio, rw);
@@ -1301,7 +1301,7 @@ static int mirror_end_io(struct dm_target *ti, struct bio *bio,
 	}
 
 out:
-	bio_record->details.bi_bdev = NULL;
+	bio_record->details.bi_disk = NULL;
 
 	return DM_ENDIO_DONE;
 }

drivers/md/dm-snap.c

@@ -1663,7 +1663,7 @@ __find_pending_exception(struct dm_snapshot *s,
 static void remap_exception(struct dm_snapshot *s, struct dm_exception *e,
 			    struct bio *bio, chunk_t chunk)
 {
-	bio->bi_bdev = s->cow->bdev;
+	bio_set_dev(bio, s->cow->bdev);
 	bio->bi_iter.bi_sector =
 		chunk_to_sector(s->store, dm_chunk_number(e->new_chunk) +
 				(chunk - e->old_chunk)) +
@@ -1681,7 +1681,7 @@ static int snapshot_map(struct dm_target *ti, struct bio *bio)
 	init_tracked_chunk(bio);
 
 	if (bio->bi_opf & REQ_PREFLUSH) {
-		bio->bi_bdev = s->cow->bdev;
+		bio_set_dev(bio, s->cow->bdev);
 		return DM_MAPIO_REMAPPED;
 	}
 
@@ -1769,7 +1769,7 @@ static int snapshot_map(struct dm_target *ti, struct bio *bio)
 			goto out;
 		}
 	} else {
-		bio->bi_bdev = s->origin->bdev;
+		bio_set_dev(bio, s->origin->bdev);
 		track_chunk(s, bio, chunk);
 	}
 
@@ -1802,9 +1802,9 @@ static int snapshot_merge_map(struct dm_target *ti, struct bio *bio)
 
 	if (bio->bi_opf & REQ_PREFLUSH) {
 		if (!dm_bio_get_target_bio_nr(bio))
-			bio->bi_bdev = s->origin->bdev;
+			bio_set_dev(bio, s->origin->bdev);
 		else
-			bio->bi_bdev = s->cow->bdev;
+			bio_set_dev(bio, s->cow->bdev);
 		return DM_MAPIO_REMAPPED;
 	}
 
@@ -1824,7 +1824,7 @@ static int snapshot_merge_map(struct dm_target *ti, struct bio *bio)
 		    chunk >= s->first_merging_chunk &&
 		    chunk < (s->first_merging_chunk +
 			     s->num_merging_chunks)) {
-			bio->bi_bdev = s->origin->bdev;
+			bio_set_dev(bio, s->origin->bdev);
 			bio_list_add(&s->bios_queued_during_merge, bio);
 			r = DM_MAPIO_SUBMITTED;
 			goto out_unlock;
@@ -1838,7 +1838,7 @@ static int snapshot_merge_map(struct dm_target *ti, struct bio *bio)
 	}
 
 redirect_to_origin:
-	bio->bi_bdev = s->origin->bdev;
+	bio_set_dev(bio, s->origin->bdev);
 
 	if (bio_data_dir(bio) == WRITE) {
 		up_write(&s->lock);
@@ -2285,7 +2285,7 @@ static int origin_map(struct dm_target *ti, struct bio *bio)
 	struct dm_origin *o = ti->private;
 	unsigned available_sectors;
 
-	bio->bi_bdev = o->dev->bdev;
+	bio_set_dev(bio, o->dev->bdev);
 
 	if (unlikely(bio->bi_opf & REQ_PREFLUSH))
 		return DM_MAPIO_REMAPPED;

drivers/md/dm-stripe.c

@@ -270,7 +270,7 @@ static int stripe_map_range(struct stripe_c *sc, struct bio *bio,
 	stripe_map_range_sector(sc, bio_end_sector(bio),
 				target_stripe, &end);
 	if (begin < end) {
-		bio->bi_bdev = sc->stripe[target_stripe].dev->bdev;
+		bio_set_dev(bio, sc->stripe[target_stripe].dev->bdev);
 		bio->bi_iter.bi_sector = begin +
 			sc->stripe[target_stripe].physical_start;
 		bio->bi_iter.bi_size = to_bytes(end - begin);
@@ -291,7 +291,7 @@ static int stripe_map(struct dm_target *ti, struct bio *bio)
 	if (bio->bi_opf & REQ_PREFLUSH) {
 		target_bio_nr = dm_bio_get_target_bio_nr(bio);
 		BUG_ON(target_bio_nr >= sc->stripes);
-		bio->bi_bdev = sc->stripe[target_bio_nr].dev->bdev;
+		bio_set_dev(bio, sc->stripe[target_bio_nr].dev->bdev);
 		return DM_MAPIO_REMAPPED;
 	}
 	if (unlikely(bio_op(bio) == REQ_OP_DISCARD) ||
@@ -306,7 +306,7 @@ static int stripe_map(struct dm_target *ti, struct bio *bio)
 			  &stripe, &bio->bi_iter.bi_sector);
 
 	bio->bi_iter.bi_sector += sc->stripe[stripe].physical_start;
-	bio->bi_bdev = sc->stripe[stripe].dev->bdev;
+	bio_set_dev(bio, sc->stripe[stripe].dev->bdev);
 
 	return DM_MAPIO_REMAPPED;
 }
@@ -430,9 +430,7 @@ static int stripe_end_io(struct dm_target *ti, struct bio *bio,
 		return DM_ENDIO_DONE;
 
 	memset(major_minor, 0, sizeof(major_minor));
-	sprintf(major_minor, "%d:%d",
-		MAJOR(disk_devt(bio->bi_bdev->bd_disk)),
-		MINOR(disk_devt(bio->bi_bdev->bd_disk)));
+	sprintf(major_minor, "%d:%d", MAJOR(bio_dev(bio)), MINOR(bio_dev(bio)));
 
 	/*
 	 * Test to see which stripe drive triggered the event

drivers/md/dm-switch.c

@@ -322,7 +322,7 @@ static int switch_map(struct dm_target *ti, struct bio *bio)
 	sector_t offset = dm_target_offset(ti, bio->bi_iter.bi_sector);
 	unsigned path_nr = switch_get_path_nr(sctx, offset);
 
-	bio->bi_bdev = sctx->path_list[path_nr].dmdev->bdev;
+	bio_set_dev(bio, sctx->path_list[path_nr].dmdev->bdev);
 	bio->bi_iter.bi_sector = sctx->path_list[path_nr].start + offset;
 
 	return DM_MAPIO_REMAPPED;

drivers/md/dm-thin.c

@@ -679,7 +679,7 @@ static void remap(struct thin_c *tc, struct bio *bio, dm_block_t block)
 	struct pool *pool = tc->pool;
 	sector_t bi_sector = bio->bi_iter.bi_sector;
 
-	bio->bi_bdev = tc->pool_dev->bdev;
+	bio_set_dev(bio, tc->pool_dev->bdev);
 	if (block_size_is_power_of_two(pool))
 		bio->bi_iter.bi_sector =
 			(block << pool->sectors_per_block_shift) |
@@ -691,7 +691,7 @@ static void remap(struct thin_c *tc, struct bio *bio, dm_block_t block)
 
 static void remap_to_origin(struct thin_c *tc, struct bio *bio)
 {
-	bio->bi_bdev = tc->origin_dev->bdev;
+	bio_set_dev(bio, tc->origin_dev->bdev);
 }
 
 static int bio_triggers_commit(struct thin_c *tc, struct bio *bio)
@@ -3313,7 +3313,7 @@ static int pool_map(struct dm_target *ti, struct bio *bio)
 	 * As this is a singleton target, ti->begin is always zero.
 	 */
 	spin_lock_irqsave(&pool->lock, flags);
-	bio->bi_bdev = pt->data_dev->bdev;
+	bio_set_dev(bio, pt->data_dev->bdev);
 	r = DM_MAPIO_REMAPPED;
 	spin_unlock_irqrestore(&pool->lock, flags);
 

drivers/md/dm-verity-target.c

@@ -637,7 +637,7 @@ static int verity_map(struct dm_target *ti, struct bio *bio)
 	struct dm_verity *v = ti->private;
 	struct dm_verity_io *io;
 
-	bio->bi_bdev = v->data_dev->bdev;
+	bio_set_dev(bio, v->data_dev->bdev);
 	bio->bi_iter.bi_sector = verity_map_sector(v, bio->bi_iter.bi_sector);
 
 	if (((unsigned)bio->bi_iter.bi_sector | bio_sectors(bio)) &

drivers/md/dm-zoned-metadata.c

@@ -409,7 +409,7 @@ static struct dmz_mblock *dmz_fetch_mblock(struct dmz_metadata *zmd,
 	}
 
 	bio->bi_iter.bi_sector = dmz_blk2sect(block);
-	bio->bi_bdev = zmd->dev->bdev;
+	bio_set_dev(bio, zmd->dev->bdev);
 	bio->bi_private = mblk;
 	bio->bi_end_io = dmz_mblock_bio_end_io;
 	bio_set_op_attrs(bio, REQ_OP_READ, REQ_META | REQ_PRIO);
@@ -564,7 +564,7 @@ static void dmz_write_mblock(struct dmz_metadata *zmd, struct dmz_mblock *mblk,
 	set_bit(DMZ_META_WRITING, &mblk->state);
 
 	bio->bi_iter.bi_sector = dmz_blk2sect(block);
-	bio->bi_bdev = zmd->dev->bdev;
+	bio_set_dev(bio, zmd->dev->bdev);
 	bio->bi_private = mblk;
 	bio->bi_end_io = dmz_mblock_bio_end_io;
 	bio_set_op_attrs(bio, REQ_OP_WRITE, REQ_META | REQ_PRIO);
@@ -586,7 +586,7 @@ static int dmz_rdwr_block(struct dmz_metadata *zmd, int op, sector_t block,
 		return -ENOMEM;
 
 	bio->bi_iter.bi_sector = dmz_blk2sect(block);
-	bio->bi_bdev = zmd->dev->bdev;
+	bio_set_dev(bio, zmd->dev->bdev);
 	bio_set_op_attrs(bio, op, REQ_SYNC | REQ_META | REQ_PRIO);
 	bio_add_page(bio, page, DMZ_BLOCK_SIZE, 0);
 	ret = submit_bio_wait(bio);

drivers/md/dm-zoned-target.c

@@ -238,7 +238,7 @@ static void dmz_submit_write_bio(struct dmz_target *dmz, struct dm_zone *zone,
 	struct dmz_bioctx *bioctx = dm_per_bio_data(bio, sizeof(struct dmz_bioctx));
 
 	/* Setup and submit the BIO */
-	bio->bi_bdev = dmz->dev->bdev;
+	bio_set_dev(bio, dmz->dev->bdev);
 	bio->bi_iter.bi_sector = dmz_start_sect(dmz->metadata, zone) + dmz_blk2sect(chunk_block);
 	atomic_inc(&bioctx->ref);
 	generic_make_request(bio);
@@ -586,7 +586,7 @@ static int dmz_map(struct dm_target *ti, struct bio *bio)
 		      (unsigned long long)dmz_chunk_block(dmz->dev, dmz_bio_block(bio)),
 		      (unsigned int)dmz_bio_blocks(bio));
 
-	bio->bi_bdev = dev->bdev;
+	bio_set_dev(bio, dev->bdev);
 
 	if (!nr_sectors && bio_op(bio) != REQ_OP_WRITE)
 		return DM_MAPIO_REMAPPED;

drivers/md/dm.c

@@ -510,7 +510,7 @@ static void start_io_acct(struct dm_io *io)
 	io->start_time = jiffies;
 
 	cpu = part_stat_lock();
-	part_round_stats(cpu, &dm_disk(md)->part0);
+	part_round_stats(md->queue, cpu, &dm_disk(md)->part0);
 	part_stat_unlock();
 	atomic_set(&dm_disk(md)->part0.in_flight[rw],
 		atomic_inc_return(&md->pending[rw]));
@@ -529,7 +529,7 @@ static void end_io_acct(struct dm_io *io)
 	int pending;
 	int rw = bio_data_dir(bio);
 
-	generic_end_io_acct(rw, &dm_disk(md)->part0, io->start_time);
+	generic_end_io_acct(md->queue, rw, &dm_disk(md)->part0, io->start_time);
 
 	if (unlikely(dm_stats_used(&md->stats)))
 		dm_stats_account_io(&md->stats, bio_data_dir(bio),
@@ -841,10 +841,10 @@ static void clone_endio(struct bio *bio)
 
 	if (unlikely(error == BLK_STS_TARGET)) {
 		if (bio_op(bio) == REQ_OP_WRITE_SAME &&
-		    !bdev_get_queue(bio->bi_bdev)->limits.max_write_same_sectors)
+		    !bio->bi_disk->queue->limits.max_write_same_sectors)
 			disable_write_same(md);
 		if (bio_op(bio) == REQ_OP_WRITE_ZEROES &&
-		    !bdev_get_queue(bio->bi_bdev)->limits.max_write_zeroes_sectors)
+		    !bio->bi_disk->queue->limits.max_write_zeroes_sectors)
 			disable_write_zeroes(md);
 	}
 
@@ -1205,8 +1205,8 @@ static void __map_bio(struct dm_target_io *tio)
 		break;
 	case DM_MAPIO_REMAPPED:
 		/* the bio has been remapped so dispatch it */
-		trace_block_bio_remap(bdev_get_queue(clone->bi_bdev), clone,
-				      tio->io->bio->bi_bdev->bd_dev, sector);
+		trace_block_bio_remap(clone->bi_disk->queue, clone,
+				      bio_dev(tio->io->bio), sector);
 		generic_make_request(clone);
 		break;
 	case DM_MAPIO_KILL:
@@ -1532,7 +1532,7 @@ static blk_qc_t dm_make_request(struct request_queue *q, struct bio *bio)
 
 	map = dm_get_live_table(md, &srcu_idx);
 
-	generic_start_io_acct(rw, bio_sectors(bio), &dm_disk(md)->part0);
+	generic_start_io_acct(q, rw, bio_sectors(bio), &dm_disk(md)->part0);
 
 	/* if we're suspended, we have to queue this io for later */
 	if (unlikely(test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags))) {
@@ -1786,7 +1786,7 @@ static struct mapped_device *alloc_dev(int minor)
 		goto bad;
 
 	bio_init(&md->flush_bio, NULL, 0);
-	md->flush_bio.bi_bdev = md->bdev;
+	bio_set_dev(&md->flush_bio, md->bdev);
 	md->flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH | REQ_SYNC;
 
 	dm_stats_init(&md->stats);

drivers/md/faulty.c

@@ -216,12 +216,12 @@ static bool faulty_make_request(struct mddev *mddev, struct bio *bio)
 	if (failit) {
 		struct bio *b = bio_clone_fast(bio, GFP_NOIO, mddev->bio_set);
 
-		b->bi_bdev = conf->rdev->bdev;
+		bio_set_dev(b, conf->rdev->bdev);
 		b->bi_private = bio;
 		b->bi_end_io = faulty_fail;
 		bio = b;
 	} else
-		bio->bi_bdev = conf->rdev->bdev;
+		bio_set_dev(bio, conf->rdev->bdev);
 
 	generic_make_request(bio);
 	return true;

drivers/md/linear.c

@@ -275,17 +275,17 @@ static bool linear_make_request(struct mddev *mddev, struct bio *bio)
 		bio = split;
 	}
 
-	bio->bi_bdev = tmp_dev->rdev->bdev;
+	bio_set_dev(bio, tmp_dev->rdev->bdev);
 	bio->bi_iter.bi_sector = bio->bi_iter.bi_sector -
 		start_sector + data_offset;
 
 	if (unlikely((bio_op(bio) == REQ_OP_DISCARD) &&
-		     !blk_queue_discard(bdev_get_queue(bio->bi_bdev)))) {
+		     !blk_queue_discard(bio->bi_disk->queue))) {
 		/* Just ignore it */
 		bio_endio(bio);
 	} else {
 		if (mddev->gendisk)
-			trace_block_bio_remap(bdev_get_queue(bio->bi_bdev),
+			trace_block_bio_remap(bio->bi_disk->queue,
 					      bio, disk_devt(mddev->gendisk),
 					      bio_sector);
 		mddev_check_writesame(mddev, bio);

drivers/md/md.c

@@ -422,7 +422,7 @@ static void submit_flushes(struct work_struct *ws)
 			bi = bio_alloc_mddev(GFP_NOIO, 0, mddev);
 			bi->bi_end_io = md_end_flush;
 			bi->bi_private = rdev;
-			bi->bi_bdev = rdev->bdev;
+			bio_set_dev(bi, rdev->bdev);
 			bi->bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
 			atomic_inc(&mddev->flush_pending);
 			submit_bio(bi);
@@ -772,7 +772,7 @@ void md_super_write(struct mddev *mddev, struct md_rdev *rdev,
 
 	atomic_inc(&rdev->nr_pending);
 
-	bio->bi_bdev = rdev->meta_bdev ? rdev->meta_bdev : rdev->bdev;
+	bio_set_dev(bio, rdev->meta_bdev ? rdev->meta_bdev : rdev->bdev);
 	bio->bi_iter.bi_sector = sector;
 	bio_add_page(bio, page, size, 0);
 	bio->bi_private = rdev;
@@ -803,8 +803,10 @@ int sync_page_io(struct md_rdev *rdev, sector_t sector, int size,
 	struct bio *bio = md_bio_alloc_sync(rdev->mddev);
 	int ret;
 
-	bio->bi_bdev = (metadata_op && rdev->meta_bdev) ?
-		rdev->meta_bdev : rdev->bdev;
+	if (metadata_op && rdev->meta_bdev)
+		bio_set_dev(bio, rdev->meta_bdev);
+	else
+		bio_set_dev(bio, rdev->bdev);
 	bio_set_op_attrs(bio, op, op_flags);
 	if (metadata_op)
 		bio->bi_iter.bi_sector = sector + rdev->sb_start;

drivers/md/md.h

@@ -509,6 +509,11 @@ static inline void md_sync_acct(struct block_device *bdev, unsigned long nr_sect
 	atomic_add(nr_sectors, &bdev->bd_contains->bd_disk->sync_io);
 }
 
+static inline void md_sync_acct_bio(struct bio *bio, unsigned long nr_sectors)
+{
+	atomic_add(nr_sectors, &bio->bi_disk->sync_io);
+}
+
 struct md_personality
 {
 	char *name;
@@ -721,14 +726,14 @@ static inline void mddev_clear_unsupported_flags(struct mddev *mddev,
 static inline void mddev_check_writesame(struct mddev *mddev, struct bio *bio)
 {
 	if (bio_op(bio) == REQ_OP_WRITE_SAME &&
-	    !bdev_get_queue(bio->bi_bdev)->limits.max_write_same_sectors)
+	    !bio->bi_disk->queue->limits.max_write_same_sectors)
 		mddev->queue->limits.max_write_same_sectors = 0;
 }
 
 static inline void mddev_check_write_zeroes(struct mddev *mddev, struct bio *bio)
 {
 	if (bio_op(bio) == REQ_OP_WRITE_ZEROES &&
-	    !bdev_get_queue(bio->bi_bdev)->limits.max_write_zeroes_sectors)
+	    !bio->bi_disk->queue->limits.max_write_zeroes_sectors)
 		mddev->queue->limits.max_write_zeroes_sectors = 0;
 }
 #endif /* _MD_MD_H */

drivers/md/multipath.c

@@ -134,7 +134,7 @@ static bool multipath_make_request(struct mddev *mddev, struct bio * bio)
 	__bio_clone_fast(&mp_bh->bio, bio);
 
 	mp_bh->bio.bi_iter.bi_sector += multipath->rdev->data_offset;
-	mp_bh->bio.bi_bdev = multipath->rdev->bdev;
+	bio_set_dev(&mp_bh->bio, multipath->rdev->bdev);
 	mp_bh->bio.bi_opf |= REQ_FAILFAST_TRANSPORT;
 	mp_bh->bio.bi_end_io = multipath_end_request;
 	mp_bh->bio.bi_private = mp_bh;
@@ -345,17 +345,17 @@ static void multipathd(struct md_thread *thread)
 
 		if ((mp_bh->path = multipath_map (conf))<0) {
 			pr_err("multipath: %s: unrecoverable IO read error for block %llu\n",
-			       bdevname(bio->bi_bdev,b),
+			       bio_devname(bio, b),
 			       (unsigned long long)bio->bi_iter.bi_sector);
 			multipath_end_bh_io(mp_bh, BLK_STS_IOERR);
 		} else {
 			pr_err("multipath: %s: redirecting sector %llu to another IO path\n",
-			       bdevname(bio->bi_bdev,b),
+			       bio_devname(bio, b),
 			       (unsigned long long)bio->bi_iter.bi_sector);
 			*bio = *(mp_bh->master_bio);
 			bio->bi_iter.bi_sector +=
 				conf->multipaths[mp_bh->path].rdev->data_offset;
-			bio->bi_bdev = conf->multipaths[mp_bh->path].rdev->bdev;
+			bio_set_dev(bio, conf->multipaths[mp_bh->path].rdev->bdev);
 			bio->bi_opf |= REQ_FAILFAST_TRANSPORT;
 			bio->bi_end_io = multipath_end_request;
 			bio->bi_private = mp_bh;

drivers/md/raid0.c

@@ -588,14 +588,13 @@ static bool raid0_make_request(struct mddev *mddev, struct bio *bio)
 
 	zone = find_zone(mddev->private, &sector);
 	tmp_dev = map_sector(mddev, zone, sector, &sector);
-	bio->bi_bdev = tmp_dev->bdev;
+	bio_set_dev(bio, tmp_dev->bdev);
 	bio->bi_iter.bi_sector = sector + zone->dev_start +
 		tmp_dev->data_offset;
 
 	if (mddev->gendisk)
-		trace_block_bio_remap(bdev_get_queue(bio->bi_bdev),
-				      bio, disk_devt(mddev->gendisk),
-				      bio_sector);
+		trace_block_bio_remap(bio->bi_disk->queue, bio,
+				disk_devt(mddev->gendisk), bio_sector);
 	mddev_check_writesame(mddev, bio);
 	mddev_check_write_zeroes(mddev, bio);
 	generic_make_request(bio);

drivers/md/raid1.c

@@ -786,13 +786,13 @@ static void flush_bio_list(struct r1conf *conf, struct bio *bio)
 
 	while (bio) { /* submit pending writes */
 		struct bio *next = bio->bi_next;
-		struct md_rdev *rdev = (void*)bio->bi_bdev;
+		struct md_rdev *rdev = (void *)bio->bi_disk;
 		bio->bi_next = NULL;
-		bio->bi_bdev = rdev->bdev;
+		bio_set_dev(bio, rdev->bdev);
 		if (test_bit(Faulty, &rdev->flags)) {
 			bio_io_error(bio);
 		} else if (unlikely((bio_op(bio) == REQ_OP_DISCARD) &&
-				    !blk_queue_discard(bdev_get_queue(bio->bi_bdev))))
+				    !blk_queue_discard(bio->bi_disk->queue)))
 			/* Just ignore it */
 			bio_endio(bio);
 		else
@@ -1273,7 +1273,7 @@ static void raid1_read_request(struct mddev *mddev, struct bio *bio,
 
 	read_bio->bi_iter.bi_sector = r1_bio->sector +
 		mirror->rdev->data_offset;
-	read_bio->bi_bdev = mirror->rdev->bdev;
+	bio_set_dev(read_bio, mirror->rdev->bdev);
 	read_bio->bi_end_io = raid1_end_read_request;
 	bio_set_op_attrs(read_bio, op, do_sync);
 	if (test_bit(FailFast, &mirror->rdev->flags) &&
@@ -1282,9 +1282,8 @@ static void raid1_read_request(struct mddev *mddev, struct bio *bio,
 	read_bio->bi_private = r1_bio;
 
 	if (mddev->gendisk)
-	        trace_block_bio_remap(bdev_get_queue(read_bio->bi_bdev),
-	                              read_bio, disk_devt(mddev->gendisk),
-	                              r1_bio->sector);
+	        trace_block_bio_remap(read_bio->bi_disk->queue, read_bio,
+				disk_devt(mddev->gendisk), r1_bio->sector);
 
 	generic_make_request(read_bio);
 }
@@ -1496,7 +1495,7 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
 
 		mbio->bi_iter.bi_sector	= (r1_bio->sector +
 				   conf->mirrors[i].rdev->data_offset);
-		mbio->bi_bdev = conf->mirrors[i].rdev->bdev;
+		bio_set_dev(mbio, conf->mirrors[i].rdev->bdev);
 		mbio->bi_end_io	= raid1_end_write_request;
 		mbio->bi_opf = bio_op(bio) | (bio->bi_opf & (REQ_SYNC | REQ_FUA));
 		if (test_bit(FailFast, &conf->mirrors[i].rdev->flags) &&
@@ -1508,11 +1507,11 @@ static void raid1_write_request(struct mddev *mddev, struct bio *bio,
 		atomic_inc(&r1_bio->remaining);
 
 		if (mddev->gendisk)
-			trace_block_bio_remap(bdev_get_queue(mbio->bi_bdev),
+			trace_block_bio_remap(mbio->bi_disk->queue,
 					      mbio, disk_devt(mddev->gendisk),
 					      r1_bio->sector);
 		/* flush_pending_writes() needs access to the rdev so...*/
-		mbio->bi_bdev = (void*)conf->mirrors[i].rdev;
+		mbio->bi_disk = (void *)conf->mirrors[i].rdev;
 
 		cb = blk_check_plugged(raid1_unplug, mddev, sizeof(*plug));
 		if (cb)
@@ -1990,8 +1989,7 @@ static int fix_sync_read_error(struct r1bio *r1_bio)
 			 * Don't fail devices as that won't really help.
 			 */
 			pr_crit_ratelimited("md/raid1:%s: %s: unrecoverable I/O read error for block %llu\n",
-					    mdname(mddev),
-					    bdevname(bio->bi_bdev, b),
+					    mdname(mddev), bio_devname(bio, b),
 					    (unsigned long long)r1_bio->sector);
 			for (d = 0; d < conf->raid_disks * 2; d++) {
 				rdev = conf->mirrors[d].rdev;
@@ -2082,7 +2080,7 @@ static void process_checks(struct r1bio *r1_bio)
 		b->bi_status = status;
 		b->bi_iter.bi_sector = r1_bio->sector +
 			conf->mirrors[i].rdev->data_offset;
-		b->bi_bdev = conf->mirrors[i].rdev->bdev;
+		bio_set_dev(b, conf->mirrors[i].rdev->bdev);
 		b->bi_end_io = end_sync_read;
 		rp->raid_bio = r1_bio;
 		b->bi_private = rp;
@@ -2350,7 +2348,7 @@ static int narrow_write_error(struct r1bio *r1_bio, int i)
 
 		bio_trim(wbio, sector - r1_bio->sector, sectors);
 		wbio->bi_iter.bi_sector += rdev->data_offset;
-		wbio->bi_bdev = rdev->bdev;
+		bio_set_dev(wbio, rdev->bdev);
 
 		if (submit_bio_wait(wbio) < 0)
 			/* failure! */
@@ -2440,7 +2438,6 @@ static void handle_read_error(struct r1conf *conf, struct r1bio *r1_bio)
 	struct mddev *mddev = conf->mddev;
 	struct bio *bio;
 	struct md_rdev *rdev;
-	dev_t bio_dev;
 	sector_t bio_sector;
 
 	clear_bit(R1BIO_ReadError, &r1_bio->state);
@@ -2454,7 +2451,6 @@ static void handle_read_error(struct r1conf *conf, struct r1bio *r1_bio)
 	 */
 
 	bio = r1_bio->bios[r1_bio->read_disk];
-	bio_dev = bio->bi_bdev->bd_dev;
 	bio_sector = conf->mirrors[r1_bio->read_disk].rdev->data_offset + r1_bio->sector;
 	bio_put(bio);
 	r1_bio->bios[r1_bio->read_disk] = NULL;
@@ -2727,7 +2723,7 @@ static sector_t raid1_sync_request(struct mddev *mddev, sector_t sector_nr,
 		if (bio->bi_end_io) {
 			atomic_inc(&rdev->nr_pending);
 			bio->bi_iter.bi_sector = sector_nr + rdev->data_offset;
-			bio->bi_bdev = rdev->bdev;
+			bio_set_dev(bio, rdev->bdev);
 			if (test_bit(FailFast, &rdev->flags))
 				bio->bi_opf |= MD_FAILFAST;
 		}
@@ -2853,7 +2849,7 @@ static sector_t raid1_sync_request(struct mddev *mddev, sector_t sector_nr,
 			bio = r1_bio->bios[i];
 			if (bio->bi_end_io == end_sync_read) {
 				read_targets--;
-				md_sync_acct(bio->bi_bdev, nr_sectors);
+				md_sync_acct_bio(bio, nr_sectors);
 				if (read_targets == 1)
 					bio->bi_opf &= ~MD_FAILFAST;
 				generic_make_request(bio);
@@ -2862,7 +2858,7 @@ static sector_t raid1_sync_request(struct mddev *mddev, sector_t sector_nr,
 	} else {
 		atomic_set(&r1_bio->remaining, 1);
 		bio = r1_bio->bios[r1_bio->read_disk];
-		md_sync_acct(bio->bi_bdev, nr_sectors);
+		md_sync_acct_bio(bio, nr_sectors);
 		if (read_targets == 1)
 			bio->bi_opf &= ~MD_FAILFAST;
 		generic_make_request(bio);

drivers/md/raid10.c

@@ -901,13 +901,13 @@ static void flush_pending_writes(struct r10conf *conf)
 
 		while (bio) { /* submit pending writes */
 			struct bio *next = bio->bi_next;
-			struct md_rdev *rdev = (void*)bio->bi_bdev;
+			struct md_rdev *rdev = (void*)bio->bi_disk;
 			bio->bi_next = NULL;
-			bio->bi_bdev = rdev->bdev;
+			bio_set_dev(bio, rdev->bdev);
 			if (test_bit(Faulty, &rdev->flags)) {
 				bio_io_error(bio);
 			} else if (unlikely((bio_op(bio) ==  REQ_OP_DISCARD) &&
-					    !blk_queue_discard(bdev_get_queue(bio->bi_bdev))))
+					    !blk_queue_discard(bio->bi_disk->queue)))
 				/* Just ignore it */
 				bio_endio(bio);
 			else
@@ -1085,13 +1085,13 @@ static void raid10_unplug(struct blk_plug_cb *cb, bool from_schedule)
 
 	while (bio) { /* submit pending writes */
 		struct bio *next = bio->bi_next;
-		struct md_rdev *rdev = (void*)bio->bi_bdev;
+		struct md_rdev *rdev = (void*)bio->bi_disk;
 		bio->bi_next = NULL;
-		bio->bi_bdev = rdev->bdev;
+		bio_set_dev(bio, rdev->bdev);
 		if (test_bit(Faulty, &rdev->flags)) {
 			bio_io_error(bio);
 		} else if (unlikely((bio_op(bio) ==  REQ_OP_DISCARD) &&
-				    !blk_queue_discard(bdev_get_queue(bio->bi_bdev))))
+				    !blk_queue_discard(bio->bi_disk->queue)))
 			/* Just ignore it */
 			bio_endio(bio);
 		else
@@ -1200,7 +1200,7 @@ static void raid10_read_request(struct mddev *mddev, struct bio *bio,
 
 	read_bio->bi_iter.bi_sector = r10_bio->devs[slot].addr +
 		choose_data_offset(r10_bio, rdev);
-	read_bio->bi_bdev = rdev->bdev;
+	bio_set_dev(read_bio, rdev->bdev);
 	read_bio->bi_end_io = raid10_end_read_request;
 	bio_set_op_attrs(read_bio, op, do_sync);
 	if (test_bit(FailFast, &rdev->flags) &&
@@ -1209,7 +1209,7 @@ static void raid10_read_request(struct mddev *mddev, struct bio *bio,
 	read_bio->bi_private = r10_bio;
 
 	if (mddev->gendisk)
-	        trace_block_bio_remap(bdev_get_queue(read_bio->bi_bdev),
+	        trace_block_bio_remap(read_bio->bi_disk->queue,
 	                              read_bio, disk_devt(mddev->gendisk),
 	                              r10_bio->sector);
 	generic_make_request(read_bio);
@@ -1249,7 +1249,7 @@ static void raid10_write_one_disk(struct mddev *mddev, struct r10bio *r10_bio,
 
 	mbio->bi_iter.bi_sector	= (r10_bio->devs[n_copy].addr +
 				   choose_data_offset(r10_bio, rdev));
-	mbio->bi_bdev = rdev->bdev;
+	bio_set_dev(mbio, rdev->bdev);
 	mbio->bi_end_io	= raid10_end_write_request;
 	bio_set_op_attrs(mbio, op, do_sync | do_fua);
 	if (!replacement && test_bit(FailFast,
@@ -1259,11 +1259,11 @@ static void raid10_write_one_disk(struct mddev *mddev, struct r10bio *r10_bio,
 	mbio->bi_private = r10_bio;
 
 	if (conf->mddev->gendisk)
-		trace_block_bio_remap(bdev_get_queue(mbio->bi_bdev),
+		trace_block_bio_remap(mbio->bi_disk->queue,
 				      mbio, disk_devt(conf->mddev->gendisk),
 				      r10_bio->sector);
 	/* flush_pending_writes() needs access to the rdev so...*/
-	mbio->bi_bdev = (void *)rdev;
+	mbio->bi_disk = (void *)rdev;
 
 	atomic_inc(&r10_bio->remaining);
 
@@ -2094,7 +2094,7 @@ static void sync_request_write(struct mddev *mddev, struct r10bio *r10_bio)
 		if (test_bit(FailFast, &conf->mirrors[d].rdev->flags))
 			tbio->bi_opf |= MD_FAILFAST;
 		tbio->bi_iter.bi_sector += conf->mirrors[d].rdev->data_offset;
-		tbio->bi_bdev = conf->mirrors[d].rdev->bdev;
+		bio_set_dev(tbio, conf->mirrors[d].rdev->bdev);
 		generic_make_request(tbio);
 	}
 
@@ -2552,7 +2552,7 @@ static int narrow_write_error(struct r10bio *r10_bio, int i)
 		wsector = r10_bio->devs[i].addr + (sector - r10_bio->sector);
 		wbio->bi_iter.bi_sector = wsector +
 				   choose_data_offset(r10_bio, rdev);
-		wbio->bi_bdev = rdev->bdev;
+		bio_set_dev(wbio, rdev->bdev);
 		bio_set_op_attrs(wbio, REQ_OP_WRITE, 0);
 
 		if (submit_bio_wait(wbio) < 0)
@@ -2575,7 +2575,6 @@ static void handle_read_error(struct mddev *mddev, struct r10bio *r10_bio)
 	struct bio *bio;
 	struct r10conf *conf = mddev->private;
 	struct md_rdev *rdev = r10_bio->devs[slot].rdev;
-	dev_t bio_dev;
 	sector_t bio_last_sector;
 
 	/* we got a read error. Maybe the drive is bad.  Maybe just
@@ -2587,7 +2586,6 @@ static void handle_read_error(struct mddev *mddev, struct r10bio *r10_bio)
 	 * frozen.
 	 */
 	bio = r10_bio->devs[slot].bio;
-	bio_dev = bio->bi_bdev->bd_dev;
 	bio_last_sector = r10_bio->devs[slot].addr + rdev->data_offset + r10_bio->sectors;
 	bio_put(bio);
 	r10_bio->devs[slot].bio = NULL;
@@ -2950,7 +2948,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 
 	/* Again, very different code for resync and recovery.
 	 * Both must result in an r10bio with a list of bios that
-	 * have bi_end_io, bi_sector, bi_bdev set,
+	 * have bi_end_io, bi_sector, bi_disk set,
 	 * and bi_private set to the r10bio.
 	 * For recovery, we may actually create several r10bios
 	 * with 2 bios in each, that correspond to the bios in the main one.
@@ -3095,7 +3093,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 				from_addr = r10_bio->devs[j].addr;
 				bio->bi_iter.bi_sector = from_addr +
 					rdev->data_offset;
-				bio->bi_bdev = rdev->bdev;
+				bio_set_dev(bio, rdev->bdev);
 				atomic_inc(&rdev->nr_pending);
 				/* and we write to 'i' (if not in_sync) */
 
@@ -3117,7 +3115,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 					bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
 					bio->bi_iter.bi_sector = to_addr
 						+ mrdev->data_offset;
-					bio->bi_bdev = mrdev->bdev;
+					bio_set_dev(bio, mrdev->bdev);
 					atomic_inc(&r10_bio->remaining);
 				} else
 					r10_bio->devs[1].bio->bi_end_io = NULL;
@@ -3143,7 +3141,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 				bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
 				bio->bi_iter.bi_sector = to_addr +
 					mreplace->data_offset;
-				bio->bi_bdev = mreplace->bdev;
+				bio_set_dev(bio, mreplace->bdev);
 				atomic_inc(&r10_bio->remaining);
 				break;
 			}
@@ -3289,7 +3287,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 			if (test_bit(FailFast, &rdev->flags))
 				bio->bi_opf |= MD_FAILFAST;
 			bio->bi_iter.bi_sector = sector + rdev->data_offset;
-			bio->bi_bdev = rdev->bdev;
+			bio_set_dev(bio, rdev->bdev);
 			count++;
 
 			rdev = rcu_dereference(conf->mirrors[d].replacement);
@@ -3311,7 +3309,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 			if (test_bit(FailFast, &rdev->flags))
 				bio->bi_opf |= MD_FAILFAST;
 			bio->bi_iter.bi_sector = sector + rdev->data_offset;
-			bio->bi_bdev = rdev->bdev;
+			bio_set_dev(bio, rdev->bdev);
 			count++;
 			rcu_read_unlock();
 		}
@@ -3367,7 +3365,7 @@ static sector_t raid10_sync_request(struct mddev *mddev, sector_t sector_nr,
 		r10_bio->sectors = nr_sectors;
 
 		if (bio->bi_end_io == end_sync_read) {
-			md_sync_acct(bio->bi_bdev, nr_sectors);
+			md_sync_acct_bio(bio, nr_sectors);
 			bio->bi_status = 0;
 			generic_make_request(bio);
 		}
@@ -4383,7 +4381,7 @@ read_more:
 
 	read_bio = bio_alloc_mddev(GFP_KERNEL, RESYNC_PAGES, mddev);
 
-	read_bio->bi_bdev = rdev->bdev;
+	bio_set_dev(read_bio, rdev->bdev);
 	read_bio->bi_iter.bi_sector = (r10_bio->devs[r10_bio->read_slot].addr
 			       + rdev->data_offset);
 	read_bio->bi_private = r10_bio;
@@ -4417,7 +4415,7 @@ read_more:
 		if (!rdev2 || test_bit(Faulty, &rdev2->flags))
 			continue;
 
-		b->bi_bdev = rdev2->bdev;
+		bio_set_dev(b, rdev2->bdev);
 		b->bi_iter.bi_sector = r10_bio->devs[s/2].addr +
 			rdev2->new_data_offset;
 		b->bi_end_io = end_reshape_write;
@@ -4449,7 +4447,7 @@ read_more:
 	r10_bio->sectors = nr_sectors;
 
 	/* Now submit the read */
-	md_sync_acct(read_bio->bi_bdev, r10_bio->sectors);
+	md_sync_acct_bio(read_bio, r10_bio->sectors);
 	atomic_inc(&r10_bio->remaining);
 	read_bio->bi_next = NULL;
 	generic_make_request(read_bio);
@@ -4511,7 +4509,7 @@ static void reshape_request_write(struct mddev *mddev, struct r10bio *r10_bio)
 		}
 		atomic_inc(&rdev->nr_pending);
 		rcu_read_unlock();
-		md_sync_acct(b->bi_bdev, r10_bio->sectors);
+		md_sync_acct_bio(b, r10_bio->sectors);
 		atomic_inc(&r10_bio->remaining);
 		b->bi_next = NULL;
 		generic_make_request(b);

drivers/md/raid5-cache.c

@@ -745,7 +745,7 @@ static struct bio *r5l_bio_alloc(struct r5l_log *log)
 	struct bio *bio = bio_alloc_bioset(GFP_NOIO, BIO_MAX_PAGES, log->bs);
 
 	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
-	bio->bi_bdev = log->rdev->bdev;
+	bio_set_dev(bio, log->rdev->bdev);
 	bio->bi_iter.bi_sector = log->rdev->data_offset + log->log_start;
 
 	return bio;
@@ -1313,7 +1313,7 @@ void r5l_flush_stripe_to_raid(struct r5l_log *log)
 	if (!do_flush)
 		return;
 	bio_reset(&log->flush_bio);
-	log->flush_bio.bi_bdev = log->rdev->bdev;
+	bio_set_dev(&log->flush_bio, log->rdev->bdev);
 	log->flush_bio.bi_end_io = r5l_log_flush_endio;
 	log->flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH;
 	submit_bio(&log->flush_bio);
@@ -1691,7 +1691,7 @@ static int r5l_recovery_fetch_ra_pool(struct r5l_log *log,
 				      sector_t offset)
 {
 	bio_reset(ctx->ra_bio);
-	ctx->ra_bio->bi_bdev = log->rdev->bdev;
+	bio_set_dev(ctx->ra_bio, log->rdev->bdev);
 	bio_set_op_attrs(ctx->ra_bio, REQ_OP_READ, 0);
 	ctx->ra_bio->bi_iter.bi_sector = log->rdev->data_offset + offset;
 

drivers/md/raid5-ppl.c

@@ -415,7 +415,7 @@ static void ppl_submit_iounit_bio(struct ppl_io_unit *io, struct bio *bio)
 	pr_debug("%s: seq: %llu size: %u sector: %llu dev: %s\n",
 		 __func__, io->seq, bio->bi_iter.bi_size,
 		 (unsigned long long)bio->bi_iter.bi_sector,
-		 bdevname(bio->bi_bdev, b));
+		 bio_devname(bio, b));
 
 	submit_bio(bio);
 }
@@ -453,7 +453,7 @@ static void ppl_submit_iounit(struct ppl_io_unit *io)
 
 	bio->bi_end_io = ppl_log_endio;
 	bio->bi_opf = REQ_OP_WRITE | REQ_FUA;
-	bio->bi_bdev = log->rdev->bdev;
+	bio_set_dev(bio, log->rdev->bdev);
 	bio->bi_iter.bi_sector = log->rdev->ppl.sector;
 	bio_add_page(bio, io->header_page, PAGE_SIZE, 0);
 
@@ -468,7 +468,7 @@ static void ppl_submit_iounit(struct ppl_io_unit *io)
 			bio = bio_alloc_bioset(GFP_NOIO, BIO_MAX_PAGES,
 					       ppl_conf->bs);
 			bio->bi_opf = prev->bi_opf;
-			bio->bi_bdev = prev->bi_bdev;
+			bio_copy_dev(bio, prev);
 			bio->bi_iter.bi_sector = bio_end_sector(prev);
 			bio_add_page(bio, sh->ppl_page, PAGE_SIZE, 0);
 

drivers/md/raid5.c

@@ -1096,7 +1096,7 @@ again:
 
 			set_bit(STRIPE_IO_STARTED, &sh->state);
 
-			bi->bi_bdev = rdev->bdev;
+			bio_set_dev(bi, rdev->bdev);
 			bio_set_op_attrs(bi, op, op_flags);
 			bi->bi_end_io = op_is_write(op)
 				? raid5_end_write_request
@@ -1145,7 +1145,7 @@ again:
 				set_bit(R5_DOUBLE_LOCKED, &sh->dev[i].flags);
 
 			if (conf->mddev->gendisk)
-				trace_block_bio_remap(bdev_get_queue(bi->bi_bdev),
+				trace_block_bio_remap(bi->bi_disk->queue,
 						      bi, disk_devt(conf->mddev->gendisk),
 						      sh->dev[i].sector);
 			if (should_defer && op_is_write(op))
@@ -1160,7 +1160,7 @@ again:
 
 			set_bit(STRIPE_IO_STARTED, &sh->state);
 
-			rbi->bi_bdev = rrdev->bdev;
+			bio_set_dev(rbi, rrdev->bdev);
 			bio_set_op_attrs(rbi, op, op_flags);
 			BUG_ON(!op_is_write(op));
 			rbi->bi_end_io = raid5_end_write_request;
@@ -1193,7 +1193,7 @@ again:
 			if (op == REQ_OP_DISCARD)
 				rbi->bi_vcnt = 0;
 			if (conf->mddev->gendisk)
-				trace_block_bio_remap(bdev_get_queue(rbi->bi_bdev),
+				trace_block_bio_remap(rbi->bi_disk->queue,
 						      rbi, disk_devt(conf->mddev->gendisk),
 						      sh->dev[i].sector);
 			if (should_defer && op_is_write(op))
@@ -5092,10 +5092,12 @@ static int raid5_congested(struct mddev *mddev, int bits)
 static int in_chunk_boundary(struct mddev *mddev, struct bio *bio)
 {
 	struct r5conf *conf = mddev->private;
-	sector_t sector = bio->bi_iter.bi_sector + get_start_sect(bio->bi_bdev);
+	sector_t sector = bio->bi_iter.bi_sector;
 	unsigned int chunk_sectors;
 	unsigned int bio_sectors = bio_sectors(bio);
 
+	WARN_ON_ONCE(bio->bi_partno);
+
 	chunk_sectors = min(conf->chunk_sectors, conf->prev_chunk_sectors);
 	return  chunk_sectors >=
 		((sector & (chunk_sectors - 1)) + bio_sectors);
@@ -5231,7 +5233,7 @@ static int raid5_read_one_chunk(struct mddev *mddev, struct bio *raid_bio)
 		atomic_inc(&rdev->nr_pending);
 		rcu_read_unlock();
 		raid_bio->bi_next = (void*)rdev;
-		align_bi->bi_bdev =  rdev->bdev;
+		bio_set_dev(align_bi, rdev->bdev);
 		bio_clear_flag(align_bi, BIO_SEG_VALID);
 
 		if (is_badblock(rdev, align_bi->bi_iter.bi_sector,
@@ -5253,7 +5255,7 @@ static int raid5_read_one_chunk(struct mddev *mddev, struct bio *raid_bio)
 		spin_unlock_irq(&conf->device_lock);
 
 		if (mddev->gendisk)
-			trace_block_bio_remap(bdev_get_queue(align_bi->bi_bdev),
+			trace_block_bio_remap(align_bi->bi_disk->queue,
 					      align_bi, disk_devt(mddev->gendisk),
 					      raid_bio->bi_iter.bi_sector);
 		generic_make_request(align_bi);

drivers/nvdimm/nd.h

@@ -390,21 +390,22 @@ int nd_region_activate(struct nd_region *nd_region);
 void __nd_iostat_start(struct bio *bio, unsigned long *start);
 static inline bool nd_iostat_start(struct bio *bio, unsigned long *start)
 {
-	struct gendisk *disk = bio->bi_bdev->bd_disk;
+	struct gendisk *disk = bio->bi_disk;
 
 	if (!blk_queue_io_stat(disk->queue))
 		return false;
 
 	*start = jiffies;
-	generic_start_io_acct(bio_data_dir(bio),
+	generic_start_io_acct(disk->queue, bio_data_dir(bio),
 			      bio_sectors(bio), &disk->part0);
 	return true;
 }
 static inline void nd_iostat_end(struct bio *bio, unsigned long start)
 {
-	struct gendisk *disk = bio->bi_bdev->bd_disk;
+	struct gendisk *disk = bio->bi_disk;
 
-	generic_end_io_acct(bio_data_dir(bio), &disk->part0, start);
+	generic_end_io_acct(disk->queue, bio_data_dir(bio), &disk->part0,
+				start);
 }
 static inline bool is_bad_pmem(struct badblocks *bb, sector_t sector,
 		unsigned int len)

drivers/nvme/host/core.c

@@ -613,11 +613,7 @@ int __nvme_submit_user_cmd(struct request_queue *q, struct nvme_command *cmd,
 
 		if (!disk)
 			goto submit;
-		bio->bi_bdev = bdget_disk(disk, 0);
-		if (!bio->bi_bdev) {
-			ret = -ENODEV;
-			goto out_unmap;
-		}
+		bio->bi_disk = disk;
 
 		if (meta_buffer && meta_len) {
 			struct bio_integrity_payload *bip;
@@ -668,11 +664,8 @@ int __nvme_submit_user_cmd(struct request_queue *q, struct nvme_command *cmd,
  out_free_meta:
 	kfree(meta);
  out_unmap:
-	if (bio) {
-		if (disk && bio->bi_bdev)
-			bdput(bio->bi_bdev);
+	if (bio)
 		blk_rq_unmap_user(bio);
-	}
  out:
 	blk_mq_free_request(req);
 	return ret;

drivers/nvme/host/fc.c

@@ -2168,7 +2168,6 @@ static const struct blk_mq_ops nvme_fc_mq_ops = {
 	.complete	= nvme_fc_complete_rq,
 	.init_request	= nvme_fc_init_request,
 	.exit_request	= nvme_fc_exit_request,
-	.reinit_request	= nvme_fc_reinit_request,
 	.init_hctx	= nvme_fc_init_hctx,
 	.poll		= nvme_fc_poll,
 	.timeout	= nvme_fc_timeout,
@@ -2269,7 +2268,7 @@ nvme_fc_reinit_io_queues(struct nvme_fc_ctrl *ctrl)
 
 	nvme_fc_init_io_queues(ctrl);
 
-	ret = blk_mq_reinit_tagset(&ctrl->tag_set);
+	ret = blk_mq_reinit_tagset(&ctrl->tag_set, nvme_fc_reinit_request);
 	if (ret)
 		goto out_free_io_queues;
 
@@ -2655,7 +2654,6 @@ static const struct blk_mq_ops nvme_fc_admin_mq_ops = {
 	.complete	= nvme_fc_complete_rq,
 	.init_request	= nvme_fc_init_request,
 	.exit_request	= nvme_fc_exit_request,
-	.reinit_request	= nvme_fc_reinit_request,
 	.init_hctx	= nvme_fc_init_admin_hctx,
 	.timeout	= nvme_fc_timeout,
 };

drivers/nvme/host/lightnvm.c

@@ -643,17 +643,9 @@ static int nvme_nvm_submit_user_cmd(struct request_queue *q,
 			vcmd->ph_rw.metadata = cpu_to_le64(metadata_dma);
 		}
 
-		if (!disk)
-			goto submit;
-
-		bio->bi_bdev = bdget_disk(disk, 0);
-		if (!bio->bi_bdev) {
-			ret = -ENODEV;
-			goto err_meta;
-		}
+		bio->bi_disk = disk;
 	}
 
-submit:
 	blk_execute_rq(q, NULL, rq, 0);
 
 	if (nvme_req(rq)->flags & NVME_REQ_CANCELLED)
@@ -673,11 +665,8 @@ err_meta:
 	if (meta_buf && meta_len)
 		dma_pool_free(dev->dma_pool, metadata, metadata_dma);
 err_map:
-	if (bio) {
-		if (disk && bio->bi_bdev)
-			bdput(bio->bi_bdev);
+	if (bio)
 		blk_rq_unmap_user(bio);
-	}
 err_ppa:
 	if (ppa_buf && ppa_len)
 		dma_pool_free(dev->dma_pool, ppa_list, ppa_dma);

drivers/nvme/host/rdma.c

@@ -711,14 +711,16 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	if (ctrl->ctrl.queue_count > 1) {
 		nvme_rdma_free_io_queues(ctrl);
 
-		ret = blk_mq_reinit_tagset(&ctrl->tag_set);
+		ret = blk_mq_reinit_tagset(&ctrl->tag_set,
+					   nvme_rdma_reinit_request);
 		if (ret)
 			goto requeue;
 	}
 
 	nvme_rdma_stop_and_free_queue(&ctrl->queues[0]);
 
-	ret = blk_mq_reinit_tagset(&ctrl->admin_tag_set);
+	ret = blk_mq_reinit_tagset(&ctrl->admin_tag_set,
+				   nvme_rdma_reinit_request);
 	if (ret)
 		goto requeue;
 
@@ -1521,7 +1523,6 @@ static const struct blk_mq_ops nvme_rdma_mq_ops = {
 	.complete	= nvme_rdma_complete_rq,
 	.init_request	= nvme_rdma_init_request,
 	.exit_request	= nvme_rdma_exit_request,
-	.reinit_request	= nvme_rdma_reinit_request,
 	.init_hctx	= nvme_rdma_init_hctx,
 	.poll		= nvme_rdma_poll,
 	.timeout	= nvme_rdma_timeout,
@@ -1533,7 +1534,6 @@ static const struct blk_mq_ops nvme_rdma_admin_mq_ops = {
 	.complete	= nvme_rdma_complete_rq,
 	.init_request	= nvme_rdma_init_request,
 	.exit_request	= nvme_rdma_exit_request,
-	.reinit_request	= nvme_rdma_reinit_request,
 	.init_hctx	= nvme_rdma_init_admin_hctx,
 	.timeout	= nvme_rdma_timeout,
 };
@@ -1731,7 +1731,8 @@ static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
 	}
 
 	if (ctrl->ctrl.queue_count > 1) {
-		ret = blk_mq_reinit_tagset(&ctrl->tag_set);
+		ret = blk_mq_reinit_tagset(&ctrl->tag_set,
+					   nvme_rdma_reinit_request);
 		if (ret)
 			goto del_dead_ctrl;