Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull s390 updates from Martin Schwidefsky:
 "The bulk of the s390 patches for 4.13. Some new things but mostly bug
  fixes and cleanups. Noteworthy changes:

   - The SCM block driver is converted to blk-mq

   - Switch s390 to 5 level page tables. The virtual address space for a
     user space process can now have up to 16EB-4KB.

   - Introduce a ELF phdr flag for qemu to avoid the global
     vm.alloc_pgste which forces all processes to large page tables

   - A couple of PCI improvements to improve error recovery

   - Included is the merge of the base support for proper machine checks
     for KVM"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (52 commits)
  s390/dasd: Fix faulty ENODEV for RO sysfs attribute
  s390/pci: recognize name clashes with uids
  s390/pci: provide more debug information
  s390/pci: fix handling of PEC 306
  s390/pci: improve pci hotplug
  s390/pci: introduce clp_get_state
  s390/pci: improve error handling during fmb (de)registration
  s390/pci: improve unreg_ioat error handling
  s390/pci: improve error handling during interrupt deregistration
  s390/pci: don't cleanup in arch_setup_msi_irqs
  KVM: s390: Backup the guest's machine check info
  s390/nmi: s390: New low level handling for machine check happening in guest
  s390/fpu: export save_fpu_regs for all configs
  s390/kvm: avoid global config of vm.alloc_pgste=1
  s390: rename struct psw_bits members
  s390: rename psw_bits enums
  s390/mm: use correct address space when enabling DAT
  s390/cio: introduce io_subchannel_type
  s390/ipl: revert Load Normal semantics for LPAR CCW-type re-IPL
  s390/dumpstack: remove raw stack dump
  ...

7 files changed, 216 insertions, 522 deletions

diff --git a/drivers/s390/block/Kconfig b/drivers/s390/block/Kconfig
index 0acb8c2f9475..31f014b57bfc 100644
--- a/drivers/s390/block/Kconfig
+++ b/drivers/s390/block/Kconfig
@@ -82,10 +82,3 @@ config SCM_BLOCK
 
 	  To compile this driver as a module, choose M here: the
 	  module will be called scm_block.
-
-config SCM_BLOCK_CLUSTER_WRITE
-	def_bool y
-	prompt "SCM force cluster writes"
-	depends on SCM_BLOCK
-	help
-	  Force writes to Storage Class Memory (SCM) to be in done in clusters.
diff --git a/drivers/s390/block/Makefile b/drivers/s390/block/Makefile
index c2f4e673e031..b64e2b32c753 100644
--- a/drivers/s390/block/Makefile
+++ b/drivers/s390/block/Makefile
@@ -19,7 +19,4 @@ obj-$(CONFIG_BLK_DEV_XPRAM) += xpram.o
 obj-$(CONFIG_DCSSBLK) += dcssblk.o
 
 scm_block-objs := scm_drv.o scm_blk.o
-ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE
-scm_block-objs += scm_blk_cluster.o
-endif
 obj-$(CONFIG_SCM_BLOCK) += scm_block.o
diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c
index b7cbd5d2cdea..0f1fe4ff7f51 100644
--- a/drivers/s390/block/dasd.c
+++ b/drivers/s390/block/dasd.c
@@ -1965,8 +1965,12 @@ static int __dasd_device_is_unusable(struct dasd_device *device,
 {
 	int mask = ~(DASD_STOPPED_DC_WAIT | DASD_UNRESUMED_PM);
 
-	if (test_bit(DASD_FLAG_OFFLINE, &device->flags)) {
-		/* dasd is being set offline. */
+	if (test_bit(DASD_FLAG_OFFLINE, &device->flags) &&
+	    !test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) {
+		/*
+		 * dasd is being set offline
+		 * but it is no safe offline where we have to allow I/O
+		 */
 		return 1;
 	}
 	if (device->stopped) {
@@ -3570,57 +3574,69 @@ int dasd_generic_set_offline(struct ccw_device *cdev)
 			else
 				pr_warn("%s: The DASD cannot be set offline while it is in use\n",
 					dev_name(&cdev->dev));
-			clear_bit(DASD_FLAG_OFFLINE, &device->flags);
-			goto out_busy;
+			rc = -EBUSY;
+			goto out_err;
 		}
 	}
 
-	if (test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) {
-		/*
-		 * safe offline already running
-		 * could only be called by normal offline so safe_offline flag
-		 * needs to be removed to run normal offline and kill all I/O
-		 */
-		if (test_and_set_bit(DASD_FLAG_OFFLINE, &device->flags))
-			/* Already doing normal offline processing */
-			goto out_busy;
-		else
-			clear_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags);
-	} else {
-		if (test_bit(DASD_FLAG_OFFLINE, &device->flags))
-			/* Already doing offline processing */
-			goto out_busy;
+	/*
+	 * Test if the offline processing is already running and exit if so.
+	 * If a safe offline is being processed this could only be a normal
+	 * offline that should be able to overtake the safe offline and
+	 * cancel any I/O we do not want to wait for any longer
+	 */
+	if (test_bit(DASD_FLAG_OFFLINE, &device->flags)) {
+		if (test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) {
+			clear_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING,
+				  &device->flags);
+		} else {
+			rc = -EBUSY;
+			goto out_err;
+		}
 	}
-
 	set_bit(DASD_FLAG_OFFLINE, &device->flags);
-	spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
 
 	/*
-	 * if safe_offline called set safe_offline_running flag and
+	 * if safe_offline is called set safe_offline_running flag and
 	 * clear safe_offline so that a call to normal offline
 	 * can overrun safe_offline processing
 	 */
 	if (test_and_clear_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags) &&
 	    !test_and_set_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) {
+		/* need to unlock here to wait for outstanding I/O */
+		spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
 		/*
 		 * If we want to set the device safe offline all IO operations
 		 * should be finished before continuing the offline process
 		 * so sync bdev first and then wait for our queues to become
 		 * empty
 		 */
-		/* sync blockdev and partitions */
 		if (device->block) {
 			rc = fsync_bdev(device->block->bdev);
 			if (rc != 0)
 				goto interrupted;
 		}
-		/* schedule device tasklet and wait for completion */
 		dasd_schedule_device_bh(device);
 		rc = wait_event_interruptible(shutdown_waitq,
 					      _wait_for_empty_queues(device));
 		if (rc != 0)
 			goto interrupted;
+
+		/*
+		 * check if a normal offline process overtook the offline
+		 * processing in this case simply do nothing beside returning
+		 * that we got interrupted
+		 * otherwise mark safe offline as not running any longer and
+		 * continue with normal offline
+		 */
+		spin_lock_irqsave(get_ccwdev_lock(cdev), flags);
+		if (!test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) {
+			rc = -ERESTARTSYS;
+			goto out_err;
+		}
+		clear_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags);
 	}
+	spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
 
 	dasd_set_target_state(device, DASD_STATE_NEW);
 	/* dasd_delete_device destroys the device reference. */
@@ -3632,22 +3648,18 @@ int dasd_generic_set_offline(struct ccw_device *cdev)
 	 */
 	if (block)
 		dasd_free_block(block);
+
 	return 0;
 
 interrupted:
 	/* interrupted by signal */
-	clear_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags);
+	spin_lock_irqsave(get_ccwdev_lock(cdev), flags);
 	clear_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags);
 	clear_bit(DASD_FLAG_OFFLINE, &device->flags);
-	dasd_put_device(device);
-
-	return rc;
-
-out_busy:
+out_err:
 	dasd_put_device(device);
 	spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
-
-	return -EBUSY;
+	return rc;
 }
 EXPORT_SYMBOL_GPL(dasd_generic_set_offline);
 
diff --git a/drivers/s390/block/dasd_devmap.c b/drivers/s390/block/dasd_devmap.c
index 1164b51d09f3..7c7351276d2e 100644
--- a/drivers/s390/block/dasd_devmap.c
+++ b/drivers/s390/block/dasd_devmap.c
@@ -315,45 +315,58 @@ static int __init dasd_parse_range(const char *range)
 	char *features_str = NULL;
 	char *from_str = NULL;
 	char *to_str = NULL;
-	size_t len = strlen(range) + 1;
-	char tmp[len];
+	int rc = 0;
+	char *tmp;
 
-	strlcpy(tmp, range, len);
+	tmp = kstrdup(range, GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
 
-	if (dasd_evaluate_range_param(tmp, &from_str, &to_str, &features_str))
-		goto out_err;
+	if (dasd_evaluate_range_param(tmp, &from_str, &to_str, &features_str)) {
+		rc = -EINVAL;
+		goto out;
+	}
 
-	if (dasd_busid(from_str, &from_id0, &from_id1, &from))
-		goto out_err;
+	if (dasd_busid(from_str, &from_id0, &from_id1, &from)) {
+		rc = -EINVAL;
+		goto out;
+	}
 
 	to = from;
 	to_id0 = from_id0;
 	to_id1 = from_id1;
 	if (to_str) {
-		if (dasd_busid(to_str, &to_id0, &to_id1, &to))
-			goto out_err;
+		if (dasd_busid(to_str, &to_id0, &to_id1, &to)) {
+			rc = -EINVAL;
+			goto out;
+		}
 		if (from_id0 != to_id0 || from_id1 != to_id1 || from > to) {
 			pr_err("%s is not a valid device range\n", range);
-			goto out_err;
+			rc = -EINVAL;
+			goto out;
 		}
 	}
 
 	features = dasd_feature_list(features_str);
-	if (features < 0)
-		goto out_err;
+	if (features < 0) {
+		rc = -EINVAL;
+		goto out;
+	}
 	/* each device in dasd= parameter should be set initially online */
 	features |= DASD_FEATURE_INITIAL_ONLINE;
 	while (from <= to) {
 		sprintf(bus_id, "%01x.%01x.%04x", from_id0, from_id1, from++);
 		devmap = dasd_add_busid(bus_id, features);
-		if (IS_ERR(devmap))
-			return PTR_ERR(devmap);
+		if (IS_ERR(devmap)) {
+			rc = PTR_ERR(devmap);
+			goto out;
+		}
 	}
 
-	return 0;
+out:
+	kfree(tmp);
 
-out_err:
-	return -EINVAL;
+	return rc;
 }
 
 /*
@@ -735,13 +748,22 @@ static ssize_t
 dasd_ro_show(struct device *dev, struct device_attribute *attr, char *buf)
 {
 	struct dasd_devmap *devmap;
-	int ro_flag;
+	struct dasd_device *device;
+	int ro_flag = 0;
 
 	devmap = dasd_find_busid(dev_name(dev));
-	if (!IS_ERR(devmap))
-		ro_flag = (devmap->features & DASD_FEATURE_READONLY) != 0;
-	else
-		ro_flag = (DASD_FEATURE_DEFAULT & DASD_FEATURE_READONLY) != 0;
+	if (IS_ERR(devmap))
+		goto out;
+
+	ro_flag = !!(devmap->features & DASD_FEATURE_READONLY);
+
+	spin_lock(&dasd_devmap_lock);
+	device = devmap->device;
+	if (device)
+		ro_flag |= test_bit(DASD_FLAG_DEVICE_RO, &device->flags);
+	spin_unlock(&dasd_devmap_lock);
+
+out:
 	return snprintf(buf, PAGE_SIZE, ro_flag ? "1\n" : "0\n");
 }
 
@@ -764,7 +786,7 @@ dasd_ro_store(struct device *dev, struct device_attribute *attr,
 
 	device = dasd_device_from_cdev(cdev);
 	if (IS_ERR(device))
-		return PTR_ERR(device);
+		return count;
 
 	spin_lock_irqsave(get_ccwdev_lock(cdev), flags);
 	val = val || test_bit(DASD_FLAG_DEVICE_RO, &device->flags);
@@ -928,11 +950,14 @@ dasd_safe_offline_store(struct device *dev, struct device_attribute *attr,
 {
 	struct ccw_device *cdev = to_ccwdev(dev);
 	struct dasd_device *device;
+	unsigned long flags;
 	int rc;
 
-	device = dasd_device_from_cdev(cdev);
+	spin_lock_irqsave(get_ccwdev_lock(cdev), flags);
+	device = dasd_device_from_cdev_locked(cdev);
 	if (IS_ERR(device)) {
 		rc = PTR_ERR(device);
+		spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
 		goto out;
 	}
 
@@ -940,12 +965,14 @@ dasd_safe_offline_store(struct device *dev, struct device_attribute *attr,
 	    test_bit(DASD_FLAG_SAFE_OFFLINE_RUNNING, &device->flags)) {
 		/* Already doing offline processing */
 		dasd_put_device(device);
+		spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
 		rc = -EBUSY;
 		goto out;
 	}
 
 	set_bit(DASD_FLAG_SAFE_OFFLINE, &device->flags);
 	dasd_put_device(device);
+	spin_unlock_irqrestore(get_ccwdev_lock(cdev), flags);
 
 	rc = ccw_device_set_offline(cdev);
 
diff --git a/drivers/s390/block/scm_blk.c b/drivers/s390/block/scm_blk.c
index 3c2c84b72877..42018a20f2b7 100644
--- a/drivers/s390/block/scm_blk.c
+++ b/drivers/s390/block/scm_blk.c
@@ -13,6 +13,7 @@
 #include <linux/mempool.h>
 #include <linux/module.h>
 #include <linux/blkdev.h>
+#include <linux/blk-mq.h>
 #include <linux/genhd.h>
 #include <linux/slab.h>
 #include <linux/list.h>
@@ -42,7 +43,6 @@ static void __scm_free_rq(struct scm_request *scmrq)
 	struct aob_rq_header *aobrq = to_aobrq(scmrq);
 
 	free_page((unsigned long) scmrq->aob);
-	__scm_free_rq_cluster(scmrq);
 	kfree(scmrq->request);
 	kfree(aobrq);
 }
@@ -82,9 +82,6 @@ static int __scm_alloc_rq(void)
 	if (!scmrq->request)
 		goto free;
 
-	if (__scm_alloc_rq_cluster(scmrq))
-		goto free;
-
 	INIT_LIST_HEAD(&scmrq->list);
 	spin_lock_irq(&list_lock);
 	list_add(&scmrq->list, &inactive_requests);
@@ -114,13 +111,13 @@ static struct scm_request *scm_request_fetch(void)
 {
 	struct scm_request *scmrq = NULL;
 
-	spin_lock(&list_lock);
+	spin_lock_irq(&list_lock);
 	if (list_empty(&inactive_requests))
 		goto out;
 	scmrq = list_first_entry(&inactive_requests, struct scm_request, list);
 	list_del(&scmrq->list);
 out:
-	spin_unlock(&list_lock);
+	spin_unlock_irq(&list_lock);
 	return scmrq;
 }
 
@@ -234,130 +231,123 @@ static inline void scm_request_init(struct scm_blk_dev *bdev,
 	scmrq->error = BLK_STS_OK;
 	/* We don't use all msbs - place aidaws at the end of the aob page. */
 	scmrq->next_aidaw = (void *) &aob->msb[nr_requests_per_io];
-	scm_request_cluster_init(scmrq);
 }
 
-static void scm_ensure_queue_restart(struct scm_blk_dev *bdev)
-{
-	if (atomic_read(&bdev->queued_reqs)) {
-		/* Queue restart is triggered by the next interrupt. */
-		return;
-	}
-	blk_delay_queue(bdev->rq, SCM_QUEUE_DELAY);
-}
-
-void scm_request_requeue(struct scm_request *scmrq)
+static void scm_request_requeue(struct scm_request *scmrq)
 {
 	struct scm_blk_dev *bdev = scmrq->bdev;
 	int i;
 
-	scm_release_cluster(scmrq);
 	for (i = 0; i < nr_requests_per_io && scmrq->request[i]; i++)
-		blk_requeue_request(bdev->rq, scmrq->request[i]);
+		blk_mq_requeue_request(scmrq->request[i], false);
 
 	atomic_dec(&bdev->queued_reqs);
 	scm_request_done(scmrq);
-	scm_ensure_queue_restart(bdev);
+	blk_mq_kick_requeue_list(bdev->rq);
 }
 
-void scm_request_finish(struct scm_request *scmrq)
+static void scm_request_finish(struct scm_request *scmrq)
 {
 	struct scm_blk_dev *bdev = scmrq->bdev;
 	int i;
 
-	scm_release_cluster(scmrq);
-	for (i = 0; i < nr_requests_per_io && scmrq->request[i]; i++)
-		blk_end_request_all(scmrq->request[i], scmrq->error);
+	for (i = 0; i < nr_requests_per_io && scmrq->request[i]; i++) {
+		if (scmrq->error)
+			blk_mq_end_request(scmrq->request[i], scmrq->error);
+		else
+			blk_mq_complete_request(scmrq->request[i]);
+	}
 
 	atomic_dec(&bdev->queued_reqs);
 	scm_request_done(scmrq);
 }
 
-static int scm_request_start(struct scm_request *scmrq)
+static void scm_request_start(struct scm_request *scmrq)
 {
 	struct scm_blk_dev *bdev = scmrq->bdev;
-	int ret;
 
 	atomic_inc(&bdev->queued_reqs);
-	if (!scmrq->aob->request.msb_count) {
-		scm_request_requeue(scmrq);
-		return -EINVAL;
-	}
-
-	ret = eadm_start_aob(scmrq->aob);
-	if (ret) {
+	if (eadm_start_aob(scmrq->aob)) {
 		SCM_LOG(5, "no subchannel");
 		scm_request_requeue(scmrq);
 	}
-	return ret;
 }
 
-static void scm_blk_request(struct request_queue *rq)
+struct scm_queue {
+	struct scm_request *scmrq;
+	spinlock_t lock;
+};
+
+static int scm_blk_request(struct blk_mq_hw_ctx *hctx,
+			   const struct blk_mq_queue_data *qd)
 {
-	struct scm_device *scmdev = rq->queuedata;
+	struct scm_device *scmdev = hctx->queue->queuedata;
 	struct scm_blk_dev *bdev = dev_get_drvdata(&scmdev->dev);
-	struct scm_request *scmrq = NULL;
-	struct request *req;
+	struct scm_queue *sq = hctx->driver_data;
+	struct request *req = qd->rq;
+	struct scm_request *scmrq;
 
-	while ((req = blk_peek_request(rq))) {
-		if (!scm_permit_request(bdev, req))
-			goto out;
+	spin_lock(&sq->lock);
+	if (!scm_permit_request(bdev, req)) {
+		spin_unlock(&sq->lock);
+		return BLK_MQ_RQ_QUEUE_BUSY;
+	}
 
+	scmrq = sq->scmrq;
+	if (!scmrq) {
+		scmrq = scm_request_fetch();
 		if (!scmrq) {
-			scmrq = scm_request_fetch();
-			if (!scmrq) {
-				SCM_LOG(5, "no request");
-				goto out;
-			}
-			scm_request_init(bdev, scmrq);
+			SCM_LOG(5, "no request");
+			spin_unlock(&sq->lock);
+			return BLK_MQ_RQ_QUEUE_BUSY;
 		}
-		scm_request_set(scmrq, req);
+		scm_request_init(bdev, scmrq);
+		sq->scmrq = scmrq;
+	}
+	scm_request_set(scmrq, req);
 
-		if (!scm_reserve_cluster(scmrq)) {
-			SCM_LOG(5, "cluster busy");
-			scm_request_set(scmrq, NULL);
-			if (scmrq->aob->request.msb_count)
-				goto out;
+	if (scm_request_prepare(scmrq)) {
+		SCM_LOG(5, "aidaw alloc failed");
+		scm_request_set(scmrq, NULL);
 
-			scm_request_done(scmrq);
-			return;
-		}
+		if (scmrq->aob->request.msb_count)
+			scm_request_start(scmrq);
 
-		if (scm_need_cluster_request(scmrq)) {
-			if (scmrq->aob->request.msb_count) {
-				/* Start cluster requests separately. */
-				scm_request_set(scmrq, NULL);
-				if (scm_request_start(scmrq))
-					return;
-			} else {
-				atomic_inc(&bdev->queued_reqs);
-				blk_start_request(req);
-				scm_initiate_cluster_request(scmrq);
-			}
-			scmrq = NULL;
-			continue;
-		}
+		sq->scmrq = NULL;
+		spin_unlock(&sq->lock);
+		return BLK_MQ_RQ_QUEUE_BUSY;
+	}
+	blk_mq_start_request(req);
 
-		if (scm_request_prepare(scmrq)) {
-			SCM_LOG(5, "aidaw alloc failed");
-			scm_request_set(scmrq, NULL);
-			goto out;
-		}
-		blk_start_request(req);
+	if (qd->last || scmrq->aob->request.msb_count == nr_requests_per_io) {
+		scm_request_start(scmrq);
+		sq->scmrq = NULL;
+	}
+	spin_unlock(&sq->lock);
+	return BLK_MQ_RQ_QUEUE_OK;
+}
 
-		if (scmrq->aob->request.msb_count < nr_requests_per_io)
-			continue;
+static int scm_blk_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
+			     unsigned int idx)
+{
+	struct scm_queue *qd = kzalloc(sizeof(*qd), GFP_KERNEL);
 
-		if (scm_request_start(scmrq))
-			return;
+	if (!qd)
+		return -ENOMEM;
 
-		scmrq = NULL;
-	}
-out:
-	if (scmrq)
-		scm_request_start(scmrq);
-	else
-		scm_ensure_queue_restart(bdev);
+	spin_lock_init(&qd->lock);
+	hctx->driver_data = qd;
+
+	return 0;
+}
+
+static void scm_blk_exit_hctx(struct blk_mq_hw_ctx *hctx, unsigned int idx)
+{
+	struct scm_queue *qd = hctx->driver_data;
+
+	WARN_ON(qd->scmrq);
+	kfree(hctx->driver_data);
+	hctx->driver_data = NULL;
 }
 
 static void __scmrq_log_error(struct scm_request *scmrq)
@@ -377,21 +367,6 @@ static void __scmrq_log_error(struct scm_request *scmrq)
 		       scmrq->error);
 }
 
-void scm_blk_irq(struct scm_device *scmdev, void *data, blk_status_t error)
-{
-	struct scm_request *scmrq = data;
-	struct scm_blk_dev *bdev = scmrq->bdev;
-
-	scmrq->error = error;
-	if (error)
-		__scmrq_log_error(scmrq);
-
-	spin_lock(&bdev->lock);
-	list_add_tail(&scmrq->list, &bdev->finished_requests);
-	spin_unlock(&bdev->lock);
-	tasklet_hi_schedule(&bdev->tasklet);
-}
-
 static void scm_blk_handle_error(struct scm_request *scmrq)
 {
 	struct scm_blk_dev *bdev = scmrq->bdev;
@@ -419,54 +394,46 @@ restart:
 		return;
 
 requeue:
-	spin_lock_irqsave(&bdev->rq_lock, flags);
 	scm_request_requeue(scmrq);
-	spin_unlock_irqrestore(&bdev->rq_lock, flags);
 }
 
-static void scm_blk_tasklet(struct scm_blk_dev *bdev)
+void scm_blk_irq(struct scm_device *scmdev, void *data, blk_status_t error)
 {
-	struct scm_request *scmrq;
-	unsigned long flags;
-
-	spin_lock_irqsave(&bdev->lock, flags);
-	while (!list_empty(&bdev->finished_requests)) {
-		scmrq = list_first_entry(&bdev->finished_requests,
-					 struct scm_request, list);
-		list_del(&scmrq->list);
-		spin_unlock_irqrestore(&bdev->lock, flags);
+	struct scm_request *scmrq = data;
 
-		if (scmrq->error && scmrq->retries-- > 0) {
+	scmrq->error = error;
+	if (error) {
+		__scmrq_log_error(scmrq);
+		if (scmrq->retries-- > 0) {
 			scm_blk_handle_error(scmrq);
-
-			/* Request restarted or requeued, handle next. */
-			spin_lock_irqsave(&bdev->lock, flags);
-			continue;
+			return;
 		}
+	}
 
-		if (scm_test_cluster_request(scmrq)) {
-			scm_cluster_request_irq(scmrq);
-			spin_lock_irqsave(&bdev->lock, flags);
-			continue;
-		}
+	scm_request_finish(scmrq);
+}
 
-		scm_request_finish(scmrq);
-		spin_lock_irqsave(&bdev->lock, flags);
-	}
-	spin_unlock_irqrestore(&bdev->lock, flags);
-	/* Look out for more requests. */
-	blk_run_queue(bdev->rq);
+static void scm_blk_request_done(struct request *req)
+{
+	blk_mq_end_request(req, 0);
 }
 
 static const struct block_device_operations scm_blk_devops = {
 	.owner = THIS_MODULE,
 };
 
+static const struct blk_mq_ops scm_mq_ops = {
+	.queue_rq = scm_blk_request,
+	.complete = scm_blk_request_done,
+	.init_hctx = scm_blk_init_hctx,
+	.exit_hctx = scm_blk_exit_hctx,
+};
+
 int scm_blk_dev_setup(struct scm_blk_dev *bdev, struct scm_device *scmdev)
 {
-	struct request_queue *rq;
-	int len, ret = -ENOMEM;
 	unsigned int devindex, nr_max_blk;
+	struct request_queue *rq;
+	int len, ret;
 
 	devindex = atomic_inc_return(&nr_devices) - 1;
 	/* scma..scmz + scmaa..scmzz */
@@ -477,18 +444,23 @@ int scm_blk_dev_setup(struct scm_blk_dev *bdev, struct scm_device *scmdev)
 
 	bdev->scmdev = scmdev;
 	bdev->state = SCM_OPER;
-	spin_lock_init(&bdev->rq_lock);
 	spin_lock_init(&bdev->lock);
-	INIT_LIST_HEAD(&bdev->finished_requests);
 	atomic_set(&bdev->queued_reqs, 0);
-	tasklet_init(&bdev->tasklet,
-		     (void (*)(unsigned long)) scm_blk_tasklet,
-		     (unsigned long) bdev);
 
-	rq = blk_init_queue(scm_blk_request, &bdev->rq_lock);
-	if (!rq)
+	bdev->tag_set.ops = &scm_mq_ops;
+	bdev->tag_set.nr_hw_queues = nr_requests;
+	bdev->tag_set.queue_depth = nr_requests_per_io * nr_requests;
+	bdev->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
+
+	ret = blk_mq_alloc_tag_set(&bdev->tag_set);
+	if (ret)
 		goto out;
 
+	rq = blk_mq_init_queue(&bdev->tag_set);
+	if (IS_ERR(rq)) {
+		ret = PTR_ERR(rq);
+		goto out_tag;
+	}
 	bdev->rq = rq;
 	nr_max_blk = min(scmdev->nr_max_block,
 			 (unsigned int) (PAGE_SIZE / sizeof(struct aidaw)));
@@ -498,12 +470,12 @@ int scm_blk_dev_setup(struct scm_blk_dev *bdev, struct scm_device *scmdev)
 	blk_queue_max_segments(rq, nr_max_blk);
 	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, rq);
 	queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, rq);
-	scm_blk_dev_cluster_setup(bdev);
 
 	bdev->gendisk = alloc_disk(SCM_NR_PARTS);
-	if (!bdev->gendisk)
+	if (!bdev->gendisk) {
+		ret = -ENOMEM;
 		goto out_queue;
-
+	}
 	rq->queuedata = scmdev;
 	bdev->gendisk->private_data = scmdev;
 	bdev->gendisk->fops = &scm_blk_devops;
@@ -528,6 +500,8 @@ int scm_blk_dev_setup(struct scm_blk_dev *bdev, struct scm_device *scmdev)
 
 out_queue:
 	blk_cleanup_queue(rq);
+out_tag:
+	blk_mq_free_tag_set(&bdev->tag_set);
 out:
 	atomic_dec(&nr_devices);
 	return ret;
@@ -535,9 +509,9 @@ out:
 
 void scm_blk_dev_cleanup(struct scm_blk_dev *bdev)
 {
-	tasklet_kill(&bdev->tasklet);
 	del_gendisk(bdev->gendisk);
 	blk_cleanup_queue(bdev->gendisk->queue);
+	blk_mq_free_tag_set(&bdev->tag_set);
 	put_disk(bdev->gendisk);
 }
 
@@ -558,7 +532,7 @@ static bool __init scm_blk_params_valid(void)
 	if (!nr_requests_per_io || nr_requests_per_io > 64)
 		return false;
 
-	return scm_cluster_size_valid();
+	return true;
 }
 
 static int __init scm_blk_init(void)
diff --git a/drivers/s390/block/scm_blk.h b/drivers/s390/block/scm_blk.h
index cd598d1a4eae..71288dd9dd7f 100644
--- a/drivers/s390/block/scm_blk.h
+++ b/drivers/s390/block/scm_blk.h
@@ -4,6 +4,7 @@
 #include <linux/interrupt.h>
 #include <linux/spinlock.h>
 #include <linux/blkdev.h>
+#include <linux/blk-mq.h>
 #include <linux/genhd.h>
 #include <linux/list.h>
 
@@ -14,18 +15,14 @@
 #define SCM_QUEUE_DELAY 5
 
 struct scm_blk_dev {
-	struct tasklet_struct tasklet;
 	struct request_queue *rq;
 	struct gendisk *gendisk;
+	struct blk_mq_tag_set tag_set;
 	struct scm_device *scmdev;
-	spinlock_t rq_lock;	/* guard the request queue */
-	spinlock_t lock;	/* guard the rest of the blockdev */
+	spinlock_t lock;
 	atomic_t queued_reqs;
 	enum {SCM_OPER, SCM_WR_PROHIBIT} state;
 	struct list_head finished_requests;
-#ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE
-	struct list_head cluster_list;
-#endif
 };
 
 struct scm_request {
@@ -36,13 +33,6 @@ struct scm_request {
 	struct list_head list;
 	u8 retries;
 	blk_status_t error;
-#ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE
-	struct {
-		enum {CLUSTER_NONE, CLUSTER_READ, CLUSTER_WRITE} state;
-		struct list_head list;
-		void **buf;
-	} cluster;
-#endif
 };
 
 #define to_aobrq(rq) container_of((void *) rq, struct aob_rq_header, data)
@@ -52,55 +42,11 @@ void scm_blk_dev_cleanup(struct scm_blk_dev *);
 void scm_blk_set_available(struct scm_blk_dev *);
 void scm_blk_irq(struct scm_device *, void *, blk_status_t);
 
-void scm_request_finish(struct scm_request *);
-void scm_request_requeue(struct scm_request *);
-
 struct aidaw *scm_aidaw_fetch(struct scm_request *scmrq, unsigned int bytes);
 
 int scm_drv_init(void);
 void scm_drv_cleanup(void);
 
-#ifdef CONFIG_SCM_BLOCK_CLUSTER_WRITE
-void __scm_free_rq_cluster(struct scm_request *);
-int __scm_alloc_rq_cluster(struct scm_request *);
-void scm_request_cluster_init(struct scm_request *);
-bool scm_reserve_cluster(struct scm_request *);
-void scm_release_cluster(struct scm_request *);
-void scm_blk_dev_cluster_setup(struct scm_blk_dev *);
-bool scm_need_cluster_request(struct scm_request *);
-void scm_initiate_cluster_request(struct scm_request *);
-void scm_cluster_request_irq(struct scm_request *);
-bool scm_test_cluster_request(struct scm_request *);
-bool scm_cluster_size_valid(void);
-#else /* CONFIG_SCM_BLOCK_CLUSTER_WRITE */
-static inline void __scm_free_rq_cluster(struct scm_request *scmrq) {}
-static inline int __scm_alloc_rq_cluster(struct scm_request *scmrq)
-{
-	return 0;
-}
-static inline void scm_request_cluster_init(struct scm_request *scmrq) {}
-static inline bool scm_reserve_cluster(struct scm_request *scmrq)
-{
-	return true;
-}
-static inline void scm_release_cluster(struct scm_request *scmrq) {}
-static inline void scm_blk_dev_cluster_setup(struct scm_blk_dev *bdev) {}
-static inline bool scm_need_cluster_request(struct scm_request *scmrq)
-{
-	return false;
-}
-static inline void scm_initiate_cluster_request(struct scm_request *scmrq) {}
-static inline void scm_cluster_request_irq(struct scm_request *scmrq) {}
-static inline bool scm_test_cluster_request(struct scm_request *scmrq)
-{
-	return false;
-}
-static inline bool scm_cluster_size_valid(void)
-{
-	return true;
-}
-#endif /* CONFIG_SCM_BLOCK_CLUSTER_WRITE */
-
 extern debug_info_t *scm_debug;
 
 #define SCM_LOG(imp, txt) do {					\
diff --git a/drivers/s390/block/scm_blk_cluster.c b/drivers/s390/block/scm_blk_cluster.c
deleted file mode 100644
index 7497ddde2dd6..000000000000
--- a/drivers/s390/block/scm_blk_cluster.c
+++ /dev/null
@@ -1,255 +0,0 @@
-/*
- * Block driver for s390 storage class memory.
- *
- * Copyright IBM Corp. 2012
- * Author(s): Sebastian Ott <sebott@linux.vnet.ibm.com>
- */
-
-#include <linux/spinlock.h>
-#include <linux/module.h>
-#include <linux/blkdev.h>
-#include <linux/genhd.h>
-#include <linux/slab.h>
-#include <linux/list.h>
-#include <asm/eadm.h>
-#include "scm_blk.h"
-
-static unsigned int write_cluster_size = 64;
-module_param(write_cluster_size, uint, S_IRUGO);
-MODULE_PARM_DESC(write_cluster_size,
-		 "Number of pages used for contiguous writes.");
-
-#define CLUSTER_SIZE (write_cluster_size * PAGE_SIZE)
-
-void __scm_free_rq_cluster(struct scm_request *scmrq)
-{
-	int i;
-
-	if (!scmrq->cluster.buf)
-		return;
-
-	for (i = 0; i < 2 * write_cluster_size; i++)
-		free_page((unsigned long) scmrq->cluster.buf[i]);
-
-	kfree(scmrq->cluster.buf);
-}
-
-int __scm_alloc_rq_cluster(struct scm_request *scmrq)
-{
-	int i;
-
-	scmrq->cluster.buf = kzalloc(sizeof(void *) * 2 * write_cluster_size,
-				 GFP_KERNEL);
-	if (!scmrq->cluster.buf)
-		return -ENOMEM;
-
-	for (i = 0; i < 2 * write_cluster_size; i++) {
-		scmrq->cluster.buf[i] = (void *) get_zeroed_page(GFP_DMA);
-		if (!scmrq->cluster.buf[i])
-			return -ENOMEM;
-	}
-	INIT_LIST_HEAD(&scmrq->cluster.list);
-	return 0;
-}
-
-void scm_request_cluster_init(struct scm_request *scmrq)
-{
-	scmrq->cluster.state = CLUSTER_NONE;
-}
-
-static bool clusters_intersect(struct request *A, struct request *B)
-{
-	unsigned long firstA, lastA, firstB, lastB;
-
-	firstA = ((u64) blk_rq_pos(A) << 9) / CLUSTER_SIZE;
-	lastA = (((u64) blk_rq_pos(A) << 9) +
-		    blk_rq_bytes(A) - 1) / CLUSTER_SIZE;
-
-	firstB = ((u64) blk_rq_pos(B) << 9) / CLUSTER_SIZE;
-	lastB = (((u64) blk_rq_pos(B) << 9) +
-		    blk_rq_bytes(B) - 1) / CLUSTER_SIZE;
-
-	return (firstB <= lastA && firstA <= lastB);
-}
-
-bool scm_reserve_cluster(struct scm_request *scmrq)
-{
-	struct request *req = scmrq->request[scmrq->aob->request.msb_count];
-	struct scm_blk_dev *bdev = scmrq->bdev;
-	struct scm_request *iter;
-	int pos, add = 1;
-
-	if (write_cluster_size == 0)
-		return true;
-
-	spin_lock(&bdev->lock);
-	list_for_each_entry(iter, &bdev->cluster_list, cluster.list) {
-		if (iter == scmrq) {
-			/*
-			 * We don't have to use clusters_intersect here, since
-			 * cluster requests are always started separately.
-			 */
-			add = 0;
-			continue;
-		}
-		for (pos = 0; pos < iter->aob->request.msb_count; pos++) {
-			if (clusters_intersect(req, iter->request[pos]) &&
-			    (rq_data_dir(req) == WRITE ||
-			     rq_data_dir(iter->request[pos]) == WRITE)) {
-				spin_unlock(&bdev->lock);
-				return false;
-			}
-		}
-	}
-	if (add)
-		list_add(&scmrq->cluster.list, &bdev->cluster_list);
-	spin_unlock(&bdev->lock);
-
-	return true;
-}
-
-void scm_release_cluster(struct scm_request *scmrq)
-{
-	struct scm_blk_dev *bdev = scmrq->bdev;
-	unsigned long flags;
-
-	if (write_cluster_size == 0)
-		return;
-
-	spin_lock_irqsave(&bdev->lock, flags);
-	list_del(&scmrq->cluster.list);
-	spin_unlock_irqrestore(&bdev->lock, flags);
-}
-
-void scm_blk_dev_cluster_setup(struct scm_blk_dev *bdev)
-{
-	INIT_LIST_HEAD(&bdev->cluster_list);
-	blk_queue_io_opt(bdev->rq, CLUSTER_SIZE);
-}
-
-static int scm_prepare_cluster_request(struct scm_request *scmrq)
-{
-	struct scm_blk_dev *bdev = scmrq->bdev;
-	struct scm_device *scmdev = bdev->gendisk->private_data;
-	struct request *req = scmrq->request[0];
-	struct msb *msb = &scmrq->aob->msb[0];
-	struct req_iterator iter;
-	struct aidaw *aidaw;
-	struct bio_vec bv;
-	int i = 0;
-	u64 addr;
-
-	switch (scmrq->cluster.state) {
-	case CLUSTER_NONE:
-		scmrq->cluster.state = CLUSTER_READ;
-		/* fall through */
-	case CLUSTER_READ:
-		msb->bs = MSB_BS_4K;
-		msb->oc = MSB_OC_READ;
-		msb->flags = MSB_FLAG_IDA;
-		msb->blk_count = write_cluster_size;
-
-		addr = scmdev->address + ((u64) blk_rq_pos(req) << 9);
-		msb->scm_addr = round_down(addr, CLUSTER_SIZE);
-
-		if (msb->scm_addr !=
-		    round_down(addr + (u64) blk_rq_bytes(req) - 1,
-			       CLUSTER_SIZE))
-			msb->blk_count = 2 * write_cluster_size;
-
-		aidaw = scm_aidaw_fetch(scmrq, msb->blk_count * PAGE_SIZE);
-		if (!aidaw)
-			return -ENOMEM;
-
-		scmrq->aob->request.msb_count = 1;
-		msb->data_addr = (u64) aidaw;
-		for (i = 0; i < msb->blk_count; i++) {
-			aidaw->data_addr = (u64) scmrq->cluster.buf[i];
-			aidaw++;
-		}
-
-		break;
-	case CLUSTER_WRITE:
-		aidaw = (void *) msb->data_addr;
-		msb->oc = MSB_OC_WRITE;
-
-		for (addr = msb->scm_addr;
-		     addr < scmdev->address + ((u64) blk_rq_pos(req) << 9);
-		     addr += PAGE_SIZE) {
-			aidaw->data_addr = (u64) scmrq->cluster.buf[i];
-			aidaw++;
-			i++;
-		}
-		rq_for_each_segment(bv, req, iter) {
-			aidaw->data_addr = (u64) page_address(bv.bv_page);
-			aidaw++;
-			i++;
-		}
-		for (; i < msb->blk_count; i++) {
-			aidaw->data_addr = (u64) scmrq->cluster.buf[i];
-			aidaw++;
-		}
-		break;
-	}
-	return 0;
-}
-
-bool scm_need_cluster_request(struct scm_request *scmrq)
-{
-	int pos = scmrq->aob->request.msb_count;
-
-	if (rq_data_dir(scmrq->request[pos]) == READ)
-		return false;
-
-	return blk_rq_bytes(scmrq->request[pos]) < CLUSTER_SIZE;
-}
-
-/* Called with queue lock held. */
-void scm_initiate_cluster_request(struct scm_request *scmrq)
-{
-	if (scm_prepare_cluster_request(scmrq))
-		goto requeue;
-	if (eadm_start_aob(scmrq->aob))
-		goto requeue;
-	return;
-requeue:
-	scm_request_requeue(scmrq);
-}
-
-bool scm_test_cluster_request(struct scm_request *scmrq)
-{
-	return scmrq->cluster.state != CLUSTER_NONE;
-}
-
-void scm_cluster_request_irq(struct scm_request *scmrq)
-{
-	struct scm_blk_dev *bdev = scmrq->bdev;
-	unsigned long flags;
-
-	switch (scmrq->cluster.state) {
-	case CLUSTER_NONE:
-		BUG();
-		break;
-	case CLUSTER_READ:
-		if (scmrq->error) {
-			scm_request_finish(scmrq);
-			break;
-		}
-		scmrq->cluster.state = CLUSTER_WRITE;
-		spin_lock_irqsave(&bdev->rq_lock, flags);
-		scm_initiate_cluster_request(scmrq);
-		spin_unlock_irqrestore(&bdev->rq_lock, flags);
-		break;
-	case CLUSTER_WRITE:
-		scm_request_finish(scmrq);
-		break;
-	}
-}
-
-bool scm_cluster_size_valid(void)
-{
-	if (write_cluster_size == 1 || write_cluster_size > 128)
-		return false;
-
-	return !(write_cluster_size & (write_cluster_size - 1));
-}