Merge branch 'for-2.6.38/core' of git://git.kernel.dk/linux-2.6-block

* 'for-2.6.38/core' of git://git.kernel.dk/linux-2.6-block: (43 commits)
  block: ensure that completion error gets properly traced
  blktrace: add missing probe argument to block_bio_complete
  block cfq: don't use atomic_t for cfq_group
  block cfq: don't use atomic_t for cfq_queue
  block: trace event block fix unassigned field
  block: add internal hd part table references
  block: fix accounting bug on cross partition merges
  kref: add kref_test_and_get
  bio-integrity: mark kintegrityd_wq highpri and CPU intensive
  block: make kblockd_workqueue smarter
  Revert "sd: implement sd_check_events()"
  block: Clean up exit_io_context() source code.
  Fix compile warnings due to missing removal of a 'ret' variable
  fs/block: type signature of major_to_index(int) to major_to_index(unsigned)
  block: convert !IS_ERR(p) && p to !IS_ERR_NOR_NULL(p)
  cfq-iosched: don't check cfqg in choose_service_tree()
  fs/splice: Pull buf->ops->confirm() from splice_from_pipe actors
  cdrom: export cdrom_check_events()
  sd: implement sd_check_events()
  sr: implement sr_check_events()
  ...

Documentation/cgroups/blkio-controller.txt

@@ -89,6 +89,33 @@ Throttling/Upper Limit policy
 
  Limits for writes can be put using blkio.write_bps_device file.
 
+Hierarchical Cgroups
+====================
+- Currently none of the IO control policy supports hierarhical groups. But
+  cgroup interface does allow creation of hierarhical cgroups and internally
+  IO policies treat them as flat hierarchy.
+
+  So this patch will allow creation of cgroup hierarhcy but at the backend
+  everything will be treated as flat. So if somebody created a hierarchy like
+  as follows.
+
+			root
+			/  \
+		     test1 test2
+			|
+		     test3
+
+  CFQ and throttling will practically treat all groups at same level.
+
+				pivot
+			     /  |   \  \
+			root  test1 test2  test3
+
+  Down the line we can implement hierarchical accounting/control support
+  and also introduce a new cgroup file "use_hierarchy" which will control
+  whether cgroup hierarchy is viewed as flat or hierarchical by the policy..
+  This is how memory controller also has implemented the things.
+
 Various user visible config options
 ===================================
 CONFIG_BLK_CGROUP

block/blk-cgroup.c

@@ -1452,10 +1452,6 @@ blkiocg_create(struct cgroup_subsys *subsys, struct cgroup *cgroup)
 		goto done;
 	}
 
-	/* Currently we do not support hierarchy deeper than two level (0,1) */
-	if (parent != cgroup->top_cgroup)
-		return ERR_PTR(-EPERM);
-
 	blkcg = kzalloc(sizeof(*blkcg), GFP_KERNEL);
 	if (!blkcg)
 		return ERR_PTR(-ENOMEM);

block/blk-core.c

@@ -33,7 +33,7 @@
 
 #include "blk.h"
 
-EXPORT_TRACEPOINT_SYMBOL_GPL(block_remap);
+EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_remap);
 EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_remap);
 EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete);
 
@@ -64,13 +64,27 @@ static void drive_stat_acct(struct request *rq, int new_io)
 		return;
 
 	cpu = part_stat_lock();
-	part = disk_map_sector_rcu(rq->rq_disk, blk_rq_pos(rq));
 
-	if (!new_io)
+	if (!new_io) {
+		part = rq->part;
 		part_stat_inc(cpu, part, merges[rw]);
-	else {
+	} else {
+		part = disk_map_sector_rcu(rq->rq_disk, blk_rq_pos(rq));
+		if (!hd_struct_try_get(part)) {
+			/*
+			 * The partition is already being removed,
+			 * the request will be accounted on the disk only
+			 *
+			 * We take a reference on disk->part0 although that
+			 * partition will never be deleted, so we can treat
+			 * it as any other partition.
+			 */
+			part = &rq->rq_disk->part0;
+			hd_struct_get(part);
+		}
 		part_round_stats(cpu, part);
 		part_inc_in_flight(part, rw);
+		rq->part = part;
 	}
 
 	part_stat_unlock();
@@ -128,6 +142,7 @@ void blk_rq_init(struct request_queue *q, struct request *rq)
 	rq->ref_count = 1;
 	rq->start_time = jiffies;
 	set_start_time_ns(rq);
+	rq->part = NULL;
 }
 EXPORT_SYMBOL(blk_rq_init);
 
@@ -1329,9 +1344,9 @@ static inline void blk_partition_remap(struct bio *bio)
 		bio->bi_sector += p->start_sect;
 		bio->bi_bdev = bdev->bd_contains;
 
-		trace_block_remap(bdev_get_queue(bio->bi_bdev), bio,
-				    bdev->bd_dev,
-				    bio->bi_sector - p->start_sect);
+		trace_block_bio_remap(bdev_get_queue(bio->bi_bdev), bio,
+				      bdev->bd_dev,
+				      bio->bi_sector - p->start_sect);
 	}
 }
 
@@ -1500,7 +1515,7 @@ static inline void __generic_make_request(struct bio *bio)
 			goto end_io;
 
 		if (old_sector != -1)
-			trace_block_remap(q, bio, old_dev, old_sector);
+			trace_block_bio_remap(q, bio, old_dev, old_sector);
 
 		old_sector = bio->bi_sector;
 		old_dev = bio->bi_bdev->bd_dev;
@@ -1776,7 +1791,7 @@ static void blk_account_io_completion(struct request *req, unsigned int bytes)
 		int cpu;
 
 		cpu = part_stat_lock();
-		part = disk_map_sector_rcu(req->rq_disk, blk_rq_pos(req));
+		part = req->part;
 		part_stat_add(cpu, part, sectors[rw], bytes >> 9);
 		part_stat_unlock();
 	}
@@ -1796,13 +1811,14 @@ static void blk_account_io_done(struct request *req)
 		int cpu;
 
 		cpu = part_stat_lock();
-		part = disk_map_sector_rcu(req->rq_disk, blk_rq_pos(req));
+		part = req->part;
 
 		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);
 
+		hd_struct_put(part);
 		part_stat_unlock();
 	}
 }
@@ -2606,7 +2622,9 @@ int __init blk_dev_init(void)
 	BUILD_BUG_ON(__REQ_NR_BITS > 8 *
 			sizeof(((struct request *)0)->cmd_flags));
 
-	kblockd_workqueue = create_workqueue("kblockd");
+	/* used for unplugging and affects IO latency/throughput - HIGHPRI */
+	kblockd_workqueue = alloc_workqueue("kblockd",
+					    WQ_MEM_RECLAIM | WQ_HIGHPRI, 0);
 	if (!kblockd_workqueue)
 		panic("Failed to create kblockd\n");
 

block/blk-ioc.c

@@ -64,7 +64,7 @@ static void cfq_exit(struct io_context *ioc)
 	rcu_read_unlock();
 }
 
-/* Called by the exitting task */
+/* Called by the exiting task */
 void exit_io_context(struct task_struct *task)
 {
 	struct io_context *ioc;
@@ -74,10 +74,9 @@ void exit_io_context(struct task_struct *task)
 	task->io_context = NULL;
 	task_unlock(task);
 
-	if (atomic_dec_and_test(&ioc->nr_tasks)) {
+	if (atomic_dec_and_test(&ioc->nr_tasks))
 		cfq_exit(ioc);
 
-	}
 	put_io_context(ioc);
 }
 

block/blk-merge.c

@@ -351,11 +351,12 @@ static void blk_account_io_merge(struct request *req)
 		int cpu;
 
 		cpu = part_stat_lock();
-		part = disk_map_sector_rcu(req->rq_disk, blk_rq_pos(req));
+		part = req->part;
 
 		part_round_stats(cpu, part);
 		part_dec_in_flight(part, rq_data_dir(req));
 
+		hd_struct_put(part);
 		part_stat_unlock();
 	}
 }

block/cfq-iosched.c

@@ -87,7 +87,6 @@ struct cfq_rb_root {
 	unsigned count;
 	unsigned total_weight;
 	u64 min_vdisktime;
-	struct rb_node *active;
 };
 #define CFQ_RB_ROOT	(struct cfq_rb_root) { .rb = RB_ROOT, .left = NULL, \
 			.count = 0, .min_vdisktime = 0, }
@@ -97,7 +96,7 @@ struct cfq_rb_root {
  */
 struct cfq_queue {
 	/* reference count */
-	atomic_t ref;
+	int ref;
 	/* various state flags, see below */
 	unsigned int flags;
 	/* parent cfq_data */
@@ -180,7 +179,6 @@ struct cfq_group {
 	/* group service_tree key */
 	u64 vdisktime;
 	unsigned int weight;
-	bool on_st;
 
 	/* number of cfqq currently on this group */
 	int nr_cfqq;
@@ -209,7 +207,7 @@ struct cfq_group {
 	struct blkio_group blkg;
 #ifdef CONFIG_CFQ_GROUP_IOSCHED
 	struct hlist_node cfqd_node;
-	atomic_t ref;
+	int ref;
 #endif
 	/* number of requests that are on the dispatch list or inside driver */
 	int dispatched;
@@ -563,11 +561,6 @@ static void update_min_vdisktime(struct cfq_rb_root *st)
 	u64 vdisktime = st->min_vdisktime;
 	struct cfq_group *cfqg;
 
-	if (st->active) {
-		cfqg = rb_entry_cfqg(st->active);
-		vdisktime = cfqg->vdisktime;
-	}
-
 	if (st->left) {
 		cfqg = rb_entry_cfqg(st->left);
 		vdisktime = min_vdisktime(vdisktime, cfqg->vdisktime);
@@ -646,11 +639,11 @@ cfq_set_prio_slice(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 static inline bool cfq_slice_used(struct cfq_queue *cfqq)
 {
 	if (cfq_cfqq_slice_new(cfqq))
-		return 0;
+		return false;
 	if (time_before(jiffies, cfqq->slice_end))
-		return 0;
+		return false;
 
-	return 1;
+	return true;
 }
 
 /*
@@ -869,7 +862,7 @@ cfq_group_service_tree_add(struct cfq_data *cfqd, struct cfq_group *cfqg)
 	struct rb_node *n;
 
 	cfqg->nr_cfqq++;
-	if (cfqg->on_st)
+	if (!RB_EMPTY_NODE(&cfqg->rb_node))
 		return;
 
 	/*
@@ -885,7 +878,6 @@ cfq_group_service_tree_add(struct cfq_data *cfqd, struct cfq_group *cfqg)
 		cfqg->vdisktime = st->min_vdisktime;
 
 	__cfq_group_service_tree_add(st, cfqg);
-	cfqg->on_st = true;
 	st->total_weight += cfqg->weight;
 }
 
@@ -894,9 +886,6 @@ cfq_group_service_tree_del(struct cfq_data *cfqd, struct cfq_group *cfqg)
 {
 	struct cfq_rb_root *st = &cfqd->grp_service_tree;
 
-	if (st->active == &cfqg->rb_node)
-		st->active = NULL;
-
 	BUG_ON(cfqg->nr_cfqq < 1);
 	cfqg->nr_cfqq--;
 
@@ -905,7 +894,6 @@ cfq_group_service_tree_del(struct cfq_data *cfqd, struct cfq_group *cfqg)
 		return;
 
 	cfq_log_cfqg(cfqd, cfqg, "del_from_rr group");
-	cfqg->on_st = false;
 	st->total_weight -= cfqg->weight;
 	if (!RB_EMPTY_NODE(&cfqg->rb_node))
 		cfq_rb_erase(&cfqg->rb_node, st);
@@ -1026,7 +1014,7 @@ cfq_find_alloc_cfqg(struct cfq_data *cfqd, struct cgroup *cgroup, int create)
 	 * elevator which will be dropped by either elevator exit
 	 * or cgroup deletion path depending on who is exiting first.
 	 */
-	atomic_set(&cfqg->ref, 1);
+	cfqg->ref = 1;
 
 	/*
 	 * Add group onto cgroup list. It might happen that bdi->dev is
@@ -1071,7 +1059,7 @@ static struct cfq_group *cfq_get_cfqg(struct cfq_data *cfqd, int create)
 
 static inline struct cfq_group *cfq_ref_get_cfqg(struct cfq_group *cfqg)
 {
-	atomic_inc(&cfqg->ref);
+	cfqg->ref++;
 	return cfqg;
 }
 
@@ -1083,7 +1071,7 @@ static void cfq_link_cfqq_cfqg(struct cfq_queue *cfqq, struct cfq_group *cfqg)
 
 	cfqq->cfqg = cfqg;
 	/* cfqq reference on cfqg */
-	atomic_inc(&cfqq->cfqg->ref);
+	cfqq->cfqg->ref++;
 }
 
 static void cfq_put_cfqg(struct cfq_group *cfqg)
@@ -1091,11 +1079,12 @@ static void cfq_put_cfqg(struct cfq_group *cfqg)
 	struct cfq_rb_root *st;
 	int i, j;
 
-	BUG_ON(atomic_read(&cfqg->ref) <= 0);
-	if (!atomic_dec_and_test(&cfqg->ref))
+	BUG_ON(cfqg->ref <= 0);
+	cfqg->ref--;
+	if (cfqg->ref)
 		return;
 	for_each_cfqg_st(cfqg, i, j, st)
-		BUG_ON(!RB_EMPTY_ROOT(&st->rb) || st->active != NULL);
+		BUG_ON(!RB_EMPTY_ROOT(&st->rb));
 	kfree(cfqg);
 }
 
@@ -1200,7 +1189,7 @@ static void cfq_service_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 			cfq_group_service_tree_del(cfqd, cfqq->cfqg);
 		cfqq->orig_cfqg = cfqq->cfqg;
 		cfqq->cfqg = &cfqd->root_group;
-		atomic_inc(&cfqd->root_group.ref);
+		cfqd->root_group.ref++;
 		group_changed = 1;
 	} else if (!cfqd->cfq_group_isolation
 		   && cfqq_type(cfqq) == SYNC_WORKLOAD && cfqq->orig_cfqg) {
@@ -1687,9 +1676,6 @@ __cfq_slice_expired(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 	if (cfqq == cfqd->active_queue)
 		cfqd->active_queue = NULL;
 
-	if (&cfqq->cfqg->rb_node == cfqd->grp_service_tree.active)
-		cfqd->grp_service_tree.active = NULL;
-
 	if (cfqd->active_cic) {
 		put_io_context(cfqd->active_cic->ioc);
 		cfqd->active_cic = NULL;
@@ -1901,10 +1887,10 @@ static bool cfq_should_idle(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 	 * in their service tree.
 	 */
 	if (service_tree->count == 1 && cfq_cfqq_sync(cfqq))
-		return 1;
+		return true;
 	cfq_log_cfqq(cfqd, cfqq, "Not idling. st->count:%d",
 			service_tree->count);
-	return 0;
+	return false;
 }
 
 static void cfq_arm_slice_timer(struct cfq_data *cfqd)
@@ -2040,7 +2026,7 @@ static int cfqq_process_refs(struct cfq_queue *cfqq)
 	int process_refs, io_refs;
 
 	io_refs = cfqq->allocated[READ] + cfqq->allocated[WRITE];
-	process_refs = atomic_read(&cfqq->ref) - io_refs;
+	process_refs = cfqq->ref - io_refs;
 	BUG_ON(process_refs < 0);
 	return process_refs;
 }
@@ -2080,10 +2066,10 @@ static void cfq_setup_merge(struct cfq_queue *cfqq, struct cfq_queue *new_cfqq)
 	 */
 	if (new_process_refs >= process_refs) {
 		cfqq->new_cfqq = new_cfqq;
-		atomic_add(process_refs, &new_cfqq->ref);
+		new_cfqq->ref += process_refs;
 	} else {
 		new_cfqq->new_cfqq = cfqq;
-		atomic_add(new_process_refs, &cfqq->ref);
+		cfqq->ref += new_process_refs;
 	}
 }
 
@@ -2116,12 +2102,7 @@ static void choose_service_tree(struct cfq_data *cfqd, struct cfq_group *cfqg)
 	unsigned count;
 	struct cfq_rb_root *st;
 	unsigned group_slice;
-
-	if (!cfqg) {
-		cfqd->serving_prio = IDLE_WORKLOAD;
-		cfqd->workload_expires = jiffies + 1;
-		return;
-	}
+	enum wl_prio_t original_prio = cfqd->serving_prio;
 
 	/* Choose next priority. RT > BE > IDLE */
 	if (cfq_group_busy_queues_wl(RT_WORKLOAD, cfqd, cfqg))
@@ -2134,6 +2115,9 @@ static void choose_service_tree(struct cfq_data *cfqd, struct cfq_group *cfqg)
 		return;
 	}
 
+	if (original_prio != cfqd->serving_prio)
+		goto new_workload;
+
 	/*
 	 * For RT and BE, we have to choose also the type
 	 * (SYNC, SYNC_NOIDLE, ASYNC), and to compute a workload
@@ -2148,6 +2132,7 @@ static void choose_service_tree(struct cfq_data *cfqd, struct cfq_group *cfqg)
 	if (count && !time_after(jiffies, cfqd->workload_expires))
 		return;
 
+new_workload:
 	/* otherwise select new workload type */
 	cfqd->serving_type =
 		cfq_choose_wl(cfqd, cfqg, cfqd->serving_prio);
@@ -2199,7 +2184,6 @@ static struct cfq_group *cfq_get_next_cfqg(struct cfq_data *cfqd)
 	if (RB_EMPTY_ROOT(&st->rb))
 		return NULL;
 	cfqg = cfq_rb_first_group(st);
-	st->active = &cfqg->rb_node;
 	update_min_vdisktime(st);
 	return cfqg;
 }
@@ -2293,6 +2277,17 @@ static struct cfq_queue *cfq_select_queue(struct cfq_data *cfqd)
 		goto keep_queue;
 	}
 
+	/*
+	 * This is a deep seek queue, but the device is much faster than
+	 * the queue can deliver, don't idle
+	 **/
+	if (CFQQ_SEEKY(cfqq) && cfq_cfqq_idle_window(cfqq) &&
+	    (cfq_cfqq_slice_new(cfqq) ||
+	    (cfqq->slice_end - jiffies > jiffies - cfqq->slice_start))) {
+		cfq_clear_cfqq_deep(cfqq);
+		cfq_clear_cfqq_idle_window(cfqq);
+	}
+
 	if (cfqq->dispatched && cfq_should_idle(cfqd, cfqq)) {
 		cfqq = NULL;
 		goto keep_queue;
@@ -2367,12 +2362,12 @@ static inline bool cfq_slice_used_soon(struct cfq_data *cfqd,
 {
 	/* the queue hasn't finished any request, can't estimate */
 	if (cfq_cfqq_slice_new(cfqq))
-		return 1;
+		return true;
 	if (time_after(jiffies + cfqd->cfq_slice_idle * cfqq->dispatched,
 		cfqq->slice_end))
-		return 1;
+		return true;
 
-	return 0;
+	return false;
 }
 
 static bool cfq_may_dispatch(struct cfq_data *cfqd, struct cfq_queue *cfqq)
@@ -2538,9 +2533,10 @@ static void cfq_put_queue(struct cfq_queue *cfqq)
 	struct cfq_data *cfqd = cfqq->cfqd;
 	struct cfq_group *cfqg, *orig_cfqg;
 
-	BUG_ON(atomic_read(&cfqq->ref) <= 0);
+	BUG_ON(cfqq->ref <= 0);
 
-	if (!atomic_dec_and_test(&cfqq->ref))
+	cfqq->ref--;
+	if (cfqq->ref)
 		return;
 
 	cfq_log_cfqq(cfqd, cfqq, "put_queue");
@@ -2843,7 +2839,7 @@ static void cfq_init_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 	RB_CLEAR_NODE(&cfqq->p_node);
 	INIT_LIST_HEAD(&cfqq->fifo);
 
-	atomic_set(&cfqq->ref, 0);
+	cfqq->ref = 0;
 	cfqq->cfqd = cfqd;
 
 	cfq_mark_cfqq_prio_changed(cfqq);
@@ -2979,11 +2975,11 @@ cfq_get_queue(struct cfq_data *cfqd, bool is_sync, struct io_context *ioc,
 	 * pin the queue now that it's allocated, scheduler exit will prune it
 	 */
 	if (!is_sync && !(*async_cfqq)) {
-		atomic_inc(&cfqq->ref);
+		cfqq->ref++;
 		*async_cfqq = cfqq;
 	}
 
-	atomic_inc(&cfqq->ref);
+	cfqq->ref++;
 	return cfqq;
 }
 
@@ -3265,6 +3261,10 @@ cfq_should_preempt(struct cfq_data *cfqd, struct cfq_queue *new_cfqq,
 	if (cfq_class_rt(new_cfqq) && !cfq_class_rt(cfqq))
 		return true;
 
+	/* An idle queue should not be idle now for some reason */
+	if (RB_EMPTY_ROOT(&cfqq->sort_list) && !cfq_should_idle(cfqd, cfqq))
+		return true;
+
 	if (!cfqd->active_cic || !cfq_cfqq_wait_request(cfqq))
 		return false;
 
@@ -3681,13 +3681,13 @@ new_queue:
 	}
 
 	cfqq->allocated[rw]++;
-	atomic_inc(&cfqq->ref);
-
-	spin_unlock_irqrestore(q->queue_lock, flags);
-
+	cfqq->ref++;
 	rq->elevator_private = cic;
 	rq->elevator_private2 = cfqq;
 	rq->elevator_private3 = cfq_ref_get_cfqg(cfqq->cfqg);
+
+	spin_unlock_irqrestore(q->queue_lock, flags);
+
 	return 0;
 
 queue_fail:
@@ -3862,6 +3862,10 @@ static void *cfq_init_queue(struct request_queue *q)
 	if (!cfqd)
 		return NULL;
 
+	/*
+	 * Don't need take queue_lock in the routine, since we are
+	 * initializing the ioscheduler, and nobody is using cfqd
+	 */
 	cfqd->cic_index = i;
 
 	/* Init root service tree */
@@ -3881,7 +3885,7 @@ static void *cfq_init_queue(struct request_queue *q)
 	 * Take a reference to root group which we never drop. This is just
 	 * to make sure that cfq_put_cfqg() does not try to kfree root group
 	 */
-	atomic_set(&cfqg->ref, 1);
+	cfqg->ref = 1;
 	rcu_read_lock();
 	cfq_blkiocg_add_blkio_group(&blkio_root_cgroup, &cfqg->blkg,
 					(void *)cfqd, 0);
@@ -3901,7 +3905,7 @@ static void *cfq_init_queue(struct request_queue *q)
 	 * will not attempt to free it.
 	 */
 	cfq_init_cfqq(cfqd, &cfqd->oom_cfqq, 1, 0);
-	atomic_inc(&cfqd->oom_cfqq.ref);
+	cfqd->oom_cfqq.ref++;
 	cfq_link_cfqq_cfqg(&cfqd->oom_cfqq, &cfqd->root_group);
 
 	INIT_LIST_HEAD(&cfqd->cic_list);

block/genhd.c

@@ -18,6 +18,7 @@
 #include <linux/buffer_head.h>
 #include <linux/mutex.h>
 #include <linux/idr.h>
+#include <linux/log2.h>
 
 #include "blk.h"
 
@@ -35,6 +36,10 @@ static DEFINE_IDR(ext_devt_idr);
 
 static struct device_type disk_type;
 
+static void disk_add_events(struct gendisk *disk);
+static void disk_del_events(struct gendisk *disk);
+static void disk_release_events(struct gendisk *disk);
+
 /**
  * disk_get_part - get partition
  * @disk: disk to look partition from
@@ -239,7 +244,7 @@ static struct blk_major_name {
 } *major_names[BLKDEV_MAJOR_HASH_SIZE];
 
 /* index in the above - for now: assume no multimajor ranges */
-static inline int major_to_index(int major)
+static inline int major_to_index(unsigned major)
 {
 	return major % BLKDEV_MAJOR_HASH_SIZE;
 }
@@ -502,6 +507,64 @@ static int exact_lock(dev_t devt, void *data)
 	return 0;
 }
 
+void register_disk(struct gendisk *disk)
+{
+	struct device *ddev = disk_to_dev(disk);
+	struct block_device *bdev;
+	struct disk_part_iter piter;
+	struct hd_struct *part;
+	int err;
+
+	ddev->parent = disk->driverfs_dev;
+
+	dev_set_name(ddev, disk->disk_name);
+
+	/* delay uevents, until we scanned partition table */
+	dev_set_uevent_suppress(ddev, 1);
+
+	if (device_add(ddev))
+		return;
+	if (!sysfs_deprecated) {
+		err = sysfs_create_link(block_depr, &ddev->kobj,
+					kobject_name(&ddev->kobj));
+		if (err) {
+			device_del(ddev);
+			return;
+		}
+	}
+	disk->part0.holder_dir = kobject_create_and_add("holders", &ddev->kobj);
+	disk->slave_dir = kobject_create_and_add("slaves", &ddev->kobj);
+
+	/* No minors to use for partitions */
+	if (!disk_partitionable(disk))
+		goto exit;
+
+	/* No such device (e.g., media were just removed) */
+	if (!get_capacity(disk))
+		goto exit;
+
+	bdev = bdget_disk(disk, 0);
+	if (!bdev)
+		goto exit;
+
+	bdev->bd_invalidated = 1;
+	err = blkdev_get(bdev, FMODE_READ, NULL);
+	if (err < 0)
+		goto exit;
+	blkdev_put(bdev, FMODE_READ);
+
+exit:
+	/* announce disk after possible partitions are created */
+	dev_set_uevent_suppress(ddev, 0);
+	kobject_uevent(&ddev->kobj, KOBJ_ADD);
+
+	/* announce possible partitions */
+	disk_part_iter_init(&piter, disk, 0);
+	while ((part = disk_part_iter_next(&piter)))
+		kobject_uevent(&part_to_dev(part)->kobj, KOBJ_ADD);
+	disk_part_iter_exit(&piter);
+}
+
 /**
  * add_disk - add partitioning information to kernel list
  * @disk: per-device partitioning information
@@ -551,18 +614,48 @@ void add_disk(struct gendisk *disk)
 	retval = sysfs_create_link(&disk_to_dev(disk)->kobj, &bdi->dev->kobj,
 				   "bdi");
 	WARN_ON(retval);
-}
 
+	disk_add_events(disk);
+}
 EXPORT_SYMBOL(add_disk);
-EXPORT_SYMBOL(del_gendisk);	/* in partitions/check.c */
 
-void unlink_gendisk(struct gendisk *disk)
+void del_gendisk(struct gendisk *disk)
 {
+	struct disk_part_iter piter;
+	struct hd_struct *part;
+
+	disk_del_events(disk);
+
+	/* invalidate stuff */
+	disk_part_iter_init(&piter, disk,
+			     DISK_PITER_INCL_EMPTY | DISK_PITER_REVERSE);
+	while ((part = disk_part_iter_next(&piter))) {
+		invalidate_partition(disk, part->partno);
+		delete_partition(disk, part->partno);
+	}
+	disk_part_iter_exit(&piter);
+
+	invalidate_partition(disk, 0);
+	blk_free_devt(disk_to_dev(disk)->devt);
+	set_capacity(disk, 0);
+	disk->flags &= ~GENHD_FL_UP;
+
 	sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi");
 	bdi_unregister(&disk->queue->backing_dev_info);
 	blk_unregister_queue(disk);
 	blk_unregister_region(disk_devt(disk), disk->minors);
+
+	part_stat_set_all(&disk->part0, 0);
+	disk->part0.stamp = 0;
+
+	kobject_put(disk->part0.holder_dir);
+	kobject_put(disk->slave_dir);
+	disk->driverfs_dev = NULL;
+	if (!sysfs_deprecated)
+		sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk)));
+	device_del(disk_to_dev(disk));
 }
+EXPORT_SYMBOL(del_gendisk);
 
 /**
  * get_gendisk - get partitioning information for a given device
@@ -735,7 +828,7 @@ static void *show_partition_start(struct seq_file *seqf, loff_t *pos)
 	static void *p;
 
 	p = disk_seqf_start(seqf, pos);
-	if (!IS_ERR(p) && p && !*pos)
+	if (!IS_ERR_OR_NULL(p) && !*pos)
 		seq_puts(seqf, "major minor  #blocks  name\n\n");
 	return p;
 }
@@ -1005,6 +1098,7 @@ static void disk_release(struct device *dev)
 {
 	struct gendisk *disk = dev_to_disk(dev);
 
+	disk_release_events(disk);
 	kfree(disk->random);
 	disk_replace_part_tbl(disk, NULL);
 	free_part_stats(&disk->part0);
@@ -1110,29 +1204,6 @@ static int __init proc_genhd_init(void)
 module_init(proc_genhd_init);
 #endif /* CONFIG_PROC_FS */
 
-static void media_change_notify_thread(struct work_struct *work)
-{
-	struct gendisk *gd = container_of(work, struct gendisk, async_notify);
-	char event[] = "MEDIA_CHANGE=1";
-	char *envp[] = { event, NULL };
-
-	/*
-	 * set enviroment vars to indicate which event this is for
-	 * so that user space will know to go check the media status.
-	 */
-	kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp);
-	put_device(gd->driverfs_dev);
-}
-
-#if 0
-void genhd_media_change_notify(struct gendisk *disk)
-{
-	get_device(disk->driverfs_dev);
-	schedule_work(&disk->async_notify);
-}
-EXPORT_SYMBOL_GPL(genhd_media_change_notify);
-#endif  /*  0  */
-
 dev_t blk_lookup_devt(const char *name, int partno)
 {
 	dev_t devt = MKDEV(0, 0);
@@ -1193,13 +1264,13 @@ struct gendisk *alloc_disk_node(int minors, int node_id)
 		}
 		disk->part_tbl->part[0] = &disk->part0;
 
+		hd_ref_init(&disk->part0);
+
 		disk->minors = minors;
 		rand_initialize_disk(disk);
 		disk_to_dev(disk)->class = &block_class;
 		disk_to_dev(disk)->type = &disk_type;
 		device_initialize(disk_to_dev(disk));
-		INIT_WORK(&disk->async_notify,
-			media_change_notify_thread);
 	}
 	return disk;
 }
@@ -1291,3 +1362,422 @@ int invalidate_partition(struct gendisk *disk, int partno)
 }
 
 EXPORT_SYMBOL(invalidate_partition);
+
+/*
+ * Disk events - monitor disk events like media change and eject request.
+ */
+struct disk_events {
+	struct list_head	node;		/* all disk_event's */
+	struct gendisk		*disk;		/* the associated disk */
+	spinlock_t		lock;
+
+	int			block;		/* event blocking depth */
+	unsigned int		pending;	/* events already sent out */
+	unsigned int		clearing;	/* events being cleared */
+
+	long			poll_msecs;	/* interval, -1 for default */
+	struct delayed_work	dwork;
+};
+
+static const char *disk_events_strs[] = {
+	[ilog2(DISK_EVENT_MEDIA_CHANGE)]	= "media_change",
+	[ilog2(DISK_EVENT_EJECT_REQUEST)]	= "eject_request",
+};
+
+static char *disk_uevents[] = {
+	[ilog2(DISK_EVENT_MEDIA_CHANGE)]	= "DISK_MEDIA_CHANGE=1",
+	[ilog2(DISK_EVENT_EJECT_REQUEST)]	= "DISK_EJECT_REQUEST=1",
+};
+
+/* list of all disk_events */
+static DEFINE_MUTEX(disk_events_mutex);
+static LIST_HEAD(disk_events);
+
+/* disable in-kernel polling by default */
+static unsigned long disk_events_dfl_poll_msecs	= 0;
+
+static unsigned long disk_events_poll_jiffies(struct gendisk *disk)
+{
+	struct disk_events *ev = disk->ev;
+	long intv_msecs = 0;
+
+	/*
+	 * If device-specific poll interval is set, always use it.  If
+	 * the default is being used, poll iff there are events which
+	 * can't be monitored asynchronously.
+	 */
+	if (ev->poll_msecs >= 0)
+		intv_msecs = ev->poll_msecs;
+	else if (disk->events & ~disk->async_events)
+		intv_msecs = disk_events_dfl_poll_msecs;
+
+	return msecs_to_jiffies(intv_msecs);
+}
+
+static void __disk_block_events(struct gendisk *disk, bool sync)
+{
+	struct disk_events *ev = disk->ev;
+	unsigned long flags;
+	bool cancel;
+
+	spin_lock_irqsave(&ev->lock, flags);
+	cancel = !ev->block++;
+	spin_unlock_irqrestore(&ev->lock, flags);
+
+	if (cancel) {
+		if (sync)
+			cancel_delayed_work_sync(&disk->ev->dwork);
+		else
+			cancel_delayed_work(&disk->ev->dwork);
+	}
+}
+
+static void __disk_unblock_events(struct gendisk *disk, bool check_now)
+{
+	struct disk_events *ev = disk->ev;
+	unsigned long intv;
+	unsigned long flags;
+
+	spin_lock_irqsave(&ev->lock, flags);
+
+	if (WARN_ON_ONCE(ev->block <= 0))
+		goto out_unlock;
+
+	if (--ev->block)
+		goto out_unlock;
+
+	/*
+	 * Not exactly a latency critical operation, set poll timer
+	 * slack to 25% and kick event check.
+	 */
+	intv = disk_events_poll_jiffies(disk);
+	set_timer_slack(&ev->dwork.timer, intv / 4);
+	if (check_now)
+		queue_delayed_work(system_nrt_wq, &ev->dwork, 0);
+	else if (intv)
+		queue_delayed_work(system_nrt_wq, &ev->dwork, intv);
+out_unlock:
+	spin_unlock_irqrestore(&ev->lock, flags);
+}
+
+/**
+ * disk_block_events - block and flush disk event checking
+ * @disk: disk to block events for
+ *
+ * On return from this function, it is guaranteed that event checking
+ * isn't in progress and won't happen until unblocked by
+ * disk_unblock_events().  Events blocking is counted and the actual
+ * unblocking happens after the matching number of unblocks are done.
+ *
+ * Note that this intentionally does not block event checking from
+ * disk_clear_events().
+ *
+ * CONTEXT:
+ * Might sleep.
+ */
+void disk_block_events(struct gendisk *disk)
+{
+	if (disk->ev)
+		__disk_block_events(disk, true);
+}
+
+/**
+ * disk_unblock_events - unblock disk event checking
+ * @disk: disk to unblock events for
+ *
+ * Undo disk_block_events().  When the block count reaches zero, it
+ * starts events polling if configured.
+ *
+ * CONTEXT:
+ * Don't care.  Safe to call from irq context.
+ */
+void disk_unblock_events(struct gendisk *disk)
+{
+	if (disk->ev)
+		__disk_unblock_events(disk, true);
+}
+
+/**
+ * disk_check_events - schedule immediate event checking
+ * @disk: disk to check events for
+ *
+ * Schedule immediate event checking on @disk if not blocked.
+ *
+ * CONTEXT:
+ * Don't care.  Safe to call from irq context.
+ */
+void disk_check_events(struct gendisk *disk)
+{
+	if (disk->ev) {
+		__disk_block_events(disk, false);
+		__disk_unblock_events(disk, true);
+	}
+}
+EXPORT_SYMBOL_GPL(disk_check_events);
+
+/**
+ * disk_clear_events - synchronously check, clear and return pending events
+ * @disk: disk to fetch and clear events from
+ * @mask: mask of events to be fetched and clearted
+ *
+ * Disk events are synchronously checked and pending events in @mask
+ * are cleared and returned.  This ignores the block count.
+ *
+ * CONTEXT:
+ * Might sleep.
+ */
+unsigned int disk_clear_events(struct gendisk *disk, unsigned int mask)
+{
+	const struct block_device_operations *bdops = disk->fops;
+	struct disk_events *ev = disk->ev;
+	unsigned int pending;
+
+	if (!ev) {
+		/* for drivers still using the old ->media_changed method */
+		if ((mask & DISK_EVENT_MEDIA_CHANGE) &&
+		    bdops->media_changed && bdops->media_changed(disk))
+			return DISK_EVENT_MEDIA_CHANGE;
+		return 0;
+	}
+
+	/* tell the workfn about the events being cleared */
+	spin_lock_irq(&ev->lock);
+	ev->clearing |= mask;
+	spin_unlock_irq(&ev->lock);
+
+	/* uncondtionally schedule event check and wait for it to finish */
+	__disk_block_events(disk, true);
+	queue_delayed_work(system_nrt_wq, &ev->dwork, 0);
+	flush_delayed_work(&ev->dwork);
+	__disk_unblock_events(disk, false);
+
+	/* then, fetch and clear pending events */
+	spin_lock_irq(&ev->lock);
+	WARN_ON_ONCE(ev->clearing & mask);	/* cleared by workfn */
+	pending = ev->pending & mask;
+	ev->pending &= ~mask;
+	spin_unlock_irq(&ev->lock);
+
+	return pending;
+}
+
+static void disk_events_workfn(struct work_struct *work)
+{
+	struct delayed_work *dwork = to_delayed_work(work);
+	struct disk_events *ev = container_of(dwork, struct disk_events, dwork);
+	struct gendisk *disk = ev->disk;
+	char *envp[ARRAY_SIZE(disk_uevents) + 1] = { };
+	unsigned int clearing = ev->clearing;
+	unsigned int events;
+	unsigned long intv;
+	int nr_events = 0, i;
+
+	/* check events */
+	events = disk->fops->check_events(disk, clearing);
+
+	/* accumulate pending events and schedule next poll if necessary */
+	spin_lock_irq(&ev->lock);
+
+	events &= ~ev->pending;
+	ev->pending |= events;
+	ev->clearing &= ~clearing;
+
+	intv = disk_events_poll_jiffies(disk);
+	if (!ev->block && intv)
+		queue_delayed_work(system_nrt_wq, &ev->dwork, intv);
+
+	spin_unlock_irq(&ev->lock);
+
+	/* tell userland about new events */
+	for (i = 0; i < ARRAY_SIZE(disk_uevents); i++)
+		if (events & (1 << i))
+			envp[nr_events++] = disk_uevents[i];
+
+	if (nr_events)
+		kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp);
+}
+
+/*
+ * A disk events enabled device has the following sysfs nodes under
+ * its /sys/block/X/ directory.
+ *
+ * events		: list of all supported events
+ * events_async		: list of events which can be detected w/o polling
+ * events_poll_msecs	: polling interval, 0: disable, -1: system default
+ */
+static ssize_t __disk_events_show(unsigned int events, char *buf)
+{
+	const char *delim = "";
+	ssize_t pos = 0;
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(disk_events_strs); i++)
+		if (events & (1 << i)) {
+			pos += sprintf(buf + pos, "%s%s",
+				       delim, disk_events_strs[i]);
+			delim = " ";
+		}
+	if (pos)
+		pos += sprintf(buf + pos, "\n");
+	return pos;
+}
+
+static ssize_t disk_events_show(struct device *dev,
+				struct device_attribute *attr, char *buf)
+{
+	struct gendisk *disk = dev_to_disk(dev);
+
+	return __disk_events_show(disk->events, buf);
+}
+
+static ssize_t disk_events_async_show(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	struct gendisk *disk = dev_to_disk(dev);
+
+	return __disk_events_show(disk->async_events, buf);
+}
+
+static ssize_t disk_events_poll_msecs_show(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	struct gendisk *disk = dev_to_disk(dev);
+
+	return sprintf(buf, "%ld\n", disk->ev->poll_msecs);
+}
+
+static ssize_t disk_events_poll_msecs_store(struct device *dev,
+					    struct device_attribute *attr,
+					    const char *buf, size_t count)
+{
+	struct gendisk *disk = dev_to_disk(dev);
+	long intv;
+
+	if (!count || !sscanf(buf, "%ld", &intv))
+		return -EINVAL;
+
+	if (intv < 0 && intv != -1)
+		return -EINVAL;
+
+	__disk_block_events(disk, true);
+	disk->ev->poll_msecs = intv;
+	__disk_unblock_events(disk, true);
+
+	return count;
+}
+
+static const DEVICE_ATTR(events, S_IRUGO, disk_events_show, NULL);
+static const DEVICE_ATTR(events_async, S_IRUGO, disk_events_async_show, NULL);
+static const DEVICE_ATTR(events_poll_msecs, S_IRUGO|S_IWUSR,
+			 disk_events_poll_msecs_show,
+			 disk_events_poll_msecs_store);
+
+static const struct attribute *disk_events_attrs[] = {
+	&dev_attr_events.attr,
+	&dev_attr_events_async.attr,
+	&dev_attr_events_poll_msecs.attr,
+	NULL,
+};
+
+/*
+ * The default polling interval can be specified by the kernel
+ * parameter block.events_dfl_poll_msecs which defaults to 0
+ * (disable).  This can also be modified runtime by writing to
+ * /sys/module/block/events_dfl_poll_msecs.
+ */
+static int disk_events_set_dfl_poll_msecs(const char *val,
+					  const struct kernel_param *kp)
+{
+	struct disk_events *ev;
+	int ret;
+
+	ret = param_set_ulong(val, kp);
+	if (ret < 0)
+		return ret;
+
+	mutex_lock(&disk_events_mutex);
+
+	list_for_each_entry(ev, &disk_events, node)
+		disk_check_events(ev->disk);
+
+	mutex_unlock(&disk_events_mutex);
+
+	return 0;
+}
+
+static const struct kernel_param_ops disk_events_dfl_poll_msecs_param_ops = {
+	.set	= disk_events_set_dfl_poll_msecs,
+	.get	= param_get_ulong,
+};
+
+#undef MODULE_PARAM_PREFIX
+#define MODULE_PARAM_PREFIX	"block."
+
+module_param_cb(events_dfl_poll_msecs, &disk_events_dfl_poll_msecs_param_ops,
+		&disk_events_dfl_poll_msecs, 0644);
+
+/*
+ * disk_{add|del|release}_events - initialize and destroy disk_events.
+ */
+static void disk_add_events(struct gendisk *disk)
+{
+	struct disk_events *ev;
+
+	if (!disk->fops->check_events || !(disk->events | disk->async_events))
+		return;
+
+	ev = kzalloc(sizeof(*ev), GFP_KERNEL);
+	if (!ev) {
+		pr_warn("%s: failed to initialize events\n", disk->disk_name);
+		return;
+	}
+
+	if (sysfs_create_files(&disk_to_dev(disk)->kobj,
+			       disk_events_attrs) < 0) {
+		pr_warn("%s: failed to create sysfs files for events\n",
+			disk->disk_name);
+		kfree(ev);
+		return;
+	}
+
+	disk->ev = ev;
+
+	INIT_LIST_HEAD(&ev->node);
+	ev->disk = disk;
+	spin_lock_init(&ev->lock);
+	ev->block = 1;
+	ev->poll_msecs = -1;
+	INIT_DELAYED_WORK(&ev->dwork, disk_events_workfn);
+
+	mutex_lock(&disk_events_mutex);
+	list_add_tail(&ev->node, &disk_events);
+	mutex_unlock(&disk_events_mutex);
+
+	/*
+	 * Block count is initialized to 1 and the following initial
+	 * unblock kicks it into action.
+	 */
+	__disk_unblock_events(disk, true);
+}
+
+static void disk_del_events(struct gendisk *disk)
+{
+	if (!disk->ev)
+		return;
+
+	__disk_block_events(disk, true);
+
+	mutex_lock(&disk_events_mutex);
+	list_del_init(&disk->ev->node);
+	mutex_unlock(&disk_events_mutex);
+
+	sysfs_remove_files(&disk_to_dev(disk)->kobj, disk_events_attrs);
+}
+
+static void disk_release_events(struct gendisk *disk)
+{
+	/* the block count should be 1 from disk_del_events() */
+	WARN_ON_ONCE(disk->ev && disk->ev->block != 1);
+	kfree(disk->ev);
+}

block/ioctl.c

@@ -294,11 +294,12 @@ int blkdev_ioctl(struct block_device *bdev, fmode_t mode, unsigned cmd,
 			return -EINVAL;
 		if (get_user(n, (int __user *) arg))
 			return -EFAULT;
-		if (!(mode & FMODE_EXCL) && bd_claim(bdev, &bdev) < 0)
+		if (!(mode & FMODE_EXCL) &&
+		    blkdev_get(bdev, mode | FMODE_EXCL, &bdev) < 0)
 			return -EBUSY;
 		ret = set_blocksize(bdev, n);
 		if (!(mode & FMODE_EXCL))
-			bd_release(bdev);
+			blkdev_put(bdev, mode | FMODE_EXCL);
 		return ret;
 	case BLKPG:
 		ret = blkpg_ioctl(bdev, (struct blkpg_ioctl_arg __user *) arg);

drivers/block/drbd/drbd_int.h

@@ -911,8 +911,6 @@ struct drbd_md {
 struct drbd_backing_dev {
 	struct block_device *backing_bdev;
 	struct block_device *md_bdev;
-	struct file *lo_file;
-	struct file *md_file;
 	struct drbd_md md;
 	struct disk_conf dc; /* The user provided config... */
 	sector_t known_size; /* last known size of that backing device */

drivers/block/drbd/drbd_main.c

@@ -3372,11 +3372,8 @@ void drbd_free_bc(struct drbd_backing_dev *ldev)
 	if (ldev == NULL)
 		return;
 
-	bd_release(ldev->backing_bdev);
-	bd_release(ldev->md_bdev);
-
-	fput(ldev->lo_file);
-	fput(ldev->md_file);
+	blkdev_put(ldev->backing_bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
+	blkdev_put(ldev->md_bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
 
 	kfree(ldev);
 }

drivers/block/drbd/drbd_nl.c

@@ -855,7 +855,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	sector_t max_possible_sectors;
 	sector_t min_md_device_sectors;
 	struct drbd_backing_dev *nbc = NULL; /* new_backing_conf */
-	struct inode *inode, *inode2;
+	struct block_device *bdev;
 	struct lru_cache *resync_lru = NULL;
 	union drbd_state ns, os;
 	unsigned int max_seg_s;
@@ -907,46 +907,40 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		}
 	}
 
-	nbc->lo_file = filp_open(nbc->dc.backing_dev, O_RDWR, 0);
-	if (IS_ERR(nbc->lo_file)) {
+	bdev = blkdev_get_by_path(nbc->dc.backing_dev,
+				  FMODE_READ | FMODE_WRITE | FMODE_EXCL, mdev);
+	if (IS_ERR(bdev)) {
 		dev_err(DEV, "open(\"%s\") failed with %ld\n", nbc->dc.backing_dev,
-		    PTR_ERR(nbc->lo_file));
-		nbc->lo_file = NULL;
+			PTR_ERR(bdev));
 		retcode = ERR_OPEN_DISK;
 		goto fail;
 	}
+	nbc->backing_bdev = bdev;
 
-	inode = nbc->lo_file->f_dentry->d_inode;
-
-	if (!S_ISBLK(inode->i_mode)) {
-		retcode = ERR_DISK_NOT_BDEV;
-		goto fail;
-	}
-
-	nbc->md_file = filp_open(nbc->dc.meta_dev, O_RDWR, 0);
-	if (IS_ERR(nbc->md_file)) {
+	/*
+	 * meta_dev_idx >= 0: external fixed size, possibly multiple
+	 * drbd sharing one meta device.  TODO in that case, paranoia
+	 * check that [md_bdev, meta_dev_idx] is not yet used by some
+	 * other drbd minor!  (if you use drbd.conf + drbdadm, that
+	 * should check it for you already; but if you don't, or
+	 * someone fooled it, we need to double check here)
+	 */
+	bdev = blkdev_get_by_path(nbc->dc.meta_dev,
+				  FMODE_READ | FMODE_WRITE | FMODE_EXCL,
+				  (nbc->dc.meta_dev_idx < 0) ?
+				  (void *)mdev : (void *)drbd_m_holder);
+	if (IS_ERR(bdev)) {
 		dev_err(DEV, "open(\"%s\") failed with %ld\n", nbc->dc.meta_dev,
-		    PTR_ERR(nbc->md_file));
-		nbc->md_file = NULL;
+			PTR_ERR(bdev));
 		retcode = ERR_OPEN_MD_DISK;
 		goto fail;
 	}
+	nbc->md_bdev = bdev;
 
-	inode2 = nbc->md_file->f_dentry->d_inode;
-
-	if (!S_ISBLK(inode2->i_mode)) {
-		retcode = ERR_MD_NOT_BDEV;
-		goto fail;
-	}
-
-	nbc->backing_bdev = inode->i_bdev;
-	if (bd_claim(nbc->backing_bdev, mdev)) {
-		printk(KERN_ERR "drbd: bd_claim(%p,%p); failed [%p;%p;%u]\n",
-		       nbc->backing_bdev, mdev,
-		       nbc->backing_bdev->bd_holder,
-		       nbc->backing_bdev->bd_contains->bd_holder,
-		       nbc->backing_bdev->bd_holders);
-		retcode = ERR_BDCLAIM_DISK;
+	if ((nbc->backing_bdev == nbc->md_bdev) !=
+	    (nbc->dc.meta_dev_idx == DRBD_MD_INDEX_INTERNAL ||
+	     nbc->dc.meta_dev_idx == DRBD_MD_INDEX_FLEX_INT)) {
+		retcode = ERR_MD_IDX_INVALID;
 		goto fail;
 	}
 
@@ -955,28 +949,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 			offsetof(struct bm_extent, lce));
 	if (!resync_lru) {
 		retcode = ERR_NOMEM;
-		goto release_bdev_fail;
-	}
-
-	/* meta_dev_idx >= 0: external fixed size,
-	 * possibly multiple drbd sharing one meta device.
-	 * TODO in that case, paranoia check that [md_bdev, meta_dev_idx] is
-	 * not yet used by some other drbd minor!
-	 * (if you use drbd.conf + drbdadm,
-	 * that should check it for you already; but if you don't, or someone
-	 * fooled it, we need to double check here) */
-	nbc->md_bdev = inode2->i_bdev;
-	if (bd_claim(nbc->md_bdev, (nbc->dc.meta_dev_idx < 0) ? (void *)mdev
-				: (void *) drbd_m_holder)) {
-		retcode = ERR_BDCLAIM_MD_DISK;
-		goto release_bdev_fail;
-	}
-
-	if ((nbc->backing_bdev == nbc->md_bdev) !=
-	    (nbc->dc.meta_dev_idx == DRBD_MD_INDEX_INTERNAL ||
-	     nbc->dc.meta_dev_idx == DRBD_MD_INDEX_FLEX_INT)) {
-		retcode = ERR_MD_IDX_INVALID;
-		goto release_bdev2_fail;
+		goto fail;
 	}
 
 	/* RT - for drbd_get_max_capacity() DRBD_MD_INDEX_FLEX_INT */
@@ -987,7 +960,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 			(unsigned long long) drbd_get_max_capacity(nbc),
 			(unsigned long long) nbc->dc.disk_size);
 		retcode = ERR_DISK_TO_SMALL;
-		goto release_bdev2_fail;
+		goto fail;
 	}
 
 	if (nbc->dc.meta_dev_idx < 0) {
@@ -1004,7 +977,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		dev_warn(DEV, "refusing attach: md-device too small, "
 		     "at least %llu sectors needed for this meta-disk type\n",
 		     (unsigned long long) min_md_device_sectors);
-		goto release_bdev2_fail;
+		goto fail;
 	}
 
 	/* Make sure the new disk is big enough
@@ -1012,7 +985,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	if (drbd_get_max_capacity(nbc) <
 	    drbd_get_capacity(mdev->this_bdev)) {
 		retcode = ERR_DISK_TO_SMALL;
-		goto release_bdev2_fail;
+		goto fail;
 	}
 
 	nbc->known_size = drbd_get_capacity(nbc->backing_bdev);
@@ -1035,7 +1008,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	retcode = _drbd_request_state(mdev, NS(disk, D_ATTACHING), CS_VERBOSE);
 	drbd_resume_io(mdev);
 	if (retcode < SS_SUCCESS)
-		goto release_bdev2_fail;
+		goto fail;
 
 	if (!get_ldev_if_state(mdev, D_ATTACHING))
 		goto force_diskless;
@@ -1269,18 +1242,14 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
  force_diskless:
 	drbd_force_state(mdev, NS(disk, D_FAILED));
 	drbd_md_sync(mdev);
- release_bdev2_fail:
-	if (nbc)
-		bd_release(nbc->md_bdev);
- release_bdev_fail:
-	if (nbc)
-		bd_release(nbc->backing_bdev);
  fail:
 	if (nbc) {
-		if (nbc->lo_file)
-			fput(nbc->lo_file);
-		if (nbc->md_file)
-			fput(nbc->md_file);
+		if (nbc->backing_bdev)
+			blkdev_put(nbc->backing_bdev,
+				   FMODE_READ | FMODE_WRITE | FMODE_EXCL);
+		if (nbc->md_bdev)
+			blkdev_put(nbc->md_bdev,
+				   FMODE_READ | FMODE_WRITE | FMODE_EXCL);
 		kfree(nbc);
 	}
 	lc_destroy(resync_lru);

drivers/block/loop.c

@@ -395,11 +395,7 @@ lo_splice_actor(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
 	struct loop_device *lo = p->lo;
 	struct page *page = buf->page;
 	sector_t IV;
-	int size, ret;
-
-	ret = buf->ops->confirm(pipe, buf);
-	if (unlikely(ret))
-		return ret;
+	int size;
 
 	IV = ((sector_t) page->index << (PAGE_CACHE_SHIFT - 9)) +
 							(buf->offset >> 9);

drivers/block/pktcdvd.c

@@ -2296,15 +2296,12 @@ static int pkt_open_dev(struct pktcdvd_device *pd, fmode_t write)
 	 * so bdget() can't fail.
 	 */
 	bdget(pd->bdev->bd_dev);
-	if ((ret = blkdev_get(pd->bdev, FMODE_READ)))
+	if ((ret = blkdev_get(pd->bdev, FMODE_READ | FMODE_EXCL, pd)))
 		goto out;
 
-	if ((ret = bd_claim(pd->bdev, pd)))
-		goto out_putdev;
-
 	if ((ret = pkt_get_last_written(pd, &lba))) {
 		printk(DRIVER_NAME": pkt_get_last_written failed\n");
-		goto out_unclaim;
+		goto out_putdev;
 	}
 
 	set_capacity(pd->disk, lba << 2);
@@ -2314,7 +2311,7 @@ static int pkt_open_dev(struct pktcdvd_device *pd, fmode_t write)
 	q = bdev_get_queue(pd->bdev);
 	if (write) {
 		if ((ret = pkt_open_write(pd)))
-			goto out_unclaim;
+			goto out_putdev;
 		/*
 		 * Some CDRW drives can not handle writes larger than one packet,
 		 * even if the size is a multiple of the packet size.
@@ -2329,23 +2326,21 @@ static int pkt_open_dev(struct pktcdvd_device *pd, fmode_t write)
 	}
 
 	if ((ret = pkt_set_segment_merging(pd, q)))
-		goto out_unclaim;
+		goto out_putdev;
 
 	if (write) {
 		if (!pkt_grow_pktlist(pd, CONFIG_CDROM_PKTCDVD_BUFFERS)) {
 			printk(DRIVER_NAME": not enough memory for buffers\n");
 			ret = -ENOMEM;
-			goto out_unclaim;
+			goto out_putdev;
 		}
 		printk(DRIVER_NAME": %lukB available on disc\n", lba << 1);
 	}
 
 	return 0;
 
-out_unclaim:
-	bd_release(pd->bdev);
 out_putdev:
-	blkdev_put(pd->bdev, FMODE_READ);
+	blkdev_put(pd->bdev, FMODE_READ | FMODE_EXCL);
 out:
 	return ret;
 }
@@ -2362,8 +2357,7 @@ static void pkt_release_dev(struct pktcdvd_device *pd, int flush)
 	pkt_lock_door(pd, 0);
 
 	pkt_set_speed(pd, MAX_SPEED, MAX_SPEED);
-	bd_release(pd->bdev);
-	blkdev_put(pd->bdev, FMODE_READ);
+	blkdev_put(pd->bdev, FMODE_READ | FMODE_EXCL);
 
 	pkt_shrink_pktlist(pd);
 }
@@ -2733,7 +2727,7 @@ static int pkt_new_dev(struct pktcdvd_device *pd, dev_t dev)
 	bdev = bdget(dev);
 	if (!bdev)
 		return -ENOMEM;
-	ret = blkdev_get(bdev, FMODE_READ | FMODE_NDELAY);
+	ret = blkdev_get(bdev, FMODE_READ | FMODE_NDELAY, NULL);
 	if (ret)
 		return ret;
 

drivers/cdrom/cdrom.c

@@ -1348,7 +1348,10 @@ static int cdrom_select_disc(struct cdrom_device_info *cdi, int slot)
 	if (!CDROM_CAN(CDC_SELECT_DISC))
 		return -EDRIVE_CANT_DO_THIS;
 
-	(void) cdi->ops->media_changed(cdi, slot);
+	if (cdi->ops->check_events)
+		cdi->ops->check_events(cdi, 0, slot);
+	else
+		cdi->ops->media_changed(cdi, slot);
 
 	if (slot == CDSL_NONE) {
 		/* set media changed bits, on both queues */
@@ -1392,6 +1395,42 @@ static int cdrom_select_disc(struct cdrom_device_info *cdi, int slot)
 	return slot;
 }
 
+/*
+ * As cdrom implements an extra ioctl consumer for media changed
+ * event, it needs to buffer ->check_events() output, such that event
+ * is not lost for both the usual VFS and ioctl paths.
+ * cdi->{vfs|ioctl}_events are used to buffer pending events for each
+ * path.
+ *
+ * XXX: Locking is non-existent.  cdi->ops->check_events() can be
+ * called in parallel and buffering fields are accessed without any
+ * exclusion.  The original media_changed code had the same problem.
+ * It might be better to simply deprecate CDROM_MEDIA_CHANGED ioctl
+ * and remove this cruft altogether.  It doesn't have much usefulness
+ * at this point.
+ */
+static void cdrom_update_events(struct cdrom_device_info *cdi,
+				unsigned int clearing)
+{
+	unsigned int events;
+
+	events = cdi->ops->check_events(cdi, clearing, CDSL_CURRENT);
+	cdi->vfs_events |= events;
+	cdi->ioctl_events |= events;
+}
+
+unsigned int cdrom_check_events(struct cdrom_device_info *cdi,
+				unsigned int clearing)
+{
+	unsigned int events;
+
+	cdrom_update_events(cdi, clearing);
+	events = cdi->vfs_events;
+	cdi->vfs_events = 0;
+	return events;
+}
+EXPORT_SYMBOL(cdrom_check_events);
+
 /* We want to make media_changed accessible to the user through an
  * ioctl. The main problem now is that we must double-buffer the
  * low-level implementation, to assure that the VFS and the user both
@@ -1403,15 +1442,26 @@ int media_changed(struct cdrom_device_info *cdi, int queue)
 {
 	unsigned int mask = (1 << (queue & 1));
 	int ret = !!(cdi->mc_flags & mask);
+	bool changed;
 
 	if (!CDROM_CAN(CDC_MEDIA_CHANGED))
-	    return ret;
+		return ret;
+
 	/* changed since last call? */
-	if (cdi->ops->media_changed(cdi, CDSL_CURRENT)) {
+	if (cdi->ops->check_events) {
+		BUG_ON(!queue);	/* shouldn't be called from VFS path */
+		cdrom_update_events(cdi, DISK_EVENT_MEDIA_CHANGE);
+		changed = cdi->ioctl_events & DISK_EVENT_MEDIA_CHANGE;
+		cdi->ioctl_events = 0;
+	} else
+		changed = cdi->ops->media_changed(cdi, CDSL_CURRENT);
+
+	if (changed) {
 		cdi->mc_flags = 0x3;    /* set bit on both queues */
 		ret |= 1;
 		cdi->media_written = 0;
 	}
+
 	cdi->mc_flags &= ~mask;         /* clear bit */
 	return ret;
 }

drivers/char/raw.c

@@ -65,15 +65,12 @@ static int raw_open(struct inode *inode, struct file *filp)
 	if (!bdev)
 		goto out;
 	igrab(bdev->bd_inode);
-	err = blkdev_get(bdev, filp->f_mode);
+	err = blkdev_get(bdev, filp->f_mode | FMODE_EXCL, raw_open);
 	if (err)
 		goto out;
-	err = bd_claim(bdev, raw_open);
-	if (err)
-		goto out1;
 	err = set_blocksize(bdev, bdev_logical_block_size(bdev));
 	if (err)
-		goto out2;
+		goto out1;
 	filp->f_flags |= O_DIRECT;
 	filp->f_mapping = bdev->bd_inode->i_mapping;
 	if (++raw_devices[minor].inuse == 1)
@@ -83,10 +80,8 @@ static int raw_open(struct inode *inode, struct file *filp)
 	mutex_unlock(&raw_mutex);
 	return 0;
 
-out2:
-	bd_release(bdev);
 out1:
-	blkdev_put(bdev, filp->f_mode);
+	blkdev_put(bdev, filp->f_mode | FMODE_EXCL);
 out:
 	mutex_unlock(&raw_mutex);
 	return err;
@@ -110,8 +105,7 @@ static int raw_release(struct inode *inode, struct file *filp)
 	}
 	mutex_unlock(&raw_mutex);
 
-	bd_release(bdev);
-	blkdev_put(bdev, filp->f_mode);
+	blkdev_put(bdev, filp->f_mode | FMODE_EXCL);
 	return 0;
 }
 

drivers/md/dm-table.c

@@ -325,15 +325,18 @@ static int open_dev(struct dm_dev_internal *d, dev_t dev,
 
 	BUG_ON(d->dm_dev.bdev);
 
-	bdev = open_by_devnum(dev, d->dm_dev.mode);
+	bdev = blkdev_get_by_dev(dev, d->dm_dev.mode | FMODE_EXCL, _claim_ptr);
 	if (IS_ERR(bdev))
 		return PTR_ERR(bdev);
-	r = bd_claim_by_disk(bdev, _claim_ptr, dm_disk(md));
-	if (r)
-		blkdev_put(bdev, d->dm_dev.mode);
-	else
-		d->dm_dev.bdev = bdev;
-	return r;
+
+	r = bd_link_disk_holder(bdev, dm_disk(md));
+	if (r) {
+		blkdev_put(bdev, d->dm_dev.mode | FMODE_EXCL);
+		return r;
+	}
+
+	d->dm_dev.bdev = bdev;
+	return 0;
 }
 
 /*
@@ -344,8 +347,7 @@ static void close_dev(struct dm_dev_internal *d, struct mapped_device *md)
 	if (!d->dm_dev.bdev)
 		return;
 
-	bd_release_from_disk(d->dm_dev.bdev, dm_disk(md));
-	blkdev_put(d->dm_dev.bdev, d->dm_dev.mode);
+	blkdev_put(d->dm_dev.bdev, d->dm_dev.mode | FMODE_EXCL);
 	d->dm_dev.bdev = NULL;
 }
 

drivers/md/dm.c

@@ -630,7 +630,7 @@ static void dec_pending(struct dm_io *io, int error)
 			queue_io(md, bio);
 		} else {
 			/* done with normal IO or empty flush */
-			trace_block_bio_complete(md->queue, bio);
+			trace_block_bio_complete(md->queue, bio, io_error);
 			bio_endio(bio, io_error);
 		}
 	}
@@ -990,8 +990,8 @@ static void __map_bio(struct dm_target *ti, struct bio *clone,
 	if (r == DM_MAPIO_REMAPPED) {
 		/* the bio has been remapped so dispatch it */
 
-		trace_block_remap(bdev_get_queue(clone->bi_bdev), clone,
-				    tio->io->bio->bi_bdev->bd_dev, sector);
+		trace_block_bio_remap(bdev_get_queue(clone->bi_bdev), clone,
+				      tio->io->bio->bi_bdev->bd_dev, sector);
 
 		generic_make_request(clone);
 	} else if (r < 0 || r == DM_MAPIO_REQUEUE) {

drivers/md/md.c

@@ -1879,7 +1879,7 @@ static int bind_rdev_to_array(mdk_rdev_t * rdev, mddev_t * mddev)
 	rdev->sysfs_state = sysfs_get_dirent_safe(rdev->kobj.sd, "state");
 
 	list_add_rcu(&rdev->same_set, &mddev->disks);
-	bd_claim_by_disk(rdev->bdev, rdev->bdev->bd_holder, mddev->gendisk);
+	bd_link_disk_holder(rdev->bdev, mddev->gendisk);
 
 	/* May as well allow recovery to be retried once */
 	mddev->recovery_disabled = 0;
@@ -1906,7 +1906,6 @@ static void unbind_rdev_from_array(mdk_rdev_t * rdev)
 		MD_BUG();
 		return;
 	}
-	bd_release_from_disk(rdev->bdev, rdev->mddev->gendisk);
 	list_del_rcu(&rdev->same_set);
 	printk(KERN_INFO "md: unbind<%s>\n", bdevname(rdev->bdev,b));
 	rdev->mddev = NULL;
@@ -1934,19 +1933,13 @@ static int lock_rdev(mdk_rdev_t *rdev, dev_t dev, int shared)
 	struct block_device *bdev;
 	char b[BDEVNAME_SIZE];
 
-	bdev = open_by_devnum(dev, FMODE_READ|FMODE_WRITE);
+	bdev = blkdev_get_by_dev(dev, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
+				 shared ? (mdk_rdev_t *)lock_rdev : rdev);
 	if (IS_ERR(bdev)) {
 		printk(KERN_ERR "md: could not open %s.\n",
 			__bdevname(dev, b));
 		return PTR_ERR(bdev);
 	}
-	err = bd_claim(bdev, shared ? (mdk_rdev_t *)lock_rdev : rdev);
-	if (err) {
-		printk(KERN_ERR "md: could not bd_claim %s.\n",
-			bdevname(bdev, b));
-		blkdev_put(bdev, FMODE_READ|FMODE_WRITE);
-		return err;
-	}
 	if (!shared)
 		set_bit(AllReserved, &rdev->flags);
 	rdev->bdev = bdev;
@@ -1959,8 +1952,7 @@ static void unlock_rdev(mdk_rdev_t *rdev)
 	rdev->bdev = NULL;
 	if (!bdev)
 		MD_BUG();
-	bd_release(bdev);
-	blkdev_put(bdev, FMODE_READ|FMODE_WRITE);
+	blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
 }
 
 void md_autodetect_dev(dev_t dev);

drivers/mtd/devices/block2mtd.c

@@ -224,7 +224,7 @@ static void block2mtd_free_device(struct block2mtd_dev *dev)
 	if (dev->blkdev) {
 		invalidate_mapping_pages(dev->blkdev->bd_inode->i_mapping,
 					0, -1);
-		close_bdev_exclusive(dev->blkdev, FMODE_READ|FMODE_WRITE);
+		blkdev_put(dev->blkdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
 	}
 
 	kfree(dev);
@@ -234,6 +234,7 @@ static void block2mtd_free_device(struct block2mtd_dev *dev)
 /* FIXME: ensure that mtd->size % erase_size == 0 */
 static struct block2mtd_dev *add_device(char *devname, int erase_size)
 {
+	const fmode_t mode = FMODE_READ | FMODE_WRITE | FMODE_EXCL;
 	struct block_device *bdev;
 	struct block2mtd_dev *dev;
 	char *name;
@@ -246,7 +247,7 @@ static struct block2mtd_dev *add_device(char *devname, int erase_size)
 		return NULL;
 
 	/* Get a handle on the device */
-	bdev = open_bdev_exclusive(devname, FMODE_READ|FMODE_WRITE, NULL);
+	bdev = blkdev_get_by_path(devname, mode, dev);
 #ifndef MODULE
 	if (IS_ERR(bdev)) {
 
@@ -254,9 +255,8 @@ static struct block2mtd_dev *add_device(char *devname, int erase_size)
 		   to resolve the device name by other means. */
 
 		dev_t devt = name_to_dev_t(devname);
-		if (devt) {
-			bdev = open_by_devnum(devt, FMODE_WRITE | FMODE_READ);
-		}
+		if (devt)
+			bdev = blkdev_get_by_dev(devt, mode, dev);
 	}
 #endif
 

drivers/s390/block/dasd_genhd.c

@@ -103,7 +103,7 @@ int dasd_scan_partitions(struct dasd_block *block)
 	struct block_device *bdev;
 
 	bdev = bdget_disk(block->gdp, 0);
-	if (!bdev || blkdev_get(bdev, FMODE_READ) < 0)
+	if (!bdev || blkdev_get(bdev, FMODE_READ, NULL) < 0)
 		return -ENODEV;
 	/*
 	 * See fs/partition/check.c:register_disk,rescan_partitions