block: remove the nr_sects field in struct hd_struct

Now that the hd_struct always has a block device attached to it, there is
no need for having two size field that just get out of sync.

Additionally the field in hd_struct did not use proper serialization,
possibly allowing for torn writes.  By only using the block_device field
this problem also gets fixed.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Acked-by: Coly Li <colyli@suse.de>			[bcache]
Acked-by: Chao Yu <yuchao0@huawei.com>			[f2fs]
Signed-off-by: Jens Axboe <axboe@kernel.dk>

5 files changed, 49 insertions, 86 deletions

diff --git a/block/bio.c b/block/bio.c
index fa01bef35bb1..669bb47a3198 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -613,8 +613,8 @@ void guard_bio_eod(struct bio *bio)
 	rcu_read_lock();
 	part = __disk_get_part(bio->bi_disk, bio->bi_partno);
 	if (part)
-		maxsector = part_nr_sects_read(part);
-	else
+		maxsector = bdev_nr_sectors(part->bdev);
+	else	
 		maxsector = get_capacity(bio->bi_disk);
 	rcu_read_unlock();
 
diff --git a/block/blk-core.c b/block/blk-core.c
index 2db8bda43b6e..988f45094a38 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -755,7 +755,7 @@ static inline int blk_partition_remap(struct bio *bio)
 		goto out;
 
 	if (bio_sectors(bio)) {
-		if (bio_check_eod(bio, part_nr_sects_read(p)))
+		if (bio_check_eod(bio, bdev_nr_sectors(p->bdev)))
 			goto out;
 		bio->bi_iter.bi_sector += p->start_sect;
 		trace_block_bio_remap(bio->bi_disk->queue, bio, part_devt(p),
diff --git a/block/blk.h b/block/blk.h
index c4839abcfa27..09cee7024fb4 100644
--- a/block/blk.h
+++ b/block/blk.h
@@ -387,59 +387,6 @@ static inline void hd_free_part(struct hd_struct *part)
 	percpu_ref_exit(&part->ref);
 }
 
-/*
- * Any access of part->nr_sects which is not protected by partition
- * bd_mutex or gendisk bdev bd_mutex, should be done using this
- * accessor function.
- *
- * Code written along the lines of i_size_read() and i_size_write().
- * CONFIG_PREEMPTION case optimizes the case of UP kernel with preemption
- * on.
- */
-static inline sector_t part_nr_sects_read(struct hd_struct *part)
-{
-#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
-	sector_t nr_sects;
-	unsigned seq;
-	do {
-		seq = read_seqcount_begin(&part->nr_sects_seq);
-		nr_sects = part->nr_sects;
-	} while (read_seqcount_retry(&part->nr_sects_seq, seq));
-	return nr_sects;
-#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
-	sector_t nr_sects;
-
-	preempt_disable();
-	nr_sects = part->nr_sects;
-	preempt_enable();
-	return nr_sects;
-#else
-	return part->nr_sects;
-#endif
-}
-
-/*
- * Should be called with mutex lock held (typically bd_mutex) of partition
- * to provide mutual exlusion among writers otherwise seqcount might be
- * left in wrong state leaving the readers spinning infinitely.
- */
-static inline void part_nr_sects_write(struct hd_struct *part, sector_t size)
-{
-#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
-	preempt_disable();
-	write_seqcount_begin(&part->nr_sects_seq);
-	part->nr_sects = size;
-	write_seqcount_end(&part->nr_sects_seq);
-	preempt_enable();
-#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPTION)
-	preempt_disable();
-	part->nr_sects = size;
-	preempt_enable();
-#else
-	part->nr_sects = size;
-#endif
-}
-
 int bio_add_hw_page(struct request_queue *q, struct bio *bio,
 		struct page *page, unsigned int len, unsigned int offset,
 		unsigned int max_sectors, bool *same_page);
diff --git a/block/genhd.c b/block/genhd.c
index bf8fa82f135f..c65f485b9db5 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -40,6 +40,16 @@ 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 set_capacity(struct gendisk *disk, sector_t sectors)
+{
+	struct block_device *bdev = disk->part0.bdev;
+
+	spin_lock(&bdev->bd_size_lock);
+	i_size_write(bdev->bd_inode, (loff_t)sectors << SECTOR_SHIFT);
+	spin_unlock(&bdev->bd_size_lock);
+}
+EXPORT_SYMBOL(set_capacity);
+
 /*
  * Set disk capacity and notify if the size is not currently zero and will not
  * be set to zero.  Returns true if a uevent was sent, otherwise false.
@@ -47,18 +57,30 @@ static void disk_release_events(struct gendisk *disk);
 bool set_capacity_and_notify(struct gendisk *disk, sector_t size)
 {
 	sector_t capacity = get_capacity(disk);
+	char *envp[] = { "RESIZE=1", NULL };
 
 	set_capacity(disk, size);
-	revalidate_disk_size(disk, true);
 
-	if (capacity != size && capacity != 0 && size != 0) {
-		char *envp[] = { "RESIZE=1", NULL };
+	/*
+	 * Only print a message and send a uevent if the gendisk is user visible
+	 * and alive.  This avoids spamming the log and udev when setting the
+	 * initial capacity during probing.
+	 */
+	if (size == capacity ||
+	    (disk->flags & (GENHD_FL_UP | GENHD_FL_HIDDEN)) != GENHD_FL_UP)
+		return false;
 
-		kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp);
-		return true;
-	}
+	pr_info("%s: detected capacity change from %lld to %lld\n",
+		disk->disk_name, size, capacity);
 
-	return false;
+	/*
+	 * Historically we did not send a uevent for changes to/from an empty
+	 * device.
+	 */
+	if (!capacity || !size)
+		return false;
+	kobject_uevent_env(&disk_to_dev(disk)->kobj, KOBJ_CHANGE, envp);
+	return true;
 }
 EXPORT_SYMBOL_GPL(set_capacity_and_notify);
 
@@ -247,7 +269,7 @@ struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter)
 		part = rcu_dereference(ptbl->part[piter->idx]);
 		if (!part)
 			continue;
-		if (!part_nr_sects_read(part) &&
+		if (!bdev_nr_sectors(part->bdev) &&
 		    !(piter->flags & DISK_PITER_INCL_EMPTY) &&
 		    !(piter->flags & DISK_PITER_INCL_EMPTY_PART0 &&
 		      piter->idx == 0))
@@ -284,7 +306,7 @@ EXPORT_SYMBOL_GPL(disk_part_iter_exit);
 static inline int sector_in_part(struct hd_struct *part, sector_t sector)
 {
 	return part->start_sect <= sector &&
-		sector < part->start_sect + part_nr_sects_read(part);
+		sector < part->start_sect + bdev_nr_sectors(part->bdev);
 }
 
 /**
@@ -986,8 +1008,8 @@ void __init printk_all_partitions(void)
 
 			printk("%s%s %10llu %s %s", is_part0 ? "" : "  ",
 			       bdevt_str(part_devt(part), devt_buf),
-			       (unsigned long long)part_nr_sects_read(part) >> 1
-			       , disk_name(disk, part->partno, name_buf),
+			       bdev_nr_sectors(part->bdev) >> 1,
+			       disk_name(disk, part->partno, name_buf),
 			       part->info ? part->info->uuid : "");
 			if (is_part0) {
 				if (dev->parent && dev->parent->driver)
@@ -1079,7 +1101,7 @@ static int show_partition(struct seq_file *seqf, void *v)
 	while ((part = disk_part_iter_next(&piter)))
 		seq_printf(seqf, "%4d  %7d %10llu %s\n",
 			   MAJOR(part_devt(part)), MINOR(part_devt(part)),
-			   (unsigned long long)part_nr_sects_read(part) >> 1,
+			   bdev_nr_sectors(part->bdev) >> 1,
 			   disk_name(sgp, part->partno, buf));
 	disk_part_iter_exit(&piter);
 
@@ -1161,8 +1183,7 @@ ssize_t part_size_show(struct device *dev,
 {
 	struct hd_struct *p = dev_to_part(dev);
 
-	return sprintf(buf, "%llu\n",
-		(unsigned long long)part_nr_sects_read(p));
+	return sprintf(buf, "%llu\n", bdev_nr_sectors(p->bdev));
 }
 
 ssize_t part_stat_show(struct device *dev,
@@ -1618,16 +1639,6 @@ struct gendisk *__alloc_disk_node(int minors, int node_id)
 	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
-	 * seqcounter to read/update the part0->nr_sects. Still init
-	 * the counter as we can read the sectors in IO submission
-	 * patch using seqence counters.
-	 *
-	 * TODO: Ideally set_capacity() and get_capacity() should be
-	 * converted to make use of bd_mutex and sequence counters.
-	 */
-	hd_sects_seq_init(&disk->part0);
 	if (hd_ref_init(&disk->part0))
 		goto out_free_bdstats;
 
diff --git a/block/partitions/core.c b/block/partitions/core.c
index 696bd9ff63c6..bcfa8215bd5e 100644
--- a/block/partitions/core.c
+++ b/block/partitions/core.c
@@ -85,6 +85,13 @@ static int (*check_part[])(struct parsed_partitions *) = {
 	NULL
 };
 
+static void bdev_set_nr_sectors(struct block_device *bdev, sector_t sectors)
+{
+	spin_lock(&bdev->bd_size_lock);
+	i_size_write(bdev->bd_inode, (loff_t)sectors << SECTOR_SHIFT);
+	spin_unlock(&bdev->bd_size_lock);
+}
+
 static struct parsed_partitions *allocate_partitions(struct gendisk *hd)
 {
 	struct parsed_partitions *state;
@@ -295,7 +302,7 @@ static void hd_struct_free_work(struct work_struct *work)
 	put_device(disk_to_dev(disk));
 
 	part->start_sect = 0;
-	part->nr_sects = 0;
+	bdev_set_nr_sectors(part->bdev, 0);
 	part_stat_set_all(part, 0);
 	put_device(part_to_dev(part));
 }
@@ -412,11 +419,10 @@ static struct hd_struct *add_partition(struct gendisk *disk, int partno,
 		goto out_free_stats;
 	p->bdev = bdev;
 
-	hd_sects_seq_init(p);
 	pdev = part_to_dev(p);
 
 	p->start_sect = start;
-	p->nr_sects = len;
+	bdev_set_nr_sectors(bdev, len);
 	p->partno = partno;
 	p->policy = get_disk_ro(disk);
 
@@ -509,7 +515,7 @@ static bool partition_overlaps(struct gendisk *disk, sector_t start,
 	disk_part_iter_init(&piter, disk, DISK_PITER_INCL_EMPTY);
 	while ((part = disk_part_iter_next(&piter))) {
 		if (part->partno == skip_partno ||
-		    start >= part->start_sect + part->nr_sects ||
+		    start >= part->start_sect + bdev_nr_sectors(part->bdev) ||
 		    start + length <= part->start_sect)
 			continue;
 		overlap = true;
@@ -600,8 +606,7 @@ int bdev_resize_partition(struct block_device *bdev, int partno,
 	if (partition_overlaps(bdev->bd_disk, start, length, partno))
 		goto out_unlock;
 
-	part_nr_sects_write(part, length);
-	bd_set_nr_sectors(bdevp, length);
+	bdev_set_nr_sectors(bdevp, length);
 
 	ret = 0;
 out_unlock: