[PATCHSET 3/3 blk-for-2.6.28] block: unify disk/part handling and improve ext devt, take #2

[PATCHSET 3/3 blk-for-2.6.28] block: unify disk/part handling and improve ext devt, take #2

[Tejun Heo]

Hello, again.

This is the second take of unify-disk-part-handling patchset.

Changes from the last take[L] are...

* adapted to blk-for-2.6.28.
* devt alloc path simplification in 0013.

For general description of the patchset, please read description of
the last take[L].

This patchset is against...

  blk-for-2.6.28 (9abd7c437c02e7448fb1d2d3cfc0b9c1ab77cf2d)
+ [1] klist-dont-iterate-over-deleted-entries
+ [2] use-klist-for-class-device-list-and-implement-iterator
+ [3] block-misc-fixes-and-improvements patchset
+ [4] block-extended-devt patchset, take #3

and available in the following git tree.

  http://git.kernel.org/?p=linux/kernel/git/tj/misc.git;a=shortlog;h=block-unify-disk-part
  git://git.kernel.org/pub/scm/linux/kernel/git/tj/misc.git block-unify-disk-part

and the combined diffstat follows.

 block/blk-core.c           |   89 ++++-----
 block/blk-integrity.c      |    5 
 block/blk-merge.c          |   12 -
 block/blk-sysfs.c          |    4 
 block/cmd-filter.c         |   12 -
 block/genhd.c              |  402 ++++++++++++++++++++++-----------------------
 block/ioctl.c              |    4 
 drivers/block/aoe/aoeblk.c |    6 
 drivers/block/aoe/aoecmd.c |   16 -
 drivers/block/aoe/aoedev.c |    2 
 drivers/block/nbd.c        |    4 
 drivers/ide/ide-cd.c       |    2 
 drivers/ide/ide-disk.c     |   16 -
 drivers/ide/ide-probe.c    |    2 
 drivers/md/dm-ioctl.c      |    2 
 drivers/md/dm.c            |   31 +--
 drivers/md/linear.c        |    9 -
 drivers/md/md.c            |   15 -
 drivers/md/multipath.c     |    9 -
 drivers/md/raid0.c         |    9 -
 drivers/md/raid1.c         |    9 -
 drivers/md/raid10.c        |    9 -
 drivers/md/raid5.c         |    9 -
 drivers/scsi/sd.c          |   83 ++++++---
 fs/block_dev.c             |   79 +++-----
 fs/partitions/check.c      |  165 +++++++++---------
 fs/partitions/check.h      |    4 
 include/linux/genhd.h      |  247 +++++++++++----------------
 28 files changed, 610 insertions(+), 646 deletions(-)

Thanks.

--
tejun

[L] http://lkml.org/lkml/2008/7/14/63
[1] http://article.gmane.org/gmane.linux.kernel/725706
[2] http://article.gmane.org/gmane.linux.kernel/725708
[3] http://thread.gmane.org/gmane.linux.kernel/725724
[4] http://lkml.org/lkml/2008/8/25/108

[PATCH 01/13] block: implement and use {disk|part}_to_dev()

[Tejun Heo]

Implement {disk|part}_to_dev() and use them to access generic device
instead of directly dereferencing {disk|part}->dev.  To make sure no
user is left behind, rename generic devices fields to __dev.

This is in preparation of unifying partition 0 handling with other
partitions.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-integrity.c      |    5 ++-
 block/blk-sysfs.c          |    4 +-
 block/genhd.c              |   27 ++++++++-------
 drivers/block/aoe/aoeblk.c |    4 +-
 drivers/block/nbd.c        |    4 +-
 drivers/ide/ide-probe.c    |    2 +-
 drivers/md/dm.c            |    4 +-
 drivers/md/md.c            |   10 +++---
 fs/block_dev.c             |    4 +-
 fs/partitions/check.c      |   79 +++++++++++++++++++++++--------------------
 include/linux/genhd.h      |   20 ++++++-----
 11 files changed, 86 insertions(+), 77 deletions(-)

diff --git a/block/blk-integrity.c b/block/blk-integrity.c
index d87606e..69023da 100644
--- a/block/blk-integrity.c
+++ b/block/blk-integrity.c
@@ -331,7 +331,8 @@ int blk_integrity_register(struct gendisk *disk, struct blk_integrity *template)
 			return -1;
 
 		if (kobject_init_and_add(&bi->kobj, &integrity_ktype,
-					 &disk->dev.kobj, "%s", "integrity")) {
+					 &disk_to_dev(disk)->kobj,
+					 "%s", "integrity")) {
 			kmem_cache_free(integrity_cachep, bi);
 			return -1;
 		}
@@ -375,7 +376,7 @@ void blk_integrity_unregister(struct gendisk *disk)
 
 	kobject_uevent(&bi->kobj, KOBJ_REMOVE);
 	kobject_del(&bi->kobj);
-	kobject_put(&disk->dev.kobj);
+	kobject_put(&disk_to_dev(disk)->kobj);
 	kmem_cache_free(integrity_cachep, bi);
 }
 EXPORT_SYMBOL(blk_integrity_unregister);
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 304ec73..b9a6ed1 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -310,7 +310,7 @@ int blk_register_queue(struct gendisk *disk)
 	if (!q->request_fn)
 		return 0;
 
-	ret = kobject_add(&q->kobj, kobject_get(&disk->dev.kobj),
+	ret = kobject_add(&q->kobj, kobject_get(&disk_to_dev(disk)->kobj),
 			  "%s", "queue");
 	if (ret < 0)
 		return ret;
@@ -339,6 +339,6 @@ void blk_unregister_queue(struct gendisk *disk)
 
 		kobject_uevent(&q->kobj, KOBJ_REMOVE);
 		kobject_del(&q->kobj);
-		kobject_put(&disk->dev.kobj);
+		kobject_put(&disk_to_dev(disk)->kobj);
 	}
 }
diff --git a/block/genhd.c b/block/genhd.c
index dbfae29..c48e878 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -59,7 +59,7 @@ struct hd_struct *disk_get_part(struct gendisk *disk, int partno)
 	rcu_read_lock();
 	part = rcu_dereference(disk->__part[partno - 1]);
 	if (part)
-		get_device(&part->dev);
+		get_device(part_to_dev(part));
 	rcu_read_unlock();
 
 	return part;
@@ -130,7 +130,7 @@ struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter)
 		if (!(piter->flags & DISK_PITER_INCL_EMPTY) && !part->nr_sects)
 			continue;
 
-		get_device(&part->dev);
+		get_device(part_to_dev(part));
 		piter->part = part;
 		piter->idx += inc;
 		break;
@@ -435,7 +435,7 @@ static struct kobject *exact_match(dev_t devt, int *partno, void *data)
 {
 	struct gendisk *p = data;
 
-	return &p->dev.kobj;
+	return &disk_to_dev(p)->kobj;
 }
 
 static int exact_lock(dev_t devt, void *data)
@@ -460,7 +460,7 @@ void add_disk(struct gendisk *disk)
 	int retval;
 
 	disk->flags |= GENHD_FL_UP;
-	disk->dev.devt = MKDEV(disk->major, disk->first_minor);
+	disk_to_dev(disk)->devt = MKDEV(disk->major, disk->first_minor);
 	blk_register_region(disk_devt(disk), disk->minors, NULL,
 			    exact_match, exact_lock, disk);
 	register_disk(disk);
@@ -469,7 +469,8 @@ void add_disk(struct gendisk *disk)
 
 	bdi = &disk->queue->backing_dev_info;
 	bdi_register_dev(bdi, disk_devt(disk));
-	retval = sysfs_create_link(&disk->dev.kobj, &bdi->dev->kobj, "bdi");
+	retval = sysfs_create_link(&disk_to_dev(disk)->kobj, &bdi->dev->kobj,
+				   "bdi");
 	WARN_ON(retval);
 }
 
@@ -479,7 +480,7 @@ EXPORT_SYMBOL(del_gendisk);	/* in partitions/check.c */
 void unlink_gendisk(struct gendisk *disk)
 {
 	blk_unregister_filter(disk);
-	sysfs_remove_link(&disk->dev.kobj, "bdi");
+	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);
@@ -904,7 +905,7 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 	int cpu;
 
 	/*
-	if (&gp->dev.kobj.entry == block_class.devices.next)
+	if (&disk_to_dev(gp)->kobj.entry == block_class.devices.next)
 		seq_puts(seqf,	"major minor name"
 				"     rio rmerge rsect ruse wio wmerge "
 				"wsect wuse running use aveq"
@@ -973,7 +974,7 @@ static void media_change_notify_thread(struct work_struct *work)
 	 * 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(&gd->dev.kobj, KOBJ_CHANGE, envp);
+	kobject_uevent_env(&disk_to_dev(gd)->kobj, KOBJ_CHANGE, envp);
 	put_device(gd->driverfs_dev);
 }
 
@@ -1063,9 +1064,9 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 		disk->minors = minors;
 		disk->ext_minors = ext_minors;
 		rand_initialize_disk(disk);
-		disk->dev.class = &block_class;
-		disk->dev.type = &disk_type;
-		device_initialize(&disk->dev);
+		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);
 	}
@@ -1087,7 +1088,7 @@ struct kobject *get_disk(struct gendisk *disk)
 	owner = disk->fops->owner;
 	if (owner && !try_module_get(owner))
 		return NULL;
-	kobj = kobject_get(&disk->dev.kobj);
+	kobj = kobject_get(&disk_to_dev(disk)->kobj);
 	if (kobj == NULL) {
 		module_put(owner);
 		return NULL;
@@ -1101,7 +1102,7 @@ EXPORT_SYMBOL(get_disk);
 void put_disk(struct gendisk *disk)
 {
 	if (disk)
-		kobject_put(&disk->dev.kobj);
+		kobject_put(&disk_to_dev(disk)->kobj);
 }
 
 EXPORT_SYMBOL(put_disk);
diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c
index 0c39782..3edb6cb 100644
--- a/drivers/block/aoe/aoeblk.c
+++ b/drivers/block/aoe/aoeblk.c
@@ -109,12 +109,12 @@ static const struct attribute_group attr_group = {
 static int
 aoedisk_add_sysfs(struct aoedev *d)
 {
-	return sysfs_create_group(&d->gd->dev.kobj, &attr_group);
+	return sysfs_create_group(&disk_to_dev(d->gd)->kobj, &attr_group);
 }
 void
 aoedisk_rm_sysfs(struct aoedev *d)
 {
-	sysfs_remove_group(&d->gd->dev.kobj, &attr_group);
+	sysfs_remove_group(&disk_to_dev(d->gd)->kobj, &attr_group);
 }
 
 static int
diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index ad98dda..00337c8 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -403,7 +403,7 @@ static int nbd_do_it(struct nbd_device *lo)
 	BUG_ON(lo->magic != LO_MAGIC);
 
 	lo->pid = current->pid;
-	ret = sysfs_create_file(&lo->disk->dev.kobj, &pid_attr.attr);
+	ret = sysfs_create_file(&disk_to_dev(lo->disk)->kobj, &pid_attr.attr);
 	if (ret) {
 		printk(KERN_ERR "nbd: sysfs_create_file failed!");
 		return ret;
@@ -412,7 +412,7 @@ static int nbd_do_it(struct nbd_device *lo)
 	while ((req = nbd_read_stat(lo)) != NULL)
 		nbd_end_request(req);
 
-	sysfs_remove_file(&lo->disk->dev.kobj, &pid_attr.attr);
+	sysfs_remove_file(&disk_to_dev(lo->disk)->kobj, &pid_attr.attr);
 	return 0;
 }
 
diff --git a/drivers/ide/ide-probe.c b/drivers/ide/ide-probe.c
index 994e410..a872625 100644
--- a/drivers/ide/ide-probe.c
+++ b/drivers/ide/ide-probe.c
@@ -1188,7 +1188,7 @@ static struct kobject *exact_match(dev_t dev, int *part, void *data)
 {
 	struct gendisk *p = data;
 	*part &= (1 << PARTN_BITS) - 1;
-	return &p->dev.kobj;
+	return &disk_to_dev(p)->kobj;
 }
 
 static int exact_lock(dev_t dev, void *data)
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index d087435..7f8d6ff 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -1182,7 +1182,7 @@ static void event_callback(void *context)
 	list_splice_init(&md->uevent_list, &uevents);
 	spin_unlock_irqrestore(&md->uevent_lock, flags);
 
-	dm_send_uevents(&uevents, &md->disk->dev.kobj);
+	dm_send_uevents(&uevents, &disk_to_dev(md->disk)->kobj);
 
 	atomic_inc(&md->event_nr);
 	wake_up(&md->eventq);
@@ -1639,7 +1639,7 @@ out:
  *---------------------------------------------------------------*/
 void dm_kobject_uevent(struct mapped_device *md)
 {
-	kobject_uevent(&md->disk->dev.kobj, KOBJ_CHANGE);
+	kobject_uevent(&disk_to_dev(md->disk)->kobj, KOBJ_CHANGE);
 }
 
 uint32_t dm_next_uevent_seq(struct mapped_device *md)
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 8cfadc5..c815b9e 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1465,9 +1465,9 @@ static int bind_rdev_to_array(mdk_rdev_t * rdev, mddev_t * mddev)
 		goto fail;
 
 	if (rdev->bdev->bd_part)
-		ko = &rdev->bdev->bd_part->dev.kobj;
+		ko = &part_to_dev(rdev->bdev->bd_part)->kobj;
 	else
-		ko = &rdev->bdev->bd_disk->dev.kobj;
+		ko = &disk_to_dev(rdev->bdev->bd_disk)->kobj;
 	if ((err = sysfs_create_link(&rdev->kobj, ko, "block"))) {
 		kobject_del(&rdev->kobj);
 		goto fail;
@@ -3470,8 +3470,8 @@ static struct kobject *md_probe(dev_t dev, int *part, void *data)
 	disk->queue = mddev->queue;
 	add_disk(disk);
 	mddev->gendisk = disk;
-	error = kobject_init_and_add(&mddev->kobj, &md_ktype, &disk->dev.kobj,
-				     "%s", "md");
+	error = kobject_init_and_add(&mddev->kobj, &md_ktype,
+				     &disk_to_dev(disk)->kobj, "%s", "md");
 	mutex_unlock(&disks_mutex);
 	if (error)
 		printk(KERN_WARNING "md: cannot register %s/md - name in use\n",
@@ -3761,7 +3761,7 @@ static int do_md_run(mddev_t * mddev)
 	sysfs_notify(&mddev->kobj, NULL, "array_state");
 	sysfs_notify(&mddev->kobj, NULL, "sync_action");
 	sysfs_notify(&mddev->kobj, NULL, "degraded");
-	kobject_uevent(&mddev->gendisk->dev.kobj, KOBJ_CHANGE);
+	kobject_uevent(&disk_to_dev(mddev->gendisk)->kobj, KOBJ_CHANGE);
 	return 0;
 }
 
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 2f2873b..a02df22 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -543,9 +543,9 @@ EXPORT_SYMBOL(bd_release);
 static struct kobject *bdev_get_kobj(struct block_device *bdev)
 {
 	if (bdev->bd_contains != bdev)
-		return kobject_get(&bdev->bd_part->dev.kobj);
+		return kobject_get(&part_to_dev(bdev->bd_part)->kobj);
 	else
-		return kobject_get(&bdev->bd_disk->dev.kobj);
+		return kobject_get(&disk_to_dev(bdev->bd_disk)->kobj);
 }
 
 static struct kobject *bdev_get_holder(struct block_device *bdev)
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 0ecc93f..caa2bdb 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -309,7 +309,7 @@ static inline void disk_sysfs_add_subdirs(struct gendisk *disk)
 {
 	struct kobject *k;
 
-	k = kobject_get(&disk->dev.kobj);
+	k = kobject_get(&disk_to_dev(disk)->kobj);
 	disk->holder_dir = kobject_create_and_add("holders", k);
 	disk->slave_dir = kobject_create_and_add("slaves", k);
 	kobject_put(k);
@@ -322,7 +322,7 @@ static void delete_partition_rcu_cb(struct rcu_head *head)
 	part->start_sect = 0;
 	part->nr_sects = 0;
 	part_stat_set_all(part, 0);
-	put_device(&part->dev);
+	put_device(part_to_dev(part));
 }
 
 void delete_partition(struct gendisk *disk, int partno)
@@ -336,7 +336,7 @@ void delete_partition(struct gendisk *disk, int partno)
 	blk_free_devt(part_devt(part));
 	rcu_assign_pointer(disk->__part[partno-1], NULL);
 	kobject_put(part->holder_dir);
-	device_del(&part->dev);
+	device_del(part_to_dev(part));
 
 	call_rcu(&part->rcu_head, delete_partition_rcu_cb);
 }
@@ -354,6 +354,9 @@ int add_partition(struct gendisk *disk, int partno,
 {
 	struct hd_struct *p;
 	dev_t devt = MKDEV(0, 0);
+	struct device *ddev = disk_to_dev(disk);
+	struct device *pdev;
+	const char *dname;
 	int err;
 
 	if (disk->__part[partno - 1])
@@ -367,42 +370,43 @@ int add_partition(struct gendisk *disk, int partno,
 		err = -ENOMEM;
 		goto out_free;
 	}
+	pdev = part_to_dev(p);
+
 	p->start_sect = start;
 	p->nr_sects = len;
 	p->partno = partno;
 	p->policy = disk->policy;
 
-	if (isdigit(disk->dev.bus_id[strlen(disk->dev.bus_id)-1]))
-		snprintf(p->dev.bus_id, BUS_ID_SIZE,
-		"%sp%d", disk->dev.bus_id, partno);
+	dname = dev_name(ddev);
+	if (isdigit(dname[strlen(dname) - 1]))
+		snprintf(pdev->bus_id, BUS_ID_SIZE, "%sp%d", dname, partno);
 	else
-		snprintf(p->dev.bus_id, BUS_ID_SIZE,
-			 "%s%d", disk->dev.bus_id, partno);
+		snprintf(pdev->bus_id, BUS_ID_SIZE, "%s%d", dname, partno);
 
-	device_initialize(&p->dev);
-	p->dev.class = &block_class;
-	p->dev.type = &part_type;
-	p->dev.parent = &disk->dev;
+	device_initialize(pdev);
+	pdev->class = &block_class;
+	pdev->type = &part_type;
+	pdev->parent = ddev;
 
 	err = blk_alloc_devt(p, &devt);
 	if (err)
-		goto out_put;
-	p->dev.devt = devt;
+		goto out_free;
+	pdev->devt = devt;
 
 	/* delay uevent until 'holders' subdir is created */
-	p->dev.uevent_suppress = 1;
-	err = device_add(&p->dev);
+	pdev->uevent_suppress = 1;
+	err = device_add(pdev);
 	if (err)
 		goto out_put;
 
 	err = -ENOMEM;
-	p->holder_dir = kobject_create_and_add("holders", &p->dev.kobj);
+	p->holder_dir = kobject_create_and_add("holders", &pdev->kobj);
 	if (!p->holder_dir)
 		goto out_del;
 
-	p->dev.uevent_suppress = 0;
+	pdev->uevent_suppress = 0;
 	if (flags & ADDPART_FLAG_WHOLEDISK) {
-		err = device_create_file(&p->dev, &dev_attr_whole_disk);
+		err = device_create_file(pdev, &dev_attr_whole_disk);
 		if (err)
 			goto out_del;
 	}
@@ -412,8 +416,8 @@ int add_partition(struct gendisk *disk, int partno,
 	rcu_assign_pointer(disk->__part[partno - 1], p);
 
 	/* suppress uevent if the disk supresses it */
-	if (!disk->dev.uevent_suppress)
-		kobject_uevent(&p->dev.kobj, KOBJ_ADD);
+	if (!ddev->uevent_suppress)
+		kobject_uevent(&pdev->kobj, KOBJ_ADD);
 
 	return 0;
 
@@ -422,9 +426,9 @@ out_free:
 	return err;
 out_del:
 	kobject_put(p->holder_dir);
-	device_del(&p->dev);
+	device_del(pdev);
 out_put:
-	put_device(&p->dev);
+	put_device(pdev);
 	blk_free_devt(devt);
 	return err;
 }
@@ -432,30 +436,31 @@ out_put:
 /* Not exported, helper to add_disk(). */
 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;
 	char *s;
 	int err;
 
-	disk->dev.parent = disk->driverfs_dev;
+	ddev->parent = disk->driverfs_dev;
 
-	strlcpy(disk->dev.bus_id, disk->disk_name, BUS_ID_SIZE);
+	strlcpy(ddev->bus_id, disk->disk_name, BUS_ID_SIZE);
 	/* ewww... some of these buggers have / in the name... */
-	s = strchr(disk->dev.bus_id, '/');
+	s = strchr(ddev->bus_id, '/');
 	if (s)
 		*s = '!';
 
 	/* delay uevents, until we scanned partition table */
-	disk->dev.uevent_suppress = 1;
+	ddev->uevent_suppress = 1;
 
-	if (device_add(&disk->dev))
+	if (device_add(ddev))
 		return;
 #ifndef CONFIG_SYSFS_DEPRECATED
-	err = sysfs_create_link(block_depr, &disk->dev.kobj,
-				kobject_name(&disk->dev.kobj));
+	err = sysfs_create_link(block_depr, &ddev->kobj,
+				kobject_name(&ddev->kobj));
 	if (err) {
-		device_del(&disk->dev);
+		device_del(ddev);
 		return;
 	}
 #endif
@@ -481,13 +486,13 @@ void register_disk(struct gendisk *disk)
 
 exit:
 	/* announce disk after possible partitions are created */
-	disk->dev.uevent_suppress = 0;
-	kobject_uevent(&disk->dev.kobj, KOBJ_ADD);
+	ddev->uevent_suppress = 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->dev.kobj, KOBJ_ADD);
+		kobject_uevent(&part_to_dev(part)->kobj, KOBJ_ADD);
 	disk_part_iter_exit(&piter);
 }
 
@@ -518,7 +523,7 @@ int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
 		return -EIO;
 
 	/* tell userspace that the media / partition table may have changed */
-	kobject_uevent(&disk->dev.kobj, KOBJ_CHANGE);
+	kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE);
 
 	for (p = 1; p < state->limit; p++) {
 		sector_t size = state->parts[p].size;
@@ -591,7 +596,7 @@ void del_gendisk(struct gendisk *disk)
 	kobject_put(disk->slave_dir);
 	disk->driverfs_dev = NULL;
 #ifndef CONFIG_SYSFS_DEPRECATED
-	sysfs_remove_link(block_depr, disk->dev.bus_id);
+	sysfs_remove_link(block_depr, dev_name(disk_to_dev(disk)));
 #endif
-	device_del(&disk->dev);
+	device_del(disk_to_dev(disk));
 }
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index c9a7315..5669e8f 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -15,9 +15,11 @@
 
 #ifdef CONFIG_BLOCK
 
-#define kobj_to_dev(k) container_of(k, struct device, kobj)
-#define dev_to_disk(device) container_of(device, struct gendisk, dev)
-#define dev_to_part(device) container_of(device, struct hd_struct, dev)
+#define kobj_to_dev(k)		container_of((k), struct device, kobj)
+#define dev_to_disk(device)	container_of((device), struct gendisk, __dev)
+#define dev_to_part(device)	container_of((device), struct hd_struct, __dev)
+#define disk_to_dev(disk)	(&((disk)->__dev))
+#define part_to_dev(part)	(&((part)->__dev))
 
 extern struct device_type part_type;
 extern struct kobject *block_depr;
@@ -88,7 +90,7 @@ struct disk_stats {
 struct hd_struct {
 	sector_t start_sect;
 	sector_t nr_sects;
-	struct device dev;
+	struct device __dev;
 	struct kobject *holder_dir;
 	int policy, partno;
 #ifdef CONFIG_FAIL_MAKE_REQUEST
@@ -149,7 +151,7 @@ struct gendisk {
 
 	int flags;
 	struct device *driverfs_dev;  // FIXME: remove
-	struct device dev;
+	struct device __dev;
 	struct kobject *holder_dir;
 	struct kobject *slave_dir;
 
@@ -173,7 +175,7 @@ struct gendisk {
 static inline struct gendisk *part_to_disk(struct hd_struct *part)
 {
 	if (likely(part))
-		return dev_to_disk((part)->dev.parent);
+		return dev_to_disk(part_to_dev(part)->parent);
 	return NULL;
 }
 
@@ -184,12 +186,12 @@ static inline int disk_max_parts(struct gendisk *disk)
 
 static inline dev_t disk_devt(struct gendisk *disk)
 {
-	return disk->dev.devt;
+	return disk_to_dev(disk)->devt;
 }
 
 static inline dev_t part_devt(struct hd_struct *part)
 {
-	return part->dev.devt;
+	return part_to_dev(part)->devt;
 }
 
 extern struct hd_struct *disk_get_part(struct gendisk *disk, int partno);
@@ -197,7 +199,7 @@ extern struct hd_struct *disk_get_part(struct gendisk *disk, int partno);
 static inline void disk_put_part(struct hd_struct *part)
 {
 	if (likely(part))
-		put_device(&part->dev);
+		put_device(part_to_dev(part));
 }
 
 /*
-- 
1.5.4.5

[PATCH 02/13] block: introduce partition 0

[Tejun Heo]

genhd and partition code handled disk and partitions separately.  All
information about the whole disk was in struct genhd and partitions in
struct hd_struct.  However, the whole disk (part0) and other
partitions have a lot in common and the data structures end up having
good number of common fields and thus separate code paths doing the
same thing.  Also, the partition array was indexed by partno - 1 which
gets pretty confusing at times.

This patch introduces partition 0 and makes the partition array
indexed by partno.  Following patches will unify the handling of disk
and parts piece-by-piece.

This patch also implements disk_partitionable() which tests whether a
disk is partitionable.  With coming dynamic partition array change,
the most common usage of disk_max_parts() will be testing whether a
disk is partitionable and the number of max partitions will become
much less important.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c         |   40 +++++++++++++++++++++++-----------------
 block/ioctl.c         |    4 ++--
 fs/block_dev.c        |    2 +-
 fs/partitions/check.c |   12 ++++++------
 include/linux/genhd.h |   11 +++++++++--
 5 files changed, 41 insertions(+), 28 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index c48e878..092f18f 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -54,10 +54,10 @@ struct hd_struct *disk_get_part(struct gendisk *disk, int partno)
 {
 	struct hd_struct *part;
 
-	if (unlikely(partno < 1 || partno > disk_max_parts(disk)))
+	if (unlikely(partno < 0 || partno >= disk_max_parts(disk)))
 		return NULL;
 	rcu_read_lock();
-	part = rcu_dereference(disk->__part[partno - 1]);
+	part = rcu_dereference(disk->__part[partno]);
 	if (part)
 		get_device(part_to_dev(part));
 	rcu_read_unlock();
@@ -85,8 +85,10 @@ void disk_part_iter_init(struct disk_part_iter *piter, struct gendisk *disk,
 
 	if (flags & DISK_PITER_REVERSE)
 		piter->idx = disk_max_parts(piter->disk) - 1;
-	else
+	else if (flags & DISK_PITER_INCL_PART0)
 		piter->idx = 0;
+	else
+		piter->idx = 1;
 
 	piter->flags = flags;
 }
@@ -114,7 +116,10 @@ struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter)
 	/* determine iteration parameters */
 	if (piter->flags & DISK_PITER_REVERSE) {
 		inc = -1;
-		end = -1;
+		if (piter->flags & DISK_PITER_INCL_PART0)
+			end = -1;
+		else
+			end = 0;
 	} else {
 		inc = 1;
 		end = disk_max_parts(piter->disk);
@@ -177,7 +182,7 @@ struct hd_struct *disk_map_sector_rcu(struct gendisk *disk, sector_t sector)
 {
 	int i;
 
-	for (i = 0; i < disk_max_parts(disk); i++) {
+	for (i = 1; i < disk_max_parts(disk); i++) {
 		struct hd_struct *part = rcu_dereference(disk->__part[i]);
 
 		if (part && part->start_sect <= sector &&
@@ -670,7 +675,7 @@ static int show_partition(struct seq_file *seqf, void *v)
 	char buf[BDEVNAME_SIZE];
 
 	/* Don't show non-partitionable removeable devices or empty devices */
-	if (!get_capacity(sgp) || (!disk_max_parts(sgp) &&
+	if (!get_capacity(sgp) || (!disk_partitionable(sgp) &&
 				   (sgp->flags & GENHD_FL_REMOVABLE)))
 		return 0;
 	if (sgp->flags & GENHD_FL_SUPPRESS_PARTITION_INFO)
@@ -743,7 +748,7 @@ static ssize_t disk_ext_range_show(struct device *dev,
 {
 	struct gendisk *disk = dev_to_disk(dev);
 
-	return sprintf(buf, "%d\n", disk_max_parts(disk) + 1);
+	return sprintf(buf, "%d\n", disk_max_parts(disk));
 }
 
 static ssize_t disk_removable_show(struct device *dev,
@@ -999,7 +1004,7 @@ dev_t blk_lookup_devt(const char *name, int partno)
 
 		if (strcmp(dev->bus_id, name))
 			continue;
-		if (partno < 0 || partno > disk_max_parts(disk))
+		if (partno < 0 || partno >= disk_max_parts(disk))
 			continue;
 
 		if (partno == 0)
@@ -1046,21 +1051,22 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 				GFP_KERNEL | __GFP_ZERO, node_id);
 	if (disk) {
 		int tot_minors = minors + ext_minors;
+		int size = tot_minors * sizeof(struct hd_struct *);
 
 		if (!init_disk_stats(disk)) {
 			kfree(disk);
 			return NULL;
 		}
-		if (tot_minors > 1) {
-			int size = (tot_minors - 1) * sizeof(struct hd_struct *);
-			disk->__part = kmalloc_node(size,
-				GFP_KERNEL | __GFP_ZERO, node_id);
-			if (!disk->__part) {
-				free_disk_stats(disk);
-				kfree(disk);
-				return NULL;
-			}
+
+		disk->__part = kmalloc_node(size, GFP_KERNEL | __GFP_ZERO,
+					    node_id);
+		if (!disk->__part) {
+			free_disk_stats(disk);
+			kfree(disk);
+			return NULL;
 		}
+		disk->__part[0] = &disk->part0;
+
 		disk->minors = minors;
 		disk->ext_minors = ext_minors;
 		rand_initialize_disk(disk);
diff --git a/block/ioctl.c b/block/ioctl.c
index a5f672a..64e7c67 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -30,7 +30,7 @@ static int blkpg_ioctl(struct block_device *bdev, struct blkpg_ioctl_arg __user
 	if (bdev != bdev->bd_contains)
 		return -EINVAL;
 	partno = p.pno;
-	if (partno <= 0 || partno > disk_max_parts(disk))
+	if (partno <= 0 || partno >= disk_max_parts(disk))
 		return -EINVAL;
 	switch (a.op) {
 		case BLKPG_ADD_PARTITION:
@@ -102,7 +102,7 @@ static int blkdev_reread_part(struct block_device *bdev)
 	struct gendisk *disk = bdev->bd_disk;
 	int res;
 
-	if (!disk_max_parts(disk) || bdev != bdev->bd_contains)
+	if (!disk_partitionable(disk) || bdev != bdev->bd_contains)
 		return -EINVAL;
 	if (!capable(CAP_SYS_ADMIN))
 		return -EACCES;
diff --git a/fs/block_dev.c b/fs/block_dev.c
index a02df22..c982a91 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -892,7 +892,7 @@ int check_disk_change(struct block_device *bdev)
 
 	if (bdops->revalidate_disk)
 		bdops->revalidate_disk(bdev->bd_disk);
-	if (disk_max_parts(bdev->bd_disk))
+	if (disk_partitionable(bdev->bd_disk))
 		bdev->bd_invalidated = 1;
 	return 1;
 }
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index caa2bdb..121d445 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -173,7 +173,7 @@ check_partition(struct gendisk *hd, struct block_device *bdev)
 	if (isdigit(state->name[strlen(state->name)-1]))
 		sprintf(state->name, "p");
 
-	state->limit = disk_max_parts(hd) + 1;
+	state->limit = disk_max_parts(hd);
 	i = res = err = 0;
 	while (!res && check_part[i]) {
 		memset(&state->parts, 0, sizeof(state->parts));
@@ -329,12 +329,12 @@ void delete_partition(struct gendisk *disk, int partno)
 {
 	struct hd_struct *part;
 
-	part = disk->__part[partno-1];
+	part = disk->__part[partno];
 	if (!part)
 		return;
 
 	blk_free_devt(part_devt(part));
-	rcu_assign_pointer(disk->__part[partno-1], NULL);
+	rcu_assign_pointer(disk->__part[partno], NULL);
 	kobject_put(part->holder_dir);
 	device_del(part_to_dev(part));
 
@@ -359,7 +359,7 @@ int add_partition(struct gendisk *disk, int partno,
 	const char *dname;
 	int err;
 
-	if (disk->__part[partno - 1])
+	if (disk->__part[partno])
 		return -EBUSY;
 
 	p = kzalloc(sizeof(*p), GFP_KERNEL);
@@ -413,7 +413,7 @@ int add_partition(struct gendisk *disk, int partno,
 
 	/* everything is up and running, commence */
 	INIT_RCU_HEAD(&p->rcu_head);
-	rcu_assign_pointer(disk->__part[partno - 1], p);
+	rcu_assign_pointer(disk->__part[partno], p);
 
 	/* suppress uevent if the disk supresses it */
 	if (!ddev->uevent_suppress)
@@ -467,7 +467,7 @@ void register_disk(struct gendisk *disk)
 	disk_sysfs_add_subdirs(disk);
 
 	/* No minors to use for partitions */
-	if (!disk_max_parts(disk))
+	if (!disk_partitionable(disk))
 		goto exit;
 
 	/* No such device (e.g., media were just removed) */
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 5669e8f..9970d96 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -136,12 +136,13 @@ struct gendisk {
 
 	char disk_name[32];		/* name of major driver */
 
-	/* Array of pointers to partitions indexed by partno - 1.
+	/* Array of pointers to partitions indexed by partno.
 	 * Protected with matching bdev lock but stat and other
 	 * non-critical accesses use RCU.  Always access through
 	 * helpers.
 	 */
 	struct hd_struct **__part;
+	struct hd_struct part0;
 
 	struct block_device_operations *fops;
 	struct request_queue *queue;
@@ -181,7 +182,12 @@ static inline struct gendisk *part_to_disk(struct hd_struct *part)
 
 static inline int disk_max_parts(struct gendisk *disk)
 {
-	return disk->minors + disk->ext_minors - 1;
+	return disk->minors + disk->ext_minors;
+}
+
+static inline bool disk_partitionable(struct gendisk *disk)
+{
+	return disk_max_parts(disk) > 1;
 }
 
 static inline dev_t disk_devt(struct gendisk *disk)
@@ -207,6 +213,7 @@ static inline void disk_put_part(struct hd_struct *part)
  */
 #define DISK_PITER_REVERSE	(1 << 0) /* iterate in the reverse direction */
 #define DISK_PITER_INCL_EMPTY	(1 << 1) /* include 0-sized parts */
+#define DISK_PITER_INCL_PART0	(1 << 2) /* include partition 0 */
 
 struct disk_part_iter {
 	struct gendisk		*disk;
-- 
1.5.4.5

[PATCH 03/13] block: move capacity from disk to part0

[Tejun Heo]

Move disk->capacity to part0->nr_sects and convert all users who
directly accessed the field to use {get|set}_capacity().  This is done
early to allow the __dev field to be moved.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 drivers/block/aoe/aoeblk.c |    2 +-
 drivers/block/aoe/aoecmd.c |    4 ++--
 drivers/block/aoe/aoedev.c |    2 +-
 fs/partitions/check.c      |    2 +-
 include/linux/genhd.h      |    5 ++---
 5 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/drivers/block/aoe/aoeblk.c b/drivers/block/aoe/aoeblk.c
index 3edb6cb..aa69556 100644
--- a/drivers/block/aoe/aoeblk.c
+++ b/drivers/block/aoe/aoeblk.c
@@ -276,7 +276,7 @@ aoeblk_gdalloc(void *vp)
 	gd->first_minor = d->sysminor * AOE_PARTITIONS;
 	gd->fops = &aoe_bdops;
 	gd->private_data = d;
-	gd->capacity = d->ssize;
+	set_capacity(gd, d->ssize);
 	snprintf(gd->disk_name, sizeof gd->disk_name, "etherd/e%ld.%d",
 		d->aoemajor, d->aoeminor);
 
diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c
index 17eed8c..934800f 100644
--- a/drivers/block/aoe/aoecmd.c
+++ b/drivers/block/aoe/aoecmd.c
@@ -645,7 +645,7 @@ aoecmd_sleepwork(struct work_struct *work)
 		unsigned long flags;
 		u64 ssize;
 
-		ssize = d->gd->capacity;
+		ssize = get_capacity(d->gd);
 		bd = bdget_disk(d->gd, 0);
 
 		if (bd) {
@@ -707,7 +707,7 @@ ataid_complete(struct aoedev *d, struct aoetgt *t, unsigned char *id)
 	if (d->flags & (DEVFL_GDALLOC|DEVFL_NEWSIZE))
 		return;
 	if (d->gd != NULL) {
-		d->gd->capacity = ssize;
+		set_capacity(d->gd, ssize);
 		d->flags |= DEVFL_NEWSIZE;
 	} else
 		d->flags |= DEVFL_GDALLOC;
diff --git a/drivers/block/aoe/aoedev.c b/drivers/block/aoe/aoedev.c
index a1d813a..6a8038d 100644
--- a/drivers/block/aoe/aoedev.c
+++ b/drivers/block/aoe/aoedev.c
@@ -91,7 +91,7 @@ aoedev_downdev(struct aoedev *d)
 	}
 
 	if (d->gd)
-		d->gd->capacity = 0;
+		set_capacity(d->gd, 0);
 
 	d->flags &= ~DEVFL_UP;
 }
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 121d445..540ef84 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -586,7 +586,7 @@ void del_gendisk(struct gendisk *disk)
 	disk_part_iter_exit(&piter);
 
 	invalidate_partition(disk, 0);
-	disk->capacity = 0;
+	set_capacity(disk, 0);
 	disk->flags &= ~GENHD_FL_UP;
 	unlink_gendisk(disk);
 	disk_stat_set_all(disk, 0);
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 9970d96..c4490f2 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -148,7 +148,6 @@ struct gendisk {
 	struct request_queue *queue;
 	struct blk_scsi_cmd_filter cmd_filter;
 	void *private_data;
-	sector_t capacity;
 
 	int flags;
 	struct device *driverfs_dev;  // FIXME: remove
@@ -421,11 +420,11 @@ static inline sector_t get_start_sect(struct block_device *bdev)
 }
 static inline sector_t get_capacity(struct gendisk *disk)
 {
-	return disk->capacity;
+	return disk->part0.nr_sects;
 }
 static inline void set_capacity(struct gendisk *disk, sector_t size)
 {
-	disk->capacity = size;
+	disk->part0.nr_sects = size;
 }
 
 #ifdef CONFIG_SOLARIS_X86_PARTITION
-- 
1.5.4.5

[PATCH 04/13] block: move __dev from disk to part0

[Tejun Heo]

Move disk->__dev to part0->__dev.  This simplifies bdget_disk() and
lookup_devt() and allows common sysfs attributes to be unified.
part_to_disk() is updated to handle part0 -> disk.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c         |   40 ++++++++++++----------------------------
 include/linux/genhd.h |   13 ++++++++-----
 2 files changed, 20 insertions(+), 33 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index 092f18f..aca135a 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -539,22 +539,15 @@ struct gendisk *get_gendisk(dev_t devt, int *partno)
  */
 extern struct block_device *bdget_disk(struct gendisk *disk, int partno)
 {
-	dev_t devt = MKDEV(0, 0);
+	struct hd_struct *part;
+	struct block_device *bdev = NULL;
 
-	if (partno == 0)
-		devt = disk_devt(disk);
-	else {
-		struct hd_struct *part;
+	part = disk_get_part(disk, partno);
+	if (part && part->nr_sects)
+		bdev = bdget(part_devt(part));
+	disk_put_part(part);
 
-		part = disk_get_part(disk, partno);
-		if (part && part->nr_sects)
-			devt = part_devt(part);
-		disk_put_part(part);
-	}
-
-	if (likely(devt != MKDEV(0, 0)))
-		return bdget(devt);
-	return NULL;
+	return bdev;
 }
 EXPORT_SYMBOL(bdget_disk);
 
@@ -1001,27 +994,18 @@ dev_t blk_lookup_devt(const char *name, int partno)
 	class_dev_iter_init(&iter, &block_class, NULL, &disk_type);
 	while ((dev = class_dev_iter_next(&iter))) {
 		struct gendisk *disk = dev_to_disk(dev);
+		struct hd_struct *part;
 
 		if (strcmp(dev->bus_id, name))
 			continue;
-		if (partno < 0 || partno >= disk_max_parts(disk))
-			continue;
-
-		if (partno == 0)
-			devt = disk_devt(disk);
-		else {
-			struct hd_struct *part;
-
-			part = disk_get_part(disk, partno);
-			if (!part || !part->nr_sects) {
-				disk_put_part(part);
-				continue;
-			}
 
+		part = disk_get_part(disk, partno);
+		if (part && part->nr_sects) {
 			devt = part_devt(part);
 			disk_put_part(part);
+			break;
 		}
-		break;
+		disk_put_part(part);
 	}
 	class_dev_iter_exit(&iter);
 	return devt;
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index c4490f2..242effa 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -16,9 +16,9 @@
 #ifdef CONFIG_BLOCK
 
 #define kobj_to_dev(k)		container_of((k), struct device, kobj)
-#define dev_to_disk(device)	container_of((device), struct gendisk, __dev)
+#define dev_to_disk(device)	container_of((device), struct gendisk, part0.__dev)
 #define dev_to_part(device)	container_of((device), struct hd_struct, __dev)
-#define disk_to_dev(disk)	(&((disk)->__dev))
+#define disk_to_dev(disk)	(&(disk)->part0.__dev)
 #define part_to_dev(part)	(&((part)->__dev))
 
 extern struct device_type part_type;
@@ -151,7 +151,6 @@ struct gendisk {
 
 	int flags;
 	struct device *driverfs_dev;  // FIXME: remove
-	struct device __dev;
 	struct kobject *holder_dir;
 	struct kobject *slave_dir;
 
@@ -174,8 +173,12 @@ struct gendisk {
 
 static inline struct gendisk *part_to_disk(struct hd_struct *part)
 {
-	if (likely(part))
-		return dev_to_disk(part_to_dev(part)->parent);
+	if (likely(part)) {
+		if (part->partno)
+			return dev_to_disk(part_to_dev(part)->parent);
+		else
+			return dev_to_disk(part_to_dev(part));
+	}
 	return NULL;
 }
 
-- 
1.5.4.5

[PATCH 05/13] block: unify sysfs size node handling

[Tejun Heo]

Now that capacity and __dev are moved to part0, part0 and others can
share the same method.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c         |   10 +---------
 fs/partitions/check.c |    4 ++--
 include/linux/genhd.h |    3 +++
 3 files changed, 6 insertions(+), 11 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index aca135a..b560d71 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -761,14 +761,6 @@ static ssize_t disk_ro_show(struct device *dev,
 	return sprintf(buf, "%d\n", disk->policy ? 1 : 0);
 }
 
-static ssize_t disk_size_show(struct device *dev,
-			      struct device_attribute *attr, char *buf)
-{
-	struct gendisk *disk = dev_to_disk(dev);
-
-	return sprintf(buf, "%llu\n", (unsigned long long)get_capacity(disk));
-}
-
 static ssize_t disk_capability_show(struct device *dev,
 				    struct device_attribute *attr, char *buf)
 {
@@ -836,7 +828,7 @@ static DEVICE_ATTR(range, S_IRUGO, disk_range_show, NULL);
 static DEVICE_ATTR(ext_range, S_IRUGO, disk_ext_range_show, NULL);
 static DEVICE_ATTR(removable, S_IRUGO, disk_removable_show, NULL);
 static DEVICE_ATTR(ro, S_IRUGO, disk_ro_show, NULL);
-static DEVICE_ATTR(size, S_IRUGO, disk_size_show, NULL);
+static DEVICE_ATTR(size, S_IRUGO, part_size_show, NULL);
 static DEVICE_ATTR(capability, S_IRUGO, disk_capability_show, NULL);
 static DEVICE_ATTR(stat, S_IRUGO, disk_stat_show, NULL);
 #ifdef CONFIG_FAIL_MAKE_REQUEST
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 540ef84..9ab3e64 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -208,8 +208,8 @@ static ssize_t part_start_show(struct device *dev,
 	return sprintf(buf, "%llu\n",(unsigned long long)p->start_sect);
 }
 
-static ssize_t part_size_show(struct device *dev,
-			      struct device_attribute *attr, char *buf)
+ssize_t part_size_show(struct device *dev,
+		       struct device_attribute *attr, char *buf)
 {
 	struct hd_struct *p = dev_to_part(dev);
 	return sprintf(buf, "%llu\n",(unsigned long long)p->nr_sects);
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 242effa..e89559a 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -601,6 +601,9 @@ extern void blk_register_region(dev_t devt, unsigned long range,
 			void *data);
 extern void blk_unregister_region(dev_t devt, unsigned long range);
 
+extern ssize_t part_size_show(struct device *dev,
+			      struct device_attribute *attr, char *buf);
+
 #else /* CONFIG_BLOCK */
 
 static inline void printk_all_partitions(void) { }
-- 
1.5.4.5

[PATCH 06/13] block: move policy from disk to part0

[Tejun Heo]

Move disk->policy to part0->policy.  Implement and use get_disk_ro().

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c         |   16 +++++-----------
 drivers/ide/ide-cd.c  |    2 +-
 drivers/md/dm-ioctl.c |    2 +-
 fs/partitions/check.c |    2 +-
 include/linux/genhd.h |    6 +++++-
 5 files changed, 13 insertions(+), 15 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index b560d71..8762f35 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -758,7 +758,7 @@ static ssize_t disk_ro_show(struct device *dev,
 {
 	struct gendisk *disk = dev_to_disk(dev);
 
-	return sprintf(buf, "%d\n", disk->policy ? 1 : 0);
+	return sprintf(buf, "%d\n", get_disk_ro(disk) ? 1 : 0);
 }
 
 static ssize_t disk_capability_show(struct device *dev,
@@ -1091,10 +1091,7 @@ EXPORT_SYMBOL(put_disk);
 
 void set_device_ro(struct block_device *bdev, int flag)
 {
-	if (bdev->bd_contains != bdev)
-		bdev->bd_part->policy = flag;
-	else
-		bdev->bd_disk->policy = flag;
+	bdev->bd_part->policy = flag;
 }
 
 EXPORT_SYMBOL(set_device_ro);
@@ -1104,8 +1101,8 @@ void set_disk_ro(struct gendisk *disk, int flag)
 	struct disk_part_iter piter;
 	struct hd_struct *part;
 
-	disk->policy = flag;
-	disk_part_iter_init(&piter, disk, DISK_PITER_INCL_EMPTY);
+	disk_part_iter_init(&piter, disk,
+			    DISK_PITER_INCL_EMPTY | DISK_PITER_INCL_PART0);
 	while ((part = disk_part_iter_next(&piter)))
 		part->policy = flag;
 	disk_part_iter_exit(&piter);
@@ -1117,10 +1114,7 @@ int bdev_read_only(struct block_device *bdev)
 {
 	if (!bdev)
 		return 0;
-	else if (bdev->bd_contains != bdev)
-		return bdev->bd_part->policy;
-	else
-		return bdev->bd_disk->policy;
+	return bdev->bd_part->policy;
 }
 
 EXPORT_SYMBOL(bdev_read_only);
diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c
index 49a8c58..62c3a79 100644
--- a/drivers/ide/ide-cd.c
+++ b/drivers/ide/ide-cd.c
@@ -1113,7 +1113,7 @@ static ide_startstop_t cdrom_start_rw(ide_drive_t *drive, struct request *rq)
 
 	if (write) {
 		/* disk has become write protected */
-		if (cd->disk->policy) {
+		if (get_disk_ro(cd->disk)) {
 			cdrom_end_request(drive, 0);
 			return ide_stopped;
 		}
diff --git a/drivers/md/dm-ioctl.c b/drivers/md/dm-ioctl.c
index c3de311..5b91915 100644
--- a/drivers/md/dm-ioctl.c
+++ b/drivers/md/dm-ioctl.c
@@ -548,7 +548,7 @@ static int __dev_status(struct mapped_device *md, struct dm_ioctl *param)
 	 */
 	param->open_count = dm_open_count(md);
 
-	if (disk->policy)
+	if (get_disk_ro(disk))
 		param->flags |= DM_READONLY_FLAG;
 
 	param->event_nr = dm_get_event_nr(md);
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 9ab3e64..cb9ec77 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -375,7 +375,7 @@ int add_partition(struct gendisk *disk, int partno,
 	p->start_sect = start;
 	p->nr_sects = len;
 	p->partno = partno;
-	p->policy = disk->policy;
+	p->policy = get_disk_ro(disk);
 
 	dname = dev_name(ddev);
 	if (isdigit(dname[strlen(dname) - 1]))
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index e89559a..461ba6d 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -155,7 +155,6 @@ struct gendisk {
 	struct kobject *slave_dir;
 
 	struct timer_rand_state *random;
-	int policy;
 
 	atomic_t sync_io;		/* RAID */
 	unsigned long stamp;
@@ -413,6 +412,11 @@ extern struct block_device *bdget_disk(struct gendisk *disk, int partno);
 extern void set_device_ro(struct block_device *bdev, int flag);
 extern void set_disk_ro(struct gendisk *disk, int flag);
 
+static inline int get_disk_ro(struct gendisk *disk)
+{
+	return disk->part0.policy;
+}
+
 /* drivers/char/random.c */
 extern void add_disk_randomness(struct gendisk *disk);
 extern void rand_initialize_disk(struct gendisk *disk);
-- 
1.5.4.5

[PATCH 07/13] block: move holder_dir from disk to part0

[Tejun Heo]

Move disk->holder_dir to part0->holder_dir.  Kill now mostly
superflous bdev_get_holder().

While at it, kill superflous kobject_get/put() around holder_dir,
slave_dir and cmd_filter creation and collapse
disk_sysfs_add_subdirs() into register_disk().  These serve no purpose
but obfuscating the code.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/cmd-filter.c    |   12 +++---------
 fs/block_dev.c        |   10 +---------
 fs/partitions/check.c |   15 +++------------
 include/linux/genhd.h |    1 -
 4 files changed, 7 insertions(+), 31 deletions(-)

diff --git a/block/cmd-filter.c b/block/cmd-filter.c
index eec4404..1ab98b2 100644
--- a/block/cmd-filter.c
+++ b/block/cmd-filter.c
@@ -309,14 +309,10 @@ int blk_register_filter(struct gendisk *disk)
 {
 	int ret;
 	struct blk_scsi_cmd_filter *filter = &disk->cmd_filter;
-	struct kobject *parent = kobject_get(disk->holder_dir->parent);
-
-	if (!parent)
-		return -ENODEV;
-
-	ret = kobject_init_and_add(&filter->kobj, &rcf_ktype, parent,
-				 "%s", "cmd_filter");
 
+	ret = kobject_init_and_add(&filter->kobj, &rcf_ktype,
+				   &disk_to_dev(disk)->kobj,
+				   "%s", "cmd_filter");
 	if (ret < 0)
 		return ret;
 
@@ -329,6 +325,4 @@ void blk_unregister_filter(struct gendisk *disk)
 	struct blk_scsi_cmd_filter *filter = &disk->cmd_filter;
 
 	kobject_put(&filter->kobj);
-	kobject_put(disk->holder_dir->parent);
 }
-
diff --git a/fs/block_dev.c b/fs/block_dev.c
index c982a91..57d5726 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -548,14 +548,6 @@ static struct kobject *bdev_get_kobj(struct block_device *bdev)
 		return kobject_get(&disk_to_dev(bdev->bd_disk)->kobj);
 }
 
-static struct kobject *bdev_get_holder(struct block_device *bdev)
-{
-	if (bdev->bd_contains != bdev)
-		return kobject_get(bdev->bd_part->holder_dir);
-	else
-		return kobject_get(bdev->bd_disk->holder_dir);
-}
-
 static int add_symlink(struct kobject *from, struct kobject *to)
 {
 	if (!from || !to)
@@ -608,7 +600,7 @@ static int bd_holder_grab_dirs(struct block_device *bdev,
 	if (!bo->sdev)
 		goto fail_put_hdev;
 
-	bo->hdir = bdev_get_holder(bdev);
+	bo->hdir = kobject_get(bdev->bd_part->holder_dir);
 	if (!bo->hdir)
 		goto fail_put_sdev;
 
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index cb9ec77..7e4a7d4 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -305,16 +305,6 @@ struct device_type part_type = {
 	.release	= part_release,
 };
 
-static inline void disk_sysfs_add_subdirs(struct gendisk *disk)
-{
-	struct kobject *k;
-
-	k = kobject_get(&disk_to_dev(disk)->kobj);
-	disk->holder_dir = kobject_create_and_add("holders", k);
-	disk->slave_dir = kobject_create_and_add("slaves", k);
-	kobject_put(k);
-}
-
 static void delete_partition_rcu_cb(struct rcu_head *head)
 {
 	struct hd_struct *part = container_of(head, struct hd_struct, rcu_head);
@@ -464,7 +454,8 @@ void register_disk(struct gendisk *disk)
 		return;
 	}
 #endif
-	disk_sysfs_add_subdirs(disk);
+	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))
@@ -592,7 +583,7 @@ void del_gendisk(struct gendisk *disk)
 	disk_stat_set_all(disk, 0);
 	disk->stamp = 0;
 
-	kobject_put(disk->holder_dir);
+	kobject_put(disk->part0.holder_dir);
 	kobject_put(disk->slave_dir);
 	disk->driverfs_dev = NULL;
 #ifndef CONFIG_SYSFS_DEPRECATED
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 461ba6d..a8573ba 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -151,7 +151,6 @@ struct gendisk {
 
 	int flags;
 	struct device *driverfs_dev;  // FIXME: remove
-	struct kobject *holder_dir;
 	struct kobject *slave_dir;
 
 	struct timer_rand_state *random;
-- 
1.5.4.5

[PATCH 08/13] block: always set bdev->bd_part

[Tejun Heo]

Till now, bdev->bd_part is set only if the bdev was for parts other
than part0.  This patch makes bdev->bd_part always set so that code
paths don't have to differenciate common handling.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-core.c      |    2 +-
 drivers/md/md.c       |    5 +---
 fs/block_dev.c        |   67 ++++++++++++++++++++++--------------------------
 fs/partitions/check.c |    7 +----
 include/linux/genhd.h |    2 +-
 5 files changed, 35 insertions(+), 48 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index 3de5610..09bba70 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1272,7 +1272,7 @@ __setup("fail_make_request=", setup_fail_make_request);
 static int should_fail_request(struct bio *bio)
 {
 	if ((bio->bi_bdev->bd_disk->flags & GENHD_FL_FAIL) ||
-	    (bio->bi_bdev->bd_part && bio->bi_bdev->bd_part->make_it_fail))
+	    bio->bi_bdev->bd_part->make_it_fail)
 		return should_fail(&fail_make_request, bio->bi_size);
 
 	return 0;
diff --git a/drivers/md/md.c b/drivers/md/md.c
index c815b9e..5d0cf63 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -1464,10 +1464,7 @@ static int bind_rdev_to_array(mdk_rdev_t * rdev, mddev_t * mddev)
 	if ((err = kobject_add(&rdev->kobj, &mddev->kobj, "dev-%s", b)))
 		goto fail;
 
-	if (rdev->bdev->bd_part)
-		ko = &part_to_dev(rdev->bdev->bd_part)->kobj;
-	else
-		ko = &disk_to_dev(rdev->bdev->bd_disk)->kobj;
+	ko = &part_to_dev(rdev->bdev->bd_part)->kobj;
 	if ((err = sysfs_create_link(&rdev->kobj, ko, "block"))) {
 		kobject_del(&rdev->kobj);
 		goto fail;
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 57d5726..c3fa19b 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -540,14 +540,6 @@ EXPORT_SYMBOL(bd_release);
  *           /sys/block/sda/holders/dm-0 --> /sys/block/dm-0
  */
 
-static struct kobject *bdev_get_kobj(struct block_device *bdev)
-{
-	if (bdev->bd_contains != bdev)
-		return kobject_get(&part_to_dev(bdev->bd_part)->kobj);
-	else
-		return kobject_get(&disk_to_dev(bdev->bd_disk)->kobj);
-}
-
 static int add_symlink(struct kobject *from, struct kobject *to)
 {
 	if (!from || !to)
@@ -596,7 +588,7 @@ static int bd_holder_grab_dirs(struct block_device *bdev,
 	if (!bo->hdev)
 		goto fail_put_sdir;
 
-	bo->sdev = bdev_get_kobj(bdev);
+	bo->sdev = kobject_get(&part_to_dev(bdev->bd_part)->kobj);
 	if (!bo->sdev)
 		goto fail_put_hdev;
 
@@ -919,7 +911,6 @@ static int __blkdev_put(struct block_device *bdev, int for_part);
 
 static int do_open(struct block_device *bdev, struct file *file, int for_part)
 {
-	struct module *owner = NULL;
 	struct gendisk *disk;
 	struct hd_struct *part = NULL;
 	int ret;
@@ -941,25 +932,27 @@ static int do_open(struct block_device *bdev, struct file *file, int for_part)
 
 	ret = -ENXIO;
 	file->f_mapping = bdev->bd_inode->i_mapping;
+
 	lock_kernel();
+
 	disk = get_gendisk(bdev->bd_dev, &partno);
-	if (!disk) {
-		unlock_kernel();
-		bdput(bdev);
-		return ret;
-	}
-	owner = disk->fops->owner;
+	if (!disk)
+		goto out_unlock_kernel;
+	part = disk_get_part(disk, partno);
+	if (!part)
+		goto out_unlock_kernel;
 
 	mutex_lock_nested(&bdev->bd_mutex, for_part);
 	if (!bdev->bd_openers) {
 		bdev->bd_disk = disk;
+		bdev->bd_part = part;
 		bdev->bd_contains = bdev;
 		if (!partno) {
 			struct backing_dev_info *bdi;
 			if (disk->fops->open) {
 				ret = disk->fops->open(bdev->bd_inode, file);
 				if (ret)
-					goto out_first;
+					goto out_clear;
 			}
 			if (!bdev->bd_openers) {
 				bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);
@@ -975,31 +968,32 @@ static int do_open(struct block_device *bdev, struct file *file, int for_part)
 			whole = bdget_disk(disk, 0);
 			ret = -ENOMEM;
 			if (!whole)
-				goto out_first;
+				goto out_clear;
 			BUG_ON(for_part);
 			ret = __blkdev_get(whole, file->f_mode, file->f_flags, 1);
 			if (ret)
-				goto out_first;
+				goto out_clear;
 			bdev->bd_contains = whole;
-			part = disk_get_part(disk, partno);
 			bdev->bd_inode->i_data.backing_dev_info =
 			   whole->bd_inode->i_data.backing_dev_info;
 			if (!(disk->flags & GENHD_FL_UP) ||
 			    !part || !part->nr_sects) {
 				ret = -ENXIO;
-				goto out_first;
+				goto out_clear;
 			}
-			bdev->bd_part = part;
 			bd_set_size(bdev, (loff_t)part->nr_sects << 9);
 		}
 	} else {
+		disk_put_part(part);
 		put_disk(disk);
-		module_put(owner);
+		module_put(disk->fops->owner);
+		part = NULL;
+		disk = NULL;
 		if (bdev->bd_contains == bdev) {
 			if (bdev->bd_disk->fops->open) {
 				ret = bdev->bd_disk->fops->open(bdev->bd_inode, file);
 				if (ret)
-					goto out;
+					goto out_unlock_bdev;
 			}
 			if (bdev->bd_invalidated)
 				rescan_partitions(bdev->bd_disk, bdev);
@@ -1012,20 +1006,24 @@ static int do_open(struct block_device *bdev, struct file *file, int for_part)
 	unlock_kernel();
 	return 0;
 
-out_first:
+ out_clear:
 	bdev->bd_disk = NULL;
+	bdev->bd_part = NULL;
 	bdev->bd_inode->i_data.backing_dev_info = &default_backing_dev_info;
 	if (bdev != bdev->bd_contains)
 		__blkdev_put(bdev->bd_contains, 1);
 	bdev->bd_contains = NULL;
-	put_disk(disk);
-	disk_put_part(part);
-	module_put(owner);
-out:
+ out_unlock_bdev:
 	mutex_unlock(&bdev->bd_mutex);
+ out_unlock_kernel:
 	unlock_kernel();
-	if (ret)
-		bdput(bdev);
+
+	disk_put_part(part);
+	if (disk)
+		module_put(disk->fops->owner);
+	put_disk(disk);
+	bdput(bdev);
+
 	return ret;
 }
 
@@ -1110,11 +1108,8 @@ static int __blkdev_put(struct block_device *bdev, int for_part)
 
 		put_disk(disk);
 		module_put(owner);
-
-		if (bdev->bd_contains != bdev) {
-			disk_put_part(bdev->bd_part);
-			bdev->bd_part = NULL;
-		}
+		disk_put_part(bdev->bd_part);
+		bdev->bd_part = NULL;
 		bdev->bd_disk = NULL;
 		bdev->bd_inode->i_data.backing_dev_info = &default_backing_dev_info;
 		if (bdev != bdev->bd_contains)
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 7e4a7d4..f99dd96 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -134,12 +134,7 @@ char *disk_name(struct gendisk *hd, int partno, char *buf)
 
 const char *bdevname(struct block_device *bdev, char *buf)
 {
-	int partno = 0;
-
-	if (bdev->bd_part)
-		partno = bdev->bd_part->partno;
-
-	return disk_name(bdev->bd_disk, partno, buf);
+	return disk_name(bdev->bd_disk, bdev->bd_part->partno, buf);
 }
 
 EXPORT_SYMBOL(bdevname);
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index a8573ba..0743324 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -422,7 +422,7 @@ extern void rand_initialize_disk(struct gendisk *disk);
 
 static inline sector_t get_start_sect(struct block_device *bdev)
 {
-	return bdev->bd_contains == bdev ? 0 : bdev->bd_part->start_sect;
+	return bdev->bd_part->start_sect;
 }
 static inline sector_t get_capacity(struct gendisk *disk)
 {
-- 
1.5.4.5

[PATCH 09/13] block: kill GENHD_FL_FAIL and use part0->make_it_fail

[Tejun Heo]

GENHD_FL_FAIL for disk is what make_it_fail is for parts.  Kill it and
use part0->make_it_fail.  Sysfs node handling is unified too.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-core.c      |    5 +++--
 block/genhd.c         |   30 +-----------------------------
 fs/partitions/check.c |   10 +++++-----
 include/linux/genhd.h |    8 +++++++-
 4 files changed, 16 insertions(+), 37 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index 09bba70..3a059cf 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -1271,8 +1271,9 @@ __setup("fail_make_request=", setup_fail_make_request);
 
 static int should_fail_request(struct bio *bio)
 {
-	if ((bio->bi_bdev->bd_disk->flags & GENHD_FL_FAIL) ||
-	    bio->bi_bdev->bd_part->make_it_fail)
+	struct hd_struct *part = bio->bi_bdev->bd_part;
+
+	if (part_to_disk(part)->part0.make_it_fail || part->make_it_fail)
 		return should_fail(&fail_make_request, bio->bi_size);
 
 	return 0;
diff --git a/block/genhd.c b/block/genhd.c
index 8762f35..b0a11c7 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -796,34 +796,6 @@ static ssize_t disk_stat_show(struct device *dev,
 		jiffies_to_msecs(disk_stat_read(disk, time_in_queue)));
 }
 
-#ifdef CONFIG_FAIL_MAKE_REQUEST
-static ssize_t disk_fail_show(struct device *dev,
-			      struct device_attribute *attr, char *buf)
-{
-	struct gendisk *disk = dev_to_disk(dev);
-
-	return sprintf(buf, "%d\n", disk->flags & GENHD_FL_FAIL ? 1 : 0);
-}
-
-static ssize_t disk_fail_store(struct device *dev,
-			       struct device_attribute *attr,
-			       const char *buf, size_t count)
-{
-	struct gendisk *disk = dev_to_disk(dev);
-	int i;
-
-	if (count > 0 && sscanf(buf, "%d", &i) > 0) {
-		if (i == 0)
-			disk->flags &= ~GENHD_FL_FAIL;
-		else
-			disk->flags |= GENHD_FL_FAIL;
-	}
-
-	return count;
-}
-
-#endif
-
 static DEVICE_ATTR(range, S_IRUGO, disk_range_show, NULL);
 static DEVICE_ATTR(ext_range, S_IRUGO, disk_ext_range_show, NULL);
 static DEVICE_ATTR(removable, S_IRUGO, disk_removable_show, NULL);
@@ -833,7 +805,7 @@ static DEVICE_ATTR(capability, S_IRUGO, disk_capability_show, NULL);
 static DEVICE_ATTR(stat, S_IRUGO, disk_stat_show, NULL);
 #ifdef CONFIG_FAIL_MAKE_REQUEST
 static struct device_attribute dev_attr_fail =
-	__ATTR(make-it-fail, S_IRUGO|S_IWUSR, disk_fail_show, disk_fail_store);
+	__ATTR(make-it-fail, S_IRUGO|S_IWUSR, part_fail_show, part_fail_store);
 #endif
 
 static struct attribute *disk_attrs[] = {
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index f99dd96..88fcf5a 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -238,17 +238,17 @@ static ssize_t part_stat_show(struct device *dev,
 }
 
 #ifdef CONFIG_FAIL_MAKE_REQUEST
-static ssize_t part_fail_show(struct device *dev,
-			      struct device_attribute *attr, char *buf)
+ssize_t part_fail_show(struct device *dev,
+		       struct device_attribute *attr, char *buf)
 {
 	struct hd_struct *p = dev_to_part(dev);
 
 	return sprintf(buf, "%d\n", p->make_it_fail);
 }
 
-static ssize_t part_fail_store(struct device *dev,
-			       struct device_attribute *attr,
-			       const char *buf, size_t count)
+ssize_t part_fail_store(struct device *dev,
+			struct device_attribute *attr,
+			const char *buf, size_t count)
 {
 	struct hd_struct *p = dev_to_part(dev);
 	int i;
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 0743324..46d2a6f 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -112,7 +112,6 @@ struct hd_struct {
 #define GENHD_FL_CD				8
 #define GENHD_FL_UP				16
 #define GENHD_FL_SUPPRESS_PARTITION_INFO	32
-#define GENHD_FL_FAIL				64
 
 #define BLK_SCSI_MAX_CMDS	(256)
 #define BLK_SCSI_CMD_PER_LONG	(BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
@@ -606,6 +605,13 @@ extern void blk_unregister_region(dev_t devt, unsigned long range);
 
 extern ssize_t part_size_show(struct device *dev,
 			      struct device_attribute *attr, char *buf);
+#ifdef CONFIG_FAIL_MAKE_REQUEST
+extern ssize_t part_fail_show(struct device *dev,
+			      struct device_attribute *attr, char *buf);
+extern ssize_t part_fail_store(struct device *dev,
+			       struct device_attribute *attr,
+			       const char *buf, size_t count);
+#endif /* CONFIG_FAIL_MAKE_REQUEST */
 
 #else /* CONFIG_BLOCK */
 
-- 
1.5.4.5

[PATCH 10/13] block: move stats from disk to part0

[Tejun Heo]

Move stats related fields - stamp, in_flight, dkstats - from disk to
part0 and unify stat handling such that...

* part_stat_*() now updates part0 together if the specified partition
  is not part0.  ie. part_stat_*() are now essentially all_stat_*().

* {disk|all}_stat_*() are gone.

* part_round_stats() is updated similary.  It handles part0 stats
  automatically and disk_round_stats() is killed.

* part_{inc|dec}_in_fligh() is implemented which automatically updates
  part0 stats for parts other than part0.

* disk_map_sector_rcu() is updated to return part0 if no part matches.
  Combined with the above changes, this makes NULL special case
  handling in callers unnecessary.

* Separate stats show code paths for disk are collapsed into part
  stats show code paths.

* Rename disk_stat_lock/unlock() to part_stat_lock/unlock()

While at it, reposition stat handling macros a bit and add missing
parentheses around macro parameters.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-core.c           |   84 +++++++++--------------
 block/blk-merge.c          |   12 +--
 block/genhd.c              |   97 +++++++--------------------
 drivers/block/aoe/aoecmd.c |   12 ++--
 drivers/md/dm.c            |   27 ++++----
 drivers/md/linear.c        |    9 ++-
 drivers/md/md.c            |    4 +-
 drivers/md/multipath.c     |    9 ++-
 drivers/md/raid0.c         |    9 ++-
 drivers/md/raid1.c         |    9 ++-
 drivers/md/raid10.c        |    9 ++-
 drivers/md/raid5.c         |    9 ++-
 fs/partitions/check.c      |   12 ++--
 include/linux/genhd.h      |  159 ++++++++++++-------------------------------
 14 files changed, 165 insertions(+), 296 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index 3a059cf..0f339ff 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -61,21 +61,17 @@ static void drive_stat_acct(struct request *rq, int new_io)
 	if (!blk_fs_request(rq) || !rq->rq_disk)
 		return;
 
-	cpu = disk_stat_lock();
+	cpu = part_stat_lock();
 	part = disk_map_sector_rcu(rq->rq_disk, rq->sector);
 
 	if (!new_io)
-		all_stat_inc(cpu, rq->rq_disk, part, merges[rw], rq->sector);
+		part_stat_inc(cpu, part, merges[rw]);
 	else {
-		disk_round_stats(cpu, rq->rq_disk);
-		rq->rq_disk->in_flight++;
-		if (part) {
-			part_round_stats(cpu, part);
-			part->in_flight++;
-		}
+		part_round_stats(cpu, part);
+		part_inc_in_flight(part);
 	}
 
-	disk_stat_unlock();
+	part_stat_unlock();
 }
 
 void blk_queue_congestion_threshold(struct request_queue *q)
@@ -981,8 +977,22 @@ static inline void add_request(struct request_queue *q, struct request *req)
 	__elv_add_request(q, req, ELEVATOR_INSERT_SORT, 0);
 }
 
-/*
- * disk_round_stats()	- Round off the performance stats on a struct
+static void part_round_stats_single(int cpu, struct hd_struct *part,
+				    unsigned long now)
+{
+	if (now == part->stamp)
+		return;
+
+	if (part->in_flight) {
+		__part_stat_add(cpu, part, time_in_queue,
+				part->in_flight * (now - part->stamp));
+		__part_stat_add(cpu, part, io_ticks, (now - part->stamp));
+	}
+	part->stamp = now;
+}
+
+/**
+ * part_round_stats()	- Round off the performance stats on a struct
  * disk_stats.
  *
  * The average IO queue length and utilisation statistics are maintained
@@ -996,36 +1006,15 @@ static inline void add_request(struct request_queue *q, struct request *req)
  * /proc/diskstats.  This accounts immediately for all queue usage up to
  * the current jiffies and restarts the counters again.
  */
-void disk_round_stats(int cpu, struct gendisk *disk)
-{
-	unsigned long now = jiffies;
-
-	if (now == disk->stamp)
-		return;
-
-	if (disk->in_flight) {
-		disk_stat_add(cpu, disk, time_in_queue,
-			      disk->in_flight * (now - disk->stamp));
-		disk_stat_add(cpu, disk, io_ticks, (now - disk->stamp));
-	}
-	disk->stamp = now;
-}
-EXPORT_SYMBOL_GPL(disk_round_stats);
-
 void part_round_stats(int cpu, struct hd_struct *part)
 {
 	unsigned long now = jiffies;
 
-	if (now == part->stamp)
-		return;
-
-	if (part->in_flight) {
-		part_stat_add(cpu, part, time_in_queue,
-			      part->in_flight * (now - part->stamp));
-		part_stat_add(cpu, part, io_ticks, (now - part->stamp));
-	}
-	part->stamp = now;
+	if (part->partno)
+		part_round_stats_single(cpu, &part_to_disk(part)->part0, now);
+	part_round_stats_single(cpu, part, now);
 }
+EXPORT_SYMBOL_GPL(part_round_stats);
 
 /*
  * queue lock must be held
@@ -1565,11 +1554,10 @@ static int __end_that_request_first(struct request *req, int error,
 		struct hd_struct *part;
 		int cpu;
 
-		cpu = disk_stat_lock();
+		cpu = part_stat_lock();
 		part = disk_map_sector_rcu(req->rq_disk, req->sector);
-		all_stat_add(cpu, req->rq_disk, part, sectors[rw],
-			     nr_bytes >> 9, req->sector);
-		disk_stat_unlock();
+		part_stat_add(cpu, part, sectors[rw], nr_bytes >> 9);
+		part_stat_unlock();
 	}
 
 	total_bytes = bio_nbytes = 0;
@@ -1756,19 +1744,15 @@ static void end_that_request_last(struct request *req, int error)
 		struct hd_struct *part;
 		int cpu;
 
-		cpu = disk_stat_lock();
+		cpu = part_stat_lock();
 		part = disk_map_sector_rcu(disk, req->sector);
 
-		all_stat_inc(cpu, disk, part, ios[rw], req->sector);
-		all_stat_add(cpu, disk, part, ticks[rw], duration, req->sector);
-		disk_round_stats(cpu, disk);
-		disk->in_flight--;
-		if (part) {
-			part_round_stats(cpu, part);
-			part->in_flight--;
-		}
+		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);
 
-		disk_stat_unlock();
+		part_stat_unlock();
 	}
 
 	if (req->end_io)
diff --git a/block/blk-merge.c b/block/blk-merge.c
index d926a24..c77196d 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -390,17 +390,13 @@ static int attempt_merge(struct request_queue *q, struct request *req,
 		struct hd_struct *part;
 		int cpu;
 
-		cpu = disk_stat_lock();
+		cpu = part_stat_lock();
 		part = disk_map_sector_rcu(req->rq_disk, req->sector);
 
-		disk_round_stats(cpu, req->rq_disk);
-		req->rq_disk->in_flight--;
-		if (part) {
-			part_round_stats(cpu, part);
-			part->in_flight--;
-		}
+		part_round_stats(cpu, part);
+		part_dec_in_flight(part);
 
-		disk_stat_unlock();
+		part_stat_unlock();
 	}
 
 	req->ioprio = ioprio_best(req->ioprio, next->ioprio);
diff --git a/block/genhd.c b/block/genhd.c
index b0a11c7..cf8d55e 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -176,7 +176,7 @@ EXPORT_SYMBOL_GPL(disk_part_iter_exit);
  * while preemption is disabled.
  *
  * RETURNS:
- * Found partition on success, NULL if there's no matching partition.
+ * Found partition on success, part0 is returned if no partition matches
  */
 struct hd_struct *disk_map_sector_rcu(struct gendisk *disk, sector_t sector)
 {
@@ -189,7 +189,7 @@ struct hd_struct *disk_map_sector_rcu(struct gendisk *disk, sector_t sector)
 		    sector < part->start_sect + part->nr_sects)
 			return part;
 	}
-	return NULL;
+	return &disk->part0;
 }
 EXPORT_SYMBOL_GPL(disk_map_sector_rcu);
 
@@ -582,24 +582,24 @@ void __init printk_all_partitions(void)
 		 * numbers in hex - the same format as the root=
 		 * option takes.
 		 */
-		printk("%s %10llu %s",
-		       bdevt_str(disk_devt(disk), devt_buf),
-		       (unsigned long long)get_capacity(disk) >> 1,
-		       disk_name(disk, 0, name_buf));
-		if (disk->driverfs_dev != NULL &&
-		    disk->driverfs_dev->driver != NULL)
-			printk(" driver: %s\n",
-			       disk->driverfs_dev->driver->name);
-		else
-			printk(" (driver?)\n");
+		disk_part_iter_init(&piter, disk, DISK_PITER_INCL_PART0);
+		while ((part = disk_part_iter_next(&piter))) {
+			bool is_part0 = part == &disk->part0;
 
-		/* now show the partitions */
-		disk_part_iter_init(&piter, disk, 0);
-		while ((part = disk_part_iter_next(&piter)))
-			printk("  %s %10llu %s\n",
+			printk("%s%s %10llu %s", is_part0 ? "" : "  ",
 			       bdevt_str(part_devt(part), devt_buf),
 			       (unsigned long long)part->nr_sects >> 1,
 			       disk_name(disk, part->partno, name_buf));
+			if (is_part0) {
+				if (disk->driverfs_dev != NULL &&
+				    disk->driverfs_dev->driver != NULL)
+					printk(" driver: %s\n",
+					      disk->driverfs_dev->driver->name);
+				else
+					printk(" (driver?)\n");
+			} else
+				printk("\n");
+		}
 		disk_part_iter_exit(&piter);
 	}
 	class_dev_iter_exit(&iter);
@@ -675,12 +675,7 @@ static int show_partition(struct seq_file *seqf, void *v)
 		return 0;
 
 	/* show the full disk and all non-0 size partitions of it */
-	seq_printf(seqf, "%4d  %7d %10llu %s\n",
-		MAJOR(disk_devt(sgp)), MINOR(disk_devt(sgp)),
-		(unsigned long long)get_capacity(sgp) >> 1,
-		disk_name(sgp, 0, buf));
-
-	disk_part_iter_init(&piter, sgp, 0);
+	disk_part_iter_init(&piter, sgp, DISK_PITER_INCL_PART0);
 	while ((part = disk_part_iter_next(&piter)))
 		seq_printf(seqf, "%4d  %7d %10llu %s\n",
 			   MAJOR(part_devt(part)), MINOR(part_devt(part)),
@@ -769,40 +764,13 @@ static ssize_t disk_capability_show(struct device *dev,
 	return sprintf(buf, "%x\n", disk->flags);
 }
 
-static ssize_t disk_stat_show(struct device *dev,
-			      struct device_attribute *attr, char *buf)
-{
-	struct gendisk *disk = dev_to_disk(dev);
-	int cpu;
-
-	cpu = disk_stat_lock();
-	disk_round_stats(cpu, disk);
-	disk_stat_unlock();
-	return sprintf(buf,
-		"%8lu %8lu %8llu %8u "
-		"%8lu %8lu %8llu %8u "
-		"%8u %8u %8u"
-		"\n",
-		disk_stat_read(disk, ios[READ]),
-		disk_stat_read(disk, merges[READ]),
-		(unsigned long long)disk_stat_read(disk, sectors[READ]),
-		jiffies_to_msecs(disk_stat_read(disk, ticks[READ])),
-		disk_stat_read(disk, ios[WRITE]),
-		disk_stat_read(disk, merges[WRITE]),
-		(unsigned long long)disk_stat_read(disk, sectors[WRITE]),
-		jiffies_to_msecs(disk_stat_read(disk, ticks[WRITE])),
-		disk->in_flight,
-		jiffies_to_msecs(disk_stat_read(disk, io_ticks)),
-		jiffies_to_msecs(disk_stat_read(disk, time_in_queue)));
-}
-
 static DEVICE_ATTR(range, S_IRUGO, disk_range_show, NULL);
 static DEVICE_ATTR(ext_range, S_IRUGO, disk_ext_range_show, NULL);
 static DEVICE_ATTR(removable, S_IRUGO, disk_removable_show, NULL);
 static DEVICE_ATTR(ro, S_IRUGO, disk_ro_show, NULL);
 static DEVICE_ATTR(size, S_IRUGO, part_size_show, NULL);
 static DEVICE_ATTR(capability, S_IRUGO, disk_capability_show, NULL);
-static DEVICE_ATTR(stat, S_IRUGO, disk_stat_show, NULL);
+static DEVICE_ATTR(stat, S_IRUGO, part_stat_show, NULL);
 #ifdef CONFIG_FAIL_MAKE_REQUEST
 static struct device_attribute dev_attr_fail =
 	__ATTR(make-it-fail, S_IRUGO|S_IWUSR, part_fail_show, part_fail_store);
@@ -837,7 +805,7 @@ static void disk_release(struct device *dev)
 
 	kfree(disk->random);
 	kfree(disk->__part);
-	free_disk_stats(disk);
+	free_part_stats(&disk->part0);
 	kfree(disk);
 }
 struct class block_class = {
@@ -874,28 +842,11 @@ static int diskstats_show(struct seq_file *seqf, void *v)
 				"\n\n");
 	*/
  
-	cpu = disk_stat_lock();
-	disk_round_stats(cpu, gp);
-	disk_stat_unlock();
-	seq_printf(seqf, "%4d %7d %s %lu %lu %llu %u %lu %lu %llu %u %u %u %u\n",
-		MAJOR(disk_devt(gp)), MINOR(disk_devt(gp)),
-		disk_name(gp, 0, buf),
-		disk_stat_read(gp, ios[0]), disk_stat_read(gp, merges[0]),
-		(unsigned long long)disk_stat_read(gp, sectors[0]),
-		jiffies_to_msecs(disk_stat_read(gp, ticks[0])),
-		disk_stat_read(gp, ios[1]), disk_stat_read(gp, merges[1]),
-		(unsigned long long)disk_stat_read(gp, sectors[1]),
-		jiffies_to_msecs(disk_stat_read(gp, ticks[1])),
-		gp->in_flight,
-		jiffies_to_msecs(disk_stat_read(gp, io_ticks)),
-		jiffies_to_msecs(disk_stat_read(gp, time_in_queue)));
-
-	/* now show all non-0 size partitions of it */
-	disk_part_iter_init(&piter, gp, 0);
+	disk_part_iter_init(&piter, gp, DISK_PITER_INCL_PART0);
 	while ((hd = disk_part_iter_next(&piter))) {
-		cpu = disk_stat_lock();
+		cpu = part_stat_lock();
 		part_round_stats(cpu, hd);
-		disk_stat_unlock();
+		part_stat_unlock();
 		seq_printf(seqf, "%4d %7d %s %lu %lu %llu "
 			   "%u %lu %lu %llu %u %u %u %u\n",
 			   MAJOR(part_devt(hd)), MINOR(part_devt(hd)),
@@ -1001,7 +952,7 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 		int tot_minors = minors + ext_minors;
 		int size = tot_minors * sizeof(struct hd_struct *);
 
-		if (!init_disk_stats(disk)) {
+		if (!init_part_stats(&disk->part0)) {
 			kfree(disk);
 			return NULL;
 		}
@@ -1009,7 +960,7 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 		disk->__part = kmalloc_node(size, GFP_KERNEL | __GFP_ZERO,
 					    node_id);
 		if (!disk->__part) {
-			free_disk_stats(disk);
+				free_part_stats(&disk->part0);
 			kfree(disk);
 			return NULL;
 		}
diff --git a/drivers/block/aoe/aoecmd.c b/drivers/block/aoe/aoecmd.c
index 934800f..961d29a 100644
--- a/drivers/block/aoe/aoecmd.c
+++ b/drivers/block/aoe/aoecmd.c
@@ -758,15 +758,15 @@ diskstats(struct gendisk *disk, struct bio *bio, ulong duration, sector_t sector
 	struct hd_struct *part;
 	int cpu;
 
-	cpu = disk_stat_lock();
+	cpu = part_stat_lock();
 	part = disk_map_sector_rcu(disk, sector);
 
-	all_stat_inc(cpu, disk, part, ios[rw], sector);
-	all_stat_add(cpu, disk, part, ticks[rw], duration, sector);
-	all_stat_add(cpu, disk, part, sectors[rw], n_sect, sector);
-	all_stat_add(cpu, disk, part, io_ticks, duration, sector);
+	part_stat_inc(cpu, part, ios[rw]);
+	part_stat_add(cpu, part, ticks[rw], duration);
+	part_stat_add(cpu, part, sectors[rw], n_sect);
+	part_stat_add(cpu, part, io_ticks, duration);
 
-	disk_stat_unlock();
+	part_stat_unlock();
 }
 
 void
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 7f8d6ff..8d40369 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -381,10 +381,10 @@ static void start_io_acct(struct dm_io *io)
 
 	io->start_time = jiffies;
 
-	cpu = disk_stat_lock();
-	disk_round_stats(cpu, dm_disk(md));
-	disk_stat_unlock();
-	dm_disk(md)->in_flight = atomic_inc_return(&md->pending);
+	cpu = part_stat_lock();
+	part_round_stats(cpu, &dm_disk(md)->part0);
+	part_stat_unlock();
+	dm_disk(md)->part0.in_flight = atomic_inc_return(&md->pending);
 }
 
 static int end_io_acct(struct dm_io *io)
@@ -395,12 +395,13 @@ static int end_io_acct(struct dm_io *io)
 	int pending, cpu;
 	int rw = bio_data_dir(bio);
 
-	cpu = disk_stat_lock();
-	disk_round_stats(cpu, dm_disk(md));
-	disk_stat_add(cpu, dm_disk(md), ticks[rw], duration);
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_round_stats(cpu, &dm_disk(md)->part0);
+	part_stat_add(cpu, &dm_disk(md)->part0, ticks[rw], duration);
+	part_stat_unlock();
 
-	dm_disk(md)->in_flight = pending = atomic_dec_return(&md->pending);
+	dm_disk(md)->part0.in_flight = pending =
+		atomic_dec_return(&md->pending);
 
 	return !pending;
 }
@@ -895,10 +896,10 @@ static int dm_request(struct request_queue *q, struct bio *bio)
 
 	down_read(&md->io_lock);
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, dm_disk(md), ios[rw]);
-	disk_stat_add(cpu, dm_disk(md), sectors[rw], bio_sectors(bio));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &dm_disk(md)->part0, ios[rw]);
+	part_stat_add(cpu, &dm_disk(md)->part0, sectors[rw], bio_sectors(bio));
+	part_stat_unlock();
 
 	/*
 	 * If we're suspended we have to queue
diff --git a/drivers/md/linear.c b/drivers/md/linear.c
index 00cbc8e..c80ea90 100644
--- a/drivers/md/linear.c
+++ b/drivers/md/linear.c
@@ -325,10 +325,11 @@ static int linear_make_request (struct request_queue *q, struct bio *bio)
 		return 0;
 	}
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, mddev->gendisk, ios[rw]);
-	disk_stat_add(cpu, mddev->gendisk, sectors[rw], bio_sectors(bio));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
+		      bio_sectors(bio));
+	part_stat_unlock();
 
 	tmp_dev = which_dev(mddev, bio->bi_sector);
 	block = bio->bi_sector >> 1;
diff --git a/drivers/md/md.c b/drivers/md/md.c
index 5d0cf63..607c302 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -5548,8 +5548,8 @@ static int is_mddev_idle(mddev_t *mddev)
 	rcu_read_lock();
 	rdev_for_each_rcu(rdev, mddev) {
 		struct gendisk *disk = rdev->bdev->bd_contains->bd_disk;
-		curr_events = disk_stat_read(disk, sectors[0]) + 
-				disk_stat_read(disk, sectors[1]) - 
+		curr_events = part_stat_read(&disk->part0, sectors[0]) +
+				part_stat_read(&disk->part0, sectors[1]) -
 				atomic_read(&disk->sync_io);
 		/* sync IO will cause sync_io to increase before the disk_stats
 		 * as sync_io is counted when a request starts, and
diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c
index 182f5a9..8bb8794 100644
--- a/drivers/md/multipath.c
+++ b/drivers/md/multipath.c
@@ -159,10 +159,11 @@ static int multipath_make_request (struct request_queue *q, struct bio * bio)
 	mp_bh->master_bio = bio;
 	mp_bh->mddev = mddev;
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, mddev->gendisk, ios[rw]);
-	disk_stat_add(cpu, mddev->gendisk, sectors[rw], bio_sectors(bio));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
+		      bio_sectors(bio));
+	part_stat_unlock();
 
 	mp_bh->path = multipath_map(conf);
 	if (mp_bh->path < 0) {
diff --git a/drivers/md/raid0.c b/drivers/md/raid0.c
index e26030f..f52f442 100644
--- a/drivers/md/raid0.c
+++ b/drivers/md/raid0.c
@@ -406,10 +406,11 @@ static int raid0_make_request (struct request_queue *q, struct bio *bio)
 		return 0;
 	}
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, mddev->gendisk, ios[rw]);
-	disk_stat_add(cpu, mddev->gendisk, sectors[rw], bio_sectors(bio));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
+		      bio_sectors(bio));
+	part_stat_unlock();
 
 	chunk_size = mddev->chunk_size >> 10;
 	chunk_sects = mddev->chunk_size >> 9;
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index babb130..b976442 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -804,10 +804,11 @@ static int make_request(struct request_queue *q, struct bio * bio)
 
 	bitmap = mddev->bitmap;
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, mddev->gendisk, ios[rw]);
-	disk_stat_add(cpu, mddev->gendisk, sectors[rw], bio_sectors(bio));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
+		      bio_sectors(bio));
+	part_stat_unlock();
 
 	/*
 	 * make_request() can abort the operation when READA is being
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 5ec80da..5f99013 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -844,10 +844,11 @@ static int make_request(struct request_queue *q, struct bio * bio)
 	 */
 	wait_barrier(conf);
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, mddev->gendisk, ios[rw]);
-	disk_stat_add(cpu, mddev->gendisk, sectors[rw], bio_sectors(bio));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
+		      bio_sectors(bio));
+	part_stat_unlock();
 
 	r10_bio = mempool_alloc(conf->r10bio_pool, GFP_NOIO);
 
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 5899f21..ae16794 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -3396,10 +3396,11 @@ static int make_request(struct request_queue *q, struct bio * bi)
 
 	md_write_start(mddev, bi);
 
-	cpu = disk_stat_lock();
-	disk_stat_inc(cpu, mddev->gendisk, ios[rw]);
-	disk_stat_add(cpu, mddev->gendisk, sectors[rw], bio_sectors(bi));
-	disk_stat_unlock();
+	cpu = part_stat_lock();
+	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
+	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
+		      bio_sectors(bi));
+	part_stat_unlock();
 
 	if (rw == READ &&
 	     mddev->reshape_position == MaxSector &&
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 88fcf5a..787dcd1 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -210,15 +210,15 @@ ssize_t part_size_show(struct device *dev,
 	return sprintf(buf, "%llu\n",(unsigned long long)p->nr_sects);
 }
 
-static ssize_t part_stat_show(struct device *dev,
-			      struct device_attribute *attr, char *buf)
+ssize_t part_stat_show(struct device *dev,
+		       struct device_attribute *attr, char *buf)
 {
 	struct hd_struct *p = dev_to_part(dev);
 	int cpu;
 
-	cpu = disk_stat_lock();
+	cpu = part_stat_lock();
 	part_round_stats(cpu, p);
-	disk_stat_unlock();
+	part_stat_unlock();
 	return sprintf(buf,
 		"%8lu %8lu %8llu %8u "
 		"%8lu %8lu %8llu %8u "
@@ -575,8 +575,8 @@ void del_gendisk(struct gendisk *disk)
 	set_capacity(disk, 0);
 	disk->flags &= ~GENHD_FL_UP;
 	unlink_gendisk(disk);
-	disk_stat_set_all(disk, 0);
-	disk->stamp = 0;
+	part_stat_set_all(&disk->part0, 0);
+	disk->part0.stamp = 0;
 
 	kobject_put(disk->part0.holder_dir);
 	kobject_put(disk->slave_dir);
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 46d2a6f..e02b09d 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -155,13 +155,6 @@ struct gendisk {
 	struct timer_rand_state *random;
 
 	atomic_t sync_io;		/* RAID */
-	unsigned long stamp;
-	int in_flight;
-#ifdef	CONFIG_SMP
-	struct disk_stats *dkstats;
-#else
-	struct disk_stats dkstats;
-#endif
 	struct work_struct async_notify;
 #ifdef  CONFIG_BLK_DEV_INTEGRITY
 	struct blk_integrity *integrity;
@@ -242,46 +235,18 @@ extern struct hd_struct *disk_map_sector_rcu(struct gendisk *disk,
  * internal use only.
  */
 #ifdef	CONFIG_SMP
-#define disk_stat_lock()	({ rcu_read_lock(); get_cpu(); })
-#define disk_stat_unlock()	do { put_cpu(); rcu_read_unlock(); } while (0)
-
-#define disk_stat_add(cpu, gendiskp, field, addnd)			\
-	(per_cpu_ptr(gendiskp->dkstats, cpu)->field += addnd)
-
-#define disk_stat_read(gendiskp, field)					\
-({									\
-	typeof(gendiskp->dkstats->field) res = 0;			\
-	int i;								\
-	for_each_possible_cpu(i)					\
-		res += per_cpu_ptr(gendiskp->dkstats, i)->field;	\
-	res;								\
-})
-
-static inline void disk_stat_set_all(struct gendisk *gendiskp, int value)
-{
-	int i;
-
-	for_each_possible_cpu(i)
-		memset(per_cpu_ptr(gendiskp->dkstats, i), value,
-				sizeof(struct disk_stats));
-}		
+#define part_stat_lock()	({ rcu_read_lock(); get_cpu(); })
+#define part_stat_unlock()	do { put_cpu(); rcu_read_unlock(); } while (0)
 
-#define part_stat_add(cpu, part, field, addnd)				\
-	(per_cpu_ptr(part->dkstats, cpu)->field += addnd)
-
-#define all_stat_add(cpu, gendiskp, part, field, addnd, sector)		\
-({									\
-	if (part)							\
-		part_stat_add(cpu, part, field, addnd);			\
-	disk_stat_add(cpu, gendiskp, field, addnd);			\
-})
+#define __part_stat_add(cpu, part, field, addnd)			\
+	(per_cpu_ptr((part)->dkstats, (cpu))->field += (addnd))
 
 #define part_stat_read(part, field)					\
 ({									\
-	typeof(part->dkstats->field) res = 0;				\
+	typeof((part)->dkstats->field) res = 0;				\
 	int i;								\
 	for_each_possible_cpu(i)					\
-		res += per_cpu_ptr(part->dkstats, i)->field;		\
+		res += per_cpu_ptr((part)->dkstats, i)->field;		\
 	res;								\
 })
 
@@ -294,109 +259,73 @@ static inline void part_stat_set_all(struct hd_struct *part, int value)
 				sizeof(struct disk_stats));
 }
 
-#else /* !CONFIG_SMP */
-#define disk_stat_lock()	({ rcu_read_lock(); 0; })
-#define disk_stat_unlock()	rcu_read_unlock()
-
-#define disk_stat_add(cpu, gendiskp, field, addnd)			\
-	(gendiskp->dkstats.field += addnd)
-#define disk_stat_read(gendiskp, field)	(gendiskp->dkstats.field)
-
-static inline void disk_stat_set_all(struct gendisk *gendiskp, int value)
+static inline int init_part_stats(struct hd_struct *part)
 {
-	memset(&gendiskp->dkstats, value, sizeof (struct disk_stats));
+	part->dkstats = alloc_percpu(struct disk_stats);
+	if (!part->dkstats)
+		return 0;
+	return 1;
 }
 
-#define part_stat_add(cpu, part, field, addnd)				\
-	(part->dkstats.field += addnd)
-
-#define all_stat_add(cpu, gendiskp, part, field, addnd, sector)		\
-({									\
-	if (part)							\
-		part_stat_add(cpu, part, field, addnd);			\
-	disk_stat_add(cpu, gendiskp, field, addnd);			\
-})
-
-#define part_stat_read(part, field)	(part->dkstats.field)
-
-static inline void part_stat_set_all(struct hd_struct *part, int value)
+static inline void free_part_stats(struct hd_struct *part)
 {
-	memset(&part->dkstats, value, sizeof(struct disk_stats));
+	free_percpu(part->dkstats);
 }
 
-#endif /* CONFIG_SMP */
-
-#define disk_stat_dec(cpu, gendiskp, field)				\
-	disk_stat_add(cpu, gendiskp, field, -1)
-#define disk_stat_inc(cpu, gendiskp, field)				\
-	disk_stat_add(cpu, gendiskp, field, 1)
-#define disk_stat_sub(cpu, gendiskp, field, subnd)			\
-	disk_stat_add(cpu, gendiskp, field, -subnd)
-
-#define part_stat_dec(cpu, gendiskp, field)				\
-	part_stat_add(cpu, gendiskp, field, -1)
-#define part_stat_inc(cpu, gendiskp, field)				\
-	part_stat_add(cpu, gendiskp, field, 1)
-#define part_stat_sub(cpu, gendiskp, field, subnd)			\
-	part_stat_add(cpu, gendiskp, field, -subnd)
+#else /* !CONFIG_SMP */
+#define part_stat_lock()	({ rcu_read_lock(); 0; })
+#define part_stat_unlock()	rcu_read_unlock()
 
-#define all_stat_dec(cpu, gendiskp, field, sector)			\
-	all_stat_add(cpu, gendiskp, field, -1, sector)
-#define all_stat_inc(cpu, gendiskp, part, field, sector)		\
-	all_stat_add(cpu, gendiskp, part, field, 1, sector)
-#define all_stat_sub(cpu, gendiskp, part, field, subnd, sector)		\
-	all_stat_add(cpu, gendiskp, part, field, -subnd, sector)
+#define __part_stat_add(cpu, part, field, addnd)				\
+	((part)->dkstats.field += addnd)
 
-/* Inlines to alloc and free disk stats in struct gendisk */
-#ifdef  CONFIG_SMP
-static inline int init_disk_stats(struct gendisk *disk)
-{
-	disk->dkstats = alloc_percpu(struct disk_stats);
-	if (!disk->dkstats)
-		return 0;
-	return 1;
-}
+#define part_stat_read(part, field)	((part)->dkstats.field)
 
-static inline void free_disk_stats(struct gendisk *disk)
+static inline void part_stat_set_all(struct hd_struct *part, int value)
 {
-	free_percpu(disk->dkstats);
+	memset(&part->dkstats, value, sizeof(struct disk_stats));
 }
 
 static inline int init_part_stats(struct hd_struct *part)
 {
-	part->dkstats = alloc_percpu(struct disk_stats);
-	if (!part->dkstats)
-		return 0;
 	return 1;
 }
 
 static inline void free_part_stats(struct hd_struct *part)
 {
-	free_percpu(part->dkstats);
 }
 
-#else	/* CONFIG_SMP */
-static inline int init_disk_stats(struct gendisk *disk)
-{
-	return 1;
-}
+#endif /* CONFIG_SMP */
 
-static inline void free_disk_stats(struct gendisk *disk)
-{
-}
+#define part_stat_add(cpu, part, field, addnd)	do {			\
+	__part_stat_add((cpu), (part), field, addnd);			\
+	if ((part)->partno)						\
+		__part_stat_add((cpu), &part_to_disk((part))->part0,	\
+				field, addnd);				\
+} while (0)
 
-static inline int init_part_stats(struct hd_struct *part)
+#define part_stat_dec(cpu, gendiskp, field)				\
+	part_stat_add(cpu, gendiskp, field, -1)
+#define part_stat_inc(cpu, gendiskp, field)				\
+	part_stat_add(cpu, gendiskp, field, 1)
+#define part_stat_sub(cpu, gendiskp, field, subnd)			\
+	part_stat_add(cpu, gendiskp, field, -subnd)
+
+static inline void part_inc_in_flight(struct hd_struct *part)
 {
-	return 1;
+	part->in_flight++;
+	if (part->partno)
+		part_to_disk(part)->part0.in_flight++;
 }
 
-static inline void free_part_stats(struct hd_struct *part)
+static inline void part_dec_in_flight(struct hd_struct *part)
 {
+	part->in_flight--;
+	if (part->partno)
+		part_to_disk(part)->part0.in_flight--;
 }
-#endif	/* CONFIG_SMP */
 
 /* drivers/block/ll_rw_blk.c */
-extern void disk_round_stats(int cpu, struct gendisk *disk);
 extern void part_round_stats(int cpu, struct hd_struct *part);
 
 /* drivers/block/genhd.c */
@@ -605,6 +534,8 @@ extern void blk_unregister_region(dev_t devt, unsigned long range);
 
 extern ssize_t part_size_show(struct device *dev,
 			      struct device_attribute *attr, char *buf);
+extern ssize_t part_stat_show(struct device *dev,
+			      struct device_attribute *attr, char *buf);
 #ifdef CONFIG_FAIL_MAKE_REQUEST
 extern ssize_t part_fail_show(struct device *dev,
 			      struct device_attribute *attr, char *buf);
-- 
1.5.4.5

[PATCH 11/13] block: make partition array dynamic

[Tejun Heo]

disk->__part used to be statically allocated to the maximum possible
number of partitions.  This patch makes partition array allocation
dynamic.  The added overhead is minimal as only real change is one
memory dereference changed to RCU one.  This saves both a bit of
memory and cpu cycles iterating through unoccupied slots and makes
increasing partition limit easier.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c         |  129 +++++++++++++++++++++++++++++++++++++++++--------
 block/ioctl.c         |    2 +-
 fs/partitions/check.c |   31 ++++++++++--
 include/linux/genhd.h |   10 ++++-
 4 files changed, 145 insertions(+), 27 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index cf8d55e..c84a021 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -52,14 +52,21 @@ static struct device_type disk_type;
  */
 struct hd_struct *disk_get_part(struct gendisk *disk, int partno)
 {
-	struct hd_struct *part;
+	struct hd_struct *part = NULL;
+	struct disk_part_tbl *ptbl;
 
-	if (unlikely(partno < 0 || partno >= disk_max_parts(disk)))
+	if (unlikely(partno < 0))
 		return NULL;
+
 	rcu_read_lock();
-	part = rcu_dereference(disk->__part[partno]);
-	if (part)
-		get_device(part_to_dev(part));
+
+	ptbl = rcu_dereference(disk->part_tbl);
+	if (likely(partno < ptbl->len)) {
+		part = rcu_dereference(ptbl->part[partno]);
+		if (part)
+			get_device(part_to_dev(part));
+	}
+
 	rcu_read_unlock();
 
 	return part;
@@ -80,17 +87,24 @@ EXPORT_SYMBOL_GPL(disk_get_part);
 void disk_part_iter_init(struct disk_part_iter *piter, struct gendisk *disk,
 			  unsigned int flags)
 {
+	struct disk_part_tbl *ptbl;
+
+	rcu_read_lock();
+	ptbl = rcu_dereference(disk->part_tbl);
+
 	piter->disk = disk;
 	piter->part = NULL;
 
 	if (flags & DISK_PITER_REVERSE)
-		piter->idx = disk_max_parts(piter->disk) - 1;
+		piter->idx = ptbl->len - 1;
 	else if (flags & DISK_PITER_INCL_PART0)
 		piter->idx = 0;
 	else
 		piter->idx = 1;
 
 	piter->flags = flags;
+
+	rcu_read_unlock();
 }
 EXPORT_SYMBOL_GPL(disk_part_iter_init);
 
@@ -105,13 +119,16 @@ EXPORT_SYMBOL_GPL(disk_part_iter_init);
  */
 struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter)
 {
+	struct disk_part_tbl *ptbl;
 	int inc, end;
 
 	/* put the last partition */
 	disk_put_part(piter->part);
 	piter->part = NULL;
 
+	/* get part_tbl */
 	rcu_read_lock();
+	ptbl = rcu_dereference(piter->disk->part_tbl);
 
 	/* determine iteration parameters */
 	if (piter->flags & DISK_PITER_REVERSE) {
@@ -122,14 +139,14 @@ struct hd_struct *disk_part_iter_next(struct disk_part_iter *piter)
 			end = 0;
 	} else {
 		inc = 1;
-		end = disk_max_parts(piter->disk);
+		end = ptbl->len;
 	}
 
 	/* iterate to the next partition */
 	for (; piter->idx != end; piter->idx += inc) {
 		struct hd_struct *part;
 
-		part = rcu_dereference(piter->disk->__part[piter->idx]);
+		part = rcu_dereference(ptbl->part[piter->idx]);
 		if (!part)
 			continue;
 		if (!(piter->flags & DISK_PITER_INCL_EMPTY) && !part->nr_sects)
@@ -180,10 +197,13 @@ EXPORT_SYMBOL_GPL(disk_part_iter_exit);
  */
 struct hd_struct *disk_map_sector_rcu(struct gendisk *disk, sector_t sector)
 {
+	struct disk_part_tbl *ptbl;
 	int i;
 
-	for (i = 1; i < disk_max_parts(disk); i++) {
-		struct hd_struct *part = rcu_dereference(disk->__part[i]);
+	ptbl = rcu_dereference(disk->part_tbl);
+
+	for (i = 1; i < ptbl->len; i++) {
+		struct hd_struct *part = rcu_dereference(ptbl->part[i]);
 
 		if (part && part->start_sect <= sector &&
 		    sector < part->start_sect + part->nr_sects)
@@ -799,12 +819,86 @@ static struct attribute_group *disk_attr_groups[] = {
 	NULL
 };
 
+static void disk_free_ptbl_rcu_cb(struct rcu_head *head)
+{
+	struct disk_part_tbl *ptbl =
+		container_of(head, struct disk_part_tbl, rcu_head);
+
+	kfree(ptbl);
+}
+
+/**
+ * disk_replace_part_tbl - replace disk->part_tbl in RCU-safe way
+ * @disk: disk to replace part_tbl for
+ * @new_ptbl: new part_tbl to install
+ *
+ * Replace disk->part_tbl with @new_ptbl in RCU-safe way.  The
+ * original ptbl is freed using RCU callback.
+ *
+ * LOCKING:
+ * Matching bd_mutx locked.
+ */
+static void disk_replace_part_tbl(struct gendisk *disk,
+				  struct disk_part_tbl *new_ptbl)
+{
+	struct disk_part_tbl *old_ptbl = disk->part_tbl;
+
+	rcu_assign_pointer(disk->part_tbl, new_ptbl);
+	if (old_ptbl)
+		call_rcu(&old_ptbl->rcu_head, disk_free_ptbl_rcu_cb);
+}
+
+/**
+ * disk_expand_part_tbl - expand disk->part_tbl
+ * @disk: disk to expand part_tbl for
+ * @partno: expand such that this partno can fit in
+ *
+ * Expand disk->part_tbl such that @partno can fit in.  disk->part_tbl
+ * uses RCU to allow unlocked dereferencing for stats and other stuff.
+ *
+ * LOCKING:
+ * Matching bd_mutex locked, might sleep.
+ *
+ * RETURNS:
+ * 0 on success, -errno on failure.
+ */
+int disk_expand_part_tbl(struct gendisk *disk, int partno)
+{
+	struct disk_part_tbl *old_ptbl = disk->part_tbl;
+	struct disk_part_tbl *new_ptbl;
+	int len = old_ptbl ? old_ptbl->len : 0;
+	int target = partno + 1;
+	size_t size;
+	int i;
+
+	/* disk_max_parts() is zero during initialization, ignore if so */
+	if (disk_max_parts(disk) && target > disk_max_parts(disk))
+		return -EINVAL;
+
+	if (target <= len)
+		return 0;
+
+	size = sizeof(*new_ptbl) + target * sizeof(new_ptbl->part[0]);
+	new_ptbl = kzalloc_node(size, GFP_KERNEL, disk->node_id);
+	if (!new_ptbl)
+		return -ENOMEM;
+
+	INIT_RCU_HEAD(&new_ptbl->rcu_head);
+	new_ptbl->len = target;
+
+	for (i = 0; i < len; i++)
+		rcu_assign_pointer(new_ptbl->part[i], old_ptbl->part[i]);
+
+	disk_replace_part_tbl(disk, new_ptbl);
+	return 0;
+}
+
 static void disk_release(struct device *dev)
 {
 	struct gendisk *disk = dev_to_disk(dev);
 
 	kfree(disk->random);
-	kfree(disk->__part);
+	disk_replace_part_tbl(disk, NULL);
 	free_part_stats(&disk->part0);
 	kfree(disk);
 }
@@ -949,22 +1043,16 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 	disk = kmalloc_node(sizeof(struct gendisk),
 				GFP_KERNEL | __GFP_ZERO, node_id);
 	if (disk) {
-		int tot_minors = minors + ext_minors;
-		int size = tot_minors * sizeof(struct hd_struct *);
-
 		if (!init_part_stats(&disk->part0)) {
 			kfree(disk);
 			return NULL;
 		}
-
-		disk->__part = kmalloc_node(size, GFP_KERNEL | __GFP_ZERO,
-					    node_id);
-		if (!disk->__part) {
-				free_part_stats(&disk->part0);
+		if (disk_expand_part_tbl(disk, 0)) {
+			free_part_stats(&disk->part0);
 			kfree(disk);
 			return NULL;
 		}
-		disk->__part[0] = &disk->part0;
+		disk->part_tbl->part[0] = &disk->part0;
 
 		disk->minors = minors;
 		disk->ext_minors = ext_minors;
@@ -974,6 +1062,7 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 		device_initialize(disk_to_dev(disk));
 		INIT_WORK(&disk->async_notify,
 			media_change_notify_thread);
+		disk->node_id = node_id;
 	}
 	return disk;
 }
diff --git a/block/ioctl.c b/block/ioctl.c
index 64e7c67..38bee32 100644
--- a/block/ioctl.c
+++ b/block/ioctl.c
@@ -30,7 +30,7 @@ static int blkpg_ioctl(struct block_device *bdev, struct blkpg_ioctl_arg __user
 	if (bdev != bdev->bd_contains)
 		return -EINVAL;
 	partno = p.pno;
-	if (partno <= 0 || partno >= disk_max_parts(disk))
+	if (partno <= 0)
 		return -EINVAL;
 	switch (a.op) {
 		case BLKPG_ADD_PARTITION:
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 787dcd1..6b7ac10 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -312,14 +312,18 @@ static void delete_partition_rcu_cb(struct rcu_head *head)
 
 void delete_partition(struct gendisk *disk, int partno)
 {
+	struct disk_part_tbl *ptbl = disk->part_tbl;
 	struct hd_struct *part;
 
-	part = disk->__part[partno];
+	if (partno >= ptbl->len)
+		return;
+
+	part = ptbl->part[partno];
 	if (!part)
 		return;
 
 	blk_free_devt(part_devt(part));
-	rcu_assign_pointer(disk->__part[partno], NULL);
+	rcu_assign_pointer(ptbl->part[partno], NULL);
 	kobject_put(part->holder_dir);
 	device_del(part_to_dev(part));
 
@@ -341,10 +345,16 @@ int add_partition(struct gendisk *disk, int partno,
 	dev_t devt = MKDEV(0, 0);
 	struct device *ddev = disk_to_dev(disk);
 	struct device *pdev;
+	struct disk_part_tbl *ptbl;
 	const char *dname;
 	int err;
 
-	if (disk->__part[partno])
+	err = disk_expand_part_tbl(disk, partno);
+	if (err)
+		return err;
+	ptbl = disk->part_tbl;
+
+	if (ptbl->part[partno])
 		return -EBUSY;
 
 	p = kzalloc(sizeof(*p), GFP_KERNEL);
@@ -398,7 +408,7 @@ int add_partition(struct gendisk *disk, int partno,
 
 	/* everything is up and running, commence */
 	INIT_RCU_HEAD(&p->rcu_head);
-	rcu_assign_pointer(disk->__part[partno], p);
+	rcu_assign_pointer(ptbl->part[partno], p);
 
 	/* suppress uevent if the disk supresses it */
 	if (!ddev->uevent_suppress)
@@ -487,7 +497,7 @@ int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
 	struct disk_part_iter piter;
 	struct hd_struct *part;
 	struct parsed_partitions *state;
-	int p, res;
+	int p, highest, res;
 
 	if (bdev->bd_part_count)
 		return -EBUSY;
@@ -511,6 +521,17 @@ int rescan_partitions(struct gendisk *disk, struct block_device *bdev)
 	/* tell userspace that the media / partition table may have changed */
 	kobject_uevent(&disk_to_dev(disk)->kobj, KOBJ_CHANGE);
 
+	/* Detect the highest partition number and preallocate
+	 * disk->part_tbl.  This is an optimization and not strictly
+	 * necessary.
+	 */
+	for (p = 1, highest = 0; p < state->limit; p++)
+		if (state->parts[p].size)
+			highest = p;
+
+	disk_expand_part_tbl(disk, highest);
+
+	/* add partitions */
 	for (p = 1; p < state->limit; p++) {
 		sector_t size = state->parts[p].size;
 		sector_t from = state->parts[p].from;
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index e02b09d..bf87b77 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -122,6 +122,12 @@ struct blk_scsi_cmd_filter {
 	struct kobject kobj;
 };
 
+struct disk_part_tbl {
+	struct rcu_head rcu_head;
+	int len;
+	struct hd_struct *part[];
+};
+
 struct gendisk {
 	/* major, first_minor, minors and ext_minors are input
 	 * parameters only, don't use directly.  Use disk_devt() and
@@ -140,7 +146,7 @@ struct gendisk {
 	 * non-critical accesses use RCU.  Always access through
 	 * helpers.
 	 */
-	struct hd_struct **__part;
+	struct disk_part_tbl *part_tbl;
 	struct hd_struct part0;
 
 	struct block_device_operations *fops;
@@ -159,6 +165,7 @@ struct gendisk {
 #ifdef  CONFIG_BLK_DEV_INTEGRITY
 	struct blk_integrity *integrity;
 #endif
+	int node_id;
 };
 
 static inline struct gendisk *part_to_disk(struct hd_struct *part)
@@ -513,6 +520,7 @@ extern void blk_free_devt(dev_t devt);
 extern dev_t blk_lookup_devt(const char *name, int partno);
 extern char *disk_name (struct gendisk *hd, int partno, char *buf);
 
+extern int disk_expand_part_tbl(struct gendisk *disk, int target);
 extern int rescan_partitions(struct gendisk *disk, struct block_device *bdev);
 extern int __must_check add_partition(struct gendisk *, int, sector_t, sector_t, int);
 extern void delete_partition(struct gendisk *, int);
-- 
1.5.4.5

[PATCH 12/13] block: replace @ext_minors with GENHD_FL_EXT_DEVT

[Tejun Heo]

With previous changes, it's meaningless to limit the number of
partitions.  Replace @ext_minors with GENHD_FL_EXT_DEVT such that
setting the flag allows the disk to have maximum number of allowed
partitions (only limited by the number of entries in parsed_partitions
as determined by MAX_PART constant).

This kills not-too-pretty alloc_disk_ext[_node]() functions and makes
@minors parameter to alloc_disk[_node]() unnecessary.  The parameter
is left alone to avoid disturbing the users.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c          |   16 +---------------
 drivers/ide/ide-disk.c |   14 +++++---------
 drivers/scsi/sd.c      |    9 ++-------
 fs/partitions/check.h  |    4 +---
 include/linux/genhd.h  |   16 ++++++++--------
 5 files changed, 17 insertions(+), 42 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index c84a021..2bf0e67 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -1025,19 +1025,10 @@ struct gendisk *alloc_disk(int minors)
 {
 	return alloc_disk_node(minors, -1);
 }
+EXPORT_SYMBOL(alloc_disk);
 
 struct gendisk *alloc_disk_node(int minors, int node_id)
 {
-	return alloc_disk_ext_node(minors, 0, node_id);
-}
-
-struct gendisk *alloc_disk_ext(int minors, int ext_minors)
-{
-	return alloc_disk_ext_node(minors, ext_minors, -1);
-}
-
-struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
-{
 	struct gendisk *disk;
 
 	disk = kmalloc_node(sizeof(struct gendisk),
@@ -1055,7 +1046,6 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 		disk->part_tbl->part[0] = &disk->part0;
 
 		disk->minors = minors;
-		disk->ext_minors = ext_minors;
 		rand_initialize_disk(disk);
 		disk_to_dev(disk)->class = &block_class;
 		disk_to_dev(disk)->type = &disk_type;
@@ -1066,11 +1056,7 @@ struct gendisk *alloc_disk_ext_node(int minors, int ext_minors, int node_id)
 	}
 	return disk;
 }
-
-EXPORT_SYMBOL(alloc_disk);
 EXPORT_SYMBOL(alloc_disk_node);
-EXPORT_SYMBOL(alloc_disk_ext);
-EXPORT_SYMBOL(alloc_disk_ext_node);
 
 struct kobject *get_disk(struct gendisk *disk)
 {
diff --git a/drivers/ide/ide-disk.c b/drivers/ide/ide-disk.c
index b4cc1ce..f584ef8 100644
--- a/drivers/ide/ide-disk.c
+++ b/drivers/ide/ide-disk.c
@@ -41,16 +41,12 @@
 #include <asm/io.h>
 #include <asm/div64.h>
 
-#define IDE_DISK_PARTS		(1 << PARTN_BITS)
-
 #if !defined(CONFIG_DEBUG_BLOCK_EXT_DEVT)
-#define IDE_DISK_MINORS		IDE_DISK_PARTS
+#define IDE_DISK_MINORS		(1 << PARTN_BITS)
 #else
 #define IDE_DISK_MINORS		1
 #endif
 
-#define IDE_DISK_EXT_MINORS	(IDE_DISK_PARTS - IDE_DISK_MINORS)
-
 struct ide_disk_obj {
 	ide_drive_t	*drive;
 	ide_driver_t	*driver;
@@ -1175,8 +1171,7 @@ static int ide_disk_probe(ide_drive_t *drive)
 	if (!idkp)
 		goto failed;
 
-	g = alloc_disk_ext_node(IDE_DISK_MINORS, IDE_DISK_EXT_MINORS,
-				hwif_to_node(drive->hwif));
+	g = alloc_disk_node(IDE_DISK_MINORS, hwif_to_node(drive->hwif));
 	if (!g)
 		goto out_free_idkp;
 
@@ -1203,9 +1198,10 @@ static int ide_disk_probe(ide_drive_t *drive)
 		drive->attach = 1;
 
 	g->minors = IDE_DISK_MINORS;
-	g->ext_minors = IDE_DISK_EXT_MINORS;
 	g->driverfs_dev = &drive->gendev;
-	g->flags = drive->removable ? GENHD_FL_REMOVABLE : 0;
+	g->flags |= GENHD_FL_EXT_DEVT;
+	if (drive->removable)
+		g->flags |= GENHD_FL_REMOVABLE;
 	set_capacity(g, idedisk_capacity(drive));
 	g->fops = &idedisk_ops;
 	add_disk(g);
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 280d231..6598024 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -86,16 +86,12 @@ MODULE_ALIAS_SCSI_DEVICE(TYPE_DISK);
 MODULE_ALIAS_SCSI_DEVICE(TYPE_MOD);
 MODULE_ALIAS_SCSI_DEVICE(TYPE_RBC);
 
-#define SD_PARTS	64
-
 #if !defined(CONFIG_DEBUG_BLOCK_EXT_DEVT)
 #define SD_MINORS	16
 #else
 #define SD_MINORS	1
 #endif
 
-#define SD_EXT_MINORS	(SD_PARTS - SD_MINORS)
-
 static int  sd_revalidate_disk(struct gendisk *);
 static int  sd_probe(struct device *);
 static int  sd_remove(struct device *);
@@ -1811,7 +1807,7 @@ static int sd_probe(struct device *dev)
 	if (!sdkp)
 		goto out;
 
-	gd = alloc_disk_ext(SD_MINORS, SD_EXT_MINORS);
+	gd = alloc_disk(SD_MINORS);
 	if (!gd)
 		goto out_free;
 
@@ -1856,7 +1852,6 @@ static int sd_probe(struct device *dev)
 	gd->major = sd_major((index & 0xf0) >> 4);
 	gd->first_minor = ((index & 0xf) << 4) | (index & 0xfff00);
 	gd->minors = SD_MINORS;
-	gd->ext_minors = SD_EXT_MINORS;
 	gd->fops = &sd_fops;
 
 	if (index < 26) {
@@ -1880,7 +1875,7 @@ static int sd_probe(struct device *dev)
 	blk_queue_prep_rq(sdp->request_queue, sd_prep_fn);
 
 	gd->driverfs_dev = &sdp->sdev_gendev;
-	gd->flags = GENHD_FL_DRIVERFS;
+	gd->flags = GENHD_FL_EXT_DEVT | GENHD_FL_DRIVERFS;
 	if (sdp->removable)
 		gd->flags |= GENHD_FL_REMOVABLE;
 
diff --git a/fs/partitions/check.h b/fs/partitions/check.h
index 17ae8ec..98dbe1a 100644
--- a/fs/partitions/check.h
+++ b/fs/partitions/check.h
@@ -5,15 +5,13 @@
  * add_gd_partition adds a partitions details to the devices partition
  * description.
  */
-enum { MAX_PART = 256 };
-
 struct parsed_partitions {
 	char name[BDEVNAME_SIZE];
 	struct {
 		sector_t from;
 		sector_t size;
 		int flags;
-	} parts[MAX_PART];
+	} parts[DISK_MAX_PARTS];
 	int next;
 	int limit;
 };
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index bf87b77..4b5d762 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -58,6 +58,8 @@ enum {
 	UNIXWARE_PARTITION = 0x63,	/* Same as GNU_HURD and SCO Unix */
 };
 
+#define DISK_MAX_PARTS			256
+
 #include <linux/major.h>
 #include <linux/device.h>
 #include <linux/smp.h>
@@ -112,6 +114,7 @@ struct hd_struct {
 #define GENHD_FL_CD				8
 #define GENHD_FL_UP				16
 #define GENHD_FL_SUPPRESS_PARTITION_INFO	32
+#define GENHD_FL_EXT_DEVT			64 /* allow extended devt */
 
 #define BLK_SCSI_MAX_CMDS	(256)
 #define BLK_SCSI_CMD_PER_LONG	(BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
@@ -129,15 +132,13 @@ struct disk_part_tbl {
 };
 
 struct gendisk {
-	/* major, first_minor, minors and ext_minors are input
-	 * parameters only, don't use directly.  Use disk_devt() and
-	 * disk_max_parts().
+	/* major, first_minor and minors are input parameters only,
+	 * don't use directly.  Use disk_devt() and disk_max_parts().
 	 */
 	int major;			/* major number of driver */
 	int first_minor;
 	int minors;                     /* maximum number of minors, =1 for
                                          * disks that can't be partitioned. */
-	int ext_minors;			/* number of extended dynamic minors */
 
 	char disk_name[32];		/* name of major driver */
 
@@ -181,7 +182,9 @@ static inline struct gendisk *part_to_disk(struct hd_struct *part)
 
 static inline int disk_max_parts(struct gendisk *disk)
 {
-	return disk->minors + disk->ext_minors;
+	if (disk->flags & GENHD_FL_EXT_DEVT)
+		return DISK_MAX_PARTS;
+	return disk->minors;
 }
 
 static inline bool disk_partitionable(struct gendisk *disk)
@@ -528,9 +531,6 @@ extern void printk_all_partitions(void);
 
 extern struct gendisk *alloc_disk_node(int minors, int node_id);
 extern struct gendisk *alloc_disk(int minors);
-extern struct gendisk *alloc_disk_ext_node(int minors, int ext_minrs,
-					   int node_id);
-extern struct gendisk *alloc_disk_ext(int minors, int ext_minors);
 extern struct kobject *get_disk(struct gendisk *disk);
 extern void put_disk(struct gendisk *disk);
 extern void blk_register_region(dev_t devt, unsigned long range,
-- 
1.5.4.5

[PATCH 13/13] block: allow disk to have extended device number

[Tejun Heo]

Now that disk and partition handlings are mostly unified, it's easy to
allow disk to have extended device number.  This patch makes
add_disk() use extended device number if disk->minors is zero.  Both
sd and ide-disk are updated to use this.

* sd_format_disk_name() is implemented which can generically determine
  the drive name.  This removes disk number restriction stemming from
  limited device names.

* If sd index goes over SD_MAX_DISKS (which can be increased now BTW),
  sd simply doesn't initialize minors letting block layer choose
  extended device number.

* If CONFIG_DEBUG_EXT_DEVT is set, both sd and ide-disk always set
  minors to 0 and use extended device numbers.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/genhd.c          |   25 +++++++++++++++-
 drivers/ide/ide-disk.c |    2 +-
 drivers/scsi/sd.c      |   74 +++++++++++++++++++++++++++++++++++-------------
 fs/partitions/check.c  |    1 +
 include/linux/genhd.h  |    3 +-
 5 files changed, 82 insertions(+), 23 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index 2bf0e67..90a986f 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -478,14 +478,37 @@ static int exact_lock(dev_t devt, void *data)
  *
  * This function registers the partitioning information in @disk
  * with the kernel.
+ *
+ * FIXME: error handling
  */
 void add_disk(struct gendisk *disk)
 {
 	struct backing_dev_info *bdi;
+	dev_t devt;
 	int retval;
 
+	/* minors == 0 indicates to use ext devt from part0 and should
+	 * be accompanied with EXT_DEVT flag.  Make sure all
+	 * parameters make sense.
+	 */
+	WARN_ON(disk->minors && !(disk->major || disk->first_minor));
+	WARN_ON(!disk->minors && !(disk->flags & GENHD_FL_EXT_DEVT));
+
 	disk->flags |= GENHD_FL_UP;
-	disk_to_dev(disk)->devt = MKDEV(disk->major, disk->first_minor);
+
+	retval = blk_alloc_devt(&disk->part0, &devt);
+	if (retval) {
+		WARN_ON(1);
+		return;
+	}
+	disk_to_dev(disk)->devt = devt;
+
+	/* ->major and ->first_minor aren't supposed to be
+	 * dereferenced from here on, but set them just in case.
+	 */
+	disk->major = MAJOR(devt);
+	disk->first_minor = MINOR(devt);
+
 	blk_register_region(disk_devt(disk), disk->minors, NULL,
 			    exact_match, exact_lock, disk);
 	register_disk(disk);
diff --git a/drivers/ide/ide-disk.c b/drivers/ide/ide-disk.c
index f584ef8..e6776c4 100644
--- a/drivers/ide/ide-disk.c
+++ b/drivers/ide/ide-disk.c
@@ -44,7 +44,7 @@
 #if !defined(CONFIG_DEBUG_BLOCK_EXT_DEVT)
 #define IDE_DISK_MINORS		(1 << PARTN_BITS)
 #else
-#define IDE_DISK_MINORS		1
+#define IDE_DISK_MINORS		0
 #endif
 
 struct ide_disk_obj {
diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 6598024..bcb04b2 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -89,7 +89,7 @@ MODULE_ALIAS_SCSI_DEVICE(TYPE_RBC);
 #if !defined(CONFIG_DEBUG_BLOCK_EXT_DEVT)
 #define SD_MINORS	16
 #else
-#define SD_MINORS	1
+#define SD_MINORS	0
 #endif
 
 static int  sd_revalidate_disk(struct gendisk *);
@@ -1770,6 +1770,52 @@ static int sd_revalidate_disk(struct gendisk *disk)
 }
 
 /**
+ *	sd_format_disk_name - format disk name
+ *	@prefix: name prefix - ie. "sd" for SCSI disks
+ *	@index: index of the disk to format name for
+ *	@buf: output buffer
+ *	@buflen: length of the output buffer
+ *
+ *	SCSI disk names starts at sda.  The 26th device is sdz and the
+ *	27th is sdaa.  The last one for two lettered suffix is sdzz
+ *	which is followed by sdaaa.
+ *
+ *	This is basically 26 base counting with one extra 'nil' entry
+ *	at the beggining from the second digit on and can be
+ *	determined using similar method as 26 base conversion with the
+ *	index shifted -1 after each digit is computed.
+ *
+ *	CONTEXT:
+ *	Don't care.
+ *
+ *	RETURNS:
+ *	0 on success, -errno on failure.
+ */
+static int sd_format_disk_name(char *prefix, int index, char *buf, int buflen)
+{
+	const int base = 'z' - 'a' + 1;
+	char *begin = buf + strlen(prefix);
+	char *end = buf + buflen;
+	char *p;
+	int unit;
+
+	p = end - 1;
+	*p = '\0';
+	unit = base;
+	do {
+		if (p == begin)
+			return -EINVAL;
+		*--p = 'a' + (index % unit);
+		index = (index / unit) - 1;
+	} while (index >= 0);
+
+	memmove(begin, p, end - p);
+	memcpy(buf, prefix, strlen(prefix));
+
+	return 0;
+}
+
+/**
  *	sd_probe - called during driver initialization and whenever a
  *	new scsi device is attached to the system. It is called once
  *	for each scsi device (not just disks) present.
@@ -1821,8 +1867,8 @@ static int sd_probe(struct device *dev)
 	if (error)
 		goto out_put;
 
-	error = -EBUSY;
-	if (index >= SD_MAX_DISKS)
+	error = sd_format_disk_name("sd", index, gd->disk_name, DISK_NAME_LEN);
+	if (error)
 		goto out_free_index;
 
 	sdkp->device = sdp;
@@ -1849,24 +1895,12 @@ static int sd_probe(struct device *dev)
 
 	get_device(&sdp->sdev_gendev);
 
-	gd->major = sd_major((index & 0xf0) >> 4);
-	gd->first_minor = ((index & 0xf) << 4) | (index & 0xfff00);
-	gd->minors = SD_MINORS;
-	gd->fops = &sd_fops;
-
-	if (index < 26) {
-		sprintf(gd->disk_name, "sd%c", 'a' + index % 26);
-	} else if (index < (26 + 1) * 26) {
-		sprintf(gd->disk_name, "sd%c%c",
-			'a' + index / 26 - 1,'a' + index % 26);
-	} else {
-		const unsigned int m1 = (index / 26 - 1) / 26 - 1;
-		const unsigned int m2 = (index / 26 - 1) % 26;
-		const unsigned int m3 =  index % 26;
-		sprintf(gd->disk_name, "sd%c%c%c",
-			'a' + m1, 'a' + m2, 'a' + m3);
+	if (index < SD_MAX_DISKS) {
+		gd->major = sd_major((index & 0xf0) >> 4);
+		gd->first_minor = ((index & 0xf) << 4) | (index & 0xfff00);
+		gd->minors = SD_MINORS;
 	}
-
+	gd->fops = &sd_fops;
 	gd->private_data = &sdkp->driver;
 	gd->queue = sdkp->device->request_queue;
 
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 6b7ac10..e6f378f 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -593,6 +593,7 @@ void del_gendisk(struct gendisk *disk)
 	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;
 	unlink_gendisk(disk);
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 4b5d762..26868ca 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -59,6 +59,7 @@ enum {
 };
 
 #define DISK_MAX_PARTS			256
+#define DISK_NAME_LEN			32
 
 #include <linux/major.h>
 #include <linux/device.h>
@@ -140,7 +141,7 @@ struct gendisk {
 	int minors;                     /* maximum number of minors, =1 for
                                          * disks that can't be partitioned. */
 
-	char disk_name[32];		/* name of major driver */
+	char disk_name[DISK_NAME_LEN];	/* name of major driver */
 
 	/* Array of pointers to partitions indexed by partno.
 	 * Protected with matching bdev lock but stat and other
-- 
1.5.4.5

Re: [PATCHSET 3/3 blk-for-2.6.28] block: unify disk/part handling and improve ext devt, take #2

[Jens Axboe]

On Mon, Aug 25 2008, Tejun Heo wrote:
> Hello, again.
> 
> This is the second take of unify-disk-part-handling patchset.
> 
> Changes from the last take[L] are...
> 
> * adapted to blk-for-2.6.28.
> * devt alloc path simplification in 0013.

Applied all four, thanks Tejun.

-- 
Jens Axboe