From a5d16333612718569ffd26064270e535cb9c3928 Mon Sep 17 00:00:00 2001
From: Li Zefan <lizf@cn.fujitsu.com>
Date: Wed, 7 Dec 2011 20:08:40 -0500
Subject: Btrfs: check if the to-be-added device is writable

If we call ioctl(BTRFS_IOC_ADD_DEV) directly, we'll succeed in adding
a readonly device to a btrfs filesystem, and btrfs will write to
that device, emitting kernel errors:

[ 3109.833692] lost page write due to I/O error on loop2
[ 3109.833720] lost page write due to I/O error on loop2
...

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/volumes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index c37433d3cd82..0a8c8f8304b1 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1611,7 +1611,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 	if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding)
 		return -EINVAL;
 
-	bdev = blkdev_get_by_path(device_path, FMODE_EXCL,
+	bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL,
 				  root->fs_info->bdev_holder);
 	if (IS_ERR(bdev))
 		return PTR_ERR(bdev);
-- 
cgit 


From 5dbc8fca8ef5d719014f22345d990e957dcfc692 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 9 Dec 2011 11:07:37 -0500
Subject: Btrfs: fix btrfs_end_bio to deal with write errors to a single mirror

btrfs_end_bio checks the number of errors on a bio against the max
number of errors allowed before sending any EIOs up to the higher
levels.

If we got enough copies of the bio done for a given raid level, it is
supposed to clear the bio error flag and return success.

We have pointers to the original bio sent down by the higher layers and
pointers to any cloned bios we made for raid purposes.  If the original
bio happens to be the one that got an io error, but not the last one to
finish, it might not have the BIO_UPTODATE bit set.

Then, when the last bio does finish, we'll call bio_end_io on the
original bio.  It won't have the uptodate bit set and we'll end up
sending EIO to the higher layers.

We already had a check for this, it just was conditional on getting the
IO error on the very last bio.  Make the check unconditional so we eat
the EIOs properly.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/volumes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 0a8c8f8304b1..91ea57a1474a 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3258,7 +3258,7 @@ static void btrfs_end_bio(struct bio *bio, int err)
 		 */
 		if (atomic_read(&bbio->error) > bbio->max_errors) {
 			err = -EIO;
-		} else if (err) {
+		} else {
 			/*
 			 * this bio is actually up to date, we didn't
 			 * go over the max number of errors
-- 
cgit 


From d85c8a6f1bc083279215ff6e79b7c292bf3ec905 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Thu, 15 Dec 2011 15:38:41 -0500
Subject: Btrfs: unplug every once and a while

The btrfs io submission threads can build up massive plug lists.  This
keeps things more reasonable so we don't hand over huge dumps of IO at
once.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/volumes.c | 6 ++++++
 1 file changed, 6 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 91ea57a1474a..f4b839fd3c9d 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -295,6 +295,12 @@ loop_lock:
 			btrfs_requeue_work(&device->work);
 			goto done;
 		}
+		/* unplug every 64 requests just for good measure */
+		if (batch_run % 64 == 0) {
+			blk_finish_plug(&plug);
+			blk_start_plug(&plug);
+			sync_pending = 0;
+		}
 	}
 
 	cond_resched();
-- 
cgit 


From 21adbd5cbb5344a3fca6bb7ddb2ab6cb03c44546 Mon Sep 17 00:00:00 2001
From: Stefan Behrens <sbehrens@giantdisaster.de>
Date: Wed, 9 Nov 2011 13:44:05 +0100
Subject: Btrfs: integrate integrity check module into btrfs

This is the last part of the patch series. It modifies the btrfs
code to use the integrity check module if configured to do so
with the define BTRFS_FS_CHECK_INTEGRITY. If this define is not set,
the only effective change is that code is added that handles the
mount option to activate the integrity check. If the mount option is
set and the define BTRFS_FS_CHECK_INTEGRITY is not set, that code
complains in the log and the mount fails with EINVAL.

Add the mount option to activate the usage of the integrity check
code.
Add invocation of btrfs integrity check code init and cleanup
function on mount and umount, respectively.
Add hook to call btrfs integrity check code version of
submit_bh/submit_bio.

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
---
 fs/btrfs/volumes.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index f4b839fd3c9d..821334f6e3a1 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -32,6 +32,7 @@
 #include "print-tree.h"
 #include "volumes.h"
 #include "async-thread.h"
+#include "check-integrity.h"
 
 static int init_first_rw_device(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
@@ -246,7 +247,7 @@ loop_lock:
 			sync_pending = 0;
 		}
 
-		submit_bio(cur->bi_rw, cur);
+		btrfsic_submit_bio(cur->bi_rw, cur);
 		num_run++;
 		batch_run++;
 		if (need_resched())
@@ -3304,7 +3305,7 @@ static noinline int schedule_bio(struct btrfs_root *root,
 	/* don't bother with additional async steps for reads, right now */
 	if (!(rw & REQ_WRITE)) {
 		bio_get(bio);
-		submit_bio(rw, bio);
+		btrfsic_submit_bio(rw, bio);
 		bio_put(bio);
 		return 0;
 	}
@@ -3399,7 +3400,7 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 			if (async_submit)
 				schedule_bio(root, dev, rw, bio);
 			else
-				submit_bio(rw, bio);
+				btrfsic_submit_bio(rw, bio);
 		} else {
 			bio->bi_bdev = root->fs_info->fs_devices->latest_bdev;
 			bio->bi_sector = logical >> 9;
-- 
cgit 


From 1100373f8aa69e377386499350496e3d8565605f Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 6 Jan 2012 15:47:38 -0500
Subject: Btrfs: use bigger metadata chunks on bigger filesystems

The 256MB chunk is a little small on a huge FS.  This scales up the
chunk size.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/volumes.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index f4b839fd3c9d..ac00e3aa80a1 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2441,7 +2441,11 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 		max_stripe_size = 1024 * 1024 * 1024;
 		max_chunk_size = 10 * max_stripe_size;
 	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
-		max_stripe_size = 256 * 1024 * 1024;
+		/* for larger filesystems, use larger metadata chunks */
+		if (fs_devices->total_rw_bytes > 50ULL * 1024 * 1024 * 1024)
+			max_stripe_size = 1024 * 1024 * 1024;
+		else
+			max_stripe_size = 256 * 1024 * 1024;
 		max_chunk_size = max_stripe_size;
 	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
 		max_stripe_size = 8 * 1024 * 1024;
-- 
cgit 


From 10f6327b5d26660e06120ba17cff993cb616b1d2 Mon Sep 17 00:00:00 2001
From: Al Viro <viro@zeniv.linux.org.uk>
Date: Thu, 17 Nov 2011 15:05:22 -0500
Subject: btrfs: fix a deadlock in btrfs_scan_one_device()

pathname resolution under a global mutex, taken on some paths in ->mount()
is a Bad Idea(tm) - think what happens if said pathname resolution triggers
automount of some btrfs instance and walks into attempt to grab the same
mutex.  Deadlock - we are waiting for daemon to finish walking the path,
daemon is waiting for us to release the mutex...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
---
 fs/btrfs/volumes.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index f4b839fd3c9d..96bb3c0a79b0 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -706,8 +706,6 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
 	u64 devid;
 	u64 transid;
 
-	mutex_lock(&uuid_mutex);
-
 	flags |= FMODE_EXCL;
 	bdev = blkdev_get_by_path(path, flags, holder);
 
@@ -716,6 +714,7 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
 		goto error;
 	}
 
+	mutex_lock(&uuid_mutex);
 	ret = set_blocksize(bdev, 4096);
 	if (ret)
 		goto error_close;
@@ -737,9 +736,9 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
 
 	brelse(bh);
 error_close:
+	mutex_unlock(&uuid_mutex);
 	blkdev_put(bdev, flags);
 error:
-	mutex_unlock(&uuid_mutex);
 	return ret;
 }
 
-- 
cgit 


From 125ccb0ae6806dbec31abf4a85448971df3b4e39 Mon Sep 17 00:00:00 2001
From: Li Zefan <lizf@cn.fujitsu.com>
Date: Thu, 8 Dec 2011 15:07:24 +0800
Subject: Btrfs: don't pass a trans handle unnecessarily in volumes.c

Some functions never use the transaction handle passed to them.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
---
 fs/btrfs/volumes.c | 18 +++++++-----------
 1 file changed, 7 insertions(+), 11 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index f4b839fd3c9d..73f673c8d8d8 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -829,7 +829,6 @@ out:
 
 /*
  * find_free_dev_extent - find free space in the specified device
- * @trans:	transaction handler
  * @device:	the device which we search the free space in
  * @num_bytes:	the size of the free space that we need
  * @start:	store the start of the free space.
@@ -848,8 +847,7 @@ out:
  * But if we don't find suitable free space, it is used to store the size of
  * the max free space.
  */
-int find_free_dev_extent(struct btrfs_trans_handle *trans,
-			 struct btrfs_device *device, u64 num_bytes,
+int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
 			 u64 *start, u64 *len)
 {
 	struct btrfs_key key;
@@ -893,7 +891,7 @@ int find_free_dev_extent(struct btrfs_trans_handle *trans,
 	key.offset = search_start;
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
-	ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 	if (ret < 0)
 		goto out;
 	if (ret > 0) {
@@ -1469,8 +1467,7 @@ error_undo:
 /*
  * does all the dirty work required for changing file system's UUID.
  */
-static int btrfs_prepare_sprout(struct btrfs_trans_handle *trans,
-				struct btrfs_root *root)
+static int btrfs_prepare_sprout(struct btrfs_root *root)
 {
 	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
 	struct btrfs_fs_devices *old_devices;
@@ -1695,7 +1692,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 
 	if (seeding_dev) {
 		sb->s_flags &= ~MS_RDONLY;
-		ret = btrfs_prepare_sprout(trans, root);
+		ret = btrfs_prepare_sprout(root);
 		BUG_ON(ret);
 	}
 
@@ -2323,8 +2320,7 @@ done:
 	return ret;
 }
 
-static int btrfs_add_system_chunk(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root,
+static int btrfs_add_system_chunk(struct btrfs_root *root,
 			   struct btrfs_key *key,
 			   struct btrfs_chunk *chunk, int item_size)
 {
@@ -2496,7 +2492,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 		if (total_avail == 0)
 			continue;
 
-		ret = find_free_dev_extent(trans, device,
+		ret = find_free_dev_extent(device,
 					   max_stripe_size * dev_stripes,
 					   &dev_offset, &max_avail);
 		if (ret && ret != -ENOSPC)
@@ -2687,7 +2683,7 @@ static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
 	BUG_ON(ret);
 
 	if (map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
-		ret = btrfs_add_system_chunk(trans, chunk_root, &key, chunk,
+		ret = btrfs_add_system_chunk(chunk_root, &key, chunk,
 					     item_size);
 		BUG_ON(ret);
 	}
-- 
cgit 


From de11cc12df17337979e0929d2831887432f236ca Mon Sep 17 00:00:00 2001
From: Li Zefan <lizf@cn.fujitsu.com>
Date: Thu, 1 Dec 2011 12:55:47 +0800
Subject: Btrfs: don't pre-allocate btrfs bio

We pre-allocate a btrfs bio with fixed size, and then may re-allocate
memory if we find stripes are bigger than the fixed size. But this
pre-allocation is not necessary.

Also we don't have to calcuate the stripe number twice.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
---
 fs/btrfs/volumes.c | 67 +++++++++++++++++-------------------------------------
 1 file changed, 21 insertions(+), 46 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 73f673c8d8d8..540fdd25fb5e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2897,26 +2897,13 @@ static int __btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
 	u64 stripe_nr;
 	u64 stripe_nr_orig;
 	u64 stripe_nr_end;
-	int stripes_allocated = 8;
-	int stripes_required = 1;
 	int stripe_index;
 	int i;
+	int ret = 0;
 	int num_stripes;
 	int max_errors = 0;
 	struct btrfs_bio *bbio = NULL;
 
-	if (bbio_ret && !(rw & (REQ_WRITE | REQ_DISCARD)))
-		stripes_allocated = 1;
-again:
-	if (bbio_ret) {
-		bbio = kzalloc(btrfs_bio_size(stripes_allocated),
-				GFP_NOFS);
-		if (!bbio)
-			return -ENOMEM;
-
-		atomic_set(&bbio->error, 0);
-	}
-
 	read_lock(&em_tree->lock);
 	em = lookup_extent_mapping(em_tree, logical, *length);
 	read_unlock(&em_tree->lock);
@@ -2935,32 +2922,6 @@ again:
 	if (mirror_num > map->num_stripes)
 		mirror_num = 0;
 
-	/* if our btrfs_bio struct is too small, back off and try again */
-	if (rw & REQ_WRITE) {
-		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
-				 BTRFS_BLOCK_GROUP_DUP)) {
-			stripes_required = map->num_stripes;
-			max_errors = 1;
-		} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
-			stripes_required = map->sub_stripes;
-			max_errors = 1;
-		}
-	}
-	if (rw & REQ_DISCARD) {
-		if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
-				 BTRFS_BLOCK_GROUP_RAID1 |
-				 BTRFS_BLOCK_GROUP_DUP |
-				 BTRFS_BLOCK_GROUP_RAID10)) {
-			stripes_required = map->num_stripes;
-		}
-	}
-	if (bbio_ret && (rw & (REQ_WRITE | REQ_DISCARD)) &&
-	    stripes_allocated < stripes_required) {
-		stripes_allocated = map->num_stripes;
-		free_extent_map(em);
-		kfree(bbio);
-		goto again;
-	}
 	stripe_nr = offset;
 	/*
 	 * stripe_nr counts the total number of stripes we have to stride
@@ -3055,6 +3016,13 @@ again:
 	}
 	BUG_ON(stripe_index >= map->num_stripes);
 
+	bbio = kzalloc(btrfs_bio_size(num_stripes), GFP_NOFS);
+	if (!bbio) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	atomic_set(&bbio->error, 0);
+
 	if (rw & REQ_DISCARD) {
 		for (i = 0; i < num_stripes; i++) {
 			bbio->stripes[i].physical =
@@ -3151,15 +3119,22 @@ again:
 			stripe_index++;
 		}
 	}
-	if (bbio_ret) {
-		*bbio_ret = bbio;
-		bbio->num_stripes = num_stripes;
-		bbio->max_errors = max_errors;
-		bbio->mirror_num = mirror_num;
+
+	if (rw & REQ_WRITE) {
+		if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
+				 BTRFS_BLOCK_GROUP_RAID10 |
+				 BTRFS_BLOCK_GROUP_DUP)) {
+			max_errors = 1;
+		}
 	}
+
+	*bbio_ret = bbio;
+	bbio->num_stripes = num_stripes;
+	bbio->max_errors = max_errors;
+	bbio->mirror_num = mirror_num;
 out:
 	free_extent_map(em);
-	return 0;
+	return ret;
 }
 
 int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
-- 
cgit 


From ec9ef7a13be4dcce964c8503e8999087945e5b9e Mon Sep 17 00:00:00 2001
From: Li Zefan <lizf@cn.fujitsu.com>
Date: Thu, 1 Dec 2011 14:06:42 +0800
Subject: Btrfs: simplfy calculation of stripe length for discard operation

For btrfs raid, while discarding a range of space, we'll need to know
the start offset and length to discard for each device, and it's done
in btrfs_map_block().

However the calculation is a bit complex for raid0 and raid10, so I
reimplement it based on a fact that:

        dev1          dev2           dev3    (raid0)
        -----------------------------------
        s0 s3 s6      s1 s4 s7       s2 s5

Each device has (total_stripes / nr_dev) stripes, or plus one.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
---
 fs/btrfs/volumes.c | 95 ++++++++++++++++++------------------------------------
 1 file changed, 31 insertions(+), 64 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 540fdd25fb5e..563ef650850e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3024,80 +3024,47 @@ static int __btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
 	atomic_set(&bbio->error, 0);
 
 	if (rw & REQ_DISCARD) {
+		int factor = 0;
+		int sub_stripes = 0;
+		u64 stripes_per_dev = 0;
+		u32 remaining_stripes = 0;
+
+		if (map->type &
+		    (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10)) {
+			if (map->type & BTRFS_BLOCK_GROUP_RAID0)
+				sub_stripes = 1;
+			else
+				sub_stripes = map->sub_stripes;
+
+			factor = map->num_stripes / sub_stripes;
+			stripes_per_dev = div_u64_rem(stripe_nr_end -
+						      stripe_nr_orig,
+						      factor,
+						      &remaining_stripes);
+		}
+
 		for (i = 0; i < num_stripes; i++) {
 			bbio->stripes[i].physical =
 				map->stripes[stripe_index].physical +
 				stripe_offset + stripe_nr * map->stripe_len;
 			bbio->stripes[i].dev = map->stripes[stripe_index].dev;
 
-			if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
-				u64 stripes;
-				u32 last_stripe = 0;
-				int j;
-
-				div_u64_rem(stripe_nr_end - 1,
-					    map->num_stripes,
-					    &last_stripe);
-
-				for (j = 0; j < map->num_stripes; j++) {
-					u32 test;
-
-					div_u64_rem(stripe_nr_end - 1 - j,
-						    map->num_stripes, &test);
-					if (test == stripe_index)
-						break;
-				}
-				stripes = stripe_nr_end - 1 - j;
-				do_div(stripes, map->num_stripes);
-				bbio->stripes[i].length = map->stripe_len *
-					(stripes - stripe_nr + 1);
-
-				if (i == 0) {
+			if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+					 BTRFS_BLOCK_GROUP_RAID10)) {
+				bbio->stripes[i].length = stripes_per_dev *
+							  map->stripe_len;
+				if (i / sub_stripes < remaining_stripes)
+					bbio->stripes[i].length +=
+						map->stripe_len;
+				if (i < sub_stripes)
 					bbio->stripes[i].length -=
 						stripe_offset;
-					stripe_offset = 0;
-				}
-				if (stripe_index == last_stripe)
-					bbio->stripes[i].length -=
-						stripe_end_offset;
-			} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
-				u64 stripes;
-				int j;
-				int factor = map->num_stripes /
-					     map->sub_stripes;
-				u32 last_stripe = 0;
-
-				div_u64_rem(stripe_nr_end - 1,
-					    factor, &last_stripe);
-				last_stripe *= map->sub_stripes;
-
-				for (j = 0; j < factor; j++) {
-					u32 test;
-
-					div_u64_rem(stripe_nr_end - 1 - j,
-						    factor, &test);
-
-					if (test ==
-					    stripe_index / map->sub_stripes)
-						break;
-				}
-				stripes = stripe_nr_end - 1 - j;
-				do_div(stripes, factor);
-				bbio->stripes[i].length = map->stripe_len *
-					(stripes - stripe_nr + 1);
-
-				if (i < map->sub_stripes) {
-					bbio->stripes[i].length -=
-						stripe_offset;
-					if (i == map->sub_stripes - 1)
-						stripe_offset = 0;
-				}
-				if (stripe_index >= last_stripe &&
-				    stripe_index <= (last_stripe +
-						     map->sub_stripes - 1)) {
+				if ((i / sub_stripes + 1) %
+				    sub_stripes == remaining_stripes)
 					bbio->stripes[i].length -=
 						stripe_end_offset;
-				}
+				if (i == sub_stripes - 1)
+					stripe_offset = 0;
 			} else
 				bbio->stripes[i].length = *length;
 
-- 
cgit 


From b367e47fb3a70f5d24ebd6faf7d42436d485fb2d Mon Sep 17 00:00:00 2001
From: Li Zefan <lizf@cn.fujitsu.com>
Date: Wed, 7 Dec 2011 11:38:24 +0800
Subject: Btrfs: fix possible deadlock when opening a seed device

The correct lock order is uuid_mutex -> volume_mutex -> chunk_mutex,
but when we mount a filesystem which has backing seed devices, we have
this lock chain:

    open_ctree()
        lock(chunk_mutex);
        read_chunk_tree();
            read_one_dev();
                open_seed_devices();
                    lock(uuid_mutex);

and then we hit a lockdep splat.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
---
 fs/btrfs/volumes.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 563ef650850e..fbb493b28d5a 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3506,7 +3506,7 @@ static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
 	struct btrfs_fs_devices *fs_devices;
 	int ret;
 
-	mutex_lock(&uuid_mutex);
+	BUG_ON(!mutex_is_locked(&uuid_mutex));
 
 	fs_devices = root->fs_info->fs_devices->seed;
 	while (fs_devices) {
@@ -3544,7 +3544,6 @@ static int open_seed_devices(struct btrfs_root *root, u8 *fsid)
 	fs_devices->seed = root->fs_info->fs_devices->seed;
 	root->fs_info->fs_devices->seed = fs_devices;
 out:
-	mutex_unlock(&uuid_mutex);
 	return ret;
 }
 
@@ -3687,6 +3686,9 @@ int btrfs_read_chunk_tree(struct btrfs_root *root)
 	if (!path)
 		return -ENOMEM;
 
+	mutex_lock(&uuid_mutex);
+	lock_chunks(root);
+
 	/* first we search for all of the device items, and then we
 	 * read in all of the chunk items.  This way we can create chunk
 	 * mappings that reference all of the devices that are afound
@@ -3737,6 +3739,9 @@ again:
 	}
 	ret = 0;
 error:
+	unlock_chunks(root);
+	mutex_unlock(&uuid_mutex);
+
 	btrfs_free_path(path);
 	return ret;
 }
-- 
cgit 


From 6fef8df1dcb9b586268caff66df1d71ce8610132 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: get rid of *_alloc_profile fields

{data,metadata,system}_alloc_profile fields have been unused for a long
time now.  Get rid of them.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index f4b839fd3c9d..89096f6d6fb4 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2752,8 +2752,7 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 		return ret;
 
 	alloc_profile = BTRFS_BLOCK_GROUP_METADATA |
-			(fs_info->metadata_alloc_profile &
-			 fs_info->avail_metadata_alloc_bits);
+				fs_info->avail_metadata_alloc_bits;
 	alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile);
 
 	ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
@@ -2763,8 +2762,7 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 	sys_chunk_offset = chunk_offset + chunk_size;
 
 	alloc_profile = BTRFS_BLOCK_GROUP_SYSTEM |
-			(fs_info->system_alloc_profile &
-			 fs_info->avail_system_alloc_bits);
+				fs_info->avail_system_alloc_bits;
 	alloc_profile = btrfs_reduce_alloc_profile(root, alloc_profile);
 
 	ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map,
-- 
cgit 


From 52ba692972532f8d652080214b6599ece3dd51b9 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: introduce masks for chunk type and profile

Chunk's type and profile are encoded in u64 flags field.  Introduce
masks to easily access them.  Also fix the type of BTRFS_BLOCK_GROUP_*
constants, it should be ULL.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 89096f6d6fb4..d5fdee56777f 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2949,12 +2949,8 @@ again:
 		}
 	}
 	if (rw & REQ_DISCARD) {
-		if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
-				 BTRFS_BLOCK_GROUP_RAID1 |
-				 BTRFS_BLOCK_GROUP_DUP |
-				 BTRFS_BLOCK_GROUP_RAID10)) {
+		if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK)
 			stripes_required = map->num_stripes;
-		}
 	}
 	if (bbio_ret && (rw & (REQ_WRITE | REQ_DISCARD)) &&
 	    stripes_allocated < stripes_required) {
@@ -2978,10 +2974,7 @@ again:
 
 	if (rw & REQ_DISCARD)
 		*length = min_t(u64, em->len - offset, *length);
-	else if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
-			      BTRFS_BLOCK_GROUP_RAID1 |
-			      BTRFS_BLOCK_GROUP_RAID10 |
-			      BTRFS_BLOCK_GROUP_DUP)) {
+	else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
 		/* we limit the length of each bio to what fits in a stripe */
 		*length = min_t(u64, em->len - offset,
 				map->stripe_len - stripe_offset);
-- 
cgit 


From c9e9f97bdfb64d06e9520f8e4f37674ac21cc9bc Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: add basic restriper infrastructure

Add basic restriper infrastructure: extended balancing ioctl and all
related ioctl data structures, add data structure for tracking
restriper's state to fs_info, etc.  The semantics of the old balancing
ioctl are fully preserved.

Explicitly disallow any volume operations when balance is in progress.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 120 ++++++++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 96 insertions(+), 24 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index d5fdee56777f..9fc06e6bc51a 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1282,7 +1282,6 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
 	bool clear_super = false;
 
 	mutex_lock(&uuid_mutex);
-	mutex_lock(&root->fs_info->volume_mutex);
 
 	all_avail = root->fs_info->avail_data_alloc_bits |
 		root->fs_info->avail_system_alloc_bits |
@@ -1452,7 +1451,6 @@ error_close:
 	if (bdev)
 		blkdev_put(bdev, FMODE_READ | FMODE_EXCL);
 out:
-	mutex_unlock(&root->fs_info->volume_mutex);
 	mutex_unlock(&uuid_mutex);
 	return ret;
 error_undo:
@@ -1629,7 +1627,6 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 	}
 
 	filemap_write_and_wait(bdev->bd_inode->i_mapping);
-	mutex_lock(&root->fs_info->volume_mutex);
 
 	devices = &root->fs_info->fs_devices->devices;
 	/*
@@ -1757,8 +1754,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 		ret = btrfs_relocate_sys_chunks(root);
 		BUG_ON(ret);
 	}
-out:
-	mutex_unlock(&root->fs_info->volume_mutex);
+
 	return ret;
 error:
 	blkdev_put(bdev, FMODE_EXCL);
@@ -1766,7 +1762,7 @@ error:
 		mutex_unlock(&uuid_mutex);
 		up_write(&sb->s_umount);
 	}
-	goto out;
+	return ret;
 }
 
 static noinline int btrfs_update_device(struct btrfs_trans_handle *trans,
@@ -2077,6 +2073,35 @@ error:
 	return ret;
 }
 
+/*
+ * Should be called with both balance and volume mutexes held to
+ * serialize other volume operations (add_dev/rm_dev/resize) with
+ * restriper.  Same goes for unset_balance_control.
+ */
+static void set_balance_control(struct btrfs_balance_control *bctl)
+{
+	struct btrfs_fs_info *fs_info = bctl->fs_info;
+
+	BUG_ON(fs_info->balance_ctl);
+
+	spin_lock(&fs_info->balance_lock);
+	fs_info->balance_ctl = bctl;
+	spin_unlock(&fs_info->balance_lock);
+}
+
+static void unset_balance_control(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
+
+	BUG_ON(!fs_info->balance_ctl);
+
+	spin_lock(&fs_info->balance_lock);
+	fs_info->balance_ctl = NULL;
+	spin_unlock(&fs_info->balance_lock);
+
+	kfree(bctl);
+}
+
 static u64 div_factor(u64 num, int factor)
 {
 	if (factor == 10)
@@ -2086,29 +2111,23 @@ static u64 div_factor(u64 num, int factor)
 	return num;
 }
 
-int btrfs_balance(struct btrfs_root *dev_root)
+static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 {
-	int ret;
-	struct list_head *devices = &dev_root->fs_info->fs_devices->devices;
+	struct btrfs_root *chunk_root = fs_info->chunk_root;
+	struct btrfs_root *dev_root = fs_info->dev_root;
+	struct list_head *devices;
 	struct btrfs_device *device;
 	u64 old_size;
 	u64 size_to_free;
 	struct btrfs_path *path;
 	struct btrfs_key key;
-	struct btrfs_root *chunk_root = dev_root->fs_info->chunk_root;
-	struct btrfs_trans_handle *trans;
 	struct btrfs_key found_key;
-
-	if (dev_root->fs_info->sb->s_flags & MS_RDONLY)
-		return -EROFS;
-
-	if (!capable(CAP_SYS_ADMIN))
-		return -EPERM;
-
-	mutex_lock(&dev_root->fs_info->volume_mutex);
-	dev_root = dev_root->fs_info->dev_root;
+	struct btrfs_trans_handle *trans;
+	int ret;
+	int enospc_errors = 0;
 
 	/* step one make some room on all the devices */
+	devices = &fs_info->fs_devices->devices;
 	list_for_each_entry(device, devices, dev_list) {
 		old_size = device->total_bytes;
 		size_to_free = div_factor(old_size, 1);
@@ -2151,12 +2170,14 @@ int btrfs_balance(struct btrfs_root *dev_root)
 		 * failed
 		 */
 		if (ret == 0)
-			break;
+			BUG(); /* FIXME break ? */
 
 		ret = btrfs_previous_item(chunk_root, path, 0,
 					  BTRFS_CHUNK_ITEM_KEY);
-		if (ret)
+		if (ret) {
+			ret = 0;
 			break;
+		}
 
 		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
 				      path->slots[0]);
@@ -2174,12 +2195,63 @@ int btrfs_balance(struct btrfs_root *dev_root)
 					   found_key.offset);
 		if (ret && ret != -ENOSPC)
 			goto error;
+		if (ret == -ENOSPC)
+			enospc_errors++;
 		key.offset = found_key.offset - 1;
 	}
-	ret = 0;
+
 error:
 	btrfs_free_path(path);
-	mutex_unlock(&dev_root->fs_info->volume_mutex);
+	if (enospc_errors) {
+		printk(KERN_INFO "btrfs: %d enospc errors during balance\n",
+		       enospc_errors);
+		if (!ret)
+			ret = -ENOSPC;
+	}
+
+	return ret;
+}
+
+static void __cancel_balance(struct btrfs_fs_info *fs_info)
+{
+	unset_balance_control(fs_info);
+}
+
+void update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
+			       struct btrfs_ioctl_balance_args *bargs);
+
+/*
+ * Should be called with both balance and volume mutexes held
+ */
+int btrfs_balance(struct btrfs_balance_control *bctl,
+		  struct btrfs_ioctl_balance_args *bargs)
+{
+	struct btrfs_fs_info *fs_info = bctl->fs_info;
+	int ret;
+
+	if (btrfs_fs_closing(fs_info)) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	set_balance_control(bctl);
+
+	mutex_unlock(&fs_info->balance_mutex);
+
+	ret = __btrfs_balance(fs_info);
+
+	mutex_lock(&fs_info->balance_mutex);
+
+	if (bargs) {
+		memset(bargs, 0, sizeof(*bargs));
+		update_ioctl_balance_args(fs_info, bargs);
+	}
+
+	__cancel_balance(fs_info);
+
+	return ret;
+out:
+	kfree(bctl);
 	return ret;
 }
 
-- 
cgit 


From f43ffb60fd94e98be02780944e182ade6653b916 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: add basic infrastructure for selective balancing

This allows to have a separate set of filters for each chunk type
(data,meta,sys).  The code however is generic and switch on chunk type
is only done once.

This commit also adds a type filter: it allows to balance for example
meta and system chunks w/o touching data ones.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 59 ++++++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 57 insertions(+), 2 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 9fc06e6bc51a..91bbf6e774c0 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2102,6 +2102,30 @@ static void unset_balance_control(struct btrfs_fs_info *fs_info)
 	kfree(bctl);
 }
 
+static int should_balance_chunk(struct btrfs_root *root,
+				struct extent_buffer *leaf,
+				struct btrfs_chunk *chunk, u64 chunk_offset)
+{
+	struct btrfs_balance_control *bctl = root->fs_info->balance_ctl;
+	struct btrfs_balance_args *bargs = NULL;
+	u64 chunk_type = btrfs_chunk_type(leaf, chunk);
+
+	/* type filter */
+	if (!((chunk_type & BTRFS_BLOCK_GROUP_TYPE_MASK) &
+	      (bctl->flags & BTRFS_BALANCE_TYPE_MASK))) {
+		return 0;
+	}
+
+	if (chunk_type & BTRFS_BLOCK_GROUP_DATA)
+		bargs = &bctl->data;
+	else if (chunk_type & BTRFS_BLOCK_GROUP_SYSTEM)
+		bargs = &bctl->sys;
+	else if (chunk_type & BTRFS_BLOCK_GROUP_METADATA)
+		bargs = &bctl->meta;
+
+	return 1;
+}
+
 static u64 div_factor(u64 num, int factor)
 {
 	if (factor == 10)
@@ -2119,10 +2143,13 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 	struct btrfs_device *device;
 	u64 old_size;
 	u64 size_to_free;
+	struct btrfs_chunk *chunk;
 	struct btrfs_path *path;
 	struct btrfs_key key;
 	struct btrfs_key found_key;
 	struct btrfs_trans_handle *trans;
+	struct extent_buffer *leaf;
+	int slot;
 	int ret;
 	int enospc_errors = 0;
 
@@ -2179,8 +2206,10 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 			break;
 		}
 
-		btrfs_item_key_to_cpu(path->nodes[0], &found_key,
-				      path->slots[0]);
+		leaf = path->nodes[0];
+		slot = path->slots[0];
+		btrfs_item_key_to_cpu(leaf, &found_key, slot);
+
 		if (found_key.objectid != key.objectid)
 			break;
 
@@ -2188,7 +2217,14 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 		if (found_key.offset == 0)
 			break;
 
+		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
+
+		ret = should_balance_chunk(chunk_root, leaf, chunk,
+					   found_key.offset);
 		btrfs_release_path(path);
+		if (!ret)
+			goto loop;
+
 		ret = btrfs_relocate_chunk(chunk_root,
 					   chunk_root->root_key.objectid,
 					   found_key.objectid,
@@ -2197,6 +2233,7 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 			goto error;
 		if (ret == -ENOSPC)
 			enospc_errors++;
+loop:
 		key.offset = found_key.offset - 1;
 	}
 
@@ -2227,6 +2264,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		  struct btrfs_ioctl_balance_args *bargs)
 {
 	struct btrfs_fs_info *fs_info = bctl->fs_info;
+	u64 allowed;
 	int ret;
 
 	if (btrfs_fs_closing(fs_info)) {
@@ -2234,6 +2272,23 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		goto out;
 	}
 
+	/*
+	 * In case of mixed groups both data and meta should be picked,
+	 * and identical options should be given for both of them.
+	 */
+	allowed = btrfs_super_incompat_flags(fs_info->super_copy);
+	if ((allowed & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
+	    (bctl->flags & (BTRFS_BALANCE_DATA | BTRFS_BALANCE_METADATA))) {
+		if (!(bctl->flags & BTRFS_BALANCE_DATA) ||
+		    !(bctl->flags & BTRFS_BALANCE_METADATA) ||
+		    memcmp(&bctl->data, &bctl->meta, sizeof(bctl->data))) {
+			printk(KERN_ERR "btrfs: with mixed groups data and "
+			       "metadata balance options must be the same\n");
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+
 	set_balance_control(bctl);
 
 	mutex_unlock(&fs_info->balance_mutex);
-- 
cgit 


From ed25e9b26f898d8d63ae4a836489f1923534143b Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: profiles filter

Select chunks based on a given profile mask.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 91bbf6e774c0..447bd422d867 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2102,6 +2102,24 @@ static void unset_balance_control(struct btrfs_fs_info *fs_info)
 	kfree(bctl);
 }
 
+/*
+ * Balance filters.  Return 1 if chunk should be filtered out
+ * (should not be balanced).
+ */
+static int chunk_profiles_filter(u64 chunk_profile,
+				 struct btrfs_balance_args *bargs)
+{
+	chunk_profile &= BTRFS_BLOCK_GROUP_PROFILE_MASK;
+
+	if (chunk_profile == 0)
+		chunk_profile = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+
+	if (bargs->profiles & chunk_profile)
+		return 0;
+
+	return 1;
+}
+
 static int should_balance_chunk(struct btrfs_root *root,
 				struct extent_buffer *leaf,
 				struct btrfs_chunk *chunk, u64 chunk_offset)
@@ -2123,6 +2141,12 @@ static int should_balance_chunk(struct btrfs_root *root,
 	else if (chunk_type & BTRFS_BLOCK_GROUP_METADATA)
 		bargs = &bctl->meta;
 
+	/* profiles filter */
+	if ((bargs->flags & BTRFS_BALANCE_ARGS_PROFILES) &&
+	    chunk_profiles_filter(chunk_type, bargs)) {
+		return 0;
+	}
+
 	return 1;
 }
 
-- 
cgit 


From 5ce5b3c0916ba3a2e34cf648b94044adc5ef9e76 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: usage filter

Select chunks that are less than X percent full.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 447bd422d867..b858242374db 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2120,6 +2120,36 @@ static int chunk_profiles_filter(u64 chunk_profile,
 	return 1;
 }
 
+static u64 div_factor_fine(u64 num, int factor)
+{
+	if (factor <= 0)
+		return 0;
+	if (factor >= 100)
+		return num;
+
+	num *= factor;
+	do_div(num, 100);
+	return num;
+}
+
+static int chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
+			      struct btrfs_balance_args *bargs)
+{
+	struct btrfs_block_group_cache *cache;
+	u64 chunk_used, user_thresh;
+	int ret = 1;
+
+	cache = btrfs_lookup_block_group(fs_info, chunk_offset);
+	chunk_used = btrfs_block_group_used(&cache->item);
+
+	user_thresh = div_factor_fine(cache->key.offset, bargs->usage);
+	if (chunk_used < user_thresh)
+		ret = 0;
+
+	btrfs_put_block_group(cache);
+	return ret;
+}
+
 static int should_balance_chunk(struct btrfs_root *root,
 				struct extent_buffer *leaf,
 				struct btrfs_chunk *chunk, u64 chunk_offset)
@@ -2147,6 +2177,12 @@ static int should_balance_chunk(struct btrfs_root *root,
 		return 0;
 	}
 
+	/* usage filter */
+	if ((bargs->flags & BTRFS_BALANCE_ARGS_USAGE) &&
+	    chunk_usage_filter(bctl->fs_info, chunk_offset, bargs)) {
+		return 0;
+	}
+
 	return 1;
 }
 
-- 
cgit 


From 409d404b461afa9738619f249fd7f62a366b68c2 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:47 +0200
Subject: Btrfs: devid filter

Relocate chunks which have at least one stripe located on a device with
devid X.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index b858242374db..9ff5cd09a256 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2150,6 +2150,23 @@ static int chunk_usage_filter(struct btrfs_fs_info *fs_info, u64 chunk_offset,
 	return ret;
 }
 
+static int chunk_devid_filter(struct extent_buffer *leaf,
+			      struct btrfs_chunk *chunk,
+			      struct btrfs_balance_args *bargs)
+{
+	struct btrfs_stripe *stripe;
+	int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
+	int i;
+
+	for (i = 0; i < num_stripes; i++) {
+		stripe = btrfs_stripe_nr(chunk, i);
+		if (btrfs_stripe_devid(leaf, stripe) == bargs->devid)
+			return 0;
+	}
+
+	return 1;
+}
+
 static int should_balance_chunk(struct btrfs_root *root,
 				struct extent_buffer *leaf,
 				struct btrfs_chunk *chunk, u64 chunk_offset)
@@ -2183,6 +2200,12 @@ static int should_balance_chunk(struct btrfs_root *root,
 		return 0;
 	}
 
+	/* devid filter */
+	if ((bargs->flags & BTRFS_BALANCE_ARGS_DEVID) &&
+	    chunk_devid_filter(leaf, chunk, bargs)) {
+		return 0;
+	}
+
 	return 1;
 }
 
-- 
cgit 


From 94e60d5a5c4b98a32b1077dec88df09ada712376 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: devid subset filter

Select chunks which have at least one byte of at least one stripe
located on a device with devid X in a given [pstart,pend) physical
address range.

This filter only works when devid filter is turned on.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 46 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 9ff5cd09a256..c60071a448a5 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2167,6 +2167,46 @@ static int chunk_devid_filter(struct extent_buffer *leaf,
 	return 1;
 }
 
+/* [pstart, pend) */
+static int chunk_drange_filter(struct extent_buffer *leaf,
+			       struct btrfs_chunk *chunk,
+			       u64 chunk_offset,
+			       struct btrfs_balance_args *bargs)
+{
+	struct btrfs_stripe *stripe;
+	int num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
+	u64 stripe_offset;
+	u64 stripe_length;
+	int factor;
+	int i;
+
+	if (!(bargs->flags & BTRFS_BALANCE_ARGS_DEVID))
+		return 0;
+
+	if (btrfs_chunk_type(leaf, chunk) & (BTRFS_BLOCK_GROUP_DUP |
+	     BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_RAID10))
+		factor = 2;
+	else
+		factor = 1;
+	factor = num_stripes / factor;
+
+	for (i = 0; i < num_stripes; i++) {
+		stripe = btrfs_stripe_nr(chunk, i);
+		if (btrfs_stripe_devid(leaf, stripe) != bargs->devid)
+			continue;
+
+		stripe_offset = btrfs_stripe_offset(leaf, stripe);
+		stripe_length = btrfs_chunk_length(leaf, chunk);
+		do_div(stripe_length, factor);
+
+		if (stripe_offset < bargs->pend &&
+		    stripe_offset + stripe_length > bargs->pstart)
+			return 0;
+	}
+
+	return 1;
+}
+
 static int should_balance_chunk(struct btrfs_root *root,
 				struct extent_buffer *leaf,
 				struct btrfs_chunk *chunk, u64 chunk_offset)
@@ -2206,6 +2246,12 @@ static int should_balance_chunk(struct btrfs_root *root,
 		return 0;
 	}
 
+	/* drange filter, makes sense only with devid filter */
+	if ((bargs->flags & BTRFS_BALANCE_ARGS_DRANGE) &&
+	    chunk_drange_filter(leaf, chunk, chunk_offset, bargs)) {
+		return 0;
+	}
+
 	return 1;
 }
 
-- 
cgit 


From ea67176ae8c024f64d85ec33873e5eadf1af7247 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: virtual address space subset filter

Select chunks which have at least one byte located inside a given
[vstart, vend) virtual address space range.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index c60071a448a5..e86c9e4fe51e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2207,6 +2207,20 @@ static int chunk_drange_filter(struct extent_buffer *leaf,
 	return 1;
 }
 
+/* [vstart, vend) */
+static int chunk_vrange_filter(struct extent_buffer *leaf,
+			       struct btrfs_chunk *chunk,
+			       u64 chunk_offset,
+			       struct btrfs_balance_args *bargs)
+{
+	if (chunk_offset < bargs->vend &&
+	    chunk_offset + btrfs_chunk_length(leaf, chunk) > bargs->vstart)
+		/* at least part of the chunk is inside this vrange */
+		return 0;
+
+	return 1;
+}
+
 static int should_balance_chunk(struct btrfs_root *root,
 				struct extent_buffer *leaf,
 				struct btrfs_chunk *chunk, u64 chunk_offset)
@@ -2252,6 +2266,12 @@ static int should_balance_chunk(struct btrfs_root *root,
 		return 0;
 	}
 
+	/* vrange filter */
+	if ((bargs->flags & BTRFS_BALANCE_ARGS_VRANGE) &&
+	    chunk_vrange_filter(leaf, chunk, chunk_offset, bargs)) {
+		return 0;
+	}
+
 	return 1;
 }
 
-- 
cgit 


From e4d8ec0f65b91bfb4984a4927632ded95f9825ad Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: implement online profile changing

Profile changing is done by launching a balance with
BTRFS_BALANCE_CONVERT bits set and target fields of respective
btrfs_balance_args structs initialized.  Profile reducing code in this
case will pick restriper's target profile if it's available instead of
doing a blind reduce.  If target profile is not yet available it goes
back to a plain reduce.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index e86c9e4fe51e..f08210e83339 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2438,6 +2438,75 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		}
 	}
 
+	/*
+	 * Profile changing sanity checks.  Skip them if a simple
+	 * balance is requested.
+	 */
+	if (!((bctl->data.flags | bctl->sys.flags | bctl->meta.flags) &
+	      BTRFS_BALANCE_ARGS_CONVERT))
+		goto do_balance;
+
+	allowed = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+	if (fs_info->fs_devices->num_devices == 1)
+		allowed |= BTRFS_BLOCK_GROUP_DUP;
+	else if (fs_info->fs_devices->num_devices < 4)
+		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1);
+	else
+		allowed |= (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
+				BTRFS_BLOCK_GROUP_RAID10);
+
+	if (!profile_is_valid(bctl->data.target, 1) ||
+	    bctl->data.target & ~allowed) {
+		printk(KERN_ERR "btrfs: unable to start balance with target "
+		       "data profile %llu\n",
+		       (unsigned long long)bctl->data.target);
+		ret = -EINVAL;
+		goto out;
+	}
+	if (!profile_is_valid(bctl->meta.target, 1) ||
+	    bctl->meta.target & ~allowed) {
+		printk(KERN_ERR "btrfs: unable to start balance with target "
+		       "metadata profile %llu\n",
+		       (unsigned long long)bctl->meta.target);
+		ret = -EINVAL;
+		goto out;
+	}
+	if (!profile_is_valid(bctl->sys.target, 1) ||
+	    bctl->sys.target & ~allowed) {
+		printk(KERN_ERR "btrfs: unable to start balance with target "
+		       "system profile %llu\n",
+		       (unsigned long long)bctl->sys.target);
+		ret = -EINVAL;
+		goto out;
+	}
+
+	if (bctl->data.target & BTRFS_BLOCK_GROUP_DUP) {
+		printk(KERN_ERR "btrfs: dup for data is not allowed\n");
+		ret = -EINVAL;
+		goto out;
+	}
+
+	/* allow to reduce meta or sys integrity only if force set */
+	allowed = BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1 |
+			BTRFS_BLOCK_GROUP_RAID10;
+	if (((bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
+	     (fs_info->avail_system_alloc_bits & allowed) &&
+	     !(bctl->sys.target & allowed)) ||
+	    ((bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) &&
+	     (fs_info->avail_metadata_alloc_bits & allowed) &&
+	     !(bctl->meta.target & allowed))) {
+		if (bctl->flags & BTRFS_BALANCE_FORCE) {
+			printk(KERN_INFO "btrfs: force reducing metadata "
+			       "integrity\n");
+		} else {
+			printk(KERN_ERR "btrfs: balance will reduce metadata "
+			       "integrity, use force if you want this\n");
+			ret = -EINVAL;
+			goto out;
+		}
+	}
+
+do_balance:
 	set_balance_control(bctl);
 
 	mutex_unlock(&fs_info->balance_mutex);
-- 
cgit 


From cfa4c961cc69ffb7bda450972320a25cbd413e19 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: soft profile changing mode (aka soft convert)

When doing convert from one profile to another if soft mode is on
restriper won't touch chunks that already have the profile we are
converting to.  This is useful if e.g. half of the FS was converted
earlier.

The soft mode switch is (like every other filter) per-type.  This means
that we can convert for example meta chunks the "hard" way while
converting data chunks selectively with soft switch.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index f08210e83339..98b4067017ff 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2221,6 +2221,23 @@ static int chunk_vrange_filter(struct extent_buffer *leaf,
 	return 1;
 }
 
+static int chunk_soft_convert_filter(u64 chunk_profile,
+				     struct btrfs_balance_args *bargs)
+{
+	if (!(bargs->flags & BTRFS_BALANCE_ARGS_CONVERT))
+		return 0;
+
+	chunk_profile &= BTRFS_BLOCK_GROUP_PROFILE_MASK;
+
+	if (chunk_profile == 0)
+		chunk_profile = BTRFS_AVAIL_ALLOC_BIT_SINGLE;
+
+	if (bargs->target & chunk_profile)
+		return 1;
+
+	return 0;
+}
+
 static int should_balance_chunk(struct btrfs_root *root,
 				struct extent_buffer *leaf,
 				struct btrfs_chunk *chunk, u64 chunk_offset)
@@ -2272,6 +2289,12 @@ static int should_balance_chunk(struct btrfs_root *root,
 		return 0;
 	}
 
+	/* soft profile changing mode */
+	if ((bargs->flags & BTRFS_BALANCE_ARGS_SOFT) &&
+	    chunk_soft_convert_filter(chunk_type, bargs)) {
+		return 0;
+	}
+
 	return 1;
 }
 
-- 
cgit 


From 0940ebf6b92ea10a6f30ae5ac3993a3b75745da6 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: save balance parameters to disk

Introduce a new btree objectid for storing balance item.  The reason is
to be able to resume restriper after a crash with the same parameters.
Balance item has a very high objectid and goes into tree of tree roots.

The key for the new item is as follows:

	[ BTRFS_BALANCE_OBJECTID ; BTRFS_BALANCE_ITEM_KEY ; 0 ]

Older kernels simply ignore it so it's safe to mount with an older
kernel and then go back to the newer one.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 99 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 99 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 98b4067017ff..4c60ca0ae90b 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2073,6 +2073,97 @@ error:
 	return ret;
 }
 
+static int insert_balance_item(struct btrfs_root *root,
+			       struct btrfs_balance_control *bctl)
+{
+	struct btrfs_trans_handle *trans;
+	struct btrfs_balance_item *item;
+	struct btrfs_disk_balance_args disk_bargs;
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_key key;
+	int ret, err;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	trans = btrfs_start_transaction(root, 0);
+	if (IS_ERR(trans)) {
+		btrfs_free_path(path);
+		return PTR_ERR(trans);
+	}
+
+	key.objectid = BTRFS_BALANCE_OBJECTID;
+	key.type = BTRFS_BALANCE_ITEM_KEY;
+	key.offset = 0;
+
+	ret = btrfs_insert_empty_item(trans, root, path, &key,
+				      sizeof(*item));
+	if (ret)
+		goto out;
+
+	leaf = path->nodes[0];
+	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_balance_item);
+
+	memset_extent_buffer(leaf, 0, (unsigned long)item, sizeof(*item));
+
+	btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->data);
+	btrfs_set_balance_data(leaf, item, &disk_bargs);
+	btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->meta);
+	btrfs_set_balance_meta(leaf, item, &disk_bargs);
+	btrfs_cpu_balance_args_to_disk(&disk_bargs, &bctl->sys);
+	btrfs_set_balance_sys(leaf, item, &disk_bargs);
+
+	btrfs_set_balance_flags(leaf, item, bctl->flags);
+
+	btrfs_mark_buffer_dirty(leaf);
+out:
+	btrfs_free_path(path);
+	err = btrfs_commit_transaction(trans, root);
+	if (err && !ret)
+		ret = err;
+	return ret;
+}
+
+static int del_balance_item(struct btrfs_root *root)
+{
+	struct btrfs_trans_handle *trans;
+	struct btrfs_path *path;
+	struct btrfs_key key;
+	int ret, err;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	trans = btrfs_start_transaction(root, 0);
+	if (IS_ERR(trans)) {
+		btrfs_free_path(path);
+		return PTR_ERR(trans);
+	}
+
+	key.objectid = BTRFS_BALANCE_OBJECTID;
+	key.type = BTRFS_BALANCE_ITEM_KEY;
+	key.offset = 0;
+
+	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+	if (ret < 0)
+		goto out;
+	if (ret > 0) {
+		ret = -ENOENT;
+		goto out;
+	}
+
+	ret = btrfs_del_item(trans, root, path);
+out:
+	btrfs_free_path(path);
+	err = btrfs_commit_transaction(trans, root);
+	if (err && !ret)
+		ret = err;
+	return ret;
+}
+
 /*
  * Should be called with both balance and volume mutexes held to
  * serialize other volume operations (add_dev/rm_dev/resize) with
@@ -2423,7 +2514,11 @@ error:
 
 static void __cancel_balance(struct btrfs_fs_info *fs_info)
 {
+	int ret;
+
 	unset_balance_control(fs_info);
+	ret = del_balance_item(fs_info->tree_root);
+	BUG_ON(ret);
 }
 
 void update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
@@ -2530,6 +2625,10 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 	}
 
 do_balance:
+	ret = insert_balance_item(fs_info->tree_root, bctl);
+	if (ret)
+		goto out;
+
 	set_balance_control(bctl);
 
 	mutex_unlock(&fs_info->balance_mutex);
-- 
cgit 


From 596410151ed71819b9e8a8018c6c9992796b256d Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: recover balance on mount

On mount, if balance item is found, resume balance in a separate
kernel thread.

Try to be smart to continue roughly where previous balance (or convert)
was interrupted.  For chunk types that were being converted to some
profile we turn on soft convert, in case of a simple balance we turn on
usage filter and relocate only less-than-90%-full chunks of that type.
These are just heuristics but they help quite a bit, and can be improved
in future.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 135 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 133 insertions(+), 2 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 4c60ca0ae90b..17e565388de0 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -23,6 +23,7 @@
 #include <linux/random.h>
 #include <linux/iocontext.h>
 #include <linux/capability.h>
+#include <linux/kthread.h>
 #include <asm/div64.h>
 #include "compat.h"
 #include "ctree.h"
@@ -2164,6 +2165,46 @@ out:
 	return ret;
 }
 
+/*
+ * This is a heuristic used to reduce the number of chunks balanced on
+ * resume after balance was interrupted.
+ */
+static void update_balance_args(struct btrfs_balance_control *bctl)
+{
+	/*
+	 * Turn on soft mode for chunk types that were being converted.
+	 */
+	if (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT)
+		bctl->data.flags |= BTRFS_BALANCE_ARGS_SOFT;
+	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT)
+		bctl->sys.flags |= BTRFS_BALANCE_ARGS_SOFT;
+	if (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT)
+		bctl->meta.flags |= BTRFS_BALANCE_ARGS_SOFT;
+
+	/*
+	 * Turn on usage filter if is not already used.  The idea is
+	 * that chunks that we have already balanced should be
+	 * reasonably full.  Don't do it for chunks that are being
+	 * converted - that will keep us from relocating unconverted
+	 * (albeit full) chunks.
+	 */
+	if (!(bctl->data.flags & BTRFS_BALANCE_ARGS_USAGE) &&
+	    !(bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
+		bctl->data.flags |= BTRFS_BALANCE_ARGS_USAGE;
+		bctl->data.usage = 90;
+	}
+	if (!(bctl->sys.flags & BTRFS_BALANCE_ARGS_USAGE) &&
+	    !(bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
+		bctl->sys.flags |= BTRFS_BALANCE_ARGS_USAGE;
+		bctl->sys.usage = 90;
+	}
+	if (!(bctl->meta.flags & BTRFS_BALANCE_ARGS_USAGE) &&
+	    !(bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT)) {
+		bctl->meta.flags |= BTRFS_BALANCE_ARGS_USAGE;
+		bctl->meta.usage = 90;
+	}
+}
+
 /*
  * Should be called with both balance and volume mutexes held to
  * serialize other volume operations (add_dev/rm_dev/resize) with
@@ -2626,10 +2667,18 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 
 do_balance:
 	ret = insert_balance_item(fs_info->tree_root, bctl);
-	if (ret)
+	if (ret && ret != -EEXIST)
 		goto out;
 
-	set_balance_control(bctl);
+	if (!(bctl->flags & BTRFS_BALANCE_RESUME)) {
+		BUG_ON(ret == -EEXIST);
+		set_balance_control(bctl);
+	} else {
+		BUG_ON(ret != -EEXIST);
+		spin_lock(&fs_info->balance_lock);
+		update_balance_args(bctl);
+		spin_unlock(&fs_info->balance_lock);
+	}
 
 	mutex_unlock(&fs_info->balance_mutex);
 
@@ -2646,7 +2695,89 @@ do_balance:
 
 	return ret;
 out:
+	if (bctl->flags & BTRFS_BALANCE_RESUME)
+		__cancel_balance(fs_info);
+	else
+		kfree(bctl);
+	return ret;
+}
+
+static int balance_kthread(void *data)
+{
+	struct btrfs_balance_control *bctl =
+			(struct btrfs_balance_control *)data;
+	struct btrfs_fs_info *fs_info = bctl->fs_info;
+	int ret;
+
+	mutex_lock(&fs_info->volume_mutex);
+	mutex_lock(&fs_info->balance_mutex);
+
+	set_balance_control(bctl);
+
+	printk(KERN_INFO "btrfs: continuing balance\n");
+	ret = btrfs_balance(bctl, NULL);
+
+	mutex_unlock(&fs_info->balance_mutex);
+	mutex_unlock(&fs_info->volume_mutex);
+	return ret;
+}
+
+int btrfs_recover_balance(struct btrfs_root *tree_root)
+{
+	struct task_struct *tsk;
+	struct btrfs_balance_control *bctl;
+	struct btrfs_balance_item *item;
+	struct btrfs_disk_balance_args disk_bargs;
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_key key;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	bctl = kzalloc(sizeof(*bctl), GFP_NOFS);
+	if (!bctl) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	key.objectid = BTRFS_BALANCE_OBJECTID;
+	key.type = BTRFS_BALANCE_ITEM_KEY;
+	key.offset = 0;
+
+	ret = btrfs_search_slot(NULL, tree_root, &key, path, 0, 0);
+	if (ret < 0)
+		goto out_bctl;
+	if (ret > 0) { /* ret = -ENOENT; */
+		ret = 0;
+		goto out_bctl;
+	}
+
+	leaf = path->nodes[0];
+	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_balance_item);
+
+	bctl->fs_info = tree_root->fs_info;
+	bctl->flags = btrfs_balance_flags(leaf, item) | BTRFS_BALANCE_RESUME;
+
+	btrfs_balance_data(leaf, item, &disk_bargs);
+	btrfs_disk_balance_args_to_cpu(&bctl->data, &disk_bargs);
+	btrfs_balance_meta(leaf, item, &disk_bargs);
+	btrfs_disk_balance_args_to_cpu(&bctl->meta, &disk_bargs);
+	btrfs_balance_sys(leaf, item, &disk_bargs);
+	btrfs_disk_balance_args_to_cpu(&bctl->sys, &disk_bargs);
+
+	tsk = kthread_run(balance_kthread, bctl, "btrfs-balance");
+	if (IS_ERR(tsk))
+		ret = PTR_ERR(tsk);
+	else
+		goto out;
+
+out_bctl:
 	kfree(bctl);
+out:
+	btrfs_free_path(path);
 	return ret;
 }
 
-- 
cgit 


From 9555c6c180600b40f6e86bd4dc53bf47e06ed663 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:48 +0200
Subject: Btrfs: add skip_balance mount option

Since restriper kthread starts involuntarily on mount and can suck cpu
and memory bandwidth add a mount option to forcefully skip it.  The
restriper in that case hangs around in paused state and can be resumed
from userspace when it's convenient.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 17e565388de0..e0160607e6e2 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2707,15 +2707,19 @@ static int balance_kthread(void *data)
 	struct btrfs_balance_control *bctl =
 			(struct btrfs_balance_control *)data;
 	struct btrfs_fs_info *fs_info = bctl->fs_info;
-	int ret;
+	int ret = 0;
 
 	mutex_lock(&fs_info->volume_mutex);
 	mutex_lock(&fs_info->balance_mutex);
 
 	set_balance_control(bctl);
 
-	printk(KERN_INFO "btrfs: continuing balance\n");
-	ret = btrfs_balance(bctl, NULL);
+	if (btrfs_test_opt(fs_info->tree_root, SKIP_BALANCE)) {
+		printk(KERN_INFO "btrfs: force skipping balance\n");
+	} else {
+		printk(KERN_INFO "btrfs: continuing balance\n");
+		ret = btrfs_balance(bctl, NULL);
+	}
 
 	mutex_unlock(&fs_info->balance_mutex);
 	mutex_unlock(&fs_info->volume_mutex);
-- 
cgit 


From 837d5b6e46d1a4af5b6cc8f2fe83cb5de79a2961 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:49 +0200
Subject: Btrfs: allow for pausing restriper

Implement an ioctl for pausing restriper.  This pauses the relocation,
but balance is still considered to be "in progress": balance item is
not deleted, other volume operations cannot be started, etc.  If paused
in the middle of profile changing operation we will continue making
allocations with the target profile.

Add a hook to close_ctree() to pause restriper and free its data
structures on unmount.  (It's safe to unmount when restriper is in
"paused" state, we will resume with the same parameters on the next
mount)

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 49 insertions(+), 2 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index e0160607e6e2..d32660ce753d 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2492,6 +2492,11 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
+		if (atomic_read(&fs_info->balance_pause_req)) {
+			ret = -ECANCELED;
+			goto error;
+		}
+
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
 		if (ret < 0)
 			goto error;
@@ -2553,6 +2558,11 @@ error:
 	return ret;
 }
 
+static inline int balance_need_close(struct btrfs_fs_info *fs_info)
+{
+	return atomic_read(&fs_info->balance_pause_req) == 0;
+}
+
 static void __cancel_balance(struct btrfs_fs_info *fs_info)
 {
 	int ret;
@@ -2575,7 +2585,8 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 	u64 allowed;
 	int ret;
 
-	if (btrfs_fs_closing(fs_info)) {
+	if (btrfs_fs_closing(fs_info) ||
+	    atomic_read(&fs_info->balance_pause_req)) {
 		ret = -EINVAL;
 		goto out;
 	}
@@ -2680,18 +2691,25 @@ do_balance:
 		spin_unlock(&fs_info->balance_lock);
 	}
 
+	atomic_inc(&fs_info->balance_running);
 	mutex_unlock(&fs_info->balance_mutex);
 
 	ret = __btrfs_balance(fs_info);
 
 	mutex_lock(&fs_info->balance_mutex);
+	atomic_dec(&fs_info->balance_running);
 
 	if (bargs) {
 		memset(bargs, 0, sizeof(*bargs));
 		update_ioctl_balance_args(fs_info, bargs);
 	}
 
-	__cancel_balance(fs_info);
+	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
+	    balance_need_close(fs_info)) {
+		__cancel_balance(fs_info);
+	}
+
+	wake_up(&fs_info->balance_wait_q);
 
 	return ret;
 out:
@@ -2785,6 +2803,35 @@ out:
 	return ret;
 }
 
+int btrfs_pause_balance(struct btrfs_fs_info *fs_info)
+{
+	int ret = 0;
+
+	mutex_lock(&fs_info->balance_mutex);
+	if (!fs_info->balance_ctl) {
+		mutex_unlock(&fs_info->balance_mutex);
+		return -ENOTCONN;
+	}
+
+	if (atomic_read(&fs_info->balance_running)) {
+		atomic_inc(&fs_info->balance_pause_req);
+		mutex_unlock(&fs_info->balance_mutex);
+
+		wait_event(fs_info->balance_wait_q,
+			   atomic_read(&fs_info->balance_running) == 0);
+
+		mutex_lock(&fs_info->balance_mutex);
+		/* we are good with balance_ctl ripped off from under us */
+		BUG_ON(atomic_read(&fs_info->balance_running));
+		atomic_dec(&fs_info->balance_pause_req);
+	} else {
+		ret = -ENOTCONN;
+	}
+
+	mutex_unlock(&fs_info->balance_mutex);
+	return ret;
+}
+
 /*
  * shrinking a device means finding all of the device extents past
  * the new size, and then following the back refs to the chunks.
-- 
cgit 


From a7e99c691af553fc15ac46a51f130b7c59a65f76 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:49 +0200
Subject: Btrfs: allow for canceling restriper

Implement an ioctl for canceling restriper.  Currently we wait until
relocation of the current block group is finished, in future this can be
done by triggering a commit.  Balance item is deleted and no memory
about the interrupted balance is kept.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 47 ++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 44 insertions(+), 3 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index d32660ce753d..c32667318ae4 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2492,7 +2492,8 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
-		if (atomic_read(&fs_info->balance_pause_req)) {
+		if (atomic_read(&fs_info->balance_pause_req) ||
+		    atomic_read(&fs_info->balance_cancel_req)) {
 			ret = -ECANCELED;
 			goto error;
 		}
@@ -2560,7 +2561,10 @@ error:
 
 static inline int balance_need_close(struct btrfs_fs_info *fs_info)
 {
-	return atomic_read(&fs_info->balance_pause_req) == 0;
+	/* cancel requested || normal exit path */
+	return atomic_read(&fs_info->balance_cancel_req) ||
+		(atomic_read(&fs_info->balance_pause_req) == 0 &&
+		 atomic_read(&fs_info->balance_cancel_req) == 0);
 }
 
 static void __cancel_balance(struct btrfs_fs_info *fs_info)
@@ -2586,7 +2590,8 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 	int ret;
 
 	if (btrfs_fs_closing(fs_info) ||
-	    atomic_read(&fs_info->balance_pause_req)) {
+	    atomic_read(&fs_info->balance_pause_req) ||
+	    atomic_read(&fs_info->balance_cancel_req)) {
 		ret = -EINVAL;
 		goto out;
 	}
@@ -2832,6 +2837,42 @@ int btrfs_pause_balance(struct btrfs_fs_info *fs_info)
 	return ret;
 }
 
+int btrfs_cancel_balance(struct btrfs_fs_info *fs_info)
+{
+	mutex_lock(&fs_info->balance_mutex);
+	if (!fs_info->balance_ctl) {
+		mutex_unlock(&fs_info->balance_mutex);
+		return -ENOTCONN;
+	}
+
+	atomic_inc(&fs_info->balance_cancel_req);
+	/*
+	 * if we are running just wait and return, balance item is
+	 * deleted in btrfs_balance in this case
+	 */
+	if (atomic_read(&fs_info->balance_running)) {
+		mutex_unlock(&fs_info->balance_mutex);
+		wait_event(fs_info->balance_wait_q,
+			   atomic_read(&fs_info->balance_running) == 0);
+		mutex_lock(&fs_info->balance_mutex);
+	} else {
+		/* __cancel_balance needs volume_mutex */
+		mutex_unlock(&fs_info->balance_mutex);
+		mutex_lock(&fs_info->volume_mutex);
+		mutex_lock(&fs_info->balance_mutex);
+
+		if (fs_info->balance_ctl)
+			__cancel_balance(fs_info);
+
+		mutex_unlock(&fs_info->volume_mutex);
+	}
+
+	BUG_ON(fs_info->balance_ctl || atomic_read(&fs_info->balance_running));
+	atomic_dec(&fs_info->balance_cancel_req);
+	mutex_unlock(&fs_info->balance_mutex);
+	return 0;
+}
+
 /*
  * shrinking a device means finding all of the device extents past
  * the new size, and then following the back refs to the chunks.
-- 
cgit 


From 19a39dce3b9bf0244d19a446718ad6f7605ff099 Mon Sep 17 00:00:00 2001
From: Ilya Dryomov <idryomov@gmail.com>
Date: Mon, 16 Jan 2012 22:04:49 +0200
Subject: Btrfs: add balance progress reporting

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
---
 fs/btrfs/volumes.c | 39 +++++++++++++++++++++++++++++++++++----
 1 file changed, 35 insertions(+), 4 deletions(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index c32667318ae4..d73439b4d7da 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2441,6 +2441,7 @@ static u64 div_factor(u64 num, int factor)
 
 static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 {
+	struct btrfs_balance_control *bctl = fs_info->balance_ctl;
 	struct btrfs_root *chunk_root = fs_info->chunk_root;
 	struct btrfs_root *dev_root = fs_info->dev_root;
 	struct list_head *devices;
@@ -2456,6 +2457,7 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 	int slot;
 	int ret;
 	int enospc_errors = 0;
+	bool counting = true;
 
 	/* step one make some room on all the devices */
 	devices = &fs_info->fs_devices->devices;
@@ -2487,12 +2489,18 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 		ret = -ENOMEM;
 		goto error;
 	}
+
+	/* zero out stat counters */
+	spin_lock(&fs_info->balance_lock);
+	memset(&bctl->stat, 0, sizeof(bctl->stat));
+	spin_unlock(&fs_info->balance_lock);
+again:
 	key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
 	key.offset = (u64)-1;
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
-		if (atomic_read(&fs_info->balance_pause_req) ||
+		if ((!counting && atomic_read(&fs_info->balance_pause_req)) ||
 		    atomic_read(&fs_info->balance_cancel_req)) {
 			ret = -ECANCELED;
 			goto error;
@@ -2529,24 +2537,47 @@ static int __btrfs_balance(struct btrfs_fs_info *fs_info)
 
 		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
 
+		if (!counting) {
+			spin_lock(&fs_info->balance_lock);
+			bctl->stat.considered++;
+			spin_unlock(&fs_info->balance_lock);
+		}
+
 		ret = should_balance_chunk(chunk_root, leaf, chunk,
 					   found_key.offset);
 		btrfs_release_path(path);
 		if (!ret)
 			goto loop;
 
+		if (counting) {
+			spin_lock(&fs_info->balance_lock);
+			bctl->stat.expected++;
+			spin_unlock(&fs_info->balance_lock);
+			goto loop;
+		}
+
 		ret = btrfs_relocate_chunk(chunk_root,
 					   chunk_root->root_key.objectid,
 					   found_key.objectid,
 					   found_key.offset);
 		if (ret && ret != -ENOSPC)
 			goto error;
-		if (ret == -ENOSPC)
+		if (ret == -ENOSPC) {
 			enospc_errors++;
+		} else {
+			spin_lock(&fs_info->balance_lock);
+			bctl->stat.completed++;
+			spin_unlock(&fs_info->balance_lock);
+		}
 loop:
 		key.offset = found_key.offset - 1;
 	}
 
+	if (counting) {
+		btrfs_release_path(path);
+		counting = false;
+		goto again;
+	}
 error:
 	btrfs_free_path(path);
 	if (enospc_errors) {
@@ -2576,7 +2607,7 @@ static void __cancel_balance(struct btrfs_fs_info *fs_info)
 	BUG_ON(ret);
 }
 
-void update_ioctl_balance_args(struct btrfs_fs_info *fs_info,
+void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
 			       struct btrfs_ioctl_balance_args *bargs);
 
 /*
@@ -2706,7 +2737,7 @@ do_balance:
 
 	if (bargs) {
 		memset(bargs, 0, sizeof(*bargs));
-		update_ioctl_balance_args(fs_info, bargs);
+		update_ioctl_balance_args(fs_info, 0, bargs);
 	}
 
 	if ((ret && ret != -ECANCELED && ret != -ENOSPC) ||
-- 
cgit 


From 96bdc7dc61fb1b1e8e858dafb13abee8482ba064 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 16 Jan 2012 08:13:11 -0500
Subject: Btrfs: use larger system chunks

system chunks by default are very small.  This makes them slightly
larger and also fixes the conditional checks to make sure we don't
allocate a billion of them at once.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/volumes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 59e878f9fdcc..7ffdb154daec 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -3166,7 +3166,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 			max_stripe_size = 256 * 1024 * 1024;
 		max_chunk_size = max_stripe_size;
 	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
-		max_stripe_size = 8 * 1024 * 1024;
+		max_stripe_size = 32 * 1024 * 1024;
 		max_chunk_size = 2 * max_stripe_size;
 	} else {
 		printk(KERN_ERR "btrfs: invalid chunk type 0x%llx requested\n",
-- 
cgit 


From 8a3344269465b26761b74493cfbeeaa75d821614 Mon Sep 17 00:00:00 2001
From: David Sterba <dsterba@suse.cz>
Date: Fri, 7 Oct 2011 18:06:13 +0200
Subject: btrfs: silence warning in raid array setup

Raid array setup code creates an extent buffer in an usual way. When the
PAGE_CACHE_SIZE is > super block size, the extent pages are not marked
up-to-date, which triggers a WARN_ON in the following
write_extent_buffer call. Add an explicit up-to-date call to silence the
warning.

Signed-off-by: David Sterba <dsterba@suse.cz>
---
 fs/btrfs/volumes.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 7ffdb154daec..d8f282b5baa9 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -4357,6 +4357,20 @@ int btrfs_read_sys_array(struct btrfs_root *root)
 		return -ENOMEM;
 	btrfs_set_buffer_uptodate(sb);
 	btrfs_set_buffer_lockdep_class(root->root_key.objectid, sb, 0);
+	/*
+	 * The sb extent buffer is artifical and just used to read the system array.
+	 * btrfs_set_buffer_uptodate() call does not properly mark all it's
+	 * pages up-to-date when the page is larger: extent does not cover the
+	 * whole page and consequently check_page_uptodate does not find all
+	 * the page's extents up-to-date (the hole beyond sb),
+	 * write_extent_buffer then triggers a WARN_ON.
+	 *
+	 * Regular short extents go through mark_extent_buffer_dirty/writeback cycle,
+	 * but sb spans only this function. Add an explicit SetPageUptodate call
+	 * to silence the warning eg. on PowerPC 64.
+	 */
+	if (PAGE_CACHE_SIZE > BTRFS_SUPER_INFO_SIZE)
+		SetPageUptodate(sb->first_page);
 
 	write_extent_buffer(sb, super_copy, 0, BTRFS_SUPER_INFO_SIZE);
 	array_size = btrfs_super_sys_array_size(super_copy);
-- 
cgit 


From 285190d99fef696ec8b0041787387f013cb71d67 Mon Sep 17 00:00:00 2001
From: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Date: Thu, 16 Feb 2012 16:23:58 +0900
Subject: Btrfs: check return value of lookup_extent_mapping() correctly

This patch corrects error checking of lookup_extent_mapping().

Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
---
 fs/btrfs/volumes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index d8f282b5baa9..cd040bf3fd87 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1954,7 +1954,7 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
 	read_unlock(&em_tree->lock);
 
-	BUG_ON(em->start > chunk_offset ||
+	BUG_ON(!em || em->start > chunk_offset ||
 	       em->start + em->len < chunk_offset);
 	map = (struct map_lookup *)em->bdev;
 
-- 
cgit 


From a6b0d5c8dbfd428717fc4db4c36757783f391c7b Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 20 Feb 2012 20:53:43 -0500
Subject: Btrfs: make sure we update latest_bdev

When we are setting up the mount, we close all the
devices that were not actually part of the metadata we found.

But, we don't make sure that one of those devices wasn't
fs_devices->latest_bdev, which means we can do a use after free
on the one we closed.

This updates latest_bdev as it goes.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/volumes.c | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

(limited to 'fs/btrfs/volumes.c')

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index cd040bf3fd87..7eefdef8a14c 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -459,12 +459,23 @@ int btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices)
 {
 	struct btrfs_device *device, *next;
 
+	struct block_device *latest_bdev = NULL;
+	u64 latest_devid = 0;
+	u64 latest_transid = 0;
+
 	mutex_lock(&uuid_mutex);
 again:
 	/* This is the initialized path, it is safe to release the devices. */
 	list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) {
-		if (device->in_fs_metadata)
+		if (device->in_fs_metadata) {
+			if (!latest_transid ||
+			    device->generation > latest_transid) {
+				latest_devid = device->devid;
+				latest_transid = device->generation;
+				latest_bdev = device->bdev;
+			}
 			continue;
+		}
 
 		if (device->bdev) {
 			blkdev_put(device->bdev, device->mode);
@@ -487,6 +498,10 @@ again:
 		goto again;
 	}
 
+	fs_devices->latest_bdev = latest_bdev;
+	fs_devices->latest_devid = latest_devid;
+	fs_devices->latest_trans = latest_transid;
+
 	mutex_unlock(&uuid_mutex);
 	return 0;
 }
-- 
cgit