From aa2ffd06168e25689e0eb9662bf4595ba2bbac14 Mon Sep 17 00:00:00 2001
From: Stefan Behrens <sbehrens@giantdisaster.de>
Date: Thu, 26 Jul 2012 03:40:35 -0600
Subject: Btrfs: fix a misplaced address operator in a condition

This should obviously not be "if (&flag)" but "if (flag)".

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
---
 fs/btrfs/locking.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/btrfs/locking.c b/fs/btrfs/locking.c
index a44eff074805..2a1762c66041 100644
--- a/fs/btrfs/locking.c
+++ b/fs/btrfs/locking.c
@@ -67,7 +67,7 @@ void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
 {
 	if (eb->lock_nested) {
 		read_lock(&eb->lock);
-		if (&eb->lock_nested && current->pid == eb->lock_owner) {
+		if (eb->lock_nested && current->pid == eb->lock_owner) {
 			read_unlock(&eb->lock);
 			return;
 		}
-- 
cgit v1.2.3


From 5986802c2fcc754040bb7ed95f30bb16c4a843b7 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Mon, 30 Jul 2012 02:16:10 -0600
Subject: Btrfs: fix some error codes in btrfs_qgroup_inherit()

These are returning zero when it should be returning a negative error
code.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
 fs/btrfs/qgroup.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index bc424ae5a81a..229ef8927e6b 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1369,8 +1369,10 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
 
 	if (srcid) {
 		srcgroup = find_qgroup_rb(fs_info, srcid);
-		if (!srcgroup)
+		if (!srcgroup) {
+			ret = -EINVAL;
 			goto unlock;
+		}
 		dstgroup->rfer = srcgroup->rfer - level_size;
 		dstgroup->rfer_cmpr = srcgroup->rfer_cmpr - level_size;
 		srcgroup->excl = level_size;
@@ -1379,8 +1381,10 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
 		qgroup_dirty(fs_info, srcgroup);
 	}
 
-	if (!inherit)
+	if (!inherit) {
+		ret = -EINVAL;
 		goto unlock;
+	}
 
 	i_qgroups = (u64 *)(inherit + 1);
 	for (i = 0; i < inherit->num_qgroups; ++i) {
-- 
cgit v1.2.3


From 57a5a882031dba5cb7bc7ebc955b897498365fe2 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Mon, 30 Jul 2012 02:15:43 -0600
Subject: Btrfs: checking for NULL instead of IS_ERR

add_qgroup_rb() never returns NULL, only error pointers.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
 fs/btrfs/qgroup.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 229ef8927e6b..38b42e7bc91d 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -1364,8 +1364,10 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
 	spin_lock(&fs_info->qgroup_lock);
 
 	dstgroup = add_qgroup_rb(fs_info, objectid);
-	if (!dstgroup)
+	if (IS_ERR(dstgroup)) {
+		ret = PTR_ERR(dstgroup);
 		goto unlock;
+	}
 
 	if (srcid) {
 		srcgroup = find_qgroup_rb(fs_info, srcid);
-- 
cgit v1.2.3


From 55e591ffde38e0088b022129e035e18a8d04c7e6 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Mon, 30 Jul 2012 02:15:15 -0600
Subject: Btrfs: unlock on error in btrfs_delalloc_reserve_metadata()

We should release this mutex before returning the error code.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
 fs/btrfs/extent-tree.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 4e1b153b7c47..45c69c4184c9 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -4571,8 +4571,10 @@ int btrfs_delalloc_reserve_metadata(struct inode *inode, u64 num_bytes)
 	if (root->fs_info->quota_enabled) {
 		ret = btrfs_qgroup_reserve(root, num_bytes +
 					   nr_extents * root->leafsize);
-		if (ret)
+		if (ret) {
+			mutex_unlock(&BTRFS_I(inode)->delalloc_mutex);
 			return ret;
+		}
 	}
 
 	ret = reserve_metadata_bytes(root, block_rsv, to_reserve, flush);
-- 
cgit v1.2.3


From dadd1105ca9a1e506c678e8e410e9623efdda821 Mon Sep 17 00:00:00 2001
From: Dan Carpenter <dan.carpenter@oracle.com>
Date: Mon, 30 Jul 2012 02:10:44 -0600
Subject: Btrfs: fix some endian bugs handling the root times

"trans->transid" is cpu endian but we want to store the data as little
endian.  "item->ctime.nsec" is only 32 bits, not 64.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
 fs/btrfs/ioctl.c       | 2 +-
 fs/btrfs/root-tree.c   | 4 ++--
 fs/btrfs/transaction.c | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 43f0012016e3..a1fbca0a1003 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -424,7 +424,7 @@ static noinline int create_subvol(struct btrfs_root *root,
 	uuid_le_gen(&new_uuid);
 	memcpy(root_item.uuid, new_uuid.b, BTRFS_UUID_SIZE);
 	root_item.otime.sec = cpu_to_le64(cur_time.tv_sec);
-	root_item.otime.nsec = cpu_to_le64(cur_time.tv_nsec);
+	root_item.otime.nsec = cpu_to_le32(cur_time.tv_nsec);
 	root_item.ctime = root_item.otime;
 	btrfs_set_root_ctransid(&root_item, trans->transid);
 	btrfs_set_root_otransid(&root_item, trans->transid);
diff --git a/fs/btrfs/root-tree.c b/fs/btrfs/root-tree.c
index 6bb465cca20f..10d8e4d88071 100644
--- a/fs/btrfs/root-tree.c
+++ b/fs/btrfs/root-tree.c
@@ -544,8 +544,8 @@ void btrfs_update_root_times(struct btrfs_trans_handle *trans,
 	struct timespec ct = CURRENT_TIME;
 
 	spin_lock(&root->root_times_lock);
-	item->ctransid = trans->transid;
+	item->ctransid = cpu_to_le64(trans->transid);
 	item->ctime.sec = cpu_to_le64(ct.tv_sec);
-	item->ctime.nsec = cpu_to_le64(ct.tv_nsec);
+	item->ctime.nsec = cpu_to_le32(ct.tv_nsec);
 	spin_unlock(&root->root_times_lock);
 }
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 7ac7cdcc294e..7208ada41e0e 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -1061,7 +1061,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	memcpy(new_root_item->parent_uuid, root->root_item.uuid,
 			BTRFS_UUID_SIZE);
 	new_root_item->otime.sec = cpu_to_le64(cur_time.tv_sec);
-	new_root_item->otime.nsec = cpu_to_le64(cur_time.tv_nsec);
+	new_root_item->otime.nsec = cpu_to_le32(cur_time.tv_nsec);
 	btrfs_set_root_otransid(new_root_item, trans->transid);
 	memset(&new_root_item->stime, 0, sizeof(new_root_item->stime));
 	memset(&new_root_item->rtime, 0, sizeof(new_root_item->rtime));
-- 
cgit v1.2.3


From eb838e73dc2121d2bae47d5678952cd7d48793b5 Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Tue, 31 Jul 2012 16:28:48 -0400
Subject: Btrfs: lock extents as we map them in DIO

A deadlock in xfstests 113 was uncovered by commit

d187663ef24cd3d033f0cbf2867e70b36a3a90b8

This is because we would not return EIOCBQUEUED for short AIO reads, instead
we'd wait for the DIO to complete and then return the amount of data we
transferred, which would allow our stuff to unlock the remaning amount.  But
with this change this no longer happens, so if we have a short AIO read (for
example if we try to read past EOF), we could leave the section from EOF to
the end of where we tried to read locked.  Fixing this is tricky since there
is no clear way to know exactly how much data DIO truly submitted for IO, so
to make this less hard on ourselves and less combersome we need to lock the
extents as we try to map them, and then we unlock any areas we didn't
actually map.  This makes us completely safe from deadlocks and reliance on
a particular behavior of the DIO code.  This also lays the groundwork for
allowing us to use the normal csum storage method for reads which means we
can remove an allocation.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/inode.c | 256 +++++++++++++++++++++++++++----------------------------
 1 file changed, 127 insertions(+), 129 deletions(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index dac1fc21d809..09182449cbdf 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -5773,18 +5773,109 @@ out:
 	return ret;
 }
 
+static int lock_extent_direct(struct inode *inode, u64 lockstart, u64 lockend,
+			      struct extent_state **cached_state, int writing)
+{
+	struct btrfs_ordered_extent *ordered;
+	int ret = 0;
+
+	while (1) {
+		lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+				 0, cached_state);
+		/*
+		 * We're concerned with the entire range that we're going to be
+		 * doing DIO to, so we need to make sure theres no ordered
+		 * extents in this range.
+		 */
+		ordered = btrfs_lookup_ordered_range(inode, lockstart,
+						     lockend - lockstart + 1);
+
+		/*
+		 * We need to make sure there are no buffered pages in this
+		 * range either, we could have raced between the invalidate in
+		 * generic_file_direct_write and locking the extent.  The
+		 * invalidate needs to happen so that reads after a write do not
+		 * get stale data.
+		 */
+		if (!ordered && (!writing ||
+		    !test_range_bit(&BTRFS_I(inode)->io_tree,
+				    lockstart, lockend, EXTENT_UPTODATE, 0,
+				    *cached_state)))
+			break;
+
+		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+				     cached_state, GFP_NOFS);
+
+		if (ordered) {
+			btrfs_start_ordered_extent(inode, ordered, 1);
+			btrfs_put_ordered_extent(ordered);
+		} else {
+			/* Screw you mmap */
+			ret = filemap_write_and_wait_range(inode->i_mapping,
+							   lockstart,
+							   lockend);
+			if (ret)
+				break;
+
+			/*
+			 * If we found a page that couldn't be invalidated just
+			 * fall back to buffered.
+			 */
+			ret = invalidate_inode_pages2_range(inode->i_mapping,
+					lockstart >> PAGE_CACHE_SHIFT,
+					lockend >> PAGE_CACHE_SHIFT);
+			if (ret)
+				break;
+		}
+
+		cond_resched();
+	}
+
+	return ret;
+}
+
 static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 				   struct buffer_head *bh_result, int create)
 {
 	struct extent_map *em;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct extent_state *cached_state = NULL;
 	u64 start = iblock << inode->i_blkbits;
+	u64 lockstart, lockend;
 	u64 len = bh_result->b_size;
 	struct btrfs_trans_handle *trans;
+	int unlock_bits = EXTENT_LOCKED;
+	int ret;
+
+	lockstart = start;
+	lockend = start + len - 1;
+	if (create) {
+		ret = btrfs_delalloc_reserve_space(inode, len);
+		if (ret)
+			return ret;
+		unlock_bits |= EXTENT_DELALLOC | EXTENT_DIRTY;
+	}
+
+	/*
+	 * If this errors out it's because we couldn't invalidate pagecache for
+	 * this range and we need to fallback to buffered.
+	 */
+	if (lock_extent_direct(inode, lockstart, lockend, &cached_state, create))
+		return -ENOTBLK;
+
+	if (create) {
+		ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
+				     lockend, EXTENT_DELALLOC, NULL,
+				     &cached_state, GFP_NOFS);
+		if (ret)
+			goto unlock_err;
+	}
 
 	em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
-	if (IS_ERR(em))
-		return PTR_ERR(em);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto unlock_err;
+	}
 
 	/*
 	 * Ok for INLINE and COMPRESSED extents we need to fallback on buffered
@@ -5803,17 +5894,16 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 	if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags) ||
 	    em->block_start == EXTENT_MAP_INLINE) {
 		free_extent_map(em);
-		return -ENOTBLK;
+		ret = -ENOTBLK;
+		goto unlock_err;
 	}
 
 	/* Just a good old fashioned hole, return */
 	if (!create && (em->block_start == EXTENT_MAP_HOLE ||
 			test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
 		free_extent_map(em);
-		/* DIO will do one hole at a time, so just unlock a sector */
-		unlock_extent(&BTRFS_I(inode)->io_tree, start,
-			      start + root->sectorsize - 1);
-		return 0;
+		ret = 0;
+		goto unlock_err;
 	}
 
 	/*
@@ -5826,8 +5916,9 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 	 *
 	 */
 	if (!create) {
-		len = em->len - (start - em->start);
-		goto map;
+		len = min(len, em->len - (start - em->start));
+		lockstart = start + len;
+		goto unlock;
 	}
 
 	if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags) ||
@@ -5859,7 +5950,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 			btrfs_end_transaction(trans, root);
 			if (ret) {
 				free_extent_map(em);
-				return ret;
+				goto unlock_err;
 			}
 			goto unlock;
 		}
@@ -5872,14 +5963,12 @@ must_cow:
 	 */
 	len = bh_result->b_size;
 	em = btrfs_new_extent_direct(inode, em, start, len);
-	if (IS_ERR(em))
-		return PTR_ERR(em);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto unlock_err;
+	}
 	len = min(len, em->len - (start - em->start));
 unlock:
-	clear_extent_bit(&BTRFS_I(inode)->io_tree, start, start + len - 1,
-			  EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DIRTY, 1,
-			  0, NULL, GFP_NOFS);
-map:
 	bh_result->b_blocknr = (em->block_start + (start - em->start)) >>
 		inode->i_blkbits;
 	bh_result->b_size = len;
@@ -5897,9 +5986,28 @@ map:
 			i_size_write(inode, start + len);
 	}
 
+	/*
+	 * In the case of write we need to clear and unlock the entire range,
+	 * in the case of read we need to unlock only the end area that we
+	 * aren't using if there is any left over space.
+	 */
+	if (lockstart < lockend)
+		clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+				 unlock_bits, 1, 0, &cached_state, GFP_NOFS);
+	else
+		free_extent_state(cached_state);
+
 	free_extent_map(em);
 
 	return 0;
+
+unlock_err:
+	if (create)
+		unlock_bits |= EXTENT_DO_ACCOUNTING;
+
+	clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend,
+			 unlock_bits, 1, 0, &cached_state, GFP_NOFS);
+	return ret;
 }
 
 struct btrfs_dio_private {
@@ -6340,132 +6448,22 @@ static ssize_t check_direct_IO(struct btrfs_root *root, int rw, struct kiocb *io
 out:
 	return retval;
 }
+
 static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
 			const struct iovec *iov, loff_t offset,
 			unsigned long nr_segs)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file->f_mapping->host;
-	struct btrfs_ordered_extent *ordered;
-	struct extent_state *cached_state = NULL;
-	u64 lockstart, lockend;
-	ssize_t ret;
-	int writing = rw & WRITE;
-	int write_bits = 0;
-	size_t count = iov_length(iov, nr_segs);
 
 	if (check_direct_IO(BTRFS_I(inode)->root, rw, iocb, iov,
-			    offset, nr_segs)) {
+			    offset, nr_segs))
 		return 0;
-	}
-
-	lockstart = offset;
-	lockend = offset + count - 1;
-
-	if (writing) {
-		ret = btrfs_delalloc_reserve_space(inode, count);
-		if (ret)
-			goto out;
-	}
 
-	while (1) {
-		lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				 0, &cached_state);
-		/*
-		 * We're concerned with the entire range that we're going to be
-		 * doing DIO to, so we need to make sure theres no ordered
-		 * extents in this range.
-		 */
-		ordered = btrfs_lookup_ordered_range(inode, lockstart,
-						     lockend - lockstart + 1);
-
-		/*
-		 * We need to make sure there are no buffered pages in this
-		 * range either, we could have raced between the invalidate in
-		 * generic_file_direct_write and locking the extent.  The
-		 * invalidate needs to happen so that reads after a write do not
-		 * get stale data.
-		 */
-		if (!ordered && (!writing ||
-		    !test_range_bit(&BTRFS_I(inode)->io_tree,
-				    lockstart, lockend, EXTENT_UPTODATE, 0,
-				    cached_state)))
-			break;
-
-		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				     &cached_state, GFP_NOFS);
-
-		if (ordered) {
-			btrfs_start_ordered_extent(inode, ordered, 1);
-			btrfs_put_ordered_extent(ordered);
-		} else {
-			/* Screw you mmap */
-			ret = filemap_write_and_wait_range(file->f_mapping,
-							   lockstart,
-							   lockend);
-			if (ret)
-				goto out;
-
-			/*
-			 * If we found a page that couldn't be invalidated just
-			 * fall back to buffered.
-			 */
-			ret = invalidate_inode_pages2_range(file->f_mapping,
-					lockstart >> PAGE_CACHE_SHIFT,
-					lockend >> PAGE_CACHE_SHIFT);
-			if (ret) {
-				if (ret == -EBUSY)
-					ret = 0;
-				goto out;
-			}
-		}
-
-		cond_resched();
-	}
-
-	/*
-	 * we don't use btrfs_set_extent_delalloc because we don't want
-	 * the dirty or uptodate bits
-	 */
-	if (writing) {
-		write_bits = EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING;
-		ret = set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				     EXTENT_DELALLOC, NULL, &cached_state,
-				     GFP_NOFS);
-		if (ret) {
-			clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
-					 lockend, EXTENT_LOCKED | write_bits,
-					 1, 0, &cached_state, GFP_NOFS);
-			goto out;
-		}
-	}
-
-	free_extent_state(cached_state);
-	cached_state = NULL;
-
-	ret = __blockdev_direct_IO(rw, iocb, inode,
+	return __blockdev_direct_IO(rw, iocb, inode,
 		   BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev,
 		   iov, offset, nr_segs, btrfs_get_blocks_direct, NULL,
 		   btrfs_submit_direct, 0);
-
-	if (ret < 0 && ret != -EIOCBQUEUED) {
-		clear_extent_bit(&BTRFS_I(inode)->io_tree, offset,
-			      offset + iov_length(iov, nr_segs) - 1,
-			      EXTENT_LOCKED | write_bits, 1, 0,
-			      &cached_state, GFP_NOFS);
-	} else if (ret >= 0 && ret < iov_length(iov, nr_segs)) {
-		/*
-		 * We're falling back to buffered, unlock the section we didn't
-		 * do IO on.
-		 */
-		clear_extent_bit(&BTRFS_I(inode)->io_tree, offset + ret,
-			      offset + iov_length(iov, nr_segs) - 1,
-			      EXTENT_LOCKED | write_bits, 1, 0,
-			      &cached_state, GFP_NOFS);
-	}
-out:
-	free_extent_state(cached_state);
-	return ret;
 }
 
 static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
-- 
cgit v1.2.3


From 3627bf4503b504077332c13496cb1bd54713bcbb Mon Sep 17 00:00:00 2001
From: Stefan Behrens <sbehrens@giantdisaster.de>
Date: Wed, 1 Aug 2012 04:28:01 -0600
Subject: Btrfs: fix that error value is changed by mistake

In iterate_inodes_from_logical() the error result from
extent_from_logical() is patched by mistake. Typically ENOENT is
patched to EINVAL because (-ENOENT & BTRFS_EXTENT_FLAG_TREE_BLOCK)
evaluates to true.

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
---
 fs/btrfs/backref.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index a256f3b2a845..ff6475f409d6 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -1438,10 +1438,10 @@ int iterate_inodes_from_logical(u64 logical, struct btrfs_fs_info *fs_info,
 	ret = extent_from_logical(fs_info, logical, path,
 					&found_key);
 	btrfs_release_path(path);
-	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
-		ret = -EINVAL;
 	if (ret < 0)
 		return ret;
+	if (ret & BTRFS_EXTENT_FLAG_TREE_BLOCK)
+		return -EINVAL;
 
 	extent_item_pos = logical - found_key.objectid;
 	ret = iterate_extent_inodes(fs_info, found_key.objectid,
-- 
cgit v1.2.3


From aa9ddcd4b5557102fa25695c11904f249b4dec49 Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Thu, 2 Aug 2012 10:22:20 -0400
Subject: Btrfs: do not use missing devices when showing devname

If you do the following

mkfs.btrfs /dev/sdb /dev/sdc
rmmod btrfs
dd if=/dev/zero of=/dev/sdb bs=1M count=1
mount -o degraded /dev/sdc /mnt/btrfs-test

the box will panic trying to deref the name for the missing dev since it is
the lower numbered devid.  So fix show_devname to not use missing devices.
Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/super.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 75ee2c7791f0..2e06f124f284 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -1505,6 +1505,8 @@ static int btrfs_show_devname(struct seq_file *m, struct dentry *root)
 	while (cur_devices) {
 		head = &cur_devices->devices;
 		list_for_each_entry(dev, head, dev_list) {
+			if (dev->missing)
+				continue;
 			if (!first_dev || dev->devid < first_dev->devid)
 				first_dev = dev;
 		}
-- 
cgit v1.2.3


From 99f5944b8477914406173b47b4f261356286730b Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Thu, 2 Aug 2012 10:23:59 -0400
Subject: Btrfs: do not strdup non existent strings

When we close devices we add back empty devices for some reason that escapes
me.  In the case of a missing dev we don't allocate an rcu_string for it's
name, so check to see if the device has a name and if it doesn't don't
bother strdup()'ing it.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/volumes.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index b8708f994e67..3b394503bd4e 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -569,9 +569,11 @@ static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
 		memcpy(new_device, device, sizeof(*new_device));
 
 		/* Safe because we are under uuid_mutex */
-		name = rcu_string_strdup(device->name->str, GFP_NOFS);
-		BUG_ON(device->name && !name); /* -ENOMEM */
-		rcu_assign_pointer(new_device->name, name);
+		if (device->name) {
+			name = rcu_string_strdup(device->name->str, GFP_NOFS);
+			BUG_ON(device->name && !name); /* -ENOMEM */
+			rcu_assign_pointer(new_device->name, name);
+		}
 		new_device->bdev = NULL;
 		new_device->writeable = 0;
 		new_device->in_fs_metadata = 0;
-- 
cgit v1.2.3


From c329861da40623cd838b8c9ee31a850242fd88cf Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Fri, 3 Aug 2012 16:49:19 -0400
Subject: Btrfs: don't allocate a seperate csums array for direct reads

We've been allocating a big array for csums instead of storing them in the
io_tree like we do for buffered reads because previously we were locking the
entire range, so we didn't have an extent state for each sector of the
range.  But now that we do the range locking as we map the buffers we can
limit the mapping lenght to sectorsize and use the private part of the
io_tree for our csums.  This allows us to avoid an extra memory allocation
for direct reads which could incur latency.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/ctree.h     |  2 +-
 fs/btrfs/file-item.c |  4 ++--
 fs/btrfs/inode.c     | 45 ++++++++++++++++-----------------------------
 3 files changed, 19 insertions(+), 32 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index adb1cd7ceb9b..348196350bf0 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -3192,7 +3192,7 @@ int btrfs_del_csums(struct btrfs_trans_handle *trans,
 int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
 			  struct bio *bio, u32 *dst);
 int btrfs_lookup_bio_sums_dio(struct btrfs_root *root, struct inode *inode,
-			      struct bio *bio, u64 logical_offset, u32 *dst);
+			      struct bio *bio, u64 logical_offset);
 int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root,
 			     u64 objectid, u64 pos,
diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
index b45b9de0c21d..857d93cd01dc 100644
--- a/fs/btrfs/file-item.c
+++ b/fs/btrfs/file-item.c
@@ -272,9 +272,9 @@ int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
 }
 
 int btrfs_lookup_bio_sums_dio(struct btrfs_root *root, struct inode *inode,
-			      struct bio *bio, u64 offset, u32 *dst)
+			      struct bio *bio, u64 offset)
 {
-	return __btrfs_lookup_bio_sums(root, inode, bio, offset, dst, 1);
+	return __btrfs_lookup_bio_sums(root, inode, bio, offset, NULL, 1);
 }
 
 int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 09182449cbdf..2d65c52b0944 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -5847,15 +5847,18 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 	int unlock_bits = EXTENT_LOCKED;
 	int ret;
 
-	lockstart = start;
-	lockend = start + len - 1;
 	if (create) {
 		ret = btrfs_delalloc_reserve_space(inode, len);
 		if (ret)
 			return ret;
 		unlock_bits |= EXTENT_DELALLOC | EXTENT_DIRTY;
+	} else {
+		len = min_t(u64, len, root->sectorsize);
 	}
 
+	lockstart = start;
+	lockend = start + len - 1;
+
 	/*
 	 * If this errors out it's because we couldn't invalidate pagecache for
 	 * this range and we need to fallback to buffered.
@@ -6015,7 +6018,6 @@ struct btrfs_dio_private {
 	u64 logical_offset;
 	u64 disk_bytenr;
 	u64 bytes;
-	u32 *csums;
 	void *private;
 
 	/* number of bios pending for this dio */
@@ -6035,7 +6037,6 @@ static void btrfs_endio_direct_read(struct bio *bio, int err)
 	struct inode *inode = dip->inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	u64 start;
-	u32 *private = dip->csums;
 
 	start = dip->logical_offset;
 	do {
@@ -6043,8 +6044,12 @@ static void btrfs_endio_direct_read(struct bio *bio, int err)
 			struct page *page = bvec->bv_page;
 			char *kaddr;
 			u32 csum = ~(u32)0;
+			u64 private = ~(u32)0;
 			unsigned long flags;
 
+			if (get_state_private(&BTRFS_I(inode)->io_tree,
+					      start, &private))
+				goto failed;
 			local_irq_save(flags);
 			kaddr = kmap_atomic(page);
 			csum = btrfs_csum_data(root, kaddr + bvec->bv_offset,
@@ -6054,18 +6059,18 @@ static void btrfs_endio_direct_read(struct bio *bio, int err)
 			local_irq_restore(flags);
 
 			flush_dcache_page(bvec->bv_page);
-			if (csum != *private) {
+			if (csum != private) {
+failed:
 				printk(KERN_ERR "btrfs csum failed ino %llu off"
 				      " %llu csum %u private %u\n",
 				      (unsigned long long)btrfs_ino(inode),
 				      (unsigned long long)start,
-				      csum, *private);
+				      csum, (unsigned)private);
 				err = -EIO;
 			}
 		}
 
 		start += bvec->bv_len;
-		private++;
 		bvec++;
 	} while (bvec <= bvec_end);
 
@@ -6073,7 +6078,6 @@ static void btrfs_endio_direct_read(struct bio *bio, int err)
 		      dip->logical_offset + dip->bytes - 1);
 	bio->bi_private = dip->private;
 
-	kfree(dip->csums);
 	kfree(dip);
 
 	/* If we had a csum failure make sure to clear the uptodate flag */
@@ -6179,7 +6183,7 @@ static struct bio *btrfs_dio_bio_alloc(struct block_device *bdev,
 
 static inline int __btrfs_submit_dio_bio(struct bio *bio, struct inode *inode,
 					 int rw, u64 file_offset, int skip_sum,
-					 u32 *csums, int async_submit)
+					 int async_submit)
 {
 	int write = rw & REQ_WRITE;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -6212,8 +6216,7 @@ static inline int __btrfs_submit_dio_bio(struct bio *bio, struct inode *inode,
 		if (ret)
 			goto err;
 	} else if (!skip_sum) {
-		ret = btrfs_lookup_bio_sums_dio(root, inode, bio,
-					  file_offset, csums);
+		ret = btrfs_lookup_bio_sums_dio(root, inode, bio, file_offset);
 		if (ret)
 			goto err;
 	}
@@ -6239,10 +6242,8 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip,
 	u64 submit_len = 0;
 	u64 map_length;
 	int nr_pages = 0;
-	u32 *csums = dip->csums;
 	int ret = 0;
 	int async_submit = 0;
-	int write = rw & REQ_WRITE;
 
 	map_length = orig_bio->bi_size;
 	ret = btrfs_map_block(map_tree, READ, start_sector << 9,
@@ -6278,16 +6279,13 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip,
 			atomic_inc(&dip->pending_bios);
 			ret = __btrfs_submit_dio_bio(bio, inode, rw,
 						     file_offset, skip_sum,
-						     csums, async_submit);
+						     async_submit);
 			if (ret) {
 				bio_put(bio);
 				atomic_dec(&dip->pending_bios);
 				goto out_err;
 			}
 
-			/* Write's use the ordered csums */
-			if (!write && !skip_sum)
-				csums = csums + nr_pages;
 			start_sector += submit_len >> 9;
 			file_offset += submit_len;
 
@@ -6317,7 +6315,7 @@ static int btrfs_submit_direct_hook(int rw, struct btrfs_dio_private *dip,
 
 submit:
 	ret = __btrfs_submit_dio_bio(bio, inode, rw, file_offset, skip_sum,
-				     csums, async_submit);
+				     async_submit);
 	if (!ret)
 		return 0;
 
@@ -6353,17 +6351,6 @@ static void btrfs_submit_direct(int rw, struct bio *bio, struct inode *inode,
 		ret = -ENOMEM;
 		goto free_ordered;
 	}
-	dip->csums = NULL;
-
-	/* Write's use the ordered csum stuff, so we don't need dip->csums */
-	if (!write && !skip_sum) {
-		dip->csums = kmalloc(sizeof(u32) * bio->bi_vcnt, GFP_NOFS);
-		if (!dip->csums) {
-			kfree(dip);
-			ret = -ENOMEM;
-			goto free_ordered;
-		}
-	}
 
 	dip->private = bio->bi_private;
 	dip->inode = inode;
-- 
cgit v1.2.3


From 6209526531e70c080f79318ab8f50e26846c40a8 Mon Sep 17 00:00:00 2001
From: Fengguang Wu <fengguang.wu@intel.com>
Date: Sat, 4 Aug 2012 01:45:02 -0600
Subject: btrfs: fix second lock in btrfs_delete_delayed_items()

Fix a real bug caught by coccinelle.

fs/btrfs/delayed-inode.c:1013:1-11: second lock on line 1013

Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
---
 fs/btrfs/delayed-inode.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index 335605c8ceab..00deed4ef3ed 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -1028,9 +1028,10 @@ do_again:
 		btrfs_release_delayed_item(prev);
 		ret = 0;
 		btrfs_release_path(path);
-		if (curr)
+		if (curr) {
+			mutex_unlock(&node->mutex);
 			goto do_again;
-		else
+		} else
 			goto delete_fail;
 	}
 
-- 
cgit v1.2.3


From 1fa11e265fa2562fb713171b6a58e72bb7afd276 Mon Sep 17 00:00:00 2001
From: Arne Jansen <sensille@gmx.net>
Date: Mon, 6 Aug 2012 14:18:51 -0600
Subject: Btrfs: fix deadlock in wait_for_more_refs

Commit a168650c introduced a waiting mechanism to prevent busy waiting in
btrfs_run_delayed_refs. This can deadlock with btrfs_run_ordered_operations,
where a tree_mod_seq is held while waiting for the io to complete, while
the end_io calls btrfs_run_delayed_refs.
This whole mechanism is unnecessary. If not enough runnable refs are
available to satisfy count, just return as count is more like a guideline
than a strict requirement.
In case we have to run all refs, commit transaction makes sure that no
other threads are working in the transaction anymore, so we just assert
here that no refs are blocked.

Signed-off-by: Arne Jansen <sensille@gmx.net>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
---
 fs/btrfs/ctree.c       |  6 ----
 fs/btrfs/ctree.h       |  1 -
 fs/btrfs/delayed-ref.c |  8 ------
 fs/btrfs/disk-io.c     |  2 --
 fs/btrfs/extent-tree.c | 77 ++++++++++++++------------------------------------
 5 files changed, 21 insertions(+), 73 deletions(-)

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 9d7621f271ff..08e0b11ba0a1 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -420,12 +420,6 @@ void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
 	}
 	spin_unlock(&fs_info->tree_mod_seq_lock);
 
-	/*
-	 * we removed the lowest blocker from the blocker list, so there may be
-	 * more processible delayed refs.
-	 */
-	wake_up(&fs_info->tree_mod_seq_wait);
-
 	/*
 	 * anything that's lower than the lowest existing (read: blocked)
 	 * sequence number can be removed from the tree.
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 348196350bf0..c38734a07a65 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -1252,7 +1252,6 @@ struct btrfs_fs_info {
 	atomic_t tree_mod_seq;
 	struct list_head tree_mod_seq_list;
 	struct seq_list tree_mod_seq_elem;
-	wait_queue_head_t tree_mod_seq_wait;
 
 	/* this protects tree_mod_log */
 	rwlock_t tree_mod_log_lock;
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index da7419ed01bb..7561431af50d 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -662,9 +662,6 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr,
 				   num_bytes, parent, ref_root, level, action,
 				   for_cow);
-	if (!need_ref_seq(for_cow, ref_root) &&
-	    waitqueue_active(&fs_info->tree_mod_seq_wait))
-		wake_up(&fs_info->tree_mod_seq_wait);
 	spin_unlock(&delayed_refs->lock);
 	if (need_ref_seq(for_cow, ref_root))
 		btrfs_qgroup_record_ref(trans, &ref->node, extent_op);
@@ -713,9 +710,6 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	add_delayed_data_ref(fs_info, trans, &ref->node, bytenr,
 				   num_bytes, parent, ref_root, owner, offset,
 				   action, for_cow);
-	if (!need_ref_seq(for_cow, ref_root) &&
-	    waitqueue_active(&fs_info->tree_mod_seq_wait))
-		wake_up(&fs_info->tree_mod_seq_wait);
 	spin_unlock(&delayed_refs->lock);
 	if (need_ref_seq(for_cow, ref_root))
 		btrfs_qgroup_record_ref(trans, &ref->node, extent_op);
@@ -744,8 +738,6 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 				   num_bytes, BTRFS_UPDATE_DELAYED_HEAD,
 				   extent_op->is_data);
 
-	if (waitqueue_active(&fs_info->tree_mod_seq_wait))
-		wake_up(&fs_info->tree_mod_seq_wait);
 	spin_unlock(&delayed_refs->lock);
 	return 0;
 }
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 502b20c56e84..a7ad8fc8dc53 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -2035,8 +2035,6 @@ int open_ctree(struct super_block *sb,
 	fs_info->free_chunk_space = 0;
 	fs_info->tree_mod_log = RB_ROOT;
 
-	init_waitqueue_head(&fs_info->tree_mod_seq_wait);
-
 	/* readahead state */
 	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_WAIT);
 	spin_lock_init(&fs_info->reada_lock);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 45c69c4184c9..d3df65f83b5c 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2318,12 +2318,6 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 		ref->in_tree = 0;
 		rb_erase(&ref->rb_node, &delayed_refs->root);
 		delayed_refs->num_entries--;
-		/*
-		 * we modified num_entries, but as we're currently running
-		 * delayed refs, skip
-		 *     wake_up(&delayed_refs->seq_wait);
-		 * here.
-		 */
 		spin_unlock(&delayed_refs->lock);
 
 		ret = run_one_delayed_ref(trans, root, ref, extent_op,
@@ -2350,22 +2344,6 @@ next:
 	return count;
 }
 
-static void wait_for_more_refs(struct btrfs_fs_info *fs_info,
-			       struct btrfs_delayed_ref_root *delayed_refs,
-			       unsigned long num_refs,
-			       struct list_head *first_seq)
-{
-	spin_unlock(&delayed_refs->lock);
-	pr_debug("waiting for more refs (num %ld, first %p)\n",
-		 num_refs, first_seq);
-	wait_event(fs_info->tree_mod_seq_wait,
-		   num_refs != delayed_refs->num_entries ||
-		   fs_info->tree_mod_seq_list.next != first_seq);
-	pr_debug("done waiting for more refs (num %ld, first %p)\n",
-		 delayed_refs->num_entries, fs_info->tree_mod_seq_list.next);
-	spin_lock(&delayed_refs->lock);
-}
-
 #ifdef SCRAMBLE_DELAYED_REFS
 /*
  * Normally delayed refs get processed in ascending bytenr order. This
@@ -2460,13 +2438,11 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 	struct btrfs_delayed_ref_root *delayed_refs;
 	struct btrfs_delayed_ref_node *ref;
 	struct list_head cluster;
-	struct list_head *first_seq = NULL;
 	int ret;
 	u64 delayed_start;
 	int run_all = count == (unsigned long)-1;
 	int run_most = 0;
-	unsigned long num_refs = 0;
-	int consider_waiting;
+	int loops;
 
 	/* We'll clean this up in btrfs_cleanup_transaction */
 	if (trans->aborted)
@@ -2484,7 +2460,7 @@ int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 	delayed_refs = &trans->transaction->delayed_refs;
 	INIT_LIST_HEAD(&cluster);
 again:
-	consider_waiting = 0;
+	loops = 0;
 	spin_lock(&delayed_refs->lock);
 
 #ifdef SCRAMBLE_DELAYED_REFS
@@ -2512,31 +2488,6 @@ again:
 		if (ret)
 			break;
 
-		if (delayed_start >= delayed_refs->run_delayed_start) {
-			if (consider_waiting == 0) {
-				/*
-				 * btrfs_find_ref_cluster looped. let's do one
-				 * more cycle. if we don't run any delayed ref
-				 * during that cycle (because we can't because
-				 * all of them are blocked) and if the number of
-				 * refs doesn't change, we avoid busy waiting.
-				 */
-				consider_waiting = 1;
-				num_refs = delayed_refs->num_entries;
-				first_seq = root->fs_info->tree_mod_seq_list.next;
-			} else {
-				wait_for_more_refs(root->fs_info, delayed_refs,
-						   num_refs, first_seq);
-				/*
-				 * after waiting, things have changed. we
-				 * dropped the lock and someone else might have
-				 * run some refs, built new clusters and so on.
-				 * therefore, we restart staleness detection.
-				 */
-				consider_waiting = 0;
-			}
-		}
-
 		ret = run_clustered_refs(trans, root, &cluster);
 		if (ret < 0) {
 			spin_unlock(&delayed_refs->lock);
@@ -2549,9 +2500,26 @@ again:
 		if (count == 0)
 			break;
 
-		if (ret || delayed_refs->run_delayed_start == 0) {
+		if (delayed_start >= delayed_refs->run_delayed_start) {
+			if (loops == 0) {
+				/*
+				 * btrfs_find_ref_cluster looped. let's do one
+				 * more cycle. if we don't run any delayed ref
+				 * during that cycle (because we can't because
+				 * all of them are blocked), bail out.
+				 */
+				loops = 1;
+			} else {
+				/*
+				 * no runnable refs left, stop trying
+				 */
+				BUG_ON(run_all);
+				break;
+			}
+		}
+		if (ret) {
 			/* refs were run, let's reset staleness detection */
-			consider_waiting = 0;
+			loops = 0;
 		}
 	}
 
@@ -5296,9 +5264,6 @@ static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
 	rb_erase(&head->node.rb_node, &delayed_refs->root);
 
 	delayed_refs->num_entries--;
-	smp_mb();
-	if (waitqueue_active(&root->fs_info->tree_mod_seq_wait))
-		wake_up(&root->fs_info->tree_mod_seq_wait);
 
 	/*
 	 * we don't take a ref on the node because we're removing it from the
-- 
cgit v1.2.3


From 66657b318e0e443ada229fccd40c8be86cfebdbf Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Wed, 1 Aug 2012 15:36:24 -0400
Subject: Btrfs: barrier before waitqueue_active

We need a barrir before calling waitqueue_active otherwise we will miss
wakeups.  So in places that do atomic_dec(); then atomic_read() use
atomic_dec_return() which imply a memory barrier (see memory-barriers.txt)
and then add an explicit memory barrier everywhere else that need them.
Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/compression.c   | 1 +
 fs/btrfs/delayed-inode.c | 7 +++----
 fs/btrfs/disk-io.c       | 7 ++++---
 fs/btrfs/inode.c         | 4 +---
 fs/btrfs/volumes.c       | 3 +--
 5 files changed, 10 insertions(+), 12 deletions(-)

diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 86eff48dab78..43d1c5a3a030 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -818,6 +818,7 @@ static void free_workspace(int type, struct list_head *workspace)
 	btrfs_compress_op[idx]->free_workspace(workspace);
 	atomic_dec(alloc_workspace);
 wake:
+	smp_mb();
 	if (waitqueue_active(workspace_wait))
 		wake_up(workspace_wait);
 }
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index 00deed4ef3ed..07d5eeb1e6f1 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -512,8 +512,8 @@ static void __btrfs_remove_delayed_item(struct btrfs_delayed_item *delayed_item)
 
 	rb_erase(&delayed_item->rb_node, root);
 	delayed_item->delayed_node->count--;
-	atomic_dec(&delayed_root->items);
-	if (atomic_read(&delayed_root->items) < BTRFS_DELAYED_BACKGROUND &&
+	if (atomic_dec_return(&delayed_root->items) <
+	    BTRFS_DELAYED_BACKGROUND &&
 	    waitqueue_active(&delayed_root->wait))
 		wake_up(&delayed_root->wait);
 }
@@ -1056,8 +1056,7 @@ static void btrfs_release_delayed_inode(struct btrfs_delayed_node *delayed_node)
 		delayed_node->count--;
 
 		delayed_root = delayed_node->root->fs_info->delayed_root;
-		atomic_dec(&delayed_root->items);
-		if (atomic_read(&delayed_root->items) <
+		if (atomic_dec_return(&delayed_root->items) <
 		    BTRFS_DELAYED_BACKGROUND &&
 		    waitqueue_active(&delayed_root->wait))
 			wake_up(&delayed_root->wait);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index a7ad8fc8dc53..dd86a5d88428 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -754,9 +754,7 @@ static void run_one_async_done(struct btrfs_work *work)
 	limit = btrfs_async_submit_limit(fs_info);
 	limit = limit * 2 / 3;
 
-	atomic_dec(&fs_info->nr_async_submits);
-
-	if (atomic_read(&fs_info->nr_async_submits) < limit &&
+	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
 	    waitqueue_active(&fs_info->async_submit_wait))
 		wake_up(&fs_info->async_submit_wait);
 
@@ -3783,14 +3781,17 @@ int btrfs_cleanup_transaction(struct btrfs_root *root)
 		/* FIXME: cleanup wait for commit */
 		t->in_commit = 1;
 		t->blocked = 1;
+		smp_mb();
 		if (waitqueue_active(&root->fs_info->transaction_blocked_wait))
 			wake_up(&root->fs_info->transaction_blocked_wait);
 
 		t->blocked = 0;
+		smp_mb();
 		if (waitqueue_active(&root->fs_info->transaction_wait))
 			wake_up(&root->fs_info->transaction_wait);
 
 		t->commit_done = 1;
+		smp_mb();
 		if (waitqueue_active(&t->commit_wait))
 			wake_up(&t->commit_wait);
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 2d65c52b0944..97baf00b40d1 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1007,9 +1007,7 @@ static noinline void async_cow_submit(struct btrfs_work *work)
 	nr_pages = (async_cow->end - async_cow->start + PAGE_CACHE_SIZE) >>
 		PAGE_CACHE_SHIFT;
 
-	atomic_sub(nr_pages, &root->fs_info->async_delalloc_pages);
-
-	if (atomic_read(&root->fs_info->async_delalloc_pages) <
+	if (atomic_sub_return(nr_pages, &root->fs_info->async_delalloc_pages) <
 	    5 * 1024 * 1024 &&
 	    waitqueue_active(&root->fs_info->async_submit_wait))
 		wake_up(&root->fs_info->async_submit_wait);
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 3b394503bd4e..0b1e69d380dd 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -227,9 +227,8 @@ loop_lock:
 		cur = pending;
 		pending = pending->bi_next;
 		cur->bi_next = NULL;
-		atomic_dec(&fs_info->nr_async_bios);
 
-		if (atomic_read(&fs_info->nr_async_bios) < limit &&
+		if (atomic_dec_return(&fs_info->nr_async_bios) < limit &&
 		    waitqueue_active(&fs_info->async_submit_wait))
 			wake_up(&fs_info->async_submit_wait);
 
-- 
cgit v1.2.3


From 6fc823b10f333313deb0b5d9069cbfd3a3f99f3a Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Mon, 6 Aug 2012 13:46:38 -0600
Subject: Btrfs: increase the size of the free space cache

Arne was complaining about the space cache having mismatching generation
numbers when debugging a deadlock.  This is because we can run out of space
in our preallocated range for our space cache if you have a pretty
fragmented amount of space in your pinned space.  So just increase the
amount of space we preallocate for space cache so we can be sure to have
enough space.  This will only really affect data ranges since their the only
chunks that end up larger than 256MB.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
---
 fs/btrfs/extent-tree.c | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index d3df65f83b5c..1bb408f737fb 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2975,17 +2975,16 @@ again:
 	}
 	spin_unlock(&block_group->lock);
 
-	num_pages = (int)div64_u64(block_group->key.offset, 1024 * 1024 * 1024);
+	/*
+	 * Try to preallocate enough space based on how big the block group is.
+	 * Keep in mind this has to include any pinned space which could end up
+	 * taking up quite a bit since it's not folded into the other space
+	 * cache.
+	 */
+	num_pages = (int)div64_u64(block_group->key.offset, 256 * 1024 * 1024);
 	if (!num_pages)
 		num_pages = 1;
 
-	/*
-	 * Just to make absolutely sure we have enough space, we're going to
-	 * preallocate 12 pages worth of space for each block group.  In
-	 * practice we ought to use at most 8, but we need extra space so we can
-	 * add our header and have a terminator between the extents and the
-	 * bitmaps.
-	 */
 	num_pages *= 16;
 	num_pages *= PAGE_CACHE_SIZE;
 
-- 
cgit v1.2.3


From b12a3b1ea209d9dec02731fba58c3dbe7d31cfd8 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@fusionio.com>
Date: Tue, 7 Aug 2012 15:34:49 -0400
Subject: Btrfs: don't run __tree_mod_log_free_eb on leaves

When we split a leaf, we may end up inserting a new root on top of that
leaf.  The reflog code was incorrectly assuming the old root was always
a node.  This makes sure we skip over leaves.

Signed-off-by: Chris Mason <chris.mason@fusionio.com>
---
 fs/btrfs/ctree.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 08e0b11ba0a1..6d183f60d63a 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -625,6 +625,9 @@ __tree_mod_log_free_eb(struct btrfs_fs_info *fs_info, struct extent_buffer *eb)
 	u32 nritems;
 	int ret;
 
+	if (btrfs_header_level(eb) == 0)
+		return;
+
 	nritems = btrfs_header_nritems(eb);
 	for (i = nritems - 1; i >= 0; i--) {
 		ret = tree_mod_log_insert_key_locked(fs_info, eb, i,
-- 
cgit v1.2.3


From 22cd2e7de7b0bd68fb668d23e1564707ca689510 Mon Sep 17 00:00:00 2001
From: Arne Jansen <sensille@gmx.net>
Date: Thu, 9 Aug 2012 00:16:53 -0600
Subject: Btrfs: fix race in run_clustered_refs

With commit

commit d1270cd91f308c9d22b2804720c36ccd32dbc35e
Author: Arne Jansen <sensille@gmx.net>
Date:   Tue Sep 13 15:16:43 2011 +0200

     Btrfs: put back delayed refs that are too new

I added a window where the delayed_ref's head->ref_mod code can diverge
from the sum of the remaining refs, because we release the head->mutex
in the middle. This leads to btrfs_lookup_extent_info returning wrong
numbers. This patch fixes this by adjusting the head's ref_mod with each
delayed ref we run.

Signed-off-by: Arne Jansen <sensille@gmx.net>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
---
 fs/btrfs/extent-tree.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 1bb408f737fb..f16411d3c252 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2318,6 +2318,23 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 		ref->in_tree = 0;
 		rb_erase(&ref->rb_node, &delayed_refs->root);
 		delayed_refs->num_entries--;
+		if (locked_ref) {
+			/*
+			 * when we play the delayed ref, also correct the
+			 * ref_mod on head
+			 */
+			switch (ref->action) {
+			case BTRFS_ADD_DELAYED_REF:
+			case BTRFS_ADD_DELAYED_EXTENT:
+				locked_ref->node.ref_mod -= ref->ref_mod;
+				break;
+			case BTRFS_DROP_DELAYED_REF:
+				locked_ref->node.ref_mod += ref->ref_mod;
+				break;
+			default:
+				WARN_ON(1);
+			}
+		}
 		spin_unlock(&delayed_refs->lock);
 
 		ret = run_one_delayed_ref(trans, root, ref, extent_op,
-- 
cgit v1.2.3


From c0f62dedd04ae0f3b8a18079db5a015af24e416f Mon Sep 17 00:00:00 2001
From: Miao Xie <miaox@cn.fujitsu.com>
Date: Wed, 8 Aug 2012 21:39:36 -0600
Subject: Btrfs: fix wrong mtime and ctime when creating snapshots

When we created a new snapshot, the mtime and ctime of its parent directory
were not updated. Fix it.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
---
 fs/btrfs/transaction.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 7208ada41e0e..3ee8d58e97ad 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -1026,6 +1026,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 
 	btrfs_i_size_write(parent_inode, parent_inode->i_size +
 					 dentry->d_name.len * 2);
+	parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
 	ret = btrfs_update_inode(trans, parent_root, parent_inode);
 	if (ret)
 		goto abort_trans_dput;
-- 
cgit v1.2.3


From 5a24e84c55f57cc49bd1cab531b6ef28b6b7bdaa Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Wed, 8 Aug 2012 10:12:59 -0600
Subject: Btrfs: fix enospc problems when deleting a subvol

Subvol delete is a special kind of awful where we use the global reserve to
cover the ENOSPC requirements.  The problem is once we're done removing
everything we do a btrfs_update_inode(), which by default will try to do the
delayed update stuff which will use it's own reserve.  There will be no
space in this reserve and we'll return ENOSPC.  So instead use
btrfs_update_inode_fallback() which will just fallback to updating the inode
item in the case of enospc.  This is fine because the global reserve covers
the space requirements for this.  With this patch I can now delete a subvol
on a problem image Dave Sterba sent me.  Thanks,

Reported-by: David Sterba <dave@jikos.cz>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
---
 fs/btrfs/inode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 97baf00b40d1..0808f483dafa 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -3171,7 +3171,7 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 	btrfs_i_size_write(dir, dir->i_size - name_len * 2);
 	inode_inc_iversion(dir);
 	dir->i_mtime = dir->i_ctime = CURRENT_TIME;
-	ret = btrfs_update_inode(trans, root, dir);
+	ret = btrfs_update_inode_fallback(trans, root, dir);
 	if (ret)
 		btrfs_abort_transaction(trans, root, ret);
 out:
-- 
cgit v1.2.3


From ae1e206b806ccc490dadff59af8a7a2477b32884 Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Tue, 7 Aug 2012 16:00:32 -0400
Subject: Btrfs: allow delayed refs to be merged

Daniel Blueman reported a bug with fio+balance on a ramdisk setup.
Basically what happens is the balance relocates a tree block which will drop
the implicit refs for all of its children and adds a full backref.  Once the
block is relocated we have to add the implicit refs back, so when we cow the
block again we add the implicit refs for its children back.  The problem
comes when the original drop ref doesn't get run before we add the implicit
refs back.  The delayed ref stuff will specifically prefer ADD operations
over DROP to keep us from freeing up an extent that will have references to
it, so we try to add the implicit ref before it is actually removed and we
panic.  This worked fine before because the add would have just canceled the
drop out and we would have been fine.  But the backref walking work needs to
be able to freeze the delayed ref stuff in time so we have this ever
increasing sequence number that gets attached to all new delayed ref updates
which makes us not merge refs and we run into this issue.

So to fix this we need to merge delayed refs.  So everytime we run a
clustered ref we need to try and merge all of its delayed refs.  The backref
walking stuff locks the delayed ref head before processing, so if we have it
locked we are safe to merge any refs inside of the sequence number.  If
there is no sequence number we can merge all refs.  Doing this not only
fixes our bug but keeps the delayed ref code from adding and removing
useless refs and batching together multiple refs into one search instead of
one search per delayed ref, which will really help our commit times.  I ran
this with Daniels test and 276 and I haven't seen any problems.  Thanks,

Reported-by: Daniel J Blueman <daniel@quora.org>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
---
 fs/btrfs/delayed-ref.c | 155 ++++++++++++++++++++++++++++++++++++++++---------
 fs/btrfs/delayed-ref.h |   4 ++
 fs/btrfs/extent-tree.c |  10 ++++
 3 files changed, 142 insertions(+), 27 deletions(-)

diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index 7561431af50d..ae9411773397 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -38,17 +38,14 @@
 static int comp_tree_refs(struct btrfs_delayed_tree_ref *ref2,
 			  struct btrfs_delayed_tree_ref *ref1)
 {
-	if (ref1->node.type == BTRFS_TREE_BLOCK_REF_KEY) {
-		if (ref1->root < ref2->root)
-			return -1;
-		if (ref1->root > ref2->root)
-			return 1;
-	} else {
-		if (ref1->parent < ref2->parent)
-			return -1;
-		if (ref1->parent > ref2->parent)
-			return 1;
-	}
+	if (ref1->root < ref2->root)
+		return -1;
+	if (ref1->root > ref2->root)
+		return 1;
+	if (ref1->parent < ref2->parent)
+		return -1;
+	if (ref1->parent > ref2->parent)
+		return 1;
 	return 0;
 }
 
@@ -85,7 +82,8 @@ static int comp_data_refs(struct btrfs_delayed_data_ref *ref2,
  * type of the delayed backrefs and content of delayed backrefs.
  */
 static int comp_entry(struct btrfs_delayed_ref_node *ref2,
-		      struct btrfs_delayed_ref_node *ref1)
+		      struct btrfs_delayed_ref_node *ref1,
+		      bool compare_seq)
 {
 	if (ref1->bytenr < ref2->bytenr)
 		return -1;
@@ -102,10 +100,12 @@ static int comp_entry(struct btrfs_delayed_ref_node *ref2,
 	if (ref1->type > ref2->type)
 		return 1;
 	/* merging of sequenced refs is not allowed */
-	if (ref1->seq < ref2->seq)
-		return -1;
-	if (ref1->seq > ref2->seq)
-		return 1;
+	if (compare_seq) {
+		if (ref1->seq < ref2->seq)
+			return -1;
+		if (ref1->seq > ref2->seq)
+			return 1;
+	}
 	if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY ||
 	    ref1->type == BTRFS_SHARED_BLOCK_REF_KEY) {
 		return comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref2),
@@ -139,7 +139,7 @@ static struct btrfs_delayed_ref_node *tree_insert(struct rb_root *root,
 		entry = rb_entry(parent_node, struct btrfs_delayed_ref_node,
 				 rb_node);
 
-		cmp = comp_entry(entry, ins);
+		cmp = comp_entry(entry, ins, 1);
 		if (cmp < 0)
 			p = &(*p)->rb_left;
 		else if (cmp > 0)
@@ -233,6 +233,114 @@ int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
+static void inline drop_delayed_ref(struct btrfs_trans_handle *trans,
+				    struct btrfs_delayed_ref_root *delayed_refs,
+				    struct btrfs_delayed_ref_node *ref)
+{
+	rb_erase(&ref->rb_node, &delayed_refs->root);
+	ref->in_tree = 0;
+	btrfs_put_delayed_ref(ref);
+	delayed_refs->num_entries--;
+	if (trans->delayed_ref_updates)
+		trans->delayed_ref_updates--;
+}
+
+static int merge_ref(struct btrfs_trans_handle *trans,
+		     struct btrfs_delayed_ref_root *delayed_refs,
+		     struct btrfs_delayed_ref_node *ref, u64 seq)
+{
+	struct rb_node *node;
+	int merged = 0;
+	int mod = 0;
+	int done = 0;
+
+	node = rb_prev(&ref->rb_node);
+	while (node) {
+		struct btrfs_delayed_ref_node *next;
+
+		next = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
+		node = rb_prev(node);
+		if (next->bytenr != ref->bytenr)
+			break;
+		if (seq && next->seq >= seq)
+			break;
+		if (comp_entry(ref, next, 0))
+			continue;
+
+		if (ref->action == next->action) {
+			mod = next->ref_mod;
+		} else {
+			if (ref->ref_mod < next->ref_mod) {
+				struct btrfs_delayed_ref_node *tmp;
+
+				tmp = ref;
+				ref = next;
+				next = tmp;
+				done = 1;
+			}
+			mod = -next->ref_mod;
+		}
+
+		merged++;
+		drop_delayed_ref(trans, delayed_refs, next);
+		ref->ref_mod += mod;
+		if (ref->ref_mod == 0) {
+			drop_delayed_ref(trans, delayed_refs, ref);
+			break;
+		} else {
+			/*
+			 * You can't have multiples of the same ref on a tree
+			 * block.
+			 */
+			WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY ||
+				ref->type == BTRFS_SHARED_BLOCK_REF_KEY);
+		}
+
+		if (done)
+			break;
+		node = rb_prev(&ref->rb_node);
+	}
+
+	return merged;
+}
+
+void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
+			      struct btrfs_fs_info *fs_info,
+			      struct btrfs_delayed_ref_root *delayed_refs,
+			      struct btrfs_delayed_ref_head *head)
+{
+	struct rb_node *node;
+	u64 seq = 0;
+
+	spin_lock(&fs_info->tree_mod_seq_lock);
+	if (!list_empty(&fs_info->tree_mod_seq_list)) {
+		struct seq_list *elem;
+
+		elem = list_first_entry(&fs_info->tree_mod_seq_list,
+					struct seq_list, list);
+		seq = elem->seq;
+	}
+	spin_unlock(&fs_info->tree_mod_seq_lock);
+
+	node = rb_prev(&head->node.rb_node);
+	while (node) {
+		struct btrfs_delayed_ref_node *ref;
+
+		ref = rb_entry(node, struct btrfs_delayed_ref_node,
+			       rb_node);
+		if (ref->bytenr != head->node.bytenr)
+			break;
+
+		/* We can't merge refs that are outside of our seq count */
+		if (seq && ref->seq >= seq)
+			break;
+		if (merge_ref(trans, delayed_refs, ref, seq))
+			node = rb_prev(&head->node.rb_node);
+		else
+			node = rb_prev(node);
+	}
+}
+
 int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info,
 			    struct btrfs_delayed_ref_root *delayed_refs,
 			    u64 seq)
@@ -336,18 +444,11 @@ update_existing_ref(struct btrfs_trans_handle *trans,
 		 * every changing the extent allocation tree.
 		 */
 		existing->ref_mod--;
-		if (existing->ref_mod == 0) {
-			rb_erase(&existing->rb_node,
-				 &delayed_refs->root);
-			existing->in_tree = 0;
-			btrfs_put_delayed_ref(existing);
-			delayed_refs->num_entries--;
-			if (trans->delayed_ref_updates)
-				trans->delayed_ref_updates--;
-		} else {
+		if (existing->ref_mod == 0)
+			drop_delayed_ref(trans, delayed_refs, existing);
+		else
 			WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY ||
 				existing->type == BTRFS_SHARED_BLOCK_REF_KEY);
-		}
 	} else {
 		WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY ||
 			existing->type == BTRFS_SHARED_BLOCK_REF_KEY);
diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
index 0d7c90c366b6..ab5300595847 100644
--- a/fs/btrfs/delayed-ref.h
+++ b/fs/btrfs/delayed-ref.h
@@ -167,6 +167,10 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 				struct btrfs_trans_handle *trans,
 				u64 bytenr, u64 num_bytes,
 				struct btrfs_delayed_extent_op *extent_op);
+void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
+			      struct btrfs_fs_info *fs_info,
+			      struct btrfs_delayed_ref_root *delayed_refs,
+			      struct btrfs_delayed_ref_head *head);
 
 struct btrfs_delayed_ref_head *
 btrfs_find_delayed_ref_head(struct btrfs_trans_handle *trans, u64 bytenr);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index f16411d3c252..ba58024d40d3 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2251,6 +2251,16 @@ static noinline int run_clustered_refs(struct btrfs_trans_handle *trans,
 			}
 		}
 
+		/*
+		 * We need to try and merge add/drops of the same ref since we
+		 * can run into issues with relocate dropping the implicit ref
+		 * and then it being added back again before the drop can
+		 * finish.  If we merged anything we need to re-loop so we can
+		 * get a good ref.
+		 */
+		btrfs_merge_delayed_refs(trans, fs_info, delayed_refs,
+					 locked_ref);
+
 		/*
 		 * locked_ref is the head node, so we have to go one
 		 * node back for any delayed ref updates
-- 
cgit v1.2.3


From 68ce9682a4bb95d6be5529cb57214bf2a1b7d20e Mon Sep 17 00:00:00 2001
From: Stefan Behrens <sbehrens@giantdisaster.de>
Date: Wed, 1 Aug 2012 05:45:52 -0600
Subject: Btrfs: remove superblock writing after fatal error

With commit acce952b0, btrfs was changed to flag the filesystem with
BTRFS_SUPER_FLAG_ERROR and switch to read-only mode after a fatal
error happened like a write I/O errors of all mirrors.
In such situations, on unmount, the superblock is written in
btrfs_error_commit_super(). This is done with the intention to be able
to evaluate the error flag on the next mount. A warning is printed
in this case during the next mount and the log tree is ignored.

The issue is that it is possible that the superblock points to a root
that was not written (due to write I/O errors).
The result is that the filesystem cannot be mounted. btrfsck also does
not start and all the other btrfs-progs tools fail to start as well.
However, mount -o recovery is working well and does the right things
to recover the filesystem (i.e., don't use the log root, clear the
free space cache and use the next mountable root that is stored in the
root backup array).

This patch removes the writing of the superblock when
BTRFS_SUPER_FLAG_ERROR is set, and removes the handling of the error
flag in the mount function.

These lines can be used to reproduce the issue (using /dev/sdm):
SCRATCH_DEV=/dev/sdm
SCRATCH_MNT=/mnt
echo 0 25165824 linear $SCRATCH_DEV 0 | dmsetup create foo
ls -alLF /dev/mapper/foo
mkfs.btrfs /dev/mapper/foo
mount /dev/mapper/foo $SCRATCH_MNT
echo bar > $SCRATCH_MNT/foo
sync
echo 0 25165824 error | dmsetup reload foo
dmsetup resume foo
ls -alF $SCRATCH_MNT
touch $SCRATCH_MNT/1
ls -alF $SCRATCH_MNT
sleep 35
echo 0 25165824 linear $SCRATCH_DEV 0 | dmsetup reload foo
dmsetup resume foo
sleep 1
umount $SCRATCH_MNT
btrfsck /dev/mapper/foo
dmsetup remove foo

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
---
 fs/btrfs/disk-io.c | 36 ++++--------------------------------
 fs/btrfs/disk-io.h |  2 +-
 2 files changed, 5 insertions(+), 33 deletions(-)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index dd86a5d88428..3c4c4397f470 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -2527,8 +2527,7 @@ retry_root_backup:
 		goto fail_trans_kthread;
 
 	/* do not make disk changes in broken FS */
-	if (btrfs_super_log_root(disk_super) != 0 &&
-	    !(fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)) {
+	if (btrfs_super_log_root(disk_super) != 0) {
 		u64 bytenr = btrfs_super_log_root(disk_super);
 
 		if (fs_devices->rw_devices == 0) {
@@ -3188,30 +3187,14 @@ int close_ctree(struct btrfs_root *root)
 	/* clear out the rbtree of defraggable inodes */
 	btrfs_run_defrag_inodes(fs_info);
 
-	/*
-	 * Here come 2 situations when btrfs is broken to flip readonly:
-	 *
-	 * 1. when btrfs flips readonly somewhere else before
-	 * btrfs_commit_super, sb->s_flags has MS_RDONLY flag,
-	 * and btrfs will skip to write sb directly to keep
-	 * ERROR state on disk.
-	 *
-	 * 2. when btrfs flips readonly just in btrfs_commit_super,
-	 * and in such case, btrfs cannot write sb via btrfs_commit_super,
-	 * and since fs_state has been set BTRFS_SUPER_FLAG_ERROR flag,
-	 * btrfs will cleanup all FS resources first and write sb then.
-	 */
 	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
 		ret = btrfs_commit_super(root);
 		if (ret)
 			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
 	}
 
-	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
-		ret = btrfs_error_commit_super(root);
-		if (ret)
-			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
-	}
+	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
+		btrfs_error_commit_super(root);
 
 	btrfs_put_block_group_cache(fs_info);
 
@@ -3433,18 +3416,11 @@ static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
 	if (read_only)
 		return 0;
 
-	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
-		printk(KERN_WARNING "warning: mount fs with errors, "
-		       "running btrfsck is recommended\n");
-	}
-
 	return 0;
 }
 
-int btrfs_error_commit_super(struct btrfs_root *root)
+void btrfs_error_commit_super(struct btrfs_root *root)
 {
-	int ret;
-
 	mutex_lock(&root->fs_info->cleaner_mutex);
 	btrfs_run_delayed_iputs(root);
 	mutex_unlock(&root->fs_info->cleaner_mutex);
@@ -3454,10 +3430,6 @@ int btrfs_error_commit_super(struct btrfs_root *root)
 
 	/* cleanup FS via transaction */
 	btrfs_cleanup_transaction(root);
-
-	ret = write_ctree_super(NULL, root, 0);
-
-	return ret;
 }
 
 static void btrfs_destroy_ordered_operations(struct btrfs_root *root)
diff --git a/fs/btrfs/disk-io.h b/fs/btrfs/disk-io.h
index 95e147eea239..c5b00a735fef 100644
--- a/fs/btrfs/disk-io.h
+++ b/fs/btrfs/disk-io.h
@@ -54,7 +54,7 @@ int write_ctree_super(struct btrfs_trans_handle *trans,
 		      struct btrfs_root *root, int max_mirrors);
 struct buffer_head *btrfs_read_dev_super(struct block_device *bdev);
 int btrfs_commit_super(struct btrfs_root *root);
-int btrfs_error_commit_super(struct btrfs_root *root);
+void btrfs_error_commit_super(struct btrfs_root *root);
 struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
 					    u64 bytenr, u32 blocksize);
 struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
-- 
cgit v1.2.3


From 5ee0844d6427e7338e0aba748f62b62d07ea2ed0 Mon Sep 17 00:00:00 2001
From: Stefan Behrens <sbehrens@giantdisaster.de>
Date: Mon, 27 Aug 2012 08:30:03 -0600
Subject: Btrfs: revert checksum error statistic which can cause a BUG()

Commit 442a4f6308e694e0fa6025708bd5e4e424bbf51c added btrfs device
statistic counters for detected IO and checksum errors to Linux 3.5.
The statistic part that counts checksum errors in
end_bio_extent_readpage() can cause a BUG() in a subfunction:
"kernel BUG at fs/btrfs/volumes.c:3762!"
That part is reverted with the current patch.
However, the counting of checksum errors in the scrub context remains
active, and the counting of detected IO errors (read, write or flush
errors) in all contexts remains active.

Cc: stable <stable@vger.kernel.org> # 3.5
Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/extent_io.c | 17 ++---------------
 fs/btrfs/volumes.c   | 22 ----------------------
 fs/btrfs/volumes.h   |  2 --
 3 files changed, 2 insertions(+), 39 deletions(-)

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 3e7c9ed6505b..49085f2336d2 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2329,23 +2329,10 @@ static void end_bio_extent_readpage(struct bio *bio, int err)
 		if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) {
 			ret = tree->ops->readpage_end_io_hook(page, start, end,
 							      state, mirror);
-			if (ret) {
-				/* no IO indicated but software detected errors
-				 * in the block, either checksum errors or
-				 * issues with the contents */
-				struct btrfs_root *root =
-					BTRFS_I(page->mapping->host)->root;
-				struct btrfs_device *device;
-
+			if (ret)
 				uptodate = 0;
-				device = btrfs_find_device_for_logical(
-						root, start, mirror);
-				if (device)
-					btrfs_dev_stat_inc_and_print(device,
-						BTRFS_DEV_STAT_CORRUPTION_ERRS);
-			} else {
+			else
 				clean_io_failure(start, page);
-			}
 		}
 
 		if (!uptodate && tree->ops && tree->ops->readpage_io_failed_hook) {
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 0b1e69d380dd..3f4e70e171ed 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -4610,28 +4610,6 @@ int btrfs_read_sys_array(struct btrfs_root *root)
 	return ret;
 }
 
-struct btrfs_device *btrfs_find_device_for_logical(struct btrfs_root *root,
-						   u64 logical, int mirror_num)
-{
-	struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
-	int ret;
-	u64 map_length = 0;
-	struct btrfs_bio *bbio = NULL;
-	struct btrfs_device *device;
-
-	BUG_ON(mirror_num == 0);
-	ret = btrfs_map_block(map_tree, WRITE, logical, &map_length, &bbio,
-			      mirror_num);
-	if (ret) {
-		BUG_ON(bbio != NULL);
-		return NULL;
-	}
-	BUG_ON(mirror_num != bbio->mirror_num);
-	device = bbio->stripes[mirror_num - 1].dev;
-	kfree(bbio);
-	return device;
-}
-
 int btrfs_read_chunk_tree(struct btrfs_root *root)
 {
 	struct btrfs_path *path;
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index 5479325987b3..53c06af92e8d 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -289,8 +289,6 @@ int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
 int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
 			 u64 *start, u64 *max_avail);
-struct btrfs_device *btrfs_find_device_for_logical(struct btrfs_root *root,
-						   u64 logical, int mirror_num);
 void btrfs_dev_stat_print_on_error(struct btrfs_device *device);
 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
 int btrfs_get_dev_stats(struct btrfs_root *root,
-- 
cgit v1.2.3


From bd7de2c9a449e26a5493d918618eb20ae60d56bd Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@fusionio.com>
Date: Fri, 24 Aug 2012 12:53:03 -0600
Subject: Btrfs: fix deadlock with freeze and sync V2

We can deadlock with freeze right now because we unconditionally start a
transaction in our ->sync_fs() call.  To fix this just check and see if we
have a running transaction to commit.  This saves us from the deadlock
because at this point we'll have the umount sem for the sb so we're safe
from freezes coming in after we've done our check.  With this patch the
freeze xfstests no longer deadlocks.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/super.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 2e06f124f284..073c2368f459 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -813,7 +813,6 @@ int btrfs_sync_fs(struct super_block *sb, int wait)
 	struct btrfs_trans_handle *trans;
 	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
 	struct btrfs_root *root = fs_info->tree_root;
-	int ret;
 
 	trace_btrfs_sync_fs(wait);
 
@@ -824,11 +823,17 @@ int btrfs_sync_fs(struct super_block *sb, int wait)
 
 	btrfs_wait_ordered_extents(root, 0, 0);
 
-	trans = btrfs_start_transaction(root, 0);
+	spin_lock(&fs_info->trans_lock);
+	if (!fs_info->running_transaction) {
+		spin_unlock(&fs_info->trans_lock);
+		return 0;
+	}
+	spin_unlock(&fs_info->trans_lock);
+
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
-	ret = btrfs_commit_transaction(trans, root);
-	return ret;
+	return btrfs_commit_transaction(trans, root);
 }
 
 static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
-- 
cgit v1.2.3


From 24c03fa5cf3d02c327cf9f2fc39f72664b1bd7e1 Mon Sep 17 00:00:00 2001
From: Liu Bo <bo.li.liu@oracle.com>
Date: Wed, 22 Aug 2012 20:10:38 -0600
Subject: Btrfs: fix a dio write regression

This bug is introduced by commit 3b8bde746f6f9bd36a9f05f5f3b6e334318176a9
(Btrfs: lock extents as we map them in DIO).

In dio write, we should unlock the section which we didn't do IO on in case that
we fall back to buffered write.  But we need to not only unlock the section
but also cleanup reserved space for the section.

This bug was found while running xfstests 133, with this 133 no longer complains.

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/inode.c | 24 ++++++++++++++++++++----
 1 file changed, 20 insertions(+), 4 deletions(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 0808f483dafa..38cda78de5e4 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -5992,11 +5992,27 @@ unlock:
 	 * in the case of read we need to unlock only the end area that we
 	 * aren't using if there is any left over space.
 	 */
-	if (lockstart < lockend)
-		clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend,
-				 unlock_bits, 1, 0, &cached_state, GFP_NOFS);
-	else
+	if (lockstart < lockend) {
+		if (create && len < lockend - lockstart) {
+			clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
+					 lockstart + len - 1, unlock_bits, 1, 0,
+					 &cached_state, GFP_NOFS);
+			/*
+			 * Beside unlock, we also need to cleanup reserved space
+			 * for the left range by attaching EXTENT_DO_ACCOUNTING.
+			 */
+			clear_extent_bit(&BTRFS_I(inode)->io_tree,
+					 lockstart + len, lockend,
+					 unlock_bits | EXTENT_DO_ACCOUNTING,
+					 1, 0, NULL, GFP_NOFS);
+		} else {
+			clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
+					 lockend, unlock_bits, 1, 0,
+					 &cached_state, GFP_NOFS);
+		}
+	} else {
 		free_extent_state(cached_state);
+	}
 
 	free_extent_map(em);
 
-- 
cgit v1.2.3


From d280e5be940931c84bb2e9831ead9d02bc785484 Mon Sep 17 00:00:00 2001
From: Liu Bo <bo.li.liu@oracle.com>
Date: Tue, 21 Aug 2012 21:13:25 -0600
Subject: Btrfs: fix ordered extent leak when failing to start a transaction

We cannot just return error before freeing ordered extent and releasing reserved
space when we fail to start a transacion.

Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/inode.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 38cda78de5e4..6ba80b902877 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1882,8 +1882,11 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 				trans = btrfs_join_transaction_nolock(root);
 			else
 				trans = btrfs_join_transaction(root);
-			if (IS_ERR(trans))
-				return PTR_ERR(trans);
+			if (IS_ERR(trans)) {
+				ret = PTR_ERR(trans);
+				trans = NULL;
+				goto out;
+			}
 			trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 			ret = btrfs_update_inode_fallback(trans, root, inode);
 			if (ret) /* -ENOMEM or corruption */
-- 
cgit v1.2.3


From 256dd1bb3750ac5ad49b40887c1691788dc44b33 Mon Sep 17 00:00:00 2001
From: Stefan Behrens <sbehrens@giantdisaster.de>
Date: Fri, 10 Aug 2012 08:58:21 -0600
Subject: Btrfs: fix that repair code is spuriously executed for transid
 failures

If verify_parent_transid() fails for all mirrors, the current code
calls repair_io_failure() anyway which means:
- that the disk block is rewritten without repairing anything and
- that a kernel log message is printed which misleadingly claims
  that a read error was corrected.

This is an example:
parent transid verify failed on 615015833600 wanted 110423 found 110424
parent transid verify failed on 615015833600 wanted 110423 found 110424
btrfs read error corrected: ino 1 off 615015833600 (dev /dev/...)

It is wrong to ignore the results from verify_parent_transid() and to
call repair_eb_io_failure() when the verification of the transids failed.
This commit fixes the issue.

Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/disk-io.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 3c4c4397f470..29c69e60d3b0 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -377,9 +377,13 @@ static int btree_read_extent_buffer_pages(struct btrfs_root *root,
 		ret = read_extent_buffer_pages(io_tree, eb, start,
 					       WAIT_COMPLETE,
 					       btree_get_extent, mirror_num);
-		if (!ret && !verify_parent_transid(io_tree, eb,
+		if (!ret) {
+			if (!verify_parent_transid(io_tree, eb,
 						   parent_transid, 0))
-			break;
+				break;
+			else
+				ret = -EIO;
+		}
 
 		/*
 		 * This buffer's crc is fine, but its contents are corrupted, so
-- 
cgit v1.2.3