From 74b2107543da4ed9607ec484f63c42362dc9fca6 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 13 Apr 2011 12:02:53 -0400
Subject: Btrfs: make sure to use the delalloc reserve when filling delalloc

In the prealloc filling code and compressed code we don't set trans->block_rsv
to the delalloc block reserve properly, which is going to make us use metadata
from the wrong pool, this patch fixes that.  Thanks,

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 7cd8ab0ef04d..3b9f1643aa57 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -619,6 +619,7 @@ retry:
 
 		trans = btrfs_join_transaction(root, 1);
 		BUG_ON(IS_ERR(trans));
+		trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 		ret = btrfs_reserve_extent(trans, root,
 					   async_extent->compressed_size,
 					   async_extent->compressed_size,
@@ -1060,6 +1061,7 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 		trans = btrfs_join_transaction(root, 1);
 	}
 	BUG_ON(IS_ERR(trans));
+	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 
 	cow_start = (u64)-1;
 	cur_offset = start;
-- 
cgit v1.2.3


From 7a7eaa40a39bde4eefc91aadeb1ce3dc4e6a1252 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 13 Apr 2011 12:54:33 -0400
Subject: Btrfs: take away the num_items argument from btrfs_join_transaction

I keep forgetting that btrfs_join_transaction() just ignores the num_items
argument, which leads me to sending pointless patches and looking stupid :).  So
just kill the num_items argument from btrfs_join_transaction and
btrfs_start_ioctl_transaction, since neither of them use it.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/disk-io.c     |  6 +++---
 fs/btrfs/extent-tree.c | 12 ++++++------
 fs/btrfs/inode.c       | 34 +++++++++++++++++-----------------
 fs/btrfs/ioctl.c       |  4 ++--
 fs/btrfs/relocation.c  | 12 ++++++------
 fs/btrfs/transaction.c | 13 +++++--------
 fs/btrfs/transaction.h |  9 +++------
 7 files changed, 42 insertions(+), 48 deletions(-)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 228cf36ece83..9d6c9e332ca3 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1568,7 +1568,7 @@ static int transaction_kthread(void *arg)
 		transid = cur->transid;
 		spin_unlock(&root->fs_info->new_trans_lock);
 
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 		BUG_ON(IS_ERR(trans));
 		if (transid == trans->transid) {
 			ret = btrfs_commit_transaction(trans, root);
@@ -2495,13 +2495,13 @@ int btrfs_commit_super(struct btrfs_root *root)
 	down_write(&root->fs_info->cleanup_work_sem);
 	up_write(&root->fs_info->cleanup_work_sem);
 
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 	ret = btrfs_commit_transaction(trans, root);
 	BUG_ON(ret);
 	/* run commit again to drop the original snapshot */
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 	btrfs_commit_transaction(trans, root);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 9ee6bd55e16c..941b28e78931 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3174,7 +3174,7 @@ again:
 			spin_unlock(&data_sinfo->lock);
 alloc:
 			alloc_target = btrfs_get_alloc_profile(root, 1);
-			trans = btrfs_join_transaction(root, 1);
+			trans = btrfs_join_transaction(root);
 			if (IS_ERR(trans))
 				return PTR_ERR(trans);
 
@@ -3202,7 +3202,7 @@ alloc:
 commit_trans:
 		if (!committed && !root->fs_info->open_ioctl_trans) {
 			committed = 1;
-			trans = btrfs_join_transaction(root, 1);
+			trans = btrfs_join_transaction(root);
 			if (IS_ERR(trans))
 				return PTR_ERR(trans);
 			ret = btrfs_commit_transaction(trans, root);
@@ -3589,7 +3589,7 @@ again:
 		goto out;
 
 	ret = -ENOSPC;
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
 		goto out;
 	ret = btrfs_commit_transaction(trans, root);
@@ -3816,7 +3816,7 @@ int btrfs_block_rsv_check(struct btrfs_trans_handle *trans,
 		if (trans)
 			return -EAGAIN;
 
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 		BUG_ON(IS_ERR(trans));
 		ret = btrfs_commit_transaction(trans, root);
 		return 0;
@@ -7649,7 +7649,7 @@ int btrfs_drop_dead_reloc_roots(struct btrfs_root *root)
 
 		BUG_ON(reloc_root->commit_root != NULL);
 		while (1) {
-			trans = btrfs_join_transaction(root, 1);
+			trans = btrfs_join_transaction(root);
 			BUG_ON(IS_ERR(trans));
 
 			mutex_lock(&root->fs_info->drop_mutex);
@@ -8176,7 +8176,7 @@ int btrfs_set_block_group_ro(struct btrfs_root *root,
 
 	BUG_ON(cache->ro);
 
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	BUG_ON(IS_ERR(trans));
 
 	alloc_flags = update_block_group_flags(root, cache->flags);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 3b9f1643aa57..e47bdf0fb75a 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -420,7 +420,7 @@ again:
 		}
 	}
 	if (start == 0) {
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 		BUG_ON(IS_ERR(trans));
 		btrfs_set_trans_block_group(trans, inode);
 		trans->block_rsv = &root->fs_info->delalloc_block_rsv;
@@ -617,7 +617,7 @@ retry:
 			    async_extent->start + async_extent->ram_size - 1,
 			    GFP_NOFS);
 
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 		BUG_ON(IS_ERR(trans));
 		trans->block_rsv = &root->fs_info->delalloc_block_rsv;
 		ret = btrfs_reserve_extent(trans, root,
@@ -779,7 +779,7 @@ static noinline int cow_file_range(struct inode *inode,
 	int ret = 0;
 
 	BUG_ON(root == root->fs_info->tree_root);
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	BUG_ON(IS_ERR(trans));
 	btrfs_set_trans_block_group(trans, inode);
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
@@ -1056,9 +1056,9 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 	BUG_ON(!path);
 	if (root == root->fs_info->tree_root) {
 		nolock = true;
-		trans = btrfs_join_transaction_nolock(root, 1);
+		trans = btrfs_join_transaction_nolock(root);
 	} else {
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 	}
 	BUG_ON(IS_ERR(trans));
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
@@ -1718,9 +1718,9 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 		ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
 		if (!ret) {
 			if (nolock)
-				trans = btrfs_join_transaction_nolock(root, 1);
+				trans = btrfs_join_transaction_nolock(root);
 			else
-				trans = btrfs_join_transaction(root, 1);
+				trans = btrfs_join_transaction(root);
 			BUG_ON(IS_ERR(trans));
 			btrfs_set_trans_block_group(trans, inode);
 			trans->block_rsv = &root->fs_info->delalloc_block_rsv;
@@ -1735,9 +1735,9 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 			 0, &cached_state, GFP_NOFS);
 
 	if (nolock)
-		trans = btrfs_join_transaction_nolock(root, 1);
+		trans = btrfs_join_transaction_nolock(root);
 	else
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 	BUG_ON(IS_ERR(trans));
 	btrfs_set_trans_block_group(trans, inode);
 	trans->block_rsv = &root->fs_info->delalloc_block_rsv;
@@ -2415,7 +2415,7 @@ int btrfs_orphan_cleanup(struct btrfs_root *root)
 					(u64)-1);
 
 	if (root->orphan_block_rsv || root->orphan_item_inserted) {
-		trans = btrfs_join_transaction(root, 1);
+		trans = btrfs_join_transaction(root);
 		if (!IS_ERR(trans))
 			btrfs_end_transaction(trans, root);
 	}
@@ -4378,9 +4378,9 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 
 	if (wbc->sync_mode == WB_SYNC_ALL) {
 		if (nolock)
-			trans = btrfs_join_transaction_nolock(root, 1);
+			trans = btrfs_join_transaction_nolock(root);
 		else
-			trans = btrfs_join_transaction(root, 1);
+			trans = btrfs_join_transaction(root);
 		if (IS_ERR(trans))
 			return PTR_ERR(trans);
 		btrfs_set_trans_block_group(trans, inode);
@@ -4407,7 +4407,7 @@ void btrfs_dirty_inode(struct inode *inode)
 	if (BTRFS_I(inode)->dummy_inode)
 		return;
 
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	BUG_ON(IS_ERR(trans));
 	btrfs_set_trans_block_group(trans, inode);
 
@@ -5226,7 +5226,7 @@ again:
 				free_extent_map(em);
 				em = NULL;
 				btrfs_release_path(root, path);
-				trans = btrfs_join_transaction(root, 1);
+				trans = btrfs_join_transaction(root);
 				if (IS_ERR(trans))
 					return ERR_CAST(trans);
 				goto again;
@@ -5470,7 +5470,7 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
 		btrfs_drop_extent_cache(inode, start, start + len - 1, 0);
 	}
 
-	trans = btrfs_join_transaction(root, 0);
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans))
 		return ERR_CAST(trans);
 
@@ -5703,7 +5703,7 @@ static int btrfs_get_blocks_direct(struct inode *inode, sector_t iblock,
 		 * to make sure the current transaction stays open
 		 * while we look for nocow cross refs
 		 */
-		trans = btrfs_join_transaction(root, 0);
+		trans = btrfs_join_transaction(root);
 		if (IS_ERR(trans))
 			goto must_cow;
 
@@ -5841,7 +5841,7 @@ again:
 
 	BUG_ON(!ordered);
 
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans)) {
 		err = -ENOMEM;
 		goto out;
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 2616f7ed4799..908c3d4b48c6 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -242,7 +242,7 @@ static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
 		ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
 	}
 
-	trans = btrfs_join_transaction(root, 1);
+	trans = btrfs_join_transaction(root);
 	BUG_ON(IS_ERR(trans));
 
 	ret = btrfs_update_inode(trans, root, inode);
@@ -2182,7 +2182,7 @@ static long btrfs_ioctl_trans_start(struct file *file)
 	mutex_unlock(&root->fs_info->trans_mutex);
 
 	ret = -ENOMEM;
-	trans = btrfs_start_ioctl_transaction(root, 0);
+	trans = btrfs_start_ioctl_transaction(root);
 	if (IS_ERR(trans))
 		goto out_drop;
 
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 199a80134312..8bb256667f2d 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -2149,7 +2149,7 @@ again:
 			err = ret;
 	}
 
-	trans = btrfs_join_transaction(rc->extent_root, 1);
+	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans)) {
 		if (!err)
 			btrfs_block_rsv_release(rc->extent_root,
@@ -3233,7 +3233,7 @@ truncate:
 		goto out;
 	}
 
-	trans = btrfs_join_transaction(root, 0);
+	trans = btrfs_join_transaction(root);
 	if (IS_ERR(trans)) {
 		btrfs_free_path(path);
 		ret = PTR_ERR(trans);
@@ -3642,7 +3642,7 @@ int prepare_to_relocate(struct reloc_control *rc)
 	rc->create_reloc_tree = 1;
 	set_reloc_control(rc);
 
-	trans = btrfs_join_transaction(rc->extent_root, 1);
+	trans = btrfs_join_transaction(rc->extent_root);
 	BUG_ON(IS_ERR(trans));
 	btrfs_commit_transaction(trans, rc->extent_root);
 	return 0;
@@ -3831,7 +3831,7 @@ restart:
 	btrfs_block_rsv_release(rc->extent_root, rc->block_rsv, (u64)-1);
 
 	/* get rid of pinned extents */
-	trans = btrfs_join_transaction(rc->extent_root, 1);
+	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans))
 		err = PTR_ERR(trans);
 	else
@@ -4156,7 +4156,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 
 	set_reloc_control(rc);
 
-	trans = btrfs_join_transaction(rc->extent_root, 1);
+	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans)) {
 		unset_reloc_control(rc);
 		err = PTR_ERR(trans);
@@ -4190,7 +4190,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 
 	unset_reloc_control(rc);
 
-	trans = btrfs_join_transaction(rc->extent_root, 1);
+	trans = btrfs_join_transaction(rc->extent_root);
 	if (IS_ERR(trans))
 		err = PTR_ERR(trans);
 	else
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index c571734d5e5a..70bfb26df967 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -257,22 +257,19 @@ struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 {
 	return start_transaction(root, num_items, TRANS_START);
 }
-struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
-						   int num_blocks)
+struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root)
 {
 	return start_transaction(root, 0, TRANS_JOIN);
 }
 
-struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root,
-							  int num_blocks)
+struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root)
 {
 	return start_transaction(root, 0, TRANS_JOIN_NOLOCK);
 }
 
-struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
-							 int num_blocks)
+struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root)
 {
-	return start_transaction(r, 0, TRANS_USERSPACE);
+	return start_transaction(root, 0, TRANS_USERSPACE);
 }
 
 /* wait for a transaction commit to be fully complete */
@@ -1171,7 +1168,7 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 
 	INIT_DELAYED_WORK(&ac->work, do_async_commit);
 	ac->root = root;
-	ac->newtrans = btrfs_join_transaction(root, 0);
+	ac->newtrans = btrfs_join_transaction(root);
 	if (IS_ERR(ac->newtrans)) {
 		int err = PTR_ERR(ac->newtrans);
 		kfree(ac);
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index e441acc6c584..1f573f09dba2 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -92,12 +92,9 @@ int btrfs_end_transaction_nolock(struct btrfs_trans_handle *trans,
 				 struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
 						   int num_items);
-struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root,
-						  int num_blocks);
-struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root,
-							  int num_blocks);
-struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
-							 int num_blocks);
+struct btrfs_trans_handle *btrfs_join_transaction(struct btrfs_root *root);
+struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root);
+struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *root);
 int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid);
 int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
 				     struct btrfs_root *root);
-- 
cgit v1.2.3


From 2a1eb4614d984d5cd4c928784e9afcf5c07f93be Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 13 Apr 2011 15:15:59 -0400
Subject: Btrfs: if we've already started a trans handle, use that one

We currently track trans handles in current->journal_info, but we don't actually
use it.  This patch fixes it.  This will cover the case where we have multiple
people starting transactions down the call chain.  This keeps us from having to
allocate a new handle and all of that, we just increase the use count of the
current handle, save the old block_rsv, and return.  I tested this with xfstests
and it worked out fine.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/transaction.c | 17 +++++++++++++++++
 fs/btrfs/transaction.h |  2 ++
 2 files changed, 19 insertions(+)

diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 70bfb26df967..46f40564c168 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -184,6 +184,15 @@ static struct btrfs_trans_handle *start_transaction(struct btrfs_root *root,
 
 	if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
 		return ERR_PTR(-EROFS);
+
+	if (current->journal_info) {
+		WARN_ON(type != TRANS_JOIN && type != TRANS_JOIN_NOLOCK);
+		h = current->journal_info;
+		h->use_count++;
+		h->orig_rsv = h->block_rsv;
+		h->block_rsv = NULL;
+		goto got_it;
+	}
 again:
 	h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
 	if (!h)
@@ -213,7 +222,9 @@ again:
 	h->block_group = 0;
 	h->bytes_reserved = 0;
 	h->delayed_ref_updates = 0;
+	h->use_count = 1;
 	h->block_rsv = NULL;
+	h->orig_rsv = NULL;
 
 	smp_mb();
 	if (cur_trans->blocked && may_wait_transaction(root, type)) {
@@ -241,6 +252,7 @@ again:
 		}
 	}
 
+got_it:
 	if (type != TRANS_JOIN_NOLOCK)
 		mutex_lock(&root->fs_info->trans_mutex);
 	record_root_in_trans(h, root);
@@ -428,6 +440,11 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	struct btrfs_fs_info *info = root->fs_info;
 	int count = 0;
 
+	if (--trans->use_count) {
+		trans->block_rsv = trans->orig_rsv;
+		return 0;
+	}
+
 	while (count < 4) {
 		unsigned long cur = trans->delayed_ref_updates;
 		trans->delayed_ref_updates = 0;
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 1f573f09dba2..154314f80f8d 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -47,11 +47,13 @@ struct btrfs_trans_handle {
 	u64 transid;
 	u64 block_group;
 	u64 bytes_reserved;
+	unsigned long use_count;
 	unsigned long blocks_reserved;
 	unsigned long blocks_used;
 	unsigned long delayed_ref_updates;
 	struct btrfs_transaction *transaction;
 	struct btrfs_block_rsv *block_rsv;
+	struct btrfs_block_rsv *orig_rsv;
 };
 
 struct btrfs_pending_snapshot {
-- 
cgit v1.2.3


From a4abeea41adfa3c143c289045f4625dfaeba2212 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Mon, 11 Apr 2011 17:25:13 -0400
Subject: Btrfs: kill trans_mutex

We use trans_mutex for lots of things, here's a basic list

1) To serialize trans_handles joining the currently running transaction
2) To make sure that no new trans handles are started while we are committing
3) To protect the dead_roots list and the transaction lists

Really the serializing trans_handles joining is not too hard, and can really get
bogged down in acquiring a reference to the transaction.  So replace the
trans_mutex with a trans_lock spinlock and use it to do the following

1) Protect fs_info->running_transaction.  All trans handles have to do is check
this, and then take a reference of the transaction and keep on going.
2) Protect the fs_info->trans_list.  This doesn't get used too much, basically
it just holds the current transactions, which will usually just be the currently
committing transaction and the currently running transaction at most.
3) Protect the dead roots list.  This is only ever processed by splicing the
list so this is relatively simple.
4) Protect the fs_info->reloc_ctl stuff.  This is very lightweight and was using
the trans_mutex before, so this is a pretty straightforward change.
5) Protect fs_info->no_trans_join.  Because we don't hold the trans_lock over
the entirety of the commit we need to have a way to block new people from
creating a new transaction while we're doing our work.  So we set no_trans_join
and in join_transaction we test to see if that is set, and if it is we do a
wait_on_commit.
6) Make the transaction use count atomic so we don't need to take locks to
modify it when we're dropping references.
7) Add a commit_lock to the transaction to make sure multiple people trying to
commit the same transaction don't race and commit at the same time.
8) Make open_ioctl_trans an atomic so we don't have to take any locks for ioctl
trans.

I have tested this with xfstests, but obviously it is a pretty hairy change so
lots of testing is greatly appreciated.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/ctree.h       |   6 +-
 fs/btrfs/disk-io.c     |  30 +++---
 fs/btrfs/extent-tree.c |   3 +-
 fs/btrfs/file.c        |   4 +-
 fs/btrfs/ioctl.c       |  12 +--
 fs/btrfs/relocation.c  |  16 +--
 fs/btrfs/transaction.c | 271 ++++++++++++++++++++++++++-----------------------
 fs/btrfs/transaction.h |   4 +-
 8 files changed, 177 insertions(+), 169 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 8f4b81de3ae2..522a39b0033d 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -919,7 +919,6 @@ struct btrfs_fs_info {
 	 * is required instead of the faster short fsync log commits
 	 */
 	u64 last_trans_log_full_commit;
-	u64 open_ioctl_trans;
 	unsigned long mount_opt:20;
 	unsigned long compress_type:4;
 	u64 max_inline;
@@ -936,7 +935,6 @@ struct btrfs_fs_info {
 	struct super_block *sb;
 	struct inode *btree_inode;
 	struct backing_dev_info bdi;
-	struct mutex trans_mutex;
 	struct mutex tree_log_mutex;
 	struct mutex transaction_kthread_mutex;
 	struct mutex cleaner_mutex;
@@ -957,6 +955,7 @@ struct btrfs_fs_info {
 	struct rw_semaphore subvol_sem;
 	struct srcu_struct subvol_srcu;
 
+	spinlock_t trans_lock;
 	struct list_head trans_list;
 	struct list_head hashers;
 	struct list_head dead_roots;
@@ -969,6 +968,7 @@ struct btrfs_fs_info {
 	atomic_t async_submit_draining;
 	atomic_t nr_async_bios;
 	atomic_t async_delalloc_pages;
+	atomic_t open_ioctl_trans;
 
 	/*
 	 * this is used by the balancing code to wait for all the pending
@@ -1032,6 +1032,7 @@ struct btrfs_fs_info {
 	int closing;
 	int log_root_recovering;
 	int enospc_unlink;
+	int trans_no_join;
 
 	u64 total_pinned;
 
@@ -1053,7 +1054,6 @@ struct btrfs_fs_info {
 	struct reloc_control *reloc_ctl;
 
 	spinlock_t delalloc_lock;
-	spinlock_t new_trans_lock;
 	u64 delalloc_bytes;
 
 	/* data_alloc_cluster is only used in ssd mode */
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 9d6c9e332ca3..93ef254ec432 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1551,22 +1551,22 @@ static int transaction_kthread(void *arg)
 		vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
 		mutex_lock(&root->fs_info->transaction_kthread_mutex);
 
-		spin_lock(&root->fs_info->new_trans_lock);
+		spin_lock(&root->fs_info->trans_lock);
 		cur = root->fs_info->running_transaction;
 		if (!cur) {
-			spin_unlock(&root->fs_info->new_trans_lock);
+			spin_unlock(&root->fs_info->trans_lock);
 			goto sleep;
 		}
 
 		now = get_seconds();
 		if (!cur->blocked &&
 		    (now < cur->start_time || now - cur->start_time < 30)) {
-			spin_unlock(&root->fs_info->new_trans_lock);
+			spin_unlock(&root->fs_info->trans_lock);
 			delay = HZ * 5;
 			goto sleep;
 		}
 		transid = cur->transid;
-		spin_unlock(&root->fs_info->new_trans_lock);
+		spin_unlock(&root->fs_info->trans_lock);
 
 		trans = btrfs_join_transaction(root);
 		BUG_ON(IS_ERR(trans));
@@ -1658,7 +1658,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	INIT_LIST_HEAD(&fs_info->ordered_operations);
 	INIT_LIST_HEAD(&fs_info->caching_block_groups);
 	spin_lock_init(&fs_info->delalloc_lock);
-	spin_lock_init(&fs_info->new_trans_lock);
+	spin_lock_init(&fs_info->trans_lock);
 	spin_lock_init(&fs_info->ref_cache_lock);
 	spin_lock_init(&fs_info->fs_roots_radix_lock);
 	spin_lock_init(&fs_info->delayed_iput_lock);
@@ -1687,6 +1687,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	fs_info->sb = sb;
 	fs_info->max_inline = 8192 * 1024;
 	fs_info->metadata_ratio = 0;
+	fs_info->trans_no_join = 0;
 
 	fs_info->thread_pool_size = min_t(unsigned long,
 					  num_online_cpus() + 2, 8);
@@ -1735,7 +1736,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	fs_info->do_barriers = 1;
 
 
-	mutex_init(&fs_info->trans_mutex);
 	mutex_init(&fs_info->ordered_operations_mutex);
 	mutex_init(&fs_info->tree_log_mutex);
 	mutex_init(&fs_info->chunk_mutex);
@@ -3006,10 +3006,13 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
 
 	WARN_ON(1);
 
-	mutex_lock(&root->fs_info->trans_mutex);
 	mutex_lock(&root->fs_info->transaction_kthread_mutex);
 
+	spin_lock(&root->fs_info->trans_lock);
 	list_splice_init(&root->fs_info->trans_list, &list);
+	root->fs_info->trans_no_join = 1;
+	spin_unlock(&root->fs_info->trans_lock);
+
 	while (!list_empty(&list)) {
 		t = list_entry(list.next, struct btrfs_transaction, list);
 		if (!t)
@@ -3034,23 +3037,18 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
 		t->blocked = 0;
 		if (waitqueue_active(&root->fs_info->transaction_wait))
 			wake_up(&root->fs_info->transaction_wait);
-		mutex_unlock(&root->fs_info->trans_mutex);
 
-		mutex_lock(&root->fs_info->trans_mutex);
 		t->commit_done = 1;
 		if (waitqueue_active(&t->commit_wait))
 			wake_up(&t->commit_wait);
-		mutex_unlock(&root->fs_info->trans_mutex);
-
-		mutex_lock(&root->fs_info->trans_mutex);
 
 		btrfs_destroy_pending_snapshots(t);
 
 		btrfs_destroy_delalloc_inodes(root);
 
-		spin_lock(&root->fs_info->new_trans_lock);
+		spin_lock(&root->fs_info->trans_lock);
 		root->fs_info->running_transaction = NULL;
-		spin_unlock(&root->fs_info->new_trans_lock);
+		spin_unlock(&root->fs_info->trans_lock);
 
 		btrfs_destroy_marked_extents(root, &t->dirty_pages,
 					     EXTENT_DIRTY);
@@ -3064,8 +3062,10 @@ static int btrfs_cleanup_transaction(struct btrfs_root *root)
 		kmem_cache_free(btrfs_transaction_cachep, t);
 	}
 
+	spin_lock(&root->fs_info->trans_lock);
+	root->fs_info->trans_no_join = 0;
+	spin_unlock(&root->fs_info->trans_lock);
 	mutex_unlock(&root->fs_info->transaction_kthread_mutex);
-	mutex_unlock(&root->fs_info->trans_mutex);
 
 	return 0;
 }
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 941b28e78931..ca599654ce19 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3200,7 +3200,8 @@ alloc:
 
 		/* commit the current transaction and try again */
 commit_trans:
-		if (!committed && !root->fs_info->open_ioctl_trans) {
+		if (!committed &&
+		    !atomic_read(&root->fs_info->open_ioctl_trans)) {
 			committed = 1;
 			trans = btrfs_join_transaction(root);
 			if (IS_ERR(trans))
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 75899a01dded..cd5e82e500cf 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1222,14 +1222,12 @@ int btrfs_sync_file(struct file *file, int datasync)
 	 * the current transaction, we can bail out now without any
 	 * syncing
 	 */
-	mutex_lock(&root->fs_info->trans_mutex);
+	smp_mb();
 	if (BTRFS_I(inode)->last_trans <=
 	    root->fs_info->last_trans_committed) {
 		BTRFS_I(inode)->last_trans = 0;
-		mutex_unlock(&root->fs_info->trans_mutex);
 		goto out;
 	}
-	mutex_unlock(&root->fs_info->trans_mutex);
 
 	/*
 	 * ok we haven't committed the transaction yet, lets do a commit
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 908c3d4b48c6..a578620e06a8 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -2177,9 +2177,7 @@ static long btrfs_ioctl_trans_start(struct file *file)
 	if (ret)
 		goto out;
 
-	mutex_lock(&root->fs_info->trans_mutex);
-	root->fs_info->open_ioctl_trans++;
-	mutex_unlock(&root->fs_info->trans_mutex);
+	atomic_inc(&root->fs_info->open_ioctl_trans);
 
 	ret = -ENOMEM;
 	trans = btrfs_start_ioctl_transaction(root);
@@ -2190,9 +2188,7 @@ static long btrfs_ioctl_trans_start(struct file *file)
 	return 0;
 
 out_drop:
-	mutex_lock(&root->fs_info->trans_mutex);
-	root->fs_info->open_ioctl_trans--;
-	mutex_unlock(&root->fs_info->trans_mutex);
+	atomic_dec(&root->fs_info->open_ioctl_trans);
 	mnt_drop_write(file->f_path.mnt);
 out:
 	return ret;
@@ -2426,9 +2422,7 @@ long btrfs_ioctl_trans_end(struct file *file)
 
 	btrfs_end_transaction(trans, root);
 
-	mutex_lock(&root->fs_info->trans_mutex);
-	root->fs_info->open_ioctl_trans--;
-	mutex_unlock(&root->fs_info->trans_mutex);
+	atomic_dec(&root->fs_info->open_ioctl_trans);
 
 	mnt_drop_write(file->f_path.mnt);
 	return 0;
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 8bb256667f2d..09c30d37d43e 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -2136,10 +2136,10 @@ int prepare_to_merge(struct reloc_control *rc, int err)
 	u64 num_bytes = 0;
 	int ret;
 
-	mutex_lock(&root->fs_info->trans_mutex);
+	spin_lock(&root->fs_info->trans_lock);
 	rc->merging_rsv_size += root->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
 	rc->merging_rsv_size += rc->nodes_relocated * 2;
-	mutex_unlock(&root->fs_info->trans_mutex);
+	spin_unlock(&root->fs_info->trans_lock);
 again:
 	if (!err) {
 		num_bytes = rc->merging_rsv_size;
@@ -2208,9 +2208,9 @@ int merge_reloc_roots(struct reloc_control *rc)
 	int ret;
 again:
 	root = rc->extent_root;
-	mutex_lock(&root->fs_info->trans_mutex);
+	spin_lock(&root->fs_info->trans_lock);
 	list_splice_init(&rc->reloc_roots, &reloc_roots);
-	mutex_unlock(&root->fs_info->trans_mutex);
+	spin_unlock(&root->fs_info->trans_lock);
 
 	while (!list_empty(&reloc_roots)) {
 		found = 1;
@@ -3583,17 +3583,17 @@ next:
 static void set_reloc_control(struct reloc_control *rc)
 {
 	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	mutex_lock(&fs_info->trans_mutex);
+	spin_lock(&fs_info->trans_lock);
 	fs_info->reloc_ctl = rc;
-	mutex_unlock(&fs_info->trans_mutex);
+	spin_unlock(&fs_info->trans_lock);
 }
 
 static void unset_reloc_control(struct reloc_control *rc)
 {
 	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	mutex_lock(&fs_info->trans_mutex);
+	spin_lock(&fs_info->trans_lock);
 	fs_info->reloc_ctl = NULL;
-	mutex_unlock(&fs_info->trans_mutex);
+	spin_unlock(&fs_info->trans_lock);
 }
 
 static int check_extent_flags(u64 flags)
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 46f40564c168..43816f8b23e7 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -34,6 +34,7 @@ static noinline void put_transaction(struct btrfs_transaction *transaction)
 {
 	WARN_ON(atomic_read(&transaction->use_count) == 0);
 	if (atomic_dec_and_test(&transaction->use_count)) {
+		BUG_ON(!list_empty(&transaction->list));
 		memset(transaction, 0, sizeof(*transaction));
 		kmem_cache_free(btrfs_transaction_cachep, transaction);
 	}
@@ -48,47 +49,73 @@ static noinline void switch_commit_root(struct btrfs_root *root)
 /*
  * either allocate a new transaction or hop into the existing one
  */
-static noinline int join_transaction(struct btrfs_root *root)
+static noinline int join_transaction(struct btrfs_root *root, int nofail)
 {
 	struct btrfs_transaction *cur_trans;
+
+	spin_lock(&root->fs_info->trans_lock);
+	if (root->fs_info->trans_no_join) {
+		if (!nofail) {
+			spin_unlock(&root->fs_info->trans_lock);
+			return -EBUSY;
+		}
+	}
+
 	cur_trans = root->fs_info->running_transaction;
-	if (!cur_trans) {
-		cur_trans = kmem_cache_alloc(btrfs_transaction_cachep,
-					     GFP_NOFS);
-		if (!cur_trans)
-			return -ENOMEM;
-		root->fs_info->generation++;
-		atomic_set(&cur_trans->num_writers, 1);
-		cur_trans->num_joined = 0;
-		cur_trans->transid = root->fs_info->generation;
-		init_waitqueue_head(&cur_trans->writer_wait);
-		init_waitqueue_head(&cur_trans->commit_wait);
-		cur_trans->in_commit = 0;
-		cur_trans->blocked = 0;
-		atomic_set(&cur_trans->use_count, 1);
-		cur_trans->commit_done = 0;
-		cur_trans->start_time = get_seconds();
-
-		cur_trans->delayed_refs.root = RB_ROOT;
-		cur_trans->delayed_refs.num_entries = 0;
-		cur_trans->delayed_refs.num_heads_ready = 0;
-		cur_trans->delayed_refs.num_heads = 0;
-		cur_trans->delayed_refs.flushing = 0;
-		cur_trans->delayed_refs.run_delayed_start = 0;
-		spin_lock_init(&cur_trans->delayed_refs.lock);
-
-		INIT_LIST_HEAD(&cur_trans->pending_snapshots);
-		list_add_tail(&cur_trans->list, &root->fs_info->trans_list);
-		extent_io_tree_init(&cur_trans->dirty_pages,
-				     root->fs_info->btree_inode->i_mapping,
-				     GFP_NOFS);
-		spin_lock(&root->fs_info->new_trans_lock);
-		root->fs_info->running_transaction = cur_trans;
-		spin_unlock(&root->fs_info->new_trans_lock);
-	} else {
+	if (cur_trans) {
+		atomic_inc(&cur_trans->use_count);
+		atomic_inc(&cur_trans->num_writers);
+		cur_trans->num_joined++;
+		spin_unlock(&root->fs_info->trans_lock);
+		return 0;
+	}
+	spin_unlock(&root->fs_info->trans_lock);
+
+	cur_trans = kmem_cache_alloc(btrfs_transaction_cachep, GFP_NOFS);
+	if (!cur_trans)
+		return -ENOMEM;
+	spin_lock(&root->fs_info->trans_lock);
+	if (root->fs_info->running_transaction) {
+		kmem_cache_free(btrfs_transaction_cachep, cur_trans);
+		cur_trans = root->fs_info->running_transaction;
+		atomic_inc(&cur_trans->use_count);
 		atomic_inc(&cur_trans->num_writers);
 		cur_trans->num_joined++;
+		spin_unlock(&root->fs_info->trans_lock);
+		return 0;
 	}
+	atomic_set(&cur_trans->num_writers, 1);
+	cur_trans->num_joined = 0;
+	init_waitqueue_head(&cur_trans->writer_wait);
+	init_waitqueue_head(&cur_trans->commit_wait);
+	cur_trans->in_commit = 0;
+	cur_trans->blocked = 0;
+	/*
+	 * One for this trans handle, one so it will live on until we
+	 * commit the transaction.
+	 */
+	atomic_set(&cur_trans->use_count, 2);
+	cur_trans->commit_done = 0;
+	cur_trans->start_time = get_seconds();
+
+	cur_trans->delayed_refs.root = RB_ROOT;
+	cur_trans->delayed_refs.num_entries = 0;
+	cur_trans->delayed_refs.num_heads_ready = 0;
+	cur_trans->delayed_refs.num_heads = 0;
+	cur_trans->delayed_refs.flushing = 0;
+	cur_trans->delayed_refs.run_delayed_start = 0;
+	spin_lock_init(&cur_trans->commit_lock);
+	spin_lock_init(&cur_trans->delayed_refs.lock);
+
+	INIT_LIST_HEAD(&cur_trans->pending_snapshots);
+	list_add_tail(&cur_trans->list, &root->fs_info->trans_list);
+	extent_io_tree_init(&cur_trans->dirty_pages,
+			     root->fs_info->btree_inode->i_mapping,
+			     GFP_NOFS);
+	root->fs_info->generation++;
+	cur_trans->transid = root->fs_info->generation;
+	root->fs_info->running_transaction = cur_trans;
+	spin_unlock(&root->fs_info->trans_lock);
 
 	return 0;
 }
@@ -99,39 +126,28 @@ static noinline int join_transaction(struct btrfs_root *root)
  * to make sure the old root from before we joined the transaction is deleted
  * when the transaction commits
  */
-static noinline int record_root_in_trans(struct btrfs_trans_handle *trans,
-					 struct btrfs_root *root)
+int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root)
 {
 	if (root->ref_cows && root->last_trans < trans->transid) {
 		WARN_ON(root == root->fs_info->extent_root);
 		WARN_ON(root->commit_root != root->node);
 
+		spin_lock(&root->fs_info->fs_roots_radix_lock);
+		if (root->last_trans == trans->transid) {
+			spin_unlock(&root->fs_info->fs_roots_radix_lock);
+			return 0;
+		}
+		root->last_trans = trans->transid;
 		radix_tree_tag_set(&root->fs_info->fs_roots_radix,
 			   (unsigned long)root->root_key.objectid,
 			   BTRFS_ROOT_TRANS_TAG);
-		root->last_trans = trans->transid;
+		spin_unlock(&root->fs_info->fs_roots_radix_lock);
 		btrfs_init_reloc_root(trans, root);
 	}
 	return 0;
 }
 
-int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root)
-{
-	if (!root->ref_cows)
-		return 0;
-
-	mutex_lock(&root->fs_info->trans_mutex);
-	if (root->last_trans == trans->transid) {
-		mutex_unlock(&root->fs_info->trans_mutex);
-		return 0;
-	}
-
-	record_root_in_trans(trans, root);
-	mutex_unlock(&root->fs_info->trans_mutex);
-	return 0;
-}
-
 /* wait for commit against the current transaction to become unblocked
  * when this is done, it is safe to start a new transaction, but the current
  * transaction might not be fully on disk.
@@ -140,21 +156,23 @@ static void wait_current_trans(struct btrfs_root *root)
 {
 	struct btrfs_transaction *cur_trans;
 
+	spin_lock(&root->fs_info->trans_lock);
 	cur_trans = root->fs_info->running_transaction;
 	if (cur_trans && cur_trans->blocked) {
 		DEFINE_WAIT(wait);
 		atomic_inc(&cur_trans->use_count);
+		spin_unlock(&root->fs_info->trans_lock);
 		while (1) {
 			prepare_to_wait(&root->fs_info->transaction_wait, &wait,
 					TASK_UNINTERRUPTIBLE);
 			if (!cur_trans->blocked)
 				break;
-			mutex_unlock(&root->fs_info->trans_mutex);
 			schedule();
-			mutex_lock(&root->fs_info->trans_mutex);
 		}
 		finish_wait(&root->fs_info->transaction_wait, &wait);
 		put_transaction(cur_trans);
+	} else {
+		spin_unlock(&root->fs_info->trans_lock);
 	}
 }
 
@@ -167,10 +185,16 @@ enum btrfs_trans_type {
 
 static int may_wait_transaction(struct btrfs_root *root, int type)
 {
-	if (!root->fs_info->log_root_recovering &&
-	    ((type == TRANS_START && !root->fs_info->open_ioctl_trans) ||
-	     type == TRANS_USERSPACE))
+	if (root->fs_info->log_root_recovering)
+		return 0;
+
+	if (type == TRANS_USERSPACE)
+		return 1;
+
+	if (type == TRANS_START &&
+	    !atomic_read(&root->fs_info->open_ioctl_trans))
 		return 1;
+
 	return 0;
 }
 
@@ -198,23 +222,21 @@ again:
 	if (!h)
 		return ERR_PTR(-ENOMEM);
 
-	if (type != TRANS_JOIN_NOLOCK)
-		mutex_lock(&root->fs_info->trans_mutex);
 	if (may_wait_transaction(root, type))
 		wait_current_trans(root);
 
-	ret = join_transaction(root);
+	do {
+		ret = join_transaction(root, type == TRANS_JOIN_NOLOCK);
+		if (ret == -EBUSY)
+			wait_current_trans(root);
+	} while (ret == -EBUSY);
+
 	if (ret < 0) {
 		kmem_cache_free(btrfs_trans_handle_cachep, h);
-		if (type != TRANS_JOIN_NOLOCK)
-			mutex_unlock(&root->fs_info->trans_mutex);
 		return ERR_PTR(ret);
 	}
 
 	cur_trans = root->fs_info->running_transaction;
-	atomic_inc(&cur_trans->use_count);
-	if (type != TRANS_JOIN_NOLOCK)
-		mutex_unlock(&root->fs_info->trans_mutex);
 
 	h->transid = cur_trans->transid;
 	h->transaction = cur_trans;
@@ -253,11 +275,7 @@ again:
 	}
 
 got_it:
-	if (type != TRANS_JOIN_NOLOCK)
-		mutex_lock(&root->fs_info->trans_mutex);
-	record_root_in_trans(h, root);
-	if (type != TRANS_JOIN_NOLOCK)
-		mutex_unlock(&root->fs_info->trans_mutex);
+	btrfs_record_root_in_trans(h, root);
 
 	if (!current->journal_info && type != TRANS_USERSPACE)
 		current->journal_info = h;
@@ -289,17 +307,13 @@ static noinline int wait_for_commit(struct btrfs_root *root,
 				    struct btrfs_transaction *commit)
 {
 	DEFINE_WAIT(wait);
-	mutex_lock(&root->fs_info->trans_mutex);
 	while (!commit->commit_done) {
 		prepare_to_wait(&commit->commit_wait, &wait,
 				TASK_UNINTERRUPTIBLE);
 		if (commit->commit_done)
 			break;
-		mutex_unlock(&root->fs_info->trans_mutex);
 		schedule();
-		mutex_lock(&root->fs_info->trans_mutex);
 	}
-	mutex_unlock(&root->fs_info->trans_mutex);
 	finish_wait(&commit->commit_wait, &wait);
 	return 0;
 }
@@ -309,50 +323,49 @@ int btrfs_wait_for_commit(struct btrfs_root *root, u64 transid)
 	struct btrfs_transaction *cur_trans = NULL, *t;
 	int ret;
 
-	mutex_lock(&root->fs_info->trans_mutex);
-
 	ret = 0;
 	if (transid) {
 		if (transid <= root->fs_info->last_trans_committed)
-			goto out_unlock;
+			goto out;
 
 		/* find specified transaction */
+		spin_lock(&root->fs_info->trans_lock);
 		list_for_each_entry(t, &root->fs_info->trans_list, list) {
 			if (t->transid == transid) {
 				cur_trans = t;
+				atomic_inc(&cur_trans->use_count);
 				break;
 			}
 			if (t->transid > transid)
 				break;
 		}
+		spin_unlock(&root->fs_info->trans_lock);
 		ret = -EINVAL;
 		if (!cur_trans)
-			goto out_unlock;  /* bad transid */
+			goto out;  /* bad transid */
 	} else {
 		/* find newest transaction that is committing | committed */
+		spin_lock(&root->fs_info->trans_lock);
 		list_for_each_entry_reverse(t, &root->fs_info->trans_list,
 					    list) {
 			if (t->in_commit) {
 				if (t->commit_done)
-					goto out_unlock;
+					goto out;
 				cur_trans = t;
+				atomic_inc(&cur_trans->use_count);
 				break;
 			}
 		}
+		spin_unlock(&root->fs_info->trans_lock);
 		if (!cur_trans)
-			goto out_unlock;  /* nothing committing|committed */
+			goto out;  /* nothing committing|committed */
 	}
 
-	atomic_inc(&cur_trans->use_count);
-	mutex_unlock(&root->fs_info->trans_mutex);
-
 	wait_for_commit(root, cur_trans);
 
-	mutex_lock(&root->fs_info->trans_mutex);
 	put_transaction(cur_trans);
 	ret = 0;
-out_unlock:
-	mutex_unlock(&root->fs_info->trans_mutex);
+out:
 	return ret;
 }
 
@@ -401,10 +414,8 @@ harder:
 
 void btrfs_throttle(struct btrfs_root *root)
 {
-	mutex_lock(&root->fs_info->trans_mutex);
-	if (!root->fs_info->open_ioctl_trans)
+	if (!atomic_read(&root->fs_info->open_ioctl_trans))
 		wait_current_trans(root);
-	mutex_unlock(&root->fs_info->trans_mutex);
 }
 
 static int should_end_transaction(struct btrfs_trans_handle *trans,
@@ -422,6 +433,7 @@ int btrfs_should_end_transaction(struct btrfs_trans_handle *trans,
 	struct btrfs_transaction *cur_trans = trans->transaction;
 	int updates;
 
+	smp_mb();
 	if (cur_trans->blocked || cur_trans->delayed_refs.flushing)
 		return 1;
 
@@ -467,9 +479,11 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 
 	btrfs_trans_release_metadata(trans, root);
 
-	if (lock && !root->fs_info->open_ioctl_trans &&
-	    should_end_transaction(trans, root))
+	if (lock && !atomic_read(&root->fs_info->open_ioctl_trans) &&
+	    should_end_transaction(trans, root)) {
 		trans->transaction->blocked = 1;
+		smp_wmb();
+	}
 
 	if (lock && cur_trans->blocked && !cur_trans->in_commit) {
 		if (throttle)
@@ -739,9 +753,9 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans,
  */
 int btrfs_add_dead_root(struct btrfs_root *root)
 {
-	mutex_lock(&root->fs_info->trans_mutex);
+	spin_lock(&root->fs_info->trans_lock);
 	list_add(&root->root_list, &root->fs_info->dead_roots);
-	mutex_unlock(&root->fs_info->trans_mutex);
+	spin_unlock(&root->fs_info->trans_lock);
 	return 0;
 }
 
@@ -757,6 +771,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 	int ret;
 	int err = 0;
 
+	spin_lock(&fs_info->fs_roots_radix_lock);
 	while (1) {
 		ret = radix_tree_gang_lookup_tag(&fs_info->fs_roots_radix,
 						 (void **)gang, 0,
@@ -769,6 +784,7 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 			radix_tree_tag_clear(&fs_info->fs_roots_radix,
 					(unsigned long)root->root_key.objectid,
 					BTRFS_ROOT_TRANS_TAG);
+			spin_unlock(&fs_info->fs_roots_radix_lock);
 
 			btrfs_free_log(trans, root);
 			btrfs_update_reloc_root(trans, root);
@@ -783,10 +799,12 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans,
 			err = btrfs_update_root(trans, fs_info->tree_root,
 						&root->root_key,
 						&root->root_item);
+			spin_lock(&fs_info->fs_roots_radix_lock);
 			if (err)
 				break;
 		}
 	}
+	spin_unlock(&fs_info->fs_roots_radix_lock);
 	return err;
 }
 
@@ -972,7 +990,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	parent = dget_parent(dentry);
 	parent_inode = parent->d_inode;
 	parent_root = BTRFS_I(parent_inode)->root;
-	record_root_in_trans(trans, parent_root);
+	btrfs_record_root_in_trans(trans, parent_root);
 
 	/*
 	 * insert the directory item
@@ -990,7 +1008,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	ret = btrfs_update_inode(trans, parent_root, parent_inode);
 	BUG_ON(ret);
 
-	record_root_in_trans(trans, root);
+	btrfs_record_root_in_trans(trans, root);
 	btrfs_set_root_last_snapshot(&root->root_item, trans->transid);
 	memcpy(new_root_item, &root->root_item, sizeof(*new_root_item));
 	btrfs_check_and_init_root_item(new_root_item);
@@ -1080,20 +1098,20 @@ static void update_super_roots(struct btrfs_root *root)
 int btrfs_transaction_in_commit(struct btrfs_fs_info *info)
 {
 	int ret = 0;
-	spin_lock(&info->new_trans_lock);
+	spin_lock(&info->trans_lock);
 	if (info->running_transaction)
 		ret = info->running_transaction->in_commit;
-	spin_unlock(&info->new_trans_lock);
+	spin_unlock(&info->trans_lock);
 	return ret;
 }
 
 int btrfs_transaction_blocked(struct btrfs_fs_info *info)
 {
 	int ret = 0;
-	spin_lock(&info->new_trans_lock);
+	spin_lock(&info->trans_lock);
 	if (info->running_transaction)
 		ret = info->running_transaction->blocked;
-	spin_unlock(&info->new_trans_lock);
+	spin_unlock(&info->trans_lock);
 	return ret;
 }
 
@@ -1117,9 +1135,7 @@ static void wait_current_trans_commit_start(struct btrfs_root *root,
 				    &wait);
 			break;
 		}
-		mutex_unlock(&root->fs_info->trans_mutex);
 		schedule();
-		mutex_lock(&root->fs_info->trans_mutex);
 		finish_wait(&root->fs_info->transaction_blocked_wait, &wait);
 	}
 }
@@ -1145,9 +1161,7 @@ static void wait_current_trans_commit_start_and_unblock(struct btrfs_root *root,
 				    &wait);
 			break;
 		}
-		mutex_unlock(&root->fs_info->trans_mutex);
 		schedule();
-		mutex_lock(&root->fs_info->trans_mutex);
 		finish_wait(&root->fs_info->transaction_wait,
 			    &wait);
 	}
@@ -1193,22 +1207,18 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 	}
 
 	/* take transaction reference */
-	mutex_lock(&root->fs_info->trans_mutex);
 	cur_trans = trans->transaction;
 	atomic_inc(&cur_trans->use_count);
-	mutex_unlock(&root->fs_info->trans_mutex);
 
 	btrfs_end_transaction(trans, root);
 	schedule_delayed_work(&ac->work, 0);
 
 	/* wait for transaction to start and unblock */
-	mutex_lock(&root->fs_info->trans_mutex);
 	if (wait_for_unblock)
 		wait_current_trans_commit_start_and_unblock(root, cur_trans);
 	else
 		wait_current_trans_commit_start(root, cur_trans);
 	put_transaction(cur_trans);
-	mutex_unlock(&root->fs_info->trans_mutex);
 
 	return 0;
 }
@@ -1252,38 +1262,41 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	ret = btrfs_run_delayed_refs(trans, root, 0);
 	BUG_ON(ret);
 
-	mutex_lock(&root->fs_info->trans_mutex);
+	spin_lock(&cur_trans->commit_lock);
 	if (cur_trans->in_commit) {
+		spin_unlock(&cur_trans->commit_lock);
 		atomic_inc(&cur_trans->use_count);
-		mutex_unlock(&root->fs_info->trans_mutex);
 		btrfs_end_transaction(trans, root);
 
 		ret = wait_for_commit(root, cur_trans);
 		BUG_ON(ret);
 
-		mutex_lock(&root->fs_info->trans_mutex);
 		put_transaction(cur_trans);
-		mutex_unlock(&root->fs_info->trans_mutex);
 
 		return 0;
 	}
 
 	trans->transaction->in_commit = 1;
 	trans->transaction->blocked = 1;
+	spin_unlock(&cur_trans->commit_lock);
 	wake_up(&root->fs_info->transaction_blocked_wait);
 
+	spin_lock(&root->fs_info->trans_lock);
 	if (cur_trans->list.prev != &root->fs_info->trans_list) {
 		prev_trans = list_entry(cur_trans->list.prev,
 					struct btrfs_transaction, list);
 		if (!prev_trans->commit_done) {
 			atomic_inc(&prev_trans->use_count);
-			mutex_unlock(&root->fs_info->trans_mutex);
+			spin_unlock(&root->fs_info->trans_lock);
 
 			wait_for_commit(root, prev_trans);
 
-			mutex_lock(&root->fs_info->trans_mutex);
 			put_transaction(prev_trans);
+		} else {
+			spin_unlock(&root->fs_info->trans_lock);
 		}
+	} else {
+		spin_unlock(&root->fs_info->trans_lock);
 	}
 
 	if (now < cur_trans->start_time || now - cur_trans->start_time < 1)
@@ -1291,12 +1304,12 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	do {
 		int snap_pending = 0;
+
 		joined = cur_trans->num_joined;
 		if (!list_empty(&trans->transaction->pending_snapshots))
 			snap_pending = 1;
 
 		WARN_ON(cur_trans != trans->transaction);
-		mutex_unlock(&root->fs_info->trans_mutex);
 
 		if (flush_on_commit || snap_pending) {
 			btrfs_start_delalloc_inodes(root, 1);
@@ -1316,14 +1329,15 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		prepare_to_wait(&cur_trans->writer_wait, &wait,
 				TASK_UNINTERRUPTIBLE);
 
-		smp_mb();
 		if (atomic_read(&cur_trans->num_writers) > 1)
 			schedule_timeout(MAX_SCHEDULE_TIMEOUT);
 		else if (should_grow)
 			schedule_timeout(1);
 
-		mutex_lock(&root->fs_info->trans_mutex);
 		finish_wait(&cur_trans->writer_wait, &wait);
+		spin_lock(&root->fs_info->trans_lock);
+		root->fs_info->trans_no_join = 1;
+		spin_unlock(&root->fs_info->trans_lock);
 	} while (atomic_read(&cur_trans->num_writers) > 1 ||
 		 (should_grow && cur_trans->num_joined != joined));
 
@@ -1364,9 +1378,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	btrfs_prepare_extent_commit(trans, root);
 
 	cur_trans = root->fs_info->running_transaction;
-	spin_lock(&root->fs_info->new_trans_lock);
-	root->fs_info->running_transaction = NULL;
-	spin_unlock(&root->fs_info->new_trans_lock);
 
 	btrfs_set_root_node(&root->fs_info->tree_root->root_item,
 			    root->fs_info->tree_root->node);
@@ -1387,10 +1398,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	       sizeof(root->fs_info->super_copy));
 
 	trans->transaction->blocked = 0;
+	spin_lock(&root->fs_info->trans_lock);
+	root->fs_info->running_transaction = NULL;
+	root->fs_info->trans_no_join = 0;
+	spin_unlock(&root->fs_info->trans_lock);
 
 	wake_up(&root->fs_info->transaction_wait);
 
-	mutex_unlock(&root->fs_info->trans_mutex);
 	ret = btrfs_write_and_wait_transaction(trans, root);
 	BUG_ON(ret);
 	write_ctree_super(trans, root, 0);
@@ -1403,22 +1417,21 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	btrfs_finish_extent_commit(trans, root);
 
-	mutex_lock(&root->fs_info->trans_mutex);
-
 	cur_trans->commit_done = 1;
 
 	root->fs_info->last_trans_committed = cur_trans->transid;
 
 	wake_up(&cur_trans->commit_wait);
 
+	spin_lock(&root->fs_info->trans_lock);
 	list_del_init(&cur_trans->list);
+	spin_unlock(&root->fs_info->trans_lock);
+
 	put_transaction(cur_trans);
 	put_transaction(cur_trans);
 
 	trace_btrfs_transaction_commit(root);
 
-	mutex_unlock(&root->fs_info->trans_mutex);
-
 	if (current->journal_info == trans)
 		current->journal_info = NULL;
 
@@ -1438,9 +1451,9 @@ int btrfs_clean_old_snapshots(struct btrfs_root *root)
 	LIST_HEAD(list);
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
-	mutex_lock(&fs_info->trans_mutex);
+	spin_lock(&fs_info->trans_lock);
 	list_splice_init(&fs_info->dead_roots, &list);
-	mutex_unlock(&fs_info->trans_mutex);
+	spin_unlock(&fs_info->trans_lock);
 
 	while (!list_empty(&list)) {
 		root = list_entry(list.next, struct btrfs_root, root_list);
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 154314f80f8d..11c6efcd4ed2 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -28,10 +28,12 @@ struct btrfs_transaction {
 	 * transaction can end
 	 */
 	atomic_t num_writers;
+	atomic_t use_count;
 
 	unsigned long num_joined;
+
+	spinlock_t commit_lock;
 	int in_commit;
-	atomic_t use_count;
 	int commit_done;
 	int blocked;
 	struct list_head list;
-- 
cgit v1.2.3


From fcb80c2affd63237cff5b34cba5756be7c976a5a Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Tue, 3 May 2011 10:40:22 -0400
Subject: Btrfs: fix how we do space reservation for truncate

The ceph guys keep running into problems where we have space reserved in our
orphan block rsv when freeing it up.  This is because they tend to do snapshots
alot, so their truncates tend to use a bunch of space, so when we go to do
things like update the inode we have to steal reservation space in order to make
the reservation happen.  This happens because truncate can use as much space as
it freaking feels like, but we still have to hold space for removing the orphan
item and updating the inode, which will definitely always happen.  So in order
to fix this we need to split all of the reservation stuf up.  So with this patch
we have

1) The orphan block reserve which only holds the space for deleting our orphan
item when everything is over.

2) The truncate block reserve which gets allocated and used specifically for the
space that the truncate will use on a per truncate basis.

3) The transaction will always have 1 item's worth of data reserved so we can
update the inode normally.

Hopefully this will make the ceph problem go away.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/ctree.h       |   3 ++
 fs/btrfs/extent-tree.c |  46 +++++++++++++++-----
 fs/btrfs/inode.c       | 111 +++++++++++++++++++++++++++++++++++++------------
 3 files changed, 123 insertions(+), 37 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 522a39b0033d..f31aed7fedd9 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2224,6 +2224,9 @@ int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv,
 void btrfs_block_rsv_release(struct btrfs_root *root,
 			     struct btrfs_block_rsv *block_rsv,
 			     u64 num_bytes);
+int btrfs_truncate_reserve_metadata(struct btrfs_trans_handle *trans,
+				    struct btrfs_root *root,
+				    struct btrfs_block_rsv *rsv);
 int btrfs_set_block_group_ro(struct btrfs_root *root,
 			     struct btrfs_block_group_cache *cache);
 int btrfs_set_block_group_rw(struct btrfs_root *root,
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index ca599654ce19..a2ca561c70f0 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3980,6 +3980,37 @@ static u64 calc_trans_metadata_size(struct btrfs_root *root, int num_items)
 		3 * num_items;
 }
 
+int btrfs_truncate_reserve_metadata(struct btrfs_trans_handle *trans,
+				    struct btrfs_root *root,
+				    struct btrfs_block_rsv *rsv)
+{
+	struct btrfs_block_rsv *trans_rsv = &root->fs_info->trans_block_rsv;
+	u64 num_bytes;
+	int ret;
+
+	/*
+	 * Truncate should be freeing data, but give us 2 items just in case it
+	 * needs to use some space.  We may want to be smarter about this in the
+	 * future.
+	 */
+	num_bytes = calc_trans_metadata_size(root, 2);
+
+	/* We already have enough bytes, just return */
+	if (rsv->reserved >= num_bytes)
+		return 0;
+
+	num_bytes -= rsv->reserved;
+
+	/*
+	 * You should have reserved enough space before hand to do this, so this
+	 * should not fail.
+	 */
+	ret = block_rsv_migrate_bytes(trans_rsv, rsv, num_bytes);
+	BUG_ON(ret);
+
+	return 0;
+}
+
 int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
 				 struct btrfs_root *root,
 				 int num_items)
@@ -4020,23 +4051,18 @@ int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
 	struct btrfs_block_rsv *dst_rsv = root->orphan_block_rsv;
 
 	/*
-	 * one for deleting orphan item, one for updating inode and
-	 * two for calling btrfs_truncate_inode_items.
-	 *
-	 * btrfs_truncate_inode_items is a delete operation, it frees
-	 * more space than it uses in most cases. So two units of
-	 * metadata space should be enough for calling it many times.
-	 * If all of the metadata space is used, we can commit
-	 * transaction and use space it freed.
+	 * We need to hold space in order to delete our orphan item once we've
+	 * added it, so this takes the reservation so we can release it later
+	 * when we are truly done with the orphan item.
 	 */
-	u64 num_bytes = calc_trans_metadata_size(root, 4);
+	u64 num_bytes = calc_trans_metadata_size(root, 1);
 	return block_rsv_migrate_bytes(src_rsv, dst_rsv, num_bytes);
 }
 
 void btrfs_orphan_release_metadata(struct inode *inode)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
-	u64 num_bytes = calc_trans_metadata_size(root, 4);
+	u64 num_bytes = calc_trans_metadata_size(root, 1);
 	btrfs_block_rsv_release(root, root->orphan_block_rsv, num_bytes);
 }
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index e47bdf0fb75a..bc12ba23db5f 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -6591,6 +6591,7 @@ out:
 static int btrfs_truncate(struct inode *inode)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_block_rsv *rsv;
 	int ret;
 	int err = 0;
 	struct btrfs_trans_handle *trans;
@@ -6604,28 +6605,83 @@ static int btrfs_truncate(struct inode *inode)
 	btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
 	btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
 
-	trans = btrfs_start_transaction(root, 5);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	/*
+	 * Yes ladies and gentelment, this is indeed ugly.  The fact is we have
+	 * 3 things going on here
+	 *
+	 * 1) We need to reserve space for our orphan item and the space to
+	 * delete our orphan item.  Lord knows we don't want to have a dangling
+	 * orphan item because we didn't reserve space to remove it.
+	 *
+	 * 2) We need to reserve space to update our inode.
+	 *
+	 * 3) We need to have something to cache all the space that is going to
+	 * be free'd up by the truncate operation, but also have some slack
+	 * space reserved in case it uses space during the truncate (thank you
+	 * very much snapshotting).
+	 *
+	 * And we need these to all be seperate.  The fact is we can use alot of
+	 * space doing the truncate, and we have no earthly idea how much space
+	 * we will use, so we need the truncate reservation to be seperate so it
+	 * doesn't end up using space reserved for updating the inode or
+	 * removing the orphan item.  We also need to be able to stop the
+	 * transaction and start a new one, which means we need to be able to
+	 * update the inode several times, and we have no idea of knowing how
+	 * many times that will be, so we can't just reserve 1 item for the
+	 * entirety of the opration, so that has to be done seperately as well.
+	 * Then there is the orphan item, which does indeed need to be held on
+	 * to for the whole operation, and we need nobody to touch this reserved
+	 * space except the orphan code.
+	 *
+	 * So that leaves us with
+	 *
+	 * 1) root->orphan_block_rsv - for the orphan deletion.
+	 * 2) rsv - for the truncate reservation, which we will steal from the
+	 * transaction reservation.
+	 * 3) fs_info->trans_block_rsv - this will have 1 items worth left for
+	 * updating the inode.
+	 */
+	rsv = btrfs_alloc_block_rsv(root);
+	if (!rsv)
+		return -ENOMEM;
+	btrfs_add_durable_block_rsv(root->fs_info, rsv);
+
+	trans = btrfs_start_transaction(root, 4);
+	if (IS_ERR(trans)) {
+		err = PTR_ERR(trans);
+		goto out;
+	}
 
 	btrfs_set_trans_block_group(trans, inode);
 
+	/*
+	 * Reserve space for the truncate process.  Truncate should be adding
+	 * space, but if there are snapshots it may end up using space.
+	 */
+	ret = btrfs_truncate_reserve_metadata(trans, root, rsv);
+	BUG_ON(ret);
+
 	ret = btrfs_orphan_add(trans, inode);
 	if (ret) {
 		btrfs_end_transaction(trans, root);
-		return ret;
+		goto out;
 	}
 
 	nr = trans->blocks_used;
 	btrfs_end_transaction(trans, root);
 	btrfs_btree_balance_dirty(root, nr);
 
-	/* Now start a transaction for the truncate */
-	trans = btrfs_start_transaction(root, 0);
-	if (IS_ERR(trans))
-		return PTR_ERR(trans);
+	/*
+	 * Ok so we've already migrated our bytes over for the truncate, so here
+	 * just reserve the one slot we need for updating the inode.
+	 */
+	trans = btrfs_start_transaction(root, 1);
+	if (IS_ERR(trans)) {
+		err = PTR_ERR(trans);
+		goto out;
+	}
 	btrfs_set_trans_block_group(trans, inode);
-	trans->block_rsv = root->orphan_block_rsv;
+	trans->block_rsv = rsv;
 
 	/*
 	 * setattr is responsible for setting the ordered_data_close flag,
@@ -6649,24 +6705,18 @@ static int btrfs_truncate(struct inode *inode)
 
 	while (1) {
 		if (!trans) {
-			trans = btrfs_start_transaction(root, 0);
-			if (IS_ERR(trans))
-				return PTR_ERR(trans);
-			btrfs_set_trans_block_group(trans, inode);
-			trans->block_rsv = root->orphan_block_rsv;
-		}
+			trans = btrfs_start_transaction(root, 3);
+			if (IS_ERR(trans)) {
+				err = PTR_ERR(trans);
+				goto out;
+			}
 
-		ret = btrfs_block_rsv_check(trans, root,
-					    root->orphan_block_rsv, 0, 5);
-		if (ret == -EAGAIN) {
-			ret = btrfs_commit_transaction(trans, root);
-			if (ret)
-				return ret;
-			trans = NULL;
-			continue;
-		} else if (ret) {
-			err = ret;
-			break;
+			ret = btrfs_truncate_reserve_metadata(trans, root,
+							      rsv);
+			BUG_ON(ret);
+
+			btrfs_set_trans_block_group(trans, inode);
+			trans->block_rsv = rsv;
 		}
 
 		ret = btrfs_truncate_inode_items(trans, root, inode,
@@ -6677,6 +6727,7 @@ static int btrfs_truncate(struct inode *inode)
 			break;
 		}
 
+		trans->block_rsv = &root->fs_info->trans_block_rsv;
 		ret = btrfs_update_inode(trans, root, inode);
 		if (ret) {
 			err = ret;
@@ -6690,6 +6741,7 @@ static int btrfs_truncate(struct inode *inode)
 	}
 
 	if (ret == 0 && inode->i_nlink > 0) {
+		trans->block_rsv = root->orphan_block_rsv;
 		ret = btrfs_orphan_del(trans, inode);
 		if (ret)
 			err = ret;
@@ -6701,15 +6753,20 @@ static int btrfs_truncate(struct inode *inode)
 		ret = btrfs_orphan_del(NULL, inode);
 	}
 
+	trans->block_rsv = &root->fs_info->trans_block_rsv;
 	ret = btrfs_update_inode(trans, root, inode);
 	if (ret && !err)
 		err = ret;
 
 	nr = trans->blocks_used;
 	ret = btrfs_end_transaction_throttle(trans, root);
+	btrfs_btree_balance_dirty(root, nr);
+
+out:
+	btrfs_free_block_rsv(root, rsv);
+
 	if (ret && !err)
 		err = ret;
-	btrfs_btree_balance_dirty(root, nr);
 
 	return err;
 }
-- 
cgit v1.2.3


From af60bed24eb0e3b6d93eaa6bb395a5721e6c09a8 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 4 May 2011 11:11:17 -0400
Subject: Btrfs: set range_start to the right start in count_range_bits

In count_range_bits we are adjusting total_bytes based on the range we are
searching for, but we don't adjust the range start according to the range we are
searching for, which makes for weird results.  For example, if the range

[0-8192]

is set DELALLOC, but I search for 4096-8192, I will get back 4096 for the number
of bytes found, but the range_start will be 0, which makes it look like the
range is [0-4096].  So instead set range_start = max(cur_start, state->start).
This makes everything come out right.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/extent_io.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index ba41da59e31b..b5f6f227a97c 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1480,7 +1480,7 @@ u64 count_range_bits(struct extent_io_tree *tree,
 			if (total_bytes >= max_bytes)
 				break;
 			if (!found) {
-				*start = state->start;
+				*start = max(cur_start, state->start);
 				found = 1;
 			}
 			last = state->end;
-- 
cgit v1.2.3


From cb25c2ea6a79702ab7895b873c6c43e0d3bc3c72 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 11 May 2011 12:17:34 -0400
Subject: Btrfs: map the node block when looking for readahead targets

If we have particularly full nodes, we could call btrfs_node_blockptr up to 32
times, which is 32 pairs of kmap/kunmap, which _sucks_.  So go ahead and map the
extent buffer while we look for readahead targets.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/ctree.c | 23 +++++++++++++++++++++--
 1 file changed, 21 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 84d7ca1fe0ba..009bcf7f1e4b 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -1229,6 +1229,7 @@ static void reada_for_search(struct btrfs_root *root,
 	u64 search;
 	u64 target;
 	u64 nread = 0;
+	u64 gen;
 	int direction = path->reada;
 	struct extent_buffer *eb;
 	u32 nr;
@@ -1256,6 +1257,15 @@ static void reada_for_search(struct btrfs_root *root,
 	nritems = btrfs_header_nritems(node);
 	nr = slot;
 	while (1) {
+		if (!node->map_token) {
+			unsigned long offset = btrfs_node_key_ptr_offset(nr);
+			map_private_extent_buffer(node, offset,
+						  sizeof(struct btrfs_key_ptr),
+						  &node->map_token,
+						  &node->kaddr,
+						  &node->map_start,
+						  &node->map_len, KM_USER1);
+		}
 		if (direction < 0) {
 			if (nr == 0)
 				break;
@@ -1273,14 +1283,23 @@ static void reada_for_search(struct btrfs_root *root,
 		search = btrfs_node_blockptr(node, nr);
 		if ((search <= target && target - search <= 65536) ||
 		    (search > target && search - target <= 65536)) {
-			readahead_tree_block(root, search, blocksize,
-				     btrfs_node_ptr_generation(node, nr));
+			gen = btrfs_node_ptr_generation(node, nr);
+			if (node->map_token) {
+				unmap_extent_buffer(node, node->map_token,
+						    KM_USER1);
+				node->map_token = NULL;
+			}
+			readahead_tree_block(root, search, blocksize, gen);
 			nread += blocksize;
 		}
 		nscan++;
 		if ((nread > 65536 || nscan > 32))
 			break;
 	}
+	if (node->map_token) {
+		unmap_extent_buffer(node, node->map_token, KM_USER1);
+		node->map_token = NULL;
+	}
 }
 
 /*
-- 
cgit v1.2.3


From 7e2355ba1a11649f0b212a29fdb9f47476f1248e Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 11 May 2011 12:25:37 -0400
Subject: Btrfs: don't look at the extent buffer level 3 times in a row

We have a bit of debugging in btrfs_search_slot to make sure the level of the
cow block is the same as the original block we were cow'ing.  I don't think I've
ever seen this tripped, so kill it.  This saves us 2 kmap's per level in our
search.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/ctree.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 009bcf7f1e4b..f7a0a64b868f 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -1672,9 +1672,6 @@ again:
 		}
 cow_done:
 		BUG_ON(!cow && ins_len);
-		if (level != btrfs_header_level(b))
-			WARN_ON(1);
-		level = btrfs_header_level(b);
 
 		p->nodes[level] = b;
 		if (!p->skip_locking)
-- 
cgit v1.2.3


From d82a6f1d7e8b61ed5996334d0db66651bb43641d Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Wed, 11 May 2011 15:26:06 -0400
Subject: Btrfs: kill BTRFS_I(inode)->block_group

Originally this was going to be used as a way to give hints to the allocator,
but frankly we can get much better hints elsewhere and it's not even used at all
for anything usefull.  In addition to be completely useless, when we initialize
an inode we try