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(+)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 7cd8ab0..3b9f164 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 v1.1


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/inode.c | 34 +++++++++++++++++-----------------
 1 file changed, 17 insertions(+), 17 deletions(-)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 3b9f164..e47bdf0 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;
-- 
cgit v1.1


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/inode.c | 111 +++++++++++++++++++++++++++++++++++++++++--------------
 1 file changed, 84 insertions(+), 27 deletions(-)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index e47bdf0..bc12ba2 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 v1.1


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 and find a freeish block group to set as the inodes block group,
and with a completely full 40gb fs this takes _forever_, so I imagine with say
1tb fs this is just unbearable.  So just axe the thing altoghether, we don't
need it and it saves us 8 bytes in the inode and saves us 500 microseconds per
inode lookup in my testcase.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/inode.c | 87 ++++++--------------------------------------------------
 1 file changed, 9 insertions(+), 78 deletions(-)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index bc12ba2..dd5938a 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -136,7 +136,6 @@ static noinline int insert_inline_extent(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 
 	path->leave_spinning = 1;
-	btrfs_set_trans_block_group(trans, inode);
 
 	key.objectid = inode->i_ino;
 	key.offset = start;
@@ -422,7 +421,6 @@ again:
 	if (start == 0) {
 		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;
 
 		/* lets try to make an inline extent */
@@ -781,7 +779,6 @@ static noinline int cow_file_range(struct inode *inode,
 	BUG_ON(root == root->fs_info->tree_root);
 	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;
 
 	num_bytes = (end - start + blocksize) & ~(blocksize - 1);
@@ -1502,8 +1499,6 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
 {
 	struct btrfs_ordered_sum *sum;
 
-	btrfs_set_trans_block_group(trans, inode);
-
 	list_for_each_entry(sum, list, list) {
 		btrfs_csum_file_blocks(trans,
 		       BTRFS_I(inode)->root->fs_info->csum_root, sum);
@@ -1722,7 +1717,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 			else
 				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;
 			ret = btrfs_update_inode(trans, root, inode);
 			BUG_ON(ret);
@@ -1739,7 +1733,6 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
 	else
 		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;
 
 	if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
@@ -2495,7 +2488,6 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_key location;
 	int maybe_acls;
-	u64 alloc_group_block;
 	u32 rdev;
 	int ret;
 
@@ -2539,8 +2531,6 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	BTRFS_I(inode)->index_cnt = (u64)-1;
 	BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
 
-	alloc_group_block = btrfs_inode_block_group(leaf, inode_item);
-
 	/*
 	 * try to precache a NULL acl entry for files that don't have
 	 * any xattrs or acls
@@ -2549,8 +2539,6 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	if (!maybe_acls)
 		cache_no_acl(inode);
 
-	BTRFS_I(inode)->block_group = btrfs_find_block_group(root, 0,
-						alloc_group_block, 0);
 	btrfs_free_path(path);
 	inode_item = NULL;
 
@@ -2630,7 +2618,7 @@ static void fill_inode_item(struct btrfs_trans_handle *trans,
 	btrfs_set_inode_transid(leaf, item, trans->transid);
 	btrfs_set_inode_rdev(leaf, item, inode->i_rdev);
 	btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags);
-	btrfs_set_inode_block_group(leaf, item, BTRFS_I(inode)->block_group);
+	btrfs_set_inode_block_group(leaf, item, 0);
 
 	if (leaf->map_token) {
 		unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
@@ -2971,8 +2959,6 @@ static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	btrfs_set_trans_block_group(trans, dir);
-
 	btrfs_record_unlink_dir(trans, dir, dentry->d_inode, 0);
 
 	ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
@@ -3068,8 +3054,6 @@ static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	btrfs_set_trans_block_group(trans, dir);
-
 	if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
 		err = btrfs_unlink_subvol(trans, root, dir,
 					  BTRFS_I(inode)->location.objectid,
@@ -3649,7 +3633,6 @@ int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
 				err = PTR_ERR(trans);
 				break;
 			}
-			btrfs_set_trans_block_group(trans, inode);
 
 			err = btrfs_drop_extents(trans, inode, cur_offset,
 						 cur_offset + hole_size,
@@ -3785,7 +3768,6 @@ void btrfs_evict_inode(struct inode *inode)
 	while (1) {
 		trans = btrfs_start_transaction(root, 0);
 		BUG_ON(IS_ERR(trans));
-		btrfs_set_trans_block_group(trans, inode);
 		trans->block_rsv = root->orphan_block_rsv;
 
 		ret = btrfs_block_rsv_check(trans, root,
@@ -4383,7 +4365,6 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 			trans = btrfs_join_transaction(root);
 		if (IS_ERR(trans))
 			return PTR_ERR(trans);
-		btrfs_set_trans_block_group(trans, inode);
 		if (nolock)
 			ret = btrfs_end_transaction_nolock(trans, root);
 		else
@@ -4409,7 +4390,6 @@ void btrfs_dirty_inode(struct inode *inode)
 
 	trans = btrfs_join_transaction(root);
 	BUG_ON(IS_ERR(trans));
-	btrfs_set_trans_block_group(trans, inode);
 
 	ret = btrfs_update_inode(trans, root, inode);
 	if (ret && ret == -ENOSPC) {
@@ -4424,7 +4404,6 @@ void btrfs_dirty_inode(struct inode *inode)
 			}
 			return;
 		}
-		btrfs_set_trans_block_group(trans, inode);
 
 		ret = btrfs_update_inode(trans, root, inode);
 		if (ret) {
@@ -4519,8 +4498,8 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 				     struct btrfs_root *root,
 				     struct inode *dir,
 				     const char *name, int name_len,
-				     u64 ref_objectid, u64 objectid,
-				     u64 alloc_hint, int mode, u64 *index)
+				     u64 ref_objectid, u64 objectid, int mode,
+				     u64 *index)
 {
 	struct inode *inode;
 	struct btrfs_inode_item *inode_item;
@@ -4567,8 +4546,6 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 		owner = 0;
 	else
 		owner = 1;
-	BTRFS_I(inode)->block_group =
-			btrfs_find_block_group(root, 0, alloc_hint, owner);
 
 	key[0].objectid = objectid;
 	btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
@@ -4729,11 +4706,9 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	btrfs_set_trans_block_group(trans, dir);
-
 	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
 				dentry->d_name.len, dir->i_ino, objectid,
-				BTRFS_I(dir)->block_group, mode, &index);
+				mode, &index);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
 		goto out_unlock;
@@ -4745,7 +4720,6 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
 		goto out_unlock;
 	}
 
-	btrfs_set_trans_block_group(trans, inode);
 	err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
 	if (err)
 		drop_inode = 1;
@@ -4754,8 +4728,6 @@ static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
 		init_special_inode(inode, inode->i_mode, rdev);
 		btrfs_update_inode(trans, root, inode);
 	}
-	btrfs_update_inode_block_group(trans, inode);
-	btrfs_update_inode_block_group(trans, dir);
 out_unlock:
 	nr = trans->blocks_used;
 	btrfs_end_transaction_throttle(trans, root);
@@ -4791,11 +4763,9 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	btrfs_set_trans_block_group(trans, dir);
-
 	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
 				dentry->d_name.len, dir->i_ino, objectid,
-				BTRFS_I(dir)->block_group, mode, &index);
+				mode, &index);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
 		goto out_unlock;
@@ -4807,7 +4777,6 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
 		goto out_unlock;
 	}
 
-	btrfs_set_trans_block_group(trans, inode);
 	err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
 	if (err)
 		drop_inode = 1;
@@ -4818,8 +4787,6 @@ static int btrfs_create(struct inode *dir, struct dentry *dentry,
 		inode->i_op = &btrfs_file_inode_operations;
 		BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
 	}
-	btrfs_update_inode_block_group(trans, inode);
-	btrfs_update_inode_block_group(trans, dir);
 out_unlock:
 	nr = trans->blocks_used;
 	btrfs_end_transaction_throttle(trans, root);
@@ -4866,8 +4833,6 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 
 	btrfs_inc_nlink(inode);
 	inode->i_ctime = CURRENT_TIME;
-
-	btrfs_set_trans_block_group(trans, dir);
 	ihold(inode);
 
 	err = btrfs_add_nondir(trans, dir, dentry, inode, 1, index);
@@ -4876,7 +4841,6 @@ static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
 		drop_inode = 1;
 	} else {
 		struct dentry *parent = dget_parent(dentry);
-		btrfs_update_inode_block_group(trans, dir);
 		err = btrfs_update_inode(trans, root, inode);
 		BUG_ON(err);
 		btrfs_log_new_name(trans, inode, NULL, parent);
@@ -4917,12 +4881,10 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 	trans = btrfs_start_transaction(root, 5);
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
-	btrfs_set_trans_block_group(trans, dir);
 
 	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
 				dentry->d_name.len, dir->i_ino, objectid,
-				BTRFS_I(dir)->block_group, S_IFDIR | mode,
-				&index);
+				S_IFDIR | mode, &index);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
 		goto out_fail;
@@ -4936,7 +4898,6 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 
 	inode->i_op = &btrfs_dir_inode_operations;
 	inode->i_fop = &btrfs_dir_file_operations;
-	btrfs_set_trans_block_group(trans, inode);
 
 	btrfs_i_size_write(inode, 0);
 	err = btrfs_update_inode(trans, root, inode);
@@ -4950,8 +4911,6 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
 
 	d_instantiate(dentry, inode);
 	drop_on_err = 0;
-	btrfs_update_inode_block_group(trans, inode);
-	btrfs_update_inode_block_group(trans, dir);
 
 out_fail:
 	nr = trans->blocks_used;
@@ -6652,8 +6611,6 @@ static int btrfs_truncate(struct inode *inode)
 		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.
@@ -6680,7 +6637,6 @@ static int btrfs_truncate(struct inode *inode)
 		err = PTR_ERR(trans);
 		goto out;
 	}
-	btrfs_set_trans_block_group(trans, inode);
 	trans->block_rsv = rsv;
 
 	/*
@@ -6715,7 +6671,6 @@ static int btrfs_truncate(struct inode *inode)
 							      rsv);
 			BUG_ON(ret);
 
-			btrfs_set_trans_block_group(trans, inode);
 			trans->block_rsv = rsv;
 		}
 
@@ -6775,15 +6730,14 @@ out:
  * create a new subvolume directory/inode (helper for the ioctl).
  */
 int btrfs_create_subvol_root(struct btrfs_trans_handle *trans,
-			     struct btrfs_root *new_root,
-			     u64 new_dirid, u64 alloc_hint)
+			     struct btrfs_root *new_root, u64 new_dirid)
 {
 	struct inode *inode;
 	int err;
 	u64 index = 0;
 
 	inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid,
-				new_dirid, alloc_hint, S_IFDIR | 0700, &index);
+				new_dirid, S_IFDIR | 0700, &index);
 	if (IS_ERR(inode))
 		return PTR_ERR(inode);
 	inode->i_op = &btrfs_dir_inode_operations;
@@ -6893,21 +6847,6 @@ void btrfs_destroy_inode(struct inode *inode)
 		spin_unlock(&root->fs_info->ordered_extent_lock);
 	}
 
-	if (root == root->fs_info->tree_root) {
-		struct btrfs_block_group_cache *block_group;
-
-		block_group = btrfs_lookup_block_group(root->fs_info,
-						BTRFS_I(inode)->block_group);
-		if (block_group && block_group->inode == inode) {
-			spin_lock(&block_group->lock);
-			block_group->inode = NULL;
-			spin_unlock(&block_group->lock);
-			btrfs_put_block_group(block_group);
-		} else if (block_group) {
-			btrfs_put_block_group(block_group);
-		}
-	}
-
 	spin_lock(&root->orphan_lock);
 	if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
 		printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n",
@@ -7091,8 +7030,6 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
                 goto out_notrans;
         }
 
-	btrfs_set_trans_block_group(trans, new_dir);
-
 	if (dest != root)
 		btrfs_record_root_in_trans(trans, dest);
 
@@ -7331,12 +7268,9 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 	if (IS_ERR(trans))
 		return PTR_ERR(trans);
 
-	btrfs_set_trans_block_group(trans, dir);
-
 	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
 				dentry->d_name.len, dir->i_ino, objectid,
-				BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO,
-				&index);
+				S_IFLNK|S_IRWXUGO, &index);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
 		goto out_unlock;
@@ -7348,7 +7282,6 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 		goto out_unlock;
 	}
 
-	btrfs_set_trans_block_group(trans, inode);
 	err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
 	if (err)
 		drop_inode = 1;
@@ -7359,8 +7292,6 @@ static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
 		inode->i_op = &btrfs_file_inode_operations;
 		BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
 	}
-	btrfs_update_inode_block_group(trans, inode);
-	btrfs_update_inode_block_group(trans, dir);
 	if (drop_inode)
 		goto out_unlock;
 
-- 
cgit v1.1


From 026fd317828500524cdc7e5ff9e8e7923abb2868 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Fri, 13 May 2011 10:32:11 -0400
Subject: Btrfs: don't always do readahead

Our readahead is sort of sloppy, and really isn't always needed.  For example if
ls is doing a stating ls (which is the default) it's going to stat in non-disk
order, so if say you have a directory with a stupid amount of files, readahead
is going to do nothing but waste time in the case of doing the stat.  Taking the
unconditional readahead out made my test go from 57 minutes to 36 minutes.  This
means that everywhere we do loop through the tree we want to make sure we do set
path->reada properly, so I went through and found all of the places where we
loop through the path and set reada to 1.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/inode.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index dd5938a..6228a30 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4242,7 +4242,9 @@ static int btrfs_real_readdir(struct file *filp, void *dirent,
 		filp->f_pos = 2;
 	}
 	path = btrfs_alloc_path();
-	path->reada = 2;
+	if (!path)
+		return -ENOMEM;
+	path->reada = 1;
 
 	btrfs_set_key_type(&key, key_type);
 	key.offset = filp->f_pos;
@@ -5043,7 +5045,15 @@ again:
 
 	if (!path) {
 		path = btrfs_alloc_path();
-		BUG_ON(!path);
+		if (!path) {
+			err = -ENOMEM;
+			goto out;
+		}
+		/*
+		 * Chances are we'll be called again, so go ahead and do
+		 * readahead
+		 */
+		path->reada = 1;
 	}
 
 	ret = btrfs_lookup_file_extent(trans, root, path,
-- 
cgit v1.1


From d90c732122a1f6d0efe388a8a204f67f144b2eb3 Mon Sep 17 00:00:00 2001
From: Josef Bacik <josef@redhat.com>
Date: Tue, 17 May 2011 09:50:54 -0400
Subject: Btrfs: leave spinning on lookup and map the leaf

On lookup we only want to read the inode item, so leave the path spinning.  Also
we're just wholesale reading the leaf off, so map the leaf so we don't do a
bunch of kmap/kunmaps.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
---
 fs/btrfs/inode.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 6228a30..dc8fb2b 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2493,6 +2493,7 @@ static void btrfs_read_locked_inode(struct inode *inode)
 
 	path = btrfs_alloc_path();
 	BUG_ON(!path);
+	path->leave_spinning = 1;
 	memcpy(&location, &BTRFS_I(inode)->location, sizeof(location));
 
 	ret = btrfs_lookup_inode(NULL, root, path, &location, 0);
@@ -2502,6 +2503,12 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	leaf = path->nodes[0];
 	inode_item = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_inode_item);
+	if (!leaf->map_token)
+		map_private_extent_buffer(leaf, (unsigned long)inode_item,
+					  sizeof(struct btrfs_inode_item),
+					  &leaf->map_token, &leaf->kaddr,
+					  &leaf->map_start, &leaf->map_len,
+					  KM_USER1);
 
 	inode->i_mode = btrfs_inode_mode(leaf, inode_item);
 	inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
@@ -2539,6 +2546,11 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	if (!maybe_acls)
 		cache_no_acl(inode);
 
+	if (leaf->map_token) {
+		unmap_extent_buffer(leaf, leaf->map_token, KM_USER1);
+		leaf->map_token = NULL;
+	}
+
 	btrfs_free_path(path);
 	inode_item = NULL;
 
-- 
cgit v1.1


From 7841cb2898f66a73062c64d0ef5733dde7279e46 Mon Sep 17 00:00:00 2001
From: David Sterba <dsterba@suse.cz>
Date: Tue, 31 May 2011 18:07:27 +0200
Subject: btrfs: add helper for fs_info->closing

wrap checking of filesystem 'closing' flag and fix a few missing memory
barriers.

Signed-off-by: David Sterba <dsterba@suse.cz>
---
 fs/btrfs/inode.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

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

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index a83e44b..02ff4a1 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4266,8 +4266,7 @@ int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc)
 	if (BTRFS_I(inode)->dummy_inode)
 		return 0;
 
-	smp_mb();
-	if (root->fs_info->closing && is_free_space_inode(root, inode))
+	if (btrfs_fs_closing(root->fs_info) && is_free_space_inode(root, inode))
 		nolock = true;
 
 	if (wbc->sync_mode == WB_SYNC_ALL) {
-- 
cgit v1.1