From 6702ed490ca0bb44e17131818a5a18b773957c5a Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Tue, 7 Aug 2007 16:15:09 -0400
Subject: Btrfs: Add run time btree defrag, and an ioctl to force btree defrag

This adds two types of btree defrag, a run time form that tries to
defrag recently allocated blocks in the btree when they are still in ram,
and an ioctl that forces defrag of all btree blocks.

File data blocks are not defragged yet, but this can make a huge difference
in sequential btree reads.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 222 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 222 insertions(+)
 create mode 100644 fs/btrfs/tree-defrag.c

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
new file mode 100644
index 000000000000..15d0a486fb59
--- /dev/null
+++ b/fs/btrfs/tree-defrag.c
@@ -0,0 +1,222 @@
+/*
+ * Copyright (C) 2007 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#include <linux/sched.h>
+#include "ctree.h"
+#include "disk-io.h"
+#include "print-tree.h"
+#include "transaction.h"
+
+static void reada_defrag(struct btrfs_root *root,
+			 struct btrfs_node *node)
+{
+	int i;
+	u32 nritems;
+	u64 blocknr;
+	int ret;
+
+	nritems = btrfs_header_nritems(&node->header);
+	for (i = 0; i < nritems; i++) {
+		blocknr = btrfs_node_blockptr(node, i);
+		ret = readahead_tree_block(root, blocknr);
+		if (ret)
+			break;
+	}
+}
+
+static int defrag_walk_down(struct btrfs_trans_handle *trans,
+			    struct btrfs_root *root,
+			    struct btrfs_path *path, int *level,
+			    int cache_only)
+{
+	struct buffer_head *next;
+	struct buffer_head *cur;
+	u64 blocknr;
+	int ret = 0;
+
+	WARN_ON(*level < 0);
+	WARN_ON(*level >= BTRFS_MAX_LEVEL);
+
+	while(*level > 0) {
+		WARN_ON(*level < 0);
+		WARN_ON(*level >= BTRFS_MAX_LEVEL);
+		cur = path->nodes[*level];
+
+		if (!cache_only && *level > 1 && path->slots[*level] == 0)
+			reada_defrag(root, btrfs_buffer_node(cur));
+
+		if (btrfs_header_level(btrfs_buffer_header(cur)) != *level)
+			WARN_ON(1);
+
+		if (path->slots[*level] >=
+		    btrfs_header_nritems(btrfs_buffer_header(cur)))
+			break;
+
+		if (*level == 1) {
+			ret = btrfs_realloc_node(trans, root,
+						 path->nodes[*level],
+						 cache_only);
+			break;
+		}
+		blocknr = btrfs_node_blockptr(btrfs_buffer_node(cur),
+					      path->slots[*level]);
+
+		if (cache_only) {
+			next = btrfs_find_tree_block(root, blocknr);
+			if (!next || !buffer_uptodate(next) ||
+			   buffer_locked(next)) {
+				brelse(next);
+				path->slots[*level]++;
+				continue;
+			}
+		} else {
+			next = read_tree_block(root, blocknr);
+		}
+		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
+				      path->slots[*level], &next);
+		BUG_ON(ret);
+		ret = btrfs_realloc_node(trans, root, next, cache_only);
+		BUG_ON(ret);
+		WARN_ON(*level <= 0);
+		if (path->nodes[*level-1])
+			btrfs_block_release(root, path->nodes[*level-1]);
+		path->nodes[*level-1] = next;
+		*level = btrfs_header_level(btrfs_buffer_header(next));
+		path->slots[*level] = 0;
+	}
+	WARN_ON(*level < 0);
+	WARN_ON(*level >= BTRFS_MAX_LEVEL);
+	btrfs_block_release(root, path->nodes[*level]);
+	path->nodes[*level] = NULL;
+	*level += 1;
+	WARN_ON(ret);
+	return 0;
+}
+
+static int defrag_walk_up(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  struct btrfs_path *path, int *level,
+			  int cache_only)
+{
+	int i;
+	int slot;
+	struct btrfs_node *node;
+
+	for(i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) {
+		slot = path->slots[i];
+		if (slot < btrfs_header_nritems(
+		    btrfs_buffer_header(path->nodes[i])) - 1) {
+			path->slots[i]++;
+			*level = i;
+			node = btrfs_buffer_node(path->nodes[i]);
+			WARN_ON(i == 0);
+			btrfs_disk_key_to_cpu(&root->defrag_progress,
+					      &node->ptrs[path->slots[i]].key);
+			root->defrag_level = i;
+			return 0;
+		} else {
+			btrfs_block_release(root, path->nodes[*level]);
+			path->nodes[*level] = NULL;
+			*level = i + 1;
+		}
+	}
+	return 1;
+}
+
+int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
+			struct btrfs_root *root, int cache_only)
+{
+	struct btrfs_path *path = NULL;
+	struct buffer_head *tmp;
+	int ret = 0;
+	int wret;
+	int level;
+	int orig_level;
+	int i;
+	int num_runs = 0;
+
+	if (root->ref_cows == 0) {
+		goto out;
+	}
+	path = btrfs_alloc_path();
+	if (!path)
+		return -ENOMEM;
+
+	level = btrfs_header_level(btrfs_buffer_header(root->node));
+	orig_level = level;
+	if (level == 0) {
+		goto out;
+	}
+	if (root->defrag_progress.objectid == 0) {
+		get_bh(root->node);
+		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
+		BUG_ON(ret);
+		ret = btrfs_realloc_node(trans, root, root->node, cache_only);
+		BUG_ON(ret);
+		path->nodes[level] = root->node;
+		path->slots[level] = 0;
+	} else {
+		level = root->defrag_level;
+		path->lowest_level = level;
+		wret = btrfs_search_slot(trans, root, &root->defrag_progress,
+					 path, 0, 1);
+
+		if (wret < 0) {
+			ret = wret;
+			goto out;
+		}
+		while(level > 0 && !path->nodes[level])
+			level--;
+		if (!path->nodes[level]) {
+			ret = 0;
+			goto out;
+		}
+	}
+
+	while(1) {
+		wret = defrag_walk_down(trans, root, path, &level, cache_only);
+		if (wret > 0)
+			break;
+		if (wret < 0)
+			ret = wret;
+
+		wret = defrag_walk_up(trans, root, path, &level, cache_only);
+		if (wret > 0)
+			break;
+		if (wret < 0)
+			ret = wret;
+		if (num_runs++ > 8) {
+			ret = -EAGAIN;
+			break;
+		}
+	}
+	for (i = 0; i <= orig_level; i++) {
+		if (path->nodes[i]) {
+			btrfs_block_release(root, path->nodes[i]);
+			path->nodes[i] = 0;
+		}
+	}
+out:
+	if (path)
+		btrfs_free_path(path);
+	if (ret != -EAGAIN) {
+		memset(&root->defrag_progress, 0,
+		       sizeof(root->defrag_progress));
+	}
+	return ret;
+}
-- 
cgit v1.2.3


From 409eb95d7f6632d5af32b795244ce68a29e49319 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Wed, 8 Aug 2007 20:17:12 -0400
Subject: Btrfs: Further reduce the concurrency penalty of defrag and
 drop_snapshot

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 15d0a486fb59..7ea66b4aa5c2 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -148,7 +148,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	int level;
 	int orig_level;
 	int i;
-	int num_runs = 0;
 
 	if (root->ref_cows == 0) {
 		goto out;
@@ -200,10 +199,8 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 			break;
 		if (wret < 0)
 			ret = wret;
-		if (num_runs++ > 8) {
-			ret = -EAGAIN;
-			break;
-		}
+		ret = -EAGAIN;
+		break;
 	}
 	for (i = 0; i <= orig_level; i++) {
 		if (path->nodes[i]) {
-- 
cgit v1.2.3


From e9d0b13b5bbb58c9b840e407a8d181442f799966 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 10 Aug 2007 14:06:19 -0400
Subject: Btrfs: Btree defrag on the extent-mapping tree as well

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 36 +++++++++++++++++++++++++++++-------
 1 file changed, 29 insertions(+), 7 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 7ea66b4aa5c2..a09064a9a41c 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -42,16 +42,20 @@ static void reada_defrag(struct btrfs_root *root,
 static int defrag_walk_down(struct btrfs_trans_handle *trans,
 			    struct btrfs_root *root,
 			    struct btrfs_path *path, int *level,
-			    int cache_only)
+			    int cache_only, u64 *last_ret)
 {
 	struct buffer_head *next;
 	struct buffer_head *cur;
 	u64 blocknr;
 	int ret = 0;
+	int is_extent = 0;
 
 	WARN_ON(*level < 0);
 	WARN_ON(*level >= BTRFS_MAX_LEVEL);
 
+	if (root->fs_info->extent_root == root)
+		is_extent = 1;
+
 	while(*level > 0) {
 		WARN_ON(*level < 0);
 		WARN_ON(*level >= BTRFS_MAX_LEVEL);
@@ -70,7 +74,10 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		if (*level == 1) {
 			ret = btrfs_realloc_node(trans, root,
 						 path->nodes[*level],
-						 cache_only);
+						 cache_only, last_ret);
+			if (is_extent)
+				btrfs_extent_post_op(trans, root);
+
 			break;
 		}
 		blocknr = btrfs_node_blockptr(btrfs_buffer_node(cur),
@@ -90,8 +97,13 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
 				      path->slots[*level], &next);
 		BUG_ON(ret);
-		ret = btrfs_realloc_node(trans, root, next, cache_only);
+		ret = btrfs_realloc_node(trans, root, next, cache_only,
+					 last_ret);
 		BUG_ON(ret);
+
+		if (is_extent)
+			btrfs_extent_post_op(trans, root);
+
 		WARN_ON(*level <= 0);
 		if (path->nodes[*level-1])
 			btrfs_block_release(root, path->nodes[*level-1]);
@@ -148,10 +160,14 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	int level;
 	int orig_level;
 	int i;
+	int is_extent = 0;
+	u64 last_ret = 0;
+
+	if (root->fs_info->extent_root == root)
+		is_extent = 1;
 
-	if (root->ref_cows == 0) {
+	if (root->ref_cows == 0 && !is_extent)
 		goto out;
-	}
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
@@ -165,16 +181,21 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		get_bh(root->node);
 		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
 		BUG_ON(ret);
-		ret = btrfs_realloc_node(trans, root, root->node, cache_only);
+		ret = btrfs_realloc_node(trans, root, root->node, cache_only,
+					 &last_ret);
 		BUG_ON(ret);
 		path->nodes[level] = root->node;
 		path->slots[level] = 0;
+		if (is_extent)
+			btrfs_extent_post_op(trans, root);
 	} else {
 		level = root->defrag_level;
 		path->lowest_level = level;
 		wret = btrfs_search_slot(trans, root, &root->defrag_progress,
 					 path, 0, 1);
 
+		if (is_extent)
+			btrfs_extent_post_op(trans, root);
 		if (wret < 0) {
 			ret = wret;
 			goto out;
@@ -188,7 +209,8 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	}
 
 	while(1) {
-		wret = defrag_walk_down(trans, root, path, &level, cache_only);
+		wret = defrag_walk_down(trans, root, path, &level, cache_only,
+					&last_ret);
 		if (wret > 0)
 			break;
 		if (wret < 0)
-- 
cgit v1.2.3


From f2183bde1a918d338337955c8e8ba29bd8f5e7b1 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 10 Aug 2007 14:42:37 -0400
Subject: Btrfs: Add BH_Defrag to mark buffers that are in need of defragging

This allows the tree walking code to defrag only the newly allocated
buffers, it seems to be a good balance between perfect defragging and the
performance hit of repeatedly reallocating blocks.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index a09064a9a41c..35fd20d24645 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -86,7 +86,7 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		if (cache_only) {
 			next = btrfs_find_tree_block(root, blocknr);
 			if (!next || !buffer_uptodate(next) ||
-			   buffer_locked(next)) {
+			   buffer_locked(next) || !buffer_defrag(next)) {
 				brelse(next);
 				path->slots[*level]++;
 				continue;
@@ -142,6 +142,7 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			root->defrag_level = i;
 			return 0;
 		} else {
+			clear_buffer_defrag(path->nodes[*level]);
 			btrfs_block_release(root, path->nodes[*level]);
 			path->nodes[*level] = NULL;
 			*level = i + 1;
-- 
cgit v1.2.3


From 86479a04eef8f304a13aeb8b64bcc8e506a68268 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 10 Sep 2007 19:58:16 -0400
Subject: Add support for defragging files via btrfsctl -d.  Avoid OOM on
 extent tree defrag.

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 35fd20d24645..420597127ed1 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -113,6 +113,8 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	}
 	WARN_ON(*level < 0);
 	WARN_ON(*level >= BTRFS_MAX_LEVEL);
+	clear_buffer_defrag(path->nodes[*level]);
+	clear_buffer_defrag_done(path->nodes[*level]);
 	btrfs_block_release(root, path->nodes[*level]);
 	path->nodes[*level] = NULL;
 	*level += 1;
@@ -143,6 +145,7 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			return 0;
 		} else {
 			clear_buffer_defrag(path->nodes[*level]);
+			clear_buffer_defrag_done(path->nodes[*level]);
 			btrfs_block_release(root, path->nodes[*level]);
 			path->nodes[*level] = NULL;
 			*level = i + 1;
-- 
cgit v1.2.3


From 5f39d397dfbe140a14edecd4e73c34ce23c4f9ee Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:14:19 -0400
Subject: Btrfs: Create extent_buffer interface for large blocksizes

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 58 ++++++++++++++++++++++++++++----------------------
 1 file changed, 32 insertions(+), 26 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 420597127ed1..daf019afa0a1 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -23,14 +23,14 @@
 #include "transaction.h"
 
 static void reada_defrag(struct btrfs_root *root,
-			 struct btrfs_node *node)
+			 struct extent_buffer *node)
 {
 	int i;
 	u32 nritems;
 	u64 blocknr;
 	int ret;
 
-	nritems = btrfs_header_nritems(&node->header);
+	nritems = btrfs_header_nritems(node);
 	for (i = 0; i < nritems; i++) {
 		blocknr = btrfs_node_blockptr(node, i);
 		ret = readahead_tree_block(root, blocknr);
@@ -44,8 +44,8 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 			    struct btrfs_path *path, int *level,
 			    int cache_only, u64 *last_ret)
 {
-	struct buffer_head *next;
-	struct buffer_head *cur;
+	struct extent_buffer *next;
+	struct extent_buffer *cur;
 	u64 blocknr;
 	int ret = 0;
 	int is_extent = 0;
@@ -62,13 +62,13 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		cur = path->nodes[*level];
 
 		if (!cache_only && *level > 1 && path->slots[*level] == 0)
-			reada_defrag(root, btrfs_buffer_node(cur));
+			reada_defrag(root, cur);
 
-		if (btrfs_header_level(btrfs_buffer_header(cur)) != *level)
+		if (btrfs_header_level(cur) != *level)
 			WARN_ON(1);
 
 		if (path->slots[*level] >=
-		    btrfs_header_nritems(btrfs_buffer_header(cur)))
+		    btrfs_header_nritems(cur))
 			break;
 
 		if (*level == 1) {
@@ -80,14 +80,13 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 
 			break;
 		}
-		blocknr = btrfs_node_blockptr(btrfs_buffer_node(cur),
-					      path->slots[*level]);
+		blocknr = btrfs_node_blockptr(cur, path->slots[*level]);
 
 		if (cache_only) {
 			next = btrfs_find_tree_block(root, blocknr);
-			if (!next || !buffer_uptodate(next) ||
-			   buffer_locked(next) || !buffer_defrag(next)) {
-				brelse(next);
+			/* FIXME, test for defrag */
+			if (!next || !btrfs_buffer_uptodate(next)) {
+				free_extent_buffer(next);
 				path->slots[*level]++;
 				continue;
 			}
@@ -106,16 +105,18 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 
 		WARN_ON(*level <= 0);
 		if (path->nodes[*level-1])
-			btrfs_block_release(root, path->nodes[*level-1]);
+			free_extent_buffer(path->nodes[*level-1]);
 		path->nodes[*level-1] = next;
-		*level = btrfs_header_level(btrfs_buffer_header(next));
+		*level = btrfs_header_level(next);
 		path->slots[*level] = 0;
 	}
 	WARN_ON(*level < 0);
 	WARN_ON(*level >= BTRFS_MAX_LEVEL);
+#if 0
 	clear_buffer_defrag(path->nodes[*level]);
 	clear_buffer_defrag_done(path->nodes[*level]);
-	btrfs_block_release(root, path->nodes[*level]);
+#endif
+	free_extent_buffer(path->nodes[*level]);
 	path->nodes[*level] = NULL;
 	*level += 1;
 	WARN_ON(ret);
@@ -129,24 +130,25 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 {
 	int i;
 	int slot;
-	struct btrfs_node *node;
+	struct extent_buffer *node;
 
 	for(i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) {
 		slot = path->slots[i];
-		if (slot < btrfs_header_nritems(
-		    btrfs_buffer_header(path->nodes[i])) - 1) {
+		if (slot < btrfs_header_nritems(path->nodes[i]) - 1) {
 			path->slots[i]++;
 			*level = i;
-			node = btrfs_buffer_node(path->nodes[i]);
+			node = path->nodes[i];
 			WARN_ON(i == 0);
-			btrfs_disk_key_to_cpu(&root->defrag_progress,
-					      &node->ptrs[path->slots[i]].key);
+			btrfs_node_key_to_cpu(node, &root->defrag_progress,
+					      path->slots[i]);
 			root->defrag_level = i;
 			return 0;
 		} else {
+			/*
 			clear_buffer_defrag(path->nodes[*level]);
 			clear_buffer_defrag_done(path->nodes[*level]);
-			btrfs_block_release(root, path->nodes[*level]);
+			*/
+			free_extent_buffer(path->nodes[*level]);
 			path->nodes[*level] = NULL;
 			*level = i + 1;
 		}
@@ -158,7 +160,7 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root, int cache_only)
 {
 	struct btrfs_path *path = NULL;
-	struct buffer_head *tmp;
+	struct extent_buffer *tmp;
 	int ret = 0;
 	int wret;
 	int level;
@@ -172,17 +174,18 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 
 	if (root->ref_cows == 0 && !is_extent)
 		goto out;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	level = btrfs_header_level(btrfs_buffer_header(root->node));
+	level = btrfs_header_level(root->node);
 	orig_level = level;
 	if (level == 0) {
 		goto out;
 	}
 	if (root->defrag_progress.objectid == 0) {
-		get_bh(root->node);
+		extent_buffer_get(root->node);
 		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
 		BUG_ON(ret);
 		ret = btrfs_realloc_node(trans, root, root->node, cache_only,
@@ -200,12 +203,15 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 
 		if (is_extent)
 			btrfs_extent_post_op(trans, root);
+
 		if (wret < 0) {
 			ret = wret;
 			goto out;
 		}
+
 		while(level > 0 && !path->nodes[level])
 			level--;
+
 		if (!path->nodes[level]) {
 			ret = 0;
 			goto out;
@@ -230,7 +236,7 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	}
 	for (i = 0; i <= orig_level; i++) {
 		if (path->nodes[i]) {
-			btrfs_block_release(root, path->nodes[i]);
+			free_extent_buffer(path->nodes[i]);
 			path->nodes[i] = 0;
 		}
 	}
-- 
cgit v1.2.3


From db94535db75e67fab12ccbb7f5ee548e33fed891 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:15:53 -0400
Subject: Btrfs: Allow tree blocks larger than the page size

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index daf019afa0a1..3feac2f28628 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -27,13 +27,15 @@ static void reada_defrag(struct btrfs_root *root,
 {
 	int i;
 	u32 nritems;
-	u64 blocknr;
+	u64 bytenr;
+	u32 blocksize;
 	int ret;
 
+	blocksize = btrfs_level_size(root, btrfs_header_level(node) - 1);
 	nritems = btrfs_header_nritems(node);
 	for (i = 0; i < nritems; i++) {
-		blocknr = btrfs_node_blockptr(node, i);
-		ret = readahead_tree_block(root, blocknr);
+		bytenr = btrfs_node_blockptr(node, i);
+		ret = readahead_tree_block(root, bytenr, blocksize);
 		if (ret)
 			break;
 	}
@@ -46,7 +48,7 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 {
 	struct extent_buffer *next;
 	struct extent_buffer *cur;
-	u64 blocknr;
+	u64 bytenr;
 	int ret = 0;
 	int is_extent = 0;
 
@@ -80,10 +82,11 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 
 			break;
 		}
-		blocknr = btrfs_node_blockptr(cur, path->slots[*level]);
+		bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
 
 		if (cache_only) {
-			next = btrfs_find_tree_block(root, blocknr);
+			next = btrfs_find_tree_block(root, bytenr,
+					   btrfs_level_size(root, *level - 1));
 			/* FIXME, test for defrag */
 			if (!next || !btrfs_buffer_uptodate(next)) {
 				free_extent_buffer(next);
@@ -91,7 +94,8 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 				continue;
 			}
 		} else {
-			next = read_tree_block(root, blocknr);
+			next = read_tree_block(root, bytenr,
+				       btrfs_level_size(root, *level - 1));
 		}
 		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
 				      path->slots[*level], &next);
-- 
cgit v1.2.3


From 6b80053d02be41886344b5007d04e345311ec0b5 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:17:34 -0400
Subject: Btrfs: Add back the online defragging code

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 3feac2f28628..d23216aa4ab5 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -116,10 +116,10 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	}
 	WARN_ON(*level < 0);
 	WARN_ON(*level >= BTRFS_MAX_LEVEL);
-#if 0
-	clear_buffer_defrag(path->nodes[*level]);
-	clear_buffer_defrag_done(path->nodes[*level]);
-#endif
+
+	btrfs_clear_buffer_defrag(path->nodes[*level]);
+	btrfs_clear_buffer_defrag_done(path->nodes[*level]);
+
 	free_extent_buffer(path->nodes[*level]);
 	path->nodes[*level] = NULL;
 	*level += 1;
@@ -148,10 +148,8 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			root->defrag_level = i;
 			return 0;
 		} else {
-			/*
-			clear_buffer_defrag(path->nodes[*level]);
-			clear_buffer_defrag_done(path->nodes[*level]);
-			*/
+			btrfs_clear_buffer_defrag(path->nodes[*level]);
+			btrfs_clear_buffer_defrag_done(path->nodes[*level]);
 			free_extent_buffer(path->nodes[*level]);
 			path->nodes[*level] = NULL;
 			*level = i + 1;
-- 
cgit v1.2.3


From 0f82731fc56448c2733f58e1f5db6c2cbfc90652 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:18:56 -0400
Subject: Breakout BTRFS_SETGET_FUNCS into a separate C file, the inlines were
 too big.

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index d23216aa4ab5..0d1a1be0995c 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -239,7 +239,7 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	for (i = 0; i <= orig_level; i++) {
 		if (path->nodes[i]) {
 			free_extent_buffer(path->nodes[i]);
-			path->nodes[i] = 0;
+			path->nodes[i] = NULL;
 		}
 	}
 out:
-- 
cgit v1.2.3


From 0f1ebbd159baade0417f8f62f0cd6810cc950832 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:19:46 -0400
Subject: Btrfs: Large block related defrag optimizations

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 0d1a1be0995c..d31c21ac6f23 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -118,7 +118,6 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	WARN_ON(*level >= BTRFS_MAX_LEVEL);
 
 	btrfs_clear_buffer_defrag(path->nodes[*level]);
-	btrfs_clear_buffer_defrag_done(path->nodes[*level]);
 
 	free_extent_buffer(path->nodes[*level]);
 	path->nodes[*level] = NULL;
@@ -148,6 +147,26 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			root->defrag_level = i;
 			return 0;
 		} else {
+			if (*level > 1 && path->nodes[*level] != root->node) {
+				struct extent_buffer *next;
+				u64 last;
+				int ret;
+				ret = btrfs_cow_block(trans, root,
+						      path->nodes[*level],
+						      path->nodes[*level + 1],
+						      path->slots[*level + 1],
+						      &next);
+				BUG_ON(ret);
+				path->nodes[*level] = next;
+				last = next->start;
+				ret = btrfs_realloc_node(trans, root, next,
+							 cache_only, &last);
+				BUG_ON(ret);
+
+				if (root == root->fs_info->extent_root)
+					btrfs_extent_post_op(trans, root);
+			}
+
 			btrfs_clear_buffer_defrag(path->nodes[*level]);
 			btrfs_clear_buffer_defrag_done(path->nodes[*level]);
 			free_extent_buffer(path->nodes[*level]);
@@ -183,9 +202,11 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 
 	level = btrfs_header_level(root->node);
 	orig_level = level;
+
 	if (level == 0) {
 		goto out;
 	}
+
 	if (root->defrag_progress.objectid == 0) {
 		extent_buffer_get(root->node);
 		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
-- 
cgit v1.2.3


From cf786e79e3f96f4bc037c69e1a7afb5fb85c6111 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:22:11 -0400
Subject: Btrfs: Defrag: only walk into nodes with the defrag bit set

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index d31c21ac6f23..f86eccf511b9 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -87,8 +87,8 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		if (cache_only) {
 			next = btrfs_find_tree_block(root, bytenr,
 					   btrfs_level_size(root, *level - 1));
-			/* FIXME, test for defrag */
-			if (!next || !btrfs_buffer_uptodate(next)) {
+			if (!next || !btrfs_buffer_uptodate(next) ||
+			    !btrfs_buffer_defrag(next)) {
 				free_extent_buffer(next);
 				path->slots[*level]++;
 				continue;
@@ -147,7 +147,8 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			root->defrag_level = i;
 			return 0;
 		} else {
-			if (*level > 1 && path->nodes[*level] != root->node) {
+			if (*level > 1 && path->nodes[*level] != root->node &&
+			    btrfs_buffer_defrag(path->nodes[*level])) {
 				struct extent_buffer *next;
 				u64 last;
 				int ret;
@@ -168,7 +169,6 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			}
 
 			btrfs_clear_buffer_defrag(path->nodes[*level]);
-			btrfs_clear_buffer_defrag_done(path->nodes[*level]);
 			free_extent_buffer(path->nodes[*level]);
 			path->nodes[*level] = NULL;
 			*level = i + 1;
-- 
cgit v1.2.3


From a6b6e75e096f436f0cc56edf5bca96301e194491 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 15 Oct 2007 16:22:39 -0400
Subject: Btrfs: Defrag only leaves, and only when the parent node has a single
 objectid

This allows us to defrag huge directories, but skip the expensive defrag
case in more common usage, where it does not help as much.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 36 +++++-------------------------------
 1 file changed, 5 insertions(+), 31 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index f86eccf511b9..b02355a7b143 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -76,7 +76,9 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		if (*level == 1) {
 			ret = btrfs_realloc_node(trans, root,
 						 path->nodes[*level],
-						 cache_only, last_ret);
+						 path->slots[*level],
+						 cache_only, last_ret,
+						 &root->defrag_progress);
 			if (is_extent)
 				btrfs_extent_post_op(trans, root);
 
@@ -100,10 +102,6 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
 				      path->slots[*level], &next);
 		BUG_ON(ret);
-		ret = btrfs_realloc_node(trans, root, next, cache_only,
-					 last_ret);
-		BUG_ON(ret);
-
 		if (is_extent)
 			btrfs_extent_post_op(trans, root);
 
@@ -122,8 +120,8 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	free_extent_buffer(path->nodes[*level]);
 	path->nodes[*level] = NULL;
 	*level += 1;
-	WARN_ON(ret);
-	return 0;
+	WARN_ON(ret && ret != -EAGAIN);
+	return ret;
 }
 
 static int defrag_walk_up(struct btrfs_trans_handle *trans,
@@ -147,27 +145,6 @@ static int defrag_walk_up(struct btrfs_trans_handle *trans,
 			root->defrag_level = i;
 			return 0;
 		} else {
-			if (*level > 1 && path->nodes[*level] != root->node &&
-			    btrfs_buffer_defrag(path->nodes[*level])) {
-				struct extent_buffer *next;
-				u64 last;
-				int ret;
-				ret = btrfs_cow_block(trans, root,
-						      path->nodes[*level],
-						      path->nodes[*level + 1],
-						      path->slots[*level + 1],
-						      &next);
-				BUG_ON(ret);
-				path->nodes[*level] = next;
-				last = next->start;
-				ret = btrfs_realloc_node(trans, root, next,
-							 cache_only, &last);
-				BUG_ON(ret);
-
-				if (root == root->fs_info->extent_root)
-					btrfs_extent_post_op(trans, root);
-			}
-
 			btrfs_clear_buffer_defrag(path->nodes[*level]);
 			free_extent_buffer(path->nodes[*level]);
 			path->nodes[*level] = NULL;
@@ -211,9 +188,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		extent_buffer_get(root->node);
 		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
 		BUG_ON(ret);
-		ret = btrfs_realloc_node(trans, root, root->node, cache_only,
-					 &last_ret);
-		BUG_ON(ret);
 		path->nodes[level] = root->node;
 		path->slots[level] = 0;
 		if (is_extent)
-- 
cgit v1.2.3


From 5708b9591617486bf1aa5b1a97f2c0549ec87933 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Thu, 25 Oct 2007 15:43:18 -0400
Subject: Btrfs: Tune the automatic defrag code

1) Forced defrag wasn't working properly (btrfsctl -d) because some
cache only checks were incorrect.

2) Defrag only the leaves unless in forced defrag mode.

3) Don't use complex logic to figure out if a leaf is needs defrag

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index b02355a7b143..65ef12351566 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -183,7 +183,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (level == 0) {
 		goto out;
 	}
-
 	if (root->defrag_progress.objectid == 0) {
 		extent_buffer_get(root->node);
 		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
-- 
cgit v1.2.3


From 081e95736d9118a96e9f6dcc5cec02dc75c9e1cb Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Tue, 6 Nov 2007 10:26:24 -0500
Subject: Btrfs: Make defrag check nodes against the progress key to prevent
 repeating work

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 65ef12351566..6ef1ba5f9c2f 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -227,7 +227,8 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 			break;
 		if (wret < 0)
 			ret = wret;
-		ret = -EAGAIN;
+		else
+			ret = -EAGAIN;
 		break;
 	}
 	for (i = 0; i <= orig_level; i++) {
-- 
cgit v1.2.3


From f84a8b362d9785ca1fa0598d8a90f35184bd8750 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Tue, 6 Nov 2007 10:26:29 -0500
Subject: Btrfs: Optimize allocations as we need to mix data and metadata into
 one group

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 6ef1ba5f9c2f..3994795edfeb 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -58,6 +58,10 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	if (root->fs_info->extent_root == root)
 		is_extent = 1;
 
+	if (*level == 1 && cache_only && path->nodes[1] &&
+	    !btrfs_buffer_defrag(path->nodes[1])) {
+		goto out;
+	}
 	while(*level > 0) {
 		WARN_ON(*level < 0);
 		WARN_ON(*level >= BTRFS_MAX_LEVEL);
@@ -116,7 +120,7 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	WARN_ON(*level >= BTRFS_MAX_LEVEL);
 
 	btrfs_clear_buffer_defrag(path->nodes[*level]);
-
+out:
 	free_extent_buffer(path->nodes[*level]);
 	path->nodes[*level] = NULL;
 	*level += 1;
-- 
cgit v1.2.3


From 7bb86316c3961d1bc401ef184fd996f999556c7f Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Tue, 11 Dec 2007 09:25:06 -0500
Subject: Btrfs: Add back pointers from extents to the btree or file
 referencing them

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 3994795edfeb..5c58630dce03 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -78,6 +78,8 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 			break;
 
 		if (*level == 1) {
+			WARN_ON(btrfs_header_generation(path->nodes[*level]) !=
+							trans->transid);
 			ret = btrfs_realloc_node(trans, root,
 						 path->nodes[*level],
 						 path->slots[*level],
-- 
cgit v1.2.3


From e18e4809b10e6c9efb5fe10c1ddcb4ebb690d517 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 18 Jan 2008 10:54:22 -0500
Subject: Btrfs: Add mount -o ssd, which includes optimizations for seek free
 storage

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 5c58630dce03..5935cbd8f2b8 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -179,6 +179,9 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (root->ref_cows == 0 && !is_extent)
 		goto out;
 
+	if (btrfs_test_opt(root, SSD))
+		goto out;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-- 
cgit v1.2.3


From b3236e68bf86b3ae87f58984a1822369225211cb Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Tue, 29 Jan 2008 16:06:37 -0500
Subject: Btrfs: Leave on the tree defragger in mount -o ssd, it still helps
 there

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 3 ---
 1 file changed, 3 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 5935cbd8f2b8..5c58630dce03 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -179,9 +179,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (root->ref_cows == 0 && !is_extent)
 		goto out;
 
-	if (btrfs_test_opt(root, SSD))
-		goto out;
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-- 
cgit v1.2.3


From 9afbb0b752ef30a429c45b9de6706e28ad1a36e1 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 15 Feb 2008 13:19:35 -0500
Subject: Btrfs: Disable tree defrag in SSD mode

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 5c58630dce03..5935cbd8f2b8 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -179,6 +179,9 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (root->ref_cows == 0 && !is_extent)
 		goto out;
 
+	if (btrfs_test_opt(root, SSD))
+		goto out;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
-- 
cgit v1.2.3


From 0999df54f850fe1aba29b10d5c869493af107478 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Tue, 1 Apr 2008 13:48:14 -0400
Subject: Btrfs: Verify checksums on tree blocks found without read_tree_block

Checksums were only verified by btrfs_read_tree_block, which meant the
functions to probe the page cache for blocks were not validating checksums.
Normally this is fine because the buffers will only be in cache if they
have already been validated.

But, there is a window while the buffer is being read from disk where
it could be up to date in the cache but not yet verified.  This patch
makes sure all buffers go through checksum verification before they
are used.

This is safer, and it prevents modification of buffers before they go
through the csum code.

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 5935cbd8f2b8..256af1870eef 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -101,6 +101,7 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 				path->slots[*level]++;
 				continue;
 			}
+			btrfs_verify_block_csum(root, next);
 		} else {
 			next = read_tree_block(root, bytenr,
 				       btrfs_level_size(root, *level - 1));
-- 
cgit v1.2.3


From ca7a79ad8dbe24669a8ced01f9fc0126b872745d Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 12 May 2008 12:59:19 -0400
Subject: Btrfs: Pass down the expected generation number when reading tree
 blocks

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 256af1870eef..5085e9e693b9 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -28,6 +28,7 @@ static void reada_defrag(struct btrfs_root *root,
 	int i;
 	u32 nritems;
 	u64 bytenr;
+	u64 gen;
 	u32 blocksize;
 	int ret;
 
@@ -35,7 +36,8 @@ static void reada_defrag(struct btrfs_root *root,
 	nritems = btrfs_header_nritems(node);
 	for (i = 0; i < nritems; i++) {
 		bytenr = btrfs_node_blockptr(node, i);
-		ret = readahead_tree_block(root, bytenr, blocksize);
+		gen = btrfs_node_ptr_generation(node, i);
+		ret = readahead_tree_block(root, bytenr, blocksize, gen);
 		if (ret)
 			break;
 	}
@@ -101,10 +103,11 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 				path->slots[*level]++;
 				continue;
 			}
-			btrfs_verify_block_csum(root, next);
 		} else {
 			next = read_tree_block(root, bytenr,
-				       btrfs_level_size(root, *level - 1));
+				       btrfs_level_size(root, *level - 1),
+				       btrfs_node_ptr_generation(cur,
+							 path->slots[*level]));
 		}
 		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
 				      path->slots[*level], &next);
-- 
cgit v1.2.3


From 1259ab75c62462b8ffad90067b5e1f6312786a18 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 12 May 2008 13:39:03 -0400
Subject: Btrfs: Handle write errors on raid1 and raid10

When duplicate copies exist, writes are allowed to fail to one of those
copies.  This changeset includes a few changes that allow the FS to
continue even when some IOs fail.

It also adds verification of the parent generation number for btree blocks.
This generation is stored in the pointer to a block, and it ensures
that missed writes to are detected.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 5085e9e693b9..c02e2bf2f028 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -51,6 +51,7 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 	struct extent_buffer *next;
 	struct extent_buffer *cur;
 	u64 bytenr;
+	u64 ptr_gen;
 	int ret = 0;
 	int is_extent = 0;
 
@@ -93,11 +94,12 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 			break;
 		}
 		bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
+		ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
 
 		if (cache_only) {
 			next = btrfs_find_tree_block(root, bytenr,
 					   btrfs_level_size(root, *level - 1));
-			if (!next || !btrfs_buffer_uptodate(next) ||
+			if (!next || !btrfs_buffer_uptodate(next, ptr_gen) ||
 			    !btrfs_buffer_defrag(next)) {
 				free_extent_buffer(next);
 				path->slots[*level]++;
@@ -106,8 +108,7 @@ static int defrag_walk_down(struct btrfs_trans_handle *trans,
 		} else {
 			next = read_tree_block(root, bytenr,
 				       btrfs_level_size(root, *level - 1),
-				       btrfs_node_ptr_generation(cur,
-							 path->slots[*level]));
+				       ptr_gen);
 		}
 		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
 				      path->slots[*level], &next);
-- 
cgit v1.2.3


From 0ef3e66b6700eb8f052daa8b89443ff872fbbdfc Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Sat, 24 May 2008 14:04:53 -0400
Subject: Btrfs: Allocator fix variety pack

* Force chunk allocation when find_free_extent has to do a full scan
* Record the max key at the start of defrag so it doesn't run forever
* Block groups might not be contiguous, make a forward search for the
  next block group in extent-tree.c
* Get rid of extra checks for total fs size
* Fix relocate_one_reference to avoid relocating the same file data block
  twice when referenced by an older transaction
* Use the open device count when allocating chunks so that we don't
  try to allocate from devices that don't exist

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index c02e2bf2f028..155961c7b4d5 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -198,6 +198,13 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 	if (root->defrag_progress.objectid == 0) {
+		u32 nritems;
+
+		nritems = btrfs_header_nritems(root->node);
+		root->defrag_max.objectid = 0;
+		/* from above we know this is not a leaf */
+		btrfs_node_key_to_cpu(root->node, &root->defrag_max,
+				      nritems - 1);
 		extent_buffer_get(root->node);
 		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
 		BUG_ON(ret);
@@ -254,6 +261,16 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 out:
 	if (path)
 		btrfs_free_path(path);
+	if (ret == -EAGAIN) {
+		if (root->defrag_max.objectid > root->defrag_progress.objectid)
+			goto done;
+		if (root->defrag_max.type > root->defrag_progress.type)
+			goto done;
+		if (root->defrag_max.offset > root->defrag_progress.offset)
+			goto done;
+		ret = 0;
+	}
+done:
 	if (ret != -EAGAIN) {
 		memset(&root->defrag_progress, 0,
 		       sizeof(root->defrag_progress));
-- 
cgit v1.2.3


From 925baeddc5b0764a53f2214a1253251bab0e0324 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Wed, 25 Jun 2008 16:01:30 -0400
Subject: Btrfs: Start btree concurrency work.

The allocation trees and the chunk trees are serialized via their own
dedicated mutexes.  This means allocation location is still not very
fine grained.

The main FS btree is protected by locks on each block in the btree.  Locks
are taken top / down, and as processing finishes on a given level of the
tree, the lock is released after locking the lower level.

The end result of a search is now a path where only the lowest level
is locked.  Releasing or freeing the path drops any locks held.

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 155961c7b4d5..fab851d85383 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -181,6 +181,8 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (root->fs_info->extent_root == root)
 		is_extent = 1;
 
+	goto out;
+
 	if (root->ref_cows == 0 && !is_extent)
 		goto out;
 
-- 
cgit v1.2.3


From e7a84565bcdb239caad29ccbe559ef978090ac7e Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Wed, 25 Jun 2008 16:01:31 -0400
Subject: Btrfs: Add btree locking to the tree defragmentation code

The online btree defragger is simplified and rewritten to use
standard btree searches instead of a walk up / down mechanism.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 239 ++++++++++---------------------------------------
 1 file changed, 47 insertions(+), 192 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index fab851d85383..1677e4edaf6f 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -21,167 +21,26 @@
 #include "disk-io.h"
 #include "print-tree.h"
 #include "transaction.h"
-
-static void reada_defrag(struct btrfs_root *root,
-			 struct extent_buffer *node)
-{
-	int i;
-	u32 nritems;
-	u64 bytenr;
-	u64 gen;
-	u32 blocksize;
-	int ret;
-
-	blocksize = btrfs_level_size(root, btrfs_header_level(node) - 1);
-	nritems = btrfs_header_nritems(node);
-	for (i = 0; i < nritems; i++) {
-		bytenr = btrfs_node_blockptr(node, i);
-		gen = btrfs_node_ptr_generation(node, i);
-		ret = readahead_tree_block(root, bytenr, blocksize, gen);
-		if (ret)
-			break;
-	}
-}
-
-static int defrag_walk_down(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root,
-			    struct btrfs_path *path, int *level,
-			    int cache_only, u64 *last_ret)
-{
-	struct extent_buffer *next;
-	struct extent_buffer *cur;
-	u64 bytenr;
-	u64 ptr_gen;
-	int ret = 0;
-	int is_extent = 0;
-
-	WARN_ON(*level < 0);
-	WARN_ON(*level >= BTRFS_MAX_LEVEL);
-
-	if (root->fs_info->extent_root == root)
-		is_extent = 1;
-
-	if (*level == 1 && cache_only && path->nodes[1] &&
-	    !btrfs_buffer_defrag(path->nodes[1])) {
-		goto out;
-	}
-	while(*level > 0) {
-		WARN_ON(*level < 0);
-		WARN_ON(*level >= BTRFS_MAX_LEVEL);
-		cur = path->nodes[*level];
-
-		if (!cache_only && *level > 1 && path->slots[*level] == 0)
-			reada_defrag(root, cur);
-
-		if (btrfs_header_level(cur) != *level)
-			WARN_ON(1);
-
-		if (path->slots[*level] >=
-		    btrfs_header_nritems(cur))
-			break;
-
-		if (*level == 1) {
-			WARN_ON(btrfs_header_generation(path->nodes[*level]) !=
-							trans->transid);
-			ret = btrfs_realloc_node(trans, root,
-						 path->nodes[*level],
-						 path->slots[*level],
-						 cache_only, last_ret,
-						 &root->defrag_progress);
-			if (is_extent)
-				btrfs_extent_post_op(trans, root);
-
-			break;
-		}
-		bytenr = btrfs_node_blockptr(cur, path->slots[*level]);
-		ptr_gen = btrfs_node_ptr_generation(cur, path->slots[*level]);
-
-		if (cache_only) {
-			next = btrfs_find_tree_block(root, bytenr,
-					   btrfs_level_size(root, *level - 1));
-			if (!next || !btrfs_buffer_uptodate(next, ptr_gen) ||
-			    !btrfs_buffer_defrag(next)) {
-				free_extent_buffer(next);
-				path->slots[*level]++;
-				continue;
-			}
-		} else {
-			next = read_tree_block(root, bytenr,
-				       btrfs_level_size(root, *level - 1),
-				       ptr_gen);
-		}
-		ret = btrfs_cow_block(trans, root, next, path->nodes[*level],
-				      path->slots[*level], &next);
-		BUG_ON(ret);
-		if (is_extent)
-			btrfs_extent_post_op(trans, root);
-
-		WARN_ON(*level <= 0);
-		if (path->nodes[*level-1])
-			free_extent_buffer(path->nodes[*level-1]);
-		path->nodes[*level-1] = next;
-		*level = btrfs_header_level(next);
-		path->slots[*level] = 0;
-	}
-	WARN_ON(*level < 0);
-	WARN_ON(*level >= BTRFS_MAX_LEVEL);
-
-	btrfs_clear_buffer_defrag(path->nodes[*level]);
-out:
-	free_extent_buffer(path->nodes[*level]);
-	path->nodes[*level] = NULL;
-	*level += 1;
-	WARN_ON(ret && ret != -EAGAIN);
-	return ret;
-}
-
-static int defrag_walk_up(struct btrfs_trans_handle *trans,
-			  struct btrfs_root *root,
-			  struct btrfs_path *path, int *level,
-			  int cache_only)
-{
-	int i;
-	int slot;
-	struct extent_buffer *node;
-
-	for(i = *level; i < BTRFS_MAX_LEVEL - 1 && path->nodes[i]; i++) {
-		slot = path->slots[i];
-		if (slot < btrfs_header_nritems(path->nodes[i]) - 1) {
-			path->slots[i]++;
-			*level = i;
-			node = path->nodes[i];
-			WARN_ON(i == 0);
-			btrfs_node_key_to_cpu(node, &root->defrag_progress,
-					      path->slots[i]);
-			root->defrag_level = i;
-			return 0;
-		} else {
-			btrfs_clear_buffer_defrag(path->nodes[*level]);
-			free_extent_buffer(path->nodes[*level]);
-			path->nodes[*level] = NULL;
-			*level = i + 1;
-		}
-	}
-	return 1;
-}
+#include "locking.h"
 
 int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root, int cache_only)
 {
 	struct btrfs_path *path = NULL;
-	struct extent_buffer *tmp;
+	struct btrfs_key key;
 	int ret = 0;
 	int wret;
 	int level;
 	int orig_level;
 	int i;
 	int is_extent = 0;
+	int next_key_ret = 0;
 	u64 last_ret = 0;
 
-	if (root->fs_info->extent_root == root)
+	if (root->fs_info->extent_root == root) {
+		mutex_lock(&root->fs_info->alloc_mutex);
 		is_extent = 1;
-
-	goto out;
+	}
 
 	if (root->ref_cows == 0 && !is_extent)
 		goto out;
@@ -200,67 +59,63 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 	if (root->defrag_progress.objectid == 0) {
+		struct extent_buffer *root_node;
 		u32 nritems;
 
-		nritems = btrfs_header_nritems(root->node);
+		root_node = btrfs_lock_root_node(root);
+		nritems = btrfs_header_nritems(root_node);
 		root->defrag_max.objectid = 0;
 		/* from above we know this is not a leaf */
-		btrfs_node_key_to_cpu(root->node, &root->defrag_max,
+		btrfs_node_key_to_cpu(root_node, &root->defrag_max,
 				      nritems - 1);
-		extent_buffer_get(root->node);
-		ret = btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
-		BUG_ON(ret);
-		path->nodes[level] = root->node;
-		path->slots[level] = 0;
-		if (is_extent)
-			btrfs_extent_post_op(trans, root);
+		btrfs_tree_unlock(root_node);
+		free_extent_buffer(root_node);
+		memset(&key, 0, sizeof(key));
 	} else {
-		level = root->defrag_level;
-		path->lowest_level = level;
-		wret = btrfs_search_slot(trans, root, &root->defrag_progress,
-					 path, 0, 1);
-
-		if (is_extent)
-			btrfs_extent_post_op(trans, root);
-
-		if (wret < 0) {
-			ret = wret;
-			goto out;
-		}
-
-		while(level > 0 && !path->nodes[level])
-			level--;
-
-		if (!path->nodes[level]) {
-			ret = 0;
-			goto out;
-		}
+		memcpy(&key, &root->defrag_progress, sizeof(key));
 	}
 
-	while(1) {
-		wret = defrag_walk_down(trans, root, path, &level, cache_only,
-					&last_ret);
-		if (wret > 0)
-			break;
-		if (wret < 0)
-			ret = wret;
+	path->lowest_level = 1;
+	path->keep_locks = 1;
+	wret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 
-		wret = defrag_walk_up(trans, root, path, &level, cache_only);
-		if (wret > 0)
-			break;
-		if (wret < 0)
-			ret = wret;
-		else
-			ret = -EAGAIN;
-		break;
+	if (wret < 0) {
+		ret = wret;
+		goto out;
+	}
+	if (!path->nodes[1]) {
+		ret = 0;
+		goto out;
+	}
+	path->slots[1] = btrfs_header_nritems(path->nodes[1]);
+	next_key_ret = btrfs_find_next_key(root, path, &key, 1);
+	ret = btrfs_realloc_node(trans, root,
+				 path->nodes[1], 0,
+				 cache_only, &last_ret,
+				 &root->defrag_progress);
+	WARN_ON(ret && ret != -EAGAIN);
+	if (next_key_ret == 0) {
+		memcpy(&root->defrag_progress, &key, sizeof(key));
+		ret = -EAGAIN;
 	}
-	for (i = 0; i <= orig_level; i++) {
+
+	for (i = 1; i < BTRFS_MAX_LEVEL; i++) {
+		if (path->locks[i]) {
+			btrfs_tree_unlock(path->nodes[i]);
+			path->locks[i] = 0;
+		}
 		if (path->nodes[i]) {
 			free_extent_buffer(path->nodes[i]);
 			path->nodes[i] = NULL;
 		}
 	}
+	if (is_extent)
+		btrfs_extent_post_op(trans, root);
+
 out:
+	if (is_extent)
+		mutex_unlock(&root->fs_info->alloc_mutex);
+
 	if (path)
 		btrfs_free_path(path);
 	if (ret == -EAGAIN) {
-- 
cgit v1.2.3


From 1b1e2135dc1e4efbcf25ac9ac9979316d4e1193e Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Wed, 25 Jun 2008 16:01:31 -0400
Subject: Btrfs: Add a per-inode csum mutex to avoid races creating csum items

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 7 +++++++
 1 file changed, 7 insertions(+)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 1677e4edaf6f..b17693f61fbc 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -38,8 +38,15 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	u64 last_ret = 0;
 
 	if (root->fs_info->extent_root == root) {
+		/*
+		 * there's recursion here right now in the tree locking,
+		 * we can't defrag the extent root without deadlock
+		 */
+		goto out;
+#if 0
 		mutex_lock(&root->fs_info->alloc_mutex);
 		is_extent = 1;
+#endif
 	}
 
 	if (root->ref_cows == 0 && !is_extent)
-- 
cgit v1.2.3


From 3f157a2fd2ad731e1ed9964fecdc5f459f04a4a4 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Wed, 25 Jun 2008 16:01:31 -0400
Subject: Btrfs: Online btree defragmentation fixes

The btree defragger wasn't making forward progress because the new key wasn't
being saved by the btrfs_search_forward function.

This also disables the automatic btree defrag, it wasn't scaling well to
huge filesystems.  The auto-defrag needs to be done differently.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 36 +++++++++++++++++++-----------------
 1 file changed, 19 insertions(+), 17 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index b17693f61fbc..cc2650b06952 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -32,10 +32,13 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	int wret;
 	int level;
 	int orig_level;
-	int i;
 	int is_extent = 0;
 	int next_key_ret = 0;
 	u64 last_ret = 0;
+	u64 min_trans = 0;
+
+	if (cache_only)
+		goto out;
 
 	if (root->fs_info->extent_root == root) {
 		/*
@@ -43,10 +46,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		 * we can't defrag the extent root without deadlock
 		 */
 		goto out;
-#if 0
-		mutex_lock(&root->fs_info->alloc_mutex);
-		is_extent = 1;
-#endif
 	}
 
 	if (root->ref_cows == 0 && !is_extent)
@@ -84,6 +83,17 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 
 	path->lowest_level = 1;
 	path->keep_locks = 1;
+	if (cache_only)
+		min_trans = root->defrag_trans_start;
+
+	ret = btrfs_search_forward(root, &key, path, cache_only, min_trans);
+	if (ret < 0)
+		goto out;
+	if (ret > 0) {
+		ret = 0;
+		goto out;
+	}
+	btrfs_release_path(root, path);
 	wret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 
 	if (wret < 0) {
@@ -95,7 +105,8 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 	path->slots[1] = btrfs_header_nritems(path->nodes[1]);
-	next_key_ret = btrfs_find_next_key(root, path, &key, 1);
+	next_key_ret = btrfs_find_next_key(root, path, &key, 1, cache_only,
+					   min_trans);
 	ret = btrfs_realloc_node(trans, root,
 				 path->nodes[1], 0,
 				 cache_only, &last_ret,
@@ -106,19 +117,9 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		ret = -EAGAIN;
 	}
 
-	for (i = 1; i < BTRFS_MAX_LEVEL; i++) {
-		if (path->locks[i]) {
-			btrfs_tree_unlock(path->nodes[i]);
-			path->locks[i] = 0;
-		}
-		if (path->nodes[i]) {
-			free_extent_buffer(path->nodes[i]);
-			path->nodes[i] = NULL;
-		}
-	}
+	btrfs_release_path(root, path);
 	if (is_extent)
 		btrfs_extent_post_op(trans, root);
-
 out:
 	if (is_extent)
 		mutex_unlock(&root->fs_info->alloc_mutex);
@@ -138,6 +139,7 @@ done:
 	if (ret != -EAGAIN) {
 		memset(&root->defrag_progress, 0,
 		       sizeof(root->defrag_progress));
+		root->defrag_trans_start = trans->transid;
 	}
 	return ret;
 }
-- 
cgit v1.2.3


From e02119d5a7b4396c5a872582fddc8bd6d305a70a Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Fri, 5 Sep 2008 16:13:11 -0400
Subject: Btrfs: Add a write ahead tree log to optimize synchronous operations

File syncs and directory syncs are optimized by copying their
items into a special (copy-on-write) log tree.  There is one log tree per
subvolume and the btrfs super block points to a tree of log tree roots.

After a crash, items are copied out of the log tree and back into the
subvolume.  See tree-log.c for all the details.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index cc2650b06952..b3bb5bbad76e 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -81,12 +81,12 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 		memcpy(&key, &root->defrag_progress, sizeof(key));
 	}
 
-	path->lowest_level = 1;
 	path->keep_locks = 1;
 	if (cache_only)
 		min_trans = root->defrag_trans_start;
 
-	ret = btrfs_search_forward(root, &key, path, cache_only, min_trans);
+	ret = btrfs_search_forward(root, &key, NULL, path,
+				   cache_only, min_trans);
 	if (ret < 0)
 		goto out;
 	if (ret > 0) {
-- 
cgit v1.2.3


From d352ac68148b69937d39ca5d48bcc4478e118dbf Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 29 Sep 2008 15:18:18 -0400
Subject: Btrfs: add and improve comments

This improves the comments at the top of many functions.  It didn't
dive into the guts of functions because I was trying to
avoid merging problems with the new allocator and back reference work.

extent-tree.c and volumes.c were both skipped, and there is definitely
more work todo in cleaning and commenting the code.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index b3bb5bbad76e..6f57d0889b1e 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -23,6 +23,10 @@
 #include "transaction.h"
 #include "locking.h"
 
+/* defrag all the leaves in a given btree.  If cache_only == 1, don't read things
+ * from disk, otherwise read all the leaves and try to get key order to
+ * better reflect disk order
+ */
 int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root, int cache_only)
 {
-- 
cgit v1.2.3


From 2517920135b0d29e70453e5b03d70d7b94207df3 Mon Sep 17 00:00:00 2001
From: Josef Bacik <jbacik@redhat.com>
Date: Wed, 29 Oct 2008 14:49:05 -0400
Subject: Btrfs: nuke fs wide allocation mutex V2

This patch removes the giant fs_info->alloc_mutex and replaces it with a bunch
of little locks.

There is now a pinned_mutex, which is used when messing with the pinned_extents
extent io tree, and the extent_ins_mutex which is used with the pending_del and
extent_ins extent io trees.

The locking for the extent tree stuff was inspired by a patch that Yan Zheng
wrote to fix a race condition, I cleaned it up some and changed the locking
around a little bit, but the idea remains the same.  Basically instead of
holding the extent_ins_mutex throughout the processing of an extent on the
extent_ins or pending_del trees, we just hold it while we're searching and when
we clear the bits on those trees, and lock the extent for the duration of the
operations on the extent.

Also to keep from getting hung up waiting to lock an extent, I've added a
try_lock_extent so if we cannot lock the extent, move on to the next one in the
tree and we'll come back to that one.  I have tested this heavily and it does
not appear to break anything.  This has to be applied on top of my
find_free_extent redo patch.

I tested this patch on top of Yan's space reblancing code and it worked fine.
The only thing that has changed since the last version is I pulled out all my
debugging stuff, apparently I forgot to run guilt refresh before I sent the
last patch out.  Thank you,

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

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index 6f57d0889b1e..a6a3956cedfb 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -125,9 +125,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (is_extent)
 		btrfs_extent_post_op(trans, root);
 out:
-	if (is_extent)
-		mutex_unlock(&root->fs_info->alloc_mutex);
-
 	if (path)
 		btrfs_free_path(path);
 	if (ret == -EAGAIN) {
-- 
cgit v1.2.3


From d397712bcc6a759a560fd247e6053ecae091f958 Mon Sep 17 00:00:00 2001
From: Chris Mason <chris.mason@oracle.com>
Date: Mon, 5 Jan 2009 21:25:51 -0500
Subject: Btrfs: Fix checkpatch.pl warnings

There were many, most are fixed now.  struct-funcs.c generates some warnings
but these are bogus.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
---
 fs/btrfs/tree-defrag.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

(limited to 'fs/btrfs/tree-defrag.c')

diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index a6a3956cedfb..3e8358c36165 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -23,10 +23,11 @@
 #include "transaction.h"
 #include "locking.h"
 
-/* defrag all the leaves in a given btree.  If cache_only == 1, don't read things
- * from disk, otherwise read all the leaves and try to get key order to
+/* defrag all the leaves in a given btree.  If cache_only == 1, don't read
+ * things from disk, otherwise read all the leaves and try to get key order to
  * better reflect disk order
  */
+
 int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 			struct btrfs_root *root, int cache_only)
 {
@@ -65,9 +66,9 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	level = btrfs_header_level(root->node);
 	orig_level = level;
 
-	if (level == 0) {
+	if (level == 0)
 		goto out;
-	}
+
 	if (root->defrag_progress.objectid == 0) {
 		struct extent_buffer *root_node;
 		u32 nritems;
-- 
cgit v1.2.3