From: Yan Zheng <zheng.yan@oracle.com>
To: linux-btrfs@vger.kernel.org
Subject: [PATCH 6.1/6] improve the balancing code
Date: Tue, 12 May 2009 14:06:02 +0800 [thread overview]
Message-ID: <4A0911CA.8020503@oracle.com> (raw)
This patch contains some random changes required by
the balancing code.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
---
diff -urpN 6/fs/btrfs/ctree.c 7/fs/btrfs/ctree.c
--- 6/fs/btrfs/ctree.c 2009-05-11 15:53:33.000000000 +0800
+++ 7/fs/btrfs/ctree.c 2009-05-11 16:02:34.000000000 +0800
@@ -594,7 +594,7 @@ static int comp_keys(struct btrfs_disk_k
/*
* same as comp_keys only with two btrfs_key's
*/
-static int comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2)
+int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2)
{
if (k1->objectid > k2->objectid)
return 1;
@@ -970,6 +970,12 @@ static int bin_search(struct extent_buff
return -1;
}
+int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
+ int level, int *slot)
+{
+ return bin_search(eb, key, level, slot);
+}
+
/* given a node and slot number, this reads the blocks it points to. The
* extent buffer is returned with a reference taken (but unlocked).
* NULL is returned on error.
@@ -1709,10 +1715,17 @@ int btrfs_search_slot(struct btrfs_trans
lowest_unlock = 2;
again:
- if (p->skip_locking)
- b = btrfs_root_node(root);
- else
- b = btrfs_lock_root_node(root);
+ if (p->search_commit_root) {
+ b = root->commit_root;
+ extent_buffer_get(b);
+ if (!p->skip_locking)
+ btrfs_tree_lock(b);
+ } else {
+ if (p->skip_locking)
+ b = btrfs_root_node(root);
+ else
+ b = btrfs_lock_root_node(root);
+ }
while (b) {
level = btrfs_header_level(b);
@@ -1852,138 +1865,6 @@ done:
return ret;
}
-int btrfs_merge_path(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_key *node_keys,
- u64 *nodes, int lowest_level)
-{
- struct extent_buffer *eb;
- struct extent_buffer *parent;
- struct btrfs_key key;
- u64 bytenr;
- u64 generation;
- u32 blocksize;
- int level;
- int slot;
- int key_match;
- int ret;
-
- eb = btrfs_lock_root_node(root);
- ret = btrfs_cow_block(trans, root, eb, NULL, 0, &eb);
- BUG_ON(ret);
-
- btrfs_set_lock_blocking(eb);
-
- parent = eb;
- while (1) {
- level = btrfs_header_level(parent);
- if (level == 0 || level <= lowest_level)
- break;
-
- ret = bin_search(parent, &node_keys[lowest_level], level,
- &slot);
- if (ret && slot > 0)
- slot--;
-
- bytenr = btrfs_node_blockptr(parent, slot);
- if (nodes[level - 1] == bytenr)
- break;
-
- blocksize = btrfs_level_size(root, level - 1);
- generation = btrfs_node_ptr_generation(parent, slot);
- btrfs_node_key_to_cpu(eb, &key, slot);
- key_match = !memcmp(&key, &node_keys[level - 1], sizeof(key));
-
- if (generation == trans->transid) {
- eb = read_tree_block(root, bytenr, blocksize,
- generation);
- btrfs_tree_lock(eb);
- btrfs_set_lock_blocking(eb);
- }
-
- /*
- * if node keys match and node pointer hasn't been modified
- * in the running transaction, we can merge the path. for
- * blocks owened by reloc trees, the node pointer check is
- * skipped, this is because these blocks are fully controlled
- * by the space balance code, no one else can modify them.
- */
- if (!nodes[level - 1] || !key_match ||
- (generation == trans->transid &&
- btrfs_header_owner(eb) != BTRFS_TREE_RELOC_OBJECTID)) {
- if (level == 1 || level == lowest_level + 1) {
- if (generation == trans->transid) {
- btrfs_tree_unlock(eb);
- free_extent_buffer(eb);
- }
- break;
- }
-
- if (generation != trans->transid) {
- eb = read_tree_block(root, bytenr, blocksize,
- generation);
- btrfs_tree_lock(eb);
- btrfs_set_lock_blocking(eb);
- }
-
- ret = btrfs_cow_block(trans, root, eb, parent, slot,
- &eb);
- BUG_ON(ret);
-
- if (root->root_key.objectid ==
- BTRFS_TREE_RELOC_OBJECTID) {
- if (!nodes[level - 1]) {
- nodes[level - 1] = eb->start;
- memcpy(&node_keys[level - 1], &key,
- sizeof(node_keys[0]));
- } else {
- WARN_ON(1);
- }
- }
-
- btrfs_tree_unlock(parent);
- free_extent_buffer(parent);
- parent = eb;
- continue;
- }
-
- btrfs_set_node_blockptr(parent, slot, nodes[level - 1]);
- btrfs_set_node_ptr_generation(parent, slot, trans->transid);
- btrfs_mark_buffer_dirty(parent);
-
- ret = btrfs_inc_extent_ref(trans, root,
- nodes[level - 1],
- blocksize, parent->start,
- btrfs_header_owner(parent),
- btrfs_header_generation(parent),
- level - 1);
- BUG_ON(ret);
-
- /*
- * If the block was created in the running transaction,
- * it's possible this is the last reference to it, so we
- * should drop the subtree.
- */
- if (generation == trans->transid) {
- ret = btrfs_drop_subtree(trans, root, eb, parent);
- BUG_ON(ret);
- btrfs_tree_unlock(eb);
- free_extent_buffer(eb);
- } else {
- ret = btrfs_free_extent(trans, root, bytenr,
- blocksize, parent->start,
- btrfs_header_owner(parent),
- btrfs_header_generation(parent),
- level - 1, 1);
- BUG_ON(ret);
- }
- break;
- }
- btrfs_tree_unlock(parent);
- free_extent_buffer(parent);
- return 0;
-}
-
/*
* adjust the pointers going up the tree, starting at level
* making sure the right key of each node is points to 'key'.
diff -urpN 6/fs/btrfs/ctree.h 7/fs/btrfs/ctree.h
--- 6/fs/btrfs/ctree.h 2009-05-11 15:53:33.000000000 +0800
+++ 7/fs/btrfs/ctree.h 2009-05-11 16:27:32.000000000 +0800
@@ -267,7 +267,8 @@ static inline unsigned long btrfs_chunk_
}
#define BTRFS_FSID_SIZE 16
-#define BTRFS_HEADER_FLAG_WRITTEN (1 << 0)
+#define BTRFS_HEADER_FLAG_WRITTEN (1 << 0)
+#define BTRFS_HEADER_FLAG_RELOC (1 << 1)
/*
* every tree block (leaf or node) starts with this header.
@@ -720,12 +721,7 @@ struct btrfs_block_group_cache {
struct list_head cluster_list;
};
-struct btrfs_leaf_ref_tree {
- struct rb_root root;
- struct list_head list;
- spinlock_t lock;
-};
-
+struct reloc_control;
struct btrfs_device;
struct btrfs_fs_devices;
struct btrfs_fs_info {
@@ -856,11 +852,6 @@ struct btrfs_fs_info {
struct task_struct *cleaner_kthread;
int thread_pool_size;
- /* tree relocation relocated fields */
- struct list_head dead_reloc_roots;
- struct btrfs_leaf_ref_tree reloc_ref_tree;
- struct btrfs_leaf_ref_tree shared_ref_tree;
-
struct kobject super_kobj;
struct completion kobj_unregister;
int do_barriers;
@@ -886,6 +877,8 @@ struct btrfs_fs_info {
*/
struct list_head space_info;
+ struct reloc_control *reloc_ctl;
+
spinlock_t delalloc_lock;
spinlock_t new_trans_lock;
u64 delalloc_bytes;
@@ -923,9 +916,6 @@ struct btrfs_root {
spinlock_t node_lock;
struct extent_buffer *commit_root;
- struct btrfs_leaf_ref_tree *ref_tree;
- struct btrfs_leaf_ref_tree ref_tree_struct;
- struct btrfs_dirty_root *dirty_root;
struct btrfs_root *log_root;
struct btrfs_root *reloc_root;
@@ -1907,16 +1897,9 @@ int btrfs_make_block_group(struct btrfs_
u64 size);
int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u64 group_start);
-int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start);
-int btrfs_free_reloc_root(struct btrfs_trans_handle *trans,
- struct btrfs_root *root);
-int btrfs_drop_dead_reloc_roots(struct btrfs_root *root);
-int btrfs_reloc_tree_cache_ref(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct extent_buffer *buf, u64 orig_start);
-int btrfs_add_dead_reloc_root(struct btrfs_root *root);
-int btrfs_cleanup_reloc_trees(struct btrfs_root *root);
-int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len);
+int btrfs_prepare_block_group_relocation(struct btrfs_root *root,
+ struct btrfs_block_group_cache *group);
+
u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags);
void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *ionde);
void btrfs_clear_space_info_full(struct btrfs_fs_info *info);
@@ -1931,13 +1914,12 @@ void btrfs_delalloc_reserve_space(struct
void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
u64 bytes);
/* ctree.c */
+int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
+ int level, int *slot);
+int btrfs_comp_cpu_keys(struct btrfs_key *k1, struct btrfs_key *k2);
int btrfs_previous_item(struct btrfs_root *root,
struct btrfs_path *path, u64 min_objectid,
int type);
-int btrfs_merge_path(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_key *node_keys,
- u64 *nodes, int lowest_level);
int btrfs_set_item_key_safe(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct btrfs_path *path,
struct btrfs_key *new_key);
@@ -2244,4 +2226,12 @@ int btrfs_check_acl(struct inode *inode,
int btrfs_init_acl(struct inode *inode, struct inode *dir);
int btrfs_acl_chmod(struct inode *inode);
+/* relocation.c */
+int btrfs_relocate_block_group(struct btrfs_root *root, u64 group_start);
+int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root);
+int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root);
+int btrfs_recover_relocation(struct btrfs_root *root);
+int btrfs_reloc_clone_csums(struct inode *inode, u64 file_pos, u64 len);
#endif
diff -urpN 6/fs/btrfs/disk-io.c 7/fs/btrfs/disk-io.c
--- 6/fs/btrfs/disk-io.c 2009-05-11 15:59:02.000000000 +0800
+++ 7/fs/btrfs/disk-io.c 2009-05-11 16:49:50.000000000 +0800
@@ -36,7 +36,6 @@
#include "print-tree.h"
#include "async-thread.h"
#include "locking.h"
-#include "ref-cache.h"
#include "tree-log.h"
#include "free-space-cache.h"
@@ -886,7 +885,6 @@ static int __setup_root(u32 nodesize, u3
{
root->node = NULL;
root->commit_root = NULL;
- root->ref_tree = NULL;
root->sectorsize = sectorsize;
root->nodesize = nodesize;
root->leafsize = leafsize;
@@ -922,9 +920,6 @@ static int __setup_root(u32 nodesize, u3
extent_io_tree_init(&root->dirty_log_pages,
fs_info->btree_inode->i_mapping, GFP_NOFS);
- btrfs_leaf_ref_tree_init(&root->ref_tree_struct);
- root->ref_tree = &root->ref_tree_struct;
-
memset(&root->root_key, 0, sizeof(root->root_key));
memset(&root->root_item, 0, sizeof(root->root_item));
memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
@@ -1553,7 +1548,6 @@ struct btrfs_root *open_ctree(struct sup
INIT_LIST_HEAD(&fs_info->ordered_operations);
spin_lock_init(&fs_info->delalloc_lock);
spin_lock_init(&fs_info->new_trans_lock);
- spin_lock_init(&fs_info->ref_cache_lock);
init_completion(&fs_info->kobj_unregister);
fs_info->tree_root = tree_root;
@@ -1614,10 +1608,6 @@ struct btrfs_root *open_ctree(struct sup
fs_info->btree_inode->i_mapping, GFP_NOFS);
fs_info->do_barriers = 1;
- INIT_LIST_HEAD(&fs_info->dead_reloc_roots);
- btrfs_leaf_ref_tree_init(&fs_info->reloc_ref_tree);
- btrfs_leaf_ref_tree_init(&fs_info->shared_ref_tree);
-
BTRFS_I(fs_info->btree_inode)->root = tree_root;
memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
sizeof(struct btrfs_key));
@@ -1885,7 +1875,7 @@ struct btrfs_root *open_ctree(struct sup
}
if (!(sb->s_flags & MS_RDONLY)) {
- ret = btrfs_cleanup_reloc_trees(tree_root);
+ ret = btrfs_recover_relocation(tree_root);
BUG_ON(ret);
}
diff -urpN 6/fs/btrfs/extent-tree.c 7/fs/btrfs/extent-tree.c
--- 6/fs/btrfs/extent-tree.c 2009-05-11 15:53:33.000000000 +0800
+++ 7/fs/btrfs/extent-tree.c 2009-05-11 16:46:47.000000000 +0800
@@ -30,7 +30,6 @@
#include "transaction.h"
#include "volumes.h"
#include "locking.h"
-#include "ref-cache.h"
#include "free-space-cache.h"
static int update_reserved_extents(struct btrfs_root *root,
@@ -5861,6 +5860,16 @@ static int __alloc_chunk_for_shrink(stru
return 0;
}
+int btrfs_prepare_block_group_relocation(struct btrfs_root *root,
+ struct btrfs_block_group_cache *group)
+
+{
+ __alloc_chunk_for_shrink(root, group, 1);
+ set_block_group_readonly(group);
+ return 0;
+}
+
+#if 0
static int __insert_orphan_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
u64 objectid, u64 size)
diff -urpN 6/fs/btrfs/inode.c 7/fs/btrfs/inode.c
--- 6/fs/btrfs/inode.c 2009-05-11 15:59:02.000000000 +0800
+++ 7/fs/btrfs/inode.c 2009-05-11 16:50:56.000000000 +0800
@@ -48,7 +48,6 @@
#include "ordered-data.h"
#include "xattr.h"
#include "tree-log.h"
-#include "ref-cache.h"
#include "compression.h"
#include "locking.h"
diff -urpN 6/fs/btrfs/Makefile 7/fs/btrfs/Makefile
--- 6/fs/btrfs/Makefile 2009-05-11 15:23:29.000000000 +0800
+++ 7/fs/btrfs/Makefile 2009-05-11 16:52:46.000000000 +0800
@@ -6,5 +6,5 @@ btrfs-y += super.o ctree.o extent-tree.o
transaction.o inode.o file.o tree-defrag.o \
extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \
extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \
- ref-cache.o export.o tree-log.o acl.o free-space-cache.o zlib.o \
- compression.o delayed-ref.o
+ export.o tree-log.o acl.o free-space-cache.o zlib.o \
+ compression.o delayed-ref.o relocation.o
diff -urpN 6/fs/btrfs/transaction.c 7/fs/btrfs/transaction.c
--- 6/fs/btrfs/transaction.c 2009-05-11 15:40:07.000000000 +0800
+++ 7/fs/btrfs/transaction.c 2009-05-11 16:50:27.000000000 +0800
@@ -25,7 +25,6 @@
#include "disk-io.h"
#include "transaction.h"
#include "locking.h"
-#include "ref-cache.h"
#include "tree-log.h"
#define BTRFS_ROOT_TRANS_TAG 0
diff -urpN 6/fs/btrfs/volumes.c 7/fs/btrfs/volumes.c
--- 6/fs/btrfs/volumes.c 2009-05-11 15:28:11.000000000 +0800
+++ 7/fs/btrfs/volumes.c 2009-05-11 16:02:34.000000000 +0800
@@ -1670,8 +1670,6 @@ static int btrfs_relocate_chunk(struct b
int ret;
int i;
- printk(KERN_INFO "btrfs relocating chunk %llu\n",
- (unsigned long long)chunk_offset);
root = root->fs_info->chunk_root;
extent_root = root->fs_info->extent_root;
em_tree = &root->fs_info->mapping_tree.map_tree;
reply other threads:[~2009-05-12 6:06 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4A0911CA.8020503@oracle.com \
--to=zheng.yan@oracle.com \
--cc=linux-btrfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.