dect
/
linux-2.6
Archived
13
0
Fork 0
Code Releases Activity

Merge branch 'for-chris' of git://git.jan-o-sch.net/btrfs-unstable into for-linus 

Conflicts:
	fs/btrfs/ulist.h

Signed-off-by: Chris Mason <chris.mason@oracle.com>
This commit is contained in:
Chris Mason 2012-05-31 16:50:28 -04:00
parent cfc442b696 c31931088f
commit 1e20932a23
14 changed files with 1368 additions and 252 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

491
fs/btrfs/backref.c
View File

@ -24,22 +24,135 @@
#include "delayed-ref.h" #include "delayed-ref.h"
#include "locking.h" #include "locking.h"
struct extent_inode_elem {
u64 inum;
u64 offset;
struct extent_inode_elem *next;
};
static int check_extent_in_eb(struct btrfs_key *key, struct extent_buffer *eb,
struct btrfs_file_extent_item *fi,
u64 extent_item_pos,
struct extent_inode_elem **eie)
{
u64 data_offset;
u64 data_len;
struct extent_inode_elem *e;
data_offset = btrfs_file_extent_offset(eb, fi);
data_len = btrfs_file_extent_num_bytes(eb, fi);
if (extent_item_pos < data_offset ||
extent_item_pos >= data_offset + data_len)
return 1;
e = kmalloc(sizeof(*e), GFP_NOFS);
if (!e)
return -ENOMEM;
e->next = *eie;
e->inum = key->objectid;
e->offset = key->offset + (extent_item_pos - data_offset);
*eie = e;
return 0;
}
static int find_extent_in_eb(struct extent_buffer *eb, u64 wanted_disk_byte,
u64 extent_item_pos,
struct extent_inode_elem **eie)
{
u64 disk_byte;
struct btrfs_key key;
struct btrfs_file_extent_item *fi;
int slot;
int nritems;
int extent_type;
int ret;
/*
* from the shared data ref, we only have the leaf but we need
* the key. thus, we must look into all items and see that we
* find one (some) with a reference to our extent item.
*/
nritems = btrfs_header_nritems(eb);
for (slot = 0; slot < nritems; ++slot) {
btrfs_item_key_to_cpu(eb, &key, slot);
if (key.type != BTRFS_EXTENT_DATA_KEY)
continue;
fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
extent_type = btrfs_file_extent_type(eb, fi);
if (extent_type == BTRFS_FILE_EXTENT_INLINE)
continue;
/* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
if (disk_byte != wanted_disk_byte)
continue;
ret = check_extent_in_eb(&key, eb, fi, extent_item_pos, eie);
if (ret < 0)
return ret;
}
return 0;
}
/* /*
* this structure records all encountered refs on the way up to the root * this structure records all encountered refs on the way up to the root
*/ */
struct __prelim_ref { struct __prelim_ref {
struct list_head list; struct list_head list;
u64 root_id; u64 root_id;
struct btrfs_key key; struct btrfs_key key_for_search;
int level; int level;
int count; int count;
struct extent_inode_elem *inode_list;
u64 parent; u64 parent;
u64 wanted_disk_byte; u64 wanted_disk_byte;
}; };
/*
* the rules for all callers of this function are:
* - obtaining the parent is the goal
* - if you add a key, you must know that it is a correct key
* - if you cannot add the parent or a correct key, then we will look into the
* block later to set a correct key
*
* delayed refs
* ============
* backref type | shared | indirect | shared | indirect
* information | tree | tree | data | data
* --------------------+--------+----------+--------+----------
* parent logical | y | - | - | -
* key to resolve | - | y | y | y
* tree block logical | - | - | - | -
* root for resolving | y | y | y | y
*
* - column 1: we've the parent -> done
* - column 2, 3, 4: we use the key to find the parent
*
* on disk refs (inline or keyed)
* ==============================
* backref type | shared | indirect | shared | indirect
* information | tree | tree | data | data
* --------------------+--------+----------+--------+----------
* parent logical | y | - | y | -
* key to resolve | - | - | - | y
* tree block logical | y | y | y | y
* root for resolving | - | y | y | y
*
* - column 1, 3: we've the parent -> done
* - column 2: we take the first key from the block to find the parent
* (see __add_missing_keys)
* - column 4: we use the key to find the parent
*
* additional information that's available but not required to find the parent
* block might help in merging entries to gain some speed.
*/
static int __add_prelim_ref(struct list_head *head, u64 root_id, static int __add_prelim_ref(struct list_head *head, u64 root_id,
struct btrfs_key *key, int level, u64 parent, struct btrfs_key *key, int level,
u64 wanted_disk_byte, int count) u64 parent, u64 wanted_disk_byte, int count)
{ {
struct __prelim_ref *ref; struct __prelim_ref *ref;
@ -50,10 +163,11 @@ static int __add_prelim_ref(struct list_head *head, u64 root_id,
ref->root_id = root_id; ref->root_id = root_id;
if (key) if (key)
ref->key = *key; ref->key_for_search = *key;
else else
memset(&ref->key, 0, sizeof(ref->key)); memset(&ref->key_for_search, 0, sizeof(ref->key_for_search));
ref->inode_list = NULL;
ref->level = level; ref->level = level;
ref->count = count; ref->count = count;
ref->parent = parent; ref->parent = parent;
@ -64,18 +178,26 @@ static int __add_prelim_ref(struct list_head *head, u64 root_id,
} }
static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path, static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
struct ulist *parents, struct ulist *parents, int level,
struct extent_buffer *eb, int level, struct btrfs_key *key, u64 wanted_disk_byte,
u64 wanted_objectid, u64 wanted_disk_byte) const u64 *extent_item_pos)
{ {
int ret; int ret;
int slot; int slot = path->slots[level];
struct extent_buffer *eb = path->nodes[level];
struct btrfs_file_extent_item *fi; struct btrfs_file_extent_item *fi;
struct btrfs_key key; struct extent_inode_elem *eie = NULL;
u64 disk_byte; u64 disk_byte;
u64 wanted_objectid = key->objectid;
add_parent: add_parent:
ret = ulist_add(parents, eb->start, 0, GFP_NOFS); if (level == 0 && extent_item_pos) {
fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
ret = check_extent_in_eb(key, eb, fi, *extent_item_pos, &eie);
if (ret < 0)
return ret;
}
ret = ulist_add(parents, eb->start, (unsigned long)eie, GFP_NOFS);
if (ret < 0) if (ret < 0)
return ret; return ret;
@ -89,6 +211,7 @@ add_parent:
* repeat this until we don't find any additional EXTENT_DATA items. * repeat this until we don't find any additional EXTENT_DATA items.
*/ */
while (1) { while (1) {
eie = NULL;
ret = btrfs_next_leaf(root, path); ret = btrfs_next_leaf(root, path);
if (ret < 0) if (ret < 0)
return ret; return ret;
@ -97,9 +220,9 @@ add_parent:
eb = path->nodes[0]; eb = path->nodes[0];
for (slot = 0; slot < btrfs_header_nritems(eb); ++slot) { for (slot = 0; slot < btrfs_header_nritems(eb); ++slot) {
btrfs_item_key_to_cpu(eb, &key, slot); btrfs_item_key_to_cpu(eb, key, slot);
if (key.objectid != wanted_objectid || if (key->objectid != wanted_objectid ||
key.type != BTRFS_EXTENT_DATA_KEY) key->type != BTRFS_EXTENT_DATA_KEY)
return 0; return 0;
fi = btrfs_item_ptr(eb, slot, fi = btrfs_item_ptr(eb, slot,
struct btrfs_file_extent_item); struct btrfs_file_extent_item);
@ -118,8 +241,10 @@ add_parent:
*/ */
static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info, static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
int search_commit_root, int search_commit_root,
u64 time_seq,
struct __prelim_ref *ref, struct __prelim_ref *ref,
struct ulist *parents) struct ulist *parents,
const u64 *extent_item_pos)
{ {
struct btrfs_path *path; struct btrfs_path *path;
struct btrfs_root *root; struct btrfs_root *root;
@ -152,12 +277,13 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
goto out; goto out;
path->lowest_level = level; path->lowest_level = level;
ret = btrfs_search_slot(NULL, root, &ref->key, path, 0, 0); ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq);
pr_debug("search slot in root %llu (level %d, ref count %d) returned " pr_debug("search slot in root %llu (level %d, ref count %d) returned "
"%d for key (%llu %u %llu)\n", "%d for key (%llu %u %llu)\n",
(unsigned long long)ref->root_id, level, ref->count, ret, (unsigned long long)ref->root_id, level, ref->count, ret,
(unsigned long long)ref->key.objectid, ref->key.type, (unsigned long long)ref->key_for_search.objectid,
(unsigned long long)ref->key.offset); ref->key_for_search.type,
(unsigned long long)ref->key_for_search.offset);
if (ret < 0) if (ret < 0)
goto out; goto out;
@ -179,9 +305,8 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
btrfs_item_key_to_cpu(eb, &key, path->slots[0]); btrfs_item_key_to_cpu(eb, &key, path->slots[0]);
} }
/* the last two parameters will only be used for level == 0 */ ret = add_all_parents(root, path, parents, level, &key,
ret = add_all_parents(root, path, parents, eb, level, key.objectid, ref->wanted_disk_byte, extent_item_pos);
ref->wanted_disk_byte);
out: out:
btrfs_free_path(path); btrfs_free_path(path);
return ret; return ret;
@ -191,8 +316,9 @@ out:
* resolve all indirect backrefs from the list * resolve all indirect backrefs from the list
*/ */
static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info, static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
int search_commit_root, int search_commit_root, u64 time_seq,
struct list_head *head) struct list_head *head,
const u64 *extent_item_pos)
{ {
int err; int err;
int ret = 0; int ret = 0;
@ -201,6 +327,7 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
struct __prelim_ref *new_ref; struct __prelim_ref *new_ref;
struct ulist *parents; struct ulist *parents;
struct ulist_node *node; struct ulist_node *node;
struct ulist_iterator uiter;
parents = ulist_alloc(GFP_NOFS); parents = ulist_alloc(GFP_NOFS);
if (!parents) if (!parents)
@ -217,7 +344,8 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
if (ref->count == 0) if (ref->count == 0)
continue; continue;
err = __resolve_indirect_ref(fs_info, search_commit_root, err = __resolve_indirect_ref(fs_info, search_commit_root,
ref, parents); time_seq, ref, parents,
extent_item_pos);
if (err) { if (err) {
if (ret == 0) if (ret == 0)
ret = err; ret = err;
@ -225,11 +353,14 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
} }
/* we put the first parent into the ref at hand */ /* we put the first parent into the ref at hand */
node = ulist_next(parents, NULL); ULIST_ITER_INIT(&uiter);
node = ulist_next(parents, &uiter);
ref->parent = node ? node->val : 0; ref->parent = node ? node->val : 0;
ref->inode_list =
node ? (struct extent_inode_elem *)node->aux : 0;
/* additional parents require new refs being added here */ /* additional parents require new refs being added here */
while ((node = ulist_next(parents, node))) { while ((node = ulist_next(parents, &uiter))) {
new_ref = kmalloc(sizeof(*new_ref), GFP_NOFS); new_ref = kmalloc(sizeof(*new_ref), GFP_NOFS);
if (!new_ref) { if (!new_ref) {
ret = -ENOMEM; ret = -ENOMEM;
@ -237,6 +368,8 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
} }
memcpy(new_ref, ref, sizeof(*ref)); memcpy(new_ref, ref, sizeof(*ref));
new_ref->parent = node->val; new_ref->parent = node->val;
new_ref->inode_list =
(struct extent_inode_elem *)node->aux;
list_add(&new_ref->list, &ref->list); list_add(&new_ref->list, &ref->list);
} }
ulist_reinit(parents); ulist_reinit(parents);
@ -246,10 +379,65 @@ static int __resolve_indirect_refs(struct btrfs_fs_info *fs_info,
return ret; return ret;
} }
static inline int ref_for_same_block(struct __prelim_ref *ref1,
struct __prelim_ref *ref2)
{
if (ref1->level != ref2->level)
return 0;
if (ref1->root_id != ref2->root_id)
return 0;
if (ref1->key_for_search.type != ref2->key_for_search.type)
return 0;
if (ref1->key_for_search.objectid != ref2->key_for_search.objectid)
return 0;
if (ref1->key_for_search.offset != ref2->key_for_search.offset)
return 0;
if (ref1->parent != ref2->parent)
return 0;
return 1;
}
/*
* read tree blocks and add keys where required.
*/
static int __add_missing_keys(struct btrfs_fs_info *fs_info,
struct list_head *head)
{
struct list_head *pos;
struct extent_buffer *eb;
list_for_each(pos, head) {
struct __prelim_ref *ref;
ref = list_entry(pos, struct __prelim_ref, list);
if (ref->parent)
continue;
if (ref->key_for_search.type)
continue;
BUG_ON(!ref->wanted_disk_byte);
eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
fs_info->tree_root->leafsize, 0);
BUG_ON(!eb);
btrfs_tree_read_lock(eb);
if (btrfs_header_level(eb) == 0)
btrfs_item_key_to_cpu(eb, &ref->key_for_search, 0);
else
btrfs_node_key_to_cpu(eb, &ref->key_for_search, 0);
btrfs_tree_read_unlock(eb);
free_extent_buffer(eb);
}
return 0;
}
/* /*
* merge two lists of backrefs and adjust counts accordingly * merge two lists of backrefs and adjust counts accordingly
* *
* mode = 1: merge identical keys, if key is set * mode = 1: merge identical keys, if key is set
* FIXME: if we add more keys in __add_prelim_ref, we can merge more here.
* additionally, we could even add a key range for the blocks we
* looked into to merge even more (-> replace unresolved refs by those
* having a parent).
* mode = 2: merge identical parents * mode = 2: merge identical parents
*/ */
static int __merge_refs(struct list_head *head, int mode) static int __merge_refs(struct list_head *head, int mode)
@ -263,20 +451,21 @@ static int __merge_refs(struct list_head *head, int mode)
ref1 = list_entry(pos1, struct __prelim_ref, list); ref1 = list_entry(pos1, struct __prelim_ref, list);
if (mode == 1 && ref1->key.type == 0)
continue;
for (pos2 = pos1->next, n2 = pos2->next; pos2 != head; for (pos2 = pos1->next, n2 = pos2->next; pos2 != head;
pos2 = n2, n2 = pos2->next) { pos2 = n2, n2 = pos2->next) {
struct __prelim_ref *ref2; struct __prelim_ref *ref2;
struct __prelim_ref *xchg;
ref2 = list_entry(pos2, struct __prelim_ref, list); ref2 = list_entry(pos2, struct __prelim_ref, list);
if (mode == 1) { if (mode == 1) {
if (memcmp(&ref1->key, &ref2->key, if (!ref_for_same_block(ref1, ref2))
sizeof(ref1->key)) ||
ref1->level != ref2->level ||
ref1->root_id != ref2->root_id)
continue; continue;
if (!ref1->parent && ref2->parent) {
xchg = ref1;
ref1 = ref2;
ref2 = xchg;
}
ref1->count += ref2->count; ref1->count += ref2->count;
} else { } else {
if (ref1->parent != ref2->parent) if (ref1->parent != ref2->parent)
@ -296,16 +485,17 @@ static int __merge_refs(struct list_head *head, int mode)
* smaller or equal that seq to the list * smaller or equal that seq to the list
*/ */
static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq, static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
struct btrfs_key *info_key,
struct list_head *prefs) struct list_head *prefs)
{ {
struct btrfs_delayed_extent_op *extent_op = head->extent_op; struct btrfs_delayed_extent_op *extent_op = head->extent_op;
struct rb_node *n = &head->node.rb_node; struct rb_node *n = &head->node.rb_node;
struct btrfs_key key;
struct btrfs_key op_key = {0};
int sgn; int sgn;
int ret = 0; int ret = 0;
if (extent_op && extent_op->update_key) if (extent_op && extent_op->update_key)
btrfs_disk_key_to_cpu(info_key, &extent_op->key); btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
while ((n = rb_prev(n))) { while ((n = rb_prev(n))) {
struct btrfs_delayed_ref_node *node; struct btrfs_delayed_ref_node *node;
@ -337,7 +527,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
struct btrfs_delayed_tree_ref *ref; struct btrfs_delayed_tree_ref *ref;
ref = btrfs_delayed_node_to_tree_ref(node); ref = btrfs_delayed_node_to_tree_ref(node);
ret = __add_prelim_ref(prefs, ref->root, info_key, ret = __add_prelim_ref(prefs, ref->root, &op_key,
ref->level + 1, 0, node->bytenr, ref->level + 1, 0, node->bytenr,
node->ref_mod * sgn); node->ref_mod * sgn);
break; break;
@ -346,7 +536,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
struct btrfs_delayed_tree_ref *ref; struct btrfs_delayed_tree_ref *ref;
ref = btrfs_delayed_node_to_tree_ref(node); ref = btrfs_delayed_node_to_tree_ref(node);
ret = __add_prelim_ref(prefs, ref->root, info_key, ret = __add_prelim_ref(prefs, ref->root, NULL,
ref->level + 1, ref->parent, ref->level + 1, ref->parent,
node->bytenr, node->bytenr,
node->ref_mod * sgn); node->ref_mod * sgn);
@ -354,8 +544,6 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
} }
case BTRFS_EXTENT_DATA_REF_KEY: { case BTRFS_EXTENT_DATA_REF_KEY: {
struct btrfs_delayed_data_ref *ref; struct btrfs_delayed_data_ref *ref;
struct btrfs_key key;
ref = btrfs_delayed_node_to_data_ref(node); ref = btrfs_delayed_node_to_data_ref(node);
key.objectid = ref->objectid; key.objectid = ref->objectid;
@ -368,7 +556,6 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
} }
case BTRFS_SHARED_DATA_REF_KEY: { case BTRFS_SHARED_DATA_REF_KEY: {
struct btrfs_delayed_data_ref *ref; struct btrfs_delayed_data_ref *ref;
struct btrfs_key key;
ref = btrfs_delayed_node_to_data_ref(node); ref = btrfs_delayed_node_to_data_ref(node);
@ -394,8 +581,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
*/ */
static int __add_inline_refs(struct btrfs_fs_info *fs_info, static int __add_inline_refs(struct btrfs_fs_info *fs_info,
struct btrfs_path *path, u64 bytenr, struct btrfs_path *path, u64 bytenr,
struct btrfs_key *info_key, int *info_level, int *info_level, struct list_head *prefs)
struct list_head *prefs)
{ {
int ret = 0; int ret = 0;
int slot; int slot;
@ -411,7 +597,7 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
* enumerate all inline refs * enumerate all inline refs
*/ */
leaf = path->nodes[0]; leaf = path->nodes[0];
slot = path->slots[0] - 1; slot = path->slots[0];
item_size = btrfs_item_size_nr(leaf, slot); item_size = btrfs_item_size_nr(leaf, slot);
BUG_ON(item_size < sizeof(*ei)); BUG_ON(item_size < sizeof(*ei));
@ -424,12 +610,9 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) { if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
struct btrfs_tree_block_info *info; struct btrfs_tree_block_info *info;
struct btrfs_disk_key disk_key;
info = (struct btrfs_tree_block_info *)ptr; info = (struct btrfs_tree_block_info *)ptr;
*info_level = btrfs_tree_block_level(leaf, info); *info_level = btrfs_tree_block_level(leaf, info);
btrfs_tree_block_key(leaf, info, &disk_key);
btrfs_disk_key_to_cpu(info_key, &disk_key);
ptr += sizeof(struct btrfs_tree_block_info); ptr += sizeof(struct btrfs_tree_block_info);
BUG_ON(ptr > end); BUG_ON(ptr > end);
} else { } else {
@ -447,7 +630,7 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
switch (type) { switch (type) {
case BTRFS_SHARED_BLOCK_REF_KEY: case BTRFS_SHARED_BLOCK_REF_KEY:
ret = __add_prelim_ref(prefs, 0, info_key, ret = __add_prelim_ref(prefs, 0, NULL,
*info_level + 1, offset, *info_level + 1, offset,
bytenr, 1); bytenr, 1);
break; break;
@ -462,8 +645,9 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
break; break;
} }
case BTRFS_TREE_BLOCK_REF_KEY: case BTRFS_TREE_BLOCK_REF_KEY:
ret = __add_prelim_ref(prefs, offset, info_key, ret = __add_prelim_ref(prefs, offset, NULL,
*info_level + 1, 0, bytenr, 1); *info_level + 1, 0,
bytenr, 1);
break; break;
case BTRFS_EXTENT_DATA_REF_KEY: { case BTRFS_EXTENT_DATA_REF_KEY: {
struct btrfs_extent_data_ref *dref; struct btrfs_extent_data_ref *dref;
@ -477,8 +661,8 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
key.type = BTRFS_EXTENT_DATA_KEY; key.type = BTRFS_EXTENT_DATA_KEY;
key.offset = btrfs_extent_data_ref_offset(leaf, dref); key.offset = btrfs_extent_data_ref_offset(leaf, dref);
root = btrfs_extent_data_ref_root(leaf, dref); root = btrfs_extent_data_ref_root(leaf, dref);
ret = __add_prelim_ref(prefs, root, &key, 0, 0, bytenr, ret = __add_prelim_ref(prefs, root, &key, 0, 0,
count); bytenr, count);
break; break;
} }
default: default:
@ -496,8 +680,7 @@ static int __add_inline_refs(struct btrfs_fs_info *fs_info,
*/ */
static int __add_keyed_refs(struct btrfs_fs_info *fs_info, static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
struct btrfs_path *path, u64 bytenr, struct btrfs_path *path, u64 bytenr,
struct btrfs_key *info_key, int info_level, int info_level, struct list_head *prefs)
struct list_head *prefs)
{ {
struct btrfs_root *extent_root = fs_info->extent_root; struct btrfs_root *extent_root = fs_info->extent_root;
int ret; int ret;
@ -527,7 +710,7 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
switch (key.type) { switch (key.type) {
case BTRFS_SHARED_BLOCK_REF_KEY: case BTRFS_SHARED_BLOCK_REF_KEY:
ret = __add_prelim_ref(prefs, 0, info_key, ret = __add_prelim_ref(prefs, 0, NULL,
info_level + 1, key.offset, info_level + 1, key.offset,
bytenr, 1); bytenr, 1);
break; break;
@ -543,8 +726,9 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
break; break;
} }
case BTRFS_TREE_BLOCK_REF_KEY: case BTRFS_TREE_BLOCK_REF_KEY:
ret = __add_prelim_ref(prefs, key.offset, info_key, ret = __add_prelim_ref(prefs, key.offset, NULL,
info_level + 1, 0, bytenr, 1); info_level + 1, 0,
bytenr, 1);
break; break;
case BTRFS_EXTENT_DATA_REF_KEY: { case BTRFS_EXTENT_DATA_REF_KEY: {
struct btrfs_extent_data_ref *dref; struct btrfs_extent_data_ref *dref;
@ -582,11 +766,12 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
*/ */
static int find_parent_nodes(struct btrfs_trans_handle *trans, static int find_parent_nodes(struct btrfs_trans_handle *trans,
struct btrfs_fs_info *fs_info, u64 bytenr, struct btrfs_fs_info *fs_info, u64 bytenr,
u64 seq, struct ulist *refs, struct ulist *roots) u64 delayed_ref_seq, u64 time_seq,
struct ulist *refs, struct ulist *roots,
const u64 *extent_item_pos)
{ {
struct btrfs_key key; struct btrfs_key key;
struct btrfs_path *path; struct btrfs_path *path;
struct btrfs_key info_key = { 0 };
struct btrfs_delayed_ref_root *delayed_refs = NULL; struct btrfs_delayed_ref_root *delayed_refs = NULL;
struct btrfs_delayed_ref_head *head; struct btrfs_delayed_ref_head *head;
int info_level = 0; int info_level = 0;
@ -645,7 +830,7 @@ again:
btrfs_put_delayed_ref(&head->node); btrfs_put_delayed_ref(&head->node);
goto again; goto again;
} }
ret = __add_delayed_refs(head, seq, &info_key, ret = __add_delayed_refs(head, delayed_ref_seq,
&prefs_delayed); &prefs_delayed);
if (ret) { if (ret) {
spin_unlock(&delayed_refs->lock); spin_unlock(&delayed_refs->lock);
@ -659,16 +844,17 @@ again:
struct extent_buffer *leaf; struct extent_buffer *leaf;
int slot; int slot;
path->slots[0]--;
leaf = path->nodes[0]; leaf = path->nodes[0];
slot = path->slots[0] - 1; slot = path->slots[0];
btrfs_item_key_to_cpu(leaf, &key, slot); btrfs_item_key_to_cpu(leaf, &key, slot);
if (key.objectid == bytenr && if (key.objectid == bytenr &&
key.type == BTRFS_EXTENT_ITEM_KEY) { key.type == BTRFS_EXTENT_ITEM_KEY) {
ret = __add_inline_refs(fs_info, path, bytenr, ret = __add_inline_refs(fs_info, path, bytenr,
&info_key, &info_level, &prefs); &info_level, &prefs);
if (ret) if (ret)
goto out; goto out;
ret = __add_keyed_refs(fs_info, path, bytenr, &info_key, ret = __add_keyed_refs(fs_info, path, bytenr,
info_level, &prefs); info_level, &prefs);
if (ret) if (ret)
goto out; goto out;
@ -676,21 +862,18 @@ again:
} }
btrfs_release_path(path); btrfs_release_path(path);
/*
* when adding the delayed refs above, the info_key might not have
* been known yet. Go over the list and replace the missing keys
*/
list_for_each_entry(ref, &prefs_delayed, list) {
if ((ref->key.offset | ref->key.type | ref->key.objectid) == 0)
memcpy(&ref->key, &info_key, sizeof(ref->key));
}
list_splice_init(&prefs_delayed, &prefs); list_splice_init(&prefs_delayed, &prefs);
ret = __add_missing_keys(fs_info, &prefs);
if (ret)
goto out;
ret = __merge_refs(&prefs, 1); ret = __merge_refs(&prefs, 1);
if (ret) if (ret)
goto out; goto out;
ret = __resolve_indirect_refs(fs_info, search_commit_root, &prefs); ret = __resolve_indirect_refs(fs_info, search_commit_root, time_seq,
&prefs, extent_item_pos);
if (ret) if (ret)
goto out; goto out;
@ -709,7 +892,33 @@ again:
BUG_ON(ret < 0); BUG_ON(ret < 0);
} }
if (ref->count && ref->parent) { if (ref->count && ref->parent) {
ret = ulist_add(refs, ref->parent, 0, GFP_NOFS); struct extent_inode_elem *eie = NULL;
if (extent_item_pos && !ref->inode_list) {
u32 bsz;
struct extent_buffer *eb;
bsz = btrfs_level_size(fs_info->extent_root,
info_level);
eb = read_tree_block(fs_info->extent_root,
ref->parent, bsz, 0);
BUG_ON(!eb);
ret = find_extent_in_eb(eb, bytenr,
*extent_item_pos, &eie);
ref->inode_list = eie;
free_extent_buffer(eb);
}
ret = ulist_add_merge(refs, ref->parent,
(unsigned long)ref->inode_list,
(unsigned long *)&eie, GFP_NOFS);
if (!ret && extent_item_pos) {
/*
* we've recorded that parent, so we must extend
* its inode list here
*/
BUG_ON(!eie);
while (eie->next)
eie = eie->next;
eie->next = ref->inode_list;
}
BUG_ON(ret < 0); BUG_ON(ret < 0);
} }
kfree(ref); kfree(ref);
@ -734,6 +943,28 @@ out:
return ret; return ret;
} }
static void free_leaf_list(struct ulist *blocks)
{
struct ulist_node *node = NULL;
struct extent_inode_elem *eie;
struct extent_inode_elem *eie_next;
struct ulist_iterator uiter;
ULIST_ITER_INIT(&uiter);
while ((node = ulist_next(blocks, &uiter))) {
if (!node->aux)
continue;
eie = (struct extent_inode_elem *)node->aux;
for (; eie; eie = eie_next) {
eie_next = eie->next;
kfree(eie);
}
node->aux = 0;
}
ulist_free(blocks);
}
/* /*
* Finds all leafs with a reference to the specified combination of bytenr and * Finds all leafs with a reference to the specified combination of bytenr and
* offset. key_list_head will point to a list of corresponding keys (caller must * offset. key_list_head will point to a list of corresponding keys (caller must
@ -744,7 +975,9 @@ out:
*/ */
static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans, static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
struct btrfs_fs_info *fs_info, u64 bytenr, struct btrfs_fs_info *fs_info, u64 bytenr,
u64 num_bytes, u64 seq, struct ulist **leafs) u64 delayed_ref_seq, u64 time_seq,
struct ulist **leafs,
const u64 *extent_item_pos)
{ {
struct ulist *tmp; struct ulist *tmp;
int ret; int ret;
@ -758,11 +991,12 @@ static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
return -ENOMEM; return -ENOMEM;
} }
ret = find_parent_nodes(trans, fs_info, bytenr, seq, *leafs, tmp); ret = find_parent_nodes(trans, fs_info, bytenr, delayed_ref_seq,
time_seq, *leafs, tmp, extent_item_pos);
ulist_free(tmp); ulist_free(tmp);
if (ret < 0 && ret != -ENOENT) { if (ret < 0 && ret != -ENOENT) {
ulist_free(*leafs); free_leaf_list(*leafs);
return ret; return ret;
} }
@ -784,10 +1018,12 @@ static int btrfs_find_all_leafs(struct btrfs_trans_handle *trans,
*/ */
int btrfs_find_all_roots(struct btrfs_trans_handle *trans, int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
struct btrfs_fs_info *fs_info, u64 bytenr, struct btrfs_fs_info *fs_info, u64 bytenr,
u64 num_bytes, u64 seq, struct ulist **roots) u64 delayed_ref_seq, u64 time_seq,
struct ulist **roots)
{ {
struct ulist *tmp; struct ulist *tmp;
struct ulist_node *node = NULL; struct ulist_node *node = NULL;
struct ulist_iterator uiter;
int ret; int ret;
tmp = ulist_alloc(GFP_NOFS); tmp = ulist_alloc(GFP_NOFS);
@ -799,15 +1035,16 @@ int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
return -ENOMEM; return -ENOMEM;
} }
ULIST_ITER_INIT(&uiter);
while (1) { while (1) {
ret = find_parent_nodes(trans, fs_info, bytenr, seq, ret = find_parent_nodes(trans, fs_info, bytenr, delayed_ref_seq,
tmp, *roots); time_seq, tmp, *roots, NULL);
if (ret < 0 && ret != -ENOENT) { if (ret < 0 && ret != -ENOENT) {
ulist_free(tmp); ulist_free(tmp);
ulist_free(*roots); ulist_free(*roots);
return ret; return ret;
} }
node = ulist_next(tmp, node); node = ulist_next(tmp, &uiter);
if (!node) if (!node)
break; break;
bytenr = node->val; bytenr = node->val;
@ -1093,67 +1330,25 @@ int tree_backref_for_extent(unsigned long *ptr, struct extent_buffer *eb,
return 0; return 0;
} }
static int iterate_leaf_refs(struct btrfs_fs_info *fs_info, u64 logical, static int iterate_leaf_refs(struct extent_inode_elem *inode_list,
u64 orig_extent_item_objectid, u64 root, u64 extent_item_objectid,
u64 extent_item_pos, u64 root,
iterate_extent_inodes_t *iterate, void *ctx) iterate_extent_inodes_t *iterate, void *ctx)
{ {
u64 disk_byte; struct extent_inode_elem *eie;
struct btrfs_key key;
struct btrfs_file_extent_item *fi;
struct extent_buffer *eb;
int slot;
int nritems;
int ret = 0; int ret = 0;
int extent_type;
u64 data_offset;
u64 data_len;
eb = read_tree_block(fs_info->tree_root, logical,
fs_info->tree_root->leafsize, 0);
if (!eb)
return -EIO;
/*
* from the shared data ref, we only have the leaf but we need
* the key. thus, we must look into all items and see that we
* find one (some) with a reference to our extent item.
*/
nritems = btrfs_header_nritems(eb);
for (slot = 0; slot < nritems; ++slot) {
btrfs_item_key_to_cpu(eb, &key, slot);
if (key.type != BTRFS_EXTENT_DATA_KEY)
continue;
fi = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
extent_type = btrfs_file_extent_type(eb, fi);
if (extent_type == BTRFS_FILE_EXTENT_INLINE)
continue;
/* don't skip BTRFS_FILE_EXTENT_PREALLOC, we can handle that */
disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
if (disk_byte != orig_extent_item_objectid)
continue;
data_offset = btrfs_file_extent_offset(eb, fi);
data_len = btrfs_file_extent_num_bytes(eb, fi);
if (extent_item_pos < data_offset ||
extent_item_pos >= data_offset + data_len)
continue;
for (eie = inode_list; eie; eie = eie->next) {
pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), " pr_debug("ref for %llu resolved, key (%llu EXTEND_DATA %llu), "
"root %llu\n", orig_extent_item_objectid, "root %llu\n", extent_item_objectid,
key.objectid, key.offset, root); eie->inum, eie->offset, root);
ret = iterate(key.objectid, ret = iterate(eie->inum, eie->offset, root, ctx);
key.offset + (extent_item_pos - data_offset),
root, ctx);
if (ret) { if (ret) {
pr_debug("stopping iteration because ret=%d\n", ret); pr_debug("stopping iteration for %llu due to ret=%d\n",
extent_item_objectid, ret);
break; break;
} }
} }
free_extent_buffer(eb);
return ret; return ret;
} }
@ -1175,7 +1370,10 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
struct ulist *roots = NULL; struct ulist *roots = NULL;
struct ulist_node *ref_node = NULL; struct ulist_node *ref_node = NULL;
struct ulist_node *root_node = NULL; struct ulist_node *root_node = NULL;
struct seq_list seq_elem; struct seq_list seq_elem = {};
struct seq_list tree_mod_seq_elem = {};
struct ulist_iterator ref_uiter;
struct ulist_iterator root_uiter;
struct btrfs_delayed_ref_root *delayed_refs = NULL; struct btrfs_delayed_ref_root *delayed_refs = NULL;
pr_debug("resolving all inodes for extent %llu\n", pr_debug("resolving all inodes for extent %llu\n",
@ -1192,34 +1390,41 @@ int iterate_extent_inodes(struct btrfs_fs_info *fs_info,
spin_lock(&delayed_refs->lock); spin_lock(&delayed_refs->lock);
btrfs_get_delayed_seq(delayed_refs, &seq_elem); btrfs_get_delayed_seq(delayed_refs, &seq_elem);
spin_unlock(&delayed_refs->lock); spin_unlock(&delayed_refs->lock);
btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
} }
ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid, ret = btrfs_find_all_leafs(trans, fs_info, extent_item_objectid,
extent_item_pos, seq_elem.seq, seq_elem.seq, tree_mod_seq_elem.seq, &refs,
&refs); &extent_item_pos);
if (ret) if (ret)
goto out; goto out;
while (!ret && (ref_node = ulist_next(refs, ref_node))) { ULIST_ITER_INIT(&ref_uiter);
ret = btrfs_find_all_roots(trans, fs_info, ref_node->val, -1, while (!ret && (ref_node = ulist_next(refs, &ref_uiter))) {
seq_elem.seq, &roots); ret = btrfs_find_all_roots(trans, fs_info, ref_node->val,
seq_elem.seq,
tree_mod_seq_elem.seq, &roots);
if (ret) if (ret)
break; break;
while (!ret && (root_node = ulist_next(roots, root_node))) { ULIST_ITER_INIT(&root_uiter);
pr_debug("root %llu references leaf %llu\n", while (!ret && (root_node = ulist_next(roots, &root_uiter))) {
root_node->val, ref_node->val); pr_debug("root %llu references leaf %llu, data list "
ret = iterate_leaf_refs(fs_info, ref_node->val, "%#lx\n", root_node->val, ref_node->val,
extent_item_objectid, ref_node->aux);
extent_item_pos, root_node->val, ret = iterate_leaf_refs(
(struct extent_inode_elem *)ref_node->aux,
root_node->val, extent_item_objectid,
iterate, ctx); iterate, ctx);
} }
ulist_free(roots);
roots = NULL;
} }
ulist_free(refs); free_leaf_list(refs);
ulist_free(roots); ulist_free(roots);
out: out:
if (!search_commit_root) { if (!search_commit_root) {
btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
btrfs_put_delayed_seq(delayed_refs, &seq_elem); btrfs_put_delayed_seq(delayed_refs, &seq_elem);
btrfs_end_transaction(trans, fs_info->extent_root); btrfs_end_transaction(trans, fs_info->extent_root);
} }

3
fs/btrfs/backref.h
View File

@ -58,7 +58,8 @@ int paths_from_inode(u64 inum, struct inode_fs_paths *ipath);
int btrfs_find_all_roots(struct btrfs_trans_handle *trans, int btrfs_find_all_roots(struct btrfs_trans_handle *trans,
struct btrfs_fs_info *fs_info, u64 bytenr, struct btrfs_fs_info *fs_info, u64 bytenr,
u64 num_bytes, u64 seq, struct ulist **roots); u64 delayed_ref_seq, u64 time_seq,
struct ulist **roots);
struct btrfs_data_container *init_data_container(u32 total_bytes); struct btrfs_data_container *init_data_container(u32 total_bytes);
struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root, struct inode_fs_paths *init_ipath(s32 total_bytes, struct btrfs_root *fs_root,

849
fs/btrfs/ctree.c
View File

File diff suppressed because it is too large Load Diff

34
fs/btrfs/ctree.h
View File

@ -1140,6 +1140,15 @@ struct btrfs_fs_info {
spinlock_t delayed_iput_lock; spinlock_t delayed_iput_lock;
struct list_head delayed_iputs; struct list_head delayed_iputs;
/* this protects tree_mod_seq_list */
spinlock_t tree_mod_seq_lock;
atomic_t tree_mod_seq;
struct list_head tree_mod_seq_list;
/* this protects tree_mod_log */
rwlock_t tree_mod_log_lock;
struct rb_root tree_mod_log;
atomic_t nr_async_submits; atomic_t nr_async_submits;
atomic_t async_submit_draining; atomic_t async_submit_draining;
atomic_t nr_async_bios; atomic_t nr_async_bios;
@ -2537,11 +2546,11 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u32 blocksize, struct btrfs_root *root, u32 blocksize,
u64 parent, u64 root_objectid, u64 parent, u64 root_objectid,
struct btrfs_disk_key *key, int level, struct btrfs_disk_key *key, int level,
u64 hint, u64 empty_size, int for_cow); u64 hint, u64 empty_size);
void btrfs_free_tree_block(struct btrfs_trans_handle *trans, void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct btrfs_root *root,
struct extent_buffer *buf, struct extent_buffer *buf,
u64 parent, int last_ref, int for_cow); u64 parent, int last_ref);
struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans, struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct btrfs_root *root,
u64 bytenr, u32 blocksize, u64 bytenr, u32 blocksize,
@ -2700,6 +2709,8 @@ int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
*root, struct btrfs_key *key, struct btrfs_path *p, int *root, struct btrfs_key *key, struct btrfs_path *p, int
ins_len, int cow); ins_len, int cow);
int btrfs_search_old_slot(struct btrfs_root *root, struct btrfs_key *key,
struct btrfs_path *p, u64 time_seq);
int btrfs_realloc_node(struct btrfs_trans_handle *trans, int btrfs_realloc_node(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct extent_buffer *parent, struct btrfs_root *root, struct extent_buffer *parent,
int start_slot, int cache_only, u64 *last_ret, int start_slot, int cache_only, u64 *last_ret,
@ -3139,4 +3150,23 @@ void btrfs_reada_detach(void *handle);
int btree_readahead_hook(struct btrfs_root *root, struct extent_buffer *eb, int btree_readahead_hook(struct btrfs_root *root, struct extent_buffer *eb,
u64 start, int err); u64 start, int err);
/* delayed seq elem */
struct seq_list {
struct list_head list;
u64 seq;
u32 flags;
};
void btrfs_get_tree_mod_seq(struct btrfs_fs_info *fs_info,
struct seq_list *elem);
void btrfs_put_tree_mod_seq(struct btrfs_fs_info *fs_info,
struct seq_list *elem);
static inline int is_fstree(u64 rootid)
{
if (rootid == BTRFS_FS_TREE_OBJECTID ||
(s64)rootid >= (s64)BTRFS_FIRST_FREE_OBJECTID)
return 1;
return 0;
}
#endif #endif

10
fs/btrfs/delayed-ref.c
View File

@ -525,7 +525,7 @@ static noinline void add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
ref->is_head = 0; ref->is_head = 0;
ref->in_tree = 1; ref->in_tree = 1;
if (need_ref_seq(for_cow, ref_root)) if (is_fstree(ref_root))
seq = inc_delayed_seq(delayed_refs); seq = inc_delayed_seq(delayed_refs);
ref->seq = seq; ref->seq = seq;
@ -584,7 +584,7 @@ static noinline void add_delayed_data_ref(struct btrfs_fs_info *fs_info,
ref->is_head = 0; ref->is_head = 0;
ref->in_tree = 1; ref->in_tree = 1;
if (need_ref_seq(for_cow, ref_root)) if (is_fstree(ref_root))
seq = inc_delayed_seq(delayed_refs); seq = inc_delayed_seq(delayed_refs);
ref->seq = seq; ref->seq = seq;
@ -658,10 +658,11 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr, add_delayed_tree_ref(fs_info, trans, &ref->node, bytenr,
num_bytes, parent, ref_root, level, action, num_bytes, parent, ref_root, level, action,
for_cow); for_cow);
if (!need_ref_seq(for_cow, ref_root) && if (!is_fstree(ref_root) &&
waitqueue_active(&delayed_refs->seq_wait)) waitqueue_active(&delayed_refs->seq_wait))
wake_up(&delayed_refs->seq_wait); wake_up(&delayed_refs->seq_wait);
spin_unlock(&delayed_refs->lock); spin_unlock(&delayed_refs->lock);
return 0; return 0;
} }
@ -706,10 +707,11 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
add_delayed_data_ref(fs_info, trans, &ref->node, bytenr, add_delayed_data_ref(fs_info, trans, &ref->node, bytenr,
num_bytes, parent, ref_root, owner, offset, num_bytes, parent, ref_root, owner, offset,
action, for_cow); action, for_cow);
if (!need_ref_seq(for_cow, ref_root) && if (!is_fstree(ref_root) &&
waitqueue_active(&delayed_refs->seq_wait)) waitqueue_active(&delayed_refs->seq_wait))
wake_up(&delayed_refs->seq_wait); wake_up(&delayed_refs->seq_wait);
spin_unlock(&delayed_refs->lock); spin_unlock(&delayed_refs->lock);
return 0; return 0;
} }

24
fs/btrfs/delayed-ref.h
View File

@ -195,11 +195,6 @@ int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
int btrfs_find_ref_cluster(struct btrfs_trans_handle *trans, int btrfs_find_ref_cluster(struct btrfs_trans_handle *trans,
struct list_head *cluster, u64 search_start); struct list_head *cluster, u64 search_start);
struct seq_list {
struct list_head list;
u64 seq;
};
static inline u64 inc_delayed_seq(struct btrfs_delayed_ref_root *delayed_refs) static inline u64 inc_delayed_seq(struct btrfs_delayed_ref_root *delayed_refs)
{ {
assert_spin_locked(&delayed_refs->lock); assert_spin_locked(&delayed_refs->lock);
@ -229,25 +224,6 @@ btrfs_put_delayed_seq(struct btrfs_delayed_ref_root *delayed_refs,
int btrfs_check_delayed_seq(struct btrfs_delayed_ref_root *delayed_refs, int btrfs_check_delayed_seq(struct btrfs_delayed_ref_root *delayed_refs,
u64 seq); u64 seq);
/*
* delayed refs with a ref_seq > 0 must be held back during backref walking.
* this only applies to items in one of the fs-trees. for_cow items never need
* to be held back, so they won't get a ref_seq number.
*/
static inline int need_ref_seq(int for_cow, u64 rootid)
{
if (for_cow)
return 0;
if (rootid == BTRFS_FS_TREE_OBJECTID)
return 1;
if ((s64)rootid >= (s64)BTRFS_FIRST_FREE_OBJECTID)
return 1;
return 0;
}
/* /*
* a node might live in a head or a regular ref, this lets you * a node might live in a head or a regular ref, this lets you
* test for the proper type to use. * test for the proper type to use.

7
fs/btrfs/disk-io.c
View File

@ -1252,7 +1252,7 @@ static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0, leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
BTRFS_TREE_LOG_OBJECTID, NULL, BTRFS_TREE_LOG_OBJECTID, NULL,
0, 0, 0, 0); 0, 0, 0);
if (IS_ERR(leaf)) { if (IS_ERR(leaf)) {
kfree(root); kfree(root);
return ERR_CAST(leaf); return ERR_CAST(leaf);
@ -1914,11 +1914,14 @@ int open_ctree(struct super_block *sb,
spin_lock_init(&fs_info->delayed_iput_lock); spin_lock_init(&fs_info->delayed_iput_lock);
spin_lock_init(&fs_info->defrag_inodes_lock); spin_lock_init(&fs_info->defrag_inodes_lock);
spin_lock_init(&fs_info->free_chunk_lock); spin_lock_init(&fs_info->free_chunk_lock);
spin_lock_init(&fs_info->tree_mod_seq_lock);
rwlock_init(&fs_info->tree_mod_log_lock);
mutex_init(&fs_info->reloc_mutex); mutex_init(&fs_info->reloc_mutex);
init_completion(&fs_info->kobj_unregister); init_completion(&fs_info->kobj_unregister);
INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots); INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
INIT_LIST_HEAD(&fs_info->space_info); INIT_LIST_HEAD(&fs_info->space_info);
INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
btrfs_mapping_init(&fs_info->mapping_tree); btrfs_mapping_init(&fs_info->mapping_tree);
btrfs_init_block_rsv(&fs_info->global_block_rsv); btrfs_init_block_rsv(&fs_info->global_block_rsv);
btrfs_init_block_rsv(&fs_info->delalloc_block_rsv); btrfs_init_block_rsv(&fs_info->delalloc_block_rsv);
@ -1931,12 +1934,14 @@ int open_ctree(struct super_block *sb,
atomic_set(&fs_info->async_submit_draining, 0); atomic_set(&fs_info->async_submit_draining, 0);
atomic_set(&fs_info->nr_async_bios, 0); atomic_set(&fs_info->nr_async_bios, 0);
atomic_set(&fs_info->defrag_running, 0); atomic_set(&fs_info->defrag_running, 0);
atomic_set(&fs_info->tree_mod_seq, 0);
fs_info->sb = sb; fs_info->sb = sb;
fs_info->max_inline = 8192 * 1024; fs_info->max_inline = 8192 * 1024;
fs_info->metadata_ratio = 0; fs_info->metadata_ratio = 0;
fs_info->defrag_inodes = RB_ROOT; fs_info->defrag_inodes = RB_ROOT;
fs_info->trans_no_join = 0; fs_info->trans_no_join = 0;
fs_info->free_chunk_space = 0; fs_info->free_chunk_space = 0;
fs_info->tree_mod_log = RB_ROOT;
/* readahead state */ /* readahead state */
INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_WAIT); INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_WAIT);

10
fs/btrfs/extent-tree.c
View File

@ -5218,7 +5218,7 @@ out:
void btrfs_free_tree_block(struct btrfs_trans_handle *trans, void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct btrfs_root *root,
struct extent_buffer *buf, struct extent_buffer *buf,
u64 parent, int last_ref, int for_cow) u64 parent, int last_ref)
{ {
struct btrfs_block_group_cache *cache = NULL; struct btrfs_block_group_cache *cache = NULL;
int ret; int ret;
@ -5228,7 +5228,7 @@ void btrfs_free_tree_block(struct btrfs_trans_handle *trans,
buf->start, buf->len, buf->start, buf->len,
parent, root->root_key.objectid, parent, root->root_key.objectid,
btrfs_header_level(buf), btrfs_header_level(buf),
BTRFS_DROP_DELAYED_REF, NULL, for_cow); BTRFS_DROP_DELAYED_REF, NULL, 0);
BUG_ON(ret); /* -ENOMEM */ BUG_ON(ret); /* -ENOMEM */
} }
@ -6250,7 +6250,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
struct btrfs_root *root, u32 blocksize, struct btrfs_root *root, u32 blocksize,
u64 parent, u64 root_objectid, u64 parent, u64 root_objectid,
struct btrfs_disk_key *key, int level, struct btrfs_disk_key *key, int level,
u64 hint, u64 empty_size, int for_cow) u64 hint, u64 empty_size)
{ {
struct btrfs_key ins; struct btrfs_key ins;
struct btrfs_block_rsv *block_rsv; struct btrfs_block_rsv *block_rsv;
@ -6298,7 +6298,7 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
ins.objectid, ins.objectid,
ins.offset, parent, root_objectid, ins.offset, parent, root_objectid,
level, BTRFS_ADD_DELAYED_EXTENT, level, BTRFS_ADD_DELAYED_EXTENT,
extent_op, for_cow); extent_op, 0);
BUG_ON(ret); /* -ENOMEM */ BUG_ON(ret); /* -ENOMEM */
} }
return buf; return buf;
@ -6716,7 +6716,7 @@ static noinline int walk_up_proc(struct btrfs_trans_handle *trans,
btrfs_header_owner(path->nodes[level + 1])); btrfs_header_owner(path->nodes[level + 1]));
} }
btrfs_free_tree_block(trans, root, eb, parent, wc->refs[level] == 1, 0); btrfs_free_tree_block(trans, root, eb, parent, wc->refs[level] == 1);
out: out:
wc->refs[level] = 0; wc->refs[level] = 0;
wc->flags[level] = 0; wc->flags[level] = 0;

68
fs/btrfs/extent_io.c
View File

@ -3924,6 +3924,7 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
eb->start = start; eb->start = start;
eb->len = len; eb->len = len;
eb->tree = tree; eb->tree = tree;
eb->bflags = 0;
rwlock_init(&eb->lock); rwlock_init(&eb->lock);
atomic_set(&eb->write_locks, 0); atomic_set(&eb->write_locks, 0);
atomic_set(&eb->read_locks, 0); atomic_set(&eb->read_locks, 0);
@ -3961,6 +3962,60 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
return eb; return eb;
} }
struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src)
{
unsigned long i;
struct page *p;
struct extent_buffer *new;
unsigned long num_pages = num_extent_pages(src->start, src->len);
new = __alloc_extent_buffer(NULL, src->start, src->len, GFP_ATOMIC);
if (new == NULL)
return NULL;
for (i = 0; i < num_pages; i++) {
p = alloc_page(GFP_ATOMIC);
BUG_ON(!p);
attach_extent_buffer_page(new, p);
WARN_ON(PageDirty(p));
SetPageUptodate(p);
new->pages[i] = p;
}
copy_extent_buffer(new, src, 0, 0, src->len);
set_bit(EXTENT_BUFFER_UPTODATE, &new->bflags);
set_bit(EXTENT_BUFFER_DUMMY, &new->bflags);
return new;
}
struct extent_buffer *alloc_dummy_extent_buffer(u64 start, unsigned long len)
{
struct extent_buffer *eb;
unsigned long num_pages = num_extent_pages(0, len);
unsigned long i;
eb = __alloc_extent_buffer(NULL, start, len, GFP_ATOMIC);
if (!eb)
return NULL;
for (i = 0; i < num_pages; i++) {
eb->pages[i] = alloc_page(GFP_ATOMIC);
if (!eb->pages[i])
goto err;
}
set_extent_buffer_uptodate(eb);
btrfs_set_header_nritems(eb, 0);
set_bit(EXTENT_BUFFER_DUMMY, &eb->bflags);
return eb;
err:
for (i--; i > 0; i--)
__free_page(eb->pages[i]);
__free_extent_buffer(eb);
return NULL;
}
static int extent_buffer_under_io(struct extent_buffer *eb) static int extent_buffer_under_io(struct extent_buffer *eb)
{ {
return (atomic_read(&eb->io_pages) || return (atomic_read(&eb->io_pages) ||
@ -3977,6 +4032,7 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
unsigned long index; unsigned long index;
unsigned long num_pages; unsigned long num_pages;
struct page *page; struct page *page;
int mapped = !test_bit(EXTENT_BUFFER_DUMMY, &eb->bflags);
BUG_ON(extent_buffer_under_io(eb)); BUG_ON(extent_buffer_under_io(eb));
@ -3988,7 +4044,7 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
do { do {
index--; index--;
page = extent_buffer_page(eb, index); page = extent_buffer_page(eb, index);
if (page) { if (page && mapped) {
spin_lock(&page->mapping->private_lock); spin_lock(&page->mapping->private_lock);
/* /*
* We do this since we'll remove the pages after we've * We do this since we'll remove the pages after we've
@ -4013,6 +4069,8 @@ static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
} }
spin_unlock(&page->mapping->private_lock); spin_unlock(&page->mapping->private_lock);
}
if (page) {
/* One for when we alloced the page */ /* One for when we alloced the page */
page_cache_release(page); page_cache_release(page);
} }
@ -4231,6 +4289,9 @@ static void release_extent_buffer(struct extent_buffer *eb, gfp_t mask)
{ {
WARN_ON(atomic_read(&eb->refs) == 0); WARN_ON(atomic_read(&eb->refs) == 0);
if (atomic_dec_and_test(&eb->refs)) { if (atomic_dec_and_test(&eb->refs)) {
if (test_bit(EXTENT_BUFFER_DUMMY, &eb->bflags)) {
spin_unlock(&eb->refs_lock);
} else {
struct extent_io_tree *tree = eb->tree; struct extent_io_tree *tree = eb->tree;
spin_unlock(&eb->refs_lock); spin_unlock(&eb->refs_lock);
@ -4239,6 +4300,7 @@ static void release_extent_buffer(struct extent_buffer *eb, gfp_t mask)
radix_tree_delete(&tree->buffer, radix_tree_delete(&tree->buffer,
eb->start >> PAGE_CACHE_SHIFT); eb->start >> PAGE_CACHE_SHIFT);
spin_unlock(&tree->buffer_lock); spin_unlock(&tree->buffer_lock);
}
/* Should be safe to release our pages at this point */ /* Should be safe to release our pages at this point */
btrfs_release_extent_buffer_page(eb, 0); btrfs_release_extent_buffer_page(eb, 0);
@ -4255,6 +4317,10 @@ void free_extent_buffer(struct extent_buffer *eb)
return; return;
spin_lock(&eb->refs_lock); spin_lock(&eb->refs_lock);
if (atomic_read(&eb->refs) == 2 &&
test_bit(EXTENT_BUFFER_DUMMY, &eb->bflags))
atomic_dec(&eb->refs);
if (atomic_read(&eb->refs) == 2 && if (atomic_read(&eb->refs) == 2 &&
test_bit(EXTENT_BUFFER_STALE, &eb->bflags) && test_bit(EXTENT_BUFFER_STALE, &eb->bflags) &&
!extent_buffer_under_io(eb) && !extent_buffer_under_io(eb) &&

3
fs/btrfs/extent_io.h
View File

@ -39,6 +39,7 @@
#define EXTENT_BUFFER_STALE 6 #define EXTENT_BUFFER_STALE 6
#define EXTENT_BUFFER_WRITEBACK 7 #define EXTENT_BUFFER_WRITEBACK 7
#define EXTENT_BUFFER_IOERR 8 #define EXTENT_BUFFER_IOERR 8
#define EXTENT_BUFFER_DUMMY 9
/* these are flags for extent_clear_unlock_delalloc */ /* these are flags for extent_clear_unlock_delalloc */
#define EXTENT_CLEAR_UNLOCK_PAGE 0x1 #define EXTENT_CLEAR_UNLOCK_PAGE 0x1
@ -264,6 +265,8 @@ void set_page_extent_mapped(struct page *page);
struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
u64 start, unsigned long len); u64 start, unsigned long len);
struct extent_buffer *alloc_dummy_extent_buffer(u64 start, unsigned long len);
struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src);
struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree, struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
u64 start, unsigned long len); u64 start, unsigned long len);
void free_extent_buffer(struct extent_buffer *eb); void free_extent_buffer(struct extent_buffer *eb);

2
fs/btrfs/ioctl.c
View File

@ -368,7 +368,7 @@ static noinline int create_subvol(struct btrfs_root *root,
return PTR_ERR(trans); return PTR_ERR(trans);
leaf = btrfs_alloc_free_block(trans, root, root->leafsize, leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
0, objectid, NULL, 0, 0, 0, 0); 0, objectid, NULL, 0, 0, 0);
if (IS_ERR(leaf)) { if (IS_ERR(leaf)) {
ret = PTR_ERR(leaf); ret = PTR_ERR(leaf);
goto fail; goto fail;

55
fs/btrfs/transaction.c
View File

@ -56,48 +56,49 @@ static noinline void switch_commit_root(struct btrfs_root *root)
static noinline int join_transaction(struct btrfs_root *root, int nofail) static noinline int join_transaction(struct btrfs_root *root, int nofail)
{ {
struct btrfs_transaction *cur_trans; struct btrfs_transaction *cur_trans;
struct btrfs_fs_info *fs_info = root->fs_info;
spin_lock(&root->fs_info->trans_lock); spin_lock(&fs_info->trans_lock);
loop: loop:
/* The file system has been taken offline. No new transactions. */ /* The file system has been taken offline. No new transactions. */
if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) { if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
spin_unlock(&root->fs_info->trans_lock); spin_unlock(&fs_info->trans_lock);
return -EROFS; return -EROFS;
} }
if (root->fs_info->trans_no_join) { if (fs_info->trans_no_join) {
if (!nofail) { if (!nofail) {
spin_unlock(&root->fs_info->trans_lock); spin_unlock(&fs_info->trans_lock);
return -EBUSY; return -EBUSY;
} }
} }
cur_trans = root->fs_info->running_transaction; cur_trans = fs_info->running_transaction;
if (cur_trans) { if (cur_trans) {
if (cur_trans->aborted) { if (cur_trans->aborted) {
spin_unlock(&root->fs_info->trans_lock); spin_unlock(&fs_info->trans_lock);
return cur_trans->aborted; return cur_trans->aborted;
} }
atomic_inc(&cur_trans->use_count); atomic_inc(&cur_trans->use_count);
atomic_inc(&cur_trans->num_writers); atomic_inc(&cur_trans->num_writers);
cur_trans->num_joined++; cur_trans->num_joined++;
spin_unlock(&root->fs_info->trans_lock); spin_unlock(&fs_info->trans_lock);
return 0; return 0;
} }
spin_unlock(&root->fs_info->trans_lock); spin_unlock(&fs_info->trans_lock);
cur_trans = kmem_cache_alloc(btrfs_transaction_cachep, GFP_NOFS); cur_trans = kmem_cache_alloc(btrfs_transaction_cachep, GFP_NOFS);
if (!cur_trans) if (!cur_trans)
return -ENOMEM; return -ENOMEM;
spin_lock(&root->fs_info->trans_lock); spin_lock(&fs_info->trans_lock);
if (root->fs_info->running_transaction) { if (fs_info->running_transaction) {
/* /*
* someone started a transaction after we unlocked. Make sure * someone started a transaction after we unlocked. Make sure
* to redo the trans_no_join checks above * to redo the trans_no_join checks above
*/ */
kmem_cache_free(btrfs_transaction_cachep, cur_trans); kmem_cache_free(btrfs_transaction_cachep, cur_trans);
cur_trans = root->fs_info->running_transaction; cur_trans = fs_info->running_transaction;
goto loop; goto loop;
} }
@ -122,20 +123,38 @@ loop:
cur_trans->delayed_refs.flushing = 0; cur_trans->delayed_refs.flushing = 0;
cur_trans->delayed_refs.run_delayed_start = 0; cur_trans->delayed_refs.run_delayed_start = 0;
cur_trans->delayed_refs.seq = 1; cur_trans->delayed_refs.seq = 1;
/*
* although the tree mod log is per file system and not per transaction,
* the log must never go across transaction boundaries.
*/
smp_mb();
if (!list_empty(&fs_info->tree_mod_seq_list)) {
printk(KERN_ERR "btrfs: tree_mod_seq_list not empty when "
"creating a fresh transaction\n");
WARN_ON(1);
}
if (!RB_EMPTY_ROOT(&fs_info->tree_mod_log)) {
printk(KERN_ERR "btrfs: tree_mod_log rb tree not empty when "
"creating a fresh transaction\n");
WARN_ON(1);
}
atomic_set(&fs_info->tree_mod_seq, 0);
init_waitqueue_head(&cur_trans->delayed_refs.seq_wait); init_waitqueue_head(&cur_trans->delayed_refs.seq_wait);
spin_lock_init(&cur_trans->commit_lock); spin_lock_init(&cur_trans->commit_lock);
spin_lock_init(&cur_trans->delayed_refs.lock); spin_lock_init(&cur_trans->delayed_refs.lock);
INIT_LIST_HEAD(&cur_trans->delayed_refs.seq_head); INIT_LIST_HEAD(&cur_trans->delayed_refs.seq_head);
INIT_LIST_HEAD(&cur_trans->pending_snapshots); INIT_LIST_HEAD(&cur_trans->pending_snapshots);
list_add_tail(&cur_trans->list, &root->fs_info->trans_list); list_add_tail(&cur_trans->list, &fs_info->trans_list);
extent_io_tree_init(&cur_trans->dirty_pages, extent_io_tree_init(&cur_trans->dirty_pages,
root->fs_info->btree_inode->i_mapping); fs_info->btree_inode->i_mapping);
root->fs_info->generation++; fs_info->generation++;
cur_trans->transid = root->fs_info->generation; cur_trans->transid = fs_info->generation;
root->fs_info->running_transaction = cur_trans; fs_info->running_transaction = cur_trans;
cur_trans->aborted = 0; cur_trans->aborted = 0;
spin_unlock(&root->fs_info->trans_lock); spin_unlock(&fs_info->trans_lock);
return 0; return 0;
} }

34
fs/btrfs/ulist.c
View File

@ -23,9 +23,9 @@
* *
* ulist = ulist_alloc(); * ulist = ulist_alloc();
* ulist_add(ulist, root); * ulist_add(ulist, root);
* elem = NULL; * ULIST_ITER_INIT(&uiter);
* *
* while ((elem = ulist_next(ulist, elem)) { * while ((elem = ulist_next(ulist, &uiter)) {
* for (all child nodes n in elem) * for (all child nodes n in elem)
* ulist_add(ulist, n); * ulist_add(ulist, n);
* do something useful with the node; * do something useful with the node;
@ -145,13 +145,22 @@ EXPORT_SYMBOL(ulist_free);
*/ */
int ulist_add(struct ulist *ulist, u64 val, unsigned long aux, int ulist_add(struct ulist *ulist, u64 val, unsigned long aux,
gfp_t gfp_mask) gfp_t gfp_mask)
{
return ulist_add_merge(ulist, val, aux, NULL, gfp_mask);
}
int ulist_add_merge(struct ulist *ulist, u64 val, unsigned long aux,
unsigned long *old_aux, gfp_t gfp_mask)
{ {
int i; int i;
for (i = 0; i < ulist->nnodes; ++i) { for (i = 0; i < ulist->nnodes; ++i) {
if (ulist->nodes[i].val == val) if (ulist->nodes[i].val == val) {
if (old_aux)
*old_aux = ulist->nodes[i].aux;
return 0; return 0;
} }
}
if (ulist->nnodes >= ulist->nodes_alloced) { if (ulist->nnodes >= ulist->nodes_alloced) {
u64 new_alloced = ulist->nodes_alloced + 128; u64 new_alloced = ulist->nodes_alloced + 128;
@ -188,33 +197,26 @@ EXPORT_SYMBOL(ulist_add);
/** /**
* ulist_next - iterate ulist * ulist_next - iterate ulist
* @ulist: ulist to iterate * @ulist: ulist to iterate
* @prev: previously returned element or %NULL to start iteration * @uiter: iterator variable, initialized with ULIST_ITER_INIT(&iterator)
* *
* Note: locking must be provided by the caller. In case of rwlocks only read * Note: locking must be provided by the caller. In case of rwlocks only read
* locking is needed * locking is needed
* *
* This function is used to iterate an ulist. The iteration is started with * This function is used to iterate an ulist.
* @prev = %NULL. It returns the next element from the ulist or %NULL when the * It returns the next element from the ulist or %NULL when the
* end is reached. No guarantee is made with respect to the order in which * end is reached. No guarantee is made with respect to the order in which
* the elements are returned. They might neither be returned in order of * the elements are returned. They might neither be returned in order of
* addition nor in ascending order. * addition nor in ascending order.
* It is allowed to call ulist_add during an enumeration. Newly added items * It is allowed to call ulist_add during an enumeration. Newly added items
* are guaranteed to show up in the running enumeration. * are guaranteed to show up in the running enumeration.
*/ */
struct ulist_node *ulist_next(struct ulist *ulist, struct ulist_node *prev) struct ulist_node *ulist_next(struct ulist *ulist, struct ulist_iterator *uiter)
{ {
int next;
if (ulist->nnodes == 0) if (ulist->nnodes == 0)
return NULL; return NULL;
if (uiter->i < 0 || uiter->i >= ulist->nnodes)
if (!prev)
return &ulist->nodes[0];
next = (prev - ulist->nodes) + 1;
if (next < 0 || next >= ulist->nnodes)
return NULL; return NULL;
return &ulist->nodes[next]; return &ulist->nodes[uiter->i++];
} }
EXPORT_SYMBOL(ulist_next); EXPORT_SYMBOL(ulist_next);

14
fs/btrfs/ulist.h
View File

@ -24,6 +24,10 @@
*/ */
#define ULIST_SIZE 16 #define ULIST_SIZE 16
struct ulist_iterator {
int i;
};
/* /*
* element of the list * element of the list
*/ */
@ -61,7 +65,13 @@ void ulist_fini(struct ulist *ulist);
void ulist_reinit(struct ulist *ulist); void ulist_reinit(struct ulist *ulist);
struct ulist *ulist_alloc(gfp_t gfp_mask); struct ulist *ulist_alloc(gfp_t gfp_mask);
void ulist_free(struct ulist *ulist); void ulist_free(struct ulist *ulist);
int ulist_add(struct ulist *ulist, u64 val, unsigned long aux, gfp_t gfp_mask); int ulist_add(struct ulist *ulist, u64 val, unsigned long aux,
struct ulist_node *ulist_next(struct ulist *ulist, struct ulist_node *prev); gfp_t gfp_mask);
int ulist_add_merge(struct ulist *ulist, u64 val, unsigned long aux,
unsigned long *old_aux, gfp_t gfp_mask);
struct ulist_node *ulist_next(struct ulist *ulist,
struct ulist_iterator *uiter);
#define ULIST_ITER_INIT(uiter) ((uiter)->i = 0)
#endif #endif