blob: 3774c191e36dcfd04ffa0cdfcf81470d0b93cdfd [file] [log] [blame]
// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (C) 2007 Oracle. All rights reserved.
*/
#include <linux/sched.h>
#include <linux/sched/signal.h>
#include <linux/pagemap.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/sort.h>
#include <linux/rcupdate.h>
#include <linux/kthread.h>
#include <linux/slab.h>
#include <linux/ratelimit.h>
#include <linux/percpu_counter.h>
#include <linux/lockdep.h>
#include <linux/crc32c.h>
#include "ctree.h"
#include "extent-tree.h"
#include "transaction.h"
#include "disk-io.h"
#include "print-tree.h"
#include "volumes.h"
#include "raid56.h"
#include "locking.h"
#include "free-space-cache.h"
#include "free-space-tree.h"
#include "qgroup.h"
#include "ref-verify.h"
#include "space-info.h"
#include "block-rsv.h"
#include "discard.h"
#include "zoned.h"
#include "dev-replace.h"
#include "fs.h"
#include "accessors.h"
#include "root-tree.h"
#include "file-item.h"
#include "orphan.h"
#include "tree-checker.h"
#include "raid-stripe-tree.h"
#undef SCRAMBLE_DELAYED_REFS
static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *href,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extra_op);
static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
struct extent_buffer *leaf,
struct btrfs_extent_item *ei);
static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
u64 parent, u64 root_objectid,
u64 flags, u64 owner, u64 offset,
struct btrfs_key *ins, int ref_mod, u64 oref_root);
static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extent_op);
static int find_next_key(struct btrfs_path *path, int level,
struct btrfs_key *key);
static int block_group_bits(struct btrfs_block_group *cache, u64 bits)
{
return (cache->flags & bits) == bits;
}
/* simple helper to search for an existing data extent at a given offset */
int btrfs_lookup_data_extent(struct btrfs_fs_info *fs_info, u64 start, u64 len)
{
struct btrfs_root *root = btrfs_extent_root(fs_info, start);
int ret;
struct btrfs_key key;
struct btrfs_path *path;
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
key.objectid = start;
key.offset = len;
key.type = BTRFS_EXTENT_ITEM_KEY;
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
btrfs_free_path(path);
return ret;
}
/*
* helper function to lookup reference count and flags of a tree block.
*
* the head node for delayed ref is used to store the sum of all the
* reference count modifications queued up in the rbtree. the head
* node may also store the extent flags to set. This way you can check
* to see what the reference count and extent flags would be if all of
* the delayed refs are not processed.
*/
int btrfs_lookup_extent_info(struct btrfs_trans_handle *trans,
struct btrfs_fs_info *fs_info, u64 bytenr,
u64 offset, int metadata, u64 *refs, u64 *flags,
u64 *owning_root)
{
struct btrfs_root *extent_root;
struct btrfs_delayed_ref_head *head;
struct btrfs_delayed_ref_root *delayed_refs;
struct btrfs_path *path;
struct btrfs_extent_item *ei;
struct extent_buffer *leaf;
struct btrfs_key key;
u32 item_size;
u64 num_refs;
u64 extent_flags;
u64 owner = 0;
int ret;
/*
* If we don't have skinny metadata, don't bother doing anything
* different
*/
if (metadata && !btrfs_fs_incompat(fs_info, SKINNY_METADATA)) {
offset = fs_info->nodesize;
metadata = 0;
}
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
if (!trans) {
path->skip_locking = 1;
path->search_commit_root = 1;
}
search_again:
key.objectid = bytenr;
key.offset = offset;
if (metadata)
key.type = BTRFS_METADATA_ITEM_KEY;
else
key.type = BTRFS_EXTENT_ITEM_KEY;
extent_root = btrfs_extent_root(fs_info, bytenr);
ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
if (ret < 0)
goto out_free;
if (ret > 0 && metadata && key.type == BTRFS_METADATA_ITEM_KEY) {
if (path->slots[0]) {
path->slots[0]--;
btrfs_item_key_to_cpu(path->nodes[0], &key,
path->slots[0]);
if (key.objectid == bytenr &&
key.type == BTRFS_EXTENT_ITEM_KEY &&
key.offset == fs_info->nodesize)
ret = 0;
}
}
if (ret == 0) {
leaf = path->nodes[0];
item_size = btrfs_item_size(leaf, path->slots[0]);
if (item_size >= sizeof(*ei)) {
ei = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_extent_item);
num_refs = btrfs_extent_refs(leaf, ei);
extent_flags = btrfs_extent_flags(leaf, ei);
owner = btrfs_get_extent_owner_root(fs_info, leaf,
path->slots[0]);
} else {
ret = -EUCLEAN;
btrfs_err(fs_info,
"unexpected extent item size, has %u expect >= %zu",
item_size, sizeof(*ei));
if (trans)
btrfs_abort_transaction(trans, ret);
else
btrfs_handle_fs_error(fs_info, ret, NULL);
goto out_free;
}
BUG_ON(num_refs == 0);
} else {
num_refs = 0;
extent_flags = 0;
ret = 0;
}
if (!trans)
goto out;
delayed_refs = &trans->transaction->delayed_refs;
spin_lock(&delayed_refs->lock);
head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
if (head) {
if (!mutex_trylock(&head->mutex)) {
refcount_inc(&head->refs);
spin_unlock(&delayed_refs->lock);
btrfs_release_path(path);
/*
* Mutex was contended, block until it's released and try
* again
*/
mutex_lock(&head->mutex);
mutex_unlock(&head->mutex);
btrfs_put_delayed_ref_head(head);
goto search_again;
}
spin_lock(&head->lock);
if (head->extent_op && head->extent_op->update_flags)
extent_flags |= head->extent_op->flags_to_set;
else
BUG_ON(num_refs == 0);
num_refs += head->ref_mod;
spin_unlock(&head->lock);
mutex_unlock(&head->mutex);
}
spin_unlock(&delayed_refs->lock);
out:
WARN_ON(num_refs == 0);
if (refs)
*refs = num_refs;
if (flags)
*flags = extent_flags;
if (owning_root)
*owning_root = owner;
out_free:
btrfs_free_path(path);
return ret;
}
/*
* Back reference rules. Back refs have three main goals:
*
* 1) differentiate between all holders of references to an extent so that
* when a reference is dropped we can make sure it was a valid reference
* before freeing the extent.
*
* 2) Provide enough information to quickly find the holders of an extent
* if we notice a given block is corrupted or bad.
*
* 3) Make it easy to migrate blocks for FS shrinking or storage pool
* maintenance. This is actually the same as #2, but with a slightly
* different use case.
*
* There are two kinds of back refs. The implicit back refs is optimized
* for pointers in non-shared tree blocks. For a given pointer in a block,
* back refs of this kind provide information about the block's owner tree
* and the pointer's key. These information allow us to find the block by
* b-tree searching. The full back refs is for pointers in tree blocks not
* referenced by their owner trees. The location of tree block is recorded
* in the back refs. Actually the full back refs is generic, and can be
* used in all cases the implicit back refs is used. The major shortcoming
* of the full back refs is its overhead. Every time a tree block gets
* COWed, we have to update back refs entry for all pointers in it.
*
* For a newly allocated tree block, we use implicit back refs for
* pointers in it. This means most tree related operations only involve
* implicit back refs. For a tree block created in old transaction, the
* only way to drop a reference to it is COW it. So we can detect the
* event that tree block loses its owner tree's reference and do the
* back refs conversion.
*
* When a tree block is COWed through a tree, there are four cases:
*
* The reference count of the block is one and the tree is the block's
* owner tree. Nothing to do in this case.
*
* The reference count of the block is one and the tree is not the
* block's owner tree. In this case, full back refs is used for pointers
* in the block. Remove these full back refs, add implicit back refs for
* every pointers in the new block.
*
* The reference count of the block is greater than one and the tree is
* the block's owner tree. In this case, implicit back refs is used for
* pointers in the block. Add full back refs for every pointers in the
* block, increase lower level extents' reference counts. The original
* implicit back refs are entailed to the new block.
*
* The reference count of the block is greater than one and the tree is
* not the block's owner tree. Add implicit back refs for every pointer in
* the new block, increase lower level extents' reference count.
*
* Back Reference Key composing:
*
* The key objectid corresponds to the first byte in the extent,
* The key type is used to differentiate between types of back refs.
* There are different meanings of the key offset for different types
* of back refs.
*
* File extents can be referenced by:
*
* - multiple snapshots, subvolumes, or different generations in one subvol
* - different files inside a single subvolume
* - different offsets inside a file (bookend extents in file.c)
*
* The extent ref structure for the implicit back refs has fields for:
*
* - Objectid of the subvolume root
* - objectid of the file holding the reference
* - original offset in the file
* - how many bookend extents
*
* The key offset for the implicit back refs is hash of the first
* three fields.
*
* The extent ref structure for the full back refs has field for:
*
* - number of pointers in the tree leaf
*
* The key offset for the implicit back refs is the first byte of
* the tree leaf
*
* When a file extent is allocated, The implicit back refs is used.
* the fields are filled in:
*
* (root_key.objectid, inode objectid, offset in file, 1)
*
* When a file extent is removed file truncation, we find the
* corresponding implicit back refs and check the following fields:
*
* (btrfs_header_owner(leaf), inode objectid, offset in file)
*
* Btree extents can be referenced by:
*
* - Different subvolumes
*
* Both the implicit back refs and the full back refs for tree blocks
* only consist of key. The key offset for the implicit back refs is
* objectid of block's owner tree. The key offset for the full back refs
* is the first byte of parent block.
*
* When implicit back refs is used, information about the lowest key and
* level of the tree block are required. These information are stored in
* tree block info structure.
*/
/*
* is_data == BTRFS_REF_TYPE_BLOCK, tree block type is required,
* is_data == BTRFS_REF_TYPE_DATA, data type is requiried,
* is_data == BTRFS_REF_TYPE_ANY, either type is OK.
*/
int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
struct btrfs_extent_inline_ref *iref,
enum btrfs_inline_ref_type is_data)
{
struct btrfs_fs_info *fs_info = eb->fs_info;
int type = btrfs_extent_inline_ref_type(eb, iref);
u64 offset = btrfs_extent_inline_ref_offset(eb, iref);
if (type == BTRFS_EXTENT_OWNER_REF_KEY) {
ASSERT(btrfs_fs_incompat(fs_info, SIMPLE_QUOTA));
return type;
}
if (type == BTRFS_TREE_BLOCK_REF_KEY ||
type == BTRFS_SHARED_BLOCK_REF_KEY ||
type == BTRFS_SHARED_DATA_REF_KEY ||
type == BTRFS_EXTENT_DATA_REF_KEY) {
if (is_data == BTRFS_REF_TYPE_BLOCK) {
if (type == BTRFS_TREE_BLOCK_REF_KEY)
return type;
if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
ASSERT(fs_info);
/*
* Every shared one has parent tree block,
* which must be aligned to sector size.
*/
if (offset && IS_ALIGNED(offset, fs_info->sectorsize))
return type;
}
} else if (is_data == BTRFS_REF_TYPE_DATA) {
if (type == BTRFS_EXTENT_DATA_REF_KEY)
return type;
if (type == BTRFS_SHARED_DATA_REF_KEY) {
ASSERT(fs_info);
/*
* Every shared one has parent tree block,
* which must be aligned to sector size.
*/
if (offset &&
IS_ALIGNED(offset, fs_info->sectorsize))
return type;
}
} else {
ASSERT(is_data == BTRFS_REF_TYPE_ANY);
return type;
}
}
WARN_ON(1);
btrfs_print_leaf(eb);
btrfs_err(fs_info,
"eb %llu iref 0x%lx invalid extent inline ref type %d",
eb->start, (unsigned long)iref, type);
return BTRFS_REF_TYPE_INVALID;
}
u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset)
{
u32 high_crc = ~(u32)0;
u32 low_crc = ~(u32)0;
__le64 lenum;
lenum = cpu_to_le64(root_objectid);
high_crc = crc32c(high_crc, &lenum, sizeof(lenum));
lenum = cpu_to_le64(owner);
low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
lenum = cpu_to_le64(offset);
low_crc = crc32c(low_crc, &lenum, sizeof(lenum));
return ((u64)high_crc << 31) ^ (u64)low_crc;
}
static u64 hash_extent_data_ref_item(struct extent_buffer *leaf,
struct btrfs_extent_data_ref *ref)
{
return hash_extent_data_ref(btrfs_extent_data_ref_root(leaf, ref),
btrfs_extent_data_ref_objectid(leaf, ref),
btrfs_extent_data_ref_offset(leaf, ref));
}
static int match_extent_data_ref(struct extent_buffer *leaf,
struct btrfs_extent_data_ref *ref,
u64 root_objectid, u64 owner, u64 offset)
{
if (btrfs_extent_data_ref_root(leaf, ref) != root_objectid ||
btrfs_extent_data_ref_objectid(leaf, ref) != owner ||
btrfs_extent_data_ref_offset(leaf, ref) != offset)
return 0;
return 1;
}
static noinline int lookup_extent_data_ref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
u64 bytenr, u64 parent,
u64 root_objectid,
u64 owner, u64 offset)
{
struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
struct btrfs_key key;
struct btrfs_extent_data_ref *ref;
struct extent_buffer *leaf;
u32 nritems;
int recow;
int ret;
key.objectid = bytenr;
if (parent) {
key.type = BTRFS_SHARED_DATA_REF_KEY;
key.offset = parent;
} else {
key.type = BTRFS_EXTENT_DATA_REF_KEY;
key.offset = hash_extent_data_ref(root_objectid,
owner, offset);
}
again:
recow = 0;
ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
if (ret < 0)
return ret;
if (parent) {
if (ret)
return -ENOENT;
return 0;
}
ret = -ENOENT;
leaf = path->nodes[0];
nritems = btrfs_header_nritems(leaf);
while (1) {
if (path->slots[0] >= nritems) {
ret = btrfs_next_leaf(root, path);
if (ret) {
if (ret > 0)
return -ENOENT;
return ret;
}
leaf = path->nodes[0];
nritems = btrfs_header_nritems(leaf);
recow = 1;
}
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (key.objectid != bytenr ||
key.type != BTRFS_EXTENT_DATA_REF_KEY)
goto fail;
ref = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_extent_data_ref);
if (match_extent_data_ref(leaf, ref, root_objectid,
owner, offset)) {
if (recow) {
btrfs_release_path(path);
goto again;
}
ret = 0;
break;
}
path->slots[0]++;
}
fail:
return ret;
}
static noinline int insert_extent_data_ref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
struct btrfs_delayed_ref_node *node,
u64 bytenr)
{
struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
struct btrfs_key key;
struct extent_buffer *leaf;
u64 owner = btrfs_delayed_ref_owner(node);
u64 offset = btrfs_delayed_ref_offset(node);
u32 size;
u32 num_refs;
int ret;
key.objectid = bytenr;
if (node->parent) {
key.type = BTRFS_SHARED_DATA_REF_KEY;
key.offset = node->parent;
size = sizeof(struct btrfs_shared_data_ref);
} else {
key.type = BTRFS_EXTENT_DATA_REF_KEY;
key.offset = hash_extent_data_ref(node->ref_root, owner, offset);
size = sizeof(struct btrfs_extent_data_ref);
}
ret = btrfs_insert_empty_item(trans, root, path, &key, size);
if (ret && ret != -EEXIST)
goto fail;
leaf = path->nodes[0];
if (node->parent) {
struct btrfs_shared_data_ref *ref;
ref = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_shared_data_ref);
if (ret == 0) {
btrfs_set_shared_data_ref_count(leaf, ref, node->ref_mod);
} else {
num_refs = btrfs_shared_data_ref_count(leaf, ref);
num_refs += node->ref_mod;
btrfs_set_shared_data_ref_count(leaf, ref, num_refs);
}
} else {
struct btrfs_extent_data_ref *ref;
while (ret == -EEXIST) {
ref = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_extent_data_ref);
if (match_extent_data_ref(leaf, ref, node->ref_root,
owner, offset))
break;
btrfs_release_path(path);
key.offset++;
ret = btrfs_insert_empty_item(trans, root, path, &key,
size);
if (ret && ret != -EEXIST)
goto fail;
leaf = path->nodes[0];
}
ref = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_extent_data_ref);
if (ret == 0) {
btrfs_set_extent_data_ref_root(leaf, ref, node->ref_root);
btrfs_set_extent_data_ref_objectid(leaf, ref, owner);
btrfs_set_extent_data_ref_offset(leaf, ref, offset);
btrfs_set_extent_data_ref_count(leaf, ref, node->ref_mod);
} else {
num_refs = btrfs_extent_data_ref_count(leaf, ref);
num_refs += node->ref_mod;
btrfs_set_extent_data_ref_count(leaf, ref, num_refs);
}
}
btrfs_mark_buffer_dirty(trans, leaf);
ret = 0;
fail:
btrfs_release_path(path);
return ret;
}
static noinline int remove_extent_data_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
int refs_to_drop)
{
struct btrfs_key key;
struct btrfs_extent_data_ref *ref1 = NULL;
struct btrfs_shared_data_ref *ref2 = NULL;
struct extent_buffer *leaf;
u32 num_refs = 0;
int ret = 0;
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
ref1 = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_extent_data_ref);
num_refs = btrfs_extent_data_ref_count(leaf, ref1);
} else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
ref2 = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_shared_data_ref);
num_refs = btrfs_shared_data_ref_count(leaf, ref2);
} else {
btrfs_err(trans->fs_info,
"unrecognized backref key (%llu %u %llu)",
key.objectid, key.type, key.offset);
btrfs_abort_transaction(trans, -EUCLEAN);
return -EUCLEAN;
}
BUG_ON(num_refs < refs_to_drop);
num_refs -= refs_to_drop;
if (num_refs == 0) {
ret = btrfs_del_item(trans, root, path);
} else {
if (key.type == BTRFS_EXTENT_DATA_REF_KEY)
btrfs_set_extent_data_ref_count(leaf, ref1, num_refs);
else if (key.type == BTRFS_SHARED_DATA_REF_KEY)
btrfs_set_shared_data_ref_count(leaf, ref2, num_refs);
btrfs_mark_buffer_dirty(trans, leaf);
}
return ret;
}
static noinline u32 extent_data_ref_count(struct btrfs_path *path,
struct btrfs_extent_inline_ref *iref)
{
struct btrfs_key key;
struct extent_buffer *leaf;
struct btrfs_extent_data_ref *ref1;
struct btrfs_shared_data_ref *ref2;
u32 num_refs = 0;
int type;
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (iref) {
/*
* If type is invalid, we should have bailed out earlier than
* this call.
*/
type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
ASSERT(type != BTRFS_REF_TYPE_INVALID);
if (type == BTRFS_EXTENT_DATA_REF_KEY) {
ref1 = (struct btrfs_extent_data_ref *)(&iref->offset);
num_refs = btrfs_extent_data_ref_count(leaf, ref1);
} else {
ref2 = (struct btrfs_shared_data_ref *)(iref + 1);
num_refs = btrfs_shared_data_ref_count(leaf, ref2);
}
} else if (key.type == BTRFS_EXTENT_DATA_REF_KEY) {
ref1 = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_extent_data_ref);
num_refs = btrfs_extent_data_ref_count(leaf, ref1);
} else if (key.type == BTRFS_SHARED_DATA_REF_KEY) {
ref2 = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_shared_data_ref);
num_refs = btrfs_shared_data_ref_count(leaf, ref2);
} else {
WARN_ON(1);
}
return num_refs;
}
static noinline int lookup_tree_block_ref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
u64 bytenr, u64 parent,
u64 root_objectid)
{
struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
struct btrfs_key key;
int ret;
key.objectid = bytenr;
if (parent) {
key.type = BTRFS_SHARED_BLOCK_REF_KEY;
key.offset = parent;
} else {
key.type = BTRFS_TREE_BLOCK_REF_KEY;
key.offset = root_objectid;
}
ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
if (ret > 0)
ret = -ENOENT;
return ret;
}
static noinline int insert_tree_block_ref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
struct btrfs_delayed_ref_node *node,
u64 bytenr)
{
struct btrfs_root *root = btrfs_extent_root(trans->fs_info, bytenr);
struct btrfs_key key;
int ret;
key.objectid = bytenr;
if (node->parent) {
key.type = BTRFS_SHARED_BLOCK_REF_KEY;
key.offset = node->parent;
} else {
key.type = BTRFS_TREE_BLOCK_REF_KEY;
key.offset = node->ref_root;
}
ret = btrfs_insert_empty_item(trans, root, path, &key, 0);
btrfs_release_path(path);
return ret;
}
static inline int extent_ref_type(u64 parent, u64 owner)
{
int type;
if (owner < BTRFS_FIRST_FREE_OBJECTID) {
if (parent > 0)
type = BTRFS_SHARED_BLOCK_REF_KEY;
else
type = BTRFS_TREE_BLOCK_REF_KEY;
} else {
if (parent > 0)
type = BTRFS_SHARED_DATA_REF_KEY;
else
type = BTRFS_EXTENT_DATA_REF_KEY;
}
return type;
}
static int find_next_key(struct btrfs_path *path, int level,
struct btrfs_key *key)
{
for (; level < BTRFS_MAX_LEVEL; level++) {
if (!path->nodes[level])
break;
if (path->slots[level] + 1 >=
btrfs_header_nritems(path->nodes[level]))
continue;
if (level == 0)
btrfs_item_key_to_cpu(path->nodes[level], key,
path->slots[level] + 1);
else
btrfs_node_key_to_cpu(path->nodes[level], key,
path->slots[level] + 1);
return 0;
}
return 1;
}
/*
* look for inline back ref. if back ref is found, *ref_ret is set
* to the address of inline back ref, and 0 is returned.
*
* if back ref isn't found, *ref_ret is set to the address where it
* should be inserted, and -ENOENT is returned.
*
* if insert is true and there are too many inline back refs, the path
* points to the extent item, and -EAGAIN is returned.
*
* NOTE: inline back refs are ordered in the same way that back ref
* items in the tree are ordered.
*/
static noinline_for_stack
int lookup_inline_extent_backref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
struct btrfs_extent_inline_ref **ref_ret,
u64 bytenr, u64 num_bytes,
u64 parent, u64 root_objectid,
u64 owner, u64 offset, int insert)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_root *root = btrfs_extent_root(fs_info, bytenr);
struct btrfs_key key;
struct extent_buffer *leaf;
struct btrfs_extent_item *ei;
struct btrfs_extent_inline_ref *iref;
u64 flags;
u64 item_size;
unsigned long ptr;
unsigned long end;
int extra_size;
int type;
int want;
int ret;
bool skinny_metadata = btrfs_fs_incompat(fs_info, SKINNY_METADATA);
int needed;
key.objectid = bytenr;
key.type = BTRFS_EXTENT_ITEM_KEY;
key.offset = num_bytes;
want = extent_ref_type(parent, owner);
if (insert) {
extra_size = btrfs_extent_inline_ref_size(want);
path->search_for_extension = 1;
path->keep_locks = 1;
} else
extra_size = -1;
/*
* Owner is our level, so we can just add one to get the level for the
* block we are interested in.
*/
if (skinny_metadata && owner < BTRFS_FIRST_FREE_OBJECTID) {
key.type = BTRFS_METADATA_ITEM_KEY;
key.offset = owner;
}
again:
ret = btrfs_search_slot(trans, root, &key, path, extra_size, 1);
if (ret < 0)
goto out;
/*
* We may be a newly converted file system which still has the old fat
* extent entries for metadata, so try and see if we have one of those.
*/
if (ret > 0 && skinny_metadata) {
skinny_metadata = false;
if (path->slots[0]) {
path->slots[0]--;
btrfs_item_key_to_cpu(path->nodes[0], &key,
path->slots[0]);
if (key.objectid == bytenr &&
key.type == BTRFS_EXTENT_ITEM_KEY &&
key.offset == num_bytes)
ret = 0;
}
if (ret) {
key.objectid = bytenr;
key.type = BTRFS_EXTENT_ITEM_KEY;
key.offset = num_bytes;
btrfs_release_path(path);
goto again;
}
}
if (ret && !insert) {
ret = -ENOENT;
goto out;
} else if (WARN_ON(ret)) {
btrfs_print_leaf(path->nodes[0]);
btrfs_err(fs_info,
"extent item not found for insert, bytenr %llu num_bytes %llu parent %llu root_objectid %llu owner %llu offset %llu",
bytenr, num_bytes, parent, root_objectid, owner,
offset);
ret = -EUCLEAN;
goto out;
}
leaf = path->nodes[0];
item_size = btrfs_item_size(leaf, path->slots[0]);
if (unlikely(item_size < sizeof(*ei))) {
ret = -EUCLEAN;
btrfs_err(fs_info,
"unexpected extent item size, has %llu expect >= %zu",
item_size, sizeof(*ei));
btrfs_abort_transaction(trans, ret);
goto out;
}
ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
flags = btrfs_extent_flags(leaf, ei);
ptr = (unsigned long)(ei + 1);
end = (unsigned long)ei + item_size;
if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK && !skinny_metadata) {
ptr += sizeof(struct btrfs_tree_block_info);
BUG_ON(ptr > end);
}
if (owner >= BTRFS_FIRST_FREE_OBJECTID)
needed = BTRFS_REF_TYPE_DATA;
else
needed = BTRFS_REF_TYPE_BLOCK;
ret = -ENOENT;
while (ptr < end) {
iref = (struct btrfs_extent_inline_ref *)ptr;
type = btrfs_get_extent_inline_ref_type(leaf, iref, needed);
if (type == BTRFS_EXTENT_OWNER_REF_KEY) {
ASSERT(btrfs_fs_incompat(fs_info, SIMPLE_QUOTA));
ptr += btrfs_extent_inline_ref_size(type);
continue;
}
if (type == BTRFS_REF_TYPE_INVALID) {
ret = -EUCLEAN;
goto out;
}
if (want < type)
break;
if (want > type) {
ptr += btrfs_extent_inline_ref_size(type);
continue;
}
if (type == BTRFS_EXTENT_DATA_REF_KEY) {
struct btrfs_extent_data_ref *dref;
dref = (struct btrfs_extent_data_ref *)(&iref->offset);
if (match_extent_data_ref(leaf, dref, root_objectid,
owner, offset)) {
ret = 0;
break;
}
if (hash_extent_data_ref_item(leaf, dref) <
hash_extent_data_ref(root_objectid, owner, offset))
break;
} else {
u64 ref_offset;
ref_offset = btrfs_extent_inline_ref_offset(leaf, iref);
if (parent > 0) {
if (parent == ref_offset) {
ret = 0;
break;
}
if (ref_offset < parent)
break;
} else {
if (root_objectid == ref_offset) {
ret = 0;
break;
}
if (ref_offset < root_objectid)
break;
}
}
ptr += btrfs_extent_inline_ref_size(type);
}
if (unlikely(ptr > end)) {
ret = -EUCLEAN;
btrfs_print_leaf(path->nodes[0]);
btrfs_crit(fs_info,
"overrun extent record at slot %d while looking for inline extent for root %llu owner %llu offset %llu parent %llu",
path->slots[0], root_objectid, owner, offset, parent);
goto out;
}
if (ret == -ENOENT && insert) {
if (item_size + extra_size >=
BTRFS_MAX_EXTENT_ITEM_SIZE(root)) {
ret = -EAGAIN;
goto out;
}
/*
* To add new inline back ref, we have to make sure
* there is no corresponding back ref item.
* For simplicity, we just do not add new inline back
* ref if there is any kind of item for this block
*/
if (find_next_key(path, 0, &key) == 0 &&
key.objectid == bytenr &&
key.type < BTRFS_BLOCK_GROUP_ITEM_KEY) {
ret = -EAGAIN;
goto out;
}
}
*ref_ret = (struct btrfs_extent_inline_ref *)ptr;
out:
if (insert) {
path->keep_locks = 0;
path->search_for_extension = 0;
btrfs_unlock_up_safe(path, 1);
}
return ret;
}
/*
* helper to add new inline back ref
*/
static noinline_for_stack
void setup_inline_extent_backref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
struct btrfs_extent_inline_ref *iref,
u64 parent, u64 root_objectid,
u64 owner, u64 offset, int refs_to_add,
struct btrfs_delayed_extent_op *extent_op)
{
struct extent_buffer *leaf;
struct btrfs_extent_item *ei;
unsigned long ptr;
unsigned long end;
unsigned long item_offset;
u64 refs;
int size;
int type;
leaf = path->nodes[0];
ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
item_offset = (unsigned long)iref - (unsigned long)ei;
type = extent_ref_type(parent, owner);
size = btrfs_extent_inline_ref_size(type);
btrfs_extend_item(trans, path, size);
ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
refs = btrfs_extent_refs(leaf, ei);
refs += refs_to_add;
btrfs_set_extent_refs(leaf, ei, refs);
if (extent_op)
__run_delayed_extent_op(extent_op, leaf, ei);
ptr = (unsigned long)ei + item_offset;
end = (unsigned long)ei + btrfs_item_size(leaf, path->slots[0]);
if (ptr < end - size)
memmove_extent_buffer(leaf, ptr + size, ptr,
end - size - ptr);
iref = (struct btrfs_extent_inline_ref *)ptr;
btrfs_set_extent_inline_ref_type(leaf, iref, type);
if (type == BTRFS_EXTENT_DATA_REF_KEY) {
struct btrfs_extent_data_ref *dref;
dref = (struct btrfs_extent_data_ref *)(&iref->offset);
btrfs_set_extent_data_ref_root(leaf, dref, root_objectid);
btrfs_set_extent_data_ref_objectid(leaf, dref, owner);
btrfs_set_extent_data_ref_offset(leaf, dref, offset);
btrfs_set_extent_data_ref_count(leaf, dref, refs_to_add);
} else if (type == BTRFS_SHARED_DATA_REF_KEY) {
struct btrfs_shared_data_ref *sref;
sref = (struct btrfs_shared_data_ref *)(iref + 1);
btrfs_set_shared_data_ref_count(leaf, sref, refs_to_add);
btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
} else if (type == BTRFS_SHARED_BLOCK_REF_KEY) {
btrfs_set_extent_inline_ref_offset(leaf, iref, parent);
} else {
btrfs_set_extent_inline_ref_offset(leaf, iref, root_objectid);
}
btrfs_mark_buffer_dirty(trans, leaf);
}
static int lookup_extent_backref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
struct btrfs_extent_inline_ref **ref_ret,
u64 bytenr, u64 num_bytes, u64 parent,
u64 root_objectid, u64 owner, u64 offset)
{
int ret;
ret = lookup_inline_extent_backref(trans, path, ref_ret, bytenr,
num_bytes, parent, root_objectid,
owner, offset, 0);
if (ret != -ENOENT)
return ret;
btrfs_release_path(path);
*ref_ret = NULL;
if (owner < BTRFS_FIRST_FREE_OBJECTID) {
ret = lookup_tree_block_ref(trans, path, bytenr, parent,
root_objectid);
} else {
ret = lookup_extent_data_ref(trans, path, bytenr, parent,
root_objectid, owner, offset);
}
return ret;
}
/*
* helper to update/remove inline back ref
*/
static noinline_for_stack int update_inline_extent_backref(
struct btrfs_trans_handle *trans,
struct btrfs_path *path,
struct btrfs_extent_inline_ref *iref,
int refs_to_mod,
struct btrfs_delayed_extent_op *extent_op)
{
struct extent_buffer *leaf = path->nodes[0];
struct btrfs_fs_info *fs_info = leaf->fs_info;
struct btrfs_extent_item *ei;
struct btrfs_extent_data_ref *dref = NULL;
struct btrfs_shared_data_ref *sref = NULL;
unsigned long ptr;
unsigned long end;
u32 item_size;
int size;
int type;
u64 refs;
ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
refs = btrfs_extent_refs(leaf, ei);
if (unlikely(refs_to_mod < 0 && refs + refs_to_mod <= 0)) {
struct btrfs_key key;
u32 extent_size;
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (key.type == BTRFS_METADATA_ITEM_KEY)
extent_size = fs_info->nodesize;
else
extent_size = key.offset;
btrfs_print_leaf(leaf);
btrfs_err(fs_info,
"invalid refs_to_mod for extent %llu num_bytes %u, has %d expect >= -%llu",
key.objectid, extent_size, refs_to_mod, refs);
return -EUCLEAN;
}
refs += refs_to_mod;
btrfs_set_extent_refs(leaf, ei, refs);
if (extent_op)
__run_delayed_extent_op(extent_op, leaf, ei);
type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_ANY);
/*
* Function btrfs_get_extent_inline_ref_type() has already printed
* error messages.
*/
if (unlikely(type == BTRFS_REF_TYPE_INVALID))
return -EUCLEAN;
if (type == BTRFS_EXTENT_DATA_REF_KEY) {
dref = (struct btrfs_extent_data_ref *)(&iref->offset);
refs = btrfs_extent_data_ref_count(leaf, dref);
} else if (type == BTRFS_SHARED_DATA_REF_KEY) {
sref = (struct btrfs_shared_data_ref *)(iref + 1);
refs = btrfs_shared_data_ref_count(leaf, sref);
} else {
refs = 1;
/*
* For tree blocks we can only drop one ref for it, and tree
* blocks should not have refs > 1.
*
* Furthermore if we're inserting a new inline backref, we
* won't reach this path either. That would be
* setup_inline_extent_backref().
*/
if (unlikely(refs_to_mod != -1)) {
struct btrfs_key key;
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
btrfs_print_leaf(leaf);
btrfs_err(fs_info,
"invalid refs_to_mod for tree block %llu, has %d expect -1",
key.objectid, refs_to_mod);
return -EUCLEAN;
}
}
if (unlikely(refs_to_mod < 0 && refs < -refs_to_mod)) {
struct btrfs_key key;
u32 extent_size;
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (key.type == BTRFS_METADATA_ITEM_KEY)
extent_size = fs_info->nodesize;
else
extent_size = key.offset;
btrfs_print_leaf(leaf);
btrfs_err(fs_info,
"invalid refs_to_mod for backref entry, iref %lu extent %llu num_bytes %u, has %d expect >= -%llu",
(unsigned long)iref, key.objectid, extent_size,
refs_to_mod, refs);
return -EUCLEAN;
}
refs += refs_to_mod;
if (refs > 0) {
if (type == BTRFS_EXTENT_DATA_REF_KEY)
btrfs_set_extent_data_ref_count(leaf, dref, refs);
else
btrfs_set_shared_data_ref_count(leaf, sref, refs);
} else {
size = btrfs_extent_inline_ref_size(type);
item_size = btrfs_item_size(leaf, path->slots[0]);
ptr = (unsigned long)iref;
end = (unsigned long)ei + item_size;
if (ptr + size < end)
memmove_extent_buffer(leaf, ptr, ptr + size,
end - ptr - size);
item_size -= size;
btrfs_truncate_item(trans, path, item_size, 1);
}
btrfs_mark_buffer_dirty(trans, leaf);
return 0;
}
static noinline_for_stack
int insert_inline_extent_backref(struct btrfs_trans_handle *trans,
struct btrfs_path *path,
u64 bytenr, u64 num_bytes, u64 parent,
u64 root_objectid, u64 owner,
u64 offset, int refs_to_add,
struct btrfs_delayed_extent_op *extent_op)
{
struct btrfs_extent_inline_ref *iref;
int ret;
ret = lookup_inline_extent_backref(trans, path, &iref, bytenr,
num_bytes, parent, root_objectid,
owner, offset, 1);
if (ret == 0) {
/*
* We're adding refs to a tree block we already own, this
* should not happen at all.
*/
if (owner < BTRFS_FIRST_FREE_OBJECTID) {
btrfs_print_leaf(path->nodes[0]);
btrfs_crit(trans->fs_info,
"adding refs to an existing tree ref, bytenr %llu num_bytes %llu root_objectid %llu slot %u",
bytenr, num_bytes, root_objectid, path->slots[0]);
return -EUCLEAN;
}
ret = update_inline_extent_backref(trans, path, iref,
refs_to_add, extent_op);
} else if (ret == -ENOENT) {
setup_inline_extent_backref(trans, path, iref, parent,
root_objectid, owner, offset,
refs_to_add, extent_op);
ret = 0;
}
return ret;
}
static int remove_extent_backref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_path *path,
struct btrfs_extent_inline_ref *iref,
int refs_to_drop, int is_data)
{
int ret = 0;
BUG_ON(!is_data && refs_to_drop != 1);
if (iref)
ret = update_inline_extent_backref(trans, path, iref,
-refs_to_drop, NULL);
else if (is_data)
ret = remove_extent_data_ref(trans, root, path, refs_to_drop);
else
ret = btrfs_del_item(trans, root, path);
return ret;
}
static int btrfs_issue_discard(struct block_device *bdev, u64 start, u64 len,
u64 *discarded_bytes)
{
int j, ret = 0;
u64 bytes_left, end;
u64 aligned_start = ALIGN(start, 1 << SECTOR_SHIFT);
/* Adjust the range to be aligned to 512B sectors if necessary. */
if (start != aligned_start) {
len -= aligned_start - start;
len = round_down(len, 1 << SECTOR_SHIFT);
start = aligned_start;
}
*discarded_bytes = 0;
if (!len)
return 0;
end = start + len;
bytes_left = len;
/* Skip any superblocks on this device. */
for (j = 0; j < BTRFS_SUPER_MIRROR_MAX; j++) {
u64 sb_start = btrfs_sb_offset(j);
u64 sb_end = sb_start + BTRFS_SUPER_INFO_SIZE;
u64 size = sb_start - start;
if (!in_range(sb_start, start, bytes_left) &&
!in_range(sb_end, start, bytes_left) &&
!in_range(start, sb_start, BTRFS_SUPER_INFO_SIZE))
continue;
/*
* Superblock spans beginning of range. Adjust start and
* try again.
*/
if (sb_start <= start) {
start += sb_end - start;
if (start > end) {
bytes_left = 0;
break;
}
bytes_left = end - start;
continue;
}
if (size) {
ret = blkdev_issue_discard(bdev, start >> SECTOR_SHIFT,
size >> SECTOR_SHIFT,
GFP_NOFS);
if (!ret)
*discarded_bytes += size;
else if (ret != -EOPNOTSUPP)
return ret;
}
start = sb_end;
if (start > end) {
bytes_left = 0;
break;
}
bytes_left = end - start;
}
if (bytes_left) {
ret = blkdev_issue_discard(bdev, start >> SECTOR_SHIFT,
bytes_left >> SECTOR_SHIFT,
GFP_NOFS);
if (!ret)
*discarded_bytes += bytes_left;
}
return ret;
}
static int do_discard_extent(struct btrfs_discard_stripe *stripe, u64 *bytes)
{
struct btrfs_device *dev = stripe->dev;
struct btrfs_fs_info *fs_info = dev->fs_info;
struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
u64 phys = stripe->physical;
u64 len = stripe->length;
u64 discarded = 0;
int ret = 0;
/* Zone reset on a zoned filesystem */
if (btrfs_can_zone_reset(dev, phys, len)) {
u64 src_disc;
ret = btrfs_reset_device_zone(dev, phys, len, &discarded);
if (ret)
goto out;
if (!btrfs_dev_replace_is_ongoing(dev_replace) ||
dev != dev_replace->srcdev)
goto out;
src_disc = discarded;
/* Send to replace target as well */
ret = btrfs_reset_device_zone(dev_replace->tgtdev, phys, len,
&discarded);
discarded += src_disc;
} else if (bdev_max_discard_sectors(stripe->dev->bdev)) {
ret = btrfs_issue_discard(dev->bdev, phys, len, &discarded);
} else {
ret = 0;
*bytes = 0;
}
out:
*bytes = discarded;
return ret;
}
int btrfs_discard_extent(struct btrfs_fs_info *fs_info, u64 bytenr,
u64 num_bytes, u64 *actual_bytes)
{
int ret = 0;
u64 discarded_bytes = 0;
u64 end = bytenr + num_bytes;
u64 cur = bytenr;
/*
* Avoid races with device replace and make sure the devices in the
* stripes don't go away while we are discarding.
*/
btrfs_bio_counter_inc_blocked(fs_info);
while (cur < end) {
struct btrfs_discard_stripe *stripes;
unsigned int num_stripes;
int i;
num_bytes = end - cur;
stripes = btrfs_map_discard(fs_info, cur, &num_bytes, &num_stripes);
if (IS_ERR(stripes)) {
ret = PTR_ERR(stripes);
if (ret == -EOPNOTSUPP)
ret = 0;
break;
}
for (i = 0; i < num_stripes; i++) {
struct btrfs_discard_stripe *stripe = stripes + i;
u64 bytes;
if (!stripe->dev->bdev) {
ASSERT(btrfs_test_opt(fs_info, DEGRADED));
continue;
}
if (!test_bit(BTRFS_DEV_STATE_WRITEABLE,
&stripe->dev->dev_state))
continue;
ret = do_discard_extent(stripe, &bytes);
if (ret) {
/*
* Keep going if discard is not supported by the
* device.
*/
if (ret != -EOPNOTSUPP)
break;
ret = 0;
} else {
discarded_bytes += bytes;
}
}
kfree(stripes);
if (ret)
break;
cur += num_bytes;
}
btrfs_bio_counter_dec(fs_info);
if (actual_bytes)
*actual_bytes = discarded_bytes;
return ret;
}
/* Can return -ENOMEM */
int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
struct btrfs_ref *generic_ref)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
int ret;
ASSERT(generic_ref->type != BTRFS_REF_NOT_SET &&
generic_ref->action);
BUG_ON(generic_ref->type == BTRFS_REF_METADATA &&
generic_ref->ref_root == BTRFS_TREE_LOG_OBJECTID);
if (generic_ref->type == BTRFS_REF_METADATA)
ret = btrfs_add_delayed_tree_ref(trans, generic_ref, NULL);
else
ret = btrfs_add_delayed_data_ref(trans, generic_ref, 0);
btrfs_ref_tree_mod(fs_info, generic_ref);
return ret;
}
/*
* Insert backreference for a given extent.
*
* The counterpart is in __btrfs_free_extent(), with examples and more details
* how it works.
*
* @trans: Handle of transaction
*
* @node: The delayed ref node used to get the bytenr/length for
* extent whose references are incremented.
*
* @extent_op Pointer to a structure, holding information necessary when
* updating a tree block's flags
*
*/
static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extent_op)
{
struct btrfs_path *path;
struct extent_buffer *leaf;
struct btrfs_extent_item *item;
struct btrfs_key key;
u64 bytenr = node->bytenr;
u64 num_bytes = node->num_bytes;
u64 owner = btrfs_delayed_ref_owner(node);
u64 offset = btrfs_delayed_ref_offset(node);
u64 refs;
int refs_to_add = node->ref_mod;
int ret;
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
/* this will setup the path even if it fails to insert the back ref */
ret = insert_inline_extent_backref(trans, path, bytenr, num_bytes,
node->parent, node->ref_root, owner,
offset, refs_to_add, extent_op);
if ((ret < 0 && ret != -EAGAIN) || !ret)
goto out;
/*
* Ok we had -EAGAIN which means we didn't have space to insert and
* inline extent ref, so just update the reference count and add a
* normal backref.
*/
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
refs = btrfs_extent_refs(leaf, item);
btrfs_set_extent_refs(leaf, item, refs + refs_to_add);
if (extent_op)
__run_delayed_extent_op(extent_op, leaf, item);
btrfs_mark_buffer_dirty(trans, leaf);
btrfs_release_path(path);
/* now insert the actual backref */
if (owner < BTRFS_FIRST_FREE_OBJECTID)
ret = insert_tree_block_ref(trans, path, node, bytenr);
else
ret = insert_extent_data_ref(trans, path, node, bytenr);
if (ret)
btrfs_abort_transaction(trans, ret);
out:
btrfs_free_path(path);
return ret;
}
static void free_head_ref_squota_rsv(struct btrfs_fs_info *fs_info,
struct btrfs_delayed_ref_head *href)
{
u64 root = href->owning_root;
/*
* Don't check must_insert_reserved, as this is called from contexts
* where it has already been unset.
*/
if (btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_SIMPLE ||
!href->is_data || !is_fstree(root))
return;
btrfs_qgroup_free_refroot(fs_info, root, href->reserved_bytes,
BTRFS_QGROUP_RSV_DATA);
}
static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *href,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extent_op,
bool insert_reserved)
{
int ret = 0;
u64 parent = 0;
u64 flags = 0;
trace_run_delayed_data_ref(trans->fs_info, node);
if (node->type == BTRFS_SHARED_DATA_REF_KEY)
parent = node->parent;
if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
struct btrfs_key key;
struct btrfs_squota_delta delta = {
.root = href->owning_root,
.num_bytes = node->num_bytes,
.is_data = true,
.is_inc = true,
.generation = trans->transid,
};
u64 owner = btrfs_delayed_ref_owner(node);
u64 offset = btrfs_delayed_ref_offset(node);
if (extent_op)
flags |= extent_op->flags_to_set;
key.objectid = node->bytenr;
key.type = BTRFS_EXTENT_ITEM_KEY;
key.offset = node->num_bytes;
ret = alloc_reserved_file_extent(trans, parent, node->ref_root,
flags, owner, offset, &key,
node->ref_mod,
href->owning_root);
free_head_ref_squota_rsv(trans->fs_info, href);
if (!ret)
ret = btrfs_record_squota_delta(trans->fs_info, &delta);
} else if (node->action == BTRFS_ADD_DELAYED_REF) {
ret = __btrfs_inc_extent_ref(trans, node, extent_op);
} else if (node->action == BTRFS_DROP_DELAYED_REF) {
ret = __btrfs_free_extent(trans, href, node, extent_op);
} else {
BUG();
}
return ret;
}
static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
struct extent_buffer *leaf,
struct btrfs_extent_item *ei)
{
u64 flags = btrfs_extent_flags(leaf, ei);
if (extent_op->update_flags) {
flags |= extent_op->flags_to_set;
btrfs_set_extent_flags(leaf, ei, flags);
}
if (extent_op->update_key) {
struct btrfs_tree_block_info *bi;
BUG_ON(!(flags & BTRFS_EXTENT_FLAG_TREE_BLOCK));
bi = (struct btrfs_tree_block_info *)(ei + 1);
btrfs_set_tree_block_key(leaf, bi, &extent_op->key);
}
}
static int run_delayed_extent_op(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *head,
struct btrfs_delayed_extent_op *extent_op)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_root *root;
struct btrfs_key key;
struct btrfs_path *path;
struct btrfs_extent_item *ei;
struct extent_buffer *leaf;
u32 item_size;
int ret;
int metadata = 1;
if (TRANS_ABORTED(trans))
return 0;
if (!btrfs_fs_incompat(fs_info, SKINNY_METADATA))
metadata = 0;
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
key.objectid = head->bytenr;
if (metadata) {
key.type = BTRFS_METADATA_ITEM_KEY;
key.offset = extent_op->level;
} else {
key.type = BTRFS_EXTENT_ITEM_KEY;
key.offset = head->num_bytes;
}
root = btrfs_extent_root(fs_info, key.objectid);
again:
ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
if (ret < 0) {
goto out;
} else if (ret > 0) {
if (metadata) {
if (path->slots[0] > 0) {
path->slots[0]--;
btrfs_item_key_to_cpu(path->nodes[0], &key,
path->slots[0]);
if (key.objectid == head->bytenr &&
key.type == BTRFS_EXTENT_ITEM_KEY &&
key.offset == head->num_bytes)
ret = 0;
}
if (ret > 0) {
btrfs_release_path(path);
metadata = 0;
key.objectid = head->bytenr;
key.offset = head->num_bytes;
key.type = BTRFS_EXTENT_ITEM_KEY;
goto again;
}
} else {
ret = -EUCLEAN;
btrfs_err(fs_info,
"missing extent item for extent %llu num_bytes %llu level %d",
head->bytenr, head->num_bytes, extent_op->level);
goto out;
}
}
leaf = path->nodes[0];
item_size = btrfs_item_size(leaf, path->slots[0]);
if (unlikely(item_size < sizeof(*ei))) {
ret = -EUCLEAN;
btrfs_err(fs_info,
"unexpected extent item size, has %u expect >= %zu",
item_size, sizeof(*ei));
btrfs_abort_transaction(trans, ret);
goto out;
}
ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
__run_delayed_extent_op(extent_op, leaf, ei);
btrfs_mark_buffer_dirty(trans, leaf);
out:
btrfs_free_path(path);
return ret;
}
static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *href,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extent_op,
bool insert_reserved)
{
int ret = 0;
struct btrfs_fs_info *fs_info = trans->fs_info;
u64 parent = 0;
u64 ref_root = 0;
trace_run_delayed_tree_ref(trans->fs_info, node);
if (node->type == BTRFS_SHARED_BLOCK_REF_KEY)
parent = node->parent;
ref_root = node->ref_root;
if (unlikely(node->ref_mod != 1)) {
btrfs_err(trans->fs_info,
"btree block %llu has %d references rather than 1: action %d ref_root %llu parent %llu",
node->bytenr, node->ref_mod, node->action, ref_root,
parent);
return -EUCLEAN;
}
if (node->action == BTRFS_ADD_DELAYED_REF && insert_reserved) {
struct btrfs_squota_delta delta = {
.root = href->owning_root,
.num_bytes = fs_info->nodesize,
.is_data = false,
.is_inc = true,
.generation = trans->transid,
};
BUG_ON(!extent_op || !extent_op->update_flags);
ret = alloc_reserved_tree_block(trans, node, extent_op);
if (!ret)
btrfs_record_squota_delta(fs_info, &delta);
} else if (node->action == BTRFS_ADD_DELAYED_REF) {
ret = __btrfs_inc_extent_ref(trans, node, extent_op);
} else if (node->action == BTRFS_DROP_DELAYED_REF) {
ret = __btrfs_free_extent(trans, href, node, extent_op);
} else {
BUG();
}
return ret;
}
/* helper function to actually process a single delayed ref entry */
static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *href,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extent_op,
bool insert_reserved)
{
int ret = 0;
if (TRANS_ABORTED(trans)) {
if (insert_reserved) {
btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
free_head_ref_squota_rsv(trans->fs_info, href);
}
return 0;
}
if (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
node->type == BTRFS_SHARED_BLOCK_REF_KEY)
ret = run_delayed_tree_ref(trans, href, node, extent_op,
insert_reserved);
else if (node->type == BTRFS_EXTENT_DATA_REF_KEY ||
node->type == BTRFS_SHARED_DATA_REF_KEY)
ret = run_delayed_data_ref(trans, href, node, extent_op,
insert_reserved);
else if (node->type == BTRFS_EXTENT_OWNER_REF_KEY)
ret = 0;
else
BUG();
if (ret && insert_reserved)
btrfs_pin_extent(trans, node->bytenr, node->num_bytes, 1);
if (ret < 0)
btrfs_err(trans->fs_info,
"failed to run delayed ref for logical %llu num_bytes %llu type %u action %u ref_mod %d: %d",
node->bytenr, node->num_bytes, node->type,
node->action, node->ref_mod, ret);
return ret;
}
static inline struct btrfs_delayed_ref_node *
select_delayed_ref(struct btrfs_delayed_ref_head *head)
{
struct btrfs_delayed_ref_node *ref;
if (RB_EMPTY_ROOT(&head->ref_tree.rb_root))
return NULL;
/*
* Select a delayed ref of type BTRFS_ADD_DELAYED_REF first.
* This is to prevent a ref count from going down to zero, which deletes
* the extent item from the extent tree, when there still are references
* to add, which would fail because they would not find the extent item.
*/
if (!list_empty(&head->ref_add_list))
return list_first_entry(&head->ref_add_list,
struct btrfs_delayed_ref_node, add_list);
ref = rb_entry(rb_first_cached(&head->ref_tree),
struct btrfs_delayed_ref_node, ref_node);
ASSERT(list_empty(&ref->add_list));
return ref;
}
static void unselect_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
struct btrfs_delayed_ref_head *head)
{
spin_lock(&delayed_refs->lock);
head->processing = false;
delayed_refs->num_heads_ready++;
spin_unlock(&delayed_refs->lock);
btrfs_delayed_ref_unlock(head);
}
static struct btrfs_delayed_extent_op *cleanup_extent_op(
struct btrfs_delayed_ref_head *head)
{
struct btrfs_delayed_extent_op *extent_op = head->extent_op;
if (!extent_op)
return NULL;
if (head->must_insert_reserved) {
head->extent_op = NULL;
btrfs_free_delayed_extent_op(extent_op);
return NULL;
}
return extent_op;
}
static int run_and_cleanup_extent_op(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *head)
{
struct btrfs_delayed_extent_op *extent_op;
int ret;
extent_op = cleanup_extent_op(head);
if (!extent_op)
return 0;
head->extent_op = NULL;
spin_unlock(&head->lock);
ret = run_delayed_extent_op(trans, head, extent_op);
btrfs_free_delayed_extent_op(extent_op);
return ret ? ret : 1;
}
u64 btrfs_cleanup_ref_head_accounting(struct btrfs_fs_info *fs_info,
struct btrfs_delayed_ref_root *delayed_refs,
struct btrfs_delayed_ref_head *head)
{
u64 ret = 0;
/*
* We had csum deletions accounted for in our delayed refs rsv, we need
* to drop the csum leaves for this update from our delayed_refs_rsv.
*/
if (head->total_ref_mod < 0 && head->is_data) {
int nr_csums;
spin_lock(&delayed_refs->lock);
delayed_refs->pending_csums -= head->num_bytes;
spin_unlock(&delayed_refs->lock);
nr_csums = btrfs_csum_bytes_to_leaves(fs_info, head->num_bytes);
btrfs_delayed_refs_rsv_release(fs_info, 0, nr_csums);
ret = btrfs_calc_delayed_ref_csum_bytes(fs_info, nr_csums);
}
/* must_insert_reserved can be set only if we didn't run the head ref. */
if (head->must_insert_reserved)
free_head_ref_squota_rsv(fs_info, head);
return ret;
}
static int cleanup_ref_head(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *head,
u64 *bytes_released)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_delayed_ref_root *delayed_refs;
int ret;
delayed_refs = &trans->transaction->delayed_refs;
ret = run_and_cleanup_extent_op(trans, head);
if (ret < 0) {
unselect_delayed_ref_head(delayed_refs, head);
btrfs_debug(fs_info, "run_delayed_extent_op returned %d", ret);
return ret;
} else if (ret) {
return ret;
}
/*
* Need to drop our head ref lock and re-acquire the delayed ref lock
* and then re-check to make sure nobody got added.
*/
spin_unlock(&head->lock);
spin_lock(&delayed_refs->lock);
spin_lock(&head->lock);
if (!RB_EMPTY_ROOT(&head->ref_tree.rb_root) || head->extent_op) {
spin_unlock(&head->lock);
spin_unlock(&delayed_refs->lock);
return 1;
}
btrfs_delete_ref_head(delayed_refs, head);
spin_unlock(&head->lock);
spin_unlock(&delayed_refs->lock);
if (head->must_insert_reserved) {
btrfs_pin_extent(trans, head->bytenr, head->num_bytes, 1);
if (head->is_data) {
struct btrfs_root *csum_root;
csum_root = btrfs_csum_root(fs_info, head->bytenr);
ret = btrfs_del_csums(trans, csum_root, head->bytenr,
head->num_bytes);
}
}
*bytes_released += btrfs_cleanup_ref_head_accounting(fs_info, delayed_refs, head);
trace_run_delayed_ref_head(fs_info, head, 0);
btrfs_delayed_ref_unlock(head);
btrfs_put_delayed_ref_head(head);
return ret;
}
static struct btrfs_delayed_ref_head *btrfs_obtain_ref_head(
struct btrfs_trans_handle *trans)
{
struct btrfs_delayed_ref_root *delayed_refs =
&trans->transaction->delayed_refs;
struct btrfs_delayed_ref_head *head = NULL;
int ret;
spin_lock(&delayed_refs->lock);
head = btrfs_select_ref_head(delayed_refs);
if (!head) {
spin_unlock(&delayed_refs->lock);
return head;
}
/*
* Grab the lock that says we are going to process all the refs for
* this head
*/
ret = btrfs_delayed_ref_lock(delayed_refs, head);
spin_unlock(&delayed_refs->lock);
/*
* We may have dropped the spin lock to get the head mutex lock, and
* that might have given someone else time to free the head. If that's
* true, it has been removed from our list and we can move on.
*/
if (ret == -EAGAIN)
head = ERR_PTR(-EAGAIN);
return head;
}
static int btrfs_run_delayed_refs_for_head(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *locked_ref,
u64 *bytes_released)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_delayed_ref_root *delayed_refs;
struct btrfs_delayed_extent_op *extent_op;
struct btrfs_delayed_ref_node *ref;
bool must_insert_reserved;
int ret;
delayed_refs = &trans->transaction->delayed_refs;
lockdep_assert_held(&locked_ref->mutex);
lockdep_assert_held(&locked_ref->lock);
while ((ref = select_delayed_ref(locked_ref))) {
if (ref->seq &&
btrfs_check_delayed_seq(fs_info, ref->seq)) {
spin_unlock(&locked_ref->lock);
unselect_delayed_ref_head(delayed_refs, locked_ref);
return -EAGAIN;
}
rb_erase_cached(&ref->ref_node, &locked_ref->ref_tree);
RB_CLEAR_NODE(&ref->ref_node);
if (!list_empty(&ref->add_list))
list_del(&ref->add_list);
/*
* When we play the delayed ref, also correct the ref_mod on
* head
*/
switch (ref->action) {
case BTRFS_ADD_DELAYED_REF:
case BTRFS_ADD_DELAYED_EXTENT:
locked_ref->ref_mod -= ref->ref_mod;
break;
case BTRFS_DROP_DELAYED_REF:
locked_ref->ref_mod += ref->ref_mod;
break;
default:
WARN_ON(1);
}
atomic_dec(&delayed_refs->num_entries);
/*
* Record the must_insert_reserved flag before we drop the
* spin lock.
*/
must_insert_reserved = locked_ref->must_insert_reserved;
/*
* Unsetting this on the head ref relinquishes ownership of
* the rsv_bytes, so it is critical that every possible code
* path from here forward frees all reserves including qgroup
* reserve.
*/
locked_ref->must_insert_reserved = false;
extent_op = locked_ref->extent_op;
locked_ref->extent_op = NULL;
spin_unlock(&locked_ref->lock);
ret = run_one_delayed_ref(trans, locked_ref, ref, extent_op,
must_insert_reserved);
btrfs_delayed_refs_rsv_release(fs_info, 1, 0);
*bytes_released += btrfs_calc_delayed_ref_bytes(fs_info, 1);
btrfs_free_delayed_extent_op(extent_op);
if (ret) {
unselect_delayed_ref_head(delayed_refs, locked_ref);
btrfs_put_delayed_ref(ref);
return ret;
}
btrfs_put_delayed_ref(ref);
cond_resched();
spin_lock(&locked_ref->lock);
btrfs_merge_delayed_refs(fs_info, delayed_refs, locked_ref);
}
return 0;
}
/*
* Returns 0 on success or if called with an already aborted transaction.
* Returns -ENOMEM or -EIO on failure and will abort the transaction.
*/
static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
u64 min_bytes)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_delayed_ref_root *delayed_refs;
struct btrfs_delayed_ref_head *locked_ref = NULL;
int ret;
unsigned long count = 0;
unsigned long max_count = 0;
u64 bytes_processed = 0;
delayed_refs = &trans->transaction->delayed_refs;
if (min_bytes == 0) {
max_count = delayed_refs->num_heads_ready;
min_bytes = U64_MAX;
}
do {
if (!locked_ref) {
locked_ref = btrfs_obtain_ref_head(trans);
if (IS_ERR_OR_NULL(locked_ref)) {
if (PTR_ERR(locked_ref) == -EAGAIN) {
continue;
} else {
break;
}
}
count++;
}
/*
* We need to try and merge add/drops of the same ref since we
* can run into issues with relocate dropping the implicit ref
* and then it being added back again before the drop can
* finish. If we merged anything we need to re-loop so we can
* get a good ref.
* Or we can get node references of the same type that weren't
* merged when created due to bumps in the tree mod seq, and
* we need to merge them to prevent adding an inline extent
* backref before dropping it (triggering a BUG_ON at
* insert_inline_extent_backref()).
*/
spin_lock(&locked_ref->lock);
btrfs_merge_delayed_refs(fs_info, delayed_refs, locked_ref);
ret = btrfs_run_delayed_refs_for_head(trans, locked_ref, &bytes_processed);
if (ret < 0 && ret != -EAGAIN) {
/*
* Error, btrfs_run_delayed_refs_for_head already
* unlocked everything so just bail out
*/
return ret;
} else if (!ret) {
/*
* Success, perform the usual cleanup of a processed
* head
*/
ret = cleanup_ref_head(trans, locked_ref, &bytes_processed);
if (ret > 0 ) {
/* We dropped our lock, we need to loop. */
ret = 0;
continue;
} else if (ret) {
return ret;
}
}
/*
* Either success case or btrfs_run_delayed_refs_for_head
* returned -EAGAIN, meaning we need to select another head
*/
locked_ref = NULL;
cond_resched();
} while ((min_bytes != U64_MAX && bytes_processed < min_bytes) ||
(max_count > 0 && count < max_count) ||
locked_ref);
return 0;
}
#ifdef SCRAMBLE_DELAYED_REFS
/*
* Normally delayed refs get processed in ascending bytenr order. This
* correlates in most cases to the order added. To expose dependencies on this
* order, we start to process the tree in the middle instead of the beginning
*/
static u64 find_middle(struct rb_root *root)
{
struct rb_node *n = root->rb_node;
struct btrfs_delayed_ref_node *entry;
int alt = 1;
u64 middle;
u64 first = 0, last = 0;
n = rb_first(root);
if (n) {
entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
first = entry->bytenr;
}
n = rb_last(root);
if (n) {
entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
last = entry->bytenr;
}
n = root->rb_node;
while (n) {
entry = rb_entry(n, struct btrfs_delayed_ref_node, rb_node);
WARN_ON(!entry->in_tree);
middle = entry->bytenr;
if (alt)
n = n->rb_left;
else
n = n->rb_right;
alt = 1 - alt;
}
return middle;
}
#endif
/*
* Start processing the delayed reference count updates and extent insertions
* we have queued up so far.
*
* @trans: Transaction handle.
* @min_bytes: How many bytes of delayed references to process. After this
* many bytes we stop processing delayed references if there are
* any more. If 0 it means to run all existing delayed references,
* but not new ones added after running all existing ones.
* Use (u64)-1 (U64_MAX) to run all existing delayed references
* plus any new ones that are added.
*
* Returns 0 on success or if called with an aborted transaction
* Returns <0 on error and aborts the transaction
*/
int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans, u64 min_bytes)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_delayed_ref_root *delayed_refs;
int ret;
/* We'll clean this up in btrfs_cleanup_transaction */
if (TRANS_ABORTED(trans))
return 0;
if (test_bit(BTRFS_FS_CREATING_FREE_SPACE_TREE, &fs_info->flags))
return 0;
delayed_refs = &trans->transaction->delayed_refs;
again:
#ifdef SCRAMBLE_DELAYED_REFS
delayed_refs->run_delayed_start = find_middle(&delayed_refs->root);
#endif
ret = __btrfs_run_delayed_refs(trans, min_bytes);
if (ret < 0) {
btrfs_abort_transaction(trans, ret);
return ret;
}
if (min_bytes == U64_MAX) {
btrfs_create_pending_block_groups(trans);
spin_lock(&delayed_refs->lock);
if (RB_EMPTY_ROOT(&delayed_refs->href_root.rb_root)) {
spin_unlock(&delayed_refs->lock);
return 0;
}
spin_unlock(&delayed_refs->lock);
cond_resched();
goto again;
}
return 0;
}
int btrfs_set_disk_extent_flags(struct btrfs_trans_handle *trans,
struct extent_buffer *eb, u64 flags)
{
struct btrfs_delayed_extent_op *extent_op;
int level = btrfs_header_level(eb);
int ret;
extent_op = btrfs_alloc_delayed_extent_op();
if (!extent_op)
return -ENOMEM;
extent_op->flags_to_set = flags;
extent_op->update_flags = true;
extent_op->update_key = false;
extent_op->level = level;
ret = btrfs_add_delayed_extent_op(trans, eb->start, eb->len, extent_op);
if (ret)
btrfs_free_delayed_extent_op(extent_op);
return ret;
}
static noinline int check_delayed_ref(struct btrfs_root *root,
struct btrfs_path *path,
u64 objectid, u64 offset, u64 bytenr)
{
struct btrfs_delayed_ref_head *head;
struct btrfs_delayed_ref_node *ref;
struct btrfs_delayed_ref_root *delayed_refs;
struct btrfs_transaction *cur_trans;
struct rb_node *node;
int ret = 0;
spin_lock(&root->fs_info->trans_lock);
cur_trans = root->fs_info->running_transaction;
if (cur_trans)
refcount_inc(&cur_trans->use_count);
spin_unlock(&root->fs_info->trans_lock);
if (!cur_trans)
return 0;
delayed_refs = &cur_trans->delayed_refs;
spin_lock(&delayed_refs->lock);
head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
if (!head) {
spin_unlock(&delayed_refs->lock);
btrfs_put_transaction(cur_trans);
return 0;
}
if (!mutex_trylock(&head->mutex)) {
if (path->nowait) {
spin_unlock(&delayed_refs->lock);
btrfs_put_transaction(cur_trans);
return -EAGAIN;
}
refcount_inc(&head->refs);
spin_unlock(&delayed_refs->lock);
btrfs_release_path(path);
/*
* Mutex was contended, block until it's released and let
* caller try again
*/
mutex_lock(&head->mutex);
mutex_unlock(&head->mutex);
btrfs_put_delayed_ref_head(head);
btrfs_put_transaction(cur_trans);
return -EAGAIN;
}
spin_unlock(&delayed_refs->lock);
spin_lock(&head->lock);
/*
* XXX: We should replace this with a proper search function in the
* future.
*/
for (node = rb_first_cached(&head->ref_tree); node;
node = rb_next(node)) {
u64 ref_owner;
u64 ref_offset;
ref = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
/* If it's a shared ref we know a cross reference exists */
if (ref->type != BTRFS_EXTENT_DATA_REF_KEY) {
ret = 1;
break;
}
ref_owner = btrfs_delayed_ref_owner(ref);
ref_offset = btrfs_delayed_ref_offset(ref);
/*
* If our ref doesn't match the one we're currently looking at
* then we have a cross reference.
*/
if (ref->ref_root != btrfs_root_id(root) ||
ref_owner != objectid || ref_offset != offset) {
ret = 1;
break;
}
}
spin_unlock(&head->lock);
mutex_unlock(&head->mutex);
btrfs_put_transaction(cur_trans);
return ret;
}
static noinline int check_committed_ref(struct btrfs_root *root,
struct btrfs_path *path,
u64 objectid, u64 offset, u64 bytenr,
bool strict)
{
struct btrfs_fs_info *fs_info = root->fs_info;
struct btrfs_root *extent_root = btrfs_extent_root(fs_info, bytenr);
struct extent_buffer *leaf;
struct btrfs_extent_data_ref *ref;
struct btrfs_extent_inline_ref *iref;
struct btrfs_extent_item *ei;
struct btrfs_key key;
u32 item_size;
u32 expected_size;
int type;
int ret;
key.objectid = bytenr;
key.offset = (u64)-1;
key.type = BTRFS_EXTENT_ITEM_KEY;
ret = btrfs_search_slot(NULL, extent_root, &key, path, 0, 0);
if (ret < 0)
goto out;
if (ret == 0) {
/*
* Key with offset -1 found, there would have to exist an extent
* item with such offset, but this is out of the valid range.
*/
ret = -EUCLEAN;
goto out;
}
ret = -ENOENT;
if (path->slots[0] == 0)
goto out;
path->slots[0]--;
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (key.objectid != bytenr || key.type != BTRFS_EXTENT_ITEM_KEY)
goto out;
ret = 1;
item_size = btrfs_item_size(leaf, path->slots[0]);
ei = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
expected_size = sizeof(*ei) + btrfs_extent_inline_ref_size(BTRFS_EXTENT_DATA_REF_KEY);
/* No inline refs; we need to bail before checking for owner ref. */
if (item_size == sizeof(*ei))
goto out;
/* Check for an owner ref; skip over it to the real inline refs. */
iref = (struct btrfs_extent_inline_ref *)(ei + 1);
type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
if (btrfs_fs_incompat(fs_info, SIMPLE_QUOTA) && type == BTRFS_EXTENT_OWNER_REF_KEY) {
expected_size += btrfs_extent_inline_ref_size(BTRFS_EXTENT_OWNER_REF_KEY);
iref = (struct btrfs_extent_inline_ref *)(iref + 1);
}
/* If extent item has more than 1 inline ref then it's shared */
if (item_size != expected_size)
goto out;
/*
* If extent created before last snapshot => it's shared unless the
* snapshot has been deleted. Use the heuristic if strict is false.
*/
if (!strict &&
(btrfs_extent_generation(leaf, ei) <=
btrfs_root_last_snapshot(&root->root_item)))
goto out;
/* If this extent has SHARED_DATA_REF then it's shared */
type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_DATA);
if (type != BTRFS_EXTENT_DATA_REF_KEY)
goto out;
ref = (struct btrfs_extent_data_ref *)(&iref->offset);
if (btrfs_extent_refs(leaf, ei) !=
btrfs_extent_data_ref_count(leaf, ref) ||
btrfs_extent_data_ref_root(leaf, ref) != btrfs_root_id(root) ||
btrfs_extent_data_ref_objectid(leaf, ref) != objectid ||
btrfs_extent_data_ref_offset(leaf, ref) != offset)
goto out;
ret = 0;
out:
return ret;
}
int btrfs_cross_ref_exist(struct btrfs_root *root, u64 objectid, u64 offset,
u64 bytenr, bool strict, struct btrfs_path *path)
{
int ret;
do {
ret = check_committed_ref(root, path, objectid,
offset, bytenr, strict);
if (ret && ret != -ENOENT)
goto out;
ret = check_delayed_ref(root, path, objectid, offset, bytenr);
} while (ret == -EAGAIN);
out:
btrfs_release_path(path);
if (btrfs_is_data_reloc_root(root))
WARN_ON(ret > 0);
return ret;
}
static int __btrfs_mod_ref(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct extent_buffer *buf,
int full_backref, int inc)
{
struct btrfs_fs_info *fs_info = root->fs_info;
u64 parent;
u64 ref_root;
u32 nritems;
struct btrfs_key key;
struct btrfs_file_extent_item *fi;
bool for_reloc = btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC);
int i;
int action;
int level;
int ret = 0;
if (btrfs_is_testing(fs_info))
return 0;
ref_root = btrfs_header_owner(buf);
nritems = btrfs_header_nritems(buf);
level = btrfs_header_level(buf);
if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state) && level == 0)
return 0;
if (full_backref)
parent = buf->start;
else
parent = 0;
if (inc)
action = BTRFS_ADD_DELAYED_REF;
else
action = BTRFS_DROP_DELAYED_REF;
for (i = 0; i < nritems; i++) {
struct btrfs_ref ref = {
.action = action,
.parent = parent,
.ref_root = ref_root,
};
if (level == 0) {
btrfs_item_key_to_cpu(buf, &key, i);
if (key.type != BTRFS_EXTENT_DATA_KEY)
continue;
fi = btrfs_item_ptr(buf, i,
struct btrfs_file_extent_item);
if (btrfs_file_extent_type(buf, fi) ==
BTRFS_FILE_EXTENT_INLINE)
continue;
ref.bytenr = btrfs_file_extent_disk_bytenr(buf, fi);
if (ref.bytenr == 0)
continue;
ref.num_bytes = btrfs_file_extent_disk_num_bytes(buf, fi);
ref.owning_root = ref_root;
key.offset -= btrfs_file_extent_offset(buf, fi);
btrfs_init_data_ref(&ref, key.objectid, key.offset,
btrfs_root_id(root), for_reloc);
if (inc)
ret = btrfs_inc_extent_ref(trans, &ref);
else
ret = btrfs_free_extent(trans, &ref);
if (ret)
goto fail;
} else {
/* We don't know the owning_root, leave as 0. */
ref.bytenr = btrfs_node_blockptr(buf, i);
ref.num_bytes = fs_info->nodesize;
btrfs_init_tree_ref(&ref, level - 1,
btrfs_root_id(root), for_reloc);
if (inc)
ret = btrfs_inc_extent_ref(trans, &ref);
else
ret = btrfs_free_extent(trans, &ref);
if (ret)
goto fail;
}
}
return 0;
fail:
return ret;
}
int btrfs_inc_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
struct extent_buffer *buf, int full_backref)
{
return __btrfs_mod_ref(trans, root, buf, full_backref, 1);
}
int btrfs_dec_ref(struct btrfs_trans_handle *trans, struct btrfs_root *root,
struct extent_buffer *buf, int full_backref)
{
return __btrfs_mod_ref(trans, root, buf, full_backref, 0);
}
static u64 get_alloc_profile_by_root(struct btrfs_root *root, int data)
{
struct btrfs_fs_info *fs_info = root->fs_info;
u64 flags;
u64 ret;
if (data)
flags = BTRFS_BLOCK_GROUP_DATA;
else if (root == fs_info->chunk_root)
flags = BTRFS_BLOCK_GROUP_SYSTEM;
else
flags = BTRFS_BLOCK_GROUP_METADATA;
ret = btrfs_get_alloc_profile(fs_info, flags);
return ret;
}
static u64 first_logical_byte(struct btrfs_fs_info *fs_info)
{
struct rb_node *leftmost;
u64 bytenr = 0;
read_lock(&fs_info->block_group_cache_lock);
/* Get the block group with the lowest logical start address. */
leftmost = rb_first_cached(&fs_info->block_group_cache_tree);
if (leftmost) {
struct btrfs_block_group *bg;
bg = rb_entry(leftmost, struct btrfs_block_group, cache_node);
bytenr = bg->start;
}
read_unlock(&fs_info->block_group_cache_lock);
return bytenr;
}
static int pin_down_extent(struct btrfs_trans_handle *trans,
struct btrfs_block_group *cache,
u64 bytenr, u64 num_bytes, int reserved)
{
struct btrfs_fs_info *fs_info = cache->fs_info;
spin_lock(&cache->space_info->lock);
spin_lock(&cache->lock);
cache->pinned += num_bytes;
btrfs_space_info_update_bytes_pinned(fs_info, cache->space_info,
num_bytes);
if (reserved) {
cache->reserved -= num_bytes;
cache->space_info->bytes_reserved -= num_bytes;
}
spin_unlock(&cache->lock);
spin_unlock(&cache->space_info->lock);
set_extent_bit(&trans->transaction->pinned_extents, bytenr,
bytenr + num_bytes - 1, EXTENT_DIRTY, NULL);
return 0;
}
int btrfs_pin_extent(struct btrfs_trans_handle *trans,
u64 bytenr, u64 num_bytes, int reserved)
{
struct btrfs_block_group *cache;
cache = btrfs_lookup_block_group(trans->fs_info, bytenr);
BUG_ON(!cache); /* Logic error */
pin_down_extent(trans, cache, bytenr, num_bytes, reserved);
btrfs_put_block_group(cache);
return 0;
}
int btrfs_pin_extent_for_log_replay(struct btrfs_trans_handle *trans,
const struct extent_buffer *eb)
{
struct btrfs_block_group *cache;
int ret;
cache = btrfs_lookup_block_group(trans->fs_info, eb->start);
if (!cache)
return -EINVAL;
/*
* Fully cache the free space first so that our pin removes the free space
* from the cache.
*/
ret = btrfs_cache_block_group(cache, true);
if (ret)
goto out;
pin_down_extent(trans, cache, eb->start, eb->len, 0);
/* remove us from the free space cache (if we're there at all) */
ret = btrfs_remove_free_space(cache, eb->start, eb->len);
out:
btrfs_put_block_group(cache);
return ret;
}
static int __exclude_logged_extent(struct btrfs_fs_info *fs_info,
u64 start, u64 num_bytes)
{
int ret;
struct btrfs_block_group *block_group;
block_group = btrfs_lookup_block_group(fs_info, start);
if (!block_group)
return -EINVAL;
ret = btrfs_cache_block_group(block_group, true);
if (ret)
goto out;
ret = btrfs_remove_free_space(block_group, start, num_bytes);
out:
btrfs_put_block_group(block_group);
return ret;
}
int btrfs_exclude_logged_extents(struct extent_buffer *eb)
{
struct btrfs_fs_info *fs_info = eb->fs_info;
struct btrfs_file_extent_item *item;
struct btrfs_key key;
int found_type;
int i;
int ret = 0;
if (!btrfs_fs_incompat(fs_info, MIXED_GROUPS))
return 0;
for (i = 0; i < btrfs_header_nritems(eb); i++) {
btrfs_item_key_to_cpu(eb, &key, i);
if (key.type != BTRFS_EXTENT_DATA_KEY)
continue;
item = btrfs_item_ptr(eb, i, struct btrfs_file_extent_item);
found_type = btrfs_file_extent_type(eb, item);
if (found_type == BTRFS_FILE_EXTENT_INLINE)
continue;
if (btrfs_file_extent_disk_bytenr(eb, item) == 0)
continue;
key.objectid = btrfs_file_extent_disk_bytenr(eb, item);
key.offset = btrfs_file_extent_disk_num_bytes(eb, item);
ret = __exclude_logged_extent(fs_info, key.objectid, key.offset);
if (ret)
break;
}
return ret;
}
static void
btrfs_inc_block_group_reservations(struct btrfs_block_group *bg)
{
atomic_inc(&bg->reservations);
}
/*
* Returns the free cluster for the given space info and sets empty_cluster to
* what it should be based on the mount options.
*/
static struct btrfs_free_cluster *
fetch_cluster_info(struct btrfs_fs_info *fs_info,
struct btrfs_space_info *space_info, u64 *empty_cluster)
{
struct btrfs_free_cluster *ret = NULL;
*empty_cluster = 0;
if (btrfs_mixed_space_info(space_info))
return ret;
if (space_info->flags & BTRFS_BLOCK_GROUP_METADATA) {
ret = &fs_info->meta_alloc_cluster;
if (btrfs_test_opt(fs_info, SSD))
*empty_cluster = SZ_2M;
else
*empty_cluster = SZ_64K;
} else if ((space_info->flags & BTRFS_BLOCK_GROUP_DATA) &&
btrfs_test_opt(fs_info, SSD_SPREAD)) {
*empty_cluster = SZ_2M;
ret = &fs_info->data_alloc_cluster;
}
return ret;
}
static int unpin_extent_range(struct btrfs_fs_info *fs_info,
u64 start, u64 end,
const bool return_free_space)
{
struct btrfs_block_group *cache = NULL;
struct btrfs_space_info *space_info;
struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
struct btrfs_free_cluster *cluster = NULL;
u64 len;
u64 total_unpinned = 0;
u64 empty_cluster = 0;
bool readonly;
int ret = 0;
while (start <= end) {
readonly = false;
if (!cache ||
start >= cache->start + cache->length) {
if (cache)
btrfs_put_block_group(cache);
total_unpinned = 0;
cache = btrfs_lookup_block_group(fs_info, start);
if (cache == NULL) {
/* Logic error, something removed the block group. */
ret = -EUCLEAN;
goto out;
}
cluster = fetch_cluster_info(fs_info,
cache->space_info,
&empty_cluster);
empty_cluster <<= 1;
}
len = cache->start + cache->length - start;
len = min(len, end + 1 - start);
if (return_free_space)
btrfs_add_free_space(cache, start, len);
start += len;
total_unpinned += len;
space_info = cache->space_info;
/*
* If this space cluster has been marked as fragmented and we've
* unpinned enough in this block group to potentially allow a
* cluster to be created inside of it go ahead and clear the
* fragmented check.
*/
if (cluster && cluster->fragmented &&
total_unpinned > empty_cluster) {
spin_lock(&cluster->lock);
cluster->fragmented = 0;
spin_unlock(&cluster->lock);
}
spin_lock(&space_info->lock);
spin_lock(&cache->lock);
cache->pinned -= len;
btrfs_space_info_update_bytes_pinned(fs_info, space_info, -len);
space_info->max_extent_size = 0;
if (cache->ro) {
space_info->bytes_readonly += len;
readonly = true;
} else if (btrfs_is_zoned(fs_info)) {
/* Need reset before reusing in a zoned block group */
space_info->bytes_zone_unusable += len;
readonly = true;
}
spin_unlock(&cache->lock);
if (!readonly && return_free_space &&
global_rsv->space_info == space_info) {
spin_lock(&global_rsv->lock);
if (!global_rsv->full) {
u64 to_add = min(len, global_rsv->size -
global_rsv->reserved);
global_rsv->reserved += to_add;
btrfs_space_info_update_bytes_may_use(fs_info,
space_info, to_add);
if (global_rsv->reserved >= global_rsv->size)
global_rsv->full = 1;
len -= to_add;
}
spin_unlock(&global_rsv->lock);
}
/* Add to any tickets we may have */
if (!readonly && return_free_space && len)
btrfs_try_granting_tickets(fs_info, space_info);
spin_unlock(&space_info->lock);
}
if (cache)
btrfs_put_block_group(cache);
out:
return ret;
}
int btrfs_finish_extent_commit(struct btrfs_trans_handle *trans)
{
struct btrfs_fs_info *fs_info = trans->fs_info;
struct btrfs_block_group *block_group, *tmp;
struct list_head *deleted_bgs;
struct extent_io_tree *unpin;
u64 start;
u64 end;
int ret;
unpin = &trans->transaction->pinned_extents;
while (!TRANS_ABORTED(trans)) {
struct extent_state *cached_state = NULL;
mutex_lock(&fs_info->unused_bg_unpin_mutex);
if (!find_first_extent_bit(unpin, 0, &start, &end,
EXTENT_DIRTY, &cached_state)) {
mutex_unlock(&fs_info->unused_bg_unpin_mutex);
break;
}
if (btrfs_test_opt(fs_info, DISCARD_SYNC))
ret = btrfs_discard_extent(fs_info, start,
end + 1 - start, NULL);
clear_extent_dirty(unpin, start, end, &cached_state);
ret = unpin_extent_range(fs_info, start, end, true);
BUG_ON(ret);
mutex_unlock(&fs_info->unused_bg_unpin_mutex);
free_extent_state(cached_state);
cond_resched();
}
if (btrfs_test_opt(fs_info, DISCARD_ASYNC)) {
btrfs_discard_calc_delay(&fs_info->discard_ctl);
btrfs_discard_schedule_work(&fs_info->discard_ctl, true);
}
/*
* Transaction is finished. We don't need the lock anymore. We
* do need to clean up the block groups in case of a transaction
* abort.
*/
deleted_bgs = &trans->transaction->deleted_bgs;
list_for_each_entry_safe(block_group, tmp, deleted_bgs, bg_list) {
u64 trimmed = 0;
ret = -EROFS;
if (!TRANS_ABORTED(trans))
ret = btrfs_discard_extent(fs_info,
block_group->start,
block_group->length,
&trimmed);
list_del_init(&block_group->bg_list);
btrfs_unfreeze_block_group(block_group);
btrfs_put_block_group(block_group);
if (ret) {
const char *errstr = btrfs_decode_error(ret);
btrfs_warn(fs_info,
"discard failed while removing blockgroup: errno=%d %s",
ret, errstr);
}
}
return 0;
}
/*
* Parse an extent item's inline extents looking for a simple quotas owner ref.
*
* @fs_info: the btrfs_fs_info for this mount
* @leaf: a leaf in the extent tree containing the extent item
* @slot: the slot in the leaf where the extent item is found
*
* Returns the objectid of the root that originally allocated the extent item
* if the inline owner ref is expected and present, otherwise 0.
*
* If an extent item has an owner ref item, it will be the first inline ref
* item. Therefore the logic is to check whether there are any inline ref
* items, then check the type of the first one.
*/
u64 btrfs_get_extent_owner_root(struct btrfs_fs_info *fs_info,
struct extent_buffer *leaf, int slot)
{
struct btrfs_extent_item *ei;
struct btrfs_extent_inline_ref *iref;
struct btrfs_extent_owner_ref *oref;
unsigned long ptr;
unsigned long end;
int type;
if (!btrfs_fs_incompat(fs_info, SIMPLE_QUOTA))
return 0;
ei = btrfs_item_ptr(leaf, slot, struct btrfs_extent_item);
ptr = (unsigned long)(ei + 1);
end = (unsigned long)ei + btrfs_item_size(leaf, slot);
/* No inline ref items of any kind, can't check type. */
if (ptr == end)
return 0;
iref = (struct btrfs_extent_inline_ref *)ptr;
type = btrfs_get_extent_inline_ref_type(leaf, iref, BTRFS_REF_TYPE_ANY);
/* We found an owner ref, get the root out of it. */
if (type == BTRFS_EXTENT_OWNER_REF_KEY) {
oref = (struct btrfs_extent_owner_ref *)(&iref->offset);
return btrfs_extent_owner_ref_root_id(leaf, oref);
}
/* We have inline refs, but not an owner ref. */
return 0;
}
static int do_free_extent_accounting(struct btrfs_trans_handle *trans,
u64 bytenr, struct btrfs_squota_delta *delta)
{
int ret;
u64 num_bytes = delta->num_bytes;
if (delta->is_data) {
struct btrfs_root *csum_root;
csum_root = btrfs_csum_root(trans->fs_info, bytenr);
ret = btrfs_del_csums(trans, csum_root, bytenr, num_bytes);
if (ret) {
btrfs_abort_transaction(trans, ret);
return ret;
}
ret = btrfs_delete_raid_extent(trans, bytenr, num_bytes);
if (ret) {
btrfs_abort_transaction(trans, ret);
return ret;
}
}
ret = btrfs_record_squota_delta(trans->fs_info, delta);
if (ret) {
btrfs_abort_transaction(trans, ret);
return ret;
}
ret = add_to_free_space_tree(trans, bytenr, num_bytes);
if (ret) {
btrfs_abort_transaction(trans, ret);
return ret;
}
ret = btrfs_update_block_group(trans, bytenr, num_bytes, false);
if (ret)
btrfs_abort_transaction(trans, ret);
return ret;
}
#define abort_and_dump(trans, path, fmt, args...) \
({ \
btrfs_abort_transaction(trans, -EUCLEAN); \
btrfs_print_leaf(path->nodes[0]); \
btrfs_crit(trans->fs_info, fmt, ##args); \
})
/*
* Drop one or more refs of @node.
*
* 1. Locate the extent refs.
* It's either inline in EXTENT/METADATA_ITEM or in keyed SHARED_* item.
* Locate it, then reduce the refs number or remove the ref line completely.
*
* 2. Update the refs count in EXTENT/METADATA_ITEM
*
* Inline backref case:
*
* in extent tree we have:
*
* item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
* refs 2 gen 6 flags DATA
* extent data backref root FS_TREE objectid 258 offset 0 count 1
* extent data backref root FS_TREE objectid 257 offset 0 count 1
*
* This function gets called with:
*
* node->bytenr = 13631488
* node->num_bytes = 1048576
* root_objectid = FS_TREE
* owner_objectid = 257
* owner_offset = 0
* refs_to_drop = 1
*
* Then we should get some like:
*
* item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 16201 itemsize 82
* refs 1 gen 6 flags DATA
* extent data backref root FS_TREE objectid 258 offset 0 count 1
*
* Keyed backref case:
*
* in extent tree we have:
*
* item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
* refs 754 gen 6 flags DATA
* [...]
* item 2 key (13631488 EXTENT_DATA_REF <HASH>) itemoff 3915 itemsize 28
* extent data backref root FS_TREE objectid 866 offset 0 count 1
*
* This function get called with:
*
* node->bytenr = 13631488
* node->num_bytes = 1048576
* root_objectid = FS_TREE
* owner_objectid = 866
* owner_offset = 0
* refs_to_drop = 1
*
* Then we should get some like:
*
* item 0 key (13631488 EXTENT_ITEM 1048576) itemoff 3971 itemsize 24
* refs 753 gen 6 flags DATA
*
* And that (13631488 EXTENT_DATA_REF <HASH>) gets removed.
*/
static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
struct btrfs_delayed_ref_head *href,
struct btrfs_delayed_ref_node *node,
struct btrfs_delayed_extent_op *extent_op)
{
struct btrfs_fs_info *info = trans->fs_info;
struct btrfs_key key;
struct btrfs_path *path;
struct btrfs_root *extent_root;
struct extent_buffer *leaf;
struct btrfs_extent_item *ei;
struct btrfs_extent_inline_ref *iref;
int ret;
int is_data;
int extent_slot = 0;
int found_extent = 0;
int num_to_del = 1;
int refs_to_drop = node->ref_mod;
u32 item_size;
u64 refs;
u64 bytenr = node->bytenr;
u64 num_bytes = node->num_bytes;
u64 owner_objectid = btrfs_delayed_ref_owner(node);
u64 owner_offset = btrfs_delayed_ref_offset(node);
bool skinny_metadata = btrfs_fs_incompat(