#include <linux/kernel.h>
#include <linux/bio.h>
#include <linux/buffer_head.h>
+#include <linux/file.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include "ioctl.h"
#include "print-tree.h"
#include "volumes.h"
+#include "ordered-data.h"
struct btrfs_iget_args {
u64 ino;
int btrfs_check_free_space(struct btrfs_root *root, u64 num_required,
int for_del)
{
- u64 total = btrfs_super_total_bytes(&root->fs_info->super_copy);
- u64 used = btrfs_super_bytes_used(&root->fs_info->super_copy);
+ u64 total;
+ u64 used;
u64 thresh;
unsigned long flags;
int ret = 0;
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ total = btrfs_super_total_bytes(&root->fs_info->super_copy);
+ used = btrfs_super_bytes_used(&root->fs_info->super_copy);
if (for_del)
thresh = total * 90;
else
do_div(thresh, 100);
- spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
if (used + root->fs_info->delalloc_bytes + num_required > thresh)
ret = -ENOSPC;
spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
u64 num_bytes;
u64 cur_alloc_size;
u64 blocksize = root->sectorsize;
- u64 orig_start = start;
u64 orig_num_bytes;
struct btrfs_key ins;
- int ret;
+ struct extent_map *em;
+ struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ int ret = 0;
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
BUG_ON(!trans);
btrfs_set_trans_block_group(trans, inode);
num_bytes = (end - start + blocksize) & ~(blocksize - 1);
num_bytes = max(blocksize, num_bytes);
- ret = btrfs_drop_extents(trans, root, inode,
- start, start + num_bytes, start, &alloc_hint);
orig_num_bytes = num_bytes;
if (alloc_hint == EXTENT_MAP_INLINE)
goto out;
BUG_ON(num_bytes > btrfs_super_total_bytes(&root->fs_info->super_copy));
+ btrfs_drop_extent_cache(inode, start, start + num_bytes - 1);
while(num_bytes > 0) {
cur_alloc_size = min(num_bytes, root->fs_info->max_extent);
- ret = btrfs_alloc_extent(trans, root, cur_alloc_size,
- root->sectorsize,
- root->root_key.objectid,
- trans->transid,
- inode->i_ino, start, 0,
- alloc_hint, (u64)-1, &ins, 1);
+ ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
+ root->sectorsize, 0, 0,
+ (u64)-1, &ins, 1);
if (ret) {
WARN_ON(1);
goto out;
}
+ em = alloc_extent_map(GFP_NOFS);
+ em->start = start;
+ em->len = ins.offset;
+ em->block_start = ins.objectid;
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
+ while(1) {
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em);
+ spin_unlock(&em_tree->lock);
+ if (ret != -EEXIST) {
+ free_extent_map(em);
+ break;
+ }
+ btrfs_drop_extent_cache(inode, start,
+ start + ins.offset - 1);
+ }
+
cur_alloc_size = ins.offset;
- ret = btrfs_insert_file_extent(trans, root, inode->i_ino,
- start, ins.objectid, ins.offset,
+ ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
ins.offset);
- inode->i_blocks += ins.offset >> 9;
- btrfs_check_file(root, inode);
+ BUG_ON(ret);
if (num_bytes < cur_alloc_size) {
printk("num_bytes %Lu cur_alloc %Lu\n", num_bytes,
cur_alloc_size);
alloc_hint = ins.objectid + ins.offset;
start += cur_alloc_size;
}
- btrfs_drop_extent_cache(inode, orig_start,
- orig_start + orig_num_bytes - 1);
- btrfs_add_ordered_inode(inode);
- btrfs_update_inode(trans, root, inode);
out:
btrfs_end_transaction(trans, root);
return ret;
u64 loops = 0;
u64 total_fs_bytes;
struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_block_group_cache *block_group;
struct extent_buffer *leaf;
int found_type;
struct btrfs_path *path;
btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
found_type = btrfs_key_type(&found_key);
if (found_key.objectid != inode->i_ino ||
- found_type != BTRFS_EXTENT_DATA_KEY) {
+ found_type != BTRFS_EXTENT_DATA_KEY)
goto not_found;
- }
found_type = btrfs_file_extent_type(leaf, item);
extent_start = found_key.offset;
if (bytenr == 0)
goto not_found;
+ if (btrfs_count_snapshots_in_path(root, path, inode->i_ino,
+ bytenr) != 1) {
+ goto not_found;
+ }
+
/*
* we may be called by the resizer, make sure we're inside
* the limits of the FS
*/
- if (bytenr + extent_num_bytes > total_fs_bytes)
+ block_group = btrfs_lookup_block_group(root->fs_info,
+ bytenr);
+ if (!block_group || block_group->ro)
goto not_found;
- if (btrfs_count_snapshots_in_path(root, path, bytenr) != 1) {
- goto not_found;
- }
-
start = extent_end;
} else {
goto not_found;
goto again;
not_found:
- cow_file_range(inode, start, cow_end);
- start = cow_end + 1;
+ cow_file_range(inode, start, end);
+ start = end + 1;
goto loop;
}
{
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret;
- mutex_lock(&root->fs_info->fs_mutex);
+
if (btrfs_test_opt(root, NODATACOW) ||
btrfs_test_flag(inode, NODATACOW))
ret = run_delalloc_nocow(inode, start, end);
else
ret = cow_file_range(inode, start, end);
- mutex_unlock(&root->fs_info->fs_mutex);
return ret;
}
int mirror_num)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
- struct btrfs_trans_handle *trans;
int ret = 0;
- char *sums = NULL;
+ struct btrfs_ordered_sum *sums;
ret = btrfs_csum_one_bio(root, bio, &sums);
BUG_ON(ret);
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
-
- btrfs_set_trans_block_group(trans, inode);
- btrfs_csum_file_blocks(trans, root, inode, bio, sums);
-
- ret = btrfs_end_transaction(trans, root);
+ ret = btrfs_add_ordered_sum(inode, sums);
BUG_ON(ret);
- mutex_unlock(&root->fs_info->fs_mutex);
-
- kfree(sums);
- return btrfs_map_bio(root, rw, bio, mirror_num);
+ return btrfs_map_bio(root, rw, bio, mirror_num, 1);
}
int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret = 0;
- if (!(rw & (1 << BIO_RW))) {
- ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
- BUG_ON(ret);
- goto mapit;
- }
+ ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
+ BUG_ON(ret);
- if (btrfs_test_opt(root, NODATASUM) ||
- btrfs_test_flag(inode, NODATASUM)) {
+ if (!(rw & (1 << BIO_RW))) {
goto mapit;
}
inode, rw, bio, mirror_num,
__btrfs_submit_bio_hook);
mapit:
- return btrfs_map_bio(root, rw, bio, mirror_num);
+ return btrfs_map_bio(root, rw, bio, mirror_num, 0);
+}
+
+static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
+ struct inode *inode, u64 file_offset,
+ struct list_head *list)
+{
+ struct list_head *cur;
+ struct btrfs_ordered_sum *sum;
+
+ btrfs_set_trans_block_group(trans, inode);
+ list_for_each(cur, list) {
+ sum = list_entry(cur, struct btrfs_ordered_sum, list);
+ mutex_lock(&BTRFS_I(inode)->csum_mutex);
+ btrfs_csum_file_blocks(trans, BTRFS_I(inode)->root,
+ inode, sum);
+ mutex_unlock(&BTRFS_I(inode)->csum_mutex);
+ }
+ return 0;
+}
+
+struct btrfs_writepage_fixup {
+ struct page *page;
+ struct btrfs_work work;
+};
+
+/* see btrfs_writepage_start_hook for details on why this is required */
+void btrfs_writepage_fixup_worker(struct btrfs_work *work)
+{
+ struct btrfs_writepage_fixup *fixup;
+ struct btrfs_ordered_extent *ordered;
+ struct page *page;
+ struct inode *inode;
+ u64 page_start;
+ u64 page_end;
+
+ fixup = container_of(work, struct btrfs_writepage_fixup, work);
+ page = fixup->page;
+
+ lock_page(page);
+ if (!page->mapping || !PageDirty(page) || !PageChecked(page)) {
+ ClearPageChecked(page);
+ goto out_page;
+ }
+
+ inode = page->mapping->host;
+ page_start = page_offset(page);
+ page_end = page_offset(page) + PAGE_CACHE_SIZE - 1;
+
+ lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered)
+ goto out;
+
+ set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start, page_end,
+ GFP_NOFS);
+ ClearPageChecked(page);
+out:
+ unlock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
+out_page:
+ unlock_page(page);
+ page_cache_release(page);
+}
+
+/*
+ * There are a few paths in the higher layers of the kernel that directly
+ * set the page dirty bit without asking the filesystem if it is a
+ * good idea. This causes problems because we want to make sure COW
+ * properly happens and the data=ordered rules are followed.
+ *
+ * In our case any range that doesn't have the EXTENT_ORDERED bit set
+ * hasn't been properly setup for IO. We kick off an async process
+ * to fix it up. The async helper will wait for ordered extents, set
+ * the delalloc bit and make it safe to write the page.
+ */
+int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end)
+{
+ struct inode *inode = page->mapping->host;
+ struct btrfs_writepage_fixup *fixup;
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret;
+
+ ret = test_range_bit(&BTRFS_I(inode)->io_tree, start, end,
+ EXTENT_ORDERED, 0);
+ if (ret)
+ return 0;
+
+ if (PageChecked(page))
+ return -EAGAIN;
+
+ fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
+ if (!fixup)
+ return -EAGAIN;
+printk("queueing worker to fixup page %lu %Lu\n", inode->i_ino, page_offset(page));
+ SetPageChecked(page);
+ page_cache_get(page);
+ fixup->work.func = btrfs_writepage_fixup_worker;
+ fixup->page = page;
+ btrfs_queue_worker(&root->fs_info->fixup_workers, &fixup->work);
+ return -EAGAIN;
+}
+
+int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
+ struct extent_state *state, int uptodate)
+{
+ struct inode *inode = page->mapping->host;
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_trans_handle *trans;
+ struct btrfs_ordered_extent *ordered_extent;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ u64 alloc_hint = 0;
+ struct list_head list;
+ struct btrfs_key ins;
+ int ret;
+
+ ret = btrfs_dec_test_ordered_pending(inode, start, end - start + 1);
+ if (!ret)
+ return 0;
+
+ trans = btrfs_join_transaction(root, 1);
+
+ ordered_extent = btrfs_lookup_ordered_extent(inode, start);
+ BUG_ON(!ordered_extent);
+
+ lock_extent(io_tree, ordered_extent->file_offset,
+ ordered_extent->file_offset + ordered_extent->len - 1,
+ GFP_NOFS);
+
+ INIT_LIST_HEAD(&list);
+
+ ins.objectid = ordered_extent->start;
+ ins.offset = ordered_extent->len;
+ ins.type = BTRFS_EXTENT_ITEM_KEY;
+ ret = btrfs_alloc_reserved_extent(trans, root, root->root_key.objectid,
+ trans->transid, inode->i_ino,
+ ordered_extent->file_offset, &ins);
+ BUG_ON(ret);
+ ret = btrfs_drop_extents(trans, root, inode,
+ ordered_extent->file_offset,
+ ordered_extent->file_offset +
+ ordered_extent->len,
+ ordered_extent->file_offset, &alloc_hint);
+ BUG_ON(ret);
+ ret = btrfs_insert_file_extent(trans, root, inode->i_ino,
+ ordered_extent->file_offset,
+ ordered_extent->start,
+ ordered_extent->len,
+ ordered_extent->len, 0);
+ BUG_ON(ret);
+ btrfs_drop_extent_cache(inode, ordered_extent->file_offset,
+ ordered_extent->file_offset +
+ ordered_extent->len - 1);
+ inode->i_blocks += ordered_extent->len >> 9;
+ unlock_extent(io_tree, ordered_extent->file_offset,
+ ordered_extent->file_offset + ordered_extent->len - 1,
+ GFP_NOFS);
+ add_pending_csums(trans, inode, ordered_extent->file_offset,
+ &ordered_extent->list);
+
+ btrfs_ordered_update_i_size(inode, ordered_extent);
+ btrfs_remove_ordered_extent(inode, ordered_extent);
+ /* once for us */
+ btrfs_put_ordered_extent(ordered_extent);
+ /* once for the tree */
+ btrfs_put_ordered_extent(ordered_extent);
+
+ btrfs_update_inode(trans, root, inode);
+ btrfs_end_transaction(trans, root);
+ return 0;
}
int btrfs_readpage_io_hook(struct page *page, u64 start, u64 end)
btrfs_test_flag(inode, NODATASUM))
return 0;
- mutex_lock(&root->fs_info->fs_mutex);
path = btrfs_alloc_path();
item = btrfs_lookup_csum(NULL, root, path, inode->i_ino, start, 0);
if (IS_ERR(item)) {
+ /*
+ * It is possible there is an ordered extent that has
+ * not yet finished for this range in the file. If so,
+ * that extent will have a csum cached, and it will insert
+ * the sum after all the blocks in the extent are fully
+ * on disk. So, look for an ordered extent and use the
+ * sum if found.
+ */
+ ret = btrfs_find_ordered_sum(inode, start, &csum);
+ if (ret == 0)
+ goto found;
+
ret = PTR_ERR(item);
/* a csum that isn't present is a preallocated region. */
if (ret == -ENOENT || ret == -EFBIG)
ret = 0;
csum = 0;
- printk("no csum found for inode %lu start %Lu\n", inode->i_ino, start);
+ printk("no csum found for inode %lu start %Lu\n", inode->i_ino,
+ start);
goto out;
}
read_extent_buffer(path->nodes[0], &csum, (unsigned long)item,
BTRFS_CRC32_SIZE);
+found:
set_state_private(io_tree, start, csum);
out:
if (path)
btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
return ret;
}
int last_mirror;
};
-int btrfs_readpage_io_failed_hook(struct bio *failed_bio,
- struct page *page, u64 start, u64 end,
- struct extent_state *state)
+int btrfs_io_failed_hook(struct bio *failed_bio,
+ struct page *page, u64 start, u64 end,
+ struct extent_state *state)
{
struct io_failure_record *failrec = NULL;
u64 private;
struct bio *bio;
int num_copies;
int ret;
+ int rw;
u64 logical;
ret = get_state_private(failure_tree, start, &private);
bio->bi_bdev = failed_bio->bi_bdev;
bio->bi_size = 0;
bio_add_page(bio, page, failrec->len, start - page_offset(page));
- btrfs_submit_bio_hook(inode, READ, bio, failrec->last_mirror);
+ if (failed_bio->bi_rw & (1 << BIO_RW))
+ rw = WRITE;
+ else
+ rw = READ;
+
+ BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio,
+ failrec->last_mirror);
+ return 0;
+}
+
+int btrfs_clean_io_failures(struct inode *inode, u64 start)
+{
+ u64 private;
+ u64 private_failure;
+ struct io_failure_record *failure;
+ int ret;
+
+ private = 0;
+ if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
+ (u64)-1, 1, EXTENT_DIRTY)) {
+ ret = get_state_private(&BTRFS_I(inode)->io_failure_tree,
+ start, &private_failure);
+ if (ret == 0) {
+ failure = (struct io_failure_record *)(unsigned long)
+ private_failure;
+ set_state_private(&BTRFS_I(inode)->io_failure_tree,
+ failure->start, 0);
+ clear_extent_bits(&BTRFS_I(inode)->io_failure_tree,
+ failure->start,
+ failure->start + failure->len - 1,
+ EXTENT_DIRTY | EXTENT_LOCKED,
+ GFP_NOFS);
+ kfree(failure);
+ }
+ }
return 0;
}
/* if the io failure tree for this inode is non-empty,
* check to see if we've recovered from a failed IO
*/
- private = 0;
- if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
- (u64)-1, 1, EXTENT_DIRTY)) {
- u64 private_failure;
- struct io_failure_record *failure;
- ret = get_state_private(&BTRFS_I(inode)->io_failure_tree,
- start, &private_failure);
- if (ret == 0) {
- failure = (struct io_failure_record *)(unsigned long)
- private_failure;
- set_state_private(&BTRFS_I(inode)->io_failure_tree,
- failure->start, 0);
- clear_extent_bits(&BTRFS_I(inode)->io_failure_tree,
- failure->start,
- failure->start + failure->len - 1,
- EXTENT_DIRTY | EXTENT_LOCKED,
- GFP_NOFS);
- kfree(failure);
- }
- }
+ btrfs_clean_io_failures(inode, start);
return 0;
zeroit:
path = btrfs_alloc_path();
BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
memcpy(&location, &BTRFS_I(inode)->location, sizeof(location));
ret = btrfs_lookup_inode(NULL, root, path, &location, 0);
inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
inode->i_uid = btrfs_inode_uid(leaf, inode_item);
inode->i_gid = btrfs_inode_gid(leaf, inode_item);
- inode->i_size = btrfs_inode_size(leaf, inode_item);
+ btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item));
tspec = btrfs_inode_atime(inode_item);
inode->i_atime.tv_sec = btrfs_timespec_sec(leaf, tspec);
btrfs_free_path(path);
inode_item = NULL;
- mutex_unlock(&root->fs_info->fs_mutex);
-
switch (inode->i_mode & S_IFMT) {
case S_IFREG:
inode->i_mapping->a_ops = &btrfs_aops;
return;
make_bad:
- btrfs_release_path(root, path);
btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
make_bad_inode(inode);
}
{
btrfs_set_inode_uid(leaf, item, inode->i_uid);
btrfs_set_inode_gid(leaf, item, inode->i_gid);
- btrfs_set_inode_size(leaf, item, inode->i_size);
+ btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size);
btrfs_set_inode_mode(leaf, item, inode->i_mode);
btrfs_set_inode_nlink(leaf, item, inode->i_nlink);
BTRFS_I(inode)->block_group->key.objectid);
}
-int btrfs_update_inode(struct btrfs_trans_handle *trans,
+int noinline btrfs_update_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct inode *inode)
{
btrfs_set_inode_last_trans(trans, inode);
ret = 0;
failed:
- btrfs_release_path(root, path);
btrfs_free_path(path);
return ret;
}
goto err;
}
ret = btrfs_delete_one_dir_name(trans, root, path, di);
+ btrfs_release_path(root, path);
dentry->d_inode->i_ctime = dir->i_ctime;
ret = btrfs_del_inode_ref(trans, root, name, name_len,
err:
btrfs_free_path(path);
if (!ret) {
- dir->i_size -= name_len * 2;
+ btrfs_i_size_write(dir, dir->i_size - name_len * 2);
dir->i_mtime = dir->i_ctime = CURRENT_TIME;
btrfs_update_inode(trans, root, dir);
#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
{
struct btrfs_root *root;
struct btrfs_trans_handle *trans;
- struct inode *inode = dentry->d_inode;
int ret;
unsigned long nr = 0;
root = BTRFS_I(dir)->root;
- mutex_lock(&root->fs_info->fs_mutex);
ret = btrfs_check_free_space(root, 1, 1);
if (ret)
ret = btrfs_unlink_trans(trans, root, dir, dentry);
nr = trans->blocks_used;
- if (inode->i_nlink == 0) {
- int found;
- /* if the inode isn't linked anywhere,
- * we don't need to worry about
- * data=ordered
- */
- found = btrfs_del_ordered_inode(inode);
- if (found == 1) {
- atomic_dec(&inode->i_count);
- }
- }
-
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
fail:
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return ret;
}
struct btrfs_trans_handle *trans;
unsigned long nr = 0;
- if (inode->i_size > BTRFS_EMPTY_DIR_SIZE)
+ if (inode->i_size > BTRFS_EMPTY_DIR_SIZE) {
return -ENOTEMPTY;
+ }
- mutex_lock(&root->fs_info->fs_mutex);
ret = btrfs_check_free_space(root, 1, 1);
if (ret)
goto fail;
/* now the directory is empty */
err = btrfs_unlink_trans(trans, root, dir, dentry);
if (!err) {
- inode->i_size = 0;
+ btrfs_i_size_write(inode, 0);
}
nr = trans->blocks_used;
- ret = btrfs_end_transaction(trans, root);
+ ret = btrfs_end_transaction_throttle(trans, root);
fail:
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
if (ret && !err)
err = ret;
ret = btrfs_del_items(trans, root, path, pending_del_slot,
pending_del_nr);
}
- btrfs_release_path(root, path);
btrfs_free_path(path);
inode->i_sb->s_dirt = 1;
return ret;
}
-static int btrfs_cow_one_page(struct inode *inode, struct page *page,
- size_t zero_start)
-{
- char *kaddr;
- struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
- u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
- u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
- int ret = 0;
-
- WARN_ON(!PageLocked(page));
- set_page_extent_mapped(page);
-
- lock_extent(io_tree, page_start, page_end, GFP_NOFS);
- set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start,
- page_end, GFP_NOFS);
-
- if (zero_start != PAGE_CACHE_SIZE) {
- kaddr = kmap(page);
- memset(kaddr + zero_start, 0, PAGE_CACHE_SIZE - zero_start);
- flush_dcache_page(page);
- kunmap(page);
- }
- set_page_dirty(page);
- unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
-
- return ret;
-}
-
/*
* taken from block_truncate_page, but does cow as it zeros out
* any bytes left in the last page in the file.
{
struct inode *inode = mapping->host;
struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct btrfs_ordered_extent *ordered;
+ char *kaddr;
u32 blocksize = root->sectorsize;
pgoff_t index = from >> PAGE_CACHE_SHIFT;
unsigned offset = from & (PAGE_CACHE_SIZE-1);
struct page *page;
int ret = 0;
u64 page_start;
+ u64 page_end;
if ((offset & (blocksize - 1)) == 0)
goto out;
ret = -ENOMEM;
+again:
page = grab_cache_page(mapping, index);
if (!page)
goto out;
+
+ page_start = page_offset(page);
+ page_end = page_start + PAGE_CACHE_SIZE - 1;
+
if (!PageUptodate(page)) {
ret = btrfs_readpage(NULL, page);
lock_page(page);
+ if (page->mapping != mapping) {
+ unlock_page(page);
+ page_cache_release(page);
+ goto again;
+ }
if (!PageUptodate(page)) {
ret = -EIO;
goto out;
}
}
- page_start = (u64)page->index << PAGE_CACHE_SHIFT;
+ wait_on_page_writeback(page);
+
+ lock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ set_page_extent_mapped(page);
+
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ unlock_page(page);
+ page_cache_release(page);
+ btrfs_wait_ordered_extent(inode, ordered);
+ btrfs_put_ordered_extent(ordered);
+ goto again;
+ }
- ret = btrfs_cow_one_page(inode, page, offset);
+ set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start,
+ page_end, GFP_NOFS);
+ ret = 0;
+ if (offset != PAGE_CACHE_SIZE) {
+ kaddr = kmap(page);
+ memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset);
+ flush_dcache_page(page);
+ kunmap(page);
+ }
+ ClearPageChecked(page);
+ set_page_dirty(page);
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
unlock_page(page);
page_cache_release(page);
if (attr->ia_size <= hole_start)
goto out;
- mutex_lock(&root->fs_info->fs_mutex);
err = btrfs_check_free_space(root, 1, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
if (err)
goto fail;
btrfs_truncate_page(inode->i_mapping, inode->i_size);
- lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
hole_size = block_end - hole_start;
+ btrfs_wait_ordered_range(inode, hole_start, hole_size);
+ lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
- mutex_lock(&root->fs_info->fs_mutex);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
err = btrfs_drop_extents(trans, root, inode,
err = btrfs_insert_file_extent(trans, root,
inode->i_ino,
hole_start, 0, 0,
- hole_size);
+ hole_size, 0);
btrfs_drop_extent_cache(inode, hole_start,
(u64)-1);
btrfs_check_file(root, inode);
}
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
if (err)
return err;
return err;
}
-void btrfs_put_inode(struct inode *inode)
-{
- int ret;
-
- if (!BTRFS_I(inode)->ordered_trans) {
- return;
- }
-
- if (mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY) ||
- mapping_tagged(inode->i_mapping, PAGECACHE_TAG_WRITEBACK))
- return;
-
- ret = btrfs_del_ordered_inode(inode);
- if (ret == 1) {
- atomic_dec(&inode->i_count);
- }
-}
-
void btrfs_delete_inode(struct inode *inode)
{
struct btrfs_trans_handle *trans;
unsigned long nr;
int ret;
+ btrfs_wait_ordered_range(inode, 0, (u64)-1);
truncate_inode_pages(&inode->i_data, 0);
if (is_bad_inode(inode)) {
goto no_delete;
}
- inode->i_size = 0;
- mutex_lock(&root->fs_info->fs_mutex);
+ btrfs_i_size_write(inode, 0);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
clear_inode(inode);
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return;
no_delete_lock:
nr = trans->blocks_used;
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
no_delete:
clear_inode(inode);
}
path = btrfs_alloc_path();
BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
*sub_root = btrfs_read_fs_root(root->fs_info, location,
dentry->d_name.name,
location->offset = 0;
btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
return 0;
}
inode->i_ino = args->ino;
BTRFS_I(inode)->root = args->root;
BTRFS_I(inode)->delalloc_bytes = 0;
+ BTRFS_I(inode)->disk_i_size = 0;
extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
extent_io_tree_init(&BTRFS_I(inode)->io_tree,
inode->i_mapping, GFP_NOFS);
extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
inode->i_mapping, GFP_NOFS);
- atomic_set(&BTRFS_I(inode)->ordered_writeback, 0);
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
return 0;
}
if (dentry->d_name.len > BTRFS_NAME_LEN)
return ERR_PTR(-ENAMETOOLONG);
- mutex_lock(&root->fs_info->fs_mutex);
ret = btrfs_inode_by_name(dir, dentry, &location);
- mutex_unlock(&root->fs_info->fs_mutex);
if (ret < 0)
return ERR_PTR(ret);
filp->f_pos = 1;
}
- mutex_lock(&root->fs_info->fs_mutex);
key.objectid = inode->i_ino;
path = btrfs_alloc_path();
path->reada = 2;
nopos:
ret = 0;
err:
- btrfs_release_path(root, path);
btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
return ret;
}
int ret = 0;
if (wait) {
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
ret = btrfs_commit_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
}
return ret;
}
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_trans_handle *trans;
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
btrfs_update_inode(trans, root, inode);
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
}
static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
inode->i_mapping, GFP_NOFS);
extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
inode->i_mapping, GFP_NOFS);
- atomic_set(&BTRFS_I(inode)->ordered_writeback, 0);
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
BTRFS_I(inode)->delalloc_bytes = 0;
+ BTRFS_I(inode)->disk_i_size = 0;
BTRFS_I(inode)->root = root;
if (mode & S_IFDIR)
dentry->d_parent->d_inode->i_ino);
}
parent_inode = dentry->d_parent->d_inode;
- parent_inode->i_size += dentry->d_name.len * 2;
+ btrfs_i_size_write(parent_inode, parent_inode->i_size +
+ dentry->d_name.len * 2);
parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
ret = btrfs_update_inode(trans, root,
dentry->d_parent->d_inode);
if (!new_valid_dev(rdev))
return -EINVAL;
- mutex_lock(&root->fs_info->fs_mutex);
err = btrfs_check_free_space(root, 1, 0);
if (err)
goto fail;
btrfs_update_inode_block_group(trans, dir);
out_unlock:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
fail:
- mutex_unlock(&root->fs_info->fs_mutex);
-
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
}
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return err;
}
unsigned long nr = 0;
u64 objectid;
- mutex_lock(&root->fs_info->fs_mutex);
err = btrfs_check_free_space(root, 1, 0);
if (err)
goto fail;
inode->i_mapping, GFP_NOFS);
extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
inode->i_mapping, GFP_NOFS);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
BTRFS_I(inode)->delalloc_bytes = 0;
- atomic_set(&BTRFS_I(inode)->ordered_writeback, 0);
+ BTRFS_I(inode)->disk_i_size = 0;
BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
}
dir->i_sb->s_dirt = 1;
btrfs_update_inode_block_group(trans, inode);
btrfs_update_inode_block_group(trans, dir);
out_unlock:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
fail:
- mutex_unlock(&root->fs_info->fs_mutex);
-
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
}
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return err;
}
#else
inc_nlink(inode);
#endif
- mutex_lock(&root->fs_info->fs_mutex);
err = btrfs_check_free_space(root, 1, 0);
if (err)
goto fail;
drop_inode = 1;
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
fail:
- mutex_unlock(&root->fs_info->fs_mutex);
-
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
}
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return err;
}
static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
{
- struct inode *inode;
+ struct inode *inode = NULL;
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(dir)->root;
int err = 0;
int drop_on_err = 0;
- u64 objectid;
+ u64 objectid = 0;
unsigned long nr = 1;
- mutex_lock(&root->fs_info->fs_mutex);
err = btrfs_check_free_space(root, 1, 0);
if (err)
goto out_unlock;
inode->i_fop = &btrfs_dir_file_operations;
btrfs_set_trans_block_group(trans, inode);
- inode->i_size = 0;
+ btrfs_i_size_write(inode, 0);
err = btrfs_update_inode(trans, root, inode);
if (err)
goto out_fail;
out_fail:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
out_unlock:
- mutex_unlock(&root->fs_info->fs_mutex);
if (drop_on_err)
iput(inode);
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return err;
}
static int merge_extent_mapping(struct extent_map_tree *em_tree,
struct extent_map *existing,
- struct extent_map *em)
+ struct extent_map *em,
+ u64 map_start, u64 map_len)
{
u64 start_diff;
- u64 new_end;
- int ret = 0;
- int real_blocks = existing->block_start < EXTENT_MAP_LAST_BYTE;
-
- if (real_blocks && em->block_start >= EXTENT_MAP_LAST_BYTE)
- goto invalid;
-
- if (!real_blocks && em->block_start != existing->block_start)
- goto invalid;
-
- new_end = max(existing->start + existing->len, em->start + em->len);
-
- if (existing->start >= em->start) {
- if (em->start + em->len < existing->start)
- goto invalid;
-
- start_diff = existing->start - em->start;
- if (real_blocks && em->block_start + start_diff !=
- existing->block_start)
- goto invalid;
-
- em->len = new_end - em->start;
-
- remove_extent_mapping(em_tree, existing);
- /* free for the tree */
- free_extent_map(existing);
- ret = add_extent_mapping(em_tree, em);
- } else if (em->start > existing->start) {
-
- if (existing->start + existing->len < em->start)
- goto invalid;
-
- start_diff = em->start - existing->start;
- if (real_blocks && existing->block_start + start_diff !=
- em->block_start)
- goto invalid;
-
- remove_extent_mapping(em_tree, existing);
- em->block_start = existing->block_start;
- em->start = existing->start;
- em->len = new_end - existing->start;
- free_extent_map(existing);
-
- ret = add_extent_mapping(em_tree, em);
- } else {
- goto invalid;
- }
- return ret;
-
-invalid:
- printk("invalid extent map merge [%Lu %Lu %Lu] [%Lu %Lu %Lu]\n",
- existing->start, existing->len, existing->block_start,
- em->start, em->len, em->block_start);
- return -EIO;
-}
+ BUG_ON(map_start < em->start || map_start >= extent_map_end(em));
+ start_diff = map_start - em->start;
+ em->start = map_start;
+ em->len = map_len;
+ if (em->block_start < EXTENT_MAP_LAST_BYTE)
+ em->block_start += start_diff;
+ return add_extent_mapping(em_tree, em);
+}
struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page,
size_t pg_offset, u64 start, u64 len,
path = btrfs_alloc_path();
BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
again:
spin_lock(&em_tree->lock);
em = lookup_extent_mapping(em_tree, start, len);
+ if (em)
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
spin_unlock(&em_tree->lock);
if (em) {
err = -ENOMEM;
goto out;
}
-
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
em->start = EXTENT_MAP_HOLE;
em->len = (u64)-1;
- em->bdev = inode->i_sb->s_bdev;
ret = btrfs_lookup_file_extent(trans, root, path,
objectid, start, trans != NULL);
if (ret < 0) {
free_extent_map(em);
em = NULL;
btrfs_release_path(root, path);
- trans = btrfs_start_transaction(root, 1);
+ trans = btrfs_join_transaction(root, 1);
goto again;
}
write_extent_buffer(leaf, map + pg_offset, ptr,
*/
if (ret == -EEXIST) {
struct extent_map *existing;
+
+ ret = 0;
+
existing = lookup_extent_mapping(em_tree, start, len);
if (existing && (existing->start > start ||
existing->start + existing->len <= start)) {
em->len);
if (existing) {
err = merge_extent_mapping(em_tree, existing,
- em);
+ em, start,
+ root->sectorsize);
free_extent_map(existing);
if (err) {
free_extent_map(em);
} else {
free_extent_map(em);
em = existing;
+ err = 0;
}
}
spin_unlock(&em_tree->lock);
btrfs_free_path(path);
if (trans) {
ret = btrfs_end_transaction(trans, root);
- if (!err)
+ if (!err) {
err = ret;
+ }
}
- mutex_unlock(&root->fs_info->fs_mutex);
if (err) {
free_extent_map(em);
WARN_ON(1);
return extent_readpages(tree, mapping, pages, nr_pages,
btrfs_get_extent);
}
-
-static int btrfs_releasepage(struct page *page, gfp_t gfp_flags)
+static int __btrfs_releasepage(struct page *page, gfp_t gfp_flags)
{
struct extent_io_tree *tree;
struct extent_map_tree *map;
return ret;
}
+static int btrfs_releasepage(struct page *page, gfp_t gfp_flags)
+{
+ return __btrfs_releasepage(page, gfp_flags);
+}
+
static void btrfs_invalidatepage(struct page *page, unsigned long offset)
{
struct extent_io_tree *tree;
+ struct btrfs_ordered_extent *ordered;
+ u64 page_start = page_offset(page);
+ u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
+ wait_on_page_writeback(page);
tree = &BTRFS_I(page->mapping->host)->io_tree;
- extent_invalidatepage(tree, page, offset);
- btrfs_releasepage(page, GFP_NOFS);
+ if (offset) {
+ btrfs_releasepage(page, GFP_NOFS);
+ return;
+ }
+
+ lock_extent(tree, page_start, page_end, GFP_NOFS);
+ ordered = btrfs_lookup_ordered_extent(page->mapping->host,
+ page_offset(page));
+ if (ordered) {
+ clear_extent_bit(tree, page_start, page_end,
+ EXTENT_DIRTY | EXTENT_DELALLOC |
+ EXTENT_LOCKED, 1, 0, GFP_NOFS);
+ btrfs_writepage_end_io_hook(page, page_start,
+ page_end, NULL, 1);
+ btrfs_put_ordered_extent(ordered);
+ lock_extent(tree, page_start, page_end, GFP_NOFS);
+ }
+ clear_extent_bit(tree, page_start, page_end,
+ EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
+ EXTENT_ORDERED,
+ 1, 1, GFP_NOFS);
+ __btrfs_releasepage(page, GFP_NOFS);
+
if (PagePrivate(page)) {
- invalidate_extent_lru(tree, page_offset(page), PAGE_CACHE_SIZE);
+ invalidate_extent_lru(tree, page_offset(page),
+ PAGE_CACHE_SIZE);
ClearPagePrivate(page);
set_page_private(page, 0);
page_cache_release(page);
{
struct inode *inode = fdentry(vma->vm_file)->d_inode;
struct btrfs_root *root = BTRFS_I(inode)->root;
- unsigned long end;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct btrfs_ordered_extent *ordered;
+ char *kaddr;
+ unsigned long zero_start;
loff_t size;
int ret;
u64 page_start;
+ u64 page_end;
- mutex_lock(&root->fs_info->fs_mutex);
ret = btrfs_check_free_space(root, PAGE_CACHE_SIZE, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
if (ret)
goto out;
ret = -EINVAL;
-
+again:
lock_page(page);
- wait_on_page_writeback(page);
size = i_size_read(inode);
- page_start = (u64)page->index << PAGE_CACHE_SHIFT;
+ page_start = page_offset(page);
+ page_end = page_start + PAGE_CACHE_SIZE - 1;
if ((page->mapping != inode->i_mapping) ||
- (page_start > size)) {
+ (page_start >= size)) {
/* page got truncated out from underneath us */
goto out_unlock;
}
+ wait_on_page_writeback(page);
+
+ lock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ set_page_extent_mapped(page);
+
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ unlock_page(page);
+ btrfs_wait_ordered_extent(inode, ordered);
+ btrfs_put_ordered_extent(ordered);
+ goto again;
+ }
+
+ set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start,
+ page_end, GFP_NOFS);
+ ret = 0;
/* page is wholly or partially inside EOF */
if (page_start + PAGE_CACHE_SIZE > size)
- end = size & ~PAGE_CACHE_MASK;
+ zero_start = size & ~PAGE_CACHE_MASK;
else
- end = PAGE_CACHE_SIZE;
+ zero_start = PAGE_CACHE_SIZE;
- ret = btrfs_cow_one_page(inode, page, end);
+ if (zero_start != PAGE_CACHE_SIZE) {
+ kaddr = kmap(page);
+ memset(kaddr + zero_start, 0, PAGE_CACHE_SIZE - zero_start);
+ flush_dcache_page(page);
+ kunmap(page);
+ }
+ ClearPageChecked(page);
+ set_page_dirty(page);
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
out_unlock:
unlock_page(page);
int ret;
struct btrfs_trans_handle *trans;
unsigned long nr;
+ u64 mask = root->sectorsize - 1;
if (!S_ISREG(inode->i_mode))
return;
btrfs_truncate_page(inode->i_mapping, inode->i_size);
- mutex_lock(&root->fs_info->fs_mutex);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
+ btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
+ btrfs_i_size_write(inode, inode->i_size);
/* FIXME, add redo link to tree so we don't leak on crash */
ret = btrfs_truncate_in_trans(trans, root, inode,
btrfs_update_inode(trans, root, inode);
nr = trans->blocks_used;
- ret = btrfs_end_transaction(trans, root);
+ ret = btrfs_end_transaction_throttle(trans, root);
BUG_ON(ret);
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
}
-static int noinline create_subvol(struct btrfs_root *root, char *name,
+/*
+ * Invalidate a single dcache entry at the root of the filesystem.
+ * Needed after creation of snapshot or subvolume.
+ */
+void btrfs_invalidate_dcache_root(struct btrfs_root *root, char *name,
int namelen)
{
- struct btrfs_trans_handle *trans;
- struct btrfs_key key;
- struct btrfs_root_item root_item;
- struct btrfs_inode_item *inode_item;
- struct extent_buffer *leaf;
- struct btrfs_root *new_root = root;
+ struct dentry *alias, *entry;
+ struct qstr qstr;
+
+ alias = d_find_alias(root->fs_info->sb->s_root->d_inode);
+ if (alias) {
+ qstr.name = name;
+ qstr.len = namelen;
+ /* change me if btrfs ever gets a d_hash operation */
+ qstr.hash = full_name_hash(qstr.name, qstr.len);
+ entry = d_lookup(alias, &qstr);
+ dput(alias);
+ if (entry) {
+ d_invalidate(entry);
+ dput(entry);
+ }
+ }
+}
+
+int btrfs_create_subvol_root(struct btrfs_root *new_root,
+ struct btrfs_trans_handle *trans, u64 new_dirid,
+ struct btrfs_block_group_cache *block_group)
+{
struct inode *inode;
- struct inode *dir;
int ret;
- int err;
- u64 objectid;
- u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
- unsigned long nr = 1;
-
- mutex_lock(&root->fs_info->fs_mutex);
- ret = btrfs_check_free_space(root, 1, 0);
- if (ret)
- goto fail_commit;
-
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
-
- ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root,
- 0, &objectid);
- if (ret)
- goto fail;
-
- leaf = __btrfs_alloc_free_block(trans, root, root->leafsize,
- objectid, trans->transid, 0, 0,
- 0, 0);
- if (IS_ERR(leaf))
- return PTR_ERR(leaf);
-
- btrfs_set_header_nritems(leaf, 0);
- btrfs_set_header_level(leaf, 0);
- btrfs_set_header_bytenr(leaf, leaf->start);
- btrfs_set_header_generation(leaf, trans->transid);
- btrfs_set_header_owner(leaf, objectid);
-
- write_extent_buffer(leaf, root->fs_info->fsid,
- (unsigned long)btrfs_header_fsid(leaf),
- BTRFS_FSID_SIZE);
- btrfs_mark_buffer_dirty(leaf);
-
- inode_item = &root_item.inode;
- memset(inode_item, 0, sizeof(*inode_item));
- inode_item->generation = cpu_to_le64(1);
- inode_item->size = cpu_to_le64(3);
- inode_item->nlink = cpu_to_le32(1);
- inode_item->nblocks = cpu_to_le64(1);
- inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
-
- btrfs_set_root_bytenr(&root_item, leaf->start);
- btrfs_set_root_level(&root_item, 0);
- btrfs_set_root_refs(&root_item, 1);
- btrfs_set_root_used(&root_item, 0);
-
- memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
- root_item.drop_level = 0;
-
- free_extent_buffer(leaf);
- leaf = NULL;
-
- btrfs_set_root_dirid(&root_item, new_dirid);
-
- key.objectid = objectid;
- key.offset = 1;
- btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
- ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
- &root_item);
- if (ret)
- goto fail;
-
- /*
- * insert the directory item
- */
- key.offset = (u64)-1;
- dir = root->fs_info->sb->s_root->d_inode;
- ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root,
- name, namelen, dir->i_ino, &key,
- BTRFS_FT_DIR);
- if (ret)
- goto fail;
-
- ret = btrfs_insert_inode_ref(trans, root->fs_info->tree_root,
- name, namelen, objectid,
- root->fs_info->sb->s_root->d_inode->i_ino);
- if (ret)
- goto fail;
-
- ret = btrfs_commit_transaction(trans, root);
- if (ret)
- goto fail_commit;
-
- new_root = btrfs_read_fs_root(root->fs_info, &key, name, namelen);
- BUG_ON(!new_root);
-
- trans = btrfs_start_transaction(new_root, 1);
- BUG_ON(!trans);
inode = btrfs_new_inode(trans, new_root, "..", 2, new_dirid,
- new_dirid,
- BTRFS_I(dir)->block_group, S_IFDIR | 0700);
+ new_dirid, block_group, S_IFDIR | 0700);
if (IS_ERR(inode))
- goto fail;
+ return PTR_ERR(inode);
inode->i_op = &btrfs_dir_inode_operations;
inode->i_fop = &btrfs_dir_file_operations;
new_root->inode = inode;
ret = btrfs_insert_inode_ref(trans, new_root, "..", 2, new_dirid,
new_dirid);
inode->i_nlink = 1;
- inode->i_size = 0;
- ret = btrfs_update_inode(trans, new_root, inode);
- if (ret)
- goto fail;
-fail:
- nr = trans->blocks_used;
- err = btrfs_commit_transaction(trans, new_root);
- if (err && !ret)
- ret = err;
-fail_commit:
- mutex_unlock(&root->fs_info->fs_mutex);
- btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
- return ret;
-}
+ btrfs_i_size_write(inode, 0);
-static int create_snapshot(struct btrfs_root *root, char *name, int namelen)
-{
- struct btrfs_pending_snapshot *pending_snapshot;
- struct btrfs_trans_handle *trans;
- int ret;
- int err;
- unsigned long nr = 0;
-
- if (!root->ref_cows)
- return -EINVAL;
-
- mutex_lock(&root->fs_info->fs_mutex);
- ret = btrfs_check_free_space(root, 1, 0);
- if (ret)
- goto fail_unlock;
-
- pending_snapshot = kmalloc(sizeof(*pending_snapshot), GFP_NOFS);
- if (!pending_snapshot) {
- ret = -ENOMEM;
- goto fail_unlock;
- }
- pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
- if (!pending_snapshot->name) {
- ret = -ENOMEM;
- kfree(pending_snapshot);
- goto fail_unlock;
- }
- memcpy(pending_snapshot->name, name, namelen);
- pending_snapshot->name[namelen] = '\0';
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
- pending_snapshot->root = root;
- list_add(&pending_snapshot->list,
- &trans->transaction->pending_snapshots);
- ret = btrfs_update_inode(trans, root, root->inode);
- err = btrfs_commit_transaction(trans, root);
-
-fail_unlock:
- mutex_unlock(&root->fs_info->fs_mutex);
- btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
- return ret;
+ return btrfs_update_inode(trans, new_root, inode);
}
unsigned long btrfs_force_ra(struct address_space *mapping,
#endif
}
-int btrfs_defrag_file(struct file *file) {
- struct inode *inode = fdentry(file)->d_inode;
- struct btrfs_root *root = BTRFS_I(inode)->root;
- struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
- struct page *page;
- unsigned long last_index;
- unsigned long ra_pages = root->fs_info->bdi.ra_pages;
- unsigned long total_read = 0;
- u64 page_start;
- u64 page_end;
- unsigned long i;
- int ret;
-
- mutex_lock(&root->fs_info->fs_mutex);
- ret = btrfs_check_free_space(root, inode->i_size, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
- if (ret)
- return -ENOSPC;
-
- mutex_lock(&inode->i_mutex);
- last_index = inode->i_size >> PAGE_CACHE_SHIFT;
- for (i = 0; i <= last_index; i++) {
- if (total_read % ra_pages == 0) {
- btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i,
- min(last_index, i + ra_pages - 1));
- }
- total_read++;
- page = grab_cache_page(inode->i_mapping, i);
- if (!page)
- goto out_unlock;
- if (!PageUptodate(page)) {
- btrfs_readpage(NULL, page);
- lock_page(page);
- if (!PageUptodate(page)) {
- unlock_page(page);
- page_cache_release(page);
- goto out_unlock;
- }
- }
- page_start = (u64)page->index << PAGE_CACHE_SHIFT;
- page_end = page_start + PAGE_CACHE_SIZE - 1;
-
- lock_extent(io_tree, page_start, page_end, GFP_NOFS);
- set_extent_delalloc(io_tree, page_start,
- page_end, GFP_NOFS);
-
- unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
- set_page_dirty(page);
- unlock_page(page);
- page_cache_release(page);
- balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
- }
-
-out_unlock:
- mutex_unlock(&inode->i_mutex);
- return 0;
-}
-
-static int btrfs_ioctl_resize(struct btrfs_root *root, void __user *arg)
-{
- u64 new_size;
- u64 old_size;
- u64 devid = 1;
- struct btrfs_ioctl_vol_args *vol_args;
- struct btrfs_trans_handle *trans;
- struct btrfs_device *device = NULL;
- char *sizestr;
- char *devstr = NULL;
- int ret = 0;
- int namelen;
- int mod = 0;
-
- vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
-
- if (!vol_args)
- return -ENOMEM;
-
- if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
- ret = -EFAULT;
- goto out;
- }
- namelen = strlen(vol_args->name);
- if (namelen > BTRFS_VOL_NAME_MAX) {
- ret = -EINVAL;
- goto out;
- }
-
- mutex_lock(&root->fs_info->fs_mutex);
- sizestr = vol_args->name;
- devstr = strchr(sizestr, ':');
- if (devstr) {
- char *end;
- sizestr = devstr + 1;
- *devstr = '\0';
- devstr = vol_args->name;
- devid = simple_strtoull(devstr, &end, 10);
-printk("resizing devid %Lu\n", devid);
- }
- device = btrfs_find_device(root, devid, NULL);
- if (!device) {
- printk("resizer unable to find device %Lu\n", devid);
- ret = -EINVAL;
- goto out_unlock;
- }
- if (!strcmp(sizestr, "max"))
- new_size = device->bdev->bd_inode->i_size;
- else {
- if (sizestr[0] == '-') {
- mod = -1;
- sizestr++;
- } else if (sizestr[0] == '+') {
- mod = 1;
- sizestr++;
- }
- new_size = btrfs_parse_size(sizestr);
- if (new_size == 0) {
- ret = -EINVAL;
- goto out_unlock;
- }
- }
-
- old_size = device->total_bytes;
-
- if (mod < 0) {
- if (new_size > old_size) {
- ret = -EINVAL;
- goto out_unlock;
- }
- new_size = old_size - new_size;
- } else if (mod > 0) {
- new_size = old_size + new_size;
- }
-
- if (new_size < 256 * 1024 * 1024) {
- ret = -EINVAL;
- goto out_unlock;
- }
- if (new_size > device->bdev->bd_inode->i_size) {
- ret = -EFBIG;
- goto out_unlock;
- }
-
- do_div(new_size, root->sectorsize);
- new_size *= root->sectorsize;
-
-printk("new size for %s is %llu\n", device->name, (unsigned long long)new_size);
-
- if (new_size > old_size) {
- trans = btrfs_start_transaction(root, 1);
- ret = btrfs_grow_device(trans, device, new_size);
- btrfs_commit_transaction(trans, root);
- } else {
- ret = btrfs_shrink_device(device, new_size);
- }
-
-out_unlock:
- mutex_unlock(&root->fs_info->fs_mutex);
-out:
- kfree(vol_args);
- return ret;
-}
-
-static int noinline btrfs_ioctl_snap_create(struct btrfs_root *root,
- void __user *arg)
-{
- struct btrfs_ioctl_vol_args *vol_args;
- struct btrfs_dir_item *di;
- struct btrfs_path *path;
- u64 root_dirid;
- int namelen;
- int ret;
-
- vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
-
- if (!vol_args)
- return -ENOMEM;
-
- if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
- ret = -EFAULT;
- goto out;
- }
-
- namelen = strlen(vol_args->name);
- if (namelen > BTRFS_VOL_NAME_MAX) {
- ret = -EINVAL;
- goto out;
- }
- if (strchr(vol_args->name, '/')) {
- ret = -EINVAL;
- goto out;
- }
-
- path = btrfs_alloc_path();
- if (!path) {
- ret = -ENOMEM;
- goto out;
- }
-
- root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
- mutex_lock(&root->fs_info->fs_mutex);
- di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
- path, root_dirid,
- vol_args->name, namelen, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
- btrfs_free_path(path);
-
- if (di && !IS_ERR(di)) {
- ret = -EEXIST;
- goto out;
- }
-
- if (IS_ERR(di)) {
- ret = PTR_ERR(di);
- goto out;
- }
-
- if (root == root->fs_info->tree_root)
- ret = create_subvol(root, vol_args->name, namelen);
- else
- ret = create_snapshot(root, vol_args->name, namelen);
-out:
- kfree(vol_args);
- return ret;
-}
-
-static int btrfs_ioctl_defrag(struct file *file)
-{
- struct inode *inode = fdentry(file)->d_inode;
- struct btrfs_root *root = BTRFS_I(inode)->root;
-
- switch (inode->i_mode & S_IFMT) {
- case S_IFDIR:
- mutex_lock(&root->fs_info->fs_mutex);
- btrfs_defrag_root(root, 0);
- btrfs_defrag_root(root->fs_info->extent_root, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
- break;
- case S_IFREG:
- btrfs_defrag_file(file);
- break;
- }
-
- return 0;
-}
-
-long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg)
-{
- struct btrfs_ioctl_vol_args *vol_args;
- int ret;
-
- vol_args = kmalloc(sizeof(*vol_args), GFP_NOFS);
-
- if (!vol_args)
- return -ENOMEM;
-
- if (copy_from_user(vol_args, arg, sizeof(*vol_args))) {
- ret = -EFAULT;
- goto out;
- }
- ret = btrfs_init_new_device(root, vol_args->name);
-
-out:
- kfree(vol_args);
- return ret;
-}
-
-long btrfs_ioctl(struct file *file, unsigned int
- cmd, unsigned long arg)
-{
- struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
-
- switch (cmd) {
- case BTRFS_IOC_SNAP_CREATE:
- return btrfs_ioctl_snap_create(root, (void __user *)arg);
- case BTRFS_IOC_DEFRAG:
- return btrfs_ioctl_defrag(file);
- case BTRFS_IOC_RESIZE:
- return btrfs_ioctl_resize(root, (void __user *)arg);
- case BTRFS_IOC_ADD_DEV:
- return btrfs_ioctl_add_dev(root, (void __user *)arg);
- }
-
- return -ENOTTY;
-}
-
-/*
- * Called inside transaction, so use GFP_NOFS
- */
struct inode *btrfs_alloc_inode(struct super_block *sb)
{
struct btrfs_inode *ei;
if (!ei)
return NULL;
ei->last_trans = 0;
- ei->ordered_trans = 0;
+ btrfs_ordered_inode_tree_init(&ei->ordered_tree);
return &ei->vfs_inode;
}
void btrfs_destroy_inode(struct inode *inode)
{
+ struct btrfs_ordered_extent *ordered;
WARN_ON(!list_empty(&inode->i_dentry));
WARN_ON(inode->i_data.nrpages);
+ while(1) {
+ ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1);
+ if (!ordered)
+ break;
+ else {
+ printk("found ordered extent %Lu %Lu\n",
+ ordered->file_offset, ordered->len);
+ btrfs_remove_ordered_extent(inode, ordered);
+ btrfs_put_ordered_extent(ordered);
+ btrfs_put_ordered_extent(ordered);
+ }
+ }
btrfs_drop_extent_cache(inode, 0, (u64)-1);
kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
}
struct inode *new_inode = new_dentry->d_inode;
struct inode *old_inode = old_dentry->d_inode;
struct timespec ctime = CURRENT_TIME;
- struct btrfs_path *path;
int ret;
if (S_ISDIR(old_inode->i_mode) && new_inode &&
return -ENOTEMPTY;
}
- mutex_lock(&root->fs_info->fs_mutex);
ret = btrfs_check_free_space(root, 1, 0);
if (ret)
goto out_unlock;
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, new_dir);
- path = btrfs_alloc_path();
- if (!path) {
- ret = -ENOMEM;
- goto out_fail;
- }
old_dentry->d_inode->i_nlink++;
old_dir->i_ctime = old_dir->i_mtime = ctime;
goto out_fail;
out_fail:
- btrfs_free_path(path);
btrfs_end_transaction(trans, root);
out_unlock:
- mutex_unlock(&root->fs_info->fs_mutex);
return ret;
}
if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root))
return -ENAMETOOLONG;
- mutex_lock(&root->fs_info->fs_mutex);
err = btrfs_check_free_space(root, 1, 0);
if (err)
goto out_fail;
inode->i_mapping, GFP_NOFS);
extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
inode->i_mapping, GFP_NOFS);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
BTRFS_I(inode)->delalloc_bytes = 0;
- atomic_set(&BTRFS_I(inode)->ordered_writeback, 0);
+ BTRFS_I(inode)->disk_i_size = 0;
BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
}
dir->i_sb->s_dirt = 1;
btrfs_update_inode_block_group(trans, inode);
inode->i_op = &btrfs_symlink_inode_operations;
inode->i_mapping->a_ops = &btrfs_symlink_aops;
inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
- inode->i_size = name_len - 1;
+ btrfs_i_size_write(inode, name_len - 1);
err = btrfs_update_inode(trans, root, inode);
if (err)
drop_inode = 1;
out_unlock:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
out_fail:
- mutex_unlock(&root->fs_info->fs_mutex);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
}
btrfs_btree_balance_dirty(root, nr);
- btrfs_throttle(root);
return err;
}
+static int btrfs_set_page_dirty(struct page *page)
+{
+ return __set_page_dirty_nobuffers(page);
+}
+
static int btrfs_permission(struct inode *inode, int mask,
struct nameidata *nd)
{
#ifdef CONFIG_COMPAT
.compat_ioctl = btrfs_ioctl,
#endif
+ .release = btrfs_release_file,
};
static struct extent_io_ops btrfs_extent_io_ops = {
.merge_bio_hook = btrfs_merge_bio_hook,
.readpage_io_hook = btrfs_readpage_io_hook,
.readpage_end_io_hook = btrfs_readpage_end_io_hook,
- .readpage_io_failed_hook = btrfs_readpage_io_failed_hook,
+ .writepage_end_io_hook = btrfs_writepage_end_io_hook,
+ .writepage_start_hook = btrfs_writepage_start_hook,
+ .readpage_io_failed_hook = btrfs_io_failed_hook,
.set_bit_hook = btrfs_set_bit_hook,
.clear_bit_hook = btrfs_clear_bit_hook,
};
.direct_IO = btrfs_direct_IO,
.invalidatepage = btrfs_invalidatepage,
.releasepage = btrfs_releasepage,
- .set_page_dirty = __set_page_dirty_nobuffers,
+ .set_page_dirty = btrfs_set_page_dirty,
};
static struct address_space_operations btrfs_symlink_aops = {
projects / linux-2.6-block.git / blobdiff