* Boston, MA 021110-1307, USA.
*/
+#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 "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"
+#include "volumes.h"
+#include "ordered-data.h"
struct btrfs_iget_args {
u64 ino;
static struct address_space_operations btrfs_aops;
static struct address_space_operations btrfs_symlink_aops;
static struct file_operations btrfs_dir_file_operations;
-static struct extent_map_ops btrfs_extent_map_ops;
+static struct extent_io_ops btrfs_extent_io_ops;
static struct kmem_cache *btrfs_inode_cachep;
struct kmem_cache *btrfs_trans_handle_cachep;
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(&root->fs_info->delalloc_lock);
if (used + root->fs_info->delalloc_bytes + num_required > thresh)
ret = -ENOSPC;
- spin_unlock(&root->fs_info->delalloc_lock);
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
return ret;
}
u64 num_bytes;
u64 cur_alloc_size;
u64 blocksize = root->sectorsize;
+ 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));
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+ btrfs_drop_extent_cache(inode, start, start + num_bytes - 1);
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
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->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;
}
- ret = btrfs_insert_file_extent(trans, root, inode->i_ino,
- start, ins.objectid, ins.offset,
+ 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;
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+ set_bit(EXTENT_FLAG_PINNED, &em->flags);
+ 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);
+ }
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
+ cur_alloc_size = ins.offset;
+ ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
ins.offset);
+ BUG_ON(ret);
+ if (num_bytes < cur_alloc_size) {
+ printk("num_bytes %Lu cur_alloc %Lu\n", num_bytes,
+ cur_alloc_size);
+ break;
+ }
num_bytes -= cur_alloc_size;
alloc_hint = ins.objectid + ins.offset;
start += cur_alloc_size;
}
- btrfs_add_ordered_inode(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;
}
static int run_delalloc_range(struct inode *inode, u64 start, u64 end)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
- u64 num_bytes;
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);
- spin_lock(&root->fs_info->delalloc_lock);
- num_bytes = end + 1 - start;
- if (root->fs_info->delalloc_bytes < num_bytes) {
- printk("delalloc accounting error total %llu sub %llu\n",
- root->fs_info->delalloc_bytes, num_bytes);
- } else {
- root->fs_info->delalloc_bytes -= num_bytes;
+ return ret;
+}
+
+int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
+ unsigned long old, unsigned long bits)
+{
+ unsigned long flags;
+ if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ BTRFS_I(inode)->delalloc_bytes += end - start + 1;
+ root->fs_info->delalloc_bytes += end - start + 1;
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
}
- spin_unlock(&root->fs_info->delalloc_lock);
+ return 0;
+}
- mutex_unlock(&root->fs_info->fs_mutex);
- return ret;
+int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end,
+ unsigned long old, unsigned long bits)
+{
+ if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ unsigned long flags;
+
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ if (end - start + 1 > root->fs_info->delalloc_bytes) {
+ printk("warning: delalloc account %Lu %Lu\n",
+ end - start + 1, root->fs_info->delalloc_bytes);
+ root->fs_info->delalloc_bytes = 0;
+ BTRFS_I(inode)->delalloc_bytes = 0;
+ } else {
+ root->fs_info->delalloc_bytes -= end - start + 1;
+ BTRFS_I(inode)->delalloc_bytes -= end - start + 1;
+ }
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
+ }
+ return 0;
+}
+
+int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
+ size_t size, struct bio *bio)
+{
+ struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
+ struct btrfs_mapping_tree *map_tree;
+ u64 logical = bio->bi_sector << 9;
+ u64 length = 0;
+ u64 map_length;
+ int ret;
+
+ length = bio->bi_size;
+ map_tree = &root->fs_info->mapping_tree;
+ map_length = length;
+ ret = btrfs_map_block(map_tree, READ, logical,
+ &map_length, NULL, 0);
+
+ if (map_length < length + size) {
+ return 1;
+ }
+ return 0;
+}
+
+int __btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
+ int mirror_num)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret = 0;
+
+ ret = btrfs_csum_one_bio(root, inode, bio);
+ BUG_ON(ret);
+
+ return btrfs_map_bio(root, rw, bio, mirror_num, 1);
+}
+
+int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
+ int mirror_num)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret = 0;
+
+ ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
+ BUG_ON(ret);
+
+ if (!(rw & (1 << BIO_RW))) {
+ goto mapit;
+ }
+
+ return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
+ inode, rw, bio, mirror_num,
+ __btrfs_submit_bio_hook);
+mapit:
+ 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;
+again:
+ 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);
+
+ /* already ordered? We're done */
+ if (test_range_bit(&BTRFS_I(inode)->io_tree, page_start, page_end,
+ EXTENT_ORDERED, 0)) {
+ goto out;
+ }
+
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(&BTRFS_I(inode)->io_tree, page_start,
+ page_end, GFP_NOFS);
+ unlock_page(page);
+ btrfs_start_ordered_extent(inode, ordered, 1);
+ goto again;
+ }
+
+ 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);
}
-int btrfs_writepage_io_hook(struct page *page, u64 start, u64 end)
+/*
+ * 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;
+}
+
+static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
+{
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_trans_handle *trans;
- char *kaddr;
- int ret = 0;
- u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
- size_t offset = start - page_start;
- if (btrfs_test_opt(root, NODATASUM) ||
- btrfs_test_flag(inode, NODATASUM))
+ struct btrfs_ordered_extent *ordered_extent;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_map *em;
+ u64 alloc_hint = 0;
+ u64 clear_start;
+ u64 clear_end;
+ 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;
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
- kaddr = kmap(page);
- btrfs_csum_file_block(trans, root, inode, inode->i_ino,
- start, kaddr + offset, end - start + 1);
- kunmap(page);
- ret = btrfs_end_transaction(trans, root);
+
+ 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);
- mutex_unlock(&root->fs_info->fs_mutex);
- return ret;
+
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+
+ 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);
+
+ spin_lock(&em_tree->lock);
+ clear_start = ordered_extent->file_offset;
+ clear_end = ordered_extent->file_offset + ordered_extent->len;
+ while(clear_start < clear_end) {
+ em = lookup_extent_mapping(em_tree, clear_start,
+ clear_end - clear_start);
+ if (em) {
+ clear_bit(EXTENT_FLAG_PINNED, &em->flags);
+ clear_start = em->start + em->len;
+ free_extent_map(em);
+ } else {
+ break;
+ }
+ }
+ spin_unlock(&em_tree->lock);
+
+ btrfs_drop_extent_cache(inode, ordered_extent->file_offset,
+ ordered_extent->file_offset +
+ ordered_extent->len - 1);
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
+ 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_writepage_end_io_hook(struct page *page, u64 start, u64 end,
+ struct extent_state *state, int uptodate)
+{
+ return btrfs_finish_ordered_io(page->mapping->host, start, end);
}
int btrfs_readpage_io_hook(struct page *page, u64 start, u64 end)
int ret = 0;
struct inode *inode = page->mapping->host;
struct btrfs_root *root = BTRFS_I(inode)->root;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
struct btrfs_csum_item *item;
struct btrfs_path *path = NULL;
u32 csum;
+
if (btrfs_test_opt(root, NODATASUM) ||
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);
goto out;
}
read_extent_buffer(path->nodes[0], &csum, (unsigned long)item,
BTRFS_CRC32_SIZE);
- set_state_private(em_tree, start, csum);
+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 btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end)
+struct io_failure_record {
+ struct page *page;
+ u64 start;
+ u64 len;
+ u64 logical;
+ int last_mirror;
+};
+
+int btrfs_io_failed_hook(struct bio *failed_bio,
+ struct page *page, u64 start, u64 end,
+ struct extent_state *state)
{
- size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT);
+ struct io_failure_record *failrec = NULL;
+ u64 private;
+ struct extent_map *em;
struct inode *inode = page->mapping->host;
+ struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- char *kaddr;
+ struct bio *bio;
+ int num_copies;
+ int ret;
+ int rw;
+ u64 logical;
+
+ ret = get_state_private(failure_tree, start, &private);
+ if (ret) {
+ failrec = kmalloc(sizeof(*failrec), GFP_NOFS);
+ if (!failrec)
+ return -ENOMEM;
+ failrec->start = start;
+ failrec->len = end - start + 1;
+ failrec->last_mirror = 0;
+
+ spin_lock(&em_tree->lock);
+ em = lookup_extent_mapping(em_tree, start, failrec->len);
+ if (em->start > start || em->start + em->len < start) {
+ free_extent_map(em);
+ em = NULL;
+ }
+ spin_unlock(&em_tree->lock);
+
+ if (!em || IS_ERR(em)) {
+ kfree(failrec);
+ return -EIO;
+ }
+ logical = start - em->start;
+ logical = em->block_start + logical;
+ failrec->logical = logical;
+ free_extent_map(em);
+ set_extent_bits(failure_tree, start, end, EXTENT_LOCKED |
+ EXTENT_DIRTY, GFP_NOFS);
+ set_state_private(failure_tree, start,
+ (u64)(unsigned long)failrec);
+ } else {
+ failrec = (struct io_failure_record *)(unsigned long)private;
+ }
+ num_copies = btrfs_num_copies(
+ &BTRFS_I(inode)->root->fs_info->mapping_tree,
+ failrec->logical, failrec->len);
+ failrec->last_mirror++;
+ if (!state) {
+ spin_lock_irq(&BTRFS_I(inode)->io_tree.lock);
+ state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
+ failrec->start,
+ EXTENT_LOCKED);
+ if (state && state->start != failrec->start)
+ state = NULL;
+ spin_unlock_irq(&BTRFS_I(inode)->io_tree.lock);
+ }
+ if (!state || failrec->last_mirror > num_copies) {
+ set_state_private(failure_tree, failrec->start, 0);
+ clear_extent_bits(failure_tree, failrec->start,
+ failrec->start + failrec->len - 1,
+ EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
+ kfree(failrec);
+ return -EIO;
+ }
+ bio = bio_alloc(GFP_NOFS, 1);
+ bio->bi_private = state;
+ bio->bi_end_io = failed_bio->bi_end_io;
+ bio->bi_sector = failrec->logical >> 9;
+ bio->bi_bdev = failed_bio->bi_bdev;
+ bio->bi_size = 0;
+ bio_add_page(bio, page, failrec->len, start - page_offset(page));
+ 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;
+}
+
+int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
+ struct extent_state *state)
+{
+ size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT);
+ struct inode *inode = page->mapping->host;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ char *kaddr;
+ u64 private = ~(u32)0;
int ret;
struct btrfs_root *root = BTRFS_I(inode)->root;
u32 csum = ~(u32)0;
unsigned long flags;
+
if (btrfs_test_opt(root, NODATASUM) ||
btrfs_test_flag(inode, NODATASUM))
return 0;
- ret = get_state_private(em_tree, start, &private);
+ if (state && state->start == start) {
+ private = state->private;
+ ret = 0;
+ } else {
+ ret = get_state_private(io_tree, start, &private);
+ }
local_irq_save(flags);
kaddr = kmap_atomic(page, KM_IRQ0);
if (ret) {
}
kunmap_atomic(kaddr, KM_IRQ0);
local_irq_restore(flags);
+
+ /* if the io failure tree for this inode is non-empty,
+ * check to see if we've recovered from a failed IO
+ */
+ btrfs_clean_io_failures(inode, start);
return 0;
zeroit:
- printk("btrfs csum failed ino %lu off %llu\n",
- page->mapping->host->i_ino, (unsigned long long)start);
+ printk("btrfs csum failed ino %lu off %llu csum %u private %Lu\n",
+ page->mapping->host->i_ino, (unsigned long long)start, csum,
+ private);
memset(kaddr + offset, 1, end - start + 1);
flush_dcache_page(page);
kunmap_atomic(kaddr, KM_IRQ0);
local_irq_restore(flags);
- return 0;
+ if (private == 0)
+ return 0;
+ return -EIO;
}
void btrfs_read_locked_inode(struct inode *inode)
struct btrfs_path *path;
struct extent_buffer *leaf;
struct btrfs_inode_item *inode_item;
- struct btrfs_inode_timespec *tspec;
+ struct btrfs_timespec *tspec;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_key location;
u64 alloc_group_block;
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_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
if (!BTRFS_I(inode)->block_group) {
BTRFS_I(inode)->block_group = btrfs_find_block_group(root,
- NULL, 0, 0, 0);
+ NULL, 0,
+ BTRFS_BLOCK_GROUP_METADATA, 0);
}
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;
- BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
+ BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
inode->i_fop = &btrfs_file_operations;
inode->i_op = &btrfs_file_inode_operations;
break;
case S_IFLNK:
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;
break;
default:
init_special_inode(inode, inode->i_mode, rdev);
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)
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;
- 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;
return err;
}
-static int btrfs_free_inode(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct inode *inode)
-{
- struct btrfs_path *path;
- int ret;
-
- clear_inode(inode);
-
- path = btrfs_alloc_path();
- BUG_ON(!path);
- ret = btrfs_lookup_inode(trans, root, path,
- &BTRFS_I(inode)->location, -1);
- if (ret > 0)
- ret = -ENOENT;
- if (!ret)
- ret = btrfs_del_item(trans, root, path);
- btrfs_free_path(path);
- return ret;
-}
-
/*
* this can truncate away extent items, csum items and directory items.
* It starts at a high offset and removes keys until it can't find
*/
static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
- struct inode *inode)
+ struct inode *inode,
+ u32 min_type)
{
int ret;
struct btrfs_path *path;
u64 root_owner = 0;
int found_extent;
int del_item;
+ int pending_del_nr = 0;
+ int pending_del_slot = 0;
int extent_type = -1;
+ u64 mask = root->sectorsize - 1;
- btrfs_drop_extent_cache(inode, inode->i_size, (u64)-1);
+ btrfs_drop_extent_cache(inode, inode->i_size & (~mask), (u64)-1);
path = btrfs_alloc_path();
path->reada = -1;
BUG_ON(!path);
key.offset = (u64)-1;
key.type = (u8)-1;
+ btrfs_init_path(path);
+search_again:
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret < 0) {
+ goto error;
+ }
+ if (ret > 0) {
+ BUG_ON(path->slots[0] == 0);
+ path->slots[0]--;
+ }
+
while(1) {
- btrfs_init_path(path);
fi = NULL;
- ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
- if (ret < 0) {
- goto error;
- }
- if (ret > 0) {
- BUG_ON(path->slots[0] == 0);
- path->slots[0]--;
- }
leaf = path->nodes[0];
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)
break;
- if (found_type != BTRFS_CSUM_ITEM_KEY &&
- found_type != BTRFS_DIR_ITEM_KEY &&
- found_type != BTRFS_DIR_INDEX_KEY &&
- found_type != BTRFS_EXTENT_DATA_KEY)
+ if (found_type < min_type)
break;
item_end = found_key.offset;
found_type = BTRFS_INODE_ITEM_KEY;
} else if (found_type == BTRFS_EXTENT_ITEM_KEY) {
found_type = BTRFS_CSUM_ITEM_KEY;
+ } else if (found_type == BTRFS_EXTENT_DATA_KEY) {
+ found_type = BTRFS_XATTR_ITEM_KEY;
+ } else if (found_type == BTRFS_XATTR_ITEM_KEY) {
+ found_type = BTRFS_INODE_REF_KEY;
} else if (found_type) {
found_type--;
} else {
break;
}
btrfs_set_key_type(&key, found_type);
- btrfs_release_path(root, path);
- continue;
+ goto next;
}
if (found_key.offset >= inode->i_size)
del_item = 1;
btrfs_file_extent_num_bytes(leaf, fi);
extent_num_bytes = inode->i_size -
found_key.offset + root->sectorsize - 1;
+ extent_num_bytes = extent_num_bytes &
+ ~((u64)root->sectorsize - 1);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_num_bytes);
num_dec = (orig_num_bytes -
- extent_num_bytes) >> 9;
- if (extent_start != 0) {
- inode->i_blocks -= num_dec;
- }
+ extent_num_bytes);
+ if (extent_start != 0)
+ dec_i_blocks(inode, num_dec);
btrfs_mark_buffer_dirty(leaf);
} else {
extent_num_bytes =
btrfs_file_extent_disk_num_bytes(leaf,
fi);
/* FIXME blocksize != 4096 */
- num_dec = btrfs_file_extent_num_bytes(leaf,
- fi) >> 9;
+ num_dec = btrfs_file_extent_num_bytes(leaf, fi);
if (extent_start != 0) {
found_extent = 1;
- inode->i_blocks -= num_dec;
+ dec_i_blocks(inode, num_dec);
}
root_gen = btrfs_header_generation(leaf);
root_owner = btrfs_header_owner(leaf);
}
- } else if (extent_type == BTRFS_FILE_EXTENT_INLINE &&
- !del_item) {
- u32 newsize = inode->i_size - found_key.offset;
- newsize = btrfs_file_extent_calc_inline_size(newsize);
- ret = btrfs_truncate_item(trans, root, path,
- newsize, 1);
- BUG_ON(ret);
+ } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+ if (!del_item) {
+ u32 newsize = inode->i_size - found_key.offset;
+ dec_i_blocks(inode, item_end + 1 -
+ found_key.offset - newsize);
+ newsize =
+ btrfs_file_extent_calc_inline_size(newsize);
+ ret = btrfs_truncate_item(trans, root, path,
+ newsize, 1);
+ BUG_ON(ret);
+ } else {
+ dec_i_blocks(inode, item_end + 1 -
+ found_key.offset);
+ }
}
delete:
if (del_item) {
- ret = btrfs_del_item(trans, root, path);
- if (ret)
- goto error;
+ if (!pending_del_nr) {
+ /* no pending yet, add ourselves */
+ pending_del_slot = path->slots[0];
+ pending_del_nr = 1;
+ } else if (pending_del_nr &&
+ path->slots[0] + 1 == pending_del_slot) {
+ /* hop on the pending chunk */
+ pending_del_nr++;
+ pending_del_slot = path->slots[0];
+ } else {
+ printk("bad pending slot %d pending_del_nr %d pending_del_slot %d\n", path->slots[0], pending_del_nr, pending_del_slot);
+ }
} else {
break;
}
- btrfs_release_path(root, path);
if (found_extent) {
ret = btrfs_free_extent(trans, root, extent_start,
extent_num_bytes,
found_key.offset, 0);
BUG_ON(ret);
}
+next:
+ if (path->slots[0] == 0) {
+ if (pending_del_nr)
+ goto del_pending;
+ btrfs_release_path(root, path);
+ goto search_again;
+ }
+
+ path->slots[0]--;
+ if (pending_del_nr &&
+ path->slots[0] + 1 != pending_del_slot) {
+ struct btrfs_key debug;
+del_pending:
+ btrfs_item_key_to_cpu(path->nodes[0], &debug,
+ pending_del_slot);
+ ret = btrfs_del_items(trans, root, path,
+ pending_del_slot,
+ pending_del_nr);
+ BUG_ON(ret);
+ pending_del_nr = 0;
+ btrfs_release_path(root, path);
+ goto search_again;
+ }
}
ret = 0;
error:
- btrfs_release_path(root, path);
+ if (pending_del_nr) {
+ ret = btrfs_del_items(trans, root, path, pending_del_slot,
+ pending_del_nr);
+ }
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)
+/*
+ * taken from block_truncate_page, but does cow as it zeros out
+ * any bytes left in the last page in the file.
+ */
+static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
{
- char *kaddr;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- struct btrfs_root *root = BTRFS_I(inode)->root;
- u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
- u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
- u64 existing_delalloc;
- u64 delalloc_start;
- int ret = 0;
-
- WARN_ON(!PageLocked(page));
- set_page_extent_mapped(page);
-
- lock_extent(em_tree, page_start, page_end, GFP_NOFS);
- delalloc_start = page_start;
- existing_delalloc = count_range_bits(&BTRFS_I(inode)->extent_tree,
- &delalloc_start, page_end,
- PAGE_CACHE_SIZE, EXTENT_DELALLOC);
- set_extent_delalloc(&BTRFS_I(inode)->extent_tree, page_start,
- page_end, GFP_NOFS);
-
- spin_lock(&root->fs_info->delalloc_lock);
- root->fs_info->delalloc_bytes += PAGE_CACHE_SIZE - existing_delalloc;
- spin_unlock(&root->fs_info->delalloc_lock);
-
- 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(em_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.
- */
-static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
-{
- struct inode *inode = mapping->host;
+ 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);
- ret = btrfs_cow_one_page(inode, page, offset);
+ 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_start_ordered_extent(inode, ordered, 1);
+ btrfs_put_ordered_extent(ordered);
+ goto again;
+ }
+
+ 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);
attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) {
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
u64 mask = root->sectorsize - 1;
- u64 pos = (inode->i_size + mask) & ~mask;
- u64 block_end = attr->ia_size | mask;
+ u64 hole_start = (inode->i_size + mask) & ~mask;
+ u64 block_end = (attr->ia_size + mask) & ~mask;
u64 hole_size;
u64 alloc_hint = 0;
- if (attr->ia_size <= pos)
+ 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(em_tree, pos, block_end, GFP_NOFS);
- hole_size = (attr->ia_size - pos + mask) & ~mask;
+ 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);
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
err = btrfs_drop_extents(trans, root, inode,
- pos, pos + hole_size, pos,
+ hole_start, block_end, hole_start,
&alloc_hint);
if (alloc_hint != EXTENT_MAP_INLINE) {
err = btrfs_insert_file_extent(trans, root,
inode->i_ino,
- pos, 0, 0, hole_size);
+ hole_start, 0, 0,
+ hole_size, 0);
+ btrfs_drop_extent_cache(inode, hole_start,
+ (u64)-1);
+ btrfs_check_file(root, inode);
}
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
- unlock_extent(em_tree, pos, block_end, GFP_NOFS);
+ unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
if (err)
return err;
}
return err;
}
-void btrfs_drop_inode(struct inode *inode)
-{
- if (!BTRFS_I(inode)->ordered_trans || inode->i_nlink) {
- generic_drop_inode(inode);
- return;
- }
- /* FIXME, make sure this delete actually ends up in the transaction */
- btrfs_del_ordered_inode(inode);
- generic_drop_inode(inode);
-}
-
void btrfs_delete_inode(struct inode *inode)
{
struct btrfs_trans_handle *trans;
if (is_bad_inode(inode)) {
goto no_delete;
}
+ btrfs_wait_ordered_range(inode, 0, (u64)-1);
- 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);
- ret = btrfs_truncate_in_trans(trans, root, inode);
- if (ret)
- goto no_delete_lock;
- ret = btrfs_delete_xattrs(trans, root, inode);
- if (ret)
- goto no_delete_lock;
- ret = btrfs_free_inode(trans, root, inode);
+ ret = btrfs_truncate_in_trans(trans, root, inode, 0);
if (ret)
goto no_delete_lock;
+
nr = trans->blocks_used;
+ 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;
}
struct btrfs_iget_args *args = p;
inode->i_ino = args->ino;
BTRFS_I(inode)->root = args->root;
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
+ 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);
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
+ mutex_init(&BTRFS_I(inode)->extent_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;
di = (struct btrfs_dir_item *)((char *)di + di_len);
}
}
- filp->f_pos++;
+ if (key_type == BTRFS_DIR_INDEX_KEY)
+ filp->f_pos = INT_LIMIT(typeof(filp->f_pos));
+ else
+ filp->f_pos++;
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,
struct btrfs_root *root,
+ const char *name, int name_len,
+ u64 ref_objectid,
u64 objectid,
struct btrfs_block_group_cache *group,
int mode)
{
struct inode *inode;
struct btrfs_inode_item *inode_item;
+ struct btrfs_block_group_cache *new_inode_group;
struct btrfs_key *location;
struct btrfs_path *path;
+ struct btrfs_inode_ref *ref;
+ struct btrfs_key key[2];
+ u32 sizes[2];
+ unsigned long ptr;
int ret;
int owner;
if (!inode)
return ERR_PTR(-ENOMEM);
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
+ 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);
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
+ mutex_init(&BTRFS_I(inode)->extent_mutex);
+ BTRFS_I(inode)->delalloc_bytes = 0;
+ BTRFS_I(inode)->disk_i_size = 0;
BTRFS_I(inode)->root = root;
if (mode & S_IFDIR)
owner = 0;
else
owner = 1;
- group = btrfs_find_block_group(root, group, 0, 0, owner);
- BTRFS_I(inode)->block_group = group;
+ new_inode_group = btrfs_find_block_group(root, group, 0,
+ BTRFS_BLOCK_GROUP_METADATA, owner);
+ if (!new_inode_group) {
+ printk("find_block group failed\n");
+ new_inode_group = group;
+ }
+ BTRFS_I(inode)->block_group = new_inode_group;
BTRFS_I(inode)->flags = 0;
- ret = btrfs_insert_empty_inode(trans, root, path, objectid);
- if (ret)
+
+ key[0].objectid = objectid;
+ btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
+ key[0].offset = 0;
+
+ key[1].objectid = objectid;
+ btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY);
+ key[1].offset = ref_objectid;
+
+ sizes[0] = sizeof(struct btrfs_inode_item);
+ sizes[1] = name_len + sizeof(*ref);
+
+ ret = btrfs_insert_empty_items(trans, root, path, key, sizes, 2);
+ if (ret != 0)
goto fail;
+ if (objectid > root->highest_inode)
+ root->highest_inode = objectid;
+
inode->i_uid = current->fsuid;
inode->i_gid = current->fsgid;
inode->i_mode = mode;
inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
struct btrfs_inode_item);
fill_inode_item(path->nodes[0], inode_item, inode);
+
+ ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
+ struct btrfs_inode_ref);
+ btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len);
+ ptr = (unsigned long)(ref + 1);
+ write_extent_buffer(path->nodes[0], name, ptr, name_len);
+
btrfs_mark_buffer_dirty(path->nodes[0]);
btrfs_free_path(path);
}
static int btrfs_add_link(struct btrfs_trans_handle *trans,
- struct dentry *dentry, struct inode *inode)
+ struct dentry *dentry, struct inode *inode,
+ int add_backref)
{
int ret;
struct btrfs_key key;
dentry->d_parent->d_inode->i_ino,
&key, btrfs_inode_type(inode));
if (ret == 0) {
- ret = btrfs_insert_inode_ref(trans, root,
- dentry->d_name.name,
- dentry->d_name.len,
- inode->i_ino,
- dentry->d_parent->d_inode->i_ino);
+ if (add_backref) {
+ ret = btrfs_insert_inode_ref(trans, root,
+ dentry->d_name.name,
+ dentry->d_name.len,
+ inode->i_ino,
+ 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);
}
static int btrfs_add_nondir(struct btrfs_trans_handle *trans,
- struct dentry *dentry, struct inode *inode)
+ struct dentry *dentry, struct inode *inode,
+ int backref)
{
- int err = btrfs_add_link(trans, dentry, inode);
+ int err = btrfs_add_link(trans, dentry, inode, backref);
if (!err) {
d_instantiate(dentry, inode);
return 0;
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;
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
+ inode = btrfs_new_inode(trans, root, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino, objectid,
BTRFS_I(dir)->block_group, mode);
err = PTR_ERR(inode);
if (IS_ERR(inode))
goto out_unlock;
btrfs_set_trans_block_group(trans, inode);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 0);
if (err)
drop_inode = 1;
else {
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;
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
- BTRFS_I(dir)->block_group, mode);
+ inode = btrfs_new_inode(trans, root, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino,
+ objectid, BTRFS_I(dir)->block_group, mode);
err = PTR_ERR(inode);
if (IS_ERR(inode))
goto out_unlock;
btrfs_set_trans_block_group(trans, inode);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 0);
if (err)
drop_inode = 1;
else {
inode->i_mapping->a_ops = &btrfs_aops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
inode->i_fop = &btrfs_file_operations;
inode->i_op = &btrfs_file_inode_operations;
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
+ 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);
- BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops;
+ extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
+ inode->i_mapping, GFP_NOFS);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
+ mutex_init(&BTRFS_I(inode)->extent_mutex);
+ BTRFS_I(inode)->delalloc_bytes = 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;
btrfs_set_trans_block_group(trans, dir);
atomic_inc(&inode->i_count);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 1);
if (err)
drop_inode = 1;
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;
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
+ inode = btrfs_new_inode(trans, root, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino, objectid,
BTRFS_I(dir)->block_group, S_IFDIR | mode);
if (IS_ERR(inode)) {
err = PTR_ERR(inode);
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;
- err = btrfs_add_link(trans, dentry, inode);
+ err = btrfs_add_link(trans, dentry, inode, 0);
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,
+ u64 map_start, u64 map_len)
+{
+ u64 start_diff;
+
+ 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 page_offset, u64 start, u64 end,
+ size_t pg_offset, u64 start, u64 len,
int create)
{
int ret;
u64 extent_end = 0;
u64 objectid = inode->i_ino;
u32 found_type;
- int failed_insert = 0;
struct btrfs_path *path;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_file_extent_item *item;
struct btrfs_key found_key;
struct extent_map *em = NULL;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
struct btrfs_trans_handle *trans = NULL;
path = btrfs_alloc_path();
BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
again:
- em = lookup_extent_mapping(em_tree, start, end);
+ 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) {
- if (em->start > start) {
- printk("get_extent start %Lu em start %Lu\n",
- start, em->start);
- WARN_ON(1);
- }
- goto out;
+ if (em->start > start || em->start + em->len <= start)
+ free_extent_map(em);
+ else if (em->block_start == EXTENT_MAP_INLINE && page)
+ free_extent_map(em);
+ else
+ goto out;
}
+ em = alloc_extent_map(GFP_NOFS);
if (!em) {
- em = alloc_extent_map(GFP_NOFS);
- if (!em) {
- err = -ENOMEM;
- goto out;
- }
- em->start = EXTENT_MAP_HOLE;
- em->end = EXTENT_MAP_HOLE;
+ err = -ENOMEM;
+ goto out;
}
- em->bdev = inode->i_sb->s_bdev;
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
+ em->start = EXTENT_MAP_HOLE;
+ em->len = (u64)-1;
ret = btrfs_lookup_file_extent(trans, root, path,
objectid, start, trans != NULL);
if (ret < 0) {
if (start < extent_start || start >= extent_end) {
em->start = start;
if (start < extent_start) {
- if (end < extent_start)
+ if (start + len <= extent_start)
goto not_found;
- em->end = extent_end - 1;
+ em->len = extent_end - extent_start;
} else {
- em->end = end;
+ em->len = len;
}
goto not_found_em;
}
bytenr = btrfs_file_extent_disk_bytenr(leaf, item);
if (bytenr == 0) {
em->start = extent_start;
- em->end = extent_end - 1;
+ em->len = extent_end - extent_start;
em->block_start = EXTENT_MAP_HOLE;
- em->block_end = EXTENT_MAP_HOLE;
goto insert;
}
bytenr += btrfs_file_extent_offset(leaf, item);
em->block_start = bytenr;
- em->block_end = em->block_start +
- btrfs_file_extent_num_bytes(leaf, item) - 1;
em->start = extent_start;
- em->end = extent_end - 1;
+ em->len = extent_end - extent_start;
goto insert;
} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
+ u64 page_start;
unsigned long ptr;
char *map;
size_t size;
size = btrfs_file_extent_inline_len(leaf, btrfs_item_nr(leaf,
path->slots[0]));
- extent_end = (extent_start + size - 1) |
- ((u64)root->sectorsize - 1);
+ extent_end = (extent_start + size + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
if (start < extent_start || start >= extent_end) {
em->start = start;
if (start < extent_start) {
- if (end < extent_start)
+ if (start + len <= extent_start)
goto not_found;
- em->end = extent_end;
+ em->len = extent_end - extent_start;
} else {
- em->end = end;
+ em->len = len;
}
goto not_found_em;
}
em->block_start = EXTENT_MAP_INLINE;
- em->block_end = EXTENT_MAP_INLINE;
if (!page) {
em->start = extent_start;
- em->end = extent_start + size - 1;
+ em->len = size;
goto out;
}
- extent_offset = ((u64)page->index << PAGE_CACHE_SHIFT) -
- extent_start + page_offset;
- copy_size = min_t(u64, PAGE_CACHE_SIZE - page_offset,
+ page_start = page_offset(page) + pg_offset;
+ extent_offset = page_start - extent_start;
+ copy_size = min_t(u64, PAGE_CACHE_SIZE - pg_offset,
size - extent_offset);
em->start = extent_start + extent_offset;
- em->end = (em->start + copy_size -1) |
- ((u64)root->sectorsize -1);
+ em->len = (copy_size + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
map = kmap(page);
ptr = btrfs_file_extent_inline_start(item) + extent_offset;
if (create == 0 && !PageUptodate(page)) {
- read_extent_buffer(leaf, map + page_offset, ptr,
+ read_extent_buffer(leaf, map + pg_offset, ptr,
copy_size);
flush_dcache_page(page);
} else if (create && PageUptodate(page)) {
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 + page_offset, ptr,
+ write_extent_buffer(leaf, map + pg_offset, ptr,
copy_size);
btrfs_mark_buffer_dirty(leaf);
}
kunmap(page);
- set_extent_uptodate(em_tree, em->start, em->end, GFP_NOFS);
+ set_extent_uptodate(io_tree, em->start,
+ extent_map_end(em) - 1, GFP_NOFS);
goto insert;
} else {
printk("unkknown found_type %d\n", found_type);
}
not_found:
em->start = start;
- em->end = end;
+ em->len = len;
not_found_em:
em->block_start = EXTENT_MAP_HOLE;
- em->block_end = EXTENT_MAP_HOLE;
insert:
btrfs_release_path(root, path);
- if (em->start > start || em->end < start) {
- printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->end, start, end);
+ if (em->start > start || extent_map_end(em) <= start) {
+ printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->len, start, len);
err = -EIO;
goto out;
}
+
+ err = 0;
+ spin_lock(&em_tree->lock);
ret = add_extent_mapping(em_tree, em);
+ /* it is possible that someone inserted the extent into the tree
+ * while we had the lock dropped. It is also possible that
+ * an overlapping map exists in the tree
+ */
if (ret == -EEXIST) {
- free_extent_map(em);
- em = NULL;
- if (0 && failed_insert == 1) {
- btrfs_drop_extent_cache(inode, start, end);
+ struct extent_map *existing;
+
+ ret = 0;
+
+ existing = lookup_extent_mapping(em_tree, start, len);
+ if (existing && (existing->start > start ||
+ existing->start + existing->len <= start)) {
+ free_extent_map(existing);
+ existing = NULL;
}
- failed_insert++;
- if (failed_insert > 5) {
- printk("failing to insert %Lu %Lu\n", start, end);
- err = -EIO;
- goto out;
+ if (!existing) {
+ existing = lookup_extent_mapping(em_tree, em->start,
+ em->len);
+ if (existing) {
+ err = merge_extent_mapping(em_tree, existing,
+ em, start,
+ root->sectorsize);
+ free_extent_map(existing);
+ if (err) {
+ free_extent_map(em);
+ em = NULL;
+ }
+ } else {
+ err = -EIO;
+ printk("failing to insert %Lu %Lu\n",
+ start, len);
+ free_extent_map(em);
+ em = NULL;
+ }
+ } else {
+ free_extent_map(em);
+ em = existing;
+ err = 0;
}
- goto again;
}
- err = 0;
+ spin_unlock(&em_tree->lock);
out:
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 em;
}
+#if 0 /* waiting for O_DIRECT reads */
+static int btrfs_get_block(struct inode *inode, sector_t iblock,
+ struct buffer_head *bh_result, int create)
+{
+ struct extent_map *em;
+ u64 start = (u64)iblock << inode->i_blkbits;
+ struct btrfs_multi_bio *multi = NULL;
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ u64 len;
+ u64 logical;
+ u64 map_length;
+ int ret = 0;
+
+ em = btrfs_get_extent(inode, NULL, 0, start, bh_result->b_size, 0);
+
+ if (!em || IS_ERR(em))
+ goto out;
+
+ if (em->start > start || em->start + em->len <= start) {
+ goto out;
+ }
+
+ if (em->block_start == EXTENT_MAP_INLINE) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ len = em->start + em->len - start;
+ len = min_t(u64, len, INT_LIMIT(typeof(bh_result->b_size)));
+
+ if (em->block_start == EXTENT_MAP_HOLE ||
+ em->block_start == EXTENT_MAP_DELALLOC) {
+ bh_result->b_size = len;
+ goto out;
+ }
+
+ logical = start - em->start;
+ logical = em->block_start + logical;
+
+ map_length = len;
+ ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
+ logical, &map_length, &multi, 0);
+ BUG_ON(ret);
+ bh_result->b_blocknr = multi->stripes[0].physical >> inode->i_blkbits;
+ bh_result->b_size = min(map_length, len);
+
+ bh_result->b_bdev = multi->stripes[0].dev->bdev;
+ set_buffer_mapped(bh_result);
+ kfree(multi);
+out:
+ free_extent_map(em);
+ return ret;
+}
+#endif
+
+static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
+ const struct iovec *iov, loff_t offset,
+ unsigned long nr_segs)
+{
+ return -EINVAL;
+#if 0
+ struct file *file = iocb->ki_filp;
+ struct inode *inode = file->f_mapping->host;
+
+ if (rw == WRITE)
+ return -EINVAL;
+
+ return blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
+ offset, nr_segs, btrfs_get_block, NULL);
+#endif
+}
+
static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock)
{
return extent_bmap(mapping, iblock, btrfs_get_extent);
int btrfs_readpage(struct file *file, struct page *page)
{
- struct extent_map_tree *tree;
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
+ struct extent_io_tree *tree;
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
return extent_read_full_page(tree, page, btrfs_get_extent);
}
static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
{
- struct extent_map_tree *tree;
+ struct extent_io_tree *tree;
if (current->flags & PF_MEMALLOC) {
unlock_page(page);
return 0;
}
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
return extent_write_full_page(tree, page, btrfs_get_extent, wbc);
}
static int btrfs_writepages(struct address_space *mapping,
struct writeback_control *wbc)
{
- struct extent_map_tree *tree;
- tree = &BTRFS_I(mapping->host)->extent_tree;
+ struct extent_io_tree *tree;
+ tree = &BTRFS_I(mapping->host)->io_tree;
return extent_writepages(tree, mapping, btrfs_get_extent, wbc);
}
btrfs_readpages(struct file *file, struct address_space *mapping,
struct list_head *pages, unsigned nr_pages)
{
- struct extent_map_tree *tree;
- tree = &BTRFS_I(mapping->host)->extent_tree;
+ struct extent_io_tree *tree;
+ tree = &BTRFS_I(mapping->host)->io_tree;
return extent_readpages(tree, mapping, pages, nr_pages,
btrfs_get_extent);
}
-
-static int btrfs_releasepage(struct page *page, gfp_t unused_gfp_flags)
+static int __btrfs_releasepage(struct page *page, gfp_t gfp_flags)
{
- struct extent_map_tree *tree;
+ struct extent_io_tree *tree;
+ struct extent_map_tree *map;
int ret;
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
- ret = try_release_extent_mapping(tree, page);
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
+ map = &BTRFS_I(page->mapping->host)->extent_tree;
+ ret = try_release_extent_mapping(map, tree, page, gfp_flags);
if (ret == 1) {
ClearPagePrivate(page);
set_page_private(page, 0);
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_map_tree *tree;
+ 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;
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
- extent_invalidatepage(tree, page, offset);
- btrfs_releasepage(page, GFP_NOFS);
+ wait_on_page_writeback(page);
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
+ 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) {
+ /*
+ * IO on this page will never be started, so we need
+ * to account for any ordered extents now
+ */
+ clear_extent_bit(tree, page_start, page_end,
+ EXTENT_DIRTY | EXTENT_DELALLOC |
+ EXTENT_LOCKED, 1, 0, GFP_NOFS);
+ btrfs_finish_ordered_io(page->mapping->host,
+ page_start, page_end);
+ 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);
+
+ ClearPageChecked(page);
+ if (PagePrivate(page)) {
+ 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);
+
+ /*
+ * we can't set the delalloc bits if there are pending ordered
+ * extents. Drop our locks and wait for them to finish
+ */
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ unlock_page(page);
+ btrfs_start_ordered_extent(inode, ordered, 1);
+ 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;
return;
btrfs_truncate_page(inode->i_mapping, inode->i_size);
+ btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
- mutex_lock(&root->fs_info->fs_mutex);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
+ 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);
+ ret = btrfs_truncate_in_trans(trans, root, inode,
+ BTRFS_EXTENT_DATA_KEY);
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, new_dirid,
- BTRFS_I(dir)->block_group, S_IFDIR | 0700);
+ inode = btrfs_new_inode(trans, new_root, "..", 2, new_dirid,
+ 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;
-}
-
-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 = kstrndup(name, namelen, GFP_NOFS);
- if (!pending_snapshot->name) {
- ret = -ENOMEM;
- kfree(pending_snapshot);
- goto fail_unlock;
- }
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
+ btrfs_i_size_write(inode, 0);
- 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,
struct file_ra_state *ra, struct file *file,
pgoff_t offset, pgoff_t last_index)
{
- pgoff_t req_size;
+ pgoff_t req_size = last_index - offset + 1;
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
- req_size = last_index - offset + 1;
offset = page_cache_readahead(mapping, ra, file, offset, req_size);
return offset;
#else
- req_size = min(last_index - offset + 1, (pgoff_t)128);
page_cache_sync_readahead(mapping, ra, file, offset, req_size);
return offset + req_size;
#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_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- struct page *page;
- unsigned long last_index;
- unsigned long ra_index = 0;
- u64 page_start;
- u64 page_end;
- u64 delalloc_start;
- u64 existing_delalloc;
- 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 (i == ra_index) {
- ra_index = btrfs_force_ra(inode->i_mapping,
- &file->f_ra,
- file, ra_index, last_index);
- }
- 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(em_tree, page_start, page_end, GFP_NOFS);
- delalloc_start = page_start;
- existing_delalloc =
- count_range_bits(&BTRFS_I(inode)->extent_tree,
- &delalloc_start, page_end,
- PAGE_CACHE_SIZE, EXTENT_DELALLOC);
- set_extent_delalloc(em_tree, page_start,
- page_end, GFP_NOFS);
-
- spin_lock(&root->fs_info->delalloc_lock);
- root->fs_info->delalloc_bytes += PAGE_CACHE_SIZE -
- existing_delalloc;
- spin_unlock(&root->fs_info->delalloc_lock);
-
- unlock_extent(em_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;
- struct btrfs_ioctl_vol_args *vol_args;
- struct btrfs_trans_handle *trans;
- char *sizestr;
- 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;
- }
-
- sizestr = vol_args->name;
- if (!strcmp(sizestr, "max"))
- new_size = root->fs_info->sb->s_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;
- }
- }
-
- mutex_lock(&root->fs_info->fs_mutex);
- old_size = btrfs_super_total_bytes(&root->fs_info->super_copy);
-
- 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 > root->fs_info->sb->s_bdev->bd_inode->i_size) {
- ret = -EFBIG;
- goto out_unlock;
- }
-
- do_div(new_size, root->sectorsize);
- new_size *= root->sectorsize;
-
-printk("new size is %Lu\n", new_size);
- if (new_size > old_size) {
- trans = btrfs_start_transaction(root, 1);
- ret = btrfs_grow_extent_tree(trans, root, new_size);
- btrfs_commit_transaction(trans, root);
- } else {
- ret = btrfs_shrink_extent_tree(root, 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(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);
- }
-
- 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 *inode = dentry->d_inode;
generic_fillattr(inode, stat);
stat->blksize = PAGE_CACHE_SIZE;
+ stat->blocks = inode->i_blocks + (BTRFS_I(inode)->delalloc_bytes >> 9);
return 0;
}
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;
if (ret)
goto out_fail;
}
- ret = btrfs_add_link(trans, new_dentry, old_inode);
+ ret = btrfs_add_link(trans, new_dentry, old_inode, 1);
if (ret)
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;
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
+ inode = btrfs_new_inode(trans, root, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino, objectid,
BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO);
err = PTR_ERR(inode);
if (IS_ERR(inode))
goto out_unlock;
btrfs_set_trans_block_group(trans, inode);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 0);
if (err)
drop_inode = 1;
else {
inode->i_mapping->a_ops = &btrfs_aops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
inode->i_fop = &btrfs_file_operations;
inode->i_op = &btrfs_file_inode_operations;
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
+ 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);
- BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops;
+ extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
+ inode->i_mapping, GFP_NOFS);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
+ mutex_init(&BTRFS_I(inode)->extent_mutex);
+ BTRFS_I(inode)->delalloc_bytes = 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_size = name_len - 1;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
+ 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_map_ops btrfs_extent_map_ops = {
+static struct extent_io_ops btrfs_extent_io_ops = {
.fill_delalloc = run_delalloc_range,
- .writepage_io_hook = btrfs_writepage_io_hook,
+ .submit_bio_hook = btrfs_submit_bio_hook,
+ .merge_bio_hook = btrfs_merge_bio_hook,
.readpage_io_hook = btrfs_readpage_io_hook,
.readpage_end_io_hook = btrfs_readpage_end_io_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,
};
static struct address_space_operations btrfs_aops = {
.readpages = btrfs_readpages,
.sync_page = block_sync_page,
.bmap = btrfs_bmap,
+ .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 = {
git.samba.org / sfrench / cifs-2.6.git / blobdiff