diff options
Diffstat (limited to 'fs/btrfs/inode.c')
| -rw-r--r-- | fs/btrfs/inode.c | 3908 |
1 files changed, 3908 insertions, 0 deletions
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c new file mode 100644 index 00000000000..bf4bed6ca4d --- /dev/null +++ b/fs/btrfs/inode.c @@ -0,0 +1,3908 @@ +/* + * Copyright (C) 2007 Oracle. All rights reserved. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public + * License v2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public + * License along with this program; if not, write to the + * Free Software Foundation, Inc., 59 Temple Place - Suite 330, + * 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 <linux/time.h> +#include <linux/init.h> +#include <linux/string.h> +#include <linux/smp_lock.h> +#include <linux/backing-dev.h> +#include <linux/mpage.h> +#include <linux/swap.h> +#include <linux/writeback.h> +#include <linux/statfs.h> +#include <linux/compat.h> +#include <linux/bit_spinlock.h> +#include <linux/version.h> +#include <linux/xattr.h> +#include <linux/posix_acl.h> +#include "ctree.h" +#include "disk-io.h" +#include "transaction.h" +#include "btrfs_inode.h" +#include "ioctl.h" +#include "print-tree.h" +#include "volumes.h" +#include "ordered-data.h" +#include "xattr.h" +#include "compat.h" +#include "tree-log.h" +#include "ref-cache.h" + +struct btrfs_iget_args { + u64 ino; + struct btrfs_root *root; +}; + +static struct inode_operations btrfs_dir_inode_operations; +static struct inode_operations btrfs_symlink_inode_operations; +static struct inode_operations btrfs_dir_ro_inode_operations; +static struct inode_operations btrfs_special_inode_operations; +static struct inode_operations btrfs_file_inode_operations; +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_io_ops btrfs_extent_io_ops; + +static struct kmem_cache *btrfs_inode_cachep; +struct kmem_cache *btrfs_trans_handle_cachep; +struct kmem_cache *btrfs_transaction_cachep; +struct kmem_cache *btrfs_bit_radix_cachep; +struct kmem_cache *btrfs_path_cachep; + +#define S_SHIFT 12 +static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = { + [S_IFREG >> S_SHIFT] = BTRFS_FT_REG_FILE, + [S_IFDIR >> S_SHIFT] = BTRFS_FT_DIR, + [S_IFCHR >> S_SHIFT] = BTRFS_FT_CHRDEV, + [S_IFBLK >> S_SHIFT] = BTRFS_FT_BLKDEV, + [S_IFIFO >> S_SHIFT] = BTRFS_FT_FIFO, + [S_IFSOCK >> S_SHIFT] = BTRFS_FT_SOCK, + [S_IFLNK >> S_SHIFT] = BTRFS_FT_SYMLINK, +}; + +static void btrfs_truncate(struct inode *inode); + +/* + * a very lame attempt at stopping writes when the FS is 85% full. There + * are countless ways this is incorrect, but it is better than nothing. + */ +int btrfs_check_free_space(struct btrfs_root *root, u64 num_required, + int for_del) +{ + 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 + thresh = total * 85; + + do_div(thresh, 100); + + if (used + root->fs_info->delalloc_bytes + num_required > thresh) + ret = -ENOSPC; + spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); + return ret; +} + +/* + * when extent_io.c finds a delayed allocation range in the file, + * the call backs end up in this code. The basic idea is to + * allocate extents on disk for the range, and create ordered data structs + * in ram to track those extents. + */ +static int cow_file_range(struct inode *inode, u64 start, u64 end) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_trans_handle *trans; + u64 alloc_hint = 0; + u64 num_bytes; + u64 cur_alloc_size; + u64 blocksize = root->sectorsize; + u64 orig_num_bytes; + struct btrfs_key ins; + struct extent_map *em; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + int ret = 0; + + 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); + 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, 0); + mutex_unlock(&BTRFS_I(inode)->extent_mutex); + + while(num_bytes > 0) { + cur_alloc_size = min(num_bytes, root->fs_info->max_extent); + ret = btrfs_reserve_extent(trans, root, cur_alloc_size, + root->sectorsize, 0, alloc_hint, + (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; + 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, 0); + } + mutex_unlock(&BTRFS_I(inode)->extent_mutex); + + cur_alloc_size = ins.offset; + ret = btrfs_add_ordered_extent(inode, start, ins.objectid, + ins.offset, 0); + 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; + } +out: + btrfs_end_transaction(trans, root); + return ret; +} + +/* + * when nowcow writeback call back. This checks for snapshots or COW copies + * of the extents that exist in the file, and COWs the file as required. + * + * If no cow copies or snapshots exist, we write directly to the existing + * blocks on disk + */ +static int run_delalloc_nocow(struct inode *inode, u64 start, u64 end) +{ + u64 extent_start; + u64 extent_end; + u64 bytenr; + u64 loops = 0; + u64 total_fs_bytes; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_block_group_cache *block_group; + struct btrfs_trans_handle *trans; + struct extent_buffer *leaf; + int found_type; + struct btrfs_path *path; + struct btrfs_file_extent_item *item; + int ret; + int err = 0; + struct btrfs_key found_key; + + total_fs_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy); + path = btrfs_alloc_path(); + BUG_ON(!path); + trans = btrfs_join_transaction(root, 1); + BUG_ON(!trans); +again: + ret = btrfs_lookup_file_extent(NULL, root, path, + inode->i_ino, start, 0); + if (ret < 0) { + err = ret; + goto out; + } + + if (ret != 0) { + if (path->slots[0] == 0) + goto not_found; + path->slots[0]--; + } + + leaf = path->nodes[0]; + item = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + + /* are we inside the extent that was found? */ + 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) + goto not_found; + + found_type = btrfs_file_extent_type(leaf, item); + extent_start = found_key.offset; + if (found_type == BTRFS_FILE_EXTENT_REG) { + u64 extent_num_bytes; + + extent_num_bytes = btrfs_file_extent_num_bytes(leaf, item); + extent_end = extent_start + extent_num_bytes; + err = 0; + + if (loops && start != extent_start) + goto not_found; + + if (start < extent_start || start >= extent_end) + goto not_found; + + bytenr = btrfs_file_extent_disk_bytenr(leaf, item); + if (bytenr == 0) + goto not_found; + + if (btrfs_cross_ref_exists(trans, root, &found_key, bytenr)) + goto not_found; + /* + * we may be called by the resizer, make sure we're inside + * the limits of the FS + */ + block_group = btrfs_lookup_block_group(root->fs_info, + bytenr); + if (!block_group || block_group->ro) + goto not_found; + + bytenr += btrfs_file_extent_offset(leaf, item); + extent_num_bytes = min(end + 1, extent_end) - start; + ret = btrfs_add_ordered_extent(inode, start, bytenr, + extent_num_bytes, 1); + if (ret) { + err = ret; + goto out; + } + + btrfs_release_path(root, path); + start = extent_end; + if (start <= end) { + loops++; + goto again; + } + } else { +not_found: + btrfs_end_transaction(trans, root); + btrfs_free_path(path); + return cow_file_range(inode, start, end); + } +out: + WARN_ON(err); + btrfs_end_transaction(trans, root); + btrfs_free_path(path); + return err; +} + +/* + * extent_io.c call back to do delayed allocation processing + */ +static int run_delalloc_range(struct inode *inode, u64 start, u64 end) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret; + + 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); + + return ret; +} + +/* + * extent_io.c set_bit_hook, used to track delayed allocation + * bytes in this file, and to maintain the list of inodes that + * have pending delalloc work to be done. + */ +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; + if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) { + list_add_tail(&BTRFS_I(inode)->delalloc_inodes, + &root->fs_info->delalloc_inodes); + } + spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); + } + return 0; +} + +/* + * extent_io.c clear_bit_hook, see set_bit_hook for why + */ +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; + } + if (BTRFS_I(inode)->delalloc_bytes == 0 && + !list_empty(&BTRFS_I(inode)->delalloc_inodes)) { + list_del_init(&BTRFS_I(inode)->delalloc_inodes); + } + spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags); + } + return 0; +} + +/* + * extent_io.c merge_bio_hook, this must check the chunk tree to make sure + * we don't create bios that span stripes or chunks + */ +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 = (u64)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; +} + +/* + * in order to insert checksums into the metadata in large chunks, + * we wait until bio submission time. All the pages in the bio are + * checksummed and sums are attached onto the ordered extent record. + * + * At IO completion time the cums attached on the ordered extent record + * are inserted into the btree + */ +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); +} + +/* + * extent_io.c submission hook. This does the right thing for csum calculation on write, + * or reading the csums from the tree before a read + */ +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 (btrfs_test_opt(root, NODATASUM) || + btrfs_test_flag(inode, NODATASUM)) { + goto mapit; + } + + if (!(rw & (1 << BIO_RW))) { + btrfs_lookup_bio_sums(root, inode, bio); + 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); +} + +/* + * given a list of ordered sums record them in the inode. This happens + * at IO completion time based on sums calculated at bio submission time. + */ +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); + btrfs_csum_file_blocks(trans, BTRFS_I(inode)->root, + inode, sum); + } + return 0; +} + +int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end) +{ + return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end, + GFP_NOFS); +} + +/* see btrfs_writepage_start_hook for details on why this is required */ +struct btrfs_writepage_fixup { + struct page *page; + struct btrfs_work work; +}; + +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; + } + + btrfs_set_extent_delalloc(inode, page_start, page_end); + 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; + + 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; +} + +/* as ordered data IO finishes, this gets called so we can finish + * an ordered extent if the range of bytes in the file it covers are + * fully written. + */ +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; + struct btrfs_ordered_extent *ordered_extent; + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct btrfs_file_extent_item *extent_item; + struct btrfs_path *path = NULL; + struct extent_buffer *leaf; + 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); + if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) + goto nocow; + + path = btrfs_alloc_path(); + BUG_ON(!path); + + lock_extent(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + ordered_extent->len - 1, + GFP_NOFS); + + INIT_LIST_HEAD(&list); + + 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); + + ins.objectid = inode->i_ino; + ins.offset = ordered_extent->file_offset; + ins.type = BTRFS_EXTENT_DATA_KEY; + ret = btrfs_insert_empty_item(trans, root, path, &ins, + sizeof(*extent_item)); + BUG_ON(ret); + leaf = path->nodes[0]; + extent_item = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + btrfs_set_file_extent_generation(leaf, extent_item, trans->transid); + btrfs_set_file_extent_type(leaf, extent_item, BTRFS_FILE_EXTENT_REG); + btrfs_set_file_extent_disk_bytenr(leaf, extent_item, + ordered_extent->start); + btrfs_set_file_extent_disk_num_bytes(leaf, extent_item, + ordered_extent->len); + btrfs_set_file_extent_offset(leaf, extent_item, 0); + btrfs_set_file_extent_num_bytes(leaf, extent_item, + ordered_extent->len); + btrfs_mark_buffer_dirty(leaf); + + btrfs_drop_extent_cache(inode, ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len - 1, 0); + mutex_unlock(&BTRFS_I(inode)->extent_mutex); + + ins.objectid = ordered_extent->start; + ins.offset = ordered_extent->len; + ins.type = BTRFS_EXTENT_ITEM_KEY; + ret = btrfs_alloc_reserved_extent(trans, root, leaf->start, + root->root_key.objectid, + trans->transid, inode->i_ino, &ins); + BUG_ON(ret); + btrfs_release_path(root, path); + + inode_add_bytes(inode, ordered_extent->len); + unlock_extent(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + ordered_extent->len - 1, + GFP_NOFS); +nocow: + add_pending_csums(trans, inode, ordered_extent->file_offset, + &ordered_extent->list); + + mutex_lock(&BTRFS_I(inode)->extent_mutex); + btrfs_ordered_update_i_size(inode, ordered_extent); + btrfs_update_inode(trans, root, inode); + btrfs_remove_ordered_extent(inode, ordered_extent); + mutex_unlock(&BTRFS_I(inode)->extent_mutex); + + /* once for us */ + btrfs_put_ordered_extent(ordered_extent); + /* once for the tree */ + btrfs_put_ordered_extent(ordered_extent); + + btrfs_end_transaction(trans, root); + if (path) + btrfs_free_path(path); + 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); +} + +/* + * When IO fails, either with EIO or csum verification fails, we + * try other mirrors that might have a good copy of the data. This + * io_failure_record is used to record state as we go through all the + * mirrors. If another mirror has good data, the page is set up to date + * and things continue. If a good mirror can't be found, the original + * bio end_io callback is called to indicate things have failed. + */ +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) +{ + 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; + 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; +} + +/* + * each time an IO finishes, we do a fast check in the IO failure tree + * to see if we need to process or clean up an io_failure_record + */ +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; +} + +/* + * when reads are done, we need to check csums to verify the data is correct + * if there's a match, we allow the bio to finish. If not, we go through + * the io_failure_record routines to find good copies + */ +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; + 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) { + goto zeroit; + } + csum = btrfs_csum_data(root, kaddr + offset, csum, end - start + 1); + btrfs_csum_final(csum, (char *)&csum); + if (csum != private) { + goto zeroit; + } + 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 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); + if (private == 0) + return 0; + return -EIO; +} + +/* + * This creates an orphan entry for the given inode in case something goes + * wrong in the middle of an unlink/truncate. + */ +int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret = 0; + + spin_lock(&root->list_lock); + + /* already on the orphan list, we're good */ + if (!list_empty(&BTRFS_I(inode)->i_orphan)) { + spin_unlock(&root->list_lock); + return 0; + } + + list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list); + + spin_unlock(&root->list_lock); + + /* + * insert an orphan item to track this unlinked/truncated file + */ + ret = btrfs_insert_orphan_item(trans, root, inode->i_ino); + + return ret; +} + +/* + * We have done the truncate/delete so we can go ahead and remove the orphan + * item for this particular inode. + */ +int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode) +{ + struct btrfs_root *root = BTRFS_I(inode)->root; + int ret = 0; + + spin_lock(&root->list_lock); + + if (list_empty(&BTRFS_I(inode)->i_orphan)) { + spin_unlock(&root->list_lock); + return 0; + } + + list_del_init(&BTRFS_I(inode)->i_orphan); + if (!trans) { + spin_unlock(&root->list_lock); + return 0; + } + + spin_unlock(&root->list_lock); + + ret = btrfs_del_orphan_item(trans, root, inode->i_ino); + + return ret; +} + +/* + * this cleans up any orphans that may be left on the list from the last use + * of this root. + */ +void btrfs_orphan_cleanup(struct btrfs_root *root) +{ + struct btrfs_path *path; + struct extent_buffer *leaf; + struct btrfs_item *item; + struct btrfs_key key, found_key; + struct btrfs_trans_handle *trans; + struct inode *inode; + int ret = 0, nr_unlink = 0, nr_truncate = 0; + + /* don't do orphan cleanup if the fs is readonly. */ + if (root->fs_info->sb->s_flags & MS_RDONLY) + return; + + path = btrfs_alloc_path(); + if (!path) + return; + path->reada = -1; + + key.objectid = BTRFS_ORPHAN_OBJECTID; + btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY); + key.offset = (u64)-1; + + + while (1) { + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) { + printk(KERN_ERR "Error searching slot for orphan: %d" + "\n", ret); + break; + } + + /* + * if ret == 0 means we found what we were searching for, which + * is weird, but possible, so only screw with path if we didnt + * find the key and see if we have stuff that matches + */ + if (ret > 0) { + if (path->slots[0] == 0) + break; + path->slots[0]--; + } + + /* pull out the item */ + leaf = path->nodes[0]; + item = btrfs_item_nr(leaf, path->slots[0]); + btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]); + + /* make sure the item matches what we want */ + if (found_key.objectid != BTRFS_ORPHAN_OBJECTID) + break; + if (btrfs_key_type(&found_key) != BTRFS_ORPHAN_ITEM_KEY) + break; + + /* release the path since we're done with it */ + btrfs_release_path(root, path); + + /* + * this is where we are basically btrfs_lookup, without the + * crossing root thing. we store the inode number in the + * offset of the orphan item. + */ + inode = btrfs_iget_locked(root->fs_info->sb, + found_key.offset, root); + if (!inode) + break; + + if (inode->i_state & I_NEW) { + BTRFS_I(inode)->root = root; + + /* have to set the location manually */ + BTRFS_I(inode)->location.objectid = inode->i_ino; + BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY; + BTRFS_I(inode)->location.offset = 0; + + btrfs_read_locked_inode(inode); + unlock_new_inode(inode); + } + + /* + * add this inode to the orphan list so btrfs_orphan_del does + * the proper thing when we hit it + */ + spin_lock(&root->list_lock); + list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list); + spin_unlock(&root->list_lock); + + /* + * if this is a bad inode, means we actually succeeded in + * removing the inode, but not the orphan record, which means + * we need to manually delete the orphan since iput will just + * do a destroy_inode + */ + if (is_bad_inode(inode)) { + trans = btrfs_start_transaction(root, 1); + btrfs_orphan_del(trans, inode); + btrfs_end_transaction(trans, root); + iput(inode); + continue; + } + + /* if we have links, this was a truncate, lets do that */ + if (inode->i_nlink) { + nr_truncate++; + btrfs_truncate(inode); + } else { + nr_unlink++; + } + + /* this will do delete_inode and everything for us */ + iput(inode); + } + + if (nr_unlink) + printk(KERN_INFO "btrfs: unlinked %d orphans\n", nr_unlink); + if (nr_truncate) + printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate); + + btrfs_free_path(path); +} + +/* + * read an inode from the btree into the in-memory inode + */ +void btrfs_read_locked_inode(struct inode *inode) +{ + struct btrfs_path *path; + struct extent_buffer *leaf; + struct btrfs_inode_item *inode_item; + struct btrfs_timespec *tspec; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_key location; + u64 alloc_group_block; + u32 rdev; + int ret; + + path = btrfs_alloc_path(); + BUG_ON(!path); + memcpy(&location, &BTRFS_I(inode)->location, sizeof(location)); + + ret = btrfs_lookup_inode(NULL, root, path, &location, 0); + if (ret) + goto make_bad; + + leaf = path->nodes[0]; + inode_item = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_inode_item); + + inode->i_mode = btrfs_inode_mode(leaf, inode_item); + 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); + 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); + inode->i_atime.tv_nsec = btrfs_timespec_nsec(leaf, tspec); + + tspec = btrfs_inode_mtime(inode_item); + inode->i_mtime.tv_sec = btrfs_timespec_sec(leaf, tspec); + inode->i_mtime.tv_nsec = btrfs_timespec_nsec(leaf, tspec); + + tspec = btrfs_inode_ctime(inode_item); + inode->i_ctime.tv_sec = btrfs_timespec_sec(leaf, tspec); + inode->i_ctime.tv_nsec = btrfs_timespec_nsec(leaf, tspec); + + inode_set_bytes(inode, btrfs_inode_nbytes(leaf, inode_item)); + BTRFS_I(inode)->generation = btrfs_inode_generation(leaf, inode_item); + inode->i_generation = BTRFS_I(inode)->generation; + inode->i_rdev = 0; + rdev = btrfs_inode_rdev(leaf, inode_item); + + BTRFS_I(inode)->index_cnt = (u64)-1; + + alloc_group_block = btrfs_inode_block_group(leaf, inode_item); + BTRFS_I(inode)->block_group = btrfs_lookup_block_group(root->fs_info, + alloc_group_block); + 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, + BTRFS_BLOCK_GROUP_METADATA, 0); + } + btrfs_free_path(path); + inode_item = NULL; + + switch (inode->i_mode & S_IFMT) { + case S_IFREG: + inode->i_mapping->a_ops = &btrfs_aops; + 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_IFDIR: |