/*
 * 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/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/xattr.h>
#include <linux/posix_acl.h>
#include <linux/falloc.h>
#include <linux/slab.h>
#include "compat.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 "tree-log.h"
#include "compression.h"
#include "locking.h"

struct btrfs_iget_args {
        u64 ino;
        struct btrfs_root *root;
};

static const struct inode_operations btrfs_dir_inode_operations;
static const struct inode_operations btrfs_symlink_inode_operations;
static const struct inode_operations btrfs_dir_ro_inode_operations;
static const struct inode_operations btrfs_special_inode_operations;
static const struct inode_operations btrfs_file_inode_operations;
static const struct address_space_operations btrfs_aops;
static const struct address_space_operations btrfs_symlink_aops;
static const 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_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);
static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end);
static noinline int cow_file_range(struct inode *inode,
                                   struct page *locked_page,
                                   u64 start, u64 end, int *page_started,
                                   unsigned long *nr_written, int unlock);

static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
                                     struct inode *inode,  struct inode *dir)
{
        int err;

        err = btrfs_init_acl(trans, inode, dir);
        if (!err)
                err = btrfs_xattr_security_init(trans, inode, dir);
        return err;
}

/*
 * this does all the hard work for inserting an inline extent into
 * the btree.  The caller should have done a btrfs_drop_extents so that
 * no overlapping inline items exist in the btree
 */
static noinline int insert_inline_extent(struct btrfs_trans_handle *trans,
                                struct btrfs_root *root, struct inode *inode,
                                u64 start, size_t size, size_t compressed_size,
                                struct page **compressed_pages)
{
        struct btrfs_key key;
        struct btrfs_path *path;
        struct extent_buffer *leaf;
        struct page *page = NULL;
        char *kaddr;
        unsigned long ptr;
        struct btrfs_file_extent_item *ei;
        int err = 0;
        int ret;
        size_t cur_size = size;
        size_t datasize;
        unsigned long offset;
        int use_compress = 0;

        if (compressed_size && compressed_pages) {
                use_compress = 1;
                cur_size = compressed_size;
        }

        path = btrfs_alloc_path();
        if (!path)
                return -ENOMEM;

        path->leave_spinning = 1;
        btrfs_set_trans_block_group(trans, inode);

        key.objectid = inode->i_ino;
        key.offset = start;
        btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY);
        datasize = btrfs_file_extent_calc_inline_size(cur_size);

        inode_add_bytes(inode, size);
        ret = btrfs_insert_empty_item(trans, root, path, &key,
                                      datasize);
        BUG_ON(ret);
        if (ret) {
                err = ret;
                goto fail;
        }
        leaf = path->nodes[0];
        ei = btrfs_item_ptr(leaf, path->slots[0],
                            struct btrfs_file_extent_item);
        btrfs_set_file_extent_generation(leaf, ei, trans->transid);
        btrfs_set_file_extent_type(leaf, ei, BTRFS_FILE_EXTENT_INLINE);
        btrfs_set_file_extent_encryption(leaf, ei, 0);
        btrfs_set_file_extent_other_encoding(leaf, ei, 0);
        btrfs_set_file_extent_ram_bytes(leaf, ei, size);
        ptr = btrfs_file_extent_inline_start(ei);

        if (use_compress) {
                struct page *cpage;
                int i = 0;
                while (compressed_size > 0) {
                        cpage = compressed_pages[i];
                        cur_size = min_t(unsigned long, compressed_size,
                                       PAGE_CACHE_SIZE);

                        kaddr = kmap_atomic(cpage, KM_USER0);
                        write_extent_buffer(leaf, kaddr, ptr, cur_size);
                        kunmap_atomic(kaddr, KM_USER0);

                        i++;
                        ptr += cur_size;
                        compressed_size -= cur_size;
                }
                btrfs_set_file_extent_compression(leaf, ei,
                                                  BTRFS_COMPRESS_ZLIB);
        } else {
                page = find_get_page(inode->i_mapping,
                                     start >> PAGE_CACHE_SHIFT);
                btrfs_set_file_extent_compression(leaf, ei, 0);
                kaddr = kmap_atomic(page, KM_USER0);
                offset = start & (PAGE_CACHE_SIZE - 1);
                write_extent_buffer(leaf, kaddr + offset, ptr, size);
                kunmap_atomic(kaddr, KM_USER0);
                page_cache_release(page);
        }
        btrfs_mark_buffer_dirty(leaf);
        btrfs_free_path(path);

        /*
         * we're an inline extent, so nobody can
         * extend the file past i_size without locking
         * a page we already have locked.
         *
         * We must do any isize and inode updates
         * before we unlock the pages.  Otherwise we
         * could end up racing with unlink.
         */
        BTRFS_I(inode)->disk_i_size = inode->i_size;
        btrfs_update_inode(trans, root, inode);

        return 0;
fail:
        btrfs_free_path(path);
        return err;
}


/*
 * conditionally insert an inline extent into the file.  This
 * does the checks required to make sure the data is small enough
 * to fit as an inline extent.
 */
static noinline int cow_file_range_inline(struct btrfs_trans_handle *trans,
                                 struct btrfs_root *root,
                                 struct inode *inode, u64 start, u64 end,
                                 size_t compressed_size,
                                 struct page **compressed_pages)
{
        u64 isize = i_size_read(inode);
        u64 actual_end = min(end + 1, isize);
        u64 inline_len = actual_end - start;
        u64 aligned_end = (end + root->sectorsize - 1) &
                        ~((u64)root->sectorsize - 1);
        u64 hint_byte;
        u64 data_len = inline_len;
        int ret;

        if (compressed_size)
                data_len = compressed_size;

        if (start > 0 ||
            actual_end >= PAGE_CACHE_SIZE ||
            data_len >= BTRFS_MAX_INLINE_DATA_SIZE(root) ||
            (!compressed_size &&
            (actual_end & (root->sectorsize - 1)) == 0) ||
            end + 1 < isize ||
            data_len > root->fs_info->max_inline) {
                return 1;
        }

        ret = btrfs_drop_extents(trans, inode, start, aligned_end,
                                 &hint_byte, 1);
        BUG_ON(ret);

        if (isize > actual_end)
                inline_len = min_t(u64, isize, actual_end);
        ret = insert_inline_extent(trans, root, inode, start,
                                   inline_len, compressed_size,
                                   compressed_pages);
        BUG_ON(ret);
        btrfs_delalloc_release_metadata(inode, end + 1 - start);
        btrfs_drop_extent_cache(inode, start, aligned_end - 1, 0);
        return 0;
}

struct async_extent {
        u64 start;
        u64 ram_size;
        u64 compressed_size;
        struct page **pages;
        unsigned long nr_pages;
        struct list_head list;
};

struct async_cow {
        struct inode *inode;
        struct btrfs_root *root;
        struct page *locked_page;
        u64 start;
        u64 end;
        struct list_head extents;
        struct btrfs_work work;
};

static noinline int add_async_extent(struct async_cow *cow,
                                     u64 start, u64 ram_size,
                                     u64 compressed_size,
                                     struct page **pages,
                                     unsigned long nr_pages)
{
        struct async_extent *async_extent;

        async_extent = kmalloc(sizeof(*async_extent), GFP_NOFS);
        async_extent->start = start;
        async_extent->ram_size = ram_size;
        async_extent->compressed_size = compressed_size;
        async_extent->pages = pages;
        async_extent->nr_pages = nr_pages;
        list_add_tail(&async_extent->list, &cow->extents);
        return 0;
}

/*
 * we create compressed extents in two phases.  The first
 * phase compresses a range of pages that have already been
 * locked (both pages and state bits are locked).
 *
 * This is done inside an ordered work queue, and the compression
 * is spread across many cpus.  The actual IO submission is step
 * two, and the ordered work queue takes care of making sure that
 * happens in the same order things were put onto the queue by
 * writepages and friends.
 *
 * If this code finds it can't get good compression, it puts an
 * entry onto the work queue to write the uncompressed bytes.  This
 * makes sure that both compressed inodes and uncompressed inodes
 * are written in the same order that pdflush sent them down.
 */
static noinline int compress_file_range(struct inode *inode,
                                        struct page *locked_page,
                                        u64 start, u64 end,
                                        struct async_cow *async_cow,
                                        int *num_added)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        struct btrfs_trans_handle *trans;
        u64 num_bytes;
        u64 orig_start;
        u64 disk_num_bytes;
        u64 blocksize = root->sectorsize;
        u64 actual_end;
        u64 isize = i_size_read(inode);
        int ret = 0;
        struct page **pages = NULL;
        unsigned long nr_pages;
        unsigned long nr_pages_ret = 0;
        unsigned long total_compressed = 0;
        unsigned long total_in = 0;
        unsigned long max_compressed = 128 * 1024;
        unsigned long max_uncompressed = 128 * 1024;
        int i;
        int will_compress;

        orig_start = start;

        actual_end = min_t(u64, isize, end + 1);
again:
        will_compress = 0;
        nr_pages = (end >> PAGE_CACHE_SHIFT) - (start >> PAGE_CACHE_SHIFT) + 1;
        nr_pages = min(nr_pages, (128 * 1024UL) / PAGE_CACHE_SIZE);

        /*
         * we don't want to send crud past the end of i_size through
         * compression, that's just a waste of CPU time.  So, if the
         * end of the file is before the start of our current
         * requested range of bytes, we bail out to the uncompressed
         * cleanup code that can deal with all of this.
         *
         * It isn't really the fastest way to fix things, but this is a
         * very uncommon corner.
         */
        if (actual_end <= start)
                goto cleanup_and_bail_uncompressed;

        total_compressed = actual_end - start;

        /* we want to make sure that amount of ram required to uncompress
         * an extent is reasonable, so we limit the total size in ram
         * of a compressed extent to 128k.  This is a crucial number
         * because it also controls how easily we can spread reads across
         * cpus for decompression.
         *
         * We also want to make sure the amount of IO required to do
         * a random read is reasonably small, so we limit the size of
         * a compressed extent to 128k.
         */
        total_compressed = min(total_compressed, max_uncompressed);
        num_bytes = (end - start + blocksize) & ~(blocksize - 1);
        num_bytes = max(blocksize,  num_bytes);
        disk_num_bytes = num_bytes;
        total_in = 0;
        ret = 0;

        /*
         * we do compression for mount -o compress and when the
         * inode has not been flagged as nocompress.  This flag can
         * change at any time if we discover bad compression ratios.
         */
        if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS) &&
            (btrfs_test_opt(root, COMPRESS) ||
             (BTRFS_I(inode)->force_compress))) {
                WARN_ON(pages);
                pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS);

                ret = btrfs_zlib_compress_pages(inode->i_mapping, start,
                                                total_compressed, pages,
                                                nr_pages, &nr_pages_ret,
                                                &total_in,
                                                &total_compressed,
                                                max_compressed);

                if (!ret) {
                        unsigned long offset = total_compressed &
                                (PAGE_CACHE_SIZE - 1);
                        struct page *page = pages[nr_pages_ret - 1];
                        char *kaddr;

                        /* zero the tail end of the last page, we might be
                         * sending it down to disk
                         */
                        if (offset) {
                                kaddr = kmap_atomic(page, KM_USER0);
                                memset(kaddr + offset, 0,
                                       PAGE_CACHE_SIZE - offset);
                                kunmap_atomic(kaddr, KM_USER0);
                        }
                        will_compress = 1;
                }
        }
        if (start == 0) {
                trans = btrfs_join_transaction(root, 1);
                BUG_ON(!trans);
                btrfs_set_trans_block_group(trans, inode);
                trans->block_rsv = &root->fs_info->delalloc_block_rsv;

                /* lets try to make an inline extent */
                if (ret || total_in < (actual_end - start)) {
                        /* we didn't compress the entire range, try
                         * to make an uncompressed inline extent.
                         */
                        ret = cow_file_range_inline(trans, root, inode,
                                                    start, end, 0, NULL);
                } else {
                        /* try making a compressed inline extent */
                        ret = cow_file_range_inline(trans, root, inode,
                                                    start, end,
                                                    total_compressed, pages);
                }
                if (ret == 0) {
                        /*
                         * inline extent creation worked, we don't need
                         * to create any more async work items.  Unlock
                         * and free up our temp pages.
                         */
                        extent_clear_unlock_delalloc(inode,
                             &BTRFS_I(inode)->io_tree,
                             start, end, NULL,
                             EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY |
                             EXTENT_CLEAR_DELALLOC |
                             EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK);

                        btrfs_end_transaction(trans, root);
                        goto free_pages_out;
                }
                btrfs_end_transaction(trans, root);
        }

        if (will_compress) {
                /*
                 * we aren't doing an inline extent round the compressed size
                 * up to a block size boundary so the allocator does sane
                 * things
                 */
                total_compressed = (total_compressed + blocksize - 1) &
                        ~(blocksize - 1);

                /*
                 * one last check to make sure the compression is really a
                 * win, compare the page count read with the blocks on disk
                 */
                total_in = (total_in + PAGE_CACHE_SIZE - 1) &
                        ~(PAGE_CACHE_SIZE - 1);
                if (total_compressed >= total_in) {
                        will_compress = 0;
                } else {
                        disk_num_bytes = total_compressed;
                        num_bytes = total_in;
                }
        }
        if (!will_compress && pages) {
                /*
                 * the compression code ran but failed to make things smaller,
                 * free any pages it allocated and our page pointer array
                 */
                for (i = 0; i < nr_pages_ret; i++) {
                        WARN_ON(pages[i]->mapping);
                        page_cache_release(pages[i]);
                }
                kfree(pages);
                pages = NULL;
                total_compressed = 0;
                nr_pages_ret = 0;

                /* flag the file so we don't compress in the future */
                if (!btrfs_test_opt(root, FORCE_COMPRESS) &&
                    !(BTRFS_I(inode)->force_compress)) {
                        BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
                }
        }
        if (will_compress) {
                *num_added += 1;

                /* the async work queues will take care of doing actual
                 * allocation on disk for these compressed pages,
                 * and will submit them to the elevator.
                 */
                add_async_extent(async_cow, start, num_bytes,
                                 total_compressed, pages, nr_pages_ret);

                if (start + num_bytes < end && start + num_bytes < actual_end) {
                        start += num_bytes;
                        pages = NULL;
                        cond_resched();
                        goto again;
                }
        } else {
cleanup_and_bail_uncompressed:
                /*
                 * No compression, but we still need to write the pages in
                 * the file we've been given so far.  redirty the locked
                 * page if it corresponds to our extent and set things up
                 * for the async work queue to run cow_file_range to do
                 * the normal delalloc dance
                 */
                if (page_offset(locked_page) >= start &&
                    page_offset(locked_page) <= end) {
                        __set_page_dirty_nobuffers(locked_page);
                        /* unlocked later on in the async handlers */
                }
                add_async_extent(async_cow, start, end - start + 1, 0, NULL, 0);
                *num_added += 1;
        }

out:
        return 0;

free_pages_out:
        for (i = 0; i < nr_pages_ret; i++) {
                WARN_ON(pages[i]->mapping);
                page_cache_release(pages[i]);
        }
        kfree(pages);

        goto out;
}

/*
 * phase two of compressed writeback.  This is the ordered portion
 * of the code, which only gets called in the order the work was
 * queued.  We walk all the async extents created by compress_file_range
 * and send them down to the disk.
 */
static noinline int submit_compressed_extents(struct inode *inode,
                                              struct async_cow *async_cow)
{
        struct async_extent *async_extent;
        u64 alloc_hint = 0;
        struct btrfs_trans_handle *trans;
        struct btrfs_key ins;
        struct extent_map *em;
        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;
        int ret = 0;

        if (list_empty(&async_cow->extents))
                return 0;


        while (!list_empty(&async_cow->extents)) {
                async_extent = list_entry(async_cow->extents.next,
                                          struct async_extent, list);
                list_del(&async_extent->list);

                io_tree = &BTRFS_I(inode)->io_tree;

retry:
                /* did the compression code fall back to uncompressed IO? */
                if (!async_extent->pages) {
                        int page_started = 0;
                        unsigned long nr_written = 0;

                        lock_extent(io_tree, async_extent->start,
                                         async_extent->start +
                                         async_extent->ram_size - 1, GFP_NOFS);

                        /* allocate blocks */
                        ret = cow_file_range(inode, async_cow->locked_page,
                                             async_extent->start,
                                             async_extent->start +
                                             async_extent->ram_size - 1,
                                             &page_started, &nr_written, 0);

                        /*
                         * if page_started, cow_file_range inserted an
                         * inline extent and took care of all the unlocking
                         * and IO for us.  Otherwise, we need to submit
                         * all those pages down to the drive.
                         */
                        if (!page_started && !ret)
                                extent_write_locked_range(io_tree,
                                                  inode, async_extent->start,
                                                  async_extent->start +
                                                  async_extent->ram_size - 1,
                                                  btrfs_get_extent,
                                                  WB_SYNC_ALL);
                        kfree(async_extent);
                        cond_resched();
                        continue;
                }

                lock_extent(io_tree, async_extent->start,
                            async_extent->start + async_extent->ram_size - 1,
                            GFP_NOFS);

                trans = btrfs_join_transaction(root, 1);
                ret = btrfs_reserve_extent(trans, root,
                                           async_extent->compressed_size,
                                           async_extent->compressed_size,
                                           0, alloc_hint,
                                           (u64)-1, &ins, 1);
                btrfs_end_transaction(trans, root);

                if (ret) {
                        int i;
                        for (i = 0; i < async_extent->nr_pages; i++) {
                                WARN_ON(async_extent->pages[i]->mapping);
                                page_cache_release(async_extent->pages[i]);
                        }
                        kfree(async_extent->pages);
                        async_extent->nr_pages = 0;
                        async_extent->pages = NULL;
                        unlock_extent(io_tree, async_extent->start,
                                      async_extent->start +
                                      async_extent->ram_size - 1, GFP_NOFS);
                        goto retry;
                }

                /*
                 * here we're doing allocation and writeback of the
                 * compressed pages
                 */
                btrfs_drop_extent_cache(inode, async_extent->start,
                                        async_extent->start +
                                        async_extent->ram_size - 1, 0);

                em = alloc_extent_map(GFP_NOFS);
                em->start = async_extent->start;
                em->len = async_extent->ram_size;
                em->orig_start = em->start;

                em->block_start = ins.objectid;
                em->block_len = ins.offset;
                em->bdev = root->fs_info->fs_devices->latest_bdev;
                set_bit(EXTENT_FLAG_PINNED, &em->flags);
                set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);

                while (1) {
                        write_lock(&em_tree->lock);
                        ret = add_extent_mapping(em_tree, em);
                        write_unlock(&em_tree->lock);
                        if (ret != -EEXIST) {
                                free_extent_map(em);
                                break;
                        }
                        btrfs_drop_extent_cache(inode, async_extent->start,
                                                async_extent->start +
                                                async_extent->ram_size - 1, 0);
                }

                ret = btrfs_add_ordered_extent(inode, async_extent->start,
                                               ins.objectid,
                                               async_extent->ram_size,
                                               ins.offset,
                                               BTRFS_ORDERED_COMPRESSED);
                BUG_ON(ret);

                /*
                 * clear dirty, set writeback and unlock the pages.
                 */
                extent_clear_unlock_delalloc(inode,
                                &BTRFS_I(inode)->io_tree,
                                async_extent->start,
                                async_extent->start +
                                async_extent->ram_size - 1,
                                NULL, EXTENT_CLEAR_UNLOCK_PAGE |
                                EXTENT_CLEAR_UNLOCK |
                                EXTENT_CLEAR_DELALLOC |
                                EXTENT_CLEAR_DIRTY | EXTENT_SET_WRITEBACK);

                ret = btrfs_submit_compressed_write(inode,
                                    async_extent->start,
                                    async_extent->ram_size,
                                    ins.objectid,
                                    ins.offset, async_extent->pages,
                                    async_extent->nr_pages);

                BUG_ON(ret);
                alloc_hint = ins.objectid + ins.offset;
                kfree(async_extent);
                cond_resched();
        }

        return 0;
}

static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
                                      u64 num_bytes)
{
        struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
        struct extent_map *em;
        u64 alloc_hint = 0;

        read_lock(&em_tree->lock);
        em = search_extent_mapping(em_tree, start, num_bytes);
        if (em) {
                /*
                 * if block start isn't an actual block number then find the
                 * first block in this inode and use that as a hint.  If that
                 * block is also bogus then just don't worry about it.
                 */
                if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
                        free_extent_map(em);
                        em = search_extent_mapping(em_tree, 0, 0);
                        if (em && em->block_start < EXTENT_MAP_LAST_BYTE)
                                alloc_hint = em->block_start;
                        if (em)
                                free_extent_map(em);
                } else {
                        alloc_hint = em->block_start;
                        free_extent_map(em);
                }
        }
        read_unlock(&em_tree->lock);

        return alloc_hint;
}

/*
 * 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.
 *
 * locked_page is the page that writepage had locked already.  We use
 * it to make sure we don't do extra locks or unlocks.
 *
 * *page_started is set to one if we unlock locked_page and do everything
 * required to start IO on it.  It may be clean and already done with
 * IO when we return.
 */
static noinline int cow_file_range(struct inode *inode,
                                   struct page *locked_page,
                                   u64 start, u64 end, int *page_started,
                                   unsigned long *nr_written,
                                   int unlock)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        struct btrfs_trans_handle *trans;
        u64 alloc_hint = 0;
        u64 num_bytes;
        unsigned long ram_size;
        u64 disk_num_bytes;
        u64 cur_alloc_size;
        u64 blocksize = root->sectorsize;
        u64 actual_end;
        u64 isize = i_size_read(inode);
        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);
        trans->block_rsv = &root->fs_info->delalloc_block_rsv;

        actual_end = min_t(u64, isize, end + 1);

        num_bytes = (end - start + blocksize) & ~(blocksize - 1);
        num_bytes = max(blocksize,  num_bytes);
        disk_num_bytes = num_bytes;
        ret = 0;

        if (start == 0) {
                /* lets try to make an inline extent */
                ret = cow_file_range_inline(trans, root, inode,
                                            start, end, 0, NULL);
                if (ret == 0) {
                        extent_clear_unlock_delalloc(inode,
                                     &BTRFS_I(inode)->io_tree,
                                     start, end, NULL,
                                     EXTENT_CLEAR_UNLOCK_PAGE |
                                     EXTENT_CLEAR_UNLOCK |
                                     EXTENT_CLEAR_DELALLOC |
                                     EXTENT_CLEAR_DIRTY |
                                     EXTENT_SET_WRITEBACK |
                                     EXTENT_END_WRITEBACK);

                        *nr_written = *nr_written +
                             (end - start + PAGE_CACHE_SIZE) / PAGE_CACHE_SIZE;
                        *page_started = 1;
                        ret = 0;
                        goto out;
                }
        }

        BUG_ON(disk_num_bytes >
               btrfs_super_total_bytes(&root->fs_info->super_copy));

        alloc_hint = get_extent_allocation_hint(inode, start, num_bytes);
        btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);

        while (disk_num_bytes > 0) {
                unsigned long op;

                cur_alloc_size = disk_num_bytes;
                ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
                                           root->sectorsize, 0, alloc_hint,
                                           (u64)-1, &ins, 1);
                BUG_ON(ret);

                em = alloc_extent_map(GFP_NOFS);
                em->start = start;
                em->orig_start = em->start;
                ram_size = ins.offset;
                em->len = ins.offset;

                em->block_start = ins.objectid;
                em->block_len = ins.offset;
                em->bdev = root->fs_info->fs_devices->latest_bdev;
                set_bit(EXTENT_FLAG_PINNED, &em->flags);

                while (1) {
                        write_lock(&em_tree->lock);
                        ret = add_extent_mapping(em_tree, em);
                        write_unlock(&em_tree->lock);
                        if (ret != -EEXIST) {
                                free_extent_map(em);
                                break;
                        }
                        btrfs_drop_extent_cache(inode, start,
                                                start + ram_size - 1, 0);
                }

                cur_alloc_size = ins.offset;
                ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
                                               ram_size, cur_alloc_size, 0);
                BUG_ON(ret);

                if (root->root_key.objectid ==
                    BTRFS_DATA_RELOC_TREE_OBJECTID) {
                        ret = btrfs_reloc_clone_csums(inode, start,
                                                      cur_alloc_size);
                        BUG_ON(ret);
                }

                if (disk_num_bytes < cur_alloc_size)
                        break;

                /* we're not doing compressed IO, don't unlock the first
                 * page (which the caller expects to stay locked), don't
                 * clear any dirty bits and don't set any writeback bits
                 *
                 * Do set the Private2 bit so we know this page was properly
                 * setup for writepage
                 */
                op = unlock ? EXTENT_CLEAR_UNLOCK_PAGE : 0;
                op |= EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC |
                        EXTENT_SET_PRIVATE2;

                extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
                                             start, start + ram_size - 1,
                                             locked_page, op);
                disk_num_bytes -= cur_alloc_size;
                num_bytes -= cur_alloc_size;
                alloc_hint = ins.objectid + ins.offset;
                start += cur_alloc_size;
        }
out:
        ret = 0;
        btrfs_end_transaction(trans, root);

        return ret;
}

/*
 * work queue call back to started compression on a file and pages
 */
static noinline void async_cow_start(struct btrfs_work *work)
{
        struct async_cow *async_cow;
        int num_added = 0;
        async_cow = container_of(work, struct async_cow, work);

        compress_file_range(async_cow->inode, async_cow->locked_page,
                            async_cow->start, async_cow->end, async_cow,
                            &num_added);
        if (num_added == 0)
                async_cow->inode = NULL;
}

/*
 * work queue call back to submit previously compressed pages
 */
static noinline void async_cow_submit(struct btrfs_work *work)
{
        struct async_cow *async_cow;
        struct btrfs_root *root;
        unsigned long nr_pages;

        async_cow = container_of(work, struct async_cow, work);

        root = async_cow->root;
        nr_pages = (async_cow->end - async_cow->start + PAGE_CACHE_SIZE) >>
                PAGE_CACHE_SHIFT;

        atomic_sub(nr_pages, &root->fs_info->async_delalloc_pages);

        if (atomic_read(&root->fs_info->async_delalloc_pages) <
            5 * 1042 * 1024 &&
            waitqueue_active(&root->fs_info->async_submit_wait))
                wake_up(&root->fs_info->async_submit_wait);

        if (async_cow->inode)
                submit_compressed_extents(async_cow->inode, async_cow);
}

static noinline void async_cow_free(struct btrfs_work *work)
{
        struct async_cow *async_cow;
        async_cow = container_of(work, struct async_cow, work);
        kfree(async_cow);
}

static int cow_file_range_async(struct inode *inode, struct page *locked_page,
                                u64 start, u64 end, int *page_started,
                                unsigned long *nr_written)
{
        struct async_cow *async_cow;
        struct btrfs_root *root = BTRFS_I(inode)->root;
        unsigned long nr_pages;
        u64 cur_end;
        int limit = 10 * 1024 * 1042;

        clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED,
                         1, 0, NULL, GFP_NOFS);
        while (start < end) {
                async_cow = kmalloc(sizeof(*async_cow), GFP_NOFS);
                async_cow->inode = inode;
                async_cow->root = root;
                async_cow->locked_page = locked_page;
                async_cow->start = start;

                if (BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS)
                        cur_end = end;
                else
                        cur_end = min(end, start + 512 * 1024 - 1);

                async_cow->end = cur_end;
                INIT_LIST_HEAD(&async_cow->extents);

                async_cow->work.func = async_cow_start;
                async_cow->work.ordered_func = async_cow_submit;
                async_cow->work.ordered_free = async_cow_free;
                async_cow->work.flags = 0;

                nr_pages = (cur_end - start + PAGE_CACHE_SIZE) >>
                        PAGE_CACHE_SHIFT;
                atomic_add(nr_pages, &root->fs_info->async_delalloc_pages);

                btrfs_queue_worker(&root->fs_info->delalloc_workers,
                                   &async_cow->work);

                if (atomic_read(&root->fs_info->async_delalloc_pages) > limit) {
                        wait_event(root->fs_info->async_submit_wait,
                           (atomic_read(&root->fs_info->async_delalloc_pages) <
                            limit));
                }

                while (atomic_read(&root->fs_info->async_submit_draining) &&
                      atomic_read(&root->fs_info->async_delalloc_pages)) {
                        wait_event(root->fs_info->async_submit_wait,
                          (atomic_read(&root->fs_info->async_delalloc_pages) ==
                           0));
                }

                *nr_written += nr_pages;
                start = cur_end + 1;
        }
        *page_started = 1;
        return 0;
}

static noinline int csum_exist_in_range(struct btrfs_root *root,
                                        u64 bytenr, u64 num_bytes)
{
        int ret;
        struct btrfs_ordered_sum *sums;
        LIST_HEAD(list);

        ret = btrfs_lookup_csums_range(root->fs_info->csum_root, bytenr,
                                       bytenr + num_bytes - 1, &list);
        if (ret == 0 && list_empty(&list))
                return 0;

        while (!list_empty(&list)) {
                sums = list_entry(list.next, struct btrfs_ordered_sum, list);
                list_del(&sums->list);
                kfree(sums);
        }
        return 1;
}

/*
 * 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 noinline int run_delalloc_nocow(struct inode *inode,
                                       struct page *locked_page,
                              u64 start, u64 end, int *page_started, int force,
                              unsigned long *nr_written)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        struct btrfs_trans_handle *trans;
        struct extent_buffer *leaf;
        struct btrfs_path *path;
        struct btrfs_file_extent_item *fi;
        struct btrfs_key found_key;
        u64 cow_start;
        u64 cur_offset;
        u64 extent_end;
        u64 extent_offset;
        u64 disk_bytenr;
        u64 num_bytes;
        int extent_type;
        int ret;
        int type;
        int nocow;
        int check_prev = 1;

        path = btrfs_alloc_path();
        BUG_ON(!path);
        trans = btrfs_join_transaction(root, 1);
        BUG_ON(!trans);

        cow_start = (u64)-1;
        cur_offset = start;
        while (1) {
                ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino,
                                               cur_offset, 0);
                BUG_ON(ret < 0);
                if (ret > 0 && path->slots[0] > 0 && check_prev) {
                        leaf = path->nodes[0];
                        btrfs_item_key_to_cpu(leaf, &found_key,
                                              path->slots[0] - 1);
                        if (found_key.objectid == inode->i_ino &&
                            found_key.type == BTRFS_EXTENT_DATA_KEY)
                                path->slots[0]--;
                }
                check_prev = 0;
next_slot:
                leaf = path->nodes[0];
                if (path->slots[0] >= btrfs_header_nritems(leaf)) {
                        ret = btrfs_next_leaf(root, path);
                        if (ret < 0)
                                BUG_ON(1);
                        if (ret > 0)
                                break;
                        leaf = path->nodes[0];
                }

                nocow = 0;
                disk_bytenr = 0;
                num_bytes = 0;
                btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);

                if (found_key.objectid > inode->i_ino ||
                    found_key.type > BTRFS_EXTENT_DATA_KEY ||
                    found_key.offset > end)
                        break;

                if (found_key.offset > cur_offset) {
                        extent_end = found_key.offset;
                        extent_type = 0;
                        goto out_check;
                }

                fi = btrfs_item_ptr(leaf, path->slots[0],
                                    struct btrfs_file_extent_item);
                extent_type = btrfs_file_extent_type(leaf, fi);

                if (extent_type == BTRFS_FILE_EXTENT_REG ||
                    extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
                        disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
                        extent_offset = btrfs_file_extent_offset(leaf, fi);
                        extent_end = found_key.offset +
                                btrfs_file_extent_num_bytes(leaf, fi);
                        if (extent_end <= start) {
                                path->slots[0]++;
                                goto next_slot;
                        }
                        if (disk_bytenr == 0)
                                goto out_check;
                        if (btrfs_file_extent_compression(leaf, fi) ||
                            btrfs_file_extent_encryption(leaf, fi) ||
                            btrfs_file_extent_other_encoding(leaf, fi))
                                goto out_check;
                        if (extent_type == BTRFS_FILE_EXTENT_REG && !force)
                                goto out_check;
                        if (btrfs_extent_readonly(root, disk_bytenr))
                                goto out_check;
                        if (btrfs_cross_ref_exist(trans, root, inode->i_ino,
                                                  found_key.offset -
                                                  extent_offset, disk_bytenr))
                                goto out_check;
                        disk_bytenr += extent_offset;
                        disk_bytenr += cur_offset - found_key.offset;
                        num_bytes = min(end + 1, extent_end) - cur_offset;
                        /*
                         * force cow if csum exists in the range.
                         * this ensure that csum for a given extent are
                         * either valid or do not exist.
                         */
                        if (csum_exist_in_range(root, disk_bytenr, num_bytes))
                                goto out_check;
                        nocow = 1;
                } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
                        extent_end = found_key.offset +
                                btrfs_file_extent_inline_len(leaf, fi);
                        extent_end = ALIGN(extent_end, root->sectorsize);
                } else {
                        BUG_ON(1);
                }
out_check:
                if (extent_end <= start) {
                        path->slots[0]++;
                        goto next_slot;
                }
                if (!nocow) {
                        if (cow_start == (u64)-1)
                                cow_start = cur_offset;
                        cur_offset = extent_end;
                        if (cur_offset > end)
                                break;
                        path->slots[0]++;
                        goto next_slot;
                }

                btrfs_release_path(root, path);
                if (cow_start != (u64)-1) {
                        ret = cow_file_range(inode, locked_page, cow_start,
                                        found_key.offset - 1, page_started,
                                        nr_written, 1);
                        BUG_ON(ret);
                        cow_start = (u64)-1;
                }

                if (extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
                        struct extent_map *em;
                        struct extent_map_tree *em_tree;
                        em_tree = &BTRFS_I(inode)->extent_tree;
                        em = alloc_extent_map(GFP_NOFS);
                        em->start = cur_offset;
                        em->orig_start = em->start;
                        em->len = num_bytes;
                        em->block_len = num_bytes;
                        em->block_start = disk_bytenr;
                        em->bdev = root->fs_info->fs_devices->latest_bdev;
                        set_bit(EXTENT_FLAG_PINNED, &em->flags);
                        while (1) {
                                write_lock(&em_tree->lock);
                                ret = add_extent_mapping(em_tree, em);
                                write_unlock(&em_tree->lock);
                                if (ret != -EEXIST) {
                                        free_extent_map(em);
                                        break;
                                }
                                btrfs_drop_extent_cache(inode, em->start,
                                                em->start + em->len - 1, 0);
                        }
                        type = BTRFS_ORDERED_PREALLOC;
                } else {
                        type = BTRFS_ORDERED_NOCOW;
                }

                ret = btrfs_add_ordered_extent(inode, cur_offset, disk_bytenr,
                                               num_bytes, num_bytes, type);
                BUG_ON(ret);

                if (root->root_key.objectid ==
                    BTRFS_DATA_RELOC_TREE_OBJECTID) {
                        ret = btrfs_reloc_clone_csums(inode, cur_offset,
                                                      num_bytes);
                        BUG_ON(ret);
                }

                extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
                                cur_offset, cur_offset + num_bytes - 1,
                                locked_page, EXTENT_CLEAR_UNLOCK_PAGE |
                                EXTENT_CLEAR_UNLOCK | EXTENT_CLEAR_DELALLOC |
                                EXTENT_SET_PRIVATE2);
                cur_offset = extent_end;
                if (cur_offset > end)
                        break;
        }
        btrfs_release_path(root, path);

        if (cur_offset <= end && cow_start == (u64)-1)
                cow_start = cur_offset;
        if (cow_start != (u64)-1) {
                ret = cow_file_range(inode, locked_page, cow_start, end,
                                     page_started, nr_written, 1);
                BUG_ON(ret);
        }

        ret = btrfs_end_transaction(trans, root);
        BUG_ON(ret);
        btrfs_free_path(path);
        return 0;
}

/*
 * extent_io.c call back to do delayed allocation processing
 */
static int run_delalloc_range(struct inode *inode, struct page *locked_page,
                              u64 start, u64 end, int *page_started,
                              unsigned long *nr_written)
{
        int ret;
        struct btrfs_root *root = BTRFS_I(inode)->root;

        if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW)
                ret = run_delalloc_nocow(inode, locked_page, start, end,
                                         page_started, 1, nr_written);
        else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC)
                ret = run_delalloc_nocow(inode, locked_page, start, end,
                                         page_started, 0, nr_written);
        else if (!btrfs_test_opt(root, COMPRESS) &&
                 !(BTRFS_I(inode)->force_compress))
                ret = cow_file_range(inode, locked_page, start, end,
                                      page_started, nr_written, 1);
        else
                ret = cow_file_range_async(inode, locked_page, start, end,
                                           page_started, nr_written);
        return ret;
}

static int btrfs_split_extent_hook(struct inode *inode,
                                   struct extent_state *orig, u64 split)
{
        /* not delalloc, ignore it */
        if (!(orig->state & EXTENT_DELALLOC))
                return 0;

        atomic_inc(&BTRFS_I(inode)->outstanding_extents);
        return 0;
}

/*
 * extent_io.c merge_extent_hook, used to track merged delayed allocation
 * extents so we can keep track of new extents that are just merged onto old
 * extents, such as when we are doing sequential writes, so we can properly
 * account for the metadata space we'll need.
 */
static int btrfs_merge_extent_hook(struct inode *inode,
                                   struct extent_state *new,
                                   struct extent_state *other)
{
        /* not delalloc, ignore it */
        if (!(other->state & EXTENT_DELALLOC))
                return 0;

        atomic_dec(&BTRFS_I(inode)->outstanding_extents);
        return 0;
}

/*
 * 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.
 */
static int btrfs_set_bit_hook(struct inode *inode,
                              struct extent_state *state, int *bits)
{

        /*
         * set_bit and clear bit hooks normally require _irqsave/restore
         * but in this case, we are only testeing for the DELALLOC
         * bit, which is only set or cleared with irqs on
         */
        if (!(state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
                struct btrfs_root *root = BTRFS_I(inode)->root;
                u64 len = state->end + 1 - state->start;

                if (*bits & EXTENT_FIRST_DELALLOC)
                        *bits &= ~EXTENT_FIRST_DELALLOC;
                else
                        atomic_inc(&BTRFS_I(inode)->outstanding_extents);

                spin_lock(&root->fs_info->delalloc_lock);
                BTRFS_I(inode)->delalloc_bytes += len;
                root->fs_info->delalloc_bytes += len;
                if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
                        list_add_tail(&BTRFS_I(inode)->delalloc_inodes,
                                      &root->fs_info->delalloc_inodes);
                }
                spin_unlock(&root->fs_info->delalloc_lock);
        }
        return 0;
}

/*
 * extent_io.c clear_bit_hook, see set_bit_hook for why
 */
static int btrfs_clear_bit_hook(struct inode *inode,
                                struct extent_state *state, int *bits)
{
        /*
         * set_bit and clear bit hooks normally require _irqsave/restore
         * but in this case, we are only testeing for the DELALLOC
         * bit, which is only set or cleared with irqs on
         */
        if ((state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
                struct btrfs_root *root = BTRFS_I(inode)->root;
                u64 len = state->end + 1 - state->start;

                if (*bits & EXTENT_FIRST_DELALLOC)
                        *bits &= ~EXTENT_FIRST_DELALLOC;
                else if (!(*bits & EXTENT_DO_ACCOUNTING))
                        atomic_dec(&BTRFS_I(inode)->outstanding_extents);

                if (*bits & EXTENT_DO_ACCOUNTING)
                        btrfs_delalloc_release_metadata(inode, len);

                if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID)
                        btrfs_free_reserved_data_space(inode, len);

                spin_lock(&root->fs_info->delalloc_lock);
                root->fs_info->delalloc_bytes -= len;
                BTRFS_I(inode)->delalloc_bytes -= len;

                if (BTRFS_I(inode)->delalloc_bytes == 0 &&
                    !list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
                        list_del_init(&BTRFS_I(inode)->delalloc_inodes);
                }
                spin_unlock(&root->fs_info->delalloc_lock);
        }
        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,
                         unsigned long bio_flags)
{
        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;

        if (bio_flags & EXTENT_BIO_COMPRESSED)
                return 0;

        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
 */
static int __btrfs_submit_bio_start(struct inode *inode, int rw,
                                    struct bio *bio, int mirror_num,
                                    unsigned long bio_flags,
                                    u64 bio_offset)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        int ret = 0;

        ret = btrfs_csum_one_bio(root, inode, bio, 0, 0);
        BUG_ON(ret);
        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
 */
static int __btrfs_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
                          int mirror_num, unsigned long bio_flags,
                          u64 bio_offset)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        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
 */
static int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
                          int mirror_num, unsigned long bio_flags,
                          u64 bio_offset)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        int ret = 0;
        int skip_sum;

        skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;

        ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
        BUG_ON(ret);

        if (!(rw & (1 << BIO_RW))) {
                if (bio_flags & EXTENT_BIO_COMPRESSED) {
                        return btrfs_submit_compressed_read(inode, bio,
                                                    mirror_num, bio_flags);
                } else if (!skip_sum)
                        btrfs_lookup_bio_sums(root, inode, bio, NULL);
                goto mapit;
        } else if (!skip_sum) {
                /* csum items have already been cloned */
                if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID)
                        goto mapit;
                /* we're doing a write, do the async checksumming */
                return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
                                   inode, rw, bio, mirror_num,
                                   bio_flags, bio_offset,
                                   __btrfs_submit_bio_start,
                                   __btrfs_submit_bio_done);
        }

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 btrfs_ordered_sum *sum;

        btrfs_set_trans_block_group(trans, inode);

        list_for_each_entry(sum, list, list) {
                btrfs_csum_file_blocks(trans,
                       BTRFS_I(inode)->root->fs_info->csum_root, sum);
        }
        return 0;
}

int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
                              struct extent_state **cached_state)
{
        if ((end & (PAGE_CACHE_SIZE - 1)) == 0)
                WARN_ON(1);
        return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end,
                                   cached_state, 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;
};

static void btrfs_writepage_fixup_worker(struct btrfs_work *work)
{
        struct btrfs_writepage_fixup *fixup;
        struct btrfs_ordered_extent *ordered;
        struct extent_state *cached_state = NULL;
        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_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0,
                         &cached_state, GFP_NOFS);

        /* already ordered? We're done */
        if (PagePrivate2(page))
                goto out;

        ordered = btrfs_lookup_ordered_extent(inode, page_start);
        if (ordered) {
                unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start,
                                     page_end, &cached_state, GFP_NOFS);
                unlock_page(page);
                btrfs_start_ordered_extent(inode, ordered, 1);
                goto again;
        }

        BUG();
        btrfs_set_extent_delalloc(inode, page_start, page_end, &cached_state);
        ClearPageChecked(page);
out:
        unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, page_end,
                             &cached_state, 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 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.
 */
static 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;

        /* this page is properly in the ordered list */
        if (TestClearPagePrivate2(page))
                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;
}

static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
                                       struct inode *inode, u64 file_pos,
                                       u64 disk_bytenr, u64 disk_num_bytes,
                                       u64 num_bytes, u64 ram_bytes,
                                       u8 compression, u8 encryption,
                                       u16 other_encoding, int extent_type)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        struct btrfs_file_extent_item *fi;
        struct btrfs_path *path;
        struct extent_buffer *leaf;
        struct btrfs_key ins;
        u64 hint;
        int ret;

        path = btrfs_alloc_path();
        BUG_ON(!path);

        path->leave_spinning = 1;

        /*
         * we may be replacing one extent in the tree with another.
         * The new extent is pinned in the extent map, and we don't want
         * to drop it from the cache until it is completely in the btree.
         *
         * So, tell btrfs_drop_extents to leave this extent in the cache.
         * the caller is expected to unpin it and allow it to be merged
         * with the others.
         */
        ret = btrfs_drop_extents(trans, inode, file_pos, file_pos + num_bytes,
                                 &hint, 0);
        BUG_ON(ret);

        ins.objectid = inode->i_ino;
        ins.offset = file_pos;
        ins.type = BTRFS_EXTENT_DATA_KEY;
        ret = btrfs_insert_empty_item(trans, root, path, &ins, sizeof(*fi));
        BUG_ON(ret);
        leaf = path->nodes[0];
        fi = btrfs_item_ptr(leaf, path->slots[0],
                            struct btrfs_file_extent_item);
        btrfs_set_file_extent_generation(leaf, fi, trans->transid);
        btrfs_set_file_extent_type(leaf, fi, extent_type);
        btrfs_set_file_extent_disk_bytenr(leaf, fi, disk_bytenr);
        btrfs_set_file_extent_disk_num_bytes(leaf, fi, disk_num_bytes);
        btrfs_set_file_extent_offset(leaf, fi, 0);
        btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
        btrfs_set_file_extent_ram_bytes(leaf, fi, ram_bytes);
        btrfs_set_file_extent_compression(leaf, fi, compression);
        btrfs_set_file_extent_encryption(leaf, fi, encryption);
        btrfs_set_file_extent_other_encoding(leaf, fi, other_encoding);

        btrfs_unlock_up_safe(path, 1);
        btrfs_set_lock_blocking(leaf);

        btrfs_mark_buffer_dirty(leaf);

        inode_add_bytes(inode, num_bytes);

        ins.objectid = disk_bytenr;
        ins.offset = disk_num_bytes;
        ins.type = BTRFS_EXTENT_ITEM_KEY;
        ret = btrfs_alloc_reserved_file_extent(trans, root,
                                        root->root_key.objectid,
                                        inode->i_ino, file_pos, &ins);
        BUG_ON(ret);
        btrfs_free_path(path);

        return 0;
}

/*
 * helper function for btrfs_finish_ordered_io, this
 * just reads in some of the csum leaves to prime them into ram
 * before we start the transaction.  It limits the amount of btree
 * reads required while inside the transaction.
 */
/* 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 = NULL;
        struct btrfs_ordered_extent *ordered_extent = NULL;
        struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
        struct extent_state *cached_state = NULL;
        int compressed = 0;
        int ret;

        ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start,
                                             end - start + 1);
        if (!ret)
                return 0;
        BUG_ON(!ordered_extent);

        if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
                BUG_ON(!list_empty(&ordered_extent->list));
                ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
                if (!ret) {
                        trans = btrfs_join_transaction(root, 1);
                        btrfs_set_trans_block_group(trans, inode);
                        trans->block_rsv = &root->fs_info->delalloc_block_rsv;
                        ret = btrfs_update_inode(trans, root, inode);
                        BUG_ON(ret);
                }
                goto out;
        }

        lock_extent_bits(io_tree, ordered_extent->file_offset,
                         ordered_extent->file_offset + ordered_extent->len - 1,
                         0, &cached_state, GFP_NOFS);

        trans = btrfs_join_transaction(root, 1);
        btrfs_set_trans_block_group(trans, inode);
        trans->block_rsv = &root->fs_info->delalloc_block_rsv;

        if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
                compressed = 1;
        if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
                BUG_ON(compressed);
                ret = btrfs_mark_extent_written(trans, inode,
                                                ordered_extent->file_offset,
                                                ordered_extent->file_offset +
                                                ordered_extent->len);
                BUG_ON(ret);
        } else {
                ret = insert_reserved_file_extent(trans, inode,
                                                ordered_extent->file_offset,
                                                ordered_extent->start,
                                                ordered_extent->disk_len,
                                                ordered_extent->len,
                                                ordered_extent->len,
                                                compressed, 0, 0,
                                                BTRFS_FILE_EXTENT_REG);
                unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
                                   ordered_extent->file_offset,
                                   ordered_extent->len);
                BUG_ON(ret);
        }
        unlock_extent_cached(io_tree, ordered_extent->file_offset,
                             ordered_extent->file_offset +
                             ordered_extent->len - 1, &cached_state, GFP_NOFS);

        add_pending_csums(trans, inode, ordered_extent->file_offset,
                          &ordered_extent->list);

        btrfs_ordered_update_i_size(inode, 0, ordered_extent);
        ret = btrfs_update_inode(trans, root, inode);
        BUG_ON(ret);
out:
        btrfs_delalloc_release_metadata(inode, ordered_extent->len);
        if (trans)
                btrfs_end_transaction(trans, root);
        /* once for us */
        btrfs_put_ordered_extent(ordered_extent);
        /* once for the tree */
        btrfs_put_ordered_extent(ordered_extent);

        return 0;
}

static int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
                                struct extent_state *state, int uptodate)
{
        ClearPagePrivate2(page);
        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;
        unsigned long bio_flags;
        int last_mirror;
};

static 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;
                failrec->bio_flags = 0;

                read_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;
                }
                read_unlock(&em_tree->lock);

                if (!em || IS_ERR(em)) {
                        kfree(failrec);
                        return -EIO;
                }
                logical = start - em->start;
                logical = em->block_start + logical;
                if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
                        logical = em->block_start;
                        failrec->bio_flags = EXTENT_BIO_COMPRESSED;
                }
                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(&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(&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,
                                                      failrec->bio_flags, 0);
        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
 */
static 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
 */
static 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;

        if (PageChecked(page)) {
                ClearPageChecked(page);
                goto good;
        }

        if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)
                return 0;

        if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID &&
            test_range_bit(io_tree, start, end, EXTENT_NODATASUM, 1, NULL)) {
                clear_extent_bits(io_tree, start, end, EXTENT_NODATASUM,
                                  GFP_NOFS);
                return 0;
        }

        if (state && state->start == start) {
                private = state->private;
                ret = 0;
        } else {
                ret = get_state_private(io_tree, start, &private);
        }
        kaddr = kmap_atomic(page, KM_USER0);
        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_USER0);
good:
        /* 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:
        if (printk_ratelimit()) {
                printk(KERN_INFO "btrfs csum failed ino %lu off %llu csum %u "
                       "private %llu\n", page->mapping->host->i_ino,
                       (unsigned long long)start, csum,
                       (unsigned long long)private);
        }
        memset(kaddr + offset, 1, end - start + 1);
        flush_dcache_page(page);
        kunmap_atomic(kaddr, KM_USER0);
        if (private == 0)
                return 0;
        return -EIO;
}

struct delayed_iput {
        struct list_head list;
        struct inode *inode;
};

void btrfs_add_delayed_iput(struct inode *inode)
{
        struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
        struct delayed_iput *delayed;

        if (atomic_add_unless(&inode->i_count, -1, 1))
                return;

        delayed = kmalloc(sizeof(*delayed), GFP_NOFS | __GFP_NOFAIL);
        delayed->inode = inode;

        spin_lock(&fs_info->delayed_iput_lock);
        list_add_tail(&delayed->list, &fs_info->delayed_iputs);
        spin_unlock(&fs_info->delayed_iput_lock);
}

void btrfs_run_delayed_iputs(struct btrfs_root *root)
{
        LIST_HEAD(list);
        struct btrfs_fs_info *fs_info = root->fs_info;
        struct delayed_iput *delayed;
        int empty;

        spin_lock(&fs_info->delayed_iput_lock);
        empty = list_empty(&fs_info->delayed_iputs);
        spin_unlock(&fs_info->delayed_iput_lock);
        if (empty)
                return;

        down_read(&root->fs_info->cleanup_work_sem);
        spin_lock(&fs_info->delayed_iput_lock);
        list_splice_init(&fs_info->delayed_iputs, &list);
        spin_unlock(&fs_info->delayed_iput_lock);

        while (!list_empty(&list)) {
                delayed = list_entry(list.next, struct delayed_iput, list);
                list_del(&delayed->list);
                iput(delayed->inode);
                kfree(delayed);

        }
        up_read(&root->fs_info->cleanup_work_sem);
}

/*
 * calculate extra metadata reservation when snapshotting a subvolume
 * contains orphan files.
 */
void btrfs_orphan_pre_snapshot(struct btrfs_trans_handle *trans,
                                struct btrfs_pending_snapshot *pending,
                                u64 *bytes_to_reserve)
{
        struct btrfs_root *root;
        struct btrfs_block_rsv *block_rsv;
        u64 num_bytes;
        int index;

        root = pending->root;
        if (!root->orphan_block_rsv || list_empty(&root->orphan_list))
                return;

        block_rsv = root->orphan_block_rsv;

        /* orphan block reservation for the snapshot */
        num_bytes = block_rsv->size;

        /*
         * after the snapshot is created, COWing tree blocks may use more
         * space than it frees. So we should make sure there is enough
         * reserved space.
         */
        index = trans->transid & 0x1;
        if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) {
                num_bytes += block_rsv->size -
                             (block_rsv->reserved + block_rsv->freed[index]);
        }

        *bytes_to_reserve += num_bytes;
}

void btrfs_orphan_post_snapshot(struct btrfs_trans_handle *trans,
                                struct btrfs_pending_snapshot *pending)
{
        struct btrfs_root *root = pending->root;
        struct btrfs_root *snap = pending->snap;
        struct btrfs_block_rsv *block_rsv;
        u64 num_bytes;
        int index;
        int ret;

        if (!root->orphan_block_rsv || list_empty(&root->orphan_list))
                return;

        /* refill source subvolume's orphan block reservation */
        block_rsv = root->orphan_block_rsv;
        index = trans->transid & 0x1;
        if (block_rsv->reserved + block_rsv->freed[index] < block_rsv->size) {
                num_bytes = block_rsv->size -
                            (block_rsv->reserved + block_rsv->freed[index]);
                ret = btrfs_block_rsv_migrate(&pending->block_rsv,
                                              root->orphan_block_rsv,
                                              num_bytes);
                BUG_ON(ret);
        }

        /* setup orphan block reservation for the snapshot */
        block_rsv = btrfs_alloc_block_rsv(snap);
        BUG_ON(!block_rsv);

        btrfs_add_durable_block_rsv(root->fs_info, block_rsv);
        snap->orphan_block_rsv = block_rsv;

        num_bytes = root->orphan_block_rsv->size;
        ret = btrfs_block_rsv_migrate(&pending->block_rsv,
                                      block_rsv, num_bytes);
        BUG_ON(ret);

#if 0
        /* insert orphan item for the snapshot */
        WARN_ON(!root->orphan_item_inserted);
        ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root,
                                       snap->root_key.objectid);
        BUG_ON(ret);
        snap->orphan_item_inserted = 1;
#endif
}

enum btrfs_orphan_cleanup_state {
        ORPHAN_CLEANUP_STARTED  = 1,
        ORPHAN_CLEANUP_DONE     = 2,
};

/*
 * This is called in transaction commmit time. If there are no orphan
 * files in the subvolume, it removes orphan item and frees block_rsv
 * structure.
 */
void btrfs_orphan_commit_root(struct btrfs_trans_handle *trans,
                              struct btrfs_root *root)
{
        int ret;

        if (!list_empty(&root->orphan_list) ||
            root->orphan_cleanup_state != ORPHAN_CLEANUP_DONE)
                return;

        if (root->orphan_item_inserted &&
            btrfs_root_refs(&root->root_item) > 0) {
                ret = btrfs_del_orphan_item(trans, root->fs_info->tree_root,
                                            root->root_key.objectid);
                BUG_ON(ret);
                root->orphan_item_inserted = 0;
        }

        if (root->orphan_block_rsv) {
                WARN_ON(root->orphan_block_rsv->size > 0);
                btrfs_free_block_rsv(root, root->orphan_block_rsv);
                root->orphan_block_rsv = NULL;
        }
}

/*
 * This creates an orphan entry for the given inode in case something goes
 * wrong in the middle of an unlink/truncate.
 *
 * NOTE: caller of this function should reserve 5 units of metadata for
 *       this function.
 */
int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
{
        struct btrfs_root *root = BTRFS_I(inode)->root;
        struct btrfs_block_rsv *block_rsv = NULL;
        int reserve = 0;
        int insert = 0;
        int ret;

        if (!root->orphan_block_rsv) {
                block_rsv = btrfs_alloc_block_rsv(root);
                BUG_ON(!block_rsv);
        }

        spin_lock(&root->orphan_lock);
        if (!root->orphan_block_rsv) {
                root->orphan_block_rsv = block_rsv;
        } else if (block_rsv) {
                btrfs_free_block_rsv(root, block_rsv);
                block_rsv = NULL;
        }

        if (list_empty(&BTRFS_I(inode)->i_orphan)) {
                list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
#if 0
                /*
                 * For proper ENOSPC handling, we should do orphan
                 * cleanup when mounting. But this introduces backward
                 * compatibility issue.
                 */
                if (!xchg(&root->orphan_item_inserted, 1))
                        insert = 2;
                else
                        insert = 1;
#endif
                insert = 1;
        } else {
                WARN_ON(!BTRFS_I(inode)->orphan_meta_reserved);
        }

        if (!BTRFS_I(inode)->orphan_meta_reserved) {
                BTRFS_I(inode)->orphan_meta_reserved = 1;
                reserve = 1;
        }
        spin_unlock(&root->orphan_lock);

        if (block_rsv)
                btrfs_add_durable_block_rsv(root->fs_info, block_rsv);

        /* grab metadata reservation from transaction handle */
        if (reserve) {
                ret = btrfs_orphan_reserve_metadata(trans, inode);
                BUG_ON(ret);
        }

        /* insert an orphan item to track this unlinked/truncated file */
        if (insert >= 1) {
                ret = btrfs_insert_orphan_item(trans, root, inode->i_ino);
                BUG_ON(ret);
        }

        /* insert an orphan item to track subvolume contains orphan files */
        if (insert >= 2) {
                ret = btrfs_insert_orphan_item(trans, root->fs_info->tree_root,
                                               root->root_key.objectid);
                BUG_ON(ret);
        }
        return 0;
}

/*
 * 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 delete_item = 0;
        int release_rsv = 0;
        int ret = 0;

        spin_lock(&root->orphan_lock);
        if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
                list_del_init(&BTRFS_I(inode)->i_orphan);
                delete_item = 1;
        }

        if (BTRFS_I(inode)->orphan_meta_reserved) {
                BTRFS_I(inode)->orphan_meta_reserved = 0;
                release_rsv = 1;
        }
        spin_unlock(&root->orphan_lock);

        if (trans && delete_item) {
                ret = btrfs_del_orphan_item(trans, root, inode->i_ino);
                BUG_ON(ret);
        }

        if (release_rsv)
                btrfs_orphan_release_metadata(inode);

        return 0;
}

/*
 * 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;

        if (cmpxchg(&root->orphan_cleanup_state, 0, ORPHAN_CLEANUP_STARTED))
                return;

        path = btrfs_alloc_path();
        BUG_ON(!path);
        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.
                 */
                found_key.objectid = found_key.offset;
                found_key.type = BTRFS_INODE_ITEM_KEY;
                found_key.offset = 0;
                inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL);
                BUG_ON(IS_ERR(inode));

                /*
                 * add this inode to the orphan list so btrfs_orphan_del does
                 * the proper thing when we hit it
                 */
                spin_lock(&root->orphan_lock);
                list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
                spin_unlock(&root->orphan_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, 0);
                        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);
        }
        btrfs_free_path(path);

        root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE;

        if (root->orphan_block_rsv)
                btrfs_block_rsv_release(root, root->orphan_block_rsv,
                                        (u64)-1);

        if (root->orphan_block_rsv || root->orphan_item_inserted) {
                trans = btrfs_join_transaction(root, 1);
                btrfs_end_transaction(trans, root);
        }

        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);
}

/*
 * very simple check to peek ahead in the leaf looking for xattrs.  If we
 * don't find any xattrs, we know there can't be any acls.
 *
 * slot is the slot the inode is in, objectid is the objectid of the inode
 */
static noinline int acls_after_inode_item(struct extent_buffer *leaf,
                                          int slot, u64 objectid)
{
        u32 nritems = btrfs_header_nritems(leaf);
        struct btrfs_key found_key;
        int scanned = 0;

        slot++;
        while (slot < nritems) {
                btrfs_item_key_to_cpu(leaf, &found_key, slot);

                /* we found a different objectid, there must not be acls */
                if (found_key.objectid != objectid)
                        return 0;

                /* we found an xattr, assume we've got an acl */
                if (found_key.type == BTRFS_XATTR_ITEM_KEY)
                        return 1;

                /*
                 * we found a key greater than an xattr key, there can't
                 * be any acls later on
                 */
                if (found_key.type > BTRFS_XATTR_ITEM_KEY)
                        return 0;

                slot++;
                scanned++;

                /*
                 * it goes inode, inode backrefs, xattrs, extents,
                 * so if there are a ton of hard links to an inode there can
                 * be a lot of backrefs.  Don't waste time searching too hard,
                 * this is just an optimization
                 */
                if (scanned >= 8)
                        break;
        }
        /* we hit the end of the leaf before we found an xattr or
         * something larger than an xattr.  We have to assume the inode
         * has acls
         */
        return 1;
}

/*
 * read an inode from the btree into the in-memory inode
 */
static 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;
        int maybe_acls;
        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);
        BTRFS_I(inode)->sequence = btrfs_inode_sequence(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;
        BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);

        alloc_group_block = btrfs_inode_block_group(leaf, inode_item);

        /*
         * try to precache a NULL acl entry for files that don't have
         * any xattrs or acls
         */
        maybe_acls = acls_after_inode_item(leaf, path->slots[0], inode->i_ino);
        if (!maybe_acls)
                cache_no_acl(inode);

        BTRFS_I(inode)->block_group = btrfs_find_block_group(root, 0,
                                                alloc_group_block, 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:
                inode->i_fop = &btrfs_dir_file_operations;
                if (root == root->fs_info->tree_root)
                        inode->i_op = &btrfs_dir_ro_inode_operations;
                else
                        inode->i_op = &btrfs_dir_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:
                inode->i_op = &btrfs_special_inode_operations;
                init_special_inode(inode, inode->i_mode, rdev);
                break;
        }

        btrfs_update_iflags(inode);
        return;

make_bad:
        btrfs_free_path(path);
        make_bad_inode(inode);
}

/*
 * given a leaf and an inode, copy the inode fields into the leaf
 */
static void fill_inode_item(struct btrfs_trans_handle *trans,
                            struct extent_buffer *leaf,
                            struct btrfs_inode_item *item,
                            struct 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, 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_set_timespec_sec(leaf, btrfs_inode_atime(item),
                               inode->i_atime.tv_sec);
        btrfs_set_timespec_nsec(leaf, btrfs_inode_atime(item),
                                inode->i_atime.tv_nsec);

        btrfs_set_timespec_sec(leaf, btrfs_inode_mtime(item),
                               inode->i_mtime.tv_sec);
        btrfs_set_timespec_nsec(leaf, btrfs_inode_mtime(item),
                                inode->i_mtime.tv_nsec);

        btrfs_set_timespec_sec(leaf, btrfs_inode_ctime(item),
                               inode->i_ctime.tv_sec);
        btrfs_set_timespec_nsec(leaf, btrfs_inode_ctime(item),
                                inode->i_ctime.tv_nsec);

        btrfs_set_inode_nbytes(leaf, item, inode_get_bytes(inode));
        btrfs_set_inode_generation(leaf, item, BTRFS_I(inode)->generation);
        btrfs_set_inode_sequence(leaf, item, BTRFS_I(inode)->sequence);
        btrfs_set_inode_transid(leaf, item, trans->transid);
        btrfs_set_inode_rdev(leaf, item, inode->i_rdev);
        btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags);
        btrfs_set_inode_block_group(leaf, item, BTRFS_I(inode)->block_group);
}

/*
 * copy everything in the in-memory inode into the btree.
 */
noinline int btrfs_update_inode(struct btrfs_trans_handle *trans,
                                struct btrfs_root *root, struct inode *inode)
{
        struct btrfs_inode_item *inode_item;
        struct btrfs_path *path;
        struct extent_buffer *leaf;
        int ret;

        path = btrfs_alloc_path();
        BUG_ON(!path);
        path->leave_spinning = 1;
        ret = btrfs_lookup_inode(trans, root, path,
                                 &BTRFS_I(inode)->location, 1);
        if (ret) {
                if (ret > 0)
                        ret = -ENOENT;
                goto failed;
        }

        btrfs_unlock_up_safe(path, 1);
        leaf = path->nodes[0];
        inode_item = btrfs_item_ptr(leaf, path->slots[0],
                                  struct btrfs_inode_item);

        fill_inode_item(trans, leaf, inode_item, inode);
        btrfs_mark_buffer_dirty(leaf);
        btrfs_set_inode_last_trans(trans, inode);
        ret = 0;
failed:
        btrfs_free_path(path);
        return ret;
}


/*
 * unlink helper that gets used here in inode.c and in the tree logging
 * recovery code.  It remove a link in a directory with a given name, and
 * also drops the back refs in the inode to the directory
 */
int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
                       struct btrfs_root *root,
                       struct inode *dir, struct inode *inode,
                       const char *name, int name_len)
{
        struct btrfs_path *path;
        int ret = 0;
        struct extent_buffer *leaf;
        struct btrfs_dir_item *di;
        struct btrfs_key key;
        u64 index;

        path = btrfs_alloc_path();
        if (!path) {
                ret = -ENOMEM;
                goto err;
        }

        path->leave_spinning = 1;
        di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino,
                                    name, name_len, -1);
        if (IS_ERR(di)) {
                ret = PTR_ERR(di);
                goto err;
        }
        if (!di) {
                ret = -ENOENT;
                goto err;
        }
        leaf = path->nodes[0];
        btrfs_dir_item_key_to_cpu(leaf, di, &key);
        ret = btrfs_delete_one_dir_name(trans, root, path, di);
        if (ret)
                goto err;
        btrfs_release_path(root, path);

        ret = btrfs_del_inode_ref(trans, root, name, name_len,
                                  inode->i_ino,
                                  dir->i_ino, &index);
        if (ret) {
                printk(KERN_INFO "btrfs failed to delete reference to %.*s, "
                       "inode %lu parent %lu\n", name_len, name,
                       inode->i_ino, dir->i_ino);
                goto err;
        }

        di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino,
                                         index, name, name_len, -1);
        if (IS_ERR(di)) {
                ret = PTR_ERR(di);
                goto err;
        }
        if (!di) {
                ret = -ENOENT;
                goto err;
        }
        ret = btrfs_delete_one_dir_name(trans, root, path, di);
        btrfs_release_path(root, path);

        ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len,
                                         inode, dir->i_ino);
        BUG_ON(ret != 0 && ret != -ENOENT);

        ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len,
                                           dir, index);
        BUG_ON(ret);
err:
        btrfs_free_path(path);
        if (ret)
                goto out;

        btrfs_i_size_write(dir, dir->i_size - name_len * 2);
        inode->i_ctime = dir->i_mtime = dir->i_ctime = CURRENT_TIME;
        btrfs_update_inode(trans, root, dir);
        btrfs_drop_nlink(inode);
        ret = btrfs_update_inode(trans, root, inode);
out:
        return ret;
}

/* helper to check if there is any shared block in the path */
static int check_path_shared(struct btrfs_root *root,
                             struct btrfs_path *path)
{
        struct extent_buffer *eb;
        int level;
        int ret;
        u64 refs = 1;

        for (level = 0; level < BTRFS_MAX_LEVEL; level++) {
                if (!path->nodes[level])
                        break;
                eb = path->nodes[level];
                if (!btrfs_block_can_be_shared(root, eb))
                        continue;
                ret = btrfs_lookup_extent_info(NULL, root, eb->start, eb->len,
                                               &refs, NULL);
                if (refs > 1)
                        return 1;
        }
        return 0;
}

/*
 * helper to start transaction for unlink and rmdir.
 *
 * unlink and rmdir are special in btrfs, they do not always free space.
 * so in enospc case, we should make sure they will free space before
 * allowing them to use the global metadata reservation.
 */
static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir,
                                                       struct dentry *dentry)
{
        struct btrfs_trans_handle *trans;
        struct btrfs_root *root = BTRFS_I(dir)->root;
        struct btrfs_path *path;
        struct btrfs_inode_ref *ref;
        struct btrfs_dir_item *di;
        struct inode *inode = dentry->d_inode;
        u64 index;
        int check_link = 1;
        int err = -ENOSPC;
        int ret;

        trans = btrfs_start_transaction(root, 10);
        if (!IS_ERR(trans) || PTR_ERR(trans) != -ENOSPC)
                return trans;

        if (inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)
                return ERR_PTR(-ENOSPC);

        /* check if there is someone else holds reference */
        if (S_ISDIR(inode->i_mode) && atomic_read(&inode->i_count) > 1)
                return ERR_PTR(-ENOSPC);

        if (atomic_read(&inode->i_count) > 2)
                return ERR_PTR(-ENOSPC);

        if (xchg(&root->fs_info->enospc_unlink, 1))
                return ERR_PTR(-ENOSPC);

        path = btrfs_alloc_path();
        if (!path) {
                root->fs_info->enospc_unlink = 0;
                return ERR_PTR(-ENOMEM);
        }

        trans = btrfs_start_transaction(root, 0);
        if (IS_ERR(trans)) {
                btrfs_free_path(path);
                root->fs_info->enospc_unlink = 0;
                return trans;
        }

        path->skip_locking = 1;
        path->search_commit_root = 1;

        ret = btrfs_lookup_inode(trans, root, path,
                                &BTRFS_I(dir)->location, 0);
        if (ret < 0) {
                err = ret;
                goto out;
        }
        if (ret == 0) {
                if (check_path_shared(root, path))
                        goto out;
        } else {
                check_link = 0;
        }
        btrfs_release_path(root, path);

        ret = btrfs_lookup_inode(trans, root, path,
                                &BTRFS_I(inode)->location, 0);
        if (ret < 0) {
                err = ret;
                goto out;
        }
        if (ret == 0) {
                if (check_path_shared(root, path))
                        goto out;
        } else {
                check_link = 0;
        }
        btrfs_release_path(root, path);

        if (ret == 0 && S_ISREG(inode->i_mode)) {
                ret = btrfs_lookup_file_extent(trans, root, path,
                                               inode->i_ino, (u64)-1, 0);
                if (ret < 0) {
                        err = ret;
                        goto out;
                }
                BUG_ON(ret == 0);
                if (check_path_shared(root, path))
                        goto out;
                btrfs_release_path(root, path);
        }

        if (!check_link) {
                err = 0;
                goto out;
        }

        di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino,
                                dentry->d_name.name, dentry->d_name.len, 0);
        if (IS_ERR(di)) {
                err = PTR_ERR(di);
                goto out;
        }
        if (di) {
                if (check_path_shared(root, path))
                        goto out;
        } else {
                err = 0;
                goto out;
        }
        btrfs_release_path(root, path);

        ref = btrfs_lookup_inode_ref(trans, root, path,
                                dentry->d_name.name, dentry->d_name.len,
                                inode->i_ino, dir->i_ino, 0);
        if (IS_ERR(ref)) {
                err = PTR_ERR(ref);
                goto out;
        }
        BUG_ON(!ref);
        if (check_path_shared(root, path))
                goto out;
        index = btrfs_inode_ref_index(path->nodes[0], ref);
        btrfs_release_path(root, path);

        di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino, index,
                                dentry->d_name.name, dentry->d_name.len, 0);
        if (IS_ERR(di)) {
                err = PTR_ERR(di);
                goto out;
        }
        BUG_ON(ret == -ENOENT);
        if (check_path_shared(root, path))
                goto out;

        err = 0;
out:
        btrfs_free_path(path);
        if (err) {
                btrfs_end_transaction(trans, root);
                root->fs_info->enospc_unlink = 0;
                return ERR_PTR(err);
        }

        trans->block_rsv = &root->fs_info->global_block_rsv;
        return trans;
}

static void __unlink_end_trans(struct btrfs_trans_handle *trans,
                               struct btrfs_root *root)
{
        if (trans->block_rsv == &root->fs_info->global_block_rsv) {
                BUG_ON(!root->fs_info->enospc_unlink);
                root->fs_info->enospc_unlink = 0;
        }
        btrfs_end_transaction_throttle(trans, root);
}

static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
{
        struct btrfs_root *root = BTRFS_I(dir)->root;
        struct btrfs_trans_handle *trans;
        struct inode *inode = dentry->d_inode;
        int ret;
        unsigned long nr = 0;

        trans = __unlink_start_trans(dir, dentry);
        if (IS_ERR(trans))
                return PTR_ERR(trans);

        btrfs_set_trans_block_group(trans, dir);

        btrfs_record_unlink_dir(trans, dir, dentry->d_inode, 0);

        ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
                                 dentry->d_name.name, dentry->d_name.len);
        BUG_ON(ret);

        if (inode->i_nlink == 0) {
                ret = btrfs_orphan_add(trans, inode);
                BUG_ON(ret);
        }

        nr = trans->blocks_used;
        __unlink_end_trans(trans, root);
        btrfs_btree_balance_dirty(root, nr);
        return ret;
}

int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
                        struct btrfs_root *root,
                        struct inode *dir, u64 objectid,
                        const char *name, int name_len)
{
        struct btrfs_path *path;
        struct extent_buffer *leaf;
        struct btrfs_dir_item *di;
        struct btrfs_key key;
        u64 index;
        int ret;

        path = btrfs_alloc_path();
        if (!path)
                return -ENOMEM;

        di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino,
                                   name, name_len, -1);
        BUG_ON(!di || IS_ERR(di));

        leaf = path->nodes[0];
        btrfs_dir_item_key_to_cpu(leaf, di, &key);
        WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid);
        ret = btrfs_delete_one_dir_name(trans, root, path, di);
        BUG_ON(ret);
        btrfs_release_path(root, path);

        ret = btrfs_del_root_ref(trans, root->fs_info->tree_root,
                                 objectid, root->root_key.objectid,
                                 dir->i_ino, &index, name, name_len);
        if (ret < 0) {
                BUG_ON(ret != -ENOENT);
                di = btrfs_search_dir_index_item(root, path, dir->i_ino,
                                                 name, name_len);
                BUG_ON(!di || IS_ERR(di));

                leaf = path->nodes[0];
                btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
                btrfs_release_path(root, path);
                index = key.offset;
        }

        di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino,
                                         index, name, name_len, -1);
        BUG_ON(!di || IS_ERR(di));

        leaf = path->nodes[0];
        btrfs_dir_item_key_to_cpu(leaf, di, &key);
        WARN_ON(key.type != BTRFS_ROOT_ITEM_KEY || key.objectid != objectid);
        ret = btrfs_delete_one_dir_name(trans, root, path, di);
        BUG_ON(ret);
        btrfs_release_path(root, path);

        btrfs_i_size_write(dir, dir->i_size - name_len * 2);
        dir->i_mtime = dir->i_ctime = CURRENT_TIME;
        ret = btrfs_update_inode(trans, root, dir);
        BUG_ON(ret);
        dir->i_sb->s_dirt = 1;

        btrfs_free_path(path);
        return 0;
}

static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
{
        struct inode *inode = dentry->d_inode;
        int err = 0;
        struct btrfs_root *root = BTRFS_I(dir)->root;
        struct btrfs_trans_handle *trans;
        unsigned long nr = 0;

        if (inode->i_size > BTRFS_EMPTY_DIR_SIZE ||
            inode->i_ino == BTRFS_FIRST_FREE_OBJECTID)
                return -ENOTEMPTY;

        trans = __unlink_start_trans(dir, dentry);
        if (IS_ERR(trans))
                return PTR_ERR(trans);

        btrfs_set_trans_block_group(trans, dir);

        if (unlikely(inode->i_ino == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
                err = btrfs_unlink_subvol(trans, root, dir,
                                          BTRFS_I(inode)->location.objectid,
                                          dentry->d_name.name,
                                          dentry->d_name.len);
                goto out;
        }

        err = btrfs_orphan_add(trans, inode);
        if (err)
                goto out;

        /* now the directory is empty */
        err = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
                                 dentry->d_name.name, dentry->d_name.len);
        if (!err)
                btrfs_i_size_write(inode, 0);
out:
        nr = trans->blocks_used;
        __unlink_end_trans(trans, root);
        btrfs_btree_balance_dirty(root, nr);

        return err;
}

#if 0
/*
 * when truncating bytes in a file, it is possible to avoid reading
 * the leaves that contain only checksum items.  This can be the
 * majority of the IO required to delete a large file, but it must
 * be done carefully.
 *
 * The keys in the level just above the leaves are checked to make sure
 * the lowest key in a given leaf is a csum key, and starts at an offset
 * after the new  size.
 *