// SPDX-License-Identifier: GPL-2.0
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 */

#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/writeback.h>
#include <linux/compat.h>
#include <linux/xattr.h>
#include <linux/posix_acl.h>
#include <linux/falloc.h>
#include <linux/slab.h>
#include <linux/ratelimit.h>
#include <linux/btrfs.h>
#include <linux/blkdev.h>
#include <linux/posix_acl_xattr.h>
#include <linux/uio.h>
#include <linux/magic.h>
#include <linux/iversion.h>
#include <linux/swap.h>
#include <linux/sched/mm.h>
#include <asm/unaligned.h>
#include "misc.h"
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "print-tree.h"
#include "ordered-data.h"
#include "xattr.h"
#include "tree-log.h"
#include "volumes.h"
#include "compression.h"
#include "locking.h"
#include "free-space-cache.h"
#include "inode-map.h"
#include "props.h"
#include "qgroup.h"
#include "delalloc-space.h"
#include "block-group.h"

struct btrfs_iget_args {
	struct btrfs_key *location;
	struct btrfs_root *root;
};

struct btrfs_dio_data {
	u64 reserve;
	u64 unsubmitted_oe_range_start;
	u64 unsubmitted_oe_range_end;
	int overwrite;
};

static const struct inode_operations btrfs_dir_inode_operations;
static const struct inode_operations btrfs_symlink_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 file_operations btrfs_dir_file_operations;
static const 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_path_cachep;
struct kmem_cache *btrfs_free_space_cachep;
struct kmem_cache *btrfs_free_space_bitmap_cachep;

static int btrfs_setsize(struct inode *inode, struct iattr *attr);
static int btrfs_truncate(struct inode *inode, bool skip_writeback);
static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent);
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 struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
				       u64 orig_start, u64 block_start,
				       u64 block_len, u64 orig_block_len,
				       u64 ram_bytes, int compress_type,
				       int type);

static void __endio_write_update_ordered(struct inode *inode,
					 const u64 offset, const u64 bytes,
					 const bool uptodate);

/*
 * Cleanup all submitted ordered extents in specified range to handle errors
 * from the btrfs_run_delalloc_range() callback.
 *
 * NOTE: caller must ensure that when an error happens, it can not call
 * extent_clear_unlock_delalloc() to clear both the bits EXTENT_DO_ACCOUNTING
 * and EXTENT_DELALLOC simultaneously, because that causes the reserved metadata
 * to be released, which we want to happen only when finishing the ordered
 * extent (btrfs_finish_ordered_io()).
 */
static inline void btrfs_cleanup_ordered_extents(struct inode *inode,
						 struct page *locked_page,
						 u64 offset, u64 bytes)
{
	unsigned long index = offset >> PAGE_SHIFT;
	unsigned long end_index = (offset + bytes - 1) >> PAGE_SHIFT;
	u64 page_start = page_offset(locked_page);
	u64 page_end = page_start + PAGE_SIZE - 1;

	struct page *page;

	while (index <= end_index) {
		page = find_get_page(inode->i_mapping, index);
		index++;
		if (!page)
			continue;
		ClearPagePrivate2(page);
		put_page(page);
	}

	/*
	 * In case this page belongs to the delalloc range being instantiated
	 * then skip it, since the first page of a range is going to be
	 * properly cleaned up by the caller of run_delalloc_range
	 */
	if (page_start >= offset && page_end <= (offset + bytes - 1)) {
		offset += PAGE_SIZE;
		bytes -= PAGE_SIZE;
	}

	return __endio_write_update_ordered(inode, offset, bytes, false);
}

static int btrfs_dirty_inode(struct inode *inode);

#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
void btrfs_test_inode_set_ops(struct inode *inode)
{
	BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
}
#endif

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

	err = btrfs_init_acl(trans, inode, dir);
	if (!err)
		err = btrfs_xattr_security_init(trans, inode, dir, qstr);
	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 int insert_inline_extent(struct btrfs_trans_handle *trans,
				struct btrfs_path *path, int extent_inserted,
				struct btrfs_root *root, struct inode *inode,
				u64 start, size_t size, size_t compressed_size,
				int compress_type,
				struct page **compressed_pages)
{
	struct extent_buffer *leaf;
	struct page *page = NULL;
	char *kaddr;
	unsigned long ptr;
	struct btrfs_file_extent_item *ei;
	int ret;
	size_t cur_size = size;
	unsigned long offset;

	ASSERT((compressed_size > 0 && compressed_pages) ||
	       (compressed_size == 0 && !compressed_pages));

	if (compressed_size && compressed_pages)
		cur_size = compressed_size;

	inode_add_bytes(inode, size);

	if (!extent_inserted) {
		struct btrfs_key key;
		size_t datasize;

		key.objectid = btrfs_ino(BTRFS_I(inode));
		key.offset = start;
		key.type = BTRFS_EXTENT_DATA_KEY;

		datasize = btrfs_file_extent_calc_inline_size(cur_size);
		path->leave_spinning = 1;
		ret = btrfs_insert_empty_item(trans, root, path, &key,
					      datasize);
		if (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 (compress_type != BTRFS_COMPRESS_NONE) {
		struct page *cpage;
		int i = 0;
		while (compressed_size > 0) {
			cpage = compressed_pages[i];
			cur_size = min_t(unsigned long, compressed_size,
				       PAGE_SIZE);

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

			i++;
			ptr += cur_size;
			compressed_size -= cur_size;
		}
		btrfs_set_file_extent_compression(leaf, ei,
						  compress_type);
	} else {
		page = find_get_page(inode->i_mapping,
				     start >> PAGE_SHIFT);
		btrfs_set_file_extent_compression(leaf, ei, 0);
		kaddr = kmap_atomic(page);
		offset = offset_in_page(start);
		write_extent_buffer(leaf, kaddr + offset, ptr, size);
		kunmap_atomic(kaddr);
		put_page(page);
	}
	btrfs_mark_buffer_dirty(leaf);
	btrfs_release_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;
	ret = btrfs_update_inode(trans, root, inode);

fail:
	return ret;
}


/*
 * 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 inode *inode, u64 start,
					  u64 end, size_t compressed_size,
					  int compress_type,
					  struct page **compressed_pages)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_trans_handle *trans;
	u64 isize = i_size_read(inode);
	u64 actual_end = min(end + 1, isize);
	u64 inline_len = actual_end - start;
	u64 aligned_end = ALIGN(end, fs_info->sectorsize);
	u64 data_len = inline_len;
	int ret;
	struct btrfs_path *path;
	int extent_inserted = 0;
	u32 extent_item_size;

	if (compressed_size)
		data_len = compressed_size;

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

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

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		btrfs_free_path(path);
		return PTR_ERR(trans);
	}
	trans->block_rsv = &BTRFS_I(inode)->block_rsv;

	if (compressed_size && compressed_pages)
		extent_item_size = btrfs_file_extent_calc_inline_size(
		   compressed_size);
	else
		extent_item_size = btrfs_file_extent_calc_inline_size(
		    inline_len);

	ret = __btrfs_drop_extents(trans, root, inode, path,
				   start, aligned_end, NULL,
				   1, 1, extent_item_size, &extent_inserted);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out;
	}

	if (isize > actual_end)
		inline_len = min_t(u64, isize, actual_end);
	ret = insert_inline_extent(trans, path, extent_inserted,
				   root, inode, start,
				   inline_len, compressed_size,
				   compress_type, compressed_pages);
	if (ret && ret != -ENOSPC) {
		btrfs_abort_transaction(trans, ret);
		goto out;
	} else if (ret == -ENOSPC) {
		ret = 1;
		goto out;
	}

	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
	btrfs_drop_extent_cache(BTRFS_I(inode), start, aligned_end - 1, 0);
out:
	/*
	 * Don't forget to free the reserved space, as for inlined extent
	 * it won't count as data extent, free them directly here.
	 * And at reserve time, it's always aligned to page size, so
	 * just free one page here.
	 */
	btrfs_qgroup_free_data(inode, NULL, 0, PAGE_SIZE);
	btrfs_free_path(path);
	btrfs_end_transaction(trans);
	return ret;
}

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

struct async_chunk {
	struct inode *inode;
	struct page *locked_page;
	u64 start;
	u64 end;
	unsigned int write_flags;
	struct list_head extents;
	struct cgroup_subsys_state *blkcg_css;
	struct btrfs_work work;
	atomic_t *pending;
};

struct async_cow {
	/* Number of chunks in flight; must be first in the structure */
	atomic_t num_chunks;
	struct async_chunk chunks[];
};

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

	async_extent = kmalloc(sizeof(*async_extent), GFP_NOFS);
	BUG_ON(!async_extent); /* -ENOMEM */
	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;
	async_extent->compress_type = compress_type;
	list_add_tail(&async_extent->list, &cow->extents);
	return 0;
}

/*
 * Check if the inode has flags compatible with compression
 */
static inline bool inode_can_compress(struct inode *inode)
{
	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW ||
	    BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)
		return false;
	return true;
}

/*
 * Check if the inode needs to be submitted to compression, based on mount
 * options, defragmentation, properties or heuristics.
 */
static inline int inode_need_compress(struct inode *inode, u64 start, u64 end)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);

	if (!inode_can_compress(inode)) {
		WARN(IS_ENABLED(CONFIG_BTRFS_DEBUG),
			KERN_ERR "BTRFS: unexpected compression for ino %llu\n",
			btrfs_ino(BTRFS_I(inode)));
		return 0;
	}
	/* force compress */
	if (btrfs_test_opt(fs_info, FORCE_COMPRESS))
		return 1;
	/* defrag ioctl */
	if (BTRFS_I(inode)->defrag_compress)
		return 1;
	/* bad compression ratios */
	if (BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS)
		return 0;
	if (btrfs_test_opt(fs_info, COMPRESS) ||
	    BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS ||
	    BTRFS_I(inode)->prop_compress)
		return btrfs_compress_heuristic(inode, start, end);
	return 0;
}

static inline void inode_should_defrag(struct btrfs_inode *inode,
		u64 start, u64 end, u64 num_bytes, u64 small_write)
{
	/* If this is a small write inside eof, kick off a defrag */
	if (num_bytes < small_write &&
	    (start > 0 || end + 1 < inode->disk_i_size))
		btrfs_add_inode_defrag(NULL, inode);
}

/*
 * 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 the flusher thread sent them
 * down.
 */
static noinline int compress_file_range(struct async_chunk *async_chunk)
{
	struct inode *inode = async_chunk->inode;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	u64 blocksize = fs_info->sectorsize;
	u64 start = async_chunk->start;
	u64 end = async_chunk->end;
	u64 actual_end;
	u64 i_size;
	int ret = 0;
	struct page **pages = NULL;
	unsigned long nr_pages;
	unsigned long total_compressed = 0;
	unsigned long total_in = 0;
	int i;
	int will_compress;
	int compress_type = fs_info->compress_type;
	int compressed_extents = 0;
	int redirty = 0;

	inode_should_defrag(BTRFS_I(inode), start, end, end - start + 1,
			SZ_16K);

	/*
	 * We need to save i_size before now because it could change in between
	 * us evaluating the size and assigning it.  This is because we lock and
	 * unlock the page in truncate and fallocate, and then modify the i_size
	 * later on.
	 *
	 * The barriers are to emulate READ_ONCE, remove that once i_size_read
	 * does that for us.
	 */
	barrier();
	i_size = i_size_read(inode);
	barrier();
	actual_end = min_t(u64, i_size, end + 1);
again:
	will_compress = 0;
	nr_pages = (end >> PAGE_SHIFT) - (start >> PAGE_SHIFT) + 1;
	BUILD_BUG_ON((BTRFS_MAX_COMPRESSED % PAGE_SIZE) != 0);
	nr_pages = min_t(unsigned long, nr_pages,
			BTRFS_MAX_COMPRESSED / PAGE_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;

	/*
	 * skip compression for a small file range(<=blocksize) that
	 * isn't an inline extent, since it doesn't save disk space at all.
	 */
	if (total_compressed <= blocksize &&
	   (start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
		goto cleanup_and_bail_uncompressed;

	total_compressed = min_t(unsigned long, total_compressed,
			BTRFS_MAX_UNCOMPRESSED);
	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 (inode_need_compress(inode, start, end)) {
		WARN_ON(pages);
		pages = kcalloc(nr_pages, sizeof(struct page *), GFP_NOFS);
		if (!pages) {
			/* just bail out to the uncompressed code */
			nr_pages = 0;
			goto cont;
		}

		if (BTRFS_I(inode)->defrag_compress)
			compress_type = BTRFS_I(inode)->defrag_compress;
		else if (BTRFS_I(inode)->prop_compress)
			compress_type = BTRFS_I(inode)->prop_compress;

		/*
		 * we need to call clear_page_dirty_for_io on each
		 * page in the range.  Otherwise applications with the file
		 * mmap'd can wander in and change the page contents while
		 * we are compressing them.
		 *
		 * If the compression fails for any reason, we set the pages
		 * dirty again later on.
		 *
		 * Note that the remaining part is redirtied, the start pointer
		 * has moved, the end is the original one.
		 */
		if (!redirty) {
			extent_range_clear_dirty_for_io(inode, start, end);
			redirty = 1;
		}

		/* Compression level is applied here and only here */
		ret = btrfs_compress_pages(
			compress_type | (fs_info->compress_level << 4),
					   inode->i_mapping, start,
					   pages,
					   &nr_pages,
					   &total_in,
					   &total_compressed);

		if (!ret) {
			unsigned long offset = offset_in_page(total_compressed);
			struct page *page = pages[nr_pages - 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);
				memset(kaddr + offset, 0,
				       PAGE_SIZE - offset);
				kunmap_atomic(kaddr);
			}
			will_compress = 1;
		}
	}
cont:
	if (start == 0) {
		/* lets try to make an inline extent */
		if (ret || total_in < actual_end) {
			/* we didn't compress the entire range, try
			 * to make an uncompressed inline extent.
			 */
			ret = cow_file_range_inline(inode, start, end, 0,
						    BTRFS_COMPRESS_NONE, NULL);
		} else {
			/* try making a compressed inline extent */
			ret = cow_file_range_inline(inode, start, end,
						    total_compressed,
						    compress_type, pages);
		}
		if (ret <= 0) {
			unsigned long clear_flags = EXTENT_DELALLOC |
				EXTENT_DELALLOC_NEW | EXTENT_DEFRAG |
				EXTENT_DO_ACCOUNTING;
			unsigned long page_error_op;

			page_error_op = ret < 0 ? PAGE_SET_ERROR : 0;

			/*
			 * inline extent creation worked or returned error,
			 * we don't need to create any more async work items.
			 * Unlock and free up our temp pages.
			 *
			 * We use DO_ACCOUNTING here because we need the
			 * delalloc_release_metadata to be done _after_ we drop
			 * our outstanding extent for clearing delalloc for this
			 * range.
			 */
			extent_clear_unlock_delalloc(inode, start, end, NULL,
						     clear_flags,
						     PAGE_UNLOCK |
						     PAGE_CLEAR_DIRTY |
						     PAGE_SET_WRITEBACK |
						     page_error_op |
						     PAGE_END_WRITEBACK);

			for (i = 0; i < nr_pages; i++) {
				WARN_ON(pages[i]->mapping);
				put_page(pages[i]);
			}
			kfree(pages);

			return 0;
		}
	}

	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 = ALIGN(total_compressed, blocksize);

		/*
		 * one last check to make sure the compression is really a
		 * win, compare the page count read with the blocks on disk,
		 * compression must free at least one sector size
		 */
		total_in = ALIGN(total_in, PAGE_SIZE);
		if (total_compressed + blocksize <= total_in) {
			compressed_extents++;

			/*
			 * 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_chunk, start, total_in,
					total_compressed, pages, nr_pages,
					compress_type);

			if (start + total_in < end) {
				start += total_in;
				pages = NULL;
				cond_resched();
				goto again;
			}
			return compressed_extents;
		}
	}
	if (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; i++) {
			WARN_ON(pages[i]->mapping);
			put_page(pages[i]);
		}
		kfree(pages);
		pages = NULL;
		total_compressed = 0;
		nr_pages = 0;

		/* flag the file so we don't compress in the future */
		if (!btrfs_test_opt(fs_info, FORCE_COMPRESS) &&
		    !(BTRFS_I(inode)->prop_compress)) {
			BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
		}
	}
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 (async_chunk->locked_page &&
	    (page_offset(async_chunk->locked_page) >= start &&
	     page_offset(async_chunk->locked_page)) <= end) {
		__set_page_dirty_nobuffers(async_chunk->locked_page);
		/* unlocked later on in the async handlers */
	}

	if (redirty)
		extent_range_redirty_for_io(inode, start, end);
	add_async_extent(async_chunk, start, end - start + 1, 0, NULL, 0,
			 BTRFS_COMPRESS_NONE);
	compressed_extents++;

	return compressed_extents;
}

static void free_async_extent_pages(struct async_extent *async_extent)
{
	int i;

	if (!async_extent->pages)
		return;

	for (i = 0; i < async_extent->nr_pages; i++) {
		WARN_ON(async_extent->pages[i]->mapping);
		put_page(async_extent->pages[i]);
	}
	kfree(async_extent->pages);
	async_extent->nr_pages = 0;
	async_extent->pages = NULL;
}

/*
 * 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 void submit_compressed_extents(struct async_chunk *async_chunk)
{
	struct inode *inode = async_chunk->inode;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct async_extent *async_extent;
	u64 alloc_hint = 0;
	struct btrfs_key ins;
	struct extent_map *em;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	int ret = 0;

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

retry:
		lock_extent(io_tree, async_extent->start,
			    async_extent->start + async_extent->ram_size - 1);
		/* did the compression code fall back to uncompressed IO? */
		if (!async_extent->pages) {
			int page_started = 0;
			unsigned long nr_written = 0;

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

			/* JDM XXX */

			/*
			 * 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(inode,
						  async_extent->start,
						  async_extent->start +
						  async_extent->ram_size - 1,
						  WB_SYNC_ALL);
			else if (ret && async_chunk->locked_page)
				unlock_page(async_chunk->locked_page);
			kfree(async_extent);
			cond_resched();
			continue;
		}

		ret = btrfs_reserve_extent(root, async_extent->ram_size,
					   async_extent->compressed_size,
					   async_extent->compressed_size,
					   0, alloc_hint, &ins, 1, 1);
		if (ret) {
			free_async_extent_pages(async_extent);

			if (ret == -ENOSPC) {
				unlock_extent(io_tree, async_extent->start,
					      async_extent->start +
					      async_extent->ram_size - 1);

				/*
				 * we need to redirty the pages if we decide to
				 * fallback to uncompressed IO, otherwise we
				 * will not submit these pages down to lower
				 * layers.
				 */
				extent_range_redirty_for_io(inode,
						async_extent->start,
						async_extent->start +
						async_extent->ram_size - 1);

				goto retry;
			}
			goto out_free;
		}
		/*
		 * here we're doing allocation and writeback of the
		 * compressed pages
		 */
		em = create_io_em(inode, async_extent->start,
				  async_extent->ram_size, /* len */
				  async_extent->start, /* orig_start */
				  ins.objectid, /* block_start */
				  ins.offset, /* block_len */
				  ins.offset, /* orig_block_len */
				  async_extent->ram_size, /* ram_bytes */
				  async_extent->compress_type,
				  BTRFS_ORDERED_COMPRESSED);
		if (IS_ERR(em))
			/* ret value is not necessary due to void function */
			goto out_free_reserve;
		free_extent_map(em);

		ret = btrfs_add_ordered_extent_compress(inode,
						async_extent->start,
						ins.objectid,
						async_extent->ram_size,
						ins.offset,
						BTRFS_ORDERED_COMPRESSED,
						async_extent->compress_type);
		if (ret) {
			btrfs_drop_extent_cache(BTRFS_I(inode),
						async_extent->start,
						async_extent->start +
						async_extent->ram_size - 1, 0);
			goto out_free_reserve;
		}
		btrfs_dec_block_group_reservations(fs_info, ins.objectid);

		/*
		 * clear dirty, set writeback and unlock the pages.
		 */
		extent_clear_unlock_delalloc(inode, async_extent->start,
				async_extent->start +
				async_extent->ram_size - 1,
				NULL, EXTENT_LOCKED | EXTENT_DELALLOC,
				PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
				PAGE_SET_WRITEBACK);
		if (btrfs_submit_compressed_write(inode,
				    async_extent->start,
				    async_extent->ram_size,
				    ins.objectid,
				    ins.offset, async_extent->pages,
				    async_extent->nr_pages,
				    async_chunk->write_flags,
				    async_chunk->blkcg_css)) {
			struct page *p = async_extent->pages[0];
			const u64 start = async_extent->start;
			const u64 end = start + async_extent->ram_size - 1;

			p->mapping = inode->i_mapping;
			btrfs_writepage_endio_finish_ordered(p, start, end, 0);

			p->mapping = NULL;
			extent_clear_unlock_delalloc(inode, start, end,
						     NULL, 0,
						     PAGE_END_WRITEBACK |
						     PAGE_SET_ERROR);
			free_async_extent_pages(async_extent);
		}
		alloc_hint = ins.objectid + ins.offset;
		kfree(async_extent);
		cond_resched();
	}
	return;
out_free_reserve:
	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
out_free:
	extent_clear_unlock_delalloc(inode, async_extent->start,
				     async_extent->start +
				     async_extent->ram_size - 1,
				     NULL, EXTENT_LOCKED | EXTENT_DELALLOC |
				     EXTENT_DELALLOC_NEW |
				     EXTENT_DEFRAG | EXTENT_DO_ACCOUNTING,
				     PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
				     PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK |
				     PAGE_SET_ERROR);
	free_async_extent_pages(async_extent);
	kfree(async_extent);
	goto again;
}

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_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	u64 alloc_hint = 0;
	u64 num_bytes;
	unsigned long ram_size;
	u64 cur_alloc_size = 0;
	u64 blocksize = fs_info->sectorsize;
	struct btrfs_key ins;
	struct extent_map *em;
	unsigned clear_bits;
	unsigned long page_ops;
	bool extent_reserved = false;
	int ret = 0;

	if (btrfs_is_free_space_inode(BTRFS_I(inode))) {
		WARN_ON_ONCE(1);
		ret = -EINVAL;
		goto out_unlock;
	}

	num_bytes = ALIGN(end - start + 1, blocksize);
	num_bytes = max(blocksize,  num_bytes);
	ASSERT(num_bytes <= btrfs_super_total_bytes(fs_info->super_copy));

	inode_should_defrag(BTRFS_I(inode), start, end, num_bytes, SZ_64K);

	if (start == 0) {
		/* lets try to make an inline extent */
		ret = cow_file_range_inline(inode, start, end, 0,
					    BTRFS_COMPRESS_NONE, NULL);
		if (ret == 0) {
			/*
			 * We use DO_ACCOUNTING here because we need the
			 * delalloc_release_metadata to be run _after_ we drop
			 * our outstanding extent for clearing delalloc for this
			 * range.
			 */
			extent_clear_unlock_delalloc(inode, start, end, NULL,
				     EXTENT_LOCKED | EXTENT_DELALLOC |
				     EXTENT_DELALLOC_NEW | EXTENT_DEFRAG |
				     EXTENT_DO_ACCOUNTING, PAGE_UNLOCK |
				     PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK |
				     PAGE_END_WRITEBACK);
			*nr_written = *nr_written +
			     (end - start + PAGE_SIZE) / PAGE_SIZE;
			*page_started = 1;
			goto out;
		} else if (ret < 0) {
			goto out_unlock;
		}
	}

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

	while (num_bytes > 0) {
		cur_alloc_size = num_bytes;
		ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size,
					   fs_info->sectorsize, 0, alloc_hint,
					   &ins, 1, 1);
		if (ret < 0)
			goto out_unlock;
		cur_alloc_size = ins.offset;
		extent_reserved = true;

		ram_size = ins.offset;
		em = create_io_em(inode, start, ins.offset, /* len */
				  start, /* orig_start */
				  ins.objectid, /* block_start */
				  ins.offset, /* block_len */
				  ins.offset, /* orig_block_len */
				  ram_size, /* ram_bytes */
				  BTRFS_COMPRESS_NONE, /* compress_type */
				  BTRFS_ORDERED_REGULAR /* type */);
		if (IS_ERR(em)) {
			ret = PTR_ERR(em);
			goto out_reserve;
		}
		free_extent_map(em);

		ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
					       ram_size, cur_alloc_size, 0);
		if (ret)
			goto out_drop_extent_cache;

		if (root->root_key.objectid ==
		    BTRFS_DATA_RELOC_TREE_OBJECTID) {
			ret = btrfs_reloc_clone_csums(inode, start,
						      cur_alloc_size);
			/*
			 * Only drop cache here, and process as normal.
			 *
			 * We must not allow extent_clear_unlock_delalloc()
			 * at out_unlock label to free meta of this ordered
			 * extent, as its meta should be freed by
			 * btrfs_finish_ordered_io().
			 *
			 * So we must continue until @start is increased to
			 * skip current ordered extent.
			 */
			if (ret)
				btrfs_drop_extent_cache(BTRFS_I(inode), start,
						start + ram_size - 1, 0);
		}

		btrfs_dec_block_group_reservations(fs_info, ins.objectid);

		/* 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
		 */
		page_ops = unlock ? PAGE_UNLOCK : 0;
		page_ops |= PAGE_SET_PRIVATE2;

		extent_clear_unlock_delalloc(inode, start,
					     start + ram_size - 1,
					     locked_page,
					     EXTENT_LOCKED | EXTENT_DELALLOC,
					     page_ops);
		if (num_bytes < cur_alloc_size)
			num_bytes = 0;
		else
			num_bytes -= cur_alloc_size;
		alloc_hint = ins.objectid + ins.offset;
		start += cur_alloc_size;
		extent_reserved = false;

		/*
		 * btrfs_reloc_clone_csums() error, since start is increased
		 * extent_clear_unlock_delalloc() at out_unlock label won't
		 * free metadata of current ordered extent, we're OK to exit.
		 */
		if (ret)
			goto out_unlock;
	}
out:
	return ret;

out_drop_extent_cache:
	btrfs_drop_extent_cache(BTRFS_I(inode), start, start + ram_size - 1, 0);
out_reserve:
	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
out_unlock:
	clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
		EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV;
	page_ops = PAGE_UNLOCK | PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK |
		PAGE_END_WRITEBACK;
	/*
	 * If we reserved an extent for our delalloc range (or a subrange) and
	 * failed to create the respective ordered extent, then it means that
	 * when we reserved the extent we decremented the extent's size from
	 * the data space_info's bytes_may_use counter and incremented the
	 * space_info's bytes_reserved counter by the same amount. We must make
	 * sure extent_clear_unlock_delalloc() does not try to decrement again
	 * the data space_info's bytes_may_use counter, therefore we do not pass
	 * it the flag EXTENT_CLEAR_DATA_RESV.
	 */
	if (extent_reserved) {
		extent_clear_unlock_delalloc(inode, start,
					     start + cur_alloc_size,
					     locked_page,
					     clear_bits,
					     page_ops);
		start += cur_alloc_size;
		if (start >= end)
			goto out;
	}
	extent_clear_unlock_delalloc(inode, start, end, locked_page,
				     clear_bits | EXTENT_CLEAR_DATA_RESV,
				     page_ops);
	goto out;
}

/*
 * work queue call back to started compression on a file and pages
 */
static noinline void async_cow_start(struct btrfs_work *work)
{
	struct async_chunk *async_chunk;
	int compressed_extents;

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

	compressed_extents = compress_file_range(async_chunk);
	if (compressed_extents == 0) {
		btrfs_add_delayed_iput(async_chunk->inode);
		async_chunk->inode = NULL;
	}
}

/*
 * work queue call back to submit previously compressed pages
 */
static noinline void async_cow_submit(struct btrfs_work *work)
{
	struct async_chunk *async_chunk = container_of(work, struct async_chunk,
						     work);
	struct btrfs_fs_info *fs_info = btrfs_work_owner(work);
	unsigned long nr_pages;

	nr_pages = (async_chunk->end - async_chunk->start + PAGE_SIZE) >>
		PAGE_SHIFT;

	/* atomic_sub_return implies a barrier */
	if (atomic_sub_return(nr_pages, &fs_info->async_delalloc_pages) <
	    5 * SZ_1M)
		cond_wake_up_nomb(&fs_info->async_submit_wait);

	/*
	 * ->inode could be NULL if async_chunk_start has failed to compress,
	 * in which case we don't have anything to submit, yet we need to
	 * always adjust ->async_delalloc_pages as its paired with the init
	 * happening in cow_file_range_async
	 */
	if (async_chunk->inode)
		submit_compressed_extents(async_chunk);
}

static noinline void async_cow_free(struct btrfs_work *work)
{
	struct async_chunk *async_chunk;

	async_chunk = container_of(work, struct async_chunk, work);
	if (async_chunk->inode)
		btrfs_add_delayed_iput(async_chunk->inode);
	if (async_chunk->blkcg_css)
		css_put(async_chunk->blkcg_css);
	/*
	 * Since the pointer to 'pending' is at the beginning of the array of
	 * async_chunk's, freeing it ensures the whole array has been freed.
	 */
	if (atomic_dec_and_test(async_chunk->pending))
		kvfree(async_chunk->pending);
}

static int cow_file_range_async(struct inode *inode,
				struct writeback_control *wbc,
				struct page *locked_page,
				u64 start, u64 end, int *page_started,
				unsigned long *nr_written)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct cgroup_subsys_state *blkcg_css = wbc_blkcg_css(wbc);
	struct async_cow *ctx;
	struct async_chunk *async_chunk;
	unsigned long nr_pages;
	u64 cur_end;
	u64 num_chunks = DIV_ROUND_UP(end - start, SZ_512K);
	int i;
	bool should_compress;
	unsigned nofs_flag;
	const unsigned int write_flags = wbc_to_write_flags(wbc);

	unlock_extent(&BTRFS_I(inode)->io_tree, start, end);

	if (BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS &&
	    !btrfs_test_opt(fs_info, FORCE_COMPRESS)) {
		num_chunks = 1;
		should_compress = false;
	} else {
		should_compress = true;
	}

	nofs_flag = memalloc_nofs_save();
	ctx = kvmalloc(struct_size(ctx, chunks, num_chunks), GFP_KERNEL);
	memalloc_nofs_restore(nofs_flag);

	if (!ctx) {
		unsigned clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC |
			EXTENT_DELALLOC_NEW | EXTENT_DEFRAG |
			EXTENT_DO_ACCOUNTING;
		unsigned long page_ops = PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
			PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK |
			PAGE_SET_ERROR;

		extent_clear_unlock_delalloc(inode, start, end, locked_page,
					     clear_bits, page_ops);
		return -ENOMEM;
	}

	async_chunk = ctx->chunks;
	atomic_set(&ctx->num_chunks, num_chunks);

	for (i = 0; i < num_chunks; i++) {
		if (should_compress)
			cur_end = min(end, start + SZ_512K - 1);
		else
			cur_end = end;

		/*
		 * igrab is called higher up in the call chain, take only the
		 * lightweight reference for the callback lifetime
		 */
		ihold(inode);
		async_chunk[i].pending = &ctx->num_chunks;
		async_chunk[i].inode = inode;
		async_chunk[i].start = start;
		async_chunk[i].end = cur_end;
		async_chunk[i].write_flags = write_flags;
		INIT_LIST_HEAD(&async_chunk[i].extents);

		/*
		 * The locked_page comes all the way from writepage and its
		 * the original page we were actually given.  As we spread
		 * this large delalloc region across multiple async_chunk
		 * structs, only the first struct needs a pointer to locked_page
		 *
		 * This way we don't need racey decisions about who is supposed
		 * to unlock it.
		 */
		if (locked_page) {
			/*
			 * Depending on the compressibility, the pages might or
			 * might not go through async.  We want all of them to
			 * be accounted against wbc once.  Let's do it here
			 * before the paths diverge.  wbc accounting is used
			 * only for foreign writeback detection and doesn't
			 * need full accuracy.  Just account the whole thing
			 * against the first page.
			 */
			wbc_account_cgroup_owner(wbc, locked_page,
						 cur_end - start);
			async_chunk[i].locked_page = locked_page;
			locked_page = NULL;
		} else {
			async_chunk[i].locked_page = NULL;
		}

		if (blkcg_css != blkcg_root_css) {
			css_get(blkcg_css);
			async_chunk[i].blkcg_css = blkcg_css;
		} else {
			async_chunk[i].blkcg_css = NULL;
		}

		btrfs_init_work(&async_chunk[i].work, async_cow_start,
				async_cow_submit, async_cow_free);

		nr_pages = DIV_ROUND_UP(cur_end - start, PAGE_SIZE);
		atomic_add(nr_pages, &fs_info->async_delalloc_pages);

		btrfs_queue_work(fs_info->delalloc_workers, &async_chunk[i].work);

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

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

	ret = btrfs_lookup_csums_range(fs_info->csum_root, bytenr,
				       bytenr + num_bytes - 1, &list, 0);
	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);
	}
	if (ret < 0)
		return ret;
	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,
				       const u64 start, const u64 end,
				       int *page_started, int force,
				       unsigned long *nr_written)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_path *path;
	u64 cow_start = (u64)-1;
	u64 cur_offset = start;
	int ret;
	bool check_prev = true;
	const bool freespace_inode = btrfs_is_free_space_inode(BTRFS_I(inode));
	u64 ino = btrfs_ino(BTRFS_I(inode));
	bool nocow = false;
	u64 disk_bytenr = 0;

	path = btrfs_alloc_path();
	if (!path) {
		extent_clear_unlock_delalloc(inode, start, end, locked_page,
					     EXTENT_LOCKED | EXTENT_DELALLOC |
					     EXTENT_DO_ACCOUNTING |
					     EXTENT_DEFRAG, PAGE_UNLOCK |
					     PAGE_CLEAR_DIRTY |
					     PAGE_SET_WRITEBACK |
					     PAGE_END_WRITEBACK);
		return -ENOMEM;
	}

	while (1) {
		struct btrfs_key found_key;
		struct btrfs_file_extent_item *fi;
		struct extent_buffer *leaf;
		u64 extent_end;
		u64 extent_offset;
		u64 num_bytes = 0;
		u64 disk_num_bytes;
		u64 ram_bytes;
		int extent_type;

		nocow = false;

		ret = btrfs_lookup_file_extent(NULL, root, path, ino,
					       cur_offset, 0);
		if (ret < 0)
			goto error;

		/*
		 * If there is no extent for our range when doing the initial
		 * search, then go back to the previous slot as it will be the
		 * one containing the search offset
		 */
		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 == ino &&
			    found_key.type == BTRFS_EXTENT_DATA_KEY)
				path->slots[0]--;
		}
		check_prev = false;
next_slot:
		/* Go to next leaf if we have exhausted the current one */
		leaf = path->nodes[0];
		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0) {
				if (cow_start != (u64)-1)
					cur_offset = cow_start;
				goto error;
			}
			if (ret > 0)
				break;
			leaf = path->nodes[0];
		}

		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);

		/* Didn't find anything for our INO */
		if (found_key.objectid > ino)
			break;
		/*
		 * Keep searching until we find an EXTENT_ITEM or there are no
		 * more extents for this inode
		 */
		if (WARN_ON_ONCE(found_key.objectid < ino) ||
		    found_key.type < BTRFS_EXTENT_DATA_KEY) {
			path->slots[0]++;
			goto next_slot;
		}

		/* Found key is not EXTENT_DATA_KEY or starts after req range */
		if (found_key.type > BTRFS_EXTENT_DATA_KEY ||
		    found_key.offset > end)
			break;

		/*
		 * If the found extent starts after requested offset, then
		 * adjust extent_end to be right before this extent begins
		 */
		if (found_key.offset > cur_offset) {
			extent_end = found_key.offset;
			extent_type = 0;
			goto out_check;
		}

		/*
		 * Found extent which begins before our range and potentially
		 * intersect it
		 */
		fi = btrfs_item_ptr(leaf, path->slots[0],
				    struct btrfs_file_extent_item);
		extent_type = btrfs_file_extent_type(leaf, fi);

		ram_bytes = btrfs_file_extent_ram_bytes(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);
			disk_num_bytes =
				btrfs_file_extent_disk_num_bytes(leaf, fi);
			/*
			 * If the extent we got ends before our current offset,
			 * skip to the next extent.
			 */
			if (extent_end <= cur_offset) {
				path->slots[0]++;
				goto next_slot;
			}
			/* Skip holes */
			if (disk_bytenr == 0)
				goto out_check;
			/* Skip compressed/encrypted/encoded extents */
			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 is created before the last volume's snapshot
			 * this implies the extent is shared, hence we can't do
			 * nocow. This is the same check as in
			 * btrfs_cross_ref_exist but without calling
			 * btrfs_search_slot.
			 */
			if (!freespace_inode &&
			    btrfs_file_extent_generation(leaf, fi) <=
			    btrfs_root_last_snapshot(&root->root_item))
				goto out_check;
			if (extent_type == BTRFS_FILE_EXTENT_REG && !force)
				goto out_check;
			/* If extent is RO, we must COW it */
			if (btrfs_extent_readonly(fs_info, disk_bytenr))
				goto out_check;
			ret = btrfs_cross_ref_exist(root, ino,
						    found_key.offset -
						    extent_offset, disk_bytenr);
			if (ret) {
				/*
				 * ret could be -EIO if the above fails to read
				 * metadata.
				 */
				if (ret < 0) {
					if (cow_start != (u64)-1)
						cur_offset = cow_start;
					goto error;
				}

				WARN_ON_ONCE(freespace_inode);
				goto out_check;
			}
			disk_bytenr += extent_offset;
			disk_bytenr += cur_offset - found_key.offset;
			num_bytes = min(end + 1, extent_end) - cur_offset;
			/*
			 * If there are pending snapshots for this root, we
			 * fall into common COW way
			 */
			if (!freespace_inode && atomic_read(&root->snapshot_force_cow))
				goto out_check;
			/*
			 * force cow if csum exists in the range.
			 * this ensure that csum for a given extent are
			 * either valid or do not exist.
			 */
			ret = csum_exist_in_range(fs_info, disk_bytenr,
						  num_bytes);
			if (ret) {
				/*
				 * ret could be -EIO if the above fails to read
				 * metadata.
				 */
				if (ret < 0) {
					if (cow_start != (u64)-1)
						cur_offset = cow_start;
					goto error;
				}
				WARN_ON_ONCE(freespace_inode);
				goto out_check;
			}
			if (!btrfs_inc_nocow_writers(fs_info, disk_bytenr))
				goto out_check;
			nocow = true;
		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
			extent_end = found_key.offset + ram_bytes;
			extent_end = ALIGN(extent_end, fs_info->sectorsize);
			/* Skip extents outside of our requested range */
			if (extent_end <= start) {
				path->slots[0]++;
				goto next_slot;
			}
		} else {
			/* If this triggers then we have a memory corruption */
			BUG();
		}
out_check:
		/*
		 * If nocow is false then record the beginning of the range
		 * that needs to be COWed
		 */
		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(path);

		/*
		 * COW range from cow_start to found_key.offset - 1. As the key
		 * will contain the beginning of the first extent that can be
		 * NOCOW, following one which needs to be COW'ed
		 */
		if (cow_start != (u64)-1) {
			ret = cow_file_range(inode, locked_page,
					     cow_start, found_key.offset - 1,
					     page_started, nr_written, 1);
			if (ret) {
				if (nocow)
					btrfs_dec_nocow_writers(fs_info,
								disk_bytenr);
				goto error;
			}
			cow_start = (u64)-1;
		}

		if (extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
			u64 orig_start = found_key.offset - extent_offset;
			struct extent_map *em;

			em = create_io_em(inode, cur_offset, num_bytes,
					  orig_start,
					  disk_bytenr, /* block_start */
					  num_bytes, /* block_len */
					  disk_num_bytes, /* orig_block_len */
					  ram_bytes, BTRFS_COMPRESS_NONE,
					  BTRFS_ORDERED_PREALLOC);
			if (IS_ERR(em)) {
				if (nocow)
					btrfs_dec_nocow_writers(fs_info,
								disk_bytenr);
				ret = PTR_ERR(em);
				goto error;
			}
			free_extent_map(em);
			ret = btrfs_add_ordered_extent(inode, cur_offset,
						       disk_bytenr, num_bytes,
						       num_bytes,
						       BTRFS_ORDERED_PREALLOC);
			if (ret) {
				btrfs_drop_extent_cache(BTRFS_I(inode),
							cur_offset,
							cur_offset + num_bytes - 1,
							0);
				goto error;
			}
		} else {
			ret = btrfs_add_ordered_extent(inode, cur_offset,
						       disk_bytenr, num_bytes,
						       num_bytes,
						       BTRFS_ORDERED_NOCOW);
			if (ret)
				goto error;
		}

		if (nocow)
			btrfs_dec_nocow_writers(fs_info, disk_bytenr);
		nocow = false;

		if (root->root_key.objectid ==
		    BTRFS_DATA_RELOC_TREE_OBJECTID)
			/*
			 * Error handled later, as we must prevent
			 * extent_clear_unlock_delalloc() in error handler
			 * from freeing metadata of created ordered extent.
			 */
			ret = btrfs_reloc_clone_csums(inode, cur_offset,
						      num_bytes);

		extent_clear_unlock_delalloc(inode, cur_offset,
					     cur_offset + num_bytes - 1,
					     locked_page, EXTENT_LOCKED |
					     EXTENT_DELALLOC |
					     EXTENT_CLEAR_DATA_RESV,
					     PAGE_UNLOCK | PAGE_SET_PRIVATE2);

		cur_offset = extent_end;

		/*
		 * btrfs_reloc_clone_csums() error, now we're OK to call error
		 * handler, as metadata for created ordered extent will only
		 * be freed by btrfs_finish_ordered_io().
		 */
		if (ret)
			goto error;
		if (cur_offset > end)
			break;
	}
	btrfs_release_path(path);

	if (cur_offset <= end && cow_start == (u64)-1)
		cow_start = cur_offset;

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

error:
	if (nocow)
		btrfs_dec_nocow_writers(fs_info, disk_bytenr);

	if (ret && cur_offset < end)
		extent_clear_unlock_delalloc(inode, cur_offset, end,
					     locked_page, EXTENT_LOCKED |
					     EXTENT_DELALLOC | EXTENT_DEFRAG |
					     EXTENT_DO_ACCOUNTING, PAGE_UNLOCK |
					     PAGE_CLEAR_DIRTY |
					     PAGE_SET_WRITEBACK |
					     PAGE_END_WRITEBACK);
	btrfs_free_path(path);
	return ret;
}

static inline int need_force_cow(struct inode *inode, u64 start, u64 end)
{

	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) &&
	    !(BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC))
		return 0;

	/*
	 * @defrag_bytes is a hint value, no spinlock held here,
	 * if is not zero, it means the file is defragging.
	 * Force cow if given extent needs to be defragged.
	 */
	if (BTRFS_I(inode)->defrag_bytes &&
	    test_range_bit(&BTRFS_I(inode)->io_tree, start, end,
			   EXTENT_DEFRAG, 0, NULL))
		return 1;

	return 0;
}

/*
 * Function to process delayed allocation (create CoW) for ranges which are
 * being touched for the first time.
 */
int btrfs_run_delalloc_range(struct inode *inode, struct page *locked_page,
		u64 start, u64 end, int *page_started, unsigned long *nr_written,
		struct writeback_control *wbc)
{
	int ret;
	int force_cow = need_force_cow(inode, start, end);

	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW && !force_cow) {
		ret = run_delalloc_nocow(inode, locked_page, start, end,
					 page_started, 1, nr_written);
	} else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC && !force_cow) {
		ret = run_delalloc_nocow(inode, locked_page, start, end,
					 page_started, 0, nr_written);
	} else if (!inode_can_compress(inode) ||
		   !inode_need_compress(inode, start, end)) {
		ret = cow_file_range(inode, locked_page, start, end,
				      page_started, nr_written, 1);
	} else {
		set_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
			&BTRFS_I(inode)->runtime_flags);
		ret = cow_file_range_async(inode, wbc, locked_page, start, end,
					   page_started, nr_written);
	}
	if (ret)
		btrfs_cleanup_ordered_extents(inode, locked_page, start,
					      end - start + 1);
	return ret;
}

void btrfs_split_delalloc_extent(struct inode *inode,
				 struct extent_state *orig, u64 split)
{
	u64 size;

	/* not delalloc, ignore it */
	if (!(orig->state & EXTENT_DELALLOC))
		return;

	size = orig->end - orig->start + 1;
	if (size > BTRFS_MAX_EXTENT_SIZE) {
		u32 num_extents;
		u64 new_size;

		/*
		 * See the explanation in btrfs_merge_delalloc_extent, the same
		 * applies here, just in reverse.
		 */
		new_size = orig->end - split + 1;
		num_extents = count_max_extents(new_size);
		new_size = split - orig->start;
		num_extents += count_max_extents(new_size);
		if (count_max_extents(size) >= num_extents)
			return;
	}

	spin_lock(&BTRFS_I(inode)->lock);
	btrfs_mod_outstanding_extents(BTRFS_I(inode), 1);
	spin_unlock(&BTRFS_I(inode)->lock);
}

/*
 * Handle 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.
 */
void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
				 struct extent_state *other)
{
	u64 new_size, old_size;
	u32 num_extents;

	/* not delalloc, ignore it */
	if (!(other->state & EXTENT_DELALLOC))
		return;

	if (new->start > other->start)
		new_size = new->end - other->start + 1;
	else
		new_size = other->end - new->start + 1;

	/* we're not bigger than the max, unreserve the space and go */
	if (new_size <= BTRFS_MAX_EXTENT_SIZE) {
		spin_lock(&BTRFS_I(inode)->lock);
		btrfs_mod_outstanding_extents(BTRFS_I(inode), -1);
		spin_unlock(&BTRFS_I(inode)->lock);
		return;
	}

	/*
	 * We have to add up either side to figure out how many extents were
	 * accounted for before we merged into one big extent.  If the number of
	 * extents we accounted for is <= the amount we need for the new range
	 * then we can return, otherwise drop.  Think of it like this
	 *
	 * [ 4k][MAX_SIZE]
	 *
	 * So we've grown the extent by a MAX_SIZE extent, this would mean we
	 * need 2 outstanding extents, on one side we have 1 and the other side
	 * we have 1 so they are == and we can return.  But in this case
	 *
	 * [MAX_SIZE+4k][MAX_SIZE+4k]
	 *
	 * Each range on their own accounts for 2 extents, but merged together
	 * they are only 3 extents worth of accounting, so we need to drop in
	 * this case.
	 */
	old_size = other->end - other->start + 1;
	num_extents = count_max_extents(old_size);
	old_size = new->end - new->start + 1;
	num_extents += count_max_extents(old_size);
	if (count_max_extents(new_size) >= num_extents)
		return;

	spin_lock(&BTRFS_I(inode)->lock);
	btrfs_mod_outstanding_extents(BTRFS_I(inode), -1);
	spin_unlock(&BTRFS_I(inode)->lock);
}

static void btrfs_add_delalloc_inodes(struct btrfs_root *root,
				      struct inode *inode)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);

	spin_lock(&root->delalloc_lock);
	if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
		list_add_tail(&BTRFS_I(inode)->delalloc_inodes,
			      &root->delalloc_inodes);
		set_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			&BTRFS_I(inode)->runtime_flags);
		root->nr_delalloc_inodes++;
		if (root->nr_delalloc_inodes == 1) {
			spin_lock(&fs_info->delalloc_root_lock);
			BUG_ON(!list_empty(&root->delalloc_root));
			list_add_tail(&root->delalloc_root,
				      &fs_info->delalloc_roots);
			spin_unlock(&fs_info->delalloc_root_lock);
		}
	}
	spin_unlock(&root->delalloc_lock);
}


void __btrfs_del_delalloc_inode(struct btrfs_root *root,
				struct btrfs_inode *inode)
{
	struct btrfs_fs_info *fs_info = root->fs_info;

	if (!list_empty(&inode->delalloc_inodes)) {
		list_del_init(&inode->delalloc_inodes);
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &inode->runtime_flags);
		root->nr_delalloc_inodes--;
		if (!root->nr_delalloc_inodes) {
			ASSERT(list_empty(&root->delalloc_inodes));
			spin_lock(&fs_info->delalloc_root_lock);
			BUG_ON(list_empty(&root->delalloc_root));
			list_del_init(&root->delalloc_root);
			spin_unlock(&fs_info->delalloc_root_lock);
		}
	}
}

static void btrfs_del_delalloc_inode(struct btrfs_root *root,
				     struct btrfs_inode *inode)
{
	spin_lock(&root->delalloc_lock);
	__btrfs_del_delalloc_inode(root, inode);
	spin_unlock(&root->delalloc_lock);
}

/*
 * Properly track delayed allocation bytes in the inode and to maintain the
 * list of inodes that have pending delalloc work to be done.
 */
void btrfs_set_delalloc_extent(struct inode *inode, struct extent_state *state,
			       unsigned *bits)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);

	if ((*bits & EXTENT_DEFRAG) && !(*bits & EXTENT_DELALLOC))
		WARN_ON(1);
	/*
	 * set_bit and clear bit hooks normally require _irqsave/restore
	 * but in this case, we are only testing 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;
		u32 num_extents = count_max_extents(len);
		bool do_list = !btrfs_is_free_space_inode(BTRFS_I(inode));

		spin_lock(&BTRFS_I(inode)->lock);
		btrfs_mod_outstanding_extents(BTRFS_I(inode), num_extents);
		spin_unlock(&BTRFS_I(inode)->lock);

		/* For sanity tests */
		if (btrfs_is_testing(fs_info))
			return;

		percpu_counter_add_batch(&fs_info->delalloc_bytes, len,
					 fs_info->delalloc_batch);
		spin_lock(&BTRFS_I(inode)->lock);
		BTRFS_I(inode)->delalloc_bytes += len;
		if (*bits & EXTENT_DEFRAG)
			BTRFS_I(inode)->defrag_bytes += len;
		if (do_list && !test_bit(BTRFS_INODE_IN_DELALLOC_LIST,
					 &BTRFS_I(inode)->runtime_flags))
			btrfs_add_delalloc_inodes(root, inode);
		spin_unlock(&BTRFS_I(inode)->lock);
	}

	if (!(state->state & EXTENT_DELALLOC_NEW) &&
	    (*bits & EXTENT_DELALLOC_NEW)) {
		spin_lock(&BTRFS_I(inode)->lock);
		BTRFS_I(inode)->new_delalloc_bytes += state->end + 1 -
			state->start;
		spin_unlock(&BTRFS_I(inode)->lock);
	}
}

/*
 * Once a range is no longer delalloc this function ensures that proper
 * accounting happens.
 */
void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
				 struct extent_state *state, unsigned *bits)
{
	struct btrfs_inode *inode = BTRFS_I(vfs_inode);
	struct btrfs_fs_info *fs_info = btrfs_sb(vfs_inode->i_sb);
	u64 len = state->end + 1 - state->start;
	u32 num_extents = count_max_extents(len);

	if ((state->state & EXTENT_DEFRAG) && (*bits & EXTENT_DEFRAG)) {
		spin_lock(&inode->lock);
		inode->defrag_bytes -= len;
		spin_unlock(&inode->lock);
	}

	/*
	 * set_bit and clear bit hooks normally require _irqsave/restore
	 * but in this case, we are only testing 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 = inode->root;
		bool do_list = !btrfs_is_free_space_inode(inode);

		spin_lock(&inode->lock);
		btrfs_mod_outstanding_extents(inode, -num_extents);
		spin_unlock(&inode->lock);

		/*
		 * We don't reserve metadata space for space cache inodes so we
		 * don't need to call delalloc_release_metadata if there is an
		 * error.
		 */
		if (*bits & EXTENT_CLEAR_META_RESV &&
		    root != fs_info->tree_root)
			btrfs_delalloc_release_metadata(inode, len, false);

		/* For sanity tests. */
		if (btrfs_is_testing(fs_info))
			return;

		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID &&
		    do_list && !(state->state & EXTENT_NORESERVE) &&
		    (*bits & EXTENT_CLEAR_DATA_RESV))
			btrfs_free_reserved_data_space_noquota(
					&inode->vfs_inode,
					state->start, len);

		percpu_counter_add_batch(&fs_info->delalloc_bytes, -len,
					 fs_info->delalloc_batch);
		spin_lock(&inode->lock);
		inode->delalloc_bytes -= len;
		if (do_list && inode->delalloc_bytes == 0 &&
		    test_bit(BTRFS_INODE_IN_DELALLOC_LIST,
					&inode->runtime_flags))
			btrfs_del_delalloc_inode(root, inode);
		spin_unlock(&inode->lock);
	}

	if ((state->state & EXTENT_DELALLOC_NEW) &&
	    (*bits & EXTENT_DELALLOC_NEW)) {
		spin_lock(&inode->lock);
		ASSERT(inode->new_delalloc_bytes >= len);
		inode->new_delalloc_bytes -= len;
		spin_unlock(&inode->lock);
	}
}

/*
 * btrfs_bio_fits_in_stripe - Checks whether the size of the given bio will fit
 * in a chunk's stripe. This function ensures that bios do not span a
 * stripe/chunk
 *
 * @page - The page we are about to add to the bio
 * @size - size we want to add to the bio
 * @bio - bio we want to ensure is smaller than a stripe
 * @bio_flags - flags of the bio
 *
 * return 1 if page cannot be added to the bio
 * return 0 if page can be added to the bio
 * return error otherwise
 */
int btrfs_bio_fits_in_stripe(struct page *page, size_t size, struct bio *bio,
			     unsigned long bio_flags)
{
	struct inode *inode = page->mapping->host;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	u64 logical = (u64)bio->bi_iter.bi_sector << 9;
	u64 length = 0;
	u64 map_length;
	int ret;
	struct btrfs_io_geometry geom;

	if (bio_flags & EXTENT_BIO_COMPRESSED)
		return 0;

	length = bio->bi_iter.bi_size;
	map_length = length;
	ret = btrfs_get_io_geometry(fs_info, btrfs_op(bio), logical, map_length,
				    &geom);
	if (ret < 0)
		return ret;

	if (geom.len < 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 blk_status_t btrfs_submit_bio_start(void *private_data, struct bio *bio,
				    u64 bio_offset)
{
	struct inode *inode = private_data;
	blk_status_t ret = 0;

	ret = btrfs_csum_one_bio(inode, bio, 0, 0);
	BUG_ON(ret); /* -ENOMEM */
	return 0;
}

/*
 * 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.
 *
 * Rules about async/sync submit,
 * a) read:				sync submit
 *
 * b) write without checksum:		sync submit
 *
 * c) write with checksum:
 *    c-1) if bio is issued by fsync:	sync submit
 *         (sync_writers != 0)
 *
 *    c-2) if root is reloc root:	sync submit
 *         (only in case of buffered IO)
 *
 *    c-3) otherwise:			async submit
 */
static blk_status_t btrfs_submit_bio_hook(struct inode *inode, struct bio *bio,
					  int mirror_num,
					  unsigned long bio_flags)

{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	enum btrfs_wq_endio_type metadata = BTRFS_WQ_ENDIO_DATA;
	blk_status_t ret = 0;
	int skip_sum;
	int async = !atomic_read(&BTRFS_I(inode)->sync_writers);

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

	if (btrfs_is_free_space_inode(BTRFS_I(inode)))
		metadata = BTRFS_WQ_ENDIO_FREE_SPACE;

	if (bio_op(bio) != REQ_OP_WRITE) {
		ret = btrfs_bio_wq_end_io(fs_info, bio, metadata);
		if (ret)
			goto out;

		if (bio_flags & EXTENT_BIO_COMPRESSED) {
			ret = btrfs_submit_compressed_read(inode, bio,
							   mirror_num,
							   bio_flags);
			goto out;
		} else if (!skip_sum) {
			ret = btrfs_lookup_bio_sums(inode, bio, (u64)-1, NULL);
			if (ret)
				goto out;
		}
		goto mapit;
	} else if (async && !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 */
		ret = btrfs_wq_submit_bio(fs_info, bio, mirror_num, bio_flags,
					  0, inode, btrfs_submit_bio_start);
		goto out;
	} else if (!skip_sum) {
		ret = btrfs_csum_one_bio(inode, bio, 0, 0);
		if (ret)
			goto out;
	}

mapit:
	ret = btrfs_map_bio(fs_info, bio, mirror_num);

out:
	if (ret) {
		bio->bi_status = ret;
		bio_endio(bio);
	}
	return ret;
}

/*
 * 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, struct list_head *list)
{
	struct btrfs_ordered_sum *sum;
	int ret;

	list_for_each_entry(sum, list, list) {
		trans->adding_csums = true;
		ret = btrfs_csum_file_blocks(trans,
		       BTRFS_I(inode)->root->fs_info->csum_root, sum);
		trans->adding_csums = false;
		if (ret)
			return ret;
	}
	return 0;
}

int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
			      unsigned int extra_bits,
			      struct extent_state **cached_state)
{
	WARN_ON(PAGE_ALIGNED(end));
	return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end,
				   extra_bits, cached_state);
}

/* see btrfs_writepage_start_hook for details on why this is required */
struct btrfs_writepage_fixup {
	struct page *page;
	struct inode *inode;
	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 extent_changeset *data_reserved = NULL;
	struct page *page;
	struct inode *inode;
	u64 page_start;
	u64 page_end;
	int ret = 0;
	bool free_delalloc_space = true;

	fixup = container_of(work, struct btrfs_writepage_fixup, work);
	page = fixup->page;
	inode = fixup->inode;
	page_start = page_offset(page);
	page_end = page_offset(page) + PAGE_SIZE - 1;

	/*
	 * This is similar to page_mkwrite, we need to reserve the space before
	 * we take the page lock.
	 */
	ret = btrfs_delalloc_reserve_space(inode, &data_reserved, page_start,
					   PAGE_SIZE);
again:
	lock_page(page);

	/*
	 * Before we queued this fixup, we took a reference on the page.
	 * page->mapping may go NULL, but it shouldn't be moved to a different
	 * address space.
	 */
	if (!page->mapping || !PageDirty(page) || !PageChecked(page)) {
		/*
		 * Unfortunately this is a little tricky, either
		 *
		 * 1) We got here and our page had already been dealt with and
		 *    we reserved our space, thus ret == 0, so we need to just
		 *    drop our space reservation and bail.  This can happen the
		 *    first time we come into the fixup worker, or could happen
		 *    while waiting for the ordered extent.
		 * 2) Our page was already dealt with, but we happened to get an
		 *    ENOSPC above from the btrfs_delalloc_reserve_space.  In
		 *    this case we obviously don't have anything to release, but
		 *    because the page was already dealt with we don't want to
		 *    mark the page with an error, so make sure we're resetting
		 *    ret to 0.  This is why we have this check _before_ the ret
		 *    check, because we do not want to have a surprise ENOSPC
		 *    when the page was already properly dealt with.
		 */
		if (!ret) {
			btrfs_delalloc_release_extents(BTRFS_I(inode),
						       PAGE_SIZE);
			btrfs_delalloc_release_space(inode, data_reserved,
						     page_start, PAGE_SIZE,
						     true);
		}
		ret = 0;
		goto out_page;
	}

	/*
	 * We can't mess with the page state unless it is locked, so now that
	 * it is locked bail if we failed to make our space reservation.
	 */
	if (ret)
		goto out_page;

	lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
			 &cached_state);

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

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

	ret = btrfs_set_extent_delalloc(inode, page_start, page_end, 0,
					&cached_state);
	if (ret)
		goto out_reserved;

	/*
	 * Everything went as planned, we're now the owner of a dirty page with
	 * delayed allocation bits set and space reserved for our COW
	 * destination.
	 *
	 * The page was dirty when we started, nothing should have cleaned it.
	 */
	BUG_ON(!PageDirty(page));
	free_delalloc_space = false;
out_reserved:
	btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE);
	if (free_delalloc_space)
		btrfs_delalloc_release_space(inode, data_reserved, page_start,
					     PAGE_SIZE, true);
	unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, page_end,
			     &cached_state);
out_page:
	if (ret) {
		/*
		 * We hit ENOSPC or other errors.  Update the mapping and page
		 * to reflect the errors and clean the page.
		 */
		mapping_set_error(page->mapping, ret);
		end_extent_writepage(page, ret, page_start, page_end);
		clear_page_dirty_for_io(page);
		SetPageError(page);
	}
	ClearPageChecked(page);
	unlock_page(page);
	put_page(page);
	kfree(fixup);
	extent_changeset_free(data_reserved);
	/*
	 * As a precaution, do a delayed iput in case it would be the last iput
	 * that could need flushing space. Recursing back to fixup worker would
	 * deadlock.
	 */
	btrfs_add_delayed_iput(inode);
}

/*
 * 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.
 */
int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end)
{
	struct inode *inode = page->mapping->host;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_writepage_fixup *fixup;

	/* this page is properly in the ordered list */
	if (TestClearPagePrivate2(page))
		return 0;

	/*
	 * PageChecked is set below when we create a fixup worker for this page,
	 * don't try to create another one if we're already PageChecked()
	 *
	 * The extent_io writepage code will redirty the page if we send back
	 * EAGAIN.
	 */
	if (PageChecked(page))
		return -EAGAIN;

	fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
	if (!fixup)
		return -EAGAIN;

	/*
	 * We are already holding a reference to this inode from
	 * write_cache_pages.  We need to hold it because the space reservation
	 * takes place outside of the page lock, and we can't trust
	 * page->mapping outside of the page lock.
	 */
	ihold(inode);
	SetPageChecked(page);
	get_page(page);
	btrfs_init_work(&fixup->work, btrfs_writepage_fixup_worker, NULL, NULL);
	fixup->page = page;
	fixup->inode = inode;
	btrfs_queue_work(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 qg_released;
	int extent_inserted = 0;
	int ret;

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

	/*
	 * 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, root, inode, path, file_pos,
				   file_pos + num_bytes, NULL, 0,
				   1, sizeof(*fi), &extent_inserted);
	if (ret)
		goto out;

	if (!extent_inserted) {
		ins.objectid = btrfs_ino(BTRFS_I(inode));
		ins.offset = file_pos;
		ins.type = BTRFS_EXTENT_DATA_KEY;

		path->leave_spinning = 1;
		ret = btrfs_insert_empty_item(trans, root, path, &ins,
					      sizeof(*fi));
		if (ret)
			goto out;
	}
	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_mark_buffer_dirty(leaf);
	btrfs_release_path(path);

	inode_add_bytes(inode, num_bytes);

	ins.objectid = disk_bytenr;
	ins.offset = disk_num_bytes;
	ins.type = BTRFS_EXTENT_ITEM_KEY;

	/*
	 * Release the reserved range from inode dirty range map, as it is
	 * already moved into delayed_ref_head
	 */
	ret = btrfs_qgroup_release_data(inode, file_pos, ram_bytes);
	if (ret < 0)
		goto out;
	qg_released = ret;
	ret = btrfs_alloc_reserved_file_extent(trans, root,
					       btrfs_ino(BTRFS_I(inode)),
					       file_pos, qg_released, &ins);
out:
	btrfs_free_path(path);

	return ret;
}

static void btrfs_release_delalloc_bytes(struct btrfs_fs_info *fs_info,
					 u64 start, u64 len)
{
	struct btrfs_block_group *cache;

	cache = btrfs_lookup_block_group(fs_info, start);
	ASSERT(cache);

	spin_lock(&cache->lock);
	cache->delalloc_bytes -= len;
	spin_unlock(&cache->lock);

	btrfs_put_block_group(cache);
}

/* 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 btrfs_ordered_extent *ordered_extent)
{
	struct inode *inode = ordered_extent->inode;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans = NULL;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_state *cached_state = NULL;
	u64 start, end;
	int compress_type = 0;
	int ret = 0;
	u64 logical_len = ordered_extent->num_bytes;
	bool freespace_inode;
	bool truncated = false;
	bool range_locked = false;
	bool clear_new_delalloc_bytes = false;
	bool clear_reserved_extent = true;
	unsigned int clear_bits;

	start = ordered_extent->file_offset;
	end = start + ordered_extent->num_bytes - 1;

	if (!test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
	    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags) &&
	    !test_bit(BTRFS_ORDERED_DIRECT, &ordered_extent->flags))
		clear_new_delalloc_bytes = true;

	freespace_inode = btrfs_is_free_space_inode(BTRFS_I(inode));

	if (test_bit(BTRFS_ORDERED_IOERR, &ordered_extent->flags)) {
		ret = -EIO;
		goto out;
	}

	btrfs_free_io_failure_record(BTRFS_I(inode), start, end);

	if (test_bit(BTRFS_ORDERED_TRUNCATED, &ordered_extent->flags)) {
		truncated = true;
		logical_len = ordered_extent->truncated_len;
		/* Truncated the entire extent, don't bother adding */
		if (!logical_len)
			goto out;
	}

	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
		BUG_ON(!list_empty(&ordered_extent->list)); /* Logic error */

		/*
		 * For mwrite(mmap + memset to write) case, we still reserve
		 * space for NOCOW range.
		 * As NOCOW won't cause a new delayed ref, just free the space
		 */
		btrfs_qgroup_free_data(inode, NULL, start,
				       ordered_extent->num_bytes);
		btrfs_ordered_update_i_size(inode, 0, ordered_extent);
		if (freespace_inode)
			trans = btrfs_join_transaction_spacecache(root);
		else
			trans = btrfs_join_transaction(root);
		if (IS_ERR(trans)) {
			ret = PTR_ERR(trans);
			trans = NULL;
			goto out;
		}
		trans->block_rsv = &BTRFS_I(inode)->block_rsv;
		ret = btrfs_update_inode_fallback(trans, root, inode);
		if (ret) /* -ENOMEM or corruption */
			btrfs_abort_transaction(trans, ret);
		goto out;
	}

	range_locked = true;
	lock_extent_bits(io_tree, start, end, &cached_state);

	if (freespace_inode)
		trans = btrfs_join_transaction_spacecache(root);
	else
		trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		trans = NULL;
		goto out;
	}

	trans->block_rsv = &BTRFS_I(inode)->block_rsv;

	if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
		compress_type = ordered_extent->compress_type;
	if (test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
		BUG_ON(compress_type);
		btrfs_qgroup_free_data(inode, NULL, start,
				       ordered_extent->num_bytes);
		ret = btrfs_mark_extent_written(trans, BTRFS_I(inode),
						ordered_extent->file_offset,
						ordered_extent->file_offset +
						logical_len);
	} else {
		BUG_ON(root == fs_info->tree_root);
		ret = insert_reserved_file_extent(trans, inode, start,
						ordered_extent->disk_bytenr,
						ordered_extent->disk_num_bytes,
						logical_len, logical_len,
						compress_type, 0, 0,
						BTRFS_FILE_EXTENT_REG);
		if (!ret) {
			clear_reserved_extent = false;
			btrfs_release_delalloc_bytes(fs_info,
						ordered_extent->disk_bytenr,
						ordered_extent->disk_num_bytes);
		}
	}
	unpin_extent_cache(&BTRFS_I(inode)->extent_tree,
			   ordered_extent->file_offset,
			   ordered_extent->num_bytes, trans->transid);
	if (ret < 0) {
		btrfs_abort_transaction(trans, ret);
		goto out;
	}

	ret = add_pending_csums(trans, inode, &ordered_extent->list);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out;
	}

	btrfs_ordered_update_i_size(inode, 0, ordered_extent);
	ret = btrfs_update_inode_fallback(trans, root, inode);
	if (ret) { /* -ENOMEM or corruption */
		btrfs_abort_transaction(trans, ret);
		goto out;
	}
	ret = 0;
out:
	clear_bits = EXTENT_DEFRAG;
	if (range_locked)
		clear_bits |= EXTENT_LOCKED;
	if (clear_new_delalloc_bytes)
		clear_bits |= EXTENT_DELALLOC_NEW;
	clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, clear_bits,
			 (clear_bits & EXTENT_LOCKED) ? 1 : 0, 0,
			 &cached_state);

	if (trans)
		btrfs_end_transaction(trans);

	if (ret || truncated) {
		u64 unwritten_start = start;

		if (truncated)
			unwritten_start += logical_len;
		clear_extent_uptodate(io_tree, unwritten_start, end, NULL);

		/* Drop the cache for the part of the extent we didn't write. */
		btrfs_drop_extent_cache(BTRFS_I(inode), unwritten_start, end, 0);

		/*
		 * If the ordered extent had an IOERR or something else went
		 * wrong we need to return the space for this ordered extent
		 * back to the allocator.  We only free the extent in the
		 * truncated case if we didn't write out the extent at all.
		 *
		 * If we made it past insert_reserved_file_extent before we
		 * errored out then we don't need to do this as the accounting
		 * has already been done.
		 */
		if ((ret || !logical_len) &&
		    clear_reserved_extent &&
		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
		    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
			/*
			 * Discard the range before returning it back to the
			 * free space pool
			 */
			if (ret && btrfs_test_opt(fs_info, DISCARD_SYNC))
				btrfs_discard_extent(fs_info,
						ordered_extent->disk_bytenr,
						ordered_extent->disk_num_bytes,
						NULL);
			btrfs_free_reserved_extent(fs_info,
					ordered_extent->disk_bytenr,
					ordered_extent->disk_num_bytes, 1);
		}
	}

	/*
	 * This needs to be done to make sure anybody waiting knows we are done
	 * updating everything for this ordered extent.
	 */
	btrfs_remove_ordered_extent(inode, ordered_extent);

	/* once for us */
	btrfs_put_ordered_extent(ordered_extent);
	/* once for the tree */
	btrfs_put_ordered_extent(ordered_extent);

	return ret;
}

static void finish_ordered_fn(struct btrfs_work *work)
{
	struct btrfs_ordered_extent *ordered_extent;
	ordered_extent = container_of(work, struct btrfs_ordered_extent, work);
	btrfs_finish_ordered_io(ordered_extent);
}

void btrfs_writepage_endio_finish_ordered(struct page *page, u64 start,
					  u64 end, int uptodate)
{
	struct inode *inode = page->mapping->host;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_ordered_extent *ordered_extent = NULL;
	struct btrfs_workqueue *wq;

	trace_btrfs_writepage_end_io_hook(page, start, end, uptodate);

	ClearPagePrivate2(page);
	if (!btrfs_dec_test_ordered_pending(inode, &ordered_extent, start,
					    end - start + 1, uptodate))
		return;

	if (btrfs_is_free_space_inode(BTRFS_I(inode)))
		wq = fs_info->endio_freespace_worker;
	else
		wq = fs_info->endio_write_workers;

	btrfs_init_work(&ordered_extent->work, finish_ordered_fn, NULL, NULL);
	btrfs_queue_work(wq, &ordered_extent->work);
}

static int __readpage_endio_check(struct inode *inode,
				  struct btrfs_io_bio *io_bio,
				  int icsum, struct page *page,
				  int pgoff, u64 start, size_t len)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
	char *kaddr;
	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
	u8 *csum_expected;
	u8 csum[BTRFS_CSUM_SIZE];

	csum_expected = ((u8 *)io_bio->csum) + icsum * csum_size;

	kaddr = kmap_atomic(page);
	shash->tfm = fs_info->csum_shash;

	crypto_shash_init(shash);
	crypto_shash_update(shash, kaddr + pgoff, len);
	crypto_shash_final(shash, csum);

	if (memcmp(csum, csum_expected, csum_size))
		goto zeroit;

	kunmap_atomic(kaddr);
	return 0;
zeroit:
	btrfs_print_data_csum_error(BTRFS_I(inode), start, csum, csum_expected,
				    io_bio->mirror_num);
	memset(kaddr + pgoff, 1, len);
	flush_dcache_page(page);
	kunmap_atomic(kaddr);
	return -EIO;
}

/*
 * 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, the code in
 * extent_io.c will try to find good copies for us.
 */
static int btrfs_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
				      u64 phy_offset, struct page *page,
				      u64 start, u64 end, int mirror)
{
	size_t offset = start - page_offset(page);
	struct inode *inode = page->mapping->host;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct btrfs_root *root = BTRFS_I(inode)->root;

	if (PageChecked(page)) {
		ClearPageChecked(page);
		return 0;
	}

	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);
		return 0;
	}

	phy_offset >>= inode->i_sb->s_blocksize_bits;
	return __readpage_endio_check(inode, io_bio, phy_offset, page, offset,
				      start, (size_t)(end - start + 1));
}

/*
 * btrfs_add_delayed_iput - perform a delayed iput on @inode
 *
 * @inode: The inode we want to perform iput on
 *
 * This function uses the generic vfs_inode::i_count to track whether we should
 * just decrement it (in case it's > 1) or if this is the last iput then link
 * the inode to the delayed iput machinery. Delayed iputs are processed at
 * transaction commit time/superblock commit/cleaner kthread.
 */
void btrfs_add_delayed_iput(struct inode *inode)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_inode *binode = BTRFS_I(inode);

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

	atomic_inc(&fs_info->nr_delayed_iputs);
	spin_lock(&fs_info->delayed_iput_lock);
	ASSERT(list_empty(&binode->delayed_iput));
	list_add_tail(&binode->delayed_iput, &fs_info->delayed_iputs);
	spin_unlock(&fs_info->delayed_iput_lock);
	if (!test_bit(BTRFS_FS_CLEANER_RUNNING, &fs_info->flags))
		wake_up_process(fs_info->cleaner_kthread);
}

static void run_delayed_iput_locked(struct btrfs_fs_info *fs_info,
				    struct btrfs_inode *inode)
{
	list_del_init(&inode->delayed_iput);
	spin_unlock(&fs_info->delayed_iput_lock);
	iput(&inode->vfs_inode);
	if (atomic_dec_and_test(&fs_info->nr_delayed_iputs))
		wake_up(&fs_info->delayed_iputs_wait);
	spin_lock(&fs_info->delayed_iput_lock);
}

static void btrfs_run_delayed_iput(struct btrfs_fs_info *fs_info,
				   struct btrfs_inode *inode)
{
	if (!list_empty(&inode->delayed_iput)) {
		spin_lock(&fs_info->delayed_iput_lock);
		if (!list_empty(&inode->delayed_iput))
			run_delayed_iput_locked(fs_info, inode);
		spin_unlock(&fs_info->delayed_iput_lock);
	}
}

void btrfs_run_delayed_iputs(struct btrfs_fs_info *fs_info)
{

	spin_lock(&fs_info->delayed_iput_lock);
	while (!list_empty(&fs_info->delayed_iputs)) {
		struct btrfs_inode *inode;

		inode = list_first_entry(&fs_info->delayed_iputs,
				struct btrfs_inode, delayed_iput);
		run_delayed_iput_locked(fs_info, inode);
	}
	spin_unlock(&fs_info->delayed_iput_lock);
}

/**
 * btrfs_wait_on_delayed_iputs - wait on the delayed iputs to be done running
 * @fs_info - the fs_info for this fs
 * @return - EINTR if we were killed, 0 if nothing's pending
 *
 * This will wait on any delayed iputs that are currently running with KILLABLE
 * set.  Once they are all done running we will return, unless we are killed in
 * which case we return EINTR. This helps in user operations like fallocate etc
 * that might get blocked on the iputs.
 */
int btrfs_wait_on_delayed_iputs(struct btrfs_fs_info *fs_info)
{
	int ret = wait_event_killable(fs_info->delayed_iputs_wait,
			atomic_read(&fs_info->nr_delayed_iputs) == 0);
	if (ret)
		return -EINTR;
	return 0;
}

/*
 * This creates an orphan entry for the given inode in case something goes wrong
 * in the middle of an unlink.
 */
int btrfs_orphan_add(struct btrfs_trans_handle *trans,
		     struct btrfs_inode *inode)
{
	int ret;

	ret = btrfs_insert_orphan_item(trans, inode->root, btrfs_ino(inode));
	if (ret && ret != -EEXIST) {
		btrfs_abort_transaction(trans, ret);
		return ret;
	}

	return 0;
}

/*
 * We have done the delete so we can go ahead and remove the orphan item for
 * this particular inode.
 */
static int btrfs_orphan_del(struct btrfs_trans_handle *trans,
			    struct btrfs_inode *inode)
{
	return btrfs_del_orphan_item(trans, inode->root, btrfs_ino(inode));
}

/*
 * this cleans up any orphans that may be left on the list from the last use
 * of this root.
 */
int btrfs_orphan_cleanup(struct btrfs_root *root)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_key key, found_key;
	struct btrfs_trans_handle *trans;
	struct inode *inode;
	u64 last_objectid = 0;
	int ret = 0, nr_unlink = 0;

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

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}
	path->reada = READA_BACK;

	key.objectid = BTRFS_ORPHAN_OBJECTID;
	key.type = BTRFS_ORPHAN_ITEM_KEY;
	key.offset = (u64)-1;

	while (1) {
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
		if (ret < 0)
			goto out;

		/*
		 * if ret == 0 means we found what we were searching for, which
		 * is weird, but possible, so only screw with path if we didn't
		 * find the key and see if we have stuff that matches
		 */
		if (ret > 0) {
			ret = 0;
			if (path->slots[0] == 0)
				break;
			path->slots[0]--;
		}

		/* pull out the item */
		leaf = path->nodes[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 (found_key.type != BTRFS_ORPHAN_ITEM_KEY)
			break;

		/* release the path since we're done with it */
		btrfs_release_path(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.
		 */

		if (found_key.offset == last_objectid) {
			btrfs_err(fs_info,
				  "Error removing orphan entry, stopping orphan cleanup");
			ret = -EINVAL;
			goto out;
		}

		last_objectid = found_key.offset;

		found_key.objectid = found_key.offset;
		found_key.type = BTRFS_INODE_ITEM_KEY;
		found_key.offset = 0;
		inode = btrfs_iget(fs_info->sb, &found_key, root);
		ret = PTR_ERR_OR_ZERO(inode);
		if (ret && ret != -ENOENT)
			goto out;

		if (ret == -ENOENT && root == fs_info->tree_root) {
			struct btrfs_root *dead_root;
			struct btrfs_fs_info *fs_info = root->fs_info;
			int is_dead_root = 0;

			/*
			 * this is an orphan in the tree root. Currently these
			 * could come from 2 sources:
			 *  a) a snapshot deletion in progress
			 *  b) a free space cache inode
			 * We need to distinguish those two, as the snapshot
			 * orphan must not get deleted.
			 * find_dead_roots already ran before us, so if this
			 * is a snapshot deletion, we should find the root
			 * in the dead_roots list
			 */
			spin_lock(&fs_info->trans_lock);
			list_for_each_entry(dead_root, &fs_info->dead_roots,
					    root_list) {
				if (dead_root->root_key.objectid ==
				    found_key.objectid) {
					is_dead_root = 1;
					break;
				}
			}
			spin_unlock(&fs_info->trans_lock);
			if (is_dead_root) {
				/* prevent this orphan from being found again */
				key.offset = found_key.objectid - 1;
				continue;
			}

		}

		/*
		 * If we have an inode with links, there are a couple of
		 * possibilities. Old kernels (before v3.12) used to create an
		 * orphan item for truncate indicating that there were possibly
		 * extent items past i_size that needed to be deleted. In v3.12,
		 * truncate was changed to update i_size in sync with the extent
		 * items, but the (useless) orphan item was still created. Since
		 * v4.18, we don't create the orphan item for truncate at all.
		 *
		 * So, this item could mean that we need to do a truncate, but
		 * only if this filesystem was last used on a pre-v3.12 kernel
		 * and was not cleanly unmounted. The odds of that are quite
		 * slim, and it's a pain to do the truncate now, so just delete
		 * the orphan item.
		 *
		 * It's also possible that this orphan item was supposed to be
		 * deleted but wasn't. The inode number may have been reused,
		 * but either way, we can delete the orphan item.
		 */
		if (ret == -ENOENT || inode->i_nlink) {
			if (!ret)
				iput(inode);
			trans = btrfs_start_transaction(root, 1);
			if (IS_ERR(trans)) {
				ret = PTR_ERR(trans);
				goto out;
			}
			btrfs_debug(fs_info, "auto deleting %Lu",
				    found_key.objectid);
			ret = btrfs_del_orphan_item(trans, root,
						    found_key.objectid);
			btrfs_end_transaction(trans);
			if (ret)
				goto out;
			continue;
		}

		nr_unlink++;

		/* this will do delete_inode and everything for us */
		iput(inode);
	}
	/* release the path since we're done with it */
	btrfs_release_path(path);

	root->orphan_cleanup_state = ORPHAN_CLEANUP_DONE;

	if (test_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state)) {
		trans = btrfs_join_transaction(root);
		if (!IS_ERR(trans))
			btrfs_end_transaction(trans);
	}

	if (nr_unlink)
		btrfs_debug(fs_info, "unlinked %d orphans", nr_unlink);

out:
	if (ret)
		btrfs_err(fs_info, "could not do orphan cleanup %d", ret);
	btrfs_free_path(path);
	return ret;
}

/*
 * 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,
					  int *first_xattr_slot)
{
	u32 nritems = btrfs_header_nritems(leaf);
	struct btrfs_key found_key;
	static u64 xattr_access = 0;
	static u64 xattr_default = 0;
	int scanned = 0;

	if (!xattr_access) {
		xattr_access = btrfs_name_hash(XATTR_NAME_POSIX_ACL_ACCESS,
					strlen(XATTR_NAME_POSIX_ACL_ACCESS));
		xattr_default = btrfs_name_hash(XATTR_NAME_POSIX_ACL_DEFAULT,
					strlen(XATTR_NAME_POSIX_ACL_DEFAULT));
	}

	slot++;
	*first_xattr_slot = -1;
	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) {
			if (*first_xattr_slot == -1)
				*first_xattr_slot = slot;
			if (found_key.offset == xattr_access ||
			    found_key.offset == xattr_default)
				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
	 */
	if (*first_xattr_slot == -1)
		*first_xattr_slot = slot;
	return 1;
}

/*
 * read an inode from the btree into the in-memory inode
 */
static int btrfs_read_locked_inode(struct inode *inode,
				   struct btrfs_path *in_path)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_path *path = in_path;
	struct extent_buffer *leaf;
	struct btrfs_inode_item *inode_item;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_key location;
	unsigned long ptr;
	int maybe_acls;
	u32 rdev;
	int ret;
	bool filled = false;
	int first_xattr_slot;

	ret = btrfs_fill_inode(inode, &rdev);
	if (!ret)
		filled = true;

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

	memcpy(&location, &BTRFS_I(inode)->location, sizeof(location));

	ret = btrfs_lookup_inode(NULL, root, path, &location, 0);
	if (ret) {
		if (path != in_path)
			btrfs_free_path(path);
		return ret;
	}

	leaf = path->nodes[0];

	if (filled)
		goto cache_index;

	inode_item = btrfs_item_ptr(leaf, path->slots[0],
				    struct btrfs_inode_item);
	inode->i_mode = btrfs_inode_mode(leaf, inode_item);
	set_nlink(inode, btrfs_inode_nlink(leaf, inode_item));
	i_uid_write(inode, btrfs_inode_uid(leaf, inode_item));
	i_gid_write(inode, btrfs_inode_gid(leaf, inode_item));
	btrfs_i_size_write(BTRFS_I(inode), btrfs_inode_size(leaf, inode_item));

	inode->i_atime.tv_sec = btrfs_timespec_sec(leaf, &inode_item->atime);
	inode->i_atime.tv_nsec = btrfs_timespec_nsec(leaf, &inode_item->atime);

	inode->i_mtime.tv_sec = btrfs_timespec_sec(leaf, &inode_item->mtime);
	inode->i_mtime.tv_nsec = btrfs_timespec_nsec(leaf, &inode_item->mtime);

	inode->i_ctime.tv_sec = btrfs_timespec_sec(leaf, &inode_item->ctime);
	inode->i_ctime.tv_nsec = btrfs_timespec_nsec(leaf, &inode_item->ctime);

	BTRFS_I(inode)->i_otime.tv_sec =
		btrfs_timespec_sec(leaf, &inode_item->otime);
	BTRFS_I(inode)->i_otime.tv_nsec =
		btrfs_timespec_nsec(leaf, &inode_item->otime);

	inode_set_bytes(inode, btrfs_inode_nbytes(leaf, inode_item));
	BTRFS_I(inode)->generation = btrfs_inode_generation(leaf, inode_item);
	BTRFS_I(inode)->last_trans = btrfs_inode_transid(leaf, inode_item);

	inode_set_iversion_queried(inode,
				   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);

cache_index:
	/*
	 * If we were modified in the current generation and evicted from memory
	 * and then re-read we need to do a full sync since we don't have any
	 * idea about which extents were modified before we were evicted from
	 * cache.
	 *
	 * This is required for both inode re-read from disk and delayed inode
	 * in delayed_nodes_tree.
	 */
	if (BTRFS_I(inode)->last_trans == fs_info->generation)
		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
			&BTRFS_I(inode)->runtime_flags);

	/*
	 * We don't persist the id of the transaction where an unlink operation
	 * against the inode was last made. So here we assume the inode might
	 * have been evicted, and therefore the exact value of last_unlink_trans
	 * lost, and set it to last_trans to avoid metadata inconsistencies
	 * between the inode and its parent if the inode is fsync'ed and the log
	 * replayed. For example, in the scenario:
	 *
	 * touch mydir/foo
	 * ln mydir/foo mydir/bar
	 * sync
	 * unlink mydir/bar
	 * echo 2 > /proc/sys/vm/drop_caches   # evicts inode
	 * xfs_io -c fsync mydir/foo
	 * <power failure>
	 * mount fs, triggers fsync log replay
	 *
	 * We must make sure that when we fsync our inode foo we also log its
	 * parent inode, otherwise after log replay the parent still has the
	 * dentry with the "bar" name but our inode foo has a link count of 1
	 * and doesn't have an inode ref with the name "bar" anymore.
	 *
	 * Setting last_unlink_trans to last_trans is a pessimistic approach,
	 * but it guarantees correctness at the expense of occasional full
	 * transaction commits on fsync if our inode is a directory, or if our
	 * inode is not a directory, logging its parent unnecessarily.
	 */
	BTRFS_I(inode)->last_unlink_trans = BTRFS_I(inode)->last_trans;

	path->slots[0]++;
	if (inode->i_nlink != 1 ||
	    path->slots[0] >= btrfs_header_nritems(leaf))
		goto cache_acl;

	btrfs_item_key_to_cpu(leaf, &location, path->slots[0]);
	if (location.objectid != btrfs_ino(BTRFS_I(inode)))
		goto cache_acl;

	ptr = btrfs_item_ptr_offset(leaf, path->slots[0]);
	if (location.type == BTRFS_INODE_REF_KEY) {
		struct btrfs_inode_ref *ref;

		ref = (struct btrfs_inode_ref *)ptr;
		BTRFS_I(inode)->dir_index = btrfs_inode_ref_index(leaf, ref);
	} else if (location.type == BTRFS_INODE_EXTREF_KEY) {
		struct btrfs_inode_extref *extref;

		extref = (struct btrfs_inode_extref *)ptr;
		BTRFS_I(inode)->dir_index = btrfs_inode_extref_index(leaf,
								     extref);
	}
cache_acl:
	/*
	 * 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],
			btrfs_ino(BTRFS_I(inode)), &first_xattr_slot);
	if (first_xattr_slot != -1) {
		path->slots[0] = first_xattr_slot;
		ret = btrfs_load_inode_props(inode, path);
		if (ret)
			btrfs_err(fs_info,
				  "error loading props for ino %llu (root %llu): %d",
				  btrfs_ino(BTRFS_I(inode)),
				  root->root_key.objectid, ret);
	}
	if (path != in_path)
		btrfs_free_path(path);

	if (!maybe_acls)
		cache_no_acl(inode);

	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
		inode->i_mapping->a_ops = &btrfs_aops;
		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;
		inode->i_op = &btrfs_dir_inode_operations;
		break;
	case S_IFLNK:
		inode->i_op = &btrfs_symlink_inode_operations;
		inode_nohighmem(inode);
		inode->i_mapping->a_ops = &btrfs_aops;
		break;
	default:
		inode->i_op = &btrfs_special_inode_operations;
		init_special_inode(inode, inode->i_mode, rdev);
		break;
	}

	btrfs_sync_inode_flags_to_i_flags(inode);
	return 0;
}

/*
 * 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)
{
	struct btrfs_map_token token;

	btrfs_init_map_token(&token, leaf);

	btrfs_set_token_inode_uid(leaf, item, i_uid_read(inode), &token);
	btrfs_set_token_inode_gid(leaf, item, i_gid_read(inode), &token);
	btrfs_set_token_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size,
				   &token);
	btrfs_set_token_inode_mode(leaf, item, inode->i_mode, &token);
	btrfs_set_token_inode_nlink(leaf, item, inode->i_nlink, &token);

	btrfs_set_token_timespec_sec(leaf, &item->atime,
				     inode->i_atime.tv_sec, &token);
	btrfs_set_token_timespec_nsec(leaf, &item->atime,
				      inode->i_atime.tv_nsec, &token);

	btrfs_set_token_timespec_sec(leaf, &item->mtime,
				     inode->i_mtime.tv_sec, &token);
	btrfs_set_token_timespec_nsec(leaf, &item->mtime,
				      inode->i_mtime.tv_nsec, &token);

	btrfs_set_token_timespec_sec(leaf, &item->ctime,
				     inode->i_ctime.tv_sec, &token);
	btrfs_set_token_timespec_nsec(leaf, &item->ctime,
				      inode->i_ctime.tv_nsec, &token);

	btrfs_set_token_timespec_sec(leaf, &item->otime,
				     BTRFS_I(inode)->i_otime.tv_sec, &token);
	btrfs_set_token_timespec_nsec(leaf, &item->otime,
				      BTRFS_I(inode)->i_otime.tv_nsec, &token);

	btrfs_set_token_inode_nbytes(leaf, item, inode_get_bytes(inode),
				     &token);
	btrfs_set_token_inode_generation(leaf, item, BTRFS_I(inode)->generation,
					 &token);
	btrfs_set_token_inode_sequence(leaf, item, inode_peek_iversion(inode),
				       &token);
	btrfs_set_token_inode_transid(leaf, item, trans->transid, &token);
	btrfs_set_token_inode_rdev(leaf, item, inode->i_rdev, &token);
	btrfs_set_token_inode_flags(leaf, item, BTRFS_I(inode)->flags, &token);
	btrfs_set_token_inode_block_group(leaf, item, 0, &token);
}

/*
 * copy everything in the in-memory inode into the btree.
 */
static noinline int btrfs_update_inode_item(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();
	if (!path)
		return -ENOMEM;

	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;
	}

	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;
}

/*
 * 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_fs_info *fs_info = root->fs_info;
	int ret;

	/*
	 * If the inode is a free space inode, we can deadlock during commit
	 * if we put it into the delayed code.
	 *
	 * The data relocation inode should also be directly updated
	 * without delay
	 */
	if (!btrfs_is_free_space_inode(BTRFS_I(inode))
	    && root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID
	    && !test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags)) {
		btrfs_update_root_times(trans, root);

		ret = btrfs_delayed_update_inode(trans, root, inode);
		if (!ret)
			btrfs_set_inode_last_trans(trans, inode);
		return ret;
	}

	return btrfs_update_inode_item(trans, root, inode);
}

noinline int btrfs_update_inode_fallback(struct btrfs_trans_handle *trans,
					 struct btrfs_root *root,
					 struct inode *inode)
{
	int ret;

	ret = btrfs_update_inode(trans, root, inode);
	if (ret == -ENOSPC)
		return btrfs_update_inode_item(trans, root, inode);
	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
 */
static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans,
				struct btrfs_root *root,
				struct btrfs_inode *dir,
				struct btrfs_inode *inode,
				const char *name, int name_len)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_path *path;
	int ret = 0;
	struct btrfs_dir_item *di;
	u64 index;
	u64 ino = btrfs_ino(inode);
	u64 dir_ino = btrfs_ino(dir);

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

	path->leave_spinning = 1;
	di = btrfs_lookup_dir_item(trans, root, path, dir_ino,
				    name, name_len, -1);
	if (IS_ERR_OR_NULL(di)) {
		ret = di ? PTR_ERR(di) : -ENOENT;
		goto err;
	}
	ret = btrfs_delete_one_dir_name(trans, root, path, di);
	if (ret)
		goto err;
	btrfs_release_path(path);

	/*
	 * If we don't have dir index, we have to get it by looking up
	 * the inode ref, since we get the inode ref, remove it directly,
	 * it is unnecessary to do delayed deletion.
	 *
	 * But if we have dir index, needn't search inode ref to get it.
	 * Since the inode ref is close to the inode item, it is better
	 * that we delay to delete it, and just do this deletion when
	 * we update the inode item.
	 */
	if (inode->dir_index) {
		ret = btrfs_delayed_delete_inode_ref(inode);
		if (!ret) {
			index = inode->dir_index;
			goto skip_backref;
		}
	}

	ret = btrfs_del_inode_ref(trans, root, name, name_len, ino,
				  dir_ino, &index);
	if (ret) {
		btrfs_info(fs_info,
			"failed to delete reference to %.*s, inode %llu parent %llu",
			name_len, name, ino, dir_ino);
		btrfs_abort_transaction(trans, ret);
		goto err;
	}
skip_backref:
	ret = btrfs_delete_delayed_dir_index(trans, dir, index);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto err;
	}

	ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len, inode,
			dir_ino);
	if (ret != 0 && ret != -ENOENT) {
		btrfs_abort_transaction(trans, ret);
		goto err;
	}

	ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len, dir,
			index);
	if (ret == -ENOENT)
		ret = 0;
	else if (ret)
		btrfs_abort_transaction(trans, ret);

	/*
	 * If we have a pending delayed iput we could end up with the final iput
	 * being run in btrfs-cleaner context.  If we have enough of these built
	 * up we can end up burning a lot of time in btrfs-cleaner without any
	 * way to throttle the unlinks.  Since we're currently holding a ref on
	 * the inode we can run the delayed iput here without any issues as the
	 * final iput won't be done until after we drop the ref we're currently
	 * holding.
	 */
	btrfs_run_delayed_iput(fs_info, inode);
err:
	btrfs_free_path(path);
	if (ret)
		goto out;

	btrfs_i_size_write(dir, dir->vfs_inode.i_size - name_len * 2);
	inode_inc_iversion(&inode->vfs_inode);
	inode_inc_iversion(&dir->vfs_inode);
	inode->vfs_inode.i_ctime = dir->vfs_inode.i_mtime =
		dir->vfs_inode.i_ctime = current_time(&inode->vfs_inode);
	ret = btrfs_update_inode(trans, root, &dir->vfs_inode);
out:
	return ret;
}

int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
		       struct btrfs_root *root,
		       struct btrfs_inode *dir, struct btrfs_inode *inode,
		       const char *name, int name_len)
{
	int ret;
	ret = __btrfs_unlink_inode(trans, root, dir, inode, name, name_len);
	if (!ret) {
		drop_nlink(&inode->vfs_inode);
		ret = btrfs_update_inode(trans, root, &inode->vfs_inode);
	}
	return ret;
}

/*
 * helper to start transaction for unlink and rmdir.
 *
 * unlink and rmdir are special in btrfs, they do not always free space, so
 * if we cannot make our reservations the normal way try and see if there is
 * plenty of slack room in the global reserve to migrate, otherwise we cannot
 * allow the unlink to occur.
 */
static struct btrfs_trans_handle *__unlink_start_trans(struct inode *dir)
{
	struct btrfs_root *root = BTRFS_I(dir)->root;

	/*
	 * 1 for the possible orphan item
	 * 1 for the dir item
	 * 1 for the dir index
	 * 1 for the inode ref
	 * 1 for the inode
	 */
	return btrfs_start_transaction_fallback_global_rsv(root, 5, 5);
}

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 = d_inode(dentry);
	int ret;

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

	btrfs_record_unlink_dir(trans, BTRFS_I(dir), BTRFS_I(d_inode(dentry)),
			0);

	ret = btrfs_unlink_inode(trans, root, BTRFS_I(dir),
			BTRFS_I(d_inode(dentry)), dentry->d_name.name,
			dentry->d_name.len);
	if (ret)
		goto out;

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

out:
	btrfs_end_transaction(trans);
	btrfs_btree_balance_dirty(root->fs_info);
	return ret;
}

static int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
			       struct inode *dir, struct dentry *dentry)
{
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_inode *inode = BTRFS_I(d_inode(dentry));
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_dir_item *di;
	struct btrfs_key key;
	const char *name = dentry->d_name.name;
	int name_len = dentry->d_name.len;
	u64 index;
	int ret;
	u64 objectid;
	u64 dir_ino = btrfs_ino(BTRFS_I(dir));

	if (btrfs_ino(inode) == BTRFS_FIRST_FREE_OBJECTID) {
		objectid = inode->root->root_key.objectid;
	} else if (btrfs_ino(inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) {
		objectid = inode->location.objectid;
	} else {
		WARN_ON(1);
		return -EINVAL;
	}

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

	di = btrfs_lookup_dir_item(trans, root, path, dir_ino,
				   name, name_len, -1);
	if (IS_ERR_OR_NULL(di)) {
		ret = di ? PTR_ERR(di) : -ENOENT;
		goto out;
	}

	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);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out;
	}
	btrfs_release_path(path);

	/*
	 * This is a placeholder inode for a subvolume we didn't have a
	 * reference to at the time of the snapshot creation.  In the meantime
	 * we could have renamed the real subvol link into our snapshot, so
	 * depending on btrfs_del_root_ref to return -ENOENT here is incorret.
	 * Instead simply lookup the dir_index_item for this entry so we can
	 * remove it.  Otherwise we know we have a ref to the root and we can
	 * call btrfs_del_root_ref, and it _shouldn't_ fail.
	 */
	if (btrfs_ino(inode) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID) {
		di = btrfs_search_dir_index_item(root, path, dir_ino,
						 name, name_len);
		if (IS_ERR_OR_NULL(di)) {
			if (!di)
				ret = -ENOENT;
			else
				ret = PTR_ERR(di);
			btrfs_abort_transaction(trans, ret);
			goto out;
		}

		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
		index = key.offset;
		btrfs_release_path(path);
	} else {
		ret = btrfs_del_root_ref(trans, objectid,
					 root->root_key.objectid, dir_ino,
					 &index, name, name_len);
		if (ret) {
			btrfs_abort_transaction(trans, ret);
			goto out;
		}
	}

	ret = btrfs_delete_delayed_dir_index(trans, BTRFS_I(dir), index);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out;
	}

	btrfs_i_size_write(BTRFS_I(dir), dir->i_size - name_len * 2);
	inode_inc_iversion(dir);
	dir->i_mtime = dir->i_ctime = current_time(dir);
	ret = btrfs_update_inode_fallback(trans, root, dir);
	if (ret)
		btrfs_abort_transaction(trans, ret);
out:
	btrfs_free_path(path);
	return ret;
}

/*
 * Helper to check if the subvolume references other subvolumes or if it's
 * default.
 */
static noinline int may_destroy_subvol(struct btrfs_root *root)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_path *path;
	struct btrfs_dir_item *di;
	struct btrfs_key key;
	u64 dir_id;
	int ret;

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

	/* Make sure this root isn't set as the default subvol */
	dir_id = btrfs_super_root_dir(fs_info->super_copy);
	di = btrfs_lookup_dir_item(NULL, fs_info->tree_root, path,
				   dir_id, "default", 7, 0);
	if (di && !IS_ERR(di)) {
		btrfs_dir_item_key_to_cpu(path->nodes[0], di, &key);
		if (key.objectid == root->root_key.objectid) {
			ret = -EPERM;
			btrfs_err(fs_info,
				  "deleting default subvolume %llu is not allowed",
				  key.objectid);
			goto out;
		}
		btrfs_release_path(path);
	}

	key.objectid = root->root_key.objectid;
	key.type = BTRFS_ROOT_REF_KEY;
	key.offset = (u64)-1;

	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
	if (ret < 0)
		goto out;
	BUG_ON(ret == 0);

	ret = 0;
	if (path->slots[0] > 0) {
		path->slots[0]--;
		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
		if (key.objectid == root->root_key.objectid &&
		    key.type == BTRFS_ROOT_REF_KEY)
			ret = -ENOTEMPTY;
	}
out:
	btrfs_free_path(path);
	return ret;
}

/* Delete all dentries for inodes belonging to the root */
static void btrfs_prune_dentries(struct btrfs_root *root)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct rb_node *node;
	struct rb_node *prev;
	struct btrfs_inode *entry;
	struct inode *inode;
	u64 objectid = 0;

	if (!test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
		WARN_ON(btrfs_root_refs(&root->root_item) != 0);

	spin_lock(&root->inode_lock);
again:
	node = root->inode_tree.rb_node;
	prev = NULL;
	while (node) {
		prev = node;
		entry = rb_entry(node, struct btrfs_inode, rb_node);

		if (objectid < btrfs_ino(entry))
			node = node->rb_left;
		else if (objectid > btrfs_ino(entry))
			node = node->rb_right;
		else
			break;
	}
	if (!node) {
		while (prev) {
			entry = rb_entry(prev, struct btrfs_inode, rb_node);
			if (objectid <= btrfs_ino(entry)) {
				node = prev;
				break;
			}
			prev = rb_next(prev);
		}
	}
	while (node) {
		entry = rb_entry(node, struct btrfs_inode, rb_node);
		objectid = btrfs_ino(entry) + 1;
		inode = igrab(&entry->vfs_inode);
		if (inode) {
			spin_unlock(&root->inode_lock);
			if (atomic_read(&inode->i_count) > 1)
				d_prune_aliases(inode);
			/*
			 * btrfs_drop_inode will have it removed from the inode
			 * cache when its usage count hits zero.
			 */
			iput(inode);
			cond_resched();
			spin_lock(&root->inode_lock);
			goto again;
		}

		if (cond_resched_lock(&root->inode_lock))
			goto again;

		node = rb_next(node);
	}
	spin_unlock(&root->inode_lock);
}

int btrfs_delete_subvolume(struct inode *dir, struct dentry *dentry)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(dentry->d_sb);
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct inode *inode = d_inode(dentry);
	struct btrfs_root *dest = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	struct btrfs_block_rsv block_rsv;
	u64 root_flags;
	int ret;
	int err;

	/*
	 * Don't allow to delete a subvolume with send in progress. This is
	 * inside the inode lock so the error handling that has to drop the bit
	 * again is not run concurrently.
	 */
	spin_lock(&dest->root_item_lock);
	if (dest->send_in_progress) {
		spin_unlock(&dest->root_item_lock);
		btrfs_warn(fs_info,
			   "attempt to delete subvolume %llu during send",
			   dest->root_key.objectid);
		return -EPERM;
	}
	root_flags = btrfs_root_flags(&dest->root_item);
	btrfs_set_root_flags(&dest->root_item,
			     root_flags | BTRFS_ROOT_SUBVOL_DEAD);
	spin_unlock(&dest->root_item_lock);

	down_write(&fs_info->subvol_sem);

	err = may_destroy_subvol(dest);
	if (err)
		goto out_up_write;

	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
	/*
	 * One for dir inode,
	 * two for dir entries,
	 * two for root ref/backref.
	 */
	err = btrfs_subvolume_reserve_metadata(root, &block_rsv, 5, true);
	if (err)
		goto out_up_write;

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		err = PTR_ERR(trans);
		goto out_release;
	}
	trans->block_rsv = &block_rsv;
	trans->bytes_reserved = block_rsv.size;

	btrfs_record_snapshot_destroy(trans, BTRFS_I(dir));

	ret = btrfs_unlink_subvol(trans, dir, dentry);
	if (ret) {
		err = ret;
		btrfs_abort_transaction(trans, ret);
		goto out_end_trans;
	}

	btrfs_record_root_in_trans(trans, dest);

	memset(&dest->root_item.drop_progress, 0,
		sizeof(dest->root_item.drop_progress));
	dest->root_item.drop_level = 0;
	btrfs_set_root_refs(&dest->root_item, 0);

	if (!test_and_set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &dest->state)) {
		ret = btrfs_insert_orphan_item(trans,
					fs_info->tree_root,
					dest->root_key.objectid);
		if (ret) {
			btrfs_abort_transaction(trans, ret);
			err = ret;
			goto out_end_trans;
		}
	}

	ret = btrfs_uuid_tree_remove(trans, dest->root_item.uuid,
				  BTRFS_UUID_KEY_SUBVOL,
				  dest->root_key.objectid);
	if (ret && ret != -ENOENT) {
		btrfs_abort_transaction(trans, ret);
		err = ret;
		goto out_end_trans;
	}
	if (!btrfs_is_empty_uuid(dest->root_item.received_uuid)) {
		ret = btrfs_uuid_tree_remove(trans,
					  dest->root_item.received_uuid,
					  BTRFS_UUID_KEY_RECEIVED_SUBVOL,
					  dest->root_key.objectid);
		if (ret && ret != -ENOENT) {
			btrfs_abort_transaction(trans, ret);
			err = ret;
			goto out_end_trans;
		}
	}

out_end_trans:
	trans->block_rsv = NULL;
	trans->bytes_reserved = 0;
	ret = btrfs_end_transaction(trans);
	if (ret && !err)
		err = ret;
	inode->i_flags |= S_DEAD;
out_release:
	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
out_up_write:
	up_write(&fs_info->subvol_sem);
	if (err) {
		spin_lock(&dest->root_item_lock);
		root_flags = btrfs_root_flags(&dest->root_item);
		btrfs_set_root_flags(&dest->root_item,
				root_flags & ~BTRFS_ROOT_SUBVOL_DEAD);
		spin_unlock(&dest->root_item_lock);
	} else {
		d_invalidate(dentry);
		btrfs_prune_dentries(dest);
		ASSERT(dest->send_in_progress == 0);

		/* the last ref */
		if (dest->ino_cache_inode) {
			iput(dest->ino_cache_inode);
			dest->ino_cache_inode = NULL;
		}
	}

	return err;
}

static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
{
	struct inode *inode = d_inode(dentry);
	int err = 0;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_trans_handle *trans;
	u64 last_unlink_trans;

	if (inode->i_size > BTRFS_EMPTY_DIR_SIZE)
		return -ENOTEMPTY;
	if (btrfs_ino(BTRFS_I(inode)) == BTRFS_FIRST_FREE_OBJECTID)
		return btrfs_delete_subvolume(dir, dentry);

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

	if (unlikely(btrfs_ino(BTRFS_I(inode)) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)) {
		err = btrfs_unlink_subvol(trans, dir, dentry);
		goto out;
	}

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

	last_unlink_trans = BTRFS_I(inode)->last_unlink_trans;

	/* now the directory is empty */
	err = btrfs_unlink_inode(trans, root, BTRFS_I(dir),
			BTRFS_I(d_inode(dentry)), dentry->d_name.name,
			dentry->d_name.len);
	if (!err) {
		btrfs_i_size_write(BTRFS_I(inode), 0);
		/*
		 * Propagate the last_unlink_trans value of the deleted dir to
		 * its parent directory. This is to prevent an unrecoverable
		 * log tree in the case we do something like this:
		 * 1) create dir foo
		 * 2) create snapshot under dir foo
		 * 3) delete the snapshot
		 * 4) rmdir foo
		 * 5) mkdir foo
		 * 6) fsync foo or some file inside foo
		 */
		if (last_unlink_trans >= trans->transid)
			BTRFS_I(dir)->last_unlink_trans = last_unlink_trans;
	}
out:
	btrfs_end_transaction(trans);
	btrfs_btree_balance_dirty(root->fs_info);

	return err;
}

/*
 * Return this if we need to call truncate_block for the last bit of the
 * truncate.
 */
#define NEED_TRUNCATE_BLOCK 1

/*
 * this can truncate away extent items, csum items and directory items.
 * It starts at a high offset and removes keys until it can't find
 * any higher than new_size
 *
 * csum items that cross the new i_size are truncated to the new size
 * as well.
 *
 * min_type is the minimum key type to truncate down to.  If set to 0, this
 * will kill all the items on this inode, including the INODE_ITEM_KEY.
 */
int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
			       struct btrfs_root *root,
			       struct inode *inode,
			       u64 new_size, u32 min_type)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_file_extent_item *fi;
	struct btrfs_key key;
	struct btrfs_key found_key;
	u64 extent_start = 0;
	u64 extent_num_bytes = 0;
	u64 extent_offset = 0;
	u64 item_end = 0;
	u64 last_size = new_size;
	u32 found_type = (u8)-1;
	int found_extent;
	int del_item;
	int pending_del_nr = 0;
	int pending_del_slot = 0;
	int extent_type = -1;
	int ret;
	u64 ino = btrfs_ino(BTRFS_I(inode));
	u64 bytes_deleted = 0;
	bool be_nice = false;
	bool should_throttle = false;
	const u64 lock_start = ALIGN_DOWN(new_size, fs_info->sectorsize);
	struct extent_state *cached_state = NULL;

	BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY);

	/*
	 * for non-free space inodes and ref cows, we want to back off from
	 * time to time
	 */
	if (!btrfs_is_free_space_inode(BTRFS_I(inode)) &&
	    test_bit(BTRFS_ROOT_REF_COWS, &root->state))
		be_nice = true;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
	path->reada = READA_BACK;

	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
		lock_extent_bits(&BTRFS_I(inode)->io_tree, lock_start, (u64)-1,
				 &cached_state);

	/*
	 * We want to drop from the next block forward in case this new size is
	 * not block aligned since we will be keeping the last block of the
	 * extent just the way it is.
	 */
	if (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
	    root == fs_info->tree_root)
		btrfs_drop_extent_cache(BTRFS_I(inode), ALIGN(new_size,
					fs_info->sectorsize),
					(u64)-1, 0);

	/*
	 * This function is also used to drop the items in the log tree before
	 * we relog the inode, so if root != BTRFS_I(inode)->root, it means
	 * it is used to drop the logged items. So we shouldn't kill the delayed
	 * items.
	 */
	if (min_type == 0 && root == BTRFS_I(inode)->root)
		btrfs_kill_delayed_inode_items(BTRFS_I(inode));

	key.objectid = ino;
	key.offset = (u64)-1;
	key.type = (u8)-1;

search_again:
	/*
	 * with a 16K leaf size and 128MB extents, you can actually queue
	 * up a huge file in a single leaf.  Most of the time that
	 * bytes_deleted is > 0, it will be huge by the time we get here
	 */
	if (be_nice && bytes_deleted > SZ_32M &&
	    btrfs_should_end_transaction(trans)) {
		ret = -EAGAIN;
		goto out;
	}

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
	if (ret < 0)
		goto out;

	if (ret > 0) {
		ret = 0;
		/* there are no items in the tree for us to truncate, we're
		 * done
		 */
		if (path->slots[0] == 0)
			goto out;
		path->slots[0]--;
	}

	while (1) {
		fi = NULL;
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
		found_type = found_key.type;

		if (found_key.objectid != ino)
			break;

		if (found_type < min_type)
			break;

		item_end = found_key.offset;
		if (found_type == BTRFS_EXTENT_DATA_KEY) {
			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_INLINE) {
				item_end +=
				    btrfs_file_extent_num_bytes(leaf, fi);

				trace_btrfs_truncate_show_fi_regular(
					BTRFS_I(inode), leaf, fi,
					found_key.offset);
			} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
				item_end += btrfs_file_extent_ram_bytes(leaf,
									fi);

				trace_btrfs_truncate_show_fi_inline(
					BTRFS_I(inode), leaf, fi, path->slots[0],
					found_key.offset);
			}
			item_end--;
		}
		if (found_type > min_type) {
			del_item = 1;
		} else {
			if (item_end < new_size)
				break;
			if (found_key.offset >= new_size)
				del_item = 1;
			else
				del_item = 0;
		}
		found_extent = 0;
		/* FIXME, shrink the extent if the ref count is only 1 */
		if (found_type != BTRFS_EXTENT_DATA_KEY)
			goto delete;

		if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
			u64 num_dec;
			extent_start = btrfs_file_extent_disk_bytenr(leaf, fi);
			if (!del_item) {
				u64 orig_num_bytes =
					btrfs_file_extent_num_bytes(leaf, fi);
				extent_num_bytes = ALIGN(new_size -
						found_key.offset,
						fs_info->sectorsize);
				btrfs_set_file_extent_num_bytes(leaf, fi,
							 extent_num_bytes);
				num_dec = (orig_num_bytes -
					   extent_num_bytes);
				if (test_bit(BTRFS_ROOT_REF_COWS,
					     &root->state) &&
				    extent_start != 0)
					inode_sub_bytes(inode, num_dec);
				btrfs_mark_buffer_dirty(leaf);
			} else {
				extent_num_bytes =
					btrfs_file_extent_disk_num_bytes(leaf,
									 fi);
				extent_offset = found_key.offset -
					btrfs_file_extent_offset(leaf, fi);

				/* FIXME blocksize != 4096 */
				num_dec = btrfs_file_extent_num_bytes(leaf, fi);
				if (extent_start != 0) {
					found_extent = 1;
					if (test_bit(BTRFS_ROOT_REF_COWS,
						     &root->state))
						inode_sub_bytes(inode, num_dec);
				}
			}
		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
			/*
			 * we can't truncate inline items that have had
			 * special encodings
			 */
			if (!del_item &&
			    btrfs_file_extent_encryption(leaf, fi) == 0 &&
			    btrfs_file_extent_other_encoding(leaf, fi) == 0 &&
			    btrfs_file_extent_compression(leaf, fi) == 0) {
				u32 size = (u32)(new_size - found_key.offset);

				btrfs_set_file_extent_ram_bytes(leaf, fi, size);
				size = btrfs_file_extent_calc_inline_size(size);
				btrfs_truncate_item(path, size, 1);
			} else if (!del_item) {
				/*
				 * We have to bail so the last_size is set to
				 * just before this extent.
				 */
				ret = NEED_TRUNCATE_BLOCK;
				break;
			}

			if (test_bit(BTRFS_ROOT_REF_COWS, &root->state))
				inode_sub_bytes(inode, item_end + 1 - new_size);
		}
delete:
		if (del_item)
			last_size = found_key.offset;
		else
			last_size = new_size;
		if (del_item) {
			if (!pending_del_nr) {
				/* no pending yet, add ourselves */
				pending_del_slot = path->slots[0];
				pending_del_nr = 1;
			} else if (pending_del_nr &&
				   path->slots[0] + 1 == pending_del_slot) {
				/* hop on the pending chunk */
				pending_del_nr++;
				pending_del_slot = path->slots[0];
			} else {
				BUG();
			}
		} else {
			break;
		}
		should_throttle = false;

		if (found_extent &&
		    (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
		     root == fs_info->tree_root)) {
			struct btrfs_ref ref = { 0 };

			bytes_deleted += extent_num_bytes;

			btrfs_init_generic_ref(&ref, BTRFS_DROP_DELAYED_REF,
					extent_start, extent_num_bytes, 0);
			ref.real_root = root->root_key.objectid;
			btrfs_init_data_ref(&ref, btrfs_header_owner(leaf),
					ino, extent_offset);
			ret = btrfs_free_extent(trans, &ref);
			if (ret) {
				btrfs_abort_transaction(trans, ret);
				break;
			}
			if (be_nice) {
				if (btrfs_should_throttle_delayed_refs(trans))
					should_throttle = true;
			}
		}

		if (found_type == BTRFS_INODE_ITEM_KEY)
			break;

		if (path->slots[0] == 0 ||
		    path->slots[0] != pending_del_slot ||
		    should_throttle) {
			if (pending_del_nr) {
				ret = btrfs_del_items(trans, root, path,
						pending_del_slot,
						pending_del_nr);
				if (ret) {
					btrfs_abort_transaction(trans, ret);
					break;
				}
				pending_del_nr = 0;
			}
			btrfs_release_path(path);

			/*
			 * We can generate a lot of delayed refs, so we need to
			 * throttle every once and a while and make sure we're
			 * adding enough space to keep up with the work we are
			 * generating.  Since we hold a transaction here we
			 * can't flush, and we don't want to FLUSH_LIMIT because
			 * we could have generated too many delayed refs to
			 * actually allocate, so just bail if we're short and
			 * let the normal reservation dance happen higher up.
			 */
			if (should_throttle) {
				ret = btrfs_delayed_refs_rsv_refill(fs_info,
							BTRFS_RESERVE_NO_FLUSH);
				if (ret) {
					ret = -EAGAIN;
					break;
				}
			}
			goto search_again;
		} else {
			path->slots[0]--;
		}
	}
out:
	if (ret >= 0 && pending_del_nr) {
		int err;

		err = btrfs_del_items(trans, root, path, pending_del_slot,
				      pending_del_nr);
		if (err) {
			btrfs_abort_transaction(trans, err);
			ret = err;
		}
	}
	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
		ASSERT(last_size >= new_size);
		if (!ret && last_size > new_size)
			last_size = new_size;
		btrfs_ordered_update_i_size(inode, last_size, NULL);
		unlock_extent_cached(&BTRFS_I(inode)->io_tree, lock_start,
				     (u64)-1, &cached_state);
	}

	btrfs_free_path(path);
	return ret;
}

/*
 * btrfs_truncate_block - read, zero a chunk and write a block
 * @inode - inode that we're zeroing
 * @from - the offset to start zeroing
 * @len - the length to zero, 0 to zero the entire range respective to the
 *	offset
 * @front - zero up to the offset instead of from the offset on
 *
 * This will find the block for the "from" offset and cow the block and zero the
 * part we want to zero.  This is used with truncate and hole punching.
 */
int btrfs_truncate_block(struct inode *inode, loff_t from, loff_t len,
			int front)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct address_space *mapping = inode->i_mapping;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct btrfs_ordered_extent *ordered;
	struct extent_state *cached_state = NULL;
	struct extent_changeset *data_reserved = NULL;
	char *kaddr;
	u32 blocksize = fs_info->sectorsize;
	pgoff_t index = from >> PAGE_SHIFT;
	unsigned offset = from & (blocksize - 1);
	struct page *page;
	gfp_t mask = btrfs_alloc_write_mask(mapping);
	int ret = 0;
	u64 block_start;
	u64 block_end;

	if (IS_ALIGNED(offset, blocksize) &&
	    (!len || IS_ALIGNED(len, blocksize)))
		goto out;

	block_start = round_down(from, blocksize);
	block_end = block_start + blocksize - 1;

	ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
					   block_start, blocksize);
	if (ret)
		goto out;

again:
	page = find_or_create_page(mapping, index, mask);
	if (!page) {
		btrfs_delalloc_release_space(inode, data_reserved,
					     block_start, blocksize, true);
		btrfs_delalloc_release_extents(BTRFS_I(inode), blocksize);
		ret = -ENOMEM;
		goto out;
	}

	if (!PageUptodate(page)) {
		ret = btrfs_readpage(NULL, page);
		lock_page(page);
		if (page->mapping != mapping) {
			unlock_page(page);
			put_page(page);
			goto again;
		}
		if (!PageUptodate(page)) {
			ret = -EIO;
			goto out_unlock;
		}
	}
	wait_on_page_writeback(page);

	lock_extent_bits(io_tree, block_start, block_end, &cached_state);
	set_page_extent_mapped(page);

	ordered = btrfs_lookup_ordered_extent(inode, block_start);
	if (ordered) {
		unlock_extent_cached(io_tree, block_start, block_end,
				     &cached_state);
		unlock_page(page);
		put_page(page);
		btrfs_start_ordered_extent(inode, ordered, 1);
		btrfs_put_ordered_extent(ordered);
		goto again;
	}

	clear_extent_bit(&BTRFS_I(inode)->io_tree, block_start, block_end,
			 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
			 0, 0, &cached_state);

	ret = btrfs_set_extent_delalloc(inode, block_start, block_end, 0,
					&cached_state);
	if (ret) {
		unlock_extent_cached(io_tree, block_start, block_end,
				     &cached_state);
		goto out_unlock;
	}

	if (offset != blocksize) {
		if (!len)
			len = blocksize - offset;
		kaddr = kmap(page);
		if (front)
			memset(kaddr + (block_start - page_offset(page)),
				0, offset);
		else
			memset(kaddr + (block_start - page_offset(page)) +  offset,
				0, len);
		flush_dcache_page(page);
		kunmap(page);
	}
	ClearPageChecked(page);
	set_page_dirty(page);
	unlock_extent_cached(io_tree, block_start, block_end, &cached_state);

out_unlock:
	if (ret)
		btrfs_delalloc_release_space(inode, data_reserved, block_start,
					     blocksize, true);
	btrfs_delalloc_release_extents(BTRFS_I(inode), blocksize);
	unlock_page(page);
	put_page(page);
out:
	extent_changeset_free(data_reserved);
	return ret;
}

static int maybe_insert_hole(struct btrfs_root *root, struct inode *inode,
			     u64 offset, u64 len)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_trans_handle *trans;
	int ret;

	/*
	 * Still need to make sure the inode looks like it's been updated so
	 * that any holes get logged if we fsync.
	 */
	if (btrfs_fs_incompat(fs_info, NO_HOLES)) {
		BTRFS_I(inode)->last_trans = fs_info->generation;
		BTRFS_I(inode)->last_sub_trans = root->log_transid;
		BTRFS_I(inode)->last_log_commit = root->last_log_commit;
		return 0;
	}

	/*
	 * 1 - for the one we're dropping
	 * 1 - for the one we're adding
	 * 1 - for updating the inode.
	 */
	trans = btrfs_start_transaction(root, 3);
	if (IS_ERR(trans))
		return PTR_ERR(trans);

	ret = btrfs_drop_extents(trans, root, inode, offset, offset + len, 1);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		btrfs_end_transaction(trans);
		return ret;
	}

	ret = btrfs_insert_file_extent(trans, root, btrfs_ino(BTRFS_I(inode)),
			offset, 0, 0, len, 0, len, 0, 0, 0);
	if (ret)
		btrfs_abort_transaction(trans, ret);
	else
		btrfs_update_inode(trans, root, inode);
	btrfs_end_transaction(trans);
	return ret;
}

/*
 * This function puts in dummy file extents for the area we're creating a hole
 * for.  So if we are truncating this file to a larger size we need to insert
 * these file extents so that btrfs_get_extent will return a EXTENT_MAP_HOLE for
 * the range between oldsize and size
 */
int btrfs_cont_expand(struct inode *inode, loff_t oldsize, loff_t size)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_map *em = NULL;
	struct extent_state *cached_state = NULL;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	u64 hole_start = ALIGN(oldsize, fs_info->sectorsize);
	u64 block_end = ALIGN(size, fs_info->sectorsize);
	u64 last_byte;
	u64 cur_offset;
	u64 hole_size;
	int err = 0;

	/*
	 * If our size started in the middle of a block we need to zero out the
	 * rest of the block before we expand the i_size, otherwise we could
	 * expose stale data.
	 */
	err = btrfs_truncate_block(inode, oldsize, 0, 0);
	if (err)
		return err;

	if (size <= hole_start)
		return 0;

	btrfs_lock_and_flush_ordered_range(io_tree, BTRFS_I(inode), hole_start,
					   block_end - 1, &cached_state);
	cur_offset = hole_start;
	while (1) {
		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, cur_offset,
				      block_end - cur_offset);
		if (IS_ERR(em)) {
			err = PTR_ERR(em);
			em = NULL;
			break;
		}
		last_byte = min(extent_map_end(em), block_end);
		last_byte = ALIGN(last_byte, fs_info->sectorsize);
		if (!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
			struct extent_map *hole_em;
			hole_size = last_byte - cur_offset;

			err = maybe_insert_hole(root, inode, cur_offset,
						hole_size);
			if (err)
				break;
			btrfs_drop_extent_cache(BTRFS_I(inode), cur_offset,
						cur_offset + hole_size - 1, 0);
			hole_em = alloc_extent_map();
			if (!hole_em) {
				set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
					&BTRFS_I(inode)->runtime_flags);
				goto next;
			}
			hole_em->start = cur_offset;
			hole_em->len = hole_size;
			hole_em->orig_start = cur_offset;

			hole_em->block_start = EXTENT_MAP_HOLE;
			hole_em->block_len = 0;
			hole_em->orig_block_len = 0;
			hole_em->ram_bytes = hole_size;
			hole_em->compress_type = BTRFS_COMPRESS_NONE;
			hole_em->generation = fs_info->generation;

			while (1) {
				write_lock(&em_tree->lock);
				err = add_extent_mapping(em_tree, hole_em, 1);
				write_unlock(&em_tree->lock);
				if (err != -EEXIST)
					break;
				btrfs_drop_extent_cache(BTRFS_I(inode),
							cur_offset,
							cur_offset +
							hole_size - 1, 0);
			}
			free_extent_map(hole_em);
		}
next:
		free_extent_map(em);
		em = NULL;
		cur_offset = last_byte;
		if (cur_offset >= block_end)
			break;
	}
	free_extent_map(em);
	unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state);
	return err;
}

static int btrfs_setsize(struct inode *inode, struct iattr *attr)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	loff_t oldsize = i_size_read(inode);
	loff_t newsize = attr->ia_size;
	int mask = attr->ia_valid;
	int ret;

	/*
	 * The regular truncate() case without ATTR_CTIME and ATTR_MTIME is a
	 * special case where we need to update the times despite not having
	 * these flags set.  For all other operations the VFS set these flags
	 * explicitly if it wants a timestamp update.
	 */
	if (newsize != oldsize) {
		inode_inc_iversion(inode);
		if (!(mask & (ATTR_CTIME | ATTR_MTIME)))
			inode->i_ctime = inode->i_mtime =
				current_time(inode);
	}

	if (newsize > oldsize) {
		/*
		 * Don't do an expanding truncate while snapshotting is ongoing.
		 * This is to ensure the snapshot captures a fully consistent
		 * state of this file - if the snapshot captures this expanding
		 * truncation, it must capture all writes that happened before
		 * this truncation.
		 */
		btrfs_wait_for_snapshot_creation(root);
		ret = btrfs_cont_expand(inode, oldsize, newsize);
		if (ret) {
			btrfs_end_write_no_snapshotting(root);
			return ret;
		}

		trans = btrfs_start_transaction(root, 1);
		if (IS_ERR(trans)) {
			btrfs_end_write_no_snapshotting(root);
			return PTR_ERR(trans);
		}

		i_size_write(inode, newsize);
		btrfs_ordered_update_i_size(inode, i_size_read(inode), NULL);
		pagecache_isize_extended(inode, oldsize, newsize);
		ret = btrfs_update_inode(trans, root, inode);
		btrfs_end_write_no_snapshotting(root);
		btrfs_end_transaction(trans);
	} else {

		/*
		 * We're truncating a file that used to have good data down to
		 * zero. Make sure it gets into the ordered flush list so that
		 * any new writes get down to disk quickly.
		 */
		if (newsize == 0)
			set_bit(BTRFS_INODE_ORDERED_DATA_CLOSE,
				&BTRFS_I(inode)->runtime_flags);

		truncate_setsize(inode, newsize);

		/* Disable nonlocked read DIO to avoid the endless truncate */
		btrfs_inode_block_unlocked_dio(BTRFS_I(inode));
		inode_dio_wait(inode);
		btrfs_inode_resume_unlocked_dio(BTRFS_I(inode));

		ret = btrfs_truncate(inode, newsize == oldsize);
		if (ret && inode->i_nlink) {
			int err;

			/*
			 * Truncate failed, so fix up the in-memory size. We
			 * adjusted disk_i_size down as we removed extents, so
			 * wait for disk_i_size to be stable and then update the
			 * in-memory size to match.
			 */
			err = btrfs_wait_ordered_range(inode, 0, (u64)-1);
			if (err)
				return err;
			i_size_write(inode, BTRFS_I(inode)->disk_i_size);
		}
	}

	return ret;
}

static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
{
	struct inode *inode = d_inode(dentry);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int err;

	if (btrfs_root_readonly(root))
		return -EROFS;

	err = setattr_prepare(dentry, attr);
	if (err)
		return err;

	if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
		err = btrfs_setsize(inode, attr);
		if (err)
			return err;
	}

	if (attr->ia_valid) {
		setattr_copy(inode, attr);
		inode_inc_iversion(inode);
		err = btrfs_dirty_inode(inode);

		if (!err && attr->ia_valid & ATTR_MODE)
			err = posix_acl_chmod(inode, inode->i_mode);
	}

	return err;
}

/*
 * While truncating the inode pages during eviction, we get the VFS calling
 * btrfs_invalidatepage() against each page of the inode. This is slow because
 * the calls to btrfs_invalidatepage() result in a huge amount of calls to
 * lock_extent_bits() and clear_extent_bit(), which keep merging and splitting
 * extent_state structures over and over, wasting lots of time.
 *
 * Therefore if the inode is being evicted, let btrfs_invalidatepage() skip all
 * those expensive operations on a per page basis and do only the ordered io
 * finishing, while we release here the extent_map and extent_state structures,
 * without the excessive merging and splitting.
 */
static void evict_inode_truncate_pages(struct inode *inode)
{
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_map_tree *map_tree = &BTRFS_I(inode)->extent_tree;
	struct rb_node *node;

	ASSERT(inode->i_state & I_FREEING);
	truncate_inode_pages_final(&inode->i_data);

	write_lock(&map_tree->lock);
	while (!RB_EMPTY_ROOT(&map_tree->map.rb_root)) {
		struct extent_map *em;

		node = rb_first_cached(&map_tree->map);
		em = rb_entry(node, struct extent_map, rb_node);
		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
		clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
		remove_extent_mapping(map_tree, em);
		free_extent_map(em);
		if (need_resched()) {
			write_unlock(&map_tree->lock);
			cond_resched();
			write_lock(&map_tree->lock);
		}
	}
	write_unlock(&map_tree->lock);

	/*
	 * Keep looping until we have no more ranges in the io tree.
	 * We can have ongoing bios started by readpages (called from readahead)
	 * that have their endio callback (extent_io.c:end_bio_extent_readpage)
	 * still in progress (unlocked the pages in the bio but did not yet
	 * unlocked the ranges in the io tree). Therefore this means some
	 * ranges can still be locked and eviction started because before
	 * submitting those bios, which are executed by a separate task (work
	 * queue kthread), inode references (inode->i_count) were not taken
	 * (which would be dropped in the end io callback of each bio).
	 * Therefore here we effectively end up waiting for those bios and
	 * anyone else holding locked ranges without having bumped the inode's
	 * reference count - if we don't do it, when they access the inode's
	 * io_tree to unlock a range it may be too late, leading to an
	 * use-after-free issue.
	 */
	spin_lock(&io_tree->lock);
	while (!RB_EMPTY_ROOT(&io_tree->state)) {
		struct extent_state *state;
		struct extent_state *cached_state = NULL;
		u64 start;
		u64 end;
		unsigned state_flags;

		node = rb_first(&io_tree->state);
		state = rb_entry(node, struct extent_state, rb_node);
		start = state->start;
		end = state->end;
		state_flags = state->state;
		spin_unlock(&io_tree->lock);

		lock_extent_bits(io_tree, start, end, &cached_state);

		/*
		 * If still has DELALLOC flag, the extent didn't reach disk,
		 * and its reserved space won't be freed by delayed_ref.
		 * So we need to free its reserved space here.
		 * (Refer to comment in btrfs_invalidatepage, case 2)
		 *
		 * Note, end is the bytenr of last byte, so we need + 1 here.
		 */
		if (state_flags & EXTENT_DELALLOC)
			btrfs_qgroup_free_data(inode, NULL, start, end - start + 1);

		clear_extent_bit(io_tree, start, end,
				 EXTENT_LOCKED | EXTENT_DELALLOC |
				 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 1, 1,
				 &cached_state);

		cond_resched();
		spin_lock(&io_tree->lock);
	}
	spin_unlock(&io_tree->lock);
}

static struct btrfs_trans_handle *evict_refill_and_join(struct btrfs_root *root,
							struct btrfs_block_rsv *rsv)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
	struct btrfs_trans_handle *trans;
	u64 delayed_refs_extra = btrfs_calc_insert_metadata_size(fs_info, 1);
	int ret;

	/*
	 * Eviction should be taking place at some place safe because of our
	 * delayed iputs.  However the normal flushing code will run delayed
	 * iputs, so we cannot use FLUSH_ALL otherwise we'll deadlock.
	 *
	 * We reserve the delayed_refs_extra here again because we can't use
	 * btrfs_start_transaction(root, 0) for the same deadlocky reason as
	 * above.  We reserve our extra bit here because we generate a ton of
	 * delayed refs activity by truncating.
	 *
	 * If we cannot make our reservation we'll attempt to steal from the
	 * global reserve, because we really want to be able to free up space.
	 */
	ret = btrfs_block_rsv_refill(root, rsv, rsv->size + delayed_refs_extra,
				     BTRFS_RESERVE_FLUSH_EVICT);
	if (ret) {
		/*
		 * Try to steal from the global reserve if there is space for
		 * it.
		 */
		if (btrfs_check_space_for_delayed_refs(fs_info) ||
		    btrfs_block_rsv_migrate(global_rsv, rsv, rsv->size, 0)) {
			btrfs_warn(fs_info,
				   "could not allocate space for delete; will truncate on mount");
			return ERR_PTR(-ENOSPC);
		}
		delayed_refs_extra = 0;
	}

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans))
		return trans;

	if (delayed_refs_extra) {
		trans->block_rsv = &fs_info->trans_block_rsv;
		trans->bytes_reserved = delayed_refs_extra;
		btrfs_block_rsv_migrate(rsv, trans->block_rsv,
					delayed_refs_extra, 1);
	}
	return trans;
}

void btrfs_evict_inode(struct inode *inode)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_block_rsv *rsv;
	int ret;

	trace_btrfs_inode_evict(inode);

	if (!root) {
		clear_inode(inode);
		return;
	}

	evict_inode_truncate_pages(inode);

	if (inode->i_nlink &&
	    ((btrfs_root_refs(&root->root_item) != 0 &&
	      root->root_key.objectid != BTRFS_ROOT_TREE_OBJECTID) ||
	     btrfs_is_free_space_inode(BTRFS_I(inode))))
		goto no_delete;

	if (is_bad_inode(inode))
		goto no_delete;

	btrfs_free_io_failure_record(BTRFS_I(inode), 0, (u64)-1);

	if (test_bit(BTRFS_FS_LOG_RECOVERING, &fs_info->flags))
		goto no_delete;

	if (inode->i_nlink > 0) {
		BUG_ON(btrfs_root_refs(&root->root_item) != 0 &&
		       root->root_key.objectid != BTRFS_ROOT_TREE_OBJECTID);
		goto no_delete;
	}

	ret = btrfs_commit_inode_delayed_inode(BTRFS_I(inode));
	if (ret)
		goto no_delete;

	rsv = btrfs_alloc_block_rsv(fs_info, BTRFS_BLOCK_RSV_TEMP);
	if (!rsv)
		goto no_delete;
	rsv->size = btrfs_calc_metadata_size(fs_info, 1);
	rsv->failfast = 1;

	btrfs_i_size_write(BTRFS_I(inode), 0);

	while (1) {
		trans = evict_refill_and_join(root, rsv);
		if (IS_ERR(trans))
			goto free_rsv;

		trans->block_rsv = rsv;

		ret = btrfs_truncate_inode_items(trans, root, inode, 0, 0);
		trans->block_rsv = &fs_info->trans_block_rsv;
		btrfs_end_transaction(trans);
		btrfs_btree_balance_dirty(fs_info);
		if (ret && ret != -ENOSPC && ret != -EAGAIN)
			goto free_rsv;
		else if (!ret)
			break;
	}

	/*
	 * Errors here aren't a big deal, it just means we leave orphan items in
	 * the tree. They will be cleaned up on the next mount. If the inode
	 * number gets reused, cleanup deletes the orphan item without doing
	 * anything, and unlink reuses the existing orphan item.
	 *
	 * If it turns out that we are dropping too many of these, we might want
	 * to add a mechanism for retrying these after a commit.
	 */
	trans = evict_refill_and_join(root, rsv);
	if (!IS_ERR(trans)) {
		trans->block_rsv = rsv;
		btrfs_orphan_del(trans, BTRFS_I(inode));
		trans->block_rsv = &fs_info->trans_block_rsv;
		btrfs_end_transaction(trans);
	}

	if (!(root == fs_info->tree_root ||
	      root->root_key.objectid == BTRFS_TREE_RELOC_OBJECTID))
		btrfs_return_ino(root, btrfs_ino(BTRFS_I(inode)));

free_rsv:
	btrfs_free_block_rsv(fs_info, rsv);
no_delete:
	/*
	 * If we didn't successfully delete, the orphan item will still be in
	 * the tree and we'll retry on the next mount. Again, we might also want
	 * to retry these periodically in the future.
	 */
	btrfs_remove_delayed_node(BTRFS_I(inode));
	clear_inode(inode);
}

/*
 * Return the key found in the dir entry in the location pointer, fill @type
 * with BTRFS_FT_*, and return 0.
 *
 * If no dir entries were found, returns -ENOENT.
 * If found a corrupted location in dir entry, returns -EUCLEAN.
 */
static int btrfs_inode_by_name(struct inode *dir, struct dentry *dentry,
			       struct btrfs_key *location, u8 *type)
{
	const char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;
	struct btrfs_dir_item *di;
	struct btrfs_path *path;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	int ret = 0;

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

	di = btrfs_lookup_dir_item(NULL, root, path, btrfs_ino(BTRFS_I(dir)),
			name, namelen, 0);
	if (IS_ERR_OR_NULL(di)) {
		ret = di ? PTR_ERR(di) : -ENOENT;
		goto out;
	}

	btrfs_dir_item_key_to_cpu(path->nodes[0], di, location);
	if (location->type != BTRFS_INODE_ITEM_KEY &&
	    location->type != BTRFS_ROOT_ITEM_KEY) {
		ret = -EUCLEAN;
		btrfs_warn(root->fs_info,
"%s gets something invalid in DIR_ITEM (name %s, directory ino %llu, location(%llu %u %llu))",
			   __func__, name, btrfs_ino(BTRFS_I(dir)),
			   location->objectid, location->type, location->offset);
	}
	if (!ret)
		*type = btrfs_dir_type(path->nodes[0], di);
out:
	btrfs_free_path(path);
	return ret;
}

/*
 * when we hit a tree root in a directory, the btrfs part of the inode
 * needs to be changed to reflect the root directory of the tree root.  This
 * is kind of like crossing a mount point.
 */
static int fixup_tree_root_location(struct btrfs_fs_info *fs_info,
				    struct inode *dir,
				    struct dentry *dentry,
				    struct btrfs_key *location,
				    struct btrfs_root **sub_root)
{
	struct btrfs_path *path;
	struct btrfs_root *new_root;
	struct btrfs_root_ref *ref;
	struct extent_buffer *leaf;
	struct btrfs_key key;
	int ret;
	int err = 0;

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

	err = -ENOENT;
	key.objectid = BTRFS_I(dir)->root->root_key.objectid;
	key.type = BTRFS_ROOT_REF_KEY;
	key.offset = location->objectid;

	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
	if (ret) {
		if (ret < 0)
			err = ret;
		goto out;
	}

	leaf = path->nodes[0];
	ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_root_ref);
	if (btrfs_root_ref_dirid(leaf, ref) != btrfs_ino(BTRFS_I(dir)) ||
	    btrfs_root_ref_name_len(leaf, ref) != dentry->d_name.len)
		goto out;

	ret = memcmp_extent_buffer(leaf, dentry->d_name.name,
				   (unsigned long)(ref + 1),
				   dentry->d_name.len);
	if (ret)
		goto out;

	btrfs_release_path(path);

	new_root = btrfs_read_fs_root_no_name(fs_info, location);
	if (IS_ERR(new_root)) {
		err = PTR_ERR(new_root);
		goto out;
	}

	*sub_root = new_root;
	location->objectid = btrfs_root_dirid(&new_root->root_item);
	location->type = BTRFS_INODE_ITEM_KEY;
	location->offset = 0;
	err = 0;
out:
	btrfs_free_path(path);
	return err;
}

static void inode_tree_add(struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_inode *entry;
	struct rb_node **p;
	struct rb_node *parent;
	struct rb_node *new = &BTRFS_I(inode)->rb_node;
	u64 ino = btrfs_ino(BTRFS_I(inode));

	if (inode_unhashed(inode))
		return;
	parent = NULL;
	spin_lock(&root->inode_lock);
	p = &root->inode_tree.rb_node;
	while (*p) {
		parent = *p;
		entry = rb_entry(parent, struct btrfs_inode, rb_node);

		if (ino < btrfs_ino(entry))
			p = &parent->rb_left;
		else if (ino > btrfs_ino(entry))
			p = &parent->rb_right;
		else {
			WARN_ON(!(entry->vfs_inode.i_state &
				  (I_WILL_FREE | I_FREEING)));
			rb_replace_node(parent, new, &root->inode_tree);
			RB_CLEAR_NODE(parent);
			spin_unlock(&root->inode_lock);
			return;
		}
	}
	rb_link_node(new, parent, p);
	rb_insert_color(new, &root->inode_tree);
	spin_unlock(&root->inode_lock);
}

static void inode_tree_del(struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int empty = 0;

	spin_lock(&root->inode_lock);
	if (!RB_EMPTY_NODE(&BTRFS_I(inode)->rb_node)) {
		rb_erase(&BTRFS_I(inode)->rb_node, &root->inode_tree);
		RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
		empty = RB_EMPTY_ROOT(&root->inode_tree);
	}
	spin_unlock(&root->inode_lock);

	if (empty && btrfs_root_refs(&root->root_item) == 0) {
		spin_lock(&root->inode_lock);
		empty = RB_EMPTY_ROOT(&root->inode_tree);
		spin_unlock(&root->inode_lock);
		if (empty)
			btrfs_add_dead_root(root);
	}
}


static int btrfs_init_locked_inode(struct inode *inode, void *p)
{
	struct btrfs_iget_args *args = p;
	inode->i_ino = args->location->objectid;
	memcpy(&BTRFS_I(inode)->location, args->location,
	       sizeof(*args->location));
	BTRFS_I(inode)->root = args->root;
	return 0;
}

static int btrfs_find_actor(struct inode *inode, void *opaque)
{
	struct btrfs_iget_args *args = opaque;
	return args->location->objectid == BTRFS_I(inode)->location.objectid &&
		args->root == BTRFS_I(inode)->root;
}

static struct inode *btrfs_iget_locked(struct super_block *s,
				       struct btrfs_key *location,
				       struct btrfs_root *root)
{
	struct inode *inode;
	struct btrfs_iget_args args;
	unsigned long hashval = btrfs_inode_hash(location->objectid, root);

	args.location = location;
	args.root = root;

	inode = iget5_locked(s, hashval, btrfs_find_actor,
			     btrfs_init_locked_inode,
			     (void *)&args);
	return inode;
}

/*
 * Get an inode object given its location and corresponding root.
 * Path can be preallocated to prevent recursing back to iget through
 * allocator. NULL is also valid but may require an additional allocation
 * later.
 */
struct inode *btrfs_iget_path(struct super_block *s, struct btrfs_key *location,
			      struct btrfs_root *root, struct btrfs_path *path)
{
	struct inode *inode;

	inode = btrfs_iget_locked(s, location, root);
	if (!inode)
		return ERR_PTR(-ENOMEM);

	if (inode->i_state & I_NEW) {
		int ret;

		ret = btrfs_read_locked_inode(inode, path);
		if (!ret) {
			inode_tree_add(inode);
			unlock_new_inode(inode);
		} else {
			iget_failed(inode);
			/*
			 * ret > 0 can come from btrfs_search_slot called by
			 * btrfs_read_locked_inode, this means the inode item
			 * was not found.
			 */
			if (ret > 0)
				ret = -ENOENT;
			inode = ERR_PTR(ret);
		}
	}

	return inode;
}

struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location,
			 struct btrfs_root *root)
{
	return btrfs_iget_path(s, location, root, NULL);
}

static struct inode *new_simple_dir(struct super_block *s,
				    struct btrfs_key *key,
				    struct btrfs_root *root)
{
	struct inode *inode = new_inode(s);

	if (!inode)
		return ERR_PTR(-ENOMEM);

	BTRFS_I(inode)->root = root;
	memcpy(&BTRFS_I(inode)->location, key, sizeof(*key));
	set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags);

	inode->i_ino = BTRFS_EMPTY_SUBVOL_DIR_OBJECTID;
	/*
	 * We only need lookup, the rest is read-only and there's no inode
	 * associated with the dentry
	 */
	inode->i_op = &simple_dir_inode_operations;
	inode->i_opflags &= ~IOP_XATTR;
	inode->i_fop = &simple_dir_operations;
	inode->i_mode = S_IFDIR | S_IRUGO | S_IWUSR | S_IXUGO;
	inode->i_mtime = current_time(inode);
	inode->i_atime = inode->i_mtime;
	inode->i_ctime = inode->i_mtime;
	BTRFS_I(inode)->i_otime = inode->i_mtime;

	return inode;
}

static inline u8 btrfs_inode_type(struct inode *inode)
{
	/*
	 * Compile-time asserts that generic FT_* types still match
	 * BTRFS_FT_* types
	 */
	BUILD_BUG_ON(BTRFS_FT_UNKNOWN != FT_UNKNOWN);
	BUILD_BUG_ON(BTRFS_FT_REG_FILE != FT_REG_FILE);
	BUILD_BUG_ON(BTRFS_FT_DIR != FT_DIR);
	BUILD_BUG_ON(BTRFS_FT_CHRDEV != FT_CHRDEV);
	BUILD_BUG_ON(BTRFS_FT_BLKDEV != FT_BLKDEV);
	BUILD_BUG_ON(BTRFS_FT_FIFO != FT_FIFO);
	BUILD_BUG_ON(BTRFS_FT_SOCK != FT_SOCK);
	BUILD_BUG_ON(BTRFS_FT_SYMLINK != FT_SYMLINK);

	return fs_umode_to_ftype(inode->i_mode);
}

struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
	struct inode *inode;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_root *sub_root = root;
	struct btrfs_key location;
	u8 di_type = 0;
	int index;
	int ret = 0;

	if (dentry->d_name.len > BTRFS_NAME_LEN)
		return ERR_PTR(-ENAMETOOLONG);

	ret = btrfs_inode_by_name(dir, dentry, &location, &di_type);
	if (ret < 0)
		return ERR_PTR(ret);

	if (location.type == BTRFS_INODE_ITEM_KEY) {
		inode = btrfs_iget(dir->i_sb, &location, root);
		if (IS_ERR(inode))
			return inode;

		/* Do extra check against inode mode with di_type */
		if (btrfs_inode_type(inode) != di_type) {
			btrfs_crit(fs_info,
"inode mode mismatch with dir: inode mode=0%o btrfs type=%u dir type=%u",
				  inode->i_mode, btrfs_inode_type(inode),
				  di_type);
			iput(inode);
			return ERR_PTR(-EUCLEAN);
		}
		return inode;
	}

	index = srcu_read_lock(&fs_info->subvol_srcu);
	ret = fixup_tree_root_location(fs_info, dir, dentry,
				       &location, &sub_root);
	if (ret < 0) {
		if (ret != -ENOENT)
			inode = ERR_PTR(ret);
		else
			inode = new_simple_dir(dir->i_sb, &location, sub_root);
	} else {
		inode = btrfs_iget(dir->i_sb, &location, sub_root);
	}
	srcu_read_unlock(&fs_info->subvol_srcu, index);

	if (!IS_ERR(inode) && root != sub_root) {
		down_read(&fs_info->cleanup_work_sem);
		if (!sb_rdonly(inode->i_sb))
			ret = btrfs_orphan_cleanup(sub_root);
		up_read(&fs_info->cleanup_work_sem);
		if (ret) {
			iput(inode);
			inode = ERR_PTR(ret);
		}
	}

	return inode;
}

static int btrfs_dentry_delete(const struct dentry *dentry)
{
	struct btrfs_root *root;
	struct inode *inode = d_inode(dentry);

	if (!inode && !IS_ROOT(dentry))
		inode = d_inode(dentry->d_parent);

	if (inode) {
		root = BTRFS_I(inode)->root;
		if (btrfs_root_refs(&root->root_item) == 0)
			return 1;

		if (btrfs_ino(BTRFS_I(inode)) == BTRFS_EMPTY_SUBVOL_DIR_OBJECTID)
			return 1;
	}
	return 0;
}

static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry,
				   unsigned int flags)
{
	struct inode *inode = btrfs_lookup_dentry(dir, dentry);

	if (inode == ERR_PTR(-ENOENT))
		inode = NULL;
	return d_splice_alias(inode, dentry);
}

/*
 * All this infrastructure exists because dir_emit can fault, and we are holding
 * the tree lock when doing readdir.  For now just allocate a buffer and copy
 * our information into that, and then dir_emit from the buffer.  This is
 * similar to what NFS does, only we don't keep the buffer around in pagecache
 * because I'm afraid I'll mess that up.  Long term we need to make filldir do
 * copy_to_user_inatomic so we don't have to worry about page faulting under the
 * tree lock.
 */
static int btrfs_opendir(struct inode *inode, struct file *file)
{
	struct btrfs_file_private *private;

	private = kzalloc(sizeof(struct btrfs_file_private), GFP_KERNEL);
	if (!private)
		return -ENOMEM;
	private->filldir_buf = kzalloc(PAGE_SIZE, GFP_KERNEL);
	if (!private->filldir_buf) {
		kfree(private);
		return -ENOMEM;
	}
	file->private_data = private;
	return 0;
}

struct dir_entry {
	u64 ino;
	u64 offset;
	unsigned type;
	int name_len;
};

static int btrfs_filldir(void *addr, int entries, struct dir_context *ctx)
{
	while (entries--) {
		struct dir_entry *entry = addr;
		char *name = (char *)(entry + 1);

		ctx->pos = get_unaligned(&entry->offset);
		if (!dir_emit(ctx, name, get_unaligned(&entry->name_len),
					 get_unaligned(&entry->ino),
					 get_unaligned(&entry->type)))
			return 1;
		addr += sizeof(struct dir_entry) +
			get_unaligned(&entry->name_len);
		ctx->pos++;
	}
	return 0;
}

static int btrfs_real_readdir(struct file *file, struct dir_context *ctx)
{
	struct inode *inode = file_inode(file);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_file_private *private = file->private_data;
	struct btrfs_dir_item *di;
	struct btrfs_key key;
	struct btrfs_key found_key;
	struct btrfs_path *path;
	void *addr;
	struct list_head ins_list;
	struct list_head del_list;
	int ret;
	struct extent_buffer *leaf;
	int slot;
	char *name_ptr;
	int name_len;
	int entries = 0;
	int total_len = 0;
	bool put = false;
	struct btrfs_key location;

	if (!dir_emit_dots(file, ctx))
		return 0;

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

	addr = private->filldir_buf;
	path->reada = READA_FORWARD;

	INIT_LIST_HEAD(&ins_list);
	INIT_LIST_HEAD(&del_list);
	put = btrfs_readdir_get_delayed_items(inode, &ins_list, &del_list);

again:
	key.type = BTRFS_DIR_INDEX_KEY;
	key.offset = ctx->pos;
	key.objectid = btrfs_ino(BTRFS_I(inode));

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto err;

	while (1) {
		struct dir_entry *entry;

		leaf = path->nodes[0];
		slot = path->slots[0];
		if (slot >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0)
				goto err;
			else if (ret > 0)
				break;
			continue;
		}

		btrfs_item_key_to_cpu(leaf, &found_key, slot);

		if (found_key.objectid != key.objectid)
			break;
		if (found_key.type != BTRFS_DIR_INDEX_KEY)
			break;
		if (found_key.offset < ctx->pos)
			goto next;
		if (btrfs_should_delete_dir_index(&del_list, found_key.offset))
			goto next;
		di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
		name_len = btrfs_dir_name_len(leaf, di);
		if ((total_len + sizeof(struct dir_entry) + name_len) >=
		    PAGE_SIZE) {
			btrfs_release_path(path);
			ret = btrfs_filldir(private->filldir_buf, entries, ctx);
			if (ret)
				goto nopos;
			addr = private->filldir_buf;
			entries = 0;
			total_len = 0;
			goto again;
		}

		entry = addr;
		put_unaligned(name_len, &entry->name_len);
		name_ptr = (char *)(entry + 1);
		read_extent_buffer(leaf, name_ptr, (unsigned long)(di + 1),
				   name_len);
		put_unaligned(fs_ftype_to_dtype(btrfs_dir_type(leaf, di)),
				&entry->type);
		btrfs_dir_item_key_to_cpu(leaf, di, &location);
		put_unaligned(location.objectid, &entry->ino);
		put_unaligned(found_key.offset, &entry->offset);
		entries++;
		addr += sizeof(struct dir_entry) + name_len;
		total_len += sizeof(struct dir_entry) + name_len;
next:
		path->slots[0]++;
	}
	btrfs_release_path(path);

	ret = btrfs_filldir(private->filldir_buf, entries, ctx);
	if (ret)
		goto nopos;

	ret = btrfs_readdir_delayed_dir_index(ctx, &ins_list);
	if (ret)
		goto nopos;

	/*
	 * Stop new entries from being returned after we return the last
	 * entry.
	 *
	 * New directory entries are assigned a strictly increasing
	 * offset.  This means that new entries created during readdir
	 * are *guaranteed* to be seen in the future by that readdir.
	 * This has broken buggy programs which operate on names as
	 * they're returned by readdir.  Until we re-use freed offsets
	 * we have this hack to stop new entries from being returned
	 * under the assumption that they'll never reach this huge
	 * offset.
	 *
	 * This is being careful not to overflow 32bit loff_t unless the
	 * last entry requires it because doing so has broken 32bit apps
	 * in the past.
	 */
	if (ctx->pos >= INT_MAX)
		ctx->pos = LLONG_MAX;
	else
		ctx->pos = INT_MAX;
nopos:
	ret = 0;
err:
	if (put)
		btrfs_readdir_put_delayed_items(inode, &ins_list, &del_list);
	btrfs_free_path(path);
	return ret;
}

/*
 * This is somewhat expensive, updating the tree every time the
 * inode changes.  But, it is most likely to find the inode in cache.
 * FIXME, needs more benchmarking...there are no reasons other than performance
 * to keep or drop this code.
 */
static int btrfs_dirty_inode(struct inode *inode)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	int ret;

	if (test_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags))
		return 0;

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans))
		return PTR_ERR(trans);

	ret = btrfs_update_inode(trans, root, inode);
	if (ret && ret == -ENOSPC) {
		/* whoops, lets try again with the full transaction */
		btrfs_end_transaction(trans);
		trans = btrfs_start_transaction(root, 1);
		if (IS_ERR(trans))
			return PTR_ERR(trans);

		ret = btrfs_update_inode(trans, root, inode);
	}
	btrfs_end_transaction(trans);
	if (BTRFS_I(inode)->delayed_node)
		btrfs_balance_delayed_items(fs_info);

	return ret;
}

/*
 * This is a copy of file_update_time.  We need this so we can return error on
 * ENOSPC for updating the inode in the case of file write and mmap writes.
 */
static int btrfs_update_time(struct inode *inode, struct timespec64 *now,
			     int flags)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	bool dirty = flags & ~S_VERSION;

	if (btrfs_root_readonly(root))
		return -EROFS;

	if (flags & S_VERSION)
		dirty |= inode_maybe_inc_iversion(inode, dirty);
	if (flags & S_CTIME)
		inode->i_ctime = *now;
	if (flags & S_MTIME)
		inode->i_mtime = *now;
	if (flags & S_ATIME)
		inode->i_atime = *now;
	return dirty ? btrfs_dirty_inode(inode) : 0;
}

/*
 * find the highest existing sequence number in a directory
 * and then set the in-memory index_cnt variable to reflect
 * free sequence numbers
 */
static int btrfs_set_inode_index_count(struct btrfs_inode *inode)
{
	struct btrfs_root *root = inode->root;
	struct btrfs_key key, found_key;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	int ret;

	key.objectid = btrfs_ino(inode);
	key.type = BTRFS_DIR_INDEX_KEY;
	key.offset = (u64)-1;

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

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto out;
	/* FIXME: we should be able to handle this */
	if (ret == 0)
		goto out;
	ret = 0;

	/*
	 * MAGIC NUMBER EXPLANATION:
	 * since we search a directory based on f_pos we have to start at 2
	 * since '.' and '..' have f_pos of 0 and 1 respectively, so everybody
	 * else has to start at 2
	 */
	if (path->slots[0] == 0) {
		inode->index_cnt = 2;
		goto out;
	}

	path->slots[0]--;

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

	if (found_key.objectid != btrfs_ino(inode) ||
	    found_key.type != BTRFS_DIR_INDEX_KEY) {
		inode->index_cnt = 2;
		goto out;
	}

	inode->index_cnt = found_key.offset + 1;
out:
	btrfs_free_path(path);
	return ret;
}

/*
 * helper to find a free sequence number in a given directory.  This current
 * code is very simple, later versions will do smarter things in the btree
 */
int btrfs_set_inode_index(struct btrfs_inode *dir, u64 *index)
{
	int ret = 0;

	if (dir->index_cnt == (u64)-1) {
		ret = btrfs_inode_delayed_dir_index_count(dir);
		if (ret) {
			ret = btrfs_set_inode_index_count(dir);
			if (ret)
				return ret;
		}
	}

	*index = dir->index_cnt;
	dir->index_cnt++;

	return ret;
}

static int btrfs_insert_inode_locked(struct inode *inode)
{
	struct btrfs_iget_args args;
	args.location = &BTRFS_I(inode)->location;
	args.root = BTRFS_I(inode)->root;

	return insert_inode_locked4(inode,
		   btrfs_inode_hash(inode->i_ino, BTRFS_I(inode)->root),
		   btrfs_find_actor, &args);
}

/*
 * Inherit flags from the parent inode.
 *
 * Currently only the compression flags and the cow flags are inherited.
 */
static void btrfs_inherit_iflags(struct inode *inode, struct inode *dir)
{
	unsigned int flags;

	if (!dir)
		return;

	flags = BTRFS_I(dir)->flags;

	if (flags & BTRFS_INODE_NOCOMPRESS) {
		BTRFS_I(inode)->flags &= ~BTRFS_INODE_COMPRESS;
		BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
	} else if (flags & BTRFS_INODE_COMPRESS) {
		BTRFS_I(inode)->flags &= ~BTRFS_INODE_NOCOMPRESS;
		BTRFS_I(inode)->flags |= BTRFS_INODE_COMPRESS;
	}

	if (flags & BTRFS_INODE_NODATACOW) {
		BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW;
		if (S_ISREG(inode->i_mode))
			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
	}

	btrfs_sync_inode_flags_to_i_flags(inode);
}

static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
				     struct btrfs_root *root,
				     struct inode *dir,
				     const char *name, int name_len,
				     u64 ref_objectid, u64 objectid,
				     umode_t mode, u64 *index)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct inode *inode;
	struct btrfs_inode_item *inode_item;
	struct btrfs_key *location;
	struct btrfs_path *path;
	struct btrfs_inode_ref *ref;
	struct btrfs_key key[2];
	u32 sizes[2];
	int nitems = name ? 2 : 1;
	unsigned long ptr;
	unsigned int nofs_flag;
	int ret;

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

	nofs_flag = memalloc_nofs_save();
	inode = new_inode(fs_info->sb);
	memalloc_nofs_restore(nofs_flag);
	if (!inode) {
		btrfs_free_path(path);
		return ERR_PTR(-ENOMEM);
	}

	/*
	 * O_TMPFILE, set link count to 0, so that after this point,
	 * we fill in an inode item with the correct link count.
	 */
	if (!name)
		set_nlink(inode, 0);

	/*
	 * we have to initialize this early, so we can reclaim the inode
	 * number if we fail afterwards in this function.
	 */
	inode->i_ino = objectid;

	if (dir && name) {
		trace_btrfs_inode_request(dir);

		ret = btrfs_set_inode_index(BTRFS_I(dir), index);
		if (ret) {
			btrfs_free_path(path);
			iput(inode);
			return ERR_PTR(ret);
		}
	} else if (dir) {
		*index = 0;
	}
	/*
	 * index_cnt is ignored for everything but a dir,
	 * btrfs_set_inode_index_count has an explanation for the magic
	 * number
	 */
	BTRFS_I(inode)->index_cnt = 2;
	BTRFS_I(inode)->dir_index = *index;
	BTRFS_I(inode)->root = root;
	BTRFS_I(inode)->generation = trans->transid;
	inode->i_generation = BTRFS_I(inode)->generation;

	/*
	 * We could have gotten an inode number from somebody who was fsynced
	 * and then removed in this same transaction, so let's just set full
	 * sync since it will be a full sync anyway and this will blow away the
	 * old info in the log.
	 */
	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);

	key[0].objectid = objectid;
	key[0].type = BTRFS_INODE_ITEM_KEY;
	key[0].offset = 0;

	sizes[0] = sizeof(struct btrfs_inode_item);

	if (name) {
		/*
		 * Start new inodes with an inode_ref. This is slightly more
		 * efficient for small numbers of hard links since they will
		 * be packed into one item. Extended refs will kick in if we
		 * add more hard links than can fit in the ref item.
		 */
		key[1].objectid = objectid;
		key[1].type = BTRFS_INODE_REF_KEY;
		key[1].offset = ref_objectid;

		sizes[1] = name_len + sizeof(*ref);
	}

	location = &BTRFS_I(inode)->location;
	location->objectid = objectid;
	location->offset = 0;
	location->type = BTRFS_INODE_ITEM_KEY;

	ret = btrfs_insert_inode_locked(inode);
	if (ret < 0) {
		iput(inode);
		goto fail;
	}

	path->leave_spinning = 1;
	ret = btrfs_insert_empty_items(trans, root, path, key, sizes, nitems);
	if (ret != 0)
		goto fail_unlock;

	inode_init_owner(inode, dir, mode);
	inode_set_bytes(inode, 0);

	inode->i_mtime = current_time(inode);
	inode->i_atime = inode->i_mtime;
	inode->i_ctime = inode->i_mtime;
	BTRFS_I(inode)->i_otime = inode->i_mtime;

	inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
				  struct btrfs_inode_item);
	memzero_extent_buffer(path->nodes[0], (unsigned long)inode_item,
			     sizeof(*inode_item));
	fill_inode_item(trans, path->nodes[0], inode_item, inode);

	if (name) {
		ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
				     struct btrfs_inode_ref);
		btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len);
		btrfs_set_inode_ref_index(path->nodes[0], ref, *index);
		ptr = (unsigned long)(ref + 1);
		write_extent_buffer(path->nodes[0], name, ptr, name_len);
	}

	btrfs_mark_buffer_dirty(path->nodes[0]);
	btrfs_free_path(path);

	btrfs_inherit_iflags(inode, dir);

	if (S_ISREG(mode)) {
		if (btrfs_test_opt(fs_info, NODATASUM))
			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
		if (btrfs_test_opt(fs_info, NODATACOW))
			BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW |
				BTRFS_INODE_NODATASUM;
	}

	inode_tree_add(inode);

	trace_btrfs_inode_new(inode);
	btrfs_set_inode_last_trans(trans, inode);

	btrfs_update_root_times(trans, root);

	ret = btrfs_inode_inherit_props(trans, inode, dir);
	if (ret)
		btrfs_err(fs_info,
			  "error inheriting props for ino %llu (root %llu): %d",
			btrfs_ino(BTRFS_I(inode)), root->root_key.objectid, ret);

	return inode;

fail_unlock:
	discard_new_inode(inode);
fail:
	if (dir && name)
		BTRFS_I(dir)->index_cnt--;
	btrfs_free_path(path);
	return ERR_PTR(ret);
}

/*
 * utility function to add 'inode' into 'parent_inode' with
 * a give name and a given sequence number.
 * if 'add_backref' is true, also insert a backref from the
 * inode to the parent directory.
 */
int btrfs_add_link(struct btrfs_trans_handle *trans,
		   struct btrfs_inode *parent_inode, struct btrfs_inode *inode,
		   const char *name, int name_len, int add_backref, u64 index)
{
	int ret = 0;
	struct btrfs_key key;
	struct btrfs_root *root = parent_inode->root;
	u64 ino = btrfs_ino(inode);
	u64 parent_ino = btrfs_ino(parent_inode);

	if (unlikely(ino == BTRFS_FIRST_FREE_OBJECTID)) {
		memcpy(&key, &inode->root->root_key, sizeof(key));
	} else {
		key.objectid = ino;
		key.type = BTRFS_INODE_ITEM_KEY;
		key.offset = 0;
	}

	if (unlikely(ino == BTRFS_FIRST_FREE_OBJECTID)) {
		ret = btrfs_add_root_ref(trans, key.objectid,
					 root->root_key.objectid, parent_ino,
					 index, name, name_len);
	} else if (add_backref) {
		ret = btrfs_insert_inode_ref(trans, root, name, name_len, ino,
					     parent_ino, index);
	}

	/* Nothing to clean up yet */
	if (ret)
		return ret;

	ret = btrfs_insert_dir_item(trans, name, name_len, parent_inode, &key,
				    btrfs_inode_type(&inode->vfs_inode), index);
	if (ret == -EEXIST || ret == -EOVERFLOW)
		goto fail_dir_item;
	else if (ret) {
		btrfs_abort_transaction(trans, ret);
		return ret;
	}

	btrfs_i_size_write(parent_inode, parent_inode->vfs_inode.i_size +
			   name_len * 2);
	inode_inc_iversion(&parent_inode->vfs_inode);
	/*
	 * If we are replaying a log tree, we do not want to update the mtime
	 * and ctime of the parent directory with the current time, since the
	 * log replay procedure is responsible for setting them to their correct
	 * values (the ones it had when the fsync was done).
	 */
	if (!test_bit(BTRFS_FS_LOG_RECOVERING, &root->fs_info->flags)) {
		struct timespec64 now = current_time(&parent_inode->vfs_inode);

		parent_inode->vfs_inode.i_mtime = now;
		parent_inode->vfs_inode.i_ctime = now;
	}
	ret = btrfs_update_inode(trans, root, &parent_inode->vfs_inode);
	if (ret)
		btrfs_abort_transaction(trans, ret);
	return ret;

fail_dir_item:
	if (unlikely(ino == BTRFS_FIRST_FREE_OBJECTID)) {
		u64 local_index;
		int err;
		err = btrfs_del_root_ref(trans, key.objectid,
					 root->root_key.objectid, parent_ino,
					 &local_index, name, name_len);
		if (err)
			btrfs_abort_transaction(trans, err);
	} else if (add_backref) {
		u64 local_index;
		int err;

		err = btrfs_del_inode_ref(trans, root, name, name_len,
					  ino, parent_ino, &local_index);
		if (err)
			btrfs_abort_transaction(trans, err);
	}

	/* Return the original error code */
	return ret;
}

static int btrfs_add_nondir(struct btrfs_trans_handle *trans,
			    struct btrfs_inode *dir, struct dentry *dentry,
			    struct btrfs_inode *inode, int backref, u64 index)
{
	int err = btrfs_add_link(trans, dir, inode,
				 dentry->d_name.name, dentry->d_name.len,
				 backref, index);
	if (err > 0)
		err = -EEXIST;
	return err;
}

static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
			umode_t mode, dev_t rdev)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct inode *inode = NULL;
	int err;
	u64 objectid;
	u64 index = 0;

	/*
	 * 2 for inode item and ref
	 * 2 for dir items
	 * 1 for xattr if selinux is on
	 */
	trans = btrfs_start_transaction(root, 5);
	if (IS_ERR(trans))
		return PTR_ERR(trans);

	err = btrfs_find_free_ino(root, &objectid);
	if (err)
		goto out_unlock;

	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
			dentry->d_name.len, btrfs_ino(BTRFS_I(dir)), objectid,
			mode, &index);
	if (IS_ERR(inode)) {
		err = PTR_ERR(inode);
		inode = NULL;
		goto out_unlock;
	}

	/*
	* If the active LSM wants to access the inode during
	* d_instantiate it needs these. Smack checks to see
	* if the filesystem supports xattrs by looking at the
	* ops vector.
	*/
	inode->i_op = &btrfs_special_inode_operations;
	init_special_inode(inode, inode->i_mode, rdev);

	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
	if (err)
		goto out_unlock;

	err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry, BTRFS_I(inode),
			0, index);
	if (err)
		goto out_unlock;

	btrfs_update_inode(trans, root, inode);
	d_instantiate_new(dentry, inode);

out_unlock:
	btrfs_end_transaction(trans);
	btrfs_btree_balance_dirty(fs_info);
	if (err && inode) {
		inode_dec_link_count(inode);
		discard_new_inode(inode);
	}
	return err;
}

static int btrfs_create(struct inode *dir, struct dentry *dentry,
			umode_t mode, bool excl)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct inode *inode = NULL;
	int err;
	u64 objectid;
	u64 index = 0;

	/*
	 * 2 for inode item and ref
	 * 2 for dir items
	 * 1 for xattr if selinux is on
	 */
	trans = btrfs_start_transaction(root, 5);
	if (IS_ERR(trans))
		return PTR_ERR(trans);

	err = btrfs_find_free_ino(root, &objectid);
	if (err)
		goto out_unlock;

	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
			dentry->d_name.len, btrfs_ino(BTRFS_I(dir)), objectid,
			mode, &index);
	if (IS_ERR(inode)) {
		err = PTR_ERR(inode);
		inode = NULL;
		goto out_unlock;
	}
	/*
	* If the active LSM wants to access the inode during
	* d_instantiate it needs these. Smack checks to see
	* if the filesystem supports xattrs by looking at the
	* ops vector.
	*/
	inode->i_fop = &btrfs_file_operations;
	inode->i_op = &btrfs_file_inode_operations;
	inode->i_mapping->a_ops = &btrfs_aops;

	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
	if (err)
		goto out_unlock;

	err = btrfs_update_inode(trans, root, inode);
	if (err)
		goto out_unlock;

	err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry, BTRFS_I(inode),
			0, index);
	if (err)
		goto out_unlock;

	BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
	d_instantiate_new(dentry, inode);

out_unlock:
	btrfs_end_transaction(trans);
	if (err && inode) {
		inode_dec_link_count(inode);
		discard_new_inode(inode);
	}
	btrfs_btree_balance_dirty(fs_info);
	return err;
}

static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
		      struct dentry *dentry)
{
	struct btrfs_trans_handle *trans = NULL;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct inode *inode = d_inode(old_dentry);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	u64 index;
	int err;
	int drop_inode = 0;

	/* do not allow sys_link's with other subvols of the same device */
	if (root->root_key.objectid != BTRFS_I(inode)->root->root_key.objectid)
		return -EXDEV;

	if (inode->i_nlink >= BTRFS_LINK_MAX)
		return -EMLINK;

	err = btrfs_set_inode_index(BTRFS_I(dir), &index);
	if (err)
		goto fail;

	/*
	 * 2 items for inode and inode ref
	 * 2 items for dir items
	 * 1 item for parent inode
	 * 1 item for orphan item deletion if O_TMPFILE
	 */
	trans = btrfs_start_transaction(root, inode->i_nlink ? 5 : 6);
	if (IS_ERR(trans)) {
		err = PTR_ERR(trans);
		trans = NULL;
		goto fail;
	}

	/* There are several dir indexes for this inode, clear the cache. */
	BTRFS_I(inode)->dir_index = 0ULL;
	inc_nlink(inode);
	inode_inc_iversion(inode);
	inode->i_ctime = current_time(inode);
	ihold(inode);
	set_bit(BTRFS_INODE_COPY_EVERYTHING, &BTRFS_I(inode)->runtime_flags);

	err = btrfs_add_nondir(trans, BTRFS_I(dir), dentry, BTRFS_I(inode),
			1, index);

	if (err) {
		drop_inode = 1;
	} else {
		struct dentry *parent = dentry->d_parent;
		int ret;

		err = btrfs_update_inode(trans, root, inode);
		if (err)
			goto fail;
		if (inode->i_nlink == 1) {
			/*
			 * If new hard link count is 1, it's a file created
			 * with open(2) O_TMPFILE flag.
			 */
			err = btrfs_orphan_del(trans, BTRFS_I(inode));
			if (err)
				goto fail;
		}
		d_instantiate(dentry, inode);
		ret = btrfs_log_new_name(trans, BTRFS_I(inode), NULL, parent,
					 true, NULL);
		if (ret == BTRFS_NEED_TRANS_COMMIT) {
			err = btrfs_commit_transaction(trans);
			trans = NULL;
		}
	}

fail:
	if (trans)
		btrfs_end_transaction(trans);
	if (drop_inode) {
		inode_dec_link_count(inode);
		iput(inode);
	}
	btrfs_btree_balance_dirty(fs_info);
	return err;
}

static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
	struct inode *inode = NULL;
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	int err = 0;
	u64 objectid = 0;
	u64 index = 0;

	/*
	 * 2 items for inode and ref
	 * 2 items for dir items
	 * 1 for xattr if selinux is on
	 */
	trans = btrfs_start_transaction(root, 5);
	if (IS_ERR(trans))
		return PTR_ERR(trans);

	err = btrfs_find_free_ino(root, &objectid);
	if (err)
		goto out_fail;

	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
			dentry->d_name.len, btrfs_ino(BTRFS_I(dir)), objectid,
			S_IFDIR | mode, &index);
	if (IS_ERR(inode)) {
		err = PTR_ERR(inode);
		inode = NULL;
		goto out_fail;
	}

	/* these must be set before we unlock the inode */
	inode->i_op = &btrfs_dir_inode_operations;
	inode->i_fop = &btrfs_dir_file_operations;

	err = btrfs_init_inode_security(trans, inode, dir, &dentry->d_name);
	if (err)
		goto out_fail;

	btrfs_i_size_write(BTRFS_I(inode), 0);
	err = btrfs_update_inode(trans, root, inode);
	if (err)
		goto out_fail;

	err = btrfs_add_link(trans, BTRFS_I(dir), BTRFS_I(inode),
			dentry->d_name.name,
			dentry->d_name.len, 0, index);
	if (err)
		goto out_fail;

	d_instantiate_new(dentry, inode);

out_fail:
	btrfs_end_transaction(trans);
	if (err && inode) {
		inode_dec_link_count(inode);
		discard_new_inode(inode);
	}
	btrfs_btree_balance_dirty(fs_info);
	return err;
}

static noinline int uncompress_inline(struct btrfs_path *path,
				      struct page *page,
				      size_t pg_offset, u64 extent_offset,
				      struct btrfs_file_extent_item *item)
{
	int ret;
	struct extent_buffer *leaf = path->nodes[0];
	char *tmp;
	size_t max_size;
	unsigned long inline_size;
	unsigned long ptr;
	int compress_type;

	WARN_ON(pg_offset != 0);
	compress_type = btrfs_file_extent_compression(leaf, item);
	max_size = btrfs_file_extent_ram_bytes(leaf, item);
	inline_size = btrfs_file_extent_inline_item_len(leaf,
					btrfs_item_nr(path->slots[0]));
	tmp = kmalloc(inline_size, GFP_NOFS);
	if (!tmp)
		return -ENOMEM;
	ptr = btrfs_file_extent_inline_start(item);

	read_extent_buffer(leaf, tmp, ptr, inline_size);

	max_size = min_t(unsigned long, PAGE_SIZE, max_size);
	ret = btrfs_decompress(compress_type, tmp, page,
			       extent_offset, inline_size, max_size);

	/*
	 * decompression code contains a memset to fill in any space between the end
	 * of the uncompressed data and the end of max_size in case the decompressed
	 * data ends up shorter than ram_bytes.  That doesn't cover the hole between
	 * the end of an inline extent and the beginning of the next block, so we
	 * cover that region here.
	 */

	if (max_size + pg_offset < PAGE_SIZE) {
		char *map = kmap(page);
		memset(map + pg_offset + max_size, 0, PAGE_SIZE - max_size - pg_offset);
		kunmap(page);
	}
	kfree(tmp);
	return ret;
}

/**
 * btrfs_get_extent - Lookup the first extent overlapping a range in a file.
 * @inode:	file to search in
 * @page:	page to read extent data into if the extent is inline
 * @pg_offset:	offset into @page to copy to
 * @start:	file offset
 * @len:	length of range starting at @start
 *
 * This returns the first &struct extent_map which overlaps with the given
 * range, reading it from the B-tree and caching it if necessary. Note that
 * there may be more extents which overlap the given range after the returned
 * extent_map.
 *
 * If @page is not NULL and the extent is inline, this also reads the extent
 * data directly into the page and marks the extent up to date in the io_tree.
 *
 * Return: ERR_PTR on error, non-NULL extent_map on success.
 */
struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
				    struct page *page, size_t pg_offset,
				    u64 start, u64 len)
{
	struct btrfs_fs_info *fs_info = inode->root->fs_info;
	int ret;
	int err = 0;
	u64 extent_start = 0;
	u64 extent_end = 0;
	u64 objectid = btrfs_ino(inode);
	int extent_type = -1;
	struct btrfs_path *path = NULL;
	struct btrfs_root *root = inode->root;
	struct btrfs_file_extent_item *item;
	struct extent_buffer *leaf;
	struct btrfs_key found_key;
	struct extent_map *em = NULL;
	struct extent_map_tree *em_tree = &inode->extent_tree;
	struct extent_io_tree *io_tree = &inode->io_tree;

	read_lock(&em_tree->lock);
	em = lookup_extent_mapping(em_tree, start, len);
	read_unlock(&em_tree->lock);

	if (em) {
		if (em->start > start || em->start + em->len <= start)
			free_extent_map(em);
		else if (em->block_start == EXTENT_MAP_INLINE && page)
			free_extent_map(em);
		else
			goto out;
	}
	em = alloc_extent_map();
	if (!em) {
		err = -ENOMEM;
		goto out;
	}
	em->start = EXTENT_MAP_HOLE;
	em->orig_start = EXTENT_MAP_HOLE;
	em->len = (u64)-1;
	em->block_len = (u64)-1;

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

	/* Chances are we'll be called again, so go ahead and do readahead */
	path->reada = READA_FORWARD;

	/*
	 * Unless we're going to uncompress the inline extent, no sleep would
	 * happen.
	 */
	path->leave_spinning = 1;

	ret = btrfs_lookup_file_extent(NULL, root, path, objectid, start, 0);
	if (ret < 0) {
		err = ret;
		goto out;
	} else if (ret > 0) {
		if (path->slots[0] == 0)
			goto not_found;
		path->slots[0]--;
	}

	leaf = path->nodes[0];
	item = btrfs_item_ptr(leaf, path->slots[0],
			      struct btrfs_file_extent_item);
	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
	if (found_key.objectid != objectid ||
	    found_key.type != BTRFS_EXTENT_DATA_KEY) {
		/*
		 * If we backup past the first extent we want to move forward
		 * and see if there is an extent in front of us, otherwise we'll
		 * say there is a hole for our whole search range which can
		 * cause problems.
		 */
		extent_end = start;
		goto next;
	}

	extent_type = btrfs_file_extent_type(leaf, item);
	extent_start = found_key.offset;
	if (extent_type == BTRFS_FILE_EXTENT_REG ||
	    extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
		/* Only regular file could have regular/prealloc extent */
		if (!S_ISREG(inode->vfs_inode.i_mode)) {
			ret = -EUCLEAN;
			btrfs_crit(fs_info,
		"regular/prealloc extent found for non-regular inode %llu",
				   btrfs_ino(inode));
			goto out;
		}