fs/btrfs/inode.c

// 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 "backref.h"
#include "props.h"
#include "qgroup.h"
#include "delalloc-space.h"
#include "block-group.h"
#include "space-info.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_dir_ro_inode_operations;
static const struct inode_operations btrfs_special_inode_operations;
static const struct inode_operations btrfs_file_inode_operations;
static const struct address_space_operations btrfs_aops;
static const struct 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 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);

			/*
			 * Ensure we only free the compressed pages if we have
			 * them allocated, as we can still reach here with
			 * inode_need_compress() == false.
			 */
			if (pages) {
				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)) {
			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 min_alloc_size;
	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);

	/*
	 * Relocation relies on the relocated extents to have exactly the same
	 * size as the original extents. Normally writeback for relocation data
	 * extents follows a NOCOW path because relocation preallocates the
	 * extents. However, due to an operation such as scrub turning a block
	 * group to RO mode, it may fallback to COW mode, so we must make sure
	 * an extent allocated during COW has exactly the requested size and can
	 * not be split into smaller extents, otherwise relocation breaks and
	 * fails during the stage where it updates the bytenr of file extent
	 * items.
	 */
	if (root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID)
		min_alloc_size = num_bytes;
	else
		min_alloc_size = fs_info->sectorsize;

	while (num_bytes > 0) {
		cur_alloc_size = num_bytes;
		ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size,
					   min_alloc_size, 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 - 1,
					     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);
	/*
	 * 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 page *locked_page,
				u64 start, u64 end, int *page_started,
				unsigned long *nr_written,
				unsigned int write_flags)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	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;

	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) {
			async_chunk[i].locked_page = locked_page;
			locked_page = NULL;
		} else {
			async_chunk[i].locked_page = 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;
}

static int fallback_to_cow(struct inode *inode, struct page *locked_page,
			   const u64 start, const u64 end,
			   int *page_started, unsigned long *nr_written)
{
	const bool is_space_ino = btrfs_is_free_space_inode(BTRFS_I(inode));
	const bool is_reloc_ino = (BTRFS_I(inode)->root->root_key.objectid ==
				   BTRFS_DATA_RELOC_TREE_OBJECTID);
	const u64 range_bytes = end + 1 - start;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	u64 range_start = start;
	u64 count;

	/*
	 * If EXTENT_NORESERVE is set it means that when the buffered write was
	 * made we had not enough available data space and therefore we did not
	 * reserve data space for it, since we though we could do NOCOW for the
	 * respective file range (either there is prealloc extent or the inode
	 * has the NOCOW bit set).
	 *
	 * However when we need to fallback to COW mode (because for example the
	 * block group for the corresponding extent was turned to RO mode by a
	 * scrub or relocation) we need to do the following:
	 *
	 * 1) We increment the bytes_may_use counter of the data space info.
	 *    If COW succeeds, it allocates a new data extent and after doing
	 *    that it decrements the space info's bytes_may_use counter and
	 *    increments its bytes_reserved counter by the same amount (we do
	 *    this at btrfs_add_reserved_bytes()). So we need to increment the
	 *    bytes_may_use counter to compensate (when space is reserved at
	 *    buffered write time, the bytes_may_use counter is incremented);
	 *
	 * 2) We clear the EXTENT_NORESERVE bit from the range. We do this so
	 *    that if the COW path fails for any reason, it decrements (through
	 *    extent_clear_unlock_delalloc()) the bytes_may_use counter of the
	 *    data space info, which we incremented in the step above.
	 *
	 * If we need to fallback to cow and the inode corresponds to a free
	 * space cache inode or an inode of the data relocation tree, we must
	 * also increment bytes_may_use of the data space_info for the same
	 * reason. Space caches and relocated data extents always get a prealloc
	 * extent for them, however scrub or balance may have set the block
	 * group that contains that extent to RO mode and therefore force COW
	 * when starting writeback.
	 */
	count = count_range_bits(io_tree, &range_start, end, range_bytes,
				 EXTENT_NORESERVE, 0);
	if (count > 0 || is_space_ino || is_reloc_ino) {
		u64 bytes = count;
		struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
		struct btrfs_space_info *sinfo = fs_info->data_sinfo;

		if (is_space_ino || is_reloc_ino)
			bytes = range_bytes;

		spin_lock(&sinfo->lock);
		btrfs_space_info_update_bytes_may_use(fs_info, sinfo, bytes);
		spin_unlock(&sinfo->lock);

		if (count > 0)
			clear_extent_bit(io_tree, start, end, EXTENT_NORESERVE,
					 0, 0, NULL);
	}

	return cow_file_range(inode, locked_page, start, end, page_started,
			      nr_written, 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, false);
			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 = fallback_to_cow(inode, locked_page, cow_start,
					      found_key.offset - 1,
					      page_started, nr_written);
			if (ret)
				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)) {
				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 = fallback_to_cow(inode, locked_page, cow_start, end,
				      page_started, nr_written);
		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);
	unsigned int write_flags = wbc_to_write_flags(wbc);

	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, locked_page, start, end,
					   page_started, nr_written,
					   write_flags);
	}
	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, 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, 0);

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;
}

/* snapshot-aware defrag */
struct sa_defrag_extent_backref {
	struct rb_node node;
	struct old_sa_defrag_extent *old;
	u64 root_id;
	u64 inum;
	u64 file_pos;
	u64 extent_offset;
	u64 num_bytes;
	u64 generation;
};

struct old_sa_defrag_extent {
	struct list_head list;
	struct new_sa_defrag_extent *new;

	u64 extent_offset;
	u64 bytenr;
	u64 offset;
	u64 len;
	int count;
};

struct new_sa_defrag_extent {
	struct rb_root root;
	struct list_head head;
	struct btrfs_path *path;
	struct inode *inode;
	u64 file_pos;
	u64 len;
	u64 bytenr;
	u64 disk_len;
	u8 compress_type;
};

static int backref_comp(struct sa_defrag_extent_backref *b1,
			struct sa_defrag_extent_backref *b2)
{
	if (b1->root_id < b2->root_id)
		return -1;
	else if (b1->root_id > b2->root_id)
		return 1;

	if (b1->inum < b2->inum)
		return -1;
	else if (b1->inum > b2->inum)
		return 1;

	if (b1->file_pos < b2->file_pos)
		return -1;
	else if (b1->file_pos > b2->file_pos)
		return 1;

	/*
	 * [------------------------------] ===> (a range of space)
	 *     |<--->|   |<---->| =============> (fs/file tree A)
	 * |<---------------------------->| ===> (fs/file tree B)
	 *
	 * A range of space can refer to two file extents in one tree while
	 * refer to only one file extent in another tree.
	 *
	 * So we may process a disk offset more than one time(two extents in A)
	 * and locate at the same extent(one extent in B), then insert two same
	 * backrefs(both refer to the extent in B).
	 */
	return 0;
}

static void backref_insert(struct rb_root *root,
			   struct sa_defrag_extent_backref *backref)
{
	struct rb_node **p = &root->rb_node;
	struct rb_node *parent = NULL;
	struct sa_defrag_extent_backref *entry;
	int ret;

	while (*p) {
		parent = *p;
		entry = rb_entry(parent, struct sa_defrag_extent_backref, node);

		ret = backref_comp(backref, entry);
		if (ret < 0)
			p = &(*p)->rb_left;
		else
			p = &(*p)->rb_right;
	}

	rb_link_node(&backref->node, parent, p);
	rb_insert_color(&backref->node, root);
}

/*
 * Note the backref might has changed, and in this case we just return 0.
 */
static noinline int record_one_backref(u64 inum, u64 offset, u64 root_id,
				       void *ctx)
{
	struct btrfs_file_extent_item *extent;
	struct old_sa_defrag_extent *old = ctx;
	struct new_sa_defrag_extent *new = old->new;
	struct btrfs_path *path = new->path;
	struct btrfs_key key;
	struct btrfs_root *root;
	struct sa_defrag_extent_backref *backref;
	struct extent_buffer *leaf;
	struct inode *inode = new->inode;
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	int slot;
	int ret;
	u64 extent_offset;
	u64 num_bytes;

	if (BTRFS_I(inode)->root->root_key.objectid == root_id &&
	    inum == btrfs_ino(BTRFS_I(inode)))
		return 0;

	key.objectid = root_id;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = (u64)-1;

	root = btrfs_read_fs_root_no_name(fs_info, &key);
	if (IS_ERR(root)) {
		if (PTR_ERR(root) == -ENOENT)
			return 0;
		WARN_ON(1);
		btrfs_debug(fs_info, "inum=%llu, offset=%llu, root_id=%llu",
			 inum, offset, root_id);
		return PTR_ERR(root);
	}

	key.objectid = inum;
	key.type = BTRFS_EXTENT_DATA_KEY;
	if (offset > (u64)-1 << 32)
		key.offset = 0;
	else
		key.offset = offset;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (WARN_ON(ret < 0))
		return ret;
	ret = 0;

	while (1) {
		cond_resched();

		leaf = path->nodes[0];
		slot = path->slots[0];

		if (slot >= btrfs_header_nritems(leaf)) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0) {
				goto out;
			} else if (ret > 0) {
				ret = 0;
				goto out;
			}
			continue;
		}

		path->slots[0]++;

		btrfs_item_key_to_cpu(leaf, &key, slot);

		if (key.objectid > inum)
			goto out;

		if (key.objectid < inum || key.type != BTRFS_EXTENT_DATA_KEY)
			continue;

		extent = btrfs_item_ptr(leaf, slot,
					struct btrfs_file_extent_item);

		if (btrfs_file_extent_disk_bytenr(leaf, extent) != old->bytenr)
			continue;

		/*
		 * 'offset' refers to the exact key.offset,
		 * NOT the 'offset' field in btrfs_extent_data_ref, ie.
		 * (key.offset - extent_offset).
		 */
		if (key.offset != offset)
			continue;

		extent_offset = btrfs_file_extent_offset(leaf, extent);
		num_bytes = btrfs_file_extent_num_bytes(leaf, extent);

		if (extent_offset >= old->extent_offset + old->offset +
		    old->len || extent_offset + num_bytes <=
		    old->extent_offset + old->offset)
			continue;
		break;
	}

	backref = kmalloc(sizeof(*backref), GFP_NOFS);
	if (!backref) {
		ret = -ENOENT;
		goto out;
	}

	backref->root_id = root_id;
	backref->inum = inum;
	backref->file_pos = offset;
	backref->num_bytes = num_bytes;
	backref->extent_offset = extent_offset;
	backref->generation = btrfs_file_extent_generation(leaf, extent);
	backref->old = old;
	backref_insert(&new->root, backref);
	old->count++;
out:
	btrfs_release_path(path);
	WARN_ON(ret);
	return ret;
}

static noinline bool record_extent_backrefs(struct btrfs_path *path,
				   struct new_sa_defrag_extent *new)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(new->inode->i_sb);
	struct old_sa_defrag_extent *old, *tmp;
	int ret;

	new->path = path;

	list_for_each_entry_safe(old, tmp, &new->head, list) {
		ret = iterate_inodes_from_logical(old->bytenr +
						  old->extent_offset, fs_info,
						  path, record_one_backref,
						  old, false);
		if (ret < 0 && ret != -ENOENT)
			return false;

		/* no backref to be processed for this extent */
		if (!old->count) {
			list_del(&old->list);
			kfree(old);
		}
	}

	if (list_empty(&new->head))
		return false;

	return true;
}

static int relink_is_mergable(struct extent_buffer *leaf,
			      struct btrfs_file_extent_item *fi,
			      struct new_sa_defrag_extent *new)
{
	if (btrfs_file_extent_disk_bytenr(leaf, fi) != new->bytenr)
		return 0;

	if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG)
		return 0;

	if (btrfs_file_extent_compression(leaf, fi) != new->compress_type)
		return 0;

	if (btrfs_file_extent_encryption(leaf, fi) ||
	    btrfs_file_extent_other_encoding(leaf, fi))
		return 0;

	return 1;
}

/*
 * Note the backref might has changed, and in this case we just return 0.
 */
static noinline int relink_extent_backref(struct btrfs_path *path,
				 struct sa_defrag_extent_backref *prev,
				 struct sa_defrag_extent_backref *backref)
{
	struct btrfs_file_extent_item *extent;
	struct btrfs_file_extent_item *item;
	struct btrfs_ordered_extent *ordered;
	struct btrfs_trans_handle *trans;
	struct btrfs_ref ref = { 0 };
	struct btrfs_root *root;
	struct btrfs_key key;
	struct extent_buffer *leaf;
	struct old_sa_defrag_extent *old = backref->old;
	struct new_sa_defrag_extent *new = old->new;
	struct btrfs_fs_info *fs_info = btrfs_sb(new->inode->i_sb);
	struct inode *inode;
	struct extent_state *cached = NULL;
	int ret = 0;
	u64 start;
	u64 len;
	u64 lock_start;
	u64 lock_end;
	bool merge = false;
	int index;

	if (prev && prev->root_id == backref->root_id &&
	    prev->inum == backref->inum &&
	    prev->file_pos + prev->num_bytes == backref->file_pos)
		merge = true;

	/* step 1: get root */
	key.objectid = backref->root_id;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = (u64)-1;

	index = srcu_read_lock(&fs_info->subvol_srcu);

	root = btrfs_read_fs_root_no_name(fs_info, &key);
	if (IS_ERR(root)) {
		srcu_read_unlock(&fs_info->subvol_srcu, index);
		if (PTR_ERR(root) == -ENOENT)
			return 0;
		return PTR_ERR(root);
	}

	if (btrfs_root_readonly(root)) {
		srcu_read_unlock(&fs_info->subvol_srcu, index);
		return 0;
	}

	/* step 2: get inode */
	key.objectid = backref->inum;
	key.type = BTRFS_INODE_ITEM_KEY;
	key.offset = 0;

	inode = btrfs_iget(fs_info->sb, &key, root, NULL);
	if (IS_ERR(inode)) {
		srcu_read_unlock(&fs_info->subvol_srcu, index);
		return 0;
	}

	srcu_read_unlock(&fs_info->subvol_srcu, index);

	/* step 3: relink backref */
	lock_start = backref->file_pos;
	lock_end = backref->file_pos + backref->num_bytes - 1;
	lock_extent_bits(&BTRFS_I(inode)->io_tree, lock_start, lock_end,
			 &cached);

	ordered = btrfs_lookup_first_ordered_extent(inode, lock_end);
	if (ordered) {
		btrfs_put_ordered_extent(ordered);
		goto out_unlock;
	}

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_unlock;
	}

	key.objectid = backref->inum;
	key.type = BTRFS_EXTENT_DATA_KEY;
	key.offset = backref->file_pos;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0) {
		goto out_free_path;
	} else if (ret > 0) {
		ret = 0;
		goto out_free_path;
	}

	extent = btrfs_item_ptr(path->nodes[0], path->slots[0],
				struct btrfs_file_extent_item);

	if (btrfs_file_extent_generation(path->nodes[0], extent) !=
	    backref->generation)
		goto out_free_path;

	btrfs_release_path(path);

	start = backref->file_pos;
	if (backref->extent_offset < old->extent_offset + old->offset)
		start += old->extent_offset + old->offset -
			 backref->extent_offset;

	len = min(backref->extent_offset + backref->num_bytes,
		  old->extent_offset + old->offset + old->len);
	len -= max(backref->extent_offset, old->extent_offset + old->offset);

	ret = btrfs_drop_extents(trans, root, inode, start,
				 start + len, 1);
	if (ret)
		goto out_free_path;
again:
	key.objectid = btrfs_ino(BTRFS_I(inode));
	key.type = BTRFS_EXTENT_DATA_KEY;
	key.offset = start;

	path->leave_spinning = 1;
	if (merge) {
		struct btrfs_file_extent_item *fi;
		u64 extent_len;
		struct btrfs_key found_key;

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

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

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

		if (extent_len + found_key.offset == start &&
		    relink_is_mergable(leaf, fi, new)) {
			btrfs_set_file_extent_num_bytes(leaf, fi,
							extent_len + len);
			btrfs_mark_buffer_dirty(leaf);
			inode_add_bytes(inode, len);

			ret = 1;
			goto out_free_path;
		} else {
			merge = false;
			btrfs_release_path(path);
			goto again;
		}
	}

	ret = btrfs_insert_empty_item(trans, root, path, &key,
					sizeof(*extent));
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out_free_path;
	}

	leaf = path->nodes[0];
	item = btrfs_item_ptr(leaf, path->slots[0],
				struct btrfs_file_extent_item);
	btrfs_set_file_extent_disk_bytenr(leaf, item, new->bytenr);
	btrfs_set_file_extent_disk_num_bytes(leaf, item, new->disk_len);
	btrfs_set_file_extent_offset(leaf, item, start - new->file_pos);
	btrfs_set_file_extent_num_bytes(leaf, item, len);
	btrfs_set_file_extent_ram_bytes(leaf, item, new->len);
	btrfs_set_file_extent_generation(leaf, item, trans->transid);
	btrfs_set_file_extent_type(leaf, item, BTRFS_FILE_EXTENT_REG);
	btrfs_set_file_extent_compression(leaf, item, new->compress_type);
	btrfs_set_file_extent_encryption(leaf, item, 0);
	btrfs_set_file_extent_other_encoding(leaf, item, 0);

	btrfs_mark_buffer_dirty(leaf);
	inode_add_bytes(inode, len);
	btrfs_release_path(path);

	btrfs_init_generic_ref(&ref, BTRFS_ADD_DELAYED_REF, new->bytenr,
			       new->disk_len, 0);
	btrfs_init_data_ref(&ref, backref->root_id, backref->inum,
			    new->file_pos);  /* start - extent_offset */
	ret = btrfs_inc_extent_ref(trans, &ref);
	if (ret) {
		btrfs_abort_transaction(trans, ret);
		goto out_free_path;
	}

	ret = 1;
out_free_path:
	btrfs_release_path(path);
	path->leave_spinning = 0;
	btrfs_end_transaction(trans);
out_unlock:
	unlock_extent_cached(&BTRFS_I(inode)->io_tree, lock_start, lock_end,
			     &cached);
	iput(inode);
	return ret;
}

static void free_sa_defrag_extent(struct new_sa_defrag_extent *new)
{
	struct old_sa_defrag_extent *old, *tmp;

	if (!new)
		return;

	list_for_each_entry_safe(old, tmp, &new->head, list) {
		kfree(old);
	}
	kfree(new);
}

static void relink_file_extents(struct new_sa_defrag_extent *new)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(new->inode->i_sb);
	struct btrfs_path *path;
	struct sa_defrag_extent_backref *backref;
	struct sa_defrag_extent_backref *prev = NULL;
	struct rb_node *node;
	int ret;

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

	if (!record_extent_backrefs(path, new)) {
		btrfs_free_path(path);
		goto out;
	}
	btrfs_release_path(path);

	while (1) {
		node = rb_first(&new->root);
		if (!node)
			break;
		rb_erase(node, &new->root);

		backref = rb_entry(node, struct sa_defrag_extent_backref, node);

		ret = relink_extent_backref(path, prev, backref);
		WARN_ON(ret < 0);

		kfree(prev);

		if (ret == 1)
			prev = backref;
		else
			prev = NULL;
		cond_resched();
	}
	kfree(prev);

	btrfs_free_path(path);
out:
	free_sa_defrag_extent(new);

	atomic_dec(&fs_info->defrag_running);
	wake_up(&fs_info->transaction_wait);
}

static struct new_sa_defrag_extent *
record_old_file_extents(struct inode *inode,
			struct btrfs_ordered_extent *ordered)
{
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_path *path;
	struct btrfs_key key;
	struct old_sa_defrag_extent *old;
	struct new_sa_defrag_extent *new;
	int ret;

	new = kmalloc(sizeof(*new), GFP_NOFS);
	if (!new)
		return NULL;

	new->inode = inode;
	new->file_pos = ordered->file_offset;
	new->len = ordered->len;
	new->bytenr = ordered->start;
	new->disk_len = ordered->disk_len;
	new->compress_type = ordered->compress_type;
	new->root = RB_ROOT;
	INIT_LIST_HEAD(&new->head);

	path = btrfs_alloc_path();
	if (!path)
		goto out_kfree;

	key.objectid = btrfs_ino(BTRFS_I(inode));
	key.type = BTRFS_EXTENT_DATA_KEY;
	key.offset = new->file_pos;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto out_free_path;
	if (ret > 0 && path->slots[0] > 0)
		path->slots[0]--;

	/* find out all the old extents for the file range */
	while (1) {
		struct btrfs_file_extent_item *extent;
		struct extent_buffer *l;
		int slot;
		u64 num_bytes;
		u64 offset;
		u64 end;
		u64 disk_bytenr;
		u64 extent_offset;

		l = path->nodes[0];
		slot = path->slots[0];

		if (slot >= btrfs_header_nritems(l)) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0)
				goto out_free_path;
			else if (ret > 0)
				break;
			continue;
		}

		btrfs_item_key_to_cpu(l, &key, slot);

		if (key.objectid != btrfs_ino(BTRFS_I(inode)))
			break;
		if (key.type != BTRFS_EXTENT_DATA_KEY)
			break;
		if (key.offset >= new->file_pos + new->len)
			break;

		extent = btrfs_item_ptr(l, slot, struct btrfs_file_extent_item);

		num_bytes = btrfs_file_extent_num_bytes(l, extent);
		if (key.offset +