[PATCH v4 00/11] ocfs2: wire up {clone,copy,dedupe}_range

[PATCH v4 00/11] ocfs2: wire up {clone,copy,dedupe}_range

[Darrick J. Wong]

Hi all,

This is the fourth submission of a series of patches that wire up the
existing ocfs2 reflinking capabilities to the new(ish) VFS
{copy,clone,dedupe}_range interface.  The first few patches hoist some
of the reflink/dedupe prep functions into the VFS since the code is
common to XFS and ocfs2.  Then there are fixes for some ocfs2 bugs that
I found; and finally the last kernel patch contains the new code.

A few minor fixes to xfstests are needed to make more of the tests run;
those have been sent separately to the fstests lists.  Some of the
xfstests will fail because ocfs2's design does not allow sharing of
blocks between disjoint sets of files with shared blocks.  That is to
say, if A shares with B and C shares with D, B and C cannot be made to
share blocks.

Andrew, could you please queue these up for testing?  I'd like to get
them into 4.10 if possible.

--D

[1] https://github.com/djwong/linux/tree/ocfs2-vfs-reflink-6

[PATCH 01/11] fs: try to clone files first in vfs_copy_file_range

[Christoph Hellwig]

A clone is a perfectly fine implementation of a file copy, so most
file systems just implement the copy that way.  Instead of duplicating
this logic move it to the VFS.  Currently btrfs and XFS implement copies
the same way as clones and there is no behavior change for them, cifs
only implements clones and grow support for copy_file_range with this
patch.  NFS implements both, so this will allow copy_file_range to work
on servers that only implement CLONE and be lot more efficient on servers
that implements CLONE and COPY.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/btrfs/ctree.h  |    3 ---
 fs/btrfs/file.c   |    1 -
 fs/btrfs/ioctl.c  |   12 ------------
 fs/read_write.c   |   27 ++++++++++++++++++++++-----
 fs/xfs/xfs_file.c |   19 -------------------
 5 files changed, 22 insertions(+), 40 deletions(-)


diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 0b8ce2b..05f75a9 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -3232,9 +3232,6 @@ int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode,
 		      loff_t pos, size_t write_bytes,
 		      struct extent_state **cached);
 int btrfs_fdatawrite_range(struct inode *inode, loff_t start, loff_t end);
-ssize_t btrfs_copy_file_range(struct file *file_in, loff_t pos_in,
-			      struct file *file_out, loff_t pos_out,
-			      size_t len, unsigned int flags);
 int btrfs_clone_file_range(struct file *file_in, loff_t pos_in,
 			   struct file *file_out, loff_t pos_out, u64 len);
 
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 3a14c87..991cc99 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -2998,7 +2998,6 @@ const struct file_operations btrfs_file_operations = {
 #ifdef CONFIG_COMPAT
 	.compat_ioctl	= btrfs_compat_ioctl,
 #endif
-	.copy_file_range = btrfs_copy_file_range,
 	.clone_file_range = btrfs_clone_file_range,
 	.dedupe_file_range = btrfs_dedupe_file_range,
 };
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 7acbd2c..dab7462 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -3980,18 +3980,6 @@ static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
 	return ret;
 }
 
-ssize_t btrfs_copy_file_range(struct file *file_in, loff_t pos_in,
-			      struct file *file_out, loff_t pos_out,
-			      size_t len, unsigned int flags)
-{
-	ssize_t ret;
-
-	ret = btrfs_clone_files(file_out, file_in, pos_in, len, pos_out);
-	if (ret == 0)
-		ret = len;
-	return ret;
-}
-
 int btrfs_clone_file_range(struct file *src_file, loff_t off,
 		struct file *dst_file, loff_t destoff, u64 len)
 {
diff --git a/fs/read_write.c b/fs/read_write.c
index 190e0d36..6674a4b 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -1542,20 +1542,37 @@ ssize_t vfs_copy_file_range(struct file *file_in, loff_t pos_in,
 	if (ret)
 		return ret;
 
-	ret = -EOPNOTSUPP;
-	if (file_out->f_op->copy_file_range)
+	/*
+	 * Try cloning first, this is supported by more file systems, and
+	 * more efficient if both clone and copy are supported (e.g. NFS).
+	 */
+	if (file_in->f_op->clone_file_range) {
+		ret = file_in->f_op->clone_file_range(file_in, pos_in,
+				file_out, pos_out, len);
+		if (ret == 0) {
+			ret = len;
+			goto done;
+		}
+	}
+
+	if (file_out->f_op->copy_file_range) {
 		ret = file_out->f_op->copy_file_range(file_in, pos_in, file_out,
 						      pos_out, len, flags);
-	if (ret == -EOPNOTSUPP)
-		ret = do_splice_direct(file_in, &pos_in, file_out, &pos_out,
-				len > MAX_RW_COUNT ? MAX_RW_COUNT : len, 0);
+		if (ret != -EOPNOTSUPP)
+			goto done;
+	}
 
+	ret = do_splice_direct(file_in, &pos_in, file_out, &pos_out,
+			len > MAX_RW_COUNT ? MAX_RW_COUNT : len, 0);
+
+done:
 	if (ret > 0) {
 		fsnotify_access(file_in);
 		add_rchar(current, ret);
 		fsnotify_modify(file_out);
 		add_wchar(current, ret);
 	}
+
 	inc_syscr(current);
 	inc_syscw(current);
 
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 6e4f7f9..86ecc9b 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -909,24 +909,6 @@ xfs_file_fallocate(
 	return error;
 }
 
-STATIC ssize_t
-xfs_file_copy_range(
-	struct file	*file_in,
-	loff_t		pos_in,
-	struct file	*file_out,
-	loff_t		pos_out,
-	size_t		len,
-	unsigned int	flags)
-{
-	int		error;
-
-	error = xfs_reflink_remap_range(file_in, pos_in, file_out, pos_out,
-				     len, false);
-	if (error)
-		return error;
-	return len;
-}
-
 STATIC int
 xfs_file_clone_range(
 	struct file	*file_in,
@@ -1625,7 +1607,6 @@ const struct file_operations xfs_file_operations = {
 	.fsync		= xfs_file_fsync,
 	.get_unmapped_area = thp_get_unmapped_area,
 	.fallocate	= xfs_file_fallocate,
-	.copy_file_range = xfs_file_copy_range,
 	.clone_file_range = xfs_file_clone_range,
 	.dedupe_file_range = xfs_file_dedupe_range,
 };

[PATCH 02/11] vfs: refactor clone/dedupe_file_range common functions

[Darrick J. Wong]

Hoist both the XFS reflink inode state and preparation code and the XFS
file blocks compare functions into the VFS so that ocfs2 can take
advantage of it for reflink and dedupe.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/read_write.c      |  204 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_reflink.c |  213 ++------------------------------------------------
 include/linux/fs.h   |    6 +
 3 files changed, 219 insertions(+), 204 deletions(-)


diff --git a/fs/read_write.c b/fs/read_write.c
index 6674a4b..dbf3f7f 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -1667,6 +1667,114 @@ static int clone_verify_area(struct file *file, loff_t pos, u64 len, bool write)
 	return security_file_permission(file, write ? MAY_WRITE : MAY_READ);
 }
 
+/*
+ * Check that the two inodes are eligible for cloning, the ranges make
+ * sense, and then flush all dirty data.  Caller must ensure that the
+ * inodes have been locked against any other modifications.
+ */
+int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
+			       struct inode *inode_out, loff_t pos_out,
+			       u64 *len, bool is_dedupe)
+{
+	loff_t bs = inode_out->i_sb->s_blocksize;
+	loff_t blen;
+	loff_t isize;
+	bool same_inode = (inode_in == inode_out);
+	int ret;
+
+	/* Don't touch certain kinds of inodes */
+	if (IS_IMMUTABLE(inode_out))
+		return -EPERM;
+
+	if (IS_SWAPFILE(inode_in) || IS_SWAPFILE(inode_out))
+		return -ETXTBSY;
+
+	/* Don't reflink dirs, pipes, sockets... */
+	if (S_ISDIR(inode_in->i_mode) || S_ISDIR(inode_out->i_mode))
+		return -EISDIR;
+	if (!S_ISREG(inode_in->i_mode) || !S_ISREG(inode_out->i_mode))
+		return -EINVAL;
+
+	/* Are we going all the way to the end? */
+	isize = i_size_read(inode_in);
+	if (isize == 0) {
+		*len = 0;
+		return 0;
+	}
+
+	/* Zero length dedupe exits immediately; reflink goes to EOF. */
+	if (*len == 0) {
+		if (is_dedupe) {
+			*len = 0;
+			return 0;
+		}
+		*len = isize - pos_in;
+	}
+
+	/* Ensure offsets don't wrap and the input is inside i_size */
+	if (pos_in + *len < pos_in || pos_out + *len < pos_out ||
+	    pos_in + *len > isize)
+		return -EINVAL;
+
+	/* Don't allow dedupe past EOF in the dest file */
+	if (is_dedupe) {
+		loff_t	disize;
+
+		disize = i_size_read(inode_out);
+		if (pos_out >= disize || pos_out + *len > disize)
+			return -EINVAL;
+	}
+
+	/* If we're linking to EOF, continue to the block boundary. */
+	if (pos_in + *len == isize)
+		blen = ALIGN(isize, bs) - pos_in;
+	else
+		blen = *len;
+
+	/* Only reflink if we're aligned to block boundaries */
+	if (!IS_ALIGNED(pos_in, bs) || !IS_ALIGNED(pos_in + blen, bs) ||
+	    !IS_ALIGNED(pos_out, bs) || !IS_ALIGNED(pos_out + blen, bs))
+		return -EINVAL;
+
+	/* Don't allow overlapped reflink within the same file */
+	if (same_inode) {
+		if (pos_out + blen > pos_in && pos_out < pos_in + blen)
+			return -EINVAL;
+	}
+
+	/* Wait for the completion of any pending IOs on both files */
+	inode_dio_wait(inode_in);
+	if (!same_inode)
+		inode_dio_wait(inode_out);
+
+	ret = filemap_write_and_wait_range(inode_in->i_mapping,
+			pos_in, pos_in + *len - 1);
+	if (ret)
+		return ret;
+
+	ret = filemap_write_and_wait_range(inode_out->i_mapping,
+			pos_out, pos_out + *len - 1);
+	if (ret)
+		return ret;
+
+	/*
+	 * Check that the extents are the same.
+	 */
+	if (is_dedupe) {
+		bool		is_same = false;
+
+		ret = vfs_dedupe_file_range_compare(inode_in, pos_in,
+				inode_out, pos_out, *len, &is_same);
+		if (ret)
+			return ret;
+		if (!is_same)
+			return -EBADE;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(vfs_clone_file_prep_inodes);
+
 int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
 		struct file *file_out, loff_t pos_out, u64 len)
 {
@@ -1718,6 +1826,102 @@ int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
 }
 EXPORT_SYMBOL(vfs_clone_file_range);
 
+/*
+ * Read a page's worth of file data into the page cache.  Return the page
+ * locked.
+ */
+static struct page *vfs_dedupe_get_page(struct inode *inode, loff_t offset)
+{
+	struct address_space *mapping;
+	struct page *page;
+	pgoff_t n;
+
+	n = offset >> PAGE_SHIFT;
+	mapping = inode->i_mapping;
+	page = read_mapping_page(mapping, n, NULL);
+	if (IS_ERR(page))
+		return page;
+	if (!PageUptodate(page)) {
+		put_page(page);
+		return ERR_PTR(-EIO);
+	}
+	lock_page(page);
+	return page;
+}
+
+/*
+ * Compare extents of two files to see if they are the same.
+ * Caller must have locked both inodes to prevent write races.
+ */
+int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
+				  struct inode *dest, loff_t destoff,
+				  loff_t len, bool *is_same)
+{
+	loff_t src_poff;
+	loff_t dest_poff;
+	void *src_addr;
+	void *dest_addr;
+	struct page *src_page;
+	struct page *dest_page;
+	loff_t cmp_len;
+	bool same;
+	int error;
+
+	error = -EINVAL;
+	same = true;
+	while (len) {
+		src_poff = srcoff & (PAGE_SIZE - 1);
+		dest_poff = destoff & (PAGE_SIZE - 1);
+		cmp_len = min(PAGE_SIZE - src_poff,
+			      PAGE_SIZE - dest_poff);
+		cmp_len = min(cmp_len, len);
+		if (cmp_len <= 0)
+			goto out_error;
+
+		src_page = vfs_dedupe_get_page(src, srcoff);
+		if (IS_ERR(src_page)) {
+			error = PTR_ERR(src_page);
+			goto out_error;
+		}
+		dest_page = vfs_dedupe_get_page(dest, destoff);
+		if (IS_ERR(dest_page)) {
+			error = PTR_ERR(dest_page);
+			unlock_page(src_page);
+			put_page(src_page);
+			goto out_error;
+		}
+		src_addr = kmap_atomic(src_page);
+		dest_addr = kmap_atomic(dest_page);
+
+		flush_dcache_page(src_page);
+		flush_dcache_page(dest_page);
+
+		if (memcmp(src_addr + src_poff, dest_addr + dest_poff, cmp_len))
+			same = false;
+
+		kunmap_atomic(dest_addr);
+		kunmap_atomic(src_addr);
+		unlock_page(dest_page);
+		unlock_page(src_page);
+		put_page(dest_page);
+		put_page(src_page);
+
+		if (!same)
+			break;
+
+		srcoff += cmp_len;
+		destoff += cmp_len;
+		len -= cmp_len;
+	}
+
+	*is_same = same;
+	return 0;
+
+out_error:
+	return error;
+}
+EXPORT_SYMBOL(vfs_dedupe_file_range_compare);
+
 int vfs_dedupe_file_range(struct file *file, struct file_dedupe_range *same)
 {
 	struct file_dedupe_range_info *info;
diff --git a/fs/xfs/xfs_reflink.c b/fs/xfs/xfs_reflink.c
index a279b4e..95d6828 100644
--- a/fs/xfs/xfs_reflink.c
+++ b/fs/xfs/xfs_reflink.c
@@ -1165,111 +1165,6 @@ xfs_reflink_remap_blocks(
 }
 
 /*
- * Read a page's worth of file data into the page cache.  Return the page
- * locked.
- */
-static struct page *
-xfs_get_page(
-	struct inode	*inode,
-	xfs_off_t	offset)
-{
-	struct address_space	*mapping;
-	struct page		*page;
-	pgoff_t			n;
-
-	n = offset >> PAGE_SHIFT;
-	mapping = inode->i_mapping;
-	page = read_mapping_page(mapping, n, NULL);
-	if (IS_ERR(page))
-		return page;
-	if (!PageUptodate(page)) {
-		put_page(page);
-		return ERR_PTR(-EIO);
-	}
-	lock_page(page);
-	return page;
-}
-
-/*
- * Compare extents of two files to see if they are the same.
- */
-static int
-xfs_compare_extents(
-	struct inode	*src,
-	xfs_off_t	srcoff,
-	struct inode	*dest,
-	xfs_off_t	destoff,
-	xfs_off_t	len,
-	bool		*is_same)
-{
-	xfs_off_t	src_poff;
-	xfs_off_t	dest_poff;
-	void		*src_addr;
-	void		*dest_addr;
-	struct page	*src_page;
-	struct page	*dest_page;
-	xfs_off_t	cmp_len;
-	bool		same;
-	int		error;
-
-	error = -EINVAL;
-	same = true;
-	while (len) {
-		src_poff = srcoff & (PAGE_SIZE - 1);
-		dest_poff = destoff & (PAGE_SIZE - 1);
-		cmp_len = min(PAGE_SIZE - src_poff,
-			      PAGE_SIZE - dest_poff);
-		cmp_len = min(cmp_len, len);
-		ASSERT(cmp_len > 0);
-
-		trace_xfs_reflink_compare_extents(XFS_I(src), srcoff, cmp_len,
-				XFS_I(dest), destoff);
-
-		src_page = xfs_get_page(src, srcoff);
-		if (IS_ERR(src_page)) {
-			error = PTR_ERR(src_page);
-			goto out_error;
-		}
-		dest_page = xfs_get_page(dest, destoff);
-		if (IS_ERR(dest_page)) {
-			error = PTR_ERR(dest_page);
-			unlock_page(src_page);
-			put_page(src_page);
-			goto out_error;
-		}
-		src_addr = kmap_atomic(src_page);
-		dest_addr = kmap_atomic(dest_page);
-
-		flush_dcache_page(src_page);
-		flush_dcache_page(dest_page);
-
-		if (memcmp(src_addr + src_poff, dest_addr + dest_poff, cmp_len))
-			same = false;
-
-		kunmap_atomic(dest_addr);
-		kunmap_atomic(src_addr);
-		unlock_page(dest_page);
-		unlock_page(src_page);
-		put_page(dest_page);
-		put_page(src_page);
-
-		if (!same)
-			break;
-
-		srcoff += cmp_len;
-		destoff += cmp_len;
-		len -= cmp_len;
-	}
-
-	*is_same = same;
-	return 0;
-
-out_error:
-	trace_xfs_reflink_compare_extents_error(XFS_I(dest), error, _RET_IP_);
-	return error;
-}
-
-/*
  * Link a range of blocks from one file to another.
  */
 int
@@ -1286,14 +1181,11 @@ xfs_reflink_remap_range(
 	struct inode		*inode_out = file_inode(file_out);
 	struct xfs_inode	*dest = XFS_I(inode_out);
 	struct xfs_mount	*mp = src->i_mount;
-	loff_t			bs = inode_out->i_sb->s_blocksize;
 	bool			same_inode = (inode_in == inode_out);
 	xfs_fileoff_t		sfsbno, dfsbno;
 	xfs_filblks_t		fsblen;
 	xfs_extlen_t		cowextsize;
-	loff_t			isize;
 	ssize_t			ret;
-	loff_t			blen;
 
 	if (!xfs_sb_version_hasreflink(&mp->m_sb))
 		return -EOPNOTSUPP;
@@ -1310,26 +1202,8 @@ xfs_reflink_remap_range(
 		xfs_lock_two_inodes(src, dest, XFS_MMAPLOCK_EXCL);
 	}
 
-	/* Don't touch certain kinds of inodes */
-	ret = -EPERM;
-	if (IS_IMMUTABLE(inode_out))
-		goto out_unlock;
-
-	ret = -ETXTBSY;
-	if (IS_SWAPFILE(inode_in) || IS_SWAPFILE(inode_out))
-		goto out_unlock;
-
-
-	/* Don't reflink dirs, pipes, sockets... */
-	ret = -EISDIR;
-	if (S_ISDIR(inode_in->i_mode) || S_ISDIR(inode_out->i_mode))
-		goto out_unlock;
+	/* Check file eligibility and prepare for block sharing. */
 	ret = -EINVAL;
-	if (S_ISFIFO(inode_in->i_mode) || S_ISFIFO(inode_out->i_mode))
-		goto out_unlock;
-	if (!S_ISREG(inode_in->i_mode) || !S_ISREG(inode_out->i_mode))
-		goto out_unlock;
-
 	/* Don't reflink realtime inodes */
 	if (XFS_IS_REALTIME_INODE(src) || XFS_IS_REALTIME_INODE(dest))
 		goto out_unlock;
@@ -1338,91 +1212,18 @@ xfs_reflink_remap_range(
 	if (IS_DAX(inode_in) || IS_DAX(inode_out))
 		goto out_unlock;
 
-	/* Are we going all the way to the end? */
-	isize = i_size_read(inode_in);
-	if (isize == 0) {
-		ret = 0;
-		goto out_unlock;
-	}
-
-	if (len == 0)
-		len = isize - pos_in;
-
-	/* Ensure offsets don't wrap and the input is inside i_size */
-	if (pos_in + len < pos_in || pos_out + len < pos_out ||
-	    pos_in + len > isize)
-		goto out_unlock;
-
-	/* Don't allow dedupe past EOF in the dest file */
-	if (is_dedupe) {
-		loff_t	disize;
-
-		disize = i_size_read(inode_out);
-		if (pos_out >= disize || pos_out + len > disize)
-			goto out_unlock;
-	}
-
-	/* If we're linking to EOF, continue to the block boundary. */
-	if (pos_in + len == isize)
-		blen = ALIGN(isize, bs) - pos_in;
-	else
-		blen = len;
-
-	/* Only reflink if we're aligned to block boundaries */
-	if (!IS_ALIGNED(pos_in, bs) || !IS_ALIGNED(pos_in + blen, bs) ||
-	    !IS_ALIGNED(pos_out, bs) || !IS_ALIGNED(pos_out + blen, bs))
-		goto out_unlock;
-
-	/* Don't allow overlapped reflink within the same file */
-	if (same_inode) {
-		if (pos_out + blen > pos_in && pos_out < pos_in + blen)
-			goto out_unlock;
-	}
-
-	/* Wait for the completion of any pending IOs on both files */
-	inode_dio_wait(inode_in);
-	if (!same_inode)
-		inode_dio_wait(inode_out);
-
-	ret = filemap_write_and_wait_range(inode_in->i_mapping,
-			pos_in, pos_in + len - 1);
-	if (ret)
-		goto out_unlock;
-
-	ret = filemap_write_and_wait_range(inode_out->i_mapping,
-			pos_out, pos_out + len - 1);
-	if (ret)
+	ret = vfs_clone_file_prep_inodes(inode_in, pos_in, inode_out, pos_out,
+			&len, is_dedupe);
+	if (ret || len == 0)
 		goto out_unlock;
 
 	trace_xfs_reflink_remap_range(src, pos_in, len, dest, pos_out);
 
-	/*
-	 * Check that the extents are the same.
-	 */
-	if (is_dedupe) {
-		bool		is_same = false;
-
-		ret = xfs_compare_extents(inode_in, pos_in, inode_out, pos_out,
-				len, &is_same);
-		if (ret)
-			goto out_unlock;
-		if (!is_same) {
-			ret = -EBADE;
-			goto out_unlock;
-		}
-	}
-
+	/* Set flags and remap blocks. */
 	ret = xfs_reflink_set_inode_flag(src, dest);
 	if (ret)
 		goto out_unlock;
 
-	/*
-	 * Invalidate the page cache so that we can clear any CoW mappings
-	 * in the destination file.
-	 */
-	truncate_inode_pages_range(&inode_out->i_data, pos_out,
-				   PAGE_ALIGN(pos_out + len) - 1);
-
 	dfsbno = XFS_B_TO_FSBT(mp, pos_out);
 	sfsbno = XFS_B_TO_FSBT(mp, pos_in);
 	fsblen = XFS_B_TO_FSB(mp, len);
@@ -1431,6 +1232,10 @@ xfs_reflink_remap_range(
 	if (ret)
 		goto out_unlock;
 
+	/* Zap any page cache for the destination file's range. */
+	truncate_inode_pages_range(&inode_out->i_data, pos_out,
+				   PAGE_ALIGN(pos_out + len) - 1);
+
 	/*
 	 * Carry the cowextsize hint from src to dest if we're sharing the
 	 * entire source file to the entire destination file, the source file
diff --git a/include/linux/fs.h b/include/linux/fs.h
index dc0478c..caea736 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1778,8 +1778,14 @@ extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
 		unsigned long, loff_t *, int);
 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
 				   loff_t, size_t, unsigned int);
+extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
+				      struct inode *inode_out, loff_t pos_out,
+				      u64 *len, bool is_dedupe);
 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
 		struct file *file_out, loff_t pos_out, u64 len);
+extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
+					 struct inode *dest, loff_t destoff,
+					 loff_t len, bool *is_same);
 extern int vfs_dedupe_file_range(struct file *file,
 				 struct file_dedupe_range *same);
 

[PATCH 03/11] ocfs2: convert inode refcount test to a helper

[Darrick J. Wong]

Replace the open-coded inode refcount flag test with a helper function
to reduce the potential for bugs.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/alloc.c        |    3 +--
 fs/ocfs2/file.c         |    7 +++----
 fs/ocfs2/inode.h        |    6 ++++++
 fs/ocfs2/move_extents.c |   10 ++--------
 fs/ocfs2/refcounttree.c |   22 +++++++++-------------
 fs/ocfs2/xattr.c        |    4 ++--
 6 files changed, 23 insertions(+), 29 deletions(-)


diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index f72712f..a0ca49f 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -5713,8 +5713,7 @@ int ocfs2_remove_btree_range(struct inode *inode,
 	struct ocfs2_refcount_tree *ref_tree = NULL;
 
 	if ((flags & OCFS2_EXT_REFCOUNTED) && len) {
-		BUG_ON(!(OCFS2_I(inode)->ip_dyn_features &
-			 OCFS2_HAS_REFCOUNT_FL));
+		BUG_ON(!ocfs2_is_refcount_inode(inode));
 
 		if (!refcount_tree_locked) {
 			ret = ocfs2_lock_refcount_tree(osb, refcount_loc, 1,
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 000c234..d261f3a 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -1030,7 +1030,7 @@ int ocfs2_extend_no_holes(struct inode *inode, struct buffer_head *di_bh,
 	 * Only quota files call this without a bh, and they can't be
 	 * refcounted.
 	 */
-	BUG_ON(!di_bh && (oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+	BUG_ON(!di_bh && ocfs2_is_refcount_inode(inode));
 	BUG_ON(!di_bh && !(oi->ip_flags & OCFS2_INODE_SYSTEM_FILE));
 
 	clusters_to_add = ocfs2_clusters_for_bytes(inode->i_sb, new_i_size);
@@ -1719,8 +1719,7 @@ static int ocfs2_remove_inode_range(struct inode *inode,
 	 * within one cluster(means is not exactly aligned to clustersize).
 	 */
 
-	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL) {
-
+	if (ocfs2_is_refcount_inode(inode)) {
 		ret = ocfs2_cow_file_pos(inode, di_bh, byte_start);
 		if (ret) {
 			mlog_errno(ret);
@@ -2036,7 +2035,7 @@ int ocfs2_check_range_for_refcount(struct inode *inode, loff_t pos,
 	struct super_block *sb = inode->i_sb;
 
 	if (!ocfs2_refcount_tree(OCFS2_SB(inode->i_sb)) ||
-	    !(OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL) ||
+	    !ocfs2_is_refcount_inode(inode) ||
 	    OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
 		return 0;
 
diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h
index 5af68fc..9b955f7 100644
--- a/fs/ocfs2/inode.h
+++ b/fs/ocfs2/inode.h
@@ -181,4 +181,10 @@ static inline struct ocfs2_inode_info *cache_info_to_inode(struct ocfs2_caching_
 	return container_of(ci, struct ocfs2_inode_info, ip_metadata_cache);
 }
 
+/* Does this inode have the reflink flag set? */
+static inline bool ocfs2_is_refcount_inode(struct inode *inode)
+{
+	return (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+}
+
 #endif /* OCFS2_INODE_H */
diff --git a/fs/ocfs2/move_extents.c b/fs/ocfs2/move_extents.c
index 4e8f32eb..e52a285 100644
--- a/fs/ocfs2/move_extents.c
+++ b/fs/ocfs2/move_extents.c
@@ -235,10 +235,7 @@ static int ocfs2_defrag_extent(struct ocfs2_move_extents_context *context,
 	u64 phys_blkno = ocfs2_clusters_to_blocks(inode->i_sb, phys_cpos);
 
 	if ((ext_flags & OCFS2_EXT_REFCOUNTED) && *len) {
-
-		BUG_ON(!(OCFS2_I(inode)->ip_dyn_features &
-			 OCFS2_HAS_REFCOUNT_FL));
-
+		BUG_ON(!ocfs2_is_refcount_inode(inode));
 		BUG_ON(!context->refcount_loc);
 
 		ret = ocfs2_lock_refcount_tree(osb, context->refcount_loc, 1,
@@ -581,10 +578,7 @@ static int ocfs2_move_extent(struct ocfs2_move_extents_context *context,
 	phys_blkno = ocfs2_clusters_to_blocks(inode->i_sb, phys_cpos);
 
 	if ((ext_flags & OCFS2_EXT_REFCOUNTED) && len) {
-
-		BUG_ON(!(OCFS2_I(inode)->ip_dyn_features &
-			 OCFS2_HAS_REFCOUNT_FL));
-
+		BUG_ON(!ocfs2_is_refcount_inode(inode));
 		BUG_ON(!context->refcount_loc);
 
 		ret = ocfs2_lock_refcount_tree(osb, context->refcount_loc, 1,
diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 1923851..3410eb1 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -410,7 +410,7 @@ static int ocfs2_get_refcount_block(struct inode *inode, u64 *ref_blkno)
 		goto out;
 	}
 
-	BUG_ON(!(OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+	BUG_ON(!ocfs2_is_refcount_inode(inode));
 
 	di = (struct ocfs2_dinode *)di_bh->b_data;
 	*ref_blkno = le64_to_cpu(di->i_refcount_loc);
@@ -570,7 +570,7 @@ static int ocfs2_create_refcount_tree(struct inode *inode,
 	u32 num_got;
 	u64 suballoc_loc, first_blkno;
 
-	BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+	BUG_ON(ocfs2_is_refcount_inode(inode));
 
 	trace_ocfs2_create_refcount_tree(
 		(unsigned long long)OCFS2_I(inode)->ip_blkno);
@@ -708,7 +708,7 @@ static int ocfs2_set_refcount_tree(struct inode *inode,
 	struct ocfs2_refcount_block *rb;
 	struct ocfs2_refcount_tree *ref_tree;
 
-	BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+	BUG_ON(ocfs2_is_refcount_inode(inode));
 
 	ret = ocfs2_lock_refcount_tree(osb, refcount_loc, 1,
 				       &ref_tree, &ref_root_bh);
@@ -775,7 +775,7 @@ int ocfs2_remove_refcount_tree(struct inode *inode, struct buffer_head *di_bh)
 	u64 blk = 0, bg_blkno = 0, ref_blkno = le64_to_cpu(di->i_refcount_loc);
 	u16 bit = 0;
 
-	if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL))
+	if (!ocfs2_is_refcount_inode(inode))
 		return 0;
 
 	BUG_ON(!ref_blkno);
@@ -2299,11 +2299,10 @@ int ocfs2_decrease_refcount(struct inode *inode,
 {
 	int ret;
 	u64 ref_blkno;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct buffer_head *ref_root_bh = NULL;
 	struct ocfs2_refcount_tree *tree;
 
-	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+	BUG_ON(!ocfs2_is_refcount_inode(inode));
 
 	ret = ocfs2_get_refcount_block(inode, &ref_blkno);
 	if (ret) {
@@ -2533,7 +2532,6 @@ int ocfs2_prepare_refcount_change_for_del(struct inode *inode,
 					  int *ref_blocks)
 {
 	int ret;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct buffer_head *ref_root_bh = NULL;
 	struct ocfs2_refcount_tree *tree;
 	u64 start_cpos = ocfs2_blocks_to_clusters(inode->i_sb, phys_blkno);
@@ -2544,7 +2542,7 @@ int ocfs2_prepare_refcount_change_for_del(struct inode *inode,
 		goto out;
 	}
 
-	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+	BUG_ON(!ocfs2_is_refcount_inode(inode));
 
 	ret = ocfs2_get_refcount_tree(OCFS2_SB(inode->i_sb),
 				      refcount_loc, &tree);
@@ -3412,14 +3410,13 @@ static int ocfs2_refcount_cow_hunk(struct inode *inode,
 {
 	int ret;
 	u32 cow_start = 0, cow_len = 0;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
 	struct buffer_head *ref_root_bh = NULL;
 	struct ocfs2_refcount_tree *ref_tree;
 	struct ocfs2_cow_context *context = NULL;
 
-	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+	BUG_ON(!ocfs2_is_refcount_inode(inode));
 
 	ret = ocfs2_refcount_cal_cow_clusters(inode, &di->id2.i_list,
 					      cpos, write_len, max_cpos,
@@ -3629,11 +3626,10 @@ int ocfs2_refcount_cow_xattr(struct inode *inode,
 {
 	int ret;
 	struct ocfs2_xattr_value_root *xv = vb->vb_xv;
-	struct ocfs2_inode_info *oi = OCFS2_I(inode);
 	struct ocfs2_cow_context *context = NULL;
 	u32 cow_start, cow_len;
 
-	BUG_ON(!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL));
+	BUG_ON(!ocfs2_is_refcount_inode(inode));
 
 	ret = ocfs2_refcount_cal_cow_clusters(inode, &xv->xr_list,
 					      cpos, write_len, UINT_MAX,
@@ -3807,7 +3803,7 @@ static int ocfs2_attach_refcount_tree(struct inode *inode,
 
 	ocfs2_init_dealloc_ctxt(&dealloc);
 
-	if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL)) {
+	if (!ocfs2_is_refcount_inode(inode)) {
 		ret = ocfs2_create_refcount_tree(inode, di_bh);
 		if (ret) {
 			mlog_errno(ret);
diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c
index cb157a3..3c5384d 100644
--- a/fs/ocfs2/xattr.c
+++ b/fs/ocfs2/xattr.c
@@ -2577,7 +2577,7 @@ int ocfs2_xattr_remove(struct inode *inode, struct buffer_head *di_bh)
 	if (!(oi->ip_dyn_features & OCFS2_HAS_XATTR_FL))
 		return 0;
 
-	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL) {
+	if (ocfs2_is_refcount_inode(inode)) {
 		ret = ocfs2_lock_refcount_tree(OCFS2_SB(inode->i_sb),
 					       le64_to_cpu(di->i_refcount_loc),
 					       1, &ref_tree, &ref_root_bh);
@@ -3608,7 +3608,7 @@ int ocfs2_xattr_set(struct inode *inode,
 	}
 
 	/* Check whether the value is refcounted and do some preparation. */
-	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL &&
+	if (ocfs2_is_refcount_inode(inode) &&
 	    (!xis.not_found || !xbs.not_found)) {
 		ret = ocfs2_prepare_refcount_xattr(inode, di, &xi,
 						   &xis, &xbs, &ref_tree,

[PATCH 04/11] ocfs2: add newlines to some error messages

[Darrick J. Wong]

These two error messages are missing the trailing newline.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/alloc.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)


diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index a0ca49f..d4ec0d8 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -5194,7 +5194,7 @@ int ocfs2_change_extent_flag(handle_t *handle,
 	rec = &el->l_recs[index];
 	if (new_flags && (rec->e_flags & new_flags)) {
 		mlog(ML_ERROR, "Owner %llu tried to set %d flags on an "
-		     "extent that already had them",
+		     "extent that already had them\n",
 		     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 		     new_flags);
 		goto out;
@@ -5202,7 +5202,7 @@ int ocfs2_change_extent_flag(handle_t *handle,
 
 	if (clear_flags && !(rec->e_flags & clear_flags)) {
 		mlog(ML_ERROR, "Owner %llu tried to clear %d flags on an "
-		     "extent that didn't have them",
+		     "extent that didn't have them\n",
 		     (unsigned long long)ocfs2_metadata_cache_owner(et->et_ci),
 		     clear_flags);
 		goto out;

[PATCH 05/11] ocfs2: prohibit refcounted swapfiles

[Darrick J. Wong]

The swapfile mechanism calls bmap once to find all the swap file
mappings, which means that we cannot properly support CoW remapping.
Therefore, error out if the swap code tries to call bmap on a
refcounted file.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/aops.c |    9 +++++++++
 1 file changed, 9 insertions(+)


diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index c5c5b97..4d037db 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -464,6 +464,15 @@ static sector_t ocfs2_bmap(struct address_space *mapping, sector_t block)
 	trace_ocfs2_bmap((unsigned long long)OCFS2_I(inode)->ip_blkno,
 			 (unsigned long long)block);
 
+	/*
+	 * The swap code (ab-)uses ->bmap to get a block mapping and then
+	 * bypasseѕ the file system for actual I/O.  We really can't allow
+	 * that on refcounted inodes, so we have to skip out here.  And yes,
+	 * 0 is the magic code for a bmap error..
+	 */
+	if (ocfs2_is_refcount_inode(inode))
+		return 0;
+
 	/* We don't need to lock journal system files, since they aren't
 	 * accessed concurrently from multiple nodes.
 	 */

[PATCH 06/11] ocfs2: budget for extent tree splits when adding refcount flag

[Darrick J. Wong]

When we're adding the refcount flag to an extent, we have to budget
enough space to handle a full extent btree split in addition to
whatever modifications have to be made to the refcount btree.  We
don't currently do this, with the result that generic/186 crashes
when we need an extent split but not a refcount split because meta_ac
never gets allocated.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/refcounttree.c |    3 +++
 1 file changed, 3 insertions(+)


diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 3410eb1..6c98d56 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -3692,6 +3692,9 @@ int ocfs2_add_refcount_flag(struct inode *inode,
 	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
 	struct ocfs2_alloc_context *meta_ac = NULL;
 
+	/* We need to be able to handle at least an extent tree split. */
+	ref_blocks = ocfs2_extend_meta_needed(data_et->et_root_el);
+
 	ret = ocfs2_calc_refcount_meta_credits(inode->i_sb,
 					       ref_ci, ref_root_bh,
 					       p_cluster, num_clusters,

[PATCH 07/11] ocfs2: don't eat io errors during _dio_end_io_write

[Darrick J. Wong]

ocfs2_dio_end_io_write eats whatever errors may happen,
which means that write errors do not propagate to userspace.
Fix that.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/aops.c |   15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)


diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 4d037db..136a49c 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -2263,10 +2263,10 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
 	return ret;
 }
 
-static void ocfs2_dio_end_io_write(struct inode *inode,
-				   struct ocfs2_dio_write_ctxt *dwc,
-				   loff_t offset,
-				   ssize_t bytes)
+static int ocfs2_dio_end_io_write(struct inode *inode,
+				  struct ocfs2_dio_write_ctxt *dwc,
+				  loff_t offset,
+				  ssize_t bytes)
 {
 	struct ocfs2_cached_dealloc_ctxt dealloc;
 	struct ocfs2_extent_tree et;
@@ -2374,6 +2374,8 @@ static void ocfs2_dio_end_io_write(struct inode *inode,
 	if (locked)
 		inode_unlock(inode);
 	ocfs2_dio_free_write_ctx(inode, dwc);
+
+	return ret;
 }
 
 /*
@@ -2388,6 +2390,7 @@ static int ocfs2_dio_end_io(struct kiocb *iocb,
 {
 	struct inode *inode = file_inode(iocb->ki_filp);
 	int level;
+	int ret = 0;
 
 	if (bytes <= 0)
 		return 0;
@@ -2396,13 +2399,13 @@ static int ocfs2_dio_end_io(struct kiocb *iocb,
 	BUG_ON(!ocfs2_iocb_is_rw_locked(iocb));
 
 	if (private)
-		ocfs2_dio_end_io_write(inode, private, offset, bytes);
+		ret = ocfs2_dio_end_io_write(inode, private, offset, bytes);
 
 	ocfs2_iocb_clear_rw_locked(iocb);
 
 	level = ocfs2_iocb_rw_locked_level(iocb);
 	ocfs2_rw_unlock(inode, level);
-	return 0;
+	return ret;
 }
 
 static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)

[PATCH 08/11] ocfs2: always unlock when completing dio writes

[Darrick J. Wong]

Always unlock the inode when completing dio writes, even if an error
has occurrred.  The caller already checks the inode and unlocks it
if needed, so we might as well reduce contention.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/aops.c |    5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)


diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 136a49c..3c531f1 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -2392,13 +2392,10 @@ static int ocfs2_dio_end_io(struct kiocb *iocb,
 	int level;
 	int ret = 0;
 
-	if (bytes <= 0)
-		return 0;
-
 	/* this io's submitter should not have unlocked this before we could */
 	BUG_ON(!ocfs2_iocb_is_rw_locked(iocb));
 
-	if (private)
+	if (bytes > 0 && private)
 		ret = ocfs2_dio_end_io_write(inode, private, offset, bytes);
 
 	ocfs2_iocb_clear_rw_locked(iocb);

[PATCH 09/11] ocfs2: fix bad pointer cast

[Darrick J. Wong]

generic/188 triggered a dmesg stack trace because the dio completion
was casting a buffer head to an on-disk inode, which is whacky.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/aops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)


diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 3c531f1..3372d82 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -2317,7 +2317,7 @@ static int ocfs2_dio_end_io_write(struct inode *inode,
 			mlog_errno(ret);
 	}
 
-	di = (struct ocfs2_dinode *)di_bh;
+	di = (struct ocfs2_dinode *)di_bh->b_data;
 
 	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(inode), di_bh);
 

[PATCH 10/11] ocfs2: charge quota for reflinked blocks

[Darrick J. Wong]

When ocfs2 shares blocks from one file to another, it's necessary to
charge that many blocks to the quota because ocfs2 tallies block charges
according to the number of blocks mapped, not the number of physical
blocks used.

Without this patch, reflinking X blocks and then CoWing all of them
causes quota usage to *decrease* by X as seen in generic/305.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ocfs2/refcounttree.c |    7 +++++++
 1 file changed, 7 insertions(+)


diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 6c98d56..dc8089a 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -3933,6 +3933,13 @@ static int ocfs2_add_refcounted_extent(struct inode *inode,
 	ret = ocfs2_increase_refcount(handle, ref_ci, ref_root_bh,
 				      p_cluster, num_clusters,
 				      meta_ac, dealloc);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+	ret = dquot_alloc_space_nodirty(inode,
+		ocfs2_clusters_to_bytes(osb->sb, num_clusters));
 	if (ret)
 		mlog_errno(ret);
 

[PATCH 11/11] ocfs2: implement the VFS clone_range, copy_range, and dedupe_range features

[Darrick J. Wong]

Connect the new VFS clone_range, copy_range, and dedupe_range features
to the existing reflink capability of ocfs2.  Compared to the existing
ocfs2 reflink ioctl We have to do things a little differently to support
the VFS semantics (we can clone subranges of a file but we don't clone
xattrs), but the VFS ioctls are more broadly supported.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
v2: Convert inline data files to extents files before reflinking,
and fix i_blocks so that stat(2) output is correct.
v3: Make zero-length dedupe consistent with btrfs behavior.
v4: Use VFS double-inode lock routines and remove MAX_DEDUPE_LEN.
---
 fs/ocfs2/file.c         |   35 +++-
 fs/ocfs2/file.h         |    3 
 fs/ocfs2/refcounttree.c |  432 +++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/refcounttree.h |    7 +
 4 files changed, 474 insertions(+), 3 deletions(-)


diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index d261f3a..c488965 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -1667,9 +1667,9 @@ static void ocfs2_calc_trunc_pos(struct inode *inode,
 	*done = ret;
 }
 
-static int ocfs2_remove_inode_range(struct inode *inode,
-				    struct buffer_head *di_bh, u64 byte_start,
-				    u64 byte_len)
+int ocfs2_remove_inode_range(struct inode *inode,
+			     struct buffer_head *di_bh, u64 byte_start,
+			     u64 byte_len)
 {
 	int ret = 0, flags = 0, done = 0, i;
 	u32 trunc_start, trunc_len, trunc_end, trunc_cpos, phys_cpos;
@@ -2439,6 +2439,31 @@ static loff_t ocfs2_file_llseek(struct file *file, loff_t offset, int whence)
 	return offset;
 }
 
+static int ocfs2_file_clone_range(struct file *file_in,
+				  loff_t pos_in,
+				  struct file *file_out,
+				  loff_t pos_out,
+				  u64 len)
+{
+	return ocfs2_reflink_remap_range(file_in, pos_in, file_out, pos_out,
+					 len, false);
+}
+
+static ssize_t ocfs2_file_dedupe_range(struct file *src_file,
+				       u64 loff,
+				       u64 len,
+				       struct file *dst_file,
+				       u64 dst_loff)
+{
+	int error;
+
+	error = ocfs2_reflink_remap_range(src_file, loff, dst_file, dst_loff,
+					  len, true);
+	if (error)
+		return error;
+	return len;
+}
+
 const struct inode_operations ocfs2_file_iops = {
 	.setattr	= ocfs2_setattr,
 	.getattr	= ocfs2_getattr,
@@ -2478,6 +2503,8 @@ const struct file_operations ocfs2_fops = {
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= iter_file_splice_write,
 	.fallocate	= ocfs2_fallocate,
+	.clone_file_range = ocfs2_file_clone_range,
+	.dedupe_file_range = ocfs2_file_dedupe_range,
 };
 
 const struct file_operations ocfs2_dops = {
@@ -2523,6 +2550,8 @@ const struct file_operations ocfs2_fops_no_plocks = {
 	.splice_read	= generic_file_splice_read,
 	.splice_write	= iter_file_splice_write,
 	.fallocate	= ocfs2_fallocate,
+	.clone_file_range = ocfs2_file_clone_range,
+	.dedupe_file_range = ocfs2_file_dedupe_range,
 };
 
 const struct file_operations ocfs2_dops_no_plocks = {
diff --git a/fs/ocfs2/file.h b/fs/ocfs2/file.h
index e8c62f2..897fd9a 100644
--- a/fs/ocfs2/file.h
+++ b/fs/ocfs2/file.h
@@ -82,4 +82,7 @@ int ocfs2_change_file_space(struct file *file, unsigned int cmd,
 
 int ocfs2_check_range_for_refcount(struct inode *inode, loff_t pos,
 				   size_t count);
+int ocfs2_remove_inode_range(struct inode *inode,
+			     struct buffer_head *di_bh, u64 byte_start,
+			     u64 byte_len);
 #endif /* OCFS2_FILE_H */
diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index dc8089a..b18465e 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -34,6 +34,7 @@
 #include "xattr.h"
 #include "namei.h"
 #include "ocfs2_trace.h"
+#include "file.h"
 
 #include <linux/bio.h>
 #include <linux/blkdev.h>
@@ -4448,3 +4449,434 @@ int ocfs2_reflink_ioctl(struct inode *inode,
 
 	return error;
 }
+
+/* Update destination inode size, if necessary. */
+static int ocfs2_reflink_update_dest(struct inode *dest,
+				     struct buffer_head *d_bh,
+				     loff_t newlen)
+{
+	handle_t *handle;
+	int ret;
+
+	dest->i_blocks = ocfs2_inode_sector_count(dest);
+
+	if (newlen <= i_size_read(dest))
+		return 0;
+
+	handle = ocfs2_start_trans(OCFS2_SB(dest->i_sb),
+				   OCFS2_INODE_UPDATE_CREDITS);
+	if (IS_ERR(handle)) {
+		ret = PTR_ERR(handle);
+		mlog_errno(ret);
+		return ret;
+	}
+
+	/* Extend i_size if needed. */
+	spin_lock(&OCFS2_I(dest)->ip_lock);
+	if (newlen > i_size_read(dest))
+		i_size_write(dest, newlen);
+	spin_unlock(&OCFS2_I(dest)->ip_lock);
+	dest->i_ctime = dest->i_mtime = current_time(dest);
+
+	ret = ocfs2_mark_inode_dirty(handle, dest, d_bh);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_commit;
+	}
+
+out_commit:
+	ocfs2_commit_trans(OCFS2_SB(dest->i_sb), handle);
+	return ret;
+}
+
+/* Remap the range pos_in:len in s_inode to pos_out:len in t_inode. */
+static int ocfs2_reflink_remap_extent(struct inode *s_inode,
+				      struct buffer_head *s_bh,
+				      loff_t pos_in,
+				      struct inode *t_inode,
+				      struct buffer_head *t_bh,
+				      loff_t pos_out,
+				      loff_t len,
+				      struct ocfs2_cached_dealloc_ctxt *dealloc)
+{
+	struct ocfs2_extent_tree s_et;
+	struct ocfs2_extent_tree t_et;
+	struct ocfs2_dinode *dis;
+	struct buffer_head *ref_root_bh = NULL;
+	struct ocfs2_refcount_tree *ref_tree;
+	struct ocfs2_super *osb;
+	loff_t pstart, plen;
+	u32 p_cluster, num_clusters, slast, spos, tpos;
+	unsigned int ext_flags;
+	int ret = 0;
+
+	osb = OCFS2_SB(s_inode->i_sb);
+	dis = (struct ocfs2_dinode *)s_bh->b_data;
+	ocfs2_init_dinode_extent_tree(&s_et, INODE_CACHE(s_inode), s_bh);
+	ocfs2_init_dinode_extent_tree(&t_et, INODE_CACHE(t_inode), t_bh);
+
+	spos = ocfs2_bytes_to_clusters(s_inode->i_sb, pos_in);
+	tpos = ocfs2_bytes_to_clusters(t_inode->i_sb, pos_out);
+	slast = ocfs2_clusters_for_bytes(s_inode->i_sb, pos_in + len);
+
+	while (spos < slast) {
+		if (fatal_signal_pending(current)) {
+			ret = -EINTR;
+			goto out;
+		}
+
+		/* Look up the extent. */
+		ret = ocfs2_get_clusters(s_inode, spos, &p_cluster,
+					 &num_clusters, &ext_flags);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		num_clusters = min_t(u32, num_clusters, slast - spos);
+
+		/* Punch out the dest range. */
+		pstart = ocfs2_clusters_to_bytes(t_inode->i_sb, tpos);
+		plen = ocfs2_clusters_to_bytes(t_inode->i_sb, num_clusters);
+		ret = ocfs2_remove_inode_range(t_inode, t_bh, pstart, plen);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		if (p_cluster == 0)
+			goto next_loop;
+
+		/* Lock the refcount btree... */
+		ret = ocfs2_lock_refcount_tree(osb,
+					       le64_to_cpu(dis->i_refcount_loc),
+					       1, &ref_tree, &ref_root_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+
+		/* Mark s_inode's extent as refcounted. */
+		if (!(ext_flags & OCFS2_EXT_REFCOUNTED)) {
+			ret = ocfs2_add_refcount_flag(s_inode, &s_et,
+						      &ref_tree->rf_ci,
+						      ref_root_bh, spos,
+						      p_cluster, num_clusters,
+						      dealloc, NULL);
+			if (ret) {
+				mlog_errno(ret);
+				goto out_unlock_refcount;
+			}
+		}
+
+		/* Map in the new extent. */
+		ext_flags |= OCFS2_EXT_REFCOUNTED;
+		ret = ocfs2_add_refcounted_extent(t_inode, &t_et,
+						  &ref_tree->rf_ci,
+						  ref_root_bh,
+						  tpos, p_cluster,
+						  num_clusters,
+						  ext_flags,
+						  dealloc);
+		if (ret) {
+			mlog_errno(ret);
+			goto out_unlock_refcount;
+		}
+
+		ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+		brelse(ref_root_bh);
+next_loop:
+		spos += num_clusters;
+		tpos += num_clusters;
+	}
+
+out:
+	return ret;
+out_unlock_refcount:
+	ocfs2_unlock_refcount_tree(osb, ref_tree, 1);
+	brelse(ref_root_bh);
+	return ret;
+}
+
+/* Set up refcount tree and remap s_inode to t_inode. */
+static int ocfs2_reflink_remap_blocks(struct inode *s_inode,
+				      struct buffer_head *s_bh,
+				      loff_t pos_in,
+				      struct inode *t_inode,
+				      struct buffer_head *t_bh,
+				      loff_t pos_out,
+				      loff_t len)
+{
+	struct ocfs2_cached_dealloc_ctxt dealloc;
+	struct ocfs2_super *osb;
+	struct ocfs2_dinode *dis;
+	struct ocfs2_dinode *dit;
+	int ret;
+
+	osb = OCFS2_SB(s_inode->i_sb);
+	dis = (struct ocfs2_dinode *)s_bh->b_data;
+	dit = (struct ocfs2_dinode *)t_bh->b_data;
+	ocfs2_init_dealloc_ctxt(&dealloc);
+
+	/*
+	 * If we're reflinking the entire file and the source is inline
+	 * data, just copy the contents.
+	 */
+	if (pos_in == pos_out && pos_in == 0 && len == i_size_read(s_inode) &&
+	    i_size_read(t_inode) <= len &&
+	    (OCFS2_I(s_inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL)) {
+		ret = ocfs2_duplicate_inline_data(s_inode, s_bh, t_inode, t_bh);
+		if (ret)
+			mlog_errno(ret);
+		goto out;
+	}
+
+	/*
+	 * If both inodes belong to two different refcount groups then
+	 * forget it because we don't know how (or want) to go merging
+	 * refcount trees.
+	 */
+	ret = -EOPNOTSUPP;
+	if (ocfs2_is_refcount_inode(s_inode) &&
+	    ocfs2_is_refcount_inode(t_inode) &&
+	    le64_to_cpu(dis->i_refcount_loc) !=
+	    le64_to_cpu(dit->i_refcount_loc))
+		goto out;
+
+	/* Neither inode has a refcount tree.  Add one to s_inode. */
+	if (!ocfs2_is_refcount_inode(s_inode) &&
+	    !ocfs2_is_refcount_inode(t_inode)) {
+		ret = ocfs2_create_refcount_tree(s_inode, s_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
+	/* Ensure that both inodes end up with the same refcount tree. */
+	if (!ocfs2_is_refcount_inode(s_inode)) {
+		ret = ocfs2_set_refcount_tree(s_inode, s_bh,
+					      le64_to_cpu(dit->i_refcount_loc));
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+	if (!ocfs2_is_refcount_inode(t_inode)) {
+		ret = ocfs2_set_refcount_tree(t_inode, t_bh,
+					      le64_to_cpu(dis->i_refcount_loc));
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
+	/* Turn off inline data in the dest file. */
+	if (OCFS2_I(t_inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
+		ret = ocfs2_convert_inline_data_to_extents(t_inode, t_bh);
+		if (ret) {
+			mlog_errno(ret);
+			goto out;
+		}
+	}
+
+	/* Actually remap extents now. */
+	ret = ocfs2_reflink_remap_extent(s_inode, s_bh, pos_in, t_inode, t_bh,
+					 pos_out, len, &dealloc);
+	if (ret) {
+		mlog_errno(ret);
+		goto out;
+	}
+
+out:
+	if (ocfs2_dealloc_has_cluster(&dealloc)) {
+		ocfs2_schedule_truncate_log_flush(osb, 1);
+		ocfs2_run_deallocs(osb, &dealloc);
+	}
+
+	return ret;
+}
+
+/* Lock an inode and grab a bh pointing to the inode. */
+static int ocfs2_reflink_inodes_lock(struct inode *s_inode,
+				     struct buffer_head **bh1,
+				     struct inode *t_inode,
+				     struct buffer_head **bh2)
+{
+	struct inode *inode1;
+	struct inode *inode2;
+	struct ocfs2_inode_info *oi1;
+	struct ocfs2_inode_info *oi2;
+	bool same_inode = (s_inode == t_inode);
+	int status;
+
+	/* First grab the VFS and rw locks. */
+	lock_two_nondirectories(s_inode, t_inode);
+	inode1 = s_inode;
+	inode2 = t_inode;
+	if (inode1->i_ino > inode2->i_ino)
+		swap(inode1, inode2);
+
+	status = ocfs2_rw_lock(inode1, 1);
+	if (status) {
+		mlog_errno(status);
+		goto out_i1;
+	}
+	if (!same_inode) {
+		status = ocfs2_rw_lock(inode2, 1);
+		if (status) {
+			mlog_errno(status);
+			goto out_i2;
+		}
+	}
+
+	/* Now go for the cluster locks */
+	oi1 = OCFS2_I(inode1);
+	oi2 = OCFS2_I(inode2);
+
+	trace_ocfs2_double_lock((unsigned long long)oi1->ip_blkno,
+				(unsigned long long)oi2->ip_blkno);
+
+	if (*bh1)
+		*bh1 = NULL;
+	if (*bh2)
+		*bh2 = NULL;
+
+	/* We always want to lock the one with the lower lockid first. */
+	if (oi1->ip_blkno > oi2->ip_blkno)
+		mlog_errno(-ENOLCK);
+
+	/* lock id1 */
+	status = ocfs2_inode_lock_nested(inode1, bh1, 1, OI_LS_REFLINK_TARGET);
+	if (status < 0) {
+		if (status != -ENOENT)
+			mlog_errno(status);
+		goto out_rw2;
+	}
+
+	/* lock id2 */
+	if (!same_inode) {
+		status = ocfs2_inode_lock_nested(inode2, bh2, 1,
+						 OI_LS_REFLINK_TARGET);
+		if (status < 0) {
+			if (status != -ENOENT)
+				mlog_errno(status);
+			goto out_cl1;
+		}
+	} else
+		*bh2 = *bh1;
+
+	trace_ocfs2_double_lock_end(
+			(unsigned long long)OCFS2_I(inode1)->ip_blkno,
+			(unsigned long long)OCFS2_I(inode2)->ip_blkno);
+
+	return 0;
+
+out_cl1:
+	ocfs2_inode_unlock(inode1, 1);
+	brelse(*bh1);
+	*bh1 = NULL;
+out_rw2:
+	ocfs2_rw_unlock(inode2, 1);
+out_i2:
+	ocfs2_rw_unlock(inode1, 1);
+out_i1:
+	unlock_two_nondirectories(s_inode, t_inode);
+	return status;
+}
+
+/* Unlock both inodes and release buffers. */
+static void ocfs2_reflink_inodes_unlock(struct inode *s_inode,
+					struct buffer_head *s_bh,
+					struct inode *t_inode,
+					struct buffer_head *t_bh)
+{
+	ocfs2_inode_unlock(s_inode, 1);
+	ocfs2_rw_unlock(s_inode, 1);
+	brelse(s_bh);
+	if (s_inode != t_inode) {
+		ocfs2_inode_unlock(t_inode, 1);
+		ocfs2_rw_unlock(t_inode, 1);
+		brelse(t_bh);
+	}
+	unlock_two_nondirectories(s_inode, t_inode);
+}
+
+/* Link a range of blocks from one file to another. */
+int ocfs2_reflink_remap_range(struct file *file_in,
+			      loff_t pos_in,
+			      struct file *file_out,
+			      loff_t pos_out,
+			      u64 len,
+			      bool is_dedupe)
+{
+	struct inode *inode_in = file_inode(file_in);
+	struct inode *inode_out = file_inode(file_out);
+	struct ocfs2_super *osb = OCFS2_SB(inode_in->i_sb);
+	struct buffer_head *in_bh = NULL, *out_bh = NULL;
+	bool same_inode = (inode_in == inode_out);
+	ssize_t ret;
+
+	if (!ocfs2_refcount_tree(osb))
+		return -EOPNOTSUPP;
+	if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
+		return -EROFS;
+
+	/* Lock both files against IO */
+	ret = ocfs2_reflink_inodes_lock(inode_in, &in_bh, inode_out, &out_bh);
+	if (ret)
+		return ret;
+
+	/* Check file eligibility and prepare for block sharing. */
+	ret = -EINVAL;
+	if ((OCFS2_I(inode_in)->ip_flags & OCFS2_INODE_SYSTEM_FILE) ||
+	    (OCFS2_I(inode_out)->ip_flags & OCFS2_INODE_SYSTEM_FILE))
+		goto out_unlock;
+
+	ret = vfs_clone_file_prep_inodes(inode_in, pos_in, inode_out, pos_out,
+			&len, is_dedupe);
+	if (ret || len == 0)
+		goto out_unlock;
+
+	/* Lock out changes to the allocation maps and remap. */
+	down_write(&OCFS2_I(inode_in)->ip_alloc_sem);
+	if (!same_inode)
+		down_write_nested(&OCFS2_I(inode_out)->ip_alloc_sem,
+				  SINGLE_DEPTH_NESTING);
+
+	ret = ocfs2_reflink_remap_blocks(inode_in, in_bh, pos_in, inode_out,
+					 out_bh, pos_out, len);
+
+	/* Zap any page cache for the destination file's range. */
+	if (!ret)
+		truncate_inode_pages_range(&inode_out->i_data, pos_out,
+					   PAGE_ALIGN(pos_out + len) - 1);
+
+	up_write(&OCFS2_I(inode_in)->ip_alloc_sem);
+	if (!same_inode)
+		up_write(&OCFS2_I(inode_out)->ip_alloc_sem);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	/*
+	 * Empty the extent map so that we may get the right extent
+	 * record from the disk.
+	 */
+	ocfs2_extent_map_trunc(inode_in, 0);
+	ocfs2_extent_map_trunc(inode_out, 0);
+
+	ret = ocfs2_reflink_update_dest(inode_out, out_bh, pos_out + len);
+	if (ret) {
+		mlog_errno(ret);
+		goto out_unlock;
+	}
+
+	ocfs2_reflink_inodes_unlock(inode_in, in_bh, inode_out, out_bh);
+	return 0;
+
+out_unlock:
+	ocfs2_reflink_inodes_unlock(inode_in, in_bh, inode_out, out_bh);
+	return ret;
+}
diff --git a/fs/ocfs2/refcounttree.h b/fs/ocfs2/refcounttree.h
index 6422bbc..4af55bf 100644
--- a/fs/ocfs2/refcounttree.h
+++ b/fs/ocfs2/refcounttree.h
@@ -115,4 +115,11 @@ int ocfs2_reflink_ioctl(struct inode *inode,
 			const char __user *oldname,
 			const char __user *newname,
 			bool preserve);
+int ocfs2_reflink_remap_range(struct file *file_in,
+			      loff_t pos_in,
+			      struct file *file_out,
+			      loff_t pos_out,
+			      u64 len,
+			      bool is_dedupe);
+
 #endif /* OCFS2_REFCOUNTTREE_H */

Re: [PATCH v4 00/11] ocfs2: wire up {clone,copy,dedupe}_range

[Al Viro]

On Sat, Dec 10, 2016 at 12:42:55PM -0800, Darrick J. Wong wrote:
> Hi all,
> 
> This is the fourth submission of a series of patches that wire up the
> existing ocfs2 reflinking capabilities to the new(ish) VFS
> {copy,clone,dedupe}_range interface.  The first few patches hoist some
> of the reflink/dedupe prep functions into the VFS since the code is
> common to XFS and ocfs2.  Then there are fixes for some ocfs2 bugs that
> I found; and finally the last kernel patch contains the new code.
> 
> A few minor fixes to xfstests are needed to make more of the tests run;
> those have been sent separately to the fstests lists.  Some of the
> xfstests will fail because ocfs2's design does not allow sharing of
> blocks between disjoint sets of files with shared blocks.  That is to
> say, if A shares with B and C shares with D, B and C cannot be made to
> share blocks.

Merged, will be in tonight's push.

Re: [PATCH v4 00/11] ocfs2: wire up {clone,copy,dedupe}_range

[Al Viro]

On Sat, Dec 10, 2016 at 11:43:25PM +0000, Al Viro wrote:
> On Sat, Dec 10, 2016 at 12:42:55PM -0800, Darrick J. Wong wrote:
> > Hi all,
> > 
> > This is the fourth submission of a series of patches that wire up the
> > existing ocfs2 reflinking capabilities to the new(ish) VFS
> > {copy,clone,dedupe}_range interface.  The first few patches hoist some
> > of the reflink/dedupe prep functions into the VFS since the code is
> > common to XFS and ocfs2.  Then there are fixes for some ocfs2 bugs that
> > I found; and finally the last kernel patch contains the new code.
> > 
> > A few minor fixes to xfstests are needed to make more of the tests run;
> > those have been sent separately to the fstests lists.  Some of the
> > xfstests will fail because ocfs2's design does not allow sharing of
> > blocks between disjoint sets of files with shared blocks.  That is to
> > say, if A shares with B and C shares with D, B and C cannot be made to
> > share blocks.
> 
> Merged, will be in tonight's push.

... of #for-next, that is - not to Linus ;-)

Re: [PATCH v4 00/11] ocfs2: wire up {clone,copy,dedupe}_range

[Darrick J. Wong]

On Sat, Dec 10, 2016 at 11:46:27PM +0000, Al Viro wrote:
> On Sat, Dec 10, 2016 at 11:43:25PM +0000, Al Viro wrote:
> > On Sat, Dec 10, 2016 at 12:42:55PM -0800, Darrick J. Wong wrote:
> > > Hi all,
> > > 
> > > This is the fourth submission of a series of patches that wire up the
> > > existing ocfs2 reflinking capabilities to the new(ish) VFS
> > > {copy,clone,dedupe}_range interface.  The first few patches hoist some
> > > of the reflink/dedupe prep functions into the VFS since the code is
> > > common to XFS and ocfs2.  Then there are fixes for some ocfs2 bugs that
> > > I found; and finally the last kernel patch contains the new code.
> > > 
> > > A few minor fixes to xfstests are needed to make more of the tests run;
> > > those have been sent separately to the fstests lists.  Some of the
> > > xfstests will fail because ocfs2's design does not allow sharing of
> > > blocks between disjoint sets of files with shared blocks.  That is to
> > > say, if A shares with B and C shares with D, B and C cannot be made to
> > > share blocks.
> > 
> > Merged, will be in tonight's push.
> 
> ... of #for-next, that is - not to Linus ;-)

Ok.  Thank you!  I'm looking forward to wider testing, particularly
the ocfs2 pieces.

--D