Btrfs: add new defrag-range ioctl.

The btrfs defrag ioctl was limited to doing the entire file.  This
commit adds a new interface that can defrag a specific range inside
the file.

It can also force compression on the file, allowing you to selectively
compress individual files after they were created, even when mount -o
compress isn't turned on.
Signed-off-by: Chris Mason <chris.mason@oracle.com>

fs/btrfs/btrfs_inode.h

@@ -153,6 +153,11 @@ struct btrfs_inode {
 	unsigned ordered_data_close:1;
 	unsigned dummy_inode:1;
 
+	/*
+	 * always compress this one file
+	 */
+	unsigned force_compress:1;
+
 	struct inode vfs_inode;
 };
 

fs/btrfs/ctree.h

@@ -1184,7 +1184,6 @@ struct btrfs_root {
 #define BTRFS_INODE_NOATIME		(1 << 9)
 #define BTRFS_INODE_DIRSYNC		(1 << 10)
 
-
 /* some macros to generate set/get funcs for the struct fields.  This
  * assumes there is a lefoo_to_cpu for every type, so lets make a simple
  * one for u8:

fs/btrfs/inode.c

@@ -379,7 +379,8 @@ static noinline int compress_file_range(struct inode *inode,
 	 * change at any time if we discover bad compression ratios.
 	 */
 	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS) &&
-	    btrfs_test_opt(root, COMPRESS)) {
+	    (btrfs_test_opt(root, COMPRESS) ||
+	     (BTRFS_I(inode)->force_compress))) {
 		WARN_ON(pages);
 		pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS);
 
@@ -483,8 +484,10 @@ static noinline int compress_file_range(struct inode *inode,
 		nr_pages_ret = 0;
 
 		/* flag the file so we don't compress in the future */
-		if (!btrfs_test_opt(root, FORCE_COMPRESS))
+		if (!btrfs_test_opt(root, FORCE_COMPRESS) &&
+		    !(BTRFS_I(inode)->force_compress)) {
 			BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS;
+		}
 	}
 	if (will_compress) {
 		*num_added += 1;
@@ -1211,7 +1214,8 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page,
 	else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC)
 		ret = run_delalloc_nocow(inode, locked_page, start, end,
 					 page_started, 0, nr_written);
-	else if (!btrfs_test_opt(root, COMPRESS))
+	else if (!btrfs_test_opt(root, COMPRESS) &&
+		 !(BTRFS_I(inode)->force_compress))
 		ret = cow_file_range(inode, locked_page, start, end,
 				      page_started, nr_written, 1);
 	else
@@ -3639,6 +3643,7 @@ static noinline void init_btrfs_i(struct inode *inode)
 	bi->index_cnt = (u64)-1;
 	bi->last_unlink_trans = 0;
 	bi->ordered_data_close = 0;
+	bi->force_compress = 0;
 	extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
 	extent_io_tree_init(&BTRFS_I(inode)->io_tree,
 			     inode->i_mapping, GFP_NOFS);

fs/btrfs/ioctl.c

@@ -476,13 +476,18 @@ static noinline int btrfs_mksubvol(struct path *parent,
 }
 
 static int should_defrag_range(struct inode *inode, u64 start, u64 len,
-			       u64 *last_len, u64 *skip, u64 *defrag_end)
+			       int thresh, u64 *last_len, u64 *skip,
+			       u64 *defrag_end)
 {
 	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
 	struct extent_map *em = NULL;
 	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 	int ret = 1;
 
+
+	if (thresh == 0)
+		thresh = 256 * 1024;
+
 	/*
 	 * make sure that once we start defragging and extent, we keep on
 	 * defragging it
@@ -517,8 +522,7 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len,
 	/*
 	 * we hit a real extent, if it is big don't bother defragging it again
 	 */
-	if ((*last_len == 0 || *last_len >= 256 * 1024) &&
-	    em->len >= 256 * 1024)
+	if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh)
 		ret = 0;
 
 	/*
@@ -542,7 +546,8 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len,
 	return ret;
 }
 
-static int btrfs_defrag_file(struct file *file)
+static int btrfs_defrag_file(struct file *file,
+			     struct btrfs_ioctl_defrag_range_args *range)
 {
 	struct inode *inode = fdentry(file)->d_inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -563,11 +568,19 @@ static int btrfs_defrag_file(struct file *file)
 	if (inode->i_size == 0)
 		return 0;
 
-	last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT;
-	i = 0;
+	if (range->start + range->len > range->start) {
+		last_index = min_t(u64, inode->i_size - 1,
+			 range->start + range->len - 1) >> PAGE_CACHE_SHIFT;
+	} else {
+		last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT;
+	}
+
+	i = range->start >> PAGE_CACHE_SHIFT;
 	while (i <= last_index) {
 		if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT,
-					PAGE_CACHE_SIZE, &last_len, &skip,
+					PAGE_CACHE_SIZE,
+					range->extent_thresh,
+					&last_len, &skip,
 					&defrag_end)) {
 			unsigned long next;
 			/*
@@ -585,6 +598,8 @@ static int btrfs_defrag_file(struct file *file)
 		}
 		total_read++;
 		mutex_lock(&inode->i_mutex);
+		if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)
+			BTRFS_I(inode)->force_compress = 1;
 
 		ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE);
 		if (ret) {
@@ -673,6 +688,28 @@ static int btrfs_defrag_file(struct file *file)
 		i++;
 	}
 
+	if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO))
+		filemap_flush(inode->i_mapping);
+
+	if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
+		/* the filemap_flush will queue IO into the worker threads, but
+		 * we have to make sure the IO is actually started and that
+		 * ordered extents get created before we return
+		 */
+		atomic_inc(&root->fs_info->async_submit_draining);
+		while (atomic_read(&root->fs_info->nr_async_submits) ||
+		      atomic_read(&root->fs_info->async_delalloc_pages)) {
+			wait_event(root->fs_info->async_submit_wait,
+			   (atomic_read(&root->fs_info->nr_async_submits) == 0 &&
+			    atomic_read(&root->fs_info->async_delalloc_pages) == 0));
+		}
+		atomic_dec(&root->fs_info->async_submit_draining);
+
+		mutex_lock(&inode->i_mutex);
+		BTRFS_I(inode)->force_compress = 0;
+		mutex_unlock(&inode->i_mutex);
+	}
+
 	return 0;
 
 err_reservations:
@@ -1284,10 +1321,11 @@ static noinline int btrfs_ioctl_snap_destroy(struct file *file,
 	return err;
 }
 
-static int btrfs_ioctl_defrag(struct file *file)
+static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
 {
 	struct inode *inode = fdentry(file)->d_inode;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct btrfs_ioctl_defrag_range_args *range;
 	int ret;
 
 	ret = mnt_want_write(file->f_path.mnt);
@@ -1308,7 +1346,30 @@ static int btrfs_ioctl_defrag(struct file *file)
 			ret = -EINVAL;
 			goto out;
 		}
-		btrfs_defrag_file(file);
+
+		range = kzalloc(sizeof(*range), GFP_KERNEL);
+		if (!range) {
+			ret = -ENOMEM;
+			goto out;
+		}
+
+		if (argp) {
+			if (copy_from_user(range, argp,
+					   sizeof(*range))) {
+				ret = -EFAULT;
+				kfree(range);
+			}
+			/* compression requires us to start the IO */
+			if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
+				range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
+				range->extent_thresh = (u32)-1;
+			}
+		} else {
+			/* the rest are all set to zero by kzalloc */
+			range->len = (u64)-1;
+		}
+		btrfs_defrag_file(file, range);
+		kfree(range);
 		break;
 	}
 out:
@@ -1831,7 +1892,9 @@ long btrfs_ioctl(struct file *file, unsigned int
 	case BTRFS_IOC_DEFAULT_SUBVOL:
 		return btrfs_ioctl_default_subvol(file, argp);
 	case BTRFS_IOC_DEFRAG:
-		return btrfs_ioctl_defrag(file);
+		return btrfs_ioctl_defrag(file, NULL);
+	case BTRFS_IOC_DEFRAG_RANGE:
+		return btrfs_ioctl_defrag(file, argp);
 	case BTRFS_IOC_RESIZE:
 		return btrfs_ioctl_resize(root, argp);
 	case BTRFS_IOC_ADD_DEV:

fs/btrfs/ioctl.h

@@ -99,6 +99,35 @@ struct btrfs_ioctl_clone_range_args {
   __u64 dest_offset;
 };
 
+/* flags for the defrag range ioctl */
+#define BTRFS_DEFRAG_RANGE_COMPRESS 1
+#define BTRFS_DEFRAG_RANGE_START_IO 2
+
+struct btrfs_ioctl_defrag_range_args {
+	/* start of the defrag operation */
+	__u64 start;
+
+	/* number of bytes to defrag, use (u64)-1 to say all */
+	__u64 len;
+
+	/*
+	 * flags for the operation, which can include turning
+	 * on compression for this one defrag
+	 */
+	__u64 flags;
+
+	/*
+	 * any extent bigger than this will be considered
+	 * already defragged.  Use 0 to take the kernel default
+	 * Use 1 to say every single extent must be rewritten
+	 */
+	__u32 extent_thresh;
+
+	/* spare for later */
+	__u32 unused[5];
+};
+
+
 #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \
 				   struct btrfs_ioctl_vol_args)
 #define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \
@@ -130,6 +159,8 @@ struct btrfs_ioctl_clone_range_args {
 				   struct btrfs_ioctl_vol_args)
 #define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \
 				struct btrfs_ioctl_vol_args)
+#define BTRFS_IOC_DEFRAG_RANGE _IOW(BTRFS_IOCTL_MAGIC, 16, \
+				struct btrfs_ioctl_defrag_range_args)
 #define BTRFS_IOC_TREE_SEARCH _IOWR(BTRFS_IOCTL_MAGIC, 17, \
 				   struct btrfs_ioctl_search_args)
 #define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \