[thirdparty/linux.git] / fs / remap_range.c

// SPDX-License-Identifier: GPL-2.0-only
#include <linux/slab.h>
#include <linux/stat.h>
#include <linux/sched/xacct.h>
#include <linux/fcntl.h>
#include <linux/file.h>
#include <linux/uio.h>
#include <linux/fsnotify.h>
#include <linux/security.h>
#include <linux/export.h>
#include <linux/syscalls.h>
#include <linux/pagemap.h>
#include <linux/splice.h>
#include <linux/compat.h>
#include <linux/mount.h>
#include <linux/fs.h>
#include <linux/dax.h>
#include <linux/overflow.h>
#include "internal.h"

#include <linux/uaccess.h>
#include <asm/unistd.h>

/*
 * Performs necessary checks before doing a clone.
 *
 * Can adjust amount of bytes to clone via @req_count argument.
 * Returns appropriate error code that caller should return or
 * zero in case the clone should be allowed.
 */
static int generic_remap_checks(struct file *file_in, loff_t pos_in,
				struct file *file_out, loff_t pos_out,
				loff_t *req_count, unsigned int remap_flags)
{
	struct inode *inode_in = file_in->f_mapping->host;
	struct inode *inode_out = file_out->f_mapping->host;
	uint64_t count = *req_count;
	uint64_t bcount;
	loff_t size_in, size_out;
	loff_t bs = inode_out->i_sb->s_blocksize;
	int ret;

	/* The start of both ranges must be aligned to an fs block. */
	if (!IS_ALIGNED(pos_in, bs) || !IS_ALIGNED(pos_out, bs))
		return -EINVAL;

	/* Ensure offsets don't wrap. */
	if (pos_in + count < pos_in || pos_out + count < pos_out)
		return -EINVAL;

	size_in = i_size_read(inode_in);
	size_out = i_size_read(inode_out);

	/* Dedupe requires both ranges to be within EOF. */
	if ((remap_flags & REMAP_FILE_DEDUP) &&
	    (pos_in >= size_in || pos_in + count > size_in ||
	     pos_out >= size_out || pos_out + count > size_out))
		return -EINVAL;

	/* Ensure the infile range is within the infile. */
	if (pos_in >= size_in)
		return -EINVAL;
	count = min(count, size_in - (uint64_t)pos_in);

	ret = generic_write_check_limits(file_out, pos_out, &count);
	if (ret)
		return ret;

	/*
	 * If the user wanted us to link to the infile's EOF, round up to the
	 * next block boundary for this check.
	 *
	 * Otherwise, make sure the count is also block-aligned, having
	 * already confirmed the starting offsets' block alignment.
	 */
	if (pos_in + count == size_in &&
	    (!(remap_flags & REMAP_FILE_DEDUP) || pos_out + count == size_out)) {
		bcount = ALIGN(size_in, bs) - pos_in;
	} else {
		if (!IS_ALIGNED(count, bs))
			count = ALIGN_DOWN(count, bs);
		bcount = count;
	}

	/* Don't allow overlapped cloning within the same file. */
	if (inode_in == inode_out &&
	    pos_out + bcount > pos_in &&
	    pos_out < pos_in + bcount)
		return -EINVAL;

	/*
	 * We shortened the request but the caller can't deal with that, so
	 * bounce the request back to userspace.
	 */
	if (*req_count != count && !(remap_flags & REMAP_FILE_CAN_SHORTEN))
		return -EINVAL;

	*req_count = count;
	return 0;
}

static int remap_verify_area(struct file *file, loff_t pos, loff_t len,
			     bool write)
{
	loff_t tmp;

	if (unlikely(pos < 0 || len < 0))
		return -EINVAL;

	if (unlikely(check_add_overflow(pos, len, &tmp)))
		return -EINVAL;

	return security_file_permission(file, write ? MAY_WRITE : MAY_READ);
}

/*
 * Ensure that we don't remap a partial EOF block in the middle of something
 * else.  Assume that the offsets have already been checked for block
 * alignment.
 *
 * For clone we only link a partial EOF block above or at the destination file's
 * EOF.  For deduplication we accept a partial EOF block only if it ends at the
 * destination file's EOF (can not link it into the middle of a file).
 *
 * Shorten the request if possible.
 */
static int generic_remap_check_len(struct inode *inode_in,
				   struct inode *inode_out,
				   loff_t pos_out,
				   loff_t *len,
				   unsigned int remap_flags)
{
	u64 blkmask = i_blocksize(inode_in) - 1;
	loff_t new_len = *len;

	if ((*len & blkmask) == 0)
		return 0;

	if (pos_out + *len < i_size_read(inode_out))
		new_len &= ~blkmask;

	if (new_len == *len)
		return 0;

	if (remap_flags & REMAP_FILE_CAN_SHORTEN) {
		*len = new_len;
		return 0;
	}

	return (remap_flags & REMAP_FILE_DEDUP) ? -EBADE : -EINVAL;
}

/* Read a page's worth of file data into the page cache. */
static struct folio *vfs_dedupe_get_folio(struct file *file, loff_t pos)
{
	return read_mapping_folio(file->f_mapping, pos >> PAGE_SHIFT, file);
}

/*
 * Lock two folios, ensuring that we lock in offset order if the folios
 * are from the same file.
 */
static void vfs_lock_two_folios(struct folio *folio1, struct folio *folio2)
{
	/* Always lock in order of increasing index. */
	if (folio1->index > folio2->index)
		swap(folio1, folio2);

	folio_lock(folio1);
	if (folio1 != folio2)
		folio_lock(folio2);
}

/* Unlock two folios, being careful not to unlock the same folio twice. */
static void vfs_unlock_two_folios(struct folio *folio1, struct folio *folio2)
{
	folio_unlock(folio1);
	if (folio1 != folio2)
		folio_unlock(folio2);
}

/*
 * Compare extents of two files to see if they are the same.
 * Caller must have locked both inodes to prevent write races.
 */
static int vfs_dedupe_file_range_compare(struct file *src, loff_t srcoff,
					 struct file *dest, loff_t dstoff,
					 loff_t len, bool *is_same)
{
	bool same = true;
	int error = -EINVAL;

	while (len) {
		struct folio *src_folio, *dst_folio;
		void *src_addr, *dst_addr;
		loff_t cmp_len = min(PAGE_SIZE - offset_in_page(srcoff),
				     PAGE_SIZE - offset_in_page(dstoff));

		cmp_len = min(cmp_len, len);
		if (cmp_len <= 0)
			goto out_error;

		src_folio = vfs_dedupe_get_folio(src, srcoff);
		if (IS_ERR(src_folio)) {
			error = PTR_ERR(src_folio);
			goto out_error;
		}
		dst_folio = vfs_dedupe_get_folio(dest, dstoff);
		if (IS_ERR(dst_folio)) {
			error = PTR_ERR(dst_folio);
			folio_put(src_folio);
			goto out_error;
		}

		vfs_lock_two_folios(src_folio, dst_folio);

		/*
		 * Now that we've locked both folios, make sure they're still
		 * mapped to the file data we're interested in.  If not,
		 * someone is invalidating pages on us and we lose.
		 */
		if (!folio_test_uptodate(src_folio) || !folio_test_uptodate(dst_folio) ||
		    src_folio->mapping != src->f_mapping ||
		    dst_folio->mapping != dest->f_mapping) {
			same = false;
			goto unlock;
		}

		src_addr = kmap_local_folio(src_folio,
					offset_in_folio(src_folio, srcoff));
		dst_addr = kmap_local_folio(dst_folio,
					offset_in_folio(dst_folio, dstoff));

		flush_dcache_folio(src_folio);
		flush_dcache_folio(dst_folio);

		if (memcmp(src_addr, dst_addr, cmp_len))
			same = false;

		kunmap_local(dst_addr);
		kunmap_local(src_addr);
unlock:
		vfs_unlock_two_folios(src_folio, dst_folio);
		folio_put(dst_folio);
		folio_put(src_folio);

		if (!same)
			break;

		srcoff += cmp_len;
		dstoff += cmp_len;
		len -= cmp_len;
	}

	*is_same = same;
	return 0;

out_error:
	return error;
}

/*
 * 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.
 *
 * If there's an error, then the usual negative error code is returned.
 * Otherwise returns 0 with *len set to the request length.
 */
int
__generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
				struct file *file_out, loff_t pos_out,
				loff_t *len, unsigned int remap_flags,
				const struct iomap_ops *dax_read_ops)
{
	struct inode *inode_in = file_inode(file_in);
	struct inode *inode_out = file_inode(file_out);
	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;

	/* Zero length dedupe exits immediately; reflink goes to EOF. */
	if (*len == 0) {
		loff_t isize = i_size_read(inode_in);

		if ((remap_flags & REMAP_FILE_DEDUP) || pos_in == isize)
			return 0;
		if (pos_in > isize)
			return -EINVAL;
		*len = isize - pos_in;
		if (*len == 0)
			return 0;
	}

	/* Check that we don't violate system file offset limits. */
	ret = generic_remap_checks(file_in, pos_in, file_out, pos_out, len,
			remap_flags);
	if (ret || *len == 0)
		return ret;

	/* 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 (remap_flags & REMAP_FILE_DEDUP) {
		bool		is_same = false;

		if (!IS_DAX(inode_in))
			ret = vfs_dedupe_file_range_compare(file_in, pos_in,
					file_out, pos_out, *len, &is_same);
		else if (dax_read_ops)
			ret = dax_dedupe_file_range_compare(inode_in, pos_in,
					inode_out, pos_out, *len, &is_same,
					dax_read_ops);
		else
			return -EINVAL;
		if (ret)
			return ret;
		if (!is_same)
			return -EBADE;
	}

	ret = generic_remap_check_len(inode_in, inode_out, pos_out, len,
			remap_flags);
	if (ret || *len == 0)
		return ret;

	/* If can't alter the file contents, we're done. */
	if (!(remap_flags & REMAP_FILE_DEDUP))
		ret = file_modified(file_out);

	return ret;
}

int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
				  struct file *file_out, loff_t pos_out,
				  loff_t *len, unsigned int remap_flags)
{
	return __generic_remap_file_range_prep(file_in, pos_in, file_out,
					       pos_out, len, remap_flags, NULL);
}
EXPORT_SYMBOL(generic_remap_file_range_prep);

loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
			   struct file *file_out, loff_t pos_out,
			   loff_t len, unsigned int remap_flags)
{
	loff_t ret;

	WARN_ON_ONCE(remap_flags & REMAP_FILE_DEDUP);

	if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
		return -EXDEV;

	ret = generic_file_rw_checks(file_in, file_out);
	if (ret < 0)
		return ret;

	if (!file_in->f_op->remap_file_range)
		return -EOPNOTSUPP;

	ret = file_in->f_op->remap_file_range(file_in, pos_in,
			file_out, pos_out, len, remap_flags);
	if (ret < 0)
		return ret;

	fsnotify_access(file_in);
	fsnotify_modify(file_out);
	return ret;
}
EXPORT_SYMBOL(do_clone_file_range);

loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
			    struct file *file_out, loff_t pos_out,
			    loff_t len, unsigned int remap_flags)
{
	loff_t ret;

	ret = remap_verify_area(file_in, pos_in, len, false);
	if (ret)
		return ret;

	ret = remap_verify_area(file_out, pos_out, len, true);
	if (ret)
		return ret;

	file_start_write(file_out);
	ret = do_clone_file_range(file_in, pos_in, file_out, pos_out, len,
				  remap_flags);
	file_end_write(file_out);

	return ret;
}
EXPORT_SYMBOL(vfs_clone_file_range);

/* Check whether we are allowed to dedupe the destination file */
static bool may_dedupe_file(struct file *file)
{
	struct mnt_idmap *idmap = file_mnt_idmap(file);
	struct inode *inode = file_inode(file);

	if (capable(CAP_SYS_ADMIN))
		return true;
	if (file->f_mode & FMODE_WRITE)
		return true;
	if (vfsuid_eq_kuid(i_uid_into_vfsuid(idmap, inode), current_fsuid()))
		return true;
	if (!inode_permission(idmap, inode, MAY_WRITE))
		return true;
	return false;
}

loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
				 struct file *dst_file, loff_t dst_pos,
				 loff_t len, unsigned int remap_flags)
{
	loff_t ret;

	WARN_ON_ONCE(remap_flags & ~(REMAP_FILE_DEDUP |
				     REMAP_FILE_CAN_SHORTEN));

	/*
	 * This is redundant if called from vfs_dedupe_file_range(), but other
	 * callers need it and it's not performance sesitive...
	 */
	ret = remap_verify_area(src_file, src_pos, len, false);
	if (ret)
		return ret;

	ret = remap_verify_area(dst_file, dst_pos, len, true);
	if (ret)
		return ret;

	/*
	 * This needs to be called after remap_verify_area() because of
	 * sb_start_write() and before may_dedupe_file() because the mount's
	 * MAY_WRITE need to be checked with mnt_get_write_access_file() held.
	 */
	ret = mnt_want_write_file(dst_file);
	if (ret)
		return ret;

	ret = -EPERM;
	if (!may_dedupe_file(dst_file))
		goto out_drop_write;

	ret = -EXDEV;
	if (file_inode(src_file)->i_sb != file_inode(dst_file)->i_sb)
		goto out_drop_write;

	ret = -EISDIR;
	if (S_ISDIR(file_inode(dst_file)->i_mode))
		goto out_drop_write;

	ret = -EINVAL;
	if (!dst_file->f_op->remap_file_range)
		goto out_drop_write;

	if (len == 0) {
		ret = 0;
		goto out_drop_write;
	}

	ret = dst_file->f_op->remap_file_range(src_file, src_pos, dst_file,
			dst_pos, len, remap_flags | REMAP_FILE_DEDUP);
out_drop_write:
	mnt_drop_write_file(dst_file);

	return ret;
}
EXPORT_SYMBOL(vfs_dedupe_file_range_one);

int vfs_dedupe_file_range(struct file *file, struct file_dedupe_range *same)
{
	struct file_dedupe_range_info *info;
	struct inode *src = file_inode(file);
	u64 off;
	u64 len;
	int i;
	int ret;
	u16 count = same->dest_count;
	loff_t deduped;

	if (!(file->f_mode & FMODE_READ))
		return -EINVAL;

	if (same->reserved1 || same->reserved2)
		return -EINVAL;

	off = same->src_offset;
	len = same->src_length;

	if (S_ISDIR(src->i_mode))
		return -EISDIR;

	if (!S_ISREG(src->i_mode))
		return -EINVAL;

	if (!file->f_op->remap_file_range)
		return -EOPNOTSUPP;

	ret = remap_verify_area(file, off, len, false);
	if (ret < 0)
		return ret;
	ret = 0;

	if (off + len > i_size_read(src))
		return -EINVAL;

	/* Arbitrary 1G limit on a single dedupe request, can be raised. */
	len = min_t(u64, len, 1 << 30);

	/* pre-format output fields to sane values */
	for (i = 0; i < count; i++) {
		same->info[i].bytes_deduped = 0ULL;
		same->info[i].status = FILE_DEDUPE_RANGE_SAME;
	}

	for (i = 0, info = same->info; i < count; i++, info++) {
		struct fd dst_fd = fdget(info->dest_fd);
		struct file *dst_file = dst_fd.file;

		if (!dst_file) {
			info->status = -EBADF;
			goto next_loop;
		}

		if (info->reserved) {
			info->status = -EINVAL;
			goto next_fdput;
		}

		deduped = vfs_dedupe_file_range_one(file, off, dst_file,
						    info->dest_offset, len,
						    REMAP_FILE_CAN_SHORTEN);
		if (deduped == -EBADE)
			info->status = FILE_DEDUPE_RANGE_DIFFERS;
		else if (deduped < 0)
			info->status = deduped;
		else
			info->bytes_deduped = len;

next_fdput:
		fdput(dst_fd);
next_loop:
		if (fatal_signal_pending(current))
			break;
	}
	return ret;
}
EXPORT_SYMBOL(vfs_dedupe_file_range);
Commit	Line	Data
02e83f46 DW	1	// SPDX-License-Identifier: GPL-2.0-only
	2	#include <linux/slab.h>
	3	#include <linux/stat.h>
	4	#include <linux/sched/xacct.h>
	5	#include <linux/fcntl.h>
	6	#include <linux/file.h>
	7	#include <linux/uio.h>
	8	#include <linux/fsnotify.h>
	9	#include <linux/security.h>
	10	#include <linux/export.h>
	11	#include <linux/syscalls.h>
	12	#include <linux/pagemap.h>
	13	#include <linux/splice.h>
	14	#include <linux/compat.h>
	15	#include <linux/mount.h>
	16	#include <linux/fs.h>
6f7db389	17	#include <linux/dax.h>
b7a9a503	18	#include <linux/overflow.h>
02e83f46 DW	19	#include "internal.h"
	20
	21	#include <linux/uaccess.h>
	22	#include <asm/unistd.h>
	23
	24	/*
	25	* Performs necessary checks before doing a clone.
	26	*
	27	* Can adjust amount of bytes to clone via @req_count argument.
	28	* Returns appropriate error code that caller should return or
	29	* zero in case the clone should be allowed.
	30	*/
1b2c54d6 DW	31	static int generic_remap_checks(struct file *file_in, loff_t pos_in,
	32	struct file *file_out, loff_t pos_out,
	33	loff_t *req_count, unsigned int remap_flags)
02e83f46 DW	34	{
	35	struct inode *inode_in = file_in->f_mapping->host;
	36	struct inode *inode_out = file_out->f_mapping->host;
	37	uint64_t count = *req_count;
	38	uint64_t bcount;
	39	loff_t size_in, size_out;
	40	loff_t bs = inode_out->i_sb->s_blocksize;
	41	int ret;
	42
	43	/* The start of both ranges must be aligned to an fs block. */
	44	if (!IS_ALIGNED(pos_in, bs) \|\| !IS_ALIGNED(pos_out, bs))
	45	return -EINVAL;
	46
	47	/* Ensure offsets don't wrap. */
	48	if (pos_in + count < pos_in \|\| pos_out + count < pos_out)
	49	return -EINVAL;
	50
	51	size_in = i_size_read(inode_in);
	52	size_out = i_size_read(inode_out);
	53
	54	/* Dedupe requires both ranges to be within EOF. */
	55	if ((remap_flags & REMAP_FILE_DEDUP) &&
	56	(pos_in >= size_in \|\| pos_in + count > size_in \|\|
	57	pos_out >= size_out \|\| pos_out + count > size_out))
	58	return -EINVAL;
	59
	60	/* Ensure the infile range is within the infile. */
	61	if (pos_in >= size_in)
	62	return -EINVAL;
	63	count = min(count, size_in - (uint64_t)pos_in);
	64
	65	ret = generic_write_check_limits(file_out, pos_out, &count);
	66	if (ret)
	67	return ret;
	68
	69	/*
	70	* If the user wanted us to link to the infile's EOF, round up to the
	71	* next block boundary for this check.
	72	*
	73	* Otherwise, make sure the count is also block-aligned, having
	74	* already confirmed the starting offsets' block alignment.
	75	*/
5750676b DC	76	if (pos_in + count == size_in &&
5750676b DC	77	(!(remap_flags & REMAP_FILE_DEDUP) \|\| pos_out + count == size_out)) {
02e83f46 DW	78	bcount = ALIGN(size_in, bs) - pos_in;
	79	} else {
	80	if (!IS_ALIGNED(count, bs))
	81	count = ALIGN_DOWN(count, bs);
	82	bcount = count;
	83	}
	84
	85	/* Don't allow overlapped cloning within the same file. */
	86	if (inode_in == inode_out &&
	87	pos_out + bcount > pos_in &&
	88	pos_out < pos_in + bcount)
	89	return -EINVAL;
	90
	91	/*
	92	* We shortened the request but the caller can't deal with that, so
	93	* bounce the request back to userspace.
	94	*/
	95	if (*req_count != count && !(remap_flags & REMAP_FILE_CAN_SHORTEN))
	96	return -EINVAL;
	97
	98	*req_count = count;
	99	return 0;
	100	}
1b2c54d6 DW	101
	102	static int remap_verify_area(struct file *file, loff_t pos, loff_t len,
	103	bool write)
	104	{
b7a9a503 DS	105	loff_t tmp;
b7a9a503 DS	106
1b2c54d6 DW	107	if (unlikely(pos < 0 \|\| len < 0))
	108	return -EINVAL;
	109
b7a9a503	110	if (unlikely(check_add_overflow(pos, len, &tmp)))
1b2c54d6 DW	111	return -EINVAL;
1b2c54d6 DW	112
1b2c54d6 DW	113	return security_file_permission(file, write ? MAY_WRITE : MAY_READ);
	114	}
	115
	116	/*
	117	* Ensure that we don't remap a partial EOF block in the middle of something
	118	* else. Assume that the offsets have already been checked for block
	119	* alignment.
	120	*
	121	* For clone we only link a partial EOF block above or at the destination file's
	122	* EOF. For deduplication we accept a partial EOF block only if it ends at the
	123	* destination file's EOF (can not link it into the middle of a file).
	124	*
	125	* Shorten the request if possible.
	126	*/
	127	static int generic_remap_check_len(struct inode *inode_in,
	128	struct inode *inode_out,
	129	loff_t pos_out,
	130	loff_t *len,
	131	unsigned int remap_flags)
	132	{
	133	u64 blkmask = i_blocksize(inode_in) - 1;
	134	loff_t new_len = *len;
	135
	136	if ((*len & blkmask) == 0)
	137	return 0;
	138
	139	if (pos_out + *len < i_size_read(inode_out))
	140	new_len &= ~blkmask;
	141
	142	if (new_len == *len)
	143	return 0;
	144
	145	if (remap_flags & REMAP_FILE_CAN_SHORTEN) {
	146	*len = new_len;
	147	return 0;
	148	}
	149
	150	return (remap_flags & REMAP_FILE_DEDUP) ? -EBADE : -EINVAL;
	151	}
	152
	153	/* Read a page's worth of file data into the page cache. */
4495a96c	154	static struct folio vfs_dedupe_get_folio(struct file file, loff_t pos)
1b2c54d6	155	{
da028b6b	156	return read_mapping_folio(file->f_mapping, pos >> PAGE_SHIFT, file);
1b2c54d6 DW	157	}
	158
	159	/*
338f379c MWO	160	* Lock two folios, ensuring that we lock in offset order if the folios
338f379c MWO	161	* are from the same file.
1b2c54d6	162	*/
338f379c	163	static void vfs_lock_two_folios(struct folio folio1, struct folio folio2)
1b2c54d6 DW	164	{
1b2c54d6 DW	165	/* Always lock in order of increasing index. */
338f379c MWO	166	if (folio1->index > folio2->index)
338f379c MWO	167	swap(folio1, folio2);
1b2c54d6	168
338f379c MWO	169	folio_lock(folio1);
	170	if (folio1 != folio2)
	171	folio_lock(folio2);
1b2c54d6 DW	172	}
1b2c54d6 DW	173
338f379c MWO	174	/* Unlock two folios, being careful not to unlock the same folio twice. */
338f379c MWO	175	static void vfs_unlock_two_folios(struct folio folio1, struct folio folio2)
1b2c54d6	176	{
338f379c MWO	177	folio_unlock(folio1);
	178	if (folio1 != folio2)
	179	folio_unlock(folio2);
1b2c54d6 DW	180	}
	181
	182	/*
	183	* Compare extents of two files to see if they are the same.
	184	* Caller must have locked both inodes to prevent write races.
	185	*/
4495a96c MWO	186	static int vfs_dedupe_file_range_compare(struct file *src, loff_t srcoff,
4495a96c MWO	187	struct file *dest, loff_t dstoff,
1b2c54d6 DW	188	loff_t len, bool *is_same)
1b2c54d6 DW	189	{
338f379c MWO	190	bool same = true;
	191	int error = -EINVAL;
	192
1b2c54d6	193	while (len) {
338f379c MWO	194	struct folio src_folio, dst_folio;
	195	void src_addr, dst_addr;
	196	loff_t cmp_len = min(PAGE_SIZE - offset_in_page(srcoff),
	197	PAGE_SIZE - offset_in_page(dstoff));
	198
1b2c54d6 DW	199	cmp_len = min(cmp_len, len);
	200	if (cmp_len <= 0)
	201	goto out_error;
	202
338f379c MWO	203	src_folio = vfs_dedupe_get_folio(src, srcoff);
	204	if (IS_ERR(src_folio)) {
	205	error = PTR_ERR(src_folio);
1b2c54d6 DW	206	goto out_error;
1b2c54d6 DW	207	}
338f379c MWO	208	dst_folio = vfs_dedupe_get_folio(dest, dstoff);
	209	if (IS_ERR(dst_folio)) {
	210	error = PTR_ERR(dst_folio);
	211	folio_put(src_folio);
1b2c54d6 DW	212	goto out_error;
	213	}
	214
338f379c	215	vfs_lock_two_folios(src_folio, dst_folio);
1b2c54d6 DW	216
1b2c54d6 DW	217	/*
338f379c	218	* Now that we've locked both folios, make sure they're still
1b2c54d6 DW	219	* mapped to the file data we're interested in. If not,
	220	* someone is invalidating pages on us and we lose.
	221	*/
338f379c	222	if (!folio_test_uptodate(src_folio) \|\| !folio_test_uptodate(dst_folio) \|\|
4495a96c MWO	223	src_folio->mapping != src->f_mapping \|\|
4495a96c MWO	224	dst_folio->mapping != dest->f_mapping) {
1b2c54d6 DW	225	same = false;
	226	goto unlock;
	227	}
	228
338f379c MWO	229	src_addr = kmap_local_folio(src_folio,
	230	offset_in_folio(src_folio, srcoff));
	231	dst_addr = kmap_local_folio(dst_folio,
	232	offset_in_folio(dst_folio, dstoff));
1b2c54d6	233
338f379c MWO	234	flush_dcache_folio(src_folio);
338f379c MWO	235	flush_dcache_folio(dst_folio);
1b2c54d6	236
338f379c	237	if (memcmp(src_addr, dst_addr, cmp_len))
1b2c54d6 DW	238	same = false;
1b2c54d6 DW	239
338f379c MWO	240	kunmap_local(dst_addr);
338f379c MWO	241	kunmap_local(src_addr);
1b2c54d6	242	unlock:
338f379c MWO	243	vfs_unlock_two_folios(src_folio, dst_folio);
	244	folio_put(dst_folio);
	245	folio_put(src_folio);
1b2c54d6 DW	246
	247	if (!same)
	248	break;
	249
	250	srcoff += cmp_len;
338f379c	251	dstoff += cmp_len;
1b2c54d6 DW	252	len -= cmp_len;
	253	}
	254
	255	*is_same = same;
	256	return 0;
	257
	258	out_error:
	259	return error;
	260	}
	261
	262	/*
	263	* Check that the two inodes are eligible for cloning, the ranges make
	264	* sense, and then flush all dirty data. Caller must ensure that the
	265	* inodes have been locked against any other modifications.
	266	*
	267	* If there's an error, then the usual negative error code is returned.
	268	* Otherwise returns 0 with *len set to the request length.
	269	*/
6f7db389 SR	270	int
	271	__generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
	272	struct file *file_out, loff_t pos_out,
	273	loff_t *len, unsigned int remap_flags,
	274	const struct iomap_ops *dax_read_ops)
1b2c54d6 DW	275	{
	276	struct inode *inode_in = file_inode(file_in);
	277	struct inode *inode_out = file_inode(file_out);
	278	bool same_inode = (inode_in == inode_out);
	279	int ret;
	280
	281	/* Don't touch certain kinds of inodes */
	282	if (IS_IMMUTABLE(inode_out))
	283	return -EPERM;
	284
	285	if (IS_SWAPFILE(inode_in) \|\| IS_SWAPFILE(inode_out))
	286	return -ETXTBSY;
	287
	288	/* Don't reflink dirs, pipes, sockets... */
	289	if (S_ISDIR(inode_in->i_mode) \|\| S_ISDIR(inode_out->i_mode))
	290	return -EISDIR;
	291	if (!S_ISREG(inode_in->i_mode) \|\| !S_ISREG(inode_out->i_mode))
	292	return -EINVAL;
	293
	294	/* Zero length dedupe exits immediately; reflink goes to EOF. */
	295	if (*len == 0) {
	296	loff_t isize = i_size_read(inode_in);
	297
	298	if ((remap_flags & REMAP_FILE_DEDUP) \|\| pos_in == isize)
	299	return 0;
	300	if (pos_in > isize)
	301	return -EINVAL;
	302	*len = isize - pos_in;
	303	if (*len == 0)
	304	return 0;
	305	}
	306
	307	/* Check that we don't violate system file offset limits. */
	308	ret = generic_remap_checks(file_in, pos_in, file_out, pos_out, len,
	309	remap_flags);
a79168a0	310	if (ret \|\| *len == 0)
1b2c54d6 DW	311	return ret;
	312
	313	/* Wait for the completion of any pending IOs on both files */
	314	inode_dio_wait(inode_in);
	315	if (!same_inode)
	316	inode_dio_wait(inode_out);
	317
	318	ret = filemap_write_and_wait_range(inode_in->i_mapping,
	319	pos_in, pos_in + *len - 1);
	320	if (ret)
	321	return ret;
	322
	323	ret = filemap_write_and_wait_range(inode_out->i_mapping,
	324	pos_out, pos_out + *len - 1);
	325	if (ret)
	326	return ret;
	327
	328	/*
	329	* Check that the extents are the same.
	330	*/
	331	if (remap_flags & REMAP_FILE_DEDUP) {
	332	bool is_same = false;
	333
6f7db389 SR	334	if (!IS_DAX(inode_in))
	335	ret = vfs_dedupe_file_range_compare(file_in, pos_in,
	336	file_out, pos_out, *len, &is_same);
	337	else if (dax_read_ops)
	338	ret = dax_dedupe_file_range_compare(inode_in, pos_in,
	339	inode_out, pos_out, *len, &is_same,
	340	dax_read_ops);
	341	else
	342	return -EINVAL;
1b2c54d6 DW	343	if (ret)
	344	return ret;
	345	if (!is_same)
	346	return -EBADE;
	347	}
	348
	349	ret = generic_remap_check_len(inode_in, inode_out, pos_out, len,
	350	remap_flags);
a79168a0	351	if (ret \|\| *len == 0)
1b2c54d6 DW	352	return ret;
	353
	354	/* If can't alter the file contents, we're done. */
	355	if (!(remap_flags & REMAP_FILE_DEDUP))
	356	ret = file_modified(file_out);
	357
	358	return ret;
	359	}
6f7db389 SR	360
	361	int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
	362	struct file *file_out, loff_t pos_out,
	363	loff_t *len, unsigned int remap_flags)
	364	{
	365	return __generic_remap_file_range_prep(file_in, pos_in, file_out,
	366	pos_out, len, remap_flags, NULL);
	367	}
1b2c54d6 DW	368	EXPORT_SYMBOL(generic_remap_file_range_prep);
	369
	370	loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
	371	struct file *file_out, loff_t pos_out,
	372	loff_t len, unsigned int remap_flags)
	373	{
	374	loff_t ret;
	375
	376	WARN_ON_ONCE(remap_flags & REMAP_FILE_DEDUP);
	377
1b2c54d6 DW	378	if (file_inode(file_in)->i_sb != file_inode(file_out)->i_sb)
	379	return -EXDEV;
	380
	381	ret = generic_file_rw_checks(file_in, file_out);
	382	if (ret < 0)
	383	return ret;
	384
	385	if (!file_in->f_op->remap_file_range)
	386	return -EOPNOTSUPP;
	387
1b2c54d6 DW	388	ret = file_in->f_op->remap_file_range(file_in, pos_in,
	389	file_out, pos_out, len, remap_flags);
	390	if (ret < 0)
	391	return ret;
	392
	393	fsnotify_access(file_in);
	394	fsnotify_modify(file_out);
	395	return ret;
	396	}
	397	EXPORT_SYMBOL(do_clone_file_range);
	398
	399	loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
	400	struct file *file_out, loff_t pos_out,
	401	loff_t len, unsigned int remap_flags)
	402	{
	403	loff_t ret;
	404
dfad3705 AG	405	ret = remap_verify_area(file_in, pos_in, len, false);
	406	if (ret)
	407	return ret;
	408
	409	ret = remap_verify_area(file_out, pos_out, len, true);
	410	if (ret)
	411	return ret;
	412
1b2c54d6 DW	413	file_start_write(file_out);
	414	ret = do_clone_file_range(file_in, pos_in, file_out, pos_out, len,
	415	remap_flags);
	416	file_end_write(file_out);
	417
	418	return ret;
	419	}
	420	EXPORT_SYMBOL(vfs_clone_file_range);
	421
	422	/* Check whether we are allowed to dedupe the destination file */
0b5263d1	423	static bool may_dedupe_file(struct file *file)
1b2c54d6	424	{
4609e1f1	425	struct mnt_idmap *idmap = file_mnt_idmap(file);
0f5d220b CB	426	struct inode *inode = file_inode(file);
0f5d220b CB	427
1b2c54d6 DW	428	if (capable(CAP_SYS_ADMIN))
	429	return true;
	430	if (file->f_mode & FMODE_WRITE)
	431	return true;
e67fe633	432	if (vfsuid_eq_kuid(i_uid_into_vfsuid(idmap, inode), current_fsuid()))
1b2c54d6	433	return true;
4609e1f1	434	if (!inode_permission(idmap, inode, MAY_WRITE))
1b2c54d6 DW	435	return true;
	436	return false;
	437	}
	438
	439	loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
	440	struct file *dst_file, loff_t dst_pos,
	441	loff_t len, unsigned int remap_flags)
	442	{
	443	loff_t ret;
	444
	445	WARN_ON_ONCE(remap_flags & ~(REMAP_FILE_DEDUP \|
	446	REMAP_FILE_CAN_SHORTEN));
	447
3078d85c MS	448	/*
	449	* This is redundant if called from vfs_dedupe_file_range(), but other
	450	* callers need it and it's not performance sesitive...
	451	*/
	452	ret = remap_verify_area(src_file, src_pos, len, false);
	453	if (ret)
0b5263d1	454	return ret;
3078d85c	455
1b2c54d6	456	ret = remap_verify_area(dst_file, dst_pos, len, true);
3078d85c	457	if (ret)
0b5263d1 AG	458	return ret;
	459
	460	/*
	461	* This needs to be called after remap_verify_area() because of
	462	* sb_start_write() and before may_dedupe_file() because the mount's
	463	* MAY_WRITE need to be checked with mnt_get_write_access_file() held.
	464	*/
	465	ret = mnt_want_write_file(dst_file);
	466	if (ret)
	467	return ret;
1b2c54d6 DW	468
1b2c54d6 DW	469	ret = -EPERM;
0b5263d1	470	if (!may_dedupe_file(dst_file))
1b2c54d6 DW	471	goto out_drop_write;
	472
	473	ret = -EXDEV;
9f5710bb	474	if (file_inode(src_file)->i_sb != file_inode(dst_file)->i_sb)
1b2c54d6 DW	475	goto out_drop_write;
	476
	477	ret = -EISDIR;
	478	if (S_ISDIR(file_inode(dst_file)->i_mode))
	479	goto out_drop_write;
	480
	481	ret = -EINVAL;
	482	if (!dst_file->f_op->remap_file_range)
	483	goto out_drop_write;
	484
	485	if (len == 0) {
	486	ret = 0;
	487	goto out_drop_write;
	488	}
	489
	490	ret = dst_file->f_op->remap_file_range(src_file, src_pos, dst_file,
	491	dst_pos, len, remap_flags \| REMAP_FILE_DEDUP);
	492	out_drop_write:
	493	mnt_drop_write_file(dst_file);
	494
	495	return ret;
	496	}
	497	EXPORT_SYMBOL(vfs_dedupe_file_range_one);
	498
	499	int vfs_dedupe_file_range(struct file file, struct file_dedupe_range same)
	500	{
	501	struct file_dedupe_range_info *info;
	502	struct inode *src = file_inode(file);
	503	u64 off;
	504	u64 len;
	505	int i;
	506	int ret;
	507	u16 count = same->dest_count;
	508	loff_t deduped;
	509
	510	if (!(file->f_mode & FMODE_READ))
	511	return -EINVAL;
	512
	513	if (same->reserved1 \|\| same->reserved2)
	514	return -EINVAL;
	515
	516	off = same->src_offset;
	517	len = same->src_length;
	518
	519	if (S_ISDIR(src->i_mode))
	520	return -EISDIR;
	521
	522	if (!S_ISREG(src->i_mode))
	523	return -EINVAL;
	524
	525	if (!file->f_op->remap_file_range)
	526	return -EOPNOTSUPP;
	527
	528	ret = remap_verify_area(file, off, len, false);
	529	if (ret < 0)
	530	return ret;
	531	ret = 0;
	532
	533	if (off + len > i_size_read(src))
	534	return -EINVAL;
	535
	536	/* Arbitrary 1G limit on a single dedupe request, can be raised. */
	537	len = min_t(u64, len, 1 << 30);
	538
539	/* pre-format output fields to sane values */
540	for (i = 0; i < count; i++) {
541	same->info[i].bytes_deduped = 0ULL;
542	same->info[i].status = FILE_DEDUPE_RANGE_SAME;
543	}
544
545	for (i = 0, info = same->info; i < count; i++, info++) {
546	struct fd dst_fd = fdget(info->dest_fd);
547	struct file *dst_file = dst_fd.file;
548
549	if (!dst_file) {
550	info->status = -EBADF;
551	goto next_loop;
552	}
553
554	if (info->reserved) {
555	info->status = -EINVAL;
556	goto next_fdput;
557	}
558
559	deduped = vfs_dedupe_file_range_one(file, off, dst_file,
560	info->dest_offset, len,
561	REMAP_FILE_CAN_SHORTEN);
562	if (deduped == -EBADE)
563	info->status = FILE_DEDUPE_RANGE_DIFFERS;
564	else if (deduped < 0)
565	info->status = deduped;
566	else
b926f2ad	567	info->bytes_deduped = len;
1b2c54d6 DW	568
	569	next_fdput:
	570	fdput(dst_fd);
	571	next_loop:
	572	if (fatal_signal_pending(current))
	573	break;
	574	}
	575	return ret;
	576	}
	577	EXPORT_SYMBOL(vfs_dedupe_file_range);