[linux-block.git] / fs / overlayfs / file.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 * Copyright (C) 2017 Red Hat, Inc.
 */

#include <linux/cred.h>
#include <linux/file.h>
#include <linux/mount.h>
#include <linux/xattr.h>
#include <linux/uio.h>
#include <linux/uaccess.h>
#include <linux/splice.h>
#include <linux/security.h>
#include <linux/mm.h>
#include <linux/fs.h>
#include "overlayfs.h"

struct ovl_aio_req {
	struct kiocb iocb;
	struct kiocb *orig_iocb;
	struct fd fd;
};

static struct kmem_cache *ovl_aio_request_cachep;

static char ovl_whatisit(struct inode *inode, struct inode *realinode)
{
	if (realinode != ovl_inode_upper(inode))
		return 'l';
	if (ovl_has_upperdata(inode))
		return 'u';
	else
		return 'm';
}

/* No atime modificaton nor notify on underlying */
#define OVL_OPEN_FLAGS (O_NOATIME | FMODE_NONOTIFY)

static struct file *ovl_open_realfile(const struct file *file,
				      struct inode *realinode)
{
	struct inode *inode = file_inode(file);
	struct file *realfile;
	const struct cred *old_cred;
	int flags = file->f_flags | OVL_OPEN_FLAGS;
	int acc_mode = ACC_MODE(flags);
	int err;

	if (flags & O_APPEND)
		acc_mode |= MAY_APPEND;

	old_cred = ovl_override_creds(inode->i_sb);
	err = inode_permission(&init_user_ns, realinode, MAY_OPEN | acc_mode);
	if (err) {
		realfile = ERR_PTR(err);
	} else {
		if (!inode_owner_or_capable(&init_user_ns, realinode))
			flags &= ~O_NOATIME;

		realfile = open_with_fake_path(&file->f_path, flags, realinode,
					       current_cred());
	}
	revert_creds(old_cred);

	pr_debug("open(%p[%pD2/%c], 0%o) -> (%p, 0%o)\n",
		 file, file, ovl_whatisit(inode, realinode), file->f_flags,
		 realfile, IS_ERR(realfile) ? 0 : realfile->f_flags);

	return realfile;
}

#define OVL_SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT)

static int ovl_change_flags(struct file *file, unsigned int flags)
{
	struct inode *inode = file_inode(file);
	int err;

	flags &= OVL_SETFL_MASK;

	if (((flags ^ file->f_flags) & O_APPEND) && IS_APPEND(inode))
		return -EPERM;

	if (flags & O_DIRECT) {
		if (!file->f_mapping->a_ops ||
		    !file->f_mapping->a_ops->direct_IO)
			return -EINVAL;
	}

	if (file->f_op->check_flags) {
		err = file->f_op->check_flags(flags);
		if (err)
			return err;
	}

	spin_lock(&file->f_lock);
	file->f_flags = (file->f_flags & ~OVL_SETFL_MASK) | flags;
	spin_unlock(&file->f_lock);

	return 0;
}

static int ovl_real_fdget_meta(const struct file *file, struct fd *real,
			       bool allow_meta)
{
	struct inode *inode = file_inode(file);
	struct inode *realinode;

	real->flags = 0;
	real->file = file->private_data;

	if (allow_meta)
		realinode = ovl_inode_real(inode);
	else
		realinode = ovl_inode_realdata(inode);

	/* Has it been copied up since we'd opened it? */
	if (unlikely(file_inode(real->file) != realinode)) {
		real->flags = FDPUT_FPUT;
		real->file = ovl_open_realfile(file, realinode);

		return PTR_ERR_OR_ZERO(real->file);
	}

	/* Did the flags change since open? */
	if (unlikely((file->f_flags ^ real->file->f_flags) & ~OVL_OPEN_FLAGS))
		return ovl_change_flags(real->file, file->f_flags);

	return 0;
}

static int ovl_real_fdget(const struct file *file, struct fd *real)
{
	if (d_is_dir(file_dentry(file))) {
		real->flags = 0;
		real->file = ovl_dir_real_file(file, false);

		return PTR_ERR_OR_ZERO(real->file);
	}

	return ovl_real_fdget_meta(file, real, false);
}

static int ovl_open(struct inode *inode, struct file *file)
{
	struct file *realfile;
	int err;

	err = ovl_maybe_copy_up(file_dentry(file), file->f_flags);
	if (err)
		return err;

	/* No longer need these flags, so don't pass them on to underlying fs */
	file->f_flags &= ~(O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC);

	realfile = ovl_open_realfile(file, ovl_inode_realdata(inode));
	if (IS_ERR(realfile))
		return PTR_ERR(realfile);

	file->private_data = realfile;

	return 0;
}

static int ovl_release(struct inode *inode, struct file *file)
{
	fput(file->private_data);

	return 0;
}

static loff_t ovl_llseek(struct file *file, loff_t offset, int whence)
{
	struct inode *inode = file_inode(file);
	struct fd real;
	const struct cred *old_cred;
	loff_t ret;

	/*
	 * The two special cases below do not need to involve real fs,
	 * so we can optimizing concurrent callers.
	 */
	if (offset == 0) {
		if (whence == SEEK_CUR)
			return file->f_pos;

		if (whence == SEEK_SET)
			return vfs_setpos(file, 0, 0);
	}

	ret = ovl_real_fdget(file, &real);
	if (ret)
		return ret;

	/*
	 * Overlay file f_pos is the master copy that is preserved
	 * through copy up and modified on read/write, but only real
	 * fs knows how to SEEK_HOLE/SEEK_DATA and real fs may impose
	 * limitations that are more strict than ->s_maxbytes for specific
	 * files, so we use the real file to perform seeks.
	 */
	ovl_inode_lock(inode);
	real.file->f_pos = file->f_pos;

	old_cred = ovl_override_creds(inode->i_sb);
	ret = vfs_llseek(real.file, offset, whence);
	revert_creds(old_cred);

	file->f_pos = real.file->f_pos;
	ovl_inode_unlock(inode);

	fdput(real);

	return ret;
}

static void ovl_file_accessed(struct file *file)
{
	struct inode *inode, *upperinode;

	if (file->f_flags & O_NOATIME)
		return;

	inode = file_inode(file);
	upperinode = ovl_inode_upper(inode);

	if (!upperinode)
		return;

	if ((!timespec64_equal(&inode->i_mtime, &upperinode->i_mtime) ||
	     !timespec64_equal(&inode->i_ctime, &upperinode->i_ctime))) {
		inode->i_mtime = upperinode->i_mtime;
		inode->i_ctime = upperinode->i_ctime;
	}

	touch_atime(&file->f_path);
}

static rwf_t ovl_iocb_to_rwf(int ifl)
{
	rwf_t flags = 0;

	if (ifl & IOCB_NOWAIT)
		flags |= RWF_NOWAIT;
	if (ifl & IOCB_HIPRI)
		flags |= RWF_HIPRI;
	if (ifl & IOCB_DSYNC)
		flags |= RWF_DSYNC;
	if (ifl & IOCB_SYNC)
		flags |= RWF_SYNC;

	return flags;
}

static void ovl_aio_cleanup_handler(struct ovl_aio_req *aio_req)
{
	struct kiocb *iocb = &aio_req->iocb;
	struct kiocb *orig_iocb = aio_req->orig_iocb;

	if (iocb->ki_flags & IOCB_WRITE) {
		struct inode *inode = file_inode(orig_iocb->ki_filp);

		/* Actually acquired in ovl_write_iter() */
		__sb_writers_acquired(file_inode(iocb->ki_filp)->i_sb,
				      SB_FREEZE_WRITE);
		file_end_write(iocb->ki_filp);
		ovl_copyattr(ovl_inode_real(inode), inode);
	}

	orig_iocb->ki_pos = iocb->ki_pos;
	fdput(aio_req->fd);
	kmem_cache_free(ovl_aio_request_cachep, aio_req);
}

static void ovl_aio_rw_complete(struct kiocb *iocb, long res, long res2)
{
	struct ovl_aio_req *aio_req = container_of(iocb,
						   struct ovl_aio_req, iocb);
	struct kiocb *orig_iocb = aio_req->orig_iocb;

	ovl_aio_cleanup_handler(aio_req);
	orig_iocb->ki_complete(orig_iocb, res, res2);
}

static ssize_t ovl_read_iter(struct kiocb *iocb, struct iov_iter *iter)
{
	struct file *file = iocb->ki_filp;
	struct fd real;
	const struct cred *old_cred;
	ssize_t ret;

	if (!iov_iter_count(iter))
		return 0;

	ret = ovl_real_fdget(file, &real);
	if (ret)
		return ret;

	ret = -EINVAL;
	if (iocb->ki_flags & IOCB_DIRECT &&
	    (!real.file->f_mapping->a_ops ||
	     !real.file->f_mapping->a_ops->direct_IO))
		goto out_fdput;

	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	if (is_sync_kiocb(iocb)) {
		ret = vfs_iter_read(real.file, iter, &iocb->ki_pos,
				    ovl_iocb_to_rwf(iocb->ki_flags));
	} else {
		struct ovl_aio_req *aio_req;

		ret = -ENOMEM;
		aio_req = kmem_cache_zalloc(ovl_aio_request_cachep, GFP_KERNEL);
		if (!aio_req)
			goto out;

		aio_req->fd = real;
		real.flags = 0;
		aio_req->orig_iocb = iocb;
		kiocb_clone(&aio_req->iocb, iocb, real.file);
		aio_req->iocb.ki_complete = ovl_aio_rw_complete;
		ret = vfs_iocb_iter_read(real.file, &aio_req->iocb, iter);
		if (ret != -EIOCBQUEUED)
			ovl_aio_cleanup_handler(aio_req);
	}
out:
	revert_creds(old_cred);
	ovl_file_accessed(file);
out_fdput:
	fdput(real);

	return ret;
}

static ssize_t ovl_write_iter(struct kiocb *iocb, struct iov_iter *iter)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file_inode(file);
	struct fd real;
	const struct cred *old_cred;
	ssize_t ret;
	int ifl = iocb->ki_flags;

	if (!iov_iter_count(iter))
		return 0;

	inode_lock(inode);
	/* Update mode */
	ovl_copyattr(ovl_inode_real(inode), inode);
	ret = file_remove_privs(file);
	if (ret)
		goto out_unlock;

	ret = ovl_real_fdget(file, &real);
	if (ret)
		goto out_unlock;

	ret = -EINVAL;
	if (iocb->ki_flags & IOCB_DIRECT &&
	    (!real.file->f_mapping->a_ops ||
	     !real.file->f_mapping->a_ops->direct_IO))
		goto out_fdput;

	if (!ovl_should_sync(OVL_FS(inode->i_sb)))
		ifl &= ~(IOCB_DSYNC | IOCB_SYNC);

	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	if (is_sync_kiocb(iocb)) {
		file_start_write(real.file);
		ret = vfs_iter_write(real.file, iter, &iocb->ki_pos,
				     ovl_iocb_to_rwf(ifl));
		file_end_write(real.file);
		/* Update size */
		ovl_copyattr(ovl_inode_real(inode), inode);
	} else {
		struct ovl_aio_req *aio_req;

		ret = -ENOMEM;
		aio_req = kmem_cache_zalloc(ovl_aio_request_cachep, GFP_KERNEL);
		if (!aio_req)
			goto out;

		file_start_write(real.file);
		/* Pacify lockdep, same trick as done in aio_write() */
		__sb_writers_release(file_inode(real.file)->i_sb,
				     SB_FREEZE_WRITE);
		aio_req->fd = real;
		real.flags = 0;
		aio_req->orig_iocb = iocb;
		kiocb_clone(&aio_req->iocb, iocb, real.file);
		aio_req->iocb.ki_flags = ifl;
		aio_req->iocb.ki_complete = ovl_aio_rw_complete;
		ret = vfs_iocb_iter_write(real.file, &aio_req->iocb, iter);
		if (ret != -EIOCBQUEUED)
			ovl_aio_cleanup_handler(aio_req);
	}
out:
	revert_creds(old_cred);
out_fdput:
	fdput(real);

out_unlock:
	inode_unlock(inode);

	return ret;
}

/*
 * Calling iter_file_splice_write() directly from overlay's f_op may deadlock
 * due to lock order inversion between pipe->mutex in iter_file_splice_write()
 * and file_start_write(real.file) in ovl_write_iter().
 *
 * So do everything ovl_write_iter() does and call iter_file_splice_write() on
 * the real file.
 */
static ssize_t ovl_splice_write(struct pipe_inode_info *pipe, struct file *out,
				loff_t *ppos, size_t len, unsigned int flags)
{
	struct fd real;
	const struct cred *old_cred;
	struct inode *inode = file_inode(out);
	struct inode *realinode = ovl_inode_real(inode);
	ssize_t ret;

	inode_lock(inode);
	/* Update mode */
	ovl_copyattr(realinode, inode);
	ret = file_remove_privs(out);
	if (ret)
		goto out_unlock;

	ret = ovl_real_fdget(out, &real);
	if (ret)
		goto out_unlock;

	old_cred = ovl_override_creds(inode->i_sb);
	file_start_write(real.file);

	ret = iter_file_splice_write(pipe, real.file, ppos, len, flags);

	file_end_write(real.file);
	/* Update size */
	ovl_copyattr(realinode, inode);
	revert_creds(old_cred);
	fdput(real);

out_unlock:
	inode_unlock(inode);

	return ret;
}

static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
	struct fd real;
	const struct cred *old_cred;
	int ret;

	ret = ovl_sync_status(OVL_FS(file_inode(file)->i_sb));
	if (ret <= 0)
		return ret;

	ret = ovl_real_fdget_meta(file, &real, !datasync);
	if (ret)
		return ret;

	/* Don't sync lower file for fear of receiving EROFS error */
	if (file_inode(real.file) == ovl_inode_upper(file_inode(file))) {
		old_cred = ovl_override_creds(file_inode(file)->i_sb);
		ret = vfs_fsync_range(real.file, start, end, datasync);
		revert_creds(old_cred);
	}

	fdput(real);

	return ret;
}

static int ovl_mmap(struct file *file, struct vm_area_struct *vma)
{
	struct file *realfile = file->private_data;
	const struct cred *old_cred;
	int ret;

	if (!realfile->f_op->mmap)
		return -ENODEV;

	if (WARN_ON(file != vma->vm_file))
		return -EIO;

	vma_set_file(vma, realfile);

	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	ret = call_mmap(vma->vm_file, vma);
	revert_creds(old_cred);
	ovl_file_accessed(file);

	return ret;
}

static long ovl_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
{
	struct inode *inode = file_inode(file);
	struct fd real;
	const struct cred *old_cred;
	int ret;

	ret = ovl_real_fdget(file, &real);
	if (ret)
		return ret;

	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	ret = vfs_fallocate(real.file, mode, offset, len);
	revert_creds(old_cred);

	/* Update size */
	ovl_copyattr(ovl_inode_real(inode), inode);

	fdput(real);

	return ret;
}

static int ovl_fadvise(struct file *file, loff_t offset, loff_t len, int advice)
{
	struct fd real;
	const struct cred *old_cred;
	int ret;

	ret = ovl_real_fdget(file, &real);
	if (ret)
		return ret;

	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	ret = vfs_fadvise(real.file, offset, len, advice);
	revert_creds(old_cred);

	fdput(real);

	return ret;
}

enum ovl_copyop {
	OVL_COPY,
	OVL_CLONE,
	OVL_DEDUPE,
};

static loff_t ovl_copyfile(struct file *file_in, loff_t pos_in,
			    struct file *file_out, loff_t pos_out,
			    loff_t len, unsigned int flags, enum ovl_copyop op)
{
	struct inode *inode_out = file_inode(file_out);
	struct fd real_in, real_out;
	const struct cred *old_cred;
	loff_t ret;

	ret = ovl_real_fdget(file_out, &real_out);
	if (ret)
		return ret;

	ret = ovl_real_fdget(file_in, &real_in);
	if (ret) {
		fdput(real_out);
		return ret;
	}

	old_cred = ovl_override_creds(file_inode(file_out)->i_sb);
	switch (op) {
	case OVL_COPY:
		ret = vfs_copy_file_range(real_in.file, pos_in,
					  real_out.file, pos_out, len, flags);
		break;

	case OVL_CLONE:
		ret = vfs_clone_file_range(real_in.file, pos_in,
					   real_out.file, pos_out, len, flags);
		break;

	case OVL_DEDUPE:
		ret = vfs_dedupe_file_range_one(real_in.file, pos_in,
						real_out.file, pos_out, len,
						flags);
		break;
	}
	revert_creds(old_cred);

	/* Update size */
	ovl_copyattr(ovl_inode_real(inode_out), inode_out);

	fdput(real_in);
	fdput(real_out);

	return ret;
}

static ssize_t ovl_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)
{
	return ovl_copyfile(file_in, pos_in, file_out, pos_out, len, flags,
			    OVL_COPY);
}

static loff_t ovl_remap_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)
{
	enum ovl_copyop op;

	if (remap_flags & ~(REMAP_FILE_DEDUP | REMAP_FILE_ADVISORY))
		return -EINVAL;

	if (remap_flags & REMAP_FILE_DEDUP)
		op = OVL_DEDUPE;
	else
		op = OVL_CLONE;

	/*
	 * Don't copy up because of a dedupe request, this wouldn't make sense
	 * most of the time (data would be duplicated instead of deduplicated).
	 */
	if (op == OVL_DEDUPE &&
	    (!ovl_inode_upper(file_inode(file_in)) ||
	     !ovl_inode_upper(file_inode(file_out))))
		return -EPERM;

	return ovl_copyfile(file_in, pos_in, file_out, pos_out, len,
			    remap_flags, op);
}

static int ovl_flush(struct file *file, fl_owner_t id)
{
	struct fd real;
	const struct cred *old_cred;
	int err;

	err = ovl_real_fdget(file, &real);
	if (err)
		return err;

	if (real.file->f_op->flush) {
		old_cred = ovl_override_creds(file_inode(file)->i_sb);
		err = real.file->f_op->flush(real.file, id);
		revert_creds(old_cred);
	}
	fdput(real);

	return err;
}

const struct file_operations ovl_file_operations = {
	.open		= ovl_open,
	.release	= ovl_release,
	.llseek		= ovl_llseek,
	.read_iter	= ovl_read_iter,
	.write_iter	= ovl_write_iter,
	.fsync		= ovl_fsync,
	.mmap		= ovl_mmap,
	.fallocate	= ovl_fallocate,
	.fadvise	= ovl_fadvise,
	.flush		= ovl_flush,
	.splice_read    = generic_file_splice_read,
	.splice_write   = ovl_splice_write,

	.copy_file_range	= ovl_copy_file_range,
	.remap_file_range	= ovl_remap_file_range,
};

int __init ovl_aio_request_cache_init(void)
{
	ovl_aio_request_cachep = kmem_cache_create("ovl_aio_req",
						   sizeof(struct ovl_aio_req),
						   0, SLAB_HWCACHE_ALIGN, NULL);
	if (!ovl_aio_request_cachep)
		return -ENOMEM;

	return 0;
}

void ovl_aio_request_cache_destroy(void)
{
	kmem_cache_destroy(ovl_aio_request_cachep);
}
Commit	Line	Data
d2912cb1	1	// SPDX-License-Identifier: GPL-2.0-only
d1d04ef8 MS	2	/*
d1d04ef8 MS	3	* Copyright (C) 2017 Red Hat, Inc.
d1d04ef8 MS	4	*/
	5
	6	#include <linux/cred.h>
	7	#include <linux/file.h>
dab5ca8f	8	#include <linux/mount.h>
d1d04ef8	9	#include <linux/xattr.h>
16914e6f	10	#include <linux/uio.h>
98487de3	11	#include <linux/uaccess.h>
1a980b8c	12	#include <linux/splice.h>
292f902a	13	#include <linux/security.h>
1a980b8c MZ	14	#include <linux/mm.h>
1a980b8c MZ	15	#include <linux/fs.h>
d1d04ef8 MS	16	#include "overlayfs.h"
d1d04ef8 MS	17
2406a307 JX	18	struct ovl_aio_req {
	19	struct kiocb iocb;
	20	struct kiocb *orig_iocb;
	21	struct fd fd;
	22	};
	23
	24	static struct kmem_cache *ovl_aio_request_cachep;
	25
8c444d2a VG	26	static char ovl_whatisit(struct inode inode, struct inode realinode)
	27	{
	28	if (realinode != ovl_inode_upper(inode))
	29	return 'l';
	30	if (ovl_has_upperdata(inode))
	31	return 'u';
	32	else
	33	return 'm';
	34	}
	35
81a33c1e AG	36	/* No atime modificaton nor notify on underlying */
	37	#define OVL_OPEN_FLAGS (O_NOATIME \| FMODE_NONOTIFY)
	38
8c444d2a VG	39	static struct file ovl_open_realfile(const struct file file,
8c444d2a VG	40	struct inode *realinode)
d1d04ef8 MS	41	{
d1d04ef8 MS	42	struct inode *inode = file_inode(file);
d1d04ef8 MS	43	struct file *realfile;
d1d04ef8 MS	44	const struct cred *old_cred;
81a33c1e	45	int flags = file->f_flags \| OVL_OPEN_FLAGS;
05acefb4 MS	46	int acc_mode = ACC_MODE(flags);
	47	int err;
	48
	49	if (flags & O_APPEND)
	50	acc_mode \|= MAY_APPEND;
d1d04ef8 MS	51
d1d04ef8 MS	52	old_cred = ovl_override_creds(inode->i_sb);
47291baa	53	err = inode_permission(&init_user_ns, realinode, MAY_OPEN \| acc_mode);
05acefb4 MS	54	if (err) {
05acefb4 MS	55	realfile = ERR_PTR(err);
05acefb4	56	} else {
21cb47be	57	if (!inode_owner_or_capable(&init_user_ns, realinode))
b6650dab MS	58	flags &= ~O_NOATIME;
b6650dab MS	59
05acefb4 MS	60	realfile = open_with_fake_path(&file->f_path, flags, realinode,
	61	current_cred());
	62	}
d1d04ef8 MS	63	revert_creds(old_cred);
	64
	65	pr_debug("open(%p[%pD2/%c], 0%o) -> (%p, 0%o)\n",
8c444d2a	66	file, file, ovl_whatisit(inode, realinode), file->f_flags,
d1d04ef8 MS	67	realfile, IS_ERR(realfile) ? 0 : realfile->f_flags);
	68
	69	return realfile;
	70	}
	71
2ef66b8a MS	72	#define OVL_SETFL_MASK (O_APPEND \| O_NONBLOCK \| O_NDELAY \| O_DIRECT)
	73
	74	static int ovl_change_flags(struct file *file, unsigned int flags)
	75	{
	76	struct inode *inode = file_inode(file);
	77	int err;
	78
2ef66b8a MS	79	flags &= OVL_SETFL_MASK;
	80
	81	if (((flags ^ file->f_flags) & O_APPEND) && IS_APPEND(inode))
	82	return -EPERM;
	83
	84	if (flags & O_DIRECT) {
	85	if (!file->f_mapping->a_ops \|\|
	86	!file->f_mapping->a_ops->direct_IO)
	87	return -EINVAL;
	88	}
	89
	90	if (file->f_op->check_flags) {
	91	err = file->f_op->check_flags(flags);
	92	if (err)
	93	return err;
	94	}
	95
	96	spin_lock(&file->f_lock);
	97	file->f_flags = (file->f_flags & ~OVL_SETFL_MASK) \| flags;
	98	spin_unlock(&file->f_lock);
	99
	100	return 0;
	101	}
	102
8c444d2a VG	103	static int ovl_real_fdget_meta(const struct file file, struct fd real,
8c444d2a VG	104	bool allow_meta)
2ef66b8a MS	105	{
2ef66b8a MS	106	struct inode *inode = file_inode(file);
8c444d2a	107	struct inode *realinode;
2ef66b8a MS	108
	109	real->flags = 0;
	110	real->file = file->private_data;
	111
8c444d2a VG	112	if (allow_meta)
	113	realinode = ovl_inode_real(inode);
	114	else
	115	realinode = ovl_inode_realdata(inode);
	116
2ef66b8a	117	/* Has it been copied up since we'd opened it? */
8c444d2a	118	if (unlikely(file_inode(real->file) != realinode)) {
2ef66b8a	119	real->flags = FDPUT_FPUT;
8c444d2a	120	real->file = ovl_open_realfile(file, realinode);
2ef66b8a MS	121
	122	return PTR_ERR_OR_ZERO(real->file);
	123	}
	124
	125	/* Did the flags change since open? */
81a33c1e	126	if (unlikely((file->f_flags ^ real->file->f_flags) & ~OVL_OPEN_FLAGS))
2ef66b8a MS	127	return ovl_change_flags(real->file, file->f_flags);
	128
	129	return 0;
	130	}
	131
8c444d2a VG	132	static int ovl_real_fdget(const struct file file, struct fd real)
8c444d2a VG	133	{
61536bed AG	134	if (d_is_dir(file_dentry(file))) {
	135	real->flags = 0;
	136	real->file = ovl_dir_real_file(file, false);
	137
	138	return PTR_ERR_OR_ZERO(real->file);
	139	}
	140
8c444d2a VG	141	return ovl_real_fdget_meta(file, real, false);
	142	}
	143
d1d04ef8 MS	144	static int ovl_open(struct inode inode, struct file file)
d1d04ef8 MS	145	{
d1d04ef8 MS	146	struct file *realfile;
	147	int err;
	148
3428030d	149	err = ovl_maybe_copy_up(file_dentry(file), file->f_flags);
d1d04ef8 MS	150	if (err)
	151	return err;
	152
	153	/* No longer need these flags, so don't pass them on to underlying fs */
	154	file->f_flags &= ~(O_CREAT \| O_EXCL \| O_NOCTTY \| O_TRUNC);
	155
8c444d2a	156	realfile = ovl_open_realfile(file, ovl_inode_realdata(inode));
d1d04ef8 MS	157	if (IS_ERR(realfile))
	158	return PTR_ERR(realfile);
	159
	160	file->private_data = realfile;
	161
	162	return 0;
	163	}
	164
	165	static int ovl_release(struct inode inode, struct file file)
	166	{
	167	fput(file->private_data);
	168
	169	return 0;
	170	}
	171
	172	static loff_t ovl_llseek(struct file *file, loff_t offset, int whence)
	173	{
9e46b840 AG	174	struct inode *inode = file_inode(file);
	175	struct fd real;
	176	const struct cred *old_cred;
a4ac9d45	177	loff_t ret;
9e46b840 AG	178
	179	/*
	180	* The two special cases below do not need to involve real fs,
	181	* so we can optimizing concurrent callers.
	182	*/
	183	if (offset == 0) {
	184	if (whence == SEEK_CUR)
	185	return file->f_pos;
	186
	187	if (whence == SEEK_SET)
	188	return vfs_setpos(file, 0, 0);
	189	}
	190
	191	ret = ovl_real_fdget(file, &real);
	192	if (ret)
	193	return ret;
	194
	195	/*
	196	* Overlay file f_pos is the master copy that is preserved
	197	* through copy up and modified on read/write, but only real
	198	* fs knows how to SEEK_HOLE/SEEK_DATA and real fs may impose
	199	* limitations that are more strict than ->s_maxbytes for specific
	200	* files, so we use the real file to perform seeks.
	201	*/
b1f9d385	202	ovl_inode_lock(inode);
9e46b840 AG	203	real.file->f_pos = file->f_pos;
	204
	205	old_cred = ovl_override_creds(inode->i_sb);
	206	ret = vfs_llseek(real.file, offset, whence);
	207	revert_creds(old_cred);
	208
	209	file->f_pos = real.file->f_pos;
b1f9d385	210	ovl_inode_unlock(inode);
9e46b840 AG	211
9e46b840 AG	212	fdput(real);
d1d04ef8	213
9e46b840	214	return ret;
d1d04ef8 MS	215	}
d1d04ef8 MS	216
16914e6f MS	217	static void ovl_file_accessed(struct file *file)
	218	{
	219	struct inode inode, upperinode;
	220
	221	if (file->f_flags & O_NOATIME)
	222	return;
	223
	224	inode = file_inode(file);
	225	upperinode = ovl_inode_upper(inode);
	226
	227	if (!upperinode)
	228	return;
	229
	230	if ((!timespec64_equal(&inode->i_mtime, &upperinode->i_mtime) \|\|
	231	!timespec64_equal(&inode->i_ctime, &upperinode->i_ctime))) {
	232	inode->i_mtime = upperinode->i_mtime;
	233	inode->i_ctime = upperinode->i_ctime;
	234	}
	235
	236	touch_atime(&file->f_path);
	237	}
	238
b778e1ee	239	static rwf_t ovl_iocb_to_rwf(int ifl)
16914e6f	240	{
16914e6f MS	241	rwf_t flags = 0;
	242
	243	if (ifl & IOCB_NOWAIT)
	244	flags \|= RWF_NOWAIT;
	245	if (ifl & IOCB_HIPRI)
	246	flags \|= RWF_HIPRI;
	247	if (ifl & IOCB_DSYNC)
	248	flags \|= RWF_DSYNC;
	249	if (ifl & IOCB_SYNC)
	250	flags \|= RWF_SYNC;
	251
	252	return flags;
	253	}
	254
2406a307 JX	255	static void ovl_aio_cleanup_handler(struct ovl_aio_req *aio_req)
	256	{
	257	struct kiocb *iocb = &aio_req->iocb;
	258	struct kiocb *orig_iocb = aio_req->orig_iocb;
	259
	260	if (iocb->ki_flags & IOCB_WRITE) {
	261	struct inode *inode = file_inode(orig_iocb->ki_filp);
	262
c8536804 MS	263	/* Actually acquired in ovl_write_iter() */
	264	__sb_writers_acquired(file_inode(iocb->ki_filp)->i_sb,
	265	SB_FREEZE_WRITE);
2406a307 JX	266	file_end_write(iocb->ki_filp);
	267	ovl_copyattr(ovl_inode_real(inode), inode);
	268	}
	269
	270	orig_iocb->ki_pos = iocb->ki_pos;
	271	fdput(aio_req->fd);
	272	kmem_cache_free(ovl_aio_request_cachep, aio_req);
	273	}
	274
	275	static void ovl_aio_rw_complete(struct kiocb *iocb, long res, long res2)
	276	{
	277	struct ovl_aio_req *aio_req = container_of(iocb,
	278	struct ovl_aio_req, iocb);
	279	struct kiocb *orig_iocb = aio_req->orig_iocb;
	280
	281	ovl_aio_cleanup_handler(aio_req);
	282	orig_iocb->ki_complete(orig_iocb, res, res2);
	283	}
	284
16914e6f MS	285	static ssize_t ovl_read_iter(struct kiocb iocb, struct iov_iter iter)
	286	{
	287	struct file *file = iocb->ki_filp;
	288	struct fd real;
	289	const struct cred *old_cred;
	290	ssize_t ret;
	291
	292	if (!iov_iter_count(iter))
	293	return 0;
	294
	295	ret = ovl_real_fdget(file, &real);
	296	if (ret)
	297	return ret;
	298
1dc1eed4 MS	299	ret = -EINVAL;
	300	if (iocb->ki_flags & IOCB_DIRECT &&
	301	(!real.file->f_mapping->a_ops \|\|
	302	!real.file->f_mapping->a_ops->direct_IO))
	303	goto out_fdput;
	304
16914e6f	305	old_cred = ovl_override_creds(file_inode(file)->i_sb);
2406a307 JX	306	if (is_sync_kiocb(iocb)) {
2406a307 JX	307	ret = vfs_iter_read(real.file, iter, &iocb->ki_pos,
b778e1ee	308	ovl_iocb_to_rwf(iocb->ki_flags));
2406a307 JX	309	} else {
	310	struct ovl_aio_req *aio_req;
	311
	312	ret = -ENOMEM;
	313	aio_req = kmem_cache_zalloc(ovl_aio_request_cachep, GFP_KERNEL);
	314	if (!aio_req)
	315	goto out;
	316
	317	aio_req->fd = real;
	318	real.flags = 0;
	319	aio_req->orig_iocb = iocb;
	320	kiocb_clone(&aio_req->iocb, iocb, real.file);
	321	aio_req->iocb.ki_complete = ovl_aio_rw_complete;
	322	ret = vfs_iocb_iter_read(real.file, &aio_req->iocb, iter);
	323	if (ret != -EIOCBQUEUED)
	324	ovl_aio_cleanup_handler(aio_req);
	325	}
	326	out:
16914e6f	327	revert_creds(old_cred);
16914e6f	328	ovl_file_accessed(file);
1dc1eed4	329	out_fdput:
16914e6f MS	330	fdput(real);
	331
	332	return ret;
	333	}
	334
2a92e07e MS	335	static ssize_t ovl_write_iter(struct kiocb iocb, struct iov_iter iter)
	336	{
	337	struct file *file = iocb->ki_filp;
	338	struct inode *inode = file_inode(file);
	339	struct fd real;
	340	const struct cred *old_cred;
	341	ssize_t ret;
c86243b0	342	int ifl = iocb->ki_flags;
2a92e07e MS	343
	344	if (!iov_iter_count(iter))
	345	return 0;
	346
	347	inode_lock(inode);
	348	/* Update mode */
	349	ovl_copyattr(ovl_inode_real(inode), inode);
	350	ret = file_remove_privs(file);
	351	if (ret)
	352	goto out_unlock;
	353
	354	ret = ovl_real_fdget(file, &real);
	355	if (ret)
	356	goto out_unlock;
	357
1dc1eed4 MS	358	ret = -EINVAL;
	359	if (iocb->ki_flags & IOCB_DIRECT &&
	360	(!real.file->f_mapping->a_ops \|\|
	361	!real.file->f_mapping->a_ops->direct_IO))
	362	goto out_fdput;
	363
c86243b0 VG	364	if (!ovl_should_sync(OVL_FS(inode->i_sb)))
	365	ifl &= ~(IOCB_DSYNC \| IOCB_SYNC);
	366
2a92e07e	367	old_cred = ovl_override_creds(file_inode(file)->i_sb);
2406a307 JX	368	if (is_sync_kiocb(iocb)) {
	369	file_start_write(real.file);
	370	ret = vfs_iter_write(real.file, iter, &iocb->ki_pos,
c86243b0	371	ovl_iocb_to_rwf(ifl));
2406a307 JX	372	file_end_write(real.file);
	373	/* Update size */
	374	ovl_copyattr(ovl_inode_real(inode), inode);
	375	} else {
	376	struct ovl_aio_req *aio_req;
	377
	378	ret = -ENOMEM;
	379	aio_req = kmem_cache_zalloc(ovl_aio_request_cachep, GFP_KERNEL);
	380	if (!aio_req)
	381	goto out;
	382
	383	file_start_write(real.file);
c8536804 MS	384	/* Pacify lockdep, same trick as done in aio_write() */
	385	__sb_writers_release(file_inode(real.file)->i_sb,
	386	SB_FREEZE_WRITE);
2406a307 JX	387	aio_req->fd = real;
	388	real.flags = 0;
	389	aio_req->orig_iocb = iocb;
	390	kiocb_clone(&aio_req->iocb, iocb, real.file);
c86243b0	391	aio_req->iocb.ki_flags = ifl;
2406a307 JX	392	aio_req->iocb.ki_complete = ovl_aio_rw_complete;
	393	ret = vfs_iocb_iter_write(real.file, &aio_req->iocb, iter);
	394	if (ret != -EIOCBQUEUED)
	395	ovl_aio_cleanup_handler(aio_req);
	396	}
	397	out:
2a92e07e	398	revert_creds(old_cred);
1dc1eed4	399	out_fdput:
2a92e07e MS	400	fdput(real);
	401
	402	out_unlock:
	403	inode_unlock(inode);
	404
	405	return ret;
	406	}
	407
9b91b6b0 MS	408	/*
	409	* Calling iter_file_splice_write() directly from overlay's f_op may deadlock
	410	* due to lock order inversion between pipe->mutex in iter_file_splice_write()
	411	* and file_start_write(real.file) in ovl_write_iter().
	412	*
	413	* So do everything ovl_write_iter() does and call iter_file_splice_write() on
	414	* the real file.
	415	*/
	416	static ssize_t ovl_splice_write(struct pipe_inode_info pipe, struct file out,
	417	loff_t *ppos, size_t len, unsigned int flags)
	418	{
	419	struct fd real;
	420	const struct cred *old_cred;
	421	struct inode *inode = file_inode(out);
	422	struct inode *realinode = ovl_inode_real(inode);
	423	ssize_t ret;
	424
	425	inode_lock(inode);
	426	/* Update mode */
	427	ovl_copyattr(realinode, inode);
	428	ret = file_remove_privs(out);
	429	if (ret)
	430	goto out_unlock;
	431
	432	ret = ovl_real_fdget(out, &real);
	433	if (ret)
	434	goto out_unlock;
	435
	436	old_cred = ovl_override_creds(inode->i_sb);
	437	file_start_write(real.file);
	438
	439	ret = iter_file_splice_write(pipe, real.file, ppos, len, flags);
	440
	441	file_end_write(real.file);
	442	/* Update size */
	443	ovl_copyattr(realinode, inode);
	444	revert_creds(old_cred);
	445	fdput(real);
	446
	447	out_unlock:
	448	inode_unlock(inode);
	449
	450	return ret;
	451	}
	452
de30dfd6 MS	453	static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync)
	454	{
	455	struct fd real;
	456	const struct cred *old_cred;
	457	int ret;
	458
335d3fc5 SD	459	ret = ovl_sync_status(OVL_FS(file_inode(file)->i_sb));
	460	if (ret <= 0)
	461	return ret;
c86243b0	462
8c444d2a	463	ret = ovl_real_fdget_meta(file, &real, !datasync);
de30dfd6 MS	464	if (ret)
	465	return ret;
	466
	467	/* Don't sync lower file for fear of receiving EROFS error */
	468	if (file_inode(real.file) == ovl_inode_upper(file_inode(file))) {
	469	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	470	ret = vfs_fsync_range(real.file, start, end, datasync);
	471	revert_creds(old_cred);
	472	}
	473
	474	fdput(real);
	475
	476	return ret;
	477	}
	478
2f502839 MS	479	static int ovl_mmap(struct file file, struct vm_area_struct vma)
	480	{
	481	struct file *realfile = file->private_data;
	482	const struct cred *old_cred;
	483	int ret;
	484
	485	if (!realfile->f_op->mmap)
	486	return -ENODEV;
	487
	488	if (WARN_ON(file != vma->vm_file))
	489	return -EIO;
	490
2896900e	491	vma_set_file(vma, realfile);
2f502839 MS	492
	493	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	494	ret = call_mmap(vma->vm_file, vma);
	495	revert_creds(old_cred);
2f502839 MS	496	ovl_file_accessed(file);
	497
	498	return ret;
	499	}
	500
aab8848c MS	501	static long ovl_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
	502	{
	503	struct inode *inode = file_inode(file);
	504	struct fd real;
	505	const struct cred *old_cred;
	506	int ret;
	507
	508	ret = ovl_real_fdget(file, &real);
	509	if (ret)
	510	return ret;
	511
	512	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	513	ret = vfs_fallocate(real.file, mode, offset, len);
	514	revert_creds(old_cred);
	515
	516	/* Update size */
	517	ovl_copyattr(ovl_inode_real(inode), inode);
	518
	519	fdput(real);
	520
	521	return ret;
	522	}
	523
b833a366 AG	524	static int ovl_fadvise(struct file *file, loff_t offset, loff_t len, int advice)
	525	{
	526	struct fd real;
	527	const struct cred *old_cred;
	528	int ret;
	529
	530	ret = ovl_real_fdget(file, &real);
	531	if (ret)
	532	return ret;
	533
	534	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	535	ret = vfs_fadvise(real.file, offset, len, advice);
	536	revert_creds(old_cred);
	537
	538	fdput(real);
	539
	540	return ret;
	541	}
	542
8ede2055 MS	543	enum ovl_copyop {
	544	OVL_COPY,
	545	OVL_CLONE,
	546	OVL_DEDUPE,
	547	};
	548
42ec3d4c	549	static loff_t ovl_copyfile(struct file *file_in, loff_t pos_in,
8ede2055	550	struct file *file_out, loff_t pos_out,
42ec3d4c	551	loff_t len, unsigned int flags, enum ovl_copyop op)
8ede2055 MS	552	{
	553	struct inode *inode_out = file_inode(file_out);
	554	struct fd real_in, real_out;
	555	const struct cred *old_cred;
42ec3d4c	556	loff_t ret;
8ede2055 MS	557
	558	ret = ovl_real_fdget(file_out, &real_out);
	559	if (ret)
	560	return ret;
	561
	562	ret = ovl_real_fdget(file_in, &real_in);
	563	if (ret) {
	564	fdput(real_out);
	565	return ret;
	566	}
	567
	568	old_cred = ovl_override_creds(file_inode(file_out)->i_sb);
	569	switch (op) {
	570	case OVL_COPY:
	571	ret = vfs_copy_file_range(real_in.file, pos_in,
	572	real_out.file, pos_out, len, flags);
	573	break;
	574
	575	case OVL_CLONE:
a725356b	576	ret = vfs_clone_file_range(real_in.file, pos_in,
452ce659	577	real_out.file, pos_out, len, flags);
8ede2055 MS	578	break;
	579
	580	case OVL_DEDUPE:
	581	ret = vfs_dedupe_file_range_one(real_in.file, pos_in,
df365836 DW	582	real_out.file, pos_out, len,
df365836 DW	583	flags);
8ede2055 MS	584	break;
	585	}
	586	revert_creds(old_cred);
	587
	588	/* Update size */
	589	ovl_copyattr(ovl_inode_real(inode_out), inode_out);
	590
	591	fdput(real_in);
	592	fdput(real_out);
	593
	594	return ret;
	595	}
	596
	597	static ssize_t ovl_copy_file_range(struct file *file_in, loff_t pos_in,
	598	struct file *file_out, loff_t pos_out,
	599	size_t len, unsigned int flags)
	600	{
	601	return ovl_copyfile(file_in, pos_in, file_out, pos_out, len, flags,
	602	OVL_COPY);
	603	}
	604
42ec3d4c DW	605	static loff_t ovl_remap_file_range(struct file *file_in, loff_t pos_in,
	606	struct file *file_out, loff_t pos_out,
	607	loff_t len, unsigned int remap_flags)
8ede2055	608	{
2e5dfc99 DW	609	enum ovl_copyop op;
	610
	611	if (remap_flags & ~(REMAP_FILE_DEDUP \| REMAP_FILE_ADVISORY))
	612	return -EINVAL;
	613
	614	if (remap_flags & REMAP_FILE_DEDUP)
	615	op = OVL_DEDUPE;
	616	else
	617	op = OVL_CLONE;
8ede2055	618
8ede2055 MS	619	/*
	620	* Don't copy up because of a dedupe request, this wouldn't make sense
	621	* most of the time (data would be duplicated instead of deduplicated).
	622	*/
2e5dfc99 DW	623	if (op == OVL_DEDUPE &&
	624	(!ovl_inode_upper(file_inode(file_in)) \|\|
	625	!ovl_inode_upper(file_inode(file_out))))
8ede2055 MS	626	return -EPERM;
8ede2055 MS	627
452ce659 DW	628	return ovl_copyfile(file_in, pos_in, file_out, pos_out, len,
452ce659 DW	629	remap_flags, op);
8ede2055 MS	630	}
8ede2055 MS	631
1f0cb8bc SD	632	static int ovl_flush(struct file *file, fl_owner_t id)
	633	{
	634	struct fd real;
	635	const struct cred *old_cred;
	636	int err;
	637
	638	err = ovl_real_fdget(file, &real);
	639	if (err)
	640	return err;
	641
	642	if (real.file->f_op->flush) {
	643	old_cred = ovl_override_creds(file_inode(file)->i_sb);
	644	err = real.file->f_op->flush(real.file, id);
	645	revert_creds(old_cred);
	646	}
	647	fdput(real);
	648
	649	return err;
	650	}
	651
d1d04ef8 MS	652	const struct file_operations ovl_file_operations = {
	653	.open = ovl_open,
	654	.release = ovl_release,
	655	.llseek = ovl_llseek,
16914e6f	656	.read_iter = ovl_read_iter,
2a92e07e	657	.write_iter = ovl_write_iter,
de30dfd6	658	.fsync = ovl_fsync,
2f502839	659	.mmap = ovl_mmap,
aab8848c	660	.fallocate = ovl_fallocate,
b833a366	661	.fadvise = ovl_fadvise,
1f0cb8bc	662	.flush = ovl_flush,
82a763e6	663	.splice_read = generic_file_splice_read,
9b91b6b0	664	.splice_write = ovl_splice_write,
8ede2055 MS	665
8ede2055 MS	666	.copy_file_range = ovl_copy_file_range,
2e5dfc99	667	.remap_file_range = ovl_remap_file_range,
d1d04ef8	668	};
2406a307 JX	669
	670	int __init ovl_aio_request_cache_init(void)
	671	{
	672	ovl_aio_request_cachep = kmem_cache_create("ovl_aio_req",
	673	sizeof(struct ovl_aio_req),
	674	0, SLAB_HWCACHE_ALIGN, NULL);
	675	if (!ovl_aio_request_cachep)
	676	return -ENOMEM;
	677
	678	return 0;
	679	}
	680
	681	void ovl_aio_request_cache_destroy(void)
	682	{
	683	kmem_cache_destroy(ovl_aio_request_cachep);
	684	}