[linux-block.git] / fs / nfs / nfs4file.c

// SPDX-License-Identifier: GPL-2.0
/*
 *  linux/fs/nfs/file.c
 *
 *  Copyright (C) 1992  Rick Sladkey
 */
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/falloc.h>
#include <linux/mount.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_ssc.h>
#include "delegation.h"
#include "internal.h"
#include "iostat.h"
#include "fscache.h"
#include "pnfs.h"

#include "nfstrace.h"

#ifdef CONFIG_NFS_V4_2
#include "nfs42.h"
#endif

#define NFSDBG_FACILITY		NFSDBG_FILE

static int
nfs4_file_open(struct inode *inode, struct file *filp)
{
	struct nfs_open_context *ctx;
	struct dentry *dentry = file_dentry(filp);
	struct dentry *parent = NULL;
	struct inode *dir;
	unsigned openflags = filp->f_flags;
	struct iattr attr;
	int err;

	/*
	 * If no cached dentry exists or if it's negative, NFSv4 handled the
	 * opens in ->lookup() or ->create().
	 *
	 * We only get this far for a cached positive dentry.  We skipped
	 * revalidation, so handle it here by dropping the dentry and returning
	 * -EOPENSTALE.  The VFS will retry the lookup/create/open.
	 */

	dprintk("NFS: open file(%pd2)\n", dentry);

	err = nfs_check_flags(openflags);
	if (err)
		return err;

	/* We can't create new files here */
	openflags &= ~(O_CREAT|O_EXCL);

	parent = dget_parent(dentry);
	dir = d_inode(parent);

	ctx = alloc_nfs_open_context(file_dentry(filp),
				     flags_to_mode(openflags), filp);
	err = PTR_ERR(ctx);
	if (IS_ERR(ctx))
		goto out;

	attr.ia_valid = ATTR_OPEN;
	if (openflags & O_TRUNC) {
		attr.ia_valid |= ATTR_SIZE;
		attr.ia_size = 0;
		filemap_write_and_wait(inode->i_mapping);
	}

	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, NULL);
	if (IS_ERR(inode)) {
		err = PTR_ERR(inode);
		switch (err) {
		default:
			goto out_put_ctx;
		case -ENOENT:
		case -ESTALE:
		case -EISDIR:
		case -ENOTDIR:
		case -ELOOP:
			goto out_drop;
		}
	}
	if (inode != d_inode(dentry))
		goto out_drop;

	nfs_file_set_open_context(filp, ctx);
	nfs_fscache_open_file(inode, filp);
	err = 0;
	filp->f_mode |= FMODE_CAN_ODIRECT;

out_put_ctx:
	put_nfs_open_context(ctx);
out:
	dput(parent);
	return err;

out_drop:
	d_drop(dentry);
	err = -EOPENSTALE;
	goto out_put_ctx;
}

/*
 * Flush all dirty pages, and check for write errors.
 */
static int
nfs4_file_flush(struct file *file, fl_owner_t id)
{
	struct inode	*inode = file_inode(file);
	errseq_t since;

	dprintk("NFS: flush(%pD2)\n", file);

	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
	if ((file->f_mode & FMODE_WRITE) == 0)
		return 0;

	/*
	 * If we're holding a write delegation, then check if we're required
	 * to flush the i/o on close. If not, then just start the i/o now.
	 */
	if (!nfs4_delegation_flush_on_close(inode))
		return filemap_fdatawrite(file->f_mapping);

	/* Flush writes to the server and return any errors */
	since = filemap_sample_wb_err(file->f_mapping);
	nfs_wb_all(inode);
	return filemap_check_wb_err(file->f_mapping, since);
}

#ifdef CONFIG_NFS_V4_2
static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
				      struct file *file_out, loff_t pos_out,
				      size_t count, unsigned int flags)
{
	struct nfs42_copy_notify_res *cn_resp = NULL;
	struct nl4_server *nss = NULL;
	nfs4_stateid *cnrs = NULL;
	ssize_t ret;
	bool sync = false;

	/* Only offload copy if superblock is the same */
	if (file_in->f_op != &nfs4_file_operations)
		return -EXDEV;
	if (!nfs_server_capable(file_inode(file_out), NFS_CAP_COPY) ||
	    !nfs_server_capable(file_inode(file_in), NFS_CAP_COPY))
		return -EOPNOTSUPP;
	if (file_inode(file_in) == file_inode(file_out))
		return -EOPNOTSUPP;
	/* if the copy size if smaller than 2 RPC payloads, make it
	 * synchronous
	 */
	if (count <= 2 * NFS_SERVER(file_inode(file_in))->rsize)
		sync = true;
retry:
	if (!nfs42_files_from_same_server(file_in, file_out)) {
		/*
		 * for inter copy, if copy size is too small
		 * then fallback to generic copy.
		 */
		if (sync)
			return -EOPNOTSUPP;
		cn_resp = kzalloc(sizeof(struct nfs42_copy_notify_res),
				  GFP_KERNEL);
		if (unlikely(cn_resp == NULL))
			return -ENOMEM;

		ret = nfs42_proc_copy_notify(file_in, file_out, cn_resp);
		if (ret) {
			ret = -EOPNOTSUPP;
			goto out;
		}
		nss = &cn_resp->cnr_src;
		cnrs = &cn_resp->cnr_stateid;
	}
	ret = nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count,
				nss, cnrs, sync);
out:
	kfree(cn_resp);

	if (ret == -EAGAIN)
		goto retry;
	return ret;
}

static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
				    struct file *file_out, loff_t pos_out,
				    size_t count, unsigned int flags)
{
	ssize_t ret;

	ret = __nfs4_copy_file_range(file_in, pos_in, file_out, pos_out, count,
				     flags);
	if (ret == -EOPNOTSUPP || ret == -EXDEV)
		ret = generic_copy_file_range(file_in, pos_in, file_out,
					      pos_out, count, flags);
	return ret;
}

static loff_t nfs4_file_llseek(struct file *filep, loff_t offset, int whence)
{
	loff_t ret;

	switch (whence) {
	case SEEK_HOLE:
	case SEEK_DATA:
		ret = nfs42_proc_llseek(filep, offset, whence);
		if (ret != -EOPNOTSUPP)
			return ret;
		fallthrough;
	default:
		return nfs_file_llseek(filep, offset, whence);
	}
}

static long nfs42_fallocate(struct file *filep, int mode, loff_t offset, loff_t len)
{
	struct inode *inode = file_inode(filep);
	long ret;

	if (!S_ISREG(inode->i_mode))
		return -EOPNOTSUPP;

	if ((mode != 0) && (mode != (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE)))
		return -EOPNOTSUPP;

	ret = inode_newsize_ok(inode, offset + len);
	if (ret < 0)
		return ret;

	if (mode & FALLOC_FL_PUNCH_HOLE)
		return nfs42_proc_deallocate(filep, offset, len);
	return nfs42_proc_allocate(filep, offset, len);
}

static loff_t nfs42_remap_file_range(struct file *src_file, loff_t src_off,
		struct file *dst_file, loff_t dst_off, loff_t count,
		unsigned int remap_flags)
{
	struct inode *dst_inode = file_inode(dst_file);
	struct nfs_server *server = NFS_SERVER(dst_inode);
	struct inode *src_inode = file_inode(src_file);
	unsigned int bs = server->clone_blksize;
	bool same_inode = false;
	int ret;

	/* NFS does not support deduplication. */
	if (remap_flags & REMAP_FILE_DEDUP)
		return -EOPNOTSUPP;

	if (remap_flags & ~REMAP_FILE_ADVISORY)
		return -EINVAL;

	if (IS_SWAPFILE(dst_inode) || IS_SWAPFILE(src_inode))
		return -ETXTBSY;

	/* check alignment w.r.t. clone_blksize */
	ret = -EINVAL;
	if (bs) {
		if (!IS_ALIGNED(src_off, bs) || !IS_ALIGNED(dst_off, bs))
			goto out;
		if (!IS_ALIGNED(count, bs) && i_size_read(src_inode) != (src_off + count))
			goto out;
	}

	if (src_inode == dst_inode)
		same_inode = true;

	/* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */
	if (same_inode) {
		inode_lock(src_inode);
	} else if (dst_inode < src_inode) {
		inode_lock_nested(dst_inode, I_MUTEX_PARENT);
		inode_lock_nested(src_inode, I_MUTEX_CHILD);
	} else {
		inode_lock_nested(src_inode, I_MUTEX_PARENT);
		inode_lock_nested(dst_inode, I_MUTEX_CHILD);
	}

	/* flush all pending writes on both src and dst so that server
	 * has the latest data */
	ret = nfs_sync_inode(src_inode);
	if (ret)
		goto out_unlock;
	ret = nfs_sync_inode(dst_inode);
	if (ret)
		goto out_unlock;

	ret = nfs42_proc_clone(src_file, dst_file, src_off, dst_off, count);

	/* truncate inode page cache of the dst range so that future reads can fetch
	 * new data from server */
	if (!ret)
		truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1);

out_unlock:
	if (same_inode) {
		inode_unlock(src_inode);
	} else if (dst_inode < src_inode) {
		inode_unlock(src_inode);
		inode_unlock(dst_inode);
	} else {
		inode_unlock(dst_inode);
		inode_unlock(src_inode);
	}
out:
	return ret < 0 ? ret : count;
}

static int read_name_gen = 1;
#define SSC_READ_NAME_BODY "ssc_read_%d"

static struct file *__nfs42_ssc_open(struct vfsmount *ss_mnt,
		struct nfs_fh *src_fh, nfs4_stateid *stateid)
{
	struct nfs_fattr *fattr = nfs_alloc_fattr();
	struct file *filep, *res;
	struct nfs_server *server;
	struct inode *r_ino = NULL;
	struct nfs_open_context *ctx;
	struct nfs4_state_owner *sp;
	char *read_name = NULL;
	int len, status = 0;

	server = NFS_SB(ss_mnt->mnt_sb);

	if (!fattr)
		return ERR_PTR(-ENOMEM);

	status = nfs4_proc_getattr(server, src_fh, fattr, NULL);
	if (status < 0) {
		res = ERR_PTR(status);
		goto out;
	}

	if (!S_ISREG(fattr->mode)) {
		res = ERR_PTR(-EBADF);
		goto out;
	}

	res = ERR_PTR(-ENOMEM);
	len = strlen(SSC_READ_NAME_BODY) + 16;
	read_name = kzalloc(len, GFP_KERNEL);
	if (read_name == NULL)
		goto out;
	snprintf(read_name, len, SSC_READ_NAME_BODY, read_name_gen++);

	r_ino = nfs_fhget(ss_mnt->mnt_sb, src_fh, fattr);
	if (IS_ERR(r_ino)) {
		res = ERR_CAST(r_ino);
		goto out_free_name;
	}

	filep = alloc_file_pseudo(r_ino, ss_mnt, read_name, O_RDONLY,
				     r_ino->i_fop);
	if (IS_ERR(filep)) {
		res = ERR_CAST(filep);
		iput(r_ino);
		goto out_free_name;
	}

	ctx = alloc_nfs_open_context(filep->f_path.dentry,
				     flags_to_mode(filep->f_flags), filep);
	if (IS_ERR(ctx)) {
		res = ERR_CAST(ctx);
		goto out_filep;
	}

	res = ERR_PTR(-EINVAL);
	sp = nfs4_get_state_owner(server, ctx->cred, GFP_KERNEL);
	if (sp == NULL)
		goto out_ctx;

	ctx->state = nfs4_get_open_state(r_ino, sp);
	if (ctx->state == NULL)
		goto out_stateowner;

	set_bit(NFS_SRV_SSC_COPY_STATE, &ctx->state->flags);
	memcpy(&ctx->state->open_stateid.other, &stateid->other,
	       NFS4_STATEID_OTHER_SIZE);
	update_open_stateid(ctx->state, stateid, NULL, filep->f_mode);
	set_bit(NFS_OPEN_STATE, &ctx->state->flags);

	nfs_file_set_open_context(filep, ctx);
	put_nfs_open_context(ctx);

	file_ra_state_init(&filep->f_ra, filep->f_mapping->host->i_mapping);
	res = filep;
out_free_name:
	kfree(read_name);
out:
	nfs_free_fattr(fattr);
	return res;
out_stateowner:
	nfs4_put_state_owner(sp);
out_ctx:
	put_nfs_open_context(ctx);
out_filep:
	fput(filep);
	goto out_free_name;
}

static void __nfs42_ssc_close(struct file *filep)
{
	struct nfs_open_context *ctx = nfs_file_open_context(filep);

	ctx->state->flags = 0;
}

static const struct nfs4_ssc_client_ops nfs4_ssc_clnt_ops_tbl = {
	.sco_open = __nfs42_ssc_open,
	.sco_close = __nfs42_ssc_close,
};

/**
 * nfs42_ssc_register_ops - Wrapper to register NFS_V4 ops in nfs_common
 *
 * Return values:
 *   None
 */
void nfs42_ssc_register_ops(void)
{
	nfs42_ssc_register(&nfs4_ssc_clnt_ops_tbl);
}

/**
 * nfs42_ssc_unregister_ops - wrapper to un-register NFS_V4 ops in nfs_common
 *
 * Return values:
 *   None.
 */
void nfs42_ssc_unregister_ops(void)
{
	nfs42_ssc_unregister(&nfs4_ssc_clnt_ops_tbl);
}
#endif /* CONFIG_NFS_V4_2 */

static int nfs4_setlease(struct file *file, long arg, struct file_lock **lease,
			 void **priv)
{
	return nfs4_proc_setlease(file, arg, lease, priv);
}

const struct file_operations nfs4_file_operations = {
	.read_iter	= nfs_file_read,
	.write_iter	= nfs_file_write,
	.mmap		= nfs_file_mmap,
	.open		= nfs4_file_open,
	.flush		= nfs4_file_flush,
	.release	= nfs_file_release,
	.fsync		= nfs_file_fsync,
	.lock		= nfs_lock,
	.flock		= nfs_flock,
	.splice_read	= generic_file_splice_read,
	.splice_write	= iter_file_splice_write,
	.check_flags	= nfs_check_flags,
	.setlease	= nfs4_setlease,
#ifdef CONFIG_NFS_V4_2
	.copy_file_range = nfs4_copy_file_range,
	.llseek		= nfs4_file_llseek,
	.fallocate	= nfs42_fallocate,
	.remap_file_range = nfs42_remap_file_range,
#else
	.llseek		= nfs_file_llseek,
#endif
};
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
ce4ef7c0 BS	2	/*
	3	* linux/fs/nfs/file.c
	4	*
	5	* Copyright (C) 1992 Rick Sladkey
	6	*/
f4ac1674	7	#include <linux/fs.h>
bea51b30	8	#include <linux/file.h>
f4ac1674	9	#include <linux/falloc.h>
f2aedb71	10	#include <linux/mount.h>
ce4ef7c0	11	#include <linux/nfs_fs.h>
0cfcd405	12	#include <linux/nfs_ssc.h>
5445b1fb	13	#include "delegation.h"
ce4ef7c0	14	#include "internal.h"
5445b1fb	15	#include "iostat.h"
a4ff1468	16	#include "fscache.h"
ce4ef7c0 BS	17	#include "pnfs.h"
ce4ef7c0 BS	18
81b79afb TM	19	#include "nfstrace.h"
81b79afb TM	20
1c6dcbe5 AS	21	#ifdef CONFIG_NFS_V4_2
	22	#include "nfs42.h"
	23	#endif
	24
ce4ef7c0 BS	25	#define NFSDBG_FACILITY NFSDBG_FILE
	26
	27	static int
	28	nfs4_file_open(struct inode inode, struct file filp)
	29	{
	30	struct nfs_open_context *ctx;
be62a1a8	31	struct dentry *dentry = file_dentry(filp);
ce4ef7c0 BS	32	struct dentry *parent = NULL;
	33	struct inode *dir;
	34	unsigned openflags = filp->f_flags;
	35	struct iattr attr;
	36	int err;
	37
ce4ef7c0 BS	38	/*
	39	* If no cached dentry exists or if it's negative, NFSv4 handled the
	40	* opens in ->lookup() or ->create().
	41	*
	42	* We only get this far for a cached positive dentry. We skipped
	43	* revalidation, so handle it here by dropping the dentry and returning
	44	* -EOPENSTALE. The VFS will retry the lookup/create/open.
	45	*/
	46
6de1472f	47	dprintk("NFS: open file(%pd2)\n", dentry);
ce4ef7c0	48
18a60089 BC	49	err = nfs_check_flags(openflags);
	50	if (err)
	51	return err;
	52
ce4ef7c0 BS	53	/* We can't create new files here */
	54	openflags &= ~(O_CREAT\|O_EXCL);
	55
	56	parent = dget_parent(dentry);
2b0143b5	57	dir = d_inode(parent);
ce4ef7c0	58
6f1c1d95 C	59	ctx = alloc_nfs_open_context(file_dentry(filp),
6f1c1d95 C	60	flags_to_mode(openflags), filp);
ce4ef7c0 BS	61	err = PTR_ERR(ctx);
	62	if (IS_ERR(ctx))
	63	goto out;
	64
	65	attr.ia_valid = ATTR_OPEN;
	66	if (openflags & O_TRUNC) {
	67	attr.ia_valid \|= ATTR_SIZE;
	68	attr.ia_size = 0;
8b7d9d09	69	filemap_write_and_wait(inode->i_mapping);
ce4ef7c0 BS	70	}
ce4ef7c0 BS	71
c5c3fb5f	72	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, NULL);
ce4ef7c0 BS	73	if (IS_ERR(inode)) {
	74	err = PTR_ERR(inode);
	75	switch (err) {
ce4ef7c0	76	default:
90cf500e TM	77	goto out_put_ctx;
	78	case -ENOENT:
	79	case -ESTALE:
	80	case -EISDIR:
	81	case -ENOTDIR:
	82	case -ELOOP:
ce4ef7c0 BS	83	goto out_drop;
	84	}
	85	}
2b0143b5	86	if (inode != d_inode(dentry))
ce4ef7c0 BS	87	goto out_drop;
ce4ef7c0 BS	88
ce4ef7c0	89	nfs_file_set_open_context(filp, ctx);
f1fe29b4	90	nfs_fscache_open_file(inode, filp);
ce4ef7c0	91	err = 0;
5ee3d10f	92	filp->f_mode \|= FMODE_CAN_ODIRECT;
ce4ef7c0 BS	93
	94	out_put_ctx:
	95	put_nfs_open_context(ctx);
	96	out:
	97	dput(parent);
	98	return err;
	99
	100	out_drop:
	101	d_drop(dentry);
	102	err = -EOPENSTALE;
	103	goto out_put_ctx;
	104	}
	105
5445b1fb TM	106	/*
	107	* Flush all dirty pages, and check for write errors.
	108	*/
	109	static int
	110	nfs4_file_flush(struct file *file, fl_owner_t id)
	111	{
	112	struct inode *inode = file_inode(file);
67dd23f9	113	errseq_t since;
5445b1fb TM	114
	115	dprintk("NFS: flush(%pD2)\n", file);
	116
	117	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
	118	if ((file->f_mode & FMODE_WRITE) == 0)
	119	return 0;
	120
	121	/*
	122	* If we're holding a write delegation, then check if we're required
	123	* to flush the i/o on close. If not, then just start the i/o now.
	124	*/
	125	if (!nfs4_delegation_flush_on_close(inode))
	126	return filemap_fdatawrite(file->f_mapping);
	127
	128	/* Flush writes to the server and return any errors */
67dd23f9 SM	129	since = filemap_sample_wb_err(file->f_mapping);
	130	nfs_wb_all(inode);
	131	return filemap_check_wb_err(file->f_mapping, since);
5445b1fb TM	132	}
5445b1fb TM	133
1c6dcbe5	134	#ifdef CONFIG_NFS_V4_2
64bf5ff5 DC	135	static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
	136	struct file *file_out, loff_t pos_out,
	137	size_t count, unsigned int flags)
2e72448b	138	{
0491567b	139	struct nfs42_copy_notify_res *cn_resp = NULL;
1d38f3f0 OK	140	struct nl4_server *nss = NULL;
1d38f3f0 OK	141	nfs4_stateid *cnrs = NULL;
0491567b	142	ssize_t ret;
12751010	143	bool sync = false;
0491567b	144
5dae222a	145	/* Only offload copy if superblock is the same */
8dff1df5	146	if (file_in->f_op != &nfs4_file_operations)
5dae222a	147	return -EXDEV;
d8a6ad91 DW	148	if (!nfs_server_capable(file_inode(file_out), NFS_CAP_COPY) \|\|
d8a6ad91 DW	149	!nfs_server_capable(file_inode(file_in), NFS_CAP_COPY))
0769663b	150	return -EOPNOTSUPP;
837bb1d7	151	if (file_inode(file_in) == file_inode(file_out))
0769663b	152	return -EOPNOTSUPP;
12751010 OK	153	/* if the copy size if smaller than 2 RPC payloads, make it
	154	* synchronous
	155	*/
	156	if (count <= 2 * NFS_SERVER(file_inode(file_in))->rsize)
	157	sync = true;
0e65a32c	158	retry:
0491567b	159	if (!nfs42_files_from_same_server(file_in, file_out)) {
ca7d1d1a DN	160	/*
	161	* for inter copy, if copy size is too small
	162	* then fallback to generic copy.
12751010	163	*/
ca7d1d1a	164	if (sync)
12751010	165	return -EOPNOTSUPP;
0491567b	166	cn_resp = kzalloc(sizeof(struct nfs42_copy_notify_res),
4fb547be	167	GFP_KERNEL);
0491567b OK	168	if (unlikely(cn_resp == NULL))
	169	return -ENOMEM;
	170
	171	ret = nfs42_proc_copy_notify(file_in, file_out, cn_resp);
	172	if (ret) {
	173	ret = -EOPNOTSUPP;
	174	goto out;
	175	}
1d38f3f0 OK	176	nss = &cn_resp->cnr_src;
1d38f3f0 OK	177	cnrs = &cn_resp->cnr_stateid;
0491567b	178	}
1d38f3f0	179	ret = nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count,
12751010	180	nss, cnrs, sync);
0491567b	181	out:
98c27f27 TR	182	kfree(cn_resp);
98c27f27 TR	183
0e65a32c OK	184	if (ret == -EAGAIN)
0e65a32c OK	185	goto retry;
0491567b	186	return ret;
2e72448b AS	187	}
2e72448b AS	188
64bf5ff5 DC	189	static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
	190	struct file *file_out, loff_t pos_out,
	191	size_t count, unsigned int flags)
	192	{
	193	ssize_t ret;
	194
	195	ret = __nfs4_copy_file_range(file_in, pos_in, file_out, pos_out, count,
	196	flags);
5dae222a	197	if (ret == -EOPNOTSUPP \|\| ret == -EXDEV)
64bf5ff5 DC	198	ret = generic_copy_file_range(file_in, pos_in, file_out,
	199	pos_out, count, flags);
	200	return ret;
	201	}
	202
1c6dcbe5 AS	203	static loff_t nfs4_file_llseek(struct file *filep, loff_t offset, int whence)
	204	{
	205	loff_t ret;
	206
	207	switch (whence) {
	208	case SEEK_HOLE:
	209	case SEEK_DATA:
	210	ret = nfs42_proc_llseek(filep, offset, whence);
e67afa7e	211	if (ret != -EOPNOTSUPP)
1c6dcbe5	212	return ret;
df561f66	213	fallthrough;
1c6dcbe5 AS	214	default:
	215	return nfs_file_llseek(filep, offset, whence);
	216	}
	217	}
f4ac1674 AS	218
	219	static long nfs42_fallocate(struct file *filep, int mode, loff_t offset, loff_t len)
	220	{
	221	struct inode *inode = file_inode(filep);
	222	long ret;
	223
	224	if (!S_ISREG(inode->i_mode))
	225	return -EOPNOTSUPP;
	226
624bd5b7	227	if ((mode != 0) && (mode != (FALLOC_FL_PUNCH_HOLE \| FALLOC_FL_KEEP_SIZE)))
f4ac1674 AS	228	return -EOPNOTSUPP;
	229
	230	ret = inode_newsize_ok(inode, offset + len);
	231	if (ret < 0)
	232	return ret;
	233
624bd5b7	234	if (mode & FALLOC_FL_PUNCH_HOLE)
f830f7dd AS	235	return nfs42_proc_deallocate(filep, offset, len);
f830f7dd AS	236	return nfs42_proc_allocate(filep, offset, len);
f4ac1674	237	}
bea51b30	238
42ec3d4c DW	239	static loff_t nfs42_remap_file_range(struct file *src_file, loff_t src_off,
42ec3d4c DW	240	struct file *dst_file, loff_t dst_off, loff_t count,
2e5dfc99	241	unsigned int remap_flags)
bea51b30 PT	242	{
bea51b30 PT	243	struct inode *dst_inode = file_inode(dst_file);
811b7b85	244	struct nfs_server *server = NFS_SERVER(dst_inode);
04b38d60	245	struct inode *src_inode = file_inode(src_file);
811b7b85	246	unsigned int bs = server->clone_blksize;
21fad313	247	bool same_inode = false;
bea51b30 PT	248	int ret;
bea51b30 PT	249
9026b3a9 DW	250	/* NFS does not support deduplication. */
	251	if (remap_flags & REMAP_FILE_DEDUP)
	252	return -EOPNOTSUPP;
	253
	254	if (remap_flags & ~REMAP_FILE_ADVISORY)
2e5dfc99 DW	255	return -EINVAL;
2e5dfc99 DW	256
f5fdf124 MZ	257	if (IS_SWAPFILE(dst_inode) \|\| IS_SWAPFILE(src_inode))
	258	return -ETXTBSY;
	259
811b7b85 PT	260	/* check alignment w.r.t. clone_blksize */
	261	ret = -EINVAL;
	262	if (bs) {
	263	if (!IS_ALIGNED(src_off, bs) \|\| !IS_ALIGNED(dst_off, bs))
04b38d60	264	goto out;
811b7b85	265	if (!IS_ALIGNED(count, bs) && i_size_read(src_inode) != (src_off + count))
04b38d60	266	goto out;
811b7b85 PT	267	}
811b7b85 PT	268
04b38d60 CH	269	if (src_inode == dst_inode)
04b38d60 CH	270	same_inode = true;
21fad313	271
bea51b30	272	/* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */
21fad313	273	if (same_inode) {
5955102c	274	inode_lock(src_inode);
21fad313	275	} else if (dst_inode < src_inode) {
5955102c AV	276	inode_lock_nested(dst_inode, I_MUTEX_PARENT);
5955102c AV	277	inode_lock_nested(src_inode, I_MUTEX_CHILD);
bea51b30	278	} else {
5955102c AV	279	inode_lock_nested(src_inode, I_MUTEX_PARENT);
5955102c AV	280	inode_lock_nested(dst_inode, I_MUTEX_CHILD);
bea51b30 PT	281	}
	282
	283	/* flush all pending writes on both src and dst so that server
	284	* has the latest data */
	285	ret = nfs_sync_inode(src_inode);
	286	if (ret)
	287	goto out_unlock;
	288	ret = nfs_sync_inode(dst_inode);
	289	if (ret)
	290	goto out_unlock;
	291
04b38d60	292	ret = nfs42_proc_clone(src_file, dst_file, src_off, dst_off, count);
bea51b30 PT	293
	294	/* truncate inode page cache of the dst range so that future reads can fetch
	295	* new data from server */
	296	if (!ret)
	297	truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1);
	298
	299	out_unlock:
21fad313	300	if (same_inode) {
5955102c	301	inode_unlock(src_inode);
21fad313	302	} else if (dst_inode < src_inode) {
5955102c AV	303	inode_unlock(src_inode);
5955102c AV	304	inode_unlock(dst_inode);
bea51b30	305	} else {
5955102c AV	306	inode_unlock(dst_inode);
5955102c AV	307	inode_unlock(src_inode);
bea51b30	308	}
04b38d60	309	out:
42ec3d4c	310	return ret < 0 ? ret : count;
bea51b30	311	}
ec4b0925 OK	312
	313	static int read_name_gen = 1;
	314	#define SSC_READ_NAME_BODY "ssc_read_%d"
	315
0cfcd405 DN	316	static struct file __nfs42_ssc_open(struct vfsmount ss_mnt,
0cfcd405 DN	317	struct nfs_fh src_fh, nfs4_stateid stateid)
ec4b0925	318	{
156cd285	319	struct nfs_fattr *fattr = nfs_alloc_fattr();
ec4b0925 OK	320	struct file filep, res;
	321	struct nfs_server *server;
	322	struct inode *r_ino = NULL;
	323	struct nfs_open_context *ctx;
	324	struct nfs4_state_owner *sp;
f751c545	325	char *read_name = NULL;
ec4b0925 OK	326	int len, status = 0;
ec4b0925 OK	327
f6957b71	328	server = NFS_SB(ss_mnt->mnt_sb);
ec4b0925	329
156cd285 TM	330	if (!fattr)
156cd285 TM	331	return ERR_PTR(-ENOMEM);
ec4b0925	332
2ef61e0e	333	status = nfs4_proc_getattr(server, src_fh, fattr, NULL);
ec4b0925 OK	334	if (status < 0) {
	335	res = ERR_PTR(status);
	336	goto out;
	337	}
	338
fcfc8be1 OK	339	if (!S_ISREG(fattr->mode)) {
	340	res = ERR_PTR(-EBADF);
	341	goto out;
	342	}
	343
ec4b0925 OK	344	res = ERR_PTR(-ENOMEM);
ec4b0925 OK	345	len = strlen(SSC_READ_NAME_BODY) + 16;
4fb547be	346	read_name = kzalloc(len, GFP_KERNEL);
ec4b0925 OK	347	if (read_name == NULL)
	348	goto out;
	349	snprintf(read_name, len, SSC_READ_NAME_BODY, read_name_gen++);
	350
f6957b71	351	r_ino = nfs_fhget(ss_mnt->mnt_sb, src_fh, fattr);
ec4b0925 OK	352	if (IS_ERR(r_ino)) {
ec4b0925 OK	353	res = ERR_CAST(r_ino);
f751c545	354	goto out_free_name;
ec4b0925 OK	355	}
ec4b0925 OK	356
aa97a3ef	357	filep = alloc_file_pseudo(r_ino, ss_mnt, read_name, O_RDONLY,
ec4b0925 OK	358	r_ino->i_fop);
	359	if (IS_ERR(filep)) {
	360	res = ERR_CAST(filep);
fcfc8be1	361	iput(r_ino);
f751c545	362	goto out_free_name;
ec4b0925	363	}
ec4b0925	364
6f1c1d95 C	365	ctx = alloc_nfs_open_context(filep->f_path.dentry,
6f1c1d95 C	366	flags_to_mode(filep->f_flags), filep);
ec4b0925 OK	367	if (IS_ERR(ctx)) {
	368	res = ERR_CAST(ctx);
	369	goto out_filep;
	370	}
	371
	372	res = ERR_PTR(-EINVAL);
	373	sp = nfs4_get_state_owner(server, ctx->cred, GFP_KERNEL);
	374	if (sp == NULL)
	375	goto out_ctx;
	376
	377	ctx->state = nfs4_get_open_state(r_ino, sp);
	378	if (ctx->state == NULL)
	379	goto out_stateowner;
	380
0b9018b9	381	set_bit(NFS_SRV_SSC_COPY_STATE, &ctx->state->flags);
ec4b0925 OK	382	memcpy(&ctx->state->open_stateid.other, &stateid->other,
	383	NFS4_STATEID_OTHER_SIZE);
	384	update_open_stateid(ctx->state, stateid, NULL, filep->f_mode);
fe8eb820	385	set_bit(NFS_OPEN_STATE, &ctx->state->flags);
ec4b0925 OK	386
	387	nfs_file_set_open_context(filep, ctx);
	388	put_nfs_open_context(ctx);
	389
	390	file_ra_state_init(&filep->f_ra, filep->f_mapping->host->i_mapping);
	391	res = filep;
f751c545 OK	392	out_free_name:
f751c545 OK	393	kfree(read_name);
ec4b0925	394	out:
156cd285	395	nfs_free_fattr(fattr);
ec4b0925 OK	396	return res;
	397	out_stateowner:
	398	nfs4_put_state_owner(sp);
	399	out_ctx:
	400	put_nfs_open_context(ctx);
	401	out_filep:
	402	fput(filep);
f751c545	403	goto out_free_name;
ec4b0925	404	}
0cfcd405 DN	405
0cfcd405 DN	406	static void __nfs42_ssc_close(struct file *filep)
ec4b0925 OK	407	{
	408	struct nfs_open_context *ctx = nfs_file_open_context(filep);
	409
	410	ctx->state->flags = 0;
	411	}
0cfcd405 DN	412
	413	static const struct nfs4_ssc_client_ops nfs4_ssc_clnt_ops_tbl = {
	414	.sco_open = __nfs42_ssc_open,
	415	.sco_close = __nfs42_ssc_close,
	416	};
	417
	418	/**
	419	* nfs42_ssc_register_ops - Wrapper to register NFS_V4 ops in nfs_common
	420	*
	421	* Return values:
	422	* None
	423	*/
	424	void nfs42_ssc_register_ops(void)
	425	{
	426	nfs42_ssc_register(&nfs4_ssc_clnt_ops_tbl);
	427	}
	428
	429	/**
	430	* nfs42_ssc_unregister_ops - wrapper to un-register NFS_V4 ops in nfs_common
	431	*
	432	* Return values:
	433	* None.
	434	*/
	435	void nfs42_ssc_unregister_ops(void)
	436	{
	437	nfs42_ssc_unregister(&nfs4_ssc_clnt_ops_tbl);
	438	}
6b7153da	439	#endif /* CONFIG_NFS_V4_2 */
bea51b30	440
e93a5e93 TM	441	static int nfs4_setlease(struct file file, long arg, struct file_lock *lease,
	442	void **priv)
	443	{
	444	return nfs4_proc_setlease(file, arg, lease, priv);
	445	}
	446
ce4ef7c0	447	const struct file_operations nfs4_file_operations = {
3aa2d199	448	.read_iter = nfs_file_read,
edaf4369	449	.write_iter = nfs_file_write,
ce4ef7c0 BS	450	.mmap = nfs_file_mmap,
ce4ef7c0 BS	451	.open = nfs4_file_open,
5445b1fb	452	.flush = nfs4_file_flush,
ce4ef7c0	453	.release = nfs_file_release,
4ff79bc7	454	.fsync = nfs_file_fsync,
ce4ef7c0 BS	455	.lock = nfs_lock,
ce4ef7c0 BS	456	.flock = nfs_flock,
82c156f8	457	.splice_read = generic_file_splice_read,
4da54c21	458	.splice_write = iter_file_splice_write,
ce4ef7c0	459	.check_flags = nfs_check_flags,
e93a5e93	460	.setlease = nfs4_setlease,
6b7153da	461	#ifdef CONFIG_NFS_V4_2
2e72448b	462	.copy_file_range = nfs4_copy_file_range,
6b7153da CH	463	.llseek = nfs4_file_llseek,
6b7153da CH	464	.fallocate = nfs42_fallocate,
2e5dfc99	465	.remap_file_range = nfs42_remap_file_range,
6b7153da CH	466	#else
	467	.llseek = nfs_file_llseek,
	468	#endif
ce4ef7c0	469	};