[linux.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/nfs_fs.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;

	if ((openflags & O_ACCMODE) == 3)
		openflags--;

	/* 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), filp->f_mode, 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) {
		case -EPERM:
		case -EACCES:
		case -EDQUOT:
		case -ENOSPC:
		case -EROFS:
			goto out_put_ctx;
		default:
			goto out_drop;
		}
	}
	if (inode != d_inode(dentry))
		goto out_drop;

	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
	nfs_file_set_open_context(filp, ctx);
	nfs_fscache_open_file(inode, filp);
	err = 0;

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);

	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 */
	return vfs_fsync(file, 0);
}

#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)
{
	ssize_t ret;

	if (file_inode(file_in) == file_inode(file_out))
		return -EINVAL;
retry:
	ret = nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count);
	if (ret == -EAGAIN)
		goto retry;
	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 != -ENOTSUPP)
			return ret;
		/* Fall through */
	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 int nfs42_clone_file_range(struct file *src_file, loff_t src_off,
		struct file *dst_file, loff_t dst_off, u64 count)
{
	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;

	/* 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;
}
#endif /* CONFIG_NFS_V4_2 */

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	= simple_nosetlease,
#ifdef CONFIG_NFS_V4_2
	.copy_file_range = nfs4_copy_file_range,
	.llseek		= nfs4_file_llseek,
	.fallocate	= nfs42_fallocate,
	.clone_file_range = nfs42_clone_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>
ce4ef7c0	10	#include <linux/nfs_fs.h>
5445b1fb	11	#include "delegation.h"
ce4ef7c0	12	#include "internal.h"
5445b1fb	13	#include "iostat.h"
a4ff1468	14	#include "fscache.h"
ce4ef7c0 BS	15	#include "pnfs.h"
ce4ef7c0 BS	16
81b79afb TM	17	#include "nfstrace.h"
81b79afb TM	18
1c6dcbe5 AS	19	#ifdef CONFIG_NFS_V4_2
	20	#include "nfs42.h"
	21	#endif
	22
ce4ef7c0 BS	23	#define NFSDBG_FACILITY NFSDBG_FILE
	24
	25	static int
	26	nfs4_file_open(struct inode inode, struct file filp)
	27	{
	28	struct nfs_open_context *ctx;
be62a1a8	29	struct dentry *dentry = file_dentry(filp);
ce4ef7c0 BS	30	struct dentry *parent = NULL;
	31	struct inode *dir;
	32	unsigned openflags = filp->f_flags;
	33	struct iattr attr;
	34	int err;
	35
ce4ef7c0 BS	36	/*
	37	* If no cached dentry exists or if it's negative, NFSv4 handled the
	38	* opens in ->lookup() or ->create().
	39	*
	40	* We only get this far for a cached positive dentry. We skipped
	41	* revalidation, so handle it here by dropping the dentry and returning
	42	* -EOPENSTALE. The VFS will retry the lookup/create/open.
	43	*/
	44
6de1472f	45	dprintk("NFS: open file(%pd2)\n", dentry);
ce4ef7c0	46
18a60089 BC	47	err = nfs_check_flags(openflags);
	48	if (err)
	49	return err;
	50
ce4ef7c0 BS	51	if ((openflags & O_ACCMODE) == 3)
	52	openflags--;
	53
	54	/* We can't create new files here */
	55	openflags &= ~(O_CREAT\|O_EXCL);
	56
	57	parent = dget_parent(dentry);
2b0143b5	58	dir = d_inode(parent);
ce4ef7c0	59
532d4def	60	ctx = alloc_nfs_open_context(file_dentry(filp), filp->f_mode, 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) {
	76	case -EPERM:
	77	case -EACCES:
	78	case -EDQUOT:
	79	case -ENOSPC:
	80	case -EROFS:
	81	goto out_put_ctx;
	82	default:
	83	goto out_drop;
	84	}
	85	}
2b0143b5	86	if (inode != d_inode(dentry))
ce4ef7c0 BS	87	goto out_drop;
	88
	89	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
	90	nfs_file_set_open_context(filp, ctx);
f1fe29b4	91	nfs_fscache_open_file(inode, filp);
ce4ef7c0 BS	92	err = 0;
	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);
	113
	114	dprintk("NFS: flush(%pD2)\n", file);
	115
	116	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
	117	if ((file->f_mode & FMODE_WRITE) == 0)
	118	return 0;
	119
	120	/*
	121	* If we're holding a write delegation, then check if we're required
	122	* to flush the i/o on close. If not, then just start the i/o now.
	123	*/
	124	if (!nfs4_delegation_flush_on_close(inode))
	125	return filemap_fdatawrite(file->f_mapping);
	126
	127	/* Flush writes to the server and return any errors */
	128	return vfs_fsync(file, 0);
	129	}
	130
1c6dcbe5	131	#ifdef CONFIG_NFS_V4_2
2e72448b AS	132	static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
	133	struct file *file_out, loff_t pos_out,
	134	size_t count, unsigned int flags)
	135	{
e4648aa4 OK	136	ssize_t ret;
e4648aa4 OK	137
837bb1d7	138	if (file_inode(file_in) == file_inode(file_out))
2e72448b	139	return -EINVAL;
e4648aa4 OK	140	retry:
	141	ret = nfs42_proc_copy(file_in, pos_in, file_out, pos_out, count);
	142	if (ret == -EAGAIN)
	143	goto retry;
	144	return ret;
2e72448b AS	145	}
2e72448b AS	146
1c6dcbe5 AS	147	static loff_t nfs4_file_llseek(struct file *filep, loff_t offset, int whence)
	148	{
	149	loff_t ret;
	150
	151	switch (whence) {
	152	case SEEK_HOLE:
	153	case SEEK_DATA:
	154	ret = nfs42_proc_llseek(filep, offset, whence);
	155	if (ret != -ENOTSUPP)
	156	return ret;
01e03bdc	157	/* Fall through */
1c6dcbe5 AS	158	default:
	159	return nfs_file_llseek(filep, offset, whence);
	160	}
	161	}
f4ac1674 AS	162
	163	static long nfs42_fallocate(struct file *filep, int mode, loff_t offset, loff_t len)
	164	{
	165	struct inode *inode = file_inode(filep);
	166	long ret;
	167
	168	if (!S_ISREG(inode->i_mode))
	169	return -EOPNOTSUPP;
	170
624bd5b7	171	if ((mode != 0) && (mode != (FALLOC_FL_PUNCH_HOLE \| FALLOC_FL_KEEP_SIZE)))
f4ac1674 AS	172	return -EOPNOTSUPP;
	173
	174	ret = inode_newsize_ok(inode, offset + len);
	175	if (ret < 0)
	176	return ret;
	177
624bd5b7	178	if (mode & FALLOC_FL_PUNCH_HOLE)
f830f7dd AS	179	return nfs42_proc_deallocate(filep, offset, len);
f830f7dd AS	180	return nfs42_proc_allocate(filep, offset, len);
f4ac1674	181	}
bea51b30	182
04b38d60 CH	183	static int nfs42_clone_file_range(struct file *src_file, loff_t src_off,
04b38d60 CH	184	struct file *dst_file, loff_t dst_off, u64 count)
bea51b30 PT	185	{
bea51b30 PT	186	struct inode *dst_inode = file_inode(dst_file);
811b7b85	187	struct nfs_server *server = NFS_SERVER(dst_inode);
04b38d60	188	struct inode *src_inode = file_inode(src_file);
811b7b85	189	unsigned int bs = server->clone_blksize;
21fad313	190	bool same_inode = false;
bea51b30 PT	191	int ret;
bea51b30 PT	192
811b7b85 PT	193	/* check alignment w.r.t. clone_blksize */
	194	ret = -EINVAL;
	195	if (bs) {
	196	if (!IS_ALIGNED(src_off, bs) \|\| !IS_ALIGNED(dst_off, bs))
04b38d60	197	goto out;
811b7b85	198	if (!IS_ALIGNED(count, bs) && i_size_read(src_inode) != (src_off + count))
04b38d60	199	goto out;
811b7b85 PT	200	}
811b7b85 PT	201
04b38d60 CH	202	if (src_inode == dst_inode)
04b38d60 CH	203	same_inode = true;
21fad313	204
bea51b30	205	/* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */
21fad313	206	if (same_inode) {
5955102c	207	inode_lock(src_inode);
21fad313	208	} else if (dst_inode < src_inode) {
5955102c AV	209	inode_lock_nested(dst_inode, I_MUTEX_PARENT);
5955102c AV	210	inode_lock_nested(src_inode, I_MUTEX_CHILD);
bea51b30	211	} else {
5955102c AV	212	inode_lock_nested(src_inode, I_MUTEX_PARENT);
5955102c AV	213	inode_lock_nested(dst_inode, I_MUTEX_CHILD);
bea51b30 PT	214	}
	215
	216	/* flush all pending writes on both src and dst so that server
	217	* has the latest data */
	218	ret = nfs_sync_inode(src_inode);
	219	if (ret)
	220	goto out_unlock;
	221	ret = nfs_sync_inode(dst_inode);
	222	if (ret)
	223	goto out_unlock;
	224
04b38d60	225	ret = nfs42_proc_clone(src_file, dst_file, src_off, dst_off, count);
bea51b30 PT	226
	227	/* truncate inode page cache of the dst range so that future reads can fetch
	228	* new data from server */
	229	if (!ret)
	230	truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1);
	231
	232	out_unlock:
21fad313	233	if (same_inode) {
5955102c	234	inode_unlock(src_inode);
21fad313	235	} else if (dst_inode < src_inode) {
5955102c AV	236	inode_unlock(src_inode);
5955102c AV	237	inode_unlock(dst_inode);
bea51b30	238	} else {
5955102c AV	239	inode_unlock(dst_inode);
5955102c AV	240	inode_unlock(src_inode);
bea51b30	241	}
04b38d60	242	out:
bea51b30 PT	243	return ret;
bea51b30 PT	244	}
6b7153da	245	#endif /* CONFIG_NFS_V4_2 */
bea51b30	246
ce4ef7c0	247	const struct file_operations nfs4_file_operations = {
3aa2d199	248	.read_iter = nfs_file_read,
edaf4369	249	.write_iter = nfs_file_write,
ce4ef7c0 BS	250	.mmap = nfs_file_mmap,
ce4ef7c0 BS	251	.open = nfs4_file_open,
5445b1fb	252	.flush = nfs4_file_flush,
ce4ef7c0	253	.release = nfs_file_release,
4ff79bc7	254	.fsync = nfs_file_fsync,
ce4ef7c0 BS	255	.lock = nfs_lock,
ce4ef7c0 BS	256	.flock = nfs_flock,
82c156f8	257	.splice_read = generic_file_splice_read,
4da54c21	258	.splice_write = iter_file_splice_write,
ce4ef7c0	259	.check_flags = nfs_check_flags,
1c994a09	260	.setlease = simple_nosetlease,
6b7153da	261	#ifdef CONFIG_NFS_V4_2
2e72448b	262	.copy_file_range = nfs4_copy_file_range,
6b7153da CH	263	.llseek = nfs4_file_llseek,
6b7153da CH	264	.fallocate = nfs42_fallocate,
04b38d60	265	.clone_file_range = nfs42_clone_file_range,
6b7153da CH	266	#else
	267	.llseek = nfs_file_llseek,
	268	#endif
ce4ef7c0	269	};