[linux-2.6-block.git] / fs / nfs / nfs4file.c

/*
 *  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 <uapi/linux/btrfs.h>	/* BTRFS_IOC_CLONE/BTRFS_IOC_CLONE_RANGE */
#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);
	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;
		nfs_sync_inode(inode);
	}

	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 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;
	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	= nfs_file_splice_read,
	.splice_write	= iter_file_splice_write,
	.check_flags	= nfs_check_flags,
	.setlease	= simple_nosetlease,
#ifdef CONFIG_NFS_V4_2
	.llseek		= nfs4_file_llseek,
	.fallocate	= nfs42_fallocate,
	.clone_file_range = nfs42_clone_file_range,
#else
	.llseek		= nfs_file_llseek,
#endif
};
Commit	Line	Data
ce4ef7c0 BS	1	/*
	2	* linux/fs/nfs/file.c
	3	*
	4	* Copyright (C) 1992 Rick Sladkey
	5	*/
f4ac1674	6	#include <linux/fs.h>
bea51b30	7	#include <linux/file.h>
f4ac1674	8	#include <linux/falloc.h>
ce4ef7c0	9	#include <linux/nfs_fs.h>
0f42a6a9	10	#include <uapi/linux/btrfs.h> /* BTRFS_IOC_CLONE/BTRFS_IOC_CLONE_RANGE */
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
be62a1a8	60	ctx = alloc_nfs_open_context(file_dentry(filp), filp->f_mode);
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;
9e1681c2	69	nfs_sync_inode(inode);
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 AS	131	#ifdef CONFIG_NFS_V4_2
	132	static loff_t nfs4_file_llseek(struct file *filep, loff_t offset, int whence)
	133	{
	134	loff_t ret;
	135
	136	switch (whence) {
	137	case SEEK_HOLE:
	138	case SEEK_DATA:
	139	ret = nfs42_proc_llseek(filep, offset, whence);
	140	if (ret != -ENOTSUPP)
	141	return ret;
	142	default:
	143	return nfs_file_llseek(filep, offset, whence);
	144	}
	145	}
f4ac1674 AS	146
	147	static long nfs42_fallocate(struct file *filep, int mode, loff_t offset, loff_t len)
	148	{
	149	struct inode *inode = file_inode(filep);
	150	long ret;
	151
	152	if (!S_ISREG(inode->i_mode))
	153	return -EOPNOTSUPP;
	154
624bd5b7	155	if ((mode != 0) && (mode != (FALLOC_FL_PUNCH_HOLE \| FALLOC_FL_KEEP_SIZE)))
f4ac1674 AS	156	return -EOPNOTSUPP;
	157
	158	ret = inode_newsize_ok(inode, offset + len);
	159	if (ret < 0)
	160	return ret;
	161
624bd5b7	162	if (mode & FALLOC_FL_PUNCH_HOLE)
f830f7dd AS	163	return nfs42_proc_deallocate(filep, offset, len);
f830f7dd AS	164	return nfs42_proc_allocate(filep, offset, len);
f4ac1674	165	}
bea51b30	166
04b38d60 CH	167	static int nfs42_clone_file_range(struct file *src_file, loff_t src_off,
04b38d60 CH	168	struct file *dst_file, loff_t dst_off, u64 count)
bea51b30 PT	169	{
bea51b30 PT	170	struct inode *dst_inode = file_inode(dst_file);
811b7b85	171	struct nfs_server *server = NFS_SERVER(dst_inode);
04b38d60	172	struct inode *src_inode = file_inode(src_file);
811b7b85	173	unsigned int bs = server->clone_blksize;
21fad313	174	bool same_inode = false;
bea51b30 PT	175	int ret;
bea51b30 PT	176
811b7b85 PT	177	/* check alignment w.r.t. clone_blksize */
	178	ret = -EINVAL;
	179	if (bs) {
	180	if (!IS_ALIGNED(src_off, bs) \|\| !IS_ALIGNED(dst_off, bs))
04b38d60	181	goto out;
811b7b85	182	if (!IS_ALIGNED(count, bs) && i_size_read(src_inode) != (src_off + count))
04b38d60	183	goto out;
811b7b85 PT	184	}
811b7b85 PT	185
04b38d60 CH	186	if (src_inode == dst_inode)
04b38d60 CH	187	same_inode = true;
21fad313	188
bea51b30	189	/* XXX: do we lock at all? what if server needs CB_RECALL_LAYOUT? */
21fad313	190	if (same_inode) {
5955102c	191	inode_lock(src_inode);
21fad313	192	} else if (dst_inode < src_inode) {
5955102c AV	193	inode_lock_nested(dst_inode, I_MUTEX_PARENT);
5955102c AV	194	inode_lock_nested(src_inode, I_MUTEX_CHILD);
bea51b30	195	} else {
5955102c AV	196	inode_lock_nested(src_inode, I_MUTEX_PARENT);
5955102c AV	197	inode_lock_nested(dst_inode, I_MUTEX_CHILD);
bea51b30 PT	198	}
	199
	200	/* flush all pending writes on both src and dst so that server
	201	* has the latest data */
	202	ret = nfs_sync_inode(src_inode);
	203	if (ret)
	204	goto out_unlock;
	205	ret = nfs_sync_inode(dst_inode);
	206	if (ret)
	207	goto out_unlock;
	208
04b38d60	209	ret = nfs42_proc_clone(src_file, dst_file, src_off, dst_off, count);
bea51b30 PT	210
	211	/* truncate inode page cache of the dst range so that future reads can fetch
	212	* new data from server */
	213	if (!ret)
	214	truncate_inode_pages_range(&dst_inode->i_data, dst_off, dst_off + count - 1);
	215
	216	out_unlock:
21fad313	217	if (same_inode) {
5955102c	218	inode_unlock(src_inode);
21fad313	219	} else if (dst_inode < src_inode) {
5955102c AV	220	inode_unlock(src_inode);
5955102c AV	221	inode_unlock(dst_inode);
bea51b30	222	} else {
5955102c AV	223	inode_unlock(dst_inode);
5955102c AV	224	inode_unlock(src_inode);
bea51b30	225	}
04b38d60	226	out:
bea51b30 PT	227	return ret;
bea51b30 PT	228	}
6b7153da	229	#endif /* CONFIG_NFS_V4_2 */
bea51b30	230
ce4ef7c0	231	const struct file_operations nfs4_file_operations = {
3aa2d199	232	.read_iter = nfs_file_read,
edaf4369	233	.write_iter = nfs_file_write,
ce4ef7c0 BS	234	.mmap = nfs_file_mmap,
ce4ef7c0 BS	235	.open = nfs4_file_open,
5445b1fb	236	.flush = nfs4_file_flush,
ce4ef7c0	237	.release = nfs_file_release,
4ff79bc7	238	.fsync = nfs_file_fsync,
ce4ef7c0 BS	239	.lock = nfs_lock,
	240	.flock = nfs_flock,
	241	.splice_read = nfs_file_splice_read,
4da54c21	242	.splice_write = iter_file_splice_write,
ce4ef7c0	243	.check_flags = nfs_check_flags,
1c994a09	244	.setlease = simple_nosetlease,
6b7153da CH	245	#ifdef CONFIG_NFS_V4_2
	246	.llseek = nfs4_file_llseek,
	247	.fallocate = nfs42_fallocate,
04b38d60	248	.clone_file_range = nfs42_clone_file_range,
6b7153da CH	249	#else
	250	.llseek = nfs_file_llseek,
	251	#endif
ce4ef7c0	252	};