[linux.git] / fs / nfs / file.c

/*
 *  linux/fs/nfs/file.c
 *
 *  Copyright (C) 1992  Rick Sladkey
 *
 *  Changes Copyright (C) 1994 by Florian La Roche
 *   - Do not copy data too often around in the kernel.
 *   - In nfs_file_read the return value of kmalloc wasn't checked.
 *   - Put in a better version of read look-ahead buffering. Original idea
 *     and implementation by Wai S Kok [email protected].
 *
 *  Expire cache on write to a file by Wai S Kok (Oct 1994).
 *
 *  Total rewrite of read side for new NFS buffer cache.. Linus.
 *
 *  nfs regular file handling functions
 */

#include <linux/time.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/fcntl.h>
#include <linux/stat.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_mount.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/smp_lock.h>

#include <asm/uaccess.h>
#include <asm/system.h>

#include "delegation.h"
#include "iostat.h"

#define NFSDBG_FACILITY		NFSDBG_FILE

static int nfs_file_open(struct inode *, struct file *);
static int nfs_file_release(struct inode *, struct file *);
static loff_t nfs_file_llseek(struct file *file, loff_t offset, int origin);
static int  nfs_file_mmap(struct file *, struct vm_area_struct *);
static ssize_t nfs_file_sendfile(struct file *, loff_t *, size_t, read_actor_t, void *);
static ssize_t nfs_file_read(struct kiocb *, char __user *, size_t, loff_t);
static ssize_t nfs_file_write(struct kiocb *, const char __user *, size_t, loff_t);
static int  nfs_file_flush(struct file *);
static int  nfs_fsync(struct file *, struct dentry *dentry, int datasync);
static int nfs_check_flags(int flags);
static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl);
static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl);

struct file_operations nfs_file_operations = {
	.llseek		= nfs_file_llseek,
	.read		= do_sync_read,
	.write		= do_sync_write,
	.aio_read		= nfs_file_read,
	.aio_write		= nfs_file_write,
	.mmap		= nfs_file_mmap,
	.open		= nfs_file_open,
	.flush		= nfs_file_flush,
	.release	= nfs_file_release,
	.fsync		= nfs_fsync,
	.lock		= nfs_lock,
	.flock		= nfs_flock,
	.sendfile	= nfs_file_sendfile,
	.check_flags	= nfs_check_flags,
};

struct inode_operations nfs_file_inode_operations = {
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
};

#ifdef CONFIG_NFS_V3
struct inode_operations nfs3_file_inode_operations = {
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
	.listxattr	= nfs3_listxattr,
	.getxattr	= nfs3_getxattr,
	.setxattr	= nfs3_setxattr,
	.removexattr	= nfs3_removexattr,
};
#endif  /* CONFIG_NFS_v3 */

/* Hack for future NFS swap support */
#ifndef IS_SWAPFILE
# define IS_SWAPFILE(inode)	(0)
#endif

static int nfs_check_flags(int flags)
{
	if ((flags & (O_APPEND | O_DIRECT)) == (O_APPEND | O_DIRECT))
		return -EINVAL;

	return 0;
}

/*
 * Open file
 */
static int
nfs_file_open(struct inode *inode, struct file *filp)
{
	int res;

	res = nfs_check_flags(filp->f_flags);
	if (res)
		return res;

	nfs_inc_stats(inode, NFSIOS_VFSOPEN);
	lock_kernel();
	res = NFS_SERVER(inode)->rpc_ops->file_open(inode, filp);
	unlock_kernel();
	return res;
}

static int
nfs_file_release(struct inode *inode, struct file *filp)
{
	/* Ensure that dirty pages are flushed out with the right creds */
	if (filp->f_mode & FMODE_WRITE)
		filemap_fdatawrite(filp->f_mapping);
	nfs_inc_stats(inode, NFSIOS_VFSRELEASE);
	return NFS_PROTO(inode)->file_release(inode, filp);
}

/**
 * nfs_revalidate_file - Revalidate the page cache & related metadata
 * @inode - pointer to inode struct
 * @file - pointer to file
 */
static int nfs_revalidate_file(struct inode *inode, struct file *filp)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	int retval = 0;

	if ((nfsi->cache_validity & (NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_ATTR))
			|| nfs_attribute_timeout(inode))
		retval = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
	nfs_revalidate_mapping(inode, filp->f_mapping);
	return 0;
}

/**
 * nfs_revalidate_size - Revalidate the file size
 * @inode - pointer to inode struct
 * @file - pointer to struct file
 *
 * Revalidates the file length. This is basically a wrapper around
 * nfs_revalidate_inode() that takes into account the fact that we may
 * have cached writes (in which case we don't care about the server's
 * idea of what the file length is), or O_DIRECT (in which case we
 * shouldn't trust the cache).
 */
static int nfs_revalidate_file_size(struct inode *inode, struct file *filp)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_inode *nfsi = NFS_I(inode);

	if (server->flags & NFS_MOUNT_NOAC)
		goto force_reval;
	if (filp->f_flags & O_DIRECT)
		goto force_reval;
	if (nfsi->npages != 0)
		return 0;
	if (!(nfsi->cache_validity & NFS_INO_REVAL_PAGECACHE) && !nfs_attribute_timeout(inode))
		return 0;
force_reval:
	return __nfs_revalidate_inode(server, inode);
}

static loff_t nfs_file_llseek(struct file *filp, loff_t offset, int origin)
{
	/* origin == SEEK_END => we must revalidate the cached file length */
	if (origin == 2) {
		struct inode *inode = filp->f_mapping->host;
		int retval = nfs_revalidate_file_size(inode, filp);
		if (retval < 0)
			return (loff_t)retval;
	}
	return remote_llseek(filp, offset, origin);
}

/*
 * Flush all dirty pages, and check for write errors.
 *
 */
static int
nfs_file_flush(struct file *file)
{
	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
	struct inode	*inode = file->f_dentry->d_inode;
	int		status;

	dfprintk(VFS, "nfs: flush(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino);

	if ((file->f_mode & FMODE_WRITE) == 0)
		return 0;
	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
	lock_kernel();
	/* Ensure that data+attribute caches are up to date after close() */
	status = nfs_wb_all(inode);
	if (!status) {
		status = ctx->error;
		ctx->error = 0;
		if (!status)
			nfs_revalidate_inode(NFS_SERVER(inode), inode);
	}
	unlock_kernel();
	return status;
}

static ssize_t
nfs_file_read(struct kiocb *iocb, char __user * buf, size_t count, loff_t pos)
{
	struct dentry * dentry = iocb->ki_filp->f_dentry;
	struct inode * inode = dentry->d_inode;
	ssize_t result;

#ifdef CONFIG_NFS_DIRECTIO
	if (iocb->ki_filp->f_flags & O_DIRECT)
		return nfs_file_direct_read(iocb, buf, count, pos);
#endif

	dfprintk(VFS, "nfs: read(%s/%s, %lu@%lu)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		(unsigned long) count, (unsigned long) pos);

	result = nfs_revalidate_file(inode, iocb->ki_filp);
	nfs_add_stats(inode, NFSIOS_NORMALREADBYTES, count);
	if (!result)
		result = generic_file_aio_read(iocb, buf, count, pos);
	return result;
}

static ssize_t
nfs_file_sendfile(struct file *filp, loff_t *ppos, size_t count,
		read_actor_t actor, void *target)
{
	struct dentry *dentry = filp->f_dentry;
	struct inode *inode = dentry->d_inode;
	ssize_t res;

	dfprintk(VFS, "nfs: sendfile(%s/%s, %lu@%Lu)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		(unsigned long) count, (unsigned long long) *ppos);

	res = nfs_revalidate_file(inode, filp);
	if (!res)
		res = generic_file_sendfile(filp, ppos, count, actor, target);
	return res;
}

static int
nfs_file_mmap(struct file * file, struct vm_area_struct * vma)
{
	struct dentry *dentry = file->f_dentry;
	struct inode *inode = dentry->d_inode;
	int	status;

	dfprintk(VFS, "nfs: mmap(%s/%s)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name);

	status = nfs_revalidate_file(inode, file);
	if (!status)
		status = generic_file_mmap(file, vma);
	return status;
}

/*
 * Flush any dirty pages for this process, and check for write errors.
 * The return status from this call provides a reliable indication of
 * whether any write errors occurred for this process.
 */
static int
nfs_fsync(struct file *file, struct dentry *dentry, int datasync)
{
	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
	struct inode *inode = dentry->d_inode;
	int status;

	dfprintk(VFS, "nfs: fsync(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino);

	nfs_inc_stats(inode, NFSIOS_VFSFSYNC);
	lock_kernel();
	status = nfs_wb_all(inode);
	if (!status) {
		status = ctx->error;
		ctx->error = 0;
	}
	unlock_kernel();
	return status;
}

/*
 * This does the "real" work of the write. The generic routine has
 * allocated the page, locked it, done all the page alignment stuff
 * calculations etc. Now we should just copy the data from user
 * space and write it back to the real medium..
 *
 * If the writer ends up delaying the write, the writer needs to
 * increment the page use counts until he is done with the page.
 */
static int nfs_prepare_write(struct file *file, struct page *page, unsigned offset, unsigned to)
{
	return nfs_flush_incompatible(file, page);
}

static int nfs_commit_write(struct file *file, struct page *page, unsigned offset, unsigned to)
{
	long status;

	lock_kernel();
	status = nfs_updatepage(file, page, offset, to-offset);
	unlock_kernel();
	return status;
}

static int nfs_invalidate_page(struct page *page, unsigned long offset)
{
	/* FIXME: we really should cancel any unstarted writes on this page */
	return 1;
}

static int nfs_release_page(struct page *page, gfp_t gfp)
{
	return !nfs_wb_page(page->mapping->host, page);
}

struct address_space_operations nfs_file_aops = {
	.readpage = nfs_readpage,
	.readpages = nfs_readpages,
	.set_page_dirty = __set_page_dirty_nobuffers,
	.writepage = nfs_writepage,
	.writepages = nfs_writepages,
	.prepare_write = nfs_prepare_write,
	.commit_write = nfs_commit_write,
	.invalidatepage = nfs_invalidate_page,
	.releasepage = nfs_release_page,
#ifdef CONFIG_NFS_DIRECTIO
	.direct_IO = nfs_direct_IO,
#endif
};

/* 
 * Write to a file (through the page cache).
 */
static ssize_t
nfs_file_write(struct kiocb *iocb, const char __user *buf, size_t count, loff_t pos)
{
	struct dentry * dentry = iocb->ki_filp->f_dentry;
	struct inode * inode = dentry->d_inode;
	ssize_t result;

#ifdef CONFIG_NFS_DIRECTIO
	if (iocb->ki_filp->f_flags & O_DIRECT)
		return nfs_file_direct_write(iocb, buf, count, pos);
#endif

	dfprintk(VFS, "nfs: write(%s/%s(%ld), %lu@%lu)\n",
		dentry->d_parent->d_name.name, dentry->d_name.name,
		inode->i_ino, (unsigned long) count, (unsigned long) pos);

	result = -EBUSY;
	if (IS_SWAPFILE(inode))
		goto out_swapfile;
	/*
	 * O_APPEND implies that we must revalidate the file length.
	 */
	if (iocb->ki_filp->f_flags & O_APPEND) {
		result = nfs_revalidate_file_size(inode, iocb->ki_filp);
		if (result)
			goto out;
	}
	nfs_revalidate_mapping(inode, iocb->ki_filp->f_mapping);

	result = count;
	if (!count)
		goto out;

	nfs_add_stats(inode, NFSIOS_NORMALWRITTENBYTES, count);
	result = generic_file_aio_write(iocb, buf, count, pos);
out:
	return result;

out_swapfile:
	printk(KERN_INFO "NFS: attempt to write to active swap file!\n");
	goto out;
}

static int do_getlk(struct file *filp, int cmd, struct file_lock *fl)
{
	struct file_lock cfl;
	struct inode *inode = filp->f_mapping->host;
	int status = 0;

	lock_kernel();
	/* Try local locking first */
	if (posix_test_lock(filp, fl, &cfl)) {
		locks_copy_lock(fl, &cfl);
		goto out;
	}

	if (nfs_have_delegation(inode, FMODE_READ))
		goto out_noconflict;

	if (NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM)
		goto out_noconflict;

	status = NFS_PROTO(inode)->lock(filp, cmd, fl);
out:
	unlock_kernel();
	return status;
out_noconflict:
	fl->fl_type = F_UNLCK;
	goto out;
}

static int do_vfs_lock(struct file *file, struct file_lock *fl)
{
	int res = 0;
	switch (fl->fl_flags & (FL_POSIX|FL_FLOCK)) {
		case FL_POSIX:
			res = posix_lock_file_wait(file, fl);
			break;
		case FL_FLOCK:
			res = flock_lock_file_wait(file, fl);
			break;
		default:
			BUG();
	}
	if (res < 0)
		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n",
				__FUNCTION__);
	return res;
}

static int do_unlk(struct file *filp, int cmd, struct file_lock *fl)
{
	struct inode *inode = filp->f_mapping->host;
	sigset_t oldset;
	int status;

	rpc_clnt_sigmask(NFS_CLIENT(inode), &oldset);
	/*
	 * Flush all pending writes before doing anything
	 * with locks..
	 */
	nfs_sync_mapping(filp->f_mapping);

	/* NOTE: special case
	 * 	If we're signalled while cleaning up locks on process exit, we
	 * 	still need to complete the unlock.
	 */
	lock_kernel();
	/* Use local locking if mounted with "-onolock" */
	if (!(NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM))
		status = NFS_PROTO(inode)->lock(filp, cmd, fl);
	else
		status = do_vfs_lock(filp, fl);
	unlock_kernel();
	rpc_clnt_sigunmask(NFS_CLIENT(inode), &oldset);
	return status;
}

static int do_setlk(struct file *filp, int cmd, struct file_lock *fl)
{
	struct inode *inode = filp->f_mapping->host;
	sigset_t oldset;
	int status;

	rpc_clnt_sigmask(NFS_CLIENT(inode), &oldset);
	/*
	 * Flush all pending writes before doing anything
	 * with locks..
	 */
	status = nfs_sync_mapping(filp->f_mapping);
	if (status != 0)
		goto out;

	lock_kernel();
	/* Use local locking if mounted with "-onolock" */
	if (!(NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM)) {
		status = NFS_PROTO(inode)->lock(filp, cmd, fl);
		/* If we were signalled we still need to ensure that
		 * we clean up any state on the server. We therefore
		 * record the lock call as having succeeded in order to
		 * ensure that locks_remove_posix() cleans it out when
		 * the process exits.
		 */
		if (status == -EINTR || status == -ERESTARTSYS)
			do_vfs_lock(filp, fl);
	} else
		status = do_vfs_lock(filp, fl);
	unlock_kernel();
	if (status < 0)
		goto out;
	/*
	 * Make sure we clear the cache whenever we try to get the lock.
	 * This makes locking act as a cache coherency point.
	 */
	nfs_sync_mapping(filp->f_mapping);
	nfs_zap_caches(inode);
out:
	rpc_clnt_sigunmask(NFS_CLIENT(inode), &oldset);
	return status;
}

/*
 * Lock a (portion of) a file
 */
static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl)
{
	struct inode * inode = filp->f_mapping->host;

	dprintk("NFS: nfs_lock(f=%s/%ld, t=%x, fl=%x, r=%Ld:%Ld)\n",
			inode->i_sb->s_id, inode->i_ino,
			fl->fl_type, fl->fl_flags,
			(long long)fl->fl_start, (long long)fl->fl_end);
	nfs_inc_stats(inode, NFSIOS_VFSLOCK);

	/* No mandatory locks over NFS */
	if ((inode->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID &&
	    fl->fl_type != F_UNLCK)
		return -ENOLCK;

	if (IS_GETLK(cmd))
		return do_getlk(filp, cmd, fl);
	if (fl->fl_type == F_UNLCK)
		return do_unlk(filp, cmd, fl);
	return do_setlk(filp, cmd, fl);
}

/*
 * Lock a (portion of) a file
 */
static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl)
{
	struct inode * inode = filp->f_mapping->host;

	dprintk("NFS: nfs_flock(f=%s/%ld, t=%x, fl=%x)\n",
			inode->i_sb->s_id, inode->i_ino,
			fl->fl_type, fl->fl_flags);

	/*
	 * No BSD flocks over NFS allowed.
	 * Note: we could try to fake a POSIX lock request here by
	 * using ((u32) filp | 0x80000000) or some such as the pid.
	 * Not sure whether that would be unique, though, or whether
	 * that would break in other places.
	 */
	if (!(fl->fl_flags & FL_FLOCK))
		return -ENOLCK;

	/* We're simulating flock() locks using posix locks on the server */
	fl->fl_owner = (fl_owner_t)filp;
	fl->fl_start = 0;
	fl->fl_end = OFFSET_MAX;

	if (fl->fl_type == F_UNLCK)
		return do_unlk(filp, cmd, fl);
	return do_setlk(filp, cmd, fl);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/nfs/file.c
	3	*
	4	* Copyright (C) 1992 Rick Sladkey
	5	*
	6	* Changes Copyright (C) 1994 by Florian La Roche
	7	* - Do not copy data too often around in the kernel.
	8	* - In nfs_file_read the return value of kmalloc wasn't checked.
	9	* - Put in a better version of read look-ahead buffering. Original idea
	10	* and implementation by Wai S Kok [email protected].
	11	*
	12	* Expire cache on write to a file by Wai S Kok (Oct 1994).
	13	*
	14	* Total rewrite of read side for new NFS buffer cache.. Linus.
	15	*
	16	* nfs regular file handling functions
	17	*/
	18
	19	#include <linux/time.h>
	20	#include <linux/kernel.h>
	21	#include <linux/errno.h>
	22	#include <linux/fcntl.h>
	23	#include <linux/stat.h>
	24	#include <linux/nfs_fs.h>
	25	#include <linux/nfs_mount.h>
	26	#include <linux/mm.h>
	27	#include <linux/slab.h>
	28	#include <linux/pagemap.h>
	29	#include <linux/smp_lock.h>
	30
	31	#include <asm/uaccess.h>
	32	#include <asm/system.h>
	33
	34	#include "delegation.h"
91d5b470	35	#include "iostat.h"
1da177e4 LT	36
	37	#define NFSDBG_FACILITY NFSDBG_FILE
	38
	39	static int nfs_file_open(struct inode , struct file );
	40	static int nfs_file_release(struct inode , struct file );
980802e3	41	static loff_t nfs_file_llseek(struct file *file, loff_t offset, int origin);
1da177e4 LT	42	static int nfs_file_mmap(struct file , struct vm_area_struct );
	43	static ssize_t nfs_file_sendfile(struct file , loff_t , size_t, read_actor_t, void *);
	44	static ssize_t nfs_file_read(struct kiocb , char __user , size_t, loff_t);
	45	static ssize_t nfs_file_write(struct kiocb , const char __user , size_t, loff_t);
	46	static int nfs_file_flush(struct file *);
	47	static int nfs_fsync(struct file , struct dentry dentry, int datasync);
	48	static int nfs_check_flags(int flags);
	49	static int nfs_lock(struct file filp, int cmd, struct file_lock fl);
	50	static int nfs_flock(struct file filp, int cmd, struct file_lock fl);
	51
	52	struct file_operations nfs_file_operations = {
980802e3	53	.llseek = nfs_file_llseek,
1da177e4 LT	54	.read = do_sync_read,
	55	.write = do_sync_write,
	56	.aio_read = nfs_file_read,
	57	.aio_write = nfs_file_write,
	58	.mmap = nfs_file_mmap,
	59	.open = nfs_file_open,
	60	.flush = nfs_file_flush,
	61	.release = nfs_file_release,
	62	.fsync = nfs_fsync,
	63	.lock = nfs_lock,
	64	.flock = nfs_flock,
	65	.sendfile = nfs_file_sendfile,
	66	.check_flags = nfs_check_flags,
	67	};
	68
	69	struct inode_operations nfs_file_inode_operations = {
	70	.permission = nfs_permission,
	71	.getattr = nfs_getattr,
	72	.setattr = nfs_setattr,
	73	};
	74
b7fa0554 AG	75	#ifdef CONFIG_NFS_V3
	76	struct inode_operations nfs3_file_inode_operations = {
	77	.permission = nfs_permission,
	78	.getattr = nfs_getattr,
	79	.setattr = nfs_setattr,
	80	.listxattr = nfs3_listxattr,
	81	.getxattr = nfs3_getxattr,
	82	.setxattr = nfs3_setxattr,
	83	.removexattr = nfs3_removexattr,
	84	};
	85	#endif /* CONFIG_NFS_v3 */
	86
1da177e4 LT	87	/* Hack for future NFS swap support */
	88	#ifndef IS_SWAPFILE
	89	# define IS_SWAPFILE(inode) (0)
	90	#endif
	91
	92	static int nfs_check_flags(int flags)
	93	{
	94	if ((flags & (O_APPEND \| O_DIRECT)) == (O_APPEND \| O_DIRECT))
	95	return -EINVAL;
	96
	97	return 0;
	98	}
	99
	100	/*
	101	* Open file
	102	*/
	103	static int
	104	nfs_file_open(struct inode inode, struct file filp)
	105	{
1da177e4 LT	106	int res;
	107
	108	res = nfs_check_flags(filp->f_flags);
	109	if (res)
	110	return res;
	111
91d5b470	112	nfs_inc_stats(inode, NFSIOS_VFSOPEN);
1da177e4	113	lock_kernel();
91d5b470	114	res = NFS_SERVER(inode)->rpc_ops->file_open(inode, filp);
1da177e4 LT	115	unlock_kernel();
	116	return res;
	117	}
	118
	119	static int
	120	nfs_file_release(struct inode inode, struct file filp)
	121	{
	122	/* Ensure that dirty pages are flushed out with the right creds */
	123	if (filp->f_mode & FMODE_WRITE)
	124	filemap_fdatawrite(filp->f_mapping);
91d5b470	125	nfs_inc_stats(inode, NFSIOS_VFSRELEASE);
1da177e4 LT	126	return NFS_PROTO(inode)->file_release(inode, filp);
	127	}
	128
fe51beec TM	129	/**
	130	* nfs_revalidate_file - Revalidate the page cache & related metadata
	131	* @inode - pointer to inode struct
	132	* @file - pointer to file
	133	*/
	134	static int nfs_revalidate_file(struct inode inode, struct file filp)
	135	{
55296809	136	struct nfs_inode *nfsi = NFS_I(inode);
fe51beec TM	137	int retval = 0;
fe51beec TM	138
b3c52da3 TM	139	if ((nfsi->cache_validity & (NFS_INO_REVAL_PAGECACHE\|NFS_INO_INVALID_ATTR))
b3c52da3 TM	140	\|\| nfs_attribute_timeout(inode))
fe51beec TM	141	retval = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
	142	nfs_revalidate_mapping(inode, filp->f_mapping);
	143	return 0;
	144	}
	145
980802e3 TM	146	/**
	147	* nfs_revalidate_size - Revalidate the file size
	148	* @inode - pointer to inode struct
	149	* @file - pointer to struct file
	150	*
	151	* Revalidates the file length. This is basically a wrapper around
	152	* nfs_revalidate_inode() that takes into account the fact that we may
	153	* have cached writes (in which case we don't care about the server's
	154	* idea of what the file length is), or O_DIRECT (in which case we
	155	* shouldn't trust the cache).
	156	*/
	157	static int nfs_revalidate_file_size(struct inode inode, struct file filp)
	158	{
	159	struct nfs_server *server = NFS_SERVER(inode);
	160	struct nfs_inode *nfsi = NFS_I(inode);
	161
	162	if (server->flags & NFS_MOUNT_NOAC)
	163	goto force_reval;
	164	if (filp->f_flags & O_DIRECT)
	165	goto force_reval;
	166	if (nfsi->npages != 0)
	167	return 0;
55296809	168	if (!(nfsi->cache_validity & NFS_INO_REVAL_PAGECACHE) && !nfs_attribute_timeout(inode))
fe51beec	169	return 0;
980802e3 TM	170	force_reval:
	171	return __nfs_revalidate_inode(server, inode);
	172	}
	173
	174	static loff_t nfs_file_llseek(struct file *filp, loff_t offset, int origin)
	175	{
	176	/* origin == SEEK_END => we must revalidate the cached file length */
	177	if (origin == 2) {
	178	struct inode *inode = filp->f_mapping->host;
	179	int retval = nfs_revalidate_file_size(inode, filp);
	180	if (retval < 0)
	181	return (loff_t)retval;
	182	}
	183	return remote_llseek(filp, offset, origin);
	184	}
	185
1da177e4 LT	186	/*
	187	* Flush all dirty pages, and check for write errors.
	188	*
	189	*/
	190	static int
	191	nfs_file_flush(struct file *file)
	192	{
	193	struct nfs_open_context ctx = (struct nfs_open_context )file->private_data;
	194	struct inode *inode = file->f_dentry->d_inode;
	195	int status;
	196
	197	dfprintk(VFS, "nfs: flush(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino);
	198
	199	if ((file->f_mode & FMODE_WRITE) == 0)
	200	return 0;
91d5b470	201	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
1da177e4 LT	202	lock_kernel();
	203	/* Ensure that data+attribute caches are up to date after close() */
	204	status = nfs_wb_all(inode);
	205	if (!status) {
	206	status = ctx->error;
	207	ctx->error = 0;
3338c143 TM	208	if (!status)
3338c143 TM	209	nfs_revalidate_inode(NFS_SERVER(inode), inode);
1da177e4 LT	210	}
	211	unlock_kernel();
	212	return status;
	213	}
	214
	215	static ssize_t
	216	nfs_file_read(struct kiocb iocb, char __user buf, size_t count, loff_t pos)
	217	{
	218	struct dentry * dentry = iocb->ki_filp->f_dentry;
	219	struct inode * inode = dentry->d_inode;
	220	ssize_t result;
	221
	222	#ifdef CONFIG_NFS_DIRECTIO
	223	if (iocb->ki_filp->f_flags & O_DIRECT)
	224	return nfs_file_direct_read(iocb, buf, count, pos);
	225	#endif
	226
	227	dfprintk(VFS, "nfs: read(%s/%s, %lu@%lu)\n",
	228	dentry->d_parent->d_name.name, dentry->d_name.name,
	229	(unsigned long) count, (unsigned long) pos);
	230
fe51beec	231	result = nfs_revalidate_file(inode, iocb->ki_filp);
91d5b470	232	nfs_add_stats(inode, NFSIOS_NORMALREADBYTES, count);
1da177e4 LT	233	if (!result)
	234	result = generic_file_aio_read(iocb, buf, count, pos);
	235	return result;
	236	}
	237
	238	static ssize_t
	239	nfs_file_sendfile(struct file filp, loff_t ppos, size_t count,
	240	read_actor_t actor, void *target)
	241	{
	242	struct dentry *dentry = filp->f_dentry;
	243	struct inode *inode = dentry->d_inode;
	244	ssize_t res;
	245
	246	dfprintk(VFS, "nfs: sendfile(%s/%s, %lu@%Lu)\n",
	247	dentry->d_parent->d_name.name, dentry->d_name.name,
	248	(unsigned long) count, (unsigned long long) *ppos);
	249
fe51beec	250	res = nfs_revalidate_file(inode, filp);
1da177e4 LT	251	if (!res)
	252	res = generic_file_sendfile(filp, ppos, count, actor, target);
	253	return res;
	254	}
	255
	256	static int
	257	nfs_file_mmap(struct file * file, struct vm_area_struct * vma)
	258	{
	259	struct dentry *dentry = file->f_dentry;
	260	struct inode *inode = dentry->d_inode;
	261	int status;
	262
	263	dfprintk(VFS, "nfs: mmap(%s/%s)\n",
	264	dentry->d_parent->d_name.name, dentry->d_name.name);
	265
fe51beec	266	status = nfs_revalidate_file(inode, file);
1da177e4 LT	267	if (!status)
	268	status = generic_file_mmap(file, vma);
	269	return status;
	270	}
	271
	272	/*
	273	* Flush any dirty pages for this process, and check for write errors.
	274	* The return status from this call provides a reliable indication of
	275	* whether any write errors occurred for this process.
	276	*/
	277	static int
	278	nfs_fsync(struct file file, struct dentry dentry, int datasync)
	279	{
	280	struct nfs_open_context ctx = (struct nfs_open_context )file->private_data;
	281	struct inode *inode = dentry->d_inode;
	282	int status;
	283
	284	dfprintk(VFS, "nfs: fsync(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino);
	285
91d5b470	286	nfs_inc_stats(inode, NFSIOS_VFSFSYNC);
1da177e4 LT	287	lock_kernel();
	288	status = nfs_wb_all(inode);
	289	if (!status) {
	290	status = ctx->error;
	291	ctx->error = 0;
	292	}
	293	unlock_kernel();
	294	return status;
	295	}
	296
	297	/*
	298	* This does the "real" work of the write. The generic routine has
	299	* allocated the page, locked it, done all the page alignment stuff
	300	* calculations etc. Now we should just copy the data from user
	301	* space and write it back to the real medium..
	302	*
	303	* If the writer ends up delaying the write, the writer needs to
	304	* increment the page use counts until he is done with the page.
	305	*/
	306	static int nfs_prepare_write(struct file file, struct page page, unsigned offset, unsigned to)
	307	{
	308	return nfs_flush_incompatible(file, page);
	309	}
	310
	311	static int nfs_commit_write(struct file file, struct page page, unsigned offset, unsigned to)
	312	{
	313	long status;
	314
	315	lock_kernel();
	316	status = nfs_updatepage(file, page, offset, to-offset);
	317	unlock_kernel();
	318	return status;
	319	}
	320
cd52ed35 TM	321	static int nfs_invalidate_page(struct page *page, unsigned long offset)
	322	{
	323	/* FIXME: we really should cancel any unstarted writes on this page */
	324	return 1;
	325	}
	326
	327	static int nfs_release_page(struct page *page, gfp_t gfp)
	328	{
	329	return !nfs_wb_page(page->mapping->host, page);
	330	}
	331
1da177e4 LT	332	struct address_space_operations nfs_file_aops = {
	333	.readpage = nfs_readpage,
	334	.readpages = nfs_readpages,
	335	.set_page_dirty = __set_page_dirty_nobuffers,
	336	.writepage = nfs_writepage,
	337	.writepages = nfs_writepages,
	338	.prepare_write = nfs_prepare_write,
	339	.commit_write = nfs_commit_write,
cd52ed35 TM	340	.invalidatepage = nfs_invalidate_page,
cd52ed35 TM	341	.releasepage = nfs_release_page,
1da177e4 LT	342	#ifdef CONFIG_NFS_DIRECTIO
	343	.direct_IO = nfs_direct_IO,
	344	#endif
	345	};
	346
	347	/*
	348	* Write to a file (through the page cache).
	349	*/
	350	static ssize_t
	351	nfs_file_write(struct kiocb iocb, const char __user buf, size_t count, loff_t pos)
	352	{
	353	struct dentry * dentry = iocb->ki_filp->f_dentry;
	354	struct inode * inode = dentry->d_inode;
	355	ssize_t result;
	356
	357	#ifdef CONFIG_NFS_DIRECTIO
	358	if (iocb->ki_filp->f_flags & O_DIRECT)
	359	return nfs_file_direct_write(iocb, buf, count, pos);
	360	#endif
	361
	362	dfprintk(VFS, "nfs: write(%s/%s(%ld), %lu@%lu)\n",
	363	dentry->d_parent->d_name.name, dentry->d_name.name,
	364	inode->i_ino, (unsigned long) count, (unsigned long) pos);
	365
	366	result = -EBUSY;
	367	if (IS_SWAPFILE(inode))
	368	goto out_swapfile;
7d52e862 TM	369	/*
	370	* O_APPEND implies that we must revalidate the file length.
	371	*/
	372	if (iocb->ki_filp->f_flags & O_APPEND) {
	373	result = nfs_revalidate_file_size(inode, iocb->ki_filp);
	374	if (result)
	375	goto out;
fe51beec TM	376	}
fe51beec TM	377	nfs_revalidate_mapping(inode, iocb->ki_filp->f_mapping);
1da177e4 LT	378
	379	result = count;
	380	if (!count)
	381	goto out;
	382
91d5b470	383	nfs_add_stats(inode, NFSIOS_NORMALWRITTENBYTES, count);
1da177e4 LT	384	result = generic_file_aio_write(iocb, buf, count, pos);
	385	out:
	386	return result;
	387
	388	out_swapfile:
	389	printk(KERN_INFO "NFS: attempt to write to active swap file!\n");
	390	goto out;
	391	}
	392
	393	static int do_getlk(struct file filp, int cmd, struct file_lock fl)
	394	{
8dc7c311	395	struct file_lock cfl;
1da177e4 LT	396	struct inode *inode = filp->f_mapping->host;
	397	int status = 0;
	398
	399	lock_kernel();
039c4d7a	400	/* Try local locking first */
8dc7c311 AA	401	if (posix_test_lock(filp, fl, &cfl)) {
8dc7c311 AA	402	locks_copy_lock(fl, &cfl);
039c4d7a	403	goto out;
1da177e4	404	}
039c4d7a TM	405
	406	if (nfs_have_delegation(inode, FMODE_READ))
	407	goto out_noconflict;
	408
	409	if (NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM)
	410	goto out_noconflict;
	411
	412	status = NFS_PROTO(inode)->lock(filp, cmd, fl);
	413	out:
1da177e4 LT	414	unlock_kernel();
1da177e4 LT	415	return status;
039c4d7a TM	416	out_noconflict:
	417	fl->fl_type = F_UNLCK;
	418	goto out;
1da177e4 LT	419	}
	420
	421	static int do_vfs_lock(struct file file, struct file_lock fl)
	422	{
	423	int res = 0;
	424	switch (fl->fl_flags & (FL_POSIX\|FL_FLOCK)) {
	425	case FL_POSIX:
	426	res = posix_lock_file_wait(file, fl);
	427	break;
	428	case FL_FLOCK:
	429	res = flock_lock_file_wait(file, fl);
	430	break;
	431	default:
	432	BUG();
	433	}
	434	if (res < 0)
	435	printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n",
	436	__FUNCTION__);
	437	return res;
	438	}
	439
	440	static int do_unlk(struct file filp, int cmd, struct file_lock fl)
	441	{
	442	struct inode *inode = filp->f_mapping->host;
	443	sigset_t oldset;
	444	int status;
	445
	446	rpc_clnt_sigmask(NFS_CLIENT(inode), &oldset);
	447	/*
	448	* Flush all pending writes before doing anything
	449	* with locks..
	450	*/
29884df0	451	nfs_sync_mapping(filp->f_mapping);
1da177e4 LT	452
	453	/* NOTE: special case
	454	* If we're signalled while cleaning up locks on process exit, we
	455	* still need to complete the unlock.
	456	*/
	457	lock_kernel();
	458	/* Use local locking if mounted with "-onolock" */
	459	if (!(NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM))
	460	status = NFS_PROTO(inode)->lock(filp, cmd, fl);
	461	else
	462	status = do_vfs_lock(filp, fl);
	463	unlock_kernel();
	464	rpc_clnt_sigunmask(NFS_CLIENT(inode), &oldset);
	465	return status;
	466	}
	467
	468	static int do_setlk(struct file filp, int cmd, struct file_lock fl)
	469	{
	470	struct inode *inode = filp->f_mapping->host;
	471	sigset_t oldset;
	472	int status;
	473
	474	rpc_clnt_sigmask(NFS_CLIENT(inode), &oldset);
	475	/*
	476	* Flush all pending writes before doing anything
	477	* with locks..
	478	*/
29884df0 TM	479	status = nfs_sync_mapping(filp->f_mapping);
29884df0 TM	480	if (status != 0)
1da177e4 LT	481	goto out;
	482
	483	lock_kernel();
	484	/* Use local locking if mounted with "-onolock" */
	485	if (!(NFS_SERVER(inode)->flags & NFS_MOUNT_NONLM)) {
	486	status = NFS_PROTO(inode)->lock(filp, cmd, fl);
	487	/* If we were signalled we still need to ensure that
	488	* we clean up any state on the server. We therefore
	489	* record the lock call as having succeeded in order to
	490	* ensure that locks_remove_posix() cleans it out when
	491	* the process exits.
	492	*/
	493	if (status == -EINTR \|\| status == -ERESTARTSYS)
	494	do_vfs_lock(filp, fl);
	495	} else
	496	status = do_vfs_lock(filp, fl);
	497	unlock_kernel();
	498	if (status < 0)
	499	goto out;
	500	/*
	501	* Make sure we clear the cache whenever we try to get the lock.
	502	* This makes locking act as a cache coherency point.
	503	*/
29884df0	504	nfs_sync_mapping(filp->f_mapping);
1da177e4 LT	505	nfs_zap_caches(inode);
	506	out:
	507	rpc_clnt_sigunmask(NFS_CLIENT(inode), &oldset);
	508	return status;
	509	}
	510
	511	/*
	512	* Lock a (portion of) a file
	513	*/
	514	static int nfs_lock(struct file filp, int cmd, struct file_lock fl)
	515	{
	516	struct inode * inode = filp->f_mapping->host;
	517
	518	dprintk("NFS: nfs_lock(f=%s/%ld, t=%x, fl=%x, r=%Ld:%Ld)\n",
	519	inode->i_sb->s_id, inode->i_ino,
	520	fl->fl_type, fl->fl_flags,
	521	(long long)fl->fl_start, (long long)fl->fl_end);
91d5b470	522	nfs_inc_stats(inode, NFSIOS_VFSLOCK);
1da177e4 LT	523
1da177e4 LT	524	/* No mandatory locks over NFS */
0800c5f7 AM	525	if ((inode->i_mode & (S_ISGID \| S_IXGRP)) == S_ISGID &&
0800c5f7 AM	526	fl->fl_type != F_UNLCK)
1da177e4 LT	527	return -ENOLCK;
	528
	529	if (IS_GETLK(cmd))
	530	return do_getlk(filp, cmd, fl);
	531	if (fl->fl_type == F_UNLCK)
	532	return do_unlk(filp, cmd, fl);
	533	return do_setlk(filp, cmd, fl);
	534	}
	535
	536	/*
	537	* Lock a (portion of) a file
	538	*/
	539	static int nfs_flock(struct file filp, int cmd, struct file_lock fl)
	540	{
	541	struct inode * inode = filp->f_mapping->host;
	542
	543	dprintk("NFS: nfs_flock(f=%s/%ld, t=%x, fl=%x)\n",
	544	inode->i_sb->s_id, inode->i_ino,
	545	fl->fl_type, fl->fl_flags);
	546
1da177e4 LT	547	/*
	548	* No BSD flocks over NFS allowed.
	549	* Note: we could try to fake a POSIX lock request here by
	550	* using ((u32) filp \| 0x80000000) or some such as the pid.
	551	* Not sure whether that would be unique, though, or whether
	552	* that would break in other places.
	553	*/
	554	if (!(fl->fl_flags & FL_FLOCK))
	555	return -ENOLCK;
	556
	557	/* We're simulating flock() locks using posix locks on the server */
	558	fl->fl_owner = (fl_owner_t)filp;
	559	fl->fl_start = 0;
	560	fl->fl_end = OFFSET_MAX;
	561
	562	if (fl->fl_type == F_UNLCK)
	563	return do_unlk(filp, cmd, fl);
	564	return do_setlk(filp, cmd, fl);
	565	}