file.c 14.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
/*
 *  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 elekokws@ee.nus.sg.
 *
 *  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"
C
Chuck Lever 已提交
35
#include "iostat.h"
L
Linus Torvalds 已提交
36 37 38 39 40

#define NFSDBG_FACILITY		NFSDBG_FILE

static int nfs_file_open(struct inode *, struct file *);
static int nfs_file_release(struct inode *, struct file *);
41
static loff_t nfs_file_llseek(struct file *file, loff_t offset, int origin);
L
Linus Torvalds 已提交
42 43
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 *);
44 45 46 47
static ssize_t nfs_file_read(struct kiocb *, const struct iovec *iov,
				unsigned long nr_segs, loff_t pos);
static ssize_t nfs_file_write(struct kiocb *, const struct iovec *iov,
				unsigned long nr_segs, loff_t pos);
48
static int  nfs_file_flush(struct file *, fl_owner_t id);
L
Linus Torvalds 已提交
49 50 51 52 53
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);

54
const struct file_operations nfs_file_operations = {
55
	.llseek		= nfs_file_llseek,
L
Linus Torvalds 已提交
56 57
	.read		= do_sync_read,
	.write		= do_sync_write,
58 59
	.aio_read	= nfs_file_read,
	.aio_write	= nfs_file_write,
L
Linus Torvalds 已提交
60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
	.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,
};

77 78 79 80 81 82 83 84 85 86 87 88
#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 */

L
Linus Torvalds 已提交
89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113
/* 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;

C
Chuck Lever 已提交
114
	nfs_inc_stats(inode, NFSIOS_VFSOPEN);
L
Linus Torvalds 已提交
115
	lock_kernel();
116
	res = NFS_PROTO(inode)->file_open(inode, filp);
L
Linus Torvalds 已提交
117 118 119 120 121 122 123 124 125 126
	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);
C
Chuck Lever 已提交
127
	nfs_inc_stats(inode, NFSIOS_VFSRELEASE);
L
Linus Torvalds 已提交
128 129 130
	return NFS_PROTO(inode)->file_release(inode, filp);
}

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
/**
 * 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;
153
	if (!(nfsi->cache_validity & NFS_INO_REVAL_PAGECACHE) && !nfs_attribute_timeout(inode))
154
		return 0;
155 156 157 158 159 160 161
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 */
162
	if (origin == SEEK_END) {
163 164 165 166 167 168 169 170
		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);
}

L
Linus Torvalds 已提交
171 172 173 174 175
/*
 * Flush all dirty pages, and check for write errors.
 *
 */
static int
176
nfs_file_flush(struct file *file, fl_owner_t id)
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184 185
{
	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;
C
Chuck Lever 已提交
186
	nfs_inc_stats(inode, NFSIOS_VFSFLUSH);
L
Linus Torvalds 已提交
187 188 189 190 191 192
	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;
193 194
		if (!status)
			nfs_revalidate_inode(NFS_SERVER(inode), inode);
L
Linus Torvalds 已提交
195 196 197 198 199 200
	}
	unlock_kernel();
	return status;
}

static ssize_t
201 202
nfs_file_read(struct kiocb *iocb, const struct iovec *iov,
		unsigned long nr_segs, loff_t pos)
L
Linus Torvalds 已提交
203 204 205 206
{
	struct dentry * dentry = iocb->ki_filp->f_dentry;
	struct inode * inode = dentry->d_inode;
	ssize_t result;
207
	size_t count = iov_length(iov, nr_segs);
L
Linus Torvalds 已提交
208 209 210

#ifdef CONFIG_NFS_DIRECTIO
	if (iocb->ki_filp->f_flags & O_DIRECT)
211
		return nfs_file_direct_read(iocb, iov, nr_segs, pos);
L
Linus Torvalds 已提交
212 213 214 215 216 217
#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);

218
	result = nfs_revalidate_mapping(inode, iocb->ki_filp->f_mapping);
C
Chuck Lever 已提交
219
	nfs_add_stats(inode, NFSIOS_NORMALREADBYTES, count);
L
Linus Torvalds 已提交
220
	if (!result)
221
		result = generic_file_aio_read(iocb, iov, nr_segs, pos);
L
Linus Torvalds 已提交
222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
	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);

237
	res = nfs_revalidate_mapping(inode, filp->f_mapping);
L
Linus Torvalds 已提交
238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
	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);

253
	status = nfs_revalidate_mapping(inode, file->f_mapping);
L
Linus Torvalds 已提交
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272
	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);

C
Chuck Lever 已提交
273
	nfs_inc_stats(inode, NFSIOS_VFSFSYNC);
L
Linus Torvalds 已提交
274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307
	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;
}

308
static void nfs_invalidate_page(struct page *page, unsigned long offset)
309
{
310 311
	if (offset != 0)
		return;
312
	/* Cancel any unstarted writes on this page */
313
	nfs_wb_page_priority(page->mapping->host, page, FLUSH_INVALIDATE);
314 315 316 317
}

static int nfs_release_page(struct page *page, gfp_t gfp)
{
T
Trond Myklebust 已提交
318 319 320 321
	/*
	 * Avoid deadlock on nfs_wait_on_request().
	 */
	if (!(gfp & __GFP_FS))
322
		return 0;
T
Trond Myklebust 已提交
323 324 325
	/* Hack... Force nfs_wb_page() to write out the page */
	SetPageDirty(page);
	return !nfs_wb_page(page->mapping->host, page);
326 327
}

328
const struct address_space_operations nfs_file_aops = {
L
Linus Torvalds 已提交
329 330
	.readpage = nfs_readpage,
	.readpages = nfs_readpages,
T
Trond Myklebust 已提交
331
	.set_page_dirty = nfs_set_page_dirty,
L
Linus Torvalds 已提交
332 333 334 335
	.writepage = nfs_writepage,
	.writepages = nfs_writepages,
	.prepare_write = nfs_prepare_write,
	.commit_write = nfs_commit_write,
336 337
	.invalidatepage = nfs_invalidate_page,
	.releasepage = nfs_release_page,
L
Linus Torvalds 已提交
338 339 340 341 342
#ifdef CONFIG_NFS_DIRECTIO
	.direct_IO = nfs_direct_IO,
#endif
};

343 344
static ssize_t nfs_file_write(struct kiocb *iocb, const struct iovec *iov,
				unsigned long nr_segs, loff_t pos)
L
Linus Torvalds 已提交
345 346 347 348
{
	struct dentry * dentry = iocb->ki_filp->f_dentry;
	struct inode * inode = dentry->d_inode;
	ssize_t result;
349
	size_t count = iov_length(iov, nr_segs);
L
Linus Torvalds 已提交
350 351 352

#ifdef CONFIG_NFS_DIRECTIO
	if (iocb->ki_filp->f_flags & O_DIRECT)
353
		return nfs_file_direct_write(iocb, iov, nr_segs, pos);
L
Linus Torvalds 已提交
354 355
#endif

356
	dfprintk(VFS, "nfs: write(%s/%s(%ld), %lu@%Ld)\n",
L
Linus Torvalds 已提交
357
		dentry->d_parent->d_name.name, dentry->d_name.name,
358
		inode->i_ino, (unsigned long) count, (long long) pos);
L
Linus Torvalds 已提交
359 360 361 362

	result = -EBUSY;
	if (IS_SWAPFILE(inode))
		goto out_swapfile;
363 364 365 366 367 368 369
	/*
	 * 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;
370
	}
L
Linus Torvalds 已提交
371 372 373 374 375

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

C
Chuck Lever 已提交
376
	nfs_add_stats(inode, NFSIOS_NORMALWRITTENBYTES, count);
377
	result = generic_file_aio_write(iocb, iov, nr_segs, pos);
T
Trond Myklebust 已提交
378 379 380 381 382 383
	/* Return error values for O_SYNC and IS_SYNC() */
	if (result >= 0 && (IS_SYNC(inode) || (iocb->ki_filp->f_flags & O_SYNC))) {
		int err = nfs_fsync(iocb->ki_filp, dentry, 1);
		if (err < 0)
			result = err;
	}
L
Linus Torvalds 已提交
384 385 386 387 388 389 390 391 392 393
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)
{
394
	struct file_lock cfl;
L
Linus Torvalds 已提交
395 396 397 398
	struct inode *inode = filp->f_mapping->host;
	int status = 0;

	lock_kernel();
399
	/* Try local locking first */
400
	if (posix_test_lock(filp, fl, &cfl)) {
401 402 403 404
		fl->fl_start = cfl.fl_start;
		fl->fl_end = cfl.fl_end;
		fl->fl_type = cfl.fl_type;
		fl->fl_pid = cfl.fl_pid;
405
		goto out;
L
Linus Torvalds 已提交
406
	}
407 408 409 410 411 412 413 414 415

	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:
L
Linus Torvalds 已提交
416 417
	unlock_kernel();
	return status;
418 419 420
out_noconflict:
	fl->fl_type = F_UNLCK;
	goto out;
L
Linus Torvalds 已提交
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
}

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;
	int status;

	/*
	 * Flush all pending writes before doing anything
	 * with locks..
	 */
T
Trond Myklebust 已提交
451
	nfs_sync_mapping(filp->f_mapping);
L
Linus Torvalds 已提交
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475

	/* 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();
	return status;
}

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

	/*
	 * Flush all pending writes before doing anything
	 * with locks..
	 */
T
Trond Myklebust 已提交
476 477
	status = nfs_sync_mapping(filp->f_mapping);
	if (status != 0)
L
Linus Torvalds 已提交
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500
		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.
	 */
T
Trond Myklebust 已提交
501
	nfs_sync_mapping(filp->f_mapping);
L
Linus Torvalds 已提交
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
	nfs_zap_caches(inode);
out:
	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);
C
Chuck Lever 已提交
518
	nfs_inc_stats(inode, NFSIOS_VFSLOCK);
L
Linus Torvalds 已提交
519 520

	/* No mandatory locks over NFS */
A
ASANO Masahiro 已提交
521 522
	if ((inode->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID &&
	    fl->fl_type != F_UNLCK)
L
Linus Torvalds 已提交
523 524 525 526 527 528 529 530 531 532 533 534 535 536 537
		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)
{
	dprintk("NFS: nfs_flock(f=%s/%ld, t=%x, fl=%x)\n",
538 539
			filp->f_dentry->d_inode->i_sb->s_id,
			filp->f_dentry->d_inode->i_ino,
L
Linus Torvalds 已提交
540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
			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);
}