dir.c 42.1 KB
Newer Older
1 2
/*
  FUSE: Filesystem in Userspace
M
Miklos Szeredi 已提交
3
  Copyright (C) 2001-2008  Miklos Szeredi <miklos@szeredi.hu>
4 5 6 7 8 9 10 11 12 13 14

  This program can be distributed under the terms of the GNU GPL.
  See the file COPYING.
*/

#include "fuse_i.h"

#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/sched.h>
#include <linux/namei.h>
15
#include <linux/slab.h>
16

M
Miklos Szeredi 已提交
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
#if BITS_PER_LONG >= 64
static inline void fuse_dentry_settime(struct dentry *entry, u64 time)
{
	entry->d_time = time;
}

static inline u64 fuse_dentry_time(struct dentry *entry)
{
	return entry->d_time;
}
#else
/*
 * On 32 bit archs store the high 32 bits of time in d_fsdata
 */
static void fuse_dentry_settime(struct dentry *entry, u64 time)
{
	entry->d_time = time;
	entry->d_fsdata = (void *) (unsigned long) (time >> 32);
}

static u64 fuse_dentry_time(struct dentry *entry)
{
	return (u64) entry->d_time +
		((u64) (unsigned long) entry->d_fsdata << 32);
}
#endif

44 45 46 47 48 49 50 51 52
/*
 * FUSE caches dentries and attributes with separate timeout.  The
 * time in jiffies until the dentry/attributes are valid is stored in
 * dentry->d_time and fuse_inode->i_time respectively.
 */

/*
 * Calculate the time in jiffies until a dentry/attributes are valid
 */
M
Miklos Szeredi 已提交
53
static u64 time_to_jiffies(unsigned long sec, unsigned long nsec)
54
{
M
Miklos Szeredi 已提交
55 56
	if (sec || nsec) {
		struct timespec ts = {sec, nsec};
M
Miklos Szeredi 已提交
57
		return get_jiffies_64() + timespec_to_jiffies(&ts);
M
Miklos Szeredi 已提交
58
	} else
M
Miklos Szeredi 已提交
59
		return 0;
60 61
}

62 63 64 65
/*
 * Set dentry and possibly attribute timeouts from the lookup/mk*
 * replies
 */
66 67
static void fuse_change_entry_timeout(struct dentry *entry,
				      struct fuse_entry_out *o)
68
{
M
Miklos Szeredi 已提交
69 70
	fuse_dentry_settime(entry,
		time_to_jiffies(o->entry_valid, o->entry_valid_nsec));
71 72 73 74 75 76 77 78 79 80
}

static u64 attr_timeout(struct fuse_attr_out *o)
{
	return time_to_jiffies(o->attr_valid, o->attr_valid_nsec);
}

static u64 entry_attr_timeout(struct fuse_entry_out *o)
{
	return time_to_jiffies(o->attr_valid, o->attr_valid_nsec);
81 82
}

83 84 85 86
/*
 * Mark the attributes as stale, so that at the next call to
 * ->getattr() they will be fetched from userspace
 */
87 88
void fuse_invalidate_attr(struct inode *inode)
{
M
Miklos Szeredi 已提交
89
	get_fuse_inode(inode)->i_time = 0;
90 91
}

92 93 94 95 96 97 98 99
/*
 * Just mark the entry as stale, so that a next attempt to look it up
 * will result in a new lookup call to userspace
 *
 * This is called when a dentry is about to become negative and the
 * timeout is unknown (unlink, rmdir, rename and in some cases
 * lookup)
 */
M
Miklos Szeredi 已提交
100
void fuse_invalidate_entry_cache(struct dentry *entry)
101
{
M
Miklos Szeredi 已提交
102
	fuse_dentry_settime(entry, 0);
103 104
}

105 106 107 108
/*
 * Same as fuse_invalidate_entry_cache(), but also try to remove the
 * dentry from the hash
 */
109 110 111 112
static void fuse_invalidate_entry(struct dentry *entry)
{
	d_invalidate(entry);
	fuse_invalidate_entry_cache(entry);
113 114
}

115 116
static void fuse_lookup_init(struct fuse_conn *fc, struct fuse_req *req,
			     u64 nodeid, struct qstr *name,
117 118
			     struct fuse_entry_out *outarg)
{
119
	memset(outarg, 0, sizeof(struct fuse_entry_out));
120
	req->in.h.opcode = FUSE_LOOKUP;
121
	req->in.h.nodeid = nodeid;
122
	req->in.numargs = 1;
123 124
	req->in.args[0].size = name->len + 1;
	req->in.args[0].value = name->name;
125
	req->out.numargs = 1;
126 127 128 129
	if (fc->minor < 9)
		req->out.args[0].size = FUSE_COMPAT_ENTRY_OUT_SIZE;
	else
		req->out.args[0].size = sizeof(struct fuse_entry_out);
130 131 132
	req->out.args[0].value = outarg;
}

133
u64 fuse_get_attr_version(struct fuse_conn *fc)
134 135 136 137 138 139 140 141 142 143 144 145 146 147
{
	u64 curr_version;

	/*
	 * The spin lock isn't actually needed on 64bit archs, but we
	 * don't yet care too much about such optimizations.
	 */
	spin_lock(&fc->lock);
	curr_version = fc->attr_version;
	spin_unlock(&fc->lock);

	return curr_version;
}

148 149 150 151 152 153 154 155 156
/*
 * Check whether the dentry is still valid
 *
 * If the entry validity timeout has expired and the dentry is
 * positive, try to redo the lookup.  If the lookup results in a
 * different inode, then let the VFS invalidate the dentry and redo
 * the lookup once more.  If the lookup results in the same inode,
 * then refresh the attributes, timeouts and mark the dentry valid.
 */
157 158
static int fuse_dentry_revalidate(struct dentry *entry, struct nameidata *nd)
{
159
	struct inode *inode;
160

161
	inode = ACCESS_ONCE(entry->d_inode);
162
	if (inode && is_bad_inode(inode))
163
		return 0;
M
Miklos Szeredi 已提交
164
	else if (fuse_dentry_time(entry) < get_jiffies_64()) {
165 166
		int err;
		struct fuse_entry_out outarg;
167 168
		struct fuse_conn *fc;
		struct fuse_req *req;
169
		struct fuse_forget_link *forget;
170
		struct dentry *parent;
171
		u64 attr_version;
172

173
		/* For negative dentries, always do a fresh lookup */
174 175 176
		if (!inode)
			return 0;

177
		if (nd && (nd->flags & LOOKUP_RCU))
178 179
			return -ECHILD;

180
		fc = get_fuse_conn(inode);
181 182
		req = fuse_get_req(fc);
		if (IS_ERR(req))
183 184
			return 0;

185 186
		forget = fuse_alloc_forget();
		if (!forget) {
M
Miklos Szeredi 已提交
187 188 189 190
			fuse_put_request(fc, req);
			return 0;
		}

191
		attr_version = fuse_get_attr_version(fc);
192

193
		parent = dget_parent(entry);
194 195
		fuse_lookup_init(fc, req, get_node_id(parent->d_inode),
				 &entry->d_name, &outarg);
196
		fuse_request_send(fc, req);
197
		dput(parent);
198
		err = req->out.h.error;
M
Miklos Szeredi 已提交
199
		fuse_put_request(fc, req);
200 201 202
		/* Zero nodeid is same as -ENOENT */
		if (!err && !outarg.nodeid)
			err = -ENOENT;
203
		if (!err) {
204
			struct fuse_inode *fi = get_fuse_inode(inode);
205
			if (outarg.nodeid != get_node_id(inode)) {
206
				fuse_queue_forget(fc, forget, outarg.nodeid, 1);
207 208
				return 0;
			}
209
			spin_lock(&fc->lock);
M
Miklos Szeredi 已提交
210
			fi->nlookup++;
211
			spin_unlock(&fc->lock);
212
		}
213
		kfree(forget);
214
		if (err || (outarg.attr.mode ^ inode->i_mode) & S_IFMT)
215 216
			return 0;

217 218 219 220
		fuse_change_attributes(inode, &outarg.attr,
				       entry_attr_timeout(&outarg),
				       attr_version);
		fuse_change_entry_timeout(entry, &outarg);
221 222 223 224
	}
	return 1;
}

225
static int invalid_nodeid(u64 nodeid)
226 227 228 229
{
	return !nodeid || nodeid == FUSE_ROOT_ID;
}

A
Al Viro 已提交
230
const struct dentry_operations fuse_dentry_operations = {
231 232 233
	.d_revalidate	= fuse_dentry_revalidate,
};

234
int fuse_valid_type(int m)
235 236 237 238 239
{
	return S_ISREG(m) || S_ISDIR(m) || S_ISLNK(m) || S_ISCHR(m) ||
		S_ISBLK(m) || S_ISFIFO(m) || S_ISSOCK(m);
}

240 241 242 243
/*
 * Add a directory inode to a dentry, ensuring that no other dentry
 * refers to this inode.  Called with fc->inst_mutex.
 */
244 245
static struct dentry *fuse_d_add_directory(struct dentry *entry,
					   struct inode *inode)
246 247
{
	struct dentry *alias = d_find_alias(inode);
248
	if (alias && !(alias->d_flags & DCACHE_DISCONNECTED)) {
249 250 251
		/* This tries to shrink the subtree below alias */
		fuse_invalidate_entry(alias);
		dput(alias);
A
Al Viro 已提交
252
		if (!hlist_empty(&inode->i_dentry))
253 254 255
			return ERR_PTR(-EBUSY);
	} else {
		dput(alias);
256
	}
257
	return d_splice_alias(inode, entry);
258 259
}

260 261
int fuse_lookup_name(struct super_block *sb, u64 nodeid, struct qstr *name,
		     struct fuse_entry_out *outarg, struct inode **inode)
262
{
263
	struct fuse_conn *fc = get_fuse_conn_super(sb);
264
	struct fuse_req *req;
265
	struct fuse_forget_link *forget;
266
	u64 attr_version;
267
	int err;
268

269 270 271 272
	*inode = NULL;
	err = -ENAMETOOLONG;
	if (name->len > FUSE_NAME_MAX)
		goto out;
273

274
	req = fuse_get_req(fc);
275
	err = PTR_ERR(req);
276
	if (IS_ERR(req))
277
		goto out;
278

279 280 281
	forget = fuse_alloc_forget();
	err = -ENOMEM;
	if (!forget) {
M
Miklos Szeredi 已提交
282
		fuse_put_request(fc, req);
283
		goto out;
M
Miklos Szeredi 已提交
284 285
	}

286
	attr_version = fuse_get_attr_version(fc);
287

288
	fuse_lookup_init(fc, req, nodeid, name, outarg);
289
	fuse_request_send(fc, req);
290
	err = req->out.h.error;
M
Miklos Szeredi 已提交
291
	fuse_put_request(fc, req);
292
	/* Zero nodeid is same as -ENOENT, but with valid timeout */
293 294 295 296 297 298 299 300 301 302 303 304 305 306
	if (err || !outarg->nodeid)
		goto out_put_forget;

	err = -EIO;
	if (!outarg->nodeid)
		goto out_put_forget;
	if (!fuse_valid_type(outarg->attr.mode))
		goto out_put_forget;

	*inode = fuse_iget(sb, outarg->nodeid, outarg->generation,
			   &outarg->attr, entry_attr_timeout(outarg),
			   attr_version);
	err = -ENOMEM;
	if (!*inode) {
307
		fuse_queue_forget(fc, forget, outarg->nodeid, 1);
308
		goto out;
309
	}
310 311 312
	err = 0;

 out_put_forget:
313
	kfree(forget);
314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
 out:
	return err;
}

static struct dentry *fuse_lookup(struct inode *dir, struct dentry *entry,
				  struct nameidata *nd)
{
	int err;
	struct fuse_entry_out outarg;
	struct inode *inode;
	struct dentry *newent;
	struct fuse_conn *fc = get_fuse_conn(dir);
	bool outarg_valid = true;

	err = fuse_lookup_name(dir->i_sb, get_node_id(dir), &entry->d_name,
			       &outarg, &inode);
	if (err == -ENOENT) {
		outarg_valid = false;
		err = 0;
	}
	if (err)
		goto out_err;

	err = -EIO;
	if (inode && get_node_id(inode) == FUSE_ROOT_ID)
		goto out_iput;
340

341 342
	if (inode && S_ISDIR(inode->i_mode)) {
		mutex_lock(&fc->inst_mutex);
343
		newent = fuse_d_add_directory(entry, inode);
344
		mutex_unlock(&fc->inst_mutex);
345 346 347 348
		err = PTR_ERR(newent);
		if (IS_ERR(newent))
			goto out_iput;
	} else {
349
		newent = d_splice_alias(inode, entry);
350
	}
351

352
	entry = newent ? newent : entry;
353
	if (outarg_valid)
354
		fuse_change_entry_timeout(entry, &outarg);
355 356
	else
		fuse_invalidate_entry_cache(entry);
357

358
	return newent;
359 360 361 362 363

 out_iput:
	iput(inode);
 out_err:
	return ERR_PTR(err);
364 365
}

366 367 368 369 370 371
/*
 * Atomic create+open operation
 *
 * If the filesystem doesn't support this, then fall back to separate
 * 'mknod' + 'open' requests.
 */
372 373
static struct file *fuse_create_open(struct inode *dir, struct dentry *entry,
				     struct opendata *od, unsigned flags,
374
				     umode_t mode, int *opened)
M
Miklos Szeredi 已提交
375 376 377 378 379
{
	int err;
	struct inode *inode;
	struct fuse_conn *fc = get_fuse_conn(dir);
	struct fuse_req *req;
380
	struct fuse_forget_link *forget;
381
	struct fuse_create_in inarg;
M
Miklos Szeredi 已提交
382 383 384 385 386
	struct fuse_open_out outopen;
	struct fuse_entry_out outentry;
	struct fuse_file *ff;
	struct file *file;

387
	forget = fuse_alloc_forget();
388
	err = -ENOMEM;
389
	if (!forget)
390
		goto out_err;
391

392
	req = fuse_get_req(fc);
393
	err = PTR_ERR(req);
394
	if (IS_ERR(req))
395
		goto out_put_forget_req;
M
Miklos Szeredi 已提交
396

397
	err = -ENOMEM;
T
Tejun Heo 已提交
398
	ff = fuse_file_alloc(fc);
M
Miklos Szeredi 已提交
399 400 401
	if (!ff)
		goto out_put_request;

402 403 404
	if (!fc->dont_mask)
		mode &= ~current_umask();

M
Miklos Szeredi 已提交
405 406
	flags &= ~O_NOCTTY;
	memset(&inarg, 0, sizeof(inarg));
407
	memset(&outentry, 0, sizeof(outentry));
M
Miklos Szeredi 已提交
408 409
	inarg.flags = flags;
	inarg.mode = mode;
410
	inarg.umask = current_umask();
M
Miklos Szeredi 已提交
411 412 413
	req->in.h.opcode = FUSE_CREATE;
	req->in.h.nodeid = get_node_id(dir);
	req->in.numargs = 2;
414 415
	req->in.args[0].size = fc->minor < 12 ? sizeof(struct fuse_open_in) :
						sizeof(inarg);
M
Miklos Szeredi 已提交
416 417 418 419
	req->in.args[0].value = &inarg;
	req->in.args[1].size = entry->d_name.len + 1;
	req->in.args[1].value = entry->d_name.name;
	req->out.numargs = 2;
420 421 422 423
	if (fc->minor < 9)
		req->out.args[0].size = FUSE_COMPAT_ENTRY_OUT_SIZE;
	else
		req->out.args[0].size = sizeof(outentry);
M
Miklos Szeredi 已提交
424 425 426
	req->out.args[0].value = &outentry;
	req->out.args[1].size = sizeof(outopen);
	req->out.args[1].value = &outopen;
427
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
428
	err = req->out.h.error;
429
	if (err)
M
Miklos Szeredi 已提交
430 431 432
		goto out_free_ff;

	err = -EIO;
433
	if (!S_ISREG(outentry.attr.mode) || invalid_nodeid(outentry.nodeid))
M
Miklos Szeredi 已提交
434 435
		goto out_free_ff;

436
	fuse_put_request(fc, req);
437 438 439
	ff->fh = outopen.fh;
	ff->nodeid = outentry.nodeid;
	ff->open_flags = outopen.open_flags;
M
Miklos Szeredi 已提交
440
	inode = fuse_iget(dir->i_sb, outentry.nodeid, outentry.generation,
441
			  &outentry.attr, entry_attr_timeout(&outentry), 0);
M
Miklos Szeredi 已提交
442 443
	if (!inode) {
		flags &= ~(O_CREAT | O_EXCL | O_TRUNC);
444
		fuse_sync_release(ff, flags);
445
		fuse_queue_forget(fc, forget, outentry.nodeid, 1);
446 447
		err = -ENOMEM;
		goto out_err;
M
Miklos Szeredi 已提交
448
	}
449
	kfree(forget);
M
Miklos Szeredi 已提交
450
	d_instantiate(entry, inode);
451
	fuse_change_entry_timeout(entry, &outentry);
452
	fuse_invalidate_attr(dir);
453
	file = finish_open(od, entry, generic_file_open, opened);
M
Miklos Szeredi 已提交
454
	if (IS_ERR(file)) {
455
		fuse_sync_release(ff, flags);
456 457 458
	} else {
		file->private_data = fuse_file_get(ff);
		fuse_finish_open(inode, file);
M
Miklos Szeredi 已提交
459
	}
460
	return file;
M
Miklos Szeredi 已提交
461

462
out_free_ff:
M
Miklos Szeredi 已提交
463
	fuse_file_free(ff);
464
out_put_request:
M
Miklos Szeredi 已提交
465
	fuse_put_request(fc, req);
466
out_put_forget_req:
467
	kfree(forget);
468 469 470 471 472 473 474
out_err:
	return ERR_PTR(err);
}

static int fuse_mknod(struct inode *, struct dentry *, umode_t, dev_t);
static struct file *fuse_atomic_open(struct inode *dir, struct dentry *entry,
				     struct opendata *od, unsigned flags,
475
				     umode_t mode, int *opened)
476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
{
	int err;
	struct fuse_conn *fc = get_fuse_conn(dir);
	struct file *file;
	struct dentry *res = NULL;

	if (d_unhashed(entry)) {
		res = fuse_lookup(dir, entry, NULL);
		if (IS_ERR(res))
			return ERR_CAST(res);

		if (res)
			entry = res;
	}

	if (!(flags & O_CREAT) || entry->d_inode)
		goto no_open;

	/* Only creates */
495
	*opened |= FILE_CREATED;
496 497 498 499

	if (fc->no_create)
		goto mknod;

500
	file = fuse_create_open(dir, entry, od, flags, mode, opened);
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
	if (PTR_ERR(file) == -ENOSYS) {
		fc->no_create = 1;
		goto mknod;
	}
out_dput:
	dput(res);
	return file;

mknod:
	err = fuse_mknod(dir, entry, mode, 0);
	if (err) {
		file = ERR_PTR(err);
		goto out_dput;
	}
no_open:
	finish_no_open(od, res);
	return NULL;
M
Miklos Szeredi 已提交
518 519
}

520 521 522
/*
 * Code shared between mknod, mkdir, symlink and link
 */
523 524
static int create_new_entry(struct fuse_conn *fc, struct fuse_req *req,
			    struct inode *dir, struct dentry *entry,
A
Al Viro 已提交
525
			    umode_t mode)
526 527 528 529
{
	struct fuse_entry_out outarg;
	struct inode *inode;
	int err;
530
	struct fuse_forget_link *forget;
M
Miklos Szeredi 已提交
531

532 533
	forget = fuse_alloc_forget();
	if (!forget) {
M
Miklos Szeredi 已提交
534
		fuse_put_request(fc, req);
535
		return -ENOMEM;
M
Miklos Szeredi 已提交
536
	}
537

538
	memset(&outarg, 0, sizeof(outarg));
539 540
	req->in.h.nodeid = get_node_id(dir);
	req->out.numargs = 1;
541 542 543 544
	if (fc->minor < 9)
		req->out.args[0].size = FUSE_COMPAT_ENTRY_OUT_SIZE;
	else
		req->out.args[0].size = sizeof(outarg);
545
	req->out.args[0].value = &outarg;
546
	fuse_request_send(fc, req);
547
	err = req->out.h.error;
M
Miklos Szeredi 已提交
548 549 550 551
	fuse_put_request(fc, req);
	if (err)
		goto out_put_forget_req;

552 553
	err = -EIO;
	if (invalid_nodeid(outarg.nodeid))
M
Miklos Szeredi 已提交
554
		goto out_put_forget_req;
555 556

	if ((outarg.attr.mode ^ mode) & S_IFMT)
M
Miklos Szeredi 已提交
557
		goto out_put_forget_req;
558

559
	inode = fuse_iget(dir->i_sb, outarg.nodeid, outarg.generation,
560
			  &outarg.attr, entry_attr_timeout(&outarg), 0);
561
	if (!inode) {
562
		fuse_queue_forget(fc, forget, outarg.nodeid, 1);
563 564
		return -ENOMEM;
	}
565
	kfree(forget);
566

567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
	if (S_ISDIR(inode->i_mode)) {
		struct dentry *alias;
		mutex_lock(&fc->inst_mutex);
		alias = d_find_alias(inode);
		if (alias) {
			/* New directory must have moved since mkdir */
			mutex_unlock(&fc->inst_mutex);
			dput(alias);
			iput(inode);
			return -EBUSY;
		}
		d_instantiate(entry, inode);
		mutex_unlock(&fc->inst_mutex);
	} else
		d_instantiate(entry, inode);
582

583
	fuse_change_entry_timeout(entry, &outarg);
584 585
	fuse_invalidate_attr(dir);
	return 0;
586

M
Miklos Szeredi 已提交
587
 out_put_forget_req:
588
	kfree(forget);
589
	return err;
590 591
}

A
Al Viro 已提交
592
static int fuse_mknod(struct inode *dir, struct dentry *entry, umode_t mode,
593 594 595 596
		      dev_t rdev)
{
	struct fuse_mknod_in inarg;
	struct fuse_conn *fc = get_fuse_conn(dir);
597 598 599
	struct fuse_req *req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
600

601 602 603
	if (!fc->dont_mask)
		mode &= ~current_umask();

604 605 606
	memset(&inarg, 0, sizeof(inarg));
	inarg.mode = mode;
	inarg.rdev = new_encode_dev(rdev);
607
	inarg.umask = current_umask();
608 609
	req->in.h.opcode = FUSE_MKNOD;
	req->in.numargs = 2;
610 611
	req->in.args[0].size = fc->minor < 12 ? FUSE_COMPAT_MKNOD_IN_SIZE :
						sizeof(inarg);
612 613 614 615 616 617
	req->in.args[0].value = &inarg;
	req->in.args[1].size = entry->d_name.len + 1;
	req->in.args[1].value = entry->d_name.name;
	return create_new_entry(fc, req, dir, entry, mode);
}

A
Al Viro 已提交
618
static int fuse_create(struct inode *dir, struct dentry *entry, umode_t mode,
619 620 621 622 623
		       struct nameidata *nd)
{
	return fuse_mknod(dir, entry, mode, 0);
}

624
static int fuse_mkdir(struct inode *dir, struct dentry *entry, umode_t mode)
625 626 627
{
	struct fuse_mkdir_in inarg;
	struct fuse_conn *fc = get_fuse_conn(dir);
628 629 630
	struct fuse_req *req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
631

632 633 634
	if (!fc->dont_mask)
		mode &= ~current_umask();

635 636
	memset(&inarg, 0, sizeof(inarg));
	inarg.mode = mode;
637
	inarg.umask = current_umask();
638 639 640 641 642 643 644 645 646 647 648 649 650 651
	req->in.h.opcode = FUSE_MKDIR;
	req->in.numargs = 2;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->in.args[1].size = entry->d_name.len + 1;
	req->in.args[1].value = entry->d_name.name;
	return create_new_entry(fc, req, dir, entry, S_IFDIR);
}

static int fuse_symlink(struct inode *dir, struct dentry *entry,
			const char *link)
{
	struct fuse_conn *fc = get_fuse_conn(dir);
	unsigned len = strlen(link) + 1;
652 653 654
	struct fuse_req *req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
655 656 657 658 659 660 661 662 663 664 665 666 667 668

	req->in.h.opcode = FUSE_SYMLINK;
	req->in.numargs = 2;
	req->in.args[0].size = entry->d_name.len + 1;
	req->in.args[0].value = entry->d_name.name;
	req->in.args[1].size = len;
	req->in.args[1].value = link;
	return create_new_entry(fc, req, dir, entry, S_IFLNK);
}

static int fuse_unlink(struct inode *dir, struct dentry *entry)
{
	int err;
	struct fuse_conn *fc = get_fuse_conn(dir);
669 670 671
	struct fuse_req *req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
672 673 674 675 676 677

	req->in.h.opcode = FUSE_UNLINK;
	req->in.h.nodeid = get_node_id(dir);
	req->in.numargs = 1;
	req->in.args[0].size = entry->d_name.len + 1;
	req->in.args[0].value = entry->d_name.name;
678
	fuse_request_send(fc, req);
679 680 681 682
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (!err) {
		struct inode *inode = entry->d_inode;
M
Miklos Szeredi 已提交
683
		struct fuse_inode *fi = get_fuse_inode(inode);
684

M
Miklos Szeredi 已提交
685 686 687 688
		spin_lock(&fc->lock);
		fi->attr_version = ++fc->attr_version;
		drop_nlink(inode);
		spin_unlock(&fc->lock);
689 690
		fuse_invalidate_attr(inode);
		fuse_invalidate_attr(dir);
691
		fuse_invalidate_entry_cache(entry);
692 693 694 695 696 697 698 699 700
	} else if (err == -EINTR)
		fuse_invalidate_entry(entry);
	return err;
}

static int fuse_rmdir(struct inode *dir, struct dentry *entry)
{
	int err;
	struct fuse_conn *fc = get_fuse_conn(dir);
701 702 703
	struct fuse_req *req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
704 705 706 707 708 709

	req->in.h.opcode = FUSE_RMDIR;
	req->in.h.nodeid = get_node_id(dir);
	req->in.numargs = 1;
	req->in.args[0].size = entry->d_name.len + 1;
	req->in.args[0].value = entry->d_name.name;
710
	fuse_request_send(fc, req);
711 712 713
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (!err) {
714
		clear_nlink(entry->d_inode);
715
		fuse_invalidate_attr(dir);
716
		fuse_invalidate_entry_cache(entry);
717 718 719 720 721 722 723 724 725 726 727
	} else if (err == -EINTR)
		fuse_invalidate_entry(entry);
	return err;
}

static int fuse_rename(struct inode *olddir, struct dentry *oldent,
		       struct inode *newdir, struct dentry *newent)
{
	int err;
	struct fuse_rename_in inarg;
	struct fuse_conn *fc = get_fuse_conn(olddir);
728
	struct fuse_req *req = fuse_get_req(fc);
729

730 731
	if (IS_ERR(req))
		return PTR_ERR(req);
732 733 734 735 736 737 738 739 740 741 742 743

	memset(&inarg, 0, sizeof(inarg));
	inarg.newdir = get_node_id(newdir);
	req->in.h.opcode = FUSE_RENAME;
	req->in.h.nodeid = get_node_id(olddir);
	req->in.numargs = 3;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->in.args[1].size = oldent->d_name.len + 1;
	req->in.args[1].value = oldent->d_name.name;
	req->in.args[2].size = newent->d_name.len + 1;
	req->in.args[2].value = newent->d_name.name;
744
	fuse_request_send(fc, req);
745 746 747
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (!err) {
748 749 750
		/* ctime changes */
		fuse_invalidate_attr(oldent->d_inode);

751 752 753
		fuse_invalidate_attr(olddir);
		if (olddir != newdir)
			fuse_invalidate_attr(newdir);
754 755

		/* newent will end up negative */
M
Miklos Szeredi 已提交
756 757
		if (newent->d_inode) {
			fuse_invalidate_attr(newent->d_inode);
758
			fuse_invalidate_entry_cache(newent);
M
Miklos Szeredi 已提交
759
		}
760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780
	} else if (err == -EINTR) {
		/* If request was interrupted, DEITY only knows if the
		   rename actually took place.  If the invalidation
		   fails (e.g. some process has CWD under the renamed
		   directory), then there can be inconsistency between
		   the dcache and the real filesystem.  Tough luck. */
		fuse_invalidate_entry(oldent);
		if (newent->d_inode)
			fuse_invalidate_entry(newent);
	}

	return err;
}

static int fuse_link(struct dentry *entry, struct inode *newdir,
		     struct dentry *newent)
{
	int err;
	struct fuse_link_in inarg;
	struct inode *inode = entry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
781 782 783
	struct fuse_req *req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799

	memset(&inarg, 0, sizeof(inarg));
	inarg.oldnodeid = get_node_id(inode);
	req->in.h.opcode = FUSE_LINK;
	req->in.numargs = 2;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->in.args[1].size = newent->d_name.len + 1;
	req->in.args[1].value = newent->d_name.name;
	err = create_new_entry(fc, req, newdir, newent, inode->i_mode);
	/* Contrary to "normal" filesystems it can happen that link
	   makes two "logical" inodes point to the same "physical"
	   inode.  We invalidate the attributes of the old one, so it
	   will reflect changes in the backing inode (link count,
	   etc.)
	*/
M
Miklos Szeredi 已提交
800 801 802 803 804 805 806
	if (!err) {
		struct fuse_inode *fi = get_fuse_inode(inode);

		spin_lock(&fc->lock);
		fi->attr_version = ++fc->attr_version;
		inc_nlink(inode);
		spin_unlock(&fc->lock);
807
		fuse_invalidate_attr(inode);
M
Miklos Szeredi 已提交
808 809 810
	} else if (err == -EINTR) {
		fuse_invalidate_attr(inode);
	}
811 812 813
	return err;
}

814 815 816
static void fuse_fillattr(struct inode *inode, struct fuse_attr *attr,
			  struct kstat *stat)
{
M
Miklos Szeredi 已提交
817 818
	unsigned int blkbits;

819 820 821 822 823 824 825 826 827 828 829 830 831 832 833
	stat->dev = inode->i_sb->s_dev;
	stat->ino = attr->ino;
	stat->mode = (inode->i_mode & S_IFMT) | (attr->mode & 07777);
	stat->nlink = attr->nlink;
	stat->uid = attr->uid;
	stat->gid = attr->gid;
	stat->rdev = inode->i_rdev;
	stat->atime.tv_sec = attr->atime;
	stat->atime.tv_nsec = attr->atimensec;
	stat->mtime.tv_sec = attr->mtime;
	stat->mtime.tv_nsec = attr->mtimensec;
	stat->ctime.tv_sec = attr->ctime;
	stat->ctime.tv_nsec = attr->ctimensec;
	stat->size = attr->size;
	stat->blocks = attr->blocks;
M
Miklos Szeredi 已提交
834 835 836 837 838 839 840

	if (attr->blksize != 0)
		blkbits = ilog2(attr->blksize);
	else
		blkbits = inode->i_sb->s_blocksize_bits;

	stat->blksize = 1 << blkbits;
841 842
}

843 844
static int fuse_do_getattr(struct inode *inode, struct kstat *stat,
			   struct file *file)
845 846
{
	int err;
847 848
	struct fuse_getattr_in inarg;
	struct fuse_attr_out outarg;
849
	struct fuse_conn *fc = get_fuse_conn(inode);
850 851 852 853
	struct fuse_req *req;
	u64 attr_version;

	req = fuse_get_req(fc);
854 855
	if (IS_ERR(req))
		return PTR_ERR(req);
856

857
	attr_version = fuse_get_attr_version(fc);
858

859
	memset(&inarg, 0, sizeof(inarg));
860
	memset(&outarg, 0, sizeof(outarg));
861 862 863 864 865 866 867
	/* Directories have separate file-handle space */
	if (file && S_ISREG(inode->i_mode)) {
		struct fuse_file *ff = file->private_data;

		inarg.getattr_flags |= FUSE_GETATTR_FH;
		inarg.fh = ff->fh;
	}
868 869
	req->in.h.opcode = FUSE_GETATTR;
	req->in.h.nodeid = get_node_id(inode);
870 871 872
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
873
	req->out.numargs = 1;
874 875 876 877
	if (fc->minor < 9)
		req->out.args[0].size = FUSE_COMPAT_ATTR_OUT_SIZE;
	else
		req->out.args[0].size = sizeof(outarg);
878
	req->out.args[0].value = &outarg;
879
	fuse_request_send(fc, req);
880 881 882
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (!err) {
883
		if ((inode->i_mode ^ outarg.attr.mode) & S_IFMT) {
884 885 886
			make_bad_inode(inode);
			err = -EIO;
		} else {
887 888
			fuse_change_attributes(inode, &outarg.attr,
					       attr_timeout(&outarg),
889 890
					       attr_version);
			if (stat)
891
				fuse_fillattr(inode, &outarg.attr, stat);
892 893 894 895 896
		}
	}
	return err;
}

M
Miklos Szeredi 已提交
897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912
int fuse_update_attributes(struct inode *inode, struct kstat *stat,
			   struct file *file, bool *refreshed)
{
	struct fuse_inode *fi = get_fuse_inode(inode);
	int err;
	bool r;

	if (fi->i_time < get_jiffies_64()) {
		r = true;
		err = fuse_do_getattr(inode, stat, file);
	} else {
		r = false;
		err = 0;
		if (stat) {
			generic_fillattr(inode, stat);
			stat->mode = fi->orig_i_mode;
913
			stat->ino = fi->orig_ino;
M
Miklos Szeredi 已提交
914 915 916 917 918 919 920 921 922
		}
	}

	if (refreshed != NULL)
		*refreshed = r;

	return err;
}

J
John Muir 已提交
923
int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
924
			     u64 child_nodeid, struct qstr *name)
J
John Muir 已提交
925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
{
	int err = -ENOTDIR;
	struct inode *parent;
	struct dentry *dir;
	struct dentry *entry;

	parent = ilookup5(sb, parent_nodeid, fuse_inode_eq, &parent_nodeid);
	if (!parent)
		return -ENOENT;

	mutex_lock(&parent->i_mutex);
	if (!S_ISDIR(parent->i_mode))
		goto unlock;

	err = -ENOENT;
	dir = d_find_alias(parent);
	if (!dir)
		goto unlock;

	entry = d_lookup(dir, name);
	dput(dir);
	if (!entry)
		goto unlock;

	fuse_invalidate_attr(parent);
	fuse_invalidate_entry(entry);
951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979

	if (child_nodeid != 0 && entry->d_inode) {
		mutex_lock(&entry->d_inode->i_mutex);
		if (get_node_id(entry->d_inode) != child_nodeid) {
			err = -ENOENT;
			goto badentry;
		}
		if (d_mountpoint(entry)) {
			err = -EBUSY;
			goto badentry;
		}
		if (S_ISDIR(entry->d_inode->i_mode)) {
			shrink_dcache_parent(entry);
			if (!simple_empty(entry)) {
				err = -ENOTEMPTY;
				goto badentry;
			}
			entry->d_inode->i_flags |= S_DEAD;
		}
		dont_mount(entry);
		clear_nlink(entry->d_inode);
		err = 0;
 badentry:
		mutex_unlock(&entry->d_inode->i_mutex);
		if (!err)
			d_delete(entry);
	} else {
		err = 0;
	}
J
John Muir 已提交
980 981 982 983 984 985 986 987
	dput(entry);

 unlock:
	mutex_unlock(&parent->i_mutex);
	iput(parent);
	return err;
}

988 989 990 991 992 993 994 995 996 997 998 999 1000
/*
 * Calling into a user-controlled filesystem gives the filesystem
 * daemon ptrace-like capabilities over the requester process.  This
 * means, that the filesystem daemon is able to record the exact
 * filesystem operations performed, and can also control the behavior
 * of the requester process in otherwise impossible ways.  For example
 * it can delay the operation for arbitrary length of time allowing
 * DoS against the requester.
 *
 * For this reason only those processes can call into the filesystem,
 * for which the owner of the mount has ptrace privilege.  This
 * excludes processes started by other users, suid or sgid processes.
 */
M
Miklos Szeredi 已提交
1001
int fuse_allow_task(struct fuse_conn *fc, struct task_struct *task)
1002
{
1003 1004
	const struct cred *cred;
	int ret;
1005

1006
	if (fc->flags & FUSE_ALLOW_OTHER)
1007 1008
		return 1;

1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
	rcu_read_lock();
	ret = 0;
	cred = __task_cred(task);
	if (cred->euid == fc->user_id &&
	    cred->suid == fc->user_id &&
	    cred->uid  == fc->user_id &&
	    cred->egid == fc->group_id &&
	    cred->sgid == fc->group_id &&
	    cred->gid  == fc->group_id)
		ret = 1;
	rcu_read_unlock();

	return ret;
1022 1023
}

M
Miklos Szeredi 已提交
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
static int fuse_access(struct inode *inode, int mask)
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	struct fuse_access_in inarg;
	int err;

	if (fc->no_access)
		return 0;

1034 1035 1036
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
M
Miklos Szeredi 已提交
1037 1038

	memset(&inarg, 0, sizeof(inarg));
1039
	inarg.mask = mask & (MAY_READ | MAY_WRITE | MAY_EXEC);
M
Miklos Szeredi 已提交
1040 1041 1042 1043 1044
	req->in.h.opcode = FUSE_ACCESS;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
1045
	fuse_request_send(fc, req);
M
Miklos Szeredi 已提交
1046 1047 1048 1049 1050 1051 1052 1053 1054
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
		fc->no_access = 1;
		err = 0;
	}
	return err;
}

1055
static int fuse_perm_getattr(struct inode *inode, int mask)
1056
{
1057
	if (mask & MAY_NOT_BLOCK)
1058 1059 1060 1061 1062
		return -ECHILD;

	return fuse_do_getattr(inode, NULL, NULL);
}

1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
/*
 * Check permission.  The two basic access models of FUSE are:
 *
 * 1) Local access checking ('default_permissions' mount option) based
 * on file mode.  This is the plain old disk filesystem permission
 * modell.
 *
 * 2) "Remote" access checking, where server is responsible for
 * checking permission in each inode operation.  An exception to this
 * is if ->permission() was invoked from sys_access() in which case an
 * access request is sent.  Execute permission is still checked
 * locally based on file mode.
 */
1076
static int fuse_permission(struct inode *inode, int mask)
1077 1078
{
	struct fuse_conn *fc = get_fuse_conn(inode);
1079 1080
	bool refreshed = false;
	int err = 0;
1081

1082
	if (!fuse_allow_task(fc, current))
1083
		return -EACCES;
1084 1085

	/*
1086
	 * If attributes are needed, refresh them before proceeding
1087
	 */
1088 1089
	if ((fc->flags & FUSE_DEFAULT_PERMISSIONS) ||
	    ((mask & MAY_EXEC) && S_ISREG(inode->i_mode))) {
1090 1091 1092 1093 1094
		struct fuse_inode *fi = get_fuse_inode(inode);

		if (fi->i_time < get_jiffies_64()) {
			refreshed = true;

1095
			err = fuse_perm_getattr(inode, mask);
1096 1097 1098
			if (err)
				return err;
		}
1099 1100 1101
	}

	if (fc->flags & FUSE_DEFAULT_PERMISSIONS) {
1102
		err = generic_permission(inode, mask);
M
Miklos Szeredi 已提交
1103 1104 1105 1106

		/* If permission is denied, try to refresh file
		   attributes.  This is also needed, because the root
		   node will at first have no permissions */
1107
		if (err == -EACCES && !refreshed) {
1108
			err = fuse_perm_getattr(inode, mask);
M
Miklos Szeredi 已提交
1109
			if (!err)
1110
				err = generic_permission(inode, mask);
M
Miklos Szeredi 已提交
1111 1112
		}

1113 1114 1115 1116
		/* Note: the opposite of the above test does not
		   exist.  So if permissions are revoked this won't be
		   noticed immediately, only after the attribute
		   timeout has expired */
E
Eric Paris 已提交
1117
	} else if (mask & (MAY_ACCESS | MAY_CHDIR)) {
1118
		if (mask & MAY_NOT_BLOCK)
1119 1120
			return -ECHILD;

1121 1122 1123 1124 1125 1126
		err = fuse_access(inode, mask);
	} else if ((mask & MAY_EXEC) && S_ISREG(inode->i_mode)) {
		if (!(inode->i_mode & S_IXUGO)) {
			if (refreshed)
				return -EACCES;

1127
			err = fuse_perm_getattr(inode, mask);
1128 1129 1130
			if (!err && !(inode->i_mode & S_IXUGO))
				return -EACCES;
		}
1131
	}
1132
	return err;
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159
}

static int parse_dirfile(char *buf, size_t nbytes, struct file *file,
			 void *dstbuf, filldir_t filldir)
{
	while (nbytes >= FUSE_NAME_OFFSET) {
		struct fuse_dirent *dirent = (struct fuse_dirent *) buf;
		size_t reclen = FUSE_DIRENT_SIZE(dirent);
		int over;
		if (!dirent->namelen || dirent->namelen > FUSE_NAME_MAX)
			return -EIO;
		if (reclen > nbytes)
			break;

		over = filldir(dstbuf, dirent->name, dirent->namelen,
			       file->f_pos, dirent->ino, dirent->type);
		if (over)
			break;

		buf += reclen;
		nbytes -= reclen;
		file->f_pos = dirent->off;
	}

	return 0;
}

1160
static int fuse_readdir(struct file *file, void *dstbuf, filldir_t filldir)
1161
{
1162 1163 1164
	int err;
	size_t nbytes;
	struct page *page;
J
Josef Sipek 已提交
1165
	struct inode *inode = file->f_path.dentry->d_inode;
1166
	struct fuse_conn *fc = get_fuse_conn(inode);
1167 1168 1169 1170 1171
	struct fuse_req *req;

	if (is_bad_inode(inode))
		return -EIO;

1172 1173 1174
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
1175

1176 1177 1178 1179 1180
	page = alloc_page(GFP_KERNEL);
	if (!page) {
		fuse_put_request(fc, req);
		return -ENOMEM;
	}
1181
	req->out.argpages = 1;
1182 1183
	req->num_pages = 1;
	req->pages[0] = page;
1184
	fuse_read_fill(req, file, file->f_pos, PAGE_SIZE, FUSE_READDIR);
1185
	fuse_request_send(fc, req);
1186
	nbytes = req->out.args[0].size;
1187 1188 1189
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (!err)
1190 1191
		err = parse_dirfile(page_address(page), nbytes, file, dstbuf,
				    filldir);
1192

1193
	__free_page(page);
1194
	fuse_invalidate_attr(inode); /* atime changed */
1195
	return err;
1196 1197 1198 1199 1200 1201
}

static char *read_link(struct dentry *dentry)
{
	struct inode *inode = dentry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
1202
	struct fuse_req *req = fuse_get_req(fc);
1203 1204
	char *link;

1205
	if (IS_ERR(req))
1206
		return ERR_CAST(req);
1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218

	link = (char *) __get_free_page(GFP_KERNEL);
	if (!link) {
		link = ERR_PTR(-ENOMEM);
		goto out;
	}
	req->in.h.opcode = FUSE_READLINK;
	req->in.h.nodeid = get_node_id(inode);
	req->out.argvar = 1;
	req->out.numargs = 1;
	req->out.args[0].size = PAGE_SIZE - 1;
	req->out.args[0].value = link;
1219
	fuse_request_send(fc, req);
1220 1221 1222 1223 1224 1225 1226
	if (req->out.h.error) {
		free_page((unsigned long) link);
		link = ERR_PTR(req->out.h.error);
	} else
		link[req->out.args[0].size] = '\0';
 out:
	fuse_put_request(fc, req);
1227
	fuse_invalidate_attr(inode); /* atime changed */
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
	return link;
}

static void free_link(char *link)
{
	if (!IS_ERR(link))
		free_page((unsigned long) link);
}

static void *fuse_follow_link(struct dentry *dentry, struct nameidata *nd)
{
	nd_set_link(nd, read_link(dentry));
	return NULL;
}

static void fuse_put_link(struct dentry *dentry, struct nameidata *nd, void *c)
{
	free_link(nd_get_link(nd));
}

static int fuse_dir_open(struct inode *inode, struct file *file)
{
1250
	return fuse_open_common(inode, file, true);
1251 1252 1253 1254
}

static int fuse_dir_release(struct inode *inode, struct file *file)
{
1255 1256 1257
	fuse_release_common(file, FUSE_RELEASEDIR);

	return 0;
1258 1259
}

1260 1261
static int fuse_dir_fsync(struct file *file, loff_t start, loff_t end,
			  int datasync)
1262
{
1263
	return fuse_fsync_common(file, start, end, datasync, 1);
1264 1265
}

1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
static long fuse_dir_ioctl(struct file *file, unsigned int cmd,
			    unsigned long arg)
{
	struct fuse_conn *fc = get_fuse_conn(file->f_mapping->host);

	/* FUSE_IOCTL_DIR only supported for API version >= 7.18 */
	if (fc->minor < 18)
		return -ENOTTY;

	return fuse_ioctl_common(file, cmd, arg, FUSE_IOCTL_DIR);
}

static long fuse_dir_compat_ioctl(struct file *file, unsigned int cmd,
				   unsigned long arg)
{
	struct fuse_conn *fc = get_fuse_conn(file->f_mapping->host);

	if (fc->minor < 18)
		return -ENOTTY;

	return fuse_ioctl_common(file, cmd, arg,
				 FUSE_IOCTL_COMPAT | FUSE_IOCTL_DIR);
}

M
Miklos Szeredi 已提交
1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
static bool update_mtime(unsigned ivalid)
{
	/* Always update if mtime is explicitly set  */
	if (ivalid & ATTR_MTIME_SET)
		return true;

	/* If it's an open(O_TRUNC) or an ftruncate(), don't update */
	if ((ivalid & ATTR_SIZE) && (ivalid & (ATTR_OPEN | ATTR_FILE)))
		return false;

	/* In all other cases update */
	return true;
}

1304
static void iattr_to_fattr(struct iattr *iattr, struct fuse_setattr_in *arg)
1305 1306 1307 1308
{
	unsigned ivalid = iattr->ia_valid;

	if (ivalid & ATTR_MODE)
1309
		arg->valid |= FATTR_MODE,   arg->mode = iattr->ia_mode;
1310
	if (ivalid & ATTR_UID)
1311
		arg->valid |= FATTR_UID,    arg->uid = iattr->ia_uid;
1312
	if (ivalid & ATTR_GID)
1313
		arg->valid |= FATTR_GID,    arg->gid = iattr->ia_gid;
1314
	if (ivalid & ATTR_SIZE)
1315
		arg->valid |= FATTR_SIZE,   arg->size = iattr->ia_size;
M
Miklos Szeredi 已提交
1316 1317
	if (ivalid & ATTR_ATIME) {
		arg->valid |= FATTR_ATIME;
1318
		arg->atime = iattr->ia_atime.tv_sec;
M
Miklos Szeredi 已提交
1319 1320 1321 1322 1323 1324
		arg->atimensec = iattr->ia_atime.tv_nsec;
		if (!(ivalid & ATTR_ATIME_SET))
			arg->valid |= FATTR_ATIME_NOW;
	}
	if ((ivalid & ATTR_MTIME) && update_mtime(ivalid)) {
		arg->valid |= FATTR_MTIME;
1325
		arg->mtime = iattr->ia_mtime.tv_sec;
M
Miklos Szeredi 已提交
1326 1327 1328
		arg->mtimensec = iattr->ia_mtime.tv_nsec;
		if (!(ivalid & ATTR_MTIME_SET))
			arg->valid |= FATTR_MTIME_NOW;
1329
	}
1330 1331
}

M
Miklos Szeredi 已提交
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
/*
 * Prevent concurrent writepages on inode
 *
 * This is done by adding a negative bias to the inode write counter
 * and waiting for all pending writes to finish.
 */
void fuse_set_nowrite(struct inode *inode)
{
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_inode *fi = get_fuse_inode(inode);

	BUG_ON(!mutex_is_locked(&inode->i_mutex));

	spin_lock(&fc->lock);
	BUG_ON(fi->writectr < 0);
	fi->writectr += FUSE_NOWRITE;
	spin_unlock(&fc->lock);
	wait_event(fi->page_waitq, fi->writectr == FUSE_NOWRITE);
}

/*
 * Allow writepages on inode
 *
 * Remove the bias from the writecounter and send any queued
 * writepages.
 */
static void __fuse_release_nowrite(struct inode *inode)
{
	struct fuse_inode *fi = get_fuse_inode(inode);

	BUG_ON(fi->writectr != FUSE_NOWRITE);
	fi->writectr = 0;
	fuse_flush_writepages(inode);
}

void fuse_release_nowrite(struct inode *inode)
{
	struct fuse_conn *fc = get_fuse_conn(inode);

	spin_lock(&fc->lock);
	__fuse_release_nowrite(inode);
	spin_unlock(&fc->lock);
}

1376 1377 1378 1379 1380
/*
 * Set attributes, and at the same time refresh them.
 *
 * Truncation is slightly complicated, because the 'truncate' request
 * may fail, in which case we don't want to touch the mapping.
M
Miklos Szeredi 已提交
1381 1382
 * vmtruncate() doesn't allow for this case, so do the rlimit checking
 * and the actual truncation by hand.
1383
 */
1384 1385
static int fuse_do_setattr(struct dentry *entry, struct iattr *attr,
			   struct file *file)
1386 1387 1388 1389 1390 1391
{
	struct inode *inode = entry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	struct fuse_setattr_in inarg;
	struct fuse_attr_out outarg;
M
Miklos Szeredi 已提交
1392 1393
	bool is_truncate = false;
	loff_t oldsize;
1394 1395
	int err;

M
Miklos Szeredi 已提交
1396 1397 1398
	if (!fuse_allow_task(fc, current))
		return -EACCES;

1399 1400 1401 1402 1403 1404
	if (!(fc->flags & FUSE_DEFAULT_PERMISSIONS))
		attr->ia_valid |= ATTR_FORCE;

	err = inode_change_ok(inode, attr);
	if (err)
		return err;
M
Miklos Szeredi 已提交
1405

M
Miklos Szeredi 已提交
1406 1407 1408 1409 1410
	if (attr->ia_valid & ATTR_OPEN) {
		if (fc->atomic_o_trunc)
			return 0;
		file = NULL;
	}
1411

1412
	if (attr->ia_valid & ATTR_SIZE)
M
Miklos Szeredi 已提交
1413
		is_truncate = true;
1414

1415 1416 1417
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
1418

M
Miklos Szeredi 已提交
1419 1420 1421
	if (is_truncate)
		fuse_set_nowrite(inode);

1422
	memset(&inarg, 0, sizeof(inarg));
1423
	memset(&outarg, 0, sizeof(outarg));
1424
	iattr_to_fattr(attr, &inarg);
1425 1426 1427 1428 1429
	if (file) {
		struct fuse_file *ff = file->private_data;
		inarg.valid |= FATTR_FH;
		inarg.fh = ff->fh;
	}
1430 1431 1432 1433 1434
	if (attr->ia_valid & ATTR_SIZE) {
		/* For mandatory locking in truncate */
		inarg.valid |= FATTR_LOCKOWNER;
		inarg.lock_owner = fuse_lock_owner_id(fc, current->files);
	}
1435 1436 1437 1438 1439 1440
	req->in.h.opcode = FUSE_SETATTR;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->out.numargs = 1;
1441 1442 1443 1444
	if (fc->minor < 9)
		req->out.args[0].size = FUSE_COMPAT_ATTR_OUT_SIZE;
	else
		req->out.args[0].size = sizeof(outarg);
1445
	req->out.args[0].value = &outarg;
1446
	fuse_request_send(fc, req);
1447 1448
	err = req->out.h.error;
	fuse_put_request(fc, req);
1449 1450 1451
	if (err) {
		if (err == -EINTR)
			fuse_invalidate_attr(inode);
M
Miklos Szeredi 已提交
1452
		goto error;
1453
	}
1454

1455 1456
	if ((inode->i_mode ^ outarg.attr.mode) & S_IFMT) {
		make_bad_inode(inode);
M
Miklos Szeredi 已提交
1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477
		err = -EIO;
		goto error;
	}

	spin_lock(&fc->lock);
	fuse_change_attributes_common(inode, &outarg.attr,
				      attr_timeout(&outarg));
	oldsize = inode->i_size;
	i_size_write(inode, outarg.attr.size);

	if (is_truncate) {
		/* NOTE: this may release/reacquire fc->lock */
		__fuse_release_nowrite(inode);
	}
	spin_unlock(&fc->lock);

	/*
	 * Only call invalidate_inode_pages2() after removing
	 * FUSE_NOWRITE, otherwise fuse_launder_page() would deadlock.
	 */
	if (S_ISREG(inode->i_mode) && oldsize != outarg.attr.size) {
N
npiggin@suse.de 已提交
1478
		truncate_pagecache(inode, oldsize, outarg.attr.size);
M
Miklos Szeredi 已提交
1479
		invalidate_inode_pages2(inode->i_mapping);
1480 1481 1482
	}

	return 0;
M
Miklos Szeredi 已提交
1483 1484 1485 1486 1487 1488

error:
	if (is_truncate)
		fuse_release_nowrite(inode);

	return err;
1489 1490
}

1491 1492 1493 1494 1495 1496 1497 1498
static int fuse_setattr(struct dentry *entry, struct iattr *attr)
{
	if (attr->ia_valid & ATTR_FILE)
		return fuse_do_setattr(entry, attr, attr->ia_file);
	else
		return fuse_do_setattr(entry, attr, NULL);
}

1499 1500 1501 1502
static int fuse_getattr(struct vfsmount *mnt, struct dentry *entry,
			struct kstat *stat)
{
	struct inode *inode = entry->d_inode;
1503 1504 1505 1506 1507
	struct fuse_conn *fc = get_fuse_conn(inode);

	if (!fuse_allow_task(fc, current))
		return -EACCES;

M
Miklos Szeredi 已提交
1508
	return fuse_update_attributes(inode, stat, NULL, NULL);
1509 1510
}

1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522
static int fuse_setxattr(struct dentry *entry, const char *name,
			 const void *value, size_t size, int flags)
{
	struct inode *inode = entry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	struct fuse_setxattr_in inarg;
	int err;

	if (fc->no_setxattr)
		return -EOPNOTSUPP;

1523 1524 1525
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538

	memset(&inarg, 0, sizeof(inarg));
	inarg.size = size;
	inarg.flags = flags;
	req->in.h.opcode = FUSE_SETXATTR;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 3;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->in.args[1].size = strlen(name) + 1;
	req->in.args[1].value = name;
	req->in.args[2].size = size;
	req->in.args[2].value = value;
1539
	fuse_request_send(fc, req);
1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
		fc->no_setxattr = 1;
		err = -EOPNOTSUPP;
	}
	return err;
}

static ssize_t fuse_getxattr(struct dentry *entry, const char *name,
			     void *value, size_t size)
{
	struct inode *inode = entry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	struct fuse_getxattr_in inarg;
	struct fuse_getxattr_out outarg;
	ssize_t ret;

	if (fc->no_getxattr)
		return -EOPNOTSUPP;

1562 1563 1564
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584

	memset(&inarg, 0, sizeof(inarg));
	inarg.size = size;
	req->in.h.opcode = FUSE_GETXATTR;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 2;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	req->in.args[1].size = strlen(name) + 1;
	req->in.args[1].value = name;
	/* This is really two different operations rolled into one */
	req->out.numargs = 1;
	if (size) {
		req->out.argvar = 1;
		req->out.args[0].size = size;
		req->out.args[0].value = value;
	} else {
		req->out.args[0].size = sizeof(outarg);
		req->out.args[0].value = &outarg;
	}
1585
	fuse_request_send(fc, req);
1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607
	ret = req->out.h.error;
	if (!ret)
		ret = size ? req->out.args[0].size : outarg.size;
	else {
		if (ret == -ENOSYS) {
			fc->no_getxattr = 1;
			ret = -EOPNOTSUPP;
		}
	}
	fuse_put_request(fc, req);
	return ret;
}

static ssize_t fuse_listxattr(struct dentry *entry, char *list, size_t size)
{
	struct inode *inode = entry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	struct fuse_getxattr_in inarg;
	struct fuse_getxattr_out outarg;
	ssize_t ret;

M
Miklos Szeredi 已提交
1608 1609 1610
	if (!fuse_allow_task(fc, current))
		return -EACCES;

1611 1612 1613
	if (fc->no_listxattr)
		return -EOPNOTSUPP;

1614 1615 1616
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634

	memset(&inarg, 0, sizeof(inarg));
	inarg.size = size;
	req->in.h.opcode = FUSE_LISTXATTR;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = sizeof(inarg);
	req->in.args[0].value = &inarg;
	/* This is really two different operations rolled into one */
	req->out.numargs = 1;
	if (size) {
		req->out.argvar = 1;
		req->out.args[0].size = size;
		req->out.args[0].value = list;
	} else {
		req->out.args[0].size = sizeof(outarg);
		req->out.args[0].value = &outarg;
	}
1635
	fuse_request_send(fc, req);
1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658
	ret = req->out.h.error;
	if (!ret)
		ret = size ? req->out.args[0].size : outarg.size;
	else {
		if (ret == -ENOSYS) {
			fc->no_listxattr = 1;
			ret = -EOPNOTSUPP;
		}
	}
	fuse_put_request(fc, req);
	return ret;
}

static int fuse_removexattr(struct dentry *entry, const char *name)
{
	struct inode *inode = entry->d_inode;
	struct fuse_conn *fc = get_fuse_conn(inode);
	struct fuse_req *req;
	int err;

	if (fc->no_removexattr)
		return -EOPNOTSUPP;

1659 1660 1661
	req = fuse_get_req(fc);
	if (IS_ERR(req))
		return PTR_ERR(req);
1662 1663 1664 1665 1666 1667

	req->in.h.opcode = FUSE_REMOVEXATTR;
	req->in.h.nodeid = get_node_id(inode);
	req->in.numargs = 1;
	req->in.args[0].size = strlen(name) + 1;
	req->in.args[0].value = name;
1668
	fuse_request_send(fc, req);
1669 1670 1671 1672 1673 1674 1675 1676 1677
	err = req->out.h.error;
	fuse_put_request(fc, req);
	if (err == -ENOSYS) {
		fc->no_removexattr = 1;
		err = -EOPNOTSUPP;
	}
	return err;
}

1678
static const struct inode_operations fuse_dir_inode_operations = {
1679
	.lookup		= fuse_lookup,
1680 1681 1682 1683 1684 1685 1686 1687
	.mkdir		= fuse_mkdir,
	.symlink	= fuse_symlink,
	.unlink		= fuse_unlink,
	.rmdir		= fuse_rmdir,
	.rename		= fuse_rename,
	.link		= fuse_link,
	.setattr	= fuse_setattr,
	.create		= fuse_create,
1688
	.atomic_open	= fuse_atomic_open,
1689
	.mknod		= fuse_mknod,
1690 1691
	.permission	= fuse_permission,
	.getattr	= fuse_getattr,
1692 1693 1694 1695
	.setxattr	= fuse_setxattr,
	.getxattr	= fuse_getxattr,
	.listxattr	= fuse_listxattr,
	.removexattr	= fuse_removexattr,
1696 1697
};

1698
static const struct file_operations fuse_dir_operations = {
M
Miklos Szeredi 已提交
1699
	.llseek		= generic_file_llseek,
1700 1701 1702 1703
	.read		= generic_read_dir,
	.readdir	= fuse_readdir,
	.open		= fuse_dir_open,
	.release	= fuse_dir_release,
1704
	.fsync		= fuse_dir_fsync,
1705 1706
	.unlocked_ioctl	= fuse_dir_ioctl,
	.compat_ioctl	= fuse_dir_compat_ioctl,
1707 1708
};

1709
static const struct inode_operations fuse_common_inode_operations = {
1710
	.setattr	= fuse_setattr,
1711 1712
	.permission	= fuse_permission,
	.getattr	= fuse_getattr,
1713 1714 1715 1716
	.setxattr	= fuse_setxattr,
	.getxattr	= fuse_getxattr,
	.listxattr	= fuse_listxattr,
	.removexattr	= fuse_removexattr,
1717 1718
};

1719
static const struct inode_operations fuse_symlink_inode_operations = {
1720
	.setattr	= fuse_setattr,
1721 1722 1723 1724
	.follow_link	= fuse_follow_link,
	.put_link	= fuse_put_link,
	.readlink	= generic_readlink,
	.getattr	= fuse_getattr,
1725 1726 1727 1728
	.setxattr	= fuse_setxattr,
	.getxattr	= fuse_getxattr,
	.listxattr	= fuse_listxattr,
	.removexattr	= fuse_removexattr,
1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745
};

void fuse_init_common(struct inode *inode)
{
	inode->i_op = &fuse_common_inode_operations;
}

void fuse_init_dir(struct inode *inode)
{
	inode->i_op = &fuse_dir_inode_operations;
	inode->i_fop = &fuse_dir_operations;
}

void fuse_init_symlink(struct inode *inode)
{
	inode->i_op = &fuse_symlink_inode_operations;
}