dir.c 20.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10
/*
 * dir.c - Operations for sysfs directories.
 */

#undef DEBUG

#include <linux/fs.h>
#include <linux/mount.h>
#include <linux/module.h>
#include <linux/kobject.h>
11
#include <linux/namei.h>
12
#include <linux/idr.h>
13
#include <linux/completion.h>
14
#include <asm/semaphore.h>
L
Linus Torvalds 已提交
15 16 17
#include "sysfs.h"

DECLARE_RWSEM(sysfs_rename_sem);
18
spinlock_t sysfs_lock = SPIN_LOCK_UNLOCKED;
19
spinlock_t kobj_sysfs_assoc_lock = SPIN_LOCK_UNLOCKED;
L
Linus Torvalds 已提交
20

21 22 23
static spinlock_t sysfs_ino_lock = SPIN_LOCK_UNLOCKED;
static DEFINE_IDA(sysfs_ino_ida);

24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
/**
 *	sysfs_link_sibling - link sysfs_dirent into sibling list
 *	@sd: sysfs_dirent of interest
 *
 *	Link @sd into its sibling list which starts from
 *	sd->s_parent->s_children.
 *
 *	Locking:
 *	mutex_lock(sd->s_parent->dentry->d_inode->i_mutex)
 */
static void sysfs_link_sibling(struct sysfs_dirent *sd)
{
	struct sysfs_dirent *parent_sd = sd->s_parent;

	BUG_ON(sd->s_sibling);
	sd->s_sibling = parent_sd->s_children;
	parent_sd->s_children = sd;
}

/**
 *	sysfs_unlink_sibling - unlink sysfs_dirent from sibling list
 *	@sd: sysfs_dirent of interest
 *
 *	Unlink @sd from its sibling list which starts from
 *	sd->s_parent->s_children.
 *
 *	Locking:
 *	mutex_lock(sd->s_parent->dentry->d_inode->i_mutex)
 */
static void sysfs_unlink_sibling(struct sysfs_dirent *sd)
{
	struct sysfs_dirent **pos;

	for (pos = &sd->s_parent->s_children; *pos; pos = &(*pos)->s_sibling) {
		if (*pos == sd) {
			*pos = sd->s_sibling;
			sd->s_sibling = NULL;
			break;
		}
	}
}

66 67 68 69 70 71 72 73 74 75 76 77
/**
 *	sysfs_get_active - get an active reference to sysfs_dirent
 *	@sd: sysfs_dirent to get an active reference to
 *
 *	Get an active reference of @sd.  This function is noop if @sd
 *	is NULL.
 *
 *	RETURNS:
 *	Pointer to @sd on success, NULL on failure.
 */
struct sysfs_dirent *sysfs_get_active(struct sysfs_dirent *sd)
{
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
	if (unlikely(!sd))
		return NULL;

	while (1) {
		int v, t;

		v = atomic_read(&sd->s_active);
		if (unlikely(v < 0))
			return NULL;

		t = atomic_cmpxchg(&sd->s_active, v, v + 1);
		if (likely(t == v))
			return sd;
		if (t < 0)
			return NULL;

		cpu_relax();
95 96 97 98 99 100 101 102 103 104 105 106
	}
}

/**
 *	sysfs_put_active - put an active reference to sysfs_dirent
 *	@sd: sysfs_dirent to put an active reference to
 *
 *	Put an active reference to @sd.  This function is noop if @sd
 *	is NULL.
 */
void sysfs_put_active(struct sysfs_dirent *sd)
{
107 108 109 110 111 112 113 114 115 116 117
	struct completion *cmpl;
	int v;

	if (unlikely(!sd))
		return;

	v = atomic_dec_return(&sd->s_active);
	if (likely(v != SD_DEACTIVATED_BIAS))
		return;

	/* atomic_dec_return() is a mb(), we'll always see the updated
118
	 * sd->s_sibling.
119
	 */
120
	cmpl = (void *)sd->s_sibling;
121
	complete(cmpl);
122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
}

/**
 *	sysfs_get_active_two - get active references to sysfs_dirent and parent
 *	@sd: sysfs_dirent of interest
 *
 *	Get active reference to @sd and its parent.  Parent's active
 *	reference is grabbed first.  This function is noop if @sd is
 *	NULL.
 *
 *	RETURNS:
 *	Pointer to @sd on success, NULL on failure.
 */
struct sysfs_dirent *sysfs_get_active_two(struct sysfs_dirent *sd)
{
	if (sd) {
		if (sd->s_parent && unlikely(!sysfs_get_active(sd->s_parent)))
			return NULL;
		if (unlikely(!sysfs_get_active(sd))) {
			sysfs_put_active(sd->s_parent);
			return NULL;
		}
	}
	return sd;
}

/**
 *	sysfs_put_active_two - put active references to sysfs_dirent and parent
 *	@sd: sysfs_dirent of interest
 *
 *	Put active references to @sd and its parent.  This function is
 *	noop if @sd is NULL.
 */
void sysfs_put_active_two(struct sysfs_dirent *sd)
{
	if (sd) {
		sysfs_put_active(sd);
		sysfs_put_active(sd->s_parent);
	}
}

/**
 *	sysfs_deactivate - deactivate sysfs_dirent
 *	@sd: sysfs_dirent to deactivate
 *
167
 *	Deny new active references and drain existing ones.
168 169 170
 */
void sysfs_deactivate(struct sysfs_dirent *sd)
{
171 172
	DECLARE_COMPLETION_ONSTACK(wait);
	int v;
173

174
	BUG_ON(sd->s_sibling || !(sd->s_flags & SYSFS_FLAG_REMOVED));
175
	sd->s_sibling = (void *)&wait;
176 177

	/* atomic_add_return() is a mb(), put_active() will always see
178
	 * the updated sd->s_sibling.
179
	 */
180 181 182 183 184
	v = atomic_add_return(SD_DEACTIVATED_BIAS, &sd->s_active);

	if (v != SD_DEACTIVATED_BIAS)
		wait_for_completion(&wait);

185
	sd->s_sibling = NULL;
186 187
}

T
Tejun Heo 已提交
188
static int sysfs_alloc_ino(ino_t *pino)
189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
{
	int ino, rc;

 retry:
	spin_lock(&sysfs_ino_lock);
	rc = ida_get_new_above(&sysfs_ino_ida, 2, &ino);
	spin_unlock(&sysfs_ino_lock);

	if (rc == -EAGAIN) {
		if (ida_pre_get(&sysfs_ino_ida, GFP_KERNEL))
			goto retry;
		rc = -ENOMEM;
	}

	*pino = ino;
	return rc;
}

static void sysfs_free_ino(ino_t ino)
{
	spin_lock(&sysfs_ino_lock);
	ida_remove(&sysfs_ino_ida, ino);
	spin_unlock(&sysfs_ino_lock);
}

214 215
void release_sysfs_dirent(struct sysfs_dirent * sd)
{
T
Tejun Heo 已提交
216 217 218 219 220
	struct sysfs_dirent *parent_sd;

 repeat:
	parent_sd = sd->s_parent;

221
	if (sysfs_type(sd) == SYSFS_KOBJ_LINK)
222
		sysfs_put(sd->s_elem.symlink.target_sd);
223
	if (sysfs_type(sd) & SYSFS_COPY_NAME)
T
Tejun Heo 已提交
224
		kfree(sd->s_name);
225
	kfree(sd->s_iattr);
226
	sysfs_free_ino(sd->s_ino);
227
	kmem_cache_free(sysfs_dir_cachep, sd);
T
Tejun Heo 已提交
228 229 230 231

	sd = parent_sd;
	if (sd && atomic_dec_and_test(&sd->s_count))
		goto repeat;
232 233
}

L
Linus Torvalds 已提交
234 235 236 237 238
static void sysfs_d_iput(struct dentry * dentry, struct inode * inode)
{
	struct sysfs_dirent * sd = dentry->d_fsdata;

	if (sd) {
239 240 241 242 243 244 245 246 247 248 249 250 251
		/* sd->s_dentry is protected with sysfs_lock.  This
		 * allows sysfs_drop_dentry() to dereference it.
		 */
		spin_lock(&sysfs_lock);

		/* The dentry might have been deleted or another
		 * lookup could have happened updating sd->s_dentry to
		 * point the new dentry.  Ignore if it isn't pointing
		 * to this dentry.
		 */
		if (sd->s_dentry == dentry)
			sd->s_dentry = NULL;
		spin_unlock(&sysfs_lock);
L
Linus Torvalds 已提交
252 253 254 255 256 257 258 259 260
		sysfs_put(sd);
	}
	iput(inode);
}

static struct dentry_operations sysfs_dentry_ops = {
	.d_iput		= sysfs_d_iput,
};

261
struct sysfs_dirent *sysfs_new_dirent(const char *name, umode_t mode, int type)
L
Linus Torvalds 已提交
262
{
T
Tejun Heo 已提交
263 264 265 266 267 268 269 270
	char *dup_name = NULL;
	struct sysfs_dirent *sd = NULL;

	if (type & SYSFS_COPY_NAME) {
		name = dup_name = kstrdup(name, GFP_KERNEL);
		if (!name)
			goto err_out;
	}
L
Linus Torvalds 已提交
271

272
	sd = kmem_cache_zalloc(sysfs_dir_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
273
	if (!sd)
T
Tejun Heo 已提交
274
		goto err_out;
L
Linus Torvalds 已提交
275

T
Tejun Heo 已提交
276 277
	if (sysfs_alloc_ino(&sd->s_ino))
		goto err_out;
278

L
Linus Torvalds 已提交
279
	atomic_set(&sd->s_count, 1);
280
	atomic_set(&sd->s_active, 0);
281
	atomic_set(&sd->s_event, 1);
282

T
Tejun Heo 已提交
283
	sd->s_name = name;
284
	sd->s_mode = mode;
285
	sd->s_flags = type;
L
Linus Torvalds 已提交
286 287

	return sd;
T
Tejun Heo 已提交
288 289 290 291 292

 err_out:
	kfree(dup_name);
	kmem_cache_free(sysfs_dir_cachep, sd);
	return NULL;
L
Linus Torvalds 已提交
293 294
}

295 296 297 298 299 300 301 302 303 304 305 306 307
static void sysfs_attach_dentry(struct sysfs_dirent *sd, struct dentry *dentry)
{
	dentry->d_op = &sysfs_dentry_ops;
	dentry->d_fsdata = sysfs_get(sd);

	/* protect sd->s_dentry against sysfs_d_iput */
	spin_lock(&sysfs_lock);
	sd->s_dentry = dentry;
	spin_unlock(&sysfs_lock);

	d_rehash(dentry);
}

308 309
void sysfs_attach_dirent(struct sysfs_dirent *sd,
			 struct sysfs_dirent *parent_sd, struct dentry *dentry)
310
{
311 312
	if (dentry)
		sysfs_attach_dentry(sd, dentry);
313

T
Tejun Heo 已提交
314 315
	if (parent_sd) {
		sd->s_parent = sysfs_get(parent_sd);
316
		sysfs_link_sibling(sd);
T
Tejun Heo 已提交
317
	}
318 319
}

320
/*
321 322 323 324 325 326 327 328 329 330 331
 *
 * Return -EEXIST if there is already a sysfs element with the same name for
 * the same parent.
 *
 * called with parent inode's i_mutex held
 */
int sysfs_dirent_exist(struct sysfs_dirent *parent_sd,
			  const unsigned char *new)
{
	struct sysfs_dirent * sd;

332
	for (sd = parent_sd->s_children; sd; sd = sd->s_sibling) {
333
		if (sysfs_type(sd)) {
T
Tejun Heo 已提交
334
			if (strcmp(sd->s_name, new))
335 336 337 338 339 340 341 342 343
				continue;
			else
				return -EEXIST;
		}
	}

	return 0;
}

344 345
static int create_dir(struct kobject *kobj, struct dentry *parent,
		      const char *name, struct dentry **p_dentry)
L
Linus Torvalds 已提交
346 347 348
{
	int error;
	umode_t mode = S_IFDIR| S_IRWXU | S_IRUGO | S_IXUGO;
349
	struct dentry *dentry;
350
	struct inode *inode;
351
	struct sysfs_dirent *sd;
L
Linus Torvalds 已提交
352

353 354
	mutex_lock(&parent->d_inode->i_mutex);

355
	/* allocate */
356 357 358 359 360 361 362
	dentry = lookup_one_len(name, parent, strlen(name));
	if (IS_ERR(dentry)) {
		error = PTR_ERR(dentry);
		goto out_unlock;
	}

	error = -EEXIST;
363
	if (dentry->d_inode)
364 365
		goto out_dput;

366
	error = -ENOMEM;
367
	sd = sysfs_new_dirent(name, mode, SYSFS_DIR);
368
	if (!sd)
369
		goto out_drop;
370
	sd->s_elem.dir.kobj = kobj;
371

372
	inode = sysfs_get_inode(sd);
373
	if (!inode)
374 375
		goto out_sput;

376 377 378 379 380 381
	if (inode->i_state & I_NEW) {
		inode->i_op = &sysfs_dir_inode_operations;
		inode->i_fop = &sysfs_dir_operations;
		/* directory inodes start off with i_nlink == 2 (for ".") */
		inc_nlink(inode);
	}
382 383 384 385 386 387 388

	/* link in */
	error = -EEXIST;
	if (sysfs_dirent_exist(parent->d_fsdata, name))
		goto out_iput;

	sysfs_instantiate(dentry, inode);
389
	inc_nlink(parent->d_inode);
390
	sysfs_attach_dirent(sd, parent->d_fsdata, dentry);
391 392 393

	*p_dentry = dentry;
	error = 0;
394
	goto out_unlock;	/* pin directory dentry in core */
395

396 397
 out_iput:
	iput(inode);
398 399 400 401 402 403 404 405
 out_sput:
	sysfs_put(sd);
 out_drop:
	d_drop(dentry);
 out_dput:
	dput(dentry);
 out_unlock:
	mutex_unlock(&parent->d_inode->i_mutex);
L
Linus Torvalds 已提交
406 407 408 409 410 411 412 413 414 415 416 417
	return error;
}


int sysfs_create_subdir(struct kobject * k, const char * n, struct dentry ** d)
{
	return create_dir(k,k->dentry,n,d);
}

/**
 *	sysfs_create_dir - create a directory for an object.
 *	@kobj:		object we're creating directory for. 
418
 *	@shadow_parent:	parent parent object.
L
Linus Torvalds 已提交
419 420
 */

421
int sysfs_create_dir(struct kobject * kobj, struct dentry *shadow_parent)
L
Linus Torvalds 已提交
422 423 424 425 426 427 428
{
	struct dentry * dentry = NULL;
	struct dentry * parent;
	int error = 0;

	BUG_ON(!kobj);

429 430 431
	if (shadow_parent)
		parent = shadow_parent;
	else if (kobj->parent)
L
Linus Torvalds 已提交
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448
		parent = kobj->parent->dentry;
	else if (sysfs_mount && sysfs_mount->mnt_sb)
		parent = sysfs_mount->mnt_sb->s_root;
	else
		return -EFAULT;

	error = create_dir(kobj,parent,kobject_name(kobj),&dentry);
	if (!error)
		kobj->dentry = dentry;
	return error;
}

static struct dentry * sysfs_lookup(struct inode *dir, struct dentry *dentry,
				struct nameidata *nd)
{
	struct sysfs_dirent * parent_sd = dentry->d_parent->d_fsdata;
	struct sysfs_dirent * sd;
449
	struct bin_attribute *bin_attr;
450 451
	struct inode *inode;
	int found = 0;
L
Linus Torvalds 已提交
452

453
	for (sd = parent_sd->s_children; sd; sd = sd->s_sibling) {
454
		if ((sysfs_type(sd) & SYSFS_NOT_PINNED) &&
455 456
		    !strcmp(sd->s_name, dentry->d_name.name)) {
			found = 1;
L
Linus Torvalds 已提交
457 458 459 460
			break;
		}
	}

461 462 463 464 465
	/* no such entry */
	if (!found)
		return NULL;

	/* attach dentry and inode */
466
	inode = sysfs_get_inode(sd);
467 468 469
	if (!inode)
		return ERR_PTR(-ENOMEM);

470 471
	if (inode->i_state & I_NEW) {
		/* initialize inode according to type */
472 473
		switch (sysfs_type(sd)) {
		case SYSFS_KOBJ_ATTR:
474 475
			inode->i_size = PAGE_SIZE;
			inode->i_fop = &sysfs_file_operations;
476 477 478
			break;
		case SYSFS_KOBJ_BIN_ATTR:
			bin_attr = sd->s_elem.bin_attr.bin_attr;
479 480
			inode->i_size = bin_attr->size;
			inode->i_fop = &bin_fops;
481 482
			break;
		case SYSFS_KOBJ_LINK:
483
			inode->i_op = &sysfs_symlink_inode_operations;
484 485 486 487
			break;
		default:
			BUG();
		}
488
	}
489 490 491 492 493

	sysfs_instantiate(dentry, inode);
	sysfs_attach_dentry(sd, dentry);

	return NULL;
L
Linus Torvalds 已提交
494 495
}

496
const struct inode_operations sysfs_dir_inode_operations = {
L
Linus Torvalds 已提交
497
	.lookup		= sysfs_lookup,
498
	.setattr	= sysfs_setattr,
L
Linus Torvalds 已提交
499 500 501 502
};

static void remove_dir(struct dentry * d)
{
503 504
	struct dentry *parent = d->d_parent;
	struct sysfs_dirent *sd = d->d_fsdata;
L
Linus Torvalds 已提交
505

506
	mutex_lock(&parent->d_inode->i_mutex);
507

508
	sysfs_unlink_sibling(sd);
509
	sd->s_flags |= SYSFS_FLAG_REMOVED;
L
Linus Torvalds 已提交
510 511 512 513

	pr_debug(" o %s removing done (%d)\n",d->d_name.name,
		 atomic_read(&d->d_count));

514
	mutex_unlock(&parent->d_inode->i_mutex);
515

516
	sysfs_drop_dentry(sd);
517 518
	sysfs_deactivate(sd);
	sysfs_put(sd);
L
Linus Torvalds 已提交
519 520 521 522 523 524 525 526
}

void sysfs_remove_subdir(struct dentry * d)
{
	remove_dir(d);
}


527
static void __sysfs_remove_dir(struct dentry *dentry)
L
Linus Torvalds 已提交
528
{
529 530 531
	struct sysfs_dirent *removed = NULL;
	struct sysfs_dirent *parent_sd;
	struct sysfs_dirent **pos;
L
Linus Torvalds 已提交
532 533 534 535 536

	if (!dentry)
		return;

	pr_debug("sysfs %s: removing dir\n",dentry->d_name.name);
537
	mutex_lock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
538
	parent_sd = dentry->d_fsdata;
539 540 541 542
	pos = &parent_sd->s_children;
	while (*pos) {
		struct sysfs_dirent *sd = *pos;

543
		if (sysfs_type(sd) && (sysfs_type(sd) & SYSFS_NOT_PINNED)) {
544
			sd->s_flags |= SYSFS_FLAG_REMOVED;
545 546 547 548 549
			*pos = sd->s_sibling;
			sd->s_sibling = removed;
			removed = sd;
		} else
			pos = &(*pos)->s_sibling;
L
Linus Torvalds 已提交
550
	}
551
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
552

553 554 555 556 557 558
	while (removed) {
		struct sysfs_dirent *sd = removed;

		removed = sd->s_sibling;
		sd->s_sibling = NULL;

559
		sysfs_drop_dentry(sd);
560 561 562 563
		sysfs_deactivate(sd);
		sysfs_put(sd);
	}

L
Linus Torvalds 已提交
564
	remove_dir(dentry);
565 566 567 568 569 570 571 572 573 574 575 576 577
}

/**
 *	sysfs_remove_dir - remove an object's directory.
 *	@kobj:	object.
 *
 *	The only thing special about this is that we remove any files in
 *	the directory before we remove the directory, and we've inlined
 *	what used to be sysfs_rmdir() below, instead of calling separately.
 */

void sysfs_remove_dir(struct kobject * kobj)
{
578 579 580
	struct dentry *d = kobj->dentry;

	spin_lock(&kobj_sysfs_assoc_lock);
581
	kobj->dentry = NULL;
582 583 584
	spin_unlock(&kobj_sysfs_assoc_lock);

	__sysfs_remove_dir(d);
L
Linus Torvalds 已提交
585 586
}

587 588
int sysfs_rename_dir(struct kobject * kobj, struct dentry *new_parent,
		     const char *new_name)
L
Linus Torvalds 已提交
589
{
T
Tejun Heo 已提交
590 591 592 593
	struct sysfs_dirent *sd = kobj->dentry->d_fsdata;
	struct sysfs_dirent *parent_sd = new_parent->d_fsdata;
	struct dentry *new_dentry;
	char *dup_name;
594
	int error;
L
Linus Torvalds 已提交
595

596 597
	if (!new_parent)
		return -EFAULT;
L
Linus Torvalds 已提交
598 599

	down_write(&sysfs_rename_sem);
600
	mutex_lock(&new_parent->d_inode->i_mutex);
L
Linus Torvalds 已提交
601

602
	new_dentry = lookup_one_len(new_name, new_parent, strlen(new_name));
603 604 605
	if (IS_ERR(new_dentry)) {
		error = PTR_ERR(new_dentry);
		goto out_unlock;
L
Linus Torvalds 已提交
606
	}
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621

	/* By allowing two different directories with the same
	 * d_parent we allow this routine to move between different
	 * shadows of the same directory
	 */
	error = -EINVAL;
	if (kobj->dentry->d_parent->d_inode != new_parent->d_inode ||
	    new_dentry->d_parent->d_inode != new_parent->d_inode ||
	    new_dentry == kobj->dentry)
		goto out_dput;

	error = -EEXIST;
	if (new_dentry->d_inode)
		goto out_dput;

T
Tejun Heo 已提交
622 623 624 625 626 627
	/* rename kobject and sysfs_dirent */
	error = -ENOMEM;
	new_name = dup_name = kstrdup(new_name, GFP_KERNEL);
	if (!new_name)
		goto out_drop;

628 629
	error = kobject_set_name(kobj, "%s", new_name);
	if (error)
T
Tejun Heo 已提交
630
		goto out_free;
631

T
Tejun Heo 已提交
632 633 634 635
	kfree(sd->s_name);
	sd->s_name = new_name;

	/* move under the new parent */
636 637 638
	d_add(new_dentry, NULL);
	d_move(kobj->dentry, new_dentry);

639
	sysfs_unlink_sibling(sd);
640 641 642
	sysfs_get(parent_sd);
	sysfs_put(sd->s_parent);
	sd->s_parent = parent_sd;
643
	sysfs_link_sibling(sd);
644 645 646 647

	error = 0;
	goto out_unlock;

T
Tejun Heo 已提交
648 649
 out_free:
	kfree(dup_name);
650 651 652 653 654
 out_drop:
	d_drop(new_dentry);
 out_dput:
	dput(new_dentry);
 out_unlock:
655
	mutex_unlock(&new_parent->d_inode->i_mutex);
L
Linus Torvalds 已提交
656 657 658 659
	up_write(&sysfs_rename_sem);
	return error;
}

660 661 662 663 664 665 666 667
int sysfs_move_dir(struct kobject *kobj, struct kobject *new_parent)
{
	struct dentry *old_parent_dentry, *new_parent_dentry, *new_dentry;
	struct sysfs_dirent *new_parent_sd, *sd;
	int error;

	old_parent_dentry = kobj->parent ?
		kobj->parent->dentry : sysfs_mount->mnt_sb->s_root;
668 669
	new_parent_dentry = new_parent ?
		new_parent->dentry : sysfs_mount->mnt_sb->s_root;
670

M
Mark Lord 已提交
671 672
	if (old_parent_dentry->d_inode == new_parent_dentry->d_inode)
		return 0;	/* nothing to move */
673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694
again:
	mutex_lock(&old_parent_dentry->d_inode->i_mutex);
	if (!mutex_trylock(&new_parent_dentry->d_inode->i_mutex)) {
		mutex_unlock(&old_parent_dentry->d_inode->i_mutex);
		goto again;
	}

	new_parent_sd = new_parent_dentry->d_fsdata;
	sd = kobj->dentry->d_fsdata;

	new_dentry = lookup_one_len(kobj->name, new_parent_dentry,
				    strlen(kobj->name));
	if (IS_ERR(new_dentry)) {
		error = PTR_ERR(new_dentry);
		goto out;
	} else
		error = 0;
	d_add(new_dentry, NULL);
	d_move(kobj->dentry, new_dentry);
	dput(new_dentry);

	/* Remove from old parent's list and insert into new parent's list. */
695
	sysfs_unlink_sibling(sd);
696 697 698
	sysfs_get(new_parent_sd);
	sysfs_put(sd->s_parent);
	sd->s_parent = new_parent_sd;
699
	sysfs_link_sibling(sd);
700 701 702 703 704 705 706 707

out:
	mutex_unlock(&new_parent_dentry->d_inode->i_mutex);
	mutex_unlock(&old_parent_dentry->d_inode->i_mutex);

	return error;
}

L
Linus Torvalds 已提交
708 709
static int sysfs_dir_open(struct inode *inode, struct file *file)
{
710
	struct dentry * dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
711
	struct sysfs_dirent * parent_sd = dentry->d_fsdata;
712
	struct sysfs_dirent * sd;
L
Linus Torvalds 已提交
713

714
	mutex_lock(&dentry->d_inode->i_mutex);
715
	sd = sysfs_new_dirent("_DIR_", 0, 0);
716 717
	if (sd)
		sysfs_attach_dirent(sd, parent_sd, NULL);
718
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
719

720 721
	file->private_data = sd;
	return sd ? 0 : -ENOMEM;
L
Linus Torvalds 已提交
722 723 724 725
}

static int sysfs_dir_close(struct inode *inode, struct file *file)
{
726
	struct dentry * dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
727 728
	struct sysfs_dirent * cursor = file->private_data;

729
	mutex_lock(&dentry->d_inode->i_mutex);
730
	sysfs_unlink_sibling(cursor);
731
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
732 733 734 735 736 737 738 739 740 741 742 743 744 745

	release_sysfs_dirent(cursor);

	return 0;
}

/* Relationship between s_mode and the DT_xxx types */
static inline unsigned char dt_type(struct sysfs_dirent *sd)
{
	return (sd->s_mode >> 12) & 15;
}

static int sysfs_readdir(struct file * filp, void * dirent, filldir_t filldir)
{
746
	struct dentry *dentry = filp->f_path.dentry;
L
Linus Torvalds 已提交
747 748
	struct sysfs_dirent * parent_sd = dentry->d_fsdata;
	struct sysfs_dirent *cursor = filp->private_data;
749
	struct sysfs_dirent **pos;
L
Linus Torvalds 已提交
750 751 752 753 754
	ino_t ino;
	int i = filp->f_pos;

	switch (i) {
		case 0:
755
			ino = parent_sd->s_ino;
L
Linus Torvalds 已提交
756 757 758 759 760 761
			if (filldir(dirent, ".", 1, i, ino, DT_DIR) < 0)
				break;
			filp->f_pos++;
			i++;
			/* fallthrough */
		case 1:
T
Tejun Heo 已提交
762 763 764 765
			if (parent_sd->s_parent)
				ino = parent_sd->s_parent->s_ino;
			else
				ino = parent_sd->s_ino;
L
Linus Torvalds 已提交
766 767 768 769 770 771
			if (filldir(dirent, "..", 2, i, ino, DT_DIR) < 0)
				break;
			filp->f_pos++;
			i++;
			/* fallthrough */
		default:
772 773 774 775 776 777 778
			pos = &parent_sd->s_children;
			while (*pos != cursor)
				pos = &(*pos)->s_sibling;

			/* unlink cursor */
			*pos = cursor->s_sibling;

A
Akinobu Mita 已提交
779
			if (filp->f_pos == 2)
780
				pos = &parent_sd->s_children;
A
Akinobu Mita 已提交
781

782 783
			for ( ; *pos; pos = &(*pos)->s_sibling) {
				struct sysfs_dirent *next = *pos;
L
Linus Torvalds 已提交
784 785 786
				const char * name;
				int len;

787
				if (!sysfs_type(next))
L
Linus Torvalds 已提交
788 789
					continue;

T
Tejun Heo 已提交
790
				name = next->s_name;
L
Linus Torvalds 已提交
791
				len = strlen(name);
792
				ino = next->s_ino;
L
Linus Torvalds 已提交
793 794 795

				if (filldir(dirent, name, len, filp->f_pos, ino,
						 dt_type(next)) < 0)
796
					break;
L
Linus Torvalds 已提交
797 798 799

				filp->f_pos++;
			}
800 801 802 803

			/* put cursor back in */
			cursor->s_sibling = *pos;
			*pos = cursor;
L
Linus Torvalds 已提交
804 805 806 807 808 809
	}
	return 0;
}

static loff_t sysfs_dir_lseek(struct file * file, loff_t offset, int origin)
{
810
	struct dentry * dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
811

812
	mutex_lock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
813 814 815 816 817 818 819
	switch (origin) {
		case 1:
			offset += file->f_pos;
		case 0:
			if (offset >= 0)
				break;
		default:
820
			mutex_unlock(&file->f_path.dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
821 822 823 824 825 826 827
			return -EINVAL;
	}
	if (offset != file->f_pos) {
		file->f_pos = offset;
		if (file->f_pos >= 2) {
			struct sysfs_dirent *sd = dentry->d_fsdata;
			struct sysfs_dirent *cursor = file->private_data;
828
			struct sysfs_dirent **pos;
L
Linus Torvalds 已提交
829 830
			loff_t n = file->f_pos - 2;

831 832 833 834 835
			sysfs_unlink_sibling(cursor);

			pos = &sd->s_children;
			while (n && *pos) {
				struct sysfs_dirent *next = *pos;
836
				if (sysfs_type(next))
L
Linus Torvalds 已提交
837
					n--;
838
				pos = &(*pos)->s_sibling;
L
Linus Torvalds 已提交
839
			}
840 841 842

			cursor->s_sibling = *pos;
			*pos = cursor;
L
Linus Torvalds 已提交
843 844
		}
	}
845
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
846 847 848
	return offset;
}

849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889

/**
 *	sysfs_make_shadowed_dir - Setup so a directory can be shadowed
 *	@kobj:	object we're creating shadow of.
 */

int sysfs_make_shadowed_dir(struct kobject *kobj,
	void * (*follow_link)(struct dentry *, struct nameidata *))
{
	struct inode *inode;
	struct inode_operations *i_op;

	inode = kobj->dentry->d_inode;
	if (inode->i_op != &sysfs_dir_inode_operations)
		return -EINVAL;

	i_op = kmalloc(sizeof(*i_op), GFP_KERNEL);
	if (!i_op)
		return -ENOMEM;

	memcpy(i_op, &sysfs_dir_inode_operations, sizeof(*i_op));
	i_op->follow_link = follow_link;

	/* Locking of inode->i_op?
	 * Since setting i_op is a single word write and they
	 * are atomic we should be ok here.
	 */
	inode->i_op = i_op;
	return 0;
}

/**
 *	sysfs_create_shadow_dir - create a shadow directory for an object.
 *	@kobj:	object we're creating directory for.
 *
 *	sysfs_make_shadowed_dir must already have been called on this
 *	directory.
 */

struct dentry *sysfs_create_shadow_dir(struct kobject *kobj)
{
T
Tejun Heo 已提交
890 891 892 893 894
	struct dentry *dir = kobj->dentry;
	struct inode *inode = dir->d_inode;
	struct dentry *parent = dir->d_parent;
	struct sysfs_dirent *parent_sd = parent->d_fsdata;
	struct dentry *shadow;
895 896 897 898 899 900 901 902 903 904
	struct sysfs_dirent *sd;

	shadow = ERR_PTR(-EINVAL);
	if (!sysfs_is_shadowed_inode(inode))
		goto out;

	shadow = d_alloc(parent, &dir->d_name);
	if (!shadow)
		goto nomem;

905
	sd = sysfs_new_dirent("_SHADOW_", inode->i_mode, SYSFS_DIR);
906 907
	if (!sd)
		goto nomem;
908
	sd->s_elem.dir.kobj = kobj;
T
Tejun Heo 已提交
909 910
	/* point to parent_sd but don't attach to it */
	sd->s_parent = sysfs_get(parent_sd);
911
	sysfs_attach_dirent(sd, NULL, shadow);
912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940

	d_instantiate(shadow, igrab(inode));
	inc_nlink(inode);
	inc_nlink(parent->d_inode);

	dget(shadow);		/* Extra count - pin the dentry in core */

out:
	return shadow;
nomem:
	dput(shadow);
	shadow = ERR_PTR(-ENOMEM);
	goto out;
}

/**
 *	sysfs_remove_shadow_dir - remove an object's directory.
 *	@shadow: dentry of shadow directory
 *
 *	The only thing special about this is that we remove any files in
 *	the directory before we remove the directory, and we've inlined
 *	what used to be sysfs_rmdir() below, instead of calling separately.
 */

void sysfs_remove_shadow_dir(struct dentry *shadow)
{
	__sysfs_remove_dir(shadow);
}

941
const struct file_operations sysfs_dir_operations = {
L
Linus Torvalds 已提交
942 943 944 945 946 947
	.open		= sysfs_dir_open,
	.release	= sysfs_dir_close,
	.llseek		= sysfs_dir_lseek,
	.read		= generic_read_dir,
	.readdir	= sysfs_readdir,
};