dir.c 19.6 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
/**
 *	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)
{
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
	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();
53 54 55 56 57 58 59 60 61 62 63 64
	}
}

/**
 *	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)
{
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79
	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
	 * sd->s_sibling.next.
	 */
	cmpl = (void *)sd->s_sibling.next;
	complete(cmpl);
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
}

/**
 *	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
 *
125
 *	Deny new active references and drain existing ones.
126 127 128
 */
void sysfs_deactivate(struct sysfs_dirent *sd)
{
129 130
	DECLARE_COMPLETION_ONSTACK(wait);
	int v;
131

132 133 134 135 136
	BUG_ON(!list_empty(&sd->s_sibling));
	sd->s_sibling.next = (void *)&wait;

	/* atomic_add_return() is a mb(), put_active() will always see
	 * the updated sd->s_sibling.next.
137
	 */
138 139 140 141 142 143
	v = atomic_add_return(SD_DEACTIVATED_BIAS, &sd->s_active);

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

	INIT_LIST_HEAD(&sd->s_sibling);
144 145
}

T
Tejun Heo 已提交
146
static int sysfs_alloc_ino(ino_t *pino)
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
{
	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);
}

172 173
void release_sysfs_dirent(struct sysfs_dirent * sd)
{
T
Tejun Heo 已提交
174 175 176 177 178
	struct sysfs_dirent *parent_sd;

 repeat:
	parent_sd = sd->s_parent;

179
	if (sd->s_type & SYSFS_KOBJ_LINK)
180
		sysfs_put(sd->s_elem.symlink.target_sd);
T
Tejun Heo 已提交
181 182
	if (sd->s_type & SYSFS_COPY_NAME)
		kfree(sd->s_name);
183
	kfree(sd->s_iattr);
184
	sysfs_free_ino(sd->s_ino);
185
	kmem_cache_free(sysfs_dir_cachep, sd);
T
Tejun Heo 已提交
186 187 188 189

	sd = parent_sd;
	if (sd && atomic_dec_and_test(&sd->s_count))
		goto repeat;
190 191
}

L
Linus Torvalds 已提交
192 193 194 195 196
static void sysfs_d_iput(struct dentry * dentry, struct inode * inode)
{
	struct sysfs_dirent * sd = dentry->d_fsdata;

	if (sd) {
197 198 199 200 201 202 203 204 205 206 207 208 209
		/* 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
210 211 212 213 214 215 216 217 218
		sysfs_put(sd);
	}
	iput(inode);
}

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

219
struct sysfs_dirent *sysfs_new_dirent(const char *name, umode_t mode, int type)
L
Linus Torvalds 已提交
220
{
T
Tejun Heo 已提交
221 222 223 224 225 226 227 228
	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
229

230
	sd = kmem_cache_zalloc(sysfs_dir_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
231
	if (!sd)
T
Tejun Heo 已提交
232
		goto err_out;
L
Linus Torvalds 已提交
233

T
Tejun Heo 已提交
234 235
	if (sysfs_alloc_ino(&sd->s_ino))
		goto err_out;
236

L
Linus Torvalds 已提交
237
	atomic_set(&sd->s_count, 1);
238
	atomic_set(&sd->s_active, 0);
239
	atomic_set(&sd->s_event, 1);
L
Linus Torvalds 已提交
240
	INIT_LIST_HEAD(&sd->s_children);
241
	INIT_LIST_HEAD(&sd->s_sibling);
242

T
Tejun Heo 已提交
243
	sd->s_name = name;
244 245
	sd->s_mode = mode;
	sd->s_type = type;
L
Linus Torvalds 已提交
246 247

	return sd;
T
Tejun Heo 已提交
248 249 250 251 252

 err_out:
	kfree(dup_name);
	kmem_cache_free(sysfs_dir_cachep, sd);
	return NULL;
L
Linus Torvalds 已提交
253 254
}

255 256 257 258 259 260 261 262 263 264 265 266 267
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);
}

268 269
void sysfs_attach_dirent(struct sysfs_dirent *sd,
			 struct sysfs_dirent *parent_sd, struct dentry *dentry)
270
{
271 272
	if (dentry)
		sysfs_attach_dentry(sd, dentry);
273

T
Tejun Heo 已提交
274 275
	if (parent_sd) {
		sd->s_parent = sysfs_get(parent_sd);
276
		list_add(&sd->s_sibling, &parent_sd->s_children);
T
Tejun Heo 已提交
277
	}
278 279
}

280
/*
281 282 283 284 285 286 287 288 289 290 291 292
 *
 * 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;

	list_for_each_entry(sd, &parent_sd->s_children, s_sibling) {
293
		if (sd->s_type) {
T
Tejun Heo 已提交
294
			if (strcmp(sd->s_name, new))
295 296 297 298 299 300 301 302 303
				continue;
			else
				return -EEXIST;
		}
	}

	return 0;
}

304 305
static int create_dir(struct kobject *kobj, struct dentry *parent,
		      const char *name, struct dentry **p_dentry)
L
Linus Torvalds 已提交
306 307 308
{
	int error;
	umode_t mode = S_IFDIR| S_IRWXU | S_IRUGO | S_IXUGO;
309
	struct dentry *dentry;
310
	struct inode *inode;
311
	struct sysfs_dirent *sd;
L
Linus Torvalds 已提交
312

313 314
	mutex_lock(&parent->d_inode->i_mutex);

315
	/* allocate */
316 317 318 319 320 321 322
	dentry = lookup_one_len(name, parent, strlen(name));
	if (IS_ERR(dentry)) {
		error = PTR_ERR(dentry);
		goto out_unlock;
	}

	error = -EEXIST;
323
	if (dentry->d_inode)
324 325
		goto out_dput;

326
	error = -ENOMEM;
327
	sd = sysfs_new_dirent(name, mode, SYSFS_DIR);
328
	if (!sd)
329
		goto out_drop;
330
	sd->s_elem.dir.kobj = kobj;
331

332
	inode = sysfs_get_inode(sd);
333
	if (!inode)
334 335
		goto out_sput;

336 337 338 339 340 341
	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);
	}
342 343 344 345 346 347 348

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

	sysfs_instantiate(dentry, inode);
349
	inc_nlink(parent->d_inode);
350
	sysfs_attach_dirent(sd, parent->d_fsdata, dentry);
351 352 353

	*p_dentry = dentry;
	error = 0;
354
	goto out_unlock;	/* pin directory dentry in core */
355

356 357
 out_iput:
	iput(inode);
358 359 360 361 362 363 364 365
 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
366 367 368 369 370 371 372 373 374 375 376 377
	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. 
378
 *	@shadow_parent:	parent parent object.
L
Linus Torvalds 已提交
379 380
 */

381
int sysfs_create_dir(struct kobject * kobj, struct dentry *shadow_parent)
L
Linus Torvalds 已提交
382 383 384 385 386 387 388
{
	struct dentry * dentry = NULL;
	struct dentry * parent;
	int error = 0;

	BUG_ON(!kobj);

389 390 391
	if (shadow_parent)
		parent = shadow_parent;
	else if (kobj->parent)
L
Linus Torvalds 已提交
392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408
		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;
409 410
	struct inode *inode;
	int found = 0;
L
Linus Torvalds 已提交
411 412

	list_for_each_entry(sd, &parent_sd->s_children, s_sibling) {
413 414 415
		if ((sd->s_type & SYSFS_NOT_PINNED) &&
		    !strcmp(sd->s_name, dentry->d_name.name)) {
			found = 1;
L
Linus Torvalds 已提交
416 417 418 419
			break;
		}
	}

420 421 422 423 424
	/* no such entry */
	if (!found)
		return NULL;

	/* attach dentry and inode */
425
	inode = sysfs_get_inode(sd);
426 427 428
	if (!inode)
		return ERR_PTR(-ENOMEM);

429 430 431 432 433 434 435 436 437 438 439 440 441
	if (inode->i_state & I_NEW) {
		/* initialize inode according to type */
		if (sd->s_type & SYSFS_KOBJ_ATTR) {
			inode->i_size = PAGE_SIZE;
			inode->i_fop = &sysfs_file_operations;
		} else if (sd->s_type & SYSFS_KOBJ_BIN_ATTR) {
			struct bin_attribute *bin_attr =
				sd->s_elem.bin_attr.bin_attr;
			inode->i_size = bin_attr->size;
			inode->i_fop = &bin_fops;
		} else if (sd->s_type & SYSFS_KOBJ_LINK)
			inode->i_op = &sysfs_symlink_inode_operations;
	}
442 443 444 445 446

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

	return NULL;
L
Linus Torvalds 已提交
447 448
}

449
const struct inode_operations sysfs_dir_inode_operations = {
L
Linus Torvalds 已提交
450
	.lookup		= sysfs_lookup,
451
	.setattr	= sysfs_setattr,
L
Linus Torvalds 已提交
452 453 454 455
};

static void remove_dir(struct dentry * d)
{
456 457
	struct dentry *parent = d->d_parent;
	struct sysfs_dirent *sd = d->d_fsdata;
L
Linus Torvalds 已提交
458

459
	mutex_lock(&parent->d_inode->i_mutex);
460

L
Linus Torvalds 已提交
461 462 463 464 465
 	list_del_init(&sd->s_sibling);

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

466
	mutex_unlock(&parent->d_inode->i_mutex);
467

468
	sysfs_drop_dentry(sd);
469 470
	sysfs_deactivate(sd);
	sysfs_put(sd);
L
Linus Torvalds 已提交
471 472 473 474 475 476 477 478
}

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


479
static void __sysfs_remove_dir(struct dentry *dentry)
L
Linus Torvalds 已提交
480
{
481
	LIST_HEAD(removed);
L
Linus Torvalds 已提交
482 483 484 485 486 487 488
	struct sysfs_dirent * parent_sd;
	struct sysfs_dirent * sd, * tmp;

	if (!dentry)
		return;

	pr_debug("sysfs %s: removing dir\n",dentry->d_name.name);
489
	mutex_lock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
490 491
	parent_sd = dentry->d_fsdata;
	list_for_each_entry_safe(sd, tmp, &parent_sd->s_children, s_sibling) {
492
		if (!sd->s_type || !(sd->s_type & SYSFS_NOT_PINNED))
L
Linus Torvalds 已提交
493
			continue;
494
		list_move(&sd->s_sibling, &removed);
L
Linus Torvalds 已提交
495
	}
496
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
497

498 499
	list_for_each_entry_safe(sd, tmp, &removed, s_sibling) {
		list_del_init(&sd->s_sibling);
500
		sysfs_drop_dentry(sd);
501 502 503 504
		sysfs_deactivate(sd);
		sysfs_put(sd);
	}

L
Linus Torvalds 已提交
505
	remove_dir(dentry);
506 507 508 509 510 511 512 513 514 515 516 517 518
}

/**
 *	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)
{
519 520 521
	struct dentry *d = kobj->dentry;

	spin_lock(&kobj_sysfs_assoc_lock);
522
	kobj->dentry = NULL;
523 524 525
	spin_unlock(&kobj_sysfs_assoc_lock);

	__sysfs_remove_dir(d);
L
Linus Torvalds 已提交
526 527
}

528 529
int sysfs_rename_dir(struct kobject * kobj, struct dentry *new_parent,
		     const char *new_name)
L
Linus Torvalds 已提交
530
{
T
Tejun Heo 已提交
531 532 533 534
	struct sysfs_dirent *sd = kobj->dentry->d_fsdata;
	struct sysfs_dirent *parent_sd = new_parent->d_fsdata;
	struct dentry *new_dentry;
	char *dup_name;
535
	int error;
L
Linus Torvalds 已提交
536

537 538
	if (!new_parent)
		return -EFAULT;
L
Linus Torvalds 已提交
539 540

	down_write(&sysfs_rename_sem);
541
	mutex_lock(&new_parent->d_inode->i_mutex);
L
Linus Torvalds 已提交
542

543
	new_dentry = lookup_one_len(new_name, new_parent, strlen(new_name));
544 545 546
	if (IS_ERR(new_dentry)) {
		error = PTR_ERR(new_dentry);
		goto out_unlock;
L
Linus Torvalds 已提交
547
	}
548 549 550 551 552 553 554 555 556 557 558 559 560 561 562

	/* 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
563 564 565 566 567 568
	/* rename kobject and sysfs_dirent */
	error = -ENOMEM;
	new_name = dup_name = kstrdup(new_name, GFP_KERNEL);
	if (!new_name)
		goto out_drop;

569 570
	error = kobject_set_name(kobj, "%s", new_name);
	if (error)
T
Tejun Heo 已提交
571
		goto out_free;
572

T
Tejun Heo 已提交
573 574 575 576
	kfree(sd->s_name);
	sd->s_name = new_name;

	/* move under the new parent */
577 578 579 580
	d_add(new_dentry, NULL);
	d_move(kobj->dentry, new_dentry);

	list_del_init(&sd->s_sibling);
581 582 583
	sysfs_get(parent_sd);
	sysfs_put(sd->s_parent);
	sd->s_parent = parent_sd;
584 585 586 587 588
	list_add(&sd->s_sibling, &parent_sd->s_children);

	error = 0;
	goto out_unlock;

T
Tejun Heo 已提交
589 590
 out_free:
	kfree(dup_name);
591 592 593 594 595
 out_drop:
	d_drop(new_dentry);
 out_dput:
	dput(new_dentry);
 out_unlock:
596
	mutex_unlock(&new_parent->d_inode->i_mutex);
L
Linus Torvalds 已提交
597 598 599 600
	up_write(&sysfs_rename_sem);
	return error;
}

601 602 603 604 605 606 607 608
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;
609 610
	new_parent_dentry = new_parent ?
		new_parent->dentry : sysfs_mount->mnt_sb->s_root;
611

M
Mark Lord 已提交
612 613
	if (old_parent_dentry->d_inode == new_parent_dentry->d_inode)
		return 0;	/* nothing to move */
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
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. */
	list_del_init(&sd->s_sibling);
637 638 639
	sysfs_get(new_parent_sd);
	sysfs_put(sd->s_parent);
	sd->s_parent = new_parent_sd;
640 641 642 643 644 645 646 647 648
	list_add(&sd->s_sibling, &new_parent_sd->s_children);

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
649 650
static int sysfs_dir_open(struct inode *inode, struct file *file)
{
651
	struct dentry * dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
652
	struct sysfs_dirent * parent_sd = dentry->d_fsdata;
653
	struct sysfs_dirent * sd;
L
Linus Torvalds 已提交
654

655
	mutex_lock(&dentry->d_inode->i_mutex);
656
	sd = sysfs_new_dirent("_DIR_", 0, 0);
657 658
	if (sd)
		sysfs_attach_dirent(sd, parent_sd, NULL);
659
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
660

661 662
	file->private_data = sd;
	return sd ? 0 : -ENOMEM;
L
Linus Torvalds 已提交
663 664 665 666
}

static int sysfs_dir_close(struct inode *inode, struct file *file)
{
667
	struct dentry * dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
668 669
	struct sysfs_dirent * cursor = file->private_data;

670
	mutex_lock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
671
	list_del_init(&cursor->s_sibling);
672
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
673 674 675 676 677 678 679 680 681 682 683 684 685 686

	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)
{
687
	struct dentry *dentry = filp->f_path.dentry;
L
Linus Torvalds 已提交
688 689 690 691 692 693 694 695
	struct sysfs_dirent * parent_sd = dentry->d_fsdata;
	struct sysfs_dirent *cursor = filp->private_data;
	struct list_head *p, *q = &cursor->s_sibling;
	ino_t ino;
	int i = filp->f_pos;

	switch (i) {
		case 0:
696
			ino = parent_sd->s_ino;
L
Linus Torvalds 已提交
697 698 699 700 701 702
			if (filldir(dirent, ".", 1, i, ino, DT_DIR) < 0)
				break;
			filp->f_pos++;
			i++;
			/* fallthrough */
		case 1:
T
Tejun Heo 已提交
703 704 705 706
			if (parent_sd->s_parent)
				ino = parent_sd->s_parent->s_ino;
			else
				ino = parent_sd->s_ino;
L
Linus Torvalds 已提交
707 708 709 710 711 712
			if (filldir(dirent, "..", 2, i, ino, DT_DIR) < 0)
				break;
			filp->f_pos++;
			i++;
			/* fallthrough */
		default:
A
Akinobu Mita 已提交
713 714 715
			if (filp->f_pos == 2)
				list_move(q, &parent_sd->s_children);

L
Linus Torvalds 已提交
716 717 718 719 720 721 722
			for (p=q->next; p!= &parent_sd->s_children; p=p->next) {
				struct sysfs_dirent *next;
				const char * name;
				int len;

				next = list_entry(p, struct sysfs_dirent,
						   s_sibling);
723
				if (!next->s_type)
L
Linus Torvalds 已提交
724 725
					continue;

T
Tejun Heo 已提交
726
				name = next->s_name;
L
Linus Torvalds 已提交
727
				len = strlen(name);
728
				ino = next->s_ino;
L
Linus Torvalds 已提交
729 730 731 732 733

				if (filldir(dirent, name, len, filp->f_pos, ino,
						 dt_type(next)) < 0)
					return 0;

A
Akinobu Mita 已提交
734
				list_move(q, p);
L
Linus Torvalds 已提交
735 736 737 738 739 740 741 742 743
				p = q;
				filp->f_pos++;
			}
	}
	return 0;
}

static loff_t sysfs_dir_lseek(struct file * file, loff_t offset, int origin)
{
744
	struct dentry * dentry = file->f_path.dentry;
L
Linus Torvalds 已提交
745

746
	mutex_lock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
747 748 749 750 751 752 753
	switch (origin) {
		case 1:
			offset += file->f_pos;
		case 0:
			if (offset >= 0)
				break;
		default:
754
			mutex_unlock(&file->f_path.dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770
			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;
			struct list_head *p;
			loff_t n = file->f_pos - 2;

			list_del(&cursor->s_sibling);
			p = sd->s_children.next;
			while (n && p != &sd->s_children) {
				struct sysfs_dirent *next;
				next = list_entry(p, struct sysfs_dirent,
						   s_sibling);
771
				if (next->s_type)
L
Linus Torvalds 已提交
772 773 774 775 776 777
					n--;
				p = p->next;
			}
			list_add_tail(&cursor->s_sibling, p);
		}
	}
778
	mutex_unlock(&dentry->d_inode->i_mutex);
L
Linus Torvalds 已提交
779 780 781
	return offset;
}

782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822

/**
 *	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
823 824 825 826 827
	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;
828 829 830 831 832 833 834 835 836 837
	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;

838
	sd = sysfs_new_dirent("_SHADOW_", inode->i_mode, SYSFS_DIR);
839 840
	if (!sd)
		goto nomem;
841
	sd->s_elem.dir.kobj = kobj;
T
Tejun Heo 已提交
842 843
	/* point to parent_sd but don't attach to it */
	sd->s_parent = sysfs_get(parent_sd);
844
	sysfs_attach_dirent(sd, NULL, shadow);
845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873

	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);
}

874
const struct file_operations sysfs_dir_operations = {
L
Linus Torvalds 已提交
875 876 877 878 879 880
	.open		= sysfs_dir_open,
	.release	= sysfs_dir_close,
	.llseek		= sysfs_dir_lseek,
	.read		= generic_read_dir,
	.readdir	= sysfs_readdir,
};