super.c 62.2 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-only
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
/*
 * This file is part of UBIFS.
 *
 * Copyright (C) 2006-2008 Nokia Corporation.
 *
 * Authors: Artem Bityutskiy (Битюцкий Артём)
 *          Adrian Hunter
 */

/*
 * This file implements UBIFS initialization and VFS superblock operations. Some
 * initialization stuff which is rather large and complex is placed at
 * corresponding subsystems, but most of it is here.
 */

#include <linux/init.h>
#include <linux/slab.h>
#include <linux/module.h>
#include <linux/ctype.h>
#include <linux/kthread.h>
#include <linux/parser.h>
#include <linux/seq_file.h>
#include <linux/mount.h>
A
Artem Bityutskiy 已提交
25
#include <linux/math64.h>
26
#include <linux/writeback.h>
27 28
#include "ubifs.h"

29 30 31 32 33 34
/*
 * Maximum amount of memory we may 'kmalloc()' without worrying that we are
 * allocating too much.
 */
#define UBIFS_KMALLOC_OK (128*1024)

35
/* Slab cache for UBIFS inodes */
36
static struct kmem_cache *ubifs_inode_slab;
37 38 39

/* UBIFS TNC shrinker description */
static struct shrinker ubifs_shrinker_info = {
40 41
	.scan_objects = ubifs_shrink_scan,
	.count_objects = ubifs_shrink_count,
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
	.seeks = DEFAULT_SEEKS,
};

/**
 * validate_inode - validate inode.
 * @c: UBIFS file-system description object
 * @inode: the inode to validate
 *
 * This is a helper function for 'ubifs_iget()' which validates various fields
 * of a newly built inode to make sure they contain sane values and prevent
 * possible vulnerabilities. Returns zero if the inode is all right and
 * a non-zero error code if not.
 */
static int validate_inode(struct ubifs_info *c, const struct inode *inode)
{
	int err;
	const struct ubifs_inode *ui = ubifs_inode(inode);

	if (inode->i_size > c->max_inode_sz) {
61
		ubifs_err(c, "inode is too large (%lld)",
62 63 64 65
			  (long long)inode->i_size);
		return 1;
	}

H
hujianyang 已提交
66
	if (ui->compr_type >= UBIFS_COMPR_TYPES_CNT) {
67
		ubifs_err(c, "unknown compression type %d", ui->compr_type);
68 69 70 71 72 73 74 75 76
		return 2;
	}

	if (ui->xattr_names + ui->xattr_cnt > XATTR_LIST_MAX)
		return 3;

	if (ui->data_len < 0 || ui->data_len > UBIFS_MAX_INO_DATA)
		return 4;

77
	if (ui->xattr && !S_ISREG(inode->i_mode))
78 79
		return 5;

80
	if (!ubifs_compr_present(c, ui->compr_type)) {
81
		ubifs_warn(c, "inode %lu uses '%s' compression, but it was not compiled in",
82
			   inode->i_ino, ubifs_compr_name(c, ui->compr_type));
83 84
	}

85
	err = dbg_check_dir(c, inode);
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
	return err;
}

struct inode *ubifs_iget(struct super_block *sb, unsigned long inum)
{
	int err;
	union ubifs_key key;
	struct ubifs_ino_node *ino;
	struct ubifs_info *c = sb->s_fs_info;
	struct inode *inode;
	struct ubifs_inode *ui;

	dbg_gen("inode %lu", inum);

	inode = iget_locked(sb, inum);
	if (!inode)
		return ERR_PTR(-ENOMEM);
	if (!(inode->i_state & I_NEW))
		return inode;
	ui = ubifs_inode(inode);

	ino = kmalloc(UBIFS_MAX_INO_NODE_SZ, GFP_NOFS);
	if (!ino) {
		err = -ENOMEM;
		goto out;
	}

	ino_key_init(c, &key, inode->i_ino);

	err = ubifs_tnc_lookup(c, &key, ino);
	if (err)
		goto out_ino;

119
	inode->i_flags |= S_NOCMTIME;
120 121 122 123

	if (!IS_ENABLED(CONFIG_UBIFS_ATIME_SUPPORT))
		inode->i_flags |= S_NOATIME;

M
Miklos Szeredi 已提交
124
	set_nlink(inode, le32_to_cpu(ino->nlink));
125 126
	i_uid_write(inode, le32_to_cpu(ino->uid));
	i_gid_write(inode, le32_to_cpu(ino->gid));
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
	inode->i_atime.tv_sec  = (int64_t)le64_to_cpu(ino->atime_sec);
	inode->i_atime.tv_nsec = le32_to_cpu(ino->atime_nsec);
	inode->i_mtime.tv_sec  = (int64_t)le64_to_cpu(ino->mtime_sec);
	inode->i_mtime.tv_nsec = le32_to_cpu(ino->mtime_nsec);
	inode->i_ctime.tv_sec  = (int64_t)le64_to_cpu(ino->ctime_sec);
	inode->i_ctime.tv_nsec = le32_to_cpu(ino->ctime_nsec);
	inode->i_mode = le32_to_cpu(ino->mode);
	inode->i_size = le64_to_cpu(ino->size);

	ui->data_len    = le32_to_cpu(ino->data_len);
	ui->flags       = le32_to_cpu(ino->flags);
	ui->compr_type  = le16_to_cpu(ino->compr_type);
	ui->creat_sqnum = le64_to_cpu(ino->creat_sqnum);
	ui->xattr_cnt   = le32_to_cpu(ino->xattr_cnt);
	ui->xattr_size  = le32_to_cpu(ino->xattr_size);
	ui->xattr_names = le32_to_cpu(ino->xattr_names);
	ui->synced_i_size = ui->ui_size = inode->i_size;

	ui->xattr = (ui->flags & UBIFS_XATTR_FL) ? 1 : 0;

	err = validate_inode(c, inode);
	if (err)
		goto out_invalid;

	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
		inode->i_mapping->a_ops = &ubifs_file_address_operations;
		inode->i_op = &ubifs_file_inode_operations;
		inode->i_fop = &ubifs_file_operations;
		if (ui->xattr) {
			ui->data = kmalloc(ui->data_len + 1, GFP_NOFS);
			if (!ui->data) {
				err = -ENOMEM;
				goto out_ino;
			}
			memcpy(ui->data, ino->data, ui->data_len);
			((char *)ui->data)[ui->data_len] = '\0';
		} else if (ui->data_len != 0) {
			err = 10;
			goto out_invalid;
		}
		break;
	case S_IFDIR:
		inode->i_op  = &ubifs_dir_inode_operations;
		inode->i_fop = &ubifs_dir_operations;
		if (ui->data_len != 0) {
			err = 11;
			goto out_invalid;
		}
		break;
	case S_IFLNK:
		inode->i_op = &ubifs_symlink_inode_operations;
		if (ui->data_len <= 0 || ui->data_len > UBIFS_MAX_INO_DATA) {
			err = 12;
			goto out_invalid;
		}
		ui->data = kmalloc(ui->data_len + 1, GFP_NOFS);
		if (!ui->data) {
			err = -ENOMEM;
			goto out_ino;
		}
		memcpy(ui->data, ino->data, ui->data_len);
		((char *)ui->data)[ui->data_len] = '\0';
		break;
	case S_IFBLK:
	case S_IFCHR:
	{
		dev_t rdev;
		union ubifs_dev_desc *dev;

		ui->data = kmalloc(sizeof(union ubifs_dev_desc), GFP_NOFS);
		if (!ui->data) {
			err = -ENOMEM;
			goto out_ino;
		}

		dev = (union ubifs_dev_desc *)ino->data;
		if (ui->data_len == sizeof(dev->new))
			rdev = new_decode_dev(le32_to_cpu(dev->new));
		else if (ui->data_len == sizeof(dev->huge))
			rdev = huge_decode_dev(le64_to_cpu(dev->huge));
		else {
			err = 13;
			goto out_invalid;
		}
		memcpy(ui->data, ino->data, ui->data_len);
		inode->i_op = &ubifs_file_inode_operations;
		init_special_inode(inode, inode->i_mode, rdev);
		break;
	}
	case S_IFSOCK:
	case S_IFIFO:
		inode->i_op = &ubifs_file_inode_operations;
		init_special_inode(inode, inode->i_mode, 0);
		if (ui->data_len != 0) {
			err = 14;
			goto out_invalid;
		}
		break;
	default:
		err = 15;
		goto out_invalid;
	}

	kfree(ino);
	ubifs_set_inode_flags(inode);
	unlock_new_inode(inode);
	return inode;

out_invalid:
237
	ubifs_err(c, "inode %lu validation failed, error %d", inode->i_ino, err);
238 239
	ubifs_dump_node(c, ino);
	ubifs_dump_inode(c, inode);
240 241 242 243
	err = -EINVAL;
out_ino:
	kfree(ino);
out:
244
	ubifs_err(c, "failed to read inode %lu, error %d", inode->i_ino, err);
245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
	iget_failed(inode);
	return ERR_PTR(err);
}

static struct inode *ubifs_alloc_inode(struct super_block *sb)
{
	struct ubifs_inode *ui;

	ui = kmem_cache_alloc(ubifs_inode_slab, GFP_NOFS);
	if (!ui)
		return NULL;

	memset((void *)ui + sizeof(struct inode), 0,
	       sizeof(struct ubifs_inode) - sizeof(struct inode));
	mutex_init(&ui->ui_mutex);
	spin_lock_init(&ui->ui_lock);
	return &ui->vfs_inode;
};

A
Al Viro 已提交
264
static void ubifs_free_inode(struct inode *inode)
N
Nick Piggin 已提交
265 266
{
	struct ubifs_inode *ui = ubifs_inode(inode);
267

268
	kfree(ui->data);
269 270
	fscrypt_free_inode(inode);

N
Nick Piggin 已提交
271 272 273
	kmem_cache_free(ubifs_inode_slab, ui);
}

274 275 276
/*
 * Note, Linux write-back code calls this without 'i_mutex'.
 */
277
static int ubifs_write_inode(struct inode *inode, struct writeback_control *wbc)
278
{
279
	int err = 0;
280 281 282
	struct ubifs_info *c = inode->i_sb->s_fs_info;
	struct ubifs_inode *ui = ubifs_inode(inode);

283
	ubifs_assert(c, !ui->xattr);
284 285 286 287 288 289
	if (is_bad_inode(inode))
		return 0;

	mutex_lock(&ui->ui_mutex);
	/*
	 * Due to races between write-back forced by budgeting
290
	 * (see 'sync_some_inodes()') and background write-back, the inode may
291 292 293 294 295 296 297 298 299
	 * have already been synchronized, do not do this again. This might
	 * also happen if it was synchronized in an VFS operation, e.g.
	 * 'ubifs_link()'.
	 */
	if (!ui->dirty) {
		mutex_unlock(&ui->ui_mutex);
		return 0;
	}

300 301 302 303 304 305 306
	/*
	 * As an optimization, do not write orphan inodes to the media just
	 * because this is not needed.
	 */
	dbg_gen("inode %lu, mode %#x, nlink %u",
		inode->i_ino, (int)inode->i_mode, inode->i_nlink);
	if (inode->i_nlink) {
307
		err = ubifs_jnl_write_inode(c, inode);
308
		if (err)
309
			ubifs_err(c, "can't write inode %lu, error %d",
310
				  inode->i_ino, err);
311 312
		else
			err = dbg_check_inode_size(c, inode, ui->ui_size);
313
	}
314 315 316 317 318 319 320

	ui->dirty = 0;
	mutex_unlock(&ui->ui_mutex);
	ubifs_release_dirty_inode_budget(c, ui);
	return err;
}

A
Al Viro 已提交
321
static void ubifs_evict_inode(struct inode *inode)
322 323 324
{
	int err;
	struct ubifs_info *c = inode->i_sb->s_fs_info;
325
	struct ubifs_inode *ui = ubifs_inode(inode);
326

327
	if (ui->xattr)
328 329 330 331 332 333 334
		/*
		 * Extended attribute inode deletions are fully handled in
		 * 'ubifs_removexattr()'. These inodes are special and have
		 * limited usage, so there is nothing to do here.
		 */
		goto out;

A
Artem Bityutskiy 已提交
335
	dbg_gen("inode %lu, mode %#x", inode->i_ino, (int)inode->i_mode);
336
	ubifs_assert(c, !atomic_read(&inode->i_count));
337

338
	truncate_inode_pages_final(&inode->i_data);
A
Al Viro 已提交
339 340 341 342

	if (inode->i_nlink)
		goto done;

343 344 345
	if (is_bad_inode(inode))
		goto out;

346
	ui->ui_size = inode->i_size = 0;
A
Artem Bityutskiy 已提交
347
	err = ubifs_jnl_delete_inode(c, inode);
348 349 350
	if (err)
		/*
		 * Worst case we have a lost orphan inode wasting space, so a
A
Artem Bityutskiy 已提交
351
		 * simple error message is OK here.
352
		 */
353
		ubifs_err(c, "can't delete inode %lu, error %d",
A
Artem Bityutskiy 已提交
354 355
			  inode->i_ino, err);

356
out:
357 358
	if (ui->dirty)
		ubifs_release_dirty_inode_budget(c, ui);
359 360
	else {
		/* We've deleted something - clean the "no space" flags */
361
		c->bi.nospace = c->bi.nospace_rp = 0;
362 363
		smp_wmb();
	}
A
Al Viro 已提交
364
done:
365
	clear_inode(inode);
366
	fscrypt_put_encryption_info(inode);
367 368
}

369
static void ubifs_dirty_inode(struct inode *inode, int flags)
370
{
371
	struct ubifs_info *c = inode->i_sb->s_fs_info;
372 373
	struct ubifs_inode *ui = ubifs_inode(inode);

374
	ubifs_assert(c, mutex_is_locked(&ui->ui_mutex));
375 376 377 378 379 380 381 382 383 384
	if (!ui->dirty) {
		ui->dirty = 1;
		dbg_gen("inode %lu",  inode->i_ino);
	}
}

static int ubifs_statfs(struct dentry *dentry, struct kstatfs *buf)
{
	struct ubifs_info *c = dentry->d_sb->s_fs_info;
	unsigned long long free;
A
Artem Bityutskiy 已提交
385
	__le32 *uuid = (__le32 *)c->uuid;
386

387
	free = ubifs_get_free_space(c);
388 389 390 391 392 393 394 395 396 397 398 399 400 401
	dbg_gen("free space %lld bytes (%lld blocks)",
		free, free >> UBIFS_BLOCK_SHIFT);

	buf->f_type = UBIFS_SUPER_MAGIC;
	buf->f_bsize = UBIFS_BLOCK_SIZE;
	buf->f_blocks = c->block_cnt;
	buf->f_bfree = free >> UBIFS_BLOCK_SHIFT;
	if (free > c->report_rp_size)
		buf->f_bavail = (free - c->report_rp_size) >> UBIFS_BLOCK_SHIFT;
	else
		buf->f_bavail = 0;
	buf->f_files = 0;
	buf->f_ffree = 0;
	buf->f_namelen = UBIFS_MAX_NLEN;
A
Artem Bityutskiy 已提交
402 403
	buf->f_fsid.val[0] = le32_to_cpu(uuid[0]) ^ le32_to_cpu(uuid[2]);
	buf->f_fsid.val[1] = le32_to_cpu(uuid[1]) ^ le32_to_cpu(uuid[3]);
404
	ubifs_assert(c, buf->f_bfree <= c->block_cnt);
405 406 407
	return 0;
}

408
static int ubifs_show_options(struct seq_file *s, struct dentry *root)
409
{
410
	struct ubifs_info *c = root->d_sb->s_fs_info;
411 412

	if (c->mount_opts.unmount_mode == 2)
413
		seq_puts(s, ",fast_unmount");
414
	else if (c->mount_opts.unmount_mode == 1)
415
		seq_puts(s, ",norm_unmount");
416

A
Adrian Hunter 已提交
417
	if (c->mount_opts.bulk_read == 2)
418
		seq_puts(s, ",bulk_read");
A
Adrian Hunter 已提交
419
	else if (c->mount_opts.bulk_read == 1)
420
		seq_puts(s, ",no_bulk_read");
A
Adrian Hunter 已提交
421

422
	if (c->mount_opts.chk_data_crc == 2)
423
		seq_puts(s, ",chk_data_crc");
424
	else if (c->mount_opts.chk_data_crc == 1)
425
		seq_puts(s, ",no_chk_data_crc");
426

427
	if (c->mount_opts.override_compr) {
H
Hunter Adrian 已提交
428
		seq_printf(s, ",compr=%s",
429
			   ubifs_compr_name(c, c->mount_opts.compr_type));
430 431
	}

432
	seq_printf(s, ",assert=%s", ubifs_assert_action_name(c));
433 434
	seq_printf(s, ",ubi=%d,vol=%d", c->vi.ubi_num, c->vi.vol_id);

435 436 437 438 439
	return 0;
}

static int ubifs_sync_fs(struct super_block *sb, int wait)
{
A
Artem Bityutskiy 已提交
440
	int i, err;
441
	struct ubifs_info *c = sb->s_fs_info;
442

443
	/*
A
Artem Bityutskiy 已提交
444 445
	 * Zero @wait is just an advisory thing to help the file system shove
	 * lots of data into the queues, and there will be the second
446 447
	 * '->sync_fs()' call, with non-zero @wait.
	 */
A
Artem Bityutskiy 已提交
448 449
	if (!wait)
		return 0;
450

451 452 453 454 455 456 457 458 459 460
	/*
	 * Synchronize write buffers, because 'ubifs_run_commit()' does not
	 * do this if it waits for an already running commit.
	 */
	for (i = 0; i < c->jhead_cnt; i++) {
		err = ubifs_wbuf_sync(&c->jheads[i].wbuf);
		if (err)
			return err;
	}

461 462 463 464 465 466 467
	/*
	 * Strictly speaking, it is not necessary to commit the journal here,
	 * synchronizing write-buffers would be enough. But committing makes
	 * UBIFS free space predictions much more accurate, so we want to let
	 * the user be able to get more accurate results of 'statfs()' after
	 * they synchronize the file system.
	 */
A
Artem Bityutskiy 已提交
468 469 470
	err = ubifs_run_commit(c);
	if (err)
		return err;
A
Artem Bityutskiy 已提交
471

A
Artem Bityutskiy 已提交
472
	return ubi_sync(c->vi.ubi_num);
473 474 475 476 477 478 479 480 481 482 483 484 485 486
}

/**
 * init_constants_early - initialize UBIFS constants.
 * @c: UBIFS file-system description object
 *
 * This function initialize UBIFS constants which do not need the superblock to
 * be read. It also checks that the UBI volume satisfies basic UBIFS
 * requirements. Returns zero in case of success and a negative error code in
 * case of failure.
 */
static int init_constants_early(struct ubifs_info *c)
{
	if (c->vi.corrupted) {
487
		ubifs_warn(c, "UBI volume is corrupted - read-only mode");
488 489 490 491
		c->ro_media = 1;
	}

	if (c->di.ro_mode) {
492
		ubifs_msg(c, "read-only UBI device");
493 494 495 496
		c->ro_media = 1;
	}

	if (c->vi.vol_type == UBI_STATIC_VOLUME) {
497
		ubifs_msg(c, "static UBI volume - read-only mode");
498 499 500 501 502
		c->ro_media = 1;
	}

	c->leb_cnt = c->vi.size;
	c->leb_size = c->vi.usable_leb_size;
503
	c->leb_start = c->di.leb_start;
504 505 506
	c->half_leb_size = c->leb_size / 2;
	c->min_io_size = c->di.min_io_size;
	c->min_io_shift = fls(c->min_io_size) - 1;
507 508
	c->max_write_size = c->di.max_write_size;
	c->max_write_shift = fls(c->max_write_size) - 1;
509 510

	if (c->leb_size < UBIFS_MIN_LEB_SZ) {
511 512
		ubifs_errc(c, "too small LEBs (%d bytes), min. is %d bytes",
			   c->leb_size, UBIFS_MIN_LEB_SZ);
513 514 515 516
		return -EINVAL;
	}

	if (c->leb_cnt < UBIFS_MIN_LEB_CNT) {
517 518
		ubifs_errc(c, "too few LEBs (%d), min. is %d",
			   c->leb_cnt, UBIFS_MIN_LEB_CNT);
519 520 521 522
		return -EINVAL;
	}

	if (!is_power_of_2(c->min_io_size)) {
523
		ubifs_errc(c, "bad min. I/O size %d", c->min_io_size);
524 525 526
		return -EINVAL;
	}

527 528 529 530 531 532 533
	/*
	 * Maximum write size has to be greater or equivalent to min. I/O
	 * size, and be multiple of min. I/O size.
	 */
	if (c->max_write_size < c->min_io_size ||
	    c->max_write_size % c->min_io_size ||
	    !is_power_of_2(c->max_write_size)) {
534 535
		ubifs_errc(c, "bad write buffer size %d for %d min. I/O unit",
			   c->max_write_size, c->min_io_size);
536 537 538
		return -EINVAL;
	}

539 540 541 542 543 544 545 546
	/*
	 * UBIFS aligns all node to 8-byte boundary, so to make function in
	 * io.c simpler, assume minimum I/O unit size to be 8 bytes if it is
	 * less than 8.
	 */
	if (c->min_io_size < 8) {
		c->min_io_size = 8;
		c->min_io_shift = 3;
547 548 549 550
		if (c->max_write_size < c->min_io_size) {
			c->max_write_size = c->min_io_size;
			c->max_write_shift = c->min_io_shift;
		}
551 552 553 554 555 556 557 558 559 560 561 562 563 564 565
	}

	c->ref_node_alsz = ALIGN(UBIFS_REF_NODE_SZ, c->min_io_size);
	c->mst_node_alsz = ALIGN(UBIFS_MST_NODE_SZ, c->min_io_size);

	/*
	 * Initialize node length ranges which are mostly needed for node
	 * length validation.
	 */
	c->ranges[UBIFS_PAD_NODE].len  = UBIFS_PAD_NODE_SZ;
	c->ranges[UBIFS_SB_NODE].len   = UBIFS_SB_NODE_SZ;
	c->ranges[UBIFS_MST_NODE].len  = UBIFS_MST_NODE_SZ;
	c->ranges[UBIFS_REF_NODE].len  = UBIFS_REF_NODE_SZ;
	c->ranges[UBIFS_TRUN_NODE].len = UBIFS_TRUN_NODE_SZ;
	c->ranges[UBIFS_CS_NODE].len   = UBIFS_CS_NODE_SZ;
566 567 568
	c->ranges[UBIFS_AUTH_NODE].min_len = UBIFS_AUTH_NODE_SZ;
	c->ranges[UBIFS_AUTH_NODE].max_len = UBIFS_AUTH_NODE_SZ +
				UBIFS_MAX_HMAC_LEN;
569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592

	c->ranges[UBIFS_INO_NODE].min_len  = UBIFS_INO_NODE_SZ;
	c->ranges[UBIFS_INO_NODE].max_len  = UBIFS_MAX_INO_NODE_SZ;
	c->ranges[UBIFS_ORPH_NODE].min_len =
				UBIFS_ORPH_NODE_SZ + sizeof(__le64);
	c->ranges[UBIFS_ORPH_NODE].max_len = c->leb_size;
	c->ranges[UBIFS_DENT_NODE].min_len = UBIFS_DENT_NODE_SZ;
	c->ranges[UBIFS_DENT_NODE].max_len = UBIFS_MAX_DENT_NODE_SZ;
	c->ranges[UBIFS_XENT_NODE].min_len = UBIFS_XENT_NODE_SZ;
	c->ranges[UBIFS_XENT_NODE].max_len = UBIFS_MAX_XENT_NODE_SZ;
	c->ranges[UBIFS_DATA_NODE].min_len = UBIFS_DATA_NODE_SZ;
	c->ranges[UBIFS_DATA_NODE].max_len = UBIFS_MAX_DATA_NODE_SZ;
	/*
	 * Minimum indexing node size is amended later when superblock is
	 * read and the key length is known.
	 */
	c->ranges[UBIFS_IDX_NODE].min_len = UBIFS_IDX_NODE_SZ + UBIFS_BRANCH_SZ;
	/*
	 * Maximum indexing node size is amended later when superblock is
	 * read and the fanout is known.
	 */
	c->ranges[UBIFS_IDX_NODE].max_len = INT_MAX;

	/*
593 594
	 * Initialize dead and dark LEB space watermarks. See gc.c for comments
	 * about these values.
595 596 597 598
	 */
	c->dead_wm = ALIGN(MIN_WRITE_SZ, c->min_io_size);
	c->dark_wm = ALIGN(UBIFS_MAX_NODE_SZ, c->min_io_size);

A
Artem Bityutskiy 已提交
599 600 601 602 603 604
	/*
	 * Calculate how many bytes would be wasted at the end of LEB if it was
	 * fully filled with data nodes of maximum size. This is used in
	 * calculations when reporting free space.
	 */
	c->leb_overhead = c->leb_size % UBIFS_MAX_DATA_NODE_SZ;
605

A
Adrian Hunter 已提交
606
	/* Buffer size for bulk-reads */
A
Artem Bityutskiy 已提交
607 608 609
	c->max_bu_buf_len = UBIFS_MAX_BULK_READ * UBIFS_MAX_DATA_NODE_SZ;
	if (c->max_bu_buf_len > c->leb_size)
		c->max_bu_buf_len = c->leb_size;
610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
	return 0;
}

/**
 * bud_wbuf_callback - bud LEB write-buffer synchronization call-back.
 * @c: UBIFS file-system description object
 * @lnum: LEB the write-buffer was synchronized to
 * @free: how many free bytes left in this LEB
 * @pad: how many bytes were padded
 *
 * This is a callback function which is called by the I/O unit when the
 * write-buffer is synchronized. We need this to correctly maintain space
 * accounting in bud logical eraseblocks. This function returns zero in case of
 * success and a negative error code in case of failure.
 *
 * This function actually belongs to the journal, but we keep it here because
 * we want to keep it static.
 */
static int bud_wbuf_callback(struct ubifs_info *c, int lnum, int free, int pad)
{
	return ubifs_update_one_lp(c, lnum, free, pad, 0, 0);
}

/*
634
 * init_constants_sb - initialize UBIFS constants.
635 636 637 638 639 640 641
 * @c: UBIFS file-system description object
 *
 * This is a helper function which initializes various UBIFS constants after
 * the superblock has been read. It also checks various UBIFS parameters and
 * makes sure they are all right. Returns zero in case of success and a
 * negative error code in case of failure.
 */
642
static int init_constants_sb(struct ubifs_info *c)
643 644
{
	int tmp, err;
A
Artem Bityutskiy 已提交
645
	long long tmp64;
646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662

	c->main_bytes = (long long)c->main_lebs * c->leb_size;
	c->max_znode_sz = sizeof(struct ubifs_znode) +
				c->fanout * sizeof(struct ubifs_zbranch);

	tmp = ubifs_idx_node_sz(c, 1);
	c->ranges[UBIFS_IDX_NODE].min_len = tmp;
	c->min_idx_node_sz = ALIGN(tmp, 8);

	tmp = ubifs_idx_node_sz(c, c->fanout);
	c->ranges[UBIFS_IDX_NODE].max_len = tmp;
	c->max_idx_node_sz = ALIGN(tmp, 8);

	/* Make sure LEB size is large enough to fit full commit */
	tmp = UBIFS_CS_NODE_SZ + UBIFS_REF_NODE_SZ * c->jhead_cnt;
	tmp = ALIGN(tmp, c->min_io_size);
	if (tmp > c->leb_size) {
663
		ubifs_err(c, "too small LEB size %d, at least %d needed",
A
Artem Bityutskiy 已提交
664
			  c->leb_size, tmp);
665 666 667 668 669 670 671
		return -EINVAL;
	}

	/*
	 * Make sure that the log is large enough to fit reference nodes for
	 * all buds plus one reserved LEB.
	 */
A
Artem Bityutskiy 已提交
672 673
	tmp64 = c->max_bud_bytes + c->leb_size - 1;
	c->max_bud_cnt = div_u64(tmp64, c->leb_size);
674 675 676 677
	tmp = (c->ref_node_alsz * c->max_bud_cnt + c->leb_size - 1);
	tmp /= c->leb_size;
	tmp += 1;
	if (c->log_lebs < tmp) {
678
		ubifs_err(c, "too small log %d LEBs, required min. %d LEBs",
A
Artem Bityutskiy 已提交
679
			  c->log_lebs, tmp);
680 681 682 683 684 685 686 687
		return -EINVAL;
	}

	/*
	 * When budgeting we assume worst-case scenarios when the pages are not
	 * be compressed and direntries are of the maximum size.
	 *
	 * Note, data, which may be stored in inodes is budgeted separately, so
688
	 * it is not included into 'c->bi.inode_budget'.
689
	 */
690 691 692
	c->bi.page_budget = UBIFS_MAX_DATA_NODE_SZ * UBIFS_BLOCKS_PER_PAGE;
	c->bi.inode_budget = UBIFS_INO_NODE_SZ;
	c->bi.dent_budget = UBIFS_MAX_DENT_NODE_SZ;
693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708

	/*
	 * When the amount of flash space used by buds becomes
	 * 'c->max_bud_bytes', UBIFS just blocks all writers and starts commit.
	 * The writers are unblocked when the commit is finished. To avoid
	 * writers to be blocked UBIFS initiates background commit in advance,
	 * when number of bud bytes becomes above the limit defined below.
	 */
	c->bg_bud_bytes = (c->max_bud_bytes * 13) >> 4;

	/*
	 * Ensure minimum journal size. All the bytes in the journal heads are
	 * considered to be used, when calculating the current journal usage.
	 * Consequently, if the journal is too small, UBIFS will treat it as
	 * always full.
	 */
A
Artem Bityutskiy 已提交
709
	tmp64 = (long long)(c->jhead_cnt + 1) * c->leb_size + 1;
710 711 712 713 714 715 716 717 718
	if (c->bg_bud_bytes < tmp64)
		c->bg_bud_bytes = tmp64;
	if (c->max_bud_bytes < tmp64 + c->leb_size)
		c->max_bud_bytes = tmp64 + c->leb_size;

	err = ubifs_calc_lpt_geom(c);
	if (err)
		return err;

719 720
	/* Initialize effective LEB size used in budgeting calculations */
	c->idx_leb_size = c->leb_size - c->max_idx_node_sz;
721 722 723 724 725 726 727 728 729 730 731 732 733 734 735
	return 0;
}

/*
 * init_constants_master - initialize UBIFS constants.
 * @c: UBIFS file-system description object
 *
 * This is a helper function which initializes various UBIFS constants after
 * the master node has been read. It also checks various UBIFS parameters and
 * makes sure they are all right.
 */
static void init_constants_master(struct ubifs_info *c)
{
	long long tmp64;

736
	c->bi.min_idx_lebs = ubifs_calc_min_idx_lebs(c);
737
	c->report_rp_size = ubifs_reported_space(c, c->rp_size);
738 739 740 741 742 743

	/*
	 * Calculate total amount of FS blocks. This number is not used
	 * internally because it does not make much sense for UBIFS, but it is
	 * necessary to report something for the 'statfs()' call.
	 *
744
	 * Subtract the LEB reserved for GC, the LEB which is reserved for
745 746
	 * deletions, minimum LEBs for the index, and assume only one journal
	 * head is available.
747
	 */
748
	tmp64 = c->main_lebs - 1 - 1 - MIN_INDEX_LEBS - c->jhead_cnt + 1;
A
Artem Bityutskiy 已提交
749
	tmp64 *= (long long)c->leb_size - c->leb_overhead;
750 751 752 753 754 755 756 757
	tmp64 = ubifs_reported_space(c, tmp64);
	c->block_cnt = tmp64 >> UBIFS_BLOCK_SHIFT;
}

/**
 * take_gc_lnum - reserve GC LEB.
 * @c: UBIFS file-system description object
 *
758 759 760 761 762 763
 * This function ensures that the LEB reserved for garbage collection is marked
 * as "taken" in lprops. We also have to set free space to LEB size and dirty
 * space to zero, because lprops may contain out-of-date information if the
 * file-system was un-mounted before it has been committed. This function
 * returns zero in case of success and a negative error code in case of
 * failure.
764 765 766 767 768 769
 */
static int take_gc_lnum(struct ubifs_info *c)
{
	int err;

	if (c->gc_lnum == -1) {
770
		ubifs_err(c, "no LEB for GC");
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790
		return -EINVAL;
	}

	/* And we have to tell lprops that this LEB is taken */
	err = ubifs_change_one_lp(c, c->gc_lnum, c->leb_size, 0,
				  LPROPS_TAKEN, 0, 0);
	return err;
}

/**
 * alloc_wbufs - allocate write-buffers.
 * @c: UBIFS file-system description object
 *
 * This helper function allocates and initializes UBIFS write-buffers. Returns
 * zero in case of success and %-ENOMEM in case of failure.
 */
static int alloc_wbufs(struct ubifs_info *c)
{
	int i, err;

791 792
	c->jheads = kcalloc(c->jhead_cnt, sizeof(struct ubifs_jhead),
			    GFP_KERNEL);
793 794 795 796 797 798 799 800 801 802 803 804
	if (!c->jheads)
		return -ENOMEM;

	/* Initialize journal heads */
	for (i = 0; i < c->jhead_cnt; i++) {
		INIT_LIST_HEAD(&c->jheads[i].buds_list);
		err = ubifs_wbuf_init(c, &c->jheads[i].wbuf);
		if (err)
			return err;

		c->jheads[i].wbuf.sync_callback = &bud_wbuf_callback;
		c->jheads[i].wbuf.jhead = i;
805
		c->jheads[i].grouped = 1;
806 807 808
		c->jheads[i].log_hash = ubifs_hash_get_desc(c);
		if (IS_ERR(c->jheads[i].log_hash))
			goto out;
809 810 811
	}

	/*
812 813
	 * Garbage Collector head does not need to be synchronized by timer.
	 * Also GC head nodes are not grouped.
814
	 */
815
	c->jheads[GCHD].wbuf.no_timer = 1;
816
	c->jheads[GCHD].grouped = 0;
817 818

	return 0;
819 820 821 822 823 824

out:
	while (i--)
		kfree(c->jheads[i].log_hash);

	return err;
825 826 827 828 829 830 831 832 833 834 835 836 837 838
}

/**
 * free_wbufs - free write-buffers.
 * @c: UBIFS file-system description object
 */
static void free_wbufs(struct ubifs_info *c)
{
	int i;

	if (c->jheads) {
		for (i = 0; i < c->jhead_cnt; i++) {
			kfree(c->jheads[i].wbuf.buf);
			kfree(c->jheads[i].wbuf.inodes);
839
			kfree(c->jheads[i].log_hash);
840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864
		}
		kfree(c->jheads);
		c->jheads = NULL;
	}
}

/**
 * free_orphans - free orphans.
 * @c: UBIFS file-system description object
 */
static void free_orphans(struct ubifs_info *c)
{
	struct ubifs_orphan *orph;

	while (c->orph_dnext) {
		orph = c->orph_dnext;
		c->orph_dnext = orph->dnext;
		list_del(&orph->list);
		kfree(orph);
	}

	while (!list_empty(&c->orph_list)) {
		orph = list_entry(c->orph_list.next, struct ubifs_orphan, list);
		list_del(&orph->list);
		kfree(orph);
865
		ubifs_err(c, "orphan list not empty at unmount");
866 867 868 869 870 871 872 873 874 875 876 877
	}

	vfree(c->orph_buf);
	c->orph_buf = NULL;
}

/**
 * free_buds - free per-bud objects.
 * @c: UBIFS file-system description object
 */
static void free_buds(struct ubifs_info *c)
{
878 879 880 881
	struct ubifs_bud *bud, *n;

	rbtree_postorder_for_each_entry_safe(bud, n, &c->buds, rb)
		kfree(bud);
882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898
}

/**
 * check_volume_empty - check if the UBI volume is empty.
 * @c: UBIFS file-system description object
 *
 * This function checks if the UBIFS volume is empty by looking if its LEBs are
 * mapped or not. The result of checking is stored in the @c->empty variable.
 * Returns zero in case of success and a negative error code in case of
 * failure.
 */
static int check_volume_empty(struct ubifs_info *c)
{
	int lnum, err;

	c->empty = 1;
	for (lnum = 0; lnum < c->leb_cnt; lnum++) {
A
Artem Bityutskiy 已提交
899
		err = ubifs_is_mapped(c, lnum);
900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917
		if (unlikely(err < 0))
			return err;
		if (err == 1) {
			c->empty = 0;
			break;
		}

		cond_resched();
	}

	return 0;
}

/*
 * UBIFS mount options.
 *
 * Opt_fast_unmount: do not run a journal commit before un-mounting
 * Opt_norm_unmount: run a journal commit before un-mounting
A
Adrian Hunter 已提交
918 919
 * Opt_bulk_read: enable bulk-reads
 * Opt_no_bulk_read: disable bulk-reads
920 921
 * Opt_chk_data_crc: check CRCs when reading data nodes
 * Opt_no_chk_data_crc: do not check CRCs when reading data nodes
922
 * Opt_override_compr: override default compressor
923
 * Opt_assert: set ubifs_assert() action
924 925
 * Opt_auth_key: The key name used for authentication
 * Opt_auth_hash_name: The hash type used for authentication
926 927 928 929 930
 * Opt_err: just end of array marker
 */
enum {
	Opt_fast_unmount,
	Opt_norm_unmount,
A
Adrian Hunter 已提交
931 932
	Opt_bulk_read,
	Opt_no_bulk_read,
933 934
	Opt_chk_data_crc,
	Opt_no_chk_data_crc,
935
	Opt_override_compr,
936
	Opt_assert,
937 938
	Opt_auth_key,
	Opt_auth_hash_name,
939
	Opt_ignore,
940 941 942
	Opt_err,
};

943
static const match_table_t tokens = {
944 945
	{Opt_fast_unmount, "fast_unmount"},
	{Opt_norm_unmount, "norm_unmount"},
A
Adrian Hunter 已提交
946 947
	{Opt_bulk_read, "bulk_read"},
	{Opt_no_bulk_read, "no_bulk_read"},
948 949
	{Opt_chk_data_crc, "chk_data_crc"},
	{Opt_no_chk_data_crc, "no_chk_data_crc"},
950
	{Opt_override_compr, "compr=%s"},
951 952
	{Opt_auth_key, "auth_key=%s"},
	{Opt_auth_hash_name, "auth_hash_name=%s"},
953 954
	{Opt_ignore, "ubi=%s"},
	{Opt_ignore, "vol=%s"},
955
	{Opt_assert, "assert=%s"},
956 957 958
	{Opt_err, NULL},
};

959 960 961 962 963 964 965 966 967 968 969 970 971 972 973
/**
 * parse_standard_option - parse a standard mount option.
 * @option: the option to parse
 *
 * Normally, standard mount options like "sync" are passed to file-systems as
 * flags. However, when a "rootflags=" kernel boot parameter is used, they may
 * be present in the options string. This function tries to deal with this
 * situation and parse standard options. Returns 0 if the option was not
 * recognized, and the corresponding integer flag if it was.
 *
 * UBIFS is only interested in the "sync" option, so do not check for anything
 * else.
 */
static int parse_standard_option(const char *option)
{
974 975

	pr_notice("UBIFS: parse %s\n", option);
976
	if (!strcmp(option, "sync"))
977
		return SB_SYNCHRONOUS;
978 979 980
	return 0;
}

981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006
/**
 * ubifs_parse_options - parse mount parameters.
 * @c: UBIFS file-system description object
 * @options: parameters to parse
 * @is_remount: non-zero if this is FS re-mount
 *
 * This function parses UBIFS mount options and returns zero in case success
 * and a negative error code in case of failure.
 */
static int ubifs_parse_options(struct ubifs_info *c, char *options,
			       int is_remount)
{
	char *p;
	substring_t args[MAX_OPT_ARGS];

	if (!options)
		return 0;

	while ((p = strsep(&options, ","))) {
		int token;

		if (!*p)
			continue;

		token = match_token(p, tokens, args);
		switch (token) {
A
Artem Bityutskiy 已提交
1007 1008
		/*
		 * %Opt_fast_unmount and %Opt_norm_unmount options are ignored.
A
Artem Bityutskiy 已提交
1009
		 * We accept them in order to be backward-compatible. But this
A
Artem Bityutskiy 已提交
1010 1011
		 * should be removed at some point.
		 */
1012 1013 1014 1015 1016 1017
		case Opt_fast_unmount:
			c->mount_opts.unmount_mode = 2;
			break;
		case Opt_norm_unmount:
			c->mount_opts.unmount_mode = 1;
			break;
A
Adrian Hunter 已提交
1018 1019 1020 1021 1022 1023 1024 1025
		case Opt_bulk_read:
			c->mount_opts.bulk_read = 2;
			c->bulk_read = 1;
			break;
		case Opt_no_bulk_read:
			c->mount_opts.bulk_read = 1;
			c->bulk_read = 0;
			break;
1026 1027 1028 1029 1030 1031 1032 1033
		case Opt_chk_data_crc:
			c->mount_opts.chk_data_crc = 2;
			c->no_chk_data_crc = 0;
			break;
		case Opt_no_chk_data_crc:
			c->mount_opts.chk_data_crc = 1;
			c->no_chk_data_crc = 1;
			break;
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
		case Opt_override_compr:
		{
			char *name = match_strdup(&args[0]);

			if (!name)
				return -ENOMEM;
			if (!strcmp(name, "none"))
				c->mount_opts.compr_type = UBIFS_COMPR_NONE;
			else if (!strcmp(name, "lzo"))
				c->mount_opts.compr_type = UBIFS_COMPR_LZO;
			else if (!strcmp(name, "zlib"))
				c->mount_opts.compr_type = UBIFS_COMPR_ZLIB;
			else {
1047
				ubifs_err(c, "unknown compressor \"%s\"", name); //FIXME: is c ready?
1048 1049 1050 1051 1052 1053 1054 1055
				kfree(name);
				return -EINVAL;
			}
			kfree(name);
			c->mount_opts.override_compr = 1;
			c->default_compr = c->mount_opts.compr_type;
			break;
		}
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
		case Opt_assert:
		{
			char *act = match_strdup(&args[0]);

			if (!act)
				return -ENOMEM;
			if (!strcmp(act, "report"))
				c->assert_action = ASSACT_REPORT;
			else if (!strcmp(act, "read-only"))
				c->assert_action = ASSACT_RO;
			else if (!strcmp(act, "panic"))
				c->assert_action = ASSACT_PANIC;
			else {
				ubifs_err(c, "unknown assert action \"%s\"", act);
				kfree(act);
				return -EINVAL;
			}
			kfree(act);
			break;
		}
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
		case Opt_auth_key:
			c->auth_key_name = kstrdup(args[0].from, GFP_KERNEL);
			if (!c->auth_key_name)
				return -ENOMEM;
			break;
		case Opt_auth_hash_name:
			c->auth_hash_name = kstrdup(args[0].from, GFP_KERNEL);
			if (!c->auth_hash_name)
				return -ENOMEM;
			break;
1086 1087
		case Opt_ignore:
			break;
1088
		default:
1089 1090 1091 1092 1093 1094
		{
			unsigned long flag;
			struct super_block *sb = c->vfs_sb;

			flag = parse_standard_option(p);
			if (!flag) {
1095
				ubifs_err(c, "unrecognized mount option \"%s\" or missing value",
1096
					  p);
1097 1098 1099 1100 1101
				return -EINVAL;
			}
			sb->s_flags |= flag;
			break;
		}
1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
		}
	}

	return 0;
}

/**
 * destroy_journal - destroy journal data structures.
 * @c: UBIFS file-system description object
 *
 * This function destroys journal data structures including those that may have
 * been created by recovery functions.
 */
static void destroy_journal(struct ubifs_info *c)
{
	while (!list_empty(&c->unclean_leb_list)) {
		struct ubifs_unclean_leb *ucleb;

		ucleb = list_entry(c->unclean_leb_list.next,
				   struct ubifs_unclean_leb, list);
		list_del(&ucleb->list);
		kfree(ucleb);
	}
	while (!list_empty(&c->old_buds)) {
		struct ubifs_bud *bud;

		bud = list_entry(c->old_buds.next, struct ubifs_bud, list);
		list_del(&bud->list);
		kfree(bud);
	}
	ubifs_destroy_idx_gc(c);
	ubifs_destroy_size_tree(c);
	ubifs_tnc_close(c);
	free_buds(c);
}

1138 1139 1140 1141 1142 1143
/**
 * bu_init - initialize bulk-read information.
 * @c: UBIFS file-system description object
 */
static void bu_init(struct ubifs_info *c)
{
1144
	ubifs_assert(c, c->bulk_read == 1);
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157

	if (c->bu.buf)
		return; /* Already initialized */

again:
	c->bu.buf = kmalloc(c->max_bu_buf_len, GFP_KERNEL | __GFP_NOWARN);
	if (!c->bu.buf) {
		if (c->max_bu_buf_len > UBIFS_KMALLOC_OK) {
			c->max_bu_buf_len = UBIFS_KMALLOC_OK;
			goto again;
		}

		/* Just disable bulk-read */
1158
		ubifs_warn(c, "cannot allocate %d bytes of memory for bulk-read, disabling it",
1159
			   c->max_bu_buf_len);
1160 1161 1162 1163 1164 1165
		c->mount_opts.bulk_read = 1;
		c->bulk_read = 0;
		return;
	}
}

1166 1167 1168 1169 1170 1171 1172 1173 1174
/**
 * check_free_space - check if there is enough free space to mount.
 * @c: UBIFS file-system description object
 *
 * This function makes sure UBIFS has enough free space to be mounted in
 * read/write mode. UBIFS must always have some free space to allow deletions.
 */
static int check_free_space(struct ubifs_info *c)
{
1175
	ubifs_assert(c, c->dark_wm > 0);
1176
	if (c->lst.total_free + c->lst.total_dirty < c->dark_wm) {
1177
		ubifs_err(c, "insufficient free space to mount in R/W mode");
1178 1179
		ubifs_dump_budg(c, &c->bi);
		ubifs_dump_lprops(c);
1180
		return -ENOSPC;
1181 1182 1183 1184
	}
	return 0;
}

1185 1186 1187 1188 1189 1190 1191 1192 1193
/**
 * mount_ubifs - mount UBIFS file-system.
 * @c: UBIFS file-system description object
 *
 * This function mounts UBIFS file system. Returns zero in case of success and
 * a negative error code in case of failure.
 */
static int mount_ubifs(struct ubifs_info *c)
{
1194
	int err;
A
Artem Bityutskiy 已提交
1195
	long long x, y;
1196 1197
	size_t sz;

1198
	c->ro_mount = !!sb_rdonly(c->vfs_sb);
1199 1200
	/* Suppress error messages while probing if SB_SILENT is set */
	c->probing = !!(c->vfs_sb->s_flags & SB_SILENT);
1201

1202 1203 1204 1205
	err = init_constants_early(c);
	if (err)
		return err;

1206 1207 1208
	err = ubifs_debugging_init(c);
	if (err)
		return err;
1209 1210 1211 1212 1213

	err = check_volume_empty(c);
	if (err)
		goto out_free;

1214
	if (c->empty && (c->ro_mount || c->ro_media)) {
1215 1216 1217 1218
		/*
		 * This UBI volume is empty, and read-only, or the file system
		 * is mounted read-only - we cannot format it.
		 */
1219
		ubifs_err(c, "can't format empty UBI volume: read-only %s",
1220 1221 1222 1223 1224
			  c->ro_media ? "UBI volume" : "mount");
		err = -EROFS;
		goto out_free;
	}

1225
	if (c->ro_media && !c->ro_mount) {
1226
		ubifs_err(c, "cannot mount read-write - read-only media");
1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
		err = -EROFS;
		goto out_free;
	}

	/*
	 * The requirement for the buffer is that it should fit indexing B-tree
	 * height amount of integers. We assume the height if the TNC tree will
	 * never exceed 64.
	 */
	err = -ENOMEM;
1237 1238
	c->bottom_up_buf = kmalloc_array(BOTTOM_UP_HEIGHT, sizeof(int),
					 GFP_KERNEL);
1239 1240 1241 1242 1243 1244 1245
	if (!c->bottom_up_buf)
		goto out_free;

	c->sbuf = vmalloc(c->leb_size);
	if (!c->sbuf)
		goto out_free;

1246
	if (!c->ro_mount) {
1247 1248 1249 1250 1251
		c->ileb_buf = vmalloc(c->leb_size);
		if (!c->ileb_buf)
			goto out_free;
	}

1252 1253 1254
	if (c->bulk_read == 1)
		bu_init(c);

1255
	if (!c->ro_mount) {
1256 1257
		c->write_reserve_buf = kmalloc(COMPRESSED_DATA_NODE_BUF_SZ + \
					       UBIFS_CIPHER_BLOCK_SIZE,
1258 1259 1260 1261 1262
					       GFP_KERNEL);
		if (!c->write_reserve_buf)
			goto out_free;
	}

1263
	c->mounting = 1;
1264

1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
	if (c->auth_key_name) {
		if (IS_ENABLED(CONFIG_UBIFS_FS_AUTHENTICATION)) {
			err = ubifs_init_authentication(c);
			if (err)
				goto out_free;
		} else {
			ubifs_err(c, "auth_key_name, but UBIFS is built without"
				  " authentication support");
			err = -EINVAL;
			goto out_free;
		}
	}

1278 1279 1280 1281
	err = ubifs_read_superblock(c);
	if (err)
		goto out_free;

1282 1283
	c->probing = 0;

1284
	/*
1285
	 * Make sure the compressor which is set as default in the superblock
1286
	 * or overridden by mount options is actually compiled in.
1287
	 */
1288
	if (!ubifs_compr_present(c, c->default_compr)) {
1289
		ubifs_err(c, "'compressor \"%s\" is not compiled in",
1290
			  ubifs_compr_name(c, c->default_compr));
1291
		err = -ENOTSUPP;
1292
		goto out_free;
1293 1294
	}

1295
	err = init_constants_sb(c);
1296
	if (err)
1297
		goto out_free;
1298

L
Liu Song 已提交
1299
	sz = ALIGN(c->max_idx_node_sz, c->min_io_size) * 2;
1300 1301 1302
	c->cbuf = kmalloc(sz, GFP_NOFS);
	if (!c->cbuf) {
		err = -ENOMEM;
1303
		goto out_free;
1304 1305
	}

1306 1307 1308 1309
	err = alloc_wbufs(c);
	if (err)
		goto out_cbuf;

1310
	sprintf(c->bgt_name, BGT_NAME_PATTERN, c->vi.ubi_num, c->vi.vol_id);
1311
	if (!c->ro_mount) {
1312
		/* Create background thread */
H
Hunter Adrian 已提交
1313
		c->bgt = kthread_create(ubifs_bg_thread, c, "%s", c->bgt_name);
1314 1315 1316
		if (IS_ERR(c->bgt)) {
			err = PTR_ERR(c->bgt);
			c->bgt = NULL;
1317
			ubifs_err(c, "cannot spawn \"%s\", error %d",
1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
				  c->bgt_name, err);
			goto out_wbufs;
		}
		wake_up_process(c->bgt);
	}

	err = ubifs_read_master(c);
	if (err)
		goto out_master;

B
Ben Gardiner 已提交
1328 1329
	init_constants_master(c);

1330
	if ((c->mst_node->flags & cpu_to_le32(UBIFS_MST_DIRTY)) != 0) {
1331
		ubifs_msg(c, "recovery needed");
1332
		c->need_recovery = 1;
B
Ben Gardiner 已提交
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344
	}

	if (c->need_recovery && !c->ro_mount) {
		err = ubifs_recover_inl_heads(c, c->sbuf);
		if (err)
			goto out_master;
	}

	err = ubifs_lpt_init(c, 1, !c->ro_mount);
	if (err)
		goto out_master;

B
Ben Gardiner 已提交
1345 1346 1347
	if (!c->ro_mount && c->space_fixup) {
		err = ubifs_fixup_free_space(c);
		if (err)
1348
			goto out_lpt;
B
Ben Gardiner 已提交
1349 1350
	}

1351
	if (!c->ro_mount && !c->need_recovery) {
1352 1353 1354 1355 1356 1357 1358
		/*
		 * Set the "dirty" flag so that if we reboot uncleanly we
		 * will notice this immediately on the next mount.
		 */
		c->mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY);
		err = ubifs_write_master(c);
		if (err)
B
Ben Gardiner 已提交
1359
			goto out_lpt;
1360 1361
	}

1362
	err = dbg_check_idx_size(c, c->bi.old_idx_sz);
1363 1364 1365 1366 1367 1368 1369
	if (err)
		goto out_lpt;

	err = ubifs_replay_journal(c);
	if (err)
		goto out_journal;

A
Artem Bityutskiy 已提交
1370
	/* Calculate 'min_idx_lebs' after journal replay */
1371
	c->bi.min_idx_lebs = ubifs_calc_min_idx_lebs(c);
A
Artem Bityutskiy 已提交
1372

1373
	err = ubifs_mount_orphans(c, c->need_recovery, c->ro_mount);
1374 1375 1376
	if (err)
		goto out_orphans;

1377
	if (!c->ro_mount) {
1378 1379
		int lnum;

1380 1381
		err = check_free_space(c);
		if (err)
1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
			goto out_orphans;

		/* Check for enough log space */
		lnum = c->lhead_lnum + 1;
		if (lnum >= UBIFS_LOG_LNUM + c->log_lebs)
			lnum = UBIFS_LOG_LNUM;
		if (lnum == c->ltail_lnum) {
			err = ubifs_consolidate_log(c);
			if (err)
				goto out_orphans;
		}

		if (c->need_recovery) {
1395 1396 1397 1398 1399 1400
			if (!ubifs_authenticated(c)) {
				err = ubifs_recover_size(c, true);
				if (err)
					goto out_orphans;
			}

1401
			err = ubifs_rcvry_gc_commit(c);
A
Artem Bityutskiy 已提交
1402 1403
			if (err)
				goto out_orphans;
1404 1405 1406 1407 1408 1409

			if (ubifs_authenticated(c)) {
				err = ubifs_recover_size(c, false);
				if (err)
					goto out_orphans;
			}
1410
		} else {
1411
			err = take_gc_lnum(c);
1412 1413 1414 1415 1416 1417 1418 1419 1420
			if (err)
				goto out_orphans;

			/*
			 * GC LEB may contain garbage if there was an unclean
			 * reboot, and it should be un-mapped.
			 */
			err = ubifs_leb_unmap(c, c->gc_lnum);
			if (err)
1421
				goto out_orphans;
1422
		}
1423 1424 1425 1426 1427

		err = dbg_check_lprops(c);
		if (err)
			goto out_orphans;
	} else if (c->need_recovery) {
1428
		err = ubifs_recover_size(c, false);
1429 1430
		if (err)
			goto out_orphans;
1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
	} else {
		/*
		 * Even if we mount read-only, we have to set space in GC LEB
		 * to proper value because this affects UBIFS free space
		 * reporting. We do not want to have a situation when
		 * re-mounting from R/O to R/W changes amount of free space.
		 */
		err = take_gc_lnum(c);
		if (err)
			goto out_orphans;
1441 1442 1443 1444 1445 1446 1447
	}

	spin_lock(&ubifs_infos_lock);
	list_add_tail(&c->infos_list, &ubifs_infos);
	spin_unlock(&ubifs_infos_lock);

	if (c->need_recovery) {
1448
		if (c->ro_mount)
1449
			ubifs_msg(c, "recovery deferred");
1450 1451
		else {
			c->need_recovery = 0;
1452
			ubifs_msg(c, "recovery completed");
A
Artem Bityutskiy 已提交
1453 1454 1455 1456 1457
			/*
			 * GC LEB has to be empty and taken at this point. But
			 * the journal head LEBs may also be accounted as
			 * "empty taken" if they are empty.
			 */
1458
			ubifs_assert(c, c->lst.taken_empty_lebs > 0);
1459
		}
A
Artem Bityutskiy 已提交
1460
	} else
1461
		ubifs_assert(c, c->lst.taken_empty_lebs > 0);
1462

A
Artem Bityutskiy 已提交
1463
	err = dbg_check_filesystem(c);
A
Artem Bityutskiy 已提交
1464 1465 1466
	if (err)
		goto out_infos;

A
Artem Bityutskiy 已提交
1467
	err = dbg_debugfs_init_fs(c);
1468 1469 1470
	if (err)
		goto out_infos;

1471
	c->mounting = 0;
1472

1473
	ubifs_msg(c, "UBIFS: mounted UBI device %d, volume %d, name \"%s\"%s",
A
Artem Bityutskiy 已提交
1474
		  c->vi.ubi_num, c->vi.vol_id, c->vi.name,
R
Richard Genoud 已提交
1475
		  c->ro_mount ? ", R/O mode" : "");
1476
	x = (long long)c->main_lebs * c->leb_size;
A
Artem Bityutskiy 已提交
1477
	y = (long long)c->log_lebs * c->leb_size + c->max_bud_bytes;
1478
	ubifs_msg(c, "LEB size: %d bytes (%d KiB), min./max. I/O unit sizes: %d bytes/%d bytes",
A
Artem Bityutskiy 已提交
1479 1480
		  c->leb_size, c->leb_size >> 10, c->min_io_size,
		  c->max_write_size);
1481
	ubifs_msg(c, "FS size: %lld bytes (%lld MiB, %d LEBs), journal size %lld bytes (%lld MiB, %d LEBs)",
A
Artem Bityutskiy 已提交
1482 1483
		  x, x >> 20, c->main_lebs,
		  y, y >> 20, c->log_lebs + c->max_bud_cnt);
1484
	ubifs_msg(c, "reserved for root: %llu bytes (%llu KiB)",
A
Artem Bityutskiy 已提交
1485
		  c->report_rp_size, c->report_rp_size >> 10);
1486
	ubifs_msg(c, "media format: w%d/r%d (latest is w%d/r%d), UUID %pUB%s",
A
Artem Bityutskiy 已提交
1487
		  c->fmt_version, c->ro_compat_version,
A
Artem Bityutskiy 已提交
1488 1489 1490
		  UBIFS_FORMAT_VERSION, UBIFS_RO_COMPAT_VERSION, c->uuid,
		  c->big_lpt ? ", big LPT model" : ", small LPT model");

1491
	dbg_gen("default compressor:  %s", ubifs_compr_name(c, c->default_compr));
A
Artem Bityutskiy 已提交
1492
	dbg_gen("data journal heads:  %d",
1493
		c->jhead_cnt - NONDATA_JHEADS_CNT);
A
Artem Bityutskiy 已提交
1494
	dbg_gen("log LEBs:            %d (%d - %d)",
1495
		c->log_lebs, UBIFS_LOG_LNUM, c->log_last);
A
Artem Bityutskiy 已提交
1496
	dbg_gen("LPT area LEBs:       %d (%d - %d)",
1497
		c->lpt_lebs, c->lpt_first, c->lpt_last);
A
Artem Bityutskiy 已提交
1498
	dbg_gen("orphan area LEBs:    %d (%d - %d)",
1499
		c->orph_lebs, c->orph_first, c->orph_last);
A
Artem Bityutskiy 已提交
1500
	dbg_gen("main area LEBs:      %d (%d - %d)",
1501
		c->main_lebs, c->main_first, c->leb_cnt - 1);
A
Artem Bityutskiy 已提交
1502 1503
	dbg_gen("index LEBs:          %d", c->lst.idx_lebs);
	dbg_gen("total index bytes:   %lld (%lld KiB, %lld MiB)",
1504 1505
		c->bi.old_idx_sz, c->bi.old_idx_sz >> 10,
		c->bi.old_idx_sz >> 20);
A
Artem Bityutskiy 已提交
1506 1507 1508 1509 1510 1511
	dbg_gen("key hash type:       %d", c->key_hash_type);
	dbg_gen("tree fanout:         %d", c->fanout);
	dbg_gen("reserved GC LEB:     %d", c->gc_lnum);
	dbg_gen("max. znode size      %d", c->max_znode_sz);
	dbg_gen("max. index node size %d", c->max_idx_node_sz);
	dbg_gen("node sizes:          data %zu, inode %zu, dentry %zu",
1512
		UBIFS_DATA_NODE_SZ, UBIFS_INO_NODE_SZ, UBIFS_DENT_NODE_SZ);
A
Artem Bityutskiy 已提交
1513
	dbg_gen("node sizes:          trun %zu, sb %zu, master %zu",
1514
		UBIFS_TRUN_NODE_SZ, UBIFS_SB_NODE_SZ, UBIFS_MST_NODE_SZ);
A
Artem Bityutskiy 已提交
1515
	dbg_gen("node sizes:          ref %zu, cmt. start %zu, orph %zu",
1516
		UBIFS_REF_NODE_SZ, UBIFS_CS_NODE_SZ, UBIFS_ORPH_NODE_SZ);
A
Artem Bityutskiy 已提交
1517
	dbg_gen("max. node sizes:     data %zu, inode %zu dentry %zu, idx %d",
1518
		UBIFS_MAX_DATA_NODE_SZ, UBIFS_MAX_INO_NODE_SZ,
1519
		UBIFS_MAX_DENT_NODE_SZ, ubifs_idx_node_sz(c, c->fanout));
A
Artem Bityutskiy 已提交
1520 1521 1522
	dbg_gen("dead watermark:      %d", c->dead_wm);
	dbg_gen("dark watermark:      %d", c->dark_wm);
	dbg_gen("LEB overhead:        %d", c->leb_overhead);
1523
	x = (long long)c->main_lebs * c->dark_wm;
A
Artem Bityutskiy 已提交
1524
	dbg_gen("max. dark space:     %lld (%lld KiB, %lld MiB)",
1525
		x, x >> 10, x >> 20);
A
Artem Bityutskiy 已提交
1526
	dbg_gen("maximum bud bytes:   %lld (%lld KiB, %lld MiB)",
1527 1528
		c->max_bud_bytes, c->max_bud_bytes >> 10,
		c->max_bud_bytes >> 20);
A
Artem Bityutskiy 已提交
1529
	dbg_gen("BG commit bud bytes: %lld (%lld KiB, %lld MiB)",
1530 1531
		c->bg_bud_bytes, c->bg_bud_bytes >> 10,
		c->bg_bud_bytes >> 20);
A
Artem Bityutskiy 已提交
1532
	dbg_gen("current bud bytes    %lld (%lld KiB, %lld MiB)",
1533
		c->bud_bytes, c->bud_bytes >> 10, c->bud_bytes >> 20);
A
Artem Bityutskiy 已提交
1534 1535
	dbg_gen("max. seq. number:    %llu", c->max_sqnum);
	dbg_gen("commit number:       %llu", c->cmt_no);
1536 1537
	dbg_gen("max. xattrs per inode: %d", ubifs_xattr_max_cnt(c));
	dbg_gen("max orphans:           %d", c->max_orphans);
1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560

	return 0;

out_infos:
	spin_lock(&ubifs_infos_lock);
	list_del(&c->infos_list);
	spin_unlock(&ubifs_infos_lock);
out_orphans:
	free_orphans(c);
out_journal:
	destroy_journal(c);
out_lpt:
	ubifs_lpt_free(c, 0);
out_master:
	kfree(c->mst_node);
	kfree(c->rcvrd_mst_node);
	if (c->bgt)
		kthread_stop(c->bgt);
out_wbufs:
	free_wbufs(c);
out_cbuf:
	kfree(c->cbuf);
out_free:
1561
	kfree(c->write_reserve_buf);
1562
	kfree(c->bu.buf);
1563 1564 1565
	vfree(c->ileb_buf);
	vfree(c->sbuf);
	kfree(c->bottom_up_buf);
1566
	ubifs_debugging_exit(c);
1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583
	return err;
}

/**
 * ubifs_umount - un-mount UBIFS file-system.
 * @c: UBIFS file-system description object
 *
 * Note, this function is called to free allocated resourced when un-mounting,
 * as well as free resources when an error occurred while we were half way
 * through mounting (error path cleanup function). So it has to make sure the
 * resource was actually allocated before freeing it.
 */
static void ubifs_umount(struct ubifs_info *c)
{
	dbg_gen("un-mounting UBI device %d, volume %d", c->vi.ubi_num,
		c->vi.vol_id);

A
Artem Bityutskiy 已提交
1584
	dbg_debugfs_exit_fs(c);
1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595
	spin_lock(&ubifs_infos_lock);
	list_del(&c->infos_list);
	spin_unlock(&ubifs_infos_lock);

	if (c->bgt)
		kthread_stop(c->bgt);

	destroy_journal(c);
	free_wbufs(c);
	free_orphans(c);
	ubifs_lpt_free(c, 0);
1596
	ubifs_exit_authentication(c);
1597

1598 1599
	kfree(c->auth_key_name);
	kfree(c->auth_hash_name);
1600 1601 1602
	kfree(c->cbuf);
	kfree(c->rcvrd_mst_node);
	kfree(c->mst_node);
1603
	kfree(c->write_reserve_buf);
1604 1605
	kfree(c->bu.buf);
	vfree(c->ileb_buf);
1606 1607
	vfree(c->sbuf);
	kfree(c->bottom_up_buf);
1608
	ubifs_debugging_exit(c);
1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622
}

/**
 * ubifs_remount_rw - re-mount in read-write mode.
 * @c: UBIFS file-system description object
 *
 * UBIFS avoids allocating many unnecessary resources when mounted in read-only
 * mode. This function allocates the needed resources and re-mounts UBIFS in
 * read-write mode.
 */
static int ubifs_remount_rw(struct ubifs_info *c)
{
	int err, lnum;

A
Artem Bityutskiy 已提交
1623
	if (c->rw_incompat) {
1624 1625
		ubifs_err(c, "the file-system is not R/W-compatible");
		ubifs_msg(c, "on-flash format version is w%d/r%d, but software only supports up to version w%d/r%d",
1626 1627
			  c->fmt_version, c->ro_compat_version,
			  UBIFS_FORMAT_VERSION, UBIFS_RO_COMPAT_VERSION);
A
Artem Bityutskiy 已提交
1628 1629 1630
		return -EROFS;
	}

1631
	mutex_lock(&c->umount_mutex);
1632
	dbg_save_space_info(c);
1633
	c->remounting_rw = 1;
A
Artem Bityutskiy 已提交
1634
	c->ro_mount = 0;
1635

1636 1637 1638
	if (c->space_fixup) {
		err = ubifs_fixup_free_space(c);
		if (err)
A
Artem Bityutskiy 已提交
1639
			goto out;
1640 1641
	}

1642 1643
	err = check_free_space(c);
	if (err)
1644 1645 1646
		goto out;

	if (c->old_leb_cnt != c->leb_cnt) {
S
Sascha Hauer 已提交
1647
		struct ubifs_sb_node *sup = c->sup_node;
1648 1649 1650 1651 1652 1653 1654 1655

		sup->leb_cnt = cpu_to_le32(c->leb_cnt);
		err = ubifs_write_sb_node(c, sup);
		if (err)
			goto out;
	}

	if (c->need_recovery) {
1656
		ubifs_msg(c, "completing deferred recovery");
1657 1658 1659
		err = ubifs_write_rcvrd_mst_node(c);
		if (err)
			goto out;
1660 1661 1662 1663 1664
		if (!ubifs_authenticated(c)) {
			err = ubifs_recover_size(c, true);
			if (err)
				goto out;
		}
1665 1666 1667 1668 1669 1670
		err = ubifs_clean_lebs(c, c->sbuf);
		if (err)
			goto out;
		err = ubifs_recover_inl_heads(c, c->sbuf);
		if (err)
			goto out;
1671 1672
	} else {
		/* A readonly mount is not allowed to have orphans */
1673
		ubifs_assert(c, c->tot_orphans == 0);
1674 1675 1676
		err = ubifs_clear_orphans(c);
		if (err)
			goto out;
1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691
	}

	if (!(c->mst_node->flags & cpu_to_le32(UBIFS_MST_DIRTY))) {
		c->mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY);
		err = ubifs_write_master(c);
		if (err)
			goto out;
	}

	c->ileb_buf = vmalloc(c->leb_size);
	if (!c->ileb_buf) {
		err = -ENOMEM;
		goto out;
	}

1692 1693
	c->write_reserve_buf = kmalloc(COMPRESSED_DATA_NODE_BUF_SZ + \
				       UBIFS_CIPHER_BLOCK_SIZE, GFP_KERNEL);
W
Wei Yongjun 已提交
1694 1695
	if (!c->write_reserve_buf) {
		err = -ENOMEM;
1696
		goto out;
W
Wei Yongjun 已提交
1697
	}
1698

1699 1700 1701 1702 1703
	err = ubifs_lpt_init(c, 0, 1);
	if (err)
		goto out;

	/* Create background thread */
H
Hunter Adrian 已提交
1704
	c->bgt = kthread_create(ubifs_bg_thread, c, "%s", c->bgt_name);
1705 1706 1707
	if (IS_ERR(c->bgt)) {
		err = PTR_ERR(c->bgt);
		c->bgt = NULL;
1708
		ubifs_err(c, "cannot spawn \"%s\", error %d",
1709
			  c->bgt_name, err);
1710
		goto out;
1711 1712 1713 1714
	}
	wake_up_process(c->bgt);

	c->orph_buf = vmalloc(c->leb_size);
1715 1716 1717 1718
	if (!c->orph_buf) {
		err = -ENOMEM;
		goto out;
	}
1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729

	/* Check for enough log space */
	lnum = c->lhead_lnum + 1;
	if (lnum >= UBIFS_LOG_LNUM + c->log_lebs)
		lnum = UBIFS_LOG_LNUM;
	if (lnum == c->ltail_lnum) {
		err = ubifs_consolidate_log(c);
		if (err)
			goto out;
	}

1730
	if (c->need_recovery) {
1731
		err = ubifs_rcvry_gc_commit(c);
1732 1733 1734 1735 1736 1737 1738 1739 1740
		if (err)
			goto out;

		if (ubifs_authenticated(c)) {
			err = ubifs_recover_size(c, false);
			if (err)
				goto out;
		}
	} else {
1741
		err = ubifs_leb_unmap(c, c->gc_lnum);
1742
	}
1743 1744 1745
	if (err)
		goto out;

1746 1747 1748
	dbg_gen("re-mounted read-write");
	c->remounting_rw = 0;

1749 1750
	if (c->need_recovery) {
		c->need_recovery = 0;
1751
		ubifs_msg(c, "deferred recovery completed");
1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
	} else {
		/*
		 * Do not run the debugging space check if the were doing
		 * recovery, because when we saved the information we had the
		 * file-system in a state where the TNC and lprops has been
		 * modified in memory, but all the I/O operations (including a
		 * commit) were deferred. So the file-system was in
		 * "non-committed" state. Now the file-system is in committed
		 * state, and of course the amount of free space will change
		 * because, for example, the old index size was imprecise.
		 */
		err = dbg_check_space_info(c);
1764
	}
1765

1766
	mutex_unlock(&c->umount_mutex);
1767
	return err;
1768 1769

out:
A
Artem Bityutskiy 已提交
1770
	c->ro_mount = 1;
1771 1772 1773 1774 1775 1776 1777
	vfree(c->orph_buf);
	c->orph_buf = NULL;
	if (c->bgt) {
		kthread_stop(c->bgt);
		c->bgt = NULL;
	}
	free_wbufs(c);
1778 1779
	kfree(c->write_reserve_buf);
	c->write_reserve_buf = NULL;
1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
	vfree(c->ileb_buf);
	c->ileb_buf = NULL;
	ubifs_lpt_free(c, 1);
	c->remounting_rw = 0;
	mutex_unlock(&c->umount_mutex);
	return err;
}

/**
 * ubifs_remount_ro - re-mount in read-only mode.
 * @c: UBIFS file-system description object
 *
1792 1793
 * We assume VFS has stopped writing. Possibly the background thread could be
 * running a commit, however kthread_stop will wait in that case.
1794 1795 1796 1797 1798
 */
static void ubifs_remount_ro(struct ubifs_info *c)
{
	int i, err;

1799 1800
	ubifs_assert(c, !c->need_recovery);
	ubifs_assert(c, !c->ro_mount);
A
Artem Bityutskiy 已提交
1801

1802 1803 1804 1805 1806 1807
	mutex_lock(&c->umount_mutex);
	if (c->bgt) {
		kthread_stop(c->bgt);
		c->bgt = NULL;
	}

1808 1809
	dbg_save_space_info(c);

1810 1811 1812 1813 1814
	for (i = 0; i < c->jhead_cnt; i++) {
		err = ubifs_wbuf_sync(&c->jheads[i].wbuf);
		if (err)
			ubifs_ro_mode(c, err);
	}
1815

A
Artem Bityutskiy 已提交
1816 1817 1818 1819 1820 1821 1822
	c->mst_node->flags &= ~cpu_to_le32(UBIFS_MST_DIRTY);
	c->mst_node->flags |= cpu_to_le32(UBIFS_MST_NO_ORPHS);
	c->mst_node->gc_lnum = cpu_to_le32(c->gc_lnum);
	err = ubifs_write_master(c);
	if (err)
		ubifs_ro_mode(c, err);

1823 1824
	vfree(c->orph_buf);
	c->orph_buf = NULL;
1825 1826
	kfree(c->write_reserve_buf);
	c->write_reserve_buf = NULL;
1827 1828 1829
	vfree(c->ileb_buf);
	c->ileb_buf = NULL;
	ubifs_lpt_free(c, 1);
1830
	c->ro_mount = 1;
1831 1832 1833
	err = dbg_check_space_info(c);
	if (err)
		ubifs_ro_mode(c, err);
1834 1835 1836 1837 1838 1839 1840 1841
	mutex_unlock(&c->umount_mutex);
}

static void ubifs_put_super(struct super_block *sb)
{
	int i;
	struct ubifs_info *c = sb->s_fs_info;

1842
	ubifs_msg(c, "un-mount UBI device %d", c->vi.ubi_num);
1843

1844 1845 1846 1847 1848
	/*
	 * The following asserts are only valid if there has not been a failure
	 * of the media. For example, there will be dirty inodes if we failed
	 * to write them back because of I/O errors.
	 */
1849
	if (!c->ro_error) {
1850 1851 1852
		ubifs_assert(c, c->bi.idx_growth == 0);
		ubifs_assert(c, c->bi.dd_growth == 0);
		ubifs_assert(c, c->bi.data_growth == 0);
1853
	}
1854 1855 1856 1857 1858 1859 1860 1861

	/*
	 * The 'c->umount_lock' prevents races between UBIFS memory shrinker
	 * and file system un-mount. Namely, it prevents the shrinker from
	 * picking this superblock for shrinking - it will be just skipped if
	 * the mutex is locked.
	 */
	mutex_lock(&c->umount_mutex);
1862
	if (!c->ro_mount) {
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872
		/*
		 * First of all kill the background thread to make sure it does
		 * not interfere with un-mounting and freeing resources.
		 */
		if (c->bgt) {
			kthread_stop(c->bgt);
			c->bgt = NULL;
		}

		/*
1873
		 * On fatal errors c->ro_error is set to 1, in which case we do
1874 1875
		 * not write the master node.
		 */
1876
		if (!c->ro_error) {
1877 1878 1879
			int err;

			/* Synchronize write-buffers */
1880 1881 1882 1883 1884
			for (i = 0; i < c->jhead_cnt; i++) {
				err = ubifs_wbuf_sync(&c->jheads[i].wbuf);
				if (err)
					ubifs_ro_mode(c, err);
			}
1885

1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900
			/*
			 * We are being cleanly unmounted which means the
			 * orphans were killed - indicate this in the master
			 * node. Also save the reserved GC LEB number.
			 */
			c->mst_node->flags &= ~cpu_to_le32(UBIFS_MST_DIRTY);
			c->mst_node->flags |= cpu_to_le32(UBIFS_MST_NO_ORPHS);
			c->mst_node->gc_lnum = cpu_to_le32(c->gc_lnum);
			err = ubifs_write_master(c);
			if (err)
				/*
				 * Recovery will attempt to fix the master area
				 * next mount, so we just print a message and
				 * continue to unmount normally.
				 */
1901
				ubifs_err(c, "failed to write master node, error %d",
1902
					  err);
1903 1904 1905 1906
		} else {
			for (i = 0; i < c->jhead_cnt; i++)
				/* Make sure write-buffer timers are canceled */
				hrtimer_cancel(&c->jheads[i].wbuf.timer);
1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919
		}
	}

	ubifs_umount(c);
	ubi_close_volume(c->ubi);
	mutex_unlock(&c->umount_mutex);
}

static int ubifs_remount_fs(struct super_block *sb, int *flags, char *data)
{
	int err;
	struct ubifs_info *c = sb->s_fs_info;

1920
	sync_filesystem(sb);
1921 1922 1923 1924
	dbg_gen("old flags %#lx, new flags %#x", sb->s_flags, *flags);

	err = ubifs_parse_options(c, data, 1);
	if (err) {
1925
		ubifs_err(c, "invalid or unknown remount parameter");
1926 1927
		return err;
	}
1928

1929
	if (c->ro_mount && !(*flags & SB_RDONLY)) {
1930
		if (c->ro_error) {
1931
			ubifs_msg(c, "cannot re-mount R/W due to prior errors");
1932 1933
			return -EROFS;
		}
A
Artem Bityutskiy 已提交
1934
		if (c->ro_media) {
1935
			ubifs_msg(c, "cannot re-mount R/W - UBI volume is R/O");
1936
			return -EROFS;
A
Artem Bityutskiy 已提交
1937
		}
1938
		err = ubifs_remount_rw(c);
A
Artem Bityutskiy 已提交
1939
		if (err)
1940
			return err;
1941
	} else if (!c->ro_mount && (*flags & SB_RDONLY)) {
1942
		if (c->ro_error) {
1943
			ubifs_msg(c, "cannot re-mount R/O due to prior errors");
1944
			return -EROFS;
A
Adrian Hunter 已提交
1945
		}
1946
		ubifs_remount_ro(c);
A
Adrian Hunter 已提交
1947
	}
1948

1949 1950 1951 1952
	if (c->bulk_read == 1)
		bu_init(c);
	else {
		dbg_gen("disable bulk-read");
1953
		mutex_lock(&c->bu_mutex);
1954 1955
		kfree(c->bu.buf);
		c->bu.buf = NULL;
1956
		mutex_unlock(&c->bu_mutex);
1957 1958
	}

1959 1960 1961
	if (!c->need_recovery)
		ubifs_assert(c, c->lst.taken_empty_lebs > 0);

1962 1963 1964
	return 0;
}

A
Artem Bityutskiy 已提交
1965
const struct super_operations ubifs_super_operations = {
1966
	.alloc_inode   = ubifs_alloc_inode,
A
Al Viro 已提交
1967
	.free_inode    = ubifs_free_inode,
1968 1969
	.put_super     = ubifs_put_super,
	.write_inode   = ubifs_write_inode,
A
Al Viro 已提交
1970
	.evict_inode   = ubifs_evict_inode,
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982
	.statfs        = ubifs_statfs,
	.dirty_inode   = ubifs_dirty_inode,
	.remount_fs    = ubifs_remount_fs,
	.show_options  = ubifs_show_options,
	.sync_fs       = ubifs_sync_fs,
};

/**
 * open_ubi - parse UBI device name string and open the UBI device.
 * @name: UBI volume name
 * @mode: UBI volume open mode
 *
1983 1984 1985 1986 1987 1988
 * The primary method of mounting UBIFS is by specifying the UBI volume
 * character device node path. However, UBIFS may also be mounted withoug any
 * character device node using one of the following methods:
 *
 * o ubiX_Y    - mount UBI device number X, volume Y;
 * o ubiY      - mount UBI device number 0, volume Y;
1989 1990 1991 1992 1993
 * o ubiX:NAME - mount UBI device X, volume with name NAME;
 * o ubi:NAME  - mount UBI device 0, volume with name NAME.
 *
 * Alternative '!' separator may be used instead of ':' (because some shells
 * like busybox may interpret ':' as an NFS host name separator). This function
1994 1995
 * returns UBI volume description object in case of success and a negative
 * error code in case of failure.
1996 1997 1998
 */
static struct ubi_volume_desc *open_ubi(const char *name, int mode)
{
1999
	struct ubi_volume_desc *ubi;
2000 2001 2002
	int dev, vol;
	char *endptr;

2003 2004 2005
	if (!name || !*name)
		return ERR_PTR(-EINVAL);

2006 2007 2008 2009 2010 2011
	/* First, try to open using the device node path method */
	ubi = ubi_open_volume_path(name, mode);
	if (!IS_ERR(ubi))
		return ubi;

	/* Try the "nodev" method */
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042
	if (name[0] != 'u' || name[1] != 'b' || name[2] != 'i')
		return ERR_PTR(-EINVAL);

	/* ubi:NAME method */
	if ((name[3] == ':' || name[3] == '!') && name[4] != '\0')
		return ubi_open_volume_nm(0, name + 4, mode);

	if (!isdigit(name[3]))
		return ERR_PTR(-EINVAL);

	dev = simple_strtoul(name + 3, &endptr, 0);

	/* ubiY method */
	if (*endptr == '\0')
		return ubi_open_volume(0, dev, mode);

	/* ubiX_Y method */
	if (*endptr == '_' && isdigit(endptr[1])) {
		vol = simple_strtoul(endptr + 1, &endptr, 0);
		if (*endptr != '\0')
			return ERR_PTR(-EINVAL);
		return ubi_open_volume(dev, vol, mode);
	}

	/* ubiX:NAME method */
	if ((*endptr == ':' || *endptr == '!') && endptr[1] != '\0')
		return ubi_open_volume_nm(dev, ++endptr, mode);

	return ERR_PTR(-EINVAL);
}

2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078
static struct ubifs_info *alloc_ubifs_info(struct ubi_volume_desc *ubi)
{
	struct ubifs_info *c;

	c = kzalloc(sizeof(struct ubifs_info), GFP_KERNEL);
	if (c) {
		spin_lock_init(&c->cnt_lock);
		spin_lock_init(&c->cs_lock);
		spin_lock_init(&c->buds_lock);
		spin_lock_init(&c->space_lock);
		spin_lock_init(&c->orphan_lock);
		init_rwsem(&c->commit_sem);
		mutex_init(&c->lp_mutex);
		mutex_init(&c->tnc_mutex);
		mutex_init(&c->log_mutex);
		mutex_init(&c->umount_mutex);
		mutex_init(&c->bu_mutex);
		mutex_init(&c->write_reserve_mutex);
		init_waitqueue_head(&c->cmt_wq);
		c->buds = RB_ROOT;
		c->old_idx = RB_ROOT;
		c->size_tree = RB_ROOT;
		c->orph_tree = RB_ROOT;
		INIT_LIST_HEAD(&c->infos_list);
		INIT_LIST_HEAD(&c->idx_gc);
		INIT_LIST_HEAD(&c->replay_list);
		INIT_LIST_HEAD(&c->replay_buds);
		INIT_LIST_HEAD(&c->uncat_list);
		INIT_LIST_HEAD(&c->empty_list);
		INIT_LIST_HEAD(&c->freeable_list);
		INIT_LIST_HEAD(&c->frdi_idx_list);
		INIT_LIST_HEAD(&c->unclean_leb_list);
		INIT_LIST_HEAD(&c->old_buds);
		INIT_LIST_HEAD(&c->orph_list);
		INIT_LIST_HEAD(&c->orph_new);
		c->no_chk_data_crc = 1;
2079
		c->assert_action = ASSACT_RO;
2080 2081 2082 2083 2084 2085 2086 2087 2088 2089

		c->highest_inum = UBIFS_FIRST_INO;
		c->lhead_lnum = c->ltail_lnum = UBIFS_LOG_LNUM;

		ubi_get_volume_info(ubi, &c->vi);
		ubi_get_device_info(c->vi.ubi_num, &c->di);
	}
	return c;
}

2090 2091
static int ubifs_fill_super(struct super_block *sb, void *data, int silent)
{
A
Al Viro 已提交
2092
	struct ubifs_info *c = sb->s_fs_info;
2093 2094 2095
	struct inode *root;
	int err;

2096
	c->vfs_sb = sb;
2097 2098 2099 2100
	/* Re-open the UBI device in read-write mode */
	c->ubi = ubi_open_volume(c->vi.ubi_num, c->vi.vol_id, UBI_READWRITE);
	if (IS_ERR(c->ubi)) {
		err = PTR_ERR(c->ubi);
A
Al Viro 已提交
2101
		goto out;
2102 2103
	}

2104 2105 2106 2107
	err = ubifs_parse_options(c, data, 0);
	if (err)
		goto out_close;

2108
	/*
A
Artem Bityutskiy 已提交
2109
	 * UBIFS provides 'backing_dev_info' in order to disable read-ahead. For
2110 2111
	 * UBIFS, I/O is not deferred, it is done immediately in readpage,
	 * which means the user would have to wait not just for their own I/O
A
Artem Bityutskiy 已提交
2112
	 * but the read-ahead I/O as well i.e. completely pointless.
2113
	 *
2114 2115 2116
	 * Read-ahead will be disabled because @sb->s_bdi->ra_pages is 0. Also
	 * @sb->s_bdi->capabilities are initialized to 0 so there won't be any
	 * writeback happening.
2117
	 */
2118 2119
	err = super_setup_bdi_name(sb, "ubifs_%d_%d", c->vi.ubi_num,
				   c->vi.vol_id);
2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130
	if (err)
		goto out_close;

	sb->s_fs_info = c;
	sb->s_magic = UBIFS_SUPER_MAGIC;
	sb->s_blocksize = UBIFS_BLOCK_SIZE;
	sb->s_blocksize_bits = UBIFS_BLOCK_SHIFT;
	sb->s_maxbytes = c->max_inode_sz = key_max_inode_size(c);
	if (c->max_inode_sz > MAX_LFS_FILESIZE)
		sb->s_maxbytes = c->max_inode_sz = MAX_LFS_FILESIZE;
	sb->s_op = &ubifs_super_operations;
2131
#ifdef CONFIG_UBIFS_FS_XATTR
2132
	sb->s_xattr = ubifs_xattr_handlers;
2133
#endif
2134
	fscrypt_set_ops(sb, &ubifs_crypt_operations);
2135 2136 2137 2138

	mutex_lock(&c->umount_mutex);
	err = mount_ubifs(c);
	if (err) {
2139
		ubifs_assert(c, err < 0);
2140 2141 2142 2143 2144 2145 2146 2147 2148 2149
		goto out_unlock;
	}

	/* Read the root inode */
	root = ubifs_iget(sb, UBIFS_ROOT_INO);
	if (IS_ERR(root)) {
		err = PTR_ERR(root);
		goto out_umount;
	}

2150
	sb->s_root = d_make_root(root);
W
Wei Yongjun 已提交
2151 2152
	if (!sb->s_root) {
		err = -ENOMEM;
2153
		goto out_umount;
W
Wei Yongjun 已提交
2154
	}
2155 2156 2157 2158 2159 2160 2161 2162 2163 2164

	mutex_unlock(&c->umount_mutex);
	return 0;

out_umount:
	ubifs_umount(c);
out_unlock:
	mutex_unlock(&c->umount_mutex);
out_close:
	ubi_close_volume(c->ubi);
A
Al Viro 已提交
2165
out:
2166 2167 2168 2169 2170
	return err;
}

static int sb_test(struct super_block *sb, void *data)
{
A
Al Viro 已提交
2171
	struct ubifs_info *c1 = data;
A
Artem Bityutskiy 已提交
2172
	struct ubifs_info *c = sb->s_fs_info;
2173

A
Al Viro 已提交
2174 2175 2176 2177 2178 2179 2180
	return c->vi.cdev == c1->vi.cdev;
}

static int sb_set(struct super_block *sb, void *data)
{
	sb->s_fs_info = data;
	return set_anon_super(sb, NULL);
2181 2182
}

A
Al Viro 已提交
2183 2184
static struct dentry *ubifs_mount(struct file_system_type *fs_type, int flags,
			const char *name, void *data)
2185 2186
{
	struct ubi_volume_desc *ubi;
A
Al Viro 已提交
2187
	struct ubifs_info *c;
2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199
	struct super_block *sb;
	int err;

	dbg_gen("name %s, flags %#x", name, flags);

	/*
	 * Get UBI device number and volume ID. Mount it read-only so far
	 * because this might be a new mount point, and UBI allows only one
	 * read-write user at a time.
	 */
	ubi = open_ubi(name, UBI_READONLY);
	if (IS_ERR(ubi)) {
2200
		if (!(flags & SB_SILENT))
2201 2202
			pr_err("UBIFS error (pid: %d): cannot open \"%s\", error %d",
			       current->pid, name, (int)PTR_ERR(ubi));
A
Al Viro 已提交
2203
		return ERR_CAST(ubi);
2204 2205
	}

A
Al Viro 已提交
2206 2207 2208 2209 2210 2211 2212
	c = alloc_ubifs_info(ubi);
	if (!c) {
		err = -ENOMEM;
		goto out_close;
	}

	dbg_gen("opened ubi%d_%d", c->vi.ubi_num, c->vi.vol_id);
2213

D
David Howells 已提交
2214
	sb = sget(fs_type, sb_test, sb_set, flags, c);
2215 2216
	if (IS_ERR(sb)) {
		err = PTR_ERR(sb);
A
Al Viro 已提交
2217
		kfree(c);
2218
		goto out_close;
2219 2220 2221
	}

	if (sb->s_root) {
2222
		struct ubifs_info *c1 = sb->s_fs_info;
A
Al Viro 已提交
2223
		kfree(c);
2224 2225
		/* A new mount point for already mounted UBIFS */
		dbg_gen("this ubi volume is already mounted");
2226
		if (!!(flags & SB_RDONLY) != c1->ro_mount) {
2227 2228 2229 2230
			err = -EBUSY;
			goto out_deact;
		}
	} else {
2231
		err = ubifs_fill_super(sb, data, flags & SB_SILENT ? 1 : 0);
2232 2233 2234
		if (err)
			goto out_deact;
		/* We do not support atime */
2235
		sb->s_flags |= SB_ACTIVE;
2236 2237 2238 2239
		if (IS_ENABLED(CONFIG_UBIFS_ATIME_SUPPORT))
			ubifs_msg(c, "full atime support is enabled.");
		else
			sb->s_flags |= SB_NOATIME;
2240 2241 2242 2243 2244
	}

	/* 'fill_super()' opens ubi again so we must close it here */
	ubi_close_volume(ubi);

A
Al Viro 已提交
2245
	return dget(sb->s_root);
2246 2247

out_deact:
2248
	deactivate_locked_super(sb);
2249 2250
out_close:
	ubi_close_volume(ubi);
A
Al Viro 已提交
2251
	return ERR_PTR(err);
2252 2253
}

A
Al Viro 已提交
2254 2255 2256 2257 2258 2259 2260
static void kill_ubifs_super(struct super_block *s)
{
	struct ubifs_info *c = s->s_fs_info;
	kill_anon_super(s);
	kfree(c);
}

2261 2262 2263
static struct file_system_type ubifs_fs_type = {
	.name    = "ubifs",
	.owner   = THIS_MODULE,
A
Al Viro 已提交
2264
	.mount   = ubifs_mount,
A
Al Viro 已提交
2265
	.kill_sb = kill_ubifs_super,
2266
};
2267
MODULE_ALIAS_FS("ubifs");
2268 2269 2270 2271

/*
 * Inode slab cache constructor.
 */
2272
static void inode_slab_ctor(void *obj)
2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320
{
	struct ubifs_inode *ui = obj;
	inode_init_once(&ui->vfs_inode);
}

static int __init ubifs_init(void)
{
	int err;

	BUILD_BUG_ON(sizeof(struct ubifs_ch) != 24);

	/* Make sure node sizes are 8-byte aligned */
	BUILD_BUG_ON(UBIFS_CH_SZ        & 7);
	BUILD_BUG_ON(UBIFS_INO_NODE_SZ  & 7);
	BUILD_BUG_ON(UBIFS_DENT_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_XENT_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_DATA_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_TRUN_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_SB_NODE_SZ   & 7);
	BUILD_BUG_ON(UBIFS_MST_NODE_SZ  & 7);
	BUILD_BUG_ON(UBIFS_REF_NODE_SZ  & 7);
	BUILD_BUG_ON(UBIFS_CS_NODE_SZ   & 7);
	BUILD_BUG_ON(UBIFS_ORPH_NODE_SZ & 7);

	BUILD_BUG_ON(UBIFS_MAX_DENT_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_MAX_XENT_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_MAX_DATA_NODE_SZ & 7);
	BUILD_BUG_ON(UBIFS_MAX_INO_NODE_SZ  & 7);
	BUILD_BUG_ON(UBIFS_MAX_NODE_SZ      & 7);
	BUILD_BUG_ON(MIN_WRITE_SZ           & 7);

	/* Check min. node size */
	BUILD_BUG_ON(UBIFS_INO_NODE_SZ  < MIN_WRITE_SZ);
	BUILD_BUG_ON(UBIFS_DENT_NODE_SZ < MIN_WRITE_SZ);
	BUILD_BUG_ON(UBIFS_XENT_NODE_SZ < MIN_WRITE_SZ);
	BUILD_BUG_ON(UBIFS_TRUN_NODE_SZ < MIN_WRITE_SZ);

	BUILD_BUG_ON(UBIFS_MAX_DENT_NODE_SZ > UBIFS_MAX_NODE_SZ);
	BUILD_BUG_ON(UBIFS_MAX_XENT_NODE_SZ > UBIFS_MAX_NODE_SZ);
	BUILD_BUG_ON(UBIFS_MAX_DATA_NODE_SZ > UBIFS_MAX_NODE_SZ);
	BUILD_BUG_ON(UBIFS_MAX_INO_NODE_SZ  > UBIFS_MAX_NODE_SZ);

	/* Defined node sizes */
	BUILD_BUG_ON(UBIFS_SB_NODE_SZ  != 4096);
	BUILD_BUG_ON(UBIFS_MST_NODE_SZ != 512);
	BUILD_BUG_ON(UBIFS_INO_NODE_SZ != 160);
	BUILD_BUG_ON(UBIFS_REF_NODE_SZ != 64);

2321 2322 2323
	/*
	 * We use 2 bit wide bit-fields to store compression type, which should
	 * be amended if more compressors are added. The bit-fields are:
2324 2325
	 * @compr_type in 'struct ubifs_inode', @default_compr in
	 * 'struct ubifs_info' and @compr_type in 'struct ubifs_mount_opts'.
2326 2327 2328
	 */
	BUILD_BUG_ON(UBIFS_COMPR_TYPES_CNT > 4);

2329
	/*
2330
	 * We require that PAGE_SIZE is greater-than-or-equal-to
2331 2332
	 * UBIFS_BLOCK_SIZE. It is assumed that both are powers of 2.
	 */
2333
	if (PAGE_SIZE < UBIFS_BLOCK_SIZE) {
2334
		pr_err("UBIFS error (pid %d): VFS page cache size is %u bytes, but UBIFS requires at least 4096 bytes",
2335
		       current->pid, (unsigned int)PAGE_SIZE);
2336 2337 2338 2339 2340
		return -EINVAL;
	}

	ubifs_inode_slab = kmem_cache_create("ubifs_inode_slab",
				sizeof(struct ubifs_inode), 0,
2341 2342
				SLAB_MEM_SPREAD | SLAB_RECLAIM_ACCOUNT |
				SLAB_ACCOUNT, &inode_slab_ctor);
2343
	if (!ubifs_inode_slab)
A
Al Viro 已提交
2344
		return -ENOMEM;
2345

2346 2347 2348
	err = register_shrinker(&ubifs_shrinker_info);
	if (err)
		goto out_slab;
2349 2350

	err = ubifs_compressors_init();
A
Artem Bityutskiy 已提交
2351 2352 2353 2354
	if (err)
		goto out_shrinker;

	err = dbg_debugfs_init();
2355 2356 2357
	if (err)
		goto out_compr;

A
Al Viro 已提交
2358 2359
	err = register_filesystem(&ubifs_fs_type);
	if (err) {
2360 2361
		pr_err("UBIFS error (pid %d): cannot register file system, error %d",
		       current->pid, err);
A
Al Viro 已提交
2362 2363
		goto out_dbg;
	}
2364 2365
	return 0;

A
Al Viro 已提交
2366 2367
out_dbg:
	dbg_debugfs_exit();
2368
out_compr:
A
Artem Bityutskiy 已提交
2369 2370
	ubifs_compressors_exit();
out_shrinker:
2371
	unregister_shrinker(&ubifs_shrinker_info);
2372
out_slab:
2373 2374 2375 2376 2377 2378 2379 2380
	kmem_cache_destroy(ubifs_inode_slab);
	return err;
}
/* late_initcall to let compressors initialize first */
late_initcall(ubifs_init);

static void __exit ubifs_exit(void)
{
2381 2382
	WARN_ON(!list_empty(&ubifs_infos));
	WARN_ON(atomic_long_read(&ubifs_clean_zn_cnt) != 0);
2383

A
Artem Bityutskiy 已提交
2384
	dbg_debugfs_exit();
2385 2386
	ubifs_compressors_exit();
	unregister_shrinker(&ubifs_shrinker_info);
2387 2388 2389 2390 2391 2392

	/*
	 * Make sure all delayed rcu free inodes are flushed before we
	 * destroy cache.
	 */
	rcu_barrier();
2393 2394 2395 2396 2397 2398 2399 2400 2401
	kmem_cache_destroy(ubifs_inode_slab);
	unregister_filesystem(&ubifs_fs_type);
}
module_exit(ubifs_exit);

MODULE_LICENSE("GPL");
MODULE_VERSION(__stringify(UBIFS_VERSION));
MODULE_AUTHOR("Artem Bityutskiy, Adrian Hunter");
MODULE_DESCRIPTION("UBIFS - UBI File System");