xref: /kernel/linux/linux-6.6/fs/erofs/super.c (revision 62306a36)
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * Copyright (C) 2017-2018 HUAWEI, Inc.
4 *             https://www.huawei.com/
5 * Copyright (C) 2021, Alibaba Cloud
6 */
7#include <linux/module.h>
8#include <linux/statfs.h>
9#include <linux/parser.h>
10#include <linux/seq_file.h>
11#include <linux/crc32c.h>
12#include <linux/fs_context.h>
13#include <linux/fs_parser.h>
14#include <linux/dax.h>
15#include <linux/exportfs.h>
16#include "xattr.h"
17
18#define CREATE_TRACE_POINTS
19#include <trace/events/erofs.h>
20
21static struct kmem_cache *erofs_inode_cachep __read_mostly;
22
23void _erofs_err(struct super_block *sb, const char *func, const char *fmt, ...)
24{
25	struct va_format vaf;
26	va_list args;
27
28	va_start(args, fmt);
29
30	vaf.fmt = fmt;
31	vaf.va = &args;
32
33	pr_err("(device %s): %s: %pV", sb->s_id, func, &vaf);
34	va_end(args);
35}
36
37void _erofs_info(struct super_block *sb, const char *func, const char *fmt, ...)
38{
39	struct va_format vaf;
40	va_list args;
41
42	va_start(args, fmt);
43
44	vaf.fmt = fmt;
45	vaf.va = &args;
46
47	pr_info("(device %s): %pV", sb->s_id, &vaf);
48	va_end(args);
49}
50
51static int erofs_superblock_csum_verify(struct super_block *sb, void *sbdata)
52{
53	size_t len = 1 << EROFS_SB(sb)->blkszbits;
54	struct erofs_super_block *dsb;
55	u32 expected_crc, crc;
56
57	if (len > EROFS_SUPER_OFFSET)
58		len -= EROFS_SUPER_OFFSET;
59
60	dsb = kmemdup(sbdata + EROFS_SUPER_OFFSET, len, GFP_KERNEL);
61	if (!dsb)
62		return -ENOMEM;
63
64	expected_crc = le32_to_cpu(dsb->checksum);
65	dsb->checksum = 0;
66	/* to allow for x86 boot sectors and other oddities. */
67	crc = crc32c(~0, dsb, len);
68	kfree(dsb);
69
70	if (crc != expected_crc) {
71		erofs_err(sb, "invalid checksum 0x%08x, 0x%08x expected",
72			  crc, expected_crc);
73		return -EBADMSG;
74	}
75	return 0;
76}
77
78static void erofs_inode_init_once(void *ptr)
79{
80	struct erofs_inode *vi = ptr;
81
82	inode_init_once(&vi->vfs_inode);
83}
84
85static struct inode *erofs_alloc_inode(struct super_block *sb)
86{
87	struct erofs_inode *vi =
88		alloc_inode_sb(sb, erofs_inode_cachep, GFP_KERNEL);
89
90	if (!vi)
91		return NULL;
92
93	/* zero out everything except vfs_inode */
94	memset(vi, 0, offsetof(struct erofs_inode, vfs_inode));
95	return &vi->vfs_inode;
96}
97
98static void erofs_free_inode(struct inode *inode)
99{
100	struct erofs_inode *vi = EROFS_I(inode);
101
102	if (inode->i_op == &erofs_fast_symlink_iops)
103		kfree(inode->i_link);
104	kfree(vi->xattr_shared_xattrs);
105	kmem_cache_free(erofs_inode_cachep, vi);
106}
107
108static bool check_layout_compatibility(struct super_block *sb,
109				       struct erofs_super_block *dsb)
110{
111	const unsigned int feature = le32_to_cpu(dsb->feature_incompat);
112
113	EROFS_SB(sb)->feature_incompat = feature;
114
115	/* check if current kernel meets all mandatory requirements */
116	if (feature & (~EROFS_ALL_FEATURE_INCOMPAT)) {
117		erofs_err(sb, "unidentified incompatible feature %x, please upgrade kernel",
118			   feature & ~EROFS_ALL_FEATURE_INCOMPAT);
119		return false;
120	}
121	return true;
122}
123
124/* read variable-sized metadata, offset will be aligned by 4-byte */
125void *erofs_read_metadata(struct super_block *sb, struct erofs_buf *buf,
126			  erofs_off_t *offset, int *lengthp)
127{
128	u8 *buffer, *ptr;
129	int len, i, cnt;
130
131	*offset = round_up(*offset, 4);
132	ptr = erofs_bread(buf, erofs_blknr(sb, *offset), EROFS_KMAP);
133	if (IS_ERR(ptr))
134		return ptr;
135
136	len = le16_to_cpu(*(__le16 *)&ptr[erofs_blkoff(sb, *offset)]);
137	if (!len)
138		len = U16_MAX + 1;
139	buffer = kmalloc(len, GFP_KERNEL);
140	if (!buffer)
141		return ERR_PTR(-ENOMEM);
142	*offset += sizeof(__le16);
143	*lengthp = len;
144
145	for (i = 0; i < len; i += cnt) {
146		cnt = min_t(int, sb->s_blocksize - erofs_blkoff(sb, *offset),
147			    len - i);
148		ptr = erofs_bread(buf, erofs_blknr(sb, *offset), EROFS_KMAP);
149		if (IS_ERR(ptr)) {
150			kfree(buffer);
151			return ptr;
152		}
153		memcpy(buffer + i, ptr + erofs_blkoff(sb, *offset), cnt);
154		*offset += cnt;
155	}
156	return buffer;
157}
158
159#ifndef CONFIG_EROFS_FS_ZIP
160static int z_erofs_parse_cfgs(struct super_block *sb,
161			      struct erofs_super_block *dsb)
162{
163	if (!dsb->u1.available_compr_algs)
164		return 0;
165
166	erofs_err(sb, "compression disabled, unable to mount compressed EROFS");
167	return -EOPNOTSUPP;
168}
169#endif
170
171static int erofs_init_device(struct erofs_buf *buf, struct super_block *sb,
172			     struct erofs_device_info *dif, erofs_off_t *pos)
173{
174	struct erofs_sb_info *sbi = EROFS_SB(sb);
175	struct erofs_fscache *fscache;
176	struct erofs_deviceslot *dis;
177	struct bdev_handle *bdev_handle;
178	void *ptr;
179
180	ptr = erofs_read_metabuf(buf, sb, erofs_blknr(sb, *pos), EROFS_KMAP);
181	if (IS_ERR(ptr))
182		return PTR_ERR(ptr);
183	dis = ptr + erofs_blkoff(sb, *pos);
184
185	if (!sbi->devs->flatdev && !dif->path) {
186		if (!dis->tag[0]) {
187			erofs_err(sb, "empty device tag @ pos %llu", *pos);
188			return -EINVAL;
189		}
190		dif->path = kmemdup_nul(dis->tag, sizeof(dis->tag), GFP_KERNEL);
191		if (!dif->path)
192			return -ENOMEM;
193	}
194
195	if (erofs_is_fscache_mode(sb)) {
196		fscache = erofs_fscache_register_cookie(sb, dif->path, 0);
197		if (IS_ERR(fscache))
198			return PTR_ERR(fscache);
199		dif->fscache = fscache;
200	} else if (!sbi->devs->flatdev) {
201		bdev_handle = bdev_open_by_path(dif->path, BLK_OPEN_READ,
202						sb->s_type, NULL);
203		if (IS_ERR(bdev_handle))
204			return PTR_ERR(bdev_handle);
205		dif->bdev_handle = bdev_handle;
206		dif->dax_dev = fs_dax_get_by_bdev(bdev_handle->bdev,
207				&dif->dax_part_off, NULL, NULL);
208	}
209
210	dif->blocks = le32_to_cpu(dis->blocks);
211	dif->mapped_blkaddr = le32_to_cpu(dis->mapped_blkaddr);
212	sbi->total_blocks += dif->blocks;
213	*pos += EROFS_DEVT_SLOT_SIZE;
214	return 0;
215}
216
217static int erofs_scan_devices(struct super_block *sb,
218			      struct erofs_super_block *dsb)
219{
220	struct erofs_sb_info *sbi = EROFS_SB(sb);
221	unsigned int ondisk_extradevs;
222	erofs_off_t pos;
223	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
224	struct erofs_device_info *dif;
225	int id, err = 0;
226
227	sbi->total_blocks = sbi->primarydevice_blocks;
228	if (!erofs_sb_has_device_table(sbi))
229		ondisk_extradevs = 0;
230	else
231		ondisk_extradevs = le16_to_cpu(dsb->extra_devices);
232
233	if (sbi->devs->extra_devices &&
234	    ondisk_extradevs != sbi->devs->extra_devices) {
235		erofs_err(sb, "extra devices don't match (ondisk %u, given %u)",
236			  ondisk_extradevs, sbi->devs->extra_devices);
237		return -EINVAL;
238	}
239	if (!ondisk_extradevs)
240		return 0;
241
242	if (!sbi->devs->extra_devices && !erofs_is_fscache_mode(sb))
243		sbi->devs->flatdev = true;
244
245	sbi->device_id_mask = roundup_pow_of_two(ondisk_extradevs + 1) - 1;
246	pos = le16_to_cpu(dsb->devt_slotoff) * EROFS_DEVT_SLOT_SIZE;
247	down_read(&sbi->devs->rwsem);
248	if (sbi->devs->extra_devices) {
249		idr_for_each_entry(&sbi->devs->tree, dif, id) {
250			err = erofs_init_device(&buf, sb, dif, &pos);
251			if (err)
252				break;
253		}
254	} else {
255		for (id = 0; id < ondisk_extradevs; id++) {
256			dif = kzalloc(sizeof(*dif), GFP_KERNEL);
257			if (!dif) {
258				err = -ENOMEM;
259				break;
260			}
261
262			err = idr_alloc(&sbi->devs->tree, dif, 0, 0, GFP_KERNEL);
263			if (err < 0) {
264				kfree(dif);
265				break;
266			}
267			++sbi->devs->extra_devices;
268
269			err = erofs_init_device(&buf, sb, dif, &pos);
270			if (err)
271				break;
272		}
273	}
274	up_read(&sbi->devs->rwsem);
275	erofs_put_metabuf(&buf);
276	return err;
277}
278
279static int erofs_read_superblock(struct super_block *sb)
280{
281	struct erofs_sb_info *sbi;
282	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
283	struct erofs_super_block *dsb;
284	void *data;
285	int ret;
286
287	data = erofs_read_metabuf(&buf, sb, 0, EROFS_KMAP);
288	if (IS_ERR(data)) {
289		erofs_err(sb, "cannot read erofs superblock");
290		return PTR_ERR(data);
291	}
292
293	sbi = EROFS_SB(sb);
294	dsb = (struct erofs_super_block *)(data + EROFS_SUPER_OFFSET);
295
296	ret = -EINVAL;
297	if (le32_to_cpu(dsb->magic) != EROFS_SUPER_MAGIC_V1) {
298		erofs_err(sb, "cannot find valid erofs superblock");
299		goto out;
300	}
301
302	sbi->blkszbits  = dsb->blkszbits;
303	if (sbi->blkszbits < 9 || sbi->blkszbits > PAGE_SHIFT) {
304		erofs_err(sb, "blkszbits %u isn't supported", sbi->blkszbits);
305		goto out;
306	}
307	if (dsb->dirblkbits) {
308		erofs_err(sb, "dirblkbits %u isn't supported", dsb->dirblkbits);
309		goto out;
310	}
311
312	sbi->feature_compat = le32_to_cpu(dsb->feature_compat);
313	if (erofs_sb_has_sb_chksum(sbi)) {
314		ret = erofs_superblock_csum_verify(sb, data);
315		if (ret)
316			goto out;
317	}
318
319	ret = -EINVAL;
320	if (!check_layout_compatibility(sb, dsb))
321		goto out;
322
323	sbi->sb_size = 128 + dsb->sb_extslots * EROFS_SB_EXTSLOT_SIZE;
324	if (sbi->sb_size > PAGE_SIZE - EROFS_SUPER_OFFSET) {
325		erofs_err(sb, "invalid sb_extslots %u (more than a fs block)",
326			  sbi->sb_size);
327		goto out;
328	}
329	sbi->primarydevice_blocks = le32_to_cpu(dsb->blocks);
330	sbi->meta_blkaddr = le32_to_cpu(dsb->meta_blkaddr);
331#ifdef CONFIG_EROFS_FS_XATTR
332	sbi->xattr_blkaddr = le32_to_cpu(dsb->xattr_blkaddr);
333	sbi->xattr_prefix_start = le32_to_cpu(dsb->xattr_prefix_start);
334	sbi->xattr_prefix_count = dsb->xattr_prefix_count;
335	sbi->xattr_filter_reserved = dsb->xattr_filter_reserved;
336#endif
337	sbi->islotbits = ilog2(sizeof(struct erofs_inode_compact));
338	sbi->root_nid = le16_to_cpu(dsb->root_nid);
339	sbi->packed_nid = le64_to_cpu(dsb->packed_nid);
340	sbi->inos = le64_to_cpu(dsb->inos);
341
342	sbi->build_time = le64_to_cpu(dsb->build_time);
343	sbi->build_time_nsec = le32_to_cpu(dsb->build_time_nsec);
344
345	memcpy(&sb->s_uuid, dsb->uuid, sizeof(dsb->uuid));
346
347	ret = strscpy(sbi->volume_name, dsb->volume_name,
348		      sizeof(dsb->volume_name));
349	if (ret < 0) {	/* -E2BIG */
350		erofs_err(sb, "bad volume name without NIL terminator");
351		ret = -EFSCORRUPTED;
352		goto out;
353	}
354
355	/* parse on-disk compression configurations */
356	ret = z_erofs_parse_cfgs(sb, dsb);
357	if (ret < 0)
358		goto out;
359
360	/* handle multiple devices */
361	ret = erofs_scan_devices(sb, dsb);
362
363	if (erofs_is_fscache_mode(sb))
364		erofs_info(sb, "EXPERIMENTAL fscache-based on-demand read feature in use. Use at your own risk!");
365out:
366	erofs_put_metabuf(&buf);
367	return ret;
368}
369
370static void erofs_default_options(struct erofs_fs_context *ctx)
371{
372#ifdef CONFIG_EROFS_FS_ZIP
373	ctx->opt.cache_strategy = EROFS_ZIP_CACHE_READAROUND;
374	ctx->opt.max_sync_decompress_pages = 3;
375	ctx->opt.sync_decompress = EROFS_SYNC_DECOMPRESS_AUTO;
376#endif
377#ifdef CONFIG_EROFS_FS_XATTR
378	set_opt(&ctx->opt, XATTR_USER);
379#endif
380#ifdef CONFIG_EROFS_FS_POSIX_ACL
381	set_opt(&ctx->opt, POSIX_ACL);
382#endif
383}
384
385enum {
386	Opt_user_xattr,
387	Opt_acl,
388	Opt_cache_strategy,
389	Opt_dax,
390	Opt_dax_enum,
391	Opt_device,
392	Opt_fsid,
393	Opt_domain_id,
394	Opt_err
395};
396
397static const struct constant_table erofs_param_cache_strategy[] = {
398	{"disabled",	EROFS_ZIP_CACHE_DISABLED},
399	{"readahead",	EROFS_ZIP_CACHE_READAHEAD},
400	{"readaround",	EROFS_ZIP_CACHE_READAROUND},
401	{}
402};
403
404static const struct constant_table erofs_dax_param_enums[] = {
405	{"always",	EROFS_MOUNT_DAX_ALWAYS},
406	{"never",	EROFS_MOUNT_DAX_NEVER},
407	{}
408};
409
410static const struct fs_parameter_spec erofs_fs_parameters[] = {
411	fsparam_flag_no("user_xattr",	Opt_user_xattr),
412	fsparam_flag_no("acl",		Opt_acl),
413	fsparam_enum("cache_strategy",	Opt_cache_strategy,
414		     erofs_param_cache_strategy),
415	fsparam_flag("dax",             Opt_dax),
416	fsparam_enum("dax",		Opt_dax_enum, erofs_dax_param_enums),
417	fsparam_string("device",	Opt_device),
418	fsparam_string("fsid",		Opt_fsid),
419	fsparam_string("domain_id",	Opt_domain_id),
420	{}
421};
422
423static bool erofs_fc_set_dax_mode(struct fs_context *fc, unsigned int mode)
424{
425#ifdef CONFIG_FS_DAX
426	struct erofs_fs_context *ctx = fc->fs_private;
427
428	switch (mode) {
429	case EROFS_MOUNT_DAX_ALWAYS:
430		warnfc(fc, "DAX enabled. Warning: EXPERIMENTAL, use at your own risk");
431		set_opt(&ctx->opt, DAX_ALWAYS);
432		clear_opt(&ctx->opt, DAX_NEVER);
433		return true;
434	case EROFS_MOUNT_DAX_NEVER:
435		set_opt(&ctx->opt, DAX_NEVER);
436		clear_opt(&ctx->opt, DAX_ALWAYS);
437		return true;
438	default:
439		DBG_BUGON(1);
440		return false;
441	}
442#else
443	errorfc(fc, "dax options not supported");
444	return false;
445#endif
446}
447
448static int erofs_fc_parse_param(struct fs_context *fc,
449				struct fs_parameter *param)
450{
451	struct erofs_fs_context *ctx = fc->fs_private;
452	struct fs_parse_result result;
453	struct erofs_device_info *dif;
454	int opt, ret;
455
456	opt = fs_parse(fc, erofs_fs_parameters, param, &result);
457	if (opt < 0)
458		return opt;
459
460	switch (opt) {
461	case Opt_user_xattr:
462#ifdef CONFIG_EROFS_FS_XATTR
463		if (result.boolean)
464			set_opt(&ctx->opt, XATTR_USER);
465		else
466			clear_opt(&ctx->opt, XATTR_USER);
467#else
468		errorfc(fc, "{,no}user_xattr options not supported");
469#endif
470		break;
471	case Opt_acl:
472#ifdef CONFIG_EROFS_FS_POSIX_ACL
473		if (result.boolean)
474			set_opt(&ctx->opt, POSIX_ACL);
475		else
476			clear_opt(&ctx->opt, POSIX_ACL);
477#else
478		errorfc(fc, "{,no}acl options not supported");
479#endif
480		break;
481	case Opt_cache_strategy:
482#ifdef CONFIG_EROFS_FS_ZIP
483		ctx->opt.cache_strategy = result.uint_32;
484#else
485		errorfc(fc, "compression not supported, cache_strategy ignored");
486#endif
487		break;
488	case Opt_dax:
489		if (!erofs_fc_set_dax_mode(fc, EROFS_MOUNT_DAX_ALWAYS))
490			return -EINVAL;
491		break;
492	case Opt_dax_enum:
493		if (!erofs_fc_set_dax_mode(fc, result.uint_32))
494			return -EINVAL;
495		break;
496	case Opt_device:
497		dif = kzalloc(sizeof(*dif), GFP_KERNEL);
498		if (!dif)
499			return -ENOMEM;
500		dif->path = kstrdup(param->string, GFP_KERNEL);
501		if (!dif->path) {
502			kfree(dif);
503			return -ENOMEM;
504		}
505		down_write(&ctx->devs->rwsem);
506		ret = idr_alloc(&ctx->devs->tree, dif, 0, 0, GFP_KERNEL);
507		up_write(&ctx->devs->rwsem);
508		if (ret < 0) {
509			kfree(dif->path);
510			kfree(dif);
511			return ret;
512		}
513		++ctx->devs->extra_devices;
514		break;
515#ifdef CONFIG_EROFS_FS_ONDEMAND
516	case Opt_fsid:
517		kfree(ctx->fsid);
518		ctx->fsid = kstrdup(param->string, GFP_KERNEL);
519		if (!ctx->fsid)
520			return -ENOMEM;
521		break;
522	case Opt_domain_id:
523		kfree(ctx->domain_id);
524		ctx->domain_id = kstrdup(param->string, GFP_KERNEL);
525		if (!ctx->domain_id)
526			return -ENOMEM;
527		break;
528#else
529	case Opt_fsid:
530	case Opt_domain_id:
531		errorfc(fc, "%s option not supported", erofs_fs_parameters[opt].name);
532		break;
533#endif
534	default:
535		return -ENOPARAM;
536	}
537	return 0;
538}
539
540static struct inode *erofs_nfs_get_inode(struct super_block *sb,
541					 u64 ino, u32 generation)
542{
543	return erofs_iget(sb, ino);
544}
545
546static struct dentry *erofs_fh_to_dentry(struct super_block *sb,
547		struct fid *fid, int fh_len, int fh_type)
548{
549	return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
550				    erofs_nfs_get_inode);
551}
552
553static struct dentry *erofs_fh_to_parent(struct super_block *sb,
554		struct fid *fid, int fh_len, int fh_type)
555{
556	return generic_fh_to_parent(sb, fid, fh_len, fh_type,
557				    erofs_nfs_get_inode);
558}
559
560static struct dentry *erofs_get_parent(struct dentry *child)
561{
562	erofs_nid_t nid;
563	unsigned int d_type;
564	int err;
565
566	err = erofs_namei(d_inode(child), &dotdot_name, &nid, &d_type);
567	if (err)
568		return ERR_PTR(err);
569	return d_obtain_alias(erofs_iget(child->d_sb, nid));
570}
571
572static const struct export_operations erofs_export_ops = {
573	.fh_to_dentry = erofs_fh_to_dentry,
574	.fh_to_parent = erofs_fh_to_parent,
575	.get_parent = erofs_get_parent,
576};
577
578static int erofs_fc_fill_super(struct super_block *sb, struct fs_context *fc)
579{
580	struct inode *inode;
581	struct erofs_sb_info *sbi;
582	struct erofs_fs_context *ctx = fc->fs_private;
583	int err;
584
585	sb->s_magic = EROFS_SUPER_MAGIC;
586	sb->s_flags |= SB_RDONLY | SB_NOATIME;
587	sb->s_maxbytes = MAX_LFS_FILESIZE;
588	sb->s_op = &erofs_sops;
589
590	sbi = kzalloc(sizeof(*sbi), GFP_KERNEL);
591	if (!sbi)
592		return -ENOMEM;
593
594	sb->s_fs_info = sbi;
595	sbi->opt = ctx->opt;
596	sbi->devs = ctx->devs;
597	ctx->devs = NULL;
598	sbi->fsid = ctx->fsid;
599	ctx->fsid = NULL;
600	sbi->domain_id = ctx->domain_id;
601	ctx->domain_id = NULL;
602
603	sbi->blkszbits = PAGE_SHIFT;
604	if (erofs_is_fscache_mode(sb)) {
605		sb->s_blocksize = PAGE_SIZE;
606		sb->s_blocksize_bits = PAGE_SHIFT;
607
608		err = erofs_fscache_register_fs(sb);
609		if (err)
610			return err;
611
612		err = super_setup_bdi(sb);
613		if (err)
614			return err;
615	} else {
616		if (!sb_set_blocksize(sb, PAGE_SIZE)) {
617			errorfc(fc, "failed to set initial blksize");
618			return -EINVAL;
619		}
620
621		sbi->dax_dev = fs_dax_get_by_bdev(sb->s_bdev,
622						  &sbi->dax_part_off,
623						  NULL, NULL);
624	}
625
626	err = erofs_read_superblock(sb);
627	if (err)
628		return err;
629
630	if (sb->s_blocksize_bits != sbi->blkszbits) {
631		if (erofs_is_fscache_mode(sb)) {
632			errorfc(fc, "unsupported blksize for fscache mode");
633			return -EINVAL;
634		}
635		if (!sb_set_blocksize(sb, 1 << sbi->blkszbits)) {
636			errorfc(fc, "failed to set erofs blksize");
637			return -EINVAL;
638		}
639	}
640
641	if (test_opt(&sbi->opt, DAX_ALWAYS)) {
642		if (!sbi->dax_dev) {
643			errorfc(fc, "DAX unsupported by block device. Turning off DAX.");
644			clear_opt(&sbi->opt, DAX_ALWAYS);
645		} else if (sbi->blkszbits != PAGE_SHIFT) {
646			errorfc(fc, "unsupported blocksize for DAX");
647			clear_opt(&sbi->opt, DAX_ALWAYS);
648		}
649	}
650
651	sb->s_time_gran = 1;
652	sb->s_xattr = erofs_xattr_handlers;
653	sb->s_export_op = &erofs_export_ops;
654
655	if (test_opt(&sbi->opt, POSIX_ACL))
656		sb->s_flags |= SB_POSIXACL;
657	else
658		sb->s_flags &= ~SB_POSIXACL;
659
660#ifdef CONFIG_EROFS_FS_ZIP
661	xa_init(&sbi->managed_pslots);
662#endif
663
664	inode = erofs_iget(sb, ROOT_NID(sbi));
665	if (IS_ERR(inode))
666		return PTR_ERR(inode);
667
668	if (!S_ISDIR(inode->i_mode)) {
669		erofs_err(sb, "rootino(nid %llu) is not a directory(i_mode %o)",
670			  ROOT_NID(sbi), inode->i_mode);
671		iput(inode);
672		return -EINVAL;
673	}
674
675	sb->s_root = d_make_root(inode);
676	if (!sb->s_root)
677		return -ENOMEM;
678
679	erofs_shrinker_register(sb);
680	if (erofs_sb_has_fragments(sbi) && sbi->packed_nid) {
681		sbi->packed_inode = erofs_iget(sb, sbi->packed_nid);
682		if (IS_ERR(sbi->packed_inode)) {
683			err = PTR_ERR(sbi->packed_inode);
684			sbi->packed_inode = NULL;
685			return err;
686		}
687	}
688	err = erofs_init_managed_cache(sb);
689	if (err)
690		return err;
691
692	err = erofs_xattr_prefixes_init(sb);
693	if (err)
694		return err;
695
696	err = erofs_register_sysfs(sb);
697	if (err)
698		return err;
699
700	erofs_info(sb, "mounted with root inode @ nid %llu.", ROOT_NID(sbi));
701	return 0;
702}
703
704static int erofs_fc_get_tree(struct fs_context *fc)
705{
706	struct erofs_fs_context *ctx = fc->fs_private;
707
708	if (IS_ENABLED(CONFIG_EROFS_FS_ONDEMAND) && ctx->fsid)
709		return get_tree_nodev(fc, erofs_fc_fill_super);
710
711	return get_tree_bdev(fc, erofs_fc_fill_super);
712}
713
714static int erofs_fc_reconfigure(struct fs_context *fc)
715{
716	struct super_block *sb = fc->root->d_sb;
717	struct erofs_sb_info *sbi = EROFS_SB(sb);
718	struct erofs_fs_context *ctx = fc->fs_private;
719
720	DBG_BUGON(!sb_rdonly(sb));
721
722	if (ctx->fsid || ctx->domain_id)
723		erofs_info(sb, "ignoring reconfiguration for fsid|domain_id.");
724
725	if (test_opt(&ctx->opt, POSIX_ACL))
726		fc->sb_flags |= SB_POSIXACL;
727	else
728		fc->sb_flags &= ~SB_POSIXACL;
729
730	sbi->opt = ctx->opt;
731
732	fc->sb_flags |= SB_RDONLY;
733	return 0;
734}
735
736static int erofs_release_device_info(int id, void *ptr, void *data)
737{
738	struct erofs_device_info *dif = ptr;
739
740	fs_put_dax(dif->dax_dev, NULL);
741	if (dif->bdev_handle)
742		bdev_release(dif->bdev_handle);
743	erofs_fscache_unregister_cookie(dif->fscache);
744	dif->fscache = NULL;
745	kfree(dif->path);
746	kfree(dif);
747	return 0;
748}
749
750static void erofs_free_dev_context(struct erofs_dev_context *devs)
751{
752	if (!devs)
753		return;
754	idr_for_each(&devs->tree, &erofs_release_device_info, NULL);
755	idr_destroy(&devs->tree);
756	kfree(devs);
757}
758
759static void erofs_fc_free(struct fs_context *fc)
760{
761	struct erofs_fs_context *ctx = fc->fs_private;
762
763	erofs_free_dev_context(ctx->devs);
764	kfree(ctx->fsid);
765	kfree(ctx->domain_id);
766	kfree(ctx);
767}
768
769static const struct fs_context_operations erofs_context_ops = {
770	.parse_param	= erofs_fc_parse_param,
771	.get_tree       = erofs_fc_get_tree,
772	.reconfigure    = erofs_fc_reconfigure,
773	.free		= erofs_fc_free,
774};
775
776static int erofs_init_fs_context(struct fs_context *fc)
777{
778	struct erofs_fs_context *ctx;
779
780	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
781	if (!ctx)
782		return -ENOMEM;
783	ctx->devs = kzalloc(sizeof(struct erofs_dev_context), GFP_KERNEL);
784	if (!ctx->devs) {
785		kfree(ctx);
786		return -ENOMEM;
787	}
788	fc->fs_private = ctx;
789
790	idr_init(&ctx->devs->tree);
791	init_rwsem(&ctx->devs->rwsem);
792	erofs_default_options(ctx);
793	fc->ops = &erofs_context_ops;
794	return 0;
795}
796
797static void erofs_kill_sb(struct super_block *sb)
798{
799	struct erofs_sb_info *sbi;
800
801	if (erofs_is_fscache_mode(sb))
802		kill_anon_super(sb);
803	else
804		kill_block_super(sb);
805
806	sbi = EROFS_SB(sb);
807	if (!sbi)
808		return;
809
810	erofs_free_dev_context(sbi->devs);
811	fs_put_dax(sbi->dax_dev, NULL);
812	erofs_fscache_unregister_fs(sb);
813	kfree(sbi->fsid);
814	kfree(sbi->domain_id);
815	kfree(sbi);
816	sb->s_fs_info = NULL;
817}
818
819static void erofs_put_super(struct super_block *sb)
820{
821	struct erofs_sb_info *const sbi = EROFS_SB(sb);
822
823	DBG_BUGON(!sbi);
824
825	erofs_unregister_sysfs(sb);
826	erofs_shrinker_unregister(sb);
827	erofs_xattr_prefixes_cleanup(sb);
828#ifdef CONFIG_EROFS_FS_ZIP
829	iput(sbi->managed_cache);
830	sbi->managed_cache = NULL;
831#endif
832	iput(sbi->packed_inode);
833	sbi->packed_inode = NULL;
834	erofs_free_dev_context(sbi->devs);
835	sbi->devs = NULL;
836	erofs_fscache_unregister_fs(sb);
837}
838
839static struct file_system_type erofs_fs_type = {
840	.owner          = THIS_MODULE,
841	.name           = "erofs",
842	.init_fs_context = erofs_init_fs_context,
843	.kill_sb        = erofs_kill_sb,
844	.fs_flags       = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
845};
846MODULE_ALIAS_FS("erofs");
847
848static int __init erofs_module_init(void)
849{
850	int err;
851
852	erofs_check_ondisk_layout_definitions();
853
854	erofs_inode_cachep = kmem_cache_create("erofs_inode",
855			sizeof(struct erofs_inode), 0,
856			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD | SLAB_ACCOUNT,
857			erofs_inode_init_once);
858	if (!erofs_inode_cachep)
859		return -ENOMEM;
860
861	err = erofs_init_shrinker();
862	if (err)
863		goto shrinker_err;
864
865	err = z_erofs_lzma_init();
866	if (err)
867		goto lzma_err;
868
869	err = z_erofs_deflate_init();
870	if (err)
871		goto deflate_err;
872
873	erofs_pcpubuf_init();
874	err = z_erofs_init_zip_subsystem();
875	if (err)
876		goto zip_err;
877
878	err = erofs_init_sysfs();
879	if (err)
880		goto sysfs_err;
881
882	err = register_filesystem(&erofs_fs_type);
883	if (err)
884		goto fs_err;
885
886	return 0;
887
888fs_err:
889	erofs_exit_sysfs();
890sysfs_err:
891	z_erofs_exit_zip_subsystem();
892zip_err:
893	z_erofs_deflate_exit();
894deflate_err:
895	z_erofs_lzma_exit();
896lzma_err:
897	erofs_exit_shrinker();
898shrinker_err:
899	kmem_cache_destroy(erofs_inode_cachep);
900	return err;
901}
902
903static void __exit erofs_module_exit(void)
904{
905	unregister_filesystem(&erofs_fs_type);
906
907	/* Ensure all RCU free inodes / pclusters are safe to be destroyed. */
908	rcu_barrier();
909
910	erofs_exit_sysfs();
911	z_erofs_exit_zip_subsystem();
912	z_erofs_deflate_exit();
913	z_erofs_lzma_exit();
914	erofs_exit_shrinker();
915	kmem_cache_destroy(erofs_inode_cachep);
916	erofs_pcpubuf_exit();
917}
918
919static int erofs_statfs(struct dentry *dentry, struct kstatfs *buf)
920{
921	struct super_block *sb = dentry->d_sb;
922	struct erofs_sb_info *sbi = EROFS_SB(sb);
923	u64 id = 0;
924
925	if (!erofs_is_fscache_mode(sb))
926		id = huge_encode_dev(sb->s_bdev->bd_dev);
927
928	buf->f_type = sb->s_magic;
929	buf->f_bsize = sb->s_blocksize;
930	buf->f_blocks = sbi->total_blocks;
931	buf->f_bfree = buf->f_bavail = 0;
932
933	buf->f_files = ULLONG_MAX;
934	buf->f_ffree = ULLONG_MAX - sbi->inos;
935
936	buf->f_namelen = EROFS_NAME_LEN;
937
938	buf->f_fsid    = u64_to_fsid(id);
939	return 0;
940}
941
942static int erofs_show_options(struct seq_file *seq, struct dentry *root)
943{
944	struct erofs_sb_info *sbi = EROFS_SB(root->d_sb);
945	struct erofs_mount_opts *opt = &sbi->opt;
946
947#ifdef CONFIG_EROFS_FS_XATTR
948	if (test_opt(opt, XATTR_USER))
949		seq_puts(seq, ",user_xattr");
950	else
951		seq_puts(seq, ",nouser_xattr");
952#endif
953#ifdef CONFIG_EROFS_FS_POSIX_ACL
954	if (test_opt(opt, POSIX_ACL))
955		seq_puts(seq, ",acl");
956	else
957		seq_puts(seq, ",noacl");
958#endif
959#ifdef CONFIG_EROFS_FS_ZIP
960	if (opt->cache_strategy == EROFS_ZIP_CACHE_DISABLED)
961		seq_puts(seq, ",cache_strategy=disabled");
962	else if (opt->cache_strategy == EROFS_ZIP_CACHE_READAHEAD)
963		seq_puts(seq, ",cache_strategy=readahead");
964	else if (opt->cache_strategy == EROFS_ZIP_CACHE_READAROUND)
965		seq_puts(seq, ",cache_strategy=readaround");
966#endif
967	if (test_opt(opt, DAX_ALWAYS))
968		seq_puts(seq, ",dax=always");
969	if (test_opt(opt, DAX_NEVER))
970		seq_puts(seq, ",dax=never");
971#ifdef CONFIG_EROFS_FS_ONDEMAND
972	if (sbi->fsid)
973		seq_printf(seq, ",fsid=%s", sbi->fsid);
974	if (sbi->domain_id)
975		seq_printf(seq, ",domain_id=%s", sbi->domain_id);
976#endif
977	return 0;
978}
979
980const struct super_operations erofs_sops = {
981	.put_super = erofs_put_super,
982	.alloc_inode = erofs_alloc_inode,
983	.free_inode = erofs_free_inode,
984	.statfs = erofs_statfs,
985	.show_options = erofs_show_options,
986};
987
988module_init(erofs_module_init);
989module_exit(erofs_module_exit);
990
991MODULE_DESCRIPTION("Enhanced ROM File System");
992MODULE_AUTHOR("Gao Xiang, Chao Yu, Miao Xie, CONSUMER BG, HUAWEI Inc.");
993MODULE_LICENSE("GPL");
994