1 // SPDX-License-Identifier: GPL-2.0
3 * Copyright (C) STRATO AG 2011. All rights reserved.
7 * This module can be used to catch cases when the btrfs kernel
8 * code executes write requests to the disk that bring the file
9 * system in an inconsistent state. In such a state, a power-loss
10 * or kernel panic event would cause that the data on disk is
11 * lost or at least damaged.
13 * Code is added that examines all block write requests during
14 * runtime (including writes of the super block). Three rules
15 * are verified and an error is printed on violation of the
17 * 1. It is not allowed to write a disk block which is
18 * currently referenced by the super block (either directly
20 * 2. When a super block is written, it is verified that all
21 * referenced (directly or indirectly) blocks fulfill the
22 * following requirements:
23 * 2a. All referenced blocks have either been present when
24 * the file system was mounted, (i.e., they have been
25 * referenced by the super block) or they have been
26 * written since then and the write completion callback
27 * was called and no write error was indicated and a
28 * FLUSH request to the device where these blocks are
29 * located was received and completed.
30 * 2b. All referenced blocks need to have a generation
31 * number which is equal to the parent's number.
33 * One issue that was found using this module was that the log
34 * tree on disk became temporarily corrupted because disk blocks
35 * that had been in use for the log tree had been freed and
36 * reused too early, while being referenced by the written super
39 * The search term in the kernel log that can be used to filter
40 * on the existence of detected integrity issues is
43 * The integrity check is enabled via mount options. These
44 * mount options are only supported if the integrity check
45 * tool is compiled by defining BTRFS_FS_CHECK_INTEGRITY.
47 * Example #1, apply integrity checks to all metadata:
48 * mount /dev/sdb1 /mnt -o check_int
50 * Example #2, apply integrity checks to all metadata and
52 * mount /dev/sdb1 /mnt -o check_int_data
54 * Example #3, apply integrity checks to all metadata and dump
55 * the tree that the super block references to kernel messages
56 * each time after a super block was written:
57 * mount /dev/sdb1 /mnt -o check_int,check_int_print_mask=263
59 * If the integrity check tool is included and activated in
60 * the mount options, plenty of kernel memory is used, and
61 * plenty of additional CPU cycles are spent. Enabling this
62 * functionality is not intended for normal use. In most
63 * cases, unless you are a btrfs developer who needs to verify
64 * the integrity of (super)-block write requests, do not
65 * enable the config option BTRFS_FS_CHECK_INTEGRITY to
66 * include and compile the integrity check tool.
68 * Expect millions of lines of information in the kernel log with an
69 * enabled check_int_print_mask. Therefore set LOG_BUF_SHIFT in the
70 * kernel config to at least 26 (which is 64MB). Usually the value is
71 * limited to 21 (which is 2MB) in init/Kconfig. The file needs to be
72 * changed like this before LOG_BUF_SHIFT can be set to a high value:
73 * config LOG_BUF_SHIFT
74 * int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
78 #include <linux/sched.h>
79 #include <linux/slab.h>
80 #include <linux/buffer_head.h>
81 #include <linux/mutex.h>
82 #include <linux/genhd.h>
83 #include <linux/blkdev.h>
85 #include <linux/string.h>
86 #include <linux/crc32c.h>
89 #include "transaction.h"
90 #include "extent_io.h"
92 #include "print-tree.h"
94 #include "check-integrity.h"
95 #include "rcu-string.h"
96 #include "compression.h"
98 #define BTRFSIC_BLOCK_HASHTABLE_SIZE 0x10000
99 #define BTRFSIC_BLOCK_LINK_HASHTABLE_SIZE 0x10000
100 #define BTRFSIC_DEV2STATE_HASHTABLE_SIZE 0x100
101 #define BTRFSIC_BLOCK_MAGIC_NUMBER 0x14491051
102 #define BTRFSIC_BLOCK_LINK_MAGIC_NUMBER 0x11070807
103 #define BTRFSIC_DEV2STATE_MAGIC_NUMBER 0x20111530
104 #define BTRFSIC_BLOCK_STACK_FRAME_MAGIC_NUMBER 20111300
105 #define BTRFSIC_TREE_DUMP_MAX_INDENT_LEVEL (200 - 6) /* in characters,
106 * excluding " [...]" */
107 #define BTRFSIC_GENERATION_UNKNOWN ((u64)-1)
110 * The definition of the bitmask fields for the print_mask.
111 * They are specified with the mount option check_integrity_print_mask.
113 #define BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE 0x00000001
114 #define BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION 0x00000002
115 #define BTRFSIC_PRINT_MASK_TREE_AFTER_SB_WRITE 0x00000004
116 #define BTRFSIC_PRINT_MASK_TREE_BEFORE_SB_WRITE 0x00000008
117 #define BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH 0x00000010
118 #define BTRFSIC_PRINT_MASK_END_IO_BIO_BH 0x00000020
119 #define BTRFSIC_PRINT_MASK_VERBOSE 0x00000040
120 #define BTRFSIC_PRINT_MASK_VERY_VERBOSE 0x00000080
121 #define BTRFSIC_PRINT_MASK_INITIAL_TREE 0x00000100
122 #define BTRFSIC_PRINT_MASK_INITIAL_ALL_TREES 0x00000200
123 #define BTRFSIC_PRINT_MASK_INITIAL_DATABASE 0x00000400
124 #define BTRFSIC_PRINT_MASK_NUM_COPIES 0x00000800
125 #define BTRFSIC_PRINT_MASK_TREE_WITH_ALL_MIRRORS 0x00001000
126 #define BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH_VERBOSE 0x00002000
128 struct btrfsic_dev_state;
129 struct btrfsic_state;
131 struct btrfsic_block {
132 u32 magic_num; /* only used for debug purposes */
133 unsigned int is_metadata:1; /* if it is meta-data, not data-data */
134 unsigned int is_superblock:1; /* if it is one of the superblocks */
135 unsigned int is_iodone:1; /* if is done by lower subsystem */
136 unsigned int iodone_w_error:1; /* error was indicated to endio */
137 unsigned int never_written:1; /* block was added because it was
138 * referenced, not because it was
140 unsigned int mirror_num; /* large enough to hold
141 * BTRFS_SUPER_MIRROR_MAX */
142 struct btrfsic_dev_state *dev_state;
143 u64 dev_bytenr; /* key, physical byte num on disk */
144 u64 logical_bytenr; /* logical byte num on disk */
146 struct btrfs_disk_key disk_key; /* extra info to print in case of
147 * issues, will not always be correct */
148 struct list_head collision_resolving_node; /* list node */
149 struct list_head all_blocks_node; /* list node */
151 /* the following two lists contain block_link items */
152 struct list_head ref_to_list; /* list */
153 struct list_head ref_from_list; /* list */
154 struct btrfsic_block *next_in_same_bio;
155 void *orig_bio_bh_private;
159 } orig_bio_bh_end_io;
160 int submit_bio_bh_rw;
161 u64 flush_gen; /* only valid if !never_written */
165 * Elements of this type are allocated dynamically and required because
166 * each block object can refer to and can be ref from multiple blocks.
167 * The key to lookup them in the hashtable is the dev_bytenr of
168 * the block ref to plus the one from the block referred from.
169 * The fact that they are searchable via a hashtable and that a
170 * ref_cnt is maintained is not required for the btrfs integrity
171 * check algorithm itself, it is only used to make the output more
172 * beautiful in case that an error is detected (an error is defined
173 * as a write operation to a block while that block is still referenced).
175 struct btrfsic_block_link {
176 u32 magic_num; /* only used for debug purposes */
178 struct list_head node_ref_to; /* list node */
179 struct list_head node_ref_from; /* list node */
180 struct list_head collision_resolving_node; /* list node */
181 struct btrfsic_block *block_ref_to;
182 struct btrfsic_block *block_ref_from;
183 u64 parent_generation;
186 struct btrfsic_dev_state {
187 u32 magic_num; /* only used for debug purposes */
188 struct block_device *bdev;
189 struct btrfsic_state *state;
190 struct list_head collision_resolving_node; /* list node */
191 struct btrfsic_block dummy_block_for_bio_bh_flush;
193 char name[BDEVNAME_SIZE];
196 struct btrfsic_block_hashtable {
197 struct list_head table[BTRFSIC_BLOCK_HASHTABLE_SIZE];
200 struct btrfsic_block_link_hashtable {
201 struct list_head table[BTRFSIC_BLOCK_LINK_HASHTABLE_SIZE];
204 struct btrfsic_dev_state_hashtable {
205 struct list_head table[BTRFSIC_DEV2STATE_HASHTABLE_SIZE];
208 struct btrfsic_block_data_ctx {
209 u64 start; /* virtual bytenr */
210 u64 dev_bytenr; /* physical bytenr on device */
212 struct btrfsic_dev_state *dev;
218 /* This structure is used to implement recursion without occupying
219 * any stack space, refer to btrfsic_process_metablock() */
220 struct btrfsic_stack_frame {
228 struct btrfsic_block *block;
229 struct btrfsic_block_data_ctx *block_ctx;
230 struct btrfsic_block *next_block;
231 struct btrfsic_block_data_ctx next_block_ctx;
232 struct btrfs_header *hdr;
233 struct btrfsic_stack_frame *prev;
236 /* Some state per mounted filesystem */
237 struct btrfsic_state {
239 int include_extent_data;
241 struct list_head all_blocks_list;
242 struct btrfsic_block_hashtable block_hashtable;
243 struct btrfsic_block_link_hashtable block_link_hashtable;
244 struct btrfs_fs_info *fs_info;
245 u64 max_superblock_generation;
246 struct btrfsic_block *latest_superblock;
251 static void btrfsic_block_init(struct btrfsic_block *b);
252 static struct btrfsic_block *btrfsic_block_alloc(void);
253 static void btrfsic_block_free(struct btrfsic_block *b);
254 static void btrfsic_block_link_init(struct btrfsic_block_link *n);
255 static struct btrfsic_block_link *btrfsic_block_link_alloc(void);
256 static void btrfsic_block_link_free(struct btrfsic_block_link *n);
257 static void btrfsic_dev_state_init(struct btrfsic_dev_state *ds);
258 static struct btrfsic_dev_state *btrfsic_dev_state_alloc(void);
259 static void btrfsic_dev_state_free(struct btrfsic_dev_state *ds);
260 static void btrfsic_block_hashtable_init(struct btrfsic_block_hashtable *h);
261 static void btrfsic_block_hashtable_add(struct btrfsic_block *b,
262 struct btrfsic_block_hashtable *h);
263 static void btrfsic_block_hashtable_remove(struct btrfsic_block *b);
264 static struct btrfsic_block *btrfsic_block_hashtable_lookup(
265 struct block_device *bdev,
267 struct btrfsic_block_hashtable *h);
268 static void btrfsic_block_link_hashtable_init(
269 struct btrfsic_block_link_hashtable *h);
270 static void btrfsic_block_link_hashtable_add(
271 struct btrfsic_block_link *l,
272 struct btrfsic_block_link_hashtable *h);
273 static void btrfsic_block_link_hashtable_remove(struct btrfsic_block_link *l);
274 static struct btrfsic_block_link *btrfsic_block_link_hashtable_lookup(
275 struct block_device *bdev_ref_to,
276 u64 dev_bytenr_ref_to,
277 struct block_device *bdev_ref_from,
278 u64 dev_bytenr_ref_from,
279 struct btrfsic_block_link_hashtable *h);
280 static void btrfsic_dev_state_hashtable_init(
281 struct btrfsic_dev_state_hashtable *h);
282 static void btrfsic_dev_state_hashtable_add(
283 struct btrfsic_dev_state *ds,
284 struct btrfsic_dev_state_hashtable *h);
285 static void btrfsic_dev_state_hashtable_remove(struct btrfsic_dev_state *ds);
286 static struct btrfsic_dev_state *btrfsic_dev_state_hashtable_lookup(dev_t dev,
287 struct btrfsic_dev_state_hashtable *h);
288 static struct btrfsic_stack_frame *btrfsic_stack_frame_alloc(void);
289 static void btrfsic_stack_frame_free(struct btrfsic_stack_frame *sf);
290 static int btrfsic_process_superblock(struct btrfsic_state *state,
291 struct btrfs_fs_devices *fs_devices);
292 static int btrfsic_process_metablock(struct btrfsic_state *state,
293 struct btrfsic_block *block,
294 struct btrfsic_block_data_ctx *block_ctx,
295 int limit_nesting, int force_iodone_flag);
296 static void btrfsic_read_from_block_data(
297 struct btrfsic_block_data_ctx *block_ctx,
298 void *dst, u32 offset, size_t len);
299 static int btrfsic_create_link_to_next_block(
300 struct btrfsic_state *state,
301 struct btrfsic_block *block,
302 struct btrfsic_block_data_ctx
303 *block_ctx, u64 next_bytenr,
305 struct btrfsic_block_data_ctx *next_block_ctx,
306 struct btrfsic_block **next_blockp,
307 int force_iodone_flag,
308 int *num_copiesp, int *mirror_nump,
309 struct btrfs_disk_key *disk_key,
310 u64 parent_generation);
311 static int btrfsic_handle_extent_data(struct btrfsic_state *state,
312 struct btrfsic_block *block,
313 struct btrfsic_block_data_ctx *block_ctx,
314 u32 item_offset, int force_iodone_flag);
315 static int btrfsic_map_block(struct btrfsic_state *state, u64 bytenr, u32 len,
316 struct btrfsic_block_data_ctx *block_ctx_out,
318 static void btrfsic_release_block_ctx(struct btrfsic_block_data_ctx *block_ctx);
319 static int btrfsic_read_block(struct btrfsic_state *state,
320 struct btrfsic_block_data_ctx *block_ctx);
321 static void btrfsic_dump_database(struct btrfsic_state *state);
322 static int btrfsic_test_for_metadata(struct btrfsic_state *state,
323 char **datav, unsigned int num_pages);
324 static void btrfsic_process_written_block(struct btrfsic_dev_state *dev_state,
325 u64 dev_bytenr, char **mapped_datav,
326 unsigned int num_pages,
327 struct bio *bio, int *bio_is_patched,
328 struct buffer_head *bh,
329 int submit_bio_bh_rw);
330 static int btrfsic_process_written_superblock(
331 struct btrfsic_state *state,
332 struct btrfsic_block *const block,
333 struct btrfs_super_block *const super_hdr);
334 static void btrfsic_bio_end_io(struct bio *bp);
335 static void btrfsic_bh_end_io(struct buffer_head *bh, int uptodate);
336 static int btrfsic_is_block_ref_by_superblock(const struct btrfsic_state *state,
337 const struct btrfsic_block *block,
338 int recursion_level);
339 static int btrfsic_check_all_ref_blocks(struct btrfsic_state *state,
340 struct btrfsic_block *const block,
341 int recursion_level);
342 static void btrfsic_print_add_link(const struct btrfsic_state *state,
343 const struct btrfsic_block_link *l);
344 static void btrfsic_print_rem_link(const struct btrfsic_state *state,
345 const struct btrfsic_block_link *l);
346 static char btrfsic_get_block_type(const struct btrfsic_state *state,
347 const struct btrfsic_block *block);
348 static void btrfsic_dump_tree(const struct btrfsic_state *state);
349 static void btrfsic_dump_tree_sub(const struct btrfsic_state *state,
350 const struct btrfsic_block *block,
352 static struct btrfsic_block_link *btrfsic_block_link_lookup_or_add(
353 struct btrfsic_state *state,
354 struct btrfsic_block_data_ctx *next_block_ctx,
355 struct btrfsic_block *next_block,
356 struct btrfsic_block *from_block,
357 u64 parent_generation);
358 static struct btrfsic_block *btrfsic_block_lookup_or_add(
359 struct btrfsic_state *state,
360 struct btrfsic_block_data_ctx *block_ctx,
361 const char *additional_string,
367 static int btrfsic_process_superblock_dev_mirror(
368 struct btrfsic_state *state,
369 struct btrfsic_dev_state *dev_state,
370 struct btrfs_device *device,
371 int superblock_mirror_num,
372 struct btrfsic_dev_state **selected_dev_state,
373 struct btrfs_super_block *selected_super);
374 static struct btrfsic_dev_state *btrfsic_dev_state_lookup(dev_t dev);
375 static void btrfsic_cmp_log_and_dev_bytenr(struct btrfsic_state *state,
377 struct btrfsic_dev_state *dev_state,
380 static struct mutex btrfsic_mutex;
381 static int btrfsic_is_initialized;
382 static struct btrfsic_dev_state_hashtable btrfsic_dev_state_hashtable;
385 static void btrfsic_block_init(struct btrfsic_block *b)
387 b->magic_num = BTRFSIC_BLOCK_MAGIC_NUMBER;
390 b->logical_bytenr = 0;
391 b->generation = BTRFSIC_GENERATION_UNKNOWN;
392 b->disk_key.objectid = 0;
393 b->disk_key.type = 0;
394 b->disk_key.offset = 0;
396 b->is_superblock = 0;
398 b->iodone_w_error = 0;
399 b->never_written = 0;
401 b->next_in_same_bio = NULL;
402 b->orig_bio_bh_private = NULL;
403 b->orig_bio_bh_end_io.bio = NULL;
404 INIT_LIST_HEAD(&b->collision_resolving_node);
405 INIT_LIST_HEAD(&b->all_blocks_node);
406 INIT_LIST_HEAD(&b->ref_to_list);
407 INIT_LIST_HEAD(&b->ref_from_list);
408 b->submit_bio_bh_rw = 0;
412 static struct btrfsic_block *btrfsic_block_alloc(void)
414 struct btrfsic_block *b;
416 b = kzalloc(sizeof(*b), GFP_NOFS);
418 btrfsic_block_init(b);
423 static void btrfsic_block_free(struct btrfsic_block *b)
425 BUG_ON(!(NULL == b || BTRFSIC_BLOCK_MAGIC_NUMBER == b->magic_num));
429 static void btrfsic_block_link_init(struct btrfsic_block_link *l)
431 l->magic_num = BTRFSIC_BLOCK_LINK_MAGIC_NUMBER;
433 INIT_LIST_HEAD(&l->node_ref_to);
434 INIT_LIST_HEAD(&l->node_ref_from);
435 INIT_LIST_HEAD(&l->collision_resolving_node);
436 l->block_ref_to = NULL;
437 l->block_ref_from = NULL;
440 static struct btrfsic_block_link *btrfsic_block_link_alloc(void)
442 struct btrfsic_block_link *l;
444 l = kzalloc(sizeof(*l), GFP_NOFS);
446 btrfsic_block_link_init(l);
451 static void btrfsic_block_link_free(struct btrfsic_block_link *l)
453 BUG_ON(!(NULL == l || BTRFSIC_BLOCK_LINK_MAGIC_NUMBER == l->magic_num));
457 static void btrfsic_dev_state_init(struct btrfsic_dev_state *ds)
459 ds->magic_num = BTRFSIC_DEV2STATE_MAGIC_NUMBER;
463 INIT_LIST_HEAD(&ds->collision_resolving_node);
464 ds->last_flush_gen = 0;
465 btrfsic_block_init(&ds->dummy_block_for_bio_bh_flush);
466 ds->dummy_block_for_bio_bh_flush.is_iodone = 1;
467 ds->dummy_block_for_bio_bh_flush.dev_state = ds;
470 static struct btrfsic_dev_state *btrfsic_dev_state_alloc(void)
472 struct btrfsic_dev_state *ds;
474 ds = kzalloc(sizeof(*ds), GFP_NOFS);
476 btrfsic_dev_state_init(ds);
481 static void btrfsic_dev_state_free(struct btrfsic_dev_state *ds)
483 BUG_ON(!(NULL == ds ||
484 BTRFSIC_DEV2STATE_MAGIC_NUMBER == ds->magic_num));
488 static void btrfsic_block_hashtable_init(struct btrfsic_block_hashtable *h)
492 for (i = 0; i < BTRFSIC_BLOCK_HASHTABLE_SIZE; i++)
493 INIT_LIST_HEAD(h->table + i);
496 static void btrfsic_block_hashtable_add(struct btrfsic_block *b,
497 struct btrfsic_block_hashtable *h)
499 const unsigned int hashval =
500 (((unsigned int)(b->dev_bytenr >> 16)) ^
501 ((unsigned int)((uintptr_t)b->dev_state->bdev))) &
502 (BTRFSIC_BLOCK_HASHTABLE_SIZE - 1);
504 list_add(&b->collision_resolving_node, h->table + hashval);
507 static void btrfsic_block_hashtable_remove(struct btrfsic_block *b)
509 list_del(&b->collision_resolving_node);
512 static struct btrfsic_block *btrfsic_block_hashtable_lookup(
513 struct block_device *bdev,
515 struct btrfsic_block_hashtable *h)
517 const unsigned int hashval =
518 (((unsigned int)(dev_bytenr >> 16)) ^
519 ((unsigned int)((uintptr_t)bdev))) &
520 (BTRFSIC_BLOCK_HASHTABLE_SIZE - 1);
521 struct btrfsic_block *b;
523 list_for_each_entry(b, h->table + hashval, collision_resolving_node) {
524 if (b->dev_state->bdev == bdev && b->dev_bytenr == dev_bytenr)
531 static void btrfsic_block_link_hashtable_init(
532 struct btrfsic_block_link_hashtable *h)
536 for (i = 0; i < BTRFSIC_BLOCK_LINK_HASHTABLE_SIZE; i++)
537 INIT_LIST_HEAD(h->table + i);
540 static void btrfsic_block_link_hashtable_add(
541 struct btrfsic_block_link *l,
542 struct btrfsic_block_link_hashtable *h)
544 const unsigned int hashval =
545 (((unsigned int)(l->block_ref_to->dev_bytenr >> 16)) ^
546 ((unsigned int)(l->block_ref_from->dev_bytenr >> 16)) ^
547 ((unsigned int)((uintptr_t)l->block_ref_to->dev_state->bdev)) ^
548 ((unsigned int)((uintptr_t)l->block_ref_from->dev_state->bdev)))
549 & (BTRFSIC_BLOCK_LINK_HASHTABLE_SIZE - 1);
551 BUG_ON(NULL == l->block_ref_to);
552 BUG_ON(NULL == l->block_ref_from);
553 list_add(&l->collision_resolving_node, h->table + hashval);
556 static void btrfsic_block_link_hashtable_remove(struct btrfsic_block_link *l)
558 list_del(&l->collision_resolving_node);
561 static struct btrfsic_block_link *btrfsic_block_link_hashtable_lookup(
562 struct block_device *bdev_ref_to,
563 u64 dev_bytenr_ref_to,
564 struct block_device *bdev_ref_from,
565 u64 dev_bytenr_ref_from,
566 struct btrfsic_block_link_hashtable *h)
568 const unsigned int hashval =
569 (((unsigned int)(dev_bytenr_ref_to >> 16)) ^
570 ((unsigned int)(dev_bytenr_ref_from >> 16)) ^
571 ((unsigned int)((uintptr_t)bdev_ref_to)) ^
572 ((unsigned int)((uintptr_t)bdev_ref_from))) &
573 (BTRFSIC_BLOCK_LINK_HASHTABLE_SIZE - 1);
574 struct btrfsic_block_link *l;
576 list_for_each_entry(l, h->table + hashval, collision_resolving_node) {
577 BUG_ON(NULL == l->block_ref_to);
578 BUG_ON(NULL == l->block_ref_from);
579 if (l->block_ref_to->dev_state->bdev == bdev_ref_to &&
580 l->block_ref_to->dev_bytenr == dev_bytenr_ref_to &&
581 l->block_ref_from->dev_state->bdev == bdev_ref_from &&
582 l->block_ref_from->dev_bytenr == dev_bytenr_ref_from)
589 static void btrfsic_dev_state_hashtable_init(
590 struct btrfsic_dev_state_hashtable *h)
594 for (i = 0; i < BTRFSIC_DEV2STATE_HASHTABLE_SIZE; i++)
595 INIT_LIST_HEAD(h->table + i);
598 static void btrfsic_dev_state_hashtable_add(
599 struct btrfsic_dev_state *ds,
600 struct btrfsic_dev_state_hashtable *h)
602 const unsigned int hashval =
603 (((unsigned int)((uintptr_t)ds->bdev->bd_dev)) &
604 (BTRFSIC_DEV2STATE_HASHTABLE_SIZE - 1));
606 list_add(&ds->collision_resolving_node, h->table + hashval);
609 static void btrfsic_dev_state_hashtable_remove(struct btrfsic_dev_state *ds)
611 list_del(&ds->collision_resolving_node);
614 static struct btrfsic_dev_state *btrfsic_dev_state_hashtable_lookup(dev_t dev,
615 struct btrfsic_dev_state_hashtable *h)
617 const unsigned int hashval =
618 dev & (BTRFSIC_DEV2STATE_HASHTABLE_SIZE - 1);
619 struct btrfsic_dev_state *ds;
621 list_for_each_entry(ds, h->table + hashval, collision_resolving_node) {
622 if (ds->bdev->bd_dev == dev)
629 static int btrfsic_process_superblock(struct btrfsic_state *state,
630 struct btrfs_fs_devices *fs_devices)
632 struct btrfs_fs_info *fs_info = state->fs_info;
633 struct btrfs_super_block *selected_super;
634 struct list_head *dev_head = &fs_devices->devices;
635 struct btrfs_device *device;
636 struct btrfsic_dev_state *selected_dev_state = NULL;
640 BUG_ON(NULL == state);
641 selected_super = kzalloc(sizeof(*selected_super), GFP_NOFS);
642 if (NULL == selected_super) {
643 pr_info("btrfsic: error, kmalloc failed!\n");
647 list_for_each_entry(device, dev_head, dev_list) {
649 struct btrfsic_dev_state *dev_state;
651 if (!device->bdev || !device->name)
654 dev_state = btrfsic_dev_state_lookup(device->bdev->bd_dev);
655 BUG_ON(NULL == dev_state);
656 for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
657 ret = btrfsic_process_superblock_dev_mirror(
658 state, dev_state, device, i,
659 &selected_dev_state, selected_super);
660 if (0 != ret && 0 == i) {
661 kfree(selected_super);
667 if (NULL == state->latest_superblock) {
668 pr_info("btrfsic: no superblock found!\n");
669 kfree(selected_super);
673 state->csum_size = btrfs_super_csum_size(selected_super);
675 for (pass = 0; pass < 3; pass++) {
682 next_bytenr = btrfs_super_root(selected_super);
683 if (state->print_mask &
684 BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION)
685 pr_info("root@%llu\n", next_bytenr);
688 next_bytenr = btrfs_super_chunk_root(selected_super);
689 if (state->print_mask &
690 BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION)
691 pr_info("chunk@%llu\n", next_bytenr);
694 next_bytenr = btrfs_super_log_root(selected_super);
695 if (0 == next_bytenr)
697 if (state->print_mask &
698 BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION)
699 pr_info("log@%llu\n", next_bytenr);
703 num_copies = btrfs_num_copies(fs_info, next_bytenr,
704 state->metablock_size);
705 if (state->print_mask & BTRFSIC_PRINT_MASK_NUM_COPIES)
706 pr_info("num_copies(log_bytenr=%llu) = %d\n",
707 next_bytenr, num_copies);
709 for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
710 struct btrfsic_block *next_block;
711 struct btrfsic_block_data_ctx tmp_next_block_ctx;
712 struct btrfsic_block_link *l;
714 ret = btrfsic_map_block(state, next_bytenr,
715 state->metablock_size,
719 pr_info("btrfsic: btrfsic_map_block(root @%llu, mirror %d) failed!\n",
720 next_bytenr, mirror_num);
721 kfree(selected_super);
725 next_block = btrfsic_block_hashtable_lookup(
726 tmp_next_block_ctx.dev->bdev,
727 tmp_next_block_ctx.dev_bytenr,
728 &state->block_hashtable);
729 BUG_ON(NULL == next_block);
731 l = btrfsic_block_link_hashtable_lookup(
732 tmp_next_block_ctx.dev->bdev,
733 tmp_next_block_ctx.dev_bytenr,
734 state->latest_superblock->dev_state->
736 state->latest_superblock->dev_bytenr,
737 &state->block_link_hashtable);
740 ret = btrfsic_read_block(state, &tmp_next_block_ctx);
741 if (ret < (int)PAGE_SIZE) {
742 pr_info("btrfsic: read @logical %llu failed!\n",
743 tmp_next_block_ctx.start);
744 btrfsic_release_block_ctx(&tmp_next_block_ctx);
745 kfree(selected_super);
749 ret = btrfsic_process_metablock(state,
752 BTRFS_MAX_LEVEL + 3, 1);
753 btrfsic_release_block_ctx(&tmp_next_block_ctx);
757 kfree(selected_super);
761 static int btrfsic_process_superblock_dev_mirror(
762 struct btrfsic_state *state,
763 struct btrfsic_dev_state *dev_state,
764 struct btrfs_device *device,
765 int superblock_mirror_num,
766 struct btrfsic_dev_state **selected_dev_state,
767 struct btrfs_super_block *selected_super)
769 struct btrfs_fs_info *fs_info = state->fs_info;
770 struct btrfs_super_block *super_tmp;
772 struct buffer_head *bh;
773 struct btrfsic_block *superblock_tmp;
775 struct block_device *const superblock_bdev = device->bdev;
777 /* super block bytenr is always the unmapped device bytenr */
778 dev_bytenr = btrfs_sb_offset(superblock_mirror_num);
779 if (dev_bytenr + BTRFS_SUPER_INFO_SIZE > device->commit_total_bytes)
781 bh = __bread(superblock_bdev, dev_bytenr / BTRFS_BDEV_BLOCKSIZE,
782 BTRFS_SUPER_INFO_SIZE);
785 super_tmp = (struct btrfs_super_block *)
786 (bh->b_data + (dev_bytenr & (BTRFS_BDEV_BLOCKSIZE - 1)));
788 if (btrfs_super_bytenr(super_tmp) != dev_bytenr ||
789 btrfs_super_magic(super_tmp) != BTRFS_MAGIC ||
790 memcmp(device->uuid, super_tmp->dev_item.uuid, BTRFS_UUID_SIZE) ||
791 btrfs_super_nodesize(super_tmp) != state->metablock_size ||
792 btrfs_super_sectorsize(super_tmp) != state->datablock_size) {
798 btrfsic_block_hashtable_lookup(superblock_bdev,
800 &state->block_hashtable);
801 if (NULL == superblock_tmp) {
802 superblock_tmp = btrfsic_block_alloc();
803 if (NULL == superblock_tmp) {
804 pr_info("btrfsic: error, kmalloc failed!\n");
808 /* for superblock, only the dev_bytenr makes sense */
809 superblock_tmp->dev_bytenr = dev_bytenr;
810 superblock_tmp->dev_state = dev_state;
811 superblock_tmp->logical_bytenr = dev_bytenr;
812 superblock_tmp->generation = btrfs_super_generation(super_tmp);
813 superblock_tmp->is_metadata = 1;
814 superblock_tmp->is_superblock = 1;
815 superblock_tmp->is_iodone = 1;
816 superblock_tmp->never_written = 0;
817 superblock_tmp->mirror_num = 1 + superblock_mirror_num;
818 if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
819 btrfs_info_in_rcu(fs_info,
820 "new initial S-block (bdev %p, %s) @%llu (%s/%llu/%d)",
822 rcu_str_deref(device->name), dev_bytenr,
823 dev_state->name, dev_bytenr,
824 superblock_mirror_num);
825 list_add(&superblock_tmp->all_blocks_node,
826 &state->all_blocks_list);
827 btrfsic_block_hashtable_add(superblock_tmp,
828 &state->block_hashtable);
831 /* select the one with the highest generation field */
832 if (btrfs_super_generation(super_tmp) >
833 state->max_superblock_generation ||
834 0 == state->max_superblock_generation) {
835 memcpy(selected_super, super_tmp, sizeof(*selected_super));
836 *selected_dev_state = dev_state;
837 state->max_superblock_generation =
838 btrfs_super_generation(super_tmp);
839 state->latest_superblock = superblock_tmp;
842 for (pass = 0; pass < 3; pass++) {
846 const char *additional_string = NULL;
847 struct btrfs_disk_key tmp_disk_key;
849 tmp_disk_key.type = BTRFS_ROOT_ITEM_KEY;
850 tmp_disk_key.offset = 0;
853 btrfs_set_disk_key_objectid(&tmp_disk_key,
854 BTRFS_ROOT_TREE_OBJECTID);
855 additional_string = "initial root ";
856 next_bytenr = btrfs_super_root(super_tmp);
859 btrfs_set_disk_key_objectid(&tmp_disk_key,
860 BTRFS_CHUNK_TREE_OBJECTID);
861 additional_string = "initial chunk ";
862 next_bytenr = btrfs_super_chunk_root(super_tmp);
865 btrfs_set_disk_key_objectid(&tmp_disk_key,
866 BTRFS_TREE_LOG_OBJECTID);
867 additional_string = "initial log ";
868 next_bytenr = btrfs_super_log_root(super_tmp);
869 if (0 == next_bytenr)
874 num_copies = btrfs_num_copies(fs_info, next_bytenr,
875 state->metablock_size);
876 if (state->print_mask & BTRFSIC_PRINT_MASK_NUM_COPIES)
877 pr_info("num_copies(log_bytenr=%llu) = %d\n",
878 next_bytenr, num_copies);
879 for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
880 struct btrfsic_block *next_block;
881 struct btrfsic_block_data_ctx tmp_next_block_ctx;
882 struct btrfsic_block_link *l;
884 if (btrfsic_map_block(state, next_bytenr,
885 state->metablock_size,
888 pr_info("btrfsic: btrfsic_map_block(bytenr @%llu, mirror %d) failed!\n",
889 next_bytenr, mirror_num);
894 next_block = btrfsic_block_lookup_or_add(
895 state, &tmp_next_block_ctx,
896 additional_string, 1, 1, 0,
898 if (NULL == next_block) {
899 btrfsic_release_block_ctx(&tmp_next_block_ctx);
904 next_block->disk_key = tmp_disk_key;
905 next_block->generation = BTRFSIC_GENERATION_UNKNOWN;
906 l = btrfsic_block_link_lookup_or_add(
907 state, &tmp_next_block_ctx,
908 next_block, superblock_tmp,
909 BTRFSIC_GENERATION_UNKNOWN);
910 btrfsic_release_block_ctx(&tmp_next_block_ctx);
917 if (state->print_mask & BTRFSIC_PRINT_MASK_INITIAL_ALL_TREES)
918 btrfsic_dump_tree_sub(state, superblock_tmp, 0);
924 static struct btrfsic_stack_frame *btrfsic_stack_frame_alloc(void)
926 struct btrfsic_stack_frame *sf;
928 sf = kzalloc(sizeof(*sf), GFP_NOFS);
930 pr_info("btrfsic: alloc memory failed!\n");
932 sf->magic = BTRFSIC_BLOCK_STACK_FRAME_MAGIC_NUMBER;
936 static void btrfsic_stack_frame_free(struct btrfsic_stack_frame *sf)
938 BUG_ON(!(NULL == sf ||
939 BTRFSIC_BLOCK_STACK_FRAME_MAGIC_NUMBER == sf->magic));
943 static int btrfsic_process_metablock(
944 struct btrfsic_state *state,
945 struct btrfsic_block *const first_block,
946 struct btrfsic_block_data_ctx *const first_block_ctx,
947 int first_limit_nesting, int force_iodone_flag)
949 struct btrfsic_stack_frame initial_stack_frame = { 0 };
950 struct btrfsic_stack_frame *sf;
951 struct btrfsic_stack_frame *next_stack;
952 struct btrfs_header *const first_hdr =
953 (struct btrfs_header *)first_block_ctx->datav[0];
956 sf = &initial_stack_frame;
959 sf->limit_nesting = first_limit_nesting;
960 sf->block = first_block;
961 sf->block_ctx = first_block_ctx;
962 sf->next_block = NULL;
966 continue_with_new_stack_frame:
967 sf->block->generation = le64_to_cpu(sf->hdr->generation);
968 if (0 == sf->hdr->level) {
969 struct btrfs_leaf *const leafhdr =
970 (struct btrfs_leaf *)sf->hdr;
973 sf->nr = btrfs_stack_header_nritems(&leafhdr->header);
975 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
976 pr_info("leaf %llu items %d generation %llu owner %llu\n",
977 sf->block_ctx->start, sf->nr,
978 btrfs_stack_header_generation(
980 btrfs_stack_header_owner(
984 continue_with_current_leaf_stack_frame:
985 if (0 == sf->num_copies || sf->mirror_num > sf->num_copies) {
990 if (sf->i < sf->nr) {
991 struct btrfs_item disk_item;
992 u32 disk_item_offset =
993 (uintptr_t)(leafhdr->items + sf->i) -
995 struct btrfs_disk_key *disk_key;
1000 if (disk_item_offset + sizeof(struct btrfs_item) >
1001 sf->block_ctx->len) {
1002 leaf_item_out_of_bounce_error:
1003 pr_info("btrfsic: leaf item out of bounce at logical %llu, dev %s\n",
1004 sf->block_ctx->start,
1005 sf->block_ctx->dev->name);
1006 goto one_stack_frame_backwards;
1008 btrfsic_read_from_block_data(sf->block_ctx,
1011 sizeof(struct btrfs_item));
1012 item_offset = btrfs_stack_item_offset(&disk_item);
1013 item_size = btrfs_stack_item_size(&disk_item);
1014 disk_key = &disk_item.key;
1015 type = btrfs_disk_key_type(disk_key);
1017 if (BTRFS_ROOT_ITEM_KEY == type) {
1018 struct btrfs_root_item root_item;
1019 u32 root_item_offset;
1022 root_item_offset = item_offset +
1023 offsetof(struct btrfs_leaf, items);
1024 if (root_item_offset + item_size >
1026 goto leaf_item_out_of_bounce_error;
1027 btrfsic_read_from_block_data(
1028 sf->block_ctx, &root_item,
1031 next_bytenr = btrfs_root_bytenr(&root_item);
1034 btrfsic_create_link_to_next_block(
1040 &sf->next_block_ctx,
1046 btrfs_root_generation(
1049 goto one_stack_frame_backwards;
1051 if (NULL != sf->next_block) {
1052 struct btrfs_header *const next_hdr =
1053 (struct btrfs_header *)
1054 sf->next_block_ctx.datav[0];
1057 btrfsic_stack_frame_alloc();
1058 if (NULL == next_stack) {
1060 btrfsic_release_block_ctx(
1063 goto one_stack_frame_backwards;
1067 next_stack->block = sf->next_block;
1068 next_stack->block_ctx =
1069 &sf->next_block_ctx;
1070 next_stack->next_block = NULL;
1071 next_stack->hdr = next_hdr;
1072 next_stack->limit_nesting =
1073 sf->limit_nesting - 1;
1074 next_stack->prev = sf;
1076 goto continue_with_new_stack_frame;
1078 } else if (BTRFS_EXTENT_DATA_KEY == type &&
1079 state->include_extent_data) {
1080 sf->error = btrfsic_handle_extent_data(
1087 goto one_stack_frame_backwards;
1090 goto continue_with_current_leaf_stack_frame;
1093 struct btrfs_node *const nodehdr = (struct btrfs_node *)sf->hdr;
1096 sf->nr = btrfs_stack_header_nritems(&nodehdr->header);
1098 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1099 pr_info("node %llu level %d items %d generation %llu owner %llu\n",
1100 sf->block_ctx->start,
1101 nodehdr->header.level, sf->nr,
1102 btrfs_stack_header_generation(
1104 btrfs_stack_header_owner(
1108 continue_with_current_node_stack_frame:
1109 if (0 == sf->num_copies || sf->mirror_num > sf->num_copies) {
1114 if (sf->i < sf->nr) {
1115 struct btrfs_key_ptr key_ptr;
1119 key_ptr_offset = (uintptr_t)(nodehdr->ptrs + sf->i) -
1121 if (key_ptr_offset + sizeof(struct btrfs_key_ptr) >
1122 sf->block_ctx->len) {
1123 pr_info("btrfsic: node item out of bounce at logical %llu, dev %s\n",
1124 sf->block_ctx->start,
1125 sf->block_ctx->dev->name);
1126 goto one_stack_frame_backwards;
1128 btrfsic_read_from_block_data(
1129 sf->block_ctx, &key_ptr, key_ptr_offset,
1130 sizeof(struct btrfs_key_ptr));
1131 next_bytenr = btrfs_stack_key_blockptr(&key_ptr);
1133 sf->error = btrfsic_create_link_to_next_block(
1139 &sf->next_block_ctx,
1145 btrfs_stack_key_generation(&key_ptr));
1147 goto one_stack_frame_backwards;
1149 if (NULL != sf->next_block) {
1150 struct btrfs_header *const next_hdr =
1151 (struct btrfs_header *)
1152 sf->next_block_ctx.datav[0];
1154 next_stack = btrfsic_stack_frame_alloc();
1155 if (NULL == next_stack) {
1157 goto one_stack_frame_backwards;
1161 next_stack->block = sf->next_block;
1162 next_stack->block_ctx = &sf->next_block_ctx;
1163 next_stack->next_block = NULL;
1164 next_stack->hdr = next_hdr;
1165 next_stack->limit_nesting =
1166 sf->limit_nesting - 1;
1167 next_stack->prev = sf;
1169 goto continue_with_new_stack_frame;
1172 goto continue_with_current_node_stack_frame;
1176 one_stack_frame_backwards:
1177 if (NULL != sf->prev) {
1178 struct btrfsic_stack_frame *const prev = sf->prev;
1180 /* the one for the initial block is freed in the caller */
1181 btrfsic_release_block_ctx(sf->block_ctx);
1184 prev->error = sf->error;
1185 btrfsic_stack_frame_free(sf);
1187 goto one_stack_frame_backwards;
1190 btrfsic_stack_frame_free(sf);
1192 goto continue_with_new_stack_frame;
1194 BUG_ON(&initial_stack_frame != sf);
1200 static void btrfsic_read_from_block_data(
1201 struct btrfsic_block_data_ctx *block_ctx,
1202 void *dstv, u32 offset, size_t len)
1205 size_t offset_in_page;
1207 char *dst = (char *)dstv;
1208 size_t start_offset = block_ctx->start & ((u64)PAGE_SIZE - 1);
1209 unsigned long i = (start_offset + offset) >> PAGE_SHIFT;
1211 WARN_ON(offset + len > block_ctx->len);
1212 offset_in_page = (start_offset + offset) & (PAGE_SIZE - 1);
1215 cur = min(len, ((size_t)PAGE_SIZE - offset_in_page));
1216 BUG_ON(i >= DIV_ROUND_UP(block_ctx->len, PAGE_SIZE));
1217 kaddr = block_ctx->datav[i];
1218 memcpy(dst, kaddr + offset_in_page, cur);
1227 static int btrfsic_create_link_to_next_block(
1228 struct btrfsic_state *state,
1229 struct btrfsic_block *block,
1230 struct btrfsic_block_data_ctx *block_ctx,
1233 struct btrfsic_block_data_ctx *next_block_ctx,
1234 struct btrfsic_block **next_blockp,
1235 int force_iodone_flag,
1236 int *num_copiesp, int *mirror_nump,
1237 struct btrfs_disk_key *disk_key,
1238 u64 parent_generation)
1240 struct btrfs_fs_info *fs_info = state->fs_info;
1241 struct btrfsic_block *next_block = NULL;
1243 struct btrfsic_block_link *l;
1244 int did_alloc_block_link;
1245 int block_was_created;
1247 *next_blockp = NULL;
1248 if (0 == *num_copiesp) {
1249 *num_copiesp = btrfs_num_copies(fs_info, next_bytenr,
1250 state->metablock_size);
1251 if (state->print_mask & BTRFSIC_PRINT_MASK_NUM_COPIES)
1252 pr_info("num_copies(log_bytenr=%llu) = %d\n",
1253 next_bytenr, *num_copiesp);
1257 if (*mirror_nump > *num_copiesp)
1260 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1261 pr_info("btrfsic_create_link_to_next_block(mirror_num=%d)\n",
1263 ret = btrfsic_map_block(state, next_bytenr,
1264 state->metablock_size,
1265 next_block_ctx, *mirror_nump);
1267 pr_info("btrfsic: btrfsic_map_block(@%llu, mirror=%d) failed!\n",
1268 next_bytenr, *mirror_nump);
1269 btrfsic_release_block_ctx(next_block_ctx);
1270 *next_blockp = NULL;
1274 next_block = btrfsic_block_lookup_or_add(state,
1275 next_block_ctx, "referenced ",
1276 1, force_iodone_flag,
1279 &block_was_created);
1280 if (NULL == next_block) {
1281 btrfsic_release_block_ctx(next_block_ctx);
1282 *next_blockp = NULL;
1285 if (block_was_created) {
1287 next_block->generation = BTRFSIC_GENERATION_UNKNOWN;
1289 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE) {
1290 if (next_block->logical_bytenr != next_bytenr &&
1291 !(!next_block->is_metadata &&
1292 0 == next_block->logical_bytenr))
1293 pr_info("Referenced block @%llu (%s/%llu/%d) found in hash table, %c, bytenr mismatch (!= stored %llu).\n",
1294 next_bytenr, next_block_ctx->dev->name,
1295 next_block_ctx->dev_bytenr, *mirror_nump,
1296 btrfsic_get_block_type(state,
1298 next_block->logical_bytenr);
1300 pr_info("Referenced block @%llu (%s/%llu/%d) found in hash table, %c.\n",
1301 next_bytenr, next_block_ctx->dev->name,
1302 next_block_ctx->dev_bytenr, *mirror_nump,
1303 btrfsic_get_block_type(state,
1306 next_block->logical_bytenr = next_bytenr;
1308 next_block->mirror_num = *mirror_nump;
1309 l = btrfsic_block_link_hashtable_lookup(
1310 next_block_ctx->dev->bdev,
1311 next_block_ctx->dev_bytenr,
1312 block_ctx->dev->bdev,
1313 block_ctx->dev_bytenr,
1314 &state->block_link_hashtable);
1317 next_block->disk_key = *disk_key;
1319 l = btrfsic_block_link_alloc();
1321 pr_info("btrfsic: error, kmalloc failed!\n");
1322 btrfsic_release_block_ctx(next_block_ctx);
1323 *next_blockp = NULL;
1327 did_alloc_block_link = 1;
1328 l->block_ref_to = next_block;
1329 l->block_ref_from = block;
1331 l->parent_generation = parent_generation;
1333 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1334 btrfsic_print_add_link(state, l);
1336 list_add(&l->node_ref_to, &block->ref_to_list);
1337 list_add(&l->node_ref_from, &next_block->ref_from_list);
1339 btrfsic_block_link_hashtable_add(l,
1340 &state->block_link_hashtable);
1342 did_alloc_block_link = 0;
1343 if (0 == limit_nesting) {
1345 l->parent_generation = parent_generation;
1346 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1347 btrfsic_print_add_link(state, l);
1351 if (limit_nesting > 0 && did_alloc_block_link) {
1352 ret = btrfsic_read_block(state, next_block_ctx);
1353 if (ret < (int)next_block_ctx->len) {
1354 pr_info("btrfsic: read block @logical %llu failed!\n",
1356 btrfsic_release_block_ctx(next_block_ctx);
1357 *next_blockp = NULL;
1361 *next_blockp = next_block;
1363 *next_blockp = NULL;
1370 static int btrfsic_handle_extent_data(
1371 struct btrfsic_state *state,
1372 struct btrfsic_block *block,
1373 struct btrfsic_block_data_ctx *block_ctx,
1374 u32 item_offset, int force_iodone_flag)
1376 struct btrfs_fs_info *fs_info = state->fs_info;
1377 struct btrfs_file_extent_item file_extent_item;
1378 u64 file_extent_item_offset;
1382 struct btrfsic_block_link *l;
1385 file_extent_item_offset = offsetof(struct btrfs_leaf, items) +
1387 if (file_extent_item_offset +
1388 offsetof(struct btrfs_file_extent_item, disk_num_bytes) >
1390 pr_info("btrfsic: file item out of bounce at logical %llu, dev %s\n",
1391 block_ctx->start, block_ctx->dev->name);
1395 btrfsic_read_from_block_data(block_ctx, &file_extent_item,
1396 file_extent_item_offset,
1397 offsetof(struct btrfs_file_extent_item, disk_num_bytes));
1398 if (BTRFS_FILE_EXTENT_REG != file_extent_item.type ||
1399 btrfs_stack_file_extent_disk_bytenr(&file_extent_item) == 0) {
1400 if (state->print_mask & BTRFSIC_PRINT_MASK_VERY_VERBOSE)
1401 pr_info("extent_data: type %u, disk_bytenr = %llu\n",
1402 file_extent_item.type,
1403 btrfs_stack_file_extent_disk_bytenr(
1404 &file_extent_item));
1408 if (file_extent_item_offset + sizeof(struct btrfs_file_extent_item) >
1410 pr_info("btrfsic: file item out of bounce at logical %llu, dev %s\n",
1411 block_ctx->start, block_ctx->dev->name);
1414 btrfsic_read_from_block_data(block_ctx, &file_extent_item,
1415 file_extent_item_offset,
1416 sizeof(struct btrfs_file_extent_item));
1417 next_bytenr = btrfs_stack_file_extent_disk_bytenr(&file_extent_item);
1418 if (btrfs_stack_file_extent_compression(&file_extent_item) ==
1419 BTRFS_COMPRESS_NONE) {
1420 next_bytenr += btrfs_stack_file_extent_offset(&file_extent_item);
1421 num_bytes = btrfs_stack_file_extent_num_bytes(&file_extent_item);
1423 num_bytes = btrfs_stack_file_extent_disk_num_bytes(&file_extent_item);
1425 generation = btrfs_stack_file_extent_generation(&file_extent_item);
1427 if (state->print_mask & BTRFSIC_PRINT_MASK_VERY_VERBOSE)
1428 pr_info("extent_data: type %u, disk_bytenr = %llu, offset = %llu, num_bytes = %llu\n",
1429 file_extent_item.type,
1430 btrfs_stack_file_extent_disk_bytenr(&file_extent_item),
1431 btrfs_stack_file_extent_offset(&file_extent_item),
1433 while (num_bytes > 0) {
1438 if (num_bytes > state->datablock_size)
1439 chunk_len = state->datablock_size;
1441 chunk_len = num_bytes;
1443 num_copies = btrfs_num_copies(fs_info, next_bytenr,
1444 state->datablock_size);
1445 if (state->print_mask & BTRFSIC_PRINT_MASK_NUM_COPIES)
1446 pr_info("num_copies(log_bytenr=%llu) = %d\n",
1447 next_bytenr, num_copies);
1448 for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
1449 struct btrfsic_block_data_ctx next_block_ctx;
1450 struct btrfsic_block *next_block;
1451 int block_was_created;
1453 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1454 pr_info("btrfsic_handle_extent_data(mirror_num=%d)\n",
1456 if (state->print_mask & BTRFSIC_PRINT_MASK_VERY_VERBOSE)
1457 pr_info("\tdisk_bytenr = %llu, num_bytes %u\n",
1458 next_bytenr, chunk_len);
1459 ret = btrfsic_map_block(state, next_bytenr,
1460 chunk_len, &next_block_ctx,
1463 pr_info("btrfsic: btrfsic_map_block(@%llu, mirror=%d) failed!\n",
1464 next_bytenr, mirror_num);
1468 next_block = btrfsic_block_lookup_or_add(
1476 &block_was_created);
1477 if (NULL == next_block) {
1478 pr_info("btrfsic: error, kmalloc failed!\n");
1479 btrfsic_release_block_ctx(&next_block_ctx);
1482 if (!block_was_created) {
1483 if ((state->print_mask &
1484 BTRFSIC_PRINT_MASK_VERBOSE) &&
1485 next_block->logical_bytenr != next_bytenr &&
1486 !(!next_block->is_metadata &&
1487 0 == next_block->logical_bytenr)) {
1488 pr_info("Referenced block @%llu (%s/%llu/%d) found in hash table, D, bytenr mismatch (!= stored %llu).\n",
1490 next_block_ctx.dev->name,
1491 next_block_ctx.dev_bytenr,
1493 next_block->logical_bytenr);
1495 next_block->logical_bytenr = next_bytenr;
1496 next_block->mirror_num = mirror_num;
1499 l = btrfsic_block_link_lookup_or_add(state,
1503 btrfsic_release_block_ctx(&next_block_ctx);
1508 next_bytenr += chunk_len;
1509 num_bytes -= chunk_len;
1515 static int btrfsic_map_block(struct btrfsic_state *state, u64 bytenr, u32 len,
1516 struct btrfsic_block_data_ctx *block_ctx_out,
1519 struct btrfs_fs_info *fs_info = state->fs_info;
1522 struct btrfs_bio *multi = NULL;
1523 struct btrfs_device *device;
1526 ret = btrfs_map_block(fs_info, BTRFS_MAP_READ,
1527 bytenr, &length, &multi, mirror_num);
1530 block_ctx_out->start = 0;
1531 block_ctx_out->dev_bytenr = 0;
1532 block_ctx_out->len = 0;
1533 block_ctx_out->dev = NULL;
1534 block_ctx_out->datav = NULL;
1535 block_ctx_out->pagev = NULL;
1536 block_ctx_out->mem_to_free = NULL;
1541 device = multi->stripes[0].dev;
1542 block_ctx_out->dev = btrfsic_dev_state_lookup(device->bdev->bd_dev);
1543 block_ctx_out->dev_bytenr = multi->stripes[0].physical;
1544 block_ctx_out->start = bytenr;
1545 block_ctx_out->len = len;
1546 block_ctx_out->datav = NULL;
1547 block_ctx_out->pagev = NULL;
1548 block_ctx_out->mem_to_free = NULL;
1551 if (NULL == block_ctx_out->dev) {
1553 pr_info("btrfsic: error, cannot lookup dev (#1)!\n");
1559 static void btrfsic_release_block_ctx(struct btrfsic_block_data_ctx *block_ctx)
1561 if (block_ctx->mem_to_free) {
1562 unsigned int num_pages;
1564 BUG_ON(!block_ctx->datav);
1565 BUG_ON(!block_ctx->pagev);
1566 num_pages = (block_ctx->len + (u64)PAGE_SIZE - 1) >>
1568 while (num_pages > 0) {
1570 if (block_ctx->datav[num_pages]) {
1571 kunmap(block_ctx->pagev[num_pages]);
1572 block_ctx->datav[num_pages] = NULL;
1574 if (block_ctx->pagev[num_pages]) {
1575 __free_page(block_ctx->pagev[num_pages]);
1576 block_ctx->pagev[num_pages] = NULL;
1580 kfree(block_ctx->mem_to_free);
1581 block_ctx->mem_to_free = NULL;
1582 block_ctx->pagev = NULL;
1583 block_ctx->datav = NULL;
1587 static int btrfsic_read_block(struct btrfsic_state *state,
1588 struct btrfsic_block_data_ctx *block_ctx)
1590 unsigned int num_pages;
1595 BUG_ON(block_ctx->datav);
1596 BUG_ON(block_ctx->pagev);
1597 BUG_ON(block_ctx->mem_to_free);
1598 if (block_ctx->dev_bytenr & ((u64)PAGE_SIZE - 1)) {
1599 pr_info("btrfsic: read_block() with unaligned bytenr %llu\n",
1600 block_ctx->dev_bytenr);
1604 num_pages = (block_ctx->len + (u64)PAGE_SIZE - 1) >>
1606 block_ctx->mem_to_free = kcalloc(sizeof(*block_ctx->datav) +
1607 sizeof(*block_ctx->pagev),
1608 num_pages, GFP_NOFS);
1609 if (!block_ctx->mem_to_free)
1611 block_ctx->datav = block_ctx->mem_to_free;
1612 block_ctx->pagev = (struct page **)(block_ctx->datav + num_pages);
1613 for (i = 0; i < num_pages; i++) {
1614 block_ctx->pagev[i] = alloc_page(GFP_NOFS);
1615 if (!block_ctx->pagev[i])
1619 dev_bytenr = block_ctx->dev_bytenr;
1620 for (i = 0; i < num_pages;) {
1624 bio = btrfs_io_bio_alloc(num_pages - i);
1625 bio_set_dev(bio, block_ctx->dev->bdev);
1626 bio->bi_iter.bi_sector = dev_bytenr >> 9;
1627 bio_set_op_attrs(bio, REQ_OP_READ, 0);
1629 for (j = i; j < num_pages; j++) {
1630 ret = bio_add_page(bio, block_ctx->pagev[j],
1632 if (PAGE_SIZE != ret)
1636 pr_info("btrfsic: error, failed to add a single page!\n");
1639 if (submit_bio_wait(bio)) {
1640 pr_info("btrfsic: read error at logical %llu dev %s!\n",
1641 block_ctx->start, block_ctx->dev->name);
1646 dev_bytenr += (j - i) * PAGE_SIZE;
1649 for (i = 0; i < num_pages; i++)
1650 block_ctx->datav[i] = kmap(block_ctx->pagev[i]);
1652 return block_ctx->len;
1655 static void btrfsic_dump_database(struct btrfsic_state *state)
1657 const struct btrfsic_block *b_all;
1659 BUG_ON(NULL == state);
1661 pr_info("all_blocks_list:\n");
1662 list_for_each_entry(b_all, &state->all_blocks_list, all_blocks_node) {
1663 const struct btrfsic_block_link *l;
1665 pr_info("%c-block @%llu (%s/%llu/%d)\n",
1666 btrfsic_get_block_type(state, b_all),
1667 b_all->logical_bytenr, b_all->dev_state->name,
1668 b_all->dev_bytenr, b_all->mirror_num);
1670 list_for_each_entry(l, &b_all->ref_to_list, node_ref_to) {
1671 pr_info(" %c @%llu (%s/%llu/%d) refers %u* to %c @%llu (%s/%llu/%d)\n",
1672 btrfsic_get_block_type(state, b_all),
1673 b_all->logical_bytenr, b_all->dev_state->name,
1674 b_all->dev_bytenr, b_all->mirror_num,
1676 btrfsic_get_block_type(state, l->block_ref_to),
1677 l->block_ref_to->logical_bytenr,
1678 l->block_ref_to->dev_state->name,
1679 l->block_ref_to->dev_bytenr,
1680 l->block_ref_to->mirror_num);
1683 list_for_each_entry(l, &b_all->ref_from_list, node_ref_from) {
1684 pr_info(" %c @%llu (%s/%llu/%d) is ref %u* from %c @%llu (%s/%llu/%d)\n",
1685 btrfsic_get_block_type(state, b_all),
1686 b_all->logical_bytenr, b_all->dev_state->name,
1687 b_all->dev_bytenr, b_all->mirror_num,
1689 btrfsic_get_block_type(state, l->block_ref_from),
1690 l->block_ref_from->logical_bytenr,
1691 l->block_ref_from->dev_state->name,
1692 l->block_ref_from->dev_bytenr,
1693 l->block_ref_from->mirror_num);
1701 * Test whether the disk block contains a tree block (leaf or node)
1702 * (note that this test fails for the super block)
1704 static int btrfsic_test_for_metadata(struct btrfsic_state *state,
1705 char **datav, unsigned int num_pages)
1707 struct btrfs_fs_info *fs_info = state->fs_info;
1708 struct btrfs_header *h;
1709 u8 csum[BTRFS_CSUM_SIZE];
1713 if (num_pages * PAGE_SIZE < state->metablock_size)
1714 return 1; /* not metadata */
1715 num_pages = state->metablock_size >> PAGE_SHIFT;
1716 h = (struct btrfs_header *)datav[0];
1718 if (memcmp(h->fsid, fs_info->fsid, BTRFS_FSID_SIZE))
1721 for (i = 0; i < num_pages; i++) {
1722 u8 *data = i ? datav[i] : (datav[i] + BTRFS_CSUM_SIZE);
1723 size_t sublen = i ? PAGE_SIZE :
1724 (PAGE_SIZE - BTRFS_CSUM_SIZE);
1726 crc = crc32c(crc, data, sublen);
1728 btrfs_csum_final(crc, csum);
1729 if (memcmp(csum, h->csum, state->csum_size))
1732 return 0; /* is metadata */
1735 static void btrfsic_process_written_block(struct btrfsic_dev_state *dev_state,
1736 u64 dev_bytenr, char **mapped_datav,
1737 unsigned int num_pages,
1738 struct bio *bio, int *bio_is_patched,
1739 struct buffer_head *bh,
1740 int submit_bio_bh_rw)
1743 struct btrfsic_block *block;
1744 struct btrfsic_block_data_ctx block_ctx;
1746 struct btrfsic_state *state = dev_state->state;
1747 struct block_device *bdev = dev_state->bdev;
1748 unsigned int processed_len;
1750 if (NULL != bio_is_patched)
1751 *bio_is_patched = 0;
1758 is_metadata = (0 == btrfsic_test_for_metadata(state, mapped_datav,
1761 block = btrfsic_block_hashtable_lookup(bdev, dev_bytenr,
1762 &state->block_hashtable);
1763 if (NULL != block) {
1765 struct btrfsic_block_link *l, *tmp;
1767 if (block->is_superblock) {
1768 bytenr = btrfs_super_bytenr((struct btrfs_super_block *)
1770 if (num_pages * PAGE_SIZE <
1771 BTRFS_SUPER_INFO_SIZE) {
1772 pr_info("btrfsic: cannot work with too short bios!\n");
1776 BUG_ON(BTRFS_SUPER_INFO_SIZE & (PAGE_SIZE - 1));
1777 processed_len = BTRFS_SUPER_INFO_SIZE;
1778 if (state->print_mask &
1779 BTRFSIC_PRINT_MASK_TREE_BEFORE_SB_WRITE) {
1780 pr_info("[before new superblock is written]:\n");
1781 btrfsic_dump_tree_sub(state, block, 0);
1785 if (!block->is_superblock) {
1786 if (num_pages * PAGE_SIZE <
1787 state->metablock_size) {
1788 pr_info("btrfsic: cannot work with too short bios!\n");
1791 processed_len = state->metablock_size;
1792 bytenr = btrfs_stack_header_bytenr(
1793 (struct btrfs_header *)
1795 btrfsic_cmp_log_and_dev_bytenr(state, bytenr,
1799 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE) {
1800 if (block->logical_bytenr != bytenr &&
1801 !(!block->is_metadata &&
1802 block->logical_bytenr == 0))
1803 pr_info("Written block @%llu (%s/%llu/%d) found in hash table, %c, bytenr mismatch (!= stored %llu).\n",
1804 bytenr, dev_state->name,
1807 btrfsic_get_block_type(state,
1809 block->logical_bytenr);
1811 pr_info("Written block @%llu (%s/%llu/%d) found in hash table, %c.\n",
1812 bytenr, dev_state->name,
1813 dev_bytenr, block->mirror_num,
1814 btrfsic_get_block_type(state,
1817 block->logical_bytenr = bytenr;
1819 if (num_pages * PAGE_SIZE <
1820 state->datablock_size) {
1821 pr_info("btrfsic: cannot work with too short bios!\n");
1824 processed_len = state->datablock_size;
1825 bytenr = block->logical_bytenr;
1826 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1827 pr_info("Written block @%llu (%s/%llu/%d) found in hash table, %c.\n",
1828 bytenr, dev_state->name, dev_bytenr,
1830 btrfsic_get_block_type(state, block));
1833 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1834 pr_info("ref_to_list: %cE, ref_from_list: %cE\n",
1835 list_empty(&block->ref_to_list) ? ' ' : '!',
1836 list_empty(&block->ref_from_list) ? ' ' : '!');
1837 if (btrfsic_is_block_ref_by_superblock(state, block, 0)) {
1838 pr_info("btrfs: attempt to overwrite %c-block @%llu (%s/%llu/%d), old(gen=%llu, objectid=%llu, type=%d, offset=%llu), new(gen=%llu), which is referenced by most recent superblock (superblockgen=%llu)!\n",
1839 btrfsic_get_block_type(state, block), bytenr,
1840 dev_state->name, dev_bytenr, block->mirror_num,
1842 btrfs_disk_key_objectid(&block->disk_key),
1843 block->disk_key.type,
1844 btrfs_disk_key_offset(&block->disk_key),
1845 btrfs_stack_header_generation(
1846 (struct btrfs_header *) mapped_datav[0]),
1847 state->max_superblock_generation);
1848 btrfsic_dump_tree(state);
1851 if (!block->is_iodone && !block->never_written) {
1852 pr_info("btrfs: attempt to overwrite %c-block @%llu (%s/%llu/%d), oldgen=%llu, newgen=%llu, which is not yet iodone!\n",
1853 btrfsic_get_block_type(state, block), bytenr,
1854 dev_state->name, dev_bytenr, block->mirror_num,
1856 btrfs_stack_header_generation(
1857 (struct btrfs_header *)
1859 /* it would not be safe to go on */
1860 btrfsic_dump_tree(state);
1865 * Clear all references of this block. Do not free
1866 * the block itself even if is not referenced anymore
1867 * because it still carries valuable information
1868 * like whether it was ever written and IO completed.
1870 list_for_each_entry_safe(l, tmp, &block->ref_to_list,
1872 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1873 btrfsic_print_rem_link(state, l);
1875 if (0 == l->ref_cnt) {
1876 list_del(&l->node_ref_to);
1877 list_del(&l->node_ref_from);
1878 btrfsic_block_link_hashtable_remove(l);
1879 btrfsic_block_link_free(l);
1883 block_ctx.dev = dev_state;
1884 block_ctx.dev_bytenr = dev_bytenr;
1885 block_ctx.start = bytenr;
1886 block_ctx.len = processed_len;
1887 block_ctx.pagev = NULL;
1888 block_ctx.mem_to_free = NULL;
1889 block_ctx.datav = mapped_datav;
1891 if (is_metadata || state->include_extent_data) {
1892 block->never_written = 0;
1893 block->iodone_w_error = 0;
1895 block->is_iodone = 0;
1896 BUG_ON(NULL == bio_is_patched);
1897 if (!*bio_is_patched) {
1898 block->orig_bio_bh_private =
1900 block->orig_bio_bh_end_io.bio =
1902 block->next_in_same_bio = NULL;
1903 bio->bi_private = block;
1904 bio->bi_end_io = btrfsic_bio_end_io;
1905 *bio_is_patched = 1;
1907 struct btrfsic_block *chained_block =
1908 (struct btrfsic_block *)
1911 BUG_ON(NULL == chained_block);
1912 block->orig_bio_bh_private =
1913 chained_block->orig_bio_bh_private;
1914 block->orig_bio_bh_end_io.bio =
1915 chained_block->orig_bio_bh_end_io.
1917 block->next_in_same_bio = chained_block;
1918 bio->bi_private = block;
1920 } else if (NULL != bh) {
1921 block->is_iodone = 0;
1922 block->orig_bio_bh_private = bh->b_private;
1923 block->orig_bio_bh_end_io.bh = bh->b_end_io;
1924 block->next_in_same_bio = NULL;
1925 bh->b_private = block;
1926 bh->b_end_io = btrfsic_bh_end_io;
1928 block->is_iodone = 1;
1929 block->orig_bio_bh_private = NULL;
1930 block->orig_bio_bh_end_io.bio = NULL;
1931 block->next_in_same_bio = NULL;
1935 block->flush_gen = dev_state->last_flush_gen + 1;
1936 block->submit_bio_bh_rw = submit_bio_bh_rw;
1938 block->logical_bytenr = bytenr;
1939 block->is_metadata = 1;
1940 if (block->is_superblock) {
1942 BTRFS_SUPER_INFO_SIZE);
1943 ret = btrfsic_process_written_superblock(
1946 (struct btrfs_super_block *)
1948 if (state->print_mask &
1949 BTRFSIC_PRINT_MASK_TREE_AFTER_SB_WRITE) {
1950 pr_info("[after new superblock is written]:\n");
1951 btrfsic_dump_tree_sub(state, block, 0);
1954 block->mirror_num = 0; /* unknown */
1955 ret = btrfsic_process_metablock(
1962 pr_info("btrfsic: btrfsic_process_metablock(root @%llu) failed!\n",
1965 block->is_metadata = 0;
1966 block->mirror_num = 0; /* unknown */
1967 block->generation = BTRFSIC_GENERATION_UNKNOWN;
1968 if (!state->include_extent_data
1969 && list_empty(&block->ref_from_list)) {
1971 * disk block is overwritten with extent
1972 * data (not meta data) and we are configured
1973 * to not include extent data: take the
1974 * chance and free the block's memory
1976 btrfsic_block_hashtable_remove(block);
1977 list_del(&block->all_blocks_node);
1978 btrfsic_block_free(block);
1981 btrfsic_release_block_ctx(&block_ctx);
1983 /* block has not been found in hash table */
1987 processed_len = state->datablock_size;
1988 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
1989 pr_info("Written block (%s/%llu/?) !found in hash table, D.\n",
1990 dev_state->name, dev_bytenr);
1991 if (!state->include_extent_data) {
1992 /* ignore that written D block */
1996 /* this is getting ugly for the
1997 * include_extent_data case... */
1998 bytenr = 0; /* unknown */
2000 processed_len = state->metablock_size;
2001 bytenr = btrfs_stack_header_bytenr(
2002 (struct btrfs_header *)
2004 btrfsic_cmp_log_and_dev_bytenr(state, bytenr, dev_state,
2006 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2007 pr_info("Written block @%llu (%s/%llu/?) !found in hash table, M.\n",
2008 bytenr, dev_state->name, dev_bytenr);
2011 block_ctx.dev = dev_state;
2012 block_ctx.dev_bytenr = dev_bytenr;
2013 block_ctx.start = bytenr;
2014 block_ctx.len = processed_len;
2015 block_ctx.pagev = NULL;
2016 block_ctx.mem_to_free = NULL;
2017 block_ctx.datav = mapped_datav;
2019 block = btrfsic_block_alloc();
2020 if (NULL == block) {
2021 pr_info("btrfsic: error, kmalloc failed!\n");
2022 btrfsic_release_block_ctx(&block_ctx);
2025 block->dev_state = dev_state;
2026 block->dev_bytenr = dev_bytenr;
2027 block->logical_bytenr = bytenr;
2028 block->is_metadata = is_metadata;
2029 block->never_written = 0;
2030 block->iodone_w_error = 0;
2031 block->mirror_num = 0; /* unknown */
2032 block->flush_gen = dev_state->last_flush_gen + 1;
2033 block->submit_bio_bh_rw = submit_bio_bh_rw;
2035 block->is_iodone = 0;
2036 BUG_ON(NULL == bio_is_patched);
2037 if (!*bio_is_patched) {
2038 block->orig_bio_bh_private = bio->bi_private;
2039 block->orig_bio_bh_end_io.bio = bio->bi_end_io;
2040 block->next_in_same_bio = NULL;
2041 bio->bi_private = block;
2042 bio->bi_end_io = btrfsic_bio_end_io;
2043 *bio_is_patched = 1;
2045 struct btrfsic_block *chained_block =
2046 (struct btrfsic_block *)
2049 BUG_ON(NULL == chained_block);
2050 block->orig_bio_bh_private =
2051 chained_block->orig_bio_bh_private;
2052 block->orig_bio_bh_end_io.bio =
2053 chained_block->orig_bio_bh_end_io.bio;
2054 block->next_in_same_bio = chained_block;
2055 bio->bi_private = block;
2057 } else if (NULL != bh) {
2058 block->is_iodone = 0;
2059 block->orig_bio_bh_private = bh->b_private;
2060 block->orig_bio_bh_end_io.bh = bh->b_end_io;
2061 block->next_in_same_bio = NULL;
2062 bh->b_private = block;
2063 bh->b_end_io = btrfsic_bh_end_io;
2065 block->is_iodone = 1;
2066 block->orig_bio_bh_private = NULL;
2067 block->orig_bio_bh_end_io.bio = NULL;
2068 block->next_in_same_bio = NULL;
2070 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2071 pr_info("New written %c-block @%llu (%s/%llu/%d)\n",
2072 is_metadata ? 'M' : 'D',
2073 block->logical_bytenr, block->dev_state->name,
2074 block->dev_bytenr, block->mirror_num);
2075 list_add(&block->all_blocks_node, &state->all_blocks_list);
2076 btrfsic_block_hashtable_add(block, &state->block_hashtable);
2079 ret = btrfsic_process_metablock(state, block,
2082 pr_info("btrfsic: process_metablock(root @%llu) failed!\n",
2085 btrfsic_release_block_ctx(&block_ctx);
2089 BUG_ON(!processed_len);
2090 dev_bytenr += processed_len;
2091 mapped_datav += processed_len >> PAGE_SHIFT;
2092 num_pages -= processed_len >> PAGE_SHIFT;
2096 static void btrfsic_bio_end_io(struct bio *bp)
2098 struct btrfsic_block *block = (struct btrfsic_block *)bp->bi_private;
2101 /* mutex is not held! This is not save if IO is not yet completed
2107 BUG_ON(NULL == block);
2108 bp->bi_private = block->orig_bio_bh_private;
2109 bp->bi_end_io = block->orig_bio_bh_end_io.bio;
2112 struct btrfsic_block *next_block;
2113 struct btrfsic_dev_state *const dev_state = block->dev_state;
2115 if ((dev_state->state->print_mask &
2116 BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
2117 pr_info("bio_end_io(err=%d) for %c @%llu (%s/%llu/%d)\n",
2119 btrfsic_get_block_type(dev_state->state, block),
2120 block->logical_bytenr, dev_state->name,
2121 block->dev_bytenr, block->mirror_num);
2122 next_block = block->next_in_same_bio;
2123 block->iodone_w_error = iodone_w_error;
2124 if (block->submit_bio_bh_rw & REQ_PREFLUSH) {
2125 dev_state->last_flush_gen++;
2126 if ((dev_state->state->print_mask &
2127 BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
2128 pr_info("bio_end_io() new %s flush_gen=%llu\n",
2130 dev_state->last_flush_gen);
2132 if (block->submit_bio_bh_rw & REQ_FUA)
2133 block->flush_gen = 0; /* FUA completed means block is
2135 block->is_iodone = 1; /* for FLUSH, this releases the block */
2137 } while (NULL != block);
2142 static void btrfsic_bh_end_io(struct buffer_head *bh, int uptodate)
2144 struct btrfsic_block *block = (struct btrfsic_block *)bh->b_private;
2145 int iodone_w_error = !uptodate;
2146 struct btrfsic_dev_state *dev_state;
2148 BUG_ON(NULL == block);
2149 dev_state = block->dev_state;
2150 if ((dev_state->state->print_mask & BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
2151 pr_info("bh_end_io(error=%d) for %c @%llu (%s/%llu/%d)\n",
2153 btrfsic_get_block_type(dev_state->state, block),
2154 block->logical_bytenr, block->dev_state->name,
2155 block->dev_bytenr, block->mirror_num);
2157 block->iodone_w_error = iodone_w_error;
2158 if (block->submit_bio_bh_rw & REQ_PREFLUSH) {
2159 dev_state->last_flush_gen++;
2160 if ((dev_state->state->print_mask &
2161 BTRFSIC_PRINT_MASK_END_IO_BIO_BH))
2162 pr_info("bh_end_io() new %s flush_gen=%llu\n",
2163 dev_state->name, dev_state->last_flush_gen);
2165 if (block->submit_bio_bh_rw & REQ_FUA)
2166 block->flush_gen = 0; /* FUA completed means block is on disk */
2168 bh->b_private = block->orig_bio_bh_private;
2169 bh->b_end_io = block->orig_bio_bh_end_io.bh;
2170 block->is_iodone = 1; /* for FLUSH, this releases the block */
2171 bh->b_end_io(bh, uptodate);
2174 static int btrfsic_process_written_superblock(
2175 struct btrfsic_state *state,
2176 struct btrfsic_block *const superblock,
2177 struct btrfs_super_block *const super_hdr)
2179 struct btrfs_fs_info *fs_info = state->fs_info;
2182 superblock->generation = btrfs_super_generation(super_hdr);
2183 if (!(superblock->generation > state->max_superblock_generation ||
2184 0 == state->max_superblock_generation)) {
2185 if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
2186 pr_info("btrfsic: superblock @%llu (%s/%llu/%d) with old gen %llu <= %llu\n",
2187 superblock->logical_bytenr,
2188 superblock->dev_state->name,
2189 superblock->dev_bytenr, superblock->mirror_num,
2190 btrfs_super_generation(super_hdr),
2191 state->max_superblock_generation);
2193 if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
2194 pr_info("btrfsic: got new superblock @%llu (%s/%llu/%d) with new gen %llu > %llu\n",
2195 superblock->logical_bytenr,
2196 superblock->dev_state->name,
2197 superblock->dev_bytenr, superblock->mirror_num,
2198 btrfs_super_generation(super_hdr),
2199 state->max_superblock_generation);
2201 state->max_superblock_generation =
2202 btrfs_super_generation(super_hdr);
2203 state->latest_superblock = superblock;
2206 for (pass = 0; pass < 3; pass++) {
2209 struct btrfsic_block *next_block;
2210 struct btrfsic_block_data_ctx tmp_next_block_ctx;
2211 struct btrfsic_block_link *l;
2214 const char *additional_string = NULL;
2215 struct btrfs_disk_key tmp_disk_key = {0};
2217 btrfs_set_disk_key_objectid(&tmp_disk_key,
2218 BTRFS_ROOT_ITEM_KEY);
2219 btrfs_set_disk_key_objectid(&tmp_disk_key, 0);
2223 btrfs_set_disk_key_objectid(&tmp_disk_key,
2224 BTRFS_ROOT_TREE_OBJECTID);
2225 additional_string = "root ";
2226 next_bytenr = btrfs_super_root(super_hdr);
2227 if (state->print_mask &
2228 BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION)
2229 pr_info("root@%llu\n", next_bytenr);
2232 btrfs_set_disk_key_objectid(&tmp_disk_key,
2233 BTRFS_CHUNK_TREE_OBJECTID);
2234 additional_string = "chunk ";
2235 next_bytenr = btrfs_super_chunk_root(super_hdr);
2236 if (state->print_mask &
2237 BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION)
2238 pr_info("chunk@%llu\n", next_bytenr);
2241 btrfs_set_disk_key_objectid(&tmp_disk_key,
2242 BTRFS_TREE_LOG_OBJECTID);
2243 additional_string = "log ";
2244 next_bytenr = btrfs_super_log_root(super_hdr);
2245 if (0 == next_bytenr)
2247 if (state->print_mask &
2248 BTRFSIC_PRINT_MASK_ROOT_CHUNK_LOG_TREE_LOCATION)
2249 pr_info("log@%llu\n", next_bytenr);
2253 num_copies = btrfs_num_copies(fs_info, next_bytenr,
2254 BTRFS_SUPER_INFO_SIZE);
2255 if (state->print_mask & BTRFSIC_PRINT_MASK_NUM_COPIES)
2256 pr_info("num_copies(log_bytenr=%llu) = %d\n",
2257 next_bytenr, num_copies);
2258 for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
2261 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2262 pr_info("btrfsic_process_written_superblock(mirror_num=%d)\n", mirror_num);
2263 ret = btrfsic_map_block(state, next_bytenr,
2264 BTRFS_SUPER_INFO_SIZE,
2265 &tmp_next_block_ctx,
2268 pr_info("btrfsic: btrfsic_map_block(@%llu, mirror=%d) failed!\n",
2269 next_bytenr, mirror_num);
2273 next_block = btrfsic_block_lookup_or_add(
2275 &tmp_next_block_ctx,
2280 if (NULL == next_block) {
2281 pr_info("btrfsic: error, kmalloc failed!\n");
2282 btrfsic_release_block_ctx(&tmp_next_block_ctx);
2286 next_block->disk_key = tmp_disk_key;
2288 next_block->generation =
2289 BTRFSIC_GENERATION_UNKNOWN;
2290 l = btrfsic_block_link_lookup_or_add(
2292 &tmp_next_block_ctx,
2295 BTRFSIC_GENERATION_UNKNOWN);
2296 btrfsic_release_block_ctx(&tmp_next_block_ctx);
2302 if (WARN_ON(-1 == btrfsic_check_all_ref_blocks(state, superblock, 0)))
2303 btrfsic_dump_tree(state);
2308 static int btrfsic_check_all_ref_blocks(struct btrfsic_state *state,
2309 struct btrfsic_block *const block,
2310 int recursion_level)
2312 const struct btrfsic_block_link *l;
2315 if (recursion_level >= 3 + BTRFS_MAX_LEVEL) {
2317 * Note that this situation can happen and does not
2318 * indicate an error in regular cases. It happens
2319 * when disk blocks are freed and later reused.
2320 * The check-integrity module is not aware of any
2321 * block free operations, it just recognizes block
2322 * write operations. Therefore it keeps the linkage
2323 * information for a block until a block is
2324 * rewritten. This can temporarily cause incorrect
2325 * and even circular linkage informations. This
2326 * causes no harm unless such blocks are referenced
2327 * by the most recent super block.
2329 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2330 pr_info("btrfsic: abort cyclic linkage (case 1).\n");
2336 * This algorithm is recursive because the amount of used stack
2337 * space is very small and the max recursion depth is limited.
2339 list_for_each_entry(l, &block->ref_to_list, node_ref_to) {
2340 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2341 pr_info("rl=%d, %c @%llu (%s/%llu/%d) %u* refers to %c @%llu (%s/%llu/%d)\n",
2343 btrfsic_get_block_type(state, block),
2344 block->logical_bytenr, block->dev_state->name,
2345 block->dev_bytenr, block->mirror_num,
2347 btrfsic_get_block_type(state, l->block_ref_to),
2348 l->block_ref_to->logical_bytenr,
2349 l->block_ref_to->dev_state->name,
2350 l->block_ref_to->dev_bytenr,
2351 l->block_ref_to->mirror_num);
2352 if (l->block_ref_to->never_written) {
2353 pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which is never written!\n",
2354 btrfsic_get_block_type(state, l->block_ref_to),
2355 l->block_ref_to->logical_bytenr,
2356 l->block_ref_to->dev_state->name,
2357 l->block_ref_to->dev_bytenr,
2358 l->block_ref_to->mirror_num);
2360 } else if (!l->block_ref_to->is_iodone) {
2361 pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which is not yet iodone!\n",
2362 btrfsic_get_block_type(state, l->block_ref_to),
2363 l->block_ref_to->logical_bytenr,
2364 l->block_ref_to->dev_state->name,
2365 l->block_ref_to->dev_bytenr,
2366 l->block_ref_to->mirror_num);
2368 } else if (l->block_ref_to->iodone_w_error) {
2369 pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which has write error!\n",
2370 btrfsic_get_block_type(state, l->block_ref_to),
2371 l->block_ref_to->logical_bytenr,
2372 l->block_ref_to->dev_state->name,
2373 l->block_ref_to->dev_bytenr,
2374 l->block_ref_to->mirror_num);
2376 } else if (l->parent_generation !=
2377 l->block_ref_to->generation &&
2378 BTRFSIC_GENERATION_UNKNOWN !=
2379 l->parent_generation &&
2380 BTRFSIC_GENERATION_UNKNOWN !=
2381 l->block_ref_to->generation) {
2382 pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) with generation %llu != parent generation %llu!\n",
2383 btrfsic_get_block_type(state, l->block_ref_to),
2384 l->block_ref_to->logical_bytenr,
2385 l->block_ref_to->dev_state->name,
2386 l->block_ref_to->dev_bytenr,
2387 l->block_ref_to->mirror_num,
2388 l->block_ref_to->generation,
2389 l->parent_generation);
2391 } else if (l->block_ref_to->flush_gen >
2392 l->block_ref_to->dev_state->last_flush_gen) {
2393 pr_info("btrfs: attempt to write superblock which references block %c @%llu (%s/%llu/%d) which is not flushed out of disk's write cache (block flush_gen=%llu, dev->flush_gen=%llu)!\n",
2394 btrfsic_get_block_type(state, l->block_ref_to),
2395 l->block_ref_to->logical_bytenr,
2396 l->block_ref_to->dev_state->name,
2397 l->block_ref_to->dev_bytenr,
2398 l->block_ref_to->mirror_num, block->flush_gen,
2399 l->block_ref_to->dev_state->last_flush_gen);
2401 } else if (-1 == btrfsic_check_all_ref_blocks(state,
2412 static int btrfsic_is_block_ref_by_superblock(
2413 const struct btrfsic_state *state,
2414 const struct btrfsic_block *block,
2415 int recursion_level)
2417 const struct btrfsic_block_link *l;
2419 if (recursion_level >= 3 + BTRFS_MAX_LEVEL) {
2420 /* refer to comment at "abort cyclic linkage (case 1)" */
2421 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2422 pr_info("btrfsic: abort cyclic linkage (case 2).\n");
2428 * This algorithm is recursive because the amount of used stack space
2429 * is very small and the max recursion depth is limited.
2431 list_for_each_entry(l, &block->ref_from_list, node_ref_from) {
2432 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2433 pr_info("rl=%d, %c @%llu (%s/%llu/%d) is ref %u* from %c @%llu (%s/%llu/%d)\n",
2435 btrfsic_get_block_type(state, block),
2436 block->logical_bytenr, block->dev_state->name,
2437 block->dev_bytenr, block->mirror_num,
2439 btrfsic_get_block_type(state, l->block_ref_from),
2440 l->block_ref_from->logical_bytenr,
2441 l->block_ref_from->dev_state->name,
2442 l->block_ref_from->dev_bytenr,
2443 l->block_ref_from->mirror_num);
2444 if (l->block_ref_from->is_superblock &&
2445 state->latest_superblock->dev_bytenr ==
2446 l->block_ref_from->dev_bytenr &&
2447 state->latest_superblock->dev_state->bdev ==
2448 l->block_ref_from->dev_state->bdev)
2450 else if (btrfsic_is_block_ref_by_superblock(state,
2460 static void btrfsic_print_add_link(const struct btrfsic_state *state,
2461 const struct btrfsic_block_link *l)
2463 pr_info("Add %u* link from %c @%llu (%s/%llu/%d) to %c @%llu (%s/%llu/%d).\n",
2465 btrfsic_get_block_type(state, l->block_ref_from),
2466 l->block_ref_from->logical_bytenr,
2467 l->block_ref_from->dev_state->name,
2468 l->block_ref_from->dev_bytenr, l->block_ref_from->mirror_num,
2469 btrfsic_get_block_type(state, l->block_ref_to),
2470 l->block_ref_to->logical_bytenr,
2471 l->block_ref_to->dev_state->name, l->block_ref_to->dev_bytenr,
2472 l->block_ref_to->mirror_num);
2475 static void btrfsic_print_rem_link(const struct btrfsic_state *state,
2476 const struct btrfsic_block_link *l)
2478 pr_info("Rem %u* link from %c @%llu (%s/%llu/%d) to %c @%llu (%s/%llu/%d).\n",
2480 btrfsic_get_block_type(state, l->block_ref_from),
2481 l->block_ref_from->logical_bytenr,
2482 l->block_ref_from->dev_state->name,
2483 l->block_ref_from->dev_bytenr, l->block_ref_from->mirror_num,
2484 btrfsic_get_block_type(state, l->block_ref_to),
2485 l->block_ref_to->logical_bytenr,
2486 l->block_ref_to->dev_state->name, l->block_ref_to->dev_bytenr,
2487 l->block_ref_to->mirror_num);
2490 static char btrfsic_get_block_type(const struct btrfsic_state *state,
2491 const struct btrfsic_block *block)
2493 if (block->is_superblock &&
2494 state->latest_superblock->dev_bytenr == block->dev_bytenr &&
2495 state->latest_superblock->dev_state->bdev == block->dev_state->bdev)
2497 else if (block->is_superblock)
2499 else if (block->is_metadata)
2505 static void btrfsic_dump_tree(const struct btrfsic_state *state)
2507 btrfsic_dump_tree_sub(state, state->latest_superblock, 0);
2510 static void btrfsic_dump_tree_sub(const struct btrfsic_state *state,
2511 const struct btrfsic_block *block,
2514 const struct btrfsic_block_link *l;
2516 static char buf[80];
2517 int cursor_position;
2520 * Should better fill an on-stack buffer with a complete line and
2521 * dump it at once when it is time to print a newline character.
2525 * This algorithm is recursive because the amount of used stack space
2526 * is very small and the max recursion depth is limited.
2528 indent_add = sprintf(buf, "%c-%llu(%s/%llu/%u)",
2529 btrfsic_get_block_type(state, block),
2530 block->logical_bytenr, block->dev_state->name,
2531 block->dev_bytenr, block->mirror_num);
2532 if (indent_level + indent_add > BTRFSIC_TREE_DUMP_MAX_INDENT_LEVEL) {
2537 indent_level += indent_add;
2538 if (list_empty(&block->ref_to_list)) {
2542 if (block->mirror_num > 1 &&
2543 !(state->print_mask & BTRFSIC_PRINT_MASK_TREE_WITH_ALL_MIRRORS)) {
2548 cursor_position = indent_level;
2549 list_for_each_entry(l, &block->ref_to_list, node_ref_to) {
2550 while (cursor_position < indent_level) {
2555 indent_add = sprintf(buf, " %d*--> ", l->ref_cnt);
2557 indent_add = sprintf(buf, " --> ");
2558 if (indent_level + indent_add >
2559 BTRFSIC_TREE_DUMP_MAX_INDENT_LEVEL) {
2561 cursor_position = 0;
2567 btrfsic_dump_tree_sub(state, l->block_ref_to,
2568 indent_level + indent_add);
2569 cursor_position = 0;
2573 static struct btrfsic_block_link *btrfsic_block_link_lookup_or_add(
2574 struct btrfsic_state *state,
2575 struct btrfsic_block_data_ctx *next_block_ctx,
2576 struct btrfsic_block *next_block,
2577 struct btrfsic_block *from_block,
2578 u64 parent_generation)
2580 struct btrfsic_block_link *l;
2582 l = btrfsic_block_link_hashtable_lookup(next_block_ctx->dev->bdev,
2583 next_block_ctx->dev_bytenr,
2584 from_block->dev_state->bdev,
2585 from_block->dev_bytenr,
2586 &state->block_link_hashtable);
2588 l = btrfsic_block_link_alloc();
2590 pr_info("btrfsic: error, kmalloc failed!\n");
2594 l->block_ref_to = next_block;
2595 l->block_ref_from = from_block;
2597 l->parent_generation = parent_generation;
2599 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2600 btrfsic_print_add_link(state, l);
2602 list_add(&l->node_ref_to, &from_block->ref_to_list);
2603 list_add(&l->node_ref_from, &next_block->ref_from_list);
2605 btrfsic_block_link_hashtable_add(l,
2606 &state->block_link_hashtable);
2609 l->parent_generation = parent_generation;
2610 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2611 btrfsic_print_add_link(state, l);
2617 static struct btrfsic_block *btrfsic_block_lookup_or_add(
2618 struct btrfsic_state *state,
2619 struct btrfsic_block_data_ctx *block_ctx,
2620 const char *additional_string,
2627 struct btrfsic_block *block;
2629 block = btrfsic_block_hashtable_lookup(block_ctx->dev->bdev,
2630 block_ctx->dev_bytenr,
2631 &state->block_hashtable);
2632 if (NULL == block) {
2633 struct btrfsic_dev_state *dev_state;
2635 block = btrfsic_block_alloc();
2636 if (NULL == block) {
2637 pr_info("btrfsic: error, kmalloc failed!\n");
2640 dev_state = btrfsic_dev_state_lookup(block_ctx->dev->bdev->bd_dev);
2641 if (NULL == dev_state) {
2642 pr_info("btrfsic: error, lookup dev_state failed!\n");
2643 btrfsic_block_free(block);
2646 block->dev_state = dev_state;
2647 block->dev_bytenr = block_ctx->dev_bytenr;
2648 block->logical_bytenr = block_ctx->start;
2649 block->is_metadata = is_metadata;
2650 block->is_iodone = is_iodone;
2651 block->never_written = never_written;
2652 block->mirror_num = mirror_num;
2653 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
2654 pr_info("New %s%c-block @%llu (%s/%llu/%d)\n",
2656 btrfsic_get_block_type(state, block),
2657 block->logical_bytenr, dev_state->name,
2658 block->dev_bytenr, mirror_num);
2659 list_add(&block->all_blocks_node, &state->all_blocks_list);
2660 btrfsic_block_hashtable_add(block, &state->block_hashtable);
2661 if (NULL != was_created)
2664 if (NULL != was_created)
2671 static void btrfsic_cmp_log_and_dev_bytenr(struct btrfsic_state *state,
2673 struct btrfsic_dev_state *dev_state,
2676 struct btrfs_fs_info *fs_info = state->fs_info;
2677 struct btrfsic_block_data_ctx block_ctx;
2683 num_copies = btrfs_num_copies(fs_info, bytenr, state->metablock_size);
2685 for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
2686 ret = btrfsic_map_block(state, bytenr, state->metablock_size,
2687 &block_ctx, mirror_num);
2689 pr_info("btrfsic: btrfsic_map_block(logical @%llu, mirror %d) failed!\n",
2690 bytenr, mirror_num);
2694 if (dev_state->bdev == block_ctx.dev->bdev &&
2695 dev_bytenr == block_ctx.dev_bytenr) {
2697 btrfsic_release_block_ctx(&block_ctx);
2700 btrfsic_release_block_ctx(&block_ctx);
2703 if (WARN_ON(!match)) {
2704 pr_info("btrfs: attempt to write M-block which contains logical bytenr that doesn't map to dev+physical bytenr of submit_bio, buffer->log_bytenr=%llu, submit_bio(bdev=%s, phys_bytenr=%llu)!\n",
2705 bytenr, dev_state->name, dev_bytenr);
2706 for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
2707 ret = btrfsic_map_block(state, bytenr,
2708 state->metablock_size,
2709 &block_ctx, mirror_num);
2713 pr_info("Read logical bytenr @%llu maps to (%s/%llu/%d)\n",
2714 bytenr, block_ctx.dev->name,
2715 block_ctx.dev_bytenr, mirror_num);
2720 static struct btrfsic_dev_state *btrfsic_dev_state_lookup(dev_t dev)
2722 return btrfsic_dev_state_hashtable_lookup(dev,
2723 &btrfsic_dev_state_hashtable);
2726 int btrfsic_submit_bh(int op, int op_flags, struct buffer_head *bh)
2728 struct btrfsic_dev_state *dev_state;
2730 if (!btrfsic_is_initialized)
2731 return submit_bh(op, op_flags, bh);
2733 mutex_lock(&btrfsic_mutex);
2734 /* since btrfsic_submit_bh() might also be called before
2735 * btrfsic_mount(), this might return NULL */
2736 dev_state = btrfsic_dev_state_lookup(bh->b_bdev->bd_dev);
2738 /* Only called to write the superblock (incl. FLUSH/FUA) */
2739 if (NULL != dev_state &&
2740 (op == REQ_OP_WRITE) && bh->b_size > 0) {
2743 dev_bytenr = BTRFS_BDEV_BLOCKSIZE * bh->b_blocknr;
2744 if (dev_state->state->print_mask &
2745 BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
2746 pr_info("submit_bh(op=0x%x,0x%x, blocknr=%llu (bytenr %llu), size=%zu, data=%p, bdev=%p)\n",
2747 op, op_flags, (unsigned long long)bh->b_blocknr,
2748 dev_bytenr, bh->b_size, bh->b_data, bh->b_bdev);
2749 btrfsic_process_written_block(dev_state, dev_bytenr,
2750 &bh->b_data, 1, NULL,
2751 NULL, bh, op_flags);
2752 } else if (NULL != dev_state && (op_flags & REQ_PREFLUSH)) {
2753 if (dev_state->state->print_mask &
2754 BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
2755 pr_info("submit_bh(op=0x%x,0x%x FLUSH, bdev=%p)\n",
2756 op, op_flags, bh->b_bdev);
2757 if (!dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
2758 if ((dev_state->state->print_mask &
2759 (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
2760 BTRFSIC_PRINT_MASK_VERBOSE)))
2761 pr_info("btrfsic_submit_bh(%s) with FLUSH but dummy block already in use (ignored)!\n",
2764 struct btrfsic_block *const block =
2765 &dev_state->dummy_block_for_bio_bh_flush;
2767 block->is_iodone = 0;
2768 block->never_written = 0;
2769 block->iodone_w_error = 0;
2770 block->flush_gen = dev_state->last_flush_gen + 1;
2771 block->submit_bio_bh_rw = op_flags;
2772 block->orig_bio_bh_private = bh->b_private;
2773 block->orig_bio_bh_end_io.bh = bh->b_end_io;
2774 block->next_in_same_bio = NULL;
2775 bh->b_private = block;
2776 bh->b_end_io = btrfsic_bh_end_io;
2779 mutex_unlock(&btrfsic_mutex);
2780 return submit_bh(op, op_flags, bh);
2783 static void __btrfsic_submit_bio(struct bio *bio)
2785 struct btrfsic_dev_state *dev_state;
2787 if (!btrfsic_is_initialized)
2790 mutex_lock(&btrfsic_mutex);
2791 /* since btrfsic_submit_bio() is also called before
2792 * btrfsic_mount(), this might return NULL */
2793 dev_state = btrfsic_dev_state_lookup(bio_dev(bio) + bio->bi_partno);
2794 if (NULL != dev_state &&
2795 (bio_op(bio) == REQ_OP_WRITE) && bio_has_data(bio)) {
2799 struct bio_vec bvec;
2800 struct bvec_iter iter;
2802 char **mapped_datav;
2803 unsigned int segs = bio_segments(bio);
2805 dev_bytenr = 512 * bio->bi_iter.bi_sector;
2807 if (dev_state->state->print_mask &
2808 BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
2809 pr_info("submit_bio(rw=%d,0x%x, bi_vcnt=%u, bi_sector=%llu (bytenr %llu), bi_disk=%p)\n",
2810 bio_op(bio), bio->bi_opf, segs,
2811 (unsigned long long)bio->bi_iter.bi_sector,
2812 dev_bytenr, bio->bi_disk);
2814 mapped_datav = kmalloc_array(segs,
2815 sizeof(*mapped_datav), GFP_NOFS);
2818 cur_bytenr = dev_bytenr;
2820 bio_for_each_segment(bvec, bio, iter) {
2821 BUG_ON(bvec.bv_len != PAGE_SIZE);
2822 mapped_datav[i] = kmap(bvec.bv_page);
2825 if (dev_state->state->print_mask &
2826 BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH_VERBOSE)
2827 pr_info("#%u: bytenr=%llu, len=%u, offset=%u\n",
2828 i, cur_bytenr, bvec.bv_len, bvec.bv_offset);
2829 cur_bytenr += bvec.bv_len;
2831 btrfsic_process_written_block(dev_state, dev_bytenr,
2833 bio, &bio_is_patched,
2835 bio_for_each_segment(bvec, bio, iter)
2836 kunmap(bvec.bv_page);
2837 kfree(mapped_datav);
2838 } else if (NULL != dev_state && (bio->bi_opf & REQ_PREFLUSH)) {
2839 if (dev_state->state->print_mask &
2840 BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
2841 pr_info("submit_bio(rw=%d,0x%x FLUSH, disk=%p)\n",
2842 bio_op(bio), bio->bi_opf, bio->bi_disk);
2843 if (!dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
2844 if ((dev_state->state->print_mask &
2845 (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
2846 BTRFSIC_PRINT_MASK_VERBOSE)))
2847 pr_info("btrfsic_submit_bio(%s) with FLUSH but dummy block already in use (ignored)!\n",
2850 struct btrfsic_block *const block =
2851 &dev_state->dummy_block_for_bio_bh_flush;
2853 block->is_iodone = 0;
2854 block->never_written = 0;
2855 block->iodone_w_error = 0;
2856 block->flush_gen = dev_state->last_flush_gen + 1;
2857 block->submit_bio_bh_rw = bio->bi_opf;
2858 block->orig_bio_bh_private = bio->bi_private;
2859 block->orig_bio_bh_end_io.bio = bio->bi_end_io;
2860 block->next_in_same_bio = NULL;
2861 bio->bi_private = block;
2862 bio->bi_end_io = btrfsic_bio_end_io;
2866 mutex_unlock(&btrfsic_mutex);
2869 void btrfsic_submit_bio(struct bio *bio)
2871 __btrfsic_submit_bio(bio);
2875 int btrfsic_submit_bio_wait(struct bio *bio)
2877 __btrfsic_submit_bio(bio);
2878 return submit_bio_wait(bio);
2881 int btrfsic_mount(struct btrfs_fs_info *fs_info,
2882 struct btrfs_fs_devices *fs_devices,
2883 int including_extent_data, u32 print_mask)
2886 struct btrfsic_state *state;
2887 struct list_head *dev_head = &fs_devices->devices;
2888 struct btrfs_device *device;
2890 if (fs_info->nodesize & ((u64)PAGE_SIZE - 1)) {
2891 pr_info("btrfsic: cannot handle nodesize %d not being a multiple of PAGE_SIZE %ld!\n",
2892 fs_info->nodesize, PAGE_SIZE);
2895 if (fs_info->sectorsize & ((u64)PAGE_SIZE - 1)) {
2896 pr_info("btrfsic: cannot handle sectorsize %d not being a multiple of PAGE_SIZE %ld!\n",
2897 fs_info->sectorsize, PAGE_SIZE);
2900 state = kvzalloc(sizeof(*state), GFP_KERNEL);
2902 pr_info("btrfs check-integrity: allocation failed!\n");
2906 if (!btrfsic_is_initialized) {
2907 mutex_init(&btrfsic_mutex);
2908 btrfsic_dev_state_hashtable_init(&btrfsic_dev_state_hashtable);
2909 btrfsic_is_initialized = 1;
2911 mutex_lock(&btrfsic_mutex);
2912 state->fs_info = fs_info;
2913 state->print_mask = print_mask;
2914 state->include_extent_data = including_extent_data;
2915 state->csum_size = 0;
2916 state->metablock_size = fs_info->nodesize;
2917 state->datablock_size = fs_info->sectorsize;
2918 INIT_LIST_HEAD(&state->all_blocks_list);
2919 btrfsic_block_hashtable_init(&state->block_hashtable);
2920 btrfsic_block_link_hashtable_init(&state->block_link_hashtable);
2921 state->max_superblock_generation = 0;
2922 state->latest_superblock = NULL;
2924 list_for_each_entry(device, dev_head, dev_list) {
2925 struct btrfsic_dev_state *ds;
2928 if (!device->bdev || !device->name)
2931 ds = btrfsic_dev_state_alloc();
2933 pr_info("btrfs check-integrity: kmalloc() failed!\n");
2934 mutex_unlock(&btrfsic_mutex);
2937 ds->bdev = device->bdev;
2939 bdevname(ds->bdev, ds->name);
2940 ds->name[BDEVNAME_SIZE - 1] = '\0';
2941 p = kbasename(ds->name);
2942 strlcpy(ds->name, p, sizeof(ds->name));
2943 btrfsic_dev_state_hashtable_add(ds,
2944 &btrfsic_dev_state_hashtable);
2947 ret = btrfsic_process_superblock(state, fs_devices);
2949 mutex_unlock(&btrfsic_mutex);
2950 btrfsic_unmount(fs_devices);
2954 if (state->print_mask & BTRFSIC_PRINT_MASK_INITIAL_DATABASE)
2955 btrfsic_dump_database(state);
2956 if (state->print_mask & BTRFSIC_PRINT_MASK_INITIAL_TREE)
2957 btrfsic_dump_tree(state);
2959 mutex_unlock(&btrfsic_mutex);
2963 void btrfsic_unmount(struct btrfs_fs_devices *fs_devices)
2965 struct btrfsic_block *b_all, *tmp_all;
2966 struct btrfsic_state *state;
2967 struct list_head *dev_head = &fs_devices->devices;
2968 struct btrfs_device *device;
2970 if (!btrfsic_is_initialized)
2973 mutex_lock(&btrfsic_mutex);
2976 list_for_each_entry(device, dev_head, dev_list) {
2977 struct btrfsic_dev_state *ds;
2979 if (!device->bdev || !device->name)
2982 ds = btrfsic_dev_state_hashtable_lookup(
2983 device->bdev->bd_dev,
2984 &btrfsic_dev_state_hashtable);
2987 btrfsic_dev_state_hashtable_remove(ds);
2988 btrfsic_dev_state_free(ds);
2992 if (NULL == state) {
2993 pr_info("btrfsic: error, cannot find state information on umount!\n");
2994 mutex_unlock(&btrfsic_mutex);
2999 * Don't care about keeping the lists' state up to date,
3000 * just free all memory that was allocated dynamically.
3001 * Free the blocks and the block_links.
3003 list_for_each_entry_safe(b_all, tmp_all, &state->all_blocks_list,
3005 struct btrfsic_block_link *l, *tmp;
3007 list_for_each_entry_safe(l, tmp, &b_all->ref_to_list,
3009 if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
3010 btrfsic_print_rem_link(state, l);
3013 if (0 == l->ref_cnt)
3014 btrfsic_block_link_free(l);
3017 if (b_all->is_iodone || b_all->never_written)
3018 btrfsic_block_free(b_all);
3020 pr_info("btrfs: attempt to free %c-block @%llu (%s/%llu/%d) on umount which is not yet iodone!\n",
3021 btrfsic_get_block_type(state, b_all),
3022 b_all->logical_bytenr, b_all->dev_state->name,
3023 b_all->dev_bytenr, b_all->mirror_num);
3026 mutex_unlock(&btrfsic_mutex);