1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * Copyright (C) 2008 Oracle. All rights reserved.
4 */
5
6 #include <linux/kernel.h>
7 #include <linux/slab.h>
8 #include <linux/mm.h>
9 #include <linux/init.h>
10 #include <linux/err.h>
11 #include <linux/sched.h>
12 #include <linux/pagemap.h>
13 #include <linux/bio.h>
14 #include <linux/lzo.h>
15 #include <linux/refcount.h>
16 #include "messages.h"
17 #include "compression.h"
18 #include "ctree.h"
19 #include "super.h"
20 #include "btrfs_inode.h"
21
22 #define LZO_LEN 4
23
24 /*
25 * Btrfs LZO compression format
26 *
27 * Regular and inlined LZO compressed data extents consist of:
28 *
29 * 1. Header
30 * Fixed size. LZO_LEN (4) bytes long, LE32.
31 * Records the total size (including the header) of compressed data.
32 *
33 * 2. Segment(s)
34 * Variable size. Each segment includes one segment header, followed by data
35 * payload.
36 * One regular LZO compressed extent can have one or more segments.
37 * For inlined LZO compressed extent, only one segment is allowed.
38 * One segment represents at most one sector of uncompressed data.
39 *
40 * 2.1 Segment header
41 * Fixed size. LZO_LEN (4) bytes long, LE32.
42 * Records the total size of the segment (not including the header).
43 * Segment header never crosses sector boundary, thus it's possible to
44 * have at most 3 padding zeros at the end of the sector.
45 *
46 * 2.2 Data Payload
47 * Variable size. Size up limit should be lzo1x_worst_compress(sectorsize)
48 * which is 4419 for a 4KiB sectorsize.
49 *
50 * Example with 4K sectorsize:
51 * Page 1:
52 * 0 0x2 0x4 0x6 0x8 0xa 0xc 0xe 0x10
53 * 0x0000 | Header | SegHdr 01 | Data payload 01 ... |
54 * ...
55 * 0x0ff0 | SegHdr N | Data payload N ... |00|
56 * ^^ padding zeros
57 * Page 2:
58 * 0x1000 | SegHdr N+1| Data payload N+1 ... |
59 */
60
61 struct workspace {
62 void *mem;
63 void *buf; /* where decompressed data goes */
64 void *cbuf; /* where compressed data goes */
65 struct list_head list;
66 };
67
workspace_buf_length(const struct btrfs_fs_info * fs_info)68 static u32 workspace_buf_length(const struct btrfs_fs_info *fs_info)
69 {
70 return lzo1x_worst_compress(fs_info->sectorsize);
71 }
workspace_cbuf_length(const struct btrfs_fs_info * fs_info)72 static u32 workspace_cbuf_length(const struct btrfs_fs_info *fs_info)
73 {
74 return lzo1x_worst_compress(fs_info->sectorsize);
75 }
76
lzo_free_workspace(struct list_head * ws)77 void lzo_free_workspace(struct list_head *ws)
78 {
79 struct workspace *workspace = list_entry(ws, struct workspace, list);
80
81 kvfree(workspace->buf);
82 kvfree(workspace->cbuf);
83 kvfree(workspace->mem);
84 kfree(workspace);
85 }
86
lzo_alloc_workspace(struct btrfs_fs_info * fs_info)87 struct list_head *lzo_alloc_workspace(struct btrfs_fs_info *fs_info)
88 {
89 struct workspace *workspace;
90
91 workspace = kzalloc_obj(*workspace);
92 if (!workspace)
93 return ERR_PTR(-ENOMEM);
94
95 workspace->mem = kvmalloc(LZO1X_MEM_COMPRESS, GFP_KERNEL | __GFP_NOWARN);
96 workspace->buf = kvmalloc(workspace_buf_length(fs_info), GFP_KERNEL | __GFP_NOWARN);
97 workspace->cbuf = kvmalloc(workspace_cbuf_length(fs_info), GFP_KERNEL | __GFP_NOWARN);
98 if (!workspace->mem || !workspace->buf || !workspace->cbuf)
99 goto fail;
100
101 INIT_LIST_HEAD(&workspace->list);
102
103 return &workspace->list;
104 fail:
105 lzo_free_workspace(&workspace->list);
106 return ERR_PTR(-ENOMEM);
107 }
108
109 /*
110 * Write data into @out_folio and queue it into @out_bio.
111 *
112 * Return 0 if everything is fine and @total_out will be increased.
113 * Return <0 for error.
114 *
115 * The @out_folio can be NULL after a full folio is queued.
116 * Thus the caller should check and allocate a new folio when needed.
117 */
write_and_queue_folio(struct bio * out_bio,struct folio ** out_folio,u32 * total_out,u32 write_len)118 static int write_and_queue_folio(struct bio *out_bio, struct folio **out_folio,
119 u32 *total_out, u32 write_len)
120 {
121 const u32 fsize = folio_size(*out_folio);
122 const u32 foffset = offset_in_folio(*out_folio, *total_out);
123
124 ASSERT(out_folio && *out_folio);
125 /* Should not cross folio boundary. */
126 ASSERT(foffset + write_len <= fsize);
127
128 /* We can not use bio_add_folio_nofail() which doesn't do any merge. */
129 if (!bio_add_folio(out_bio, *out_folio, write_len, foffset)) {
130 /*
131 * We have allocated a bio that havs BTRFS_MAX_COMPRESSED_PAGES
132 * vecs, and all ranges inside the same folio should have been
133 * merged. If bio_add_folio() still failed, that means we have
134 * reached the bvec limits.
135 *
136 * This should only happen at the beginning of a folio, and
137 * caller is responsible for releasing the folio, since it's
138 * not yet queued into the bio.
139 */
140 ASSERT(IS_ALIGNED(*total_out, fsize));
141 return -E2BIG;
142 }
143
144 *total_out += write_len;
145 /*
146 * The full folio has been filled and queued, reset @out_folio to NULL,
147 * so that error handling is fully handled by the bio.
148 */
149 if (IS_ALIGNED(*total_out, fsize))
150 *out_folio = NULL;
151 return 0;
152 }
153
154 /*
155 * Copy compressed data to bio.
156 *
157 * @out_bio: The bio that will contain all the compressed data.
158 * @compressed_data: The compressed data of this segment.
159 * @compressed_size: The size of the compressed data.
160 * @out_folio: The current output folio, will be updated if a new
161 * folio is allocated.
162 * @total_out: The total bytes of current output.
163 * @max_out: The maximum size of the compressed data.
164 *
165 * Will do:
166 *
167 * - Write a segment header into the destination
168 * - Copy the compressed buffer into the destination
169 * - Make sure we have enough space in the last sector to fit a segment header
170 * If not, we will pad at most (LZO_LEN (4)) - 1 bytes of zeros.
171 * - If a full folio is filled, it will be queued into @out_bio, and @out_folio
172 * will be updated.
173 *
174 * Will allocate new pages when needed.
175 */
copy_compressed_data_to_bio(struct btrfs_fs_info * fs_info,struct bio * out_bio,const char * compressed_data,size_t compressed_size,struct folio ** out_folio,u32 * total_out,u32 max_out)176 static int copy_compressed_data_to_bio(struct btrfs_fs_info *fs_info,
177 struct bio *out_bio,
178 const char *compressed_data,
179 size_t compressed_size,
180 struct folio **out_folio,
181 u32 *total_out, u32 max_out)
182 {
183 const u32 sectorsize = fs_info->sectorsize;
184 const u32 sectorsize_bits = fs_info->sectorsize_bits;
185 const u32 fsize = btrfs_min_folio_size(fs_info);
186 const u32 old_size = out_bio->bi_iter.bi_size;
187 u32 copy_start;
188 u32 sector_bytes_left;
189 char *kaddr;
190 int ret;
191
192 ASSERT(out_folio);
193
194 /* There should be at least a lzo header queued. */
195 ASSERT(old_size);
196 ASSERT(old_size == *total_out);
197
198 /*
199 * We never allow a segment header crossing sector boundary, previous
200 * run should ensure we have enough space left inside the sector.
201 */
202 ASSERT((old_size >> sectorsize_bits) == (old_size + LZO_LEN - 1) >> sectorsize_bits);
203
204 if (!*out_folio) {
205 *out_folio = btrfs_alloc_compr_folio(fs_info, GFP_NOFS);
206 if (!*out_folio)
207 return -ENOMEM;
208 }
209
210 /* Write the segment header first. */
211 kaddr = kmap_local_folio(*out_folio, offset_in_folio(*out_folio, *total_out));
212 put_unaligned_le32(compressed_size, kaddr);
213 kunmap_local(kaddr);
214 ret = write_and_queue_folio(out_bio, out_folio, total_out, LZO_LEN);
215 if (ret < 0)
216 return ret;
217
218 copy_start = *total_out;
219
220 /* Copy compressed data. */
221 while (*total_out - copy_start < compressed_size) {
222 u32 copy_len = min_t(u32, sectorsize - *total_out % sectorsize,
223 copy_start + compressed_size - *total_out);
224 u32 foffset = *total_out & (fsize - 1);
225
226 /* With the range copied, we're larger than the original range. */
227 if (((*total_out + copy_len) >> sectorsize_bits) >=
228 max_out >> sectorsize_bits)
229 return -E2BIG;
230
231 if (!*out_folio) {
232 *out_folio = btrfs_alloc_compr_folio(fs_info, GFP_NOFS);
233 if (!*out_folio)
234 return -ENOMEM;
235 }
236
237 kaddr = kmap_local_folio(*out_folio, foffset);
238 memcpy(kaddr, compressed_data + *total_out - copy_start, copy_len);
239 kunmap_local(kaddr);
240 ret = write_and_queue_folio(out_bio, out_folio, total_out, copy_len);
241 if (ret < 0)
242 return ret;
243 }
244
245 /*
246 * Check if we can fit the next segment header into the remaining space
247 * of the sector.
248 */
249 sector_bytes_left = round_up(*total_out, sectorsize) - *total_out;
250 if (sector_bytes_left >= LZO_LEN || sector_bytes_left == 0)
251 return 0;
252
253 ASSERT(*out_folio);
254
255 /* The remaining size is not enough, pad it with zeros */
256 folio_zero_range(*out_folio, offset_in_folio(*out_folio, *total_out), sector_bytes_left);
257 return write_and_queue_folio(out_bio, out_folio, total_out, sector_bytes_left);
258 }
259
lzo_compress_bio(struct list_head * ws,struct compressed_bio * cb)260 int lzo_compress_bio(struct list_head *ws, struct compressed_bio *cb)
261 {
262 struct btrfs_inode *inode = cb->bbio.inode;
263 struct btrfs_fs_info *fs_info = inode->root->fs_info;
264 struct workspace *workspace = list_entry(ws, struct workspace, list);
265 struct bio *bio = &cb->bbio.bio;
266 const u64 start = cb->start;
267 const u32 len = cb->len;
268 const u32 sectorsize = fs_info->sectorsize;
269 const u32 min_folio_size = btrfs_min_folio_size(fs_info);
270 struct address_space *mapping = inode->vfs_inode.i_mapping;
271 struct folio *folio_in = NULL;
272 struct folio *folio_out = NULL;
273 char *sizes_ptr;
274 int ret = 0;
275 /* Points to the file offset of input data. */
276 u64 cur_in = start;
277 /* Points to the current output byte. */
278 u32 total_out = 0;
279
280 ASSERT(bio->bi_iter.bi_size == 0);
281 ASSERT(len);
282
283 folio_out = btrfs_alloc_compr_folio(fs_info, GFP_NOFS);
284 if (!folio_out)
285 return -ENOMEM;
286
287 /* Queue a segment header first. */
288 ret = write_and_queue_folio(bio, &folio_out, &total_out, LZO_LEN);
289 /* The first header should not fail. */
290 ASSERT(ret == 0);
291
292 while (cur_in < start + len) {
293 char *data_in;
294 const u32 sectorsize_mask = sectorsize - 1;
295 u32 sector_off = (cur_in - start) & sectorsize_mask;
296 u32 in_len;
297 size_t out_len;
298
299 /* Get the input page first. */
300 if (!folio_in) {
301 ret = btrfs_compress_filemap_get_folio(mapping, cur_in, &folio_in);
302 if (ret < 0)
303 goto out;
304 }
305
306 /* Compress at most one sector of data each time. */
307 in_len = min_t(u32, start + len - cur_in, sectorsize - sector_off);
308 ASSERT(in_len);
309 data_in = kmap_local_folio(folio_in, offset_in_folio(folio_in, cur_in));
310 ret = lzo1x_1_compress(data_in, in_len, workspace->cbuf, &out_len,
311 workspace->mem);
312 kunmap_local(data_in);
313 if (unlikely(ret < 0)) {
314 /* lzo1x_1_compress never fails. */
315 ret = -EIO;
316 goto out;
317 }
318
319 ret = copy_compressed_data_to_bio(fs_info, bio, workspace->cbuf, out_len,
320 &folio_out, &total_out, len);
321 if (ret < 0)
322 goto out;
323
324 cur_in += in_len;
325
326 /*
327 * Check if we're making it bigger after two sectors. And if
328 * it is so, give up.
329 */
330 if (cur_in - start > sectorsize * 2 && cur_in - start < total_out) {
331 ret = -E2BIG;
332 goto out;
333 }
334
335 /* Check if we have reached input folio boundary. */
336 if (IS_ALIGNED(cur_in, min_folio_size)) {
337 folio_put(folio_in);
338 folio_in = NULL;
339 }
340 }
341 /*
342 * The last folio is already queued. Bio is responsible for freeing
343 * those folios now.
344 */
345 folio_out = NULL;
346
347 /* Store the size of all chunks of compressed data */
348 sizes_ptr = kmap_local_folio(bio_first_folio_all(bio), 0);
349 put_unaligned_le32(total_out, sizes_ptr);
350 kunmap_local(sizes_ptr);
351 out:
352 /*
353 * We can only free the folio that has no part queued into the bio.
354 *
355 * As any folio that is already queued into bio will be released by
356 * the endio function of bio.
357 */
358 if (folio_out && IS_ALIGNED(total_out, min_folio_size)) {
359 btrfs_free_compr_folio(folio_out);
360 folio_out = NULL;
361 }
362 if (folio_in)
363 folio_put(folio_in);
364 return ret;
365 }
366
get_current_folio(struct compressed_bio * cb,struct folio_iter * fi,u32 * cur_folio_index,u32 cur_in)367 static struct folio *get_current_folio(struct compressed_bio *cb, struct folio_iter *fi,
368 u32 *cur_folio_index, u32 cur_in)
369 {
370 struct btrfs_fs_info *fs_info = cb_to_fs_info(cb);
371 const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
372
373 ASSERT(cur_folio_index);
374
375 /* Need to switch to the next folio. */
376 if (cur_in >> min_folio_shift != *cur_folio_index) {
377 /* We can only do the switch one folio a time. */
378 ASSERT(cur_in >> min_folio_shift == *cur_folio_index + 1);
379
380 bio_next_folio(fi, &cb->bbio.bio);
381 (*cur_folio_index)++;
382 }
383 return fi->folio;
384 }
385
386 /*
387 * Copy the compressed segment payload into @dest.
388 *
389 * For the payload there will be no padding, just need to do page switching.
390 */
copy_compressed_segment(struct compressed_bio * cb,struct folio_iter * fi,u32 * cur_folio_index,char * dest,u32 len,u32 * cur_in)391 static void copy_compressed_segment(struct compressed_bio *cb,
392 struct folio_iter *fi, u32 *cur_folio_index,
393 char *dest, u32 len, u32 *cur_in)
394 {
395 u32 orig_in = *cur_in;
396
397 while (*cur_in < orig_in + len) {
398 struct folio *cur_folio = get_current_folio(cb, fi, cur_folio_index, *cur_in);
399 u32 copy_len;
400
401 ASSERT(cur_folio);
402 copy_len = min_t(u32, orig_in + len - *cur_in,
403 folio_size(cur_folio) - offset_in_folio(cur_folio, *cur_in));
404 ASSERT(copy_len);
405
406 memcpy_from_folio(dest + *cur_in - orig_in, cur_folio,
407 offset_in_folio(cur_folio, *cur_in), copy_len);
408
409 *cur_in += copy_len;
410 }
411 }
412
lzo_decompress_bio(struct list_head * ws,struct compressed_bio * cb)413 int lzo_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
414 {
415 struct workspace *workspace = list_entry(ws, struct workspace, list);
416 struct btrfs_fs_info *fs_info = cb->bbio.inode->root->fs_info;
417 const u32 sectorsize = fs_info->sectorsize;
418 const u32 compressed_len = bio_get_size(&cb->bbio.bio);
419 struct folio_iter fi;
420 char *kaddr;
421 int ret;
422 /* Compressed data length, can be unaligned */
423 u32 len_in;
424 /* Offset inside the compressed data */
425 u32 cur_in = 0;
426 /* Bytes decompressed so far */
427 u32 cur_out = 0;
428 /* The current folio index number inside the bio. */
429 u32 cur_folio_index = 0;
430
431 bio_first_folio(&fi, &cb->bbio.bio, 0);
432 /* There must be a compressed folio and matches the sectorsize. */
433 if (unlikely(!fi.folio))
434 return -EINVAL;
435 ASSERT(folio_size(fi.folio) == btrfs_min_folio_size(fs_info));
436 kaddr = kmap_local_folio(fi.folio, 0);
437 len_in = get_unaligned_le32(kaddr);
438 kunmap_local(kaddr);
439 cur_in += LZO_LEN;
440
441 /*
442 * LZO header length check
443 *
444 * The total length should not exceed the maximum extent length,
445 * and all sectors should be used.
446 * If this happens, it means the compressed extent is corrupted.
447 */
448 if (unlikely(len_in > min_t(size_t, BTRFS_MAX_COMPRESSED, compressed_len) ||
449 round_up(len_in, sectorsize) < compressed_len)) {
450 struct btrfs_inode *inode = cb->bbio.inode;
451
452 btrfs_err(fs_info,
453 "lzo header invalid, root %llu inode %llu offset %llu lzo len %u compressed len %u",
454 btrfs_root_id(inode->root), btrfs_ino(inode),
455 cb->start, len_in, compressed_len);
456 return -EUCLEAN;
457 }
458
459 /* Go through each lzo segment */
460 while (cur_in < len_in) {
461 struct folio *cur_folio;
462 /* Length of the compressed segment */
463 u32 seg_len;
464 u32 sector_bytes_left;
465 size_t out_len = lzo1x_worst_compress(sectorsize);
466
467 /*
468 * We should always have enough space for one segment header
469 * inside current sector.
470 */
471 ASSERT(cur_in / sectorsize ==
472 (cur_in + LZO_LEN - 1) / sectorsize);
473 cur_folio = get_current_folio(cb, &fi, &cur_folio_index, cur_in);
474 ASSERT(cur_folio);
475 kaddr = kmap_local_folio(cur_folio, 0);
476 seg_len = get_unaligned_le32(kaddr + offset_in_folio(cur_folio, cur_in));
477 kunmap_local(kaddr);
478 cur_in += LZO_LEN;
479
480 if (unlikely(seg_len > workspace_cbuf_length(fs_info))) {
481 struct btrfs_inode *inode = cb->bbio.inode;
482
483 /*
484 * seg_len shouldn't be larger than we have allocated
485 * for workspace->cbuf
486 */
487 btrfs_err(fs_info,
488 "lzo segment too big, root %llu inode %llu offset %llu len %u",
489 btrfs_root_id(inode->root), btrfs_ino(inode),
490 cb->start, seg_len);
491 return -EIO;
492 }
493
494 /* Copy the compressed segment payload into workspace */
495 copy_compressed_segment(cb, &fi, &cur_folio_index, workspace->cbuf,
496 seg_len, &cur_in);
497
498 /* Decompress the data */
499 ret = lzo1x_decompress_safe(workspace->cbuf, seg_len,
500 workspace->buf, &out_len);
501 if (unlikely(ret != LZO_E_OK)) {
502 struct btrfs_inode *inode = cb->bbio.inode;
503
504 btrfs_err(fs_info,
505 "lzo decompression failed, error %d root %llu inode %llu offset %llu",
506 ret, btrfs_root_id(inode->root), btrfs_ino(inode),
507 cb->start);
508 return -EIO;
509 }
510
511 /* Copy the data into inode pages */
512 ret = btrfs_decompress_buf2page(workspace->buf, out_len, cb, cur_out);
513 cur_out += out_len;
514
515 /* All data read, exit */
516 if (ret == 0)
517 return 0;
518 ret = 0;
519
520 /* Check if the sector has enough space for a segment header */
521 sector_bytes_left = sectorsize - (cur_in % sectorsize);
522 if (sector_bytes_left >= LZO_LEN)
523 continue;
524
525 /* Skip the padding zeros */
526 cur_in += sector_bytes_left;
527 }
528
529 return 0;
530 }
531
lzo_decompress(struct list_head * ws,const u8 * data_in,struct folio * dest_folio,unsigned long dest_pgoff,size_t srclen,size_t destlen)532 int lzo_decompress(struct list_head *ws, const u8 *data_in,
533 struct folio *dest_folio, unsigned long dest_pgoff, size_t srclen,
534 size_t destlen)
535 {
536 struct workspace *workspace = list_entry(ws, struct workspace, list);
537 struct btrfs_fs_info *fs_info = folio_to_fs_info(dest_folio);
538 const u32 sectorsize = fs_info->sectorsize;
539 size_t in_len;
540 size_t out_len;
541 size_t max_segment_len = workspace_buf_length(fs_info);
542 int ret;
543
544 if (unlikely(srclen < LZO_LEN || srclen > max_segment_len + LZO_LEN * 2))
545 return -EUCLEAN;
546
547 in_len = get_unaligned_le32(data_in);
548 if (unlikely(in_len != srclen))
549 return -EUCLEAN;
550 data_in += LZO_LEN;
551
552 in_len = get_unaligned_le32(data_in);
553 if (unlikely(in_len != srclen - LZO_LEN * 2))
554 return -EUCLEAN;
555 data_in += LZO_LEN;
556
557 out_len = sectorsize;
558 ret = lzo1x_decompress_safe(data_in, in_len, workspace->buf, &out_len);
559 if (unlikely(ret != LZO_E_OK)) {
560 struct btrfs_inode *inode = folio_to_inode(dest_folio);
561
562 btrfs_err(fs_info,
563 "lzo decompression failed, error %d root %llu inode %llu offset %llu",
564 ret, btrfs_root_id(inode->root), btrfs_ino(inode),
565 folio_pos(dest_folio));
566 return -EIO;
567 }
568
569 ASSERT(out_len <= sectorsize);
570 memcpy_to_folio(dest_folio, dest_pgoff, workspace->buf, out_len);
571 /* Early end, considered as an error. */
572 if (unlikely(out_len < destlen)) {
573 folio_zero_range(dest_folio, dest_pgoff + out_len, destlen - out_len);
574 return -EIO;
575 }
576
577 return 0;
578 }
579
580 const struct btrfs_compress_levels btrfs_lzo_compress = {
581 .max_level = 1,
582 .default_level = 1,
583 };
584