]>
Commit | Line | Data |
---|---|---|
c8b97818 CM |
1 | /* |
2 | * Copyright (C) 2008 Oracle. All rights reserved. | |
3 | * | |
4 | * This program is free software; you can redistribute it and/or | |
5 | * modify it under the terms of the GNU General Public | |
6 | * License v2 as published by the Free Software Foundation. | |
7 | * | |
8 | * This program is distributed in the hope that it will be useful, | |
9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
11 | * General Public License for more details. | |
12 | * | |
13 | * You should have received a copy of the GNU General Public | |
14 | * License along with this program; if not, write to the | |
15 | * Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
16 | * Boston, MA 021110-1307, USA. | |
17 | */ | |
18 | ||
19 | #include <linux/kernel.h> | |
20 | #include <linux/bio.h> | |
21 | #include <linux/buffer_head.h> | |
22 | #include <linux/file.h> | |
23 | #include <linux/fs.h> | |
24 | #include <linux/pagemap.h> | |
25 | #include <linux/highmem.h> | |
26 | #include <linux/time.h> | |
27 | #include <linux/init.h> | |
28 | #include <linux/string.h> | |
c8b97818 CM |
29 | #include <linux/backing-dev.h> |
30 | #include <linux/mpage.h> | |
31 | #include <linux/swap.h> | |
32 | #include <linux/writeback.h> | |
33 | #include <linux/bit_spinlock.h> | |
5a0e3ad6 | 34 | #include <linux/slab.h> |
fe308533 | 35 | #include <linux/sched/mm.h> |
858177d3 | 36 | #include <linux/sort.h> |
19562430 | 37 | #include <linux/log2.h> |
c8b97818 CM |
38 | #include "ctree.h" |
39 | #include "disk-io.h" | |
40 | #include "transaction.h" | |
41 | #include "btrfs_inode.h" | |
42 | #include "volumes.h" | |
43 | #include "ordered-data.h" | |
c8b97818 CM |
44 | #include "compression.h" |
45 | #include "extent_io.h" | |
46 | #include "extent_map.h" | |
47 | ||
8140dc30 | 48 | static int btrfs_decompress_bio(struct compressed_bio *cb); |
48a3b636 | 49 | |
2ff7e61e | 50 | static inline int compressed_bio_size(struct btrfs_fs_info *fs_info, |
d20f7043 CM |
51 | unsigned long disk_size) |
52 | { | |
0b246afa | 53 | u16 csum_size = btrfs_super_csum_size(fs_info->super_copy); |
6c41761f | 54 | |
d20f7043 | 55 | return sizeof(struct compressed_bio) + |
0b246afa | 56 | (DIV_ROUND_UP(disk_size, fs_info->sectorsize)) * csum_size; |
d20f7043 CM |
57 | } |
58 | ||
f898ac6a | 59 | static int check_compressed_csum(struct btrfs_inode *inode, |
d20f7043 CM |
60 | struct compressed_bio *cb, |
61 | u64 disk_start) | |
62 | { | |
63 | int ret; | |
d20f7043 CM |
64 | struct page *page; |
65 | unsigned long i; | |
66 | char *kaddr; | |
67 | u32 csum; | |
68 | u32 *cb_sum = &cb->sums; | |
69 | ||
f898ac6a | 70 | if (inode->flags & BTRFS_INODE_NODATASUM) |
d20f7043 CM |
71 | return 0; |
72 | ||
73 | for (i = 0; i < cb->nr_pages; i++) { | |
74 | page = cb->compressed_pages[i]; | |
75 | csum = ~(u32)0; | |
76 | ||
7ac687d9 | 77 | kaddr = kmap_atomic(page); |
09cbfeaf | 78 | csum = btrfs_csum_data(kaddr, csum, PAGE_SIZE); |
0b5e3daf | 79 | btrfs_csum_final(csum, (u8 *)&csum); |
7ac687d9 | 80 | kunmap_atomic(kaddr); |
d20f7043 CM |
81 | |
82 | if (csum != *cb_sum) { | |
f898ac6a | 83 | btrfs_print_data_csum_error(inode, disk_start, csum, |
0970a22e | 84 | *cb_sum, cb->mirror_num); |
d20f7043 CM |
85 | ret = -EIO; |
86 | goto fail; | |
87 | } | |
88 | cb_sum++; | |
89 | ||
90 | } | |
91 | ret = 0; | |
92 | fail: | |
93 | return ret; | |
94 | } | |
95 | ||
c8b97818 CM |
96 | /* when we finish reading compressed pages from the disk, we |
97 | * decompress them and then run the bio end_io routines on the | |
98 | * decompressed pages (in the inode address space). | |
99 | * | |
100 | * This allows the checksumming and other IO error handling routines | |
101 | * to work normally | |
102 | * | |
103 | * The compressed pages are freed here, and it must be run | |
104 | * in process context | |
105 | */ | |
4246a0b6 | 106 | static void end_compressed_bio_read(struct bio *bio) |
c8b97818 | 107 | { |
c8b97818 CM |
108 | struct compressed_bio *cb = bio->bi_private; |
109 | struct inode *inode; | |
110 | struct page *page; | |
111 | unsigned long index; | |
cf1167d5 | 112 | unsigned int mirror = btrfs_io_bio(bio)->mirror_num; |
e6311f24 | 113 | int ret = 0; |
c8b97818 | 114 | |
4e4cbee9 | 115 | if (bio->bi_status) |
c8b97818 CM |
116 | cb->errors = 1; |
117 | ||
118 | /* if there are more bios still pending for this compressed | |
119 | * extent, just exit | |
120 | */ | |
a50299ae | 121 | if (!refcount_dec_and_test(&cb->pending_bios)) |
c8b97818 CM |
122 | goto out; |
123 | ||
cf1167d5 LB |
124 | /* |
125 | * Record the correct mirror_num in cb->orig_bio so that | |
126 | * read-repair can work properly. | |
127 | */ | |
128 | ASSERT(btrfs_io_bio(cb->orig_bio)); | |
129 | btrfs_io_bio(cb->orig_bio)->mirror_num = mirror; | |
130 | cb->mirror_num = mirror; | |
131 | ||
e6311f24 LB |
132 | /* |
133 | * Some IO in this cb have failed, just skip checksum as there | |
134 | * is no way it could be correct. | |
135 | */ | |
136 | if (cb->errors == 1) | |
137 | goto csum_failed; | |
138 | ||
d20f7043 | 139 | inode = cb->inode; |
f898ac6a | 140 | ret = check_compressed_csum(BTRFS_I(inode), cb, |
4f024f37 | 141 | (u64)bio->bi_iter.bi_sector << 9); |
d20f7043 CM |
142 | if (ret) |
143 | goto csum_failed; | |
144 | ||
c8b97818 CM |
145 | /* ok, we're the last bio for this extent, lets start |
146 | * the decompression. | |
147 | */ | |
8140dc30 AJ |
148 | ret = btrfs_decompress_bio(cb); |
149 | ||
d20f7043 | 150 | csum_failed: |
c8b97818 CM |
151 | if (ret) |
152 | cb->errors = 1; | |
153 | ||
154 | /* release the compressed pages */ | |
155 | index = 0; | |
156 | for (index = 0; index < cb->nr_pages; index++) { | |
157 | page = cb->compressed_pages[index]; | |
158 | page->mapping = NULL; | |
09cbfeaf | 159 | put_page(page); |
c8b97818 CM |
160 | } |
161 | ||
162 | /* do io completion on the original bio */ | |
771ed689 | 163 | if (cb->errors) { |
c8b97818 | 164 | bio_io_error(cb->orig_bio); |
d20f7043 | 165 | } else { |
2c30c71b KO |
166 | int i; |
167 | struct bio_vec *bvec; | |
d20f7043 CM |
168 | |
169 | /* | |
170 | * we have verified the checksum already, set page | |
171 | * checked so the end_io handlers know about it | |
172 | */ | |
c09abff8 | 173 | ASSERT(!bio_flagged(bio, BIO_CLONED)); |
2c30c71b | 174 | bio_for_each_segment_all(bvec, cb->orig_bio, i) |
d20f7043 | 175 | SetPageChecked(bvec->bv_page); |
2c30c71b | 176 | |
4246a0b6 | 177 | bio_endio(cb->orig_bio); |
d20f7043 | 178 | } |
c8b97818 CM |
179 | |
180 | /* finally free the cb struct */ | |
181 | kfree(cb->compressed_pages); | |
182 | kfree(cb); | |
183 | out: | |
184 | bio_put(bio); | |
185 | } | |
186 | ||
187 | /* | |
188 | * Clear the writeback bits on all of the file | |
189 | * pages for a compressed write | |
190 | */ | |
7bdcefc1 FM |
191 | static noinline void end_compressed_writeback(struct inode *inode, |
192 | const struct compressed_bio *cb) | |
c8b97818 | 193 | { |
09cbfeaf KS |
194 | unsigned long index = cb->start >> PAGE_SHIFT; |
195 | unsigned long end_index = (cb->start + cb->len - 1) >> PAGE_SHIFT; | |
c8b97818 CM |
196 | struct page *pages[16]; |
197 | unsigned long nr_pages = end_index - index + 1; | |
198 | int i; | |
199 | int ret; | |
200 | ||
7bdcefc1 FM |
201 | if (cb->errors) |
202 | mapping_set_error(inode->i_mapping, -EIO); | |
203 | ||
d397712b | 204 | while (nr_pages > 0) { |
c8b97818 | 205 | ret = find_get_pages_contig(inode->i_mapping, index, |
5b050f04 CM |
206 | min_t(unsigned long, |
207 | nr_pages, ARRAY_SIZE(pages)), pages); | |
c8b97818 CM |
208 | if (ret == 0) { |
209 | nr_pages -= 1; | |
210 | index += 1; | |
211 | continue; | |
212 | } | |
213 | for (i = 0; i < ret; i++) { | |
7bdcefc1 FM |
214 | if (cb->errors) |
215 | SetPageError(pages[i]); | |
c8b97818 | 216 | end_page_writeback(pages[i]); |
09cbfeaf | 217 | put_page(pages[i]); |
c8b97818 CM |
218 | } |
219 | nr_pages -= ret; | |
220 | index += ret; | |
221 | } | |
222 | /* the inode may be gone now */ | |
c8b97818 CM |
223 | } |
224 | ||
225 | /* | |
226 | * do the cleanup once all the compressed pages hit the disk. | |
227 | * This will clear writeback on the file pages and free the compressed | |
228 | * pages. | |
229 | * | |
230 | * This also calls the writeback end hooks for the file pages so that | |
231 | * metadata and checksums can be updated in the file. | |
232 | */ | |
4246a0b6 | 233 | static void end_compressed_bio_write(struct bio *bio) |
c8b97818 CM |
234 | { |
235 | struct extent_io_tree *tree; | |
236 | struct compressed_bio *cb = bio->bi_private; | |
237 | struct inode *inode; | |
238 | struct page *page; | |
239 | unsigned long index; | |
240 | ||
4e4cbee9 | 241 | if (bio->bi_status) |
c8b97818 CM |
242 | cb->errors = 1; |
243 | ||
244 | /* if there are more bios still pending for this compressed | |
245 | * extent, just exit | |
246 | */ | |
a50299ae | 247 | if (!refcount_dec_and_test(&cb->pending_bios)) |
c8b97818 CM |
248 | goto out; |
249 | ||
250 | /* ok, we're the last bio for this extent, step one is to | |
251 | * call back into the FS and do all the end_io operations | |
252 | */ | |
253 | inode = cb->inode; | |
254 | tree = &BTRFS_I(inode)->io_tree; | |
70b99e69 | 255 | cb->compressed_pages[0]->mapping = cb->inode->i_mapping; |
c8b97818 CM |
256 | tree->ops->writepage_end_io_hook(cb->compressed_pages[0], |
257 | cb->start, | |
258 | cb->start + cb->len - 1, | |
7bdcefc1 | 259 | NULL, |
2dbe0c77 AJ |
260 | bio->bi_status ? |
261 | BLK_STS_OK : BLK_STS_NOTSUPP); | |
70b99e69 | 262 | cb->compressed_pages[0]->mapping = NULL; |
c8b97818 | 263 | |
7bdcefc1 | 264 | end_compressed_writeback(inode, cb); |
c8b97818 CM |
265 | /* note, our inode could be gone now */ |
266 | ||
267 | /* | |
268 | * release the compressed pages, these came from alloc_page and | |
269 | * are not attached to the inode at all | |
270 | */ | |
271 | index = 0; | |
272 | for (index = 0; index < cb->nr_pages; index++) { | |
273 | page = cb->compressed_pages[index]; | |
274 | page->mapping = NULL; | |
09cbfeaf | 275 | put_page(page); |
c8b97818 CM |
276 | } |
277 | ||
278 | /* finally free the cb struct */ | |
279 | kfree(cb->compressed_pages); | |
280 | kfree(cb); | |
281 | out: | |
282 | bio_put(bio); | |
283 | } | |
284 | ||
285 | /* | |
286 | * worker function to build and submit bios for previously compressed pages. | |
287 | * The corresponding pages in the inode should be marked for writeback | |
288 | * and the compressed pages should have a reference on them for dropping | |
289 | * when the IO is complete. | |
290 | * | |
291 | * This also checksums the file bytes and gets things ready for | |
292 | * the end io hooks. | |
293 | */ | |
4e4cbee9 | 294 | blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start, |
c8b97818 CM |
295 | unsigned long len, u64 disk_start, |
296 | unsigned long compressed_len, | |
297 | struct page **compressed_pages, | |
f82b7359 LB |
298 | unsigned long nr_pages, |
299 | unsigned int write_flags) | |
c8b97818 | 300 | { |
0b246afa | 301 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
c8b97818 | 302 | struct bio *bio = NULL; |
c8b97818 CM |
303 | struct compressed_bio *cb; |
304 | unsigned long bytes_left; | |
305 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
306e16ce | 306 | int pg_index = 0; |
c8b97818 CM |
307 | struct page *page; |
308 | u64 first_byte = disk_start; | |
309 | struct block_device *bdev; | |
4e4cbee9 | 310 | blk_status_t ret; |
e55179b3 | 311 | int skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM; |
c8b97818 | 312 | |
09cbfeaf | 313 | WARN_ON(start & ((u64)PAGE_SIZE - 1)); |
2ff7e61e | 314 | cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS); |
dac97e51 | 315 | if (!cb) |
4e4cbee9 | 316 | return BLK_STS_RESOURCE; |
a50299ae | 317 | refcount_set(&cb->pending_bios, 0); |
c8b97818 CM |
318 | cb->errors = 0; |
319 | cb->inode = inode; | |
320 | cb->start = start; | |
321 | cb->len = len; | |
d20f7043 | 322 | cb->mirror_num = 0; |
c8b97818 CM |
323 | cb->compressed_pages = compressed_pages; |
324 | cb->compressed_len = compressed_len; | |
325 | cb->orig_bio = NULL; | |
326 | cb->nr_pages = nr_pages; | |
327 | ||
0b246afa | 328 | bdev = fs_info->fs_devices->latest_bdev; |
c8b97818 | 329 | |
c821e7f3 | 330 | bio = btrfs_bio_alloc(bdev, first_byte); |
f82b7359 | 331 | bio->bi_opf = REQ_OP_WRITE | write_flags; |
c8b97818 CM |
332 | bio->bi_private = cb; |
333 | bio->bi_end_io = end_compressed_bio_write; | |
a50299ae | 334 | refcount_set(&cb->pending_bios, 1); |
c8b97818 CM |
335 | |
336 | /* create and submit bios for the compressed pages */ | |
337 | bytes_left = compressed_len; | |
306e16ce | 338 | for (pg_index = 0; pg_index < cb->nr_pages; pg_index++) { |
4e4cbee9 CH |
339 | int submit = 0; |
340 | ||
306e16ce | 341 | page = compressed_pages[pg_index]; |
c8b97818 | 342 | page->mapping = inode->i_mapping; |
4f024f37 | 343 | if (bio->bi_iter.bi_size) |
4e4cbee9 | 344 | submit = io_tree->ops->merge_bio_hook(page, 0, |
09cbfeaf | 345 | PAGE_SIZE, |
c8b97818 | 346 | bio, 0); |
c8b97818 | 347 | |
70b99e69 | 348 | page->mapping = NULL; |
4e4cbee9 | 349 | if (submit || bio_add_page(bio, page, PAGE_SIZE, 0) < |
09cbfeaf | 350 | PAGE_SIZE) { |
c8b97818 CM |
351 | bio_get(bio); |
352 | ||
af09abfe CM |
353 | /* |
354 | * inc the count before we submit the bio so | |
355 | * we know the end IO handler won't happen before | |
356 | * we inc the count. Otherwise, the cb might get | |
357 | * freed before we're done setting it up | |
358 | */ | |
a50299ae | 359 | refcount_inc(&cb->pending_bios); |
0b246afa JM |
360 | ret = btrfs_bio_wq_end_io(fs_info, bio, |
361 | BTRFS_WQ_ENDIO_DATA); | |
79787eaa | 362 | BUG_ON(ret); /* -ENOMEM */ |
c8b97818 | 363 | |
e55179b3 | 364 | if (!skip_sum) { |
2ff7e61e | 365 | ret = btrfs_csum_one_bio(inode, bio, start, 1); |
79787eaa | 366 | BUG_ON(ret); /* -ENOMEM */ |
e55179b3 | 367 | } |
d20f7043 | 368 | |
2ff7e61e | 369 | ret = btrfs_map_bio(fs_info, bio, 0, 1); |
f5daf2c7 | 370 | if (ret) { |
4e4cbee9 | 371 | bio->bi_status = ret; |
f5daf2c7 LB |
372 | bio_endio(bio); |
373 | } | |
c8b97818 CM |
374 | |
375 | bio_put(bio); | |
376 | ||
c821e7f3 | 377 | bio = btrfs_bio_alloc(bdev, first_byte); |
f82b7359 | 378 | bio->bi_opf = REQ_OP_WRITE | write_flags; |
c8b97818 CM |
379 | bio->bi_private = cb; |
380 | bio->bi_end_io = end_compressed_bio_write; | |
09cbfeaf | 381 | bio_add_page(bio, page, PAGE_SIZE, 0); |
c8b97818 | 382 | } |
09cbfeaf | 383 | if (bytes_left < PAGE_SIZE) { |
0b246afa | 384 | btrfs_info(fs_info, |
efe120a0 | 385 | "bytes left %lu compress len %lu nr %lu", |
cfbc246e CM |
386 | bytes_left, cb->compressed_len, cb->nr_pages); |
387 | } | |
09cbfeaf KS |
388 | bytes_left -= PAGE_SIZE; |
389 | first_byte += PAGE_SIZE; | |
771ed689 | 390 | cond_resched(); |
c8b97818 CM |
391 | } |
392 | bio_get(bio); | |
393 | ||
0b246afa | 394 | ret = btrfs_bio_wq_end_io(fs_info, bio, BTRFS_WQ_ENDIO_DATA); |
79787eaa | 395 | BUG_ON(ret); /* -ENOMEM */ |
c8b97818 | 396 | |
e55179b3 | 397 | if (!skip_sum) { |
2ff7e61e | 398 | ret = btrfs_csum_one_bio(inode, bio, start, 1); |
79787eaa | 399 | BUG_ON(ret); /* -ENOMEM */ |
e55179b3 | 400 | } |
d20f7043 | 401 | |
2ff7e61e | 402 | ret = btrfs_map_bio(fs_info, bio, 0, 1); |
f5daf2c7 | 403 | if (ret) { |
4e4cbee9 | 404 | bio->bi_status = ret; |
f5daf2c7 LB |
405 | bio_endio(bio); |
406 | } | |
c8b97818 CM |
407 | |
408 | bio_put(bio); | |
409 | return 0; | |
410 | } | |
411 | ||
2a4d0c90 CH |
412 | static u64 bio_end_offset(struct bio *bio) |
413 | { | |
414 | struct bio_vec *last = &bio->bi_io_vec[bio->bi_vcnt - 1]; | |
415 | ||
416 | return page_offset(last->bv_page) + last->bv_len + last->bv_offset; | |
417 | } | |
418 | ||
771ed689 CM |
419 | static noinline int add_ra_bio_pages(struct inode *inode, |
420 | u64 compressed_end, | |
421 | struct compressed_bio *cb) | |
422 | { | |
423 | unsigned long end_index; | |
306e16ce | 424 | unsigned long pg_index; |
771ed689 CM |
425 | u64 last_offset; |
426 | u64 isize = i_size_read(inode); | |
427 | int ret; | |
428 | struct page *page; | |
429 | unsigned long nr_pages = 0; | |
430 | struct extent_map *em; | |
431 | struct address_space *mapping = inode->i_mapping; | |
771ed689 CM |
432 | struct extent_map_tree *em_tree; |
433 | struct extent_io_tree *tree; | |
434 | u64 end; | |
435 | int misses = 0; | |
436 | ||
2a4d0c90 | 437 | last_offset = bio_end_offset(cb->orig_bio); |
771ed689 CM |
438 | em_tree = &BTRFS_I(inode)->extent_tree; |
439 | tree = &BTRFS_I(inode)->io_tree; | |
440 | ||
441 | if (isize == 0) | |
442 | return 0; | |
443 | ||
09cbfeaf | 444 | end_index = (i_size_read(inode) - 1) >> PAGE_SHIFT; |
771ed689 | 445 | |
d397712b | 446 | while (last_offset < compressed_end) { |
09cbfeaf | 447 | pg_index = last_offset >> PAGE_SHIFT; |
771ed689 | 448 | |
306e16ce | 449 | if (pg_index > end_index) |
771ed689 CM |
450 | break; |
451 | ||
452 | rcu_read_lock(); | |
306e16ce | 453 | page = radix_tree_lookup(&mapping->page_tree, pg_index); |
771ed689 | 454 | rcu_read_unlock(); |
0cd6144a | 455 | if (page && !radix_tree_exceptional_entry(page)) { |
771ed689 CM |
456 | misses++; |
457 | if (misses > 4) | |
458 | break; | |
459 | goto next; | |
460 | } | |
461 | ||
c62d2555 MH |
462 | page = __page_cache_alloc(mapping_gfp_constraint(mapping, |
463 | ~__GFP_FS)); | |
771ed689 CM |
464 | if (!page) |
465 | break; | |
466 | ||
c62d2555 | 467 | if (add_to_page_cache_lru(page, mapping, pg_index, GFP_NOFS)) { |
09cbfeaf | 468 | put_page(page); |
771ed689 CM |
469 | goto next; |
470 | } | |
471 | ||
09cbfeaf | 472 | end = last_offset + PAGE_SIZE - 1; |
771ed689 CM |
473 | /* |
474 | * at this point, we have a locked page in the page cache | |
475 | * for these bytes in the file. But, we have to make | |
476 | * sure they map to this compressed extent on disk. | |
477 | */ | |
478 | set_page_extent_mapped(page); | |
d0082371 | 479 | lock_extent(tree, last_offset, end); |
890871be | 480 | read_lock(&em_tree->lock); |
771ed689 | 481 | em = lookup_extent_mapping(em_tree, last_offset, |
09cbfeaf | 482 | PAGE_SIZE); |
890871be | 483 | read_unlock(&em_tree->lock); |
771ed689 CM |
484 | |
485 | if (!em || last_offset < em->start || | |
09cbfeaf | 486 | (last_offset + PAGE_SIZE > extent_map_end(em)) || |
4f024f37 | 487 | (em->block_start >> 9) != cb->orig_bio->bi_iter.bi_sector) { |
771ed689 | 488 | free_extent_map(em); |
d0082371 | 489 | unlock_extent(tree, last_offset, end); |
771ed689 | 490 | unlock_page(page); |
09cbfeaf | 491 | put_page(page); |
771ed689 CM |
492 | break; |
493 | } | |
494 | free_extent_map(em); | |
495 | ||
496 | if (page->index == end_index) { | |
497 | char *userpage; | |
09cbfeaf | 498 | size_t zero_offset = isize & (PAGE_SIZE - 1); |
771ed689 CM |
499 | |
500 | if (zero_offset) { | |
501 | int zeros; | |
09cbfeaf | 502 | zeros = PAGE_SIZE - zero_offset; |
7ac687d9 | 503 | userpage = kmap_atomic(page); |
771ed689 CM |
504 | memset(userpage + zero_offset, 0, zeros); |
505 | flush_dcache_page(page); | |
7ac687d9 | 506 | kunmap_atomic(userpage); |
771ed689 CM |
507 | } |
508 | } | |
509 | ||
510 | ret = bio_add_page(cb->orig_bio, page, | |
09cbfeaf | 511 | PAGE_SIZE, 0); |
771ed689 | 512 | |
09cbfeaf | 513 | if (ret == PAGE_SIZE) { |
771ed689 | 514 | nr_pages++; |
09cbfeaf | 515 | put_page(page); |
771ed689 | 516 | } else { |
d0082371 | 517 | unlock_extent(tree, last_offset, end); |
771ed689 | 518 | unlock_page(page); |
09cbfeaf | 519 | put_page(page); |
771ed689 CM |
520 | break; |
521 | } | |
522 | next: | |
09cbfeaf | 523 | last_offset += PAGE_SIZE; |
771ed689 | 524 | } |
771ed689 CM |
525 | return 0; |
526 | } | |
527 | ||
c8b97818 CM |
528 | /* |
529 | * for a compressed read, the bio we get passed has all the inode pages | |
530 | * in it. We don't actually do IO on those pages but allocate new ones | |
531 | * to hold the compressed pages on disk. | |
532 | * | |
4f024f37 | 533 | * bio->bi_iter.bi_sector points to the compressed extent on disk |
c8b97818 | 534 | * bio->bi_io_vec points to all of the inode pages |
c8b97818 CM |
535 | * |
536 | * After the compressed pages are read, we copy the bytes into the | |
537 | * bio we were passed and then call the bio end_io calls | |
538 | */ | |
4e4cbee9 | 539 | blk_status_t btrfs_submit_compressed_read(struct inode *inode, struct bio *bio, |
c8b97818 CM |
540 | int mirror_num, unsigned long bio_flags) |
541 | { | |
0b246afa | 542 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
c8b97818 CM |
543 | struct extent_io_tree *tree; |
544 | struct extent_map_tree *em_tree; | |
545 | struct compressed_bio *cb; | |
c8b97818 CM |
546 | unsigned long compressed_len; |
547 | unsigned long nr_pages; | |
306e16ce | 548 | unsigned long pg_index; |
c8b97818 CM |
549 | struct page *page; |
550 | struct block_device *bdev; | |
551 | struct bio *comp_bio; | |
4f024f37 | 552 | u64 cur_disk_byte = (u64)bio->bi_iter.bi_sector << 9; |
e04ca626 CM |
553 | u64 em_len; |
554 | u64 em_start; | |
c8b97818 | 555 | struct extent_map *em; |
4e4cbee9 | 556 | blk_status_t ret = BLK_STS_RESOURCE; |
15e3004a | 557 | int faili = 0; |
d20f7043 | 558 | u32 *sums; |
c8b97818 CM |
559 | |
560 | tree = &BTRFS_I(inode)->io_tree; | |
561 | em_tree = &BTRFS_I(inode)->extent_tree; | |
562 | ||
563 | /* we need the actual starting offset of this extent in the file */ | |
890871be | 564 | read_lock(&em_tree->lock); |
c8b97818 CM |
565 | em = lookup_extent_mapping(em_tree, |
566 | page_offset(bio->bi_io_vec->bv_page), | |
09cbfeaf | 567 | PAGE_SIZE); |
890871be | 568 | read_unlock(&em_tree->lock); |
285190d9 | 569 | if (!em) |
4e4cbee9 | 570 | return BLK_STS_IOERR; |
c8b97818 | 571 | |
d20f7043 | 572 | compressed_len = em->block_len; |
2ff7e61e | 573 | cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS); |
6b82ce8d | 574 | if (!cb) |
575 | goto out; | |
576 | ||
a50299ae | 577 | refcount_set(&cb->pending_bios, 0); |
c8b97818 CM |
578 | cb->errors = 0; |
579 | cb->inode = inode; | |
d20f7043 CM |
580 | cb->mirror_num = mirror_num; |
581 | sums = &cb->sums; | |
c8b97818 | 582 | |
ff5b7ee3 | 583 | cb->start = em->orig_start; |
e04ca626 CM |
584 | em_len = em->len; |
585 | em_start = em->start; | |
d20f7043 | 586 | |
c8b97818 | 587 | free_extent_map(em); |
e04ca626 | 588 | em = NULL; |
c8b97818 | 589 | |
81381053 | 590 | cb->len = bio->bi_iter.bi_size; |
c8b97818 | 591 | cb->compressed_len = compressed_len; |
261507a0 | 592 | cb->compress_type = extent_compress_type(bio_flags); |
c8b97818 CM |
593 | cb->orig_bio = bio; |
594 | ||
09cbfeaf | 595 | nr_pages = DIV_ROUND_UP(compressed_len, PAGE_SIZE); |
31e818fe | 596 | cb->compressed_pages = kcalloc(nr_pages, sizeof(struct page *), |
c8b97818 | 597 | GFP_NOFS); |
6b82ce8d | 598 | if (!cb->compressed_pages) |
599 | goto fail1; | |
600 | ||
0b246afa | 601 | bdev = fs_info->fs_devices->latest_bdev; |
c8b97818 | 602 | |
306e16ce DS |
603 | for (pg_index = 0; pg_index < nr_pages; pg_index++) { |
604 | cb->compressed_pages[pg_index] = alloc_page(GFP_NOFS | | |
c8b97818 | 605 | __GFP_HIGHMEM); |
15e3004a JB |
606 | if (!cb->compressed_pages[pg_index]) { |
607 | faili = pg_index - 1; | |
0e9350de | 608 | ret = BLK_STS_RESOURCE; |
6b82ce8d | 609 | goto fail2; |
15e3004a | 610 | } |
c8b97818 | 611 | } |
15e3004a | 612 | faili = nr_pages - 1; |
c8b97818 CM |
613 | cb->nr_pages = nr_pages; |
614 | ||
7f042a83 | 615 | add_ra_bio_pages(inode, em_start + em_len, cb); |
771ed689 | 616 | |
771ed689 | 617 | /* include any pages we added in add_ra-bio_pages */ |
81381053 | 618 | cb->len = bio->bi_iter.bi_size; |
771ed689 | 619 | |
c821e7f3 | 620 | comp_bio = btrfs_bio_alloc(bdev, cur_disk_byte); |
37226b21 | 621 | bio_set_op_attrs (comp_bio, REQ_OP_READ, 0); |
c8b97818 CM |
622 | comp_bio->bi_private = cb; |
623 | comp_bio->bi_end_io = end_compressed_bio_read; | |
a50299ae | 624 | refcount_set(&cb->pending_bios, 1); |
c8b97818 | 625 | |
306e16ce | 626 | for (pg_index = 0; pg_index < nr_pages; pg_index++) { |
4e4cbee9 CH |
627 | int submit = 0; |
628 | ||
306e16ce | 629 | page = cb->compressed_pages[pg_index]; |
c8b97818 | 630 | page->mapping = inode->i_mapping; |
09cbfeaf | 631 | page->index = em_start >> PAGE_SHIFT; |
d20f7043 | 632 | |
4f024f37 | 633 | if (comp_bio->bi_iter.bi_size) |
4e4cbee9 | 634 | submit = tree->ops->merge_bio_hook(page, 0, |
09cbfeaf | 635 | PAGE_SIZE, |
c8b97818 | 636 | comp_bio, 0); |
c8b97818 | 637 | |
70b99e69 | 638 | page->mapping = NULL; |
4e4cbee9 | 639 | if (submit || bio_add_page(comp_bio, page, PAGE_SIZE, 0) < |
09cbfeaf | 640 | PAGE_SIZE) { |
c8b97818 CM |
641 | bio_get(comp_bio); |
642 | ||
0b246afa JM |
643 | ret = btrfs_bio_wq_end_io(fs_info, comp_bio, |
644 | BTRFS_WQ_ENDIO_DATA); | |
79787eaa | 645 | BUG_ON(ret); /* -ENOMEM */ |
c8b97818 | 646 | |
af09abfe CM |
647 | /* |
648 | * inc the count before we submit the bio so | |
649 | * we know the end IO handler won't happen before | |
650 | * we inc the count. Otherwise, the cb might get | |
651 | * freed before we're done setting it up | |
652 | */ | |
a50299ae | 653 | refcount_inc(&cb->pending_bios); |
af09abfe | 654 | |
6cbff00f | 655 | if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) { |
2ff7e61e JM |
656 | ret = btrfs_lookup_bio_sums(inode, comp_bio, |
657 | sums); | |
79787eaa | 658 | BUG_ON(ret); /* -ENOMEM */ |
d20f7043 | 659 | } |
ed6078f7 | 660 | sums += DIV_ROUND_UP(comp_bio->bi_iter.bi_size, |
0b246afa | 661 | fs_info->sectorsize); |
d20f7043 | 662 | |
2ff7e61e | 663 | ret = btrfs_map_bio(fs_info, comp_bio, mirror_num, 0); |
4246a0b6 | 664 | if (ret) { |
4e4cbee9 | 665 | comp_bio->bi_status = ret; |
4246a0b6 CH |
666 | bio_endio(comp_bio); |
667 | } | |
c8b97818 CM |
668 | |
669 | bio_put(comp_bio); | |
670 | ||
c821e7f3 | 671 | comp_bio = btrfs_bio_alloc(bdev, cur_disk_byte); |
37226b21 | 672 | bio_set_op_attrs(comp_bio, REQ_OP_READ, 0); |
771ed689 CM |
673 | comp_bio->bi_private = cb; |
674 | comp_bio->bi_end_io = end_compressed_bio_read; | |
675 | ||
09cbfeaf | 676 | bio_add_page(comp_bio, page, PAGE_SIZE, 0); |
c8b97818 | 677 | } |
09cbfeaf | 678 | cur_disk_byte += PAGE_SIZE; |
c8b97818 CM |
679 | } |
680 | bio_get(comp_bio); | |
681 | ||
0b246afa | 682 | ret = btrfs_bio_wq_end_io(fs_info, comp_bio, BTRFS_WQ_ENDIO_DATA); |
79787eaa | 683 | BUG_ON(ret); /* -ENOMEM */ |
c8b97818 | 684 | |
c2db1073 | 685 | if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) { |
2ff7e61e | 686 | ret = btrfs_lookup_bio_sums(inode, comp_bio, sums); |
79787eaa | 687 | BUG_ON(ret); /* -ENOMEM */ |
c2db1073 | 688 | } |
d20f7043 | 689 | |
2ff7e61e | 690 | ret = btrfs_map_bio(fs_info, comp_bio, mirror_num, 0); |
4246a0b6 | 691 | if (ret) { |
4e4cbee9 | 692 | comp_bio->bi_status = ret; |
4246a0b6 CH |
693 | bio_endio(comp_bio); |
694 | } | |
c8b97818 CM |
695 | |
696 | bio_put(comp_bio); | |
697 | return 0; | |
6b82ce8d | 698 | |
699 | fail2: | |
15e3004a JB |
700 | while (faili >= 0) { |
701 | __free_page(cb->compressed_pages[faili]); | |
702 | faili--; | |
703 | } | |
6b82ce8d | 704 | |
705 | kfree(cb->compressed_pages); | |
706 | fail1: | |
707 | kfree(cb); | |
708 | out: | |
709 | free_extent_map(em); | |
710 | return ret; | |
c8b97818 | 711 | } |
261507a0 | 712 | |
17b5a6c1 TT |
713 | /* |
714 | * Heuristic uses systematic sampling to collect data from the input data | |
715 | * range, the logic can be tuned by the following constants: | |
716 | * | |
717 | * @SAMPLING_READ_SIZE - how many bytes will be copied from for each sample | |
718 | * @SAMPLING_INTERVAL - range from which the sampled data can be collected | |
719 | */ | |
720 | #define SAMPLING_READ_SIZE (16) | |
721 | #define SAMPLING_INTERVAL (256) | |
722 | ||
723 | /* | |
724 | * For statistical analysis of the input data we consider bytes that form a | |
725 | * Galois Field of 256 objects. Each object has an attribute count, ie. how | |
726 | * many times the object appeared in the sample. | |
727 | */ | |
728 | #define BUCKET_SIZE (256) | |
729 | ||
730 | /* | |
731 | * The size of the sample is based on a statistical sampling rule of thumb. | |
732 | * The common way is to perform sampling tests as long as the number of | |
733 | * elements in each cell is at least 5. | |
734 | * | |
735 | * Instead of 5, we choose 32 to obtain more accurate results. | |
736 | * If the data contain the maximum number of symbols, which is 256, we obtain a | |
737 | * sample size bound by 8192. | |
738 | * | |
739 | * For a sample of at most 8KB of data per data range: 16 consecutive bytes | |
740 | * from up to 512 locations. | |
741 | */ | |
742 | #define MAX_SAMPLE_SIZE (BTRFS_MAX_UNCOMPRESSED * \ | |
743 | SAMPLING_READ_SIZE / SAMPLING_INTERVAL) | |
744 | ||
745 | struct bucket_item { | |
746 | u32 count; | |
747 | }; | |
4e439a0b TT |
748 | |
749 | struct heuristic_ws { | |
17b5a6c1 TT |
750 | /* Partial copy of input data */ |
751 | u8 *sample; | |
a440d48c | 752 | u32 sample_size; |
17b5a6c1 TT |
753 | /* Buckets store counters for each byte value */ |
754 | struct bucket_item *bucket; | |
4e439a0b TT |
755 | struct list_head list; |
756 | }; | |
757 | ||
758 | static void free_heuristic_ws(struct list_head *ws) | |
759 | { | |
760 | struct heuristic_ws *workspace; | |
761 | ||
762 | workspace = list_entry(ws, struct heuristic_ws, list); | |
763 | ||
17b5a6c1 TT |
764 | kvfree(workspace->sample); |
765 | kfree(workspace->bucket); | |
4e439a0b TT |
766 | kfree(workspace); |
767 | } | |
768 | ||
769 | static struct list_head *alloc_heuristic_ws(void) | |
770 | { | |
771 | struct heuristic_ws *ws; | |
772 | ||
773 | ws = kzalloc(sizeof(*ws), GFP_KERNEL); | |
774 | if (!ws) | |
775 | return ERR_PTR(-ENOMEM); | |
776 | ||
17b5a6c1 TT |
777 | ws->sample = kvmalloc(MAX_SAMPLE_SIZE, GFP_KERNEL); |
778 | if (!ws->sample) | |
779 | goto fail; | |
780 | ||
781 | ws->bucket = kcalloc(BUCKET_SIZE, sizeof(*ws->bucket), GFP_KERNEL); | |
782 | if (!ws->bucket) | |
783 | goto fail; | |
4e439a0b | 784 | |
17b5a6c1 | 785 | INIT_LIST_HEAD(&ws->list); |
4e439a0b | 786 | return &ws->list; |
17b5a6c1 TT |
787 | fail: |
788 | free_heuristic_ws(&ws->list); | |
789 | return ERR_PTR(-ENOMEM); | |
4e439a0b TT |
790 | } |
791 | ||
792 | struct workspaces_list { | |
d9187649 BL |
793 | struct list_head idle_ws; |
794 | spinlock_t ws_lock; | |
6ac10a6a DS |
795 | /* Number of free workspaces */ |
796 | int free_ws; | |
797 | /* Total number of allocated workspaces */ | |
798 | atomic_t total_ws; | |
799 | /* Waiters for a free workspace */ | |
d9187649 | 800 | wait_queue_head_t ws_wait; |
4e439a0b TT |
801 | }; |
802 | ||
803 | static struct workspaces_list btrfs_comp_ws[BTRFS_COMPRESS_TYPES]; | |
804 | ||
805 | static struct workspaces_list btrfs_heuristic_ws; | |
261507a0 | 806 | |
e8c9f186 | 807 | static const struct btrfs_compress_op * const btrfs_compress_op[] = { |
261507a0 | 808 | &btrfs_zlib_compress, |
a6fa6fae | 809 | &btrfs_lzo_compress, |
5c1aab1d | 810 | &btrfs_zstd_compress, |
261507a0 LZ |
811 | }; |
812 | ||
143bede5 | 813 | void __init btrfs_init_compress(void) |
261507a0 | 814 | { |
4e439a0b | 815 | struct list_head *workspace; |
261507a0 LZ |
816 | int i; |
817 | ||
4e439a0b TT |
818 | INIT_LIST_HEAD(&btrfs_heuristic_ws.idle_ws); |
819 | spin_lock_init(&btrfs_heuristic_ws.ws_lock); | |
820 | atomic_set(&btrfs_heuristic_ws.total_ws, 0); | |
821 | init_waitqueue_head(&btrfs_heuristic_ws.ws_wait); | |
f77dd0d6 | 822 | |
4e439a0b TT |
823 | workspace = alloc_heuristic_ws(); |
824 | if (IS_ERR(workspace)) { | |
825 | pr_warn( | |
826 | "BTRFS: cannot preallocate heuristic workspace, will try later\n"); | |
827 | } else { | |
828 | atomic_set(&btrfs_heuristic_ws.total_ws, 1); | |
829 | btrfs_heuristic_ws.free_ws = 1; | |
830 | list_add(workspace, &btrfs_heuristic_ws.idle_ws); | |
831 | } | |
832 | ||
833 | for (i = 0; i < BTRFS_COMPRESS_TYPES; i++) { | |
d9187649 BL |
834 | INIT_LIST_HEAD(&btrfs_comp_ws[i].idle_ws); |
835 | spin_lock_init(&btrfs_comp_ws[i].ws_lock); | |
6ac10a6a | 836 | atomic_set(&btrfs_comp_ws[i].total_ws, 0); |
d9187649 | 837 | init_waitqueue_head(&btrfs_comp_ws[i].ws_wait); |
f77dd0d6 DS |
838 | |
839 | /* | |
840 | * Preallocate one workspace for each compression type so | |
841 | * we can guarantee forward progress in the worst case | |
842 | */ | |
843 | workspace = btrfs_compress_op[i]->alloc_workspace(); | |
844 | if (IS_ERR(workspace)) { | |
62e85577 | 845 | pr_warn("BTRFS: cannot preallocate compression workspace, will try later\n"); |
f77dd0d6 DS |
846 | } else { |
847 | atomic_set(&btrfs_comp_ws[i].total_ws, 1); | |
848 | btrfs_comp_ws[i].free_ws = 1; | |
849 | list_add(workspace, &btrfs_comp_ws[i].idle_ws); | |
850 | } | |
261507a0 | 851 | } |
261507a0 LZ |
852 | } |
853 | ||
854 | /* | |
e721e49d DS |
855 | * This finds an available workspace or allocates a new one. |
856 | * If it's not possible to allocate a new one, waits until there's one. | |
857 | * Preallocation makes a forward progress guarantees and we do not return | |
858 | * errors. | |
261507a0 | 859 | */ |
4e439a0b | 860 | static struct list_head *__find_workspace(int type, bool heuristic) |
261507a0 LZ |
861 | { |
862 | struct list_head *workspace; | |
863 | int cpus = num_online_cpus(); | |
864 | int idx = type - 1; | |
fe308533 | 865 | unsigned nofs_flag; |
4e439a0b TT |
866 | struct list_head *idle_ws; |
867 | spinlock_t *ws_lock; | |
868 | atomic_t *total_ws; | |
869 | wait_queue_head_t *ws_wait; | |
870 | int *free_ws; | |
871 | ||
872 | if (heuristic) { | |
873 | idle_ws = &btrfs_heuristic_ws.idle_ws; | |
874 | ws_lock = &btrfs_heuristic_ws.ws_lock; | |
875 | total_ws = &btrfs_heuristic_ws.total_ws; | |
876 | ws_wait = &btrfs_heuristic_ws.ws_wait; | |
877 | free_ws = &btrfs_heuristic_ws.free_ws; | |
878 | } else { | |
879 | idle_ws = &btrfs_comp_ws[idx].idle_ws; | |
880 | ws_lock = &btrfs_comp_ws[idx].ws_lock; | |
881 | total_ws = &btrfs_comp_ws[idx].total_ws; | |
882 | ws_wait = &btrfs_comp_ws[idx].ws_wait; | |
883 | free_ws = &btrfs_comp_ws[idx].free_ws; | |
884 | } | |
261507a0 | 885 | |
261507a0 | 886 | again: |
d9187649 BL |
887 | spin_lock(ws_lock); |
888 | if (!list_empty(idle_ws)) { | |
889 | workspace = idle_ws->next; | |
261507a0 | 890 | list_del(workspace); |
6ac10a6a | 891 | (*free_ws)--; |
d9187649 | 892 | spin_unlock(ws_lock); |
261507a0 LZ |
893 | return workspace; |
894 | ||
895 | } | |
6ac10a6a | 896 | if (atomic_read(total_ws) > cpus) { |
261507a0 LZ |
897 | DEFINE_WAIT(wait); |
898 | ||
d9187649 BL |
899 | spin_unlock(ws_lock); |
900 | prepare_to_wait(ws_wait, &wait, TASK_UNINTERRUPTIBLE); | |
6ac10a6a | 901 | if (atomic_read(total_ws) > cpus && !*free_ws) |
261507a0 | 902 | schedule(); |
d9187649 | 903 | finish_wait(ws_wait, &wait); |
261507a0 LZ |
904 | goto again; |
905 | } | |
6ac10a6a | 906 | atomic_inc(total_ws); |
d9187649 | 907 | spin_unlock(ws_lock); |
261507a0 | 908 | |
fe308533 DS |
909 | /* |
910 | * Allocation helpers call vmalloc that can't use GFP_NOFS, so we have | |
911 | * to turn it off here because we might get called from the restricted | |
912 | * context of btrfs_compress_bio/btrfs_compress_pages | |
913 | */ | |
914 | nofs_flag = memalloc_nofs_save(); | |
4e439a0b TT |
915 | if (heuristic) |
916 | workspace = alloc_heuristic_ws(); | |
917 | else | |
918 | workspace = btrfs_compress_op[idx]->alloc_workspace(); | |
fe308533 DS |
919 | memalloc_nofs_restore(nofs_flag); |
920 | ||
261507a0 | 921 | if (IS_ERR(workspace)) { |
6ac10a6a | 922 | atomic_dec(total_ws); |
d9187649 | 923 | wake_up(ws_wait); |
e721e49d DS |
924 | |
925 | /* | |
926 | * Do not return the error but go back to waiting. There's a | |
927 | * workspace preallocated for each type and the compression | |
928 | * time is bounded so we get to a workspace eventually. This | |
929 | * makes our caller's life easier. | |
52356716 DS |
930 | * |
931 | * To prevent silent and low-probability deadlocks (when the | |
932 | * initial preallocation fails), check if there are any | |
933 | * workspaces at all. | |
e721e49d | 934 | */ |
52356716 DS |
935 | if (atomic_read(total_ws) == 0) { |
936 | static DEFINE_RATELIMIT_STATE(_rs, | |
937 | /* once per minute */ 60 * HZ, | |
938 | /* no burst */ 1); | |
939 | ||
940 | if (__ratelimit(&_rs)) { | |
ab8d0fc4 | 941 | pr_warn("BTRFS: no compression workspaces, low memory, retrying\n"); |
52356716 DS |
942 | } |
943 | } | |
e721e49d | 944 | goto again; |
261507a0 LZ |
945 | } |
946 | return workspace; | |
947 | } | |
948 | ||
4e439a0b TT |
949 | static struct list_head *find_workspace(int type) |
950 | { | |
951 | return __find_workspace(type, false); | |
952 | } | |
953 | ||
261507a0 LZ |
954 | /* |
955 | * put a workspace struct back on the list or free it if we have enough | |
956 | * idle ones sitting around | |
957 | */ | |
4e439a0b TT |
958 | static void __free_workspace(int type, struct list_head *workspace, |
959 | bool heuristic) | |
261507a0 LZ |
960 | { |
961 | int idx = type - 1; | |
4e439a0b TT |
962 | struct list_head *idle_ws; |
963 | spinlock_t *ws_lock; | |
964 | atomic_t *total_ws; | |
965 | wait_queue_head_t *ws_wait; | |
966 | int *free_ws; | |
967 | ||
968 | if (heuristic) { | |
969 | idle_ws = &btrfs_heuristic_ws.idle_ws; | |
970 | ws_lock = &btrfs_heuristic_ws.ws_lock; | |
971 | total_ws = &btrfs_heuristic_ws.total_ws; | |
972 | ws_wait = &btrfs_heuristic_ws.ws_wait; | |
973 | free_ws = &btrfs_heuristic_ws.free_ws; | |
974 | } else { | |
975 | idle_ws = &btrfs_comp_ws[idx].idle_ws; | |
976 | ws_lock = &btrfs_comp_ws[idx].ws_lock; | |
977 | total_ws = &btrfs_comp_ws[idx].total_ws; | |
978 | ws_wait = &btrfs_comp_ws[idx].ws_wait; | |
979 | free_ws = &btrfs_comp_ws[idx].free_ws; | |
980 | } | |
d9187649 BL |
981 | |
982 | spin_lock(ws_lock); | |
26b28dce | 983 | if (*free_ws <= num_online_cpus()) { |
d9187649 | 984 | list_add(workspace, idle_ws); |
6ac10a6a | 985 | (*free_ws)++; |
d9187649 | 986 | spin_unlock(ws_lock); |
261507a0 LZ |
987 | goto wake; |
988 | } | |
d9187649 | 989 | spin_unlock(ws_lock); |
261507a0 | 990 | |
4e439a0b TT |
991 | if (heuristic) |
992 | free_heuristic_ws(workspace); | |
993 | else | |
994 | btrfs_compress_op[idx]->free_workspace(workspace); | |
6ac10a6a | 995 | atomic_dec(total_ws); |
261507a0 | 996 | wake: |
a83342aa DS |
997 | /* |
998 | * Make sure counter is updated before we wake up waiters. | |
999 | */ | |
66657b31 | 1000 | smp_mb(); |
d9187649 BL |
1001 | if (waitqueue_active(ws_wait)) |
1002 | wake_up(ws_wait); | |
261507a0 LZ |
1003 | } |
1004 | ||
4e439a0b TT |
1005 | static void free_workspace(int type, struct list_head *ws) |
1006 | { | |
1007 | return __free_workspace(type, ws, false); | |
1008 | } | |
1009 | ||
261507a0 LZ |
1010 | /* |
1011 | * cleanup function for module exit | |
1012 | */ | |
1013 | static void free_workspaces(void) | |
1014 | { | |
1015 | struct list_head *workspace; | |
1016 | int i; | |
1017 | ||
4e439a0b TT |
1018 | while (!list_empty(&btrfs_heuristic_ws.idle_ws)) { |
1019 | workspace = btrfs_heuristic_ws.idle_ws.next; | |
1020 | list_del(workspace); | |
1021 | free_heuristic_ws(workspace); | |
1022 | atomic_dec(&btrfs_heuristic_ws.total_ws); | |
1023 | } | |
1024 | ||
261507a0 | 1025 | for (i = 0; i < BTRFS_COMPRESS_TYPES; i++) { |
d9187649 BL |
1026 | while (!list_empty(&btrfs_comp_ws[i].idle_ws)) { |
1027 | workspace = btrfs_comp_ws[i].idle_ws.next; | |
261507a0 LZ |
1028 | list_del(workspace); |
1029 | btrfs_compress_op[i]->free_workspace(workspace); | |
6ac10a6a | 1030 | atomic_dec(&btrfs_comp_ws[i].total_ws); |
261507a0 LZ |
1031 | } |
1032 | } | |
1033 | } | |
1034 | ||
1035 | /* | |
38c31464 DS |
1036 | * Given an address space and start and length, compress the bytes into @pages |
1037 | * that are allocated on demand. | |
261507a0 | 1038 | * |
f51d2b59 DS |
1039 | * @type_level is encoded algorithm and level, where level 0 means whatever |
1040 | * default the algorithm chooses and is opaque here; | |
1041 | * - compression algo are 0-3 | |
1042 | * - the level are bits 4-7 | |
1043 | * | |
4d3a800e DS |
1044 | * @out_pages is an in/out parameter, holds maximum number of pages to allocate |
1045 | * and returns number of actually allocated pages | |
261507a0 | 1046 | * |
38c31464 DS |
1047 | * @total_in is used to return the number of bytes actually read. It |
1048 | * may be smaller than the input length if we had to exit early because we | |
261507a0 LZ |
1049 | * ran out of room in the pages array or because we cross the |
1050 | * max_out threshold. | |
1051 | * | |
38c31464 DS |
1052 | * @total_out is an in/out parameter, must be set to the input length and will |
1053 | * be also used to return the total number of compressed bytes | |
261507a0 | 1054 | * |
38c31464 | 1055 | * @max_out tells us the max number of bytes that we're allowed to |
261507a0 LZ |
1056 | * stuff into pages |
1057 | */ | |
f51d2b59 | 1058 | int btrfs_compress_pages(unsigned int type_level, struct address_space *mapping, |
38c31464 | 1059 | u64 start, struct page **pages, |
261507a0 LZ |
1060 | unsigned long *out_pages, |
1061 | unsigned long *total_in, | |
e5d74902 | 1062 | unsigned long *total_out) |
261507a0 LZ |
1063 | { |
1064 | struct list_head *workspace; | |
1065 | int ret; | |
f51d2b59 | 1066 | int type = type_level & 0xF; |
261507a0 LZ |
1067 | |
1068 | workspace = find_workspace(type); | |
261507a0 | 1069 | |
f51d2b59 | 1070 | btrfs_compress_op[type - 1]->set_level(workspace, type_level); |
261507a0 | 1071 | ret = btrfs_compress_op[type-1]->compress_pages(workspace, mapping, |
38c31464 | 1072 | start, pages, |
4d3a800e | 1073 | out_pages, |
e5d74902 | 1074 | total_in, total_out); |
261507a0 LZ |
1075 | free_workspace(type, workspace); |
1076 | return ret; | |
1077 | } | |
1078 | ||
1079 | /* | |
1080 | * pages_in is an array of pages with compressed data. | |
1081 | * | |
1082 | * disk_start is the starting logical offset of this array in the file | |
1083 | * | |
974b1adc | 1084 | * orig_bio contains the pages from the file that we want to decompress into |
261507a0 LZ |
1085 | * |
1086 | * srclen is the number of bytes in pages_in | |
1087 | * | |
1088 | * The basic idea is that we have a bio that was created by readpages. | |
1089 | * The pages in the bio are for the uncompressed data, and they may not | |
1090 | * be contiguous. They all correspond to the range of bytes covered by | |
1091 | * the compressed extent. | |
1092 | */ | |
8140dc30 | 1093 | static int btrfs_decompress_bio(struct compressed_bio *cb) |
261507a0 LZ |
1094 | { |
1095 | struct list_head *workspace; | |
1096 | int ret; | |
8140dc30 | 1097 | int type = cb->compress_type; |
261507a0 LZ |
1098 | |
1099 | workspace = find_workspace(type); | |
e1ddce71 | 1100 | ret = btrfs_compress_op[type - 1]->decompress_bio(workspace, cb); |
261507a0 | 1101 | free_workspace(type, workspace); |
e1ddce71 | 1102 | |
261507a0 LZ |
1103 | return ret; |
1104 | } | |
1105 | ||
1106 | /* | |
1107 | * a less complex decompression routine. Our compressed data fits in a | |
1108 | * single page, and we want to read a single page out of it. | |
1109 | * start_byte tells us the offset into the compressed data we're interested in | |
1110 | */ | |
1111 | int btrfs_decompress(int type, unsigned char *data_in, struct page *dest_page, | |
1112 | unsigned long start_byte, size_t srclen, size_t destlen) | |
1113 | { | |
1114 | struct list_head *workspace; | |
1115 | int ret; | |
1116 | ||
1117 | workspace = find_workspace(type); | |
261507a0 LZ |
1118 | |
1119 | ret = btrfs_compress_op[type-1]->decompress(workspace, data_in, | |
1120 | dest_page, start_byte, | |
1121 | srclen, destlen); | |
1122 | ||
1123 | free_workspace(type, workspace); | |
1124 | return ret; | |
1125 | } | |
1126 | ||
8e4eef7a | 1127 | void btrfs_exit_compress(void) |
261507a0 LZ |
1128 | { |
1129 | free_workspaces(); | |
1130 | } | |
3a39c18d LZ |
1131 | |
1132 | /* | |
1133 | * Copy uncompressed data from working buffer to pages. | |
1134 | * | |
1135 | * buf_start is the byte offset we're of the start of our workspace buffer. | |
1136 | * | |
1137 | * total_out is the last byte of the buffer | |
1138 | */ | |
14a3357b | 1139 | int btrfs_decompress_buf2page(const char *buf, unsigned long buf_start, |
3a39c18d | 1140 | unsigned long total_out, u64 disk_start, |
974b1adc | 1141 | struct bio *bio) |
3a39c18d LZ |
1142 | { |
1143 | unsigned long buf_offset; | |
1144 | unsigned long current_buf_start; | |
1145 | unsigned long start_byte; | |
6e78b3f7 | 1146 | unsigned long prev_start_byte; |
3a39c18d LZ |
1147 | unsigned long working_bytes = total_out - buf_start; |
1148 | unsigned long bytes; | |
1149 | char *kaddr; | |
974b1adc | 1150 | struct bio_vec bvec = bio_iter_iovec(bio, bio->bi_iter); |
3a39c18d LZ |
1151 | |
1152 | /* | |
1153 | * start byte is the first byte of the page we're currently | |
1154 | * copying into relative to the start of the compressed data. | |
1155 | */ | |
974b1adc | 1156 | start_byte = page_offset(bvec.bv_page) - disk_start; |
3a39c18d LZ |
1157 | |
1158 | /* we haven't yet hit data corresponding to this page */ | |
1159 | if (total_out <= start_byte) | |
1160 | return 1; | |
1161 | ||
1162 | /* | |
1163 | * the start of the data we care about is offset into | |
1164 | * the middle of our working buffer | |
1165 | */ | |
1166 | if (total_out > start_byte && buf_start < start_byte) { | |
1167 | buf_offset = start_byte - buf_start; | |
1168 | working_bytes -= buf_offset; | |
1169 | } else { | |
1170 | buf_offset = 0; | |
1171 | } | |
1172 | current_buf_start = buf_start; | |
1173 | ||
1174 | /* copy bytes from the working buffer into the pages */ | |
1175 | while (working_bytes > 0) { | |
974b1adc CH |
1176 | bytes = min_t(unsigned long, bvec.bv_len, |
1177 | PAGE_SIZE - buf_offset); | |
3a39c18d | 1178 | bytes = min(bytes, working_bytes); |
974b1adc CH |
1179 | |
1180 | kaddr = kmap_atomic(bvec.bv_page); | |
1181 | memcpy(kaddr + bvec.bv_offset, buf + buf_offset, bytes); | |
7ac687d9 | 1182 | kunmap_atomic(kaddr); |
974b1adc | 1183 | flush_dcache_page(bvec.bv_page); |
3a39c18d | 1184 | |
3a39c18d LZ |
1185 | buf_offset += bytes; |
1186 | working_bytes -= bytes; | |
1187 | current_buf_start += bytes; | |
1188 | ||
1189 | /* check if we need to pick another page */ | |
974b1adc CH |
1190 | bio_advance(bio, bytes); |
1191 | if (!bio->bi_iter.bi_size) | |
1192 | return 0; | |
1193 | bvec = bio_iter_iovec(bio, bio->bi_iter); | |
6e78b3f7 | 1194 | prev_start_byte = start_byte; |
974b1adc | 1195 | start_byte = page_offset(bvec.bv_page) - disk_start; |
3a39c18d | 1196 | |
974b1adc | 1197 | /* |
6e78b3f7 OS |
1198 | * We need to make sure we're only adjusting |
1199 | * our offset into compression working buffer when | |
1200 | * we're switching pages. Otherwise we can incorrectly | |
1201 | * keep copying when we were actually done. | |
974b1adc | 1202 | */ |
6e78b3f7 OS |
1203 | if (start_byte != prev_start_byte) { |
1204 | /* | |
1205 | * make sure our new page is covered by this | |
1206 | * working buffer | |
1207 | */ | |
1208 | if (total_out <= start_byte) | |
1209 | return 1; | |
3a39c18d | 1210 | |
6e78b3f7 OS |
1211 | /* |
1212 | * the next page in the biovec might not be adjacent | |
1213 | * to the last page, but it might still be found | |
1214 | * inside this working buffer. bump our offset pointer | |
1215 | */ | |
1216 | if (total_out > start_byte && | |
1217 | current_buf_start < start_byte) { | |
1218 | buf_offset = start_byte - buf_start; | |
1219 | working_bytes = total_out - start_byte; | |
1220 | current_buf_start = buf_start + buf_offset; | |
1221 | } | |
3a39c18d LZ |
1222 | } |
1223 | } | |
1224 | ||
1225 | return 1; | |
1226 | } | |
c2fcdcdf | 1227 | |
19562430 TT |
1228 | /* |
1229 | * Shannon Entropy calculation | |
1230 | * | |
1231 | * Pure byte distribution analysis fails to determine compressiability of data. | |
1232 | * Try calculating entropy to estimate the average minimum number of bits | |
1233 | * needed to encode the sampled data. | |
1234 | * | |
1235 | * For convenience, return the percentage of needed bits, instead of amount of | |
1236 | * bits directly. | |
1237 | * | |
1238 | * @ENTROPY_LVL_ACEPTABLE - below that threshold, sample has low byte entropy | |
1239 | * and can be compressible with high probability | |
1240 | * | |
1241 | * @ENTROPY_LVL_HIGH - data are not compressible with high probability | |
1242 | * | |
1243 | * Use of ilog2() decreases precision, we lower the LVL to 5 to compensate. | |
1244 | */ | |
1245 | #define ENTROPY_LVL_ACEPTABLE (65) | |
1246 | #define ENTROPY_LVL_HIGH (80) | |
1247 | ||
1248 | /* | |
1249 | * For increasead precision in shannon_entropy calculation, | |
1250 | * let's do pow(n, M) to save more digits after comma: | |
1251 | * | |
1252 | * - maximum int bit length is 64 | |
1253 | * - ilog2(MAX_SAMPLE_SIZE) -> 13 | |
1254 | * - 13 * 4 = 52 < 64 -> M = 4 | |
1255 | * | |
1256 | * So use pow(n, 4). | |
1257 | */ | |
1258 | static inline u32 ilog2_w(u64 n) | |
1259 | { | |
1260 | return ilog2(n * n * n * n); | |
1261 | } | |
1262 | ||
1263 | static u32 shannon_entropy(struct heuristic_ws *ws) | |
1264 | { | |
1265 | const u32 entropy_max = 8 * ilog2_w(2); | |
1266 | u32 entropy_sum = 0; | |
1267 | u32 p, p_base, sz_base; | |
1268 | u32 i; | |
1269 | ||
1270 | sz_base = ilog2_w(ws->sample_size); | |
1271 | for (i = 0; i < BUCKET_SIZE && ws->bucket[i].count > 0; i++) { | |
1272 | p = ws->bucket[i].count; | |
1273 | p_base = ilog2_w(p); | |
1274 | entropy_sum += p * (sz_base - p_base); | |
1275 | } | |
1276 | ||
1277 | entropy_sum /= ws->sample_size; | |
1278 | return entropy_sum * 100 / entropy_max; | |
1279 | } | |
1280 | ||
858177d3 TT |
1281 | /* Compare buckets by size, ascending */ |
1282 | static int bucket_comp_rev(const void *lv, const void *rv) | |
1283 | { | |
1284 | const struct bucket_item *l = (const struct bucket_item *)lv; | |
1285 | const struct bucket_item *r = (const struct bucket_item *)rv; | |
1286 | ||
1287 | return r->count - l->count; | |
1288 | } | |
1289 | ||
1290 | /* | |
1291 | * Size of the core byte set - how many bytes cover 90% of the sample | |
1292 | * | |
1293 | * There are several types of structured binary data that use nearly all byte | |
1294 | * values. The distribution can be uniform and counts in all buckets will be | |
1295 | * nearly the same (eg. encrypted data). Unlikely to be compressible. | |
1296 | * | |
1297 | * Other possibility is normal (Gaussian) distribution, where the data could | |
1298 | * be potentially compressible, but we have to take a few more steps to decide | |
1299 | * how much. | |
1300 | * | |
1301 | * @BYTE_CORE_SET_LOW - main part of byte values repeated frequently, | |
1302 | * compression algo can easy fix that | |
1303 | * @BYTE_CORE_SET_HIGH - data have uniform distribution and with high | |
1304 | * probability is not compressible | |
1305 | */ | |
1306 | #define BYTE_CORE_SET_LOW (64) | |
1307 | #define BYTE_CORE_SET_HIGH (200) | |
1308 | ||
1309 | static int byte_core_set_size(struct heuristic_ws *ws) | |
1310 | { | |
1311 | u32 i; | |
1312 | u32 coreset_sum = 0; | |
1313 | const u32 core_set_threshold = ws->sample_size * 90 / 100; | |
1314 | struct bucket_item *bucket = ws->bucket; | |
1315 | ||
1316 | /* Sort in reverse order */ | |
1317 | sort(bucket, BUCKET_SIZE, sizeof(*bucket), &bucket_comp_rev, NULL); | |
1318 | ||
1319 | for (i = 0; i < BYTE_CORE_SET_LOW; i++) | |
1320 | coreset_sum += bucket[i].count; | |
1321 | ||
1322 | if (coreset_sum > core_set_threshold) | |
1323 | return i; | |
1324 | ||
1325 | for (; i < BYTE_CORE_SET_HIGH && bucket[i].count > 0; i++) { | |
1326 | coreset_sum += bucket[i].count; | |
1327 | if (coreset_sum > core_set_threshold) | |
1328 | break; | |
1329 | } | |
1330 | ||
1331 | return i; | |
1332 | } | |
1333 | ||
a288e92c TT |
1334 | /* |
1335 | * Count byte values in buckets. | |
1336 | * This heuristic can detect textual data (configs, xml, json, html, etc). | |
1337 | * Because in most text-like data byte set is restricted to limited number of | |
1338 | * possible characters, and that restriction in most cases makes data easy to | |
1339 | * compress. | |
1340 | * | |
1341 | * @BYTE_SET_THRESHOLD - consider all data within this byte set size: | |
1342 | * less - compressible | |
1343 | * more - need additional analysis | |
1344 | */ | |
1345 | #define BYTE_SET_THRESHOLD (64) | |
1346 | ||
1347 | static u32 byte_set_size(const struct heuristic_ws *ws) | |
1348 | { | |
1349 | u32 i; | |
1350 | u32 byte_set_size = 0; | |
1351 | ||
1352 | for (i = 0; i < BYTE_SET_THRESHOLD; i++) { | |
1353 | if (ws->bucket[i].count > 0) | |
1354 | byte_set_size++; | |
1355 | } | |
1356 | ||
1357 | /* | |
1358 | * Continue collecting count of byte values in buckets. If the byte | |
1359 | * set size is bigger then the threshold, it's pointless to continue, | |
1360 | * the detection technique would fail for this type of data. | |
1361 | */ | |
1362 | for (; i < BUCKET_SIZE; i++) { | |
1363 | if (ws->bucket[i].count > 0) { | |
1364 | byte_set_size++; | |
1365 | if (byte_set_size > BYTE_SET_THRESHOLD) | |
1366 | return byte_set_size; | |
1367 | } | |
1368 | } | |
1369 | ||
1370 | return byte_set_size; | |
1371 | } | |
1372 | ||
1fe4f6fa TT |
1373 | static bool sample_repeated_patterns(struct heuristic_ws *ws) |
1374 | { | |
1375 | const u32 half_of_sample = ws->sample_size / 2; | |
1376 | const u8 *data = ws->sample; | |
1377 | ||
1378 | return memcmp(&data[0], &data[half_of_sample], half_of_sample) == 0; | |
1379 | } | |
1380 | ||
a440d48c TT |
1381 | static void heuristic_collect_sample(struct inode *inode, u64 start, u64 end, |
1382 | struct heuristic_ws *ws) | |
1383 | { | |
1384 | struct page *page; | |
1385 | u64 index, index_end; | |
1386 | u32 i, curr_sample_pos; | |
1387 | u8 *in_data; | |
1388 | ||
1389 | /* | |
1390 | * Compression handles the input data by chunks of 128KiB | |
1391 | * (defined by BTRFS_MAX_UNCOMPRESSED) | |
1392 | * | |
1393 | * We do the same for the heuristic and loop over the whole range. | |
1394 | * | |
1395 | * MAX_SAMPLE_SIZE - calculated under assumption that heuristic will | |
1396 | * process no more than BTRFS_MAX_UNCOMPRESSED at a time. | |
1397 | */ | |
1398 | if (end - start > BTRFS_MAX_UNCOMPRESSED) | |
1399 | end = start + BTRFS_MAX_UNCOMPRESSED; | |
1400 | ||
1401 | index = start >> PAGE_SHIFT; | |
1402 | index_end = end >> PAGE_SHIFT; | |
1403 | ||
1404 | /* Don't miss unaligned end */ | |
1405 | if (!IS_ALIGNED(end, PAGE_SIZE)) | |
1406 | index_end++; | |
1407 | ||
1408 | curr_sample_pos = 0; | |
1409 | while (index < index_end) { | |
1410 | page = find_get_page(inode->i_mapping, index); | |
1411 | in_data = kmap(page); | |
1412 | /* Handle case where the start is not aligned to PAGE_SIZE */ | |
1413 | i = start % PAGE_SIZE; | |
1414 | while (i < PAGE_SIZE - SAMPLING_READ_SIZE) { | |
1415 | /* Don't sample any garbage from the last page */ | |
1416 | if (start > end - SAMPLING_READ_SIZE) | |
1417 | break; | |
1418 | memcpy(&ws->sample[curr_sample_pos], &in_data[i], | |
1419 | SAMPLING_READ_SIZE); | |
1420 | i += SAMPLING_INTERVAL; | |
1421 | start += SAMPLING_INTERVAL; | |
1422 | curr_sample_pos += SAMPLING_READ_SIZE; | |
1423 | } | |
1424 | kunmap(page); | |
1425 | put_page(page); | |
1426 | ||
1427 | index++; | |
1428 | } | |
1429 | ||
1430 | ws->sample_size = curr_sample_pos; | |
1431 | } | |
1432 | ||
c2fcdcdf TT |
1433 | /* |
1434 | * Compression heuristic. | |
1435 | * | |
1436 | * For now is's a naive and optimistic 'return true', we'll extend the logic to | |
1437 | * quickly (compared to direct compression) detect data characteristics | |
1438 | * (compressible/uncompressible) to avoid wasting CPU time on uncompressible | |
1439 | * data. | |
1440 | * | |
1441 | * The following types of analysis can be performed: | |
1442 | * - detect mostly zero data | |
1443 | * - detect data with low "byte set" size (text, etc) | |
1444 | * - detect data with low/high "core byte" set | |
1445 | * | |
1446 | * Return non-zero if the compression should be done, 0 otherwise. | |
1447 | */ | |
1448 | int btrfs_compress_heuristic(struct inode *inode, u64 start, u64 end) | |
1449 | { | |
4e439a0b TT |
1450 | struct list_head *ws_list = __find_workspace(0, true); |
1451 | struct heuristic_ws *ws; | |
a440d48c TT |
1452 | u32 i; |
1453 | u8 byte; | |
19562430 | 1454 | int ret = 0; |
c2fcdcdf | 1455 | |
4e439a0b TT |
1456 | ws = list_entry(ws_list, struct heuristic_ws, list); |
1457 | ||
a440d48c TT |
1458 | heuristic_collect_sample(inode, start, end, ws); |
1459 | ||
1fe4f6fa TT |
1460 | if (sample_repeated_patterns(ws)) { |
1461 | ret = 1; | |
1462 | goto out; | |
1463 | } | |
1464 | ||
a440d48c TT |
1465 | memset(ws->bucket, 0, sizeof(*ws->bucket)*BUCKET_SIZE); |
1466 | ||
1467 | for (i = 0; i < ws->sample_size; i++) { | |
1468 | byte = ws->sample[i]; | |
1469 | ws->bucket[byte].count++; | |
c2fcdcdf TT |
1470 | } |
1471 | ||
a288e92c TT |
1472 | i = byte_set_size(ws); |
1473 | if (i < BYTE_SET_THRESHOLD) { | |
1474 | ret = 2; | |
1475 | goto out; | |
1476 | } | |
1477 | ||
858177d3 TT |
1478 | i = byte_core_set_size(ws); |
1479 | if (i <= BYTE_CORE_SET_LOW) { | |
1480 | ret = 3; | |
1481 | goto out; | |
1482 | } | |
1483 | ||
1484 | if (i >= BYTE_CORE_SET_HIGH) { | |
1485 | ret = 0; | |
1486 | goto out; | |
1487 | } | |
1488 | ||
19562430 TT |
1489 | i = shannon_entropy(ws); |
1490 | if (i <= ENTROPY_LVL_ACEPTABLE) { | |
1491 | ret = 4; | |
1492 | goto out; | |
1493 | } | |
1494 | ||
1495 | /* | |
1496 | * For the levels below ENTROPY_LVL_HIGH, additional analysis would be | |
1497 | * needed to give green light to compression. | |
1498 | * | |
1499 | * For now just assume that compression at that level is not worth the | |
1500 | * resources because: | |
1501 | * | |
1502 | * 1. it is possible to defrag the data later | |
1503 | * | |
1504 | * 2. the data would turn out to be hardly compressible, eg. 150 byte | |
1505 | * values, every bucket has counter at level ~54. The heuristic would | |
1506 | * be confused. This can happen when data have some internal repeated | |
1507 | * patterns like "abbacbbc...". This can be detected by analyzing | |
1508 | * pairs of bytes, which is too costly. | |
1509 | */ | |
1510 | if (i < ENTROPY_LVL_HIGH) { | |
1511 | ret = 5; | |
1512 | goto out; | |
1513 | } else { | |
1514 | ret = 0; | |
1515 | goto out; | |
1516 | } | |
1517 | ||
1fe4f6fa | 1518 | out: |
4e439a0b | 1519 | __free_workspace(0, ws_list, true); |
c2fcdcdf TT |
1520 | return ret; |
1521 | } | |
f51d2b59 DS |
1522 | |
1523 | unsigned int btrfs_compress_str2level(const char *str) | |
1524 | { | |
1525 | if (strncmp(str, "zlib", 4) != 0) | |
1526 | return 0; | |
1527 | ||
fa4d885a AB |
1528 | /* Accepted form: zlib:1 up to zlib:9 and nothing left after the number */ |
1529 | if (str[4] == ':' && '1' <= str[5] && str[5] <= '9' && str[6] == 0) | |
1530 | return str[5] - '0'; | |
f51d2b59 | 1531 | |
eae8d825 | 1532 | return BTRFS_ZLIB_DEFAULT_LEVEL; |
f51d2b59 | 1533 | } |