Commit | Line | Data |
---|---|---|
c1d7c514 | 1 | // SPDX-License-Identifier: GPL-2.0 |
6cbd5570 CM |
2 | /* |
3 | * Copyright (C) 2007 Oracle. All rights reserved. | |
6cbd5570 CM |
4 | */ |
5 | ||
39279cc3 CM |
6 | #include <linux/fs.h> |
7 | #include <linux/pagemap.h> | |
39279cc3 CM |
8 | #include <linux/time.h> |
9 | #include <linux/init.h> | |
10 | #include <linux/string.h> | |
39279cc3 | 11 | #include <linux/backing-dev.h> |
2fe17c10 | 12 | #include <linux/falloc.h> |
39279cc3 | 13 | #include <linux/writeback.h> |
39279cc3 | 14 | #include <linux/compat.h> |
5a0e3ad6 | 15 | #include <linux/slab.h> |
55e301fd | 16 | #include <linux/btrfs.h> |
e2e40f2c | 17 | #include <linux/uio.h> |
ae5e165d | 18 | #include <linux/iversion.h> |
39279cc3 CM |
19 | #include "ctree.h" |
20 | #include "disk-io.h" | |
21 | #include "transaction.h" | |
22 | #include "btrfs_inode.h" | |
39279cc3 | 23 | #include "print-tree.h" |
e02119d5 CM |
24 | #include "tree-log.h" |
25 | #include "locking.h" | |
2aaa6655 | 26 | #include "volumes.h" |
fcebe456 | 27 | #include "qgroup.h" |
ebb8765b | 28 | #include "compression.h" |
39279cc3 | 29 | |
9247f317 | 30 | static struct kmem_cache *btrfs_inode_defrag_cachep; |
4cb5300b CM |
31 | /* |
32 | * when auto defrag is enabled we | |
33 | * queue up these defrag structs to remember which | |
34 | * inodes need defragging passes | |
35 | */ | |
36 | struct inode_defrag { | |
37 | struct rb_node rb_node; | |
38 | /* objectid */ | |
39 | u64 ino; | |
40 | /* | |
41 | * transid where the defrag was added, we search for | |
42 | * extents newer than this | |
43 | */ | |
44 | u64 transid; | |
45 | ||
46 | /* root objectid */ | |
47 | u64 root; | |
48 | ||
49 | /* last offset we were able to defrag */ | |
50 | u64 last_offset; | |
51 | ||
52 | /* if we've wrapped around back to zero once already */ | |
53 | int cycled; | |
54 | }; | |
55 | ||
762f2263 MX |
56 | static int __compare_inode_defrag(struct inode_defrag *defrag1, |
57 | struct inode_defrag *defrag2) | |
58 | { | |
59 | if (defrag1->root > defrag2->root) | |
60 | return 1; | |
61 | else if (defrag1->root < defrag2->root) | |
62 | return -1; | |
63 | else if (defrag1->ino > defrag2->ino) | |
64 | return 1; | |
65 | else if (defrag1->ino < defrag2->ino) | |
66 | return -1; | |
67 | else | |
68 | return 0; | |
69 | } | |
70 | ||
4cb5300b CM |
71 | /* pop a record for an inode into the defrag tree. The lock |
72 | * must be held already | |
73 | * | |
74 | * If you're inserting a record for an older transid than an | |
75 | * existing record, the transid already in the tree is lowered | |
76 | * | |
77 | * If an existing record is found the defrag item you | |
78 | * pass in is freed | |
79 | */ | |
6158e1ce | 80 | static int __btrfs_add_inode_defrag(struct btrfs_inode *inode, |
4cb5300b CM |
81 | struct inode_defrag *defrag) |
82 | { | |
3ffbd68c | 83 | struct btrfs_fs_info *fs_info = inode->root->fs_info; |
4cb5300b CM |
84 | struct inode_defrag *entry; |
85 | struct rb_node **p; | |
86 | struct rb_node *parent = NULL; | |
762f2263 | 87 | int ret; |
4cb5300b | 88 | |
0b246afa | 89 | p = &fs_info->defrag_inodes.rb_node; |
4cb5300b CM |
90 | while (*p) { |
91 | parent = *p; | |
92 | entry = rb_entry(parent, struct inode_defrag, rb_node); | |
93 | ||
762f2263 MX |
94 | ret = __compare_inode_defrag(defrag, entry); |
95 | if (ret < 0) | |
4cb5300b | 96 | p = &parent->rb_left; |
762f2263 | 97 | else if (ret > 0) |
4cb5300b CM |
98 | p = &parent->rb_right; |
99 | else { | |
100 | /* if we're reinserting an entry for | |
101 | * an old defrag run, make sure to | |
102 | * lower the transid of our existing record | |
103 | */ | |
104 | if (defrag->transid < entry->transid) | |
105 | entry->transid = defrag->transid; | |
106 | if (defrag->last_offset > entry->last_offset) | |
107 | entry->last_offset = defrag->last_offset; | |
8ddc4734 | 108 | return -EEXIST; |
4cb5300b CM |
109 | } |
110 | } | |
6158e1ce | 111 | set_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags); |
4cb5300b | 112 | rb_link_node(&defrag->rb_node, parent, p); |
0b246afa | 113 | rb_insert_color(&defrag->rb_node, &fs_info->defrag_inodes); |
8ddc4734 MX |
114 | return 0; |
115 | } | |
4cb5300b | 116 | |
2ff7e61e | 117 | static inline int __need_auto_defrag(struct btrfs_fs_info *fs_info) |
8ddc4734 | 118 | { |
0b246afa | 119 | if (!btrfs_test_opt(fs_info, AUTO_DEFRAG)) |
8ddc4734 MX |
120 | return 0; |
121 | ||
0b246afa | 122 | if (btrfs_fs_closing(fs_info)) |
8ddc4734 | 123 | return 0; |
4cb5300b | 124 | |
8ddc4734 | 125 | return 1; |
4cb5300b CM |
126 | } |
127 | ||
128 | /* | |
129 | * insert a defrag record for this inode if auto defrag is | |
130 | * enabled | |
131 | */ | |
132 | int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans, | |
6158e1ce | 133 | struct btrfs_inode *inode) |
4cb5300b | 134 | { |
6158e1ce | 135 | struct btrfs_root *root = inode->root; |
3ffbd68c | 136 | struct btrfs_fs_info *fs_info = root->fs_info; |
4cb5300b | 137 | struct inode_defrag *defrag; |
4cb5300b | 138 | u64 transid; |
8ddc4734 | 139 | int ret; |
4cb5300b | 140 | |
2ff7e61e | 141 | if (!__need_auto_defrag(fs_info)) |
4cb5300b CM |
142 | return 0; |
143 | ||
6158e1ce | 144 | if (test_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags)) |
4cb5300b CM |
145 | return 0; |
146 | ||
147 | if (trans) | |
148 | transid = trans->transid; | |
149 | else | |
6158e1ce | 150 | transid = inode->root->last_trans; |
4cb5300b | 151 | |
9247f317 | 152 | defrag = kmem_cache_zalloc(btrfs_inode_defrag_cachep, GFP_NOFS); |
4cb5300b CM |
153 | if (!defrag) |
154 | return -ENOMEM; | |
155 | ||
6158e1ce | 156 | defrag->ino = btrfs_ino(inode); |
4cb5300b CM |
157 | defrag->transid = transid; |
158 | defrag->root = root->root_key.objectid; | |
159 | ||
0b246afa | 160 | spin_lock(&fs_info->defrag_inodes_lock); |
6158e1ce | 161 | if (!test_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags)) { |
8ddc4734 MX |
162 | /* |
163 | * If we set IN_DEFRAG flag and evict the inode from memory, | |
164 | * and then re-read this inode, this new inode doesn't have | |
165 | * IN_DEFRAG flag. At the case, we may find the existed defrag. | |
166 | */ | |
167 | ret = __btrfs_add_inode_defrag(inode, defrag); | |
168 | if (ret) | |
169 | kmem_cache_free(btrfs_inode_defrag_cachep, defrag); | |
170 | } else { | |
9247f317 | 171 | kmem_cache_free(btrfs_inode_defrag_cachep, defrag); |
8ddc4734 | 172 | } |
0b246afa | 173 | spin_unlock(&fs_info->defrag_inodes_lock); |
a0f98dde | 174 | return 0; |
4cb5300b CM |
175 | } |
176 | ||
177 | /* | |
8ddc4734 MX |
178 | * Requeue the defrag object. If there is a defrag object that points to |
179 | * the same inode in the tree, we will merge them together (by | |
180 | * __btrfs_add_inode_defrag()) and free the one that we want to requeue. | |
4cb5300b | 181 | */ |
46e59791 | 182 | static void btrfs_requeue_inode_defrag(struct btrfs_inode *inode, |
48a3b636 | 183 | struct inode_defrag *defrag) |
8ddc4734 | 184 | { |
3ffbd68c | 185 | struct btrfs_fs_info *fs_info = inode->root->fs_info; |
8ddc4734 MX |
186 | int ret; |
187 | ||
2ff7e61e | 188 | if (!__need_auto_defrag(fs_info)) |
8ddc4734 MX |
189 | goto out; |
190 | ||
191 | /* | |
192 | * Here we don't check the IN_DEFRAG flag, because we need merge | |
193 | * them together. | |
194 | */ | |
0b246afa | 195 | spin_lock(&fs_info->defrag_inodes_lock); |
8ddc4734 | 196 | ret = __btrfs_add_inode_defrag(inode, defrag); |
0b246afa | 197 | spin_unlock(&fs_info->defrag_inodes_lock); |
8ddc4734 MX |
198 | if (ret) |
199 | goto out; | |
200 | return; | |
201 | out: | |
202 | kmem_cache_free(btrfs_inode_defrag_cachep, defrag); | |
203 | } | |
204 | ||
4cb5300b | 205 | /* |
26176e7c MX |
206 | * pick the defragable inode that we want, if it doesn't exist, we will get |
207 | * the next one. | |
4cb5300b | 208 | */ |
26176e7c MX |
209 | static struct inode_defrag * |
210 | btrfs_pick_defrag_inode(struct btrfs_fs_info *fs_info, u64 root, u64 ino) | |
4cb5300b CM |
211 | { |
212 | struct inode_defrag *entry = NULL; | |
762f2263 | 213 | struct inode_defrag tmp; |
4cb5300b CM |
214 | struct rb_node *p; |
215 | struct rb_node *parent = NULL; | |
762f2263 MX |
216 | int ret; |
217 | ||
218 | tmp.ino = ino; | |
219 | tmp.root = root; | |
4cb5300b | 220 | |
26176e7c MX |
221 | spin_lock(&fs_info->defrag_inodes_lock); |
222 | p = fs_info->defrag_inodes.rb_node; | |
4cb5300b CM |
223 | while (p) { |
224 | parent = p; | |
225 | entry = rb_entry(parent, struct inode_defrag, rb_node); | |
226 | ||
762f2263 MX |
227 | ret = __compare_inode_defrag(&tmp, entry); |
228 | if (ret < 0) | |
4cb5300b | 229 | p = parent->rb_left; |
762f2263 | 230 | else if (ret > 0) |
4cb5300b CM |
231 | p = parent->rb_right; |
232 | else | |
26176e7c | 233 | goto out; |
4cb5300b CM |
234 | } |
235 | ||
26176e7c MX |
236 | if (parent && __compare_inode_defrag(&tmp, entry) > 0) { |
237 | parent = rb_next(parent); | |
238 | if (parent) | |
4cb5300b | 239 | entry = rb_entry(parent, struct inode_defrag, rb_node); |
26176e7c MX |
240 | else |
241 | entry = NULL; | |
4cb5300b | 242 | } |
26176e7c MX |
243 | out: |
244 | if (entry) | |
245 | rb_erase(parent, &fs_info->defrag_inodes); | |
246 | spin_unlock(&fs_info->defrag_inodes_lock); | |
247 | return entry; | |
4cb5300b CM |
248 | } |
249 | ||
26176e7c | 250 | void btrfs_cleanup_defrag_inodes(struct btrfs_fs_info *fs_info) |
4cb5300b CM |
251 | { |
252 | struct inode_defrag *defrag; | |
26176e7c MX |
253 | struct rb_node *node; |
254 | ||
255 | spin_lock(&fs_info->defrag_inodes_lock); | |
256 | node = rb_first(&fs_info->defrag_inodes); | |
257 | while (node) { | |
258 | rb_erase(node, &fs_info->defrag_inodes); | |
259 | defrag = rb_entry(node, struct inode_defrag, rb_node); | |
260 | kmem_cache_free(btrfs_inode_defrag_cachep, defrag); | |
261 | ||
351810c1 | 262 | cond_resched_lock(&fs_info->defrag_inodes_lock); |
26176e7c MX |
263 | |
264 | node = rb_first(&fs_info->defrag_inodes); | |
265 | } | |
266 | spin_unlock(&fs_info->defrag_inodes_lock); | |
267 | } | |
268 | ||
269 | #define BTRFS_DEFRAG_BATCH 1024 | |
270 | ||
271 | static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info, | |
272 | struct inode_defrag *defrag) | |
273 | { | |
4cb5300b CM |
274 | struct btrfs_root *inode_root; |
275 | struct inode *inode; | |
4cb5300b CM |
276 | struct btrfs_key key; |
277 | struct btrfs_ioctl_defrag_range_args range; | |
4cb5300b | 278 | int num_defrag; |
6f1c3605 LB |
279 | int index; |
280 | int ret; | |
4cb5300b | 281 | |
26176e7c MX |
282 | /* get the inode */ |
283 | key.objectid = defrag->root; | |
962a298f | 284 | key.type = BTRFS_ROOT_ITEM_KEY; |
26176e7c | 285 | key.offset = (u64)-1; |
6f1c3605 LB |
286 | |
287 | index = srcu_read_lock(&fs_info->subvol_srcu); | |
288 | ||
26176e7c MX |
289 | inode_root = btrfs_read_fs_root_no_name(fs_info, &key); |
290 | if (IS_ERR(inode_root)) { | |
6f1c3605 LB |
291 | ret = PTR_ERR(inode_root); |
292 | goto cleanup; | |
293 | } | |
26176e7c MX |
294 | |
295 | key.objectid = defrag->ino; | |
962a298f | 296 | key.type = BTRFS_INODE_ITEM_KEY; |
26176e7c MX |
297 | key.offset = 0; |
298 | inode = btrfs_iget(fs_info->sb, &key, inode_root, NULL); | |
299 | if (IS_ERR(inode)) { | |
6f1c3605 LB |
300 | ret = PTR_ERR(inode); |
301 | goto cleanup; | |
26176e7c | 302 | } |
6f1c3605 | 303 | srcu_read_unlock(&fs_info->subvol_srcu, index); |
26176e7c MX |
304 | |
305 | /* do a chunk of defrag */ | |
306 | clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags); | |
4cb5300b CM |
307 | memset(&range, 0, sizeof(range)); |
308 | range.len = (u64)-1; | |
26176e7c | 309 | range.start = defrag->last_offset; |
b66f00da MX |
310 | |
311 | sb_start_write(fs_info->sb); | |
26176e7c MX |
312 | num_defrag = btrfs_defrag_file(inode, NULL, &range, defrag->transid, |
313 | BTRFS_DEFRAG_BATCH); | |
b66f00da | 314 | sb_end_write(fs_info->sb); |
26176e7c MX |
315 | /* |
316 | * if we filled the whole defrag batch, there | |
317 | * must be more work to do. Queue this defrag | |
318 | * again | |
319 | */ | |
320 | if (num_defrag == BTRFS_DEFRAG_BATCH) { | |
321 | defrag->last_offset = range.start; | |
46e59791 | 322 | btrfs_requeue_inode_defrag(BTRFS_I(inode), defrag); |
26176e7c MX |
323 | } else if (defrag->last_offset && !defrag->cycled) { |
324 | /* | |
325 | * we didn't fill our defrag batch, but | |
326 | * we didn't start at zero. Make sure we loop | |
327 | * around to the start of the file. | |
328 | */ | |
329 | defrag->last_offset = 0; | |
330 | defrag->cycled = 1; | |
46e59791 | 331 | btrfs_requeue_inode_defrag(BTRFS_I(inode), defrag); |
26176e7c MX |
332 | } else { |
333 | kmem_cache_free(btrfs_inode_defrag_cachep, defrag); | |
334 | } | |
335 | ||
336 | iput(inode); | |
337 | return 0; | |
6f1c3605 LB |
338 | cleanup: |
339 | srcu_read_unlock(&fs_info->subvol_srcu, index); | |
340 | kmem_cache_free(btrfs_inode_defrag_cachep, defrag); | |
341 | return ret; | |
26176e7c MX |
342 | } |
343 | ||
344 | /* | |
345 | * run through the list of inodes in the FS that need | |
346 | * defragging | |
347 | */ | |
348 | int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info) | |
349 | { | |
350 | struct inode_defrag *defrag; | |
351 | u64 first_ino = 0; | |
352 | u64 root_objectid = 0; | |
4cb5300b CM |
353 | |
354 | atomic_inc(&fs_info->defrag_running); | |
67871254 | 355 | while (1) { |
dc81cdc5 MX |
356 | /* Pause the auto defragger. */ |
357 | if (test_bit(BTRFS_FS_STATE_REMOUNTING, | |
358 | &fs_info->fs_state)) | |
359 | break; | |
360 | ||
2ff7e61e | 361 | if (!__need_auto_defrag(fs_info)) |
26176e7c | 362 | break; |
4cb5300b CM |
363 | |
364 | /* find an inode to defrag */ | |
26176e7c MX |
365 | defrag = btrfs_pick_defrag_inode(fs_info, root_objectid, |
366 | first_ino); | |
4cb5300b | 367 | if (!defrag) { |
26176e7c | 368 | if (root_objectid || first_ino) { |
762f2263 | 369 | root_objectid = 0; |
4cb5300b CM |
370 | first_ino = 0; |
371 | continue; | |
372 | } else { | |
373 | break; | |
374 | } | |
375 | } | |
376 | ||
4cb5300b | 377 | first_ino = defrag->ino + 1; |
762f2263 | 378 | root_objectid = defrag->root; |
4cb5300b | 379 | |
26176e7c | 380 | __btrfs_run_defrag_inode(fs_info, defrag); |
4cb5300b | 381 | } |
4cb5300b CM |
382 | atomic_dec(&fs_info->defrag_running); |
383 | ||
384 | /* | |
385 | * during unmount, we use the transaction_wait queue to | |
386 | * wait for the defragger to stop | |
387 | */ | |
388 | wake_up(&fs_info->transaction_wait); | |
389 | return 0; | |
390 | } | |
39279cc3 | 391 | |
d352ac68 CM |
392 | /* simple helper to fault in pages and copy. This should go away |
393 | * and be replaced with calls into generic code. | |
394 | */ | |
ee22f0c4 | 395 | static noinline int btrfs_copy_from_user(loff_t pos, size_t write_bytes, |
a1b32a59 | 396 | struct page **prepared_pages, |
11c65dcc | 397 | struct iov_iter *i) |
39279cc3 | 398 | { |
914ee295 | 399 | size_t copied = 0; |
d0215f3e | 400 | size_t total_copied = 0; |
11c65dcc | 401 | int pg = 0; |
09cbfeaf | 402 | int offset = pos & (PAGE_SIZE - 1); |
39279cc3 | 403 | |
11c65dcc | 404 | while (write_bytes > 0) { |
39279cc3 | 405 | size_t count = min_t(size_t, |
09cbfeaf | 406 | PAGE_SIZE - offset, write_bytes); |
11c65dcc | 407 | struct page *page = prepared_pages[pg]; |
914ee295 XZ |
408 | /* |
409 | * Copy data from userspace to the current page | |
914ee295 | 410 | */ |
914ee295 | 411 | copied = iov_iter_copy_from_user_atomic(page, i, offset, count); |
11c65dcc | 412 | |
39279cc3 CM |
413 | /* Flush processor's dcache for this page */ |
414 | flush_dcache_page(page); | |
31339acd CM |
415 | |
416 | /* | |
417 | * if we get a partial write, we can end up with | |
418 | * partially up to date pages. These add | |
419 | * a lot of complexity, so make sure they don't | |
420 | * happen by forcing this copy to be retried. | |
421 | * | |
422 | * The rest of the btrfs_file_write code will fall | |
423 | * back to page at a time copies after we return 0. | |
424 | */ | |
425 | if (!PageUptodate(page) && copied < count) | |
426 | copied = 0; | |
427 | ||
11c65dcc JB |
428 | iov_iter_advance(i, copied); |
429 | write_bytes -= copied; | |
914ee295 | 430 | total_copied += copied; |
39279cc3 | 431 | |
b30ac0fc | 432 | /* Return to btrfs_file_write_iter to fault page */ |
9f570b8d | 433 | if (unlikely(copied == 0)) |
914ee295 | 434 | break; |
11c65dcc | 435 | |
09cbfeaf | 436 | if (copied < PAGE_SIZE - offset) { |
11c65dcc JB |
437 | offset += copied; |
438 | } else { | |
439 | pg++; | |
440 | offset = 0; | |
441 | } | |
39279cc3 | 442 | } |
914ee295 | 443 | return total_copied; |
39279cc3 CM |
444 | } |
445 | ||
d352ac68 CM |
446 | /* |
447 | * unlocks pages after btrfs_file_write is done with them | |
448 | */ | |
48a3b636 | 449 | static void btrfs_drop_pages(struct page **pages, size_t num_pages) |
39279cc3 CM |
450 | { |
451 | size_t i; | |
452 | for (i = 0; i < num_pages; i++) { | |
d352ac68 CM |
453 | /* page checked is some magic around finding pages that |
454 | * have been modified without going through btrfs_set_page_dirty | |
2457aec6 MG |
455 | * clear it here. There should be no need to mark the pages |
456 | * accessed as prepare_pages should have marked them accessed | |
457 | * in prepare_pages via find_or_create_page() | |
d352ac68 | 458 | */ |
4a096752 | 459 | ClearPageChecked(pages[i]); |
39279cc3 | 460 | unlock_page(pages[i]); |
09cbfeaf | 461 | put_page(pages[i]); |
39279cc3 CM |
462 | } |
463 | } | |
464 | ||
f48bf66b FM |
465 | static int btrfs_find_new_delalloc_bytes(struct btrfs_inode *inode, |
466 | const u64 start, | |
467 | const u64 len, | |
468 | struct extent_state **cached_state) | |
469 | { | |
470 | u64 search_start = start; | |
471 | const u64 end = start + len - 1; | |
472 | ||
473 | while (search_start < end) { | |
474 | const u64 search_len = end - search_start + 1; | |
475 | struct extent_map *em; | |
476 | u64 em_len; | |
477 | int ret = 0; | |
478 | ||
479 | em = btrfs_get_extent(inode, NULL, 0, search_start, | |
480 | search_len, 0); | |
481 | if (IS_ERR(em)) | |
482 | return PTR_ERR(em); | |
483 | ||
484 | if (em->block_start != EXTENT_MAP_HOLE) | |
485 | goto next; | |
486 | ||
487 | em_len = em->len; | |
488 | if (em->start < search_start) | |
489 | em_len -= search_start - em->start; | |
490 | if (em_len > search_len) | |
491 | em_len = search_len; | |
492 | ||
493 | ret = set_extent_bit(&inode->io_tree, search_start, | |
494 | search_start + em_len - 1, | |
495 | EXTENT_DELALLOC_NEW, | |
496 | NULL, cached_state, GFP_NOFS); | |
497 | next: | |
498 | search_start = extent_map_end(em); | |
499 | free_extent_map(em); | |
500 | if (ret) | |
501 | return ret; | |
502 | } | |
503 | return 0; | |
504 | } | |
505 | ||
d352ac68 CM |
506 | /* |
507 | * after copy_from_user, pages need to be dirtied and we need to make | |
508 | * sure holes are created between the current EOF and the start of | |
509 | * any next extents (if required). | |
510 | * | |
511 | * this also makes the decision about creating an inline extent vs | |
512 | * doing real data extents, marking pages dirty and delalloc as required. | |
513 | */ | |
2ff7e61e JM |
514 | int btrfs_dirty_pages(struct inode *inode, struct page **pages, |
515 | size_t num_pages, loff_t pos, size_t write_bytes, | |
516 | struct extent_state **cached) | |
39279cc3 | 517 | { |
0b246afa | 518 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
39279cc3 | 519 | int err = 0; |
a52d9a80 | 520 | int i; |
db94535d | 521 | u64 num_bytes; |
a52d9a80 CM |
522 | u64 start_pos; |
523 | u64 end_of_last_block; | |
524 | u64 end_pos = pos + write_bytes; | |
525 | loff_t isize = i_size_read(inode); | |
e3b8a485 | 526 | unsigned int extra_bits = 0; |
39279cc3 | 527 | |
0b246afa | 528 | start_pos = pos & ~((u64) fs_info->sectorsize - 1); |
da17066c | 529 | num_bytes = round_up(write_bytes + pos - start_pos, |
0b246afa | 530 | fs_info->sectorsize); |
39279cc3 | 531 | |
db94535d | 532 | end_of_last_block = start_pos + num_bytes - 1; |
e3b8a485 FM |
533 | |
534 | if (!btrfs_is_free_space_inode(BTRFS_I(inode))) { | |
535 | if (start_pos >= isize && | |
536 | !(BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC)) { | |
537 | /* | |
538 | * There can't be any extents following eof in this case | |
539 | * so just set the delalloc new bit for the range | |
540 | * directly. | |
541 | */ | |
542 | extra_bits |= EXTENT_DELALLOC_NEW; | |
543 | } else { | |
544 | err = btrfs_find_new_delalloc_bytes(BTRFS_I(inode), | |
545 | start_pos, | |
546 | num_bytes, cached); | |
547 | if (err) | |
548 | return err; | |
549 | } | |
550 | } | |
551 | ||
2ac55d41 | 552 | err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block, |
e3b8a485 | 553 | extra_bits, cached, 0); |
d0215f3e JB |
554 | if (err) |
555 | return err; | |
9ed74f2d | 556 | |
c8b97818 CM |
557 | for (i = 0; i < num_pages; i++) { |
558 | struct page *p = pages[i]; | |
559 | SetPageUptodate(p); | |
560 | ClearPageChecked(p); | |
561 | set_page_dirty(p); | |
a52d9a80 | 562 | } |
9f570b8d JB |
563 | |
564 | /* | |
565 | * we've only changed i_size in ram, and we haven't updated | |
566 | * the disk i_size. There is no need to log the inode | |
567 | * at this time. | |
568 | */ | |
569 | if (end_pos > isize) | |
a52d9a80 | 570 | i_size_write(inode, end_pos); |
a22285a6 | 571 | return 0; |
39279cc3 CM |
572 | } |
573 | ||
d352ac68 CM |
574 | /* |
575 | * this drops all the extents in the cache that intersect the range | |
576 | * [start, end]. Existing extents are split as required. | |
577 | */ | |
dcdbc059 | 578 | void btrfs_drop_extent_cache(struct btrfs_inode *inode, u64 start, u64 end, |
7014cdb4 | 579 | int skip_pinned) |
a52d9a80 CM |
580 | { |
581 | struct extent_map *em; | |
3b951516 CM |
582 | struct extent_map *split = NULL; |
583 | struct extent_map *split2 = NULL; | |
dcdbc059 | 584 | struct extent_map_tree *em_tree = &inode->extent_tree; |
39b5637f | 585 | u64 len = end - start + 1; |
5dc562c5 | 586 | u64 gen; |
3b951516 CM |
587 | int ret; |
588 | int testend = 1; | |
5b21f2ed | 589 | unsigned long flags; |
c8b97818 | 590 | int compressed = 0; |
09a2a8f9 | 591 | bool modified; |
a52d9a80 | 592 | |
e6dcd2dc | 593 | WARN_ON(end < start); |
3b951516 | 594 | if (end == (u64)-1) { |
39b5637f | 595 | len = (u64)-1; |
3b951516 CM |
596 | testend = 0; |
597 | } | |
d397712b | 598 | while (1) { |
7014cdb4 JB |
599 | int no_splits = 0; |
600 | ||
09a2a8f9 | 601 | modified = false; |
3b951516 | 602 | if (!split) |
172ddd60 | 603 | split = alloc_extent_map(); |
3b951516 | 604 | if (!split2) |
172ddd60 | 605 | split2 = alloc_extent_map(); |
7014cdb4 JB |
606 | if (!split || !split2) |
607 | no_splits = 1; | |
3b951516 | 608 | |
890871be | 609 | write_lock(&em_tree->lock); |
39b5637f | 610 | em = lookup_extent_mapping(em_tree, start, len); |
d1310b2e | 611 | if (!em) { |
890871be | 612 | write_unlock(&em_tree->lock); |
a52d9a80 | 613 | break; |
d1310b2e | 614 | } |
5b21f2ed | 615 | flags = em->flags; |
5dc562c5 | 616 | gen = em->generation; |
5b21f2ed | 617 | if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) { |
55ef6899 | 618 | if (testend && em->start + em->len >= start + len) { |
5b21f2ed | 619 | free_extent_map(em); |
a1ed835e | 620 | write_unlock(&em_tree->lock); |
5b21f2ed ZY |
621 | break; |
622 | } | |
55ef6899 YZ |
623 | start = em->start + em->len; |
624 | if (testend) | |
5b21f2ed | 625 | len = start + len - (em->start + em->len); |
5b21f2ed | 626 | free_extent_map(em); |
a1ed835e | 627 | write_unlock(&em_tree->lock); |
5b21f2ed ZY |
628 | continue; |
629 | } | |
c8b97818 | 630 | compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags); |
3ce7e67a | 631 | clear_bit(EXTENT_FLAG_PINNED, &em->flags); |
3b277594 | 632 | clear_bit(EXTENT_FLAG_LOGGING, &flags); |
09a2a8f9 | 633 | modified = !list_empty(&em->list); |
7014cdb4 JB |
634 | if (no_splits) |
635 | goto next; | |
3b951516 | 636 | |
ee20a983 | 637 | if (em->start < start) { |
3b951516 CM |
638 | split->start = em->start; |
639 | split->len = start - em->start; | |
ee20a983 JB |
640 | |
641 | if (em->block_start < EXTENT_MAP_LAST_BYTE) { | |
642 | split->orig_start = em->orig_start; | |
643 | split->block_start = em->block_start; | |
644 | ||
645 | if (compressed) | |
646 | split->block_len = em->block_len; | |
647 | else | |
648 | split->block_len = split->len; | |
649 | split->orig_block_len = max(split->block_len, | |
650 | em->orig_block_len); | |
651 | split->ram_bytes = em->ram_bytes; | |
652 | } else { | |
653 | split->orig_start = split->start; | |
654 | split->block_len = 0; | |
655 | split->block_start = em->block_start; | |
656 | split->orig_block_len = 0; | |
657 | split->ram_bytes = split->len; | |
658 | } | |
659 | ||
5dc562c5 | 660 | split->generation = gen; |
3b951516 | 661 | split->bdev = em->bdev; |
5b21f2ed | 662 | split->flags = flags; |
261507a0 | 663 | split->compress_type = em->compress_type; |
176840b3 | 664 | replace_extent_mapping(em_tree, em, split, modified); |
3b951516 CM |
665 | free_extent_map(split); |
666 | split = split2; | |
667 | split2 = NULL; | |
668 | } | |
ee20a983 | 669 | if (testend && em->start + em->len > start + len) { |
3b951516 CM |
670 | u64 diff = start + len - em->start; |
671 | ||
672 | split->start = start + len; | |
673 | split->len = em->start + em->len - (start + len); | |
674 | split->bdev = em->bdev; | |
5b21f2ed | 675 | split->flags = flags; |
261507a0 | 676 | split->compress_type = em->compress_type; |
5dc562c5 | 677 | split->generation = gen; |
ee20a983 JB |
678 | |
679 | if (em->block_start < EXTENT_MAP_LAST_BYTE) { | |
680 | split->orig_block_len = max(em->block_len, | |
b4939680 | 681 | em->orig_block_len); |
3b951516 | 682 | |
ee20a983 JB |
683 | split->ram_bytes = em->ram_bytes; |
684 | if (compressed) { | |
685 | split->block_len = em->block_len; | |
686 | split->block_start = em->block_start; | |
687 | split->orig_start = em->orig_start; | |
688 | } else { | |
689 | split->block_len = split->len; | |
690 | split->block_start = em->block_start | |
691 | + diff; | |
692 | split->orig_start = em->orig_start; | |
693 | } | |
c8b97818 | 694 | } else { |
ee20a983 JB |
695 | split->ram_bytes = split->len; |
696 | split->orig_start = split->start; | |
697 | split->block_len = 0; | |
698 | split->block_start = em->block_start; | |
699 | split->orig_block_len = 0; | |
c8b97818 | 700 | } |
3b951516 | 701 | |
176840b3 FM |
702 | if (extent_map_in_tree(em)) { |
703 | replace_extent_mapping(em_tree, em, split, | |
704 | modified); | |
705 | } else { | |
706 | ret = add_extent_mapping(em_tree, split, | |
707 | modified); | |
708 | ASSERT(ret == 0); /* Logic error */ | |
709 | } | |
3b951516 CM |
710 | free_extent_map(split); |
711 | split = NULL; | |
712 | } | |
7014cdb4 | 713 | next: |
176840b3 FM |
714 | if (extent_map_in_tree(em)) |
715 | remove_extent_mapping(em_tree, em); | |
890871be | 716 | write_unlock(&em_tree->lock); |
d1310b2e | 717 | |
a52d9a80 CM |
718 | /* once for us */ |
719 | free_extent_map(em); | |
720 | /* once for the tree*/ | |
721 | free_extent_map(em); | |
722 | } | |
3b951516 CM |
723 | if (split) |
724 | free_extent_map(split); | |
725 | if (split2) | |
726 | free_extent_map(split2); | |
a52d9a80 CM |
727 | } |
728 | ||
39279cc3 CM |
729 | /* |
730 | * this is very complex, but the basic idea is to drop all extents | |
731 | * in the range start - end. hint_block is filled in with a block number | |
732 | * that would be a good hint to the block allocator for this file. | |
733 | * | |
734 | * If an extent intersects the range but is not entirely inside the range | |
735 | * it is either truncated or split. Anything entirely inside the range | |
736 | * is deleted from the tree. | |
737 | */ | |
5dc562c5 JB |
738 | int __btrfs_drop_extents(struct btrfs_trans_handle *trans, |
739 | struct btrfs_root *root, struct inode *inode, | |
740 | struct btrfs_path *path, u64 start, u64 end, | |
1acae57b FDBM |
741 | u64 *drop_end, int drop_cache, |
742 | int replace_extent, | |
743 | u32 extent_item_size, | |
744 | int *key_inserted) | |
39279cc3 | 745 | { |
0b246afa | 746 | struct btrfs_fs_info *fs_info = root->fs_info; |
5f39d397 | 747 | struct extent_buffer *leaf; |
920bbbfb | 748 | struct btrfs_file_extent_item *fi; |
00f5c795 | 749 | struct btrfs_key key; |
920bbbfb | 750 | struct btrfs_key new_key; |
4a0cc7ca | 751 | u64 ino = btrfs_ino(BTRFS_I(inode)); |
920bbbfb YZ |
752 | u64 search_start = start; |
753 | u64 disk_bytenr = 0; | |
754 | u64 num_bytes = 0; | |
755 | u64 extent_offset = 0; | |
756 | u64 extent_end = 0; | |
62fe51c1 | 757 | u64 last_end = start; |
920bbbfb YZ |
758 | int del_nr = 0; |
759 | int del_slot = 0; | |
760 | int extent_type; | |
ccd467d6 | 761 | int recow; |
00f5c795 | 762 | int ret; |
dc7fdde3 | 763 | int modify_tree = -1; |
27cdeb70 | 764 | int update_refs; |
c3308f84 | 765 | int found = 0; |
1acae57b | 766 | int leafs_visited = 0; |
39279cc3 | 767 | |
a1ed835e | 768 | if (drop_cache) |
dcdbc059 | 769 | btrfs_drop_extent_cache(BTRFS_I(inode), start, end - 1, 0); |
a52d9a80 | 770 | |
d5f37527 | 771 | if (start >= BTRFS_I(inode)->disk_i_size && !replace_extent) |
dc7fdde3 CM |
772 | modify_tree = 0; |
773 | ||
27cdeb70 | 774 | update_refs = (test_bit(BTRFS_ROOT_REF_COWS, &root->state) || |
0b246afa | 775 | root == fs_info->tree_root); |
d397712b | 776 | while (1) { |
ccd467d6 | 777 | recow = 0; |
33345d01 | 778 | ret = btrfs_lookup_file_extent(trans, root, path, ino, |
dc7fdde3 | 779 | search_start, modify_tree); |
39279cc3 | 780 | if (ret < 0) |
920bbbfb YZ |
781 | break; |
782 | if (ret > 0 && path->slots[0] > 0 && search_start == start) { | |
783 | leaf = path->nodes[0]; | |
784 | btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1); | |
33345d01 | 785 | if (key.objectid == ino && |
920bbbfb YZ |
786 | key.type == BTRFS_EXTENT_DATA_KEY) |
787 | path->slots[0]--; | |
39279cc3 | 788 | } |
920bbbfb | 789 | ret = 0; |
1acae57b | 790 | leafs_visited++; |
8c2383c3 | 791 | next_slot: |
5f39d397 | 792 | leaf = path->nodes[0]; |
920bbbfb YZ |
793 | if (path->slots[0] >= btrfs_header_nritems(leaf)) { |
794 | BUG_ON(del_nr > 0); | |
795 | ret = btrfs_next_leaf(root, path); | |
796 | if (ret < 0) | |
797 | break; | |
798 | if (ret > 0) { | |
799 | ret = 0; | |
800 | break; | |
8c2383c3 | 801 | } |
1acae57b | 802 | leafs_visited++; |
920bbbfb YZ |
803 | leaf = path->nodes[0]; |
804 | recow = 1; | |
805 | } | |
806 | ||
807 | btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); | |
aeafbf84 FM |
808 | |
809 | if (key.objectid > ino) | |
810 | break; | |
811 | if (WARN_ON_ONCE(key.objectid < ino) || | |
812 | key.type < BTRFS_EXTENT_DATA_KEY) { | |
813 | ASSERT(del_nr == 0); | |
814 | path->slots[0]++; | |
815 | goto next_slot; | |
816 | } | |
817 | if (key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end) | |
920bbbfb YZ |
818 | break; |
819 | ||
820 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
821 | struct btrfs_file_extent_item); | |
822 | extent_type = btrfs_file_extent_type(leaf, fi); | |
823 | ||
824 | if (extent_type == BTRFS_FILE_EXTENT_REG || | |
825 | extent_type == BTRFS_FILE_EXTENT_PREALLOC) { | |
826 | disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); | |
827 | num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); | |
828 | extent_offset = btrfs_file_extent_offset(leaf, fi); | |
829 | extent_end = key.offset + | |
830 | btrfs_file_extent_num_bytes(leaf, fi); | |
831 | } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { | |
832 | extent_end = key.offset + | |
e41ca589 | 833 | btrfs_file_extent_ram_bytes(leaf, fi); |
8c2383c3 | 834 | } else { |
aeafbf84 FM |
835 | /* can't happen */ |
836 | BUG(); | |
39279cc3 CM |
837 | } |
838 | ||
fc19c5e7 FM |
839 | /* |
840 | * Don't skip extent items representing 0 byte lengths. They | |
841 | * used to be created (bug) if while punching holes we hit | |
842 | * -ENOSPC condition. So if we find one here, just ensure we | |
843 | * delete it, otherwise we would insert a new file extent item | |
844 | * with the same key (offset) as that 0 bytes length file | |
845 | * extent item in the call to setup_items_for_insert() later | |
846 | * in this function. | |
847 | */ | |
62fe51c1 JB |
848 | if (extent_end == key.offset && extent_end >= search_start) { |
849 | last_end = extent_end; | |
fc19c5e7 | 850 | goto delete_extent_item; |
62fe51c1 | 851 | } |
fc19c5e7 | 852 | |
920bbbfb YZ |
853 | if (extent_end <= search_start) { |
854 | path->slots[0]++; | |
8c2383c3 | 855 | goto next_slot; |
39279cc3 CM |
856 | } |
857 | ||
c3308f84 | 858 | found = 1; |
920bbbfb | 859 | search_start = max(key.offset, start); |
dc7fdde3 CM |
860 | if (recow || !modify_tree) { |
861 | modify_tree = -1; | |
b3b4aa74 | 862 | btrfs_release_path(path); |
920bbbfb | 863 | continue; |
39279cc3 | 864 | } |
6643558d | 865 | |
920bbbfb YZ |
866 | /* |
867 | * | - range to drop - | | |
868 | * | -------- extent -------- | | |
869 | */ | |
870 | if (start > key.offset && end < extent_end) { | |
871 | BUG_ON(del_nr > 0); | |
00fdf13a | 872 | if (extent_type == BTRFS_FILE_EXTENT_INLINE) { |
3f9e3df8 | 873 | ret = -EOPNOTSUPP; |
00fdf13a LB |
874 | break; |
875 | } | |
920bbbfb YZ |
876 | |
877 | memcpy(&new_key, &key, sizeof(new_key)); | |
878 | new_key.offset = start; | |
879 | ret = btrfs_duplicate_item(trans, root, path, | |
880 | &new_key); | |
881 | if (ret == -EAGAIN) { | |
b3b4aa74 | 882 | btrfs_release_path(path); |
920bbbfb | 883 | continue; |
6643558d | 884 | } |
920bbbfb YZ |
885 | if (ret < 0) |
886 | break; | |
887 | ||
888 | leaf = path->nodes[0]; | |
889 | fi = btrfs_item_ptr(leaf, path->slots[0] - 1, | |
890 | struct btrfs_file_extent_item); | |
891 | btrfs_set_file_extent_num_bytes(leaf, fi, | |
892 | start - key.offset); | |
893 | ||
894 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
895 | struct btrfs_file_extent_item); | |
896 | ||
897 | extent_offset += start - key.offset; | |
898 | btrfs_set_file_extent_offset(leaf, fi, extent_offset); | |
899 | btrfs_set_file_extent_num_bytes(leaf, fi, | |
900 | extent_end - start); | |
901 | btrfs_mark_buffer_dirty(leaf); | |
902 | ||
5dc562c5 | 903 | if (update_refs && disk_bytenr > 0) { |
84f7d8e6 | 904 | ret = btrfs_inc_extent_ref(trans, root, |
920bbbfb YZ |
905 | disk_bytenr, num_bytes, 0, |
906 | root->root_key.objectid, | |
907 | new_key.objectid, | |
b06c4bf5 | 908 | start - extent_offset); |
79787eaa | 909 | BUG_ON(ret); /* -ENOMEM */ |
771ed689 | 910 | } |
920bbbfb | 911 | key.offset = start; |
6643558d | 912 | } |
62fe51c1 JB |
913 | /* |
914 | * From here on out we will have actually dropped something, so | |
915 | * last_end can be updated. | |
916 | */ | |
917 | last_end = extent_end; | |
918 | ||
920bbbfb YZ |
919 | /* |
920 | * | ---- range to drop ----- | | |
921 | * | -------- extent -------- | | |
922 | */ | |
923 | if (start <= key.offset && end < extent_end) { | |
00fdf13a | 924 | if (extent_type == BTRFS_FILE_EXTENT_INLINE) { |
3f9e3df8 | 925 | ret = -EOPNOTSUPP; |
00fdf13a LB |
926 | break; |
927 | } | |
6643558d | 928 | |
920bbbfb YZ |
929 | memcpy(&new_key, &key, sizeof(new_key)); |
930 | new_key.offset = end; | |
0b246afa | 931 | btrfs_set_item_key_safe(fs_info, path, &new_key); |
6643558d | 932 | |
920bbbfb YZ |
933 | extent_offset += end - key.offset; |
934 | btrfs_set_file_extent_offset(leaf, fi, extent_offset); | |
935 | btrfs_set_file_extent_num_bytes(leaf, fi, | |
936 | extent_end - end); | |
937 | btrfs_mark_buffer_dirty(leaf); | |
2671485d | 938 | if (update_refs && disk_bytenr > 0) |
920bbbfb | 939 | inode_sub_bytes(inode, end - key.offset); |
920bbbfb | 940 | break; |
39279cc3 | 941 | } |
771ed689 | 942 | |
920bbbfb YZ |
943 | search_start = extent_end; |
944 | /* | |
945 | * | ---- range to drop ----- | | |
946 | * | -------- extent -------- | | |
947 | */ | |
948 | if (start > key.offset && end >= extent_end) { | |
949 | BUG_ON(del_nr > 0); | |
00fdf13a | 950 | if (extent_type == BTRFS_FILE_EXTENT_INLINE) { |
3f9e3df8 | 951 | ret = -EOPNOTSUPP; |
00fdf13a LB |
952 | break; |
953 | } | |
8c2383c3 | 954 | |
920bbbfb YZ |
955 | btrfs_set_file_extent_num_bytes(leaf, fi, |
956 | start - key.offset); | |
957 | btrfs_mark_buffer_dirty(leaf); | |
2671485d | 958 | if (update_refs && disk_bytenr > 0) |
920bbbfb | 959 | inode_sub_bytes(inode, extent_end - start); |
920bbbfb YZ |
960 | if (end == extent_end) |
961 | break; | |
c8b97818 | 962 | |
920bbbfb YZ |
963 | path->slots[0]++; |
964 | goto next_slot; | |
31840ae1 ZY |
965 | } |
966 | ||
920bbbfb YZ |
967 | /* |
968 | * | ---- range to drop ----- | | |
969 | * | ------ extent ------ | | |
970 | */ | |
971 | if (start <= key.offset && end >= extent_end) { | |
fc19c5e7 | 972 | delete_extent_item: |
920bbbfb YZ |
973 | if (del_nr == 0) { |
974 | del_slot = path->slots[0]; | |
975 | del_nr = 1; | |
976 | } else { | |
977 | BUG_ON(del_slot + del_nr != path->slots[0]); | |
978 | del_nr++; | |
979 | } | |
31840ae1 | 980 | |
5dc562c5 JB |
981 | if (update_refs && |
982 | extent_type == BTRFS_FILE_EXTENT_INLINE) { | |
a76a3cd4 | 983 | inode_sub_bytes(inode, |
920bbbfb YZ |
984 | extent_end - key.offset); |
985 | extent_end = ALIGN(extent_end, | |
0b246afa | 986 | fs_info->sectorsize); |
5dc562c5 | 987 | } else if (update_refs && disk_bytenr > 0) { |
84f7d8e6 | 988 | ret = btrfs_free_extent(trans, root, |
920bbbfb YZ |
989 | disk_bytenr, num_bytes, 0, |
990 | root->root_key.objectid, | |
5d4f98a2 | 991 | key.objectid, key.offset - |
b06c4bf5 | 992 | extent_offset); |
79787eaa | 993 | BUG_ON(ret); /* -ENOMEM */ |
920bbbfb YZ |
994 | inode_sub_bytes(inode, |
995 | extent_end - key.offset); | |
31840ae1 | 996 | } |
31840ae1 | 997 | |
920bbbfb YZ |
998 | if (end == extent_end) |
999 | break; | |
1000 | ||
1001 | if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) { | |
1002 | path->slots[0]++; | |
1003 | goto next_slot; | |
1004 | } | |
1005 | ||
1006 | ret = btrfs_del_items(trans, root, path, del_slot, | |
1007 | del_nr); | |
79787eaa | 1008 | if (ret) { |
66642832 | 1009 | btrfs_abort_transaction(trans, ret); |
5dc562c5 | 1010 | break; |
79787eaa | 1011 | } |
920bbbfb YZ |
1012 | |
1013 | del_nr = 0; | |
1014 | del_slot = 0; | |
1015 | ||
b3b4aa74 | 1016 | btrfs_release_path(path); |
920bbbfb | 1017 | continue; |
39279cc3 | 1018 | } |
920bbbfb YZ |
1019 | |
1020 | BUG_ON(1); | |
39279cc3 | 1021 | } |
920bbbfb | 1022 | |
79787eaa | 1023 | if (!ret && del_nr > 0) { |
1acae57b FDBM |
1024 | /* |
1025 | * Set path->slots[0] to first slot, so that after the delete | |
1026 | * if items are move off from our leaf to its immediate left or | |
1027 | * right neighbor leafs, we end up with a correct and adjusted | |
d5f37527 | 1028 | * path->slots[0] for our insertion (if replace_extent != 0). |
1acae57b FDBM |
1029 | */ |
1030 | path->slots[0] = del_slot; | |
920bbbfb | 1031 | ret = btrfs_del_items(trans, root, path, del_slot, del_nr); |
79787eaa | 1032 | if (ret) |
66642832 | 1033 | btrfs_abort_transaction(trans, ret); |
d5f37527 | 1034 | } |
1acae57b | 1035 | |
d5f37527 FDBM |
1036 | leaf = path->nodes[0]; |
1037 | /* | |
1038 | * If btrfs_del_items() was called, it might have deleted a leaf, in | |
1039 | * which case it unlocked our path, so check path->locks[0] matches a | |
1040 | * write lock. | |
1041 | */ | |
1042 | if (!ret && replace_extent && leafs_visited == 1 && | |
1043 | (path->locks[0] == BTRFS_WRITE_LOCK_BLOCKING || | |
1044 | path->locks[0] == BTRFS_WRITE_LOCK) && | |
2ff7e61e | 1045 | btrfs_leaf_free_space(fs_info, leaf) >= |
d5f37527 FDBM |
1046 | sizeof(struct btrfs_item) + extent_item_size) { |
1047 | ||
1048 | key.objectid = ino; | |
1049 | key.type = BTRFS_EXTENT_DATA_KEY; | |
1050 | key.offset = start; | |
1051 | if (!del_nr && path->slots[0] < btrfs_header_nritems(leaf)) { | |
1052 | struct btrfs_key slot_key; | |
1053 | ||
1054 | btrfs_item_key_to_cpu(leaf, &slot_key, path->slots[0]); | |
1055 | if (btrfs_comp_cpu_keys(&key, &slot_key) > 0) | |
1056 | path->slots[0]++; | |
1acae57b | 1057 | } |
d5f37527 FDBM |
1058 | setup_items_for_insert(root, path, &key, |
1059 | &extent_item_size, | |
1060 | extent_item_size, | |
1061 | sizeof(struct btrfs_item) + | |
1062 | extent_item_size, 1); | |
1063 | *key_inserted = 1; | |
6643558d | 1064 | } |
920bbbfb | 1065 | |
1acae57b FDBM |
1066 | if (!replace_extent || !(*key_inserted)) |
1067 | btrfs_release_path(path); | |
2aaa6655 | 1068 | if (drop_end) |
62fe51c1 | 1069 | *drop_end = found ? min(end, last_end) : end; |
5dc562c5 JB |
1070 | return ret; |
1071 | } | |
1072 | ||
1073 | int btrfs_drop_extents(struct btrfs_trans_handle *trans, | |
1074 | struct btrfs_root *root, struct inode *inode, u64 start, | |
2671485d | 1075 | u64 end, int drop_cache) |
5dc562c5 JB |
1076 | { |
1077 | struct btrfs_path *path; | |
1078 | int ret; | |
1079 | ||
1080 | path = btrfs_alloc_path(); | |
1081 | if (!path) | |
1082 | return -ENOMEM; | |
2aaa6655 | 1083 | ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL, |
1acae57b | 1084 | drop_cache, 0, 0, NULL); |
920bbbfb | 1085 | btrfs_free_path(path); |
39279cc3 CM |
1086 | return ret; |
1087 | } | |
1088 | ||
d899e052 | 1089 | static int extent_mergeable(struct extent_buffer *leaf, int slot, |
6c7d54ac YZ |
1090 | u64 objectid, u64 bytenr, u64 orig_offset, |
1091 | u64 *start, u64 *end) | |
d899e052 YZ |
1092 | { |
1093 | struct btrfs_file_extent_item *fi; | |
1094 | struct btrfs_key key; | |
1095 | u64 extent_end; | |
1096 | ||
1097 | if (slot < 0 || slot >= btrfs_header_nritems(leaf)) | |
1098 | return 0; | |
1099 | ||
1100 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
1101 | if (key.objectid != objectid || key.type != BTRFS_EXTENT_DATA_KEY) | |
1102 | return 0; | |
1103 | ||
1104 | fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); | |
1105 | if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG || | |
1106 | btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr || | |
6c7d54ac | 1107 | btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset || |
d899e052 YZ |
1108 | btrfs_file_extent_compression(leaf, fi) || |
1109 | btrfs_file_extent_encryption(leaf, fi) || | |
1110 | btrfs_file_extent_other_encoding(leaf, fi)) | |
1111 | return 0; | |
1112 | ||
1113 | extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); | |
1114 | if ((*start && *start != key.offset) || (*end && *end != extent_end)) | |
1115 | return 0; | |
1116 | ||
1117 | *start = key.offset; | |
1118 | *end = extent_end; | |
1119 | return 1; | |
1120 | } | |
1121 | ||
1122 | /* | |
1123 | * Mark extent in the range start - end as written. | |
1124 | * | |
1125 | * This changes extent type from 'pre-allocated' to 'regular'. If only | |
1126 | * part of extent is marked as written, the extent will be split into | |
1127 | * two or three. | |
1128 | */ | |
1129 | int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, | |
7a6d7067 | 1130 | struct btrfs_inode *inode, u64 start, u64 end) |
d899e052 | 1131 | { |
3ffbd68c | 1132 | struct btrfs_fs_info *fs_info = trans->fs_info; |
7a6d7067 | 1133 | struct btrfs_root *root = inode->root; |
d899e052 YZ |
1134 | struct extent_buffer *leaf; |
1135 | struct btrfs_path *path; | |
1136 | struct btrfs_file_extent_item *fi; | |
1137 | struct btrfs_key key; | |
920bbbfb | 1138 | struct btrfs_key new_key; |
d899e052 YZ |
1139 | u64 bytenr; |
1140 | u64 num_bytes; | |
1141 | u64 extent_end; | |
5d4f98a2 | 1142 | u64 orig_offset; |
d899e052 YZ |
1143 | u64 other_start; |
1144 | u64 other_end; | |
920bbbfb YZ |
1145 | u64 split; |
1146 | int del_nr = 0; | |
1147 | int del_slot = 0; | |
6c7d54ac | 1148 | int recow; |
d899e052 | 1149 | int ret; |
7a6d7067 | 1150 | u64 ino = btrfs_ino(inode); |
d899e052 | 1151 | |
d899e052 | 1152 | path = btrfs_alloc_path(); |
d8926bb3 MF |
1153 | if (!path) |
1154 | return -ENOMEM; | |
d899e052 | 1155 | again: |
6c7d54ac | 1156 | recow = 0; |
920bbbfb | 1157 | split = start; |
33345d01 | 1158 | key.objectid = ino; |
d899e052 | 1159 | key.type = BTRFS_EXTENT_DATA_KEY; |
920bbbfb | 1160 | key.offset = split; |
d899e052 YZ |
1161 | |
1162 | ret = btrfs_search_slot(trans, root, &key, path, -1, 1); | |
41415730 JB |
1163 | if (ret < 0) |
1164 | goto out; | |
d899e052 YZ |
1165 | if (ret > 0 && path->slots[0] > 0) |
1166 | path->slots[0]--; | |
1167 | ||
1168 | leaf = path->nodes[0]; | |
1169 | btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); | |
9c8e63db JB |
1170 | if (key.objectid != ino || |
1171 | key.type != BTRFS_EXTENT_DATA_KEY) { | |
1172 | ret = -EINVAL; | |
1173 | btrfs_abort_transaction(trans, ret); | |
1174 | goto out; | |
1175 | } | |
d899e052 YZ |
1176 | fi = btrfs_item_ptr(leaf, path->slots[0], |
1177 | struct btrfs_file_extent_item); | |
9c8e63db JB |
1178 | if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_PREALLOC) { |
1179 | ret = -EINVAL; | |
1180 | btrfs_abort_transaction(trans, ret); | |
1181 | goto out; | |
1182 | } | |
d899e052 | 1183 | extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); |
9c8e63db JB |
1184 | if (key.offset > start || extent_end < end) { |
1185 | ret = -EINVAL; | |
1186 | btrfs_abort_transaction(trans, ret); | |
1187 | goto out; | |
1188 | } | |
d899e052 YZ |
1189 | |
1190 | bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); | |
1191 | num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); | |
5d4f98a2 | 1192 | orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi); |
6c7d54ac YZ |
1193 | memcpy(&new_key, &key, sizeof(new_key)); |
1194 | ||
1195 | if (start == key.offset && end < extent_end) { | |
1196 | other_start = 0; | |
1197 | other_end = start; | |
1198 | if (extent_mergeable(leaf, path->slots[0] - 1, | |
33345d01 | 1199 | ino, bytenr, orig_offset, |
6c7d54ac YZ |
1200 | &other_start, &other_end)) { |
1201 | new_key.offset = end; | |
0b246afa | 1202 | btrfs_set_item_key_safe(fs_info, path, &new_key); |
6c7d54ac YZ |
1203 | fi = btrfs_item_ptr(leaf, path->slots[0], |
1204 | struct btrfs_file_extent_item); | |
224ecce5 JB |
1205 | btrfs_set_file_extent_generation(leaf, fi, |
1206 | trans->transid); | |
6c7d54ac YZ |
1207 | btrfs_set_file_extent_num_bytes(leaf, fi, |
1208 | extent_end - end); | |
1209 | btrfs_set_file_extent_offset(leaf, fi, | |
1210 | end - orig_offset); | |
1211 | fi = btrfs_item_ptr(leaf, path->slots[0] - 1, | |
1212 | struct btrfs_file_extent_item); | |
224ecce5 JB |
1213 | btrfs_set_file_extent_generation(leaf, fi, |
1214 | trans->transid); | |
6c7d54ac YZ |
1215 | btrfs_set_file_extent_num_bytes(leaf, fi, |
1216 | end - other_start); | |
1217 | btrfs_mark_buffer_dirty(leaf); | |
1218 | goto out; | |
1219 | } | |
1220 | } | |
1221 | ||
1222 | if (start > key.offset && end == extent_end) { | |
1223 | other_start = end; | |
1224 | other_end = 0; | |
1225 | if (extent_mergeable(leaf, path->slots[0] + 1, | |
33345d01 | 1226 | ino, bytenr, orig_offset, |
6c7d54ac YZ |
1227 | &other_start, &other_end)) { |
1228 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
1229 | struct btrfs_file_extent_item); | |
1230 | btrfs_set_file_extent_num_bytes(leaf, fi, | |
1231 | start - key.offset); | |
224ecce5 JB |
1232 | btrfs_set_file_extent_generation(leaf, fi, |
1233 | trans->transid); | |
6c7d54ac YZ |
1234 | path->slots[0]++; |
1235 | new_key.offset = start; | |
0b246afa | 1236 | btrfs_set_item_key_safe(fs_info, path, &new_key); |
6c7d54ac YZ |
1237 | |
1238 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
1239 | struct btrfs_file_extent_item); | |
224ecce5 JB |
1240 | btrfs_set_file_extent_generation(leaf, fi, |
1241 | trans->transid); | |
6c7d54ac YZ |
1242 | btrfs_set_file_extent_num_bytes(leaf, fi, |
1243 | other_end - start); | |
1244 | btrfs_set_file_extent_offset(leaf, fi, | |
1245 | start - orig_offset); | |
1246 | btrfs_mark_buffer_dirty(leaf); | |
1247 | goto out; | |
1248 | } | |
1249 | } | |
d899e052 | 1250 | |
920bbbfb YZ |
1251 | while (start > key.offset || end < extent_end) { |
1252 | if (key.offset == start) | |
1253 | split = end; | |
1254 | ||
920bbbfb YZ |
1255 | new_key.offset = split; |
1256 | ret = btrfs_duplicate_item(trans, root, path, &new_key); | |
1257 | if (ret == -EAGAIN) { | |
b3b4aa74 | 1258 | btrfs_release_path(path); |
920bbbfb | 1259 | goto again; |
d899e052 | 1260 | } |
79787eaa | 1261 | if (ret < 0) { |
66642832 | 1262 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
1263 | goto out; |
1264 | } | |
d899e052 | 1265 | |
920bbbfb YZ |
1266 | leaf = path->nodes[0]; |
1267 | fi = btrfs_item_ptr(leaf, path->slots[0] - 1, | |
d899e052 | 1268 | struct btrfs_file_extent_item); |
224ecce5 | 1269 | btrfs_set_file_extent_generation(leaf, fi, trans->transid); |
d899e052 | 1270 | btrfs_set_file_extent_num_bytes(leaf, fi, |
920bbbfb YZ |
1271 | split - key.offset); |
1272 | ||
1273 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
1274 | struct btrfs_file_extent_item); | |
1275 | ||
224ecce5 | 1276 | btrfs_set_file_extent_generation(leaf, fi, trans->transid); |
920bbbfb YZ |
1277 | btrfs_set_file_extent_offset(leaf, fi, split - orig_offset); |
1278 | btrfs_set_file_extent_num_bytes(leaf, fi, | |
1279 | extent_end - split); | |
d899e052 YZ |
1280 | btrfs_mark_buffer_dirty(leaf); |
1281 | ||
84f7d8e6 | 1282 | ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, |
2ff7e61e | 1283 | 0, root->root_key.objectid, |
b06c4bf5 | 1284 | ino, orig_offset); |
9c8e63db JB |
1285 | if (ret) { |
1286 | btrfs_abort_transaction(trans, ret); | |
1287 | goto out; | |
1288 | } | |
d899e052 | 1289 | |
920bbbfb YZ |
1290 | if (split == start) { |
1291 | key.offset = start; | |
1292 | } else { | |
9c8e63db JB |
1293 | if (start != key.offset) { |
1294 | ret = -EINVAL; | |
1295 | btrfs_abort_transaction(trans, ret); | |
1296 | goto out; | |
1297 | } | |
d899e052 | 1298 | path->slots[0]--; |
920bbbfb | 1299 | extent_end = end; |
d899e052 | 1300 | } |
6c7d54ac | 1301 | recow = 1; |
d899e052 YZ |
1302 | } |
1303 | ||
920bbbfb YZ |
1304 | other_start = end; |
1305 | other_end = 0; | |
6c7d54ac | 1306 | if (extent_mergeable(leaf, path->slots[0] + 1, |
33345d01 | 1307 | ino, bytenr, orig_offset, |
6c7d54ac YZ |
1308 | &other_start, &other_end)) { |
1309 | if (recow) { | |
b3b4aa74 | 1310 | btrfs_release_path(path); |
6c7d54ac YZ |
1311 | goto again; |
1312 | } | |
920bbbfb YZ |
1313 | extent_end = other_end; |
1314 | del_slot = path->slots[0] + 1; | |
1315 | del_nr++; | |
84f7d8e6 | 1316 | ret = btrfs_free_extent(trans, root, bytenr, num_bytes, |
920bbbfb | 1317 | 0, root->root_key.objectid, |
b06c4bf5 | 1318 | ino, orig_offset); |
9c8e63db JB |
1319 | if (ret) { |
1320 | btrfs_abort_transaction(trans, ret); | |
1321 | goto out; | |
1322 | } | |
d899e052 | 1323 | } |
920bbbfb YZ |
1324 | other_start = 0; |
1325 | other_end = start; | |
6c7d54ac | 1326 | if (extent_mergeable(leaf, path->slots[0] - 1, |
33345d01 | 1327 | ino, bytenr, orig_offset, |
6c7d54ac YZ |
1328 | &other_start, &other_end)) { |
1329 | if (recow) { | |
b3b4aa74 | 1330 | btrfs_release_path(path); |
6c7d54ac YZ |
1331 | goto again; |
1332 | } | |
920bbbfb YZ |
1333 | key.offset = other_start; |
1334 | del_slot = path->slots[0]; | |
1335 | del_nr++; | |
84f7d8e6 | 1336 | ret = btrfs_free_extent(trans, root, bytenr, num_bytes, |
920bbbfb | 1337 | 0, root->root_key.objectid, |
b06c4bf5 | 1338 | ino, orig_offset); |
9c8e63db JB |
1339 | if (ret) { |
1340 | btrfs_abort_transaction(trans, ret); | |
1341 | goto out; | |
1342 | } | |
920bbbfb YZ |
1343 | } |
1344 | if (del_nr == 0) { | |
3f6fae95 SL |
1345 | fi = btrfs_item_ptr(leaf, path->slots[0], |
1346 | struct btrfs_file_extent_item); | |
920bbbfb YZ |
1347 | btrfs_set_file_extent_type(leaf, fi, |
1348 | BTRFS_FILE_EXTENT_REG); | |
224ecce5 | 1349 | btrfs_set_file_extent_generation(leaf, fi, trans->transid); |
920bbbfb | 1350 | btrfs_mark_buffer_dirty(leaf); |
6c7d54ac | 1351 | } else { |
3f6fae95 SL |
1352 | fi = btrfs_item_ptr(leaf, del_slot - 1, |
1353 | struct btrfs_file_extent_item); | |
6c7d54ac YZ |
1354 | btrfs_set_file_extent_type(leaf, fi, |
1355 | BTRFS_FILE_EXTENT_REG); | |
224ecce5 | 1356 | btrfs_set_file_extent_generation(leaf, fi, trans->transid); |
6c7d54ac YZ |
1357 | btrfs_set_file_extent_num_bytes(leaf, fi, |
1358 | extent_end - key.offset); | |
1359 | btrfs_mark_buffer_dirty(leaf); | |
920bbbfb | 1360 | |
6c7d54ac | 1361 | ret = btrfs_del_items(trans, root, path, del_slot, del_nr); |
79787eaa | 1362 | if (ret < 0) { |
66642832 | 1363 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
1364 | goto out; |
1365 | } | |
6c7d54ac | 1366 | } |
920bbbfb | 1367 | out: |
d899e052 YZ |
1368 | btrfs_free_path(path); |
1369 | return 0; | |
1370 | } | |
1371 | ||
b1bf862e CM |
1372 | /* |
1373 | * on error we return an unlocked page and the error value | |
1374 | * on success we return a locked page and 0 | |
1375 | */ | |
bb1591b4 CM |
1376 | static int prepare_uptodate_page(struct inode *inode, |
1377 | struct page *page, u64 pos, | |
b6316429 | 1378 | bool force_uptodate) |
b1bf862e CM |
1379 | { |
1380 | int ret = 0; | |
1381 | ||
09cbfeaf | 1382 | if (((pos & (PAGE_SIZE - 1)) || force_uptodate) && |
b6316429 | 1383 | !PageUptodate(page)) { |
b1bf862e CM |
1384 | ret = btrfs_readpage(NULL, page); |
1385 | if (ret) | |
1386 | return ret; | |
1387 | lock_page(page); | |
1388 | if (!PageUptodate(page)) { | |
1389 | unlock_page(page); | |
1390 | return -EIO; | |
1391 | } | |
bb1591b4 CM |
1392 | if (page->mapping != inode->i_mapping) { |
1393 | unlock_page(page); | |
1394 | return -EAGAIN; | |
1395 | } | |
b1bf862e CM |
1396 | } |
1397 | return 0; | |
1398 | } | |
1399 | ||
39279cc3 | 1400 | /* |
376cc685 | 1401 | * this just gets pages into the page cache and locks them down. |
39279cc3 | 1402 | */ |
b37392ea MX |
1403 | static noinline int prepare_pages(struct inode *inode, struct page **pages, |
1404 | size_t num_pages, loff_t pos, | |
1405 | size_t write_bytes, bool force_uptodate) | |
39279cc3 CM |
1406 | { |
1407 | int i; | |
09cbfeaf | 1408 | unsigned long index = pos >> PAGE_SHIFT; |
3b16a4e3 | 1409 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
fc28b62d | 1410 | int err = 0; |
376cc685 | 1411 | int faili; |
8c2383c3 | 1412 | |
39279cc3 | 1413 | for (i = 0; i < num_pages; i++) { |
bb1591b4 | 1414 | again: |
a94733d0 | 1415 | pages[i] = find_or_create_page(inode->i_mapping, index + i, |
e3a41a5b | 1416 | mask | __GFP_WRITE); |
39279cc3 | 1417 | if (!pages[i]) { |
b1bf862e CM |
1418 | faili = i - 1; |
1419 | err = -ENOMEM; | |
1420 | goto fail; | |
1421 | } | |
1422 | ||
1423 | if (i == 0) | |
bb1591b4 | 1424 | err = prepare_uptodate_page(inode, pages[i], pos, |
b6316429 | 1425 | force_uptodate); |
bb1591b4 CM |
1426 | if (!err && i == num_pages - 1) |
1427 | err = prepare_uptodate_page(inode, pages[i], | |
b6316429 | 1428 | pos + write_bytes, false); |
b1bf862e | 1429 | if (err) { |
09cbfeaf | 1430 | put_page(pages[i]); |
bb1591b4 CM |
1431 | if (err == -EAGAIN) { |
1432 | err = 0; | |
1433 | goto again; | |
1434 | } | |
b1bf862e CM |
1435 | faili = i - 1; |
1436 | goto fail; | |
39279cc3 | 1437 | } |
ccd467d6 | 1438 | wait_on_page_writeback(pages[i]); |
39279cc3 | 1439 | } |
376cc685 MX |
1440 | |
1441 | return 0; | |
1442 | fail: | |
1443 | while (faili >= 0) { | |
1444 | unlock_page(pages[faili]); | |
09cbfeaf | 1445 | put_page(pages[faili]); |
376cc685 MX |
1446 | faili--; |
1447 | } | |
1448 | return err; | |
1449 | ||
1450 | } | |
1451 | ||
1452 | /* | |
1453 | * This function locks the extent and properly waits for data=ordered extents | |
1454 | * to finish before allowing the pages to be modified if need. | |
1455 | * | |
1456 | * The return value: | |
1457 | * 1 - the extent is locked | |
1458 | * 0 - the extent is not locked, and everything is OK | |
1459 | * -EAGAIN - need re-prepare the pages | |
1460 | * the other < 0 number - Something wrong happens | |
1461 | */ | |
1462 | static noinline int | |
2cff578c | 1463 | lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages, |
376cc685 | 1464 | size_t num_pages, loff_t pos, |
2e78c927 | 1465 | size_t write_bytes, |
376cc685 MX |
1466 | u64 *lockstart, u64 *lockend, |
1467 | struct extent_state **cached_state) | |
1468 | { | |
3ffbd68c | 1469 | struct btrfs_fs_info *fs_info = inode->root->fs_info; |
376cc685 MX |
1470 | u64 start_pos; |
1471 | u64 last_pos; | |
1472 | int i; | |
1473 | int ret = 0; | |
1474 | ||
0b246afa | 1475 | start_pos = round_down(pos, fs_info->sectorsize); |
2e78c927 | 1476 | last_pos = start_pos |
da17066c | 1477 | + round_up(pos + write_bytes - start_pos, |
0b246afa | 1478 | fs_info->sectorsize) - 1; |
376cc685 | 1479 | |
e3b8a485 | 1480 | if (start_pos < inode->vfs_inode.i_size) { |
e6dcd2dc | 1481 | struct btrfs_ordered_extent *ordered; |
a7e3b975 | 1482 | |
2cff578c NB |
1483 | lock_extent_bits(&inode->io_tree, start_pos, last_pos, |
1484 | cached_state); | |
b88935bf MX |
1485 | ordered = btrfs_lookup_ordered_range(inode, start_pos, |
1486 | last_pos - start_pos + 1); | |
e6dcd2dc CM |
1487 | if (ordered && |
1488 | ordered->file_offset + ordered->len > start_pos && | |
376cc685 | 1489 | ordered->file_offset <= last_pos) { |
2cff578c | 1490 | unlock_extent_cached(&inode->io_tree, start_pos, |
e43bbe5e | 1491 | last_pos, cached_state); |
e6dcd2dc CM |
1492 | for (i = 0; i < num_pages; i++) { |
1493 | unlock_page(pages[i]); | |
09cbfeaf | 1494 | put_page(pages[i]); |
e6dcd2dc | 1495 | } |
2cff578c NB |
1496 | btrfs_start_ordered_extent(&inode->vfs_inode, |
1497 | ordered, 1); | |
b88935bf MX |
1498 | btrfs_put_ordered_extent(ordered); |
1499 | return -EAGAIN; | |
e6dcd2dc CM |
1500 | } |
1501 | if (ordered) | |
1502 | btrfs_put_ordered_extent(ordered); | |
e3b8a485 FM |
1503 | clear_extent_bit(&inode->io_tree, start_pos, last_pos, |
1504 | EXTENT_DIRTY | EXTENT_DELALLOC | | |
1505 | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, | |
ae0f1625 | 1506 | 0, 0, cached_state); |
376cc685 MX |
1507 | *lockstart = start_pos; |
1508 | *lockend = last_pos; | |
1509 | ret = 1; | |
0762704b | 1510 | } |
376cc685 | 1511 | |
e6dcd2dc | 1512 | for (i = 0; i < num_pages; i++) { |
32c7f202 WF |
1513 | if (clear_page_dirty_for_io(pages[i])) |
1514 | account_page_redirty(pages[i]); | |
e6dcd2dc CM |
1515 | set_page_extent_mapped(pages[i]); |
1516 | WARN_ON(!PageLocked(pages[i])); | |
1517 | } | |
b1bf862e | 1518 | |
376cc685 | 1519 | return ret; |
39279cc3 CM |
1520 | } |
1521 | ||
85b7ab67 | 1522 | static noinline int check_can_nocow(struct btrfs_inode *inode, loff_t pos, |
7ee9e440 JB |
1523 | size_t *write_bytes) |
1524 | { | |
3ffbd68c | 1525 | struct btrfs_fs_info *fs_info = inode->root->fs_info; |
85b7ab67 | 1526 | struct btrfs_root *root = inode->root; |
7ee9e440 JB |
1527 | struct btrfs_ordered_extent *ordered; |
1528 | u64 lockstart, lockend; | |
1529 | u64 num_bytes; | |
1530 | int ret; | |
1531 | ||
ea14b57f | 1532 | ret = btrfs_start_write_no_snapshotting(root); |
8257b2dc MX |
1533 | if (!ret) |
1534 | return -ENOSPC; | |
1535 | ||
0b246afa | 1536 | lockstart = round_down(pos, fs_info->sectorsize); |
da17066c | 1537 | lockend = round_up(pos + *write_bytes, |
0b246afa | 1538 | fs_info->sectorsize) - 1; |
7ee9e440 JB |
1539 | |
1540 | while (1) { | |
85b7ab67 | 1541 | lock_extent(&inode->io_tree, lockstart, lockend); |
7ee9e440 JB |
1542 | ordered = btrfs_lookup_ordered_range(inode, lockstart, |
1543 | lockend - lockstart + 1); | |
1544 | if (!ordered) { | |
1545 | break; | |
1546 | } | |
85b7ab67 NB |
1547 | unlock_extent(&inode->io_tree, lockstart, lockend); |
1548 | btrfs_start_ordered_extent(&inode->vfs_inode, ordered, 1); | |
7ee9e440 JB |
1549 | btrfs_put_ordered_extent(ordered); |
1550 | } | |
1551 | ||
7ee9e440 | 1552 | num_bytes = lockend - lockstart + 1; |
85b7ab67 NB |
1553 | ret = can_nocow_extent(&inode->vfs_inode, lockstart, &num_bytes, |
1554 | NULL, NULL, NULL); | |
7ee9e440 JB |
1555 | if (ret <= 0) { |
1556 | ret = 0; | |
ea14b57f | 1557 | btrfs_end_write_no_snapshotting(root); |
7ee9e440 | 1558 | } else { |
c933956d MX |
1559 | *write_bytes = min_t(size_t, *write_bytes , |
1560 | num_bytes - pos + lockstart); | |
7ee9e440 JB |
1561 | } |
1562 | ||
85b7ab67 | 1563 | unlock_extent(&inode->io_tree, lockstart, lockend); |
7ee9e440 JB |
1564 | |
1565 | return ret; | |
1566 | } | |
1567 | ||
e4af400a GR |
1568 | static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb, |
1569 | struct iov_iter *i) | |
4b46fce2 | 1570 | { |
e4af400a GR |
1571 | struct file *file = iocb->ki_filp; |
1572 | loff_t pos = iocb->ki_pos; | |
496ad9aa | 1573 | struct inode *inode = file_inode(file); |
0b246afa | 1574 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
11c65dcc | 1575 | struct btrfs_root *root = BTRFS_I(inode)->root; |
11c65dcc | 1576 | struct page **pages = NULL; |
376cc685 | 1577 | struct extent_state *cached_state = NULL; |
364ecf36 | 1578 | struct extent_changeset *data_reserved = NULL; |
7ee9e440 | 1579 | u64 release_bytes = 0; |
376cc685 MX |
1580 | u64 lockstart; |
1581 | u64 lockend; | |
d0215f3e JB |
1582 | size_t num_written = 0; |
1583 | int nrptrs; | |
c9149235 | 1584 | int ret = 0; |
7ee9e440 | 1585 | bool only_release_metadata = false; |
b6316429 | 1586 | bool force_page_uptodate = false; |
4b46fce2 | 1587 | |
09cbfeaf KS |
1588 | nrptrs = min(DIV_ROUND_UP(iov_iter_count(i), PAGE_SIZE), |
1589 | PAGE_SIZE / (sizeof(struct page *))); | |
142349f5 WF |
1590 | nrptrs = min(nrptrs, current->nr_dirtied_pause - current->nr_dirtied); |
1591 | nrptrs = max(nrptrs, 8); | |
31e818fe | 1592 | pages = kmalloc_array(nrptrs, sizeof(struct page *), GFP_KERNEL); |
d0215f3e JB |
1593 | if (!pages) |
1594 | return -ENOMEM; | |
ab93dbec | 1595 | |
d0215f3e | 1596 | while (iov_iter_count(i) > 0) { |
09cbfeaf | 1597 | size_t offset = pos & (PAGE_SIZE - 1); |
2e78c927 | 1598 | size_t sector_offset; |
d0215f3e | 1599 | size_t write_bytes = min(iov_iter_count(i), |
09cbfeaf | 1600 | nrptrs * (size_t)PAGE_SIZE - |
8c2383c3 | 1601 | offset); |
ed6078f7 | 1602 | size_t num_pages = DIV_ROUND_UP(write_bytes + offset, |
09cbfeaf | 1603 | PAGE_SIZE); |
7ee9e440 | 1604 | size_t reserve_bytes; |
d0215f3e JB |
1605 | size_t dirty_pages; |
1606 | size_t copied; | |
2e78c927 CR |
1607 | size_t dirty_sectors; |
1608 | size_t num_sectors; | |
79f015f2 | 1609 | int extents_locked; |
39279cc3 | 1610 | |
8c2383c3 | 1611 | WARN_ON(num_pages > nrptrs); |
1832a6d5 | 1612 | |
914ee295 XZ |
1613 | /* |
1614 | * Fault pages before locking them in prepare_pages | |
1615 | * to avoid recursive lock | |
1616 | */ | |
d0215f3e | 1617 | if (unlikely(iov_iter_fault_in_readable(i, write_bytes))) { |
914ee295 | 1618 | ret = -EFAULT; |
d0215f3e | 1619 | break; |
914ee295 XZ |
1620 | } |
1621 | ||
da17066c | 1622 | sector_offset = pos & (fs_info->sectorsize - 1); |
2e78c927 | 1623 | reserve_bytes = round_up(write_bytes + sector_offset, |
da17066c | 1624 | fs_info->sectorsize); |
d9d8b2a5 | 1625 | |
364ecf36 QW |
1626 | extent_changeset_release(data_reserved); |
1627 | ret = btrfs_check_data_free_space(inode, &data_reserved, pos, | |
1628 | write_bytes); | |
c6887cd1 JB |
1629 | if (ret < 0) { |
1630 | if ((BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW | | |
1631 | BTRFS_INODE_PREALLOC)) && | |
85b7ab67 NB |
1632 | check_can_nocow(BTRFS_I(inode), pos, |
1633 | &write_bytes) > 0) { | |
c6887cd1 JB |
1634 | /* |
1635 | * For nodata cow case, no need to reserve | |
1636 | * data space. | |
1637 | */ | |
1638 | only_release_metadata = true; | |
1639 | /* | |
1640 | * our prealloc extent may be smaller than | |
1641 | * write_bytes, so scale down. | |
1642 | */ | |
1643 | num_pages = DIV_ROUND_UP(write_bytes + offset, | |
1644 | PAGE_SIZE); | |
1645 | reserve_bytes = round_up(write_bytes + | |
1646 | sector_offset, | |
da17066c | 1647 | fs_info->sectorsize); |
c6887cd1 JB |
1648 | } else { |
1649 | break; | |
1650 | } | |
1651 | } | |
1832a6d5 | 1652 | |
8b62f87b | 1653 | WARN_ON(reserve_bytes == 0); |
9f3db423 NB |
1654 | ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), |
1655 | reserve_bytes); | |
7ee9e440 JB |
1656 | if (ret) { |
1657 | if (!only_release_metadata) | |
bc42bda2 QW |
1658 | btrfs_free_reserved_data_space(inode, |
1659 | data_reserved, pos, | |
1660 | write_bytes); | |
8257b2dc | 1661 | else |
ea14b57f | 1662 | btrfs_end_write_no_snapshotting(root); |
7ee9e440 JB |
1663 | break; |
1664 | } | |
1665 | ||
1666 | release_bytes = reserve_bytes; | |
376cc685 | 1667 | again: |
4a64001f JB |
1668 | /* |
1669 | * This is going to setup the pages array with the number of | |
1670 | * pages we want, so we don't really need to worry about the | |
1671 | * contents of pages from loop to loop | |
1672 | */ | |
b37392ea MX |
1673 | ret = prepare_pages(inode, pages, num_pages, |
1674 | pos, write_bytes, | |
b6316429 | 1675 | force_page_uptodate); |
8b62f87b JB |
1676 | if (ret) { |
1677 | btrfs_delalloc_release_extents(BTRFS_I(inode), | |
43b18595 | 1678 | reserve_bytes, true); |
d0215f3e | 1679 | break; |
8b62f87b | 1680 | } |
39279cc3 | 1681 | |
79f015f2 GR |
1682 | extents_locked = lock_and_cleanup_extent_if_need( |
1683 | BTRFS_I(inode), pages, | |
2cff578c NB |
1684 | num_pages, pos, write_bytes, &lockstart, |
1685 | &lockend, &cached_state); | |
79f015f2 GR |
1686 | if (extents_locked < 0) { |
1687 | if (extents_locked == -EAGAIN) | |
376cc685 | 1688 | goto again; |
8b62f87b | 1689 | btrfs_delalloc_release_extents(BTRFS_I(inode), |
43b18595 | 1690 | reserve_bytes, true); |
79f015f2 | 1691 | ret = extents_locked; |
376cc685 | 1692 | break; |
376cc685 MX |
1693 | } |
1694 | ||
ee22f0c4 | 1695 | copied = btrfs_copy_from_user(pos, write_bytes, pages, i); |
b1bf862e | 1696 | |
0b246afa | 1697 | num_sectors = BTRFS_BYTES_TO_BLKS(fs_info, reserve_bytes); |
56244ef1 | 1698 | dirty_sectors = round_up(copied + sector_offset, |
0b246afa JM |
1699 | fs_info->sectorsize); |
1700 | dirty_sectors = BTRFS_BYTES_TO_BLKS(fs_info, dirty_sectors); | |
56244ef1 | 1701 | |
b1bf862e CM |
1702 | /* |
1703 | * if we have trouble faulting in the pages, fall | |
1704 | * back to one page at a time | |
1705 | */ | |
1706 | if (copied < write_bytes) | |
1707 | nrptrs = 1; | |
1708 | ||
b6316429 JB |
1709 | if (copied == 0) { |
1710 | force_page_uptodate = true; | |
56244ef1 | 1711 | dirty_sectors = 0; |
b1bf862e | 1712 | dirty_pages = 0; |
b6316429 JB |
1713 | } else { |
1714 | force_page_uptodate = false; | |
ed6078f7 | 1715 | dirty_pages = DIV_ROUND_UP(copied + offset, |
09cbfeaf | 1716 | PAGE_SIZE); |
b6316429 | 1717 | } |
914ee295 | 1718 | |
2e78c927 | 1719 | if (num_sectors > dirty_sectors) { |
8b8b08cb CM |
1720 | /* release everything except the sectors we dirtied */ |
1721 | release_bytes -= dirty_sectors << | |
0b246afa | 1722 | fs_info->sb->s_blocksize_bits; |
485290a7 | 1723 | if (only_release_metadata) { |
691fa059 | 1724 | btrfs_delalloc_release_metadata(BTRFS_I(inode), |
43b18595 | 1725 | release_bytes, true); |
485290a7 QW |
1726 | } else { |
1727 | u64 __pos; | |
1728 | ||
da17066c | 1729 | __pos = round_down(pos, |
0b246afa | 1730 | fs_info->sectorsize) + |
09cbfeaf | 1731 | (dirty_pages << PAGE_SHIFT); |
bc42bda2 QW |
1732 | btrfs_delalloc_release_space(inode, |
1733 | data_reserved, __pos, | |
43b18595 | 1734 | release_bytes, true); |
485290a7 | 1735 | } |
914ee295 XZ |
1736 | } |
1737 | ||
2e78c927 | 1738 | release_bytes = round_up(copied + sector_offset, |
0b246afa | 1739 | fs_info->sectorsize); |
376cc685 MX |
1740 | |
1741 | if (copied > 0) | |
2ff7e61e | 1742 | ret = btrfs_dirty_pages(inode, pages, dirty_pages, |
94f45071 | 1743 | pos, copied, &cached_state); |
79f015f2 | 1744 | if (extents_locked) |
376cc685 | 1745 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, |
e43bbe5e | 1746 | lockstart, lockend, &cached_state); |
43b18595 | 1747 | btrfs_delalloc_release_extents(BTRFS_I(inode), reserve_bytes, |
336a8bb8 | 1748 | true); |
f1de9683 MX |
1749 | if (ret) { |
1750 | btrfs_drop_pages(pages, num_pages); | |
376cc685 | 1751 | break; |
f1de9683 | 1752 | } |
39279cc3 | 1753 | |
376cc685 | 1754 | release_bytes = 0; |
8257b2dc | 1755 | if (only_release_metadata) |
ea14b57f | 1756 | btrfs_end_write_no_snapshotting(root); |
8257b2dc | 1757 | |
7ee9e440 | 1758 | if (only_release_metadata && copied > 0) { |
da17066c | 1759 | lockstart = round_down(pos, |
0b246afa | 1760 | fs_info->sectorsize); |
da17066c | 1761 | lockend = round_up(pos + copied, |
0b246afa | 1762 | fs_info->sectorsize) - 1; |
7ee9e440 JB |
1763 | |
1764 | set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, | |
1765 | lockend, EXTENT_NORESERVE, NULL, | |
1766 | NULL, GFP_NOFS); | |
1767 | only_release_metadata = false; | |
1768 | } | |
1769 | ||
f1de9683 MX |
1770 | btrfs_drop_pages(pages, num_pages); |
1771 | ||
d0215f3e JB |
1772 | cond_resched(); |
1773 | ||
d0e1d66b | 1774 | balance_dirty_pages_ratelimited(inode->i_mapping); |
0b246afa | 1775 | if (dirty_pages < (fs_info->nodesize >> PAGE_SHIFT) + 1) |
2ff7e61e | 1776 | btrfs_btree_balance_dirty(fs_info); |
cb843a6f | 1777 | |
914ee295 XZ |
1778 | pos += copied; |
1779 | num_written += copied; | |
d0215f3e | 1780 | } |
39279cc3 | 1781 | |
d0215f3e JB |
1782 | kfree(pages); |
1783 | ||
7ee9e440 | 1784 | if (release_bytes) { |
8257b2dc | 1785 | if (only_release_metadata) { |
ea14b57f | 1786 | btrfs_end_write_no_snapshotting(root); |
691fa059 | 1787 | btrfs_delalloc_release_metadata(BTRFS_I(inode), |
43b18595 | 1788 | release_bytes, true); |
8257b2dc | 1789 | } else { |
bc42bda2 QW |
1790 | btrfs_delalloc_release_space(inode, data_reserved, |
1791 | round_down(pos, fs_info->sectorsize), | |
43b18595 | 1792 | release_bytes, true); |
8257b2dc | 1793 | } |
7ee9e440 JB |
1794 | } |
1795 | ||
364ecf36 | 1796 | extent_changeset_free(data_reserved); |
d0215f3e JB |
1797 | return num_written ? num_written : ret; |
1798 | } | |
1799 | ||
1af5bb49 | 1800 | static ssize_t __btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from) |
d0215f3e JB |
1801 | { |
1802 | struct file *file = iocb->ki_filp; | |
728404da | 1803 | struct inode *inode = file_inode(file); |
e4af400a | 1804 | loff_t pos; |
d0215f3e JB |
1805 | ssize_t written; |
1806 | ssize_t written_buffered; | |
1807 | loff_t endbyte; | |
1808 | int err; | |
1809 | ||
1af5bb49 | 1810 | written = generic_file_direct_write(iocb, from); |
d0215f3e | 1811 | |
0c949334 | 1812 | if (written < 0 || !iov_iter_count(from)) |
d0215f3e JB |
1813 | return written; |
1814 | ||
e4af400a GR |
1815 | pos = iocb->ki_pos; |
1816 | written_buffered = btrfs_buffered_write(iocb, from); | |
d0215f3e JB |
1817 | if (written_buffered < 0) { |
1818 | err = written_buffered; | |
1819 | goto out; | |
39279cc3 | 1820 | } |
075bdbdb FM |
1821 | /* |
1822 | * Ensure all data is persisted. We want the next direct IO read to be | |
1823 | * able to read what was just written. | |
1824 | */ | |
d0215f3e | 1825 | endbyte = pos + written_buffered - 1; |
728404da | 1826 | err = btrfs_fdatawrite_range(inode, pos, endbyte); |
075bdbdb FM |
1827 | if (err) |
1828 | goto out; | |
728404da | 1829 | err = filemap_fdatawait_range(inode->i_mapping, pos, endbyte); |
d0215f3e JB |
1830 | if (err) |
1831 | goto out; | |
1832 | written += written_buffered; | |
867c4f93 | 1833 | iocb->ki_pos = pos + written_buffered; |
09cbfeaf KS |
1834 | invalidate_mapping_pages(file->f_mapping, pos >> PAGE_SHIFT, |
1835 | endbyte >> PAGE_SHIFT); | |
39279cc3 | 1836 | out: |
d0215f3e JB |
1837 | return written ? written : err; |
1838 | } | |
5b92ee72 | 1839 | |
6c760c07 JB |
1840 | static void update_time_for_write(struct inode *inode) |
1841 | { | |
95582b00 | 1842 | struct timespec64 now; |
6c760c07 JB |
1843 | |
1844 | if (IS_NOCMTIME(inode)) | |
1845 | return; | |
1846 | ||
c2050a45 | 1847 | now = current_time(inode); |
95582b00 | 1848 | if (!timespec64_equal(&inode->i_mtime, &now)) |
6c760c07 JB |
1849 | inode->i_mtime = now; |
1850 | ||
95582b00 | 1851 | if (!timespec64_equal(&inode->i_ctime, &now)) |
6c760c07 JB |
1852 | inode->i_ctime = now; |
1853 | ||
1854 | if (IS_I_VERSION(inode)) | |
1855 | inode_inc_iversion(inode); | |
1856 | } | |
1857 | ||
b30ac0fc AV |
1858 | static ssize_t btrfs_file_write_iter(struct kiocb *iocb, |
1859 | struct iov_iter *from) | |
d0215f3e JB |
1860 | { |
1861 | struct file *file = iocb->ki_filp; | |
496ad9aa | 1862 | struct inode *inode = file_inode(file); |
0b246afa | 1863 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
d0215f3e | 1864 | struct btrfs_root *root = BTRFS_I(inode)->root; |
0c1a98c8 | 1865 | u64 start_pos; |
3ac0d7b9 | 1866 | u64 end_pos; |
d0215f3e | 1867 | ssize_t num_written = 0; |
b812ce28 | 1868 | bool sync = (file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host); |
3309dd04 | 1869 | ssize_t err; |
ff0fa732 | 1870 | loff_t pos; |
edf064e7 | 1871 | size_t count = iov_iter_count(from); |
27772b68 CR |
1872 | loff_t oldsize; |
1873 | int clean_page = 0; | |
d0215f3e | 1874 | |
91f9943e CH |
1875 | if (!(iocb->ki_flags & IOCB_DIRECT) && |
1876 | (iocb->ki_flags & IOCB_NOWAIT)) | |
1877 | return -EOPNOTSUPP; | |
1878 | ||
ff0fa732 GR |
1879 | if (!inode_trylock(inode)) { |
1880 | if (iocb->ki_flags & IOCB_NOWAIT) | |
edf064e7 | 1881 | return -EAGAIN; |
ff0fa732 GR |
1882 | inode_lock(inode); |
1883 | } | |
1884 | ||
1885 | err = generic_write_checks(iocb, from); | |
1886 | if (err <= 0) { | |
1887 | inode_unlock(inode); | |
1888 | return err; | |
1889 | } | |
1890 | ||
1891 | pos = iocb->ki_pos; | |
1892 | if (iocb->ki_flags & IOCB_NOWAIT) { | |
edf064e7 GR |
1893 | /* |
1894 | * We will allocate space in case nodatacow is not set, | |
1895 | * so bail | |
1896 | */ | |
1897 | if (!(BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW | | |
1898 | BTRFS_INODE_PREALLOC)) || | |
1899 | check_can_nocow(BTRFS_I(inode), pos, &count) <= 0) { | |
1900 | inode_unlock(inode); | |
1901 | return -EAGAIN; | |
1902 | } | |
d0215f3e JB |
1903 | } |
1904 | ||
3309dd04 | 1905 | current->backing_dev_info = inode_to_bdi(inode); |
5fa8e0a1 | 1906 | err = file_remove_privs(file); |
d0215f3e | 1907 | if (err) { |
5955102c | 1908 | inode_unlock(inode); |
d0215f3e JB |
1909 | goto out; |
1910 | } | |
1911 | ||
1912 | /* | |
1913 | * If BTRFS flips readonly due to some impossible error | |
1914 | * (fs_info->fs_state now has BTRFS_SUPER_FLAG_ERROR), | |
1915 | * although we have opened a file as writable, we have | |
1916 | * to stop this write operation to ensure FS consistency. | |
1917 | */ | |
0b246afa | 1918 | if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) { |
5955102c | 1919 | inode_unlock(inode); |
d0215f3e JB |
1920 | err = -EROFS; |
1921 | goto out; | |
1922 | } | |
1923 | ||
6c760c07 JB |
1924 | /* |
1925 | * We reserve space for updating the inode when we reserve space for the | |
1926 | * extent we are going to write, so we will enospc out there. We don't | |
1927 | * need to start yet another transaction to update the inode as we will | |
1928 | * update the inode when we finish writing whatever data we write. | |
1929 | */ | |
1930 | update_time_for_write(inode); | |
d0215f3e | 1931 | |
0b246afa | 1932 | start_pos = round_down(pos, fs_info->sectorsize); |
27772b68 CR |
1933 | oldsize = i_size_read(inode); |
1934 | if (start_pos > oldsize) { | |
3ac0d7b9 | 1935 | /* Expand hole size to cover write data, preventing empty gap */ |
da17066c | 1936 | end_pos = round_up(pos + count, |
0b246afa | 1937 | fs_info->sectorsize); |
27772b68 | 1938 | err = btrfs_cont_expand(inode, oldsize, end_pos); |
0c1a98c8 | 1939 | if (err) { |
5955102c | 1940 | inode_unlock(inode); |
0c1a98c8 MX |
1941 | goto out; |
1942 | } | |
0b246afa | 1943 | if (start_pos > round_up(oldsize, fs_info->sectorsize)) |
27772b68 | 1944 | clean_page = 1; |
0c1a98c8 MX |
1945 | } |
1946 | ||
b812ce28 JB |
1947 | if (sync) |
1948 | atomic_inc(&BTRFS_I(inode)->sync_writers); | |
1949 | ||
2ba48ce5 | 1950 | if (iocb->ki_flags & IOCB_DIRECT) { |
1af5bb49 | 1951 | num_written = __btrfs_direct_write(iocb, from); |
d0215f3e | 1952 | } else { |
e4af400a | 1953 | num_written = btrfs_buffered_write(iocb, from); |
d0215f3e | 1954 | if (num_written > 0) |
867c4f93 | 1955 | iocb->ki_pos = pos + num_written; |
27772b68 CR |
1956 | if (clean_page) |
1957 | pagecache_isize_extended(inode, oldsize, | |
1958 | i_size_read(inode)); | |
d0215f3e JB |
1959 | } |
1960 | ||
5955102c | 1961 | inode_unlock(inode); |
2ff3e9b6 | 1962 | |
5a3f23d5 | 1963 | /* |
6c760c07 JB |
1964 | * We also have to set last_sub_trans to the current log transid, |
1965 | * otherwise subsequent syncs to a file that's been synced in this | |
bb7ab3b9 | 1966 | * transaction will appear to have already occurred. |
5a3f23d5 | 1967 | */ |
2f2ff0ee | 1968 | spin_lock(&BTRFS_I(inode)->lock); |
6c760c07 | 1969 | BTRFS_I(inode)->last_sub_trans = root->log_transid; |
2f2ff0ee | 1970 | spin_unlock(&BTRFS_I(inode)->lock); |
e2592217 CH |
1971 | if (num_written > 0) |
1972 | num_written = generic_write_sync(iocb, num_written); | |
0a3404dc | 1973 | |
b812ce28 JB |
1974 | if (sync) |
1975 | atomic_dec(&BTRFS_I(inode)->sync_writers); | |
0a3404dc | 1976 | out: |
39279cc3 | 1977 | current->backing_dev_info = NULL; |
39279cc3 CM |
1978 | return num_written ? num_written : err; |
1979 | } | |
1980 | ||
d397712b | 1981 | int btrfs_release_file(struct inode *inode, struct file *filp) |
e1b81e67 | 1982 | { |
23b5ec74 JB |
1983 | struct btrfs_file_private *private = filp->private_data; |
1984 | ||
23b5ec74 JB |
1985 | if (private && private->filldir_buf) |
1986 | kfree(private->filldir_buf); | |
1987 | kfree(private); | |
1988 | filp->private_data = NULL; | |
1989 | ||
f6dc45c7 CM |
1990 | /* |
1991 | * ordered_data_close is set by settattr when we are about to truncate | |
1992 | * a file from a non-zero size to a zero size. This tries to | |
1993 | * flush down new bytes that may have been written if the | |
1994 | * application were using truncate to replace a file in place. | |
1995 | */ | |
1996 | if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, | |
1997 | &BTRFS_I(inode)->runtime_flags)) | |
1998 | filemap_flush(inode->i_mapping); | |
e1b81e67 M |
1999 | return 0; |
2000 | } | |
2001 | ||
669249ee FM |
2002 | static int start_ordered_ops(struct inode *inode, loff_t start, loff_t end) |
2003 | { | |
2004 | int ret; | |
343e4fc1 | 2005 | struct blk_plug plug; |
669249ee | 2006 | |
343e4fc1 LB |
2007 | /* |
2008 | * This is only called in fsync, which would do synchronous writes, so | |
2009 | * a plug can merge adjacent IOs as much as possible. Esp. in case of | |
2010 | * multiple disks using raid profile, a large IO can be split to | |
2011 | * several segments of stripe length (currently 64K). | |
2012 | */ | |
2013 | blk_start_plug(&plug); | |
669249ee | 2014 | atomic_inc(&BTRFS_I(inode)->sync_writers); |
728404da | 2015 | ret = btrfs_fdatawrite_range(inode, start, end); |
669249ee | 2016 | atomic_dec(&BTRFS_I(inode)->sync_writers); |
343e4fc1 | 2017 | blk_finish_plug(&plug); |
669249ee FM |
2018 | |
2019 | return ret; | |
2020 | } | |
2021 | ||
d352ac68 CM |
2022 | /* |
2023 | * fsync call for both files and directories. This logs the inode into | |
2024 | * the tree log instead of forcing full commits whenever possible. | |
2025 | * | |
2026 | * It needs to call filemap_fdatawait so that all ordered extent updates are | |
2027 | * in the metadata btree are up to date for copying to the log. | |
2028 | * | |
2029 | * It drops the inode mutex before doing the tree log commit. This is an | |
2030 | * important optimization for directories because holding the mutex prevents | |
2031 | * new operations on the dir while we write to disk. | |
2032 | */ | |
02c24a82 | 2033 | int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) |
39279cc3 | 2034 | { |
de17e793 | 2035 | struct dentry *dentry = file_dentry(file); |
2b0143b5 | 2036 | struct inode *inode = d_inode(dentry); |
0b246afa | 2037 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
39279cc3 | 2038 | struct btrfs_root *root = BTRFS_I(inode)->root; |
39279cc3 | 2039 | struct btrfs_trans_handle *trans; |
8b050d35 | 2040 | struct btrfs_log_ctx ctx; |
333427a5 | 2041 | int ret = 0, err; |
9dcbeed4 | 2042 | u64 len; |
39279cc3 | 2043 | |
9dcbeed4 DS |
2044 | /* |
2045 | * The range length can be represented by u64, we have to do the typecasts | |
2046 | * to avoid signed overflow if it's [0, LLONG_MAX] eg. from fsync() | |
2047 | */ | |
2048 | len = (u64)end - (u64)start + 1; | |
1abe9b8a | 2049 | trace_btrfs_sync_file(file, datasync); |
257c62e1 | 2050 | |
ebb70442 LB |
2051 | btrfs_init_log_ctx(&ctx, inode); |
2052 | ||
90abccf2 MX |
2053 | /* |
2054 | * We write the dirty pages in the range and wait until they complete | |
2055 | * out of the ->i_mutex. If so, we can flush the dirty pages by | |
2ab28f32 JB |
2056 | * multi-task, and make the performance up. See |
2057 | * btrfs_wait_ordered_range for an explanation of the ASYNC check. | |
90abccf2 | 2058 | */ |
669249ee | 2059 | ret = start_ordered_ops(inode, start, end); |
90abccf2 | 2060 | if (ret) |
333427a5 | 2061 | goto out; |
90abccf2 | 2062 | |
5955102c | 2063 | inode_lock(inode); |
2ecb7923 | 2064 | atomic_inc(&root->log_batch); |
b5e6c3e1 | 2065 | |
669249ee | 2066 | /* |
b5e6c3e1 JB |
2067 | * We have to do this here to avoid the priority inversion of waiting on |
2068 | * IO of a lower priority task while holding a transaciton open. | |
669249ee | 2069 | */ |
b5e6c3e1 | 2070 | ret = btrfs_wait_ordered_range(inode, start, len); |
669249ee | 2071 | if (ret) { |
5955102c | 2072 | inode_unlock(inode); |
669249ee | 2073 | goto out; |
0ef8b726 | 2074 | } |
2ecb7923 | 2075 | atomic_inc(&root->log_batch); |
257c62e1 | 2076 | |
a4abeea4 | 2077 | smp_mb(); |
0f8939b8 | 2078 | if (btrfs_inode_in_log(BTRFS_I(inode), fs_info->generation) || |
ca5788ab | 2079 | BTRFS_I(inode)->last_trans <= fs_info->last_trans_committed) { |
5dc562c5 | 2080 | /* |
01327610 | 2081 | * We've had everything committed since the last time we were |
5dc562c5 JB |
2082 | * modified so clear this flag in case it was set for whatever |
2083 | * reason, it's no longer relevant. | |
2084 | */ | |
2085 | clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC, | |
2086 | &BTRFS_I(inode)->runtime_flags); | |
0596a904 FM |
2087 | /* |
2088 | * An ordered extent might have started before and completed | |
2089 | * already with io errors, in which case the inode was not | |
2090 | * updated and we end up here. So check the inode's mapping | |
333427a5 JL |
2091 | * for any errors that might have happened since we last |
2092 | * checked called fsync. | |
0596a904 | 2093 | */ |
333427a5 | 2094 | ret = filemap_check_wb_err(inode->i_mapping, file->f_wb_err); |
5955102c | 2095 | inode_unlock(inode); |
15ee9bc7 JB |
2096 | goto out; |
2097 | } | |
15ee9bc7 | 2098 | |
5039eddc JB |
2099 | /* |
2100 | * We use start here because we will need to wait on the IO to complete | |
2101 | * in btrfs_sync_log, which could require joining a transaction (for | |
2102 | * example checking cross references in the nocow path). If we use join | |
2103 | * here we could get into a situation where we're waiting on IO to | |
2104 | * happen that is blocked on a transaction trying to commit. With start | |
2105 | * we inc the extwriter counter, so we wait for all extwriters to exit | |
2106 | * before we start blocking join'ers. This comment is to keep somebody | |
2107 | * from thinking they are super smart and changing this to | |
2108 | * btrfs_join_transaction *cough*Josef*cough*. | |
2109 | */ | |
a22285a6 YZ |
2110 | trans = btrfs_start_transaction(root, 0); |
2111 | if (IS_ERR(trans)) { | |
2112 | ret = PTR_ERR(trans); | |
5955102c | 2113 | inode_unlock(inode); |
39279cc3 CM |
2114 | goto out; |
2115 | } | |
5039eddc | 2116 | trans->sync = true; |
e02119d5 | 2117 | |
e5b84f7a | 2118 | ret = btrfs_log_dentry_safe(trans, dentry, start, end, &ctx); |
02c24a82 | 2119 | if (ret < 0) { |
a0634be5 FDBM |
2120 | /* Fallthrough and commit/free transaction. */ |
2121 | ret = 1; | |
02c24a82 | 2122 | } |
49eb7e46 CM |
2123 | |
2124 | /* we've logged all the items and now have a consistent | |
2125 | * version of the file in the log. It is possible that | |
2126 | * someone will come in and modify the file, but that's | |
2127 | * fine because the log is consistent on disk, and we | |
2128 | * have references to all of the file's extents | |
2129 | * | |
2130 | * It is possible that someone will come in and log the | |
2131 | * file again, but that will end up using the synchronization | |
2132 | * inside btrfs_sync_log to keep things safe. | |
2133 | */ | |
5955102c | 2134 | inode_unlock(inode); |
49eb7e46 | 2135 | |
8407f553 FM |
2136 | /* |
2137 | * If any of the ordered extents had an error, just return it to user | |
2138 | * space, so that the application knows some writes didn't succeed and | |
2139 | * can take proper action (retry for e.g.). Blindly committing the | |
2140 | * transaction in this case, would fool userspace that everything was | |
2141 | * successful. And we also want to make sure our log doesn't contain | |
2142 | * file extent items pointing to extents that weren't fully written to - | |
2143 | * just like in the non fast fsync path, where we check for the ordered | |
2144 | * operation's error flag before writing to the log tree and return -EIO | |
2145 | * if any of them had this flag set (btrfs_wait_ordered_range) - | |
2146 | * therefore we need to check for errors in the ordered operations, | |
2147 | * which are indicated by ctx.io_err. | |
2148 | */ | |
2149 | if (ctx.io_err) { | |
3a45bb20 | 2150 | btrfs_end_transaction(trans); |
8407f553 FM |
2151 | ret = ctx.io_err; |
2152 | goto out; | |
2153 | } | |
2154 | ||
257c62e1 | 2155 | if (ret != BTRFS_NO_LOG_SYNC) { |
0ef8b726 | 2156 | if (!ret) { |
8b050d35 | 2157 | ret = btrfs_sync_log(trans, root, &ctx); |
0ef8b726 | 2158 | if (!ret) { |
3a45bb20 | 2159 | ret = btrfs_end_transaction(trans); |
0ef8b726 | 2160 | goto out; |
2ab28f32 | 2161 | } |
257c62e1 | 2162 | } |
3a45bb20 | 2163 | ret = btrfs_commit_transaction(trans); |
257c62e1 | 2164 | } else { |
3a45bb20 | 2165 | ret = btrfs_end_transaction(trans); |
e02119d5 | 2166 | } |
39279cc3 | 2167 | out: |
ebb70442 | 2168 | ASSERT(list_empty(&ctx.list)); |
333427a5 JL |
2169 | err = file_check_and_advance_wb_err(file); |
2170 | if (!ret) | |
2171 | ret = err; | |
014e4ac4 | 2172 | return ret > 0 ? -EIO : ret; |
39279cc3 CM |
2173 | } |
2174 | ||
f0f37e2f | 2175 | static const struct vm_operations_struct btrfs_file_vm_ops = { |
92fee66d | 2176 | .fault = filemap_fault, |
f1820361 | 2177 | .map_pages = filemap_map_pages, |
9ebefb18 CM |
2178 | .page_mkwrite = btrfs_page_mkwrite, |
2179 | }; | |
2180 | ||
2181 | static int btrfs_file_mmap(struct file *filp, struct vm_area_struct *vma) | |
2182 | { | |
058a457e MX |
2183 | struct address_space *mapping = filp->f_mapping; |
2184 | ||
2185 | if (!mapping->a_ops->readpage) | |
2186 | return -ENOEXEC; | |
2187 | ||
9ebefb18 | 2188 | file_accessed(filp); |
058a457e | 2189 | vma->vm_ops = &btrfs_file_vm_ops; |
058a457e | 2190 | |
9ebefb18 CM |
2191 | return 0; |
2192 | } | |
2193 | ||
35339c24 | 2194 | static int hole_mergeable(struct btrfs_inode *inode, struct extent_buffer *leaf, |
2aaa6655 JB |
2195 | int slot, u64 start, u64 end) |
2196 | { | |
2197 | struct btrfs_file_extent_item *fi; | |
2198 | struct btrfs_key key; | |
2199 | ||
2200 | if (slot < 0 || slot >= btrfs_header_nritems(leaf)) | |
2201 | return 0; | |
2202 | ||
2203 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
35339c24 | 2204 | if (key.objectid != btrfs_ino(inode) || |
2aaa6655 JB |
2205 | key.type != BTRFS_EXTENT_DATA_KEY) |
2206 | return 0; | |
2207 | ||
2208 | fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); | |
2209 | ||
2210 | if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG) | |
2211 | return 0; | |
2212 | ||
2213 | if (btrfs_file_extent_disk_bytenr(leaf, fi)) | |
2214 | return 0; | |
2215 | ||
2216 | if (key.offset == end) | |
2217 | return 1; | |
2218 | if (key.offset + btrfs_file_extent_num_bytes(leaf, fi) == start) | |
2219 | return 1; | |
2220 | return 0; | |
2221 | } | |
2222 | ||
a012a74e NB |
2223 | static int fill_holes(struct btrfs_trans_handle *trans, |
2224 | struct btrfs_inode *inode, | |
2225 | struct btrfs_path *path, u64 offset, u64 end) | |
2aaa6655 | 2226 | { |
3ffbd68c | 2227 | struct btrfs_fs_info *fs_info = trans->fs_info; |
a012a74e | 2228 | struct btrfs_root *root = inode->root; |
2aaa6655 JB |
2229 | struct extent_buffer *leaf; |
2230 | struct btrfs_file_extent_item *fi; | |
2231 | struct extent_map *hole_em; | |
a012a74e | 2232 | struct extent_map_tree *em_tree = &inode->extent_tree; |
2aaa6655 JB |
2233 | struct btrfs_key key; |
2234 | int ret; | |
2235 | ||
0b246afa | 2236 | if (btrfs_fs_incompat(fs_info, NO_HOLES)) |
16e7549f JB |
2237 | goto out; |
2238 | ||
a012a74e | 2239 | key.objectid = btrfs_ino(inode); |
2aaa6655 JB |
2240 | key.type = BTRFS_EXTENT_DATA_KEY; |
2241 | key.offset = offset; | |
2242 | ||
2aaa6655 | 2243 | ret = btrfs_search_slot(trans, root, &key, path, 0, 1); |
f94480bd JB |
2244 | if (ret <= 0) { |
2245 | /* | |
2246 | * We should have dropped this offset, so if we find it then | |
2247 | * something has gone horribly wrong. | |
2248 | */ | |
2249 | if (ret == 0) | |
2250 | ret = -EINVAL; | |
2aaa6655 | 2251 | return ret; |
f94480bd | 2252 | } |
2aaa6655 JB |
2253 | |
2254 | leaf = path->nodes[0]; | |
a012a74e | 2255 | if (hole_mergeable(inode, leaf, path->slots[0] - 1, offset, end)) { |
2aaa6655 JB |
2256 | u64 num_bytes; |
2257 | ||
2258 | path->slots[0]--; | |
2259 | fi = btrfs_item_ptr(leaf, path->slots[0], | |
2260 | struct btrfs_file_extent_item); | |
2261 | num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + | |
2262 | end - offset; | |
2263 | btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes); | |
2264 | btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes); | |
2265 | btrfs_set_file_extent_offset(leaf, fi, 0); | |
2266 | btrfs_mark_buffer_dirty(leaf); | |
2267 | goto out; | |
2268 | } | |
2269 | ||
1707e26d | 2270 | if (hole_mergeable(inode, leaf, path->slots[0], offset, end)) { |
2aaa6655 JB |
2271 | u64 num_bytes; |
2272 | ||
2aaa6655 | 2273 | key.offset = offset; |
0b246afa | 2274 | btrfs_set_item_key_safe(fs_info, path, &key); |
2aaa6655 JB |
2275 | fi = btrfs_item_ptr(leaf, path->slots[0], |
2276 | struct btrfs_file_extent_item); | |
2277 | num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + end - | |
2278 | offset; | |
2279 | btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes); | |
2280 | btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes); | |
2281 | btrfs_set_file_extent_offset(leaf, fi, 0); | |
2282 | btrfs_mark_buffer_dirty(leaf); | |
2283 | goto out; | |
2284 | } | |
2285 | btrfs_release_path(path); | |
2286 | ||
a012a74e | 2287 | ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), |
f85b7379 | 2288 | offset, 0, 0, end - offset, 0, end - offset, 0, 0, 0); |
2aaa6655 JB |
2289 | if (ret) |
2290 | return ret; | |
2291 | ||
2292 | out: | |
2293 | btrfs_release_path(path); | |
2294 | ||
2295 | hole_em = alloc_extent_map(); | |
2296 | if (!hole_em) { | |
2297 | btrfs_drop_extent_cache(inode, offset, end - 1, 0); | |
a012a74e | 2298 | set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags); |
2aaa6655 JB |
2299 | } else { |
2300 | hole_em->start = offset; | |
2301 | hole_em->len = end - offset; | |
cc95bef6 | 2302 | hole_em->ram_bytes = hole_em->len; |
2aaa6655 JB |
2303 | hole_em->orig_start = offset; |
2304 | ||
2305 | hole_em->block_start = EXTENT_MAP_HOLE; | |
2306 | hole_em->block_len = 0; | |
b4939680 | 2307 | hole_em->orig_block_len = 0; |
0b246afa | 2308 | hole_em->bdev = fs_info->fs_devices->latest_bdev; |
2aaa6655 JB |
2309 | hole_em->compress_type = BTRFS_COMPRESS_NONE; |
2310 | hole_em->generation = trans->transid; | |
2311 | ||
2312 | do { | |
2313 | btrfs_drop_extent_cache(inode, offset, end - 1, 0); | |
2314 | write_lock(&em_tree->lock); | |
09a2a8f9 | 2315 | ret = add_extent_mapping(em_tree, hole_em, 1); |
2aaa6655 JB |
2316 | write_unlock(&em_tree->lock); |
2317 | } while (ret == -EEXIST); | |
2318 | free_extent_map(hole_em); | |
2319 | if (ret) | |
2320 | set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, | |
a012a74e | 2321 | &inode->runtime_flags); |
2aaa6655 JB |
2322 | } |
2323 | ||
2324 | return 0; | |
2325 | } | |
2326 | ||
d7781546 QW |
2327 | /* |
2328 | * Find a hole extent on given inode and change start/len to the end of hole | |
2329 | * extent.(hole/vacuum extent whose em->start <= start && | |
2330 | * em->start + em->len > start) | |
2331 | * When a hole extent is found, return 1 and modify start/len. | |
2332 | */ | |
2333 | static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len) | |
2334 | { | |
609805d8 | 2335 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
d7781546 QW |
2336 | struct extent_map *em; |
2337 | int ret = 0; | |
2338 | ||
609805d8 FM |
2339 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, |
2340 | round_down(*start, fs_info->sectorsize), | |
2341 | round_up(*len, fs_info->sectorsize), 0); | |
9986277e DC |
2342 | if (IS_ERR(em)) |
2343 | return PTR_ERR(em); | |
d7781546 QW |
2344 | |
2345 | /* Hole or vacuum extent(only exists in no-hole mode) */ | |
2346 | if (em->block_start == EXTENT_MAP_HOLE) { | |
2347 | ret = 1; | |
2348 | *len = em->start + em->len > *start + *len ? | |
2349 | 0 : *start + *len - em->start - em->len; | |
2350 | *start = em->start + em->len; | |
2351 | } | |
2352 | free_extent_map(em); | |
2353 | return ret; | |
2354 | } | |
2355 | ||
f27451f2 FM |
2356 | static int btrfs_punch_hole_lock_range(struct inode *inode, |
2357 | const u64 lockstart, | |
2358 | const u64 lockend, | |
2359 | struct extent_state **cached_state) | |
2360 | { | |
2361 | while (1) { | |
2362 | struct btrfs_ordered_extent *ordered; | |
2363 | int ret; | |
2364 | ||
2365 | truncate_pagecache_range(inode, lockstart, lockend); | |
2366 | ||
2367 | lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, | |
2368 | cached_state); | |
2369 | ordered = btrfs_lookup_first_ordered_extent(inode, lockend); | |
2370 | ||
2371 | /* | |
2372 | * We need to make sure we have no ordered extents in this range | |
2373 | * and nobody raced in and read a page in this range, if we did | |
2374 | * we need to try again. | |
2375 | */ | |
2376 | if ((!ordered || | |
2377 | (ordered->file_offset + ordered->len <= lockstart || | |
2378 | ordered->file_offset > lockend)) && | |
051c98eb DS |
2379 | !filemap_range_has_page(inode->i_mapping, |
2380 | lockstart, lockend)) { | |
f27451f2 FM |
2381 | if (ordered) |
2382 | btrfs_put_ordered_extent(ordered); | |
2383 | break; | |
2384 | } | |
2385 | if (ordered) | |
2386 | btrfs_put_ordered_extent(ordered); | |
2387 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, | |
2388 | lockend, cached_state); | |
2389 | ret = btrfs_wait_ordered_range(inode, lockstart, | |
2390 | lockend - lockstart + 1); | |
2391 | if (ret) | |
2392 | return ret; | |
2393 | } | |
2394 | return 0; | |
2395 | } | |
2396 | ||
2aaa6655 JB |
2397 | static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) |
2398 | { | |
0b246afa | 2399 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
2aaa6655 JB |
2400 | struct btrfs_root *root = BTRFS_I(inode)->root; |
2401 | struct extent_state *cached_state = NULL; | |
2402 | struct btrfs_path *path; | |
2403 | struct btrfs_block_rsv *rsv; | |
2404 | struct btrfs_trans_handle *trans; | |
d7781546 QW |
2405 | u64 lockstart; |
2406 | u64 lockend; | |
2407 | u64 tail_start; | |
2408 | u64 tail_len; | |
2409 | u64 orig_start = offset; | |
2410 | u64 cur_offset; | |
5f52a2c5 | 2411 | u64 min_size = btrfs_calc_trans_metadata_size(fs_info, 1); |
2aaa6655 | 2412 | u64 drop_end; |
2aaa6655 JB |
2413 | int ret = 0; |
2414 | int err = 0; | |
6e4d6fa1 | 2415 | unsigned int rsv_count; |
9703fefe | 2416 | bool same_block; |
0b246afa | 2417 | bool no_holes = btrfs_fs_incompat(fs_info, NO_HOLES); |
a1a50f60 | 2418 | u64 ino_size; |
9703fefe | 2419 | bool truncated_block = false; |
e8c1c76e | 2420 | bool updated_inode = false; |
2aaa6655 | 2421 | |
0ef8b726 JB |
2422 | ret = btrfs_wait_ordered_range(inode, offset, len); |
2423 | if (ret) | |
2424 | return ret; | |
2aaa6655 | 2425 | |
5955102c | 2426 | inode_lock(inode); |
0b246afa | 2427 | ino_size = round_up(inode->i_size, fs_info->sectorsize); |
d7781546 QW |
2428 | ret = find_first_non_hole(inode, &offset, &len); |
2429 | if (ret < 0) | |
2430 | goto out_only_mutex; | |
2431 | if (ret && !len) { | |
2432 | /* Already in a large hole */ | |
2433 | ret = 0; | |
2434 | goto out_only_mutex; | |
2435 | } | |
2436 | ||
da17066c | 2437 | lockstart = round_up(offset, btrfs_inode_sectorsize(inode)); |
d7781546 | 2438 | lockend = round_down(offset + len, |
da17066c | 2439 | btrfs_inode_sectorsize(inode)) - 1; |
0b246afa JM |
2440 | same_block = (BTRFS_BYTES_TO_BLKS(fs_info, offset)) |
2441 | == (BTRFS_BYTES_TO_BLKS(fs_info, offset + len - 1)); | |
7426cc04 | 2442 | /* |
9703fefe | 2443 | * We needn't truncate any block which is beyond the end of the file |
7426cc04 MX |
2444 | * because we are sure there is no data there. |
2445 | */ | |
2aaa6655 | 2446 | /* |
9703fefe CR |
2447 | * Only do this if we are in the same block and we aren't doing the |
2448 | * entire block. | |
2aaa6655 | 2449 | */ |
0b246afa | 2450 | if (same_block && len < fs_info->sectorsize) { |
e8c1c76e | 2451 | if (offset < ino_size) { |
9703fefe CR |
2452 | truncated_block = true; |
2453 | ret = btrfs_truncate_block(inode, offset, len, 0); | |
e8c1c76e FM |
2454 | } else { |
2455 | ret = 0; | |
2456 | } | |
d7781546 | 2457 | goto out_only_mutex; |
2aaa6655 JB |
2458 | } |
2459 | ||
9703fefe | 2460 | /* zero back part of the first block */ |
12870f1c | 2461 | if (offset < ino_size) { |
9703fefe CR |
2462 | truncated_block = true; |
2463 | ret = btrfs_truncate_block(inode, offset, 0, 0); | |
7426cc04 | 2464 | if (ret) { |
5955102c | 2465 | inode_unlock(inode); |
7426cc04 MX |
2466 | return ret; |
2467 | } | |
2aaa6655 JB |
2468 | } |
2469 | ||
d7781546 QW |
2470 | /* Check the aligned pages after the first unaligned page, |
2471 | * if offset != orig_start, which means the first unaligned page | |
01327610 | 2472 | * including several following pages are already in holes, |
d7781546 QW |
2473 | * the extra check can be skipped */ |
2474 | if (offset == orig_start) { | |
2475 | /* after truncate page, check hole again */ | |
2476 | len = offset + len - lockstart; | |
2477 | offset = lockstart; | |
2478 | ret = find_first_non_hole(inode, &offset, &len); | |
2479 | if (ret < 0) | |
2480 | goto out_only_mutex; | |
2481 | if (ret && !len) { | |
2482 | ret = 0; | |
2483 | goto out_only_mutex; | |
2484 | } | |
2485 | lockstart = offset; | |
2486 | } | |
2487 | ||
2488 | /* Check the tail unaligned part is in a hole */ | |
2489 | tail_start = lockend + 1; | |
2490 | tail_len = offset + len - tail_start; | |
2491 | if (tail_len) { | |
2492 | ret = find_first_non_hole(inode, &tail_start, &tail_len); | |
2493 | if (unlikely(ret < 0)) | |
2494 | goto out_only_mutex; | |
2495 | if (!ret) { | |
2496 | /* zero the front end of the last page */ | |
2497 | if (tail_start + tail_len < ino_size) { | |
9703fefe CR |
2498 | truncated_block = true; |
2499 | ret = btrfs_truncate_block(inode, | |
2500 | tail_start + tail_len, | |
2501 | 0, 1); | |
d7781546 QW |
2502 | if (ret) |
2503 | goto out_only_mutex; | |
51f395ad | 2504 | } |
0061280d | 2505 | } |
2aaa6655 JB |
2506 | } |
2507 | ||
2508 | if (lockend < lockstart) { | |
e8c1c76e FM |
2509 | ret = 0; |
2510 | goto out_only_mutex; | |
2aaa6655 JB |
2511 | } |
2512 | ||
f27451f2 FM |
2513 | ret = btrfs_punch_hole_lock_range(inode, lockstart, lockend, |
2514 | &cached_state); | |
2515 | if (ret) { | |
2516 | inode_unlock(inode); | |
2517 | goto out_only_mutex; | |
2aaa6655 JB |
2518 | } |
2519 | ||
2520 | path = btrfs_alloc_path(); | |
2521 | if (!path) { | |
2522 | ret = -ENOMEM; | |
2523 | goto out; | |
2524 | } | |
2525 | ||
2ff7e61e | 2526 | rsv = btrfs_alloc_block_rsv(fs_info, BTRFS_BLOCK_RSV_TEMP); |
2aaa6655 JB |
2527 | if (!rsv) { |
2528 | ret = -ENOMEM; | |
2529 | goto out_free; | |
2530 | } | |
5f52a2c5 | 2531 | rsv->size = btrfs_calc_trans_metadata_size(fs_info, 1); |
2aaa6655 JB |
2532 | rsv->failfast = 1; |
2533 | ||
2534 | /* | |
2535 | * 1 - update the inode | |
2536 | * 1 - removing the extents in the range | |
16e7549f | 2537 | * 1 - adding the hole extent if no_holes isn't set |
2aaa6655 | 2538 | */ |
16e7549f JB |
2539 | rsv_count = no_holes ? 2 : 3; |
2540 | trans = btrfs_start_transaction(root, rsv_count); | |
2aaa6655 JB |
2541 | if (IS_ERR(trans)) { |
2542 | err = PTR_ERR(trans); | |
2543 | goto out_free; | |
2544 | } | |
2545 | ||
0b246afa | 2546 | ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv, rsv, |
25d609f8 | 2547 | min_size, 0); |
2aaa6655 JB |
2548 | BUG_ON(ret); |
2549 | trans->block_rsv = rsv; | |
2550 | ||
d7781546 QW |
2551 | cur_offset = lockstart; |
2552 | len = lockend - cur_offset; | |
2aaa6655 JB |
2553 | while (cur_offset < lockend) { |
2554 | ret = __btrfs_drop_extents(trans, root, inode, path, | |
2555 | cur_offset, lockend + 1, | |
1acae57b | 2556 | &drop_end, 1, 0, 0, NULL); |
2aaa6655 JB |
2557 | if (ret != -ENOSPC) |
2558 | break; | |
2559 | ||
0b246afa | 2560 | trans->block_rsv = &fs_info->trans_block_rsv; |
2aaa6655 | 2561 | |
62fe51c1 | 2562 | if (cur_offset < drop_end && cur_offset < ino_size) { |
a012a74e NB |
2563 | ret = fill_holes(trans, BTRFS_I(inode), path, |
2564 | cur_offset, drop_end); | |
12870f1c | 2565 | if (ret) { |
f94480bd JB |
2566 | /* |
2567 | * If we failed then we didn't insert our hole | |
2568 | * entries for the area we dropped, so now the | |
2569 | * fs is corrupted, so we must abort the | |
2570 | * transaction. | |
2571 | */ | |
2572 | btrfs_abort_transaction(trans, ret); | |
12870f1c FM |
2573 | err = ret; |
2574 | break; | |
2575 | } | |
2aaa6655 JB |
2576 | } |
2577 | ||
2578 | cur_offset = drop_end; | |
2579 | ||
2580 | ret = btrfs_update_inode(trans, root, inode); | |
2581 | if (ret) { | |
2582 | err = ret; | |
2583 | break; | |
2584 | } | |
2585 | ||
3a45bb20 | 2586 | btrfs_end_transaction(trans); |
2ff7e61e | 2587 | btrfs_btree_balance_dirty(fs_info); |
2aaa6655 | 2588 | |
16e7549f | 2589 | trans = btrfs_start_transaction(root, rsv_count); |
2aaa6655 JB |
2590 | if (IS_ERR(trans)) { |
2591 | ret = PTR_ERR(trans); | |
2592 | trans = NULL; | |
2593 | break; | |
2594 | } | |
2595 | ||
0b246afa | 2596 | ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv, |
25d609f8 | 2597 | rsv, min_size, 0); |
2aaa6655 JB |
2598 | BUG_ON(ret); /* shouldn't happen */ |
2599 | trans->block_rsv = rsv; | |
d7781546 QW |
2600 | |
2601 | ret = find_first_non_hole(inode, &cur_offset, &len); | |
2602 | if (unlikely(ret < 0)) | |
2603 | break; | |
2604 | if (ret && !len) { | |
2605 | ret = 0; | |
2606 | break; | |
2607 | } | |
2aaa6655 JB |
2608 | } |
2609 | ||
2610 | if (ret) { | |
2611 | err = ret; | |
2612 | goto out_trans; | |
2613 | } | |
2614 | ||
0b246afa | 2615 | trans->block_rsv = &fs_info->trans_block_rsv; |
2959a32a FM |
2616 | /* |
2617 | * If we are using the NO_HOLES feature we might have had already an | |
2618 | * hole that overlaps a part of the region [lockstart, lockend] and | |
2619 | * ends at (or beyond) lockend. Since we have no file extent items to | |
2620 | * represent holes, drop_end can be less than lockend and so we must | |
2621 | * make sure we have an extent map representing the existing hole (the | |
2622 | * call to __btrfs_drop_extents() might have dropped the existing extent | |
2623 | * map representing the existing hole), otherwise the fast fsync path | |
2624 | * will not record the existence of the hole region | |
2625 | * [existing_hole_start, lockend]. | |
2626 | */ | |
2627 | if (drop_end <= lockend) | |
2628 | drop_end = lockend + 1; | |
fc19c5e7 FM |
2629 | /* |
2630 | * Don't insert file hole extent item if it's for a range beyond eof | |
2631 | * (because it's useless) or if it represents a 0 bytes range (when | |
2632 | * cur_offset == drop_end). | |
2633 | */ | |
2634 | if (cur_offset < ino_size && cur_offset < drop_end) { | |
a012a74e NB |
2635 | ret = fill_holes(trans, BTRFS_I(inode), path, |
2636 | cur_offset, drop_end); | |
12870f1c | 2637 | if (ret) { |
f94480bd JB |
2638 | /* Same comment as above. */ |
2639 | btrfs_abort_transaction(trans, ret); | |
12870f1c FM |
2640 | err = ret; |
2641 | goto out_trans; | |
2642 | } | |
2aaa6655 JB |
2643 | } |
2644 | ||
2645 | out_trans: | |
2646 | if (!trans) | |
2647 | goto out_free; | |
2648 | ||
e1f5790e | 2649 | inode_inc_iversion(inode); |
c2050a45 | 2650 | inode->i_mtime = inode->i_ctime = current_time(inode); |
e1f5790e | 2651 | |
0b246afa | 2652 | trans->block_rsv = &fs_info->trans_block_rsv; |
2aaa6655 | 2653 | ret = btrfs_update_inode(trans, root, inode); |
e8c1c76e | 2654 | updated_inode = true; |
3a45bb20 | 2655 | btrfs_end_transaction(trans); |
2ff7e61e | 2656 | btrfs_btree_balance_dirty(fs_info); |
2aaa6655 JB |
2657 | out_free: |
2658 | btrfs_free_path(path); | |
2ff7e61e | 2659 | btrfs_free_block_rsv(fs_info, rsv); |
2aaa6655 JB |
2660 | out: |
2661 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, | |
e43bbe5e | 2662 | &cached_state); |
d7781546 | 2663 | out_only_mutex: |
9703fefe | 2664 | if (!updated_inode && truncated_block && !ret && !err) { |
e8c1c76e FM |
2665 | /* |
2666 | * If we only end up zeroing part of a page, we still need to | |
2667 | * update the inode item, so that all the time fields are | |
2668 | * updated as well as the necessary btrfs inode in memory fields | |
2669 | * for detecting, at fsync time, if the inode isn't yet in the | |
2670 | * log tree or it's there but not up to date. | |
2671 | */ | |
2672 | trans = btrfs_start_transaction(root, 1); | |
2673 | if (IS_ERR(trans)) { | |
2674 | err = PTR_ERR(trans); | |
2675 | } else { | |
2676 | err = btrfs_update_inode(trans, root, inode); | |
3a45bb20 | 2677 | ret = btrfs_end_transaction(trans); |
e8c1c76e FM |
2678 | } |
2679 | } | |
5955102c | 2680 | inode_unlock(inode); |
2aaa6655 JB |
2681 | if (ret && !err) |
2682 | err = ret; | |
2683 | return err; | |
2684 | } | |
2685 | ||
14524a84 QW |
2686 | /* Helper structure to record which range is already reserved */ |
2687 | struct falloc_range { | |
2688 | struct list_head list; | |
2689 | u64 start; | |
2690 | u64 len; | |
2691 | }; | |
2692 | ||
2693 | /* | |
2694 | * Helper function to add falloc range | |
2695 | * | |
2696 | * Caller should have locked the larger range of extent containing | |
2697 | * [start, len) | |
2698 | */ | |
2699 | static int add_falloc_range(struct list_head *head, u64 start, u64 len) | |
2700 | { | |
2701 | struct falloc_range *prev = NULL; | |
2702 | struct falloc_range *range = NULL; | |
2703 | ||
2704 | if (list_empty(head)) | |
2705 | goto insert; | |
2706 | ||
2707 | /* | |
2708 | * As fallocate iterate by bytenr order, we only need to check | |
2709 | * the last range. | |
2710 | */ | |
2711 | prev = list_entry(head->prev, struct falloc_range, list); | |
2712 | if (prev->start + prev->len == start) { | |
2713 | prev->len += len; | |
2714 | return 0; | |
2715 | } | |
2716 | insert: | |
32fc932e | 2717 | range = kmalloc(sizeof(*range), GFP_KERNEL); |
14524a84 QW |
2718 | if (!range) |
2719 | return -ENOMEM; | |
2720 | range->start = start; | |
2721 | range->len = len; | |
2722 | list_add_tail(&range->list, head); | |
2723 | return 0; | |
2724 | } | |
2725 | ||
f27451f2 FM |
2726 | static int btrfs_fallocate_update_isize(struct inode *inode, |
2727 | const u64 end, | |
2728 | const int mode) | |
2729 | { | |
2730 | struct btrfs_trans_handle *trans; | |
2731 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
2732 | int ret; | |
2733 | int ret2; | |
2734 | ||
2735 | if (mode & FALLOC_FL_KEEP_SIZE || end <= i_size_read(inode)) | |
2736 | return 0; | |
2737 | ||
2738 | trans = btrfs_start_transaction(root, 1); | |
2739 | if (IS_ERR(trans)) | |
2740 | return PTR_ERR(trans); | |
2741 | ||
2742 | inode->i_ctime = current_time(inode); | |
2743 | i_size_write(inode, end); | |
2744 | btrfs_ordered_update_i_size(inode, end, NULL); | |
2745 | ret = btrfs_update_inode(trans, root, inode); | |
2746 | ret2 = btrfs_end_transaction(trans); | |
2747 | ||
2748 | return ret ? ret : ret2; | |
2749 | } | |
2750 | ||
81fdf638 FM |
2751 | enum { |
2752 | RANGE_BOUNDARY_WRITTEN_EXTENT = 0, | |
2753 | RANGE_BOUNDARY_PREALLOC_EXTENT = 1, | |
2754 | RANGE_BOUNDARY_HOLE = 2, | |
2755 | }; | |
2756 | ||
f27451f2 FM |
2757 | static int btrfs_zero_range_check_range_boundary(struct inode *inode, |
2758 | u64 offset) | |
2759 | { | |
2760 | const u64 sectorsize = btrfs_inode_sectorsize(inode); | |
2761 | struct extent_map *em; | |
81fdf638 | 2762 | int ret; |
f27451f2 FM |
2763 | |
2764 | offset = round_down(offset, sectorsize); | |
2765 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, offset, sectorsize, 0); | |
2766 | if (IS_ERR(em)) | |
2767 | return PTR_ERR(em); | |
2768 | ||
2769 | if (em->block_start == EXTENT_MAP_HOLE) | |
81fdf638 FM |
2770 | ret = RANGE_BOUNDARY_HOLE; |
2771 | else if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) | |
2772 | ret = RANGE_BOUNDARY_PREALLOC_EXTENT; | |
2773 | else | |
2774 | ret = RANGE_BOUNDARY_WRITTEN_EXTENT; | |
f27451f2 FM |
2775 | |
2776 | free_extent_map(em); | |
2777 | return ret; | |
2778 | } | |
2779 | ||
2780 | static int btrfs_zero_range(struct inode *inode, | |
2781 | loff_t offset, | |
2782 | loff_t len, | |
2783 | const int mode) | |
2784 | { | |
2785 | struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; | |
2786 | struct extent_map *em; | |
2787 | struct extent_changeset *data_reserved = NULL; | |
2788 | int ret; | |
2789 | u64 alloc_hint = 0; | |
2790 | const u64 sectorsize = btrfs_inode_sectorsize(inode); | |
2791 | u64 alloc_start = round_down(offset, sectorsize); | |
2792 | u64 alloc_end = round_up(offset + len, sectorsize); | |
2793 | u64 bytes_to_reserve = 0; | |
2794 | bool space_reserved = false; | |
2795 | ||
2796 | inode_dio_wait(inode); | |
2797 | ||
2798 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, | |
2799 | alloc_start, alloc_end - alloc_start, 0); | |
2800 | if (IS_ERR(em)) { | |
2801 | ret = PTR_ERR(em); | |
2802 | goto out; | |
2803 | } | |
2804 | ||
2805 | /* | |
2806 | * Avoid hole punching and extent allocation for some cases. More cases | |
2807 | * could be considered, but these are unlikely common and we keep things | |
2808 | * as simple as possible for now. Also, intentionally, if the target | |
2809 | * range contains one or more prealloc extents together with regular | |
2810 | * extents and holes, we drop all the existing extents and allocate a | |
2811 | * new prealloc extent, so that we get a larger contiguous disk extent. | |
2812 | */ | |
2813 | if (em->start <= alloc_start && | |
2814 | test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) { | |
2815 | const u64 em_end = em->start + em->len; | |
2816 | ||
2817 | if (em_end >= offset + len) { | |
2818 | /* | |
2819 | * The whole range is already a prealloc extent, | |
2820 | * do nothing except updating the inode's i_size if | |
2821 | * needed. | |
2822 | */ | |
2823 | free_extent_map(em); | |
2824 | ret = btrfs_fallocate_update_isize(inode, offset + len, | |
2825 | mode); | |
2826 | goto out; | |
2827 | } | |
2828 | /* | |
2829 | * Part of the range is already a prealloc extent, so operate | |
2830 | * only on the remaining part of the range. | |
2831 | */ | |
2832 | alloc_start = em_end; | |
2833 | ASSERT(IS_ALIGNED(alloc_start, sectorsize)); | |
2834 | len = offset + len - alloc_start; | |
2835 | offset = alloc_start; | |
2836 | alloc_hint = em->block_start + em->len; | |
2837 | } | |
2838 | free_extent_map(em); | |
2839 | ||
2840 | if (BTRFS_BYTES_TO_BLKS(fs_info, offset) == | |
2841 | BTRFS_BYTES_TO_BLKS(fs_info, offset + len - 1)) { | |
2842 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, | |
2843 | alloc_start, sectorsize, 0); | |
2844 | if (IS_ERR(em)) { | |
2845 | ret = PTR_ERR(em); | |
2846 | goto out; | |
2847 | } | |
2848 | ||
2849 | if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) { | |
2850 | free_extent_map(em); | |
2851 | ret = btrfs_fallocate_update_isize(inode, offset + len, | |
2852 | mode); | |
2853 | goto out; | |
2854 | } | |
2855 | if (len < sectorsize && em->block_start != EXTENT_MAP_HOLE) { | |
2856 | free_extent_map(em); | |
2857 | ret = btrfs_truncate_block(inode, offset, len, 0); | |
2858 | if (!ret) | |
2859 | ret = btrfs_fallocate_update_isize(inode, | |
2860 | offset + len, | |
2861 | mode); | |
2862 | return ret; | |
2863 | } | |
2864 | free_extent_map(em); | |
2865 | alloc_start = round_down(offset, sectorsize); | |
2866 | alloc_end = alloc_start + sectorsize; | |
2867 | goto reserve_space; | |
2868 | } | |
2869 | ||
2870 | alloc_start = round_up(offset, sectorsize); | |
2871 | alloc_end = round_down(offset + len, sectorsize); | |
2872 | ||
2873 | /* | |
2874 | * For unaligned ranges, check the pages at the boundaries, they might | |
2875 | * map to an extent, in which case we need to partially zero them, or | |
2876 | * they might map to a hole, in which case we need our allocation range | |
2877 | * to cover them. | |
2878 | */ | |
2879 | if (!IS_ALIGNED(offset, sectorsize)) { | |
2880 | ret = btrfs_zero_range_check_range_boundary(inode, offset); | |
2881 | if (ret < 0) | |
2882 | goto out; | |
81fdf638 | 2883 | if (ret == RANGE_BOUNDARY_HOLE) { |
f27451f2 FM |
2884 | alloc_start = round_down(offset, sectorsize); |
2885 | ret = 0; | |
81fdf638 | 2886 | } else if (ret == RANGE_BOUNDARY_WRITTEN_EXTENT) { |
f27451f2 FM |
2887 | ret = btrfs_truncate_block(inode, offset, 0, 0); |
2888 | if (ret) | |
2889 | goto out; | |
81fdf638 FM |
2890 | } else { |
2891 | ret = 0; | |
f27451f2 FM |
2892 | } |
2893 | } | |
2894 | ||
2895 | if (!IS_ALIGNED(offset + len, sectorsize)) { | |
2896 | ret = btrfs_zero_range_check_range_boundary(inode, | |
2897 | offset + len); | |
2898 | if (ret < 0) | |
2899 | goto out; | |
81fdf638 | 2900 | if (ret == RANGE_BOUNDARY_HOLE) { |
f27451f2 FM |
2901 | alloc_end = round_up(offset + len, sectorsize); |
2902 | ret = 0; | |
81fdf638 | 2903 | } else if (ret == RANGE_BOUNDARY_WRITTEN_EXTENT) { |
f27451f2 FM |
2904 | ret = btrfs_truncate_block(inode, offset + len, 0, 1); |
2905 | if (ret) | |
2906 | goto out; | |
81fdf638 FM |
2907 | } else { |
2908 | ret = 0; | |
f27451f2 FM |
2909 | } |
2910 | } | |
2911 | ||
2912 | reserve_space: | |
2913 | if (alloc_start < alloc_end) { | |
2914 | struct extent_state *cached_state = NULL; | |
2915 | const u64 lockstart = alloc_start; | |
2916 | const u64 lockend = alloc_end - 1; | |
2917 | ||
2918 | bytes_to_reserve = alloc_end - alloc_start; | |
2919 | ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode), | |
2920 | bytes_to_reserve); | |
2921 | if (ret < 0) | |
2922 | goto out; | |
2923 | space_reserved = true; | |
2924 | ret = btrfs_qgroup_reserve_data(inode, &data_reserved, | |
2925 | alloc_start, bytes_to_reserve); | |
2926 | if (ret) | |
2927 | goto out; | |
2928 | ret = btrfs_punch_hole_lock_range(inode, lockstart, lockend, | |
2929 | &cached_state); | |
2930 | if (ret) | |
2931 | goto out; | |
2932 | ret = btrfs_prealloc_file_range(inode, mode, alloc_start, | |
2933 | alloc_end - alloc_start, | |
2934 | i_blocksize(inode), | |
2935 | offset + len, &alloc_hint); | |
2936 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, | |
2937 | lockend, &cached_state); | |
2938 | /* btrfs_prealloc_file_range releases reserved space on error */ | |
9f13ce74 | 2939 | if (ret) { |
f27451f2 | 2940 | space_reserved = false; |
9f13ce74 FM |
2941 | goto out; |
2942 | } | |
f27451f2 | 2943 | } |
9f13ce74 | 2944 | ret = btrfs_fallocate_update_isize(inode, offset + len, mode); |
f27451f2 FM |
2945 | out: |
2946 | if (ret && space_reserved) | |
2947 | btrfs_free_reserved_data_space(inode, data_reserved, | |
2948 | alloc_start, bytes_to_reserve); | |
2949 | extent_changeset_free(data_reserved); | |
2950 | ||
2951 | return ret; | |
2952 | } | |
2953 | ||
2fe17c10 CH |
2954 | static long btrfs_fallocate(struct file *file, int mode, |
2955 | loff_t offset, loff_t len) | |
2956 | { | |
496ad9aa | 2957 | struct inode *inode = file_inode(file); |
2fe17c10 | 2958 | struct extent_state *cached_state = NULL; |
364ecf36 | 2959 | struct extent_changeset *data_reserved = NULL; |
14524a84 QW |
2960 | struct falloc_range *range; |
2961 | struct falloc_range *tmp; | |
2962 | struct list_head reserve_list; | |
2fe17c10 CH |
2963 | u64 cur_offset; |
2964 | u64 last_byte; | |
2965 | u64 alloc_start; | |
2966 | u64 alloc_end; | |
2967 | u64 alloc_hint = 0; | |
2968 | u64 locked_end; | |
14524a84 | 2969 | u64 actual_end = 0; |
2fe17c10 | 2970 | struct extent_map *em; |
da17066c | 2971 | int blocksize = btrfs_inode_sectorsize(inode); |
2fe17c10 CH |
2972 | int ret; |
2973 | ||
797f4277 MX |
2974 | alloc_start = round_down(offset, blocksize); |
2975 | alloc_end = round_up(offset + len, blocksize); | |
18513091 | 2976 | cur_offset = alloc_start; |
2fe17c10 | 2977 | |
2aaa6655 | 2978 | /* Make sure we aren't being give some crap mode */ |
f27451f2 FM |
2979 | if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE | |
2980 | FALLOC_FL_ZERO_RANGE)) | |
2fe17c10 CH |
2981 | return -EOPNOTSUPP; |
2982 | ||
2aaa6655 JB |
2983 | if (mode & FALLOC_FL_PUNCH_HOLE) |
2984 | return btrfs_punch_hole(inode, offset, len); | |
2985 | ||
d98456fc | 2986 | /* |
14524a84 QW |
2987 | * Only trigger disk allocation, don't trigger qgroup reserve |
2988 | * | |
2989 | * For qgroup space, it will be checked later. | |
d98456fc | 2990 | */ |
f27451f2 FM |
2991 | if (!(mode & FALLOC_FL_ZERO_RANGE)) { |
2992 | ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode), | |
2993 | alloc_end - alloc_start); | |
2994 | if (ret < 0) | |
2995 | return ret; | |
2996 | } | |
d98456fc | 2997 | |
5955102c | 2998 | inode_lock(inode); |
2a162ce9 DI |
2999 | |
3000 | if (!(mode & FALLOC_FL_KEEP_SIZE) && offset + len > inode->i_size) { | |
3001 | ret = inode_newsize_ok(inode, offset + len); | |
3002 | if (ret) | |
3003 | goto out; | |
3004 | } | |
2fe17c10 | 3005 | |
14524a84 QW |
3006 | /* |
3007 | * TODO: Move these two operations after we have checked | |
3008 | * accurate reserved space, or fallocate can still fail but | |
3009 | * with page truncated or size expanded. | |
3010 | * | |
3011 | * But that's a minor problem and won't do much harm BTW. | |
3012 | */ | |
2fe17c10 | 3013 | if (alloc_start > inode->i_size) { |
a41ad394 JB |
3014 | ret = btrfs_cont_expand(inode, i_size_read(inode), |
3015 | alloc_start); | |
2fe17c10 CH |
3016 | if (ret) |
3017 | goto out; | |
0f6925fa | 3018 | } else if (offset + len > inode->i_size) { |
a71754fc JB |
3019 | /* |
3020 | * If we are fallocating from the end of the file onward we | |
9703fefe CR |
3021 | * need to zero out the end of the block if i_size lands in the |
3022 | * middle of a block. | |
a71754fc | 3023 | */ |
9703fefe | 3024 | ret = btrfs_truncate_block(inode, inode->i_size, 0, 0); |
a71754fc JB |
3025 | if (ret) |
3026 | goto out; | |
2fe17c10 CH |
3027 | } |
3028 | ||
a71754fc JB |
3029 | /* |
3030 | * wait for ordered IO before we have any locks. We'll loop again | |
3031 | * below with the locks held. | |
3032 | */ | |
0ef8b726 JB |
3033 | ret = btrfs_wait_ordered_range(inode, alloc_start, |
3034 | alloc_end - alloc_start); | |
3035 | if (ret) | |
3036 | goto out; | |
a71754fc | 3037 | |
f27451f2 FM |
3038 | if (mode & FALLOC_FL_ZERO_RANGE) { |
3039 | ret = btrfs_zero_range(inode, offset, len, mode); | |
3040 | inode_unlock(inode); | |
3041 | return ret; | |
3042 | } | |
3043 | ||
2fe17c10 CH |
3044 | locked_end = alloc_end - 1; |
3045 | while (1) { | |
3046 | struct btrfs_ordered_extent *ordered; | |
3047 | ||
3048 | /* the extent lock is ordered inside the running | |
3049 | * transaction | |
3050 | */ | |
3051 | lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, | |
ff13db41 | 3052 | locked_end, &cached_state); |
96b09dde NB |
3053 | ordered = btrfs_lookup_first_ordered_extent(inode, locked_end); |
3054 | ||
2fe17c10 CH |
3055 | if (ordered && |
3056 | ordered->file_offset + ordered->len > alloc_start && | |
3057 | ordered->file_offset < alloc_end) { | |
3058 | btrfs_put_ordered_extent(ordered); | |
3059 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
3060 | alloc_start, locked_end, | |
e43bbe5e | 3061 | &cached_state); |
2fe17c10 CH |
3062 | /* |
3063 | * we can't wait on the range with the transaction | |
3064 | * running or with the extent lock held | |
3065 | */ | |
0ef8b726 JB |
3066 | ret = btrfs_wait_ordered_range(inode, alloc_start, |
3067 | alloc_end - alloc_start); | |
3068 | if (ret) | |
3069 | goto out; | |
2fe17c10 CH |
3070 | } else { |
3071 | if (ordered) | |
3072 | btrfs_put_ordered_extent(ordered); | |
3073 | break; | |
3074 | } | |
3075 | } | |
3076 | ||
14524a84 QW |
3077 | /* First, check if we exceed the qgroup limit */ |
3078 | INIT_LIST_HEAD(&reserve_list); | |
6b7d6e93 | 3079 | while (cur_offset < alloc_end) { |
fc4f21b1 | 3080 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, cur_offset, |
2fe17c10 | 3081 | alloc_end - cur_offset, 0); |
9986277e DC |
3082 | if (IS_ERR(em)) { |
3083 | ret = PTR_ERR(em); | |
79787eaa JM |
3084 | break; |
3085 | } | |
2fe17c10 | 3086 | last_byte = min(extent_map_end(em), alloc_end); |
f1e490a7 | 3087 | actual_end = min_t(u64, extent_map_end(em), offset + len); |
797f4277 | 3088 | last_byte = ALIGN(last_byte, blocksize); |
2fe17c10 CH |
3089 | if (em->block_start == EXTENT_MAP_HOLE || |
3090 | (cur_offset >= inode->i_size && | |
3091 | !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { | |
14524a84 QW |
3092 | ret = add_falloc_range(&reserve_list, cur_offset, |
3093 | last_byte - cur_offset); | |
3094 | if (ret < 0) { | |
3095 | free_extent_map(em); | |
3096 | break; | |
3d850dd4 | 3097 | } |
364ecf36 QW |
3098 | ret = btrfs_qgroup_reserve_data(inode, &data_reserved, |
3099 | cur_offset, last_byte - cur_offset); | |
be2d253c FM |
3100 | if (ret < 0) { |
3101 | free_extent_map(em); | |
14524a84 | 3102 | break; |
be2d253c | 3103 | } |
18513091 WX |
3104 | } else { |
3105 | /* | |
3106 | * Do not need to reserve unwritten extent for this | |
3107 | * range, free reserved data space first, otherwise | |
3108 | * it'll result in false ENOSPC error. | |
3109 | */ | |
bc42bda2 QW |
3110 | btrfs_free_reserved_data_space(inode, data_reserved, |
3111 | cur_offset, last_byte - cur_offset); | |
2fe17c10 CH |
3112 | } |
3113 | free_extent_map(em); | |
2fe17c10 | 3114 | cur_offset = last_byte; |
14524a84 QW |
3115 | } |
3116 | ||
3117 | /* | |
3118 | * If ret is still 0, means we're OK to fallocate. | |
3119 | * Or just cleanup the list and exit. | |
3120 | */ | |
3121 | list_for_each_entry_safe(range, tmp, &reserve_list, list) { | |
3122 | if (!ret) | |
3123 | ret = btrfs_prealloc_file_range(inode, mode, | |
3124 | range->start, | |
93407472 | 3125 | range->len, i_blocksize(inode), |
14524a84 | 3126 | offset + len, &alloc_hint); |
18513091 | 3127 | else |
bc42bda2 QW |
3128 | btrfs_free_reserved_data_space(inode, |
3129 | data_reserved, range->start, | |
3130 | range->len); | |
14524a84 QW |
3131 | list_del(&range->list); |
3132 | kfree(range); | |
3133 | } | |
3134 | if (ret < 0) | |
3135 | goto out_unlock; | |
3136 | ||
f27451f2 FM |
3137 | /* |
3138 | * We didn't need to allocate any more space, but we still extended the | |
3139 | * size of the file so we need to update i_size and the inode item. | |
3140 | */ | |
3141 | ret = btrfs_fallocate_update_isize(inode, actual_end, mode); | |
14524a84 | 3142 | out_unlock: |
2fe17c10 | 3143 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, |
e43bbe5e | 3144 | &cached_state); |
2fe17c10 | 3145 | out: |
5955102c | 3146 | inode_unlock(inode); |
d98456fc | 3147 | /* Let go of our reservation. */ |
f27451f2 | 3148 | if (ret != 0 && !(mode & FALLOC_FL_ZERO_RANGE)) |
bc42bda2 QW |
3149 | btrfs_free_reserved_data_space(inode, data_reserved, |
3150 | alloc_start, alloc_end - cur_offset); | |
364ecf36 | 3151 | extent_changeset_free(data_reserved); |
2fe17c10 CH |
3152 | return ret; |
3153 | } | |
3154 | ||
965c8e59 | 3155 | static int find_desired_extent(struct inode *inode, loff_t *offset, int whence) |
b2675157 | 3156 | { |
0b246afa | 3157 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
7f4ca37c | 3158 | struct extent_map *em = NULL; |
b2675157 | 3159 | struct extent_state *cached_state = NULL; |
4d1a40c6 LB |
3160 | u64 lockstart; |
3161 | u64 lockend; | |
3162 | u64 start; | |
3163 | u64 len; | |
b2675157 JB |
3164 | int ret = 0; |
3165 | ||
4d1a40c6 LB |
3166 | if (inode->i_size == 0) |
3167 | return -ENXIO; | |
3168 | ||
3169 | /* | |
3170 | * *offset can be negative, in this case we start finding DATA/HOLE from | |
3171 | * the very start of the file. | |
3172 | */ | |
3173 | start = max_t(loff_t, 0, *offset); | |
3174 | ||
0b246afa | 3175 | lockstart = round_down(start, fs_info->sectorsize); |
da17066c | 3176 | lockend = round_up(i_size_read(inode), |
0b246afa | 3177 | fs_info->sectorsize); |
b2675157 | 3178 | if (lockend <= lockstart) |
0b246afa | 3179 | lockend = lockstart + fs_info->sectorsize; |
1214b53f | 3180 | lockend--; |
b2675157 JB |
3181 | len = lockend - lockstart + 1; |
3182 | ||
ff13db41 | 3183 | lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, |
d0082371 | 3184 | &cached_state); |
b2675157 | 3185 | |
7f4ca37c | 3186 | while (start < inode->i_size) { |
fc4f21b1 NB |
3187 | em = btrfs_get_extent_fiemap(BTRFS_I(inode), NULL, 0, |
3188 | start, len, 0); | |
b2675157 | 3189 | if (IS_ERR(em)) { |
6af021d8 | 3190 | ret = PTR_ERR(em); |
7f4ca37c | 3191 | em = NULL; |
b2675157 JB |
3192 | break; |
3193 | } | |
3194 | ||
7f4ca37c JB |
3195 | if (whence == SEEK_HOLE && |
3196 | (em->block_start == EXTENT_MAP_HOLE || | |
3197 | test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) | |
3198 | break; | |
3199 | else if (whence == SEEK_DATA && | |
3200 | (em->block_start != EXTENT_MAP_HOLE && | |
3201 | !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) | |
3202 | break; | |
b2675157 JB |
3203 | |
3204 | start = em->start + em->len; | |
b2675157 | 3205 | free_extent_map(em); |
7f4ca37c | 3206 | em = NULL; |
b2675157 JB |
3207 | cond_resched(); |
3208 | } | |
7f4ca37c JB |
3209 | free_extent_map(em); |
3210 | if (!ret) { | |
3211 | if (whence == SEEK_DATA && start >= inode->i_size) | |
3212 | ret = -ENXIO; | |
3213 | else | |
3214 | *offset = min_t(loff_t, start, inode->i_size); | |
3215 | } | |
b2675157 | 3216 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, |
e43bbe5e | 3217 | &cached_state); |
b2675157 JB |
3218 | return ret; |
3219 | } | |
3220 | ||
965c8e59 | 3221 | static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence) |
b2675157 JB |
3222 | { |
3223 | struct inode *inode = file->f_mapping->host; | |
3224 | int ret; | |
3225 | ||
5955102c | 3226 | inode_lock(inode); |
965c8e59 | 3227 | switch (whence) { |
b2675157 JB |
3228 | case SEEK_END: |
3229 | case SEEK_CUR: | |
965c8e59 | 3230 | offset = generic_file_llseek(file, offset, whence); |
b2675157 JB |
3231 | goto out; |
3232 | case SEEK_DATA: | |
3233 | case SEEK_HOLE: | |
48802c8a | 3234 | if (offset >= i_size_read(inode)) { |
5955102c | 3235 | inode_unlock(inode); |
48802c8a JL |
3236 | return -ENXIO; |
3237 | } | |
3238 | ||
965c8e59 | 3239 | ret = find_desired_extent(inode, &offset, whence); |
b2675157 | 3240 | if (ret) { |
5955102c | 3241 | inode_unlock(inode); |
b2675157 JB |
3242 | return ret; |
3243 | } | |
3244 | } | |
3245 | ||
46a1c2c7 | 3246 | offset = vfs_setpos(file, offset, inode->i_sb->s_maxbytes); |
b2675157 | 3247 | out: |
5955102c | 3248 | inode_unlock(inode); |
b2675157 JB |
3249 | return offset; |
3250 | } | |
3251 | ||
edf064e7 GR |
3252 | static int btrfs_file_open(struct inode *inode, struct file *filp) |
3253 | { | |
91f9943e | 3254 | filp->f_mode |= FMODE_NOWAIT; |
edf064e7 GR |
3255 | return generic_file_open(inode, filp); |
3256 | } | |
3257 | ||
828c0950 | 3258 | const struct file_operations btrfs_file_operations = { |
b2675157 | 3259 | .llseek = btrfs_file_llseek, |
aad4f8bb | 3260 | .read_iter = generic_file_read_iter, |
e9906a98 | 3261 | .splice_read = generic_file_splice_read, |
b30ac0fc | 3262 | .write_iter = btrfs_file_write_iter, |
9ebefb18 | 3263 | .mmap = btrfs_file_mmap, |
edf064e7 | 3264 | .open = btrfs_file_open, |
e1b81e67 | 3265 | .release = btrfs_release_file, |
39279cc3 | 3266 | .fsync = btrfs_sync_file, |
2fe17c10 | 3267 | .fallocate = btrfs_fallocate, |
34287aa3 | 3268 | .unlocked_ioctl = btrfs_ioctl, |
39279cc3 | 3269 | #ifdef CONFIG_COMPAT |
4c63c245 | 3270 | .compat_ioctl = btrfs_compat_ioctl, |
39279cc3 | 3271 | #endif |
04b38d60 | 3272 | .clone_file_range = btrfs_clone_file_range, |
2b3909f8 | 3273 | .dedupe_file_range = btrfs_dedupe_file_range, |
39279cc3 | 3274 | }; |
9247f317 | 3275 | |
e67c718b | 3276 | void __cold btrfs_auto_defrag_exit(void) |
9247f317 | 3277 | { |
5598e900 | 3278 | kmem_cache_destroy(btrfs_inode_defrag_cachep); |
9247f317 MX |
3279 | } |
3280 | ||
f5c29bd9 | 3281 | int __init btrfs_auto_defrag_init(void) |
9247f317 MX |
3282 | { |
3283 | btrfs_inode_defrag_cachep = kmem_cache_create("btrfs_inode_defrag", | |
3284 | sizeof(struct inode_defrag), 0, | |
fba4b697 | 3285 | SLAB_MEM_SPREAD, |
9247f317 MX |
3286 | NULL); |
3287 | if (!btrfs_inode_defrag_cachep) | |
3288 | return -ENOMEM; | |
3289 | ||
3290 | return 0; | |
3291 | } | |
728404da FM |
3292 | |
3293 | int btrfs_fdatawrite_range(struct inode *inode, loff_t start, loff_t end) | |
3294 | { | |
3295 | int ret; | |
3296 | ||
3297 | /* | |
3298 | * So with compression we will find and lock a dirty page and clear the | |
3299 | * first one as dirty, setup an async extent, and immediately return | |
3300 | * with the entire range locked but with nobody actually marked with | |
3301 | * writeback. So we can't just filemap_write_and_wait_range() and | |
3302 | * expect it to work since it will just kick off a thread to do the | |
3303 | * actual work. So we need to call filemap_fdatawrite_range _again_ | |
3304 | * since it will wait on the page lock, which won't be unlocked until | |
3305 | * after the pages have been marked as writeback and so we're good to go | |
3306 | * from there. We have to do this otherwise we'll miss the ordered | |
3307 | * extents and that results in badness. Please Josef, do not think you | |
3308 | * know better and pull this out at some point in the future, it is | |
3309 | * right and you are wrong. | |
3310 | */ | |
3311 | ret = filemap_fdatawrite_range(inode->i_mapping, start, end); | |
3312 | if (!ret && test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, | |
3313 | &BTRFS_I(inode)->runtime_flags)) | |
3314 | ret = filemap_fdatawrite_range(inode->i_mapping, start, end); | |
3315 | ||
3316 | return ret; | |
3317 | } |