]>
Commit | Line | Data |
---|---|---|
ccd979bd MF |
1 | /* -*- mode: c; c-basic-offset: 8; -*- |
2 | * vim: noexpandtab sw=8 ts=8 sts=0: | |
3 | * | |
4 | * alloc.h | |
5 | * | |
6 | * Function prototypes | |
7 | * | |
8 | * Copyright (C) 2002, 2004 Oracle. All rights reserved. | |
9 | * | |
10 | * This program is free software; you can redistribute it and/or | |
11 | * modify it under the terms of the GNU General Public | |
12 | * License as published by the Free Software Foundation; either | |
13 | * version 2 of the License, or (at your option) any later version. | |
14 | * | |
15 | * This program is distributed in the hope that it will be useful, | |
16 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
18 | * General Public License for more details. | |
19 | * | |
20 | * You should have received a copy of the GNU General Public | |
21 | * License along with this program; if not, write to the | |
22 | * Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
23 | * Boston, MA 021110-1307, USA. | |
24 | */ | |
25 | ||
26 | #ifndef OCFS2_ALLOC_H | |
27 | #define OCFS2_ALLOC_H | |
28 | ||
e7d4cb6b | 29 | |
ca12b7c4 TM |
30 | /* |
31 | * For xattr tree leaf, we limit the leaf byte size to be 64K. | |
32 | */ | |
33 | #define OCFS2_MAX_XATTR_TREE_LEAF_SIZE 65536 | |
34 | ||
f99b9b7c JB |
35 | /* |
36 | * ocfs2_extent_tree and ocfs2_extent_tree_operations are used to abstract | |
37 | * the b-tree operations in ocfs2. Now all the b-tree operations are not | |
38 | * limited to ocfs2_dinode only. Any data which need to allocate clusters | |
39 | * to store can use b-tree. And it only needs to implement its ocfs2_extent_tree | |
40 | * and operation. | |
41 | * | |
42 | * ocfs2_extent_tree becomes the first-class object for extent tree | |
43 | * manipulation. Callers of the alloc.c code need to fill it via one of | |
8d6220d6 | 44 | * the ocfs2_init_*_extent_tree() operations below. |
f99b9b7c JB |
45 | * |
46 | * ocfs2_extent_tree contains info for the root of the b-tree, it must have a | |
47 | * root ocfs2_extent_list and a root_bh so that they can be used in the b-tree | |
d9a0a1f8 JB |
48 | * functions. It needs the ocfs2_caching_info structure associated with |
49 | * I/O on the tree. With metadata ecc, we now call different journal_access | |
13723d00 JB |
50 | * functions for each type of metadata, so it must have the |
51 | * root_journal_access function. | |
f99b9b7c JB |
52 | * ocfs2_extent_tree_operations abstract the normal operations we do for |
53 | * the root of extent b-tree. | |
54 | */ | |
55 | struct ocfs2_extent_tree_operations; | |
56 | struct ocfs2_extent_tree { | |
9e62dc09 | 57 | const struct ocfs2_extent_tree_operations *et_ops; |
f99b9b7c JB |
58 | struct buffer_head *et_root_bh; |
59 | struct ocfs2_extent_list *et_root_el; | |
d9a0a1f8 | 60 | struct ocfs2_caching_info *et_ci; |
13723d00 | 61 | ocfs2_journal_access_func et_root_journal_access; |
f99b9b7c JB |
62 | void *et_object; |
63 | unsigned int et_max_leaf_clusters; | |
64 | }; | |
65 | ||
66 | /* | |
8d6220d6 JB |
67 | * ocfs2_init_*_extent_tree() will fill an ocfs2_extent_tree from the |
68 | * specified object buffer. | |
f99b9b7c | 69 | */ |
8d6220d6 | 70 | void ocfs2_init_dinode_extent_tree(struct ocfs2_extent_tree *et, |
5e404e9e | 71 | struct ocfs2_caching_info *ci, |
8d6220d6 JB |
72 | struct buffer_head *bh); |
73 | void ocfs2_init_xattr_tree_extent_tree(struct ocfs2_extent_tree *et, | |
5e404e9e | 74 | struct ocfs2_caching_info *ci, |
8d6220d6 | 75 | struct buffer_head *bh); |
2a50a743 | 76 | struct ocfs2_xattr_value_buf; |
8d6220d6 | 77 | void ocfs2_init_xattr_value_extent_tree(struct ocfs2_extent_tree *et, |
5e404e9e | 78 | struct ocfs2_caching_info *ci, |
2a50a743 | 79 | struct ocfs2_xattr_value_buf *vb); |
9b7895ef | 80 | void ocfs2_init_dx_root_extent_tree(struct ocfs2_extent_tree *et, |
5e404e9e | 81 | struct ocfs2_caching_info *ci, |
9b7895ef | 82 | struct buffer_head *bh); |
fe924415 TM |
83 | void ocfs2_init_refcount_extent_tree(struct ocfs2_extent_tree *et, |
84 | struct ocfs2_caching_info *ci, | |
85 | struct buffer_head *bh); | |
f99b9b7c | 86 | |
5e96581a JB |
87 | /* |
88 | * Read an extent block into *bh. If *bh is NULL, a bh will be | |
89 | * allocated. This is a cached read. The extent block will be validated | |
90 | * with ocfs2_validate_extent_block(). | |
91 | */ | |
3d03a305 | 92 | int ocfs2_read_extent_block(struct ocfs2_caching_info *ci, u64 eb_blkno, |
5e96581a JB |
93 | struct buffer_head **bh); |
94 | ||
ccd979bd | 95 | struct ocfs2_alloc_context; |
cc79d8c1 | 96 | int ocfs2_insert_extent(handle_t *handle, |
f99b9b7c JB |
97 | struct ocfs2_extent_tree *et, |
98 | u32 cpos, | |
99 | u64 start_blk, | |
100 | u32 new_clusters, | |
101 | u8 flags, | |
102 | struct ocfs2_alloc_context *meta_ac); | |
103 | ||
0eb8d47e TM |
104 | enum ocfs2_alloc_restarted { |
105 | RESTART_NONE = 0, | |
106 | RESTART_TRANS, | |
107 | RESTART_META | |
108 | }; | |
cbee7e1a JB |
109 | int ocfs2_add_clusters_in_btree(handle_t *handle, |
110 | struct ocfs2_extent_tree *et, | |
0eb8d47e TM |
111 | u32 *logical_offset, |
112 | u32 clusters_to_add, | |
113 | int mark_unwritten, | |
0eb8d47e TM |
114 | struct ocfs2_alloc_context *data_ac, |
115 | struct ocfs2_alloc_context *meta_ac, | |
f99b9b7c | 116 | enum ocfs2_alloc_restarted *reason_ret); |
328d5752 | 117 | struct ocfs2_cached_dealloc_ctxt; |
e2e9f608 TM |
118 | struct ocfs2_path; |
119 | int ocfs2_split_extent(handle_t *handle, | |
120 | struct ocfs2_extent_tree *et, | |
121 | struct ocfs2_path *path, | |
122 | int split_index, | |
123 | struct ocfs2_extent_rec *split_rec, | |
124 | struct ocfs2_alloc_context *meta_ac, | |
125 | struct ocfs2_cached_dealloc_ctxt *dealloc); | |
f99b9b7c JB |
126 | int ocfs2_mark_extent_written(struct inode *inode, |
127 | struct ocfs2_extent_tree *et, | |
328d5752 MF |
128 | handle_t *handle, u32 cpos, u32 len, u32 phys, |
129 | struct ocfs2_alloc_context *meta_ac, | |
f99b9b7c | 130 | struct ocfs2_cached_dealloc_ctxt *dealloc); |
1aa75fea TM |
131 | int ocfs2_change_extent_flag(handle_t *handle, |
132 | struct ocfs2_extent_tree *et, | |
133 | u32 cpos, u32 len, u32 phys, | |
134 | struct ocfs2_alloc_context *meta_ac, | |
135 | struct ocfs2_cached_dealloc_ctxt *dealloc, | |
136 | int new_flags, int clear_flags); | |
dbdcf6a4 JB |
137 | int ocfs2_remove_extent(handle_t *handle, struct ocfs2_extent_tree *et, |
138 | u32 cpos, u32 len, | |
063c4561 | 139 | struct ocfs2_alloc_context *meta_ac, |
f99b9b7c | 140 | struct ocfs2_cached_dealloc_ctxt *dealloc); |
fecc0112 MF |
141 | int ocfs2_remove_btree_range(struct inode *inode, |
142 | struct ocfs2_extent_tree *et, | |
78f94673 TY |
143 | u32 cpos, u32 phys_cpos, u32 len, int flags, |
144 | struct ocfs2_cached_dealloc_ctxt *dealloc, | |
f62f12b3 | 145 | u64 refcount_loc, bool refcount_tree_locked); |
fecc0112 | 146 | |
964f14a0 | 147 | int ocfs2_num_free_extents(struct ocfs2_extent_tree *et); |
e7d4cb6b | 148 | |
811f933d TM |
149 | /* |
150 | * how many new metadata chunks would an allocation need at maximum? | |
151 | * | |
152 | * Please note that the caller must make sure that root_el is the root | |
153 | * of extent tree. So for an inode, it should be &fe->id2.i_list. Otherwise | |
154 | * the result may be wrong. | |
155 | */ | |
156 | static inline int ocfs2_extend_meta_needed(struct ocfs2_extent_list *root_el) | |
ccd979bd MF |
157 | { |
158 | /* | |
159 | * Rather than do all the work of determining how much we need | |
160 | * (involves a ton of reads and locks), just ask for the | |
161 | * maximal limit. That's a tree depth shift. So, one block for | |
162 | * level of the tree (current l_tree_depth), one block for the | |
163 | * new tree_depth==0 extent_block, and one block at the new | |
164 | * top-of-the tree. | |
165 | */ | |
811f933d | 166 | return le16_to_cpu(root_el->l_tree_depth) + 2; |
ccd979bd MF |
167 | } |
168 | ||
5b6a3a2b | 169 | void ocfs2_dinode_new_extent_list(struct inode *inode, struct ocfs2_dinode *di); |
1afc32b9 MF |
170 | void ocfs2_set_inode_data_inline(struct inode *inode, struct ocfs2_dinode *di); |
171 | int ocfs2_convert_inline_data_to_extents(struct inode *inode, | |
172 | struct buffer_head *di_bh); | |
173 | ||
ccd979bd MF |
174 | int ocfs2_truncate_log_init(struct ocfs2_super *osb); |
175 | void ocfs2_truncate_log_shutdown(struct ocfs2_super *osb); | |
176 | void ocfs2_schedule_truncate_log_flush(struct ocfs2_super *osb, | |
177 | int cancel); | |
178 | int ocfs2_flush_truncate_log(struct ocfs2_super *osb); | |
179 | int ocfs2_begin_truncate_log_recovery(struct ocfs2_super *osb, | |
180 | int slot_num, | |
181 | struct ocfs2_dinode **tl_copy); | |
182 | int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb, | |
183 | struct ocfs2_dinode *tl_copy); | |
063c4561 MF |
184 | int ocfs2_truncate_log_needs_flush(struct ocfs2_super *osb); |
185 | int ocfs2_truncate_log_append(struct ocfs2_super *osb, | |
186 | handle_t *handle, | |
187 | u64 start_blk, | |
188 | unsigned int num_clusters); | |
189 | int __ocfs2_flush_truncate_log(struct ocfs2_super *osb); | |
2070ad1a ER |
190 | int ocfs2_try_to_free_truncate_log(struct ocfs2_super *osb, |
191 | unsigned int needed); | |
ccd979bd | 192 | |
2b604351 MF |
193 | /* |
194 | * Process local structure which describes the block unlinks done | |
195 | * during an operation. This is populated via | |
196 | * ocfs2_cache_block_dealloc(). | |
197 | * | |
198 | * ocfs2_run_deallocs() should be called after the potentially | |
199 | * de-allocating routines. No journal handles should be open, and most | |
200 | * locks should have been dropped. | |
201 | */ | |
202 | struct ocfs2_cached_dealloc_ctxt { | |
203 | struct ocfs2_per_slot_free_list *c_first_suballocator; | |
2891d290 | 204 | struct ocfs2_cached_block_free *c_global_allocator; |
2b604351 MF |
205 | }; |
206 | static inline void ocfs2_init_dealloc_ctxt(struct ocfs2_cached_dealloc_ctxt *c) | |
207 | { | |
208 | c->c_first_suballocator = NULL; | |
2891d290 | 209 | c->c_global_allocator = NULL; |
2b604351 | 210 | } |
2891d290 TM |
211 | int ocfs2_cache_cluster_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt, |
212 | u64 blkno, unsigned int bit); | |
1823cb0b | 213 | int ocfs2_cache_block_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt, |
74380c47 | 214 | int type, int slot, u64 suballoc, u64 blkno, |
1823cb0b | 215 | unsigned int bit); |
78f30c31 TM |
216 | static inline int ocfs2_dealloc_has_cluster(struct ocfs2_cached_dealloc_ctxt *c) |
217 | { | |
218 | return c->c_global_allocator != NULL; | |
219 | } | |
2b604351 MF |
220 | int ocfs2_run_deallocs(struct ocfs2_super *osb, |
221 | struct ocfs2_cached_dealloc_ctxt *ctxt); | |
222 | ||
ccd979bd | 223 | struct ocfs2_truncate_context { |
59a5e416 | 224 | struct ocfs2_cached_dealloc_ctxt tc_dealloc; |
ccd979bd MF |
225 | int tc_ext_alloc_locked; /* is it cluster locked? */ |
226 | /* these get destroyed once it's passed to ocfs2_commit_truncate. */ | |
227 | struct buffer_head *tc_last_eb_bh; | |
228 | }; | |
229 | ||
35edec1d MF |
230 | int ocfs2_zero_range_for_truncate(struct inode *inode, handle_t *handle, |
231 | u64 range_start, u64 range_end); | |
ccd979bd MF |
232 | int ocfs2_commit_truncate(struct ocfs2_super *osb, |
233 | struct inode *inode, | |
78f94673 | 234 | struct buffer_head *di_bh); |
1afc32b9 MF |
235 | int ocfs2_truncate_inline(struct inode *inode, struct buffer_head *di_bh, |
236 | unsigned int start, unsigned int end, int trunc); | |
ccd979bd | 237 | |
facdb77f JB |
238 | int ocfs2_find_leaf(struct ocfs2_caching_info *ci, |
239 | struct ocfs2_extent_list *root_el, u32 cpos, | |
240 | struct buffer_head **leaf_bh); | |
328d5752 | 241 | int ocfs2_search_extent_list(struct ocfs2_extent_list *el, u32 v_cluster); |
363041a5 | 242 | |
e80de36d | 243 | int ocfs2_trim_fs(struct super_block *sb, struct fstrim_range *range); |
e48edee2 MF |
244 | /* |
245 | * Helper function to look at the # of clusters in an extent record. | |
246 | */ | |
247 | static inline unsigned int ocfs2_rec_clusters(struct ocfs2_extent_list *el, | |
248 | struct ocfs2_extent_rec *rec) | |
249 | { | |
250 | /* | |
251 | * Cluster count in extent records is slightly different | |
252 | * between interior nodes and leaf nodes. This is to support | |
253 | * unwritten extents which need a flags field in leaf node | |
254 | * records, thus shrinking the available space for a clusters | |
255 | * field. | |
256 | */ | |
257 | if (el->l_tree_depth) | |
258 | return le32_to_cpu(rec->e_int_clusters); | |
259 | else | |
260 | return le16_to_cpu(rec->e_leaf_clusters); | |
261 | } | |
262 | ||
00dc417f MF |
263 | /* |
264 | * This is only valid for leaf nodes, which are the only ones that can | |
265 | * have empty extents anyway. | |
266 | */ | |
267 | static inline int ocfs2_is_empty_extent(struct ocfs2_extent_rec *rec) | |
268 | { | |
269 | return !rec->e_leaf_clusters; | |
270 | } | |
271 | ||
6f70fa51 TM |
272 | int ocfs2_grab_pages(struct inode *inode, loff_t start, loff_t end, |
273 | struct page **pages, int *num); | |
274 | void ocfs2_map_and_dirty_page(struct inode *inode, handle_t *handle, | |
275 | unsigned int from, unsigned int to, | |
276 | struct page *page, int zero, u64 *phys); | |
e2e9f608 TM |
277 | /* |
278 | * Structures which describe a path through a btree, and functions to | |
279 | * manipulate them. | |
280 | * | |
281 | * The idea here is to be as generic as possible with the tree | |
282 | * manipulation code. | |
283 | */ | |
284 | struct ocfs2_path_item { | |
285 | struct buffer_head *bh; | |
286 | struct ocfs2_extent_list *el; | |
287 | }; | |
288 | ||
289 | #define OCFS2_MAX_PATH_DEPTH 5 | |
290 | ||
291 | struct ocfs2_path { | |
292 | int p_tree_depth; | |
293 | ocfs2_journal_access_func p_root_access; | |
294 | struct ocfs2_path_item p_node[OCFS2_MAX_PATH_DEPTH]; | |
295 | }; | |
296 | ||
297 | #define path_root_bh(_path) ((_path)->p_node[0].bh) | |
298 | #define path_root_el(_path) ((_path)->p_node[0].el) | |
299 | #define path_root_access(_path)((_path)->p_root_access) | |
300 | #define path_leaf_bh(_path) ((_path)->p_node[(_path)->p_tree_depth].bh) | |
301 | #define path_leaf_el(_path) ((_path)->p_node[(_path)->p_tree_depth].el) | |
302 | #define path_num_items(_path) ((_path)->p_tree_depth + 1) | |
303 | ||
304 | void ocfs2_reinit_path(struct ocfs2_path *path, int keep_root); | |
305 | void ocfs2_free_path(struct ocfs2_path *path); | |
306 | int ocfs2_find_path(struct ocfs2_caching_info *ci, | |
307 | struct ocfs2_path *path, | |
308 | u32 cpos); | |
309 | struct ocfs2_path *ocfs2_new_path_from_path(struct ocfs2_path *path); | |
310 | struct ocfs2_path *ocfs2_new_path_from_et(struct ocfs2_extent_tree *et); | |
311 | int ocfs2_path_bh_journal_access(handle_t *handle, | |
312 | struct ocfs2_caching_info *ci, | |
313 | struct ocfs2_path *path, | |
314 | int idx); | |
315 | int ocfs2_journal_access_path(struct ocfs2_caching_info *ci, | |
316 | handle_t *handle, | |
317 | struct ocfs2_path *path); | |
38a04e43 TM |
318 | int ocfs2_find_cpos_for_right_leaf(struct super_block *sb, |
319 | struct ocfs2_path *path, u32 *cpos); | |
ee149a7c TY |
320 | int ocfs2_find_cpos_for_left_leaf(struct super_block *sb, |
321 | struct ocfs2_path *path, u32 *cpos); | |
38a04e43 TM |
322 | int ocfs2_find_subtree_root(struct ocfs2_extent_tree *et, |
323 | struct ocfs2_path *left, | |
324 | struct ocfs2_path *right); | |
ccd979bd | 325 | #endif /* OCFS2_ALLOC_H */ |