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 { | |
57 | struct ocfs2_extent_tree_operations *et_ops; | |
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 JB |
70 | void ocfs2_init_dinode_extent_tree(struct ocfs2_extent_tree *et, |
71 | struct inode *inode, | |
72 | struct buffer_head *bh); | |
73 | void ocfs2_init_xattr_tree_extent_tree(struct ocfs2_extent_tree *et, | |
f99b9b7c | 74 | struct inode *inode, |
8d6220d6 | 75 | struct buffer_head *bh); |
2a50a743 | 76 | struct ocfs2_xattr_value_buf; |
8d6220d6 JB |
77 | void ocfs2_init_xattr_value_extent_tree(struct ocfs2_extent_tree *et, |
78 | struct inode *inode, | |
2a50a743 | 79 | struct ocfs2_xattr_value_buf *vb); |
9b7895ef MF |
80 | void ocfs2_init_dx_root_extent_tree(struct ocfs2_extent_tree *et, |
81 | struct inode *inode, | |
82 | struct buffer_head *bh); | |
f99b9b7c | 83 | |
5e96581a JB |
84 | /* |
85 | * Read an extent block into *bh. If *bh is NULL, a bh will be | |
86 | * allocated. This is a cached read. The extent block will be validated | |
87 | * with ocfs2_validate_extent_block(). | |
88 | */ | |
89 | int ocfs2_read_extent_block(struct inode *inode, u64 eb_blkno, | |
90 | struct buffer_head **bh); | |
91 | ||
ccd979bd | 92 | struct ocfs2_alloc_context; |
f99b9b7c JB |
93 | int ocfs2_insert_extent(struct ocfs2_super *osb, |
94 | handle_t *handle, | |
95 | struct inode *inode, | |
96 | struct ocfs2_extent_tree *et, | |
97 | u32 cpos, | |
98 | u64 start_blk, | |
99 | u32 new_clusters, | |
100 | u8 flags, | |
101 | struct ocfs2_alloc_context *meta_ac); | |
102 | ||
0eb8d47e TM |
103 | enum ocfs2_alloc_restarted { |
104 | RESTART_NONE = 0, | |
105 | RESTART_TRANS, | |
106 | RESTART_META | |
107 | }; | |
108 | int ocfs2_add_clusters_in_btree(struct ocfs2_super *osb, | |
109 | struct inode *inode, | |
110 | u32 *logical_offset, | |
111 | u32 clusters_to_add, | |
112 | int mark_unwritten, | |
f99b9b7c | 113 | struct ocfs2_extent_tree *et, |
0eb8d47e TM |
114 | handle_t *handle, |
115 | struct ocfs2_alloc_context *data_ac, | |
116 | struct ocfs2_alloc_context *meta_ac, | |
f99b9b7c | 117 | enum ocfs2_alloc_restarted *reason_ret); |
328d5752 | 118 | struct ocfs2_cached_dealloc_ctxt; |
f99b9b7c JB |
119 | int ocfs2_mark_extent_written(struct inode *inode, |
120 | struct ocfs2_extent_tree *et, | |
328d5752 MF |
121 | handle_t *handle, u32 cpos, u32 len, u32 phys, |
122 | struct ocfs2_alloc_context *meta_ac, | |
f99b9b7c JB |
123 | struct ocfs2_cached_dealloc_ctxt *dealloc); |
124 | int ocfs2_remove_extent(struct inode *inode, | |
125 | struct ocfs2_extent_tree *et, | |
063c4561 MF |
126 | u32 cpos, u32 len, handle_t *handle, |
127 | struct ocfs2_alloc_context *meta_ac, | |
f99b9b7c | 128 | struct ocfs2_cached_dealloc_ctxt *dealloc); |
fecc0112 MF |
129 | int ocfs2_remove_btree_range(struct inode *inode, |
130 | struct ocfs2_extent_tree *et, | |
131 | u32 cpos, u32 phys_cpos, u32 len, | |
132 | struct ocfs2_cached_dealloc_ctxt *dealloc); | |
133 | ||
ccd979bd MF |
134 | int ocfs2_num_free_extents(struct ocfs2_super *osb, |
135 | struct inode *inode, | |
f99b9b7c | 136 | struct ocfs2_extent_tree *et); |
e7d4cb6b | 137 | |
811f933d TM |
138 | /* |
139 | * how many new metadata chunks would an allocation need at maximum? | |
140 | * | |
141 | * Please note that the caller must make sure that root_el is the root | |
142 | * of extent tree. So for an inode, it should be &fe->id2.i_list. Otherwise | |
143 | * the result may be wrong. | |
144 | */ | |
145 | static inline int ocfs2_extend_meta_needed(struct ocfs2_extent_list *root_el) | |
ccd979bd MF |
146 | { |
147 | /* | |
148 | * Rather than do all the work of determining how much we need | |
149 | * (involves a ton of reads and locks), just ask for the | |
150 | * maximal limit. That's a tree depth shift. So, one block for | |
151 | * level of the tree (current l_tree_depth), one block for the | |
152 | * new tree_depth==0 extent_block, and one block at the new | |
153 | * top-of-the tree. | |
154 | */ | |
811f933d | 155 | return le16_to_cpu(root_el->l_tree_depth) + 2; |
ccd979bd MF |
156 | } |
157 | ||
5b6a3a2b | 158 | void ocfs2_dinode_new_extent_list(struct inode *inode, struct ocfs2_dinode *di); |
1afc32b9 MF |
159 | void ocfs2_set_inode_data_inline(struct inode *inode, struct ocfs2_dinode *di); |
160 | int ocfs2_convert_inline_data_to_extents(struct inode *inode, | |
161 | struct buffer_head *di_bh); | |
162 | ||
ccd979bd MF |
163 | int ocfs2_truncate_log_init(struct ocfs2_super *osb); |
164 | void ocfs2_truncate_log_shutdown(struct ocfs2_super *osb); | |
165 | void ocfs2_schedule_truncate_log_flush(struct ocfs2_super *osb, | |
166 | int cancel); | |
167 | int ocfs2_flush_truncate_log(struct ocfs2_super *osb); | |
168 | int ocfs2_begin_truncate_log_recovery(struct ocfs2_super *osb, | |
169 | int slot_num, | |
170 | struct ocfs2_dinode **tl_copy); | |
171 | int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb, | |
172 | struct ocfs2_dinode *tl_copy); | |
063c4561 MF |
173 | int ocfs2_truncate_log_needs_flush(struct ocfs2_super *osb); |
174 | int ocfs2_truncate_log_append(struct ocfs2_super *osb, | |
175 | handle_t *handle, | |
176 | u64 start_blk, | |
177 | unsigned int num_clusters); | |
178 | int __ocfs2_flush_truncate_log(struct ocfs2_super *osb); | |
ccd979bd | 179 | |
2b604351 MF |
180 | /* |
181 | * Process local structure which describes the block unlinks done | |
182 | * during an operation. This is populated via | |
183 | * ocfs2_cache_block_dealloc(). | |
184 | * | |
185 | * ocfs2_run_deallocs() should be called after the potentially | |
186 | * de-allocating routines. No journal handles should be open, and most | |
187 | * locks should have been dropped. | |
188 | */ | |
189 | struct ocfs2_cached_dealloc_ctxt { | |
190 | struct ocfs2_per_slot_free_list *c_first_suballocator; | |
2891d290 | 191 | struct ocfs2_cached_block_free *c_global_allocator; |
2b604351 MF |
192 | }; |
193 | static inline void ocfs2_init_dealloc_ctxt(struct ocfs2_cached_dealloc_ctxt *c) | |
194 | { | |
195 | c->c_first_suballocator = NULL; | |
2891d290 | 196 | c->c_global_allocator = NULL; |
2b604351 | 197 | } |
2891d290 TM |
198 | int ocfs2_cache_cluster_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt, |
199 | u64 blkno, unsigned int bit); | |
78f30c31 TM |
200 | static inline int ocfs2_dealloc_has_cluster(struct ocfs2_cached_dealloc_ctxt *c) |
201 | { | |
202 | return c->c_global_allocator != NULL; | |
203 | } | |
2b604351 MF |
204 | int ocfs2_run_deallocs(struct ocfs2_super *osb, |
205 | struct ocfs2_cached_dealloc_ctxt *ctxt); | |
206 | ||
ccd979bd | 207 | struct ocfs2_truncate_context { |
59a5e416 | 208 | struct ocfs2_cached_dealloc_ctxt tc_dealloc; |
ccd979bd MF |
209 | int tc_ext_alloc_locked; /* is it cluster locked? */ |
210 | /* these get destroyed once it's passed to ocfs2_commit_truncate. */ | |
211 | struct buffer_head *tc_last_eb_bh; | |
212 | }; | |
213 | ||
35edec1d MF |
214 | int ocfs2_zero_range_for_truncate(struct inode *inode, handle_t *handle, |
215 | u64 range_start, u64 range_end); | |
ccd979bd MF |
216 | int ocfs2_prepare_truncate(struct ocfs2_super *osb, |
217 | struct inode *inode, | |
218 | struct buffer_head *fe_bh, | |
219 | struct ocfs2_truncate_context **tc); | |
220 | int ocfs2_commit_truncate(struct ocfs2_super *osb, | |
221 | struct inode *inode, | |
222 | struct buffer_head *fe_bh, | |
223 | struct ocfs2_truncate_context *tc); | |
1afc32b9 MF |
224 | int ocfs2_truncate_inline(struct inode *inode, struct buffer_head *di_bh, |
225 | unsigned int start, unsigned int end, int trunc); | |
ccd979bd | 226 | |
363041a5 MF |
227 | int ocfs2_find_leaf(struct inode *inode, struct ocfs2_extent_list *root_el, |
228 | u32 cpos, struct buffer_head **leaf_bh); | |
328d5752 | 229 | int ocfs2_search_extent_list(struct ocfs2_extent_list *el, u32 v_cluster); |
363041a5 | 230 | |
e48edee2 MF |
231 | /* |
232 | * Helper function to look at the # of clusters in an extent record. | |
233 | */ | |
234 | static inline unsigned int ocfs2_rec_clusters(struct ocfs2_extent_list *el, | |
235 | struct ocfs2_extent_rec *rec) | |
236 | { | |
237 | /* | |
238 | * Cluster count in extent records is slightly different | |
239 | * between interior nodes and leaf nodes. This is to support | |
240 | * unwritten extents which need a flags field in leaf node | |
241 | * records, thus shrinking the available space for a clusters | |
242 | * field. | |
243 | */ | |
244 | if (el->l_tree_depth) | |
245 | return le32_to_cpu(rec->e_int_clusters); | |
246 | else | |
247 | return le16_to_cpu(rec->e_leaf_clusters); | |
248 | } | |
249 | ||
00dc417f MF |
250 | /* |
251 | * This is only valid for leaf nodes, which are the only ones that can | |
252 | * have empty extents anyway. | |
253 | */ | |
254 | static inline int ocfs2_is_empty_extent(struct ocfs2_extent_rec *rec) | |
255 | { | |
256 | return !rec->e_leaf_clusters; | |
257 | } | |
258 | ||
ccd979bd | 259 | #endif /* OCFS2_ALLOC_H */ |