Commit | Line | Data |
---|---|---|
0a8165d7 | 1 | /* |
39a53e0c JK |
2 | * fs/f2fs/node.h |
3 | * | |
4 | * Copyright (c) 2012 Samsung Electronics Co., Ltd. | |
5 | * http://www.samsung.com/ | |
6 | * | |
7 | * This program is free software; you can redistribute it and/or modify | |
8 | * it under the terms of the GNU General Public License version 2 as | |
9 | * published by the Free Software Foundation. | |
10 | */ | |
11 | /* start node id of a node block dedicated to the given node id */ | |
12 | #define START_NID(nid) ((nid / NAT_ENTRY_PER_BLOCK) * NAT_ENTRY_PER_BLOCK) | |
13 | ||
14 | /* node block offset on the NAT area dedicated to the given start node id */ | |
15 | #define NAT_BLOCK_OFFSET(start_nid) (start_nid / NAT_ENTRY_PER_BLOCK) | |
16 | ||
17 | /* # of pages to perform readahead before building free nids */ | |
18 | #define FREE_NID_PAGES 4 | |
19 | ||
39a53e0c JK |
20 | /* maximum readahead size for node during getting data blocks */ |
21 | #define MAX_RA_NODE 128 | |
22 | ||
cdfc41c1 JK |
23 | /* control the memory footprint threshold (10MB per 1GB ram) */ |
24 | #define DEF_RAM_THRESHOLD 10 | |
25 | ||
39a53e0c JK |
26 | /* vector size for gang look-up from nat cache that consists of radix tree */ |
27 | #define NATVEC_SIZE 64 | |
7aed0d45 | 28 | #define SETVEC_SIZE 32 |
39a53e0c | 29 | |
56ae674c JK |
30 | /* return value for read_node_page */ |
31 | #define LOCKED_PAGE 1 | |
32 | ||
5c27f4ee CY |
33 | /* For flag in struct node_info */ |
34 | enum { | |
35 | IS_CHECKPOINTED, /* is it checkpointed before? */ | |
36 | HAS_FSYNCED_INODE, /* is the inode fsynced before? */ | |
37 | HAS_LAST_FSYNC, /* has the latest node fsync mark? */ | |
38 | IS_DIRTY, /* this nat entry is dirty? */ | |
39 | }; | |
40 | ||
39a53e0c JK |
41 | /* |
42 | * For node information | |
43 | */ | |
44 | struct node_info { | |
45 | nid_t nid; /* node id */ | |
46 | nid_t ino; /* inode number of the node's owner */ | |
47 | block_t blk_addr; /* block address of the node */ | |
48 | unsigned char version; /* version of the node */ | |
5c27f4ee | 49 | unsigned char flag; /* for node information bits */ |
7ef35e3b JK |
50 | }; |
51 | ||
39a53e0c JK |
52 | struct nat_entry { |
53 | struct list_head list; /* for clean or dirty nat list */ | |
39a53e0c JK |
54 | struct node_info ni; /* in-memory node information */ |
55 | }; | |
56 | ||
57 | #define nat_get_nid(nat) (nat->ni.nid) | |
58 | #define nat_set_nid(nat, n) (nat->ni.nid = n) | |
59 | #define nat_get_blkaddr(nat) (nat->ni.blk_addr) | |
60 | #define nat_set_blkaddr(nat, b) (nat->ni.blk_addr = b) | |
61 | #define nat_get_ino(nat) (nat->ni.ino) | |
62 | #define nat_set_ino(nat, i) (nat->ni.ino = i) | |
63 | #define nat_get_version(nat) (nat->ni.version) | |
64 | #define nat_set_version(nat, v) (nat->ni.version = v) | |
65 | ||
39a53e0c JK |
66 | #define inc_node_version(version) (++version) |
67 | ||
5c27f4ee CY |
68 | static inline void copy_node_info(struct node_info *dst, |
69 | struct node_info *src) | |
70 | { | |
71 | dst->nid = src->nid; | |
72 | dst->ino = src->ino; | |
73 | dst->blk_addr = src->blk_addr; | |
74 | dst->version = src->version; | |
75 | /* should not copy flag here */ | |
76 | } | |
77 | ||
7ef35e3b JK |
78 | static inline void set_nat_flag(struct nat_entry *ne, |
79 | unsigned int type, bool set) | |
80 | { | |
81 | unsigned char mask = 0x01 << type; | |
82 | if (set) | |
5c27f4ee | 83 | ne->ni.flag |= mask; |
7ef35e3b | 84 | else |
5c27f4ee | 85 | ne->ni.flag &= ~mask; |
7ef35e3b JK |
86 | } |
87 | ||
88 | static inline bool get_nat_flag(struct nat_entry *ne, unsigned int type) | |
89 | { | |
90 | unsigned char mask = 0x01 << type; | |
5c27f4ee | 91 | return ne->ni.flag & mask; |
7ef35e3b JK |
92 | } |
93 | ||
88bd02c9 JK |
94 | static inline void nat_reset_flag(struct nat_entry *ne) |
95 | { | |
96 | /* these states can be set only after checkpoint was done */ | |
97 | set_nat_flag(ne, IS_CHECKPOINTED, true); | |
98 | set_nat_flag(ne, HAS_FSYNCED_INODE, false); | |
99 | set_nat_flag(ne, HAS_LAST_FSYNC, true); | |
100 | } | |
101 | ||
39a53e0c JK |
102 | static inline void node_info_from_raw_nat(struct node_info *ni, |
103 | struct f2fs_nat_entry *raw_ne) | |
104 | { | |
105 | ni->ino = le32_to_cpu(raw_ne->ino); | |
106 | ni->blk_addr = le32_to_cpu(raw_ne->block_addr); | |
107 | ni->version = raw_ne->version; | |
108 | } | |
109 | ||
94dac22e CY |
110 | static inline void raw_nat_from_node_info(struct f2fs_nat_entry *raw_ne, |
111 | struct node_info *ni) | |
112 | { | |
113 | raw_ne->ino = cpu_to_le32(ni->ino); | |
114 | raw_ne->block_addr = cpu_to_le32(ni->blk_addr); | |
115 | raw_ne->version = ni->version; | |
116 | } | |
117 | ||
6fb03f3a | 118 | enum mem_type { |
cdfc41c1 | 119 | FREE_NIDS, /* indicates the free nid list */ |
6fb03f3a | 120 | NAT_ENTRIES, /* indicates the cached nat entry */ |
e5e7ea3c JK |
121 | DIRTY_DENTS, /* indicates dirty dentry pages */ |
122 | INO_ENTRIES, /* indicates inode entries */ | |
1e84371f | 123 | BASE_CHECK, /* check kernel status */ |
cdfc41c1 JK |
124 | }; |
125 | ||
aec71382 | 126 | struct nat_entry_set { |
309cc2b6 | 127 | struct list_head set_list; /* link with other nat sets */ |
aec71382 | 128 | struct list_head entry_list; /* link with dirty nat entries */ |
309cc2b6 | 129 | nid_t set; /* set number*/ |
aec71382 CY |
130 | unsigned int entry_cnt; /* the # of nat entries in set */ |
131 | }; | |
132 | ||
39a53e0c JK |
133 | /* |
134 | * For free nid mangement | |
135 | */ | |
136 | enum nid_state { | |
137 | NID_NEW, /* newly added to free nid list */ | |
138 | NID_ALLOC /* it is allocated */ | |
139 | }; | |
140 | ||
141 | struct free_nid { | |
142 | struct list_head list; /* for free node id list */ | |
143 | nid_t nid; /* node id */ | |
144 | int state; /* in use or not: NID_NEW or NID_ALLOC */ | |
145 | }; | |
146 | ||
120c2cba | 147 | static inline void next_free_nid(struct f2fs_sb_info *sbi, nid_t *nid) |
39a53e0c JK |
148 | { |
149 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
150 | struct free_nid *fnid; | |
151 | ||
39a53e0c | 152 | spin_lock(&nm_i->free_nid_list_lock); |
c6e48930 HY |
153 | if (nm_i->fcnt <= 0) { |
154 | spin_unlock(&nm_i->free_nid_list_lock); | |
120c2cba | 155 | return; |
c6e48930 | 156 | } |
39a53e0c JK |
157 | fnid = list_entry(nm_i->free_nid_list.next, struct free_nid, list); |
158 | *nid = fnid->nid; | |
159 | spin_unlock(&nm_i->free_nid_list_lock); | |
39a53e0c JK |
160 | } |
161 | ||
162 | /* | |
163 | * inline functions | |
164 | */ | |
165 | static inline void get_nat_bitmap(struct f2fs_sb_info *sbi, void *addr) | |
166 | { | |
167 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
168 | memcpy(addr, nm_i->nat_bitmap, nm_i->bitmap_size); | |
169 | } | |
170 | ||
171 | static inline pgoff_t current_nat_addr(struct f2fs_sb_info *sbi, nid_t start) | |
172 | { | |
173 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
174 | pgoff_t block_off; | |
175 | pgoff_t block_addr; | |
176 | int seg_off; | |
177 | ||
178 | block_off = NAT_BLOCK_OFFSET(start); | |
179 | seg_off = block_off >> sbi->log_blocks_per_seg; | |
180 | ||
181 | block_addr = (pgoff_t)(nm_i->nat_blkaddr + | |
182 | (seg_off << sbi->log_blocks_per_seg << 1) + | |
183 | (block_off & ((1 << sbi->log_blocks_per_seg) - 1))); | |
184 | ||
185 | if (f2fs_test_bit(block_off, nm_i->nat_bitmap)) | |
186 | block_addr += sbi->blocks_per_seg; | |
187 | ||
188 | return block_addr; | |
189 | } | |
190 | ||
191 | static inline pgoff_t next_nat_addr(struct f2fs_sb_info *sbi, | |
192 | pgoff_t block_addr) | |
193 | { | |
194 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
195 | ||
196 | block_addr -= nm_i->nat_blkaddr; | |
197 | if ((block_addr >> sbi->log_blocks_per_seg) % 2) | |
198 | block_addr -= sbi->blocks_per_seg; | |
199 | else | |
200 | block_addr += sbi->blocks_per_seg; | |
201 | ||
202 | return block_addr + nm_i->nat_blkaddr; | |
203 | } | |
204 | ||
205 | static inline void set_to_next_nat(struct f2fs_nm_info *nm_i, nid_t start_nid) | |
206 | { | |
207 | unsigned int block_off = NAT_BLOCK_OFFSET(start_nid); | |
208 | ||
c6ac4c0e | 209 | f2fs_change_bit(block_off, nm_i->nat_bitmap); |
39a53e0c JK |
210 | } |
211 | ||
212 | static inline void fill_node_footer(struct page *page, nid_t nid, | |
213 | nid_t ino, unsigned int ofs, bool reset) | |
214 | { | |
45590710 | 215 | struct f2fs_node *rn = F2FS_NODE(page); |
09eb483e JK |
216 | unsigned int old_flag = 0; |
217 | ||
39a53e0c JK |
218 | if (reset) |
219 | memset(rn, 0, sizeof(*rn)); | |
09eb483e JK |
220 | else |
221 | old_flag = le32_to_cpu(rn->footer.flag); | |
222 | ||
39a53e0c JK |
223 | rn->footer.nid = cpu_to_le32(nid); |
224 | rn->footer.ino = cpu_to_le32(ino); | |
09eb483e JK |
225 | |
226 | /* should remain old flag bits such as COLD_BIT_SHIFT */ | |
227 | rn->footer.flag = cpu_to_le32((ofs << OFFSET_BIT_SHIFT) | | |
228 | (old_flag & OFFSET_BIT_MASK)); | |
39a53e0c JK |
229 | } |
230 | ||
231 | static inline void copy_node_footer(struct page *dst, struct page *src) | |
232 | { | |
45590710 GZ |
233 | struct f2fs_node *src_rn = F2FS_NODE(src); |
234 | struct f2fs_node *dst_rn = F2FS_NODE(dst); | |
39a53e0c JK |
235 | memcpy(&dst_rn->footer, &src_rn->footer, sizeof(struct node_footer)); |
236 | } | |
237 | ||
238 | static inline void fill_node_footer_blkaddr(struct page *page, block_t blkaddr) | |
239 | { | |
4081363f | 240 | struct f2fs_checkpoint *ckpt = F2FS_CKPT(F2FS_P_SB(page)); |
45590710 GZ |
241 | struct f2fs_node *rn = F2FS_NODE(page); |
242 | ||
39a53e0c | 243 | rn->footer.cp_ver = ckpt->checkpoint_ver; |
25ca923b | 244 | rn->footer.next_blkaddr = cpu_to_le32(blkaddr); |
39a53e0c JK |
245 | } |
246 | ||
247 | static inline nid_t ino_of_node(struct page *node_page) | |
248 | { | |
45590710 | 249 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
250 | return le32_to_cpu(rn->footer.ino); |
251 | } | |
252 | ||
253 | static inline nid_t nid_of_node(struct page *node_page) | |
254 | { | |
45590710 | 255 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
256 | return le32_to_cpu(rn->footer.nid); |
257 | } | |
258 | ||
259 | static inline unsigned int ofs_of_node(struct page *node_page) | |
260 | { | |
45590710 | 261 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
262 | unsigned flag = le32_to_cpu(rn->footer.flag); |
263 | return flag >> OFFSET_BIT_SHIFT; | |
264 | } | |
265 | ||
266 | static inline unsigned long long cpver_of_node(struct page *node_page) | |
267 | { | |
45590710 | 268 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
269 | return le64_to_cpu(rn->footer.cp_ver); |
270 | } | |
271 | ||
272 | static inline block_t next_blkaddr_of_node(struct page *node_page) | |
273 | { | |
45590710 | 274 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
275 | return le32_to_cpu(rn->footer.next_blkaddr); |
276 | } | |
277 | ||
278 | /* | |
279 | * f2fs assigns the following node offsets described as (num). | |
280 | * N = NIDS_PER_BLOCK | |
281 | * | |
282 | * Inode block (0) | |
283 | * |- direct node (1) | |
284 | * |- direct node (2) | |
285 | * |- indirect node (3) | |
286 | * | `- direct node (4 => 4 + N - 1) | |
287 | * |- indirect node (4 + N) | |
288 | * | `- direct node (5 + N => 5 + 2N - 1) | |
289 | * `- double indirect node (5 + 2N) | |
290 | * `- indirect node (6 + 2N) | |
4f4124d0 CY |
291 | * `- direct node |
292 | * ...... | |
293 | * `- indirect node ((6 + 2N) + x(N + 1)) | |
294 | * `- direct node | |
295 | * ...... | |
296 | * `- indirect node ((6 + 2N) + (N - 1)(N + 1)) | |
297 | * `- direct node | |
39a53e0c JK |
298 | */ |
299 | static inline bool IS_DNODE(struct page *node_page) | |
300 | { | |
301 | unsigned int ofs = ofs_of_node(node_page); | |
dbe6a5ff | 302 | |
4bc8e9bc | 303 | if (f2fs_has_xattr_block(ofs)) |
dbe6a5ff JK |
304 | return false; |
305 | ||
39a53e0c JK |
306 | if (ofs == 3 || ofs == 4 + NIDS_PER_BLOCK || |
307 | ofs == 5 + 2 * NIDS_PER_BLOCK) | |
308 | return false; | |
309 | if (ofs >= 6 + 2 * NIDS_PER_BLOCK) { | |
310 | ofs -= 6 + 2 * NIDS_PER_BLOCK; | |
3315101f | 311 | if (!((long int)ofs % (NIDS_PER_BLOCK + 1))) |
39a53e0c JK |
312 | return false; |
313 | } | |
314 | return true; | |
315 | } | |
316 | ||
317 | static inline void set_nid(struct page *p, int off, nid_t nid, bool i) | |
318 | { | |
45590710 | 319 | struct f2fs_node *rn = F2FS_NODE(p); |
39a53e0c | 320 | |
54b591df | 321 | f2fs_wait_on_page_writeback(p, NODE); |
39a53e0c JK |
322 | |
323 | if (i) | |
324 | rn->i.i_nid[off - NODE_DIR1_BLOCK] = cpu_to_le32(nid); | |
325 | else | |
326 | rn->in.nid[off] = cpu_to_le32(nid); | |
327 | set_page_dirty(p); | |
328 | } | |
329 | ||
330 | static inline nid_t get_nid(struct page *p, int off, bool i) | |
331 | { | |
45590710 GZ |
332 | struct f2fs_node *rn = F2FS_NODE(p); |
333 | ||
39a53e0c JK |
334 | if (i) |
335 | return le32_to_cpu(rn->i.i_nid[off - NODE_DIR1_BLOCK]); | |
336 | return le32_to_cpu(rn->in.nid[off]); | |
337 | } | |
338 | ||
339 | /* | |
340 | * Coldness identification: | |
341 | * - Mark cold files in f2fs_inode_info | |
342 | * - Mark cold node blocks in their node footer | |
343 | * - Mark cold data pages in page cache | |
344 | */ | |
a06a2416 | 345 | static inline int is_file(struct inode *inode, int type) |
39a53e0c | 346 | { |
a06a2416 | 347 | return F2FS_I(inode)->i_advise & type; |
39a53e0c JK |
348 | } |
349 | ||
a06a2416 | 350 | static inline void set_file(struct inode *inode, int type) |
953a3e27 | 351 | { |
a06a2416 | 352 | F2FS_I(inode)->i_advise |= type; |
953a3e27 JK |
353 | } |
354 | ||
354a3399 JK |
355 | static inline void clear_file(struct inode *inode, int type) |
356 | { | |
357 | F2FS_I(inode)->i_advise &= ~type; | |
358 | } | |
359 | ||
360 | #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) | |
361 | #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) | |
362 | #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) | |
363 | #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) | |
364 | #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) | |
365 | #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) | |
953a3e27 | 366 | |
39a53e0c JK |
367 | static inline int is_cold_data(struct page *page) |
368 | { | |
369 | return PageChecked(page); | |
370 | } | |
371 | ||
372 | static inline void set_cold_data(struct page *page) | |
373 | { | |
374 | SetPageChecked(page); | |
375 | } | |
376 | ||
377 | static inline void clear_cold_data(struct page *page) | |
378 | { | |
379 | ClearPageChecked(page); | |
380 | } | |
381 | ||
a06a2416 | 382 | static inline int is_node(struct page *page, int type) |
39a53e0c | 383 | { |
45590710 | 384 | struct f2fs_node *rn = F2FS_NODE(page); |
a06a2416 | 385 | return le32_to_cpu(rn->footer.flag) & (1 << type); |
39a53e0c JK |
386 | } |
387 | ||
a06a2416 NJ |
388 | #define is_cold_node(page) is_node(page, COLD_BIT_SHIFT) |
389 | #define is_fsync_dnode(page) is_node(page, FSYNC_BIT_SHIFT) | |
390 | #define is_dent_dnode(page) is_node(page, DENT_BIT_SHIFT) | |
39a53e0c JK |
391 | |
392 | static inline void set_cold_node(struct inode *inode, struct page *page) | |
393 | { | |
45590710 | 394 | struct f2fs_node *rn = F2FS_NODE(page); |
39a53e0c JK |
395 | unsigned int flag = le32_to_cpu(rn->footer.flag); |
396 | ||
397 | if (S_ISDIR(inode->i_mode)) | |
398 | flag &= ~(0x1 << COLD_BIT_SHIFT); | |
399 | else | |
400 | flag |= (0x1 << COLD_BIT_SHIFT); | |
401 | rn->footer.flag = cpu_to_le32(flag); | |
402 | } | |
403 | ||
a06a2416 | 404 | static inline void set_mark(struct page *page, int mark, int type) |
39a53e0c | 405 | { |
45590710 | 406 | struct f2fs_node *rn = F2FS_NODE(page); |
39a53e0c JK |
407 | unsigned int flag = le32_to_cpu(rn->footer.flag); |
408 | if (mark) | |
a06a2416 | 409 | flag |= (0x1 << type); |
39a53e0c | 410 | else |
a06a2416 | 411 | flag &= ~(0x1 << type); |
39a53e0c JK |
412 | rn->footer.flag = cpu_to_le32(flag); |
413 | } | |
a06a2416 NJ |
414 | #define set_dentry_mark(page, mark) set_mark(page, mark, DENT_BIT_SHIFT) |
415 | #define set_fsync_mark(page, mark) set_mark(page, mark, FSYNC_BIT_SHIFT) |