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 | ||
ea1a29a0 | 17 | /* # of pages to perform synchronous readahead before building free nids */ |
39a53e0c JK |
18 | #define FREE_NID_PAGES 4 |
19 | ||
ea1a29a0 CY |
20 | #define DEF_RA_NID_PAGES 4 /* # of nid pages to be readaheaded */ |
21 | ||
39a53e0c JK |
22 | /* maximum readahead size for node during getting data blocks */ |
23 | #define MAX_RA_NODE 128 | |
24 | ||
cdfc41c1 JK |
25 | /* control the memory footprint threshold (10MB per 1GB ram) */ |
26 | #define DEF_RAM_THRESHOLD 10 | |
27 | ||
7d768d2c CY |
28 | /* control dirty nats ratio threshold (default: 10% over max nid count) */ |
29 | #define DEF_DIRTY_NAT_RATIO_THRESHOLD 10 | |
30 | ||
39a53e0c JK |
31 | /* vector size for gang look-up from nat cache that consists of radix tree */ |
32 | #define NATVEC_SIZE 64 | |
7aed0d45 | 33 | #define SETVEC_SIZE 32 |
39a53e0c | 34 | |
56ae674c JK |
35 | /* return value for read_node_page */ |
36 | #define LOCKED_PAGE 1 | |
37 | ||
5c27f4ee CY |
38 | /* For flag in struct node_info */ |
39 | enum { | |
40 | IS_CHECKPOINTED, /* is it checkpointed before? */ | |
41 | HAS_FSYNCED_INODE, /* is the inode fsynced before? */ | |
42 | HAS_LAST_FSYNC, /* has the latest node fsync mark? */ | |
43 | IS_DIRTY, /* this nat entry is dirty? */ | |
44 | }; | |
45 | ||
39a53e0c JK |
46 | /* |
47 | * For node information | |
48 | */ | |
49 | struct node_info { | |
50 | nid_t nid; /* node id */ | |
51 | nid_t ino; /* inode number of the node's owner */ | |
52 | block_t blk_addr; /* block address of the node */ | |
53 | unsigned char version; /* version of the node */ | |
5c27f4ee | 54 | unsigned char flag; /* for node information bits */ |
7ef35e3b JK |
55 | }; |
56 | ||
39a53e0c JK |
57 | struct nat_entry { |
58 | struct list_head list; /* for clean or dirty nat list */ | |
39a53e0c JK |
59 | struct node_info ni; /* in-memory node information */ |
60 | }; | |
61 | ||
62 | #define nat_get_nid(nat) (nat->ni.nid) | |
63 | #define nat_set_nid(nat, n) (nat->ni.nid = n) | |
64 | #define nat_get_blkaddr(nat) (nat->ni.blk_addr) | |
65 | #define nat_set_blkaddr(nat, b) (nat->ni.blk_addr = b) | |
66 | #define nat_get_ino(nat) (nat->ni.ino) | |
67 | #define nat_set_ino(nat, i) (nat->ni.ino = i) | |
68 | #define nat_get_version(nat) (nat->ni.version) | |
69 | #define nat_set_version(nat, v) (nat->ni.version = v) | |
70 | ||
39a53e0c JK |
71 | #define inc_node_version(version) (++version) |
72 | ||
5c27f4ee CY |
73 | static inline void copy_node_info(struct node_info *dst, |
74 | struct node_info *src) | |
75 | { | |
76 | dst->nid = src->nid; | |
77 | dst->ino = src->ino; | |
78 | dst->blk_addr = src->blk_addr; | |
79 | dst->version = src->version; | |
80 | /* should not copy flag here */ | |
81 | } | |
82 | ||
7ef35e3b JK |
83 | static inline void set_nat_flag(struct nat_entry *ne, |
84 | unsigned int type, bool set) | |
85 | { | |
86 | unsigned char mask = 0x01 << type; | |
87 | if (set) | |
5c27f4ee | 88 | ne->ni.flag |= mask; |
7ef35e3b | 89 | else |
5c27f4ee | 90 | ne->ni.flag &= ~mask; |
7ef35e3b JK |
91 | } |
92 | ||
93 | static inline bool get_nat_flag(struct nat_entry *ne, unsigned int type) | |
94 | { | |
95 | unsigned char mask = 0x01 << type; | |
5c27f4ee | 96 | return ne->ni.flag & mask; |
7ef35e3b JK |
97 | } |
98 | ||
88bd02c9 JK |
99 | static inline void nat_reset_flag(struct nat_entry *ne) |
100 | { | |
101 | /* these states can be set only after checkpoint was done */ | |
102 | set_nat_flag(ne, IS_CHECKPOINTED, true); | |
103 | set_nat_flag(ne, HAS_FSYNCED_INODE, false); | |
104 | set_nat_flag(ne, HAS_LAST_FSYNC, true); | |
105 | } | |
106 | ||
39a53e0c JK |
107 | static inline void node_info_from_raw_nat(struct node_info *ni, |
108 | struct f2fs_nat_entry *raw_ne) | |
109 | { | |
110 | ni->ino = le32_to_cpu(raw_ne->ino); | |
111 | ni->blk_addr = le32_to_cpu(raw_ne->block_addr); | |
112 | ni->version = raw_ne->version; | |
113 | } | |
114 | ||
94dac22e CY |
115 | static inline void raw_nat_from_node_info(struct f2fs_nat_entry *raw_ne, |
116 | struct node_info *ni) | |
117 | { | |
118 | raw_ne->ino = cpu_to_le32(ni->ino); | |
119 | raw_ne->block_addr = cpu_to_le32(ni->blk_addr); | |
120 | raw_ne->version = ni->version; | |
121 | } | |
122 | ||
7d768d2c CY |
123 | static inline bool excess_dirty_nats(struct f2fs_sb_info *sbi) |
124 | { | |
125 | return NM_I(sbi)->dirty_nat_cnt >= NM_I(sbi)->max_nid * | |
2304cb0c | 126 | NM_I(sbi)->dirty_nats_ratio / 100; |
7d768d2c CY |
127 | } |
128 | ||
6fb03f3a | 129 | enum mem_type { |
cdfc41c1 | 130 | FREE_NIDS, /* indicates the free nid list */ |
6fb03f3a | 131 | NAT_ENTRIES, /* indicates the cached nat entry */ |
a1257023 | 132 | DIRTY_DENTS, /* indicates dirty dentry pages */ |
e5e7ea3c | 133 | INO_ENTRIES, /* indicates inode entries */ |
13054c54 | 134 | EXTENT_CACHE, /* indicates extent cache */ |
1e84371f | 135 | BASE_CHECK, /* check kernel status */ |
cdfc41c1 JK |
136 | }; |
137 | ||
aec71382 | 138 | struct nat_entry_set { |
309cc2b6 | 139 | struct list_head set_list; /* link with other nat sets */ |
aec71382 | 140 | struct list_head entry_list; /* link with dirty nat entries */ |
309cc2b6 | 141 | nid_t set; /* set number*/ |
aec71382 CY |
142 | unsigned int entry_cnt; /* the # of nat entries in set */ |
143 | }; | |
144 | ||
39a53e0c JK |
145 | /* |
146 | * For free nid mangement | |
147 | */ | |
148 | enum nid_state { | |
149 | NID_NEW, /* newly added to free nid list */ | |
150 | NID_ALLOC /* it is allocated */ | |
151 | }; | |
152 | ||
153 | struct free_nid { | |
154 | struct list_head list; /* for free node id list */ | |
155 | nid_t nid; /* node id */ | |
156 | int state; /* in use or not: NID_NEW or NID_ALLOC */ | |
157 | }; | |
158 | ||
120c2cba | 159 | static inline void next_free_nid(struct f2fs_sb_info *sbi, nid_t *nid) |
39a53e0c JK |
160 | { |
161 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
162 | struct free_nid *fnid; | |
163 | ||
39a53e0c | 164 | spin_lock(&nm_i->free_nid_list_lock); |
c6e48930 HY |
165 | if (nm_i->fcnt <= 0) { |
166 | spin_unlock(&nm_i->free_nid_list_lock); | |
120c2cba | 167 | return; |
c6e48930 | 168 | } |
39a53e0c JK |
169 | fnid = list_entry(nm_i->free_nid_list.next, struct free_nid, list); |
170 | *nid = fnid->nid; | |
171 | spin_unlock(&nm_i->free_nid_list_lock); | |
39a53e0c JK |
172 | } |
173 | ||
174 | /* | |
175 | * inline functions | |
176 | */ | |
177 | static inline void get_nat_bitmap(struct f2fs_sb_info *sbi, void *addr) | |
178 | { | |
179 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
180 | memcpy(addr, nm_i->nat_bitmap, nm_i->bitmap_size); | |
181 | } | |
182 | ||
183 | static inline pgoff_t current_nat_addr(struct f2fs_sb_info *sbi, nid_t start) | |
184 | { | |
185 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
186 | pgoff_t block_off; | |
187 | pgoff_t block_addr; | |
188 | int seg_off; | |
189 | ||
190 | block_off = NAT_BLOCK_OFFSET(start); | |
191 | seg_off = block_off >> sbi->log_blocks_per_seg; | |
192 | ||
193 | block_addr = (pgoff_t)(nm_i->nat_blkaddr + | |
194 | (seg_off << sbi->log_blocks_per_seg << 1) + | |
3519e3f9 | 195 | (block_off & (sbi->blocks_per_seg - 1))); |
39a53e0c JK |
196 | |
197 | if (f2fs_test_bit(block_off, nm_i->nat_bitmap)) | |
198 | block_addr += sbi->blocks_per_seg; | |
199 | ||
200 | return block_addr; | |
201 | } | |
202 | ||
203 | static inline pgoff_t next_nat_addr(struct f2fs_sb_info *sbi, | |
204 | pgoff_t block_addr) | |
205 | { | |
206 | struct f2fs_nm_info *nm_i = NM_I(sbi); | |
207 | ||
208 | block_addr -= nm_i->nat_blkaddr; | |
209 | if ((block_addr >> sbi->log_blocks_per_seg) % 2) | |
210 | block_addr -= sbi->blocks_per_seg; | |
211 | else | |
212 | block_addr += sbi->blocks_per_seg; | |
213 | ||
214 | return block_addr + nm_i->nat_blkaddr; | |
215 | } | |
216 | ||
217 | static inline void set_to_next_nat(struct f2fs_nm_info *nm_i, nid_t start_nid) | |
218 | { | |
219 | unsigned int block_off = NAT_BLOCK_OFFSET(start_nid); | |
220 | ||
c6ac4c0e | 221 | f2fs_change_bit(block_off, nm_i->nat_bitmap); |
39a53e0c JK |
222 | } |
223 | ||
224 | static inline void fill_node_footer(struct page *page, nid_t nid, | |
225 | nid_t ino, unsigned int ofs, bool reset) | |
226 | { | |
45590710 | 227 | struct f2fs_node *rn = F2FS_NODE(page); |
09eb483e JK |
228 | unsigned int old_flag = 0; |
229 | ||
39a53e0c JK |
230 | if (reset) |
231 | memset(rn, 0, sizeof(*rn)); | |
09eb483e JK |
232 | else |
233 | old_flag = le32_to_cpu(rn->footer.flag); | |
234 | ||
39a53e0c JK |
235 | rn->footer.nid = cpu_to_le32(nid); |
236 | rn->footer.ino = cpu_to_le32(ino); | |
09eb483e JK |
237 | |
238 | /* should remain old flag bits such as COLD_BIT_SHIFT */ | |
239 | rn->footer.flag = cpu_to_le32((ofs << OFFSET_BIT_SHIFT) | | |
240 | (old_flag & OFFSET_BIT_MASK)); | |
39a53e0c JK |
241 | } |
242 | ||
243 | static inline void copy_node_footer(struct page *dst, struct page *src) | |
244 | { | |
45590710 GZ |
245 | struct f2fs_node *src_rn = F2FS_NODE(src); |
246 | struct f2fs_node *dst_rn = F2FS_NODE(dst); | |
39a53e0c JK |
247 | memcpy(&dst_rn->footer, &src_rn->footer, sizeof(struct node_footer)); |
248 | } | |
249 | ||
250 | static inline void fill_node_footer_blkaddr(struct page *page, block_t blkaddr) | |
251 | { | |
4081363f | 252 | struct f2fs_checkpoint *ckpt = F2FS_CKPT(F2FS_P_SB(page)); |
45590710 GZ |
253 | struct f2fs_node *rn = F2FS_NODE(page); |
254 | ||
39a53e0c | 255 | rn->footer.cp_ver = ckpt->checkpoint_ver; |
25ca923b | 256 | rn->footer.next_blkaddr = cpu_to_le32(blkaddr); |
39a53e0c JK |
257 | } |
258 | ||
259 | static inline nid_t ino_of_node(struct page *node_page) | |
260 | { | |
45590710 | 261 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
262 | return le32_to_cpu(rn->footer.ino); |
263 | } | |
264 | ||
265 | static inline nid_t nid_of_node(struct page *node_page) | |
266 | { | |
45590710 | 267 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
268 | return le32_to_cpu(rn->footer.nid); |
269 | } | |
270 | ||
271 | static inline unsigned int ofs_of_node(struct page *node_page) | |
272 | { | |
45590710 | 273 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
274 | unsigned flag = le32_to_cpu(rn->footer.flag); |
275 | return flag >> OFFSET_BIT_SHIFT; | |
276 | } | |
277 | ||
278 | static inline unsigned long long cpver_of_node(struct page *node_page) | |
279 | { | |
45590710 | 280 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
281 | return le64_to_cpu(rn->footer.cp_ver); |
282 | } | |
283 | ||
284 | static inline block_t next_blkaddr_of_node(struct page *node_page) | |
285 | { | |
45590710 | 286 | struct f2fs_node *rn = F2FS_NODE(node_page); |
39a53e0c JK |
287 | return le32_to_cpu(rn->footer.next_blkaddr); |
288 | } | |
289 | ||
290 | /* | |
291 | * f2fs assigns the following node offsets described as (num). | |
292 | * N = NIDS_PER_BLOCK | |
293 | * | |
294 | * Inode block (0) | |
295 | * |- direct node (1) | |
296 | * |- direct node (2) | |
297 | * |- indirect node (3) | |
298 | * | `- direct node (4 => 4 + N - 1) | |
299 | * |- indirect node (4 + N) | |
300 | * | `- direct node (5 + N => 5 + 2N - 1) | |
301 | * `- double indirect node (5 + 2N) | |
302 | * `- indirect node (6 + 2N) | |
4f4124d0 CY |
303 | * `- direct node |
304 | * ...... | |
305 | * `- indirect node ((6 + 2N) + x(N + 1)) | |
306 | * `- direct node | |
307 | * ...... | |
308 | * `- indirect node ((6 + 2N) + (N - 1)(N + 1)) | |
309 | * `- direct node | |
39a53e0c JK |
310 | */ |
311 | static inline bool IS_DNODE(struct page *node_page) | |
312 | { | |
313 | unsigned int ofs = ofs_of_node(node_page); | |
dbe6a5ff | 314 | |
4bc8e9bc | 315 | if (f2fs_has_xattr_block(ofs)) |
dbe6a5ff JK |
316 | return false; |
317 | ||
39a53e0c JK |
318 | if (ofs == 3 || ofs == 4 + NIDS_PER_BLOCK || |
319 | ofs == 5 + 2 * NIDS_PER_BLOCK) | |
320 | return false; | |
321 | if (ofs >= 6 + 2 * NIDS_PER_BLOCK) { | |
322 | ofs -= 6 + 2 * NIDS_PER_BLOCK; | |
3315101f | 323 | if (!((long int)ofs % (NIDS_PER_BLOCK + 1))) |
39a53e0c JK |
324 | return false; |
325 | } | |
326 | return true; | |
327 | } | |
328 | ||
12719ae1 | 329 | static inline int set_nid(struct page *p, int off, nid_t nid, bool i) |
39a53e0c | 330 | { |
45590710 | 331 | struct f2fs_node *rn = F2FS_NODE(p); |
39a53e0c | 332 | |
fec1d657 | 333 | f2fs_wait_on_page_writeback(p, NODE, true); |
39a53e0c JK |
334 | |
335 | if (i) | |
336 | rn->i.i_nid[off - NODE_DIR1_BLOCK] = cpu_to_le32(nid); | |
337 | else | |
338 | rn->in.nid[off] = cpu_to_le32(nid); | |
12719ae1 | 339 | return set_page_dirty(p); |
39a53e0c JK |
340 | } |
341 | ||
342 | static inline nid_t get_nid(struct page *p, int off, bool i) | |
343 | { | |
45590710 GZ |
344 | struct f2fs_node *rn = F2FS_NODE(p); |
345 | ||
39a53e0c JK |
346 | if (i) |
347 | return le32_to_cpu(rn->i.i_nid[off - NODE_DIR1_BLOCK]); | |
348 | return le32_to_cpu(rn->in.nid[off]); | |
349 | } | |
350 | ||
351 | /* | |
352 | * Coldness identification: | |
353 | * - Mark cold files in f2fs_inode_info | |
354 | * - Mark cold node blocks in their node footer | |
355 | * - Mark cold data pages in page cache | |
356 | */ | |
39a53e0c JK |
357 | static inline int is_cold_data(struct page *page) |
358 | { | |
359 | return PageChecked(page); | |
360 | } | |
361 | ||
362 | static inline void set_cold_data(struct page *page) | |
363 | { | |
364 | SetPageChecked(page); | |
365 | } | |
366 | ||
367 | static inline void clear_cold_data(struct page *page) | |
368 | { | |
369 | ClearPageChecked(page); | |
370 | } | |
371 | ||
a06a2416 | 372 | static inline int is_node(struct page *page, int type) |
39a53e0c | 373 | { |
45590710 | 374 | struct f2fs_node *rn = F2FS_NODE(page); |
a06a2416 | 375 | return le32_to_cpu(rn->footer.flag) & (1 << type); |
39a53e0c JK |
376 | } |
377 | ||
a06a2416 NJ |
378 | #define is_cold_node(page) is_node(page, COLD_BIT_SHIFT) |
379 | #define is_fsync_dnode(page) is_node(page, FSYNC_BIT_SHIFT) | |
380 | #define is_dent_dnode(page) is_node(page, DENT_BIT_SHIFT) | |
39a53e0c | 381 | |
2049d4fc JK |
382 | static inline int is_inline_node(struct page *page) |
383 | { | |
384 | return PageChecked(page); | |
385 | } | |
386 | ||
387 | static inline void set_inline_node(struct page *page) | |
388 | { | |
389 | SetPageChecked(page); | |
390 | } | |
391 | ||
392 | static inline void clear_inline_node(struct page *page) | |
393 | { | |
394 | ClearPageChecked(page); | |
395 | } | |
396 | ||
39a53e0c JK |
397 | static inline void set_cold_node(struct inode *inode, struct page *page) |
398 | { | |
45590710 | 399 | struct f2fs_node *rn = F2FS_NODE(page); |
39a53e0c JK |
400 | unsigned int flag = le32_to_cpu(rn->footer.flag); |
401 | ||
402 | if (S_ISDIR(inode->i_mode)) | |
403 | flag &= ~(0x1 << COLD_BIT_SHIFT); | |
404 | else | |
405 | flag |= (0x1 << COLD_BIT_SHIFT); | |
406 | rn->footer.flag = cpu_to_le32(flag); | |
407 | } | |
408 | ||
a06a2416 | 409 | static inline void set_mark(struct page *page, int mark, int type) |
39a53e0c | 410 | { |
45590710 | 411 | struct f2fs_node *rn = F2FS_NODE(page); |
39a53e0c JK |
412 | unsigned int flag = le32_to_cpu(rn->footer.flag); |
413 | if (mark) | |
a06a2416 | 414 | flag |= (0x1 << type); |
39a53e0c | 415 | else |
a06a2416 | 416 | flag &= ~(0x1 << type); |
39a53e0c JK |
417 | rn->footer.flag = cpu_to_le32(flag); |
418 | } | |
a06a2416 NJ |
419 | #define set_dentry_mark(page, mark) set_mark(page, mark, DENT_BIT_SHIFT) |
420 | #define set_fsync_mark(page, mark) set_mark(page, mark, FSYNC_BIT_SHIFT) |