node.h 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396
  1. /*
  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. /* node block offset on the NAT area dedicated to the given start node id */
  14. #define NAT_BLOCK_OFFSET(start_nid) (start_nid / NAT_ENTRY_PER_BLOCK)
  15. /* # of pages to perform readahead before building free nids */
  16. #define FREE_NID_PAGES 4
  17. /* maximum readahead size for node during getting data blocks */
  18. #define MAX_RA_NODE 128
  19. /* control the memory footprint threshold (10MB per 1GB ram) */
  20. #define DEF_RAM_THRESHOLD 10
  21. /* vector size for gang look-up from nat cache that consists of radix tree */
  22. #define NATVEC_SIZE 64
  23. /* return value for read_node_page */
  24. #define LOCKED_PAGE 1
  25. /*
  26. * For node information
  27. */
  28. struct node_info {
  29. nid_t nid; /* node id */
  30. nid_t ino; /* inode number of the node's owner */
  31. block_t blk_addr; /* block address of the node */
  32. unsigned char version; /* version of the node */
  33. };
  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. struct nat_entry {
  41. struct list_head list; /* for clean or dirty nat list */
  42. unsigned char flag; /* for node information bits */
  43. struct node_info ni; /* in-memory node information */
  44. };
  45. #define nat_get_nid(nat) (nat->ni.nid)
  46. #define nat_set_nid(nat, n) (nat->ni.nid = n)
  47. #define nat_get_blkaddr(nat) (nat->ni.blk_addr)
  48. #define nat_set_blkaddr(nat, b) (nat->ni.blk_addr = b)
  49. #define nat_get_ino(nat) (nat->ni.ino)
  50. #define nat_set_ino(nat, i) (nat->ni.ino = i)
  51. #define nat_get_version(nat) (nat->ni.version)
  52. #define nat_set_version(nat, v) (nat->ni.version = v)
  53. #define inc_node_version(version) (++version)
  54. static inline void set_nat_flag(struct nat_entry *ne,
  55. unsigned int type, bool set)
  56. {
  57. unsigned char mask = 0x01 << type;
  58. if (set)
  59. ne->flag |= mask;
  60. else
  61. ne->flag &= ~mask;
  62. }
  63. static inline bool get_nat_flag(struct nat_entry *ne, unsigned int type)
  64. {
  65. unsigned char mask = 0x01 << type;
  66. return ne->flag & mask;
  67. }
  68. static inline void nat_reset_flag(struct nat_entry *ne)
  69. {
  70. /* these states can be set only after checkpoint was done */
  71. set_nat_flag(ne, IS_CHECKPOINTED, true);
  72. set_nat_flag(ne, HAS_FSYNCED_INODE, false);
  73. set_nat_flag(ne, HAS_LAST_FSYNC, true);
  74. }
  75. static inline void node_info_from_raw_nat(struct node_info *ni,
  76. struct f2fs_nat_entry *raw_ne)
  77. {
  78. ni->ino = le32_to_cpu(raw_ne->ino);
  79. ni->blk_addr = le32_to_cpu(raw_ne->block_addr);
  80. ni->version = raw_ne->version;
  81. }
  82. static inline void raw_nat_from_node_info(struct f2fs_nat_entry *raw_ne,
  83. struct node_info *ni)
  84. {
  85. raw_ne->ino = cpu_to_le32(ni->ino);
  86. raw_ne->block_addr = cpu_to_le32(ni->blk_addr);
  87. raw_ne->version = ni->version;
  88. }
  89. enum mem_type {
  90. FREE_NIDS, /* indicates the free nid list */
  91. NAT_ENTRIES, /* indicates the cached nat entry */
  92. DIRTY_DENTS /* indicates dirty dentry pages */
  93. };
  94. struct nat_entry_set {
  95. struct list_head set_list; /* link with other nat sets */
  96. struct list_head entry_list; /* link with dirty nat entries */
  97. nid_t set; /* set number*/
  98. unsigned int entry_cnt; /* the # of nat entries in set */
  99. };
  100. /*
  101. * For free nid mangement
  102. */
  103. enum nid_state {
  104. NID_NEW, /* newly added to free nid list */
  105. NID_ALLOC /* it is allocated */
  106. };
  107. struct free_nid {
  108. struct list_head list; /* for free node id list */
  109. nid_t nid; /* node id */
  110. int state; /* in use or not: NID_NEW or NID_ALLOC */
  111. };
  112. static inline void next_free_nid(struct f2fs_sb_info *sbi, nid_t *nid)
  113. {
  114. struct f2fs_nm_info *nm_i = NM_I(sbi);
  115. struct free_nid *fnid;
  116. spin_lock(&nm_i->free_nid_list_lock);
  117. if (nm_i->fcnt <= 0) {
  118. spin_unlock(&nm_i->free_nid_list_lock);
  119. return;
  120. }
  121. fnid = list_entry(nm_i->free_nid_list.next, struct free_nid, list);
  122. *nid = fnid->nid;
  123. spin_unlock(&nm_i->free_nid_list_lock);
  124. }
  125. /*
  126. * inline functions
  127. */
  128. static inline void get_nat_bitmap(struct f2fs_sb_info *sbi, void *addr)
  129. {
  130. struct f2fs_nm_info *nm_i = NM_I(sbi);
  131. memcpy(addr, nm_i->nat_bitmap, nm_i->bitmap_size);
  132. }
  133. static inline pgoff_t current_nat_addr(struct f2fs_sb_info *sbi, nid_t start)
  134. {
  135. struct f2fs_nm_info *nm_i = NM_I(sbi);
  136. pgoff_t block_off;
  137. pgoff_t block_addr;
  138. int seg_off;
  139. block_off = NAT_BLOCK_OFFSET(start);
  140. seg_off = block_off >> sbi->log_blocks_per_seg;
  141. block_addr = (pgoff_t)(nm_i->nat_blkaddr +
  142. (seg_off << sbi->log_blocks_per_seg << 1) +
  143. (block_off & ((1 << sbi->log_blocks_per_seg) - 1)));
  144. if (f2fs_test_bit(block_off, nm_i->nat_bitmap))
  145. block_addr += sbi->blocks_per_seg;
  146. return block_addr;
  147. }
  148. static inline pgoff_t next_nat_addr(struct f2fs_sb_info *sbi,
  149. pgoff_t block_addr)
  150. {
  151. struct f2fs_nm_info *nm_i = NM_I(sbi);
  152. block_addr -= nm_i->nat_blkaddr;
  153. if ((block_addr >> sbi->log_blocks_per_seg) % 2)
  154. block_addr -= sbi->blocks_per_seg;
  155. else
  156. block_addr += sbi->blocks_per_seg;
  157. return block_addr + nm_i->nat_blkaddr;
  158. }
  159. static inline void set_to_next_nat(struct f2fs_nm_info *nm_i, nid_t start_nid)
  160. {
  161. unsigned int block_off = NAT_BLOCK_OFFSET(start_nid);
  162. if (f2fs_test_bit(block_off, nm_i->nat_bitmap))
  163. f2fs_clear_bit(block_off, nm_i->nat_bitmap);
  164. else
  165. f2fs_set_bit(block_off, nm_i->nat_bitmap);
  166. }
  167. static inline void fill_node_footer(struct page *page, nid_t nid,
  168. nid_t ino, unsigned int ofs, bool reset)
  169. {
  170. struct f2fs_node *rn = F2FS_NODE(page);
  171. if (reset)
  172. memset(rn, 0, sizeof(*rn));
  173. rn->footer.nid = cpu_to_le32(nid);
  174. rn->footer.ino = cpu_to_le32(ino);
  175. rn->footer.flag = cpu_to_le32(ofs << OFFSET_BIT_SHIFT);
  176. }
  177. static inline void copy_node_footer(struct page *dst, struct page *src)
  178. {
  179. struct f2fs_node *src_rn = F2FS_NODE(src);
  180. struct f2fs_node *dst_rn = F2FS_NODE(dst);
  181. memcpy(&dst_rn->footer, &src_rn->footer, sizeof(struct node_footer));
  182. }
  183. static inline void fill_node_footer_blkaddr(struct page *page, block_t blkaddr)
  184. {
  185. struct f2fs_checkpoint *ckpt = F2FS_CKPT(F2FS_P_SB(page));
  186. struct f2fs_node *rn = F2FS_NODE(page);
  187. rn->footer.cp_ver = ckpt->checkpoint_ver;
  188. rn->footer.next_blkaddr = cpu_to_le32(blkaddr);
  189. }
  190. static inline nid_t ino_of_node(struct page *node_page)
  191. {
  192. struct f2fs_node *rn = F2FS_NODE(node_page);
  193. return le32_to_cpu(rn->footer.ino);
  194. }
  195. static inline nid_t nid_of_node(struct page *node_page)
  196. {
  197. struct f2fs_node *rn = F2FS_NODE(node_page);
  198. return le32_to_cpu(rn->footer.nid);
  199. }
  200. static inline unsigned int ofs_of_node(struct page *node_page)
  201. {
  202. struct f2fs_node *rn = F2FS_NODE(node_page);
  203. unsigned flag = le32_to_cpu(rn->footer.flag);
  204. return flag >> OFFSET_BIT_SHIFT;
  205. }
  206. static inline unsigned long long cpver_of_node(struct page *node_page)
  207. {
  208. struct f2fs_node *rn = F2FS_NODE(node_page);
  209. return le64_to_cpu(rn->footer.cp_ver);
  210. }
  211. static inline block_t next_blkaddr_of_node(struct page *node_page)
  212. {
  213. struct f2fs_node *rn = F2FS_NODE(node_page);
  214. return le32_to_cpu(rn->footer.next_blkaddr);
  215. }
  216. /*
  217. * f2fs assigns the following node offsets described as (num).
  218. * N = NIDS_PER_BLOCK
  219. *
  220. * Inode block (0)
  221. * |- direct node (1)
  222. * |- direct node (2)
  223. * |- indirect node (3)
  224. * | `- direct node (4 => 4 + N - 1)
  225. * |- indirect node (4 + N)
  226. * | `- direct node (5 + N => 5 + 2N - 1)
  227. * `- double indirect node (5 + 2N)
  228. * `- indirect node (6 + 2N)
  229. * `- direct node
  230. * ......
  231. * `- indirect node ((6 + 2N) + x(N + 1))
  232. * `- direct node
  233. * ......
  234. * `- indirect node ((6 + 2N) + (N - 1)(N + 1))
  235. * `- direct node
  236. */
  237. static inline bool IS_DNODE(struct page *node_page)
  238. {
  239. unsigned int ofs = ofs_of_node(node_page);
  240. if (f2fs_has_xattr_block(ofs))
  241. return false;
  242. if (ofs == 3 || ofs == 4 + NIDS_PER_BLOCK ||
  243. ofs == 5 + 2 * NIDS_PER_BLOCK)
  244. return false;
  245. if (ofs >= 6 + 2 * NIDS_PER_BLOCK) {
  246. ofs -= 6 + 2 * NIDS_PER_BLOCK;
  247. if (!((long int)ofs % (NIDS_PER_BLOCK + 1)))
  248. return false;
  249. }
  250. return true;
  251. }
  252. static inline void set_nid(struct page *p, int off, nid_t nid, bool i)
  253. {
  254. struct f2fs_node *rn = F2FS_NODE(p);
  255. f2fs_wait_on_page_writeback(p, NODE);
  256. if (i)
  257. rn->i.i_nid[off - NODE_DIR1_BLOCK] = cpu_to_le32(nid);
  258. else
  259. rn->in.nid[off] = cpu_to_le32(nid);
  260. set_page_dirty(p);
  261. }
  262. static inline nid_t get_nid(struct page *p, int off, bool i)
  263. {
  264. struct f2fs_node *rn = F2FS_NODE(p);
  265. if (i)
  266. return le32_to_cpu(rn->i.i_nid[off - NODE_DIR1_BLOCK]);
  267. return le32_to_cpu(rn->in.nid[off]);
  268. }
  269. /*
  270. * Coldness identification:
  271. * - Mark cold files in f2fs_inode_info
  272. * - Mark cold node blocks in their node footer
  273. * - Mark cold data pages in page cache
  274. */
  275. static inline int is_file(struct inode *inode, int type)
  276. {
  277. return F2FS_I(inode)->i_advise & type;
  278. }
  279. static inline void set_file(struct inode *inode, int type)
  280. {
  281. F2FS_I(inode)->i_advise |= type;
  282. }
  283. static inline void clear_file(struct inode *inode, int type)
  284. {
  285. F2FS_I(inode)->i_advise &= ~type;
  286. }
  287. #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
  288. #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
  289. #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
  290. #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
  291. #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
  292. #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
  293. static inline int is_cold_data(struct page *page)
  294. {
  295. return PageChecked(page);
  296. }
  297. static inline void set_cold_data(struct page *page)
  298. {
  299. SetPageChecked(page);
  300. }
  301. static inline void clear_cold_data(struct page *page)
  302. {
  303. ClearPageChecked(page);
  304. }
  305. static inline int is_node(struct page *page, int type)
  306. {
  307. struct f2fs_node *rn = F2FS_NODE(page);
  308. return le32_to_cpu(rn->footer.flag) & (1 << type);
  309. }
  310. #define is_cold_node(page) is_node(page, COLD_BIT_SHIFT)
  311. #define is_fsync_dnode(page) is_node(page, FSYNC_BIT_SHIFT)
  312. #define is_dent_dnode(page) is_node(page, DENT_BIT_SHIFT)
  313. static inline void set_cold_node(struct inode *inode, struct page *page)
  314. {
  315. struct f2fs_node *rn = F2FS_NODE(page);
  316. unsigned int flag = le32_to_cpu(rn->footer.flag);
  317. if (S_ISDIR(inode->i_mode))
  318. flag &= ~(0x1 << COLD_BIT_SHIFT);
  319. else
  320. flag |= (0x1 << COLD_BIT_SHIFT);
  321. rn->footer.flag = cpu_to_le32(flag);
  322. }
  323. static inline void set_mark(struct page *page, int mark, int type)
  324. {
  325. struct f2fs_node *rn = F2FS_NODE(page);
  326. unsigned int flag = le32_to_cpu(rn->footer.flag);
  327. if (mark)
  328. flag |= (0x1 << type);
  329. else
  330. flag &= ~(0x1 << type);
  331. rn->footer.flag = cpu_to_le32(flag);
  332. }
  333. #define set_dentry_mark(page, mark) set_mark(page, mark, DENT_BIT_SHIFT)
  334. #define set_fsync_mark(page, mark) set_mark(page, mark, FSYNC_BIT_SHIFT)