4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
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.
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)
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)
17 /* # of pages to perform readahead before building free nids */
18 #define FREE_NID_PAGES 4
20 /* maximum readahead size for node during getting data blocks */
21 #define MAX_RA_NODE 128
23 /* control the memory footprint threshold (10MB per 1GB ram) */
24 #define DEF_RAM_THRESHOLD 10
26 /* vector size for gang look-up from nat cache that consists of radix tree */
27 #define NATVEC_SIZE 64
29 /* return value for read_node_page */
33 * For node information
36 nid_t nid
; /* node id */
37 nid_t ino
; /* inode number of the node's owner */
38 block_t blk_addr
; /* block address of the node */
39 unsigned char version
; /* version of the node */
43 struct list_head list
; /* for clean or dirty nat list */
44 bool checkpointed
; /* whether it is checkpointed or not */
45 bool fsync_done
; /* whether the latest node has fsync mark */
46 struct node_info ni
; /* in-memory node information */
49 #define nat_get_nid(nat) (nat->ni.nid)
50 #define nat_set_nid(nat, n) (nat->ni.nid = n)
51 #define nat_get_blkaddr(nat) (nat->ni.blk_addr)
52 #define nat_set_blkaddr(nat, b) (nat->ni.blk_addr = b)
53 #define nat_get_ino(nat) (nat->ni.ino)
54 #define nat_set_ino(nat, i) (nat->ni.ino = i)
55 #define nat_get_version(nat) (nat->ni.version)
56 #define nat_set_version(nat, v) (nat->ni.version = v)
58 #define __set_nat_cache_dirty(nm_i, ne) \
60 ne->checkpointed = false; \
61 list_move_tail(&ne->list, &nm_i->dirty_nat_entries); \
63 #define __clear_nat_cache_dirty(nm_i, ne) \
65 ne->checkpointed = true; \
66 list_move_tail(&ne->list, &nm_i->nat_entries); \
68 #define inc_node_version(version) (++version)
70 static inline void node_info_from_raw_nat(struct node_info
*ni
,
71 struct f2fs_nat_entry
*raw_ne
)
73 ni
->ino
= le32_to_cpu(raw_ne
->ino
);
74 ni
->blk_addr
= le32_to_cpu(raw_ne
->block_addr
);
75 ni
->version
= raw_ne
->version
;
78 static inline void raw_nat_from_node_info(struct f2fs_nat_entry
*raw_ne
,
81 raw_ne
->ino
= cpu_to_le32(ni
->ino
);
82 raw_ne
->block_addr
= cpu_to_le32(ni
->blk_addr
);
83 raw_ne
->version
= ni
->version
;
87 FREE_NIDS
, /* indicates the free nid list */
88 NAT_ENTRIES
, /* indicates the cached nat entry */
89 DIRTY_DENTS
/* indicates dirty dentry pages */
92 struct nat_entry_set
{
93 struct list_head set_list
; /* link with all nat sets */
94 struct list_head entry_list
; /* link with dirty nat entries */
95 nid_t start_nid
; /* start nid of nats in set */
96 unsigned int entry_cnt
; /* the # of nat entries in set */
100 * For free nid mangement
103 NID_NEW
, /* newly added to free nid list */
104 NID_ALLOC
/* it is allocated */
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 */
113 static inline int next_free_nid(struct f2fs_sb_info
*sbi
, nid_t
*nid
)
115 struct f2fs_nm_info
*nm_i
= NM_I(sbi
);
116 struct free_nid
*fnid
;
120 spin_lock(&nm_i
->free_nid_list_lock
);
121 fnid
= list_entry(nm_i
->free_nid_list
.next
, struct free_nid
, list
);
123 spin_unlock(&nm_i
->free_nid_list_lock
);
130 static inline void get_nat_bitmap(struct f2fs_sb_info
*sbi
, void *addr
)
132 struct f2fs_nm_info
*nm_i
= NM_I(sbi
);
133 memcpy(addr
, nm_i
->nat_bitmap
, nm_i
->bitmap_size
);
136 static inline pgoff_t
current_nat_addr(struct f2fs_sb_info
*sbi
, nid_t start
)
138 struct f2fs_nm_info
*nm_i
= NM_I(sbi
);
143 block_off
= NAT_BLOCK_OFFSET(start
);
144 seg_off
= block_off
>> sbi
->log_blocks_per_seg
;
146 block_addr
= (pgoff_t
)(nm_i
->nat_blkaddr
+
147 (seg_off
<< sbi
->log_blocks_per_seg
<< 1) +
148 (block_off
& ((1 << sbi
->log_blocks_per_seg
) - 1)));
150 if (f2fs_test_bit(block_off
, nm_i
->nat_bitmap
))
151 block_addr
+= sbi
->blocks_per_seg
;
156 static inline pgoff_t
next_nat_addr(struct f2fs_sb_info
*sbi
,
159 struct f2fs_nm_info
*nm_i
= NM_I(sbi
);
161 block_addr
-= nm_i
->nat_blkaddr
;
162 if ((block_addr
>> sbi
->log_blocks_per_seg
) % 2)
163 block_addr
-= sbi
->blocks_per_seg
;
165 block_addr
+= sbi
->blocks_per_seg
;
167 return block_addr
+ nm_i
->nat_blkaddr
;
170 static inline void set_to_next_nat(struct f2fs_nm_info
*nm_i
, nid_t start_nid
)
172 unsigned int block_off
= NAT_BLOCK_OFFSET(start_nid
);
174 if (f2fs_test_bit(block_off
, nm_i
->nat_bitmap
))
175 f2fs_clear_bit(block_off
, nm_i
->nat_bitmap
);
177 f2fs_set_bit(block_off
, nm_i
->nat_bitmap
);
180 static inline void fill_node_footer(struct page
*page
, nid_t nid
,
181 nid_t ino
, unsigned int ofs
, bool reset
)
183 struct f2fs_node
*rn
= F2FS_NODE(page
);
185 memset(rn
, 0, sizeof(*rn
));
186 rn
->footer
.nid
= cpu_to_le32(nid
);
187 rn
->footer
.ino
= cpu_to_le32(ino
);
188 rn
->footer
.flag
= cpu_to_le32(ofs
<< OFFSET_BIT_SHIFT
);
191 static inline void copy_node_footer(struct page
*dst
, struct page
*src
)
193 struct f2fs_node
*src_rn
= F2FS_NODE(src
);
194 struct f2fs_node
*dst_rn
= F2FS_NODE(dst
);
195 memcpy(&dst_rn
->footer
, &src_rn
->footer
, sizeof(struct node_footer
));
198 static inline void fill_node_footer_blkaddr(struct page
*page
, block_t blkaddr
)
200 struct f2fs_sb_info
*sbi
= F2FS_SB(page
->mapping
->host
->i_sb
);
201 struct f2fs_checkpoint
*ckpt
= F2FS_CKPT(sbi
);
202 struct f2fs_node
*rn
= F2FS_NODE(page
);
204 rn
->footer
.cp_ver
= ckpt
->checkpoint_ver
;
205 rn
->footer
.next_blkaddr
= cpu_to_le32(blkaddr
);
208 static inline nid_t
ino_of_node(struct page
*node_page
)
210 struct f2fs_node
*rn
= F2FS_NODE(node_page
);
211 return le32_to_cpu(rn
->footer
.ino
);
214 static inline nid_t
nid_of_node(struct page
*node_page
)
216 struct f2fs_node
*rn
= F2FS_NODE(node_page
);
217 return le32_to_cpu(rn
->footer
.nid
);
220 static inline unsigned int ofs_of_node(struct page
*node_page
)
222 struct f2fs_node
*rn
= F2FS_NODE(node_page
);
223 unsigned flag
= le32_to_cpu(rn
->footer
.flag
);
224 return flag
>> OFFSET_BIT_SHIFT
;
227 static inline unsigned long long cpver_of_node(struct page
*node_page
)
229 struct f2fs_node
*rn
= F2FS_NODE(node_page
);
230 return le64_to_cpu(rn
->footer
.cp_ver
);
233 static inline block_t
next_blkaddr_of_node(struct page
*node_page
)
235 struct f2fs_node
*rn
= F2FS_NODE(node_page
);
236 return le32_to_cpu(rn
->footer
.next_blkaddr
);
240 * f2fs assigns the following node offsets described as (num).
246 * |- indirect node (3)
247 * | `- direct node (4 => 4 + N - 1)
248 * |- indirect node (4 + N)
249 * | `- direct node (5 + N => 5 + 2N - 1)
250 * `- double indirect node (5 + 2N)
251 * `- indirect node (6 + 2N)
254 * `- indirect node ((6 + 2N) + x(N + 1))
257 * `- indirect node ((6 + 2N) + (N - 1)(N + 1))
260 static inline bool IS_DNODE(struct page
*node_page
)
262 unsigned int ofs
= ofs_of_node(node_page
);
264 if (f2fs_has_xattr_block(ofs
))
267 if (ofs
== 3 || ofs
== 4 + NIDS_PER_BLOCK
||
268 ofs
== 5 + 2 * NIDS_PER_BLOCK
)
270 if (ofs
>= 6 + 2 * NIDS_PER_BLOCK
) {
271 ofs
-= 6 + 2 * NIDS_PER_BLOCK
;
272 if (!((long int)ofs
% (NIDS_PER_BLOCK
+ 1)))
278 static inline void set_nid(struct page
*p
, int off
, nid_t nid
, bool i
)
280 struct f2fs_node
*rn
= F2FS_NODE(p
);
282 f2fs_wait_on_page_writeback(p
, NODE
);
285 rn
->i
.i_nid
[off
- NODE_DIR1_BLOCK
] = cpu_to_le32(nid
);
287 rn
->in
.nid
[off
] = cpu_to_le32(nid
);
291 static inline nid_t
get_nid(struct page
*p
, int off
, bool i
)
293 struct f2fs_node
*rn
= F2FS_NODE(p
);
296 return le32_to_cpu(rn
->i
.i_nid
[off
- NODE_DIR1_BLOCK
]);
297 return le32_to_cpu(rn
->in
.nid
[off
]);
301 * Coldness identification:
302 * - Mark cold files in f2fs_inode_info
303 * - Mark cold node blocks in their node footer
304 * - Mark cold data pages in page cache
306 static inline int is_file(struct inode
*inode
, int type
)
308 return F2FS_I(inode
)->i_advise
& type
;
311 static inline void set_file(struct inode
*inode
, int type
)
313 F2FS_I(inode
)->i_advise
|= type
;
316 static inline void clear_file(struct inode
*inode
, int type
)
318 F2FS_I(inode
)->i_advise
&= ~type
;
321 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT)
322 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT)
323 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT)
324 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT)
325 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT)
326 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT)
328 static inline int is_cold_data(struct page
*page
)
330 return PageChecked(page
);
333 static inline void set_cold_data(struct page
*page
)
335 SetPageChecked(page
);
338 static inline void clear_cold_data(struct page
*page
)
340 ClearPageChecked(page
);
343 static inline int is_node(struct page
*page
, int type
)
345 struct f2fs_node
*rn
= F2FS_NODE(page
);
346 return le32_to_cpu(rn
->footer
.flag
) & (1 << type
);
349 #define is_cold_node(page) is_node(page, COLD_BIT_SHIFT)
350 #define is_fsync_dnode(page) is_node(page, FSYNC_BIT_SHIFT)
351 #define is_dent_dnode(page) is_node(page, DENT_BIT_SHIFT)
353 static inline void set_cold_node(struct inode
*inode
, struct page
*page
)
355 struct f2fs_node
*rn
= F2FS_NODE(page
);
356 unsigned int flag
= le32_to_cpu(rn
->footer
.flag
);
358 if (S_ISDIR(inode
->i_mode
))
359 flag
&= ~(0x1 << COLD_BIT_SHIFT
);
361 flag
|= (0x1 << COLD_BIT_SHIFT
);
362 rn
->footer
.flag
= cpu_to_le32(flag
);
365 static inline void set_mark(struct page
*page
, int mark
, int type
)
367 struct f2fs_node
*rn
= F2FS_NODE(page
);
368 unsigned int flag
= le32_to_cpu(rn
->footer
.flag
);
370 flag
|= (0x1 << type
);
372 flag
&= ~(0x1 << type
);
373 rn
->footer
.flag
= cpu_to_le32(flag
);
375 #define set_dentry_mark(page, mark) set_mark(page, mark, DENT_BIT_SHIFT)
376 #define set_fsync_mark(page, mark) set_mark(page, mark, FSYNC_BIT_SHIFT)