1 #ifndef _WIMLIB_LOOKUP_TABLE_H
2 #define _WIMLIB_LOOKUP_TABLE_H
3 #include "wimlib_internal.h"
8 /* Size of each lookup table entry in the WIM file. */
9 #define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
11 #define LOOKUP_FLAG_ADS_OK 0x00000001
12 #define LOOKUP_FLAG_DIRECTORY_OK 0x00000002
15 //#define LOOKUP_FLAG_FOLLOW_SYMLINKS 0x00000004
18 /* A lookup table that is used to translate the hash codes of dentries into the
19 * offsets and sizes of uncompressed or compressed file resources. It is
20 * implemented as a hash table. */
22 struct hlist_head *array;
30 struct ntfs_location {
32 char *stream_name_utf16;
33 u16 stream_name_utf16_num_chars;
34 struct _ntfs_volume **ntfs_vol_p;
35 bool is_reparse_point;
40 * An entry in the lookup table in the WIM file.
42 * It is used to find data streams for files in the WIM.
44 * Metadata resources and reparse point data buffers will also have lookup table
45 * entries associated with the data.
47 * The lookup_table_entry for a given dentry or alternate stream entry in the
48 * WIM is found using the SHA1 message digest field.
50 struct lookup_table_entry {
52 /* List of lookup table entries in this hash bucket */
53 struct hlist_node hash_list;
55 /* Location and size of the stream in the WIM, whether it is compressed
56 * or not, and whether it's a metadata resource or not. This is an
58 struct resource_entry resource_entry;
60 /* Specifies which part of the split WIM the resource is located in.
61 * This is on on-disk field.
63 * In stand-alone WIMs, this must be 1.
65 * In split WIMs, every split WIM part has its own lookup table, and in
66 * read_lookup_table() it's currently expected that the part number of
67 * each lookup table entry in a split WIM part's lookup table is the
68 * same as the part number of that split WIM part. So this makes this
69 * field redundant since we store a pointer to the corresponding
70 * WIMStruct in the lookup table entry anyway.
74 /* An enumerated type that identifies where the stream corresponding to
75 * this lookup table entry is actually located.
77 * Obviously if we open a WIM and read its lookup table, the location is
78 * set to RESOURCE_IN_WIM since all the streams will initially be
79 * located in the WIM. However, to deal with problems such as image
80 * capture and image mount, we allow the actual location of the stream
81 * to be somewhere else, such as an external file.
84 /* The lookup table entry does not correspond to a stream (this
85 * state should exist only temporarily) */
86 RESOURCE_NONEXISTENT = 0,
88 /* The stream resource is located in a WIM file. The WIMStruct
89 * for the WIM file will be pointed to by the @wim member. */
92 /* The stream resource is located in an external file. The
93 * name of the file will be provided by @file_on_disk member.
94 * In addition, if @file_on_disk_fp is not NULL, it will be an
95 * open FILE * to the file. */
96 RESOURCE_IN_FILE_ON_DISK,
98 /* The stream resource is located in an external file in the
99 * staging directory for a read-write mount. */
100 RESOURCE_IN_STAGING_FILE,
102 /* The stream resource is directly attached in an in-memory
103 * buffer pointed to by @attached_buffer. */
104 RESOURCE_IN_ATTACHED_BUFFER,
106 /* The stream resource is located in an NTFS volume. It is
107 * identified by volume, filename, data stream name, and by
108 * whether it is a reparse point or not. @ntfs_loc points to a
109 * structure containing this information. */
110 RESOURCE_IN_NTFS_VOLUME,
114 * Number of times this lookup table entry is referenced by dentries.
115 * Unfortunately, this field is not always set correctly in Microsoft's
116 * WIMs, so we have no choice but to fix it if more references to the
117 * lookup table entry are found than stated here. */
121 /* (On-disk field) SHA1 message digest of the stream referenced
122 * by this lookup table entry */
123 u8 hash[SHA1_HASH_SIZE];
125 /* First 4 or 8 bytes of the SHA1 message digest, used for
126 * inserting the entry into the hash table. Since the SHA1
127 * message digest can be considered random, we don't really need
128 * the full 20 byte hash just to insert the entry in a hash
133 /* Pointers to somewhere where the stream is actually located. See the
134 * comments for the @resource_location field above. */
138 char *staging_file_name;
141 struct ntfs_location *ntfs_loc;
145 /* @file_on_disk_fp and @attr are both used to cache file/stream
146 * handles so we don't have re-open them on every read */
148 /* Valid iff resource_location == RESOURCE_IN_FILE_ON_DISK */
149 FILE *file_on_disk_fp;
151 /* Valid iff resource_location == RESOURCE_IN_NTFS_VOLUME */
152 struct _ntfs_attr *attr;
155 /* Pointer to inode that contains the opened file descriptors to
156 * this stream (valid iff resource_location ==
157 * RESOURCE_IN_STAGING_FILE) */
158 struct inode *lte_inode;
164 /* When a WIM file is written, out_refcnt starts at 0 and is incremented
165 * whenever the file resource pointed to by this lookup table entry
166 * needs to be written. The file resource only need to be written when
167 * out_refcnt is nonzero, since otherwise it is not referenced by any
173 /* When a WIM file is written, @output_resource_entry is filled
174 * in with the resource entry for the output WIM. This will not
175 * necessarily be the same as the @resource_entry since:
176 * - The stream may have a different offset in the new WIM
177 * - The stream may have a different compressed size in the
178 * new WIM if the compression type changed
180 struct resource_entry output_resource_entry;
182 /* This field is used for the special hardlink or symlink image
183 * extraction mode. In these mode, all identical files are linked
184 * together, and @extracted_file will be set to the filename of the
185 * first extracted file containing this stream. */
186 char *extracted_file;
189 /* List of lookup table entries that correspond to streams that have
190 * been extracted to the staging directory when modifying a read-write
192 struct list_head staging_list;
194 /* Temporary field for creating a singly linked list. */
195 struct lookup_table_entry *next_lte_in_swm;
199 static inline u64 wim_resource_size(const struct lookup_table_entry *lte)
201 return lte->resource_entry.original_size;
205 wim_resource_compressed_size(const struct lookup_table_entry *lte)
207 return lte->resource_entry.size;
211 * XXX Probably should store the compression type directly in the lookup table
215 wim_resource_compression_type(const struct lookup_table_entry *lte)
217 if (!(lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
218 || lte->resource_location != RESOURCE_IN_WIM)
219 return WIM_COMPRESSION_TYPE_NONE;
220 return wimlib_get_compression_type(lte->wim);
224 extern struct lookup_table *new_lookup_table(size_t capacity);
226 extern void lookup_table_insert(struct lookup_table *table,
227 struct lookup_table_entry *lte);
229 /* Unlinks a lookup table entry from the table; does not free it. */
230 static inline void lookup_table_unlink(struct lookup_table *table,
231 struct lookup_table_entry *lte)
233 hlist_del(<e->hash_list);
234 table->num_entries--;
238 extern struct lookup_table_entry *
239 lookup_table_decrement_refcnt(struct lookup_table* table, const u8 hash[]);
242 extern struct lookup_table_entry *
243 lte_decrement_num_opened_fds(struct lookup_table_entry *lte,
244 struct lookup_table *table);
247 extern struct lookup_table_entry *
248 lte_decrement_refcnt(struct lookup_table_entry *lte,
249 struct lookup_table *table);
252 extern struct lookup_table_entry *new_lookup_table_entry();
254 extern int for_lookup_table_entry(struct lookup_table *table,
255 int (*visitor)(struct lookup_table_entry *, void *),
258 extern struct lookup_table_entry *
259 __lookup_resource(const struct lookup_table *table, const u8 hash[]);
261 extern int lookup_resource(WIMStruct *w, const char *path,
262 int lookup_flags, struct dentry **dentry_ret,
263 struct lookup_table_entry **lte_ret,
264 u16 *stream_idx_ret);
266 extern int lte_zero_out_refcnt(struct lookup_table_entry *entry, void *ignore);
267 extern int lte_zero_real_refcnt(struct lookup_table_entry *entry, void *ignore);
268 extern int lte_free_extracted_file(struct lookup_table_entry *lte, void *ignone);
270 extern void print_lookup_table_entry(const struct lookup_table_entry *entry);
272 extern int read_lookup_table(WIMStruct *w);
274 extern void free_lookup_table(struct lookup_table *table);
276 extern int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out);
278 extern void free_lookup_table_entry(struct lookup_table_entry *lte);
280 extern int dentry_resolve_ltes(struct dentry *dentry, void *__table);
282 /* Writes the lookup table to the output file. */
283 static inline int write_lookup_table(struct lookup_table *table, FILE *out)
285 return for_lookup_table_entry(table, write_lookup_table_entry, out);
288 /* Unlinks and frees an entry from a lookup table. */
289 static inline void lookup_table_remove(struct lookup_table *table,
290 struct lookup_table_entry *lte)
292 lookup_table_unlink(table, lte);
293 free_lookup_table_entry(lte);
296 static inline struct resource_entry* wim_metadata_resource_entry(WIMStruct *w)
298 return &w->image_metadata[
299 w->current_image - 1].metadata_lte->resource_entry;
302 static inline struct lookup_table_entry *
303 inode_stream_lte_resolved(const struct inode *inode, unsigned stream_idx)
305 wimlib_assert(inode->resolved);
306 wimlib_assert(stream_idx <= inode->num_ads);
310 return inode->ads_entries[stream_idx - 1].lte;
313 static inline struct lookup_table_entry *
314 inode_stream_lte_unresolved(const struct inode *inode, unsigned stream_idx,
315 const struct lookup_table *table)
317 wimlib_assert(!inode->resolved);
318 wimlib_assert(stream_idx <= inode->num_ads);
322 return __lookup_resource(table, inode->hash);
324 return __lookup_resource(table,
326 stream_idx - 1].hash);
329 * Returns the lookup table entry for stream @stream_idx of the inode, where
330 * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
331 * corresponds to an alternate data stream.
333 * This works for both resolved and un-resolved dentries.
335 static inline struct lookup_table_entry *
336 inode_stream_lte(const struct inode *inode, unsigned stream_idx,
337 const struct lookup_table *table)
340 return inode_stream_lte_resolved(inode, stream_idx);
342 return inode_stream_lte_unresolved(inode, stream_idx, table);
346 static inline const u8 *inode_stream_hash_unresolved(const struct inode *inode,
349 wimlib_assert(!inode->resolved);
350 wimlib_assert(stream_idx <= inode->num_ads);
354 return inode->ads_entries[stream_idx - 1].hash;
358 static inline const u8 *inode_stream_hash_resolved(const struct inode *inode,
361 struct lookup_table_entry *lte;
362 lte = inode_stream_lte_resolved(inode, stream_idx);
370 * Returns the hash for stream @stream_idx of the inode, where stream_idx = 0
371 * means the default un-named file stream, and stream_idx >= 1 corresponds to an
372 * alternate data stream.
374 * This works for both resolved and un-resolved dentries.
376 static inline const u8 *inode_stream_hash(const struct inode *inode,
380 return inode_stream_hash_resolved(inode, stream_idx);
382 return inode_stream_hash_unresolved(inode, stream_idx);
385 static inline u16 inode_stream_name_len(const struct inode *inode,
388 wimlib_assert(stream_idx <= inode->num_ads);
392 return inode->ads_entries[stream_idx - 1].stream_name_len;
395 static inline struct lookup_table_entry *
396 inode_unnamed_lte_resolved(const struct inode *inode)
398 wimlib_assert(inode->resolved);
399 for (unsigned i = 0; i <= inode->num_ads; i++)
400 if (inode_stream_name_len(inode, i) == 0 &&
401 !is_zero_hash(inode_stream_hash_resolved(inode, i)))
402 return inode_stream_lte_resolved(inode, i);
406 static inline struct lookup_table_entry *
407 inode_unnamed_lte_unresolved(const struct inode *inode,
408 const struct lookup_table *table)
410 wimlib_assert(!inode->resolved);
411 for (unsigned i = 0; i <= inode->num_ads; i++)
412 if (inode_stream_name_len(inode, i) == 0 &&
413 !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
414 return inode_stream_lte_unresolved(inode, i, table);
418 extern struct lookup_table_entry *
419 inode_unnamed_lte(const struct inode *inode,
420 const struct lookup_table *table);