1 #ifndef _WIMLIB_LOOKUP_TABLE_H
2 #define _WIMLIB_LOOKUP_TABLE_H
4 #include "wimlib/list.h"
5 #include "wimlib/resource.h"
6 #include "wimlib/sha1.h"
7 #include "wimlib/types.h"
9 /* An enumerated type that identifies where the stream corresponding to this
10 * lookup table entry is actually located.
12 * If we open a WIM and read its lookup table, the location is set to
13 * RESOURCE_IN_WIM since all the streams will initially be located in the WIM.
14 * However, to handle situations such as image capture and image mount, we allow
15 * the actual location of the stream to be somewhere else, such as an external
17 enum resource_location {
18 /* The lookup table entry does not yet correspond to a stream; this is a
19 * temporary state only. */
20 RESOURCE_NONEXISTENT = 0,
22 /* The stream is located in a resource in a WIM file identified by the
23 * `struct wim_resource_spec' pointed to by @rspec. @offset_in_res
24 * identifies the offset at which this particular stream begins in the
25 * uncompressed data of the resource; this is normally 0, but in general
26 * a WIM resource may be "packed" and potentially contain multiple
30 /* The stream is located in the external file named by @file_on_disk.
31 * On Windows, @file_on_disk may actually specify a named data stream
32 * (file path, then colon, then name of the stream). */
33 RESOURCE_IN_FILE_ON_DISK,
35 /* The stream is directly attached in the in-memory buffer pointed to by
36 * @attached_buffer. */
37 RESOURCE_IN_ATTACHED_BUFFER,
40 /* The stream is located in the external file named by
41 * @staging_file_name, located in the staging directory for a read-write
43 RESOURCE_IN_STAGING_FILE,
47 /* The stream is located in an NTFS volume. It is identified by volume,
48 * filename, data stream name, and by whether it is a reparse point or
49 * not. @ntfs_loc points to a structure containing this information.
51 RESOURCE_IN_NTFS_VOLUME,
55 /* Windows only: the stream is located in the external file named by
56 * @file_on_disk, but the file is encrypted and must be read using the
57 * appropriate Windows API. */
58 RESOURCE_WIN32_ENCRYPTED,
62 /* Specification for a stream, which may be the contents of a file (unnamed data
63 * stream), a named data stream, reparse point data, or a WIM metadata resource.
65 * One instance of this structure is created for each entry in the WIM's lookup
66 * table, hence the name of the struct. Each of these entries contains the SHA1
67 * message digest of a stream and the location of the stream data in the WIM
68 * file (size, location, flags). The in-memory lookup table is a map from SHA1
69 * message digests to stream locations. */
70 struct wim_lookup_table_entry {
72 /* List node for a hash bucket of the lookup table. */
73 struct hlist_node hash_list;
75 /* Uncompressed size of this stream. */
78 /* Stream flags (WIM_RESHDR_FLAG_*). */
81 /* One of the `enum resource_location' values documented above. */
82 u32 resource_location : 4;
84 /* 1 if this stream has not had a SHA1 message digest calculated for it
88 /* Temoorary fields used when writing streams; set as documented for
89 * prepare_stream_list_for_write(). */
91 u32 will_be_in_output_wim : 1;
93 /* Set to 1 when a metadata entry has its checksum changed; in such
94 * cases the hash cannot be used to verify the data if the metadata
95 * resource is read again. (This could be avoided if we used separate
96 * fields for input/output checksum, but most stream entries wouldn't
98 u32 dont_check_metadata_hash : 1;
101 /* (On-disk field) SHA1 message digest of the stream referenced
102 * by this lookup table entry. */
103 u8 hash[SHA1_HASH_SIZE];
105 /* First 4 or 8 bytes of the SHA1 message digest, used for
106 * inserting the entry into the hash table. Since the SHA1
107 * message digest can be considered random, we don't really need
108 * the full 20 byte hash just to insert the entry in a hash
112 /* Unhashed entries only (unhashed == 1): these variables make
113 * it possible to find the pointer to this 'struct
114 * wim_lookup_table_entry' contained in either 'struct
115 * wim_ads_entry' or 'struct wim_inode'. There can be at most 1
116 * such pointer, as we can only join duplicate streams after
117 * they have been hashed. */
119 struct wim_inode *back_inode;
124 /* Number of times this lookup table entry is referenced by dentries in
125 * the WIM. When a WIM's lookup table is read, this field is
126 * initialized from a corresponding entry; while it should be correct,
127 * in general it may not be. wim_recalculate_refcnts() recalculates the
128 * reference counts for all streams and is run before doing any
132 /* When a WIM file is written, this is set to the number of references
133 * (by dentries) to this stream in the output WIM file.
135 * During extraction, this is set to the number of times the stream must
138 * During image export, this is set to the number of references of this
139 * stream that originated from the source WIM.
141 * When mounting a WIM image read-write, this is set to the number of
142 * extra references to this stream preemptively taken to allow later
143 * saving the modified image as a new image and leaving the original
148 /* Number of open file descriptors to this stream during a FUSE mount of
149 * the containing image. */
153 /* Specification of where this stream is actually located. Which member
154 * is valid is determined by the @resource_location field. */
157 struct wim_resource_spec *rspec;
161 void *attached_buffer;
163 tchar *staging_file_name;
166 struct ntfs_location *ntfs_loc;
170 /* Links together streams that share the same underlying WIM resource.
171 * The head is the `stream_list' member of `struct wim_resource_spec'.
173 struct list_head rspec_node;
175 /* This field is used during the hardlink and symlink image extraction
176 * modes. In these modes, all identical files are linked together, and
177 * @extracted_file will be set to the filename of the first extracted
178 * file containing this stream. */
179 tchar *extracted_file;
181 /* Temporary fields */
183 /* Fields used temporarily during WIM file writing. */
186 /* List node used for stream size table. */
187 struct hlist_node hash_list_2;
189 /* Metadata for the underlying packed resource
190 * in the WIM being written (only valid if
191 * WIM_RESHDR_FLAG_PACKED_STREAMS set in
192 * out_reshdr.flags). */
194 u64 out_res_offset_in_wim;
195 u64 out_res_size_in_wim;
199 /* Links streams being written to the WIM. */
200 struct list_head write_streams_list;
202 /* Metadata for this stream in the WIM being written.
204 struct wim_reshdr out_reshdr;
207 /* Used temporarily during extraction */
209 /* Dentries to extract that reference this stream.
210 * out_refcnt tracks the number of slots filled. */
211 struct wim_dentry *inline_lte_dentries[7];
213 struct wim_dentry **lte_dentries;
214 size_t alloc_lte_dentries;
218 /* Actual reference count to this stream (only used while
219 * verifying an image). */
223 /* Temporary list fields. */
225 /* Links streams for writing lookup table. */
226 struct list_head lookup_table_list;
228 /* Links streams being extracted. */
229 struct list_head extraction_list;
231 /* Links streams being exported. */
232 struct list_head export_stream_list;
234 /* Links original list of streams in the read-write mounted image. */
235 struct list_head orig_stream_list;
238 /* Links streams that are still unhashed after being been added to a
240 struct list_head unhashed_list;
243 /* Functions to allocate and free lookup tables */
245 extern struct wim_lookup_table *
246 new_lookup_table(size_t capacity) _malloc_attribute;
249 free_lookup_table(struct wim_lookup_table *table);
251 /* Functions to read or write the lookup table from/to a WIM file */
254 read_wim_lookup_table(WIMStruct *wim);
257 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
258 struct filedes *out_fd,
260 struct wim_reshdr *out_reshdr,
261 int write_resource_flags);
263 /* Functions to create, clone, print, and free lookup table entries */
265 extern struct wim_lookup_table_entry *
266 new_lookup_table_entry(void) _malloc_attribute;
268 extern struct wim_lookup_table_entry *
269 clone_lookup_table_entry(const struct wim_lookup_table_entry *lte)
273 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
274 struct wim_lookup_table *table);
277 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte);
281 free_lookup_table_entry(struct wim_lookup_table_entry *lte);
283 /* Functions to insert and delete entries from a lookup table */
286 lookup_table_insert(struct wim_lookup_table *table,
287 struct wim_lookup_table_entry *lte);
290 lookup_table_unlink(struct wim_lookup_table *table,
291 struct wim_lookup_table_entry *lte);
293 /* Function to lookup a stream by SHA1 message digest */
294 extern struct wim_lookup_table_entry *
295 lookup_stream(const struct wim_lookup_table *table, const u8 hash[]);
297 /* Functions to iterate through the entries of a lookup table */
300 for_lookup_table_entry(struct wim_lookup_table *table,
301 int (*visitor)(struct wim_lookup_table_entry *, void *),
305 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
306 int (*visitor)(struct wim_lookup_table_entry *,
312 /* Function to get a resource entry in stable format */
314 struct wimlib_resource_entry;
317 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
318 struct wimlib_resource_entry *wentry);
320 /* Functions to sort a list of lookup table entries */
322 sort_stream_list(struct list_head *stream_list,
323 size_t list_head_offset,
324 int (*compar)(const void *, const void*));
327 sort_stream_list_by_sequential_order(struct list_head *stream_list,
328 size_t list_head_offset);
330 /* Utility functions */
333 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *ignore);
336 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *ignore);
339 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore);
342 lte_is_partial(const struct wim_lookup_table_entry * lte)
344 return lte->resource_location == RESOURCE_IN_WIM &&
345 lte->size != lte->rspec->uncompressed_size;
349 lte_filename_valid(const struct wim_lookup_table_entry *lte)
351 return lte->resource_location == RESOURCE_IN_FILE_ON_DISK
353 || lte->resource_location == RESOURCE_WIN32_ENCRYPTED
356 || lte->resource_location == RESOURCE_IN_STAGING_FILE
362 lte_bind_wim_resource_spec(struct wim_lookup_table_entry *lte,
363 struct wim_resource_spec *rspec)
365 lte->resource_location = RESOURCE_IN_WIM;
367 list_add_tail(<e->rspec_node, &rspec->stream_list);
371 lte_unbind_wim_resource_spec(struct wim_lookup_table_entry *lte)
373 list_del(<e->rspec_node);
374 lte->resource_location = RESOURCE_NONEXISTENT;
378 lte_put_resource(struct wim_lookup_table_entry *lte);
380 extern struct wim_lookup_table_entry *
381 new_stream_from_data_buffer(const void *buffer, size_t size,
382 struct wim_lookup_table *lookup_table);
385 add_unhashed_stream(struct wim_lookup_table_entry *lte,
386 struct wim_inode *back_inode,
388 struct list_head *unhashed_streams)
391 lte->back_inode = back_inode;
392 lte->back_stream_id = back_stream_id;
393 list_add_tail(<e->unhashed_list, unhashed_streams);
397 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
398 struct wim_lookup_table *lookup_table,
399 struct wim_lookup_table_entry **lte_ret);
401 extern struct wim_lookup_table_entry **
402 retrieve_lte_pointer(struct wim_lookup_table_entry *lte);
404 #endif /* _WIMLIB_LOOKUP_TABLE_H */