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.
32 RESOURCE_IN_FILE_ON_DISK,
34 /* The stream is directly attached in the in-memory buffer pointed to by
35 * @attached_buffer. */
36 RESOURCE_IN_ATTACHED_BUFFER,
39 /* The stream is located in the external file named by
40 * @staging_file_name, located in the staging directory for a read-write
42 RESOURCE_IN_STAGING_FILE,
46 /* The stream is located in an NTFS volume. It is identified by volume,
47 * filename, data stream name, and by whether it is a reparse point or
48 * not. @ntfs_loc points to a structure containing this information.
50 RESOURCE_IN_NTFS_VOLUME,
54 /* Windows only: the stream is located in the external file named by
55 * @file_on_disk, which is in the Windows NT namespace and may specify a
56 * named data stream. */
57 RESOURCE_IN_WINNT_FILE_ON_DISK,
59 /* Windows only: the stream is located in the external file named by
60 * @file_on_disk, but the file is encrypted and must be read using the
61 * appropriate Windows API. */
62 RESOURCE_WIN32_ENCRYPTED,
66 /* Specification for a stream, which may be the contents of a file (unnamed data
67 * stream), a named data stream, reparse point data, or a WIM metadata resource.
69 * One instance of this structure is created for each entry in the WIM's lookup
70 * table, hence the name of the struct. Each of these entries contains the SHA1
71 * message digest of a stream and the location of the stream data in the WIM
72 * file (size, location, flags). The in-memory lookup table is a map from SHA1
73 * message digests to stream locations. */
74 struct wim_lookup_table_entry {
76 /* List node for a hash bucket of the lookup table. */
77 struct hlist_node hash_list;
79 /* Uncompressed size of this stream. */
82 /* Stream flags (WIM_RESHDR_FLAG_*). */
85 /* One of the `enum resource_location' values documented above. */
86 u32 resource_location : 4;
88 /* 1 if this stream has not had a SHA1 message digest calculated for it
92 /* Temoorary fields used when writing streams; set as documented for
93 * prepare_stream_list_for_write(). */
95 u32 will_be_in_output_wim : 1;
97 /* Set to 1 when a metadata entry has its checksum changed; in such
98 * cases the hash cannot be used to verify the data if the metadata
99 * resource is read again. (This could be avoided if we used separate
100 * fields for input/output checksum, but most stream entries wouldn't
102 u32 dont_check_metadata_hash : 1;
105 /* (On-disk field) SHA1 message digest of the stream referenced
106 * by this lookup table entry. */
107 u8 hash[SHA1_HASH_SIZE];
109 /* First 4 or 8 bytes of the SHA1 message digest, used for
110 * inserting the entry into the hash table. Since the SHA1
111 * message digest can be considered random, we don't really need
112 * the full 20 byte hash just to insert the entry in a hash
116 /* Unhashed entries only (unhashed == 1): these variables make
117 * it possible to find the pointer to this 'struct
118 * wim_lookup_table_entry' contained in either 'struct
119 * wim_ads_entry' or 'struct wim_inode'. There can be at most 1
120 * such pointer, as we can only join duplicate streams after
121 * they have been hashed. */
123 struct wim_inode *back_inode;
128 /* Number of times this lookup table entry is referenced by dentries in
129 * the WIM. When a WIM's lookup table is read, this field is
130 * initialized from a corresponding entry; while it should be correct,
131 * in general it may not be. wim_recalculate_refcnts() recalculates the
132 * reference counts for all streams and is run before doing any
136 /* When a WIM file is written, this is set to the number of references
137 * (by dentries) to this stream in the output WIM file.
139 * During extraction, this is set to the number of times the stream must
142 * During image export, this is set to the number of references of this
143 * stream that originated from the source WIM.
145 * When mounting a WIM image read-write, this is set to the number of
146 * extra references to this stream preemptively taken to allow later
147 * saving the modified image as a new image and leaving the original
152 /* Number of open file descriptors to this stream during a FUSE mount of
153 * the containing image. */
157 /* Specification of where this stream is actually located. Which member
158 * is valid is determined by the @resource_location field. */
161 struct wim_resource_spec *rspec;
165 void *attached_buffer;
167 tchar *staging_file_name;
170 struct ntfs_location *ntfs_loc;
174 /* Links together streams that share the same underlying WIM resource.
175 * The head is the `stream_list' member of `struct wim_resource_spec'.
177 struct list_head rspec_node;
179 /* This field is used during the hardlink and symlink image extraction
180 * modes. In these modes, all identical files are linked together, and
181 * @extracted_file will be set to the filename of the first extracted
182 * file containing this stream. */
183 tchar *extracted_file;
185 /* Temporary fields */
187 /* Fields used temporarily during WIM file writing. */
190 /* List node used for stream size table. */
191 struct hlist_node hash_list_2;
193 /* Metadata for the underlying packed resource
194 * in the WIM being written (only valid if
195 * WIM_RESHDR_FLAG_PACKED_STREAMS set in
196 * out_reshdr.flags). */
198 u64 out_res_offset_in_wim;
199 u64 out_res_size_in_wim;
200 u64 out_res_uncompressed_size;
204 /* Links streams being written to the WIM. */
205 struct list_head write_streams_list;
207 /* Metadata for this stream in the WIM being written.
209 struct wim_reshdr out_reshdr;
212 /* Used temporarily during extraction */
214 /* Dentries to extract that reference this stream.
215 * out_refcnt tracks the number of slots filled. */
216 struct wim_dentry *inline_lte_dentries[7];
218 struct wim_dentry **lte_dentries;
219 size_t alloc_lte_dentries;
223 /* Actual reference count to this stream (only used while
224 * verifying an image). */
228 /* Temporary list fields. */
230 /* Links streams for writing lookup table. */
231 struct list_head lookup_table_list;
233 /* Links streams being extracted. */
234 struct list_head extraction_list;
236 /* Links streams being exported. */
237 struct list_head export_stream_list;
239 /* Links original list of streams in the read-write mounted image. */
240 struct list_head orig_stream_list;
243 /* Links streams that are still unhashed after being been added to a
245 struct list_head unhashed_list;
248 /* Functions to allocate and free lookup tables */
250 extern struct wim_lookup_table *
251 new_lookup_table(size_t capacity) _malloc_attribute;
254 free_lookup_table(struct wim_lookup_table *table);
256 /* Functions to read or write the lookup table from/to a WIM file */
259 read_wim_lookup_table(WIMStruct *wim);
262 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
263 struct filedes *out_fd,
265 struct wim_reshdr *out_reshdr,
266 int write_resource_flags);
268 /* Functions to create, clone, print, and free lookup table entries */
270 extern struct wim_lookup_table_entry *
271 new_lookup_table_entry(void) _malloc_attribute;
273 extern struct wim_lookup_table_entry *
274 clone_lookup_table_entry(const struct wim_lookup_table_entry *lte)
278 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
279 struct wim_lookup_table *table);
282 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte);
286 free_lookup_table_entry(struct wim_lookup_table_entry *lte);
288 /* Functions to insert and delete entries from a lookup table */
291 lookup_table_insert(struct wim_lookup_table *table,
292 struct wim_lookup_table_entry *lte);
295 lookup_table_unlink(struct wim_lookup_table *table,
296 struct wim_lookup_table_entry *lte);
298 /* Function to lookup a stream by SHA1 message digest */
299 extern struct wim_lookup_table_entry *
300 lookup_stream(const struct wim_lookup_table *table, const u8 hash[]);
302 /* Functions to iterate through the entries of a lookup table */
305 for_lookup_table_entry(struct wim_lookup_table *table,
306 int (*visitor)(struct wim_lookup_table_entry *, void *),
310 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
311 int (*visitor)(struct wim_lookup_table_entry *,
317 /* Function to get a resource entry in stable format */
319 struct wimlib_resource_entry;
322 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
323 struct wimlib_resource_entry *wentry);
325 /* Functions to sort a list of lookup table entries */
327 sort_stream_list(struct list_head *stream_list,
328 size_t list_head_offset,
329 int (*compar)(const void *, const void*));
332 sort_stream_list_by_sequential_order(struct list_head *stream_list,
333 size_t list_head_offset);
335 /* Utility functions */
338 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *ignore);
341 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *ignore);
344 lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore);
347 lte_is_partial(const struct wim_lookup_table_entry * lte)
349 return lte->resource_location == RESOURCE_IN_WIM &&
350 lte->size != lte->rspec->uncompressed_size;
354 lte_filename_valid(const struct wim_lookup_table_entry *lte)
356 return lte->resource_location == RESOURCE_IN_FILE_ON_DISK
358 || lte->resource_location == RESOURCE_WIN32_ENCRYPTED
361 || lte->resource_location == RESOURCE_IN_STAGING_FILE
367 lte_bind_wim_resource_spec(struct wim_lookup_table_entry *lte,
368 struct wim_resource_spec *rspec)
370 lte->resource_location = RESOURCE_IN_WIM;
372 list_add_tail(<e->rspec_node, &rspec->stream_list);
376 lte_unbind_wim_resource_spec(struct wim_lookup_table_entry *lte)
378 list_del(<e->rspec_node);
379 lte->resource_location = RESOURCE_NONEXISTENT;
383 lte_put_resource(struct wim_lookup_table_entry *lte);
385 extern struct wim_lookup_table_entry *
386 new_stream_from_data_buffer(const void *buffer, size_t size,
387 struct wim_lookup_table *lookup_table);
390 add_unhashed_stream(struct wim_lookup_table_entry *lte,
391 struct wim_inode *back_inode,
393 struct list_head *unhashed_streams)
396 lte->back_inode = back_inode;
397 lte->back_stream_id = back_stream_id;
398 list_add_tail(<e->unhashed_list, unhashed_streams);
402 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
403 struct wim_lookup_table *lookup_table,
404 struct wim_lookup_table_entry **lte_ret);
406 extern struct wim_lookup_table_entry **
407 retrieve_lte_pointer(struct wim_lookup_table_entry *lte);
409 #endif /* _WIMLIB_LOOKUP_TABLE_H */