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,
67 struct wim_inode *inode;
68 const utf16lechar *stream_name;
71 /* Specification for a stream, which may be the contents of a file (unnamed data
72 * stream), a named data stream, reparse point data, or a WIM metadata resource.
74 * One instance of this structure is created for each entry in the WIM's lookup
75 * table, hence the name of the struct. Each of these entries contains the SHA1
76 * message digest of a stream and the location of the stream data in the WIM
77 * file (size, location, flags). The in-memory lookup table is a map from SHA1
78 * message digests to stream locations. */
79 struct wim_lookup_table_entry {
81 /* List node for a hash bucket of the lookup table. */
82 struct hlist_node hash_list;
84 /* Uncompressed size of this stream. */
87 /* Stream flags (WIM_RESHDR_FLAG_*). */
90 /* One of the `enum resource_location' values documented above. */
91 u32 resource_location : 4;
93 /* 1 if this stream has not had a SHA1 message digest calculated for it
97 /* Temoorary fields used when writing streams; set as documented for
98 * prepare_stream_list_for_write(). */
100 u32 will_be_in_output_wim : 1;
102 /* Set to 1 when a metadata entry has its checksum changed; in such
103 * cases the hash cannot be used to verify the data if the metadata
104 * resource is read again. (This could be avoided if we used separate
105 * fields for input/output checksum, but most stream entries wouldn't
107 u32 dont_check_metadata_hash : 1;
109 u32 may_send_done_with_file : 1;
111 /* Only used by wimlib_export_image() */
112 u32 was_exported : 1;
115 /* (On-disk field) SHA1 message digest of the stream referenced
116 * by this lookup table entry. */
117 u8 hash[SHA1_HASH_SIZE];
119 /* First 4 or 8 bytes of the SHA1 message digest, used for
120 * inserting the entry into the hash table. Since the SHA1
121 * message digest can be considered random, we don't really need
122 * the full 20 byte hash just to insert the entry in a hash
126 /* Unhashed entries only (unhashed == 1): these variables make
127 * it possible to find the pointer to this 'struct
128 * wim_lookup_table_entry' contained in either 'struct
129 * wim_ads_entry' or 'struct wim_inode'. There can be at most 1
130 * such pointer, as we can only join duplicate streams after
131 * they have been hashed. */
133 struct wim_inode *back_inode;
138 /* Number of times this lookup table entry is referenced by dentries in
139 * the WIM. When a WIM's lookup table is read, this field is
140 * initialized from a corresponding entry.
142 * However, see lte_decrement_refcnt() for information about the
143 * limitations of this field. */
146 /* When a WIM file is written, this is set to the number of references
147 * (by dentries) to this stream in the output WIM file.
149 * During extraction, this is the number of slots in stream_owners (or
150 * inline_stream_owners) that have been filled.
152 * During image export, this is set to the number of references of this
153 * stream that originated from the source WIM.
155 * When mounting a WIM image read-write, this is set to the number of
156 * extra references to this stream preemptively taken to allow later
157 * saving the modified image as a new image and leaving the original
162 /* Number of open file descriptors to this stream during a FUSE mount of
163 * the containing image. */
167 /* Specification of where this stream is actually located. Which member
168 * is valid is determined by the @resource_location field. */
171 struct wim_resource_spec *rspec;
176 struct wim_inode *file_inode;
178 void *attached_buffer;
181 char *staging_file_name;
186 struct ntfs_location *ntfs_loc;
190 /* Links together streams that share the same underlying WIM resource.
191 * The head is the `stream_list' member of `struct wim_resource_spec'.
193 struct list_head rspec_node;
195 /* Temporary fields */
197 /* Fields used temporarily during WIM file writing. */
200 /* List node used for stream size table. */
201 struct hlist_node hash_list_2;
203 /* Metadata for the underlying packed resource
204 * in the WIM being written (only valid if
205 * WIM_RESHDR_FLAG_PACKED_STREAMS set in
206 * out_reshdr.flags). */
208 u64 out_res_offset_in_wim;
209 u64 out_res_size_in_wim;
210 u64 out_res_uncompressed_size;
214 /* Links streams being written to the WIM. */
215 struct list_head write_streams_list;
217 /* Metadata for this stream in the WIM being written.
219 struct wim_reshdr out_reshdr;
222 /* Used temporarily during extraction. This is an array of
223 * pointers to the inodes being extracted that use this stream.
226 /* Inodes to extract that reference this stream.
227 * out_refcnt tracks the number of slots filled. */
228 struct stream_owner inline_stream_owners[3];
230 struct stream_owner *stream_owners;
231 u32 alloc_stream_owners;
236 /* Temporary list fields. */
238 /* Links streams for writing lookup table. */
239 struct list_head lookup_table_list;
241 /* Links streams being extracted. */
242 struct list_head extraction_list;
244 /* Links streams being exported. */
245 struct list_head export_stream_list;
247 /* Links original list of streams in the read-write mounted image. */
248 struct list_head orig_stream_list;
251 /* Links streams that are still unhashed after being been added to a
253 struct list_head unhashed_list;
256 /* Functions to allocate and free lookup tables */
258 extern struct wim_lookup_table *
259 new_lookup_table(size_t capacity) _malloc_attribute;
262 free_lookup_table(struct wim_lookup_table *table);
264 /* Functions to read or write the lookup table from/to a WIM file */
267 read_wim_lookup_table(WIMStruct *wim);
270 write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
271 struct filedes *out_fd,
273 struct wim_reshdr *out_reshdr,
274 int write_resource_flags);
276 /* Functions to create, clone, print, and free lookup table entries */
278 extern struct wim_lookup_table_entry *
279 new_lookup_table_entry(void) _malloc_attribute;
281 extern struct wim_lookup_table_entry *
282 clone_lookup_table_entry(const struct wim_lookup_table_entry *lte)
286 lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
287 struct wim_lookup_table *table);
290 lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte);
294 free_lookup_table_entry(struct wim_lookup_table_entry *lte);
296 /* Functions to insert and delete entries from a lookup table */
299 lookup_table_insert(struct wim_lookup_table *table,
300 struct wim_lookup_table_entry *lte);
303 lookup_table_unlink(struct wim_lookup_table *table,
304 struct wim_lookup_table_entry *lte);
306 /* Function to lookup a stream by SHA1 message digest */
307 extern struct wim_lookup_table_entry *
308 lookup_stream(const struct wim_lookup_table *table, const u8 hash[]);
310 /* Functions to iterate through the entries of a lookup table */
313 for_lookup_table_entry(struct wim_lookup_table *table,
314 int (*visitor)(struct wim_lookup_table_entry *, void *),
318 for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
319 int (*visitor)(struct wim_lookup_table_entry *,
325 /* Function to get a resource entry in stable format */
327 struct wimlib_resource_entry;
330 lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
331 struct wimlib_resource_entry *wentry);
333 /* Functions to sort a list of lookup table entries */
335 sort_stream_list(struct list_head *stream_list,
336 size_t list_head_offset,
337 int (*compar)(const void *, const void*));
340 sort_stream_list_by_sequential_order(struct list_head *stream_list,
341 size_t list_head_offset);
343 /* Utility functions */
346 lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *ignore);
349 lte_is_partial(const struct wim_lookup_table_entry * lte)
351 return lte->resource_location == RESOURCE_IN_WIM &&
352 lte->size != lte->rspec->uncompressed_size;
355 static inline const struct stream_owner *
356 stream_owners(struct wim_lookup_table_entry *stream)
358 if (stream->out_refcnt <= ARRAY_LEN(stream->inline_stream_owners))
359 return stream->inline_stream_owners;
361 return stream->stream_owners;
365 lte_bind_wim_resource_spec(struct wim_lookup_table_entry *lte,
366 struct wim_resource_spec *rspec)
368 lte->resource_location = RESOURCE_IN_WIM;
370 list_add_tail(<e->rspec_node, &rspec->stream_list);
374 lte_unbind_wim_resource_spec(struct wim_lookup_table_entry *lte)
376 list_del(<e->rspec_node);
377 lte->resource_location = RESOURCE_NONEXISTENT;
381 lte_put_resource(struct wim_lookup_table_entry *lte);
383 extern struct wim_lookup_table_entry *
384 new_stream_from_data_buffer(const void *buffer, size_t size,
385 struct wim_lookup_table *lookup_table);
388 add_unhashed_stream(struct wim_lookup_table_entry *lte,
389 struct wim_inode *back_inode,
391 struct list_head *unhashed_streams)
394 lte->back_inode = back_inode;
395 lte->back_stream_id = back_stream_id;
396 list_add_tail(<e->unhashed_list, unhashed_streams);
400 hash_unhashed_stream(struct wim_lookup_table_entry *lte,
401 struct wim_lookup_table *lookup_table,
402 struct wim_lookup_table_entry **lte_ret);
404 extern struct wim_lookup_table_entry **
405 retrieve_lte_pointer(struct wim_lookup_table_entry *lte);
407 #endif /* _WIMLIB_LOOKUP_TABLE_H */