1 #ifndef _WIMLIB_BLOB_TABLE_H
2 #define _WIMLIB_BLOB_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 a blob's data is located. */
12 /* The blob's data does not exist. This is a temporary state only. */
15 /* The blob's data is available in the WIM resource identified by the
16 * `struct wim_resource_descriptor' pointed to by @rdesc.
17 * @offset_in_res identifies the offset at which this particular blob
18 * begins in the uncompressed data of the resource. */
21 /* The blob's data is available as the contents of the file named by
25 /* The blob's data is available as the contents of the in-memory buffer
26 * pointed to by @attached_buffer. */
27 BLOB_IN_ATTACHED_BUFFER,
30 /* The blob's data is available as the contents of the file with name
31 * @staging_file_name relative to the open directory file descriptor
37 /* The blob's data is available as the contents of an NTFS attribute
38 * accessible through libntfs-3g. The attribute is identified by
39 * volume, path to an inode, attribute name, and attribute type.
40 * @ntfs_loc points to a structure containing this information. */
45 /* Windows only: the blob's data is available as the contents of the
46 * data stream named by @file_on_disk. @file_on_disk is an NT namespace
47 * path that may be longer than the Win32-level MAX_PATH. Furthermore,
48 * the stream may require "backup semantics" to access. */
49 BLOB_IN_WINNT_FILE_ON_DISK,
51 /* Windows only: the blob's data is available as the raw encrypted data
52 * of the external file named by @file_on_disk. @file_on_disk is a
53 * Win32 namespace path. */
58 /* A "blob extraction target" is a stream, and the inode to which that stream
59 * belongs, to which a blob needs to be extracted as part of an extraction
60 * operation. Since blobs are single-instanced, a blob may have multiple
61 * extraction targets. */
62 struct blob_extraction_target {
63 struct wim_inode *inode;
64 struct wim_inode_stream *stream;
68 * Descriptor for a "blob", which is a known length sequence of binary data.
70 * Within a WIM file, blobs are single instanced and are identified by SHA-1
73 struct blob_descriptor {
75 /* List node for a hash bucket of the blob table */
76 struct hlist_node hash_list;
78 /* Uncompressed size of this blob */
83 * For unhashed == 0: 'hash' is the SHA-1 message digest of the
84 * blob's data. 'hash_short' allows accessing just a prefix of
85 * the SHA-1 message digest, which is useful for getting a "hash
86 * code" for hash table lookup/insertion.
88 u8 hash[SHA1_HASH_SIZE];
91 /* For unhashed == 1: these variables make it possible to find
92 * the stream that references this blob. There can be at most
93 * one such reference, since duplicate blobs can only be joined
94 * after they have been hashed. */
96 struct wim_inode *back_inode;
101 /* Number of times this blob is referenced by file streams in WIM
102 * images. See blob_decrement_refcnt() for information about the
103 * limitations of this field. */
107 * When a WIM file is written, this is set to the number of references
108 * (from file streams) to this blob in the output WIM file.
110 * During extraction, this is set to the number of targets to which this
111 * blob is being extracted.
113 * During image export, this is set to the number of references of this
114 * blob that originated from the source WIM.
116 * When mounting a WIM image read-write, this is set to the number of
117 * extra references to this blob preemptively taken to allow later
118 * saving the modified image as a new image and leaving the original
124 /* Number of open file descriptors to this blob during a FUSE mount of
129 /* One of the `enum blob_location' values documented above. */
130 u16 blob_location : 4;
132 /* 1 iff this blob contains "metadata" as opposed to data. */
135 /* 1 iff the SHA-1 message digest of this blob is unknown. */
138 /* Temporary fields used when writing blobs; set as documented for
139 * prepare_blob_list_for_write(). */
141 u16 will_be_in_output_wim : 1;
143 /* Set to 1 if this blob represents a metadata resource that has been
144 * changed. In such cases, the hash cannot be used to verify the data
145 * if the metadata resource is read again. (This could be avoided if we
146 * used separate fields for input/output checksum, but most blobs
147 * wouldn't need this.) */
148 u16 dont_check_metadata_hash : 1;
150 u16 may_send_done_with_file : 1;
152 /* Only used by wimlib_export_image() */
153 u16 was_exported : 1;
155 /* Specification of where this blob's data is located. Which member of
156 * this union is valid is determined by the @blob_location field. */
160 struct wim_resource_descriptor *rdesc;
163 /* Links together blobs that share the same underlying
164 * WIM resource. The head is rdesc->blob_list. */
165 struct list_head rdesc_node;
172 /* BLOB_IN_FILE_ON_DISK
173 * BLOB_IN_WINNT_FILE_ON_DISK
174 * BLOB_WIN32_ENCRYPTED */
177 struct wim_inode *file_inode;
180 /* BLOB_IN_ATTACHED_BUFFER */
181 void *attached_buffer;
184 /* BLOB_IN_STAGING_FILE */
186 char *staging_file_name;
192 /* BLOB_IN_NTFS_VOLUME */
193 struct ntfs_location *ntfs_loc;
197 /* List link for per-WIM-image list of unhashed blobs */
198 struct list_head unhashed_list;
202 /* Temporary fields */
204 /* Fields used temporarily during WIM file writing. */
207 /* List node used for blob size table. */
208 struct hlist_node hash_list_2;
210 /* Metadata for the underlying solid resource in
211 * the WIM being written (only valid if
212 * WIM_RESHDR_FLAG_SOLID set in
213 * out_reshdr.flags). */
215 u64 out_res_offset_in_wim;
216 u64 out_res_size_in_wim;
217 u64 out_res_uncompressed_size;
221 /* Links blobs being written to the WIM. */
222 struct list_head write_blobs_list;
225 /* Metadata for this blob in the WIM being
227 struct wim_reshdr out_reshdr;
230 /* Name under which this blob is being
231 * sorted; used only when sorting blobs
232 * for solid compression. */
233 utf16lechar *solid_sort_name;
234 size_t solid_sort_name_nbytes;
239 /* Used temporarily during extraction. This is an array of
240 * references to the streams being extracted that use this blob.
241 * out_refcnt tracks the number of slots filled. */
243 struct blob_extraction_target inline_blob_extraction_targets[3];
245 struct blob_extraction_target *blob_extraction_targets;
246 u32 alloc_blob_extraction_targets;
251 /* Temporary list fields. */
253 /* Links blobs for writing blob table. */
254 struct list_head blob_table_list;
256 /* Links blobs being extracted. */
257 struct list_head extraction_list;
259 /* Links blobs being exported. */
260 struct list_head export_blob_list;
262 /* Links original list of blobs in the read-write mounted image. */
263 struct list_head orig_blob_list;
267 extern struct blob_table *
268 new_blob_table(size_t capacity) _malloc_attribute;
271 free_blob_table(struct blob_table *table);
274 read_blob_table(WIMStruct *wim);
277 write_blob_table_from_blob_list(struct list_head *blob_list,
278 struct filedes *out_fd,
280 struct wim_reshdr *out_reshdr,
281 int write_resource_flags);
283 extern struct blob_descriptor *
284 new_blob_descriptor(void) _malloc_attribute;
286 extern struct blob_descriptor *
287 clone_blob_descriptor(const struct blob_descriptor *blob) _malloc_attribute;
290 blob_decrement_refcnt(struct blob_descriptor *blob, struct blob_table *table);
293 blob_subtract_refcnt(struct blob_descriptor *blob, struct blob_table *table,
298 blob_decrement_num_opened_fds(struct blob_descriptor *blob);
302 free_blob_descriptor(struct blob_descriptor *blob);
305 blob_table_insert(struct blob_table *table, struct blob_descriptor *blob);
308 blob_table_unlink(struct blob_table *table, struct blob_descriptor *blob);
310 extern struct blob_descriptor *
311 lookup_blob(const struct blob_table *table, const u8 *hash);
314 for_blob_in_table(struct blob_table *table,
315 int (*visitor)(struct blob_descriptor *, void *), void *arg);
318 for_blob_in_table_sorted_by_sequential_order(struct blob_table *table,
319 int (*visitor)(struct blob_descriptor *, void *),
322 struct wimlib_resource_entry;
325 blob_to_wimlib_resource_entry(const struct blob_descriptor *blob,
326 struct wimlib_resource_entry *wentry);
329 sort_blob_list(struct list_head *blob_list,
330 size_t list_head_offset,
331 int (*compar)(const void *, const void*));
334 sort_blob_list_by_sequential_order(struct list_head *blob_list,
335 size_t list_head_offset);
338 cmp_blobs_by_sequential_order(const void *p1, const void *p2);
340 static inline const struct blob_extraction_target *
341 blob_extraction_targets(struct blob_descriptor *blob)
343 if (blob->out_refcnt <= ARRAY_LEN(blob->inline_blob_extraction_targets))
344 return blob->inline_blob_extraction_targets;
346 return blob->blob_extraction_targets;
350 * Declare that the specified blob is located in the specified WIM resource at
351 * the specified offset. The caller is expected to set blob->size if required.
354 blob_set_is_located_in_wim_resource(struct blob_descriptor *blob,
355 struct wim_resource_descriptor *rdesc,
358 blob->blob_location = BLOB_IN_WIM;
360 list_add_tail(&blob->rdesc_node, &rdesc->blob_list);
361 blob->offset_in_res = offset_in_res;
365 * Declare that the specified blob is located in the specified non-solid WIM
366 * resource. In this case, the blob data is the entire uncompressed resource.
369 blob_set_is_located_in_nonsolid_wim_resource(struct blob_descriptor *blob,
370 struct wim_resource_descriptor *rdesc)
372 blob_set_is_located_in_wim_resource(blob, rdesc, 0);
373 blob->size = rdesc->uncompressed_size;
377 blob_unset_is_located_in_wim_resource(struct blob_descriptor *blob)
379 list_del(&blob->rdesc_node);
380 blob->blob_location = BLOB_NONEXISTENT;
384 blob_set_is_located_in_attached_buffer(struct blob_descriptor *blob,
385 void *buffer, size_t size)
387 blob->blob_location = BLOB_IN_ATTACHED_BUFFER;
388 blob->attached_buffer = buffer;
392 extern struct blob_descriptor *
393 new_blob_from_data_buffer(const void *buffer, size_t size,
394 struct blob_table *blob_table);
396 extern struct blob_descriptor *
397 after_blob_hashed(struct blob_descriptor *blob,
398 struct blob_descriptor **back_ptr,
399 struct blob_table *blob_table);
402 hash_unhashed_blob(struct blob_descriptor *blob,
403 struct blob_table *blob_table,
404 struct blob_descriptor **blob_ret);
406 extern struct blob_descriptor **
407 retrieve_pointer_to_unhashed_blob(struct blob_descriptor *blob);
410 prepare_unhashed_blob(struct blob_descriptor *blob,
411 struct wim_inode *back_inode, u32 stream_id,
412 struct list_head *unhashed_blobs)
417 blob->back_inode = back_inode;
418 blob->back_stream_id = stream_id;
419 list_add_tail(&blob->unhashed_list, unhashed_blobs);
422 #endif /* _WIMLIB_BLOB_TABLE_H */