#define _WIMLIB_LOOKUP_TABLE_H
#include "wimlib_internal.h"
#include "dentry.h"
+#include "sha1.h"
#include <sys/types.h>
/* Size of each lookup table entry in the WIM file. */
#define LOOKUP_FLAG_ADS_OK 0x00000001
#define LOOKUP_FLAG_DIRECTORY_OK 0x00000002
-#define LOOKUP_FLAG_FOLLOW_SYMLINKS 0x00000004
+
+#ifdef __WIN32__
+#include <windef.h>
+#endif
-/* A lookup table that is used to translate the hash codes of dentries into the
- * offsets and sizes of uncompressed or compressed file resources. It is
- * implemented as a hash table. */
-struct lookup_table {
- struct lookup_table_entry **array;
+/* The lookup table of a WIM file maps SHA1 message digests to streams of data.
+ * Here, the in-memory structure is implemented as a hash table.
+ *
+ * Given a SHA1 message digest, the mapped-to stream is specified by an offset
+ * in the WIM, an uncompressed and compressed size, and resource flags (see
+ * 'struct resource_entry'). But, we associate additional information, such as
+ * a reference count, with each stream, so the actual mapping is from SHA1
+ * message digests to 'struct wim_lookup_table_entry's, each of which contains
+ * an embedded 'struct resource_entry'.
+ *
+ * Note: Everything will break horribly if there is a SHA1 collision.
+ */
+struct wim_lookup_table {
+ struct hlist_head *array;
u64 num_entries;
u64 capacity;
+ struct list_head *unhashed_streams;
};
-/* An entry in the lookup table in the WIM file. */
-struct lookup_table_entry {
+#ifdef WITH_NTFS_3G
+struct ntfs_location {
+ tchar *path;
+ utf16lechar *stream_name;
+ u16 stream_name_nchars;
+ struct _ntfs_volume *ntfs_vol;
+ bool is_reparse_point;
+};
+#endif
- /* The next struct lookup_table_entry in the hash bucket. NULL if this is the
- * last one. */
- struct lookup_table_entry *next;
+/* An enumerated type that identifies where the stream corresponding to this
+ * lookup table entry is actually located.
+ *
+ * If we open a WIM and read its lookup table, the location is set to
+ * RESOURCE_IN_WIM since all the streams will initially be located in the WIM.
+ * However, to deal with problems such as image capture and image mount, we
+ * allow the actual location of the stream to be somewhere else, such as an
+ * external file.
+ */
+enum resource_location {
+ /* The lookup table entry does not correspond to a stream (this state
+ * should exist only temporarily) */
+ RESOURCE_NONEXISTENT = 0,
+
+ /* The stream resource is located in a WIM file. The WIMStruct for the
+ * WIM file will be pointed to by the @wim member. */
+ RESOURCE_IN_WIM,
+
+ /* The stream resource is located in an external file. The name of the
+ * file will be provided by @file_on_disk member. In addition, if
+ * @file_on_disk_fp is not NULL, it will be an open FILE * to the file.
+ * */
+ RESOURCE_IN_FILE_ON_DISK,
+
+ /* The stream resource is directly attached in an in-memory buffer
+ * pointed to by @attached_buffer. */
+ RESOURCE_IN_ATTACHED_BUFFER,
+
+#ifdef WITH_FUSE
+ /* The stream resource is located in an external file in the staging
+ * directory for a read-write mount. */
+ RESOURCE_IN_STAGING_FILE,
+#endif
- /* @resource_entry is read from the lookup table in the WIM
- * file; it says where to find the file resource in the WIM
- * file, and whether it is compressed or not. */
+#ifdef WITH_NTFS_3G
+ /* The stream resource is located in an NTFS volume. It is identified
+ * by volume, filename, data stream name, and by whether it is a reparse
+ * point or not. @ntfs_loc points to a structure containing this
+ * information. */
+ RESOURCE_IN_NTFS_VOLUME,
+#endif
+
+#ifdef __WIN32__
+ /* Resource must be accessed using Win32 API (may be a named data
+ * stream) */
+ RESOURCE_WIN32,
+
+ /* Windows only: the file is on disk in the file named @file_on_disk,
+ * but the file is encrypted and must be read using special functions.
+ * */
+ RESOURCE_WIN32_ENCRYPTED,
+#endif
+
+};
+
+/*
+ * An entry in the lookup table in the WIM file.
+ *
+ * It is used to find data streams for files in the WIM.
+ *
+ * Metadata resources and reparse point data buffers will also have lookup table
+ * entries associated with the data.
+ *
+ * The lookup_table_entry for a given dentry or alternate stream entry in the
+ * WIM is found using the SHA1 message digest field.
+ */
+struct wim_lookup_table_entry {
+
+ /* List of lookup table entries in this hash bucket */
+ struct hlist_node hash_list;
+
+ /* Location and size of the stream in the WIM, whether it is compressed
+ * or not, and whether it's a metadata resource or not. This is an
+ * on-disk field. */
struct resource_entry resource_entry;
- /* Currently ignored; set to 1 in new lookup table entries. */
+ /* Specifies which part of the split WIM the resource is located in.
+ * This is on on-disk field.
+ *
+ * In stand-alone WIMs, this must be 1.
+ *
+ * In split WIMs, every split WIM part has its own lookup table, and in
+ * read_lookup_table() it's currently expected that the part number of
+ * each lookup table entry in a split WIM part's lookup table is the
+ * same as the part number of that split WIM part. So this makes this
+ * field redundant since we store a pointer to the corresponding
+ * WIMStruct in the lookup table entry anyway.
+ */
u16 part_number;
- /* Number of times this lookup table entry is referenced by dentries. */
- u32 refcnt;
+ /* One of the `enum resource_location' values documented above. */
+ u16 resource_location : 5;
+
+ /* 1 if this stream is a unique size (only set while writing streams). */
+ u8 unique_size : 1;
- /* If %true, this lookup table entry corresponds to a symbolic link
- * reparse buffer. @symlink_reparse_data_buf will give the target of
- * the symbolic link. */
- bool is_symlink;
+ /* 1 if this stream had a SHA1-message digest calculated for it yet? */
+ u8 unhashed : 1;
+
+ u8 deferred : 1;
+
+ u8 no_progress : 1;
+
+ /* (On-disk field)
+ * Number of times this lookup table entry is referenced by dentries.
+ * Unfortunately, this field is not always set correctly in Microsoft's
+ * WIMs, so we have no choice but to fix it if more references to the
+ * lookup table entry are found than stated here. */
+ u32 refcnt;
union {
- /* SHA1 hash of the file resource pointed to by this lookup
- * table entry */
- u8 hash[WIM_HASH_SIZE];
-
- /* First 4 or 8 bytes of the SHA1 hash, used for inserting the
- * entry into the hash table. Since the SHA1 hashes can be
- * considered random, we don't really need the full 20 byte hash
- * just to insert the entry in a hash table. */
+ /* (On-disk field) SHA1 message digest of the stream referenced
+ * by this lookup table entry */
+ u8 hash[SHA1_HASH_SIZE];
+
+ /* First 4 or 8 bytes of the SHA1 message digest, used for
+ * inserting the entry into the hash table. Since the SHA1
+ * message digest can be considered random, we don't really need
+ * the full 20 byte hash just to insert the entry in a hash
+ * table. */
size_t hash_short;
+
+ /* Unhashed entries only (unhashed == 1): these variables make
+ * it possible to find the pointer to this 'struct
+ * wim_lookup_table_entry' contained in either 'struct
+ * wim_ads_entry' or 'struct wim_inode'. There can be at most 1
+ * such pointer, as we can only join duplicate streams after
+ * they have been hashed. */
+ struct {
+ struct wim_inode *back_inode;
+ u32 back_stream_id;
+ };
};
- /* If @file_on_disk != NULL, the file resource indicated by this lookup
- * table entry is not in the WIM file, but rather a file on disk; this
- * occurs for files that are added to the WIM. In that case,
- * file_on_disk is the name of the file in the outside filesystem.
- * It will not be compressed, and its size will be given by
- * resource_entry.size and resource_entry.original_size. */
+ /* When a WIM file is written, out_refcnt starts at 0 and is incremented
+ * whenever the stream pointed to by this lookup table entry needs to be
+ * written. The stream only need to be written when out_refcnt is
+ * nonzero, since otherwise it is not referenced by any dentries. */
+ u32 out_refcnt;
+
+ /* Pointers to somewhere where the stream is actually located. See the
+ * comments for the @resource_location field above. */
union {
- char *file_on_disk;
- char *staging_file_name;
- void *symlink_buf;
- struct lookup_table_entry *next_lte_in_swm;
+ WIMStruct *wim;
+ tchar *file_on_disk;
+ void *attached_buffer;
+ #ifdef WITH_FUSE
+ tchar *staging_file_name;
+ #endif
+ #ifdef WITH_NTFS_3G
+ struct ntfs_location *ntfs_loc;
+ #endif
};
- union {
- struct { /* Used for wimlib_export. */
+ /* Actual reference count to this stream (only used while
+ * verifying an image). */
+ u32 real_refcnt;
- /* If (other_wim_fp != NULL), the file resource indicated
- * by this lookup table entry is in a different WIM
- * file, and other_wim_fp is the FILE* for it. */
- FILE *other_wim_fp;
+ union {
+ #ifdef WITH_FUSE
+ /* Number of times this stream has been opened (used only during
+ * mounting) */
+ u16 num_opened_fds;
+ #endif
+
+ /* This field is used for the special hardlink or symlink image
+ * extraction mode. In these mode, all identical files are linked
+ * together, and @extracted_file will be set to the filename of the
+ * first extracted file containing this stream. */
+ tchar *extracted_file;
+ };
- /* Compression type used in other WIM. */
- int other_wim_ctype;
+ union {
+ /* When a WIM file is written, @output_resource_entry is filled
+ * in with the resource entry for the output WIM. This will not
+ * necessarily be the same as the @resource_entry since: - The
+ * stream may have a different offset in the new WIM - The
+ * stream may have a different compressed size in the new WIM if
+ * the compression type changed
+ */
+ struct resource_entry output_resource_entry;
+
+ struct {
+ struct list_head msg_list;
+ struct list_head being_compressed_list;
};
+ struct list_head inode_list;
- struct { /* Used for read-write mounts. */
+ struct {
+ struct hlist_node hash_list_2;
-
- /* Offset of the stream file_on_disk_fd. */
- off_t staging_offset;
-
-
- /* If file_on_disk_fd, if it is not -1, is the file
- * descriptor, opened for reading, for file_on_disk. */
- int staging_fd;
-
- /* Number of times the file has been opened.
- * file_on_disk_fd can be closed when num_times_opened
- * is decremented to 0. */
- int staging_num_times_opened;
+ struct list_head write_streams_list;
};
};
- /* When a WIM file is written, out_refcnt starts at 0 and is incremented
- * whenever the file resource pointed to by this lookup table entry
- * needs to be written. Naturally, the file resource only need to be
- * written when out_refcnt is 0. Incrementing it further is needed to
- * find the correct reference count to write to the lookup table in the
- * output file, which may be less than the regular refcnt if not all
- * images in the WIM file are written.
- *
- * output_resource_entry is the struct resource_entry for the position of the
- * file resource when written to the output file. */
+ /* Temporary list fields */
union {
- u32 out_refcnt;
- bool refcnt_is_incremented;
+ struct list_head unhashed_list;
+ struct list_head swm_stream_list;
+ struct list_head extraction_list;
+ struct list_head export_stream_list;
};
- struct resource_entry output_resource_entry;
- struct dentry *hard_link_sets;
};
-extern struct lookup_table *new_lookup_table(size_t capacity);
+static inline u64
+wim_resource_size(const struct wim_lookup_table_entry *lte)
+{
+ return lte->resource_entry.original_size;
+}
+
+static inline u64
+wim_resource_chunks(const struct wim_lookup_table_entry *lte)
+{
+ return (wim_resource_size(lte) + WIM_CHUNK_SIZE - 1) / WIM_CHUNK_SIZE;
+}
+
+static inline u64
+wim_resource_compressed_size(const struct wim_lookup_table_entry *lte)
+{
+ return lte->resource_entry.size;
+}
+
+/*
+ * XXX Probably should store the compression type directly in the lookup table
+ * entry
+ */
+static inline int
+wim_resource_compression_type(const struct wim_lookup_table_entry *lte)
+{
+ if (!(lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
+ || lte->resource_location != RESOURCE_IN_WIM)
+ return WIMLIB_COMPRESSION_TYPE_NONE;
+ return wimlib_get_compression_type(lte->wim);
+}
+
+static inline bool
+lte_filename_valid(const struct wim_lookup_table_entry *lte)
+{
+ return lte->resource_location == RESOURCE_IN_FILE_ON_DISK
+ #ifdef __WIN32__
+ || lte->resource_location == RESOURCE_WIN32
+ || lte->resource_location == RESOURCE_WIN32_ENCRYPTED
+ #endif
+ #ifdef WITH_FUSE
+ || lte->resource_location == RESOURCE_IN_STAGING_FILE
+ #endif
+ ;
+}
+
+extern struct wim_lookup_table *
+new_lookup_table(size_t capacity);
-extern void lookup_table_insert(struct lookup_table *table,
- struct lookup_table_entry *lte);
+extern int
+read_lookup_table(WIMStruct *w);
-extern void lookup_table_unlink(struct lookup_table *table,
- struct lookup_table_entry *lte);
+extern int
+write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry);
+
+extern void
+free_lookup_table(struct wim_lookup_table *table);
+
+extern void
+lookup_table_insert(struct wim_lookup_table *table, struct wim_lookup_table_entry *lte);
+
+/* Unlinks a lookup table entry from the table; does not free it. */
+static inline void
+lookup_table_unlink(struct wim_lookup_table *table, struct wim_lookup_table_entry *lte)
+{
+ wimlib_assert(!lte->unhashed);
+ hlist_del(<e->hash_list);
+ wimlib_assert(table->num_entries != 0);
+ table->num_entries--;
+}
-extern bool lookup_table_decrement_refcnt(struct lookup_table* table,
- const u8 hash[]);
+extern struct wim_lookup_table_entry *
+new_lookup_table_entry();
+extern struct wim_lookup_table_entry *
+clone_lookup_table_entry(const struct wim_lookup_table_entry *lte);
-extern struct lookup_table_entry *new_lookup_table_entry();
+extern void
+print_lookup_table_entry(const struct wim_lookup_table_entry *entry,
+ FILE *out);
-extern int for_lookup_table_entry(struct lookup_table *table,
- int (*visitor)(struct lookup_table_entry *, void *),
+extern void
+free_lookup_table_entry(struct wim_lookup_table_entry *lte);
+
+extern int
+for_lookup_table_entry(struct wim_lookup_table *table,
+ int (*visitor)(struct wim_lookup_table_entry *, void *),
+ void *arg);
+
+extern int
+cmp_streams_by_wim_position(const void *p1, const void *p2);
+
+extern int
+for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
+ int (*visitor)(struct wim_lookup_table_entry *,
+ void *),
void *arg);
-extern struct lookup_table_entry *
-__lookup_resource(const struct lookup_table *lookup_table, const u8 hash[]);
+extern struct wim_lookup_table_entry *
+__lookup_resource(const struct wim_lookup_table *table, const u8 hash[]);
+
+extern int
+lookup_resource(WIMStruct *w, const tchar *path,
+ int lookup_flags, struct wim_dentry **dentry_ret,
+ struct wim_lookup_table_entry **lte_ret, u16 *stream_idx_ret);
+
+extern void
+lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
+ struct wim_lookup_table *table);
+#ifdef WITH_FUSE
+extern void
+lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte);
+#endif
-extern int lookup_resource(WIMStruct *w, const char *path,
- int lookup_flags, struct dentry **dentry_ret,
- struct lookup_table_entry **lte_ret,
- u8 **hash_ret);
+extern int
+lte_zero_out_refcnt(struct wim_lookup_table_entry *entry, void *ignore);
-extern int zero_out_refcnts(struct lookup_table_entry *entry, void *ignore);
+extern int
+lte_zero_real_refcnt(struct wim_lookup_table_entry *entry, void *ignore);
-extern int print_lookup_table_entry(struct lookup_table_entry *entry, void *ignore);
+extern int
+lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore);
-extern int read_lookup_table(FILE *fp, u64 offset, u64 size,
- struct lookup_table **table_ret);
+extern void
+inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table);
-extern void free_lookup_table(struct lookup_table *table);
+extern void
+inode_unresolve_ltes(struct wim_inode *inode);
-extern int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out);
+extern int
+write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *__out);
-static inline void free_lookup_table_entry(struct lookup_table_entry *lte)
+static inline struct wim_lookup_table_entry *
+inode_stream_lte_resolved(const struct wim_inode *inode, unsigned stream_idx)
{
- if (lte) {
- FREE(lte->file_on_disk);
- FREE(lte);
- }
+ wimlib_assert(inode->i_resolved);
+ wimlib_assert(stream_idx <= inode->i_num_ads);
+ if (stream_idx == 0)
+ return inode->i_lte;
+ else
+ return inode->i_ads_entries[stream_idx - 1].lte;
+}
+
+static inline struct wim_lookup_table_entry *
+inode_stream_lte_unresolved(const struct wim_inode *inode, unsigned stream_idx,
+ const struct wim_lookup_table *table)
+{
+ wimlib_assert(!inode->i_resolved);
+ wimlib_assert(stream_idx <= inode->i_num_ads);
+ if (!table)
+ return NULL;
+ if (stream_idx == 0)
+ return __lookup_resource(table, inode->i_hash);
+ else
+ return __lookup_resource(table,
+ inode->i_ads_entries[
+ stream_idx - 1].hash);
}
-/* Writes the lookup table to the output file. */
-static inline int write_lookup_table(struct lookup_table *table, FILE *out)
+extern struct wim_lookup_table_entry *
+inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
+ const struct wim_lookup_table *table);
+
+static inline const u8 *
+inode_stream_hash_unresolved(const struct wim_inode *inode, unsigned stream_idx)
+{
+ wimlib_assert(!inode->i_resolved);
+ wimlib_assert(stream_idx <= inode->i_num_ads);
+ if (stream_idx == 0)
+ return inode->i_hash;
+ else
+ return inode->i_ads_entries[stream_idx - 1].hash;
+}
+
+
+static inline const u8 *
+inode_stream_hash_resolved(const struct wim_inode *inode, unsigned stream_idx)
{
- return for_lookup_table_entry(table, write_lookup_table_entry, out);
+ struct wim_lookup_table_entry *lte;
+ lte = inode_stream_lte_resolved(inode, stream_idx);
+ if (lte)
+ return lte->hash;
+ else
+ return zero_hash;
}
-/* Unlinks and frees an entry from a lookup table. */
-static inline void lookup_table_remove(struct lookup_table *table,
- struct lookup_table_entry *lte)
+/*
+ * Returns the hash for stream @stream_idx of the inode, where stream_idx = 0
+ * means the default un-named file stream, and stream_idx >= 1 corresponds to an
+ * alternate data stream.
+ *
+ * This works for both resolved and un-resolved dentries.
+ */
+static inline const u8 *
+inode_stream_hash(const struct wim_inode *inode, unsigned stream_idx)
{
- lookup_table_unlink(table, lte);
- free_lookup_table_entry(lte);
+ if (inode->i_resolved)
+ return inode_stream_hash_resolved(inode, stream_idx);
+ else
+ return inode_stream_hash_unresolved(inode, stream_idx);
}
-static inline struct resource_entry* wim_metadata_resource_entry(WIMStruct *w)
+static inline u16
+inode_stream_name_nbytes(const struct wim_inode *inode, unsigned stream_idx)
{
- return &w->image_metadata[
- w->current_image - 1].metadata_lte->resource_entry;
+ wimlib_assert(stream_idx <= inode->i_num_ads);
+ if (stream_idx == 0)
+ return 0;
+ else
+ return inode->i_ads_entries[stream_idx - 1].stream_name_nbytes;
}
+static inline struct wim_lookup_table_entry *
+inode_unnamed_lte_resolved(const struct wim_inode *inode)
+{
+ wimlib_assert(inode->i_resolved);
+ for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+ if (inode_stream_name_nbytes(inode, i) == 0 &&
+ !is_zero_hash(inode_stream_hash_resolved(inode, i)))
+ {
+ return inode_stream_lte_resolved(inode, i);
+ }
+ }
+ return NULL;
+}
+
+static inline struct wim_lookup_table_entry *
+inode_unnamed_lte_unresolved(const struct wim_inode *inode,
+ const struct wim_lookup_table *table)
+{
+ wimlib_assert(!inode->i_resolved);
+ for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+ if (inode_stream_name_nbytes(inode, i) == 0 &&
+ !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
+ {
+ return inode_stream_lte_unresolved(inode, i, table);
+ }
+ }
+ return NULL;
+}
+
+extern struct wim_lookup_table_entry *
+inode_unnamed_lte(const struct wim_inode *inode, const struct wim_lookup_table *table);
+
+extern u64
+lookup_table_total_stream_size(struct wim_lookup_table *table);
+
+
+static inline void
+lookup_table_insert_unhashed(struct wim_lookup_table *table,
+ struct wim_lookup_table_entry *lte,
+ struct wim_inode *back_inode,
+ u32 back_stream_id)
+{
+ lte->unhashed = 1;
+ lte->back_inode = back_inode;
+ lte->back_stream_id = back_stream_id;
+ list_add_tail(<e->unhashed_list, table->unhashed_streams);
+}
+
+extern int
+hash_unhashed_stream(struct wim_lookup_table_entry *lte,
+ struct wim_lookup_table *lookup_table,
+ struct wim_lookup_table_entry **lte_ret);
+
+extern struct wim_lookup_table_entry **
+retrieve_lte_pointer(struct wim_lookup_table_entry *lte);
+
#endif