X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.h;h=633c874e17d1f7d1fc8ec03e6c41f722ddd135ff;hp=2665c39c3de1029c6e7fad98037c0561b04e9d95;hb=062465170ddb4520ca6fbc64a0f6e27e786de101;hpb=e6227e4978dd41e0726ef1b54713089d343b57de diff --git a/src/lookup_table.h b/src/lookup_table.h index 2665c39c..633c874e 100644 --- a/src/lookup_table.h +++ b/src/lookup_table.h @@ -11,26 +11,35 @@ #define LOOKUP_FLAG_ADS_OK 0x00000001 #define LOOKUP_FLAG_DIRECTORY_OK 0x00000002 -/* Not yet used */ -//#define LOOKUP_FLAG_FOLLOW_SYMLINKS 0x00000004 +#ifdef __WIN32__ +#include +#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 { +/* 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; }; -struct wimlib_fd; - #ifdef WITH_NTFS_3G struct ntfs_location { - char *path_utf8; - char *stream_name_utf16; - u16 stream_name_utf16_num_chars; + tchar *path; + utf16lechar *stream_name; + u16 stream_name_nchars; struct _ntfs_volume **ntfs_vol_p; bool is_reparse_point; }; @@ -60,19 +69,35 @@ enum resource_location { * */ RESOURCE_IN_FILE_ON_DISK, - /* The stream resource is located in an external file in the staging - * directory for a read-write mount. */ - RESOURCE_IN_STAGING_FILE, - /* 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 + +#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 + }; /* @@ -86,7 +111,7 @@ enum resource_location { * The lookup_table_entry for a given dentry or alternate stream entry in the * WIM is found using the SHA1 message digest field. */ -struct lookup_table_entry { +struct wim_lookup_table_entry { /* List of lookup table entries in this hash bucket */ struct hlist_node hash_list; @@ -111,7 +136,9 @@ struct lookup_table_entry { u16 part_number; /* See enum resource_location above */ - u16 resource_location; + u16 resource_location : 5; + u8 unique_size : 1; + u8 unhashed : 1; /* (On-disk field) * Number of times this lookup table entry is referenced by dentries. @@ -131,48 +158,51 @@ struct lookup_table_entry { * the full 20 byte hash just to insert the entry in a hash * table. */ size_t hash_short; + + /* Unhashed entries only (unhashed == 1): this points directly + * to the pointer to this 'struct wim_lookup_table_entry' + * contained in a 'struct wim_ads_entry' or 'struct wim_inode'. + * */ + struct wim_lookup_table_entry **my_ptr; }; - #ifdef WITH_FUSE - u16 num_opened_fds; - #endif + /* 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. The file resource 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 { WIMStruct *wim; - char *file_on_disk; - char *staging_file_name; - u8 *attached_buffer; + tchar *file_on_disk; + tchar *staging_file_name; + void *attached_buffer; #ifdef WITH_NTFS_3G struct ntfs_location *ntfs_loc; #endif }; - union { - /* @file_on_disk_fp and @attr are both used to cache file/stream - * handles so we don't have re-open them on every read */ - /* Valid iff resource_location == RESOURCE_IN_FILE_ON_DISK */ - FILE *file_on_disk_fp; - #ifdef WITH_NTFS_3G - /* Valid iff resource_location == RESOURCE_IN_NTFS_VOLUME */ - struct _ntfs_attr *attr; - #endif + /* Pointer to inode that contains the opened file descriptors to + * this stream (valid iff resource_location == + * RESOURCE_IN_STAGING_FILE) */ + struct wim_inode *lte_inode; - /* Pointer to inode that contains the opened file descriptors to - * this stream (valid iff resource_location == - * RESOURCE_IN_STAGING_FILE) */ - struct inode *lte_inode; - }; + u32 real_refcnt; - /* 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. The file resource only need to be written when - * out_refcnt is nonzero, since otherwise it is not referenced by any - * dentries. */ - u32 out_refcnt; + union { + #ifdef WITH_FUSE + u16 num_opened_fds; + #endif - u32 real_refcnt; + /* 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; + }; union { /* When a WIM file is written, @output_resource_entry is filled @@ -185,34 +215,41 @@ struct lookup_table_entry { struct resource_entry output_resource_entry; struct list_head msg_list; + struct list_head inode_list; + + struct { + struct hlist_node hash_list_2; + struct list_head write_streams_list; + }; }; + /* List of lookup table entries that correspond to streams that have + * been extracted to the staging directory when modifying a read-write + * mounted WIM. + * + * This field is also used to make other lists of lookup table entries. + * */ union { - /* 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. */ - char *extracted_file; - - /* List of lookup table entries that correspond to streams that have - * been extracted to the staging directory when modifying a read-write - * mounted WIM. */ + struct list_head unhashed_list; struct list_head staging_list; + struct list_head extraction_list; }; }; -static inline u64 wim_resource_size(const struct lookup_table_entry *lte) +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 lookup_table_entry *lte) +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 lookup_table_entry *lte) +wim_resource_compressed_size(const struct wim_lookup_table_entry *lte) { return lte->resource_entry.size; } @@ -222,146 +259,160 @@ wim_resource_compressed_size(const struct lookup_table_entry *lte) * entry */ static inline int -wim_resource_compression_type(const struct lookup_table_entry *lte) +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 WIM_COMPRESSION_TYPE_NONE; + 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 int +read_lookup_table(WIMStruct *w); + +extern int +write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry); -extern struct lookup_table *new_lookup_table(size_t capacity); +extern void +free_lookup_table(struct wim_lookup_table *table); -extern void lookup_table_insert(struct lookup_table *table, - struct lookup_table_entry *lte); +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 lookup_table *table, - struct lookup_table_entry *lte) +static inline void +lookup_table_unlink(struct wim_lookup_table *table, struct wim_lookup_table_entry *lte) { hlist_del(<e->hash_list); + wimlib_assert(table->num_entries != 0); table->num_entries--; } -extern struct lookup_table_entry *new_lookup_table_entry(); +extern struct wim_lookup_table_entry * +new_lookup_table_entry(); -extern struct lookup_table_entry * -clone_lookup_table_entry(const struct lookup_table_entry *lte); +extern struct wim_lookup_table_entry * +clone_lookup_table_entry(const struct wim_lookup_table_entry *lte); -extern int for_lookup_table_entry(struct lookup_table *table, - int (*visitor)(struct lookup_table_entry *, void *), - void *arg); +extern void +print_lookup_table_entry(const struct wim_lookup_table_entry *entry, + FILE *out); -extern struct lookup_table_entry * -__lookup_resource(const struct lookup_table *table, const u8 hash[]); +extern void +free_lookup_table_entry(struct wim_lookup_table_entry *lte); -extern int lookup_resource(WIMStruct *w, const char *path, - int lookup_flags, struct dentry **dentry_ret, - struct lookup_table_entry **lte_ret, - u16 *stream_idx_ret); +extern int +for_lookup_table_entry(struct wim_lookup_table *table, + int (*visitor)(struct wim_lookup_table_entry *, void *), + void *arg); -extern void lte_decrement_refcnt(struct lookup_table_entry *lte, - struct lookup_table *table); -#ifdef WITH_FUSE -extern void lte_decrement_num_opened_fds(struct lookup_table_entry *lte); -#endif +extern int +cmp_streams_by_wim_position(const void *p1, const void *p2); -extern int lte_zero_out_refcnt(struct lookup_table_entry *entry, void *ignore); -extern int lte_zero_real_refcnt(struct lookup_table_entry *entry, void *ignore); -extern int lte_zero_extracted_file(struct lookup_table_entry *lte, void *ignore); -extern int lte_free_extracted_file(struct lookup_table_entry *lte, void *ignore); +extern int +for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table, + int (*visitor)(struct wim_lookup_table_entry *, + void *), + void *arg); -extern void print_lookup_table_entry(const struct lookup_table_entry *entry); +extern struct wim_lookup_table_entry * +__lookup_resource(const struct wim_lookup_table *table, const u8 hash[]); -extern int read_lookup_table(WIMStruct *w); +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 free_lookup_table(struct lookup_table *table); +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 write_lookup_table_entry(struct lookup_table_entry *lte, void *__out); +extern int +lte_zero_out_refcnt(struct wim_lookup_table_entry *entry, void *ignore); -extern void free_lookup_table_entry(struct lookup_table_entry *lte); +extern int +lte_zero_real_refcnt(struct wim_lookup_table_entry *entry, void *ignore); -extern int dentry_resolve_ltes(struct dentry *dentry, void *__table); -extern int dentry_unresolve_ltes(struct dentry *dentry, void *ignore); +extern int +lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore); -int write_lookup_table(struct lookup_table *table, FILE *out, - struct resource_entry *out_res_entry); +extern void +inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table); -/* Unlinks and frees an entry from a lookup table. */ -static inline void lookup_table_remove(struct lookup_table *table, - struct lookup_table_entry *lte) -{ - lookup_table_unlink(table, lte); - free_lookup_table_entry(lte); -} +extern void +inode_unresolve_ltes(struct wim_inode *inode); -static inline struct resource_entry* wim_metadata_resource_entry(WIMStruct *w) -{ - return &w->image_metadata[ - w->current_image - 1].metadata_lte->resource_entry; -} +extern int +write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *__out); -static inline struct lookup_table_entry * -inode_stream_lte_resolved(const struct inode *inode, unsigned stream_idx) +static inline struct wim_lookup_table_entry * +inode_stream_lte_resolved(const struct wim_inode *inode, unsigned stream_idx) { - wimlib_assert(inode->resolved); - wimlib_assert(stream_idx <= inode->num_ads); + wimlib_assert(inode->i_resolved); + wimlib_assert(stream_idx <= inode->i_num_ads); if (stream_idx == 0) - return inode->lte; + return inode->i_lte; else - return inode->ads_entries[stream_idx - 1].lte; + return inode->i_ads_entries[stream_idx - 1].lte; } -static inline struct lookup_table_entry * -inode_stream_lte_unresolved(const struct inode *inode, unsigned stream_idx, - const struct lookup_table *table) +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->resolved); - wimlib_assert(stream_idx <= inode->num_ads); + 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->hash); + return __lookup_resource(table, inode->i_hash); else return __lookup_resource(table, - inode->ads_entries[ + inode->i_ads_entries[ stream_idx - 1].hash); } -/* - * Returns the lookup table entry 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 struct lookup_table_entry * -inode_stream_lte(const struct inode *inode, unsigned stream_idx, - const struct lookup_table *table) -{ - if (inode->resolved) - return inode_stream_lte_resolved(inode, stream_idx); - else - return inode_stream_lte_unresolved(inode, stream_idx, table); -} +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 inode *inode, - unsigned stream_idx) +static inline const u8 * +inode_stream_hash_unresolved(const struct wim_inode *inode, unsigned stream_idx) { - wimlib_assert(!inode->resolved); - wimlib_assert(stream_idx <= inode->num_ads); + wimlib_assert(!inode->i_resolved); + wimlib_assert(stream_idx <= inode->i_num_ads); if (stream_idx == 0) - return inode->hash; + return inode->i_hash; else - return inode->ads_entries[stream_idx - 1].hash; + return inode->i_ads_entries[stream_idx - 1].hash; } -static inline const u8 *inode_stream_hash_resolved(const struct inode *inode, - unsigned stream_idx) +static inline const u8 * +inode_stream_hash_resolved(const struct wim_inode *inode, unsigned stream_idx) { - struct lookup_table_entry *lte; + struct wim_lookup_table_entry *lte; lte = inode_stream_lte_resolved(inode, stream_idx); if (lte) return lte->hash; @@ -376,51 +427,73 @@ static inline const u8 *inode_stream_hash_resolved(const struct inode *inode, * * This works for both resolved and un-resolved dentries. */ -static inline const u8 *inode_stream_hash(const struct inode *inode, - unsigned stream_idx) +static inline const u8 * +inode_stream_hash(const struct wim_inode *inode, unsigned stream_idx) { - if (inode->resolved) + if (inode->i_resolved) return inode_stream_hash_resolved(inode, stream_idx); else return inode_stream_hash_unresolved(inode, stream_idx); } -static inline u16 inode_stream_name_len(const struct inode *inode, - unsigned stream_idx) +static inline u16 +inode_stream_name_nbytes(const struct wim_inode *inode, unsigned stream_idx) { - wimlib_assert(stream_idx <= inode->num_ads); + wimlib_assert(stream_idx <= inode->i_num_ads); if (stream_idx == 0) return 0; else - return inode->ads_entries[stream_idx - 1].stream_name_len; + return inode->i_ads_entries[stream_idx - 1].stream_name_nbytes; } -static inline struct lookup_table_entry * -inode_unnamed_lte_resolved(const struct inode *inode) +static inline struct wim_lookup_table_entry * +inode_unnamed_lte_resolved(const struct wim_inode *inode) { - wimlib_assert(inode->resolved); - for (unsigned i = 0; i <= inode->num_ads; i++) - if (inode_stream_name_len(inode, i) == 0 && - !is_zero_hash(inode_stream_hash_resolved(inode, i))) + 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 lookup_table_entry * -inode_unnamed_lte_unresolved(const struct inode *inode, - const struct lookup_table *table) +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->resolved); - for (unsigned i = 0; i <= inode->num_ads; i++) - if (inode_stream_name_len(inode, i) == 0 && - !is_zero_hash(inode_stream_hash_unresolved(inode, i))) + 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 lookup_table_entry * -inode_unnamed_lte(const struct inode *inode, - const struct lookup_table *table); +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_lookup_table_entry **my_ptr) +{ + lte->unhashed = 1; + list_add_tail(<e->unhashed_list, table->unhashed_streams); + lte->my_ptr = my_ptr; + *my_ptr = lte; +} +extern void +free_lte_list(struct list_head *list); #endif