X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.h;h=dbcf55543254fc18ca48e78d7b18ff11a493b87e;hp=280376994176c36c8ef44a999902b72fe7253f7d;hb=6cca349b45e66a2b2b82aa5dcd269a4bf61c50db;hpb=3efd3c749f6cef737951769ace2fc802d8d8e61c diff --git a/src/lookup_table.h b/src/lookup_table.h index 28037699..dbcf5554 100644 --- a/src/lookup_table.h +++ b/src/lookup_table.h @@ -10,7 +10,9 @@ #define LOOKUP_FLAG_ADS_OK 0x00000001 #define LOOKUP_FLAG_DIRECTORY_OK 0x00000002 -#define LOOKUP_FLAG_FOLLOW_SYMLINKS 0x00000004 + +/* Not yet used */ +//#define LOOKUP_FLAG_FOLLOW_SYMLINKS 0x00000004 /* A lookup table that is used to translate the hash codes of dentries into the @@ -24,98 +26,180 @@ struct lookup_table { struct wimlib_fd; -/* - * An entry in the lookup table in the WIM file. +#ifdef WITH_NTFS_3G +struct ntfs_location { + char *path_utf8; + char *stream_name_utf16; + u16 stream_name_utf16_num_chars; + struct _ntfs_volume **ntfs_vol_p; + bool is_reparse_point; +}; +#endif + +/* + * An entry in the lookup table in the WIM file. + * + * It is used to find data streams for files in the WIM. * - * 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 in the WIM is found using the SHA1 - * message digest field. + * 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 { /* List of lookup table entries in this hash bucket */ - struct hlist_node hash_list; + union { + struct hlist_node hash_list; + struct list_head list; + }; - /* @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. */ + /* 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; - /* 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. */ + /* An enumerated type that identifies where the stream corresponding to + * this lookup table entry is actually located. + * + * Obviously 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 { + /* 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 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, + + /* 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, } resource_location; - /* Number of times this lookup table entry is referenced by dentries. */ + /* (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 */ + /* (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 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. */ + /* 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; }; - /* 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. */ + /* 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; - struct lookup_table_entry *next_lte_in_swm; + #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 inode *lte_inode; }; - FILE *file_on_disk_fp; #ifdef WITH_FUSE - /* File descriptors table for this data stream */ u16 num_opened_fds; - u16 num_allocated_fds; - struct wimlib_fd **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. 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. */ + * 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; - struct resource_entry output_resource_entry; - - /* Circular linked list of streams that share the same lookup table - * entry - * - * This list of streams may include streams from different hard link - * sets that happen to be the same. */ - struct list_head lte_group_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. */ - struct list_head staging_list; + + u32 real_refcnt; + + /* 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 + */ + union { + struct resource_entry output_resource_entry; + struct list_head msg_list; + }; + + /* 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; + + union { + /* 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 staging_list; + + /* Temporary field for creating a singly linked list. */ + struct lookup_table_entry *next_lte_in_swm; + }; }; static inline u64 wim_resource_size(const struct lookup_table_entry *lte) @@ -123,6 +207,11 @@ static inline u64 wim_resource_size(const struct lookup_table_entry *lte) return lte->resource_entry.original_size; } +static inline u64 wim_resource_chunks(const struct 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) { @@ -145,30 +234,24 @@ wim_resource_compression_type(const struct lookup_table_entry *lte) extern struct lookup_table *new_lookup_table(size_t capacity); -extern void lookup_table_insert(struct lookup_table *table, +extern void lookup_table_insert(struct lookup_table *table, struct 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, +static inline void lookup_table_unlink(struct lookup_table *table, struct lookup_table_entry *lte) { hlist_del(<e->hash_list); table->num_entries--; } +extern struct lookup_table_entry *new_lookup_table_entry(); extern struct lookup_table_entry * -lookup_table_decrement_refcnt(struct lookup_table* table, const u8 hash[]); - -extern struct lookup_table_entry * -lte_decrement_refcnt(struct lookup_table_entry *lte, - struct lookup_table *table); +clone_lookup_table_entry(const struct lookup_table_entry *lte); - -extern struct lookup_table_entry *new_lookup_table_entry(); - -extern int for_lookup_table_entry(struct lookup_table *table, - int (*visitor)(struct lookup_table_entry *, void *), +extern int for_lookup_table_entry(struct lookup_table *table, + int (*visitor)(struct lookup_table_entry *, void *), void *arg); extern struct lookup_table_entry * @@ -177,9 +260,17 @@ __lookup_resource(const struct lookup_table *table, const u8 hash[]); extern int lookup_resource(WIMStruct *w, const char *path, int lookup_flags, struct dentry **dentry_ret, struct lookup_table_entry **lte_ret, - unsigned *stream_idx_ret); + u16 *stream_idx_ret); + +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 zero_out_refcnts(struct lookup_table_entry *entry, void *ignore); +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_free_extracted_file(struct lookup_table_entry *lte, void *ignone); extern void print_lookup_table_entry(const struct lookup_table_entry *entry); @@ -192,6 +283,7 @@ extern int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out) extern void free_lookup_table_entry(struct lookup_table_entry *lte); extern int dentry_resolve_ltes(struct dentry *dentry, void *__table); +extern int dentry_unresolve_ltes(struct dentry *dentry, void *ignore); /* Writes the lookup table to the output file. */ static inline int write_lookup_table(struct lookup_table *table, FILE *out) @@ -200,7 +292,7 @@ static inline int write_lookup_table(struct lookup_table *table, FILE *out) } /* Unlinks and frees an entry from a lookup table. */ -static inline void lookup_table_remove(struct lookup_table *table, +static inline void lookup_table_remove(struct lookup_table *table, struct lookup_table_entry *lte) { lookup_table_unlink(table, lte); @@ -214,117 +306,124 @@ static inline struct resource_entry* wim_metadata_resource_entry(WIMStruct *w) } static inline struct lookup_table_entry * -dentry_stream_lte_resolved(const struct dentry *dentry, unsigned stream_idx) +inode_stream_lte_resolved(const struct inode *inode, unsigned stream_idx) { - wimlib_assert(dentry->resolved); - wimlib_assert(stream_idx <= dentry->num_ads); + wimlib_assert(inode->resolved); + wimlib_assert(stream_idx <= inode->num_ads); if (stream_idx == 0) - return dentry->lte; + return inode->lte; else - return dentry->ads_entries[stream_idx - 1].lte; + return inode->ads_entries[stream_idx - 1].lte; } static inline struct lookup_table_entry * -dentry_stream_lte_unresolved(const struct dentry *dentry, unsigned stream_idx, - const struct lookup_table *table) +inode_stream_lte_unresolved(const struct inode *inode, unsigned stream_idx, + const struct lookup_table *table) { - wimlib_assert(!dentry->resolved); - wimlib_assert(stream_idx <= dentry->num_ads); + wimlib_assert(!inode->resolved); + wimlib_assert(stream_idx <= inode->num_ads); if (!table) return NULL; if (stream_idx == 0) - return __lookup_resource(table, dentry->hash); + return __lookup_resource(table, inode->hash); else return __lookup_resource(table, - dentry->ads_entries[ + inode->ads_entries[ stream_idx - 1].hash); } -/* - * Returns the lookup table entry for stream @stream_idx of the dentry, where +/* + * 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 * -dentry_stream_lte(const struct dentry *dentry, unsigned stream_idx, - const struct lookup_table *table) +inode_stream_lte(const struct inode *inode, unsigned stream_idx, + const struct lookup_table *table) { - if (dentry->resolved) - return dentry_stream_lte_resolved(dentry, stream_idx); + if (inode->resolved) + return inode_stream_lte_resolved(inode, stream_idx); else - return dentry_stream_lte_unresolved(dentry, stream_idx, table); + return inode_stream_lte_unresolved(inode, stream_idx, table); } -static inline const u8 *dentry_stream_hash_unresolved(const struct dentry *dentry, - unsigned stream_idx) +static inline const u8 *inode_stream_hash_unresolved(const struct inode *inode, + unsigned stream_idx) { - wimlib_assert(!dentry->resolved); - wimlib_assert(stream_idx <= dentry->num_ads); + wimlib_assert(!inode->resolved); + wimlib_assert(stream_idx <= inode->num_ads); if (stream_idx == 0) - return dentry->hash; + return inode->hash; else - return dentry->ads_entries[stream_idx - 1].hash; + return inode->ads_entries[stream_idx - 1].hash; } -static inline const u8 *dentry_stream_hash_resolved(const struct dentry *dentry, - unsigned stream_idx) + +static inline const u8 *inode_stream_hash_resolved(const struct inode *inode, + unsigned stream_idx) { struct lookup_table_entry *lte; - lte = dentry_stream_lte_resolved(dentry, stream_idx); + lte = inode_stream_lte_resolved(inode, stream_idx); if (lte) return lte->hash; else - return NULL; + return zero_hash; } -/* - * Returns the hash for stream @stream_idx of the dentry, where stream_idx = 0 +/* + * 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 *dentry_stream_hash(const struct dentry *dentry, - unsigned stream_idx) +static inline const u8 *inode_stream_hash(const struct inode *inode, + unsigned stream_idx) { - if (dentry->resolved) - return dentry_stream_hash_resolved(dentry, stream_idx); + if (inode->resolved) + return inode_stream_hash_resolved(inode, stream_idx); else - return dentry_stream_hash_unresolved(dentry, stream_idx); + return inode_stream_hash_unresolved(inode, stream_idx); } -static inline struct lookup_table_entry * -dentry_first_lte_resolved(const struct dentry *dentry) +static inline u16 inode_stream_name_len(const struct inode *inode, + unsigned stream_idx) { - struct lookup_table_entry *lte; - wimlib_assert(dentry->resolved); + wimlib_assert(stream_idx <= inode->num_ads); + if (stream_idx == 0) + return 0; + else + return inode->ads_entries[stream_idx - 1].stream_name_len; +} - for (unsigned i = 0; i <= dentry->num_ads; i++) { - lte = dentry_stream_lte_resolved(dentry, i); - if (lte) - return lte; - } +static inline struct lookup_table_entry * +inode_unnamed_lte_resolved(const struct 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))) + return inode_stream_lte_resolved(inode, i); return NULL; } static inline struct lookup_table_entry * -dentry_first_lte_unresolved(const struct dentry *dentry, - const struct lookup_table *table) +inode_unnamed_lte_unresolved(const struct inode *inode, + const struct lookup_table *table) { - struct lookup_table_entry *lte; - wimlib_assert(!dentry->resolved); - - for (unsigned i = 0; i <= dentry->num_ads; i++) { - lte = dentry_stream_lte_unresolved(dentry, i, table); - if (lte) - return lte; - } + 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))) + return inode_stream_lte_unresolved(inode, i, table); return NULL; } extern struct lookup_table_entry * -dentry_first_lte(const struct dentry *dentry, const struct lookup_table *table); +inode_unnamed_lte(const struct inode *inode, + const struct lookup_table *table); + #endif