X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fdentry.h;h=b51000a28ff871949caae9f28a1439903fdbb34b;hp=b6aa966f0501f2d568f879bcfc17c0ee3385c1d8;hb=aa882e9bb92d998707484d7f6f56e9cf109a7c62;hpb=0dd190af48ac572f9e33ff9436ec8900fa6281a3

diff --git a/src/dentry.h b/src/dentry.h
index b6aa966f..b51000a2 100644
--- a/src/dentry.h
+++ b/src/dentry.h
@@ -4,22 +4,24 @@
 #include "util.h"
 #include "config.h"
 #include "list.h"
+#include "sha1.h"
 #include <string.h>
 
-
 struct stat;
 struct lookup_table;
-typedef struct WIMStruct WIMStruct;
+struct WIMStruct;
+struct lookup_table_entry;
+struct wimlib_fd;
+struct inode;
+struct dentry;
 
 /* Size of the struct dentry up to and including the file_name_len. */
 #define WIM_DENTRY_DISK_SIZE    102
 
+/* Size of on-disk WIM alternate data stream entry, in bytes, up to and
+ * including the stream length field (see below). */
 #define WIM_ADS_ENTRY_DISK_SIZE 38
 
-#ifndef WIM_HASH_SIZE
-#define WIM_HASH_SIZE 20
-#endif
-
 /* 
  * Reparse tags documented at 
  * http://msdn.microsoft.com/en-us/library/dd541667(v=prot.10).aspx
@@ -52,20 +54,23 @@ typedef struct WIMStruct WIMStruct;
 #define FILE_ATTRIBUTE_ENCRYPTED           0x00004000
 #define FILE_ATTRIBUTE_VIRTUAL             0x00010000
 
-struct lookup_table_entry;
 
-/* Alternate data stream entry */
+/* Alternate data stream entry.
+ *
+ * We read this from disk in the read_ads_entries() function; see that function
+ * for more explanation. */
 struct ads_entry {
 	union {
 		/* SHA-1 message digest of stream contents */
-		u8 hash[WIM_HASH_SIZE];
+		u8 hash[SHA1_HASH_SIZE];
 
 		/* The corresponding lookup table entry (only for resolved
 		 * streams) */
 		struct lookup_table_entry *lte;
 	};
 
-	/* Length of stream name (UTF-16) */
+	/* Length of stream name (UTF-16).  This is in bytes, not characters,
+	 * and does not include the terminating null character   */
 	u16 stream_name_len;
 
 	/* Length of stream name (UTF-8) */
@@ -77,102 +82,87 @@ struct ads_entry {
 	/* Stream name (UTF-8) */
 	char *stream_name_utf8;
 
-	struct stream_list_head lte_group_list;
+#ifdef WITH_FUSE
+	/* Number to identify an alternate data stream even after it's possibly
+	 * been moved or renamed. */
+	u32 stream_id;
+#endif
 };
 
-static inline u64 ads_entry_length(const struct ads_entry *entry)
-{
-	u64 len = WIM_ADS_ENTRY_DISK_SIZE + entry->stream_name_len + 2;
-	return (len + 7) & ~7;
-}
-
-static inline void destroy_ads_entry(struct ads_entry *entry)
-{
-	FREE(entry->stream_name);
-	FREE(entry->stream_name_utf8);
-	memset(entry, 0, sizeof(entry));
-}
 
-static inline bool ads_entry_has_name(const struct ads_entry *entry,
-				      const char *name, size_t name_len)
+static inline bool ads_entries_have_same_name(const struct ads_entry *entry_1,
+					      const struct ads_entry *entry_2)
 {
-	if (entry->stream_name_utf8_len != name_len)
+	if (entry_1->stream_name_len != entry_2->stream_name_len)
 		return false;
-	return memcmp(entry->stream_name_utf8, name, name_len) == 0;
+	return memcmp(entry_1->stream_name, entry_2->stream_name,
+		      entry_1->stream_name_len) == 0;
 }
 
-
-/* In-memory structure for a directory entry.  There is a directory tree for
- * each image in the WIM.  */
+/* 
+ * In-memory structure for a WIM directory entry (dentry).  There is a directory
+ * tree for each image in the WIM. 
+ *
+ * Note that this is a directory entry and not an inode.  Since NTFS allows hard
+ * links, it's possible for a NTFS inode to correspond to multiple WIM dentries.
+ * The hard_link field on the on-disk WIM dentry tells us the number of the NTFS
+ * inode that the dentry corresponds to.
+ *
+ * Unfortunately, WIM files do not have an analogue to an inode; instead certain
+ * information, such as file attributes, the security descriptor, and file
+ * streams is replicated in each hard-linked dentry, even though this
+ * information really is associated with an inode.  In-memory, we fix up this
+ * flaw by allocating a `struct inode' for each dentry that contains some of
+ * this duplicated information, then combining the inodes for each hard link
+ * group together.
+ *
+ * Confusingly, it's possible for stream information to be missing from a dentry
+ * in a hard link set, in which case the stream information needs to be gotten
+ * from one of the other dentries in the hard link set.  In addition, it is
+ * possible for dentries to have inconsistent security IDs, file attributes, or
+ * file streams when they share the same hard link ID (don't even ask.  I hope
+ * that Microsoft may have fixed this problem, since I've only noticed it in the
+ * 'install.wim' for Windows 7).  For those dentries, we have to use the
+ * conflicting fields to split up the hard link groups.  (See fix_inodes() in
+ * hardlink.c).
+ */
 struct dentry {
+	/* The inode for this dentry */
+	struct inode *d_inode;
+
 	/* The parent of this directory entry. */
 	struct dentry *parent;
 
 	/* Linked list of sibling directory entries. */
 	struct dentry *next;
-
 	struct dentry *prev;
 
-	/* Pointer to a child of this directory entry. */
-	struct dentry *children;
-
-	/* Size of directory entry, in bytes.  Typical size is around 104 to 120
-	 * bytes. */
-	/* It is possible for the length field to be 0.  This situation, which
+	/* 
+	 * Size of directory entry on disk, in bytes.  Typical size is around
+	 * 104 to 120 bytes.
+	 *
+	 * It is possible for the length field to be 0.  This situation, which
 	 * is undocumented, indicates the end of a list of sibling nodes in a
 	 * directory.  It also means the real length is 8, because the dentry
-	 * included only the length field, but that takes up 8 bytes. */
+	 * included only the length field, but that takes up 8 bytes.
+	 *
+	 * The length here includes the base directory entry on disk as well as
+	 * the long and short filenames.  It does NOT include any alternate
+	 * stream entries that may follow the directory entry, even though the
+	 * size of those needs to be considered.  The length SHOULD be 8-byte
+	 * aligned, although we don't require it to be.  We do require the
+	 * length to be large enough to hold the file name(s) of the dentry;
+	 * additionally, a warning is issued if this field is larger than the
+	 * aligned size.
+	 */
 	u64 length;
 
-	/* The file attributes associated with this file. */
-	u32 attributes;
-
-	/* The index of the node in the security table that contains this file's
-	 * security information.  If -1, no security information exists for this
-	 * file.  */
-	int32_t security_id;
-
-	/* The offset, from the start of the metadata section, of this directory
-	 * entry's child files.  0 if the directory entry has no children. */
+	/* The offset, from the start of the uncompressed WIM metadata resource
+	 * for this image, of this dentry's child dentries.  0 if the directory
+	 * entry has no children, which is the case for regular files or reparse
+	 * points. */
 	u64 subdir_offset;
 
-	/* Timestamps for the entry.  The timestamps are the number of
-	 * 100-nanosecond intervals that have elapsed since 12:00 A.M., January
-	 * 1st, 1601, UTC. */
-	u64 creation_time;
-	u64 last_access_time;
-	u64 last_write_time;
-
-	/* true if the dentry's lookup table entry has been resolved (i.e. the
-	 * @lte field is invalid, but the @hash field is not valid) */
-	bool resolved;
-
-	/* A hash of the file's contents, or a pointer to the lookup table entry
-	 * for this dentry if the lookup table entries have been resolved.
-	 *
-	 * More specifically, this is for the un-named default file stream, as
-	 * opposed to the alternate file streams, which may have their own
-	 * lookup table entries.  */
-	union {
-		u8 hash[WIM_HASH_SIZE];
-		struct lookup_table_entry *lte;
-	};
-
-	/* Identity of a reparse point.  See
-	 * http://msdn.microsoft.com/en-us/library/windows/desktop/aa365503(v=vs.85).aspx
-	 * for what a reparse point is. */
-	u32 reparse_tag;
-
-	/* Although M$'s documentation does not tell you this, it seems that the
-	 * reparse_reserved field does not actually exist.  So the hard_link
-	 * field directly follows the reparse_tag on disk.  EXCEPT when the
-	 * dentry is actually a reparse point... well, just take a look at the
-	 * read_dentry() function. */
-	//u32 reparse_reserved;
-
-	/* Number of alternate data streams associated with this file. */
-	u16 num_ads;
-
 	/* Length of short filename, in bytes, not including the terminating
 	 * zero wide-character. */
 	u16 short_name_len;
@@ -185,111 +175,139 @@ struct dentry {
 	 * the terminating zero byte. */
 	u16 file_name_utf8_len;
 
-	/* Pointer to the short filename */
+	/* Pointer to the short filename (malloc()ed buffer) */
 	char *short_name;
 
-	/* Pointer to the filename. */
+	/* Pointer to the filename (malloc()ed buffer). */
 	char *file_name;
 
-	/* Pointer to the filename converted to UTF-8. */
+	/* Pointer to the filename converted to UTF-8 (malloc()ed buffer). */
 	char *file_name_utf8;
 
-	/* Full path to this dentry. */
+	/* Full path to this dentry (malloc()ed buffer). */
 	char *full_path_utf8;
 	u32   full_path_utf8_len;
 
-	/* Alternate stream entries for this dentry. */
-	struct ads_entry *ads_entries;
+	/* Number of references to the dentry tree itself, as in multiple
+	 * WIMStructs */
+	u32 refcnt;
 
-	union {
-		/* Number of references to the dentry tree itself, as in multiple
-		 * WIMStructs */
-		int refcnt;
+	/* List of dentries in the inode (hard link set)  */
+	struct list_head inode_dentry_list;
 
-		/* Number of times this dentry has been opened (only for
-		 * directories!) */
-		u32 num_times_opened;
+	union {
+		struct list_head tmp_list;
+		bool is_extracted;
 	};
+};
 
-	/* If the file is part of a hard link set, all the directory entries in
-	 * the set will share the same value for this field. */
-	u64 hard_link;
+/*
+ * WIM inode.
+ *
+ * As mentioned above, in the WIM file that is no on-disk analogue of a real
+ * inode, as most of these fields are duplicated in the dentries.
+ */
+struct inode {
+	/* Timestamps for the inode.  The timestamps are the number of
+	 * 100-nanosecond intervals that have elapsed since 12:00 A.M., January
+	 * 1st, 1601, UTC.  This is the same format used in NTFS inodes. */
+	u64 creation_time;
+	u64 last_access_time;
+	u64 last_write_time;
 
-	enum {
-		/* This dentry is the owner of its ads_entries, although it may
-		 * be in a hard link set */
-		GROUP_INDEPENDENT = 0,
+	/* The file attributes associated with this inode.  This is a bitwise OR
+	 * of the FILE_ATTRIBUTE_* flags. */
+	u32 attributes;
 
-		/* This dentry is the owner of the ads_entries in the hard link
-		 * set */
-		GROUP_MASTER,
+	/* The index of the security descriptor in the WIM image's table of
+	 * security descriptors that contains this file's security information.
+	 * If -1, no security information exists for this file.  */
+	int32_t security_id;
+
+	/* %true iff the inode's lookup table entries has been resolved (i.e.
+	 * the @lte field is valid, but the @hash field is not valid) 
+	 *
+	 * (This is not an on-disk field.) */
+	u8 resolved : 1;
+
+	/* %true iff verify_inode() has run on this dentry. */
+	u8 verified : 1;
+
+	/* Number of alternate data streams associated with this inode */
+	u16 num_ads;
+
+	/* A hash of the file's contents, or a pointer to the lookup table entry
+	 * for this dentry if the lookup table entries have been resolved.
+	 *
+	 * More specifically, this is for the un-named default file stream, as
+	 * opposed to the alternate (named) file streams, which may have their
+	 * own lookup table entries.  */
+	union {
+		u8 hash[SHA1_HASH_SIZE];
+		struct lookup_table_entry *lte;
+	};
 
-		/* This dentry shares its ads_entries with a dentry in the hard
-		 * link set that has GROUP_MASTER set. */
-		GROUP_SLAVE
-	} link_group_master_status;
+	/* Identity of a reparse point.  See
+	 * http://msdn.microsoft.com/en-us/library/windows/desktop/aa365503(v=vs.85).aspx
+	 * for what a reparse point is. */
+	u32 reparse_tag;
 
+	/* Number of dentries that reference this inode */
+	u32 link_count;
 
-	/* List of dentries in the hard link set */
-	struct list_head link_group_list;
+	/* Alternate data stream entries. */
+	struct ads_entry *ads_entries;
 
-	/* List of dentries sharing the same lookup table entry */
-	struct stream_list_head lte_group_list;
+	/* Inode number */
+	u64 ino;
 
-	/* Path to extracted file on disk (used during extraction only) */
+	/* List of dentries that reference this inode (there should be
+	 * link_count of them) */
+	struct list_head dentry_list;
+	struct hlist_node hlist;
 	char *extracted_file;
+
+	/* If non-NULL, the children of this inode (implies the inode is a
+	 * directory) */
+	struct dentry *children;
+
+#ifdef WITH_FUSE
+	/* wimfs file descriptors table for the inode */
+	u16 num_opened_fds;
+	u16 num_allocated_fds;
+	struct wimlib_fd **fds;
+
+	/* Next alternate data stream ID to be assigned */
+	u32 next_stream_id;
+#endif
 };
 
-/* Return hash of the "unnamed" (default) data stream. */
-static inline const u8 *dentry_hash(const struct dentry *dentry)
-{
-	wimlib_assert(!dentry->resolved);
-	/* If there are alternate data streams, the dentry hash field is zeroed
-	 * out, and we need to find the hash in the un-named data stream (should
-	 * be the first one, but check them in order just in case, and fall back
-	 * to the dentry hash field if we can't find an unnamed data stream). */
-	for (u16 i = 0; i < dentry->num_ads; i++)
-		if (dentry->ads_entries[i].stream_name_len == 0)
-			return dentry->ads_entries[i].hash;
-	return dentry->hash;
-}
+#define inode_for_each_dentry(dentry, inode) \
+		list_for_each_entry((dentry), &(inode)->dentry_list, inode_dentry_list)
+
+#define inode_add_dentry(dentry, inode) \
+	({								\
+	 	wimlib_assert((inode)->dentry_list.next != NULL);		\
+		list_add(&(dentry)->inode_dentry_list, &(inode)->dentry_list);	\
+	})
 
-/* Return lte for the "unnamed" (default) data stream.  Only for resolved
- * dentries */
-static inline struct lookup_table_entry *
-dentry_lte(const struct dentry *dentry)
+static inline bool dentry_is_extracted(const struct dentry *dentry)
 {
-	wimlib_assert(dentry->resolved);
-	for (u16 i = 0; i < dentry->num_ads; i++)
-		if (dentry->ads_entries[i].stream_name_len == 0)
-			return dentry->ads_entries[i].lte;
-	return dentry->lte;
+	return dentry->is_extracted;
 }
 
-static inline size_t dentry_link_group_size(const struct dentry *dentry)
+static inline bool dentry_is_first_in_inode(const struct dentry *dentry)
 {
-	const struct list_head *cur = &dentry->link_group_list;
-	size_t size = 0;
-	do {
-		size++;
-		cur = cur->next;
-	} while (cur != &dentry->link_group_list);
-	return size;
+	return container_of(dentry->d_inode->dentry_list.next,
+			    struct dentry,
+			    inode_dentry_list) == dentry;
 }
 
-extern struct ads_entry *dentry_get_ads_entry(struct dentry *dentry,
-					      const char *stream_name);
-
-extern struct ads_entry *dentry_add_ads(struct dentry *dentry,
-					const char *stream_name);
+extern u64 dentry_correct_total_length(const struct dentry *dentry);
 
-extern void dentry_remove_ads(struct dentry *dentry, struct ads_entry *entry);
-
-extern const char *path_stream_name(const char *path);
-
-extern u64 dentry_total_length(const struct dentry *dentry);
-
-extern void stbuf_to_dentry(const struct stat *stbuf, struct dentry *dentry);
+extern void stbuf_to_inode(const struct stat *stbuf, struct inode *inode);
+extern int inode_to_stbuf(const struct inode *inode,
+			  struct lookup_table_entry *lte, struct stat *stbuf);
 
 extern int for_dentry_in_tree(struct dentry *root, 
 			      int (*visitor)(struct dentry*, void*), 
@@ -304,54 +322,53 @@ extern void calculate_subdir_offsets(struct dentry *dentry, u64 *subdir_offset_p
 extern int get_names(char **name_utf16_ret, char **name_utf8_ret,
 	      	     u16 *name_utf16_len_ret, u16 *name_utf8_len_ret,
 	             const char *name);
-extern int change_dentry_name(struct dentry *dentry, const char *new_name);
-extern int change_ads_name(struct ads_entry *entry, const char *new_name);
 
-extern void unlink_dentry(struct dentry *dentry);
-extern void link_dentry(struct dentry *dentry, struct dentry *parent);
+extern struct dentry *get_dentry(struct WIMStruct *w, const char *path);
+extern struct inode *wim_pathname_to_inode(struct WIMStruct *w,
+					   const char *path);
+extern struct dentry *get_dentry_child_with_name(const struct dentry *dentry, 
+						 const char *name);
+extern struct dentry *get_parent_dentry(struct WIMStruct *w, const char *path);
 
 extern int print_dentry(struct dentry *dentry, void *lookup_table);
 extern int print_dentry_full_path(struct dentry *entry, void *ignore);
 
-extern struct dentry *get_dentry(WIMStruct *w, const char *path);
-extern struct dentry *get_parent_dentry(WIMStruct *w, const char *path);
-extern struct dentry *get_dentry_child_with_name(const struct dentry *dentry, 
-							const char *name);
-extern void dentry_update_all_timestamps(struct dentry *dentry);
-extern void init_dentry(struct dentry *dentry, const char *name);
 extern struct dentry *new_dentry(const char *name);
+extern struct dentry *new_dentry_with_inode(const char *name);
+extern struct dentry *new_dentry_with_timeless_inode(const char *name);
 
-extern void dentry_free_ads_entries(struct dentry *dentry);
+extern void free_inode(struct inode *inode);
 extern void free_dentry(struct dentry *dentry);
 extern void put_dentry(struct dentry *dentry);
-extern int share_dentry_ads(struct dentry *master,
-			    struct dentry *slave);
-extern struct dentry *clone_dentry(struct dentry *old);
+
 extern void free_dentry_tree(struct dentry *root,
-			     struct lookup_table *lookup_table, 
-			     bool lt_decrement_refcnt);
+			     struct lookup_table *lookup_table);
 extern int increment_dentry_refcnt(struct dentry *dentry, void *ignore);
-extern int decrement_dentry_refcnt(struct dentry *dentry, void *ignore);
 
-extern void calculate_dir_tree_statistics(struct dentry *root, 
-					  struct lookup_table *table, 
-					  u64 *dir_count_ret, 
-					  u64 *file_count_ret, 
-					  u64 *total_bytes_ret, 
-					  u64 *hard_link_bytes_ret);
+extern void unlink_dentry(struct dentry *dentry);
+extern void link_dentry(struct dentry *dentry, struct dentry *parent);
+
+extern int verify_dentry(struct dentry *dentry, void *wim);
+
+
+extern struct ads_entry *inode_get_ads_entry(struct inode *inode,
+					     const char *stream_name,
+					     u16 *idx_ret);
+extern struct ads_entry *inode_add_ads(struct inode *dentry,
+				       const char *stream_name);
+
+extern void inode_remove_ads(struct inode *inode, u16 idx,
+			     struct lookup_table *lookup_table);
 
 extern int read_dentry(const u8 metadata_resource[], u64 metadata_resource_len, 
 		       u64 offset, struct dentry *dentry);
 
+
 extern int read_dentry_tree(const u8 metadata_resource[], 
 			    u64 metadata_resource_len, struct dentry *dentry);
 
 extern u8 *write_dentry_tree(const struct dentry *tree, u8 *p);
 
-
-/* Inline utility functions for WIMDentries */
-
-
 static inline bool dentry_is_root(const struct dentry *dentry)
 {
 	return dentry->parent == dentry;
@@ -359,7 +376,7 @@ static inline bool dentry_is_root(const struct dentry *dentry)
 
 static inline bool dentry_is_first_sibling(const struct dentry *dentry)
 {
-	return dentry_is_root(dentry) || dentry->parent->children == dentry;
+	return dentry_is_root(dentry) || dentry->parent->d_inode->children == dentry;
 }
 
 static inline bool dentry_is_only_child(const struct dentry *dentry)
@@ -367,29 +384,44 @@ static inline bool dentry_is_only_child(const struct dentry *dentry)
 	return dentry->next == dentry;
 }
 
+static inline bool inode_is_directory(const struct inode *inode)
+{
+	return (inode->attributes & FILE_ATTRIBUTE_DIRECTORY)
+		&& !(inode->attributes & FILE_ATTRIBUTE_REPARSE_POINT);
+}
+
 static inline bool dentry_is_directory(const struct dentry *dentry)
 {
-	return (dentry->attributes & FILE_ATTRIBUTE_DIRECTORY)
-		&& !(dentry->attributes & FILE_ATTRIBUTE_REPARSE_POINT);
+	return inode_is_directory(dentry->d_inode);
 }
 
 /* For our purposes, we consider "real" symlinks and "junction points" to both
  * be symlinks. */
+static inline bool inode_is_symlink(const struct inode *inode)
+{
+	return (inode->attributes & FILE_ATTRIBUTE_REPARSE_POINT)
+		&& ((inode->reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK) ||
+		     inode->reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT);
+}
+
 static inline bool dentry_is_symlink(const struct dentry *dentry)
 {
-	return (dentry->attributes & FILE_ATTRIBUTE_REPARSE_POINT)
-		&& ((dentry->reparse_tag == WIM_IO_REPARSE_TAG_SYMLINK) ||
-		     dentry->reparse_tag == WIM_IO_REPARSE_TAG_MOUNT_POINT);
+	return inode_is_symlink(dentry->d_inode);
+}
+
+static inline bool inode_is_regular_file(const struct inode *inode)
+{
+	return !inode_is_directory(inode) && !inode_is_symlink(inode);
 }
 
 static inline bool dentry_is_regular_file(const struct dentry *dentry)
 {
-	return !dentry_is_directory(dentry) && !dentry_is_symlink(dentry);
+	return inode_is_regular_file(dentry->d_inode);
 }
 
 static inline bool dentry_is_empty_directory(const struct dentry *dentry)
 {
-	return dentry_is_directory(dentry) && dentry->children == NULL;
+	return dentry_is_directory(dentry) && dentry->d_inode->children == NULL;
 }
 
 #endif