X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fhardlink.c;h=3f316f1d993d1f84fa9852fde53e9ec900ad3f09;hp=35fb1acd04c5cd261ea5832d962ceb74458fb50b;hb=48fbbd68e82d879e2cb6f384c1440e81373338c7;hpb=10a87017a0a82d34ed3981e1f5e586b5b8613e3f

diff --git a/src/hardlink.c b/src/hardlink.c
index 35fb1acd..3f316f1d 100644
--- a/src/hardlink.c
+++ b/src/hardlink.c
@@ -5,7 +5,7 @@
  */
 
 /*
- * Copyright (C) 2012, 2013 Biggers
+ * Copyright (C) 2012, 2013 Eric Biggers
  *
  * This file is part of wimlib, a library for working with WIM files.
  *
@@ -23,10 +23,14 @@
  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  */
 
-#include "wimlib_internal.h"
-#include "dentry.h"
-#include "list.h"
-#include "lookup_table.h"
+#ifdef HAVE_CONFIG_H
+#  include "config.h"
+#endif
+
+#include "wimlib/capture.h"
+#include "wimlib/dentry.h"
+#include "wimlib/error.h"
+#include "wimlib/lookup_table.h"
 
 /*                             NULL        NULL
  *                              ^           ^
@@ -49,35 +53,9 @@
  *                            -----------------
  */
 
-/* Hash table to find inodes, identified by their inode ID.
- * */
-struct wim_inode_table {
-	/* Fields for the hash table */
-	struct hlist_head *array;
-	u64 num_entries;
-	u64 capacity;
-
-	/*
-	 * Linked list of "extra" inodes.  These may be:
-	 *
-	 * - inodes with link count 1, which are all allowed to have 0 for their
-	 *   inode number, meaning we cannot insert them into the hash table
-	 *   before calling assign_inode_numbers().
-         *
-	 * - Groups we create ourselves by splitting a nominal inode due to
-	 *   inconsistencies in the dentries.  These inodes will share a inode
-	 *   number with some other inode until assign_inode_numbers() is
-	 *   called.
-	 */
-	struct hlist_head extra_inodes;
-};
-
-static inline void destroy_inode_table(struct wim_inode_table *table)
-{
-	FREE(table->array);
-}
 
-static int init_inode_table(struct wim_inode_table *table, size_t capacity)
+int
+init_inode_table(struct wim_inode_table *table, size_t capacity)
 {
 	table->array = CALLOC(capacity, sizeof(table->array[0]));
 	if (!table->array) {
@@ -86,11 +64,12 @@ static int init_inode_table(struct wim_inode_table *table, size_t capacity)
 	}
 	table->num_entries  = 0;
 	table->capacity     = capacity;
-	INIT_HLIST_HEAD(&table->extra_inodes);
+	INIT_LIST_HEAD(&table->extra_inodes);
 	return 0;
 }
 
-static inline size_t inode_link_count(const struct wim_inode *inode)
+static inline size_t
+inode_link_count(const struct wim_inode *inode)
 {
 	const struct list_head *cur;
 	size_t size = 0;
@@ -102,20 +81,17 @@ static inline size_t inode_link_count(const struct wim_inode *inode)
 /* Insert a dentry into the inode table based on the inode number of the
  * attached inode (which came from the hard link group ID field of the on-disk
  * WIM dentry) */
-static int inode_table_insert(struct wim_dentry *dentry, void *__table)
+static int
+inode_table_insert(struct wim_dentry *dentry, void *_table)
 {
-	struct wim_inode_table *table = __table;
+	struct wim_inode_table *table = _table;
 	struct wim_inode *d_inode = dentry->d_inode;
 
 	if (d_inode->i_ino == 0) {
 		/* A dentry with a hard link group ID of 0 indicates that it's
 		 * in a hard link group by itself.  Add it to the list of extra
 		 * inodes rather than inserting it into the hash lists. */
-		hlist_add_head(&d_inode->i_hlist, &table->extra_inodes);
-
-		wimlib_assert(d_inode->i_dentry.next == &dentry->d_alias);
-		wimlib_assert(d_inode->i_dentry.prev == &dentry->d_alias);
-		wimlib_assert(d_inode->i_nlink == 1);
+		list_add_tail(&d_inode->i_list, &table->extra_inodes);
 	} else {
 		size_t pos;
 		struct wim_inode *inode;
@@ -125,8 +101,16 @@ static int inode_table_insert(struct wim_dentry *dentry, void *__table)
 		pos = d_inode->i_ino % table->capacity;
 		hlist_for_each_entry(inode, cur, &table->array[pos], i_hlist) {
 			if (inode->i_ino == d_inode->i_ino) {
+				if (unlikely((inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) ||
+					     (d_inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)))
+				{
+					ERROR("Unsupported directory hard link "
+					      "\"%"TS"\" <=> \"%"TS"\"",
+					      dentry_full_path(dentry),
+					      dentry_full_path(inode_first_dentry(inode)));
+					return WIMLIB_ERR_INVALID_METADATA_RESOURCE;
+				}
 				inode_add_dentry(dentry, inode);
-				inode->i_nlink++;
 				return 0;
 			}
 		}
@@ -135,10 +119,6 @@ static int inode_table_insert(struct wim_dentry *dentry, void *__table)
 		 * it to the table. */
 		hlist_add_head(&d_inode->i_hlist, &table->array[pos]);
 
-		wimlib_assert(d_inode->i_dentry.next == &dentry->d_alias);
-		wimlib_assert(d_inode->i_dentry.prev == &dentry->d_alias);
-		wimlib_assert(d_inode->i_nlink == 1);
-
 		/* XXX Make the table grow when too many entries have been
 		 * inserted. */
 		table->num_entries++;
@@ -146,25 +126,111 @@ static int inode_table_insert(struct wim_dentry *dentry, void *__table)
 	return 0;
 }
 
-static void print_inode_dentries(const struct wim_inode *inode)
+static struct wim_inode *
+inode_table_get_inode(struct wim_inode_table *table, u64 ino, u64 devno)
+{
+	u64 hash = hash_u64(hash_u64(ino) + hash_u64(devno));
+	size_t pos = hash % table->capacity;
+	struct wim_inode *inode;
+	struct hlist_node *cur;
+
+	hlist_for_each_entry(inode, cur, &table->array[pos], i_hlist) {
+		if (inode->i_ino == ino && inode->i_devno == devno) {
+			DEBUG("Using existing inode {devno=%"PRIu64", ino=%"PRIu64"}",
+			      devno, ino);
+			inode->i_nlink++;
+			return inode;
+		}
+	}
+	inode = new_timeless_inode();
+	if (inode) {
+		inode->i_ino = ino;
+		inode->i_devno = devno;
+		hlist_add_head(&inode->i_hlist, &table->array[pos]);
+		table->num_entries++;
+	}
+	return inode;
+}
+
+void
+inode_ref_streams(struct wim_inode *inode)
+{
+	for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+		struct wim_lookup_table_entry *lte;
+		lte = inode_stream_lte_resolved(inode, i);
+		if (lte)
+			lte->refcnt++;
+	}
+}
+
+/* Given a directory entry with the name @name for the file with the inode
+ * number @ino and device number @devno, create a new WIM dentry with an
+ * associated inode, where the inode is shared if an inode with the same @ino
+ * and @devno has already been created.  On success, the new WIM dentry is
+ * written to *dentry_ret, and its inode has i_nlink > 1 if a previously
+ * existing inode was used.
+ */
+int
+inode_table_new_dentry(struct wim_inode_table *table, const tchar *name,
+		       u64 ino, u64 devno, bool noshare,
+		       struct wim_dentry **dentry_ret)
+{
+	struct wim_dentry *dentry;
+	struct wim_inode *inode;
+	int ret;
+
+	if (noshare) {
+		/* File that cannot be hardlinked--- Return a new inode with its
+		 * inode and device numbers left at 0. */
+		ret = new_dentry_with_timeless_inode(name, &dentry);
+		if (ret)
+			return ret;
+		list_add_tail(&dentry->d_inode->i_list, &table->extra_inodes);
+	} else {
+		/* File that can be hardlinked--- search the table for an
+		 * existing inode matching the inode number and device;
+		 * otherwise create a new inode. */
+		ret = new_dentry(name, &dentry);
+		if (ret)
+			return ret;
+		inode = inode_table_get_inode(table, ino, devno);
+		if (!inode) {
+			free_dentry(dentry);
+			return WIMLIB_ERR_NOMEM;
+		}
+		/* If using an existing inode, we need to gain a reference to
+		 * each of its streams. */
+		if (inode->i_nlink > 1)
+			inode_ref_streams(inode);
+		dentry->d_inode = inode;
+		inode_add_dentry(dentry, inode);
+	}
+	*dentry_ret = dentry;
+	return 0;
+}
+
+static inline void
+print_inode_dentries(const struct wim_inode *inode)
 {
 	struct wim_dentry *dentry;
 	inode_for_each_dentry(dentry, inode)
-		printf("`%s'\n", dentry->full_path_utf8);
+		tfprintf(stderr, T("%"TS"\n"), dentry_full_path(dentry));
 }
 
-static void inconsistent_inode(const struct wim_inode *inode)
+static void
+inconsistent_inode(const struct wim_inode *inode)
 {
-	ERROR("An inconsistent hard link group that cannot be corrected has "
-	      "been detected");
-	ERROR("The dentries are located at the following paths:");
-#ifdef ENABLE_ERROR_MESSAGES
-	print_inode_dentries(inode);
-#endif
+	if (wimlib_print_errors) {
+		ERROR("An inconsistent hard link group that cannot be corrected has "
+		      "been detected");
+		ERROR("The dentries are located at the following paths:");
+		print_inode_dentries(inode);
+	}
 }
 
-static bool ref_inodes_consistent(const struct wim_inode * restrict ref_inode_1,
-				  const struct wim_inode * restrict ref_inode_2)
+static bool
+ref_inodes_consistent(const struct wim_inode * restrict ref_inode_1,
+		      const struct wim_inode * restrict ref_inode_2)
 {
 	wimlib_assert(ref_inode_1 != ref_inode_2);
 
@@ -187,8 +253,9 @@ static bool ref_inodes_consistent(const struct wim_inode * restrict ref_inode_1,
 	return true;
 }
 
-static bool inodes_consistent(const struct wim_inode * restrict ref_inode,
-			      const struct wim_inode * restrict inode)
+static bool
+inodes_consistent(const struct wim_inode * restrict ref_inode,
+		  const struct wim_inode * restrict inode)
 {
 	wimlib_assert(ref_inode != inode);
 
@@ -212,7 +279,8 @@ static bool inodes_consistent(const struct wim_inode * restrict ref_inode,
 }
 
 /* Fix up a "true" inode and check for inconsistencies */
-static int fix_true_inode(struct wim_inode *inode, struct hlist_head *inode_list)
+static int
+fix_true_inode(struct wim_inode *inode, struct list_head *inode_list)
 {
 	struct wim_dentry *dentry;
 	struct wim_dentry *ref_dentry = NULL;
@@ -233,8 +301,8 @@ static int fix_true_inode(struct wim_inode *inode, struct hlist_head *inode_list
 	}
 
 	ref_inode = ref_dentry->d_inode;
-	ref_inode->i_nlink = 1;
-	hlist_add_head(&ref_inode->i_hlist, inode_list);
+	wimlib_assert(ref_inode->i_nlink == 1);
+	list_add_tail(&ref_inode->i_list, inode_list);
 
 	list_del(&inode->i_dentry);
 	list_add(&ref_inode->i_dentry, &ref_dentry->d_alias);
@@ -243,11 +311,10 @@ static int fix_true_inode(struct wim_inode *inode, struct hlist_head *inode_list
 		if (dentry != ref_dentry) {
 			if (!inodes_consistent(ref_inode, dentry->d_inode)) {
 				inconsistent_inode(ref_inode);
-				return WIMLIB_ERR_INVALID_DENTRY;
+				return WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 			}
 			/* Free the unneeded `struct wim_inode'. */
-			dentry->d_inode->i_hlist.next = NULL;
-			dentry->d_inode->i_hlist.pprev = NULL;
+			wimlib_assert(dentry->d_inode->i_nlink == 1);
 			free_inode(dentry->d_inode);
 			dentry->d_inode = ref_inode;
 			ref_inode->i_nlink++;
@@ -274,16 +341,15 @@ static int fix_true_inode(struct wim_inode *inode, struct hlist_head *inode_list
  * wim_inode's.  There will be just one `struct wim_inode' for each hard link
  * group remaining.
  */
-static int fix_nominal_inode(struct wim_inode *inode,
-			     struct hlist_head *inode_list)
+static int
+fix_nominal_inode(struct wim_inode *inode, struct list_head *inode_list,
+		  bool *ino_changes_needed)
 {
 	struct wim_dentry *dentry;
 	struct hlist_node *cur, *tmp;
 	int ret;
 	size_t num_true_inodes;
 
-	wimlib_assert(inode->i_nlink == inode_link_count(inode));
-
 	LIST_HEAD(dentries_with_data_streams);
 	LIST_HEAD(dentries_with_no_data_streams);
 	HLIST_HEAD(true_inodes);
@@ -311,9 +377,10 @@ static int fix_nominal_inode(struct wim_inode *inode,
 	 * inode to be a true inode */
 	if (list_empty(&dentries_with_data_streams)) {
 	#ifdef ENABLE_DEBUG
-		if (inode->i_nlink > 1) {
+		unsigned nominal_group_size = inode_link_count(inode);
+		if (nominal_group_size > 1) {
 			DEBUG("Found link group of size %u without "
-			      "any data streams:", inode->i_nlink);
+			      "any data streams:", nominal_group_size);
 			print_inode_dentries(inode);
 			DEBUG("We are going to interpret it as true "
 			      "link group, provided that the dentries "
@@ -353,13 +420,14 @@ next_dentry_2:
          * streamless dentries to. */
 	if (!list_empty(&dentries_with_no_data_streams)) {
 		if (num_true_inodes != 1) {
-			ERROR("Hard inode ambiguity detected!");
+			ERROR("Hard link ambiguity detected!");
 			ERROR("We split up inode 0x%"PRIx64" due to "
 			      "inconsistencies,", inode->i_ino);
 			ERROR("but dentries with no stream information remained. "
 			      "We don't know which inode");
 			ERROR("to assign them to.");
-			return WIMLIB_ERR_INVALID_DENTRY;
+			ret = WIMLIB_ERR_INVALID_METADATA_RESOURCE;
+			goto out_cleanup_true_inode_list;
 		}
 		inode = container_of(true_inodes.first, struct wim_inode, i_hlist);
 		/* Assign the streamless dentries to the one and only true
@@ -367,48 +435,58 @@ next_dentry_2:
 		list_for_each_entry(dentry, &dentries_with_no_data_streams, tmp_list)
 			inode_add_dentry(dentry, inode);
 	}
-	#ifdef ENABLE_DEBUG
         if (num_true_inodes != 1) {
+	#ifdef ENABLE_DEBUG
 		inode = container_of(true_inodes.first, struct wim_inode, i_hlist);
 
-		printf("Split nominal inode 0x%"PRIx64" into %zu "
-		       "inodes:\n",
-		       inode->i_ino, num_true_inodes);
-		puts("------------------------------------------------------------------------------");
+		tprintf(T("Split nominal inode 0x%"PRIx64" into %zu "
+			  "inodes:\n"), inode->i_ino, num_true_inodes);
+		tputs(T("----------------------------------------------------"
+			"--------------------------"));
 		size_t i = 1;
 		hlist_for_each_entry(inode, cur, &true_inodes, i_hlist) {
-			printf("[Split inode %zu]\n", i++);
+			tprintf(T("[Split inode %zu]\n"), i++);
 			print_inode_dentries(inode);
-			putchar('\n');
+			tputchar(T('\n'));
 		}
-		puts("------------------------------------------------------------------------------");
-        }
+		tputs(T("----------------------------------------------------"
+			"--------------------------"));
 	#endif
+		*ino_changes_needed = true;
+        }
 
 	hlist_for_each_entry_safe(inode, cur, tmp, &true_inodes, i_hlist) {
+		hlist_del_init(&inode->i_hlist);
 		ret = fix_true_inode(inode, inode_list);
-		if (ret != 0)
-			return ret;
+		if (ret)
+			goto out_cleanup_true_inode_list;
 	}
-	return 0;
+	ret = 0;
+	goto out;
+out_cleanup_true_inode_list:
+	hlist_for_each_entry_safe(inode, cur, tmp, &true_inodes, i_hlist)
+		hlist_del_init(&inode->i_hlist);
+out:
+	return ret;
 }
 
-static int fix_inodes(struct wim_inode_table *table,
-		      struct hlist_head *inode_list)
+static int
+fix_inodes(struct wim_inode_table *table, struct list_head *inode_list,
+	   bool *ino_changes_needed)
 {
 	struct wim_inode *inode;
 	struct hlist_node *cur, *tmp;
 	int ret;
-	INIT_HLIST_HEAD(inode_list);
+	INIT_LIST_HEAD(inode_list);
 	for (u64 i = 0; i < table->capacity; i++) {
 		hlist_for_each_entry_safe(inode, cur, tmp, &table->array[i], i_hlist) {
-			ret = fix_nominal_inode(inode, inode_list);
-			if (ret != 0)
+			hlist_del_init(&inode->i_hlist);
+			ret = fix_nominal_inode(inode, inode_list, ino_changes_needed);
+			if (ret)
 				return ret;
 		}
 	}
-	hlist_for_each_safe(cur, tmp, &table->extra_inodes)
-		hlist_add_head(cur, inode_list);
+	list_splice_tail(&table->extra_inodes, inode_list);
 	return 0;
 }
 
@@ -436,39 +514,102 @@ static int fix_inodes(struct wim_inode_table *table,
  * WIM) is examined for consistency and may be split into multiple "true" inodes
  * that are maximally sized consistent sets of dentries.
  *
- * Return 0 on success; WIMLIB_ERR_NOMEM or WIMLIB_ERR_INVALID_DENTRY on
- * failure.  On success, the list of "true" inodes, linked by the i_hlist field,
+ * On success, the list of "true" inodes, linked by the i_hlist field,
  * is returned in the hlist @inode_list.
+ *
+ * Return values:
+ *	WIMLIB_ERR_SUCCESS (0)
+ *	WIMLIB_ERR_INVALID_METADATA_RESOURCE
+ *	WIMLIB_ERR_NOMEM
  */
-int dentry_tree_fix_inodes(struct wim_dentry *root, struct hlist_head *inode_list)
+int
+dentry_tree_fix_inodes(struct wim_dentry *root, struct list_head *inode_list)
 {
 	struct wim_inode_table inode_tab;
 	int ret;
+	bool ino_changes_needed;
+	struct wim_inode *inode;
 
 	DEBUG("Inserting dentries into inode table");
 	ret = init_inode_table(&inode_tab, 9001);
-	if (ret != 0)
-		return ret;
+	if (ret)
+		goto out;
 
-	for_dentry_in_tree(root, inode_table_insert, &inode_tab);
+	ret = for_dentry_in_tree(root, inode_table_insert, &inode_tab);
+	if (ret)
+		goto out_destroy_inode_table;
 
 	DEBUG("Cleaning up the hard link groups");
-	ret = fix_inodes(&inode_tab, inode_list);
+	ino_changes_needed = false;
+	ret = fix_inodes(&inode_tab, inode_list, &ino_changes_needed);
+	if (ret)
+		goto out_destroy_inode_table;
+
+	if (ino_changes_needed) {
+		u64 cur_ino = 1;
+
+		WARNING("Re-assigning inode numbers due to inode inconsistencies");
+		list_for_each_entry(inode, inode_list, i_list) {
+			if (inode->i_nlink > 1)
+				inode->i_ino = cur_ino++;
+			else
+				inode->i_ino = 0;
+		}
+	}
+	/* On success, all the inodes have been moved to the image inode list,
+	 * so there's no need to delete from from the hash lists in the inode
+	 * table before freeing the hash buckets array directly. */
+	ret = 0;
+	goto out_destroy_inode_table_raw;
+out_destroy_inode_table:
+	for (size_t i = 0; i < inode_tab.capacity; i++) {
+		struct hlist_node *cur, *tmp;
+		hlist_for_each_entry_safe(inode, cur, tmp, &inode_tab.array[i], i_hlist)
+			hlist_del_init(&inode->i_hlist);
+	}
+	{
+		struct wim_inode *tmp;
+		list_for_each_entry_safe(inode, tmp, &inode_tab.extra_inodes, i_list)
+			list_del_init(&inode->i_list);
+	}
+out_destroy_inode_table_raw:
 	destroy_inode_table(&inode_tab);
+out:
 	return ret;
 }
 
-/* Assign inode numbers to a list of inodes and return the next available
- * number. */
-u64 assign_inode_numbers(struct hlist_head *inode_list)
+/* Assign consecutive inode numbers to a new set of inodes from the inode table,
+ * and append the inodes to a single list @head that contains the inodes already
+ * existing in the WIM image.  */
+void
+inode_table_prepare_inode_list(struct wim_inode_table *table,
+			       struct list_head *head)
 {
-	DEBUG("Assigning inode numbers");
-	struct wim_inode *inode;
-	struct hlist_node *cur;
+	struct wim_inode *inode, *tmp_inode;
+	struct hlist_node *cur, *tmp;
 	u64 cur_ino = 1;
-	hlist_for_each_entry(inode, cur, inode_list, i_hlist) {
-		inode->i_ino = cur_ino;
-		cur_ino++;
+
+	/* Re-assign inode numbers in the existing list to avoid duplicates. */
+	list_for_each_entry(inode, head, i_list)
+		inode->i_ino = cur_ino++;
+
+	/* Assign inode numbers to the new inodes and move them to the image's
+	 * inode list. */
+	for (size_t i = 0; i < table->capacity; i++) {
+		hlist_for_each_entry_safe(inode, cur, tmp, &table->array[i], i_hlist)
+		{
+			inode->i_ino = cur_ino++;
+			inode->i_devno = 0;
+			list_add_tail(&inode->i_list, head);
+		}
+		INIT_HLIST_HEAD(&table->array[i]);
+	}
+	list_for_each_entry_safe(inode, tmp_inode, &table->extra_inodes, i_list)
+	{
+		inode->i_ino = cur_ino++;
+		inode->i_devno = 0;
+		list_add_tail(&inode->i_list, head);
 	}
-	return cur_ino;
+	INIT_LIST_HEAD(&table->extra_inodes);
+	table->num_entries = 0;
 }