X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.c;h=89b57b4e65ee757dc0485c61903750b866319091;hp=5b57e3ac2967bcb260c1c9b893a091aaa5a44840;hb=1f2091f5bb860769e20105af2d9eff3a0872c57b;hpb=13c6ce3160fce7c40008d1d182325c8b42450d1e

diff --git a/src/lookup_table.c b/src/lookup_table.c
index 5b57e3ac..89b57b4e 100644
--- a/src/lookup_table.c
+++ b/src/lookup_table.c
@@ -88,8 +88,11 @@ clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
 	switch (new->resource_location) {
 #ifdef __WIN32__
 	case RESOURCE_WIN32:
+	case RESOURCE_WIN32_ENCRYPTED:
 #endif
+#ifdef WITH_FUSE
 	case RESOURCE_IN_STAGING_FILE:
+#endif
 	case RESOURCE_IN_FILE_ON_DISK:
 		BUILD_BUG_ON((void*)&old->file_on_disk !=
 			     (void*)&old->staging_file_name);
@@ -215,7 +218,7 @@ finalize_lte(struct wim_lookup_table_entry *lte)
 	#ifdef WITH_FUSE
 	if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 		unlink(lte->staging_file_name);
-		list_del(&lte->staging_list);
+		list_del(&lte->unhashed_list);
 	}
 	#endif
 	free_lookup_table_entry(lte);
@@ -275,7 +278,7 @@ for_lookup_table_entry(struct wim_lookup_table *table,
 	return 0;
 }
 
-static int
+int
 cmp_streams_by_wim_position(const void *p1, const void *p2)
 {
 	const struct wim_lookup_table_entry *lte1, *lte2;
@@ -289,40 +292,6 @@ cmp_streams_by_wim_position(const void *p1, const void *p2)
 		return 0;
 }
 
-int
-sort_stream_list_by_wim_position(struct list_head *stream_list)
-{
-	struct list_head *cur;
-	size_t num_streams;
-	struct wim_lookup_table_entry **array;
-	size_t i;
-	size_t array_size;
-
-	num_streams = 0;
-	list_for_each(cur, stream_list)
-		num_streams++;
-	array_size = num_streams * sizeof(array[0]);
-	array = MALLOC(array_size);
-	if (!array) {
-		ERROR("Failed to allocate %zu bytes to sort stream entries",
-		      array_size);
-		return WIMLIB_ERR_NOMEM;
-	}
-	cur = stream_list->next;
-	for (i = 0; i < num_streams; i++) {
-		array[i] = container_of(cur, struct wim_lookup_table_entry, staging_list);
-		cur = cur->next;
-	}
-
-	qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
-
-	INIT_LIST_HEAD(stream_list);
-	for (i = 0; i < num_streams; i++)
-		list_add_tail(&array[i]->staging_list, stream_list);
-	FREE(array);
-	return 0;
-}
-
 
 static int
 add_lte_to_array(struct wim_lookup_table_entry *lte,
@@ -357,8 +326,11 @@ for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 	qsort(lte_array, num_streams, sizeof(lte_array[0]),
 	      cmp_streams_by_wim_position);
 	ret = 0;
-	for (size_t i = 0; i < num_streams && ret == 0; i++)
+	for (size_t i = 0; i < num_streams; i++) {
 		ret = visitor(lte_array[i], arg);
+		if (ret)
+			break;
+	}
 	FREE(lte_array);
 	return ret;
 }
@@ -623,23 +595,22 @@ write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry
 	return 0;
 }
 
-
 int
-lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *ignore)
+lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	lte->real_refcnt = 0;
 	return 0;
 }
 
 int
-lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *ignore)
+lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	lte->out_refcnt = 0;
 	return 0;
 }
 
 int
-lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore)
+lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	if (lte->extracted_file != NULL) {
 		FREE(lte->extracted_file);
@@ -667,9 +638,14 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 	tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 	tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 
-	tfprintf(out, T("Hash              = 0x"));
-	print_hash(lte->hash, out);
-	tputc(T('\n'), out);
+	if (lte->unhashed) {
+		tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
+			 lte->back_inode, lte->back_stream_id);
+	} else {
+		tfprintf(out, T("Hash              = 0x"));
+		print_hash(lte->hash, out);
+		tputc(T('\n'), out);
+	}
 
 	tfprintf(out, T("Flags             = "));
 	u8 flags = lte->resource_entry.flags;
@@ -691,6 +667,7 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 		break;
 #ifdef __WIN32__
 	case RESOURCE_WIN32:
+	case RESOURCE_WIN32_ENCRYPTED:
 #endif
 	case RESOURCE_IN_FILE_ON_DISK:
 		tfprintf(out, T("File on Disk      = `%"TS"'\n"),
@@ -922,14 +899,78 @@ lookup_table_total_stream_size(struct wim_lookup_table *table)
 	return total_size;
 }
 
-void
-free_lte_list(struct list_head *list)
+struct wim_lookup_table_entry **
+retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
+{
+	wimlib_assert(lte->unhashed);
+	struct wim_inode *inode = lte->back_inode;
+	u32 stream_id = lte->back_stream_id;
+	if (stream_id == 0)
+		return &inode->i_lte;
+	else
+		for (u16 i = 0; i < inode->i_num_ads; i++)
+			if (inode->i_ads_entries[i].stream_id == stream_id)
+				return &inode->i_ads_entries[i].lte;
+	wimlib_assert(0);
+	return NULL;
+}
+
+/* Calculate the SHA1 message digest of a stream and move it from the list of
+ * unhashed streams to the stream lookup table, possibly joining it with an
+ * existing lookup table entry for an identical stream.
+ *
+ * @lte:  An unhashed lookup table entry.
+ * @lookup_table:  Lookup table for the WIM.
+ * @lte_ret:  On success, write a pointer to the resulting lookup table
+ *            entry to this location.  This will be the same as @lte
+ *            if it was inserted into the lookup table, or different if
+ *            a duplicate stream was found.
+ *
+ * Returns 0 on success; nonzero if there is an error reading the stream.
+ */
+int
+hash_unhashed_stream(struct wim_lookup_table_entry *lte,
+		     struct wim_lookup_table *lookup_table,
+		     struct wim_lookup_table_entry **lte_ret)
 {
-	struct wim_lookup_table_entry *lte, *tmp;
+	int ret;
+	struct wim_lookup_table_entry *duplicate_lte;
+	struct wim_lookup_table_entry **back_ptr;
+
+	wimlib_assert(lte->unhashed);
 
-	list_for_each_entry_safe(lte, tmp, list, staging_list) {
-		DEBUG("%p", lte);
+	/* back_ptr must be saved because @back_inode and @back_stream_id are in
+	 * union with the SHA1 message digest and will no longer be valid once
+	 * the SHA1 has been calculated. */
+	back_ptr = retrieve_lte_pointer(lte);
+
+	ret = sha1_resource(lte);
+	if (ret)
+		return ret;
+
+	/* Look for a duplicate stream */
+	duplicate_lte = __lookup_resource(lookup_table, lte->hash);
+	list_del(&lte->unhashed_list);
+	if (duplicate_lte) {
+		/* We have a duplicate stream.  Transfer the reference counts
+		 * from this stream to the duplicate, update the reference to
+		 * this stream (in an inode or ads_entry) to point to the
+		 * duplicate, then free this stream. */
+		wimlib_assert(!(duplicate_lte->unhashed));
+		duplicate_lte->refcnt += lte->refcnt;
+		duplicate_lte->out_refcnt += lte->refcnt;
+		*back_ptr = duplicate_lte;
 		free_lookup_table_entry(lte);
+		lte = duplicate_lte;
+	} else {
+		/* No duplicate stream, so we need to insert
+		 * this stream into the lookup table and treat
+		 * it as a hashed stream. */
+		lookup_table_insert(lookup_table, lte);
+		lte->unhashed = 0;
+	}
+	if (lte_ret)
+		*lte_ret = lte;
+	return 0;
 }
-	INIT_LIST_HEAD(list);
-}
+