Fix some issues with ACL handling:

  * ACL storage is no longer erased before a group of entries are added.
  * ACL text creation no longer tries to skip over non-existent text.
  * UTF8 encoder no longer blows up on invalid wide characters.
  * Fixed ACL state management for default ACLs.
Also, publicize function for obtaining text-format ACL in various
formats.  The interface is now extensible through a "flags" argument
that allows you to select a variant format.

5 files changed, 89 insertions, 53 deletions

diff --git a/lib/libarchive/archive_entry.c b/lib/libarchive/archive_entry.c
index 80e3da1c5afc..6ddbd959a97c 100644
--- a/lib/libarchive/archive_entry.c
+++ b/lib/libarchive/archive_entry.c
@@ -747,7 +747,7 @@ archive_entry_acl_reset(struct archive_entry *entry, int want_type)
 		entry->acl_state = ARCHIVE_ENTRY_ACL_USER_OBJ;
 	else
 		entry->acl_state = 0;
-	entry->acl_p = NULL;
+	entry->acl_p = entry->acl_head;
 	return (count);
 }
 
@@ -768,30 +768,34 @@ archive_entry_acl_next(struct archive_entry *entry, int want_type, int *type,
 	 * (reading from list), or an entry type (retrieve that type
 	 * from ae_stat.st_mode).
 	 */
-
 	if (entry->acl_state == 0)
 		return (ARCHIVE_WARN);
 
-	if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0  &&
-	    entry->acl_state > 0) {
-		*type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS;
-		*tag = entry->acl_state;
+	/* The first three access entries are special. */
+	if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) {
 		switch (entry->acl_state) {
 		case ARCHIVE_ENTRY_ACL_USER_OBJ:
 			*permset = (entry->ae_stat.st_mode >> 6) & 7;
+			*type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS;
+			*tag = ARCHIVE_ENTRY_ACL_USER_OBJ;
 			entry->acl_state = ARCHIVE_ENTRY_ACL_GROUP_OBJ;
-			break;
+			return (ARCHIVE_OK);
 		case ARCHIVE_ENTRY_ACL_GROUP_OBJ:
 			*permset = (entry->ae_stat.st_mode >> 3) & 7;
+			*type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS;
+			*tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ;
 			entry->acl_state = ARCHIVE_ENTRY_ACL_OTHER;
-			break;
+			return (ARCHIVE_OK);
 		case ARCHIVE_ENTRY_ACL_OTHER:
 			*permset = entry->ae_stat.st_mode & 7;
+			*type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS;
+			*tag = ARCHIVE_ENTRY_ACL_OTHER;
 			entry->acl_state = -1;
 			entry->acl_p = entry->acl_head;
+			return (ARCHIVE_OK);
+		default:
 			break;
 		}
-		return (ARCHIVE_OK);
 	}
 
 	while (entry->acl_p != NULL && (entry->acl_p->type & want_type) == 0)
@@ -810,24 +814,19 @@ archive_entry_acl_next(struct archive_entry *entry, int want_type, int *type,
 }
 
 /*
- * Generate a text version of the ACL.  The format here varies
- * from POSIX.1e in a couple of useful ways:
- *
- *  * An additional colon-delimited field holds the numeric uid
- *    or gid.  For proper archiving, it is essential to have both
- *    the uname/gname and the uid/gid.
- *
- *  * You can request a single text holding both access and default
- *    entries.  In this case, each default entry is prefixed with
- *    "default:".
+ * Generate a text version of the ACL.  The flags parameter controls
+ * the style of the generated ACL.
  */
 const wchar_t *
-__archive_entry_acl_text_w(struct archive_entry *entry, int type)
+archive_entry_acl_text_w(struct archive_entry *entry, int flags)
 {
 	int count;
 	int length;
 	const wchar_t *wname;
+	const wchar_t *prefix;
+	wchar_t separator;
 	struct ae_acl *ap;
+	int id;
 	wchar_t *wp;
 
 	if (entry->acl_text_w != NULL) {
@@ -835,14 +834,15 @@ __archive_entry_acl_text_w(struct archive_entry *entry, int type)
 		entry->acl_text_w = NULL;
 	}
 
+	separator = L',';
 	count = 0;
 	length = 0;
 	ap = entry->acl_head;
 	while (ap != NULL) {
-		if ((ap->type & type) != 0) {
+		if ((ap->type & flags) != 0) {
 			count++;
-			if (type & (ARCHIVE_ENTRY_ACL_TYPE_ACCESS |
-				ARCHIVE_ENTRY_ACL_TYPE_DEFAULT))
+			if ((flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) &&
+			    (ap->type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT))
 				length += 8; /* "default:" */
 			length += 5; /* tag name */
 			length += 1; /* colon */
@@ -858,7 +858,7 @@ __archive_entry_acl_text_w(struct archive_entry *entry, int type)
 		ap = ap->next;
 	}
 
-	if (count > 0 && ((type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0)) {
+	if (count > 0 && ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0)) {
 		length += 10; /* "user::rwx\n" */
 		length += 11; /* "group::rwx\n" */
 		length += 11; /* "other::rwx\n" */
@@ -870,7 +870,7 @@ __archive_entry_acl_text_w(struct archive_entry *entry, int type)
 	/* Now, allocate the string and actually populate it. */
 	wp = entry->acl_text_w = malloc(length * sizeof(wchar_t));
 	count = 0;
-	if ((type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) {
+	if ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) {
 		append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_USER_OBJ, NULL,
 		    entry->ae_stat.st_mode & 0700, -1);
 		*wp++ = ',';
@@ -885,24 +885,39 @@ __archive_entry_acl_text_w(struct archive_entry *entry, int type)
 		while (ap != NULL) {
 			if ((ap->type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) {
 				wname = aes_get_wcs(&ap->name);
-				*wp++ = ',';
+				*wp++ = separator;
+				if (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID)
+					id = ap->id;
+				else
+					id = -1;
 				append_entry_w(&wp, NULL, ap->tag, wname,
-				    ap->permset, ap->id);
+				    ap->permset, id);
 				count++;
 			}
 			ap = ap->next;
 		}
 	}
 
-	if ((type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) {
+
+	if ((flags & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) {
+		if (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT)
+			prefix = L"default:";
+		else
+			prefix = NULL;
 		ap = entry->acl_head;
+		count = 0;
 		while (ap != NULL) {
 			if ((ap->type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) {
 				wname = aes_get_wcs(&ap->name);
 				if (count > 0)
-					*wp++ = ',';
-				append_entry_w(&wp, L"default:", ap->tag,
-				    wname, ap->permset, ap->id);
+					*wp++ = separator;
+				if (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID)
+					id = ap->id;
+				else
+					id = -1;
+				append_entry_w(&wp, prefix, ap->tag,
+				    wname, ap->permset, id);
+				count ++;
 			}
 			ap = ap->next;
 		}
@@ -955,9 +970,10 @@ append_entry_w(wchar_t **wp, const wchar_t *prefix, int tag,
 	}
 	*wp += wcslen(*wp);
 	*(*wp)++ = L':';
-	if (wname != NULL)
+	if (wname != NULL) {
 		wcscpy(*wp, wname);
-	*wp += wcslen(*wp);
+		*wp += wcslen(*wp);
+	}
 	*(*wp)++ = L':';
 	*(*wp)++ = (perm & 0444) ? L'r' : L'-';
 	*(*wp)++ = (perm & 0222) ? L'w' : L'-';
@@ -990,8 +1006,6 @@ __archive_entry_acl_parse_w(struct archive_entry *entry,
 	namebuff = NULL;
 	namebuff_length = 0;
 
-	archive_entry_acl_clear(entry);
-
 	while (text != NULL  &&  *text != L'\0') {
 		next_field_w(&text, &start, &end, &sep);
 		if (sep != L':')
diff --git a/lib/libarchive/archive_entry.h b/lib/libarchive/archive_entry.h
index 0471f011fe1b..d749e9b7ac87 100644
--- a/lib/libarchive/archive_entry.h
+++ b/lib/libarchive/archive_entry.h
@@ -167,6 +167,21 @@ int	 archive_entry_acl_next_w(struct archive_entry *, int want_type,
 	     int *type, int *permset, int *tag, int *qual,
 	     const wchar_t **name);
 
+/*
+ * Construct a text-format ACL.  The flags argument is a bitmask that
+ * can include any of the following:
+ *
+ * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - Include access entries.
+ * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - Include default entries.
+ * ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID - Include extra numeric ID field in
+ *    each ACL entry.  (As used by 'star'.)
+ * ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT - Include "default:" before each
+ *    default ACL entry.
+ */
+#define	ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID	1024
+#define	ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT	2048
+const wchar_t	*archive_entry_acl_text_w(struct archive_entry *, int flags);
+
 /* Return a count of entries matching 'want_type' */
 int	 archive_entry_acl_count(struct archive_entry *, int want_type);
 
diff --git a/lib/libarchive/archive_private.h b/lib/libarchive/archive_private.h
index 21655e7a2614..7cb654ed5e97 100644
--- a/lib/libarchive/archive_private.h
+++ b/lib/libarchive/archive_private.h
@@ -243,6 +243,5 @@ int	__archive_read_register_compression(struct archive *a,
  */
 int		 __archive_entry_acl_parse_w(struct archive_entry *,
 		     const wchar_t *, int type);
-const wchar_t	*__archive_entry_acl_text_w(struct archive_entry *, int type);
 
 #endif
diff --git a/lib/libarchive/archive_read_extract.c b/lib/libarchive/archive_read_extract.c
index f1e38cf7377c..034e5b7e0880 100644
--- a/lib/libarchive/archive_read_extract.c
+++ b/lib/libarchive/archive_read_extract.c
@@ -81,7 +81,7 @@ static int	mkdirpath_recursive(char *path);
 static int	mksubdir(char *path);
 #ifdef HAVE_POSIX_ACL
 static int	set_acl(struct archive *, struct archive_entry *,
-		    acl_type_t, int archive_entry_acl_type);
+		    acl_type_t, int archive_entry_acl_type, const char *tn);
 #endif
 static int	set_acls(struct archive *, struct archive_entry *);
 static int	set_extended_perm(struct archive *, struct archive_entry *,
@@ -845,18 +845,18 @@ set_acls(struct archive *a, struct archive_entry *entry)
 	int		 ret;
 
 	ret = set_acl(a, entry, ACL_TYPE_ACCESS,
-	    ARCHIVE_ENTRY_ACL_TYPE_ACCESS);
+	    ARCHIVE_ENTRY_ACL_TYPE_ACCESS, "access");
 	if (ret != ARCHIVE_OK)
 		return (ret);
 	ret = set_acl(a, entry, ACL_TYPE_DEFAULT,
-	    ARCHIVE_ENTRY_ACL_TYPE_DEFAULT);
+	    ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, "default");
 	return (ret);
 }
 
 
 static int
 set_acl(struct archive *a, struct archive_entry *entry, acl_type_t acl_type,
-    int ae_requested_type)
+    int ae_requested_type, const char *typename)
 {
 	acl_t		 acl;
 	acl_entry_t	 acl_entry;
@@ -907,7 +907,7 @@ set_acl(struct archive *a, struct archive_entry *entry, acl_type_t acl_type,
 
 	name = archive_entry_pathname(entry);
 	if (acl_set_file(name, acl_type, acl) != 0) {
-		archive_set_error(a, errno, "Failed to set acl");
+		archive_set_error(a, errno, "Failed to set %s acl", typename);
 		ret = ARCHIVE_WARN;
 	}
 	acl_free(acl);
diff --git a/lib/libarchive/archive_write_set_format_pax.c b/lib/libarchive/archive_write_set_format_pax.c
index d0cec2c16a0b..ab060fd50726 100644
--- a/lib/libarchive/archive_write_set_format_pax.c
+++ b/lib/libarchive/archive_write_set_format_pax.c
@@ -197,7 +197,9 @@ add_pax_attr_w(struct archive_string *as, const char *key, const wchar_t *wval)
 	utf8len = 0;
 	for (wp = wval; *wp != L'\0'; ) {
 		wc = *wp++;
-		if (wc <= 0x7f)
+		if (wc <= 0) {
+			/* Ignore negative values. */
+		} else if (wc <= 0x7f)
 			utf8len++;
 		else if (wc <= 0x7ff)
 			utf8len += 2;
@@ -214,7 +216,9 @@ add_pax_attr_w(struct archive_string *as, const char *key, const wchar_t *wval)
 	utf8_value = malloc(utf8len + 1);
 	for (wp = wval, p = utf8_value; *wp != L'\0'; ) {
 		wc = *wp++;
-		if (wc <= 0x7f) {
+		if (wc <= 0) {
+			/* Ignore negative values. */
+		} else if (wc <= 0x7f) {
 			*p++ = (char)wc;
 		} else if (wc <= 0x7ff) {
 			p[0] = 0xc0 | ((wc >> 6) & 0x1f);
@@ -485,16 +489,17 @@ archive_write_pax_header(struct archive *a,
 	 * avoid writing an mtime attribute just to handle a
 	 * high-resolution timestamp in "restricted pax" mode.
 	 */
-	if ((st_main->st_mtime < 0) || (st_main->st_mtime >= 0x7fffffff))
+	if (!need_extension &&
+	    ((st_main->st_mtime < 0) || (st_main->st_mtime >= 0x7fffffff)))
 		need_extension = 1;
 
 	/* If there are non-trivial ACL entries, we need an extension. */
-	if (archive_entry_acl_count(entry_original,
+	if (!need_extension && archive_entry_acl_count(entry_original,
 		ARCHIVE_ENTRY_ACL_TYPE_ACCESS) > 0)
 		need_extension = 1;
 
 	/* If there are non-trivial ACL entries, we need an extension. */
-	if (archive_entry_acl_count(entry_original,
+	if (!need_extension && archive_entry_acl_count(entry_original,
 		ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) > 0)
 		need_extension = 1;
 
@@ -530,15 +535,18 @@ archive_write_pax_header(struct archive *a,
 			add_pax_attr(&(pax->pax_header), "SCHILY.fflags", p);
 
 		/* I use star-compatible ACL attributes. */
-		wp = __archive_entry_acl_text_w(entry_original,
-		    ARCHIVE_ENTRY_ACL_TYPE_ACCESS);
+		wp = archive_entry_acl_text_w(entry_original,
+		    ARCHIVE_ENTRY_ACL_TYPE_ACCESS |
+		    ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID);
 		if (wp != NULL && *wp != L'\0')
-			add_pax_attr_w(&(pax->pax_header), "SCHILY.acl.access", wp);
-		wp = __archive_entry_acl_text_w(entry_original,
-		    ARCHIVE_ENTRY_ACL_TYPE_DEFAULT);
+			add_pax_attr_w(&(pax->pax_header),
+			    "SCHILY.acl.access", wp);
+		wp = archive_entry_acl_text_w(entry_original,
+		    ARCHIVE_ENTRY_ACL_TYPE_DEFAULT |
+		    ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID);
 		if (wp != NULL && *wp != L'\0')
-			add_pax_attr_w(&(pax->pax_header), "SCHILY.acl.default",
-			    wp);
+			add_pax_attr_w(&(pax->pax_header),
+			    "SCHILY.acl.default", wp);
 
 		/* Include star-compatible metadata info. */
 		add_pax_attr_int(&(pax->pax_header), "SCHILY.dev",