tar: handle pax-encoded utf8 filenames and link names. Closes 9406

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

3 files changed, 45 insertions, 7 deletions

diff --git a/archival/libarchive/get_header_tar.c b/archival/libarchive/get_header_tar.c
index ea91a88..c7e3bc1 100644
--- a/archival/libarchive/get_header_tar.c
+++ b/archival/libarchive/get_header_tar.c
@@ -113,11 +113,19 @@ static void process_pax_hdr(archive_handle_t *archive_handle, unsigned sz, int g
 		value = end + 1;
 
 # if ENABLE_FEATURE_TAR_GNU_EXTENSIONS
-		if (!global && is_prefixed_with(value, "path=")) {
-			value += sizeof("path=") - 1;
-			free(archive_handle->tar__longname);
-			archive_handle->tar__longname = xstrdup(value);
-			continue;
+		if (!global) {
+			if (is_prefixed_with(value, "path=")) {
+				value += sizeof("path=") - 1;
+				free(archive_handle->tar__longname);
+				archive_handle->tar__longname = xstrdup(value);
+				continue;
+			}
+			if (is_prefixed_with(value, "linkpath=")) {
+				value += sizeof("linkpath=") - 1;
+				free(archive_handle->tar__linkname);
+				archive_handle->tar__linkname = xstrdup(value);
+				continue;
+			}
 		}
 # endif
 
@@ -179,7 +187,13 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 	 * the message and we don't check whether we indeed
 	 * saw zero block directly before this. */
 	if (i == 0) {
-		bb_error_msg("short read");
+		/* GNU tar 1.29 will be silent if tar archive ends abruptly
+		 * (if there are no zero blocks at all, and last read returns zero,
+		 * not short read 0 < len < 512). Complain only if
+		 * the very first read fails. Grrr.
+		 */
+		if (archive_handle->offset == 0)
+			bb_error_msg("short read");
 		/* this merely signals end of archive, not exit(1): */
 		return EXIT_FAILURE;
 	}
@@ -195,7 +209,11 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 	archive_handle->offset += i;
 
 	/* If there is no filename its an empty header */
-	if (tar.name[0] == 0 && tar.prefix[0] == 0) {
+	if (tar.name[0] == 0 && tar.prefix[0] == 0
+	/* Have seen a tar archive with pax 'x' header supplying UTF8 filename,
+	 * with actual file having all name fields NUL-filled. Check this: */
+	 && !p_longname
+	) {
 		if (archive_handle->tar__end) {
 			/* Second consecutive empty header - end of archive.
 			 * Read until the end to empty the pipe from gz or bz2
diff --git a/testsuite/tar.tests b/testsuite/tar.tests
index c44b7ad..ffcbe9b 100755
--- a/testsuite/tar.tests
+++ b/testsuite/tar.tests
@@ -299,6 +299,26 @@ l4/V8LDoe90yiWJhOJvIypgEfxdyRThQkBVn/bI=
 "
 SKIP=
 
+optional UNICODE_SUPPORT
+testing "Pax-encoded UTF8 names and symlinks" '\
+tar xvf ../tar.utf8.tar.bz2 2>&1; echo $?
+export LANG=en_US.UTF-8
+ls -l etc/ssl/certs/* | sed "s:.*etc/:etc/:"
+unset LANG
+rm -rf etc usr
+' "\
+etc/ssl/certs/3b2716e5.0
+etc/ssl/certs/EBG_Elektronik_Sertifika_Hizmet_Sağlayıcısı.pem
+etc/ssl/certs/f80cc7f6.0
+usr/share/ca-certificates/mozilla/EBG_Elektronik_Sertifika_Hizmet_Sağlayıcısı.crt
+0
+etc/ssl/certs/3b2716e5.0 -> EBG_Elektronik_Sertifika_Hizmet_Sağlayıcısı.pem
+etc/ssl/certs/EBG_Elektronik_Sertifika_Hizmet_Sağlayıcısı.pem -> /usr/share/ca-certificates/mozilla/EBG_Elektronik_Sertifika_Hizmet_Sağlayıcısı.crt
+etc/ssl/certs/f80cc7f6.0 -> EBG_Elektronik_Sertifika_Hizmet_Sağlayıcısı.pem
+" \
+"" ""
+SKIP=
+
 
 cd .. && rm -rf tar.tempdir || exit 1
 
diff --git a/testsuite/tar.utf8.tar.bz2 b/testsuite/tar.utf8.tar.bz2
new file mode 100644
index 0000000..0398e1a
--- /dev/null
+++ b/testsuite/tar.utf8.tar.bz2