From f48fc107cbc1cf28cbd305b37c399cd180cefeed Mon Sep 17 00:00:00 2001
From: Patrick McDermott <patrick.mcdermott@libiquity.com>
Date: Tue, 02 May 2023 22:30:32 -0400
Subject: ustar: Improve compatibility of written headers

---
(limited to 'src')

diff --git a/src/ustar.c b/src/ustar.c
index d70a616..b5b0310 100644
--- a/src/ustar.c
+++ b/src/ustar.c
@@ -28,6 +28,8 @@
 
 #define OPKG_OPK_USTAR_NUM_BASE_ 8
 
+/* #define OPKG_OPK_USTAR_BB_EXACT_HEADER_ 1 */
+
 struct _opkg_opk_ustar_header {
 	char name     [100];
 	char mode       [8];
@@ -284,13 +286,16 @@ opkg_opk_ustar_write_header(struct opkg_opk_ustar *ustar,
 				sizeof(ustar->header.prefix)) != OPKG_OPK_OK) {
 		return OPKG_OPK_ERROR;
 	}
-	sprintf(ustar->header.mode, "%o", mode);
-	sprintf(ustar->header.uid,  "%o", uid);
-	sprintf(ustar->header.gid,  "%o", gid);
-	strncpy(ustar->header.uname, uname, sizeof(ustar->header.uname));
-	strncpy(ustar->header.gname, gname, sizeof(ustar->header.gname));
-	sprintf(ustar->header.size, "%o", size);
-	sprintf(ustar->header.mtime, "%o", mtime);
+	/* POSIX doesn't say to NUL-terminate mode, uid, gid, size, or mtime.
+	 * GNU and BB tar accept values that aren't (but generate values that
+	 * ARE) NUL-terminated.  But more importantly, opkg-lede (see file
+	 * libbb/unarchive.c) parses these with simple strtol() calls, without
+	 * even checking endptr, so we have to NUL-terminate these fields. */
+	sprintf(ustar->header.mode,   "%07o", mode);
+	sprintf(ustar->header.uid,    "%07o", uid);
+	sprintf(ustar->header.gid,    "%07o", gid);
+	sprintf(ustar->header.size,  "%011o", size);
+	sprintf(ustar->header.mtime, "%011o", mtime);
 	switch (type) {
 		case '-':  /* Regular file */
 			*ustar->header.typeflag = '0';
@@ -315,9 +320,24 @@ opkg_opk_ustar_write_header(struct opkg_opk_ustar *ustar,
 		default:
 			return OPKG_OPK_ERROR;  /* Unsupported */
 	}
+	strncpy(ustar->header.uname, uname, sizeof(ustar->header.uname));
+	strncpy(ustar->header.gname, gname, sizeof(ustar->header.gname));
+
+	/* In these fields POSIX says to write:     But GNU and BB tar write: */
+	strcpy(ustar->header.magic,   "ustar");  /* "ustar " */
+	memcpy(ustar->header.version, "00", 2);  /* " \0" i.e. 0x20 0x00 */
+	/* See files in BusyBox:
+	 *     include/bb_archive.h
+	 *     archival/chksum_and_xwrite_tar_header.c
+	 *     archival/libarchive/get_header_tar.c
+	 * GNU and BB tar accept POSIX-conformant magic and version fields.
+	 * opkg-lede libbb/unarchive.c only checks first 5 bytes of magic and
+	 * ignores version.  So conforming to POSIX seems safe.
+	 */
+#if OPKG_OPK_USTAR_BB_EXACT_HEADER_
+	strcpy(ustar->header.magic, "ustar  ");
+#endif
 
-	strcpy(ustar->header.magic, "ustar");
-	memcpy(ustar->header.version, "00", 2);
 	chksum = 0;
 	for (i = 0; i < sizeof(ustar->header.chksum); ++i) {
 		ustar->header.chksum[i] = ' ';
@@ -326,7 +346,17 @@ opkg_opk_ustar_write_header(struct opkg_opk_ustar *ustar,
 	for (i = 0; i < OPKG_OPK_USTAR_RECORD_SIZE; ++i) {
 		chksum += header_uc[i];
 	}
-	sprintf(ustar->header.chksum, "%o", chksum);
+	/* See above about NUL-terminating and strtol() in opkg-lede. */
+	/* Since commit 5661fe078eed752780b11f3f4fdd33bbd76a6c5e, BB tar has
+	 * filled chksum with "[6] digits, a null, then a space -- rather than
+	 * digits, followed by a null like the other fields...".  The maxiumum
+	 * checksum is (255 x 512 = octal 377000), i.e. a maximum of 6 octal
+	 * digits, so it doesn't matter whether we use 6 or 7 digits. */
+#if OPKG_OPK_USTAR_BB_EXACT_HEADER_
+	sprintf(ustar->header.chksum, "%06o", chksum);
+#else
+	sprintf(ustar->header.chksum, "%07o", chksum);
+#endif
 
 	if (opkg_opk_gzip_write(ustar->gzip, &ustar->header,
 				OPKG_OPK_USTAR_RECORD_SIZE) != OPKG_OPK_OK) {
--
cgit v0.9.1