[git commit] unzip: fix a case where we find wrong CDE. Closes 8821

[Denys Vlasenko]

commit: https://git.busybox.net/busybox/commit/?id=0ccf52a9fb9fa2f76eacbb6f1ef8419220557a40
branch: https://git.busybox.net/busybox/commit/?id=refs/heads/master

function                                             old     new   delta
unzip_main                                          2472    2490     +18

Signed-off-by: Denys Vlasenko <vda.linux at googlemail.com>
---
 archival/unzip.c | 43 +++++++++++++++++++++++++++++++++++--------
 1 file changed, 35 insertions(+), 8 deletions(-)

diff --git a/archival/unzip.c b/archival/unzip.c
index f41ab6f..b0a6ca8 100644
--- a/archival/unzip.c
+++ b/archival/unzip.c
@@ -45,6 +45,12 @@
 #include "libbb.h"
 #include "bb_archive.h"
 
+#if 0
+# define dbg(...) bb_error_msg(__VA_ARGS__)
+#else
+# define dbg(...) ((void)0)
+#endif
+
 enum {
 #if BB_BIG_ENDIAN
 	ZIP_FILEHEADER_MAGIC = 0x504b0304,
@@ -193,15 +199,17 @@ static uint32_t find_cdf_offset(void)
 	unsigned char *p;
 	off_t end;
 	unsigned char *buf = xzalloc(PEEK_FROM_END);
+	uint32_t found;
 
 	end = xlseek(zip_fd, 0, SEEK_END);
 	end -= PEEK_FROM_END;
 	if (end < 0)
 		end = 0;
-	xlseek(zip_fd, end, SEEK_SET);
+	dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
+ 	xlseek(zip_fd, end, SEEK_SET);
 	full_read(zip_fd, buf, PEEK_FROM_END);
 
-	cde_header.formatted.cdf_offset = BAD_CDF_OFFSET;
+	found = BAD_CDF_OFFSET;
 	p = buf;
 	while (p <= buf + PEEK_FROM_END - CDE_HEADER_LEN - 4) {
 		if (*p != 'P') {
@@ -220,14 +228,25 @@ static uint32_t find_cdf_offset(void)
 		/*
 		 * I've seen .ZIP files with seemingly valid CDEs
 		 * where cdf_offset points past EOF - ??
-		 * Ignore such CDEs:
+		 * This check ignores such CDEs:
 		 */
-		if (cde_header.formatted.cdf_offset < end + (p - buf))
-			break;
-		cde_header.formatted.cdf_offset = BAD_CDF_OFFSET;
+		if (cde_header.formatted.cdf_offset < end + (p - buf)) {
+			found = cde_header.formatted.cdf_offset;
+			dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
+				(unsigned)found, end + (p-3 - buf));
+			dbg("  cdf_offset+cdf_size:0x%x",
+				(unsigned)(found + SWAP_LE32(cde_header.formatted.cdf_size)));
+			/*
+			 * We do not "break" here because only the last CDE is valid.
+			 * I've seen a .zip archive which contained a .zip file,
+			 * uncompressed, and taking the first CDE was using
+			 * the CDE inside that file!
+			 */
+		}
 	}
 	free(buf);
-	return cde_header.formatted.cdf_offset;
+	dbg("Found cdf_offset:0x%x", (unsigned)found);
+	return found;
 };
 
 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf_ptr)
@@ -240,9 +259,13 @@ static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf_ptr)
 		cdf_offset = find_cdf_offset();
 
 	if (cdf_offset != BAD_CDF_OFFSET) {
+		dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 		xlseek(zip_fd, cdf_offset + 4, SEEK_SET);
 		xread(zip_fd, cdf_ptr->raw, CDF_HEADER_LEN);
 		FIX_ENDIANNESS_CDF(*cdf_ptr);
+		dbg("file_name_length:%u", (unsigned)cdf_ptr->formatted.file_name_length);
+		dbg("extra_field_length:%u", (unsigned)cdf_ptr->formatted.extra_field_length);
+		dbg("file_comment_length:%u", (unsigned)cdf_ptr->formatted.file_comment_length);
 		cdf_offset += 4 + CDF_HEADER_LEN
 			+ cdf_ptr->formatted.file_name_length
 			+ cdf_ptr->formatted.extra_field_length
@@ -532,11 +555,14 @@ int unzip_main(int argc, char **argv)
 		/* Check magic number */
 		xread(zip_fd, &magic, 4);
 		/* Central directory? It's at the end, so exit */
-		if (magic == ZIP_CDF_MAGIC)
+		if (magic == ZIP_CDF_MAGIC) {
+			dbg("got ZIP_CDF_MAGIC");
 			break;
+		}
 #if ENABLE_DESKTOP
 		/* Data descriptor? It was a streaming file, go on */
 		if (magic == ZIP_DD_MAGIC) {
+			dbg("got ZIP_DD_MAGIC");
 			/* skip over duplicate crc32, cmpsize and ucmpsize */
 			unzip_skip(3 * 4);
 			continue;
@@ -544,6 +570,7 @@ int unzip_main(int argc, char **argv)
 #endif
 		if (magic != ZIP_FILEHEADER_MAGIC)
 			bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
+		dbg("got ZIP_FILEHEADER_MAGIC");
 
 		/* Read the file header */
 		xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);