xz compression detection: avoid the need to seek

function                                             old     new   delta
unpack_unxz                                           12      67     +55
unpack_xz_stream                                    2357    2373     +16
xmalloc_read                                         197     199      +2
setup_unzip_on_fd                                    118      99     -19
rpm2cpio_main                                        222     203     -19
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 3/2 up/down: 73/-38)             Total: 35 bytes

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

5 files changed, 30 insertions, 14 deletions

diff --git a/archival/bbunzip.c b/archival/bbunzip.c
index b243afb2e..832a7bbf3 100644
--- a/archival/bbunzip.c
+++ b/archival/bbunzip.c
@@ -373,6 +373,15 @@ int unlzma_main(int argc UNUSED_PARAM, char **argv)
 static
 IF_DESKTOP(long long) int FAST_FUNC unpack_unxz(unpack_info_t *info UNUSED_PARAM)
 {
+        struct {
+                uint32_t v1;
+                uint16_t v2;
+        } magic;
+        xread(STDIN_FILENO, &magic, 6);
+        if (magic.v1 != XZ_MAGIC1a || magic.v2 != XZ_MAGIC2a) {
+                bb_error_msg("invalid magic");
+                return -1;
+        }
         return unpack_xz_stream(STDIN_FILENO, STDOUT_FILENO);
 }
 int unxz_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
diff --git a/archival/libunarchive/decompress_unxz.c b/archival/libunarchive/decompress_unxz.c
index 5d7513a2e..721acd907 100644
--- a/archival/libunarchive/decompress_unxz.c
+++ b/archival/libunarchive/decompress_unxz.c
@@ -56,9 +56,11 @@ unpack_xz_stream(int src_fd, int dst_fd)
         if (!crc32_table)
                 crc32_table = crc32_filltable(NULL, /*endian:*/ 0);
 
-        membuf = xmalloc(2 * BUFSIZ);
         memset(&iobuf, 0, sizeof(iobuf));
+        /* Preload XZ file signature */
+        membuf = (void*) strcpy(xmalloc(2 * BUFSIZ), HEADER_MAGIC);
         iobuf.in = membuf;
+        iobuf.in_size = HEADER_MAGIC_SIZE;
         iobuf.out = membuf + BUFSIZ;
         iobuf.out_size = BUFSIZ;
 
diff --git a/archival/rpm2cpio.c b/archival/rpm2cpio.c
index 1f67fa887..b23d5afa9 100644
--- a/archival/rpm2cpio.c
+++ b/archival/rpm2cpio.c
@@ -92,8 +92,8 @@ int rpm2cpio_main(int argc UNUSED_PARAM, char **argv)
                         xread(rpm_fd, magic.b32, sizeof(magic.b32[0]));
                         if (magic.b32[0] != XZ_MAGIC2)
                                 goto no_magic;
-                        /* unpack_xz_stream wants fd at position 0 */
-                        xlseek(rpm_fd, -6, SEEK_CUR);
+                        /* unpack_xz_stream wants fd at position 6, no need to seek */
+                        //xlseek(rpm_fd, -6, SEEK_CUR);
                         unpack = unpack_xz_stream;
                 } else {
  no_magic:
diff --git a/include/unarchive.h b/include/unarchive.h
index b4cf16082..3b8e1575a 100644
--- a/include/unarchive.h
+++ b/include/unarchive.h
@@ -7,16 +7,21 @@ PUSH_AND_SET_FUNCTION_VISIBILITY_TO_HIDDEN
 enum {
 #if BB_BIG_ENDIAN
         COMPRESS_MAGIC = 0x1f9d,
-        GZIP_MAGIC = 0x1f8b,
-        BZIP2_MAGIC = ('B'<<8) + 'Z',
-        XZ_MAGIC1 = (0xfd<<8) + '7',
-        XZ_MAGIC2 = ((((('z'<<8) + 'X')<<8) + 'Z')<<8) + 0,
+        GZIP_MAGIC  = 0x1f8b,
+        BZIP2_MAGIC = 'B' * 256 + 'Z',
+        XZ_MAGIC1   = 0xfd * 256 + '7',
+        XZ_MAGIC2   = (('z' * 256 + 'X') * 256 + 'Z') * 256 + 0,
+        /* Different form: 32 bits, then 16 bits: */
+        XZ_MAGIC1a  = ((0xfd * 256 + '7') * 256 + 'z') * 256 + 'X',
+        XZ_MAGIC2a  = 'Z' * 256 + 0,
 #else
         COMPRESS_MAGIC = 0x9d1f,
-        GZIP_MAGIC = 0x8b1f,
-        BZIP2_MAGIC = ('Z'<<8) + 'B',
-        XZ_MAGIC1 = ('7'<<8) + 0xfd,
-        XZ_MAGIC2 = (((((0<<8) + 'Z')<<8) + 'X')<<8) + 'z',
+        GZIP_MAGIC  = 0x8b1f,
+        BZIP2_MAGIC = 'Z' * 256 + 'B',
+        XZ_MAGIC1   = '7' * 256 + 0xfd,
+        XZ_MAGIC2   = ((0 * 256 + 'Z') * 256 + 'X') * 256 + 'z',
+        XZ_MAGIC1a  = (('X' * 256 + 'z') * 256 + '7') * 256 + 0xfd,
+        XZ_MAGIC2a  = 0 * 256 + 'Z',
 #endif
 };
 
@@ -196,7 +201,7 @@ typedef struct inflate_unzip_result {
 } inflate_unzip_result;
 
 IF_DESKTOP(long long) int inflate_unzip(inflate_unzip_result *res, off_t compr_size, int src_fd, int dst_fd) FAST_FUNC;
-/* xz unpacker takes .xz stream from offset 0 */
+/* xz unpacker takes .xz stream from offset 6 */
 IF_DESKTOP(long long) int unpack_xz_stream(int src_fd, int dst_fd) FAST_FUNC;
 /* lzma unpacker takes .lzma stream from offset 0 */
 IF_DESKTOP(long long) int unpack_lzma_stream(int src_fd, int dst_fd) FAST_FUNC;
diff --git a/libbb/read_printf.c b/libbb/read_printf.c
index 3aee075c6..6780b94ef 100644
--- a/libbb/read_printf.c
+++ b/libbb/read_printf.c
@@ -296,8 +296,8 @@ void FAST_FUNC setup_unzip_on_fd(int fd /*, int fail_if_not_detected*/)
                 if (magic.b32[0] == XZ_MAGIC2) {
 # if BB_MMU
                         xformer = unpack_xz_stream;
-                        /* unpack_xz_stream wants fd at position 0 */
-                        xlseek(fd, offset, SEEK_CUR);
+                        /* unpack_xz_stream wants fd at position 6, no need to seek */
+                        //xlseek(fd, offset, SEEK_CUR);
 # else
                         xformer_prog = "unxz";
 # endif