Index: vendor/libarchive/dist/configure.ac =================================================================== --- vendor/libarchive/dist/configure.ac (revision 302055) +++ vendor/libarchive/dist/configure.ac (revision 302056) @@ -1,878 +1,880 @@ dnl Process this file with autoconf to produce a configure script. dnl First, define all of the version numbers up front. dnl In particular, this allows the version macro to be used in AC_INIT dnl These first two version numbers are updated automatically on each release. m4_define([LIBARCHIVE_VERSION_S],[3.2.1]) m4_define([LIBARCHIVE_VERSION_N],[3002001]) dnl bsdtar and bsdcpio versioning tracks libarchive m4_define([BSDTAR_VERSION_S],LIBARCHIVE_VERSION_S()) m4_define([BSDCPIO_VERSION_S],LIBARCHIVE_VERSION_S()) m4_define([BSDCAT_VERSION_S],LIBARCHIVE_VERSION_S()) AC_PREREQ([2.69]) # # Now starts the "real" configure script. # AC_INIT([libarchive],[LIBARCHIVE_VERSION_S()],[libarchive-discuss@googlegroups.com]) # Make sure the srcdir contains "libarchive" directory AC_CONFIG_SRCDIR([libarchive]) # Use auxiliary subscripts from this subdirectory (cleans up root) AC_CONFIG_AUX_DIR([build/autoconf]) # M4 scripts AC_CONFIG_MACRO_DIR([build/autoconf]) # Must follow AC_CONFIG macros above... AM_INIT_AUTOMAKE() +AM_MAINTAINER_MODE([enable]) m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])]) # Libtool's "interface version" can be computed from the libarchive version. # Libtool interface version bumps on any API change, so increments # whenever libarchive minor version does. ARCHIVE_MINOR=$(( (LIBARCHIVE_VERSION_N() / 1000) % 1000 )) # Libarchive 2.7 == libtool interface 9 = 2 + 7 # Libarchive 2.8 == libtool interface 10 = 2 + 8 # Libarchive 2.9 == libtool interface 11 = 2 + 8 # Libarchive 3.0 == libtool interface 12 # Libarchive 3.1 == libtool interface 13 ARCHIVE_INTERFACE=`echo $((13 + ${ARCHIVE_MINOR}))` # Libarchive revision is bumped on any source change === libtool revision ARCHIVE_REVISION=$(( LIBARCHIVE_VERSION_N() % 1000 )) # Libarchive minor is bumped on any interface addition === libtool age ARCHIVE_LIBTOOL_VERSION=$ARCHIVE_INTERFACE:$ARCHIVE_REVISION:$ARCHIVE_MINOR # Stick the version numbers into config.h AC_DEFINE([LIBARCHIVE_VERSION_STRING],"LIBARCHIVE_VERSION_S()", [Version number of libarchive]) AC_DEFINE_UNQUOTED([LIBARCHIVE_VERSION_NUMBER],"LIBARCHIVE_VERSION_N()", [Version number of libarchive as a single integer]) AC_DEFINE([BSDCPIO_VERSION_STRING],"BSDCPIO_VERSION_S()", [Version number of bsdcpio]) AC_DEFINE([BSDTAR_VERSION_STRING],"BSDTAR_VERSION_S()", [Version number of bsdtar]) AC_DEFINE([BSDCAT_VERSION_STRING],"BSDTAR_VERSION_S()", [Version number of bsdcat]) # The shell variables here must be the same as the AC_SUBST() variables # below, but the shell variable names apparently cannot be the same as # the m4 macro names above. Why? Ask autoconf. BSDCPIO_VERSION_STRING=BSDCPIO_VERSION_S() BSDTAR_VERSION_STRING=BSDTAR_VERSION_S() BSDCAT_VERSION_STRING=BSDCAT_VERSION_S() LIBARCHIVE_VERSION_STRING=LIBARCHIVE_VERSION_S() LIBARCHIVE_VERSION_NUMBER=LIBARCHIVE_VERSION_N() # Substitute the above version numbers into the various files below. # Yes, I believe this is the fourth time we define what are essentially # the same symbols. Why? Ask autoconf. AC_SUBST(ARCHIVE_LIBTOOL_VERSION) AC_SUBST(BSDCPIO_VERSION_STRING) AC_SUBST(BSDTAR_VERSION_STRING) AC_SUBST(BSDCAT_VERSION_STRING) AC_SUBST(LIBARCHIVE_VERSION_STRING) AC_SUBST(LIBARCHIVE_VERSION_NUMBER) AC_CONFIG_HEADERS([config.h]) AC_CONFIG_FILES([Makefile]) AC_CONFIG_FILES([build/pkgconfig/libarchive.pc]) # Check for host type AC_CANONICAL_HOST dnl Compilation on mingw and Cygwin needs special Makefile rules inc_windows_files=no inc_cygwin_files=no case "$host_os" in *mingw* ) inc_windows_files=yes ;; *cygwin*) inc_cygwin_files=yes ;; esac AM_CONDITIONAL([INC_WINDOWS_FILES], [test $inc_windows_files = yes]) AM_CONDITIONAL([INC_CYGWIN_FILES], [test $inc_cygwin_files = yes]) dnl Defines that are required for specific platforms (e.g. -D_POSIX_SOURCE, etc) PLATFORMCPPFLAGS= case "$host_os" in *mingw* ) PLATFORMCPPFLAGS=-D__USE_MINGW_ANSI_STDIO ;; esac AC_SUBST(PLATFORMCPPFLAGS) # Checks for programs. AC_PROG_CC AM_PROG_CC_C_O AC_USE_SYSTEM_EXTENSIONS AC_LIBTOOL_WIN32_DLL AC_PROG_LIBTOOL AC_CHECK_TOOL([STRIP],[strip]) AC_PROG_MKDIR_P # # Options for building bsdtar. # # Default is to build bsdtar, but allow people to override that. # AC_ARG_ENABLE([bsdtar], [AS_HELP_STRING([--enable-bsdtar], [enable build of bsdtar (default)]) AS_HELP_STRING([--enable-bsdtar=static], [force static build of bsdtar]) AS_HELP_STRING([--enable-bsdtar=shared], [force dynamic build of bsdtar]) AS_HELP_STRING([--disable-bsdtar], [disable build of bsdtar])], [], [enable_bsdtar=yes]) case "$enable_bsdtar" in yes) if test "$enable_static" = "no"; then static_bsdtar=no else static_bsdtar=yes fi build_bsdtar=yes ;; dynamic|shared) if test "$enable_shared" = "no"; then AC_MSG_FAILURE([Shared linking of bsdtar requires shared libarchive]) fi build_bsdtar=yes static_bsdtar=no ;; static) build_bsdtar=yes static_bsdtar=yes ;; no) build_bsdtar=no static_bsdtar=no ;; *) AC_MSG_FAILURE([Unsupported value for --enable-bsdtar]) ;; esac AM_CONDITIONAL([BUILD_BSDTAR], [ test "$build_bsdtar" = yes ]) AM_CONDITIONAL([STATIC_BSDTAR], [ test "$static_bsdtar" = yes ]) # # Options for building bsdcat. # # Default is to build bsdcat, but allow people to override that. # AC_ARG_ENABLE([bsdcat], [AS_HELP_STRING([--enable-bsdcat], [enable build of bsdcat (default)]) AS_HELP_STRING([--enable-bsdcat=static], [force static build of bsdcat]) AS_HELP_STRING([--enable-bsdcat=shared], [force dynamic build of bsdcat]) AS_HELP_STRING([--disable-bsdcat], [disable build of bsdcat])], [], [enable_bsdcat=yes]) case "$enable_bsdcat" in yes) if test "$enable_static" = "no"; then static_bsdcat=no else static_bsdcat=yes fi build_bsdcat=yes ;; dynamic|shared) if test "$enable_shared" = "no"; then AC_MSG_FAILURE([Shared linking of bsdcat requires shared libarchive]) fi build_bsdcat=yes static_bsdcat=no ;; static) build_bsdcat=yes static_bsdcat=yes ;; no) build_bsdcat=no static_bsdcat=no ;; *) AC_MSG_FAILURE([Unsupported value for --enable-bsdcat]) ;; esac AM_CONDITIONAL([BUILD_BSDCAT], [ test "$build_bsdcat" = yes ]) AM_CONDITIONAL([STATIC_BSDCAT], [ test "$static_bsdcat" = yes ]) # # Options for building bsdcpio. # # Default is not to build bsdcpio, but that can be overridden. # AC_ARG_ENABLE([bsdcpio], [AS_HELP_STRING([--enable-bsdcpio], [enable build of bsdcpio (default)]) AS_HELP_STRING([--enable-bsdcpio=static], [static build of bsdcpio]) AS_HELP_STRING([--enable-bsdcpio=shared], [dynamic build of bsdcpio]) AS_HELP_STRING([--disable-bsdcpio], [disable build of bsdcpio])], [], [enable_bsdcpio=yes]) case "$enable_bsdcpio" in yes) if test "$enable_static" = "no"; then static_bsdcpio=no else static_bsdcpio=yes fi build_bsdcpio=yes ;; dynamic|shared) if test "$enabled_shared" = "no"; then AC_MSG_FAILURE([Shared linking of bsdcpio requires shared libarchive]) fi build_bsdcpio=yes ;; static) build_bsdcpio=yes static_bsdcpio=yes ;; no) build_bsdcpio=no static_bsdcpio=no ;; *) AC_MSG_FAILURE([Unsupported value for --enable-bsdcpio]) ;; esac AM_CONDITIONAL([BUILD_BSDCPIO], [ test "$build_bsdcpio" = yes ]) AM_CONDITIONAL([STATIC_BSDCPIO], [ test "$static_bsdcpio" = yes ]) # Set up defines needed before including any headers case $host in *mingw* | *cygwin* ) AC_DEFINE([_WIN32_WINNT], 0x0502, [Define to '0x0502' for Windows Server 2003 APIs.]) AC_DEFINE([WINVER], 0x0502, [Define to '0x0502' for Windows Server 2003 APIs.]) AC_DEFINE([NTDDI_VERSION], 0x05020000, [Define to '0x05020000' for Windows Server 2003 APIs.]) ;; esac # Checks for header files. AC_HEADER_DIRENT AC_HEADER_SYS_WAIT AC_CHECK_HEADERS([copyfile.h ctype.h]) AC_CHECK_HEADERS([errno.h ext2fs/ext2_fs.h fcntl.h grp.h]) AC_CACHE_CHECK([whether EXT2_IOC_GETFLAGS is usable], [ac_cv_have_decl_EXT2_IOC_GETFLAGS], [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([@%:@include @%:@include ], [int x = EXT2_IOC_GETFLAGS])], [AS_VAR_SET([ac_cv_have_decl_EXT2_IOC_GETFLAGS], [yes])], [AS_VAR_SET([ac_cv_have_decl_EXT2_IOC_GETFLAGS], [no])])]) AS_VAR_IF([ac_cv_have_decl_EXT2_IOC_GETFLAGS], [yes], [AC_DEFINE_UNQUOTED([HAVE_WORKING_EXT2_IOC_GETFLAGS], [1], [Define to 1 if you have a working EXT2_IOC_GETFLAGS])]) AC_CHECK_HEADERS([inttypes.h io.h langinfo.h limits.h]) AC_CHECK_HEADERS([linux/fiemap.h linux/fs.h linux/magic.h linux/types.h]) AC_CHECK_HEADERS([locale.h paths.h poll.h pthread.h pwd.h]) AC_CHECK_HEADERS([readpassphrase.h signal.h spawn.h]) AC_CHECK_HEADERS([stdarg.h stdint.h stdlib.h string.h]) AC_CHECK_HEADERS([sys/cdefs.h sys/extattr.h]) AC_CHECK_HEADERS([sys/ioctl.h sys/mkdev.h sys/mount.h]) AC_CHECK_HEADERS([sys/param.h sys/poll.h sys/select.h sys/statfs.h sys/statvfs.h]) AC_CHECK_HEADERS([sys/time.h sys/utime.h sys/utsname.h sys/vfs.h]) AC_CHECK_HEADERS([time.h unistd.h utime.h wchar.h wctype.h]) AC_CHECK_HEADERS([windows.h]) AC_CHECK_HEADERS([Bcrypt.h]) # check windows.h first; the other headers require it. AC_CHECK_HEADERS([wincrypt.h winioctl.h],[],[], [[#ifdef HAVE_WINDOWS_H # include #endif ]]) # Checks for libraries. AC_ARG_WITH([zlib], AS_HELP_STRING([--without-zlib], [Don't build support for gzip through zlib])) if test "x$with_zlib" != "xno"; then AC_CHECK_HEADERS([zlib.h]) AC_CHECK_LIB(z,inflate) fi AC_ARG_WITH([bz2lib], AS_HELP_STRING([--without-bz2lib], [Don't build support for bzip2 through bz2lib])) if test "x$with_bz2lib" != "xno"; then AC_CHECK_HEADERS([bzlib.h]) case "$host_os" in *mingw* | *cygwin*) dnl AC_CHECK_LIB cannot be used on the Windows port of libbz2, therefore dnl use AC_LINK_IFELSE. AC_MSG_CHECKING([for BZ2_bzDecompressInit in -lbz2]) old_LIBS="$LIBS" LIBS="-lbz2 $LIBS" AC_LINK_IFELSE( [AC_LANG_SOURCE(#include int main() { return BZ2_bzDecompressInit(NULL, 0, 0); })], [ac_cv_lib_bz2_BZ2_bzDecompressInit=yes], [ac_cv_lib_bz2_BZ2_bzDecompressInit=no]) LIBS="$old_LIBS" AC_MSG_RESULT($ac_cv_lib_bz2_BZ2_bzDecompressInit) if test "x$ac_cv_lib_bz2_BZ2_bzDecompressInit" = xyes; then AC_DEFINE([HAVE_LIBBZ2], [1], [Define to 1 if you have the `bz2' library (-lbz2).]) LIBS="-lbz2 $LIBS" fi ;; *) AC_CHECK_LIB(bz2,BZ2_bzDecompressInit) ;; esac fi AC_ARG_WITH([lzmadec], AS_HELP_STRING([--without-lzmadec], [Don't build support for lzma through lzmadec])) if test "x$with_lzmadec" != "xno"; then AC_CHECK_HEADERS([lzmadec.h]) AC_CHECK_LIB(lzmadec,lzmadec_decode) fi AC_ARG_WITH([iconv], AS_HELP_STRING([--without-iconv], [Don't try to link against iconv])) if test "x$with_iconv" != "xno"; then AM_ICONV AC_CHECK_HEADERS([iconv.h],[],[],[#include ]) if test "x$am_cv_func_iconv" = "xyes"; then AC_CHECK_HEADERS([localcharset.h]) am_save_LIBS="$LIBS" LIBS="${LIBS} ${LIBICONV}" AC_CHECK_FUNCS([locale_charset]) LIBS="${am_save_LIBS}" if test "x$ac_cv_func_locale_charset" != "xyes"; then # If locale_charset() is not in libiconv, we have to find libcharset. AC_CHECK_LIB(charset,locale_charset) fi fi fi AC_ARG_WITH([lz4], AS_HELP_STRING([--without-lz4], [Don't build support for lz4 through liblz4])) if test "x$with_lz4" != "xno"; then AC_CHECK_HEADERS([lz4.h lz4hc.h]) AC_CHECK_LIB(lz4,LZ4_decompress_safe) fi AC_ARG_WITH([lzma], AS_HELP_STRING([--without-lzma], [Don't build support for xz through lzma])) if test "x$with_lzma" != "xno"; then AC_CHECK_HEADERS([lzma.h]) AC_CHECK_LIB(lzma,lzma_stream_decoder) # Some pre-release (but widely distributed) versions of liblzma # included a disabled version of lzma_stream_encoder_mt that # fools a naive AC_CHECK_LIB or AC_CHECK_FUNC, so we need # to do something more complex here: AC_CACHE_CHECK( [whether we have multithread support in lzma], ac_cv_lzma_has_mt, [AC_COMPILE_IFELSE([ AC_LANG_PROGRAM([[#include ]], [[lzma_stream_encoder_mt(0, 0);]])], [ac_cv_lzma_has_mt=yes], [ac_cv_lzma_has_mt=no])]) if test "x$ac_cv_lzma_has_mt" != xno; then AC_DEFINE([HAVE_LZMA_STREAM_ENCODER_MT], [1], [Define to 1 if you have the `lzma_stream_encoder_mt' function.]) fi fi AC_ARG_WITH([lzo2], AS_HELP_STRING([--without-lzo2], [Don't build support for lzop through liblzo2])) if test "x$with_lzo2" != "xno"; then AC_CHECK_HEADERS([lzo/lzoconf.h lzo/lzo1x.h]) AC_CHECK_LIB(lzo2,lzo1x_decompress_safe) fi AC_ARG_WITH([nettle], AS_HELP_STRING([--without-nettle], [Don't build with crypto support from Nettle])) AC_ARG_WITH([openssl], AS_HELP_STRING([--without-openssl], [Don't build support for mtree and xar hashes through openssl])) case "$host_os" in *darwin* ) with_openssl=no ;; esac AC_ARG_WITH([xml2], AS_HELP_STRING([--without-xml2], [Don't build support for xar through libxml2])) AC_ARG_WITH([expat], AS_HELP_STRING([--without-expat], [Don't build support for xar through expat])) if test "x$with_xml2" != "xno"; then PKG_PROG_PKG_CONFIG PKG_CHECK_MODULES(LIBXML2_PC, [libxml-2.0], [ CPPFLAGS="${CPPFLAGS} ${LIBXML2_PC_CFLAGS}" LIBS="${LIBS} ${LIBXML2_PC_LIBS}" AC_CHECK_LIB(xml2,xmlInitParser,[true],AC_MSG_FAILURE(Missing xml2 library)) ], [ AC_CHECK_LIB(xml2,xmlInitParser) ]) AC_CHECK_HEADERS([libxml/xmlreader.h libxml/xmlwriter.h]) fi if test "x$ac_cv_header_libxml_xmlreader_h" != "xyes"; then if test "x$with_expat" != "xno"; then AC_CHECK_HEADERS([expat.h]) AC_CHECK_LIB(expat,XML_ParserCreate) fi fi AC_ARG_ENABLE([posix-regex-lib], [AS_HELP_STRING([--enable-posix-regex-lib], [choose what library to use for POSIX regular expression support (default: auto)]) AS_HELP_STRING([--enable-posix-regex-lib=libc], [use libc POSIX regular expression support]) AS_HELP_STRING([--enable-posix-regex-lib=libregex], [use libregex POSIX regular expression support]) AS_HELP_STRING([--enable-posix-regex-lib=libpcreposix], [use libpcreposix POSIX regular expression support]) AS_HELP_STRING([--disable-posix-regex-lib], [don't enable POSIX regular expression support])], [], [enable_posix_regex_lib=auto]) posix_regex_lib_found= if test "$enable_posix_regex_lib" = "auto" || test "$enable_posix_regex_lib" = "libc" || test "$enable_posix_regex_lib" = "libregex"; then AC_CHECK_HEADERS([regex.h]) if test "x$ac_cv_header_regex_h" != "xno"; then AC_CHECK_FUNC(regcomp) if test "x$ac_cv_func_regcomp" = xyes; then posix_regex_lib_found=1 else AC_CHECK_LIB(regex,regcomp) if test "x$ac_cv_lib_regex_regcomp" = xyes; then posix_regex_lib_found=1 fi fi fi fi if test -z $posix_regex_lib_found && (test "$enable_posix_regex_lib" = "auto" || test "$enable_posix_regex_lib" = "libpcreposix"); then AC_CHECK_HEADERS([pcreposix.h]) AC_CHECK_LIB(pcreposix,regcomp) if test "x$ac_cv_lib_pcreposix_regcomp" != xyes; then AC_MSG_NOTICE(trying libpcreposix check again with libpcre) unset ac_cv_lib_pcreposix_regcomp AC_CHECK_LIB(pcre,pcre_exec) AC_CHECK_LIB(pcreposix,regcomp) if test "x$ac_cv_lib_pcre_pcre_exec" = xyes && test "x$ac_cv_lib_pcreposix_regcomp" = xyes; then AC_MSG_CHECKING(if PCRE_STATIC needs to be defined) AC_LINK_IFELSE( [AC_LANG_SOURCE(#include int main() { return regcomp(NULL, NULL, 0); })], [without_pcre_static=yes], [without_pcre_static=no]) AC_LINK_IFELSE( [AC_LANG_SOURCE(#define PCRE_STATIC #include int main() { return regcomp(NULL, NULL, 0); })], [with_pcre_static=yes], [with_pcre_static=no]) if test "x$without_pcre_static" != xyes && test "x$with_pcre_static" = xyes; then AC_MSG_RESULT(yes) AC_DEFINE([PCRE_STATIC], [1], [Define to 1 if PCRE_STATIC needs to be defined.]) elif test "x$without_pcre_static" = xyes || test "x$with_pcre_static" = xyes; then AC_MSG_RESULT(no) fi posix_regex_lib_found=1 fi else posix_regex_lib_found=1 fi fi # TODO: Give the user the option of using a pre-existing system # libarchive. This will define HAVE_LIBARCHIVE which will cause # bsdtar_platform.h to use #include <...> for the libarchive headers. # Need to include Makefile.am magic to link against system # -larchive in that case. #AC_CHECK_LIB(archive,archive_version) # Checks for supported compiler flags AX_APPEND_COMPILE_FLAGS([-Wall -Wformat -Wformat-security]) # Checks for typedefs, structures, and compiler characteristics. AC_C_CONST # la_TYPE_UID_T defaults to "int", which is incorrect for MinGW # and MSVC. Use a customized version. la_TYPE_UID_T AC_TYPE_MODE_T # AC_TYPE_OFF_T defaults to "long", which limits us to 4GB files on # most systems... default to "long long" instead. AC_CHECK_TYPE(off_t, [long long]) AC_TYPE_SIZE_T AC_CHECK_TYPE(id_t, [unsigned long]) AC_CHECK_TYPE(uintptr_t, [unsigned int]) # Check for tm_gmtoff in struct tm AC_CHECK_MEMBERS([struct tm.tm_gmtoff, struct tm.__tm_gmtoff],,, [ #include ]) # Check for f_namemax in struct statfs AC_CHECK_MEMBERS([struct statfs.f_namemax],,, [ #include #include ]) # Check for f_iosize in struct statvfs AC_CHECK_MEMBERS([struct statvfs.f_iosize],,, [ #include ]) # Check for birthtime in struct stat AC_CHECK_MEMBERS([struct stat.st_birthtime]) # Check for high-resolution timestamps in struct stat AC_CHECK_MEMBERS([struct stat.st_birthtimespec.tv_nsec]) AC_CHECK_MEMBERS([struct stat.st_mtimespec.tv_nsec]) AC_CHECK_MEMBERS([struct stat.st_mtim.tv_nsec]) AC_CHECK_MEMBERS([struct stat.st_mtime_n]) # AIX AC_CHECK_MEMBERS([struct stat.st_umtime]) # Tru64 AC_CHECK_MEMBERS([struct stat.st_mtime_usec]) # Hurd # Check for block size support in struct stat AC_CHECK_MEMBERS([struct stat.st_blksize]) # Check for st_flags in struct stat (BSD fflags) AC_CHECK_MEMBERS([struct stat.st_flags]) # If you have uintmax_t, we assume printf supports %ju # If you have unsigned long long, we assume printf supports %llu # TODO: Check for %ju and %llu support directly. AC_CHECK_TYPES([uintmax_t, unsigned long long]) # We use C99-style integer types # Declare them if the local platform doesn't already do so. AC_TYPE_INTMAX_T AC_TYPE_UINTMAX_T AC_TYPE_INT64_T AC_TYPE_UINT64_T AC_TYPE_INT32_T AC_TYPE_UINT32_T AC_TYPE_INT16_T AC_TYPE_UINT16_T AC_TYPE_UINT8_T AC_CHECK_DECLS([SIZE_MAX, INT32_MAX, INT32_MIN]) AC_CHECK_DECLS([INT64_MAX, INT64_MIN, UINT64_MAX, UINT32_MAX]) AC_CHECK_DECLS([INTMAX_MAX, INTMAX_MIN, UINTMAX_MAX]) AC_CHECK_DECL([SSIZE_MAX], [AC_DEFINE(HAVE_DECL_SSIZE_MAX, 1, [Define to 1 if you have the declaration of `SSIZE_MAX', and to 0 if you don't.])], [], [#include ]) AC_CHECK_DECL([EFTYPE], [AC_DEFINE(HAVE_EFTYPE, 1, [A possible errno value for invalid file format errors])], [], [#include ]) AC_CHECK_DECL([EILSEQ], [AC_DEFINE(HAVE_EILSEQ, 1, [A possible errno value for invalid file format errors])], [], [#include ]) AC_CHECK_TYPE([wchar_t], [AC_DEFINE_UNQUOTED(AS_TR_CPP(HAVE_[]wchar_t), 1, [Define to 1 if the system has the type `wchar_t'.])dnl AC_CHECK_SIZEOF([wchar_t])], []) AC_HEADER_TIME # Checks for library functions. AC_PROG_GCC_TRADITIONAL AC_HEADER_MAJOR AC_FUNC_FSEEKO AC_FUNC_MEMCMP AC_FUNC_LSTAT AC_FUNC_STAT AC_FUNC_STRERROR_R AC_FUNC_STRFTIME AC_FUNC_VPRINTF # check for: # CreateHardLinkA(LPCSTR, LPCSTR, LPSECURITY_ATTRIBUTES) # To avoid necessity for including windows.h or special forward declaration # workarounds, we use 'void *' for 'struct SECURITY_ATTRIBUTES *' AC_CHECK_STDCALL_FUNC([CreateHardLinkA],[const char *, const char *, void *]) AC_CHECK_FUNCS([arc4random_buf chflags chown chroot ctime_r dirfd]) AC_CHECK_FUNCS([fchdir fchflags fchmod fchown fcntl fdopendir fork]) AC_CHECK_FUNCS([fstat fstatat fstatfs fstatvfs ftruncate]) AC_CHECK_FUNCS([futimens futimes futimesat]) AC_CHECK_FUNCS([geteuid getpid getgrgid_r getgrnam_r]) AC_CHECK_FUNCS([getpwnam_r getpwuid_r getvfsbyname gmtime_r]) AC_CHECK_FUNCS([lchflags lchmod lchown link localtime_r lstat lutimes]) AC_CHECK_FUNCS([mbrtowc memmove memset]) AC_CHECK_FUNCS([mkdir mkfifo mknod mkstemp]) AC_CHECK_FUNCS([nl_langinfo openat pipe poll posix_spawnp readlink readlinkat]) AC_CHECK_FUNCS([readpassphrase]) AC_CHECK_FUNCS([select setenv setlocale sigaction statfs statvfs]) AC_CHECK_FUNCS([strchr strdup strerror strncpy_s strrchr symlink timegm]) AC_CHECK_FUNCS([tzset unsetenv utime utimensat utimes vfork]) AC_CHECK_FUNCS([wcrtomb wcscmp wcscpy wcslen wctomb wmemcmp wmemcpy wmemmove]) AC_CHECK_FUNCS([_ctime64_s _fseeki64]) AC_CHECK_FUNCS([_get_timezone _localtime64_s _mkgmtime64]) # detects cygwin-1.7, as opposed to older versions AC_CHECK_FUNCS([cygwin_conv_path]) # There are several variants of readdir_r around; we only # accept the POSIX-compliant version. AC_COMPILE_IFELSE( [AC_LANG_PROGRAM([[#include ]], [[DIR *dir; struct dirent e, *r; return(readdir_r(dir, &e, &r));]])], [AC_DEFINE(HAVE_READDIR_R,1,[Define to 1 if you have a POSIX compatible readdir_r])] ) # FreeBSD's nl_langinfo supports an option to specify whether the # current locale uses month/day or day/month ordering. It makes the # output a little prettier... AC_CHECK_DECL([D_MD_ORDER], [AC_DEFINE(HAVE_D_MD_ORDER, 1, [Define to 1 if nl_langinfo supports D_MD_ORDER])], [], [#if HAVE_LANGINFO_H #include #endif ]) # Check for dirent.d_namlen field explicitly # (This is a bit more straightforward than, if not quite as portable as, # the recipe given by the autoconf maintainers.) AC_CHECK_MEMBER(struct dirent.d_namlen,,, [#if HAVE_DIRENT_H #include #endif ]) # Check for Extended Attributes support AC_ARG_ENABLE([xattr], AS_HELP_STRING([--disable-xattr], [Disable Extended Attributes support (default: check)])) if test "x$enable_xattr" != "xno"; then AC_CHECK_HEADERS([attr/xattr.h]) AC_CHECK_HEADERS([sys/xattr.h sys/ea.h]) AC_SEARCH_LIBS([setxattr], [attr]) AC_CHECK_FUNCS([extattr_get_file extattr_list_file]) AC_CHECK_FUNCS([extattr_set_fd extattr_set_file]) AC_CHECK_FUNCS([fgetxattr flistxattr fsetxattr getxattr]) AC_CHECK_FUNCS([lgetxattr listxattr llistxattr lsetxattr]) AC_CHECK_FUNCS([fgetea flistea fsetea getea]) AC_CHECK_FUNCS([lgetea listea llistea lsetea]) AC_CHECK_DECLS([EXTATTR_NAMESPACE_USER], [], [], [#include #include ]) fi # Check for ACL support # # The ACL support in libarchive is written against the POSIX1e draft, # which was never officially approved and varies quite a bit across # platforms. Worse, some systems have completely non-POSIX acl functions, # which makes the following checks rather more complex than I would like. # AC_ARG_ENABLE([acl], AS_HELP_STRING([--disable-acl], [Disable ACL support (default: check)])) if test "x$enable_acl" != "xno"; then AC_CHECK_HEADERS([acl/libacl.h]) AC_CHECK_HEADERS([sys/acl.h]) AC_CHECK_LIB([acl],[acl_get_file]) AC_CHECK_FUNCS([acl_create_entry acl_init acl_set_fd acl_set_fd_np acl_set_file]) AC_CHECK_TYPES(acl_permset_t,,, [#if HAVE_SYS_TYPES_H #include #endif #if HAVE_SYS_ACL_H #include #endif ]) # The "acl_get_perm()" function was omitted from the POSIX draft. # (It's a pretty obvious oversight; otherwise, there's no way to # test for specific permissions in a permset.) Linux uses the obvious # name, FreeBSD adds _np to mark it as "non-Posix extension." # Test for both as a double-check that we really have POSIX-style ACL support. AC_CHECK_FUNCS(acl_get_perm_np acl_get_perm acl_get_link acl_get_link_np,,, [#if HAVE_SYS_TYPES_H #include #endif #if HAVE_SYS_ACL_H #include #endif ]) # MacOS has an acl.h that isn't POSIX. It can be detected by # checking for ACL_USER AC_CHECK_DECL([ACL_USER], [AC_DEFINE(HAVE_ACL_USER, 1, [True for systems with POSIX ACL support])], [], [#include ]) fi # Additional requirements AC_SYS_LARGEFILE dnl NOTE: Crypto checks must run last. AC_DEFUN([CRYPTO_CHECK], [ if test "$found_$1" != yes; then saved_CPPFLAGS="$CPPFLAGS" CPPFLAGS="$CPPFLAGS -I. -I$srcdir -I$srcdir/libarchive" touch "check_crypto_md.h" AC_MSG_CHECKING([support for ARCHIVE_CRYPTO_$1_$2]) AC_LINK_IFELSE([AC_LANG_SOURCE([ #define ARCHIVE_$1_COMPILE_TEST #define ARCHIVE_CRYPTO_$1_$2 #define PLATFORM_CONFIG_H "check_crypto_md.h" $(cat "$srcdir/libarchive/archive_digest.c") int main(int argc, char **argv) { archive_$3_ctx ctx; archive_$3_init(&ctx); archive_$3_update(&ctx, *argv, argc); archive_$3_final(&ctx, NULL); return 0; } ])], [ AC_MSG_RESULT([yes]) found_$1=yes found_$2=yes AC_DEFINE(ARCHIVE_CRYPTO_$1_$2, 1, [ $1 via ARCHIVE_CRYPTO_$1_$2 supported.]) ], [ AC_MSG_RESULT([no])]) CPPFLAGS="$saved_CPPFLAGS" rm "check_crypto_md.h" fi ]) AC_DEFUN([CRYPTO_CHECK_WIN], [ if test "$found_$1" != yes; then AC_MSG_CHECKING([support for ARCHIVE_CRYPTO_$1_WIN]) AC_LINK_IFELSE([AC_LANG_SOURCE([ #define ARCHIVE_$1_COMPILE_TEST #include #include int main(int argc, char **argv) { (void)argc; (void)argv; return ($2); } ])], [ AC_MSG_RESULT([yes]) found_$1=yes found_WIN=yes AC_DEFINE(ARCHIVE_CRYPTO_$1_WIN, 1, [ $1 via ARCHIVE_CRYPTO_$1_WIN supported.]) ], [ AC_MSG_RESULT([no])]) fi ]) case "$host_os" in *mingw* | *cygwin*) ;; *) CRYPTO_CHECK(MD5, LIBC, md5) CRYPTO_CHECK(MD5, LIBSYSTEM, md5) CRYPTO_CHECK(RMD160, LIBC, rmd160) CRYPTO_CHECK(SHA1, LIBC, sha1) CRYPTO_CHECK(SHA1, LIBSYSTEM, sha1) CRYPTO_CHECK(SHA256, LIBC, sha256) CRYPTO_CHECK(SHA256, LIBC2, sha256) CRYPTO_CHECK(SHA256, LIBC3, sha256) CRYPTO_CHECK(SHA256, LIBSYSTEM, sha256) CRYPTO_CHECK(SHA384, LIBC, sha384) CRYPTO_CHECK(SHA384, LIBC2, sha384) CRYPTO_CHECK(SHA384, LIBC3, sha384) CRYPTO_CHECK(SHA384, LIBSYSTEM, sha384) CRYPTO_CHECK(SHA512, LIBC, sha512) CRYPTO_CHECK(SHA512, LIBC2, sha512) CRYPTO_CHECK(SHA512, LIBC3, sha512) CRYPTO_CHECK(SHA512, LIBSYSTEM, sha512) ;; esac if test "x$with_nettle" != "xno"; then AC_CHECK_HEADERS([nettle/md5.h nettle/ripemd160.h nettle/sha.h]) AC_CHECK_HEADERS([nettle/pbkdf2.h nettle/aes.h nettle/hmac.h]) saved_LIBS=$LIBS AC_CHECK_LIB(nettle,nettle_sha1_init) CRYPTO_CHECK(MD5, NETTLE, md5) CRYPTO_CHECK(RMD160, NETTLE, rmd160) CRYPTO_CHECK(SHA1, NETTLE, sha1) CRYPTO_CHECK(SHA256, NETTLE, sha256) CRYPTO_CHECK(SHA384, NETTLE, sha384) CRYPTO_CHECK(SHA512, NETTLE, sha512) if test "x$found_NETTLE" != "xyes"; then LIBS=$saved_LIBS fi fi if test "x$with_openssl" != "xno"; then AC_CHECK_HEADERS([openssl/evp.h]) saved_LIBS=$LIBS case "$host_os" in *mingw* | *cygwin*) case "$host_cpu" in x86_64) AC_CHECK_LIB(eay64,OPENSSL_config) if test "x$ac_cv_lib_eay64_main" != "xyes"; then AC_CHECK_LIB(eay32,OPENSSL_config) fi ;; *) AC_CHECK_LIB(eay32,OPENSSL_config) ;; esac ;; *) AC_CHECK_LIB(crypto,OPENSSL_config) ;; esac CRYPTO_CHECK(MD5, OPENSSL, md5) CRYPTO_CHECK(RMD160, OPENSSL, rmd160) CRYPTO_CHECK(SHA1, OPENSSL, sha1) CRYPTO_CHECK(SHA256, OPENSSL, sha256) CRYPTO_CHECK(SHA384, OPENSSL, sha384) CRYPTO_CHECK(SHA512, OPENSSL, sha512) if test "x$found_OPENSSL" != "xyes"; then LIBS=$saved_LIBS else AC_CHECK_FUNCS([PKCS5_PBKDF2_HMAC_SHA1]) fi + AC_CHECK_LIB(crypto,EVP_CIPHER_CTX_init) fi # Probe libmd AFTER OpenSSL/libcrypto. # The two are incompatible and OpenSSL is more complete. AC_CHECK_HEADERS([md5.h ripemd.h sha.h sha256.h sha512.h]) saved_LIBS=$LIBS AC_CHECK_LIB(md,MD5Init) CRYPTO_CHECK(MD5, LIBMD, md5) CRYPTO_CHECK(RMD160, LIBMD, rmd160) CRYPTO_CHECK(SHA1, LIBMD, sha1) CRYPTO_CHECK(SHA256, LIBMD, sha256) CRYPTO_CHECK(SHA512, LIBMD, sha512) if test "x$found_LIBMD" != "xyes"; then LIBS=$saved_LIBS fi case "$host_os" in *mingw* | *cygwin*) CRYPTO_CHECK_WIN(MD5, CALG_MD5) CRYPTO_CHECK_WIN(SHA1, CALG_SHA1) CRYPTO_CHECK_WIN(SHA256, CALG_SHA_256) CRYPTO_CHECK_WIN(SHA384, CALG_SHA_384) CRYPTO_CHECK_WIN(SHA512, CALG_SHA_512) ;; esac # Ensure test directories are present if building out-of-tree AC_CONFIG_COMMANDS([mkdirs], [mkdir -p libarchive/test tar/test cat/test cpio/test]) AC_OUTPUT Index: vendor/libarchive/dist/cpio/cpio.c =================================================================== --- vendor/libarchive/dist/cpio/cpio.c (revision 302055) +++ vendor/libarchive/dist/cpio/cpio.c (revision 302056) @@ -1,1471 +1,1471 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "cpio_platform.h" __FBSDID("$FreeBSD: src/usr.bin/cpio/cpio.c,v 1.15 2008/12/06 07:30:40 kientzle Exp $"); #include #include #include #ifdef HAVE_SYS_MKDEV_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_GRP_H #include #endif #ifdef HAVE_LOCALE_H #include #endif #ifdef HAVE_PWD_H #include #endif #ifdef HAVE_SIGNAL_H #include #endif #ifdef HAVE_STDARG_H #include #endif #ifdef HAVE_STDINT_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_TIME_H #include #endif #include "cpio.h" #include "err.h" #include "line_reader.h" #include "passphrase.h" /* Fixed size of uname/gname caches. */ #define name_cache_size 101 #ifndef O_BINARY #define O_BINARY 0 #endif struct name_cache { int probes; int hits; size_t size; struct { id_t id; char *name; } cache[name_cache_size]; }; static int extract_data(struct archive *, struct archive *); const char * cpio_i64toa(int64_t); static const char *cpio_rename(const char *name); static int entry_to_archive(struct cpio *, struct archive_entry *); static int file_to_archive(struct cpio *, const char *); static void free_cache(struct name_cache *cache); static void list_item_verbose(struct cpio *, struct archive_entry *); static void long_help(void); static const char *lookup_gname(struct cpio *, gid_t gid); static int lookup_gname_helper(struct cpio *, const char **name, id_t gid); static const char *lookup_uname(struct cpio *, uid_t uid); static int lookup_uname_helper(struct cpio *, const char **name, id_t uid); static void mode_in(struct cpio *); static void mode_list(struct cpio *); static void mode_out(struct cpio *); static void mode_pass(struct cpio *, const char *); static const char *remove_leading_slash(const char *); static int restore_time(struct cpio *, struct archive_entry *, const char *, int fd); static void usage(void); static void version(void); static const char * passphrase_callback(struct archive *, void *); static void passphrase_free(char *); int main(int argc, char *argv[]) { static char buff[16384]; struct cpio _cpio; /* Allocated on stack. */ struct cpio *cpio; const char *errmsg; int uid, gid; int opt; cpio = &_cpio; memset(cpio, 0, sizeof(*cpio)); cpio->buff = buff; cpio->buff_size = sizeof(buff); #if defined(HAVE_SIGACTION) && defined(SIGPIPE) { /* Ignore SIGPIPE signals. */ struct sigaction sa; sigemptyset(&sa.sa_mask); sa.sa_flags = 0; sa.sa_handler = SIG_IGN; sigaction(SIGPIPE, &sa, NULL); } #endif /* Set lafe_progname before calling lafe_warnc. */ lafe_setprogname(*argv, "bsdcpio"); #if HAVE_SETLOCALE if (setlocale(LC_ALL, "") == NULL) lafe_warnc(0, "Failed to set default locale"); #endif cpio->uid_override = -1; cpio->gid_override = -1; cpio->argv = argv; cpio->argc = argc; cpio->mode = '\0'; cpio->verbose = 0; cpio->compress = '\0'; cpio->extract_flags = ARCHIVE_EXTRACT_NO_AUTODIR; cpio->extract_flags |= ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER; cpio->extract_flags |= ARCHIVE_EXTRACT_SECURE_SYMLINKS; cpio->extract_flags |= ARCHIVE_EXTRACT_SECURE_NODOTDOT; cpio->extract_flags |= ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS; cpio->extract_flags |= ARCHIVE_EXTRACT_PERM; cpio->extract_flags |= ARCHIVE_EXTRACT_FFLAGS; cpio->extract_flags |= ARCHIVE_EXTRACT_ACL; #if !defined(_WIN32) && !defined(__CYGWIN__) if (geteuid() == 0) cpio->extract_flags |= ARCHIVE_EXTRACT_OWNER; #endif cpio->bytes_per_block = 512; cpio->filename = NULL; cpio->matching = archive_match_new(); if (cpio->matching == NULL) lafe_errc(1, 0, "Out of memory"); while ((opt = cpio_getopt(cpio)) != -1) { switch (opt) { case '0': /* GNU convention: --null, -0 */ cpio->option_null = 1; break; case 'A': /* NetBSD/OpenBSD */ cpio->option_append = 1; break; case 'a': /* POSIX 1997 */ cpio->option_atime_restore = 1; break; case 'B': /* POSIX 1997 */ cpio->bytes_per_block = 5120; break; case OPTION_B64ENCODE: cpio->add_filter = opt; break; case 'C': /* NetBSD/OpenBSD */ cpio->bytes_per_block = atoi(cpio->argument); if (cpio->bytes_per_block <= 0) lafe_errc(1, 0, "Invalid blocksize %s", cpio->argument); break; case 'c': /* POSIX 1997 */ cpio->format = "odc"; break; case 'd': /* POSIX 1997 */ cpio->extract_flags &= ~ARCHIVE_EXTRACT_NO_AUTODIR; break; case 'E': /* NetBSD/OpenBSD */ if (archive_match_include_pattern_from_file( cpio->matching, cpio->argument, cpio->option_null) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(cpio->matching)); break; case 'F': /* NetBSD/OpenBSD/GNU cpio */ cpio->filename = cpio->argument; break; case 'f': /* POSIX 1997 */ if (archive_match_exclude_pattern(cpio->matching, cpio->argument) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(cpio->matching)); break; case OPTION_GRZIP: cpio->compress = opt; break; case 'H': /* GNU cpio (also --format) */ cpio->format = cpio->argument; break; case 'h': long_help(); break; case 'I': /* NetBSD/OpenBSD */ cpio->filename = cpio->argument; break; case 'i': /* POSIX 1997 */ if (cpio->mode != '\0') lafe_errc(1, 0, "Cannot use both -i and -%c", cpio->mode); cpio->mode = opt; break; case 'J': /* GNU tar, others */ cpio->compress = opt; break; case 'j': /* GNU tar, others */ cpio->compress = opt; break; case OPTION_INSECURE: cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_SYMLINKS; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NODOTDOT; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS; break; case 'L': /* GNU cpio */ cpio->option_follow_links = 1; break; case 'l': /* POSIX 1997 */ cpio->option_link = 1; break; case OPTION_LRZIP: case OPTION_LZ4: case OPTION_LZMA: /* GNU tar, others */ case OPTION_LZOP: /* GNU tar, others */ cpio->compress = opt; break; case 'm': /* POSIX 1997 */ cpio->extract_flags |= ARCHIVE_EXTRACT_TIME; break; case 'n': /* GNU cpio */ cpio->option_numeric_uid_gid = 1; break; case OPTION_NO_PRESERVE_OWNER: /* GNU cpio */ cpio->extract_flags &= ~ARCHIVE_EXTRACT_OWNER; break; case 'O': /* GNU cpio */ cpio->filename = cpio->argument; break; case 'o': /* POSIX 1997 */ if (cpio->mode != '\0') lafe_errc(1, 0, "Cannot use both -o and -%c", cpio->mode); cpio->mode = opt; break; case 'p': /* POSIX 1997 */ if (cpio->mode != '\0') lafe_errc(1, 0, "Cannot use both -p and -%c", cpio->mode); cpio->mode = opt; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NODOTDOT; cpio->extract_flags &= ~ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS; break; case OPTION_PASSPHRASE: cpio->passphrase = cpio->argument; break; case OPTION_PRESERVE_OWNER: cpio->extract_flags |= ARCHIVE_EXTRACT_OWNER; break; case OPTION_QUIET: /* GNU cpio */ cpio->quiet = 1; break; case 'R': /* GNU cpio, also --owner */ /* TODO: owner_parse should return uname/gname * also; use that to set [ug]name_override. */ errmsg = owner_parse(cpio->argument, &uid, &gid); if (errmsg) { lafe_warnc(-1, "%s", errmsg); usage(); } if (uid != -1) { cpio->uid_override = uid; cpio->uname_override = NULL; } if (gid != -1) { cpio->gid_override = gid; cpio->gname_override = NULL; } break; case 'r': /* POSIX 1997 */ cpio->option_rename = 1; break; case 't': /* POSIX 1997 */ cpio->option_list = 1; break; case 'u': /* POSIX 1997 */ cpio->extract_flags &= ~ARCHIVE_EXTRACT_NO_OVERWRITE_NEWER; break; case OPTION_UUENCODE: cpio->add_filter = opt; break; case 'v': /* POSIX 1997 */ cpio->verbose++; break; case 'V': /* GNU cpio */ cpio->dot++; break; case OPTION_VERSION: /* GNU convention */ version(); break; #if 0 /* * cpio_getopt() handles -W specially, so it's not * available here. */ case 'W': /* Obscure, but useful GNU convention. */ break; #endif case 'y': /* tar convention */ cpio->compress = opt; break; case 'Z': /* tar convention */ cpio->compress = opt; break; case 'z': /* tar convention */ cpio->compress = opt; break; default: usage(); } } /* * Sanity-check args, error out on nonsensical combinations. */ /* -t implies -i if no mode was specified. */ if (cpio->option_list && cpio->mode == '\0') cpio->mode = 'i'; /* -t requires -i */ if (cpio->option_list && cpio->mode != 'i') lafe_errc(1, 0, "Option -t requires -i"); /* -n requires -it */ if (cpio->option_numeric_uid_gid && !cpio->option_list) lafe_errc(1, 0, "Option -n requires -it"); /* Can only specify format when writing */ if (cpio->format != NULL && cpio->mode != 'o') lafe_errc(1, 0, "Option --format requires -o"); /* -l requires -p */ if (cpio->option_link && cpio->mode != 'p') lafe_errc(1, 0, "Option -l requires -p"); /* -v overrides -V */ if (cpio->dot && cpio->verbose) cpio->dot = 0; /* TODO: Flag other nonsensical combinations. */ switch (cpio->mode) { case 'o': /* TODO: Implement old binary format in libarchive, use that here. */ if (cpio->format == NULL) cpio->format = "odc"; /* Default format */ mode_out(cpio); break; case 'i': while (*cpio->argv != NULL) { if (archive_match_include_pattern(cpio->matching, *cpio->argv) != ARCHIVE_OK) lafe_errc(1, 0, "Error : %s", archive_error_string(cpio->matching)); --cpio->argc; ++cpio->argv; } if (cpio->option_list) mode_list(cpio); else mode_in(cpio); break; case 'p': if (*cpio->argv == NULL || **cpio->argv == '\0') lafe_errc(1, 0, "-p mode requires a target directory"); mode_pass(cpio, *cpio->argv); break; default: lafe_errc(1, 0, "Must specify at least one of -i, -o, or -p"); } archive_match_free(cpio->matching); free_cache(cpio->gname_cache); free_cache(cpio->uname_cache); free(cpio->destdir); passphrase_free(cpio->ppbuff); return (cpio->return_value); } static void usage(void) { const char *p; p = lafe_getprogname(); fprintf(stderr, "Brief Usage:\n"); fprintf(stderr, " List: %s -it < archive\n", p); fprintf(stderr, " Extract: %s -i < archive\n", p); fprintf(stderr, " Create: %s -o < filenames > archive\n", p); fprintf(stderr, " Help: %s --help\n", p); exit(1); } static const char *long_help_msg = "First option must be a mode specifier:\n" " -i Input -o Output -p Pass\n" "Common Options:\n" " -v Verbose filenames -V one dot per file\n" "Create: %p -o [options] < [list of files] > [archive]\n" " -J,-y,-z,--lzma Compress archive with xz/bzip2/gzip/lzma\n" " --format {odc|newc|ustar} Select archive format\n" "List: %p -it < [archive]\n" "Extract: %p -i [options] < [archive]\n"; /* * Note that the word 'bsdcpio' will always appear in the first line * of output. * * In particular, /bin/sh scripts that need to test for the presence * of bsdcpio can use the following template: * * if (cpio --help 2>&1 | grep bsdcpio >/dev/null 2>&1 ) then \ * echo bsdcpio; else echo not bsdcpio; fi */ static void long_help(void) { const char *prog; const char *p; prog = lafe_getprogname(); fflush(stderr); p = (strcmp(prog,"bsdcpio") != 0) ? "(bsdcpio)" : ""; printf("%s%s: manipulate archive files\n", prog, p); for (p = long_help_msg; *p != '\0'; p++) { if (*p == '%') { if (p[1] == 'p') { fputs(prog, stdout); p++; } else putchar('%'); } else putchar(*p); } version(); } static void version(void) { - fprintf(stdout,"bsdcpio %s -- %s\n", + fprintf(stdout,"bsdcpio %s - %s\n", BSDCPIO_VERSION_STRING, archive_version_details()); exit(0); } static void mode_out(struct cpio *cpio) { struct archive_entry *entry, *spare; struct lafe_line_reader *lr; const char *p; int r; if (cpio->option_append) lafe_errc(1, 0, "Append mode not yet supported."); cpio->archive_read_disk = archive_read_disk_new(); if (cpio->archive_read_disk == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); if (cpio->option_follow_links) archive_read_disk_set_symlink_logical(cpio->archive_read_disk); else archive_read_disk_set_symlink_physical(cpio->archive_read_disk); archive_read_disk_set_standard_lookup(cpio->archive_read_disk); cpio->archive = archive_write_new(); if (cpio->archive == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); switch (cpio->compress) { case OPTION_GRZIP: r = archive_write_add_filter_grzip(cpio->archive); break; case 'J': r = archive_write_add_filter_xz(cpio->archive); break; case OPTION_LRZIP: r = archive_write_add_filter_lrzip(cpio->archive); break; case OPTION_LZ4: r = archive_write_add_filter_lz4(cpio->archive); break; case OPTION_LZMA: r = archive_write_add_filter_lzma(cpio->archive); break; case OPTION_LZOP: r = archive_write_add_filter_lzop(cpio->archive); break; case 'j': case 'y': r = archive_write_add_filter_bzip2(cpio->archive); break; case 'z': r = archive_write_add_filter_gzip(cpio->archive); break; case 'Z': r = archive_write_add_filter_compress(cpio->archive); break; default: r = archive_write_add_filter_none(cpio->archive); break; } if (r < ARCHIVE_WARN) lafe_errc(1, 0, "Requested compression not available"); switch (cpio->add_filter) { case 0: r = ARCHIVE_OK; break; case OPTION_B64ENCODE: r = archive_write_add_filter_b64encode(cpio->archive); break; case OPTION_UUENCODE: r = archive_write_add_filter_uuencode(cpio->archive); break; } if (r < ARCHIVE_WARN) lafe_errc(1, 0, "Requested filter not available"); r = archive_write_set_format_by_name(cpio->archive, cpio->format); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); archive_write_set_bytes_per_block(cpio->archive, cpio->bytes_per_block); cpio->linkresolver = archive_entry_linkresolver_new(); archive_entry_linkresolver_set_strategy(cpio->linkresolver, archive_format(cpio->archive)); if (cpio->passphrase != NULL) r = archive_write_set_passphrase(cpio->archive, cpio->passphrase); else r = archive_write_set_passphrase_callback(cpio->archive, cpio, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); /* * The main loop: Copy each file into the output archive. */ r = archive_write_open_filename(cpio->archive, cpio->filename); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); lr = lafe_line_reader("-", cpio->option_null); while ((p = lafe_line_reader_next(lr)) != NULL) file_to_archive(cpio, p); lafe_line_reader_free(lr); /* * The hardlink detection may have queued up a couple of entries * that can now be flushed. */ entry = NULL; archive_entry_linkify(cpio->linkresolver, &entry, &spare); while (entry != NULL) { entry_to_archive(cpio, entry); archive_entry_free(entry); entry = NULL; archive_entry_linkify(cpio->linkresolver, &entry, &spare); } r = archive_write_close(cpio->archive); if (cpio->dot) fprintf(stderr, "\n"); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(cpio->archive, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_write_free(cpio->archive); } static const char * remove_leading_slash(const char *p) { const char *rp; /* Remove leading "//./" or "//?/" or "//?/UNC/" * (absolute path prefixes used by Windows API) */ if ((p[0] == '/' || p[0] == '\\') && (p[1] == '/' || p[1] == '\\') && (p[2] == '.' || p[2] == '?') && (p[3] == '/' || p[3] == '\\')) { if (p[2] == '?' && (p[4] == 'U' || p[4] == 'u') && (p[5] == 'N' || p[5] == 'n') && (p[6] == 'C' || p[6] == 'c') && (p[7] == '/' || p[7] == '\\')) p += 8; else p += 4; } do { rp = p; /* Remove leading drive letter from archives created * on Windows. */ if (((p[0] >= 'a' && p[0] <= 'z') || (p[0] >= 'A' && p[0] <= 'Z')) && p[1] == ':') { p += 2; } /* Remove leading "/../", "//", etc. */ while (p[0] == '/' || p[0] == '\\') { if (p[1] == '.' && p[2] == '.' && (p[3] == '/' || p[3] == '\\')) { p += 3; /* Remove "/..", leave "/" * for next pass. */ } else p += 1; /* Remove "/". */ } } while (rp != p); return (p); } /* * This is used by both out mode (to copy objects from disk into * an archive) and pass mode (to copy objects from disk to * an archive_write_disk "archive"). */ static int file_to_archive(struct cpio *cpio, const char *srcpath) { const char *destpath; struct archive_entry *entry, *spare; size_t len; int r; /* * Create an archive_entry describing the source file. * */ entry = archive_entry_new(); if (entry == NULL) lafe_errc(1, 0, "Couldn't allocate entry"); archive_entry_copy_sourcepath(entry, srcpath); r = archive_read_disk_entry_from_file(cpio->archive_read_disk, entry, -1, NULL); if (r < ARCHIVE_FAILED) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive_read_disk)); if (r < ARCHIVE_OK) lafe_warnc(0, "%s", archive_error_string(cpio->archive_read_disk)); if (r <= ARCHIVE_FAILED) { cpio->return_value = 1; return (r); } if (cpio->uid_override >= 0) { archive_entry_set_uid(entry, cpio->uid_override); archive_entry_set_uname(entry, cpio->uname_override); } if (cpio->gid_override >= 0) { archive_entry_set_gid(entry, cpio->gid_override); archive_entry_set_gname(entry, cpio->gname_override); } /* * Generate a destination path for this entry. * "destination path" is the name to which it will be copied in * pass mode or the name that will go into the archive in * output mode. */ destpath = srcpath; if (cpio->destdir) { len = strlen(cpio->destdir) + strlen(srcpath) + 8; if (len >= cpio->pass_destpath_alloc) { while (len >= cpio->pass_destpath_alloc) { cpio->pass_destpath_alloc += 512; cpio->pass_destpath_alloc *= 2; } free(cpio->pass_destpath); cpio->pass_destpath = malloc(cpio->pass_destpath_alloc); if (cpio->pass_destpath == NULL) lafe_errc(1, ENOMEM, "Can't allocate path buffer"); } strcpy(cpio->pass_destpath, cpio->destdir); strcat(cpio->pass_destpath, remove_leading_slash(srcpath)); destpath = cpio->pass_destpath; } if (cpio->option_rename) destpath = cpio_rename(destpath); if (destpath == NULL) return (0); archive_entry_copy_pathname(entry, destpath); /* * If we're trying to preserve hardlinks, match them here. */ spare = NULL; if (cpio->linkresolver != NULL && archive_entry_filetype(entry) != AE_IFDIR) { archive_entry_linkify(cpio->linkresolver, &entry, &spare); } if (entry != NULL) { r = entry_to_archive(cpio, entry); archive_entry_free(entry); if (spare != NULL) { if (r == 0) r = entry_to_archive(cpio, spare); archive_entry_free(spare); } } return (r); } static int entry_to_archive(struct cpio *cpio, struct archive_entry *entry) { const char *destpath = archive_entry_pathname(entry); const char *srcpath = archive_entry_sourcepath(entry); int fd = -1; ssize_t bytes_read; int r; /* Print out the destination name to the user. */ if (cpio->verbose) fprintf(stderr,"%s", destpath); if (cpio->dot) fprintf(stderr, "."); /* * Option_link only makes sense in pass mode and for * regular files. Also note: if a link operation fails * because of cross-device restrictions, we'll fall back * to copy mode for that entry. * * TODO: Test other cpio implementations to see if they * hard-link anything other than regular files here. */ if (cpio->option_link && archive_entry_filetype(entry) == AE_IFREG) { struct archive_entry *t; /* Save the original entry in case we need it later. */ t = archive_entry_clone(entry); if (t == NULL) lafe_errc(1, ENOMEM, "Can't create link"); /* Note: link(2) doesn't create parent directories, * so we use archive_write_header() instead as a * convenience. */ archive_entry_set_hardlink(t, srcpath); /* This is a straight link that carries no data. */ archive_entry_set_size(t, 0); r = archive_write_header(cpio->archive, t); archive_entry_free(t); if (r != ARCHIVE_OK) lafe_warnc(archive_errno(cpio->archive), "%s", archive_error_string(cpio->archive)); if (r == ARCHIVE_FATAL) exit(1); #ifdef EXDEV if (r != ARCHIVE_OK && archive_errno(cpio->archive) == EXDEV) { /* Cross-device link: Just fall through and use * the original entry to copy the file over. */ lafe_warnc(0, "Copying file instead"); } else #endif return (0); } /* * Make sure we can open the file (if necessary) before * trying to write the header. */ if (archive_entry_filetype(entry) == AE_IFREG) { if (archive_entry_size(entry) > 0) { fd = open(srcpath, O_RDONLY | O_BINARY); if (fd < 0) { lafe_warnc(errno, "%s: could not open file", srcpath); goto cleanup; } } } else { archive_entry_set_size(entry, 0); } r = archive_write_header(cpio->archive, entry); if (r != ARCHIVE_OK) lafe_warnc(archive_errno(cpio->archive), "%s: %s", srcpath, archive_error_string(cpio->archive)); if (r == ARCHIVE_FATAL) exit(1); if (r >= ARCHIVE_WARN && archive_entry_size(entry) > 0 && fd >= 0) { bytes_read = read(fd, cpio->buff, (unsigned)cpio->buff_size); while (bytes_read > 0) { ssize_t bytes_write; bytes_write = archive_write_data(cpio->archive, cpio->buff, bytes_read); if (bytes_write < 0) lafe_errc(1, archive_errno(cpio->archive), "%s", archive_error_string(cpio->archive)); if (bytes_write < bytes_read) { lafe_warnc(0, "Truncated write; file may have " "grown while being archived."); } bytes_read = read(fd, cpio->buff, (unsigned)cpio->buff_size); } } fd = restore_time(cpio, entry, srcpath, fd); cleanup: if (cpio->verbose) fprintf(stderr,"\n"); if (fd >= 0) close(fd); return (0); } static int restore_time(struct cpio *cpio, struct archive_entry *entry, const char *name, int fd) { #ifndef HAVE_UTIMES static int warned = 0; (void)cpio; /* UNUSED */ (void)entry; /* UNUSED */ (void)name; /* UNUSED */ if (!warned) lafe_warnc(0, "Can't restore access times on this platform"); warned = 1; return (fd); #else #if defined(_WIN32) && !defined(__CYGWIN__) struct __timeval times[2]; #else struct timeval times[2]; #endif if (!cpio->option_atime_restore) return (fd); times[1].tv_sec = archive_entry_mtime(entry); times[1].tv_usec = archive_entry_mtime_nsec(entry) / 1000; times[0].tv_sec = archive_entry_atime(entry); times[0].tv_usec = archive_entry_atime_nsec(entry) / 1000; #if defined(HAVE_FUTIMES) && !defined(__CYGWIN__) if (fd >= 0 && futimes(fd, times) == 0) return (fd); #endif /* * Some platform cannot restore access times if the file descriptor * is still opened. */ if (fd >= 0) { close(fd); fd = -1; } #ifdef HAVE_LUTIMES if (lutimes(name, times) != 0) #else if ((AE_IFLNK != archive_entry_filetype(entry)) && utimes(name, times) != 0) #endif lafe_warnc(errno, "Can't update time for %s", name); #endif return (fd); } static void mode_in(struct cpio *cpio) { struct archive *a; struct archive_entry *entry; struct archive *ext; const char *destpath; int r; ext = archive_write_disk_new(); if (ext == NULL) lafe_errc(1, 0, "Couldn't allocate restore object"); r = archive_write_disk_set_options(ext, cpio->extract_flags); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(ext)); a = archive_read_new(); if (a == NULL) lafe_errc(1, 0, "Couldn't allocate archive object"); archive_read_support_filter_all(a); archive_read_support_format_all(a); if (cpio->passphrase != NULL) r = archive_read_add_passphrase(a, cpio->passphrase); else r = archive_read_set_passphrase_callback(a, cpio, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (archive_read_open_filename(a, cpio->filename, cpio->bytes_per_block)) lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); for (;;) { r = archive_read_next_header(a, &entry); if (r == ARCHIVE_EOF) break; if (r != ARCHIVE_OK) { lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); } if (archive_match_path_excluded(cpio->matching, entry)) continue; if (cpio->option_rename) { destpath = cpio_rename(archive_entry_pathname(entry)); archive_entry_set_pathname(entry, destpath); } else destpath = archive_entry_pathname(entry); if (destpath == NULL) continue; if (cpio->verbose) fprintf(stderr, "%s\n", destpath); if (cpio->dot) fprintf(stderr, "."); if (cpio->uid_override >= 0) archive_entry_set_uid(entry, cpio->uid_override); if (cpio->gid_override >= 0) archive_entry_set_gid(entry, cpio->gid_override); r = archive_write_header(ext, entry); if (r != ARCHIVE_OK) { fprintf(stderr, "%s: %s\n", archive_entry_pathname(entry), archive_error_string(ext)); } else if (!archive_entry_size_is_set(entry) || archive_entry_size(entry) > 0) { r = extract_data(a, ext); if (r != ARCHIVE_OK) cpio->return_value = 1; } } r = archive_read_close(a); if (cpio->dot) fprintf(stderr, "\n"); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); r = archive_write_close(ext); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(ext)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(a, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_read_free(a); archive_write_free(ext); exit(cpio->return_value); } /* * Exits if there's a fatal error. Returns ARCHIVE_OK * if everything is kosher. */ static int extract_data(struct archive *ar, struct archive *aw) { int r; size_t size; const void *block; int64_t offset; for (;;) { r = archive_read_data_block(ar, &block, &size, &offset); if (r == ARCHIVE_EOF) return (ARCHIVE_OK); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(ar), "%s", archive_error_string(ar)); exit(1); } r = (int)archive_write_data_block(aw, block, size, offset); if (r != ARCHIVE_OK) { lafe_warnc(archive_errno(aw), "%s", archive_error_string(aw)); return (r); } } } static void mode_list(struct cpio *cpio) { struct archive *a; struct archive_entry *entry; int r; a = archive_read_new(); if (a == NULL) lafe_errc(1, 0, "Couldn't allocate archive object"); archive_read_support_filter_all(a); archive_read_support_format_all(a); if (cpio->passphrase != NULL) r = archive_read_add_passphrase(a, cpio->passphrase); else r = archive_read_set_passphrase_callback(a, cpio, &passphrase_callback); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (archive_read_open_filename(a, cpio->filename, cpio->bytes_per_block)) lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); for (;;) { r = archive_read_next_header(a, &entry); if (r == ARCHIVE_EOF) break; if (r != ARCHIVE_OK) { lafe_errc(1, archive_errno(a), "%s", archive_error_string(a)); } if (archive_match_path_excluded(cpio->matching, entry)) continue; if (cpio->verbose) list_item_verbose(cpio, entry); else fprintf(stdout, "%s\n", archive_entry_pathname(entry)); } r = archive_read_close(a); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(a)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(a, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_read_free(a); exit(0); } /* * Display information about the current file. * * The format here roughly duplicates the output of 'ls -l'. * This is based on SUSv2, where 'tar tv' is documented as * listing additional information in an "unspecified format," * and 'pax -l' is documented as using the same format as 'ls -l'. */ static void list_item_verbose(struct cpio *cpio, struct archive_entry *entry) { char size[32]; char date[32]; char uids[16], gids[16]; const char *uname, *gname; FILE *out = stdout; const char *fmt; time_t mtime; static time_t now; if (!now) time(&now); if (cpio->option_numeric_uid_gid) { /* Format numeric uid/gid for display. */ strcpy(uids, cpio_i64toa(archive_entry_uid(entry))); uname = uids; strcpy(gids, cpio_i64toa(archive_entry_gid(entry))); gname = gids; } else { /* Use uname if it's present, else lookup name from uid. */ uname = archive_entry_uname(entry); if (uname == NULL) uname = lookup_uname(cpio, (uid_t)archive_entry_uid(entry)); /* Use gname if it's present, else lookup name from gid. */ gname = archive_entry_gname(entry); if (gname == NULL) gname = lookup_gname(cpio, (uid_t)archive_entry_gid(entry)); } /* Print device number or file size. */ if (archive_entry_filetype(entry) == AE_IFCHR || archive_entry_filetype(entry) == AE_IFBLK) { snprintf(size, sizeof(size), "%lu,%lu", (unsigned long)archive_entry_rdevmajor(entry), (unsigned long)archive_entry_rdevminor(entry)); } else { strcpy(size, cpio_i64toa(archive_entry_size(entry))); } /* Format the time using 'ls -l' conventions. */ mtime = archive_entry_mtime(entry); #if defined(_WIN32) && !defined(__CYGWIN__) /* Windows' strftime function does not support %e format. */ if (mtime - now > 365*86400/2 || mtime - now < -365*86400/2) fmt = cpio->day_first ? "%d %b %Y" : "%b %d %Y"; else fmt = cpio->day_first ? "%d %b %H:%M" : "%b %d %H:%M"; #else if (mtime - now > 365*86400/2 || mtime - now < -365*86400/2) fmt = cpio->day_first ? "%e %b %Y" : "%b %e %Y"; else fmt = cpio->day_first ? "%e %b %H:%M" : "%b %e %H:%M"; #endif strftime(date, sizeof(date), fmt, localtime(&mtime)); fprintf(out, "%s%3d %-8s %-8s %8s %12s %s", archive_entry_strmode(entry), archive_entry_nlink(entry), uname, gname, size, date, archive_entry_pathname(entry)); /* Extra information for links. */ if (archive_entry_hardlink(entry)) /* Hard link */ fprintf(out, " link to %s", archive_entry_hardlink(entry)); else if (archive_entry_symlink(entry)) /* Symbolic link */ fprintf(out, " -> %s", archive_entry_symlink(entry)); fprintf(out, "\n"); } static void mode_pass(struct cpio *cpio, const char *destdir) { struct lafe_line_reader *lr; const char *p; int r; /* Ensure target dir has a trailing '/' to simplify path surgery. */ cpio->destdir = malloc(strlen(destdir) + 8); strcpy(cpio->destdir, destdir); if (destdir[strlen(destdir) - 1] != '/') strcat(cpio->destdir, "/"); cpio->archive = archive_write_disk_new(); if (cpio->archive == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); r = archive_write_disk_set_options(cpio->archive, cpio->extract_flags); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); cpio->linkresolver = archive_entry_linkresolver_new(); archive_write_disk_set_standard_lookup(cpio->archive); cpio->archive_read_disk = archive_read_disk_new(); if (cpio->archive_read_disk == NULL) lafe_errc(1, 0, "Failed to allocate archive object"); if (cpio->option_follow_links) archive_read_disk_set_symlink_logical(cpio->archive_read_disk); else archive_read_disk_set_symlink_physical(cpio->archive_read_disk); archive_read_disk_set_standard_lookup(cpio->archive_read_disk); lr = lafe_line_reader("-", cpio->option_null); while ((p = lafe_line_reader_next(lr)) != NULL) file_to_archive(cpio, p); lafe_line_reader_free(lr); archive_entry_linkresolver_free(cpio->linkresolver); r = archive_write_close(cpio->archive); if (cpio->dot) fprintf(stderr, "\n"); if (r != ARCHIVE_OK) lafe_errc(1, 0, "%s", archive_error_string(cpio->archive)); if (!cpio->quiet) { int64_t blocks = (archive_filter_bytes(cpio->archive, 0) + 511) / 512; fprintf(stderr, "%lu %s\n", (unsigned long)blocks, blocks == 1 ? "block" : "blocks"); } archive_write_free(cpio->archive); } /* * Prompt for a new name for this entry. Returns a pointer to the * new name or NULL if the entry should not be copied. This * implements the semantics defined in POSIX.1-1996, which specifies * that an input of '.' means the name should be unchanged. GNU cpio * treats '.' as a literal new name. */ static const char * cpio_rename(const char *name) { static char buff[1024]; FILE *t; char *p, *ret; #if defined(_WIN32) && !defined(__CYGWIN__) FILE *to; t = fopen("CONIN$", "r"); if (t == NULL) return (name); to = fopen("CONOUT$", "w"); if (to == NULL) { fclose(t); return (name); } fprintf(to, "%s (Enter/./(new name))? ", name); fclose(to); #else t = fopen("/dev/tty", "r+"); if (t == NULL) return (name); fprintf(t, "%s (Enter/./(new name))? ", name); fflush(t); #endif p = fgets(buff, sizeof(buff), t); fclose(t); if (p == NULL) /* End-of-file is a blank line. */ return (NULL); while (*p == ' ' || *p == '\t') ++p; if (*p == '\n' || *p == '\0') /* Empty line. */ return (NULL); if (*p == '.' && p[1] == '\n') /* Single period preserves original name. */ return (name); ret = p; /* Trim the final newline. */ while (*p != '\0' && *p != '\n') ++p; /* Overwrite the final \n with a null character. */ *p = '\0'; return (ret); } static void free_cache(struct name_cache *cache) { size_t i; if (cache != NULL) { for (i = 0; i < cache->size; i++) free(cache->cache[i].name); free(cache); } } /* * Lookup uname/gname from uid/gid, return NULL if no match. */ static const char * lookup_name(struct cpio *cpio, struct name_cache **name_cache_variable, int (*lookup_fn)(struct cpio *, const char **, id_t), id_t id) { char asnum[16]; struct name_cache *cache; const char *name; int slot; if (*name_cache_variable == NULL) { *name_cache_variable = malloc(sizeof(struct name_cache)); if (*name_cache_variable == NULL) lafe_errc(1, ENOMEM, "No more memory"); memset(*name_cache_variable, 0, sizeof(struct name_cache)); (*name_cache_variable)->size = name_cache_size; } cache = *name_cache_variable; cache->probes++; slot = id % cache->size; if (cache->cache[slot].name != NULL) { if (cache->cache[slot].id == id) { cache->hits++; return (cache->cache[slot].name); } free(cache->cache[slot].name); cache->cache[slot].name = NULL; } if (lookup_fn(cpio, &name, id) == 0) { if (name == NULL || name[0] == '\0') { /* If lookup failed, format it as a number. */ snprintf(asnum, sizeof(asnum), "%u", (unsigned)id); name = asnum; } cache->cache[slot].name = strdup(name); if (cache->cache[slot].name != NULL) { cache->cache[slot].id = id; return (cache->cache[slot].name); } /* * Conveniently, NULL marks an empty slot, so * if the strdup() fails, we've just failed to * cache it. No recovery necessary. */ } return (NULL); } static const char * lookup_uname(struct cpio *cpio, uid_t uid) { return (lookup_name(cpio, &cpio->uname_cache, &lookup_uname_helper, (id_t)uid)); } static int lookup_uname_helper(struct cpio *cpio, const char **name, id_t id) { struct passwd *pwent; (void)cpio; /* UNUSED */ errno = 0; pwent = getpwuid((uid_t)id); if (pwent == NULL) { *name = NULL; if (errno != 0 && errno != ENOENT) lafe_warnc(errno, "getpwuid(%s) failed", cpio_i64toa((int64_t)id)); return (errno); } *name = pwent->pw_name; return (0); } static const char * lookup_gname(struct cpio *cpio, gid_t gid) { return (lookup_name(cpio, &cpio->gname_cache, &lookup_gname_helper, (id_t)gid)); } static int lookup_gname_helper(struct cpio *cpio, const char **name, id_t id) { struct group *grent; (void)cpio; /* UNUSED */ errno = 0; grent = getgrgid((gid_t)id); if (grent == NULL) { *name = NULL; if (errno != 0) lafe_warnc(errno, "getgrgid(%s) failed", cpio_i64toa((int64_t)id)); return (errno); } *name = grent->gr_name; return (0); } /* * It would be nice to just use printf() for formatting large numbers, * but the compatibility problems are a big headache. Hence the * following simple utility function. */ const char * cpio_i64toa(int64_t n0) { /* 2^64 =~ 1.8 * 10^19, so 20 decimal digits suffice. * We also need 1 byte for '-' and 1 for '\0'. */ static char buff[22]; int64_t n = n0 < 0 ? -n0 : n0; char *p = buff + sizeof(buff); *--p = '\0'; do { *--p = '0' + (int)(n % 10); n /= 10; } while (n > 0); if (n0 < 0) *--p = '-'; return p; } #define PPBUFF_SIZE 1024 static const char * passphrase_callback(struct archive *a, void *_client_data) { struct cpio *cpio = (struct cpio *)_client_data; (void)a; /* UNUSED */ if (cpio->ppbuff == NULL) { cpio->ppbuff = malloc(PPBUFF_SIZE); if (cpio->ppbuff == NULL) lafe_errc(1, errno, "Out of memory"); } return lafe_readpassphrase("Enter passphrase:", cpio->ppbuff, PPBUFF_SIZE); } static void passphrase_free(char *ppbuff) { if (ppbuff != NULL) { memset(ppbuff, 0, PPBUFF_SIZE); free(ppbuff); } } Index: vendor/libarchive/dist/cpio/test/test_option_version.c =================================================================== --- vendor/libarchive/dist/cpio/test/test_option_version.c (revision 302055) +++ vendor/libarchive/dist/cpio/test/test_option_version.c (revision 302056) @@ -1,114 +1,114 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "test.h" __FBSDID("$FreeBSD$"); /* * Test that --version option works and generates reasonable output. */ static void verify(const char *p, size_t s) { const char *q = p; /* Version message should start with name of program, then space. */ failure("version message too short:", p); if (!assert(s > 6)) return; failure("Version message should begin with 'bsdcpio': %s", p); if (!assertEqualMem(q, "bsdcpio ", 8)) /* If we're not testing bsdcpio, don't keep going. */ return; q += 8; s -= 8; /* Version number is a series of digits and periods. */ while (s > 0 && (*q == '.' || (*q >= '0' && *q <= '9'))) { ++q; --s; } /* Version number terminated by space. */ failure("Version: %s", p); assert(s > 1); /* Skip a single trailing a,b,c, or d. */ if (*q == 'a' || *q == 'b' || *q == 'c' || *q == 'd') ++q; failure("Version: %s", p); assert(*q == ' '); ++q; --s; /* Separator. */ failure("Version: %s", p); - assertEqualMem(q, "-- ", 3); - q += 3; s -= 3; + assertEqualMem(q, "- ", 2); + q += 2; s -= 2; /* libarchive name and version number */ assert(s > 11); failure("Version: %s", p); assertEqualMem(q, "libarchive ", 11); q += 11; s -= 11; /* Version number is a series of digits and periods. */ while (s > 0 && (*q == '.' || (*q >= '0' && *q <= '9'))) { ++q; --s; } /* Skip a single trailing a,b,c, or d. */ if (*q == 'a' || *q == 'b' || *q == 'c' || *q == 'd') ++q; /* Skip arbitrary third-party version numbers. */ while (s > 0 && (*q == ' ' || *q == '/' || *q == '.' || isalnum(*q))) { ++q; --s; } /* All terminated by end-of-line: \r, \r\n, or \n */ assert(s >= 1); failure("Version: %s", p); if (*q == '\x0d') { if (q[1] != '\0') assertEqualMem(q, "\x0d\x0a", 2); } else assertEqualMem(q, "\x0a", 1); } DEFINE_TEST(test_option_version) { int r; char *p; size_t s; r = systemf("%s --version >version.stdout 2>version.stderr", testprog); if (r != 0) r = systemf("%s -W version >version.stdout 2>version.stderr", testprog); failure("Unable to run either %s --version or %s -W version", testprog, testprog); if (!assert(r == 0)) return; /* --version should generate nothing to stderr. */ assertEmptyFile("version.stderr"); /* Verify format of version message. */ p = slurpfile(&s, "version.stdout"); verify(p, s); free(p); } Index: vendor/libarchive/dist/libarchive/archive_read_support_format_lha.c =================================================================== --- vendor/libarchive/dist/libarchive/archive_read_support_format_lha.c (revision 302055) +++ vendor/libarchive/dist/libarchive/archive_read_support_format_lha.c (revision 302056) @@ -1,2807 +1,2808 @@ /*- * Copyright (c) 2008-2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_endian.h" #define MAXMATCH 256 /* Maximum match length. */ #define MINMATCH 3 /* Minimum match length. */ /* * Literal table format: * +0 +256 +510 * +---------------+-------------------------+ * | literal code | match length | * | 0 ... 255 | MINMATCH ... MAXMATCH | * +---------------+-------------------------+ * <--- LT_BITLEN_SIZE ---> */ /* Literal table size. */ #define LT_BITLEN_SIZE (UCHAR_MAX + 1 + MAXMATCH - MINMATCH + 1) /* Position table size. * Note: this used for both position table and pre literal table.*/ #define PT_BITLEN_SIZE (3 + 16) struct lzh_dec { /* Decoding status. */ int state; /* * Window to see last 8Ki(lh5),32Ki(lh6),64Ki(lh7) bytes of decoded * data. */ int w_size; int w_mask; /* Window buffer, which is a loop buffer. */ unsigned char *w_buff; /* The insert position to the window. */ int w_pos; /* The position where we can copy decoded code from the window. */ int copy_pos; /* The length how many bytes we can copy decoded code from * the window. */ int copy_len; /* * Bit stream reader. */ struct lzh_br { #define CACHE_TYPE uint64_t #define CACHE_BITS (8 * sizeof(CACHE_TYPE)) /* Cache buffer. */ CACHE_TYPE cache_buffer; /* Indicates how many bits avail in cache_buffer. */ int cache_avail; } br; /* * Huffman coding. */ struct huffman { int len_size; int len_avail; int len_bits; int freq[17]; unsigned char *bitlen; /* * Use a index table. It's faster than searching a huffman * coding tree, which is a binary tree. But a use of a large * index table causes L1 cache read miss many times. */ #define HTBL_BITS 10 int max_bits; int shift_bits; int tbl_bits; int tree_used; int tree_avail; /* Direct access table. */ uint16_t *tbl; /* Binary tree table for extra bits over the direct access. */ struct htree_t { uint16_t left; uint16_t right; } *tree; } lt, pt; int blocks_avail; int pos_pt_len_size; int pos_pt_len_bits; int literal_pt_len_size; int literal_pt_len_bits; int reading_position; int loop; int error; }; struct lzh_stream { const unsigned char *next_in; int avail_in; int64_t total_in; const unsigned char *ref_ptr; int avail_out; int64_t total_out; struct lzh_dec *ds; }; struct lha { /* entry_bytes_remaining is the number of bytes we expect. */ int64_t entry_offset; int64_t entry_bytes_remaining; int64_t entry_unconsumed; uint16_t entry_crc_calculated; size_t header_size; /* header size */ unsigned char level; /* header level */ char method[3]; /* compress type */ int64_t compsize; /* compressed data size */ int64_t origsize; /* original file size */ int setflag; #define BIRTHTIME_IS_SET 1 #define ATIME_IS_SET 2 #define UNIX_MODE_IS_SET 4 #define CRC_IS_SET 8 time_t birthtime; long birthtime_tv_nsec; time_t mtime; long mtime_tv_nsec; time_t atime; long atime_tv_nsec; mode_t mode; int64_t uid; int64_t gid; struct archive_string uname; struct archive_string gname; uint16_t header_crc; uint16_t crc; struct archive_string_conv *sconv; struct archive_string_conv *opt_sconv; struct archive_string dirname; struct archive_string filename; struct archive_wstring ws; unsigned char dos_attr; /* Flag to mark progress that an archive was read their first header.*/ char found_first_header; /* Flag to mark that indicates an empty directory. */ char directory; /* Flags to mark progress of decompression. */ char decompress_init; char end_of_entry; char end_of_entry_cleanup; char entry_is_compressed; char format_name[64]; struct lzh_stream strm; }; /* * LHA header common member offset. */ #define H_METHOD_OFFSET 2 /* Compress type. */ #define H_ATTR_OFFSET 19 /* DOS attribute. */ #define H_LEVEL_OFFSET 20 /* Header Level. */ #define H_SIZE 22 /* Minimum header size. */ static int archive_read_format_lha_bid(struct archive_read *, int); static int archive_read_format_lha_options(struct archive_read *, const char *, const char *); static int archive_read_format_lha_read_header(struct archive_read *, struct archive_entry *); static int archive_read_format_lha_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_lha_read_data_skip(struct archive_read *); static int archive_read_format_lha_cleanup(struct archive_read *); static void lha_replace_path_separator(struct lha *, struct archive_entry *); static int lha_read_file_header_0(struct archive_read *, struct lha *); static int lha_read_file_header_1(struct archive_read *, struct lha *); static int lha_read_file_header_2(struct archive_read *, struct lha *); static int lha_read_file_header_3(struct archive_read *, struct lha *); static int lha_read_file_extended_header(struct archive_read *, struct lha *, uint16_t *, int, size_t, size_t *); static size_t lha_check_header_format(const void *); static int lha_skip_sfx(struct archive_read *); static time_t lha_dos_time(const unsigned char *); static time_t lha_win_time(uint64_t, long *); static unsigned char lha_calcsum(unsigned char, const void *, int, size_t); static int lha_parse_linkname(struct archive_string *, struct archive_string *); static int lha_read_data_none(struct archive_read *, const void **, size_t *, int64_t *); static int lha_read_data_lzh(struct archive_read *, const void **, size_t *, int64_t *); static void lha_crc16_init(void); static uint16_t lha_crc16(uint16_t, const void *, size_t); static int lzh_decode_init(struct lzh_stream *, const char *); static void lzh_decode_free(struct lzh_stream *); static int lzh_decode(struct lzh_stream *, int); static int lzh_br_fillup(struct lzh_stream *, struct lzh_br *); static int lzh_huffman_init(struct huffman *, size_t, int); static void lzh_huffman_free(struct huffman *); static int lzh_read_pt_bitlen(struct lzh_stream *, int start, int end); static int lzh_make_fake_table(struct huffman *, uint16_t); static int lzh_make_huffman_table(struct huffman *); static inline int lzh_decode_huffman(struct huffman *, unsigned); static int lzh_decode_huffman_tree(struct huffman *, unsigned, int); int archive_read_support_format_lha(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct lha *lha; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_lha"); lha = (struct lha *)calloc(1, sizeof(*lha)); if (lha == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate lha data"); return (ARCHIVE_FATAL); } archive_string_init(&lha->ws); r = __archive_read_register_format(a, lha, "lha", archive_read_format_lha_bid, archive_read_format_lha_options, archive_read_format_lha_read_header, archive_read_format_lha_read_data, archive_read_format_lha_read_data_skip, NULL, archive_read_format_lha_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(lha); return (ARCHIVE_OK); } static size_t lha_check_header_format(const void *h) { const unsigned char *p = h; size_t next_skip_bytes; switch (p[H_METHOD_OFFSET+3]) { /* * "-lh0-" ... "-lh7-" "-lhd-" * "-lzs-" "-lz5-" */ case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case 'd': case 's': next_skip_bytes = 4; /* b0 == 0 means the end of an LHa archive file. */ if (p[0] == 0) break; if (p[H_METHOD_OFFSET] != '-' || p[H_METHOD_OFFSET+1] != 'l' || p[H_METHOD_OFFSET+4] != '-') break; if (p[H_METHOD_OFFSET+2] == 'h') { /* "-lh?-" */ if (p[H_METHOD_OFFSET+3] == 's') break; if (p[H_LEVEL_OFFSET] == 0) return (0); if (p[H_LEVEL_OFFSET] <= 3 && p[H_ATTR_OFFSET] == 0x20) return (0); } if (p[H_METHOD_OFFSET+2] == 'z') { /* LArc extensions: -lzs-,-lz4- and -lz5- */ if (p[H_LEVEL_OFFSET] != 0) break; if (p[H_METHOD_OFFSET+3] == 's' || p[H_METHOD_OFFSET+3] == '4' || p[H_METHOD_OFFSET+3] == '5') return (0); } break; case 'h': next_skip_bytes = 1; break; case 'z': next_skip_bytes = 1; break; case 'l': next_skip_bytes = 2; break; case '-': next_skip_bytes = 3; break; default : next_skip_bytes = 4; break; } return (next_skip_bytes); } static int archive_read_format_lha_bid(struct archive_read *a, int best_bid) { const char *p; const void *buff; ssize_t bytes_avail, offset, window; size_t next; /* If there's already a better bid than we can ever make, don't bother testing. */ if (best_bid > 30) return (-1); if ((p = __archive_read_ahead(a, H_SIZE, NULL)) == NULL) return (-1); if (lha_check_header_format(p) == 0) return (30); if (p[0] == 'M' && p[1] == 'Z') { /* PE file */ offset = 0; window = 4096; while (offset < (1024 * 20)) { buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < (H_SIZE + 3)) return (0); continue; } p = (const char *)buff + offset; while (p + H_SIZE < (const char *)buff + bytes_avail) { if ((next = lha_check_header_format(p)) == 0) return (30); p += next; } offset = p - (const char *)buff; } } return (0); } static int archive_read_format_lha_options(struct archive_read *a, const char *key, const char *val) { struct lha *lha; int ret = ARCHIVE_FAILED; lha = (struct lha *)(a->format->data); if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lha: hdrcharset option needs a character-set name"); else { lha->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (lha->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int lha_skip_sfx(struct archive_read *a) { const void *h; const char *p, *q; size_t next, skip; ssize_t bytes, window; window = 4096; for (;;) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < (H_SIZE + 3)) goto fatal; continue; } if (bytes < H_SIZE) goto fatal; p = h; q = p + bytes; /* * Scan ahead until we find something that looks * like the lha header. */ while (p + H_SIZE < q) { if ((next = lha_check_header_format(p)) == 0) { skip = p - (const char *)h; __archive_read_consume(a, skip); return (ARCHIVE_OK); } p += next; } skip = p - (const char *)h; __archive_read_consume(a, skip); } fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out LHa header"); return (ARCHIVE_FATAL); } static int truncated_error(struct archive_read *a) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated LHa header"); return (ARCHIVE_FATAL); } static int archive_read_format_lha_read_header(struct archive_read *a, struct archive_entry *entry) { struct archive_string linkname; struct archive_string pathname; struct lha *lha; const unsigned char *p; const char *signature; int err; lha_crc16_init(); a->archive.archive_format = ARCHIVE_FORMAT_LHA; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "lha"; lha = (struct lha *)(a->format->data); lha->decompress_init = 0; lha->end_of_entry = 0; lha->end_of_entry_cleanup = 0; lha->entry_unconsumed = 0; if ((p = __archive_read_ahead(a, H_SIZE, NULL)) == NULL) { /* * LHa archiver added 0 to the tail of its archive file as * the mark of the end of the archive. */ signature = __archive_read_ahead(a, sizeof(signature[0]), NULL); if (signature == NULL || signature[0] == 0) return (ARCHIVE_EOF); return (truncated_error(a)); } signature = (const char *)p; if (lha->found_first_header == 0 && signature[0] == 'M' && signature[1] == 'Z') { /* This is an executable? Must be self-extracting... */ err = lha_skip_sfx(a); if (err < ARCHIVE_WARN) return (err); if ((p = __archive_read_ahead(a, sizeof(*p), NULL)) == NULL) return (truncated_error(a)); signature = (const char *)p; } /* signature[0] == 0 means the end of an LHa archive file. */ if (signature[0] == 0) return (ARCHIVE_EOF); /* * Check the header format and method type. */ if (lha_check_header_format(p) != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad LHa file"); return (ARCHIVE_FATAL); } /* We've found the first header. */ lha->found_first_header = 1; /* Set a default value and common data */ lha->header_size = 0; lha->level = p[H_LEVEL_OFFSET]; lha->method[0] = p[H_METHOD_OFFSET+1]; lha->method[1] = p[H_METHOD_OFFSET+2]; lha->method[2] = p[H_METHOD_OFFSET+3]; if (memcmp(lha->method, "lhd", 3) == 0) lha->directory = 1; else lha->directory = 0; if (memcmp(lha->method, "lh0", 3) == 0 || memcmp(lha->method, "lz4", 3) == 0) lha->entry_is_compressed = 0; else lha->entry_is_compressed = 1; lha->compsize = 0; lha->origsize = 0; lha->setflag = 0; lha->birthtime = 0; lha->birthtime_tv_nsec = 0; lha->mtime = 0; lha->mtime_tv_nsec = 0; lha->atime = 0; lha->atime_tv_nsec = 0; lha->mode = (lha->directory)? 0777 : 0666; lha->uid = 0; lha->gid = 0; archive_string_empty(&lha->dirname); archive_string_empty(&lha->filename); lha->dos_attr = 0; if (lha->opt_sconv != NULL) lha->sconv = lha->opt_sconv; else lha->sconv = NULL; switch (p[H_LEVEL_OFFSET]) { case 0: err = lha_read_file_header_0(a, lha); break; case 1: err = lha_read_file_header_1(a, lha); break; case 2: err = lha_read_file_header_2(a, lha); break; case 3: err = lha_read_file_header_3(a, lha); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported LHa header level %d", p[H_LEVEL_OFFSET]); err = ARCHIVE_FATAL; break; } if (err < ARCHIVE_WARN) return (err); if (!lha->directory && archive_strlen(&lha->filename) == 0) /* The filename has not been set */ return (truncated_error(a)); /* * Make a pathname from a dirname and a filename. */ archive_string_concat(&lha->dirname, &lha->filename); archive_string_init(&pathname); archive_string_init(&linkname); archive_string_copy(&pathname, &lha->dirname); if ((lha->mode & AE_IFMT) == AE_IFLNK) { /* * Extract the symlink-name if it's included in the pathname. */ if (!lha_parse_linkname(&linkname, &pathname)) { /* We couldn't get the symlink-name. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown symlink-name"); archive_string_free(&pathname); archive_string_free(&linkname); return (ARCHIVE_FAILED); } } else { /* * Make sure a file-type is set. * The mode has been overridden if it is in the extended data. */ lha->mode = (lha->mode & ~AE_IFMT) | ((lha->directory)? AE_IFDIR: AE_IFREG); } if ((lha->setflag & UNIX_MODE_IS_SET) == 0 && (lha->dos_attr & 1) != 0) lha->mode &= ~(0222);/* read only. */ /* * Set basic file parameters. */ if (archive_entry_copy_pathname_l(entry, pathname.s, pathname.length, lha->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(lha->sconv)); err = ARCHIVE_WARN; } archive_string_free(&pathname); if (archive_strlen(&linkname) > 0) { if (archive_entry_copy_symlink_l(entry, linkname.s, linkname.length, lha->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Linkname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(lha->sconv)); err = ARCHIVE_WARN; } } else archive_entry_set_symlink(entry, NULL); archive_string_free(&linkname); /* * When a header level is 0, there is a possibility that * a pathname and a symlink has '\' character, a directory * separator in DOS/Windows. So we should convert it to '/'. */ if (p[H_LEVEL_OFFSET] == 0) lha_replace_path_separator(lha, entry); archive_entry_set_mode(entry, lha->mode); archive_entry_set_uid(entry, lha->uid); archive_entry_set_gid(entry, lha->gid); if (archive_strlen(&lha->uname) > 0) archive_entry_set_uname(entry, lha->uname.s); if (archive_strlen(&lha->gname) > 0) archive_entry_set_gname(entry, lha->gname.s); if (lha->setflag & BIRTHTIME_IS_SET) { archive_entry_set_birthtime(entry, lha->birthtime, lha->birthtime_tv_nsec); archive_entry_set_ctime(entry, lha->birthtime, lha->birthtime_tv_nsec); } else { archive_entry_unset_birthtime(entry); archive_entry_unset_ctime(entry); } archive_entry_set_mtime(entry, lha->mtime, lha->mtime_tv_nsec); if (lha->setflag & ATIME_IS_SET) archive_entry_set_atime(entry, lha->atime, lha->atime_tv_nsec); else archive_entry_unset_atime(entry); if (lha->directory || archive_entry_symlink(entry) != NULL) archive_entry_unset_size(entry); else archive_entry_set_size(entry, lha->origsize); /* * Prepare variables used to read a file content. */ lha->entry_bytes_remaining = lha->compsize; lha->entry_offset = 0; lha->entry_crc_calculated = 0; /* * This file does not have a content. */ if (lha->directory || lha->compsize == 0) lha->end_of_entry = 1; sprintf(lha->format_name, "lha -%c%c%c-", lha->method[0], lha->method[1], lha->method[2]); a->archive.archive_format_name = lha->format_name; return (err); } /* * Replace a DOS path separator '\' by a character '/'. * Some multi-byte character set have a character '\' in its second byte. */ static void lha_replace_path_separator(struct lha *lha, struct archive_entry *entry) { const wchar_t *wp; size_t i; if ((wp = archive_entry_pathname_w(entry)) != NULL) { archive_wstrcpy(&(lha->ws), wp); for (i = 0; i < archive_strlen(&(lha->ws)); i++) { if (lha->ws.s[i] == L'\\') lha->ws.s[i] = L'/'; } archive_entry_copy_pathname_w(entry, lha->ws.s); } if ((wp = archive_entry_symlink_w(entry)) != NULL) { archive_wstrcpy(&(lha->ws), wp); for (i = 0; i < archive_strlen(&(lha->ws)); i++) { if (lha->ws.s[i] == L'\\') lha->ws.s[i] = L'/'; } archive_entry_copy_symlink_w(entry, lha->ws.s); } } /* * Header 0 format * * +0 +1 +2 +7 +11 * +---------------+----------+----------------+-------------------+ * |header size(*1)|header sum|compression type|compressed size(*2)| * +---------------+----------+----------------+-------------------+ * <---------------------(*1)----------* * * +11 +15 +17 +19 +20 +21 * +-----------------+---------+---------+--------------+----------------+ * |uncompressed size|time(DOS)|date(DOS)|attribute(DOS)|header level(=0)| * +-----------------+---------+---------+--------------+----------------+ * *--------------------------------(*1)---------------------------------* * * +21 +22 +22+(*3) +22+(*3)+2 +22+(*3)+2+(*4) * +---------------+---------+----------+----------------+------------------+ * |name length(*3)|file name|file CRC16|extra header(*4)| compressed data | * +---------------+---------+----------+----------------+------------------+ * <--(*3)-> <------(*2)------> * *----------------------(*1)--------------------------> * */ #define H0_HEADER_SIZE_OFFSET 0 #define H0_HEADER_SUM_OFFSET 1 #define H0_COMP_SIZE_OFFSET 7 #define H0_ORIG_SIZE_OFFSET 11 #define H0_DOS_TIME_OFFSET 15 #define H0_NAME_LEN_OFFSET 21 #define H0_FILE_NAME_OFFSET 22 #define H0_FIXED_SIZE 24 static int lha_read_file_header_0(struct archive_read *a, struct lha *lha) { const unsigned char *p; int extdsize, namelen; unsigned char headersum, sum_calculated; if ((p = __archive_read_ahead(a, H0_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); lha->header_size = p[H0_HEADER_SIZE_OFFSET] + 2; headersum = p[H0_HEADER_SUM_OFFSET]; lha->compsize = archive_le32dec(p + H0_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H0_ORIG_SIZE_OFFSET); lha->mtime = lha_dos_time(p + H0_DOS_TIME_OFFSET); namelen = p[H0_NAME_LEN_OFFSET]; extdsize = (int)lha->header_size - H0_FIXED_SIZE - namelen; if ((namelen > 221 || extdsize < 0) && extdsize != -2) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header"); return (ARCHIVE_FATAL); } if ((p = __archive_read_ahead(a, lha->header_size, NULL)) == NULL) return (truncated_error(a)); archive_strncpy(&lha->filename, p + H0_FILE_NAME_OFFSET, namelen); /* When extdsize == -2, A CRC16 value is not present in the header. */ if (extdsize >= 0) { lha->crc = archive_le16dec(p + H0_FILE_NAME_OFFSET + namelen); lha->setflag |= CRC_IS_SET; } sum_calculated = lha_calcsum(0, p, 2, lha->header_size - 2); /* Read an extended header */ if (extdsize > 0) { /* This extended data is set by 'LHa for UNIX' only. * Maybe fixed size. */ p += H0_FILE_NAME_OFFSET + namelen + 2; if (p[0] == 'U' && extdsize == 12) { /* p[1] is a minor version. */ lha->mtime = archive_le32dec(&p[2]); lha->mode = archive_le16dec(&p[6]); lha->uid = archive_le16dec(&p[8]); lha->gid = archive_le16dec(&p[10]); lha->setflag |= UNIX_MODE_IS_SET; } } __archive_read_consume(a, lha->header_size); if (sum_calculated != headersum) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LHa header sum error"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } /* * Header 1 format * * +0 +1 +2 +7 +11 * +---------------+----------+----------------+-------------+ * |header size(*1)|header sum|compression type|skip size(*2)| * +---------------+----------+----------------+-------------+ * <---------------(*1)----------* * * +11 +15 +17 +19 +20 +21 * +-----------------+---------+---------+--------------+----------------+ * |uncompressed size|time(DOS)|date(DOS)|attribute(DOS)|header level(=1)| * +-----------------+---------+---------+--------------+----------------+ * *-------------------------------(*1)----------------------------------* * * +21 +22 +22+(*3) +22+(*3)+2 +22+(*3)+3 +22+(*3)+3+(*4) * +---------------+---------+----------+-----------+-----------+ * |name length(*3)|file name|file CRC16| creator |padding(*4)| * +---------------+---------+----------+-----------+-----------+ * <--(*3)-> * *----------------------------(*1)----------------------------* * * +22+(*3)+3+(*4) +22+(*3)+3+(*4)+2 +22+(*3)+3+(*4)+2+(*5) * +----------------+---------------------+------------------------+ * |next header size| extended header(*5) | compressed data | * +----------------+---------------------+------------------------+ * *------(*1)-----> <--------------------(*2)--------------------> */ #define H1_HEADER_SIZE_OFFSET 0 #define H1_HEADER_SUM_OFFSET 1 #define H1_COMP_SIZE_OFFSET 7 #define H1_ORIG_SIZE_OFFSET 11 #define H1_DOS_TIME_OFFSET 15 #define H1_NAME_LEN_OFFSET 21 #define H1_FILE_NAME_OFFSET 22 #define H1_FIXED_SIZE 27 static int lha_read_file_header_1(struct archive_read *a, struct lha *lha) { const unsigned char *p; size_t extdsize; int i, err, err2; int namelen, padding; unsigned char headersum, sum_calculated; err = ARCHIVE_OK; if ((p = __archive_read_ahead(a, H1_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); lha->header_size = p[H1_HEADER_SIZE_OFFSET] + 2; headersum = p[H1_HEADER_SUM_OFFSET]; /* Note: An extended header size is included in a compsize. */ lha->compsize = archive_le32dec(p + H1_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H1_ORIG_SIZE_OFFSET); lha->mtime = lha_dos_time(p + H1_DOS_TIME_OFFSET); namelen = p[H1_NAME_LEN_OFFSET]; /* Calculate a padding size. The result will be normally 0 only(?) */ padding = ((int)lha->header_size) - H1_FIXED_SIZE - namelen; if (namelen > 230 || padding < 0) goto invalid; if ((p = __archive_read_ahead(a, lha->header_size, NULL)) == NULL) return (truncated_error(a)); for (i = 0; i < namelen; i++) { if (p[i + H1_FILE_NAME_OFFSET] == 0xff) goto invalid;/* Invalid filename. */ } archive_strncpy(&lha->filename, p + H1_FILE_NAME_OFFSET, namelen); lha->crc = archive_le16dec(p + H1_FILE_NAME_OFFSET + namelen); lha->setflag |= CRC_IS_SET; sum_calculated = lha_calcsum(0, p, 2, lha->header_size - 2); /* Consume used bytes but not include `next header size' data * since it will be consumed in lha_read_file_extended_header(). */ __archive_read_consume(a, lha->header_size - 2); /* Read extended headers */ err2 = lha_read_file_extended_header(a, lha, NULL, 2, (size_t)(lha->compsize + 2), &extdsize); if (err2 < ARCHIVE_WARN) return (err2); if (err2 < err) err = err2; /* Get a real compressed file size. */ lha->compsize -= extdsize - 2; if (sum_calculated != headersum) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LHa header sum error"); return (ARCHIVE_FATAL); } return (err); invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header"); return (ARCHIVE_FATAL); } /* * Header 2 format * * +0 +2 +7 +11 +15 * +---------------+----------------+-------------------+-----------------+ * |header size(*1)|compression type|compressed size(*2)|uncompressed size| * +---------------+----------------+-------------------+-----------------+ * <--------------------------------(*1)---------------------------------* * * +15 +19 +20 +21 +23 +24 * +-----------------+------------+----------------+----------+-----------+ * |data/time(time_t)| 0x20 fixed |header level(=2)|file CRC16| creator | * +-----------------+------------+----------------+----------+-----------+ * *---------------------------------(*1)---------------------------------* * * +24 +26 +26+(*3) +26+(*3)+(*4) * +----------------+-------------------+-------------+-------------------+ * |next header size|extended header(*3)| padding(*4) | compressed data | * +----------------+-------------------+-------------+-------------------+ * *--------------------------(*1)-------------------> <------(*2)-------> * */ #define H2_HEADER_SIZE_OFFSET 0 #define H2_COMP_SIZE_OFFSET 7 #define H2_ORIG_SIZE_OFFSET 11 #define H2_TIME_OFFSET 15 #define H2_CRC_OFFSET 21 #define H2_FIXED_SIZE 24 static int lha_read_file_header_2(struct archive_read *a, struct lha *lha) { const unsigned char *p; size_t extdsize; int err, padding; uint16_t header_crc; if ((p = __archive_read_ahead(a, H2_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); lha->header_size =archive_le16dec(p + H2_HEADER_SIZE_OFFSET); lha->compsize = archive_le32dec(p + H2_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H2_ORIG_SIZE_OFFSET); lha->mtime = archive_le32dec(p + H2_TIME_OFFSET); lha->crc = archive_le16dec(p + H2_CRC_OFFSET); lha->setflag |= CRC_IS_SET; if (lha->header_size < H2_FIXED_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header size"); return (ARCHIVE_FATAL); } header_crc = lha_crc16(0, p, H2_FIXED_SIZE); __archive_read_consume(a, H2_FIXED_SIZE); /* Read extended headers */ err = lha_read_file_extended_header(a, lha, &header_crc, 2, lha->header_size - H2_FIXED_SIZE, &extdsize); if (err < ARCHIVE_WARN) return (err); /* Calculate a padding size. The result will be normally 0 or 1. */ padding = (int)lha->header_size - (int)(H2_FIXED_SIZE + extdsize); if (padding > 0) { if ((p = __archive_read_ahead(a, padding, NULL)) == NULL) return (truncated_error(a)); header_crc = lha_crc16(header_crc, p, padding); __archive_read_consume(a, padding); } if (header_crc != lha->header_crc) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "LHa header CRC error"); return (ARCHIVE_FATAL); } return (err); } /* * Header 3 format * * +0 +2 +7 +11 +15 * +------------+----------------+-------------------+-----------------+ * | 0x04 fixed |compression type|compressed size(*2)|uncompressed size| * +------------+----------------+-------------------+-----------------+ * <-------------------------------(*1)-------------------------------* * * +15 +19 +20 +21 +23 +24 * +-----------------+------------+----------------+----------+-----------+ * |date/time(time_t)| 0x20 fixed |header level(=3)|file CRC16| creator | * +-----------------+------------+----------------+----------+-----------+ * *--------------------------------(*1)----------------------------------* * * +24 +28 +32 +32+(*3) * +---------------+----------------+-------------------+-----------------+ * |header size(*1)|next header size|extended header(*3)| compressed data | * +---------------+----------------+-------------------+-----------------+ * *------------------------(*1)-----------------------> <------(*2)-----> * */ #define H3_FIELD_LEN_OFFSET 0 #define H3_COMP_SIZE_OFFSET 7 #define H3_ORIG_SIZE_OFFSET 11 #define H3_TIME_OFFSET 15 #define H3_CRC_OFFSET 21 #define H3_HEADER_SIZE_OFFSET 24 #define H3_FIXED_SIZE 28 static int lha_read_file_header_3(struct archive_read *a, struct lha *lha) { const unsigned char *p; size_t extdsize; int err; uint16_t header_crc; if ((p = __archive_read_ahead(a, H3_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); if (archive_le16dec(p + H3_FIELD_LEN_OFFSET) != 4) goto invalid; lha->header_size =archive_le32dec(p + H3_HEADER_SIZE_OFFSET); lha->compsize = archive_le32dec(p + H3_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H3_ORIG_SIZE_OFFSET); lha->mtime = archive_le32dec(p + H3_TIME_OFFSET); lha->crc = archive_le16dec(p + H3_CRC_OFFSET); lha->setflag |= CRC_IS_SET; if (lha->header_size < H3_FIXED_SIZE + 4) goto invalid; header_crc = lha_crc16(0, p, H3_FIXED_SIZE); __archive_read_consume(a, H3_FIXED_SIZE); /* Read extended headers */ err = lha_read_file_extended_header(a, lha, &header_crc, 4, lha->header_size - H3_FIXED_SIZE, &extdsize); if (err < ARCHIVE_WARN) return (err); if (header_crc != lha->header_crc) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "LHa header CRC error"); return (ARCHIVE_FATAL); } return (err); invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header"); return (ARCHIVE_FATAL); } /* * Extended header format * * +0 +2 +3 -- used in header 1 and 2 * +0 +4 +5 -- used in header 3 * +--------------+---------+-------------------+--------------+-- * |ex-header size|header id| data |ex-header size| ....... * +--------------+---------+-------------------+--------------+-- * <-------------( ex-header size)------------> <-- next extended header --* * * If the ex-header size is zero, it is the make of the end of extended * headers. * */ static int lha_read_file_extended_header(struct archive_read *a, struct lha *lha, uint16_t *crc, int sizefield_length, size_t limitsize, size_t *total_size) { const void *h; const unsigned char *extdheader; size_t extdsize; size_t datasize; unsigned int i; unsigned char extdtype; #define EXT_HEADER_CRC 0x00 /* Header CRC and information*/ #define EXT_FILENAME 0x01 /* Filename */ #define EXT_DIRECTORY 0x02 /* Directory name */ #define EXT_DOS_ATTR 0x40 /* MS-DOS attribute */ #define EXT_TIMESTAMP 0x41 /* Windows time stamp */ #define EXT_FILESIZE 0x42 /* Large file size */ #define EXT_TIMEZONE 0x43 /* Time zone */ #define EXT_UTF16_FILENAME 0x44 /* UTF-16 filename */ #define EXT_UTF16_DIRECTORY 0x45 /* UTF-16 directory name */ #define EXT_CODEPAGE 0x46 /* Codepage */ #define EXT_UNIX_MODE 0x50 /* File permission */ #define EXT_UNIX_GID_UID 0x51 /* gid,uid */ #define EXT_UNIX_GNAME 0x52 /* Group name */ #define EXT_UNIX_UNAME 0x53 /* User name */ #define EXT_UNIX_MTIME 0x54 /* Modified time */ #define EXT_OS2_NEW_ATTR 0x7f /* new attribute(OS/2 only) */ #define EXT_NEW_ATTR 0xff /* new attribute */ *total_size = sizefield_length; for (;;) { /* Read an extended header size. */ if ((h = __archive_read_ahead(a, sizefield_length, NULL)) == NULL) return (truncated_error(a)); /* Check if the size is the zero indicates the end of the * extended header. */ if (sizefield_length == sizeof(uint16_t)) extdsize = archive_le16dec(h); else extdsize = archive_le32dec(h); if (extdsize == 0) { /* End of extended header */ if (crc != NULL) *crc = lha_crc16(*crc, h, sizefield_length); __archive_read_consume(a, sizefield_length); return (ARCHIVE_OK); } /* Sanity check to the extended header size. */ if (((uint64_t)*total_size + extdsize) > (uint64_t)limitsize || extdsize <= (size_t)sizefield_length) goto invalid; /* Read the extended header. */ if ((h = __archive_read_ahead(a, extdsize, NULL)) == NULL) return (truncated_error(a)); *total_size += extdsize; extdheader = (const unsigned char *)h; /* Get the extended header type. */ extdtype = extdheader[sizefield_length]; /* Calculate an extended data size. */ datasize = extdsize - (1 + sizefield_length); /* Skip an extended header size field and type field. */ extdheader += sizefield_length + 1; if (crc != NULL && extdtype != EXT_HEADER_CRC) *crc = lha_crc16(*crc, h, extdsize); switch (extdtype) { case EXT_HEADER_CRC: /* We only use a header CRC. Following data will not * be used. */ if (datasize >= 2) { lha->header_crc = archive_le16dec(extdheader); if (crc != NULL) { static const char zeros[2] = {0, 0}; *crc = lha_crc16(*crc, h, extdsize - datasize); /* CRC value itself as zero */ *crc = lha_crc16(*crc, zeros, 2); *crc = lha_crc16(*crc, extdheader+2, datasize - 2); } } break; case EXT_FILENAME: if (datasize == 0) { /* maybe directory header */ archive_string_empty(&lha->filename); break; } if (extdheader[0] == '\0') goto invalid; archive_strncpy(&lha->filename, (const char *)extdheader, datasize); break; case EXT_DIRECTORY: if (datasize == 0 || extdheader[0] == '\0') /* no directory name data. exit this case. */ goto invalid; archive_strncpy(&lha->dirname, (const char *)extdheader, datasize); /* * Convert directory delimiter from 0xFF * to '/' for local system. */ for (i = 0; i < lha->dirname.length; i++) { if ((unsigned char)lha->dirname.s[i] == 0xFF) lha->dirname.s[i] = '/'; } /* Is last character directory separator? */ if (lha->dirname.s[lha->dirname.length-1] != '/') /* invalid directory data */ goto invalid; break; case EXT_DOS_ATTR: if (datasize == 2) lha->dos_attr = (unsigned char) (archive_le16dec(extdheader) & 0xff); break; case EXT_TIMESTAMP: if (datasize == (sizeof(uint64_t) * 3)) { lha->birthtime = lha_win_time( archive_le64dec(extdheader), &lha->birthtime_tv_nsec); extdheader += sizeof(uint64_t); lha->mtime = lha_win_time( archive_le64dec(extdheader), &lha->mtime_tv_nsec); extdheader += sizeof(uint64_t); lha->atime = lha_win_time( archive_le64dec(extdheader), &lha->atime_tv_nsec); lha->setflag |= BIRTHTIME_IS_SET | ATIME_IS_SET; } break; case EXT_FILESIZE: if (datasize == sizeof(uint64_t) * 2) { lha->compsize = archive_le64dec(extdheader); extdheader += sizeof(uint64_t); lha->origsize = archive_le64dec(extdheader); } break; case EXT_CODEPAGE: /* Get an archived filename charset from codepage. * This overwrites the charset specified by * hdrcharset option. */ if (datasize == sizeof(uint32_t)) { struct archive_string cp; const char *charset; archive_string_init(&cp); switch (archive_le32dec(extdheader)) { case 65001: /* UTF-8 */ charset = "UTF-8"; break; default: archive_string_sprintf(&cp, "CP%d", (int)archive_le32dec(extdheader)); charset = cp.s; break; } lha->sconv = archive_string_conversion_from_charset( &(a->archive), charset, 1); archive_string_free(&cp); if (lha->sconv == NULL) return (ARCHIVE_FATAL); } break; case EXT_UNIX_MODE: if (datasize == sizeof(uint16_t)) { lha->mode = archive_le16dec(extdheader); lha->setflag |= UNIX_MODE_IS_SET; } break; case EXT_UNIX_GID_UID: if (datasize == (sizeof(uint16_t) * 2)) { lha->gid = archive_le16dec(extdheader); lha->uid = archive_le16dec(extdheader+2); } break; case EXT_UNIX_GNAME: if (datasize > 0) archive_strncpy(&lha->gname, (const char *)extdheader, datasize); break; case EXT_UNIX_UNAME: if (datasize > 0) archive_strncpy(&lha->uname, (const char *)extdheader, datasize); break; case EXT_UNIX_MTIME: if (datasize == sizeof(uint32_t)) lha->mtime = archive_le32dec(extdheader); break; case EXT_OS2_NEW_ATTR: /* This extended header is OS/2 depend. */ if (datasize == 16) { lha->dos_attr = (unsigned char) (archive_le16dec(extdheader) & 0xff); lha->mode = archive_le16dec(extdheader+2); lha->gid = archive_le16dec(extdheader+4); lha->uid = archive_le16dec(extdheader+6); lha->birthtime = archive_le32dec(extdheader+8); lha->atime = archive_le32dec(extdheader+12); lha->setflag |= UNIX_MODE_IS_SET | BIRTHTIME_IS_SET | ATIME_IS_SET; } break; case EXT_NEW_ATTR: if (datasize == 20) { lha->mode = (mode_t)archive_le32dec(extdheader); lha->gid = archive_le32dec(extdheader+4); lha->uid = archive_le32dec(extdheader+8); lha->birthtime = archive_le32dec(extdheader+12); lha->atime = archive_le32dec(extdheader+16); lha->setflag |= UNIX_MODE_IS_SET | BIRTHTIME_IS_SET | ATIME_IS_SET; } break; case EXT_TIMEZONE: /* Not supported */ case EXT_UTF16_FILENAME: /* Not supported */ case EXT_UTF16_DIRECTORY: /* Not supported */ default: break; } __archive_read_consume(a, extdsize); } invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid extended LHa header"); return (ARCHIVE_FATAL); } static int lha_end_of_entry(struct archive_read *a) { struct lha *lha = (struct lha *)(a->format->data); int r = ARCHIVE_EOF; if (!lha->end_of_entry_cleanup) { if ((lha->setflag & CRC_IS_SET) && lha->crc != lha->entry_crc_calculated) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LHa data CRC error"); r = ARCHIVE_WARN; } /* End-of-entry cleanup done. */ lha->end_of_entry_cleanup = 1; } return (r); } static int archive_read_format_lha_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct lha *lha = (struct lha *)(a->format->data); int r; if (lha->entry_unconsumed) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, lha->entry_unconsumed); lha->entry_unconsumed = 0; } if (lha->end_of_entry) { *offset = lha->entry_offset; *size = 0; *buff = NULL; return (lha_end_of_entry(a)); } if (lha->entry_is_compressed) r = lha_read_data_lzh(a, buff, size, offset); else /* No compression. */ r = lha_read_data_none(a, buff, size, offset); return (r); } /* * Read a file content in no compression. * * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets * lha->end_of_entry if it consumes all of the data. */ static int lha_read_data_none(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct lha *lha = (struct lha *)(a->format->data); ssize_t bytes_avail; if (lha->entry_bytes_remaining == 0) { *buff = NULL; *size = 0; *offset = lha->entry_offset; lha->end_of_entry = 1; return (ARCHIVE_OK); } /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ *buff = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated LHa file data"); return (ARCHIVE_FATAL); } if (bytes_avail > lha->entry_bytes_remaining) bytes_avail = (ssize_t)lha->entry_bytes_remaining; lha->entry_crc_calculated = lha_crc16(lha->entry_crc_calculated, *buff, bytes_avail); *size = bytes_avail; *offset = lha->entry_offset; lha->entry_offset += bytes_avail; lha->entry_bytes_remaining -= bytes_avail; if (lha->entry_bytes_remaining == 0) lha->end_of_entry = 1; lha->entry_unconsumed = bytes_avail; return (ARCHIVE_OK); } /* * Read a file content in LZHUFF encoding. * * Returns ARCHIVE_OK if successful, returns ARCHIVE_WARN if compression is * unsupported, ARCHIVE_FATAL otherwise, sets lha->end_of_entry if it consumes * all of the data. */ static int lha_read_data_lzh(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct lha *lha = (struct lha *)(a->format->data); ssize_t bytes_avail; int r; /* If we haven't yet read any data, initialize the decompressor. */ if (!lha->decompress_init) { r = lzh_decode_init(&(lha->strm), lha->method); switch (r) { case ARCHIVE_OK: break; case ARCHIVE_FAILED: /* Unsupported compression. */ *buff = NULL; *size = 0; *offset = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported lzh compression method -%c%c%c-", lha->method[0], lha->method[1], lha->method[2]); /* We know compressed size; just skip it. */ archive_read_format_lha_read_data_skip(a); return (ARCHIVE_WARN); default: archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory " "for lzh decompression"); return (ARCHIVE_FATAL); } /* We've initialized decompression for this stream. */ lha->decompress_init = 1; lha->strm.avail_out = 0; lha->strm.total_out = 0; } /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ lha->strm.next_in = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated LHa file body"); return (ARCHIVE_FATAL); } if (bytes_avail > lha->entry_bytes_remaining) bytes_avail = (ssize_t)lha->entry_bytes_remaining; lha->strm.avail_in = (int)bytes_avail; lha->strm.total_in = 0; lha->strm.avail_out = 0; r = lzh_decode(&(lha->strm), bytes_avail == lha->entry_bytes_remaining); switch (r) { case ARCHIVE_OK: break; case ARCHIVE_EOF: lha->end_of_entry = 1; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Bad lzh data"); return (ARCHIVE_FAILED); } lha->entry_unconsumed = lha->strm.total_in; lha->entry_bytes_remaining -= lha->strm.total_in; if (lha->strm.avail_out) { *offset = lha->entry_offset; *size = lha->strm.avail_out; *buff = lha->strm.ref_ptr; lha->entry_crc_calculated = lha_crc16(lha->entry_crc_calculated, *buff, *size); lha->entry_offset += *size; } else { *offset = lha->entry_offset; *size = 0; *buff = NULL; if (lha->end_of_entry) return (lha_end_of_entry(a)); } return (ARCHIVE_OK); } /* * Skip a file content. */ static int archive_read_format_lha_read_data_skip(struct archive_read *a) { struct lha *lha; int64_t bytes_skipped; lha = (struct lha *)(a->format->data); if (lha->entry_unconsumed) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, lha->entry_unconsumed); lha->entry_unconsumed = 0; } /* if we've already read to end of data, we're done. */ if (lha->end_of_entry_cleanup) return (ARCHIVE_OK); /* * If the length is at the beginning, we can skip the * compressed data much more quickly. */ bytes_skipped = __archive_read_consume(a, lha->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); /* This entry is finished and done. */ lha->end_of_entry_cleanup = lha->end_of_entry = 1; return (ARCHIVE_OK); } static int archive_read_format_lha_cleanup(struct archive_read *a) { struct lha *lha = (struct lha *)(a->format->data); lzh_decode_free(&(lha->strm)); archive_string_free(&(lha->dirname)); archive_string_free(&(lha->filename)); archive_string_free(&(lha->uname)); archive_string_free(&(lha->gname)); archive_wstring_free(&(lha->ws)); free(lha); (a->format->data) = NULL; return (ARCHIVE_OK); } /* * 'LHa for UNIX' utility has archived a symbolic-link name after * a pathname with '|' character. * This function extracts the symbolic-link name from the pathname. * * example. * 1. a symbolic-name is 'aaa/bb/cc' * 2. a filename is 'xxx/bbb' * then a archived pathname is 'xxx/bbb|aaa/bb/cc' */ static int lha_parse_linkname(struct archive_string *linkname, struct archive_string *pathname) { char * linkptr; size_t symlen; linkptr = strchr(pathname->s, '|'); if (linkptr != NULL) { symlen = strlen(linkptr + 1); archive_strncpy(linkname, linkptr+1, symlen); *linkptr = 0; pathname->length = strlen(pathname->s); return (1); } return (0); } /* Convert an MSDOS-style date/time into Unix-style time. */ static time_t lha_dos_time(const unsigned char *p) { int msTime, msDate; struct tm ts; msTime = archive_le16dec(p); msDate = archive_le16dec(p+2); memset(&ts, 0, sizeof(ts)); ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ ts.tm_mday = msDate & 0x1f; /* Day of month. */ ts.tm_hour = (msTime >> 11) & 0x1f; ts.tm_min = (msTime >> 5) & 0x3f; ts.tm_sec = (msTime << 1) & 0x3e; ts.tm_isdst = -1; return (mktime(&ts)); } /* Convert an MS-Windows-style date/time into Unix-style time. */ static time_t lha_win_time(uint64_t wintime, long *ns) { #define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) if (wintime >= EPOC_TIME) { wintime -= EPOC_TIME; /* 1970-01-01 00:00:00 (UTC) */ if (ns != NULL) *ns = (long)(wintime % 10000000) * 100; return (wintime / 10000000); } else { if (ns != NULL) *ns = 0; return (0); } } static unsigned char lha_calcsum(unsigned char sum, const void *pp, int offset, size_t size) { unsigned char const *p = (unsigned char const *)pp; p += offset; for (;size > 0; --size) sum += *p++; return (sum); } static uint16_t crc16tbl[2][256]; static void lha_crc16_init(void) { unsigned int i; static int crc16init = 0; if (crc16init) return; crc16init = 1; for (i = 0; i < 256; i++) { unsigned int j; uint16_t crc = (uint16_t)i; for (j = 8; j; j--) crc = (crc >> 1) ^ ((crc & 1) * 0xA001); crc16tbl[0][i] = crc; } for (i = 0; i < 256; i++) { crc16tbl[1][i] = (crc16tbl[0][i] >> 8) ^ crc16tbl[0][crc16tbl[0][i] & 0xff]; } } static uint16_t lha_crc16(uint16_t crc, const void *pp, size_t len) { const unsigned char *p = (const unsigned char *)pp; const uint16_t *buff; const union { uint32_t i; char c[4]; } u = { 0x01020304 }; if (len == 0) return crc; /* Process unaligned address. */ if (((uintptr_t)p) & (uintptr_t)0x1) { crc = (crc >> 8) ^ crc16tbl[0][(crc ^ *p++) & 0xff]; len--; } buff = (const uint16_t *)p; /* * Modern C compiler such as GCC does not unroll automatically yet * without unrolling pragma, and Clang is so. So we should * unroll this loop for its performance. */ for (;len >= 8; len -= 8) { /* This if statement expects compiler optimization will * remove the stament which will not be executed. */ +#undef bswap16 #if defined(_MSC_VER) && _MSC_VER >= 1400 /* Visual Studio */ # define bswap16(x) _byteswap_ushort(x) #elif (defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 8) \ || defined(__clang__) # define bswap16(x) __builtin_bswap16(x) #else # define bswap16(x) ((((x) >> 8) & 0xff) | ((x) << 8)) #endif #define CRC16W do { \ if(u.c[0] == 1) { /* Big endian */ \ crc ^= bswap16(*buff); buff++; \ } else \ crc ^= *buff++; \ crc = crc16tbl[1][crc & 0xff] ^ crc16tbl[0][crc >> 8];\ } while (0) CRC16W; CRC16W; CRC16W; CRC16W; #undef CRC16W #undef bswap16 } p = (const unsigned char *)buff; for (;len; len--) { crc = (crc >> 8) ^ crc16tbl[0][(crc ^ *p++) & 0xff]; } return crc; } /* * Initialize LZHUF decoder. * * Returns ARCHIVE_OK if initialization was successful. * Returns ARCHIVE_FAILED if method is unsupported. * Returns ARCHIVE_FATAL if initialization failed; memory allocation * error occurred. */ static int lzh_decode_init(struct lzh_stream *strm, const char *method) { struct lzh_dec *ds; int w_bits, w_size; if (strm->ds == NULL) { strm->ds = calloc(1, sizeof(*strm->ds)); if (strm->ds == NULL) return (ARCHIVE_FATAL); } ds = strm->ds; ds->error = ARCHIVE_FAILED; if (method == NULL || method[0] != 'l' || method[1] != 'h') return (ARCHIVE_FAILED); switch (method[2]) { case '5': w_bits = 13;/* 8KiB for window */ break; case '6': w_bits = 15;/* 32KiB for window */ break; case '7': w_bits = 16;/* 64KiB for window */ break; default: return (ARCHIVE_FAILED);/* Not supported. */ } ds->error = ARCHIVE_FATAL; /* Expand a window size up to 128 KiB for decompressing process * performance whatever its original window size is. */ ds->w_size = 1U << 17; ds->w_mask = ds->w_size -1; if (ds->w_buff == NULL) { ds->w_buff = malloc(ds->w_size); if (ds->w_buff == NULL) return (ARCHIVE_FATAL); } w_size = 1U << w_bits; memset(ds->w_buff + ds->w_size - w_size, 0x20, w_size); ds->w_pos = 0; ds->state = 0; ds->pos_pt_len_size = w_bits + 1; ds->pos_pt_len_bits = (w_bits == 15 || w_bits == 16)? 5: 4; ds->literal_pt_len_size = PT_BITLEN_SIZE; ds->literal_pt_len_bits = 5; ds->br.cache_buffer = 0; ds->br.cache_avail = 0; if (lzh_huffman_init(&(ds->lt), LT_BITLEN_SIZE, 16) != ARCHIVE_OK) return (ARCHIVE_FATAL); ds->lt.len_bits = 9; if (lzh_huffman_init(&(ds->pt), PT_BITLEN_SIZE, 16) != ARCHIVE_OK) return (ARCHIVE_FATAL); ds->error = 0; return (ARCHIVE_OK); } /* * Release LZHUF decoder. */ static void lzh_decode_free(struct lzh_stream *strm) { if (strm->ds == NULL) return; free(strm->ds->w_buff); lzh_huffman_free(&(strm->ds->lt)); lzh_huffman_free(&(strm->ds->pt)); free(strm->ds); strm->ds = NULL; } /* * Bit stream reader. */ /* Check that the cache buffer has enough bits. */ #define lzh_br_has(br, n) ((br)->cache_avail >= n) /* Get compressed data by bit. */ #define lzh_br_bits(br, n) \ (((uint16_t)((br)->cache_buffer >> \ ((br)->cache_avail - (n)))) & cache_masks[n]) #define lzh_br_bits_forced(br, n) \ (((uint16_t)((br)->cache_buffer << \ ((n) - (br)->cache_avail))) & cache_masks[n]) /* Read ahead to make sure the cache buffer has enough compressed data we * will use. * True : completed, there is enough data in the cache buffer. * False : we met that strm->next_in is empty, we have to get following * bytes. */ #define lzh_br_read_ahead_0(strm, br, n) \ (lzh_br_has(br, (n)) || lzh_br_fillup(strm, br)) /* True : the cache buffer has some bits as much as we need. * False : there are no enough bits in the cache buffer to be used, * we have to get following bytes if we could. */ #define lzh_br_read_ahead(strm, br, n) \ (lzh_br_read_ahead_0((strm), (br), (n)) || lzh_br_has((br), (n))) /* Notify how many bits we consumed. */ #define lzh_br_consume(br, n) ((br)->cache_avail -= (n)) #define lzh_br_unconsume(br, n) ((br)->cache_avail += (n)) static const uint16_t cache_masks[] = { 0x0000, 0x0001, 0x0003, 0x0007, 0x000F, 0x001F, 0x003F, 0x007F, 0x00FF, 0x01FF, 0x03FF, 0x07FF, 0x0FFF, 0x1FFF, 0x3FFF, 0x7FFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF }; /* * Shift away used bits in the cache data and fill it up with following bits. * Call this when cache buffer does not have enough bits you need. * * Returns 1 if the cache buffer is full. * Returns 0 if the cache buffer is not full; input buffer is empty. */ static int lzh_br_fillup(struct lzh_stream *strm, struct lzh_br *br) { int n = CACHE_BITS - br->cache_avail; for (;;) { const int x = n >> 3; if (strm->avail_in >= x) { switch (x) { case 8: br->cache_buffer = ((uint64_t)strm->next_in[0]) << 56 | ((uint64_t)strm->next_in[1]) << 48 | ((uint64_t)strm->next_in[2]) << 40 | ((uint64_t)strm->next_in[3]) << 32 | ((uint32_t)strm->next_in[4]) << 24 | ((uint32_t)strm->next_in[5]) << 16 | ((uint32_t)strm->next_in[6]) << 8 | (uint32_t)strm->next_in[7]; strm->next_in += 8; strm->avail_in -= 8; br->cache_avail += 8 * 8; return (1); case 7: br->cache_buffer = (br->cache_buffer << 56) | ((uint64_t)strm->next_in[0]) << 48 | ((uint64_t)strm->next_in[1]) << 40 | ((uint64_t)strm->next_in[2]) << 32 | ((uint32_t)strm->next_in[3]) << 24 | ((uint32_t)strm->next_in[4]) << 16 | ((uint32_t)strm->next_in[5]) << 8 | (uint32_t)strm->next_in[6]; strm->next_in += 7; strm->avail_in -= 7; br->cache_avail += 7 * 8; return (1); case 6: br->cache_buffer = (br->cache_buffer << 48) | ((uint64_t)strm->next_in[0]) << 40 | ((uint64_t)strm->next_in[1]) << 32 | ((uint32_t)strm->next_in[2]) << 24 | ((uint32_t)strm->next_in[3]) << 16 | ((uint32_t)strm->next_in[4]) << 8 | (uint32_t)strm->next_in[5]; strm->next_in += 6; strm->avail_in -= 6; br->cache_avail += 6 * 8; return (1); case 0: /* We have enough compressed data in * the cache buffer.*/ return (1); default: break; } } if (strm->avail_in == 0) { /* There is not enough compressed data to fill up the * cache buffer. */ return (0); } br->cache_buffer = (br->cache_buffer << 8) | *strm->next_in++; strm->avail_in--; br->cache_avail += 8; n -= 8; } } /* * Decode LZHUF. * * 1. Returns ARCHIVE_OK if output buffer or input buffer are empty. * Please set available buffer and call this function again. * 2. Returns ARCHIVE_EOF if decompression has been completed. * 3. Returns ARCHIVE_FAILED if an error occurred; compressed data * is broken or you do not set 'last' flag properly. * 4. 'last' flag is very important, you must set 1 to the flag if there * is no input data. The lha compressed data format does not provide how * to know the compressed data is really finished. * Note: lha command utility check if the total size of output bytes is * reached the uncompressed size recorded in its header. it does not mind * that the decoding process is properly finished. * GNU ZIP can decompress another compressed file made by SCO LZH compress. * it handles EOF as null to fill read buffer with zero until the decoding * process meet 2 bytes of zeros at reading a size of a next chunk, so the * zeros are treated as the mark of the end of the data although the zeros * is dummy, not the file data. */ static int lzh_read_blocks(struct lzh_stream *, int); static int lzh_decode_blocks(struct lzh_stream *, int); #define ST_RD_BLOCK 0 #define ST_RD_PT_1 1 #define ST_RD_PT_2 2 #define ST_RD_PT_3 3 #define ST_RD_PT_4 4 #define ST_RD_LITERAL_1 5 #define ST_RD_LITERAL_2 6 #define ST_RD_LITERAL_3 7 #define ST_RD_POS_DATA_1 8 #define ST_GET_LITERAL 9 #define ST_GET_POS_1 10 #define ST_GET_POS_2 11 #define ST_COPY_DATA 12 static int lzh_decode(struct lzh_stream *strm, int last) { struct lzh_dec *ds = strm->ds; int avail_in; int r; if (ds->error) return (ds->error); avail_in = strm->avail_in; do { if (ds->state < ST_GET_LITERAL) r = lzh_read_blocks(strm, last); else r = lzh_decode_blocks(strm, last); } while (r == 100); strm->total_in += avail_in - strm->avail_in; return (r); } static void lzh_emit_window(struct lzh_stream *strm, size_t s) { strm->ref_ptr = strm->ds->w_buff; strm->avail_out = (int)s; strm->total_out += s; } static int lzh_read_blocks(struct lzh_stream *strm, int last) { struct lzh_dec *ds = strm->ds; struct lzh_br *br = &(ds->br); int c = 0, i; unsigned rbits; for (;;) { switch (ds->state) { case ST_RD_BLOCK: /* * Read a block number indicates how many blocks * we will handle. The block is composed of a * literal and a match, sometimes a literal only * in particular, there are no reference data at * the beginning of the decompression. */ if (!lzh_br_read_ahead_0(strm, br, 16)) { if (!last) /* We need following data. */ return (ARCHIVE_OK); if (lzh_br_has(br, 8)) { /* * It seems there are extra bits. * 1. Compressed data is broken. * 2. `last' flag does not properly * set. */ goto failed; } if (ds->w_pos > 0) { lzh_emit_window(strm, ds->w_pos); ds->w_pos = 0; return (ARCHIVE_OK); } /* End of compressed data; we have completely * handled all compressed data. */ return (ARCHIVE_EOF); } ds->blocks_avail = lzh_br_bits(br, 16); if (ds->blocks_avail == 0) goto failed; lzh_br_consume(br, 16); /* * Read a literal table compressed in huffman * coding. */ ds->pt.len_size = ds->literal_pt_len_size; ds->pt.len_bits = ds->literal_pt_len_bits; ds->reading_position = 0; /* FALL THROUGH */ case ST_RD_PT_1: /* Note: ST_RD_PT_1, ST_RD_PT_2 and ST_RD_PT_4 are * used in reading both a literal table and a * position table. */ if (!lzh_br_read_ahead(strm, br, ds->pt.len_bits)) { if (last) goto failed;/* Truncated data. */ ds->state = ST_RD_PT_1; return (ARCHIVE_OK); } ds->pt.len_avail = lzh_br_bits(br, ds->pt.len_bits); lzh_br_consume(br, ds->pt.len_bits); /* FALL THROUGH */ case ST_RD_PT_2: if (ds->pt.len_avail == 0) { /* There is no bitlen. */ if (!lzh_br_read_ahead(strm, br, ds->pt.len_bits)) { if (last) goto failed;/* Truncated data.*/ ds->state = ST_RD_PT_2; return (ARCHIVE_OK); } if (!lzh_make_fake_table(&(ds->pt), lzh_br_bits(br, ds->pt.len_bits))) goto failed;/* Invalid data. */ lzh_br_consume(br, ds->pt.len_bits); if (ds->reading_position) ds->state = ST_GET_LITERAL; else ds->state = ST_RD_LITERAL_1; break; } else if (ds->pt.len_avail > ds->pt.len_size) goto failed;/* Invalid data. */ ds->loop = 0; memset(ds->pt.freq, 0, sizeof(ds->pt.freq)); if (ds->pt.len_avail < 3 || ds->pt.len_size == ds->pos_pt_len_size) { ds->state = ST_RD_PT_4; break; } /* FALL THROUGH */ case ST_RD_PT_3: ds->loop = lzh_read_pt_bitlen(strm, ds->loop, 3); if (ds->loop < 3) { if (ds->loop < 0 || last) goto failed;/* Invalid data. */ /* Not completed, get following data. */ ds->state = ST_RD_PT_3; return (ARCHIVE_OK); } /* There are some null in bitlen of the literal. */ if (!lzh_br_read_ahead(strm, br, 2)) { if (last) goto failed;/* Truncated data. */ ds->state = ST_RD_PT_3; return (ARCHIVE_OK); } c = lzh_br_bits(br, 2); lzh_br_consume(br, 2); if (c > ds->pt.len_avail - 3) goto failed;/* Invalid data. */ for (i = 3; c-- > 0 ;) ds->pt.bitlen[i++] = 0; ds->loop = i; /* FALL THROUGH */ case ST_RD_PT_4: ds->loop = lzh_read_pt_bitlen(strm, ds->loop, ds->pt.len_avail); if (ds->loop < ds->pt.len_avail) { if (ds->loop < 0 || last) goto failed;/* Invalid data. */ /* Not completed, get following data. */ ds->state = ST_RD_PT_4; return (ARCHIVE_OK); } if (!lzh_make_huffman_table(&(ds->pt))) goto failed;/* Invalid data */ if (ds->reading_position) { ds->state = ST_GET_LITERAL; break; } /* FALL THROUGH */ case ST_RD_LITERAL_1: if (!lzh_br_read_ahead(strm, br, ds->lt.len_bits)) { if (last) goto failed;/* Truncated data. */ ds->state = ST_RD_LITERAL_1; return (ARCHIVE_OK); } ds->lt.len_avail = lzh_br_bits(br, ds->lt.len_bits); lzh_br_consume(br, ds->lt.len_bits); /* FALL THROUGH */ case ST_RD_LITERAL_2: if (ds->lt.len_avail == 0) { /* There is no bitlen. */ if (!lzh_br_read_ahead(strm, br, ds->lt.len_bits)) { if (last) goto failed;/* Truncated data.*/ ds->state = ST_RD_LITERAL_2; return (ARCHIVE_OK); } if (!lzh_make_fake_table(&(ds->lt), lzh_br_bits(br, ds->lt.len_bits))) goto failed;/* Invalid data */ lzh_br_consume(br, ds->lt.len_bits); ds->state = ST_RD_POS_DATA_1; break; } else if (ds->lt.len_avail > ds->lt.len_size) goto failed;/* Invalid data */ ds->loop = 0; memset(ds->lt.freq, 0, sizeof(ds->lt.freq)); /* FALL THROUGH */ case ST_RD_LITERAL_3: i = ds->loop; while (i < ds->lt.len_avail) { if (!lzh_br_read_ahead(strm, br, ds->pt.max_bits)) { if (last) goto failed;/* Truncated data.*/ ds->loop = i; ds->state = ST_RD_LITERAL_3; return (ARCHIVE_OK); } rbits = lzh_br_bits(br, ds->pt.max_bits); c = lzh_decode_huffman(&(ds->pt), rbits); if (c > 2) { /* Note: 'c' will never be more than * eighteen since it's limited by * PT_BITLEN_SIZE, which is being set * to ds->pt.len_size through * ds->literal_pt_len_size. */ lzh_br_consume(br, ds->pt.bitlen[c]); c -= 2; ds->lt.freq[c]++; ds->lt.bitlen[i++] = c; } else if (c == 0) { lzh_br_consume(br, ds->pt.bitlen[c]); ds->lt.bitlen[i++] = 0; } else { /* c == 1 or c == 2 */ int n = (c == 1)?4:9; if (!lzh_br_read_ahead(strm, br, ds->pt.bitlen[c] + n)) { if (last) /* Truncated data. */ goto failed; ds->loop = i; ds->state = ST_RD_LITERAL_3; return (ARCHIVE_OK); } lzh_br_consume(br, ds->pt.bitlen[c]); c = lzh_br_bits(br, n); lzh_br_consume(br, n); c += (n == 4)?3:20; if (i + c > ds->lt.len_avail) goto failed;/* Invalid data */ memset(&(ds->lt.bitlen[i]), 0, c); i += c; } } if (i > ds->lt.len_avail || !lzh_make_huffman_table(&(ds->lt))) goto failed;/* Invalid data */ /* FALL THROUGH */ case ST_RD_POS_DATA_1: /* * Read a position table compressed in huffman * coding. */ ds->pt.len_size = ds->pos_pt_len_size; ds->pt.len_bits = ds->pos_pt_len_bits; ds->reading_position = 1; ds->state = ST_RD_PT_1; break; case ST_GET_LITERAL: return (100); } } failed: return (ds->error = ARCHIVE_FAILED); } static int lzh_decode_blocks(struct lzh_stream *strm, int last) { struct lzh_dec *ds = strm->ds; struct lzh_br bre = ds->br; struct huffman *lt = &(ds->lt); struct huffman *pt = &(ds->pt); unsigned char *w_buff = ds->w_buff; unsigned char *lt_bitlen = lt->bitlen; unsigned char *pt_bitlen = pt->bitlen; int blocks_avail = ds->blocks_avail, c = 0; int copy_len = ds->copy_len, copy_pos = ds->copy_pos; int w_pos = ds->w_pos, w_mask = ds->w_mask, w_size = ds->w_size; int lt_max_bits = lt->max_bits, pt_max_bits = pt->max_bits; int state = ds->state; for (;;) { switch (state) { case ST_GET_LITERAL: for (;;) { if (blocks_avail == 0) { /* We have decoded all blocks. * Let's handle next blocks. */ ds->state = ST_RD_BLOCK; ds->br = bre; ds->blocks_avail = 0; ds->w_pos = w_pos; ds->copy_pos = 0; return (100); } /* lzh_br_read_ahead() always try to fill the * cache buffer up. In specific situation we * are close to the end of the data, the cache * buffer will not be full and thus we have to * determine if the cache buffer has some bits * as much as we need after lzh_br_read_ahead() * failed. */ if (!lzh_br_read_ahead(strm, &bre, lt_max_bits)) { if (!last) goto next_data; /* Remaining bits are less than * maximum bits(lt.max_bits) but maybe * it still remains as much as we need, * so we should try to use it with * dummy bits. */ c = lzh_decode_huffman(lt, lzh_br_bits_forced(&bre, lt_max_bits)); lzh_br_consume(&bre, lt_bitlen[c]); if (!lzh_br_has(&bre, 0)) goto failed;/* Over read. */ } else { c = lzh_decode_huffman(lt, lzh_br_bits(&bre, lt_max_bits)); lzh_br_consume(&bre, lt_bitlen[c]); } blocks_avail--; if (c > UCHAR_MAX) /* Current block is a match data. */ break; /* * 'c' is exactly a literal code. */ /* Save a decoded code to reference it * afterward. */ w_buff[w_pos] = c; if (++w_pos >= w_size) { w_pos = 0; lzh_emit_window(strm, w_size); goto next_data; } } /* 'c' is the length of a match pattern we have * already extracted, which has be stored in * window(ds->w_buff). */ copy_len = c - (UCHAR_MAX + 1) + MINMATCH; /* FALL THROUGH */ case ST_GET_POS_1: /* * Get a reference position. */ if (!lzh_br_read_ahead(strm, &bre, pt_max_bits)) { if (!last) { state = ST_GET_POS_1; ds->copy_len = copy_len; goto next_data; } copy_pos = lzh_decode_huffman(pt, lzh_br_bits_forced(&bre, pt_max_bits)); lzh_br_consume(&bre, pt_bitlen[copy_pos]); if (!lzh_br_has(&bre, 0)) goto failed;/* Over read. */ } else { copy_pos = lzh_decode_huffman(pt, lzh_br_bits(&bre, pt_max_bits)); lzh_br_consume(&bre, pt_bitlen[copy_pos]); } /* FALL THROUGH */ case ST_GET_POS_2: if (copy_pos > 1) { /* We need an additional adjustment number to * the position. */ int p = copy_pos - 1; if (!lzh_br_read_ahead(strm, &bre, p)) { if (last) goto failed;/* Truncated data.*/ state = ST_GET_POS_2; ds->copy_len = copy_len; ds->copy_pos = copy_pos; goto next_data; } copy_pos = (1 << p) + lzh_br_bits(&bre, p); lzh_br_consume(&bre, p); } /* The position is actually a distance from the last * code we had extracted and thus we have to convert * it to a position of the window. */ copy_pos = (w_pos - copy_pos - 1) & w_mask; /* FALL THROUGH */ case ST_COPY_DATA: /* * Copy `copy_len' bytes as extracted data from * the window into the output buffer. */ for (;;) { int l; l = copy_len; if (copy_pos > w_pos) { if (l > w_size - copy_pos) l = w_size - copy_pos; } else { if (l > w_size - w_pos) l = w_size - w_pos; } if ((copy_pos + l < w_pos) || (w_pos + l < copy_pos)) { /* No overlap. */ memcpy(w_buff + w_pos, w_buff + copy_pos, l); } else { const unsigned char *s; unsigned char *d; int li; d = w_buff + w_pos; s = w_buff + copy_pos; for (li = 0; li < l-1;) { d[li] = s[li];li++; d[li] = s[li];li++; } if (li < l) d[li] = s[li]; } w_pos += l; if (w_pos == w_size) { w_pos = 0; lzh_emit_window(strm, w_size); if (copy_len <= l) state = ST_GET_LITERAL; else { state = ST_COPY_DATA; ds->copy_len = copy_len - l; ds->copy_pos = (copy_pos + l) & w_mask; } goto next_data; } if (copy_len <= l) /* A copy of current pattern ended. */ break; copy_len -= l; copy_pos = (copy_pos + l) & w_mask; } state = ST_GET_LITERAL; break; } } failed: return (ds->error = ARCHIVE_FAILED); next_data: ds->br = bre; ds->blocks_avail = blocks_avail; ds->state = state; ds->w_pos = w_pos; return (ARCHIVE_OK); } static int lzh_huffman_init(struct huffman *hf, size_t len_size, int tbl_bits) { int bits; if (hf->bitlen == NULL) { hf->bitlen = malloc(len_size * sizeof(hf->bitlen[0])); if (hf->bitlen == NULL) return (ARCHIVE_FATAL); } if (hf->tbl == NULL) { if (tbl_bits < HTBL_BITS) bits = tbl_bits; else bits = HTBL_BITS; hf->tbl = malloc(((size_t)1 << bits) * sizeof(hf->tbl[0])); if (hf->tbl == NULL) return (ARCHIVE_FATAL); } if (hf->tree == NULL && tbl_bits > HTBL_BITS) { hf->tree_avail = 1 << (tbl_bits - HTBL_BITS + 4); hf->tree = malloc(hf->tree_avail * sizeof(hf->tree[0])); if (hf->tree == NULL) return (ARCHIVE_FATAL); } hf->len_size = (int)len_size; hf->tbl_bits = tbl_bits; return (ARCHIVE_OK); } static void lzh_huffman_free(struct huffman *hf) { free(hf->bitlen); free(hf->tbl); free(hf->tree); } static char bitlen_tbl[0x400] = { 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 15, 15, 16, 0 }; static int lzh_read_pt_bitlen(struct lzh_stream *strm, int start, int end) { struct lzh_dec *ds = strm->ds; struct lzh_br *br = &(ds->br); int c, i; for (i = start; i < end; ) { /* * bit pattern the number we need * 000 -> 0 * 001 -> 1 * 010 -> 2 * ... * 110 -> 6 * 1110 -> 7 * 11110 -> 8 * ... * 1111111111110 -> 16 */ if (!lzh_br_read_ahead(strm, br, 3)) return (i); if ((c = lzh_br_bits(br, 3)) == 7) { if (!lzh_br_read_ahead(strm, br, 13)) return (i); c = bitlen_tbl[lzh_br_bits(br, 13) & 0x3FF]; if (c) lzh_br_consume(br, c - 3); else return (-1);/* Invalid data. */ } else lzh_br_consume(br, 3); ds->pt.bitlen[i++] = c; ds->pt.freq[c]++; } return (i); } static int lzh_make_fake_table(struct huffman *hf, uint16_t c) { if (c >= hf->len_size) return (0); hf->tbl[0] = c; hf->max_bits = 0; hf->shift_bits = 0; hf->bitlen[hf->tbl[0]] = 0; return (1); } /* * Make a huffman coding table. */ static int lzh_make_huffman_table(struct huffman *hf) { uint16_t *tbl; const unsigned char *bitlen; int bitptn[17], weight[17]; int i, maxbits = 0, ptn, tbl_size, w; int diffbits, len_avail; /* * Initialize bit patterns. */ ptn = 0; for (i = 1, w = 1 << 15; i <= 16; i++, w >>= 1) { bitptn[i] = ptn; weight[i] = w; if (hf->freq[i]) { ptn += hf->freq[i] * w; maxbits = i; } } if (ptn != 0x10000 || maxbits > hf->tbl_bits) return (0);/* Invalid */ hf->max_bits = maxbits; /* * Cut out extra bits which we won't house in the table. * This preparation reduces the same calculation in the for-loop * making the table. */ if (maxbits < 16) { int ebits = 16 - maxbits; for (i = 1; i <= maxbits; i++) { bitptn[i] >>= ebits; weight[i] >>= ebits; } } if (maxbits > HTBL_BITS) { unsigned htbl_max; uint16_t *p; diffbits = maxbits - HTBL_BITS; for (i = 1; i <= HTBL_BITS; i++) { bitptn[i] >>= diffbits; weight[i] >>= diffbits; } htbl_max = bitptn[HTBL_BITS] + weight[HTBL_BITS] * hf->freq[HTBL_BITS]; p = &(hf->tbl[htbl_max]); while (p < &hf->tbl[1U<shift_bits = diffbits; /* * Make the table. */ tbl_size = 1 << HTBL_BITS; tbl = hf->tbl; bitlen = hf->bitlen; len_avail = hf->len_avail; hf->tree_used = 0; for (i = 0; i < len_avail; i++) { uint16_t *p; int len, cnt; uint16_t bit; int extlen; struct htree_t *ht; if (bitlen[i] == 0) continue; /* Get a bit pattern */ len = bitlen[i]; ptn = bitptn[len]; cnt = weight[len]; if (len <= HTBL_BITS) { /* Calculate next bit pattern */ if ((bitptn[len] = ptn + cnt) > tbl_size) return (0);/* Invalid */ /* Update the table */ p = &(tbl[ptn]); if (cnt > 7) { uint16_t *pc; cnt -= 8; pc = &p[cnt]; pc[0] = (uint16_t)i; pc[1] = (uint16_t)i; pc[2] = (uint16_t)i; pc[3] = (uint16_t)i; pc[4] = (uint16_t)i; pc[5] = (uint16_t)i; pc[6] = (uint16_t)i; pc[7] = (uint16_t)i; if (cnt > 7) { cnt -= 8; memcpy(&p[cnt], pc, 8 * sizeof(uint16_t)); pc = &p[cnt]; while (cnt > 15) { cnt -= 16; memcpy(&p[cnt], pc, 16 * sizeof(uint16_t)); } } if (cnt) memcpy(p, pc, cnt * sizeof(uint16_t)); } else { while (cnt > 1) { p[--cnt] = (uint16_t)i; p[--cnt] = (uint16_t)i; } if (cnt) p[--cnt] = (uint16_t)i; } continue; } /* * A bit length is too big to be housed to a direct table, * so we use a tree model for its extra bits. */ bitptn[len] = ptn + cnt; bit = 1U << (diffbits -1); extlen = len - HTBL_BITS; p = &(tbl[ptn >> diffbits]); if (*p == 0) { *p = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { if (*p < len_avail || *p >= (len_avail + hf->tree_used)) return (0);/* Invalid */ ht = &(hf->tree[*p - len_avail]); } while (--extlen > 0) { if (ptn & bit) { if (ht->left < len_avail) { ht->left = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { ht = &(hf->tree[ht->left - len_avail]); } } else { if (ht->right < len_avail) { ht->right = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { ht = &(hf->tree[ht->right - len_avail]); } } bit >>= 1; } if (ptn & bit) { if (ht->left != 0) return (0);/* Invalid */ ht->left = (uint16_t)i; } else { if (ht->right != 0) return (0);/* Invalid */ ht->right = (uint16_t)i; } } return (1); } static int lzh_decode_huffman_tree(struct huffman *hf, unsigned rbits, int c) { struct htree_t *ht; int extlen; ht = hf->tree; extlen = hf->shift_bits; while (c >= hf->len_avail) { c -= hf->len_avail; if (extlen-- <= 0 || c >= hf->tree_used) return (0); if (rbits & (1U << extlen)) c = ht[c].left; else c = ht[c].right; } return (c); } static inline int lzh_decode_huffman(struct huffman *hf, unsigned rbits) { int c; /* * At first search an index table for a bit pattern. * If it fails, search a huffman tree for. */ c = hf->tbl[rbits >> hf->shift_bits]; if (c < hf->len_avail || hf->len_avail == 0) return (c); /* This bit pattern needs to be found out at a huffman tree. */ return (lzh_decode_huffman_tree(hf, rbits, c)); } Index: vendor/libarchive/dist/libarchive/archive_read_support_format_warc.c =================================================================== --- vendor/libarchive/dist/libarchive/archive_read_support_format_warc.c (revision 302055) +++ vendor/libarchive/dist/libarchive/archive_read_support_format_warc.c (revision 302056) @@ -1,795 +1,801 @@ /*- * Copyright (c) 2014 Sebastian Freundt * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); /** * WARC is standardised by ISO TC46/SC4/WG12 and currently available as * ISO 28500:2009. * For the purposes of this file we used the final draft from: * http://bibnum.bnf.fr/warc/WARC_ISO_28500_version1_latestdraft.pdf * * Todo: * [ ] real-world warcs can contain resources at endpoints ending in / * e.g. http://bibnum.bnf.fr/warc/ * if you're lucky their response contains a Content-Location: header * pointing to a unix-compliant filename, in the example above it's * Content-Location: http://bibnum.bnf.fr/warc/index.html * however, that's not mandated and github for example doesn't follow * this convention. * We need a set of archive options to control what to do with * entries like these, at the moment care is taken to skip them. * **/ #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_CTYPE_H #include #endif #ifdef HAVE_TIME_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_private.h" typedef enum { WT_NONE, /* warcinfo */ WT_INFO, /* metadata */ WT_META, /* resource */ WT_RSRC, /* request, unsupported */ WT_REQ, /* response, unsupported */ WT_RSP, /* revisit, unsupported */ WT_RVIS, /* conversion, unsupported */ WT_CONV, /* continutation, unsupported at the moment */ WT_CONT, /* invalid type */ LAST_WT } warc_type_t; typedef struct { size_t len; const char *str; } warc_string_t; typedef struct { size_t len; char *str; } warc_strbuf_t; struct warc_s { /* content length ahead */ size_t cntlen; /* and how much we've processed so far */ size_t cntoff; /* and how much we need to consume between calls */ size_t unconsumed; /* string pool */ warc_strbuf_t pool; /* previous version */ unsigned int pver; /* stringified format name */ struct archive_string sver; }; static int _warc_bid(struct archive_read *a, int); static int _warc_cleanup(struct archive_read *a); static int _warc_read(struct archive_read*, const void**, size_t*, int64_t*); static int _warc_skip(struct archive_read *a); static int _warc_rdhdr(struct archive_read *a, struct archive_entry *e); /* private routines */ static unsigned int _warc_rdver(const char buf[10], size_t bsz); static unsigned int _warc_rdtyp(const char *buf, size_t bsz); static warc_string_t _warc_rduri(const char *buf, size_t bsz); static ssize_t _warc_rdlen(const char *buf, size_t bsz); static time_t _warc_rdrtm(const char *buf, size_t bsz); static time_t _warc_rdmtm(const char *buf, size_t bsz); static const char *_warc_find_eoh(const char *buf, size_t bsz); int archive_read_support_format_warc(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct warc_s *w; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_warc"); if ((w = malloc(sizeof(*w))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate warc data"); return (ARCHIVE_FATAL); } memset(w, 0, sizeof(*w)); r = __archive_read_register_format( a, w, "warc", _warc_bid, NULL, _warc_rdhdr, _warc_read, _warc_skip, NULL, _warc_cleanup, NULL, NULL); if (r != ARCHIVE_OK) { free(w); return (r); } return (ARCHIVE_OK); } static int _warc_cleanup(struct archive_read *a) { struct warc_s *w = a->format->data; if (w->pool.len > 0U) { free(w->pool.str); } archive_string_free(&w->sver); free(w); a->format->data = NULL; return (ARCHIVE_OK); } static int _warc_bid(struct archive_read *a, int best_bid) { const char *hdr; ssize_t nrd; unsigned int ver; (void)best_bid; /* UNUSED */ /* check first line of file, it should be a record already */ if ((hdr = __archive_read_ahead(a, 12U, &nrd)) == NULL) { /* no idea what to do */ return -1; } else if (nrd < 12) { /* nah, not for us, our magic cookie is at least 12 bytes */ return -1; } /* otherwise snarf the record's version number */ ver = _warc_rdver(hdr, nrd); if (ver == 0U || ver > 10000U) { /* oh oh oh, best not to wager ... */ return -1; } /* otherwise be confident */ return (64); } static int _warc_rdhdr(struct archive_read *a, struct archive_entry *entry) { #define HDR_PROBE_LEN (12U) struct warc_s *w = a->format->data; unsigned int ver; const char *buf; ssize_t nrd; const char *eoh; /* for the file name, saves some strndup()'ing */ warc_string_t fnam; /* warc record type, not that we really use it a lot */ warc_type_t ftyp; /* content-length+error monad */ ssize_t cntlen; /* record time is the WARC-Date time we reinterpret it as ctime */ time_t rtime; /* mtime is the Last-Modified time which will be the entry's mtime */ time_t mtime; start_over: /* just use read_ahead() they keep track of unconsumed * bits and bobs for us; no need to put an extra shift in * and reproduce that functionality here */ buf = __archive_read_ahead(a, HDR_PROBE_LEN, &nrd); if (nrd < 0) { /* no good */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Bad record header"); return (ARCHIVE_FATAL); } else if (buf == NULL) { /* there should be room for at least WARC/bla\r\n * must be EOF therefore */ return (ARCHIVE_EOF); } /* looks good so far, try and find the end of the header now */ eoh = _warc_find_eoh(buf, nrd); if (eoh == NULL) { /* still no good, the header end might be beyond the * probe we've requested, but then again who'd cram * so much stuff into the header *and* be 28500-compliant */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Bad record header"); return (ARCHIVE_FATAL); } else if ((ver = _warc_rdver(buf, eoh - buf)) > 10000U) { /* nawww, I wish they promised backward compatibility * anyhoo, in their infinite wisdom the 28500 guys might * come up with something we can't possibly handle so * best end things here */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Unsupported record version"); return (ARCHIVE_FATAL); } else if ((cntlen = _warc_rdlen(buf, eoh - buf)) < 0) { /* nightmare! the specs say content-length is mandatory * so I don't feel overly bad stopping the reader here */ archive_set_error( &a->archive, EINVAL, "Bad content length"); return (ARCHIVE_FATAL); } else if ((rtime = _warc_rdrtm(buf, eoh - buf)) == (time_t)-1) { /* record time is mandatory as per WARC/1.0, * so just barf here, fast and loud */ archive_set_error( &a->archive, EINVAL, "Bad record time"); return (ARCHIVE_FATAL); } /* let the world know we're a WARC archive */ a->archive.archive_format = ARCHIVE_FORMAT_WARC; if (ver != w->pver) { /* stringify this entry's version */ archive_string_sprintf(&w->sver, "WARC/%u.%u", ver / 10000, ver % 10000); /* remember the version */ w->pver = ver; } /* start off with the type */ ftyp = _warc_rdtyp(buf, eoh - buf); /* and let future calls know about the content */ w->cntlen = cntlen; w->cntoff = 0U; mtime = 0;/* Avoid compiling error on some platform. */ switch (ftyp) { case WT_RSRC: case WT_RSP: /* only try and read the filename in the cases that are * guaranteed to have one */ fnam = _warc_rduri(buf, eoh - buf); /* check the last character in the URI to avoid creating * directory endpoints as files, see Todo above */ if (fnam.len == 0 || fnam.str[fnam.len - 1] == '/') { /* break here for now */ fnam.len = 0U; fnam.str = NULL; break; } /* bang to our string pool, so we save a * malloc()+free() roundtrip */ if (fnam.len + 1U > w->pool.len) { w->pool.len = ((fnam.len + 64U) / 64U) * 64U; w->pool.str = realloc(w->pool.str, w->pool.len); } memcpy(w->pool.str, fnam.str, fnam.len); w->pool.str[fnam.len] = '\0'; /* let noone else know about the pool, it's a secret, shhh */ fnam.str = w->pool.str; /* snarf mtime or deduce from rtime * this is a custom header added by our writer, it's quite * hard to believe anyone else would go through with it * (apart from being part of some http responses of course) */ if ((mtime = _warc_rdmtm(buf, eoh - buf)) == (time_t)-1) { mtime = rtime; } break; default: fnam.len = 0U; fnam.str = NULL; break; } /* now eat some of those delicious buffer bits */ __archive_read_consume(a, eoh - buf); switch (ftyp) { case WT_RSRC: case WT_RSP: if (fnam.len > 0U) { /* populate entry object */ archive_entry_set_filetype(entry, AE_IFREG); archive_entry_copy_pathname(entry, fnam.str); archive_entry_set_size(entry, cntlen); archive_entry_set_perm(entry, 0644); /* rtime is the new ctime, mtime stays mtime */ archive_entry_set_ctime(entry, rtime, 0L); archive_entry_set_mtime(entry, mtime, 0L); break; } /* FALLTHROUGH */ default: /* consume the content and start over */ _warc_skip(a); goto start_over; } return (ARCHIVE_OK); } static int _warc_read(struct archive_read *a, const void **buf, size_t *bsz, int64_t *off) { struct warc_s *w = a->format->data; const char *rab; ssize_t nrd; if (w->cntoff >= w->cntlen) { eof: /* it's our lucky day, no work, we can leave early */ *buf = NULL; *bsz = 0U; *off = w->cntoff + 4U/*for \r\n\r\n separator*/; w->unconsumed = 0U; return (ARCHIVE_EOF); } rab = __archive_read_ahead(a, 1U, &nrd); if (nrd < 0) { *bsz = 0U; /* big catastrophe */ return (int)nrd; } else if (nrd == 0) { goto eof; } else if ((size_t)nrd > w->cntlen - w->cntoff) { /* clamp to content-length */ nrd = w->cntlen - w->cntoff; } *off = w->cntoff; *bsz = nrd; *buf = rab; w->cntoff += nrd; w->unconsumed = (size_t)nrd; return (ARCHIVE_OK); } static int _warc_skip(struct archive_read *a) { struct warc_s *w = a->format->data; __archive_read_consume(a, w->cntlen + 4U/*\r\n\r\n separator*/); w->cntlen = 0U; w->cntoff = 0U; return (ARCHIVE_OK); } /* private routines */ static void* deconst(const void *c) { return (char *)0x1 + (((const char *)c) - (const char *)0x1); } static char* xmemmem(const char *hay, const size_t haysize, const char *needle, const size_t needlesize) { const char *const eoh = hay + haysize; const char *const eon = needle + needlesize; const char *hp; const char *np; const char *cand; unsigned int hsum; unsigned int nsum; unsigned int eqp; /* trivial checks first * a 0-sized needle is defined to be found anywhere in haystack * then run strchr() to find a candidate in HAYSTACK (i.e. a portion * that happens to begin with *NEEDLE) */ if (needlesize == 0UL) { return deconst(hay); } else if ((hay = memchr(hay, *needle, haysize)) == NULL) { /* trivial */ return NULL; } /* First characters of haystack and needle are the same now. Both are * guaranteed to be at least one character long. Now computes the sum * of characters values of needle together with the sum of the first * needle_len characters of haystack. */ for (hp = hay + 1U, np = needle + 1U, hsum = *hay, nsum = *hay, eqp = 1U; hp < eoh && np < eon; hsum ^= *hp, nsum ^= *np, eqp &= *hp == *np, hp++, np++); /* HP now references the (NEEDLESIZE + 1)-th character. */ if (np < eon) { /* haystack is smaller than needle, :O */ return NULL; } else if (eqp) { /* found a match */ return deconst(hay); } /* now loop through the rest of haystack, * updating the sum iteratively */ for (cand = hay; hp < eoh; hp++) { hsum ^= *cand++; hsum ^= *hp; /* Since the sum of the characters is already known to be * equal at that point, it is enough to check just NEEDLESIZE - 1 * characters for equality, * also CAND is by design < HP, so no need for range checks */ if (hsum == nsum && memcmp(cand, needle, needlesize - 1U) == 0) { return deconst(cand); } } return NULL; } static int strtoi_lim(const char *str, const char **ep, int llim, int ulim) { int res = 0; const char *sp; /* we keep track of the number of digits via rulim */ int rulim; for (sp = str, rulim = ulim > 10 ? ulim : 10; res * 10 <= ulim && rulim && *sp >= '0' && *sp <= '9'; sp++, rulim /= 10) { res *= 10; res += *sp - '0'; } if (sp == str) { res = -1; } else if (res < llim || res > ulim) { res = -2; } *ep = (const char*)sp; return res; } static time_t time_from_tm(struct tm *t) { #if HAVE_TIMEGM /* Use platform timegm() if available. */ return (timegm(t)); #elif HAVE__MKGMTIME64 return (_mkgmtime64(t)); #else /* Else use direct calculation using POSIX assumptions. */ /* First, fix up tm_yday based on the year/month/day. */ if (mktime(t) == (time_t)-1) return ((time_t)-1); /* Then we can compute timegm() from first principles. */ return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 + ((t->tm_year - 69) / 4) * 86400 - ((t->tm_year - 1) / 100) * 86400 + ((t->tm_year + 299) / 400) * 86400); #endif } static time_t xstrpisotime(const char *s, char **endptr) { /** like strptime() but strictly for ISO 8601 Zulu strings */ struct tm tm; time_t res = (time_t)-1; /* make sure tm is clean */ memset(&tm, 0, sizeof(tm)); /* as a courtesy to our callers, and since this is a non-standard * routine, we skip leading whitespace */ - for (; isspace(*s); s++); + while (isspace((unsigned char)*s)) + ++s; /* read year */ if ((tm.tm_year = strtoi_lim(s, &s, 1583, 4095)) < 0 || *s++ != '-') { goto out; } /* read month */ if ((tm.tm_mon = strtoi_lim(s, &s, 1, 12)) < 0 || *s++ != '-') { goto out; } /* read day-of-month */ if ((tm.tm_mday = strtoi_lim(s, &s, 1, 31)) < 0 || *s++ != 'T') { goto out; } /* read hour */ if ((tm.tm_hour = strtoi_lim(s, &s, 0, 23)) < 0 || *s++ != ':') { goto out; } /* read minute */ if ((tm.tm_min = strtoi_lim(s, &s, 0, 59)) < 0 || *s++ != ':') { goto out; } /* read second */ if ((tm.tm_sec = strtoi_lim(s, &s, 0, 60)) < 0 || *s++ != 'Z') { goto out; } /* massage TM to fulfill some of POSIX' contraints */ tm.tm_year -= 1900; tm.tm_mon--; /* now convert our custom tm struct to a unix stamp using UTC */ res = time_from_tm(&tm); out: if (endptr != NULL) { *endptr = deconst(s); } return res; } static unsigned int _warc_rdver(const char buf[10], size_t bsz) { static const char magic[] = "WARC/"; unsigned int ver; (void)bsz; /* UNUSED */ if (memcmp(buf, magic, sizeof(magic) - 1U) != 0) { /* nope */ return 99999U; } /* looks good so far, read the version number for a laugh */ buf += sizeof(magic) - 1U; /* most common case gets a quick-check here */ if (memcmp(buf, "1.0\r\n", 5U) == 0) { ver = 10000U; } else { switch (*buf) { case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': if (buf[1U] == '.') { char *on; /* set up major version */ ver = (buf[0U] - '0') * 10000U; /* minor version, anyone? */ ver += (strtol(buf + 2U, &on, 10)) * 100U; /* don't parse anything else */ if (on > buf + 2U) { break; } } /* FALLTHROUGH */ case '9': default: /* just make the version ridiculously high */ ver = 999999U; break; } } return ver; } static unsigned int _warc_rdtyp(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Type:"; const char *const eob = buf + bsz; const char *val; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return WT_NONE; } /* overread whitespace */ - for (val += sizeof(_key) - 1U; val < eob && isspace(*val); val++); + val += sizeof(_key) - 1U; + while (val < eob && isspace((unsigned char)*val)) + ++val; if (val + 8U > eob) { ; } else if (memcmp(val, "resource", 8U) == 0) { return WT_RSRC; } else if (memcmp(val, "warcinfo", 8U) == 0) { return WT_INFO; } else if (memcmp(val, "metadata", 8U) == 0) { return WT_META; } else if (memcmp(val, "request", 7U) == 0) { return WT_REQ; } else if (memcmp(val, "response", 8U) == 0) { return WT_RSP; } else if (memcmp(val, "conversi", 8U) == 0) { return WT_CONV; } else if (memcmp(val, "continua", 8U) == 0) { return WT_CONT; } return WT_NONE; } static warc_string_t _warc_rduri(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Target-URI:"; const char *const eob = buf + bsz; const char *val; const char *uri; const char *eol; warc_string_t res = {0U, NULL}; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return res; } /* overread whitespace */ - for (val += sizeof(_key) - 1U; val < eob && isspace(*val); val++); + val += sizeof(_key) - 1U; + while (val < eob && isspace((unsigned char)*val)) + ++val; /* overread URL designators */ if ((uri = xmemmem(val, eob - val, "://", 3U)) == NULL) { /* not touching that! */ return res; } else if ((eol = memchr(uri, '\n', eob - uri)) == NULL) { /* no end of line? :O */ return res; } /* massage uri to point to after :// */ uri += 3U; /* also massage eol to point to the first whitespace * after the last non-whitespace character before * the end of the line */ - for (; eol > uri && isspace(eol[-1]); eol--); + while (eol > uri && isspace((unsigned char)eol[-1])) + --eol; /* now then, inspect the URI */ if (memcmp(val, "file", 4U) == 0) { /* perfect, nothing left to do here */ } else if (memcmp(val, "http", 4U) == 0 || memcmp(val, "ftp", 3U) == 0) { /* overread domain, and the first / */ while (uri < eol && *uri++ != '/'); } else { /* not sure what to do? best to bugger off */ return res; } res.str = uri; res.len = eol - uri; return res; } static ssize_t _warc_rdlen(const char *buf, size_t bsz) { static const char _key[] = "\r\nContent-Length:"; const char *val; char *on = NULL; long int len; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return -1; } /* strtol kindly overreads whitespace for us, so use that */ val += sizeof(_key) - 1U; len = strtol(val, &on, 10); - if (on == NULL || !isspace(*on)) { + if (on == NULL || !isspace((unsigned char)*on)) { /* hm, can we trust that number? Best not. */ return -1; } return (size_t)len; } static time_t _warc_rdrtm(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Date:"; const char *val; char *on = NULL; time_t res; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return (time_t)-1; } /* xstrpisotime() kindly overreads whitespace for us, so use that */ val += sizeof(_key) - 1U; res = xstrpisotime(val, &on); - if (on == NULL || !isspace(*on)) { + if (on == NULL || !isspace((unsigned char)*on)) { /* hm, can we trust that number? Best not. */ return (time_t)-1; } return res; } static time_t _warc_rdmtm(const char *buf, size_t bsz) { static const char _key[] = "\r\nLast-Modified:"; const char *val; char *on = NULL; time_t res; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return (time_t)-1; } /* xstrpisotime() kindly overreads whitespace for us, so use that */ val += sizeof(_key) - 1U; res = xstrpisotime(val, &on); - if (on == NULL || !isspace(*on)) { + if (on == NULL || !isspace((unsigned char)*on)) { /* hm, can we trust that number? Best not. */ return (time_t)-1; } return res; } static const char* _warc_find_eoh(const char *buf, size_t bsz) { static const char _marker[] = "\r\n\r\n"; const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); if (hit != NULL) { hit += sizeof(_marker) - 1U; } return hit; } /* archive_read_support_format_warc.c ends here */ Index: vendor/libarchive/dist/libarchive_fe/passphrase.c =================================================================== --- vendor/libarchive/dist/libarchive_fe/passphrase.c (revision 302055) +++ vendor/libarchive/dist/libarchive_fe/passphrase.c (revision 302056) @@ -1,317 +1,322 @@ /*- * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ /* $OpenBSD: readpassphrase.c,v 1.22 2010/01/13 10:20:54 dtucker Exp $ */ /* * Copyright (c) 2000-2002, 2007 Todd C. Miller * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. * * Sponsored in part by the Defense Advanced Research Projects * Agency (DARPA) and Air Force Research Laboratory, Air Force * Materiel Command, USAF, under agreement number F39502-99-1-0512. */ /* OPENBSD ORIGINAL: lib/libc/gen/readpassphrase.c */ #include "lafe_platform.h" __FBSDID("$FreeBSD$"); #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_READPASSPHRASE_H #include #endif #include "err.h" #include "passphrase.h" #ifndef HAVE_READPASSPHRASE #define RPP_ECHO_OFF 0x00 /* Turn off echo (default). */ #define RPP_ECHO_ON 0x01 /* Leave echo on. */ #define RPP_REQUIRE_TTY 0x02 /* Fail if there is no tty. */ #define RPP_FORCELOWER 0x04 /* Force input to lower case. */ #define RPP_FORCEUPPER 0x08 /* Force input to upper case. */ #define RPP_SEVENBIT 0x10 /* Strip the high bit from input. */ #define RPP_STDIN 0x20 /* Read from stdin, not /dev/tty */ #if defined(_WIN32) && !defined(__CYGWIN__) #include static char * readpassphrase(const char *prompt, char *buf, size_t bufsiz, int flags) { HANDLE hStdin, hStdout; DWORD mode, rbytes; BOOL success; (void)flags; hStdin = GetStdHandle(STD_INPUT_HANDLE); if (hStdin == INVALID_HANDLE_VALUE) return (NULL); hStdout = GetStdHandle(STD_OUTPUT_HANDLE); if (hStdout == INVALID_HANDLE_VALUE) return (NULL); success = GetConsoleMode(hStdin, &mode); if (!success) return (NULL); mode &= ~ENABLE_ECHO_INPUT; mode |= ENABLE_LINE_INPUT | ENABLE_PROCESSED_INPUT; success = SetConsoleMode(hStdin, mode); if (!success) return (NULL); success = WriteFile(hStdout, prompt, (DWORD)strlen(prompt), NULL, NULL); if (!success) return (NULL); success = ReadFile(hStdin, buf, (DWORD)bufsiz - 1, &rbytes, NULL); if (!success) return (NULL); WriteFile(hStdout, "\r\n", 2, NULL, NULL); buf[rbytes] = '\0'; /* Remove trailing carriage return(s). */ if (rbytes > 2 && buf[rbytes - 2] == '\r' && buf[rbytes - 1] == '\n') buf[rbytes - 2] = '\0'; return (buf); } #else /* _WIN32 && !__CYGWIN__ */ -#include -#include +#include #include #include #ifdef HAVE_PATHS_H #include #endif +#include #include +#include #include #ifdef TCSASOFT # define _T_FLUSH (TCSAFLUSH|TCSASOFT) #else # define _T_FLUSH (TCSAFLUSH) #endif /* SunOS 4.x which lacks _POSIX_VDISABLE, but has VDISABLE */ #if !defined(_POSIX_VDISABLE) && defined(VDISABLE) # define _POSIX_VDISABLE VDISABLE #endif -static volatile sig_atomic_t *signo; +#define M(a,b) (a > b ? a : b) +#define MAX_SIGNO M(M(M(SIGALRM, SIGHUP), \ + M(SIGINT, SIGPIPE)), \ + M(M(SIGQUIT, SIGTERM), \ + M(M(SIGTSTP, SIGTTIN), SIGTTOU))) +static volatile sig_atomic_t signo[MAX_SIGNO + 1]; + static void handler(int s) { + assert(s <= MAX_SIGNO); signo[s] = 1; } static char * readpassphrase(const char *prompt, char *buf, size_t bufsiz, int flags) { ssize_t nr; int input, output, save_errno, i, need_restart; char ch, *p, *end; struct termios term, oterm; struct sigaction sa, savealrm, saveint, savehup, savequit, saveterm; struct sigaction savetstp, savettin, savettou, savepipe; /* I suppose we could alloc on demand in this case (XXX). */ if (bufsiz == 0) { errno = EINVAL; return(NULL); } - if (signo == NULL) { - signo = calloc(SIGRTMAX, sizeof(sig_atomic_t)); - } - restart: - for (i = 0; i < SIGRTMAX; i++) + for (i = 0; i <= MAX_SIGNO; i++) signo[i] = 0; nr = -1; save_errno = 0; need_restart = 0; /* * Read and write to /dev/tty if available. If not, read from * stdin and write to stderr unless a tty is required. */ if ((flags & RPP_STDIN) || (input = output = open(_PATH_TTY, O_RDWR)) == -1) { if (flags & RPP_REQUIRE_TTY) { errno = ENOTTY; return(NULL); } input = STDIN_FILENO; output = STDERR_FILENO; } /* * Catch signals that would otherwise cause the user to end * up with echo turned off in the shell. Don't worry about * things like SIGXCPU and SIGVTALRM for now. */ sigemptyset(&sa.sa_mask); sa.sa_flags = 0; /* don't restart system calls */ sa.sa_handler = handler; + /* Keep this list in sync with MAX_SIGNO! */ (void)sigaction(SIGALRM, &sa, &savealrm); (void)sigaction(SIGHUP, &sa, &savehup); (void)sigaction(SIGINT, &sa, &saveint); (void)sigaction(SIGPIPE, &sa, &savepipe); (void)sigaction(SIGQUIT, &sa, &savequit); (void)sigaction(SIGTERM, &sa, &saveterm); (void)sigaction(SIGTSTP, &sa, &savetstp); (void)sigaction(SIGTTIN, &sa, &savettin); (void)sigaction(SIGTTOU, &sa, &savettou); /* Turn off echo if possible. */ if (input != STDIN_FILENO && tcgetattr(input, &oterm) == 0) { memcpy(&term, &oterm, sizeof(term)); if (!(flags & RPP_ECHO_ON)) term.c_lflag &= ~(ECHO | ECHONL); #ifdef VSTATUS if (term.c_cc[VSTATUS] != _POSIX_VDISABLE) term.c_cc[VSTATUS] = _POSIX_VDISABLE; #endif (void)tcsetattr(input, _T_FLUSH, &term); } else { memset(&term, 0, sizeof(term)); term.c_lflag |= ECHO; memset(&oterm, 0, sizeof(oterm)); oterm.c_lflag |= ECHO; } /* No I/O if we are already backgrounded. */ if (signo[SIGTTOU] != 1 && signo[SIGTTIN] != 1) { if (!(flags & RPP_STDIN)) { int r = write(output, prompt, strlen(prompt)); (void)r; } end = buf + bufsiz - 1; p = buf; while ((nr = read(input, &ch, 1)) == 1 && ch != '\n' && ch != '\r') { if (p < end) { if ((flags & RPP_SEVENBIT)) ch &= 0x7f; - if (isalpha(ch)) { + if (isalpha((unsigned char)ch)) { if ((flags & RPP_FORCELOWER)) - ch = (char)tolower(ch); + ch = (char)tolower((unsigned char)ch); if ((flags & RPP_FORCEUPPER)) - ch = (char)toupper(ch); + ch = (char)toupper((unsigned char)ch); } *p++ = ch; } } *p = '\0'; save_errno = errno; if (!(term.c_lflag & ECHO)) { int r = write(output, "\n", 1); (void)r; } } /* Restore old terminal settings and signals. */ if (memcmp(&term, &oterm, sizeof(term)) != 0) { while (tcsetattr(input, _T_FLUSH, &oterm) == -1 && errno == EINTR) continue; } (void)sigaction(SIGALRM, &savealrm, NULL); (void)sigaction(SIGHUP, &savehup, NULL); (void)sigaction(SIGINT, &saveint, NULL); (void)sigaction(SIGQUIT, &savequit, NULL); (void)sigaction(SIGPIPE, &savepipe, NULL); (void)sigaction(SIGTERM, &saveterm, NULL); (void)sigaction(SIGTSTP, &savetstp, NULL); (void)sigaction(SIGTTIN, &savettin, NULL); (void)sigaction(SIGTTOU, &savettou, NULL); if (input != STDIN_FILENO) (void)close(input); /* * If we were interrupted by a signal, resend it to ourselves * now that we have restored the signal handlers. */ - for (i = 0; i < SIGRTMAX; i++) { + for (i = 0; i <= MAX_SIGNO; i++) { if (signo[i]) { kill(getpid(), i); switch (i) { case SIGTSTP: case SIGTTIN: case SIGTTOU: need_restart = 1; } } } if (need_restart) goto restart; if (save_errno) errno = save_errno; return(nr == -1 ? NULL : buf); } #endif /* _WIN32 && !__CYGWIN__ */ #endif /* HAVE_READPASSPHRASE */ char * lafe_readpassphrase(const char *prompt, char *buf, size_t bufsiz) { char *p; p = readpassphrase(prompt, buf, bufsiz, RPP_ECHO_OFF); if (p == NULL) { switch (errno) { case EINTR: break; default: lafe_errc(1, errno, "Couldn't read passphrase"); break; } } return (p); }