diff options
| -rw-r--r-- | ChangeLog | 9 | ||||
| -rw-r--r-- | INSTALL | 2 | ||||
| -rw-r--r-- | Makefile.in | 31 | ||||
| -rw-r--r-- | NEWS | 17 | ||||
| -rw-r--r-- | README | 35 | ||||
| -rw-r--r-- | common.cc | 149 | ||||
| -rw-r--r-- | common_decode.cc | 200 | ||||
| -rwxr-xr-x | configure | 2 | ||||
| -rw-r--r-- | create.cc | 2 | ||||
| -rw-r--r-- | delete.cc | 8 | ||||
| -rw-r--r-- | delete_lz.cc | 9 | ||||
| -rw-r--r-- | doc/tarlz.1 | 22 | ||||
| -rw-r--r-- | doc/tarlz.info | 232 | ||||
| -rw-r--r-- | doc/tarlz.texi | 171 | ||||
| -rw-r--r-- | exclude.cc | 1 | ||||
| -rw-r--r-- | extract.cc | 203 | ||||
| -rw-r--r-- | list_lz.cc | 267 | ||||
| -rw-r--r-- | main.cc | 23 | ||||
| -rw-r--r-- | tarlz.h | 48 | ||||
| -rwxr-xr-x | testsuite/check.sh | 85 |
20 files changed, 854 insertions, 662 deletions
@@ -1,3 +1,12 @@ +2019-10-08 Antonio Diaz Diaz <antonio@gnu.org> + + * Version 0.16 released. + * extract.cc (extract_member): Fixed call order of chown, chmod. + * delete_lz.cc (delete_members_lz): Return 2 if collective member. + * main.cc: Set a valid invocation_name even if argc == 0. + * #include <sys/sysmacros.h> unconditionally. + * tarlz.texi: Added new chapter 'Portable character set'. + 2019-04-11 Antonio Diaz Diaz <antonio@gnu.org> * Version 0.15 released. @@ -1,7 +1,7 @@ Requirements ------------ You will need a C++ compiler and the lzlib compression library installed. -I use gcc 5.3.0 and 4.1.2, but the code should compile with any standards +I use gcc 6.1.0 and 4.1.2, but the code should compile with any standards compliant compiler. Lzlib must be version 1.0 or newer, but --keep-damaged requires lzlib 1.11 or newer to recover as much data as possible from each damaged member. diff --git a/Makefile.in b/Makefile.in index 9756ab6..9697efe 100644 --- a/Makefile.in +++ b/Makefile.in @@ -8,8 +8,9 @@ LIBS = -llz -lpthread SHELL = /bin/sh CAN_RUN_INSTALLINFO = $(SHELL) -c "install-info --version" > /dev/null 2>&1 -objs = arg_parser.o lzip_index.o create.o create_lz.o delete.o delete_lz.o \ - exclude.o extended.o extract.o list_lz.o main.o +objs = arg_parser.o lzip_index.o common.o common_decode.o create.o \ + create_lz.o delete.o delete_lz.o exclude.o extended.o extract.o \ + list_lz.o main.o .PHONY : all install install-bin install-info install-man \ @@ -29,18 +30,20 @@ main.o : main.cc %.o : %.cc $(CXX) $(CPPFLAGS) $(CXXFLAGS) -c -o $@ $< -$(objs) : Makefile -arg_parser.o : arg_parser.h -create.o : arg_parser.h tarlz.h -create_lz.o : arg_parser.h tarlz.h -delete.o : arg_parser.h lzip_index.h tarlz.h -delete_lz.o : arg_parser.h lzip_index.h tarlz.h -exclude.o : tarlz.h -extended.o : tarlz.h -extract.o : arg_parser.h lzip_index.h tarlz.h -list_lz.o : arg_parser.h lzip_index.h tarlz.h -lzip_index.o : lzip_index.h tarlz.h -main.o : arg_parser.h tarlz.h +$(objs) : Makefile +arg_parser.o : arg_parser.h +common.o : tarlz.h +common_decode.o : arg_parser.h tarlz.h +create.o : arg_parser.h tarlz.h +create_lz.o : arg_parser.h tarlz.h +delete.o : arg_parser.h lzip_index.h tarlz.h +delete_lz.o : arg_parser.h lzip_index.h tarlz.h +exclude.o : tarlz.h +extended.o : tarlz.h +extract.o : arg_parser.h lzip_index.h tarlz.h +list_lz.o : arg_parser.h lzip_index.h tarlz.h +lzip_index.o : lzip_index.h tarlz.h +main.o : arg_parser.h tarlz.h doc : info man @@ -1,10 +1,11 @@ -Changes in version 0.15: +Changes in version 0.16: -The new option '--delete', which deletes files and directories from an -archive in place, has been added. It currently can delete only from -uncompressed archives and from archives with individually compressed files -('--no-solid' archives). +'chown' and 'chmod' are now called in the right order on extracion to +preserve the S_ISUID and S_ISGID bits of executable files. -Multi-threaded listing of compressed archives with format violations (for -example, an extended header without the corresponding ustar header) has been -fixed. +The return value of '--delete' when failing to delete a tar member not +individually compressed has been fixed. It returned 0, but should be 2. + +The header <sys/sysmacros.h> is now #included unconditionally. + +The new chapter 'Portable character set' has been added to the manual. @@ -2,13 +2,19 @@ Description Tarlz is a massively parallel (multi-threaded) combined implementation of the tar archiver and the lzip compressor. Tarlz creates, lists and extracts -archives in a simplified posix pax format compressed with lzip, keeping the -alignment between tar members and lzip members. This method adds an indexed -lzip layer on top of the tar archive, making it possible to decode the -archive safely in parallel. The resulting multimember tar.lz archive is -fully backward compatible with standard tar tools like GNU tar, which treat -it like any other tar.lz archive. Tarlz can append files to the end of such -compressed archives. +archives in a simplified and safer variant of the POSIX pax format +compressed with lzip, keeping the alignment between tar members and lzip +members. The resulting multimember tar.lz archive is fully backward +compatible with standard tar tools like GNU tar, which treat it like any +other tar.lz archive. Tarlz can append files to the end of such compressed +archives. + +Keeping the alignment between tar members and lzip members has two +advantages. It adds an indexed lzip layer on top of the tar archive, making +it possible to decode the archive safely in parallel. It also minimizes the +amount of data lost in case of corruption. Compressing a tar archive with +plzip may even double the amount of files lost for each lzip member damaged +because it does not keep the members aligned. Tarlz can create tar archives with five levels of compression granularity; per file (--no-solid), per block (--bsolid, default), per directory @@ -25,7 +31,7 @@ archive, but it has the following advantages: member), and unwanted members can be deleted from the archive. Just like an uncompressed tar archive. - * It is a safe posix-style backup format. In case of corruption, + * It is a safe POSIX-style backup format. In case of corruption, tarlz can extract all the undamaged members from the tar.lz archive, skipping over the damaged members, just like the standard (uncompressed) tar. Moreover, the option '--keep-damaged' can be @@ -36,24 +42,31 @@ archive, but it has the following advantages: corresponding solidly compressed tar.gz archive, except when individually compressing files smaller than about 32 KiB. -Note that the posix pax format has a serious flaw. The metadata stored in +Note that the POSIX pax format has a serious flaw. The metadata stored in pax extended records are not protected by any kind of check sequence. -Corruption in a long filename may cause the extraction of the file in the +Corruption in a long file name may cause the extraction of the file in the wrong place without warning. Corruption in a large file size may cause the truncation of the file or the appending of garbage to the file, both followed by a spurious warning about a corrupt header far from the place of the undetected corruption. -Metadata like filename and file size must be always protected in an archive +Metadata like file name and file size must be always protected in an archive format because of the adverse effects of undetected corruption in them, potentially much worse that undetected corruption in the data. Even more so in the case of pax because the amount of metadata it stores is potentially large, making undetected corruption more probable. +Headers and metadata must be protected separately from data because the +integrity checking of lzip may not be able to detect the corruption before +the metadata has been used, for example, to create a new file in the wrong +place. + Because of the above, tarlz protects the extended records with a CRC in a way compatible with standard tar tools. Tarlz does not understand other tar formats like gnu, oldgnu, star or v7. +'tarlz -tf archive.tar.lz > /dev/null' can be used to verify that the format +of the archive is compatible with tarlz. The diagram below shows the correspondence between each tar member (formed by one or two headers plus optional data) in the tar archive and each lzip diff --git a/common.cc b/common.cc new file mode 100644 index 0000000..2e5d231 --- /dev/null +++ b/common.cc @@ -0,0 +1,149 @@ +/* Tarlz - Archiver with multimember lzip compression + Copyright (C) 2013-2019 Antonio Diaz Diaz. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +#define _FILE_OFFSET_BITS 64 + +#include <cctype> +#include <cerrno> +#include <climits> +#include <cstdlib> +#include <cstring> +#include <string> +#include <vector> +#include <pthread.h> +#include <stdint.h> +#include <unistd.h> + +#include "tarlz.h" + + +void xinit_mutex( pthread_mutex_t * const mutex ) + { + const int errcode = pthread_mutex_init( mutex, 0 ); + if( errcode ) + { show_error( "pthread_mutex_init", errcode ); cleanup_and_fail(); } + } + +void xinit_cond( pthread_cond_t * const cond ) + { + const int errcode = pthread_cond_init( cond, 0 ); + if( errcode ) + { show_error( "pthread_cond_init", errcode ); cleanup_and_fail(); } + } + + +void xdestroy_mutex( pthread_mutex_t * const mutex ) + { + const int errcode = pthread_mutex_destroy( mutex ); + if( errcode ) + { show_error( "pthread_mutex_destroy", errcode ); cleanup_and_fail(); } + } + +void xdestroy_cond( pthread_cond_t * const cond ) + { + const int errcode = pthread_cond_destroy( cond ); + if( errcode ) + { show_error( "pthread_cond_destroy", errcode ); cleanup_and_fail(); } + } + + +void xlock( pthread_mutex_t * const mutex ) + { + const int errcode = pthread_mutex_lock( mutex ); + if( errcode ) + { show_error( "pthread_mutex_lock", errcode ); cleanup_and_fail(); } + } + + +void xunlock( pthread_mutex_t * const mutex ) + { + const int errcode = pthread_mutex_unlock( mutex ); + if( errcode ) + { show_error( "pthread_mutex_unlock", errcode ); cleanup_and_fail(); } + } + + +void xwait( pthread_cond_t * const cond, pthread_mutex_t * const mutex ) + { + const int errcode = pthread_cond_wait( cond, mutex ); + if( errcode ) + { show_error( "pthread_cond_wait", errcode ); cleanup_and_fail(); } + } + + +void xsignal( pthread_cond_t * const cond ) + { + const int errcode = pthread_cond_signal( cond ); + if( errcode ) + { show_error( "pthread_cond_signal", errcode ); cleanup_and_fail(); } + } + + +void xbroadcast( pthread_cond_t * const cond ) + { + const int errcode = pthread_cond_broadcast( cond ); + if( errcode ) + { show_error( "pthread_cond_broadcast", errcode ); cleanup_and_fail(); } + } + + +unsigned long long parse_octal( const uint8_t * const ptr, const int size ) + { + unsigned long long result = 0; + int i = 0; + while( i < size && std::isspace( ptr[i] ) ) ++i; + for( ; i < size && ptr[i] >= '0' && ptr[i] <= '7'; ++i ) + { result <<= 3; result += ptr[i] - '0'; } + return result; + } + + +/* Returns the number of bytes really read. + If (returned value < size) and (errno == 0), means EOF was reached. +*/ +int readblock( const int fd, uint8_t * const buf, const int size ) + { + int sz = 0; + errno = 0; + while( sz < size ) + { + const int n = read( fd, buf + sz, size - sz ); + if( n > 0 ) sz += n; + else if( n == 0 ) break; // EOF + else if( errno != EINTR ) break; + errno = 0; + } + return sz; + } + + +/* Returns the number of bytes really written. + If (returned value < size), it is always an error. +*/ +int writeblock( const int fd, const uint8_t * const buf, const int size ) + { + int sz = 0; + errno = 0; + while( sz < size ) + { + const int n = write( fd, buf + sz, size - sz ); + if( n > 0 ) sz += n; + else if( n < 0 && errno != EINTR ) break; + errno = 0; + } + return sz; + } diff --git a/common_decode.cc b/common_decode.cc new file mode 100644 index 0000000..791d6e5 --- /dev/null +++ b/common_decode.cc @@ -0,0 +1,200 @@ +/* Tarlz - Archiver with multimember lzip compression + Copyright (C) 2013-2019 Antonio Diaz Diaz. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +#define _FILE_OFFSET_BITS 64 + +#include <climits> +#include <cstdio> +#include <cstdlib> +#include <cstring> +#include <ctime> +#include <string> +#include <vector> +#include <pthread.h> +#include <stdint.h> +#include <sys/stat.h> + +#include "arg_parser.h" +#include "tarlz.h" + + +namespace { + +enum { mode_string_size = 10, + group_string_size = 1 + uname_l + 1 + gname_l + 1 }; // 67 + +void format_mode_string( const Tar_header header, char buf[mode_string_size] ) + { + const Typeflag typeflag = (Typeflag)header[typeflag_o]; + + std::memcpy( buf, "----------", mode_string_size ); + switch( typeflag ) + { + case tf_regular: break; + case tf_link: buf[0] = 'h'; break; + case tf_symlink: buf[0] = 'l'; break; + case tf_chardev: buf[0] = 'c'; break; + case tf_blockdev: buf[0] = 'b'; break; + case tf_directory: buf[0] = 'd'; break; + case tf_fifo: buf[0] = 'p'; break; + case tf_hiperf: buf[0] = 'C'; break; + default: buf[0] = '?'; + } + const mode_t mode = parse_octal( header + mode_o, mode_l ); // 12 bits + const bool setuid = mode & S_ISUID; + const bool setgid = mode & S_ISGID; + const bool sticky = mode & S_ISVTX; + if( mode & S_IRUSR ) buf[1] = 'r'; + if( mode & S_IWUSR ) buf[2] = 'w'; + if( mode & S_IXUSR ) buf[3] = setuid ? 's' : 'x'; + else if( setuid ) buf[3] = 'S'; + if( mode & S_IRGRP ) buf[4] = 'r'; + if( mode & S_IWGRP ) buf[5] = 'w'; + if( mode & S_IXGRP ) buf[6] = setgid ? 's' : 'x'; + else if( setgid ) buf[6] = 'S'; + if( mode & S_IROTH ) buf[7] = 'r'; + if( mode & S_IWOTH ) buf[8] = 'w'; + if( mode & S_IXOTH ) buf[9] = sticky ? 't' : 'x'; + else if( sticky ) buf[9] = 'T'; + } + + +int format_user_group_string( const Tar_header header, + char buf[group_string_size] ) + { + int len; + if( header[uname_o] && header[gname_o] ) + len = snprintf( buf, group_string_size, + " %.32s/%.32s", header + uname_o, header + gname_o ); + else + { + const unsigned uid = parse_octal( header + uid_o, uid_l ); + const unsigned gid = parse_octal( header + gid_o, gid_l ); + len = snprintf( buf, group_string_size, " %u/%u", uid, gid ); + } + return len; + } + + +// return true if dir is a parent directory of name +bool compare_prefix_dir( const char * const dir, const char * const name ) + { + int len = 0; + while( dir[len] && dir[len] == name[len] ) ++len; + return ( !dir[len] && len > 0 && ( dir[len-1] == '/' || name[len] == '/' ) ); + } + + +// compare two file names ignoring trailing slashes +bool compare_tslash( const char * const name1, const char * const name2 ) + { + const char * p = name1; + const char * q = name2; + while( *p && *p == *q ) { ++p; ++q; } + while( *p == '/' ) ++p; + while( *q == '/' ) ++q; + return ( !*p && !*q ); + } + +} // end namespace + + +bool block_is_zero( const uint8_t * const buf, const int size ) + { + for( int i = 0; i < size; ++i ) if( buf[i] != 0 ) return false; + return true; + } + + +bool format_member_name( const Extended & extended, const Tar_header header, + Resizable_buffer & rbuf, const bool long_format ) + { + if( long_format ) + { + format_mode_string( header, rbuf() ); + const int group_string_len = + format_user_group_string( header, rbuf() + mode_string_size ); + int offset = mode_string_size + group_string_len; + const time_t mtime = parse_octal( header + mtime_o, mtime_l ); // 33 bits + struct tm tms; + const struct tm * tm = localtime_r( &mtime, &tms ); + if( !tm ) + { time_t z = 0; tm = localtime_r( &z, &tms ); if( !tm ) tm = &tms; } + const Typeflag typeflag = (Typeflag)header[typeflag_o]; + const bool islink = ( typeflag == tf_link || typeflag == tf_symlink ); + const char * const link_string = !islink ? "" : + ( ( typeflag == tf_link ) ? " link to " : " -> " ); + if( typeflag == tf_chardev || typeflag == tf_blockdev ) + offset += snprintf( rbuf() + offset, rbuf.size() - offset, " %5u,%u", + (unsigned)parse_octal( header + devmajor_o, devmajor_l ), + (unsigned)parse_octal( header + devminor_o, devminor_l ) ); + else + offset += snprintf( rbuf() + offset, rbuf.size() - offset, " %9llu", + extended.file_size() ); + for( int i = 0; i < 2; ++i ) + { + const int len = snprintf( rbuf() + offset, rbuf.size() - offset, + " %4d-%02u-%02u %02u:%02u %s%s%s\n", + 1900 + tm->tm_year, 1 + tm->tm_mon, tm->tm_mday, + tm->tm_hour, tm->tm_min, extended.path().c_str(), + link_string, islink ? extended.linkpath().c_str() : "" ); + if( (int)rbuf.size() > len + offset ) break; + if( !rbuf.resize( len + offset + 1 ) ) return false; + } + } + else + { + if( rbuf.size() < extended.path().size() + 2 && + !rbuf.resize( extended.path().size() + 2 ) ) return false; + snprintf( rbuf(), rbuf.size(), "%s\n", extended.path().c_str() ); + } + return true; + } + + +bool show_member_name( const Extended & extended, const Tar_header header, + const int vlevel, Resizable_buffer & rbuf ) + { + if( verbosity >= vlevel ) + { + if( !format_member_name( extended, header, rbuf, verbosity > vlevel ) ) + { show_error( mem_msg ); return false; } + std::fputs( rbuf(), stdout ); + std::fflush( stdout ); + } + return true; + } + + +bool check_skip_filename( const Arg_parser & parser, + std::vector< char > & name_pending, + const char * const filename, const int filenames ) + { + if( Exclude::excluded( filename ) ) return true; // skip excluded files + bool skip = filenames > 0; + if( skip ) + for( int i = 0; i < parser.arguments(); ++i ) + if( !parser.code( i ) && parser.argument( i ).size() ) + { + const char * const name = + remove_leading_dotslash( parser.argument( i ).c_str() ); + if( compare_prefix_dir( name, filename ) || + compare_tslash( name, filename ) ) + { skip = false; name_pending[i] = false; break; } + } + return skip; + } @@ -6,7 +6,7 @@ # to copy, distribute and modify it. pkgname=tarlz -pkgversion=0.15 +pkgversion=0.16 progname=tarlz srctrigger=doc/${pkgname}.texi @@ -30,9 +30,7 @@ #include <unistd.h> #include <sys/stat.h> #include <sys/types.h> -#if defined(__GNU_LIBRARY__) #include <sys/sysmacros.h> // for major, minor -#endif #include <ftw.h> #include <grp.h> #include <pwd.h> @@ -75,10 +75,10 @@ int tail_copy( const char * const archive_namep, const Arg_parser & parser, if( ostream_pos < 0 ) { show_error( "Seek error", errno ); retval = 1; } else if( ostream_pos > 0 && ostream_pos < lzip_index.file_size() ) { - int result; - do result = ftruncate( outfd, ostream_pos ); - while( result != 0 && errno == EINTR ); - if( result != 0 ) + int ret; + do ret = ftruncate( outfd, ostream_pos ); + while( ret != 0 && errno == EINTR ); + if( ret != 0 || lseek( outfd, 0, SEEK_END ) != ostream_pos ) { show_file_error( archive_namep, "Can't truncate archive", errno ); if( retval < 1 ) retval = 1; diff --git a/delete_lz.cc b/delete_lz.cc index 11c3a14..6888520 100644 --- a/delete_lz.cc +++ b/delete_lz.cc @@ -51,7 +51,7 @@ int delete_members_lz( const char * const archive_namep, long long istream_pos = 0; // source of next data move const long long cdata_size = lzip_index.cdata_size(); - int retval = 0; + int retval = 0, retval2 = 0; for( long i = 0; i < lzip_index.members(); ++i ) { const long long mdata_pos = lzip_index.dblock( i ).pos(); @@ -142,16 +142,16 @@ int delete_members_lz( const char * const archive_namep, if( member_begin != mdata_pos || data_pos != mdata_end ) { show_file_error( extended.path().c_str(), "Can't delete: not individually compressed." ); - retval = 2; extended.reset(); continue; } + retval2 = 2; extended.reset(); continue; } if( !show_member_name( extended, header, 1, rbuf ) ) { retval = 1; goto done; } const long long size = member_pos - istream_pos; if( size > 0 ) // move pending data each time a member is deleted { if( istream_pos == 0 ) - { if( !safe_seek( outfd, size ) ) { retval = 1; break; } } + { if( !safe_seek( outfd, size ) ) { retval = 1; goto done; } } else if( !safe_seek( infd, istream_pos ) || - !copy_file( infd, outfd, size ) ) { retval = 1; break; } + !copy_file( infd, outfd, size ) ) { retval = 1; goto done; } } istream_pos = member_end; } @@ -159,6 +159,7 @@ int delete_members_lz( const char * const archive_namep, } } done: + if( retval < retval2 ) retval = retval2; if( LZ_decompress_close( decoder ) < 0 && !retval ) { show_error( "LZ_decompress_close failed." ); retval = 1; } // tail copy keeps trailing data diff --git a/doc/tarlz.1 b/doc/tarlz.1 index 4308fc2..83325dd 100644 --- a/doc/tarlz.1 +++ b/doc/tarlz.1 @@ -1,5 +1,5 @@ .\" DO NOT MODIFY THIS FILE! It was generated by help2man 1.46.1. -.TH TARLZ "1" "April 2019" "tarlz 0.15" "User Commands" +.TH TARLZ "1" "October 2019" "tarlz 0.16" "User Commands" .SH NAME tarlz \- creates tar archives with multimember lzip compression .SH SYNOPSIS @@ -8,15 +8,19 @@ tarlz \- creates tar archives with multimember lzip compression .SH DESCRIPTION Tarlz is a massively parallel (multi\-threaded) combined implementation of the tar archiver and the lzip compressor. Tarlz creates, lists and extracts -archives in a simplified posix pax format compressed with lzip, keeping the -alignment between tar members and lzip members. This method adds an indexed -lzip layer on top of the tar archive, making it possible to decode the -archive safely in parallel. The resulting multimember tar.lz archive is -fully backward compatible with standard tar tools like GNU tar, which treat -it like any other tar.lz archive. Tarlz can append files to the end of such -compressed archives. +archives in a simplified and safer variant of the POSIX pax format +compressed with lzip, keeping the alignment between tar members and lzip +members. The resulting multimember tar.lz archive is fully backward +compatible with standard tar tools like GNU tar, which treat it like any +other tar.lz archive. Tarlz can append files to the end of such compressed +archives. .PP -The tarlz file format is a safe posix\-style backup format. In case of +Keeping the alignment between tar members and lzip members has two +advantages. It adds an indexed lzip layer on top of the tar archive, making +it possible to decode the archive safely in parallel. It also minimizes the +amount of data lost in case of corruption. +.PP +The tarlz file format is a safe POSIX\-style backup format. In case of corruption, tarlz can extract all the undamaged members from the tar.lz archive, skipping over the damaged members, just like the standard (uncompressed) tar. Moreover, the option '\-\-keep\-damaged' can be used to diff --git a/doc/tarlz.info b/doc/tarlz.info index f49bb43..edcee44 100644 --- a/doc/tarlz.info +++ b/doc/tarlz.info @@ -11,12 +11,13 @@ File: tarlz.info, Node: Top, Next: Introduction, Up: (dir) Tarlz Manual ************ -This manual is for Tarlz (version 0.15, 11 April 2019). +This manual is for Tarlz (version 0.16, 8 October 2019). * Menu: * Introduction:: Purpose and features of tarlz * Invoking tarlz:: Command line interface +* Portable character set:: POSIX portable filename character set * File format:: Detailed format of the compressed archive * Amendments to pax format:: The reasons for the differences with pax * Multi-threaded tar:: Limitations of parallel tar decoding @@ -39,13 +40,19 @@ File: tarlz.info, Node: Introduction, Next: Invoking tarlz, Prev: Top, Up: T Tarlz is a massively parallel (multi-threaded) combined implementation of the tar archiver and the lzip compressor. Tarlz creates, lists and -extracts archives in a simplified posix pax format compressed with -lzip, keeping the alignment between tar members and lzip members. This -method adds an indexed lzip layer on top of the tar archive, making it -possible to decode the archive safely in parallel. The resulting -multimember tar.lz archive is fully backward compatible with standard -tar tools like GNU tar, which treat it like any other tar.lz archive. -Tarlz can append files to the end of such compressed archives. +extracts archives in a simplified and safer variant of the POSIX pax +format compressed with lzip, keeping the alignment between tar members +and lzip members. The resulting multimember tar.lz archive is fully +backward compatible with standard tar tools like GNU tar, which treat +it like any other tar.lz archive. Tarlz can append files to the end of +such compressed archives. + + Keeping the alignment between tar members and lzip members has two +advantages. It adds an indexed lzip layer on top of the tar archive, +making it possible to decode the archive safely in parallel. It also +minimizes the amount of data lost in case of corruption. Compressing a +tar archive with plzip may even double the amount of files lost for +each lzip member damaged because it does not keep the members aligned. Tarlz can create tar archives with five levels of compression granularity; per file (--no-solid), per block (--bsolid, default), per @@ -62,7 +69,7 @@ archive, but it has the following advantages: member), and unwanted members can be deleted from the archive. Just like an uncompressed tar archive. - * It is a safe posix-style backup format. In case of corruption, + * It is a safe POSIX-style backup format. In case of corruption, tarlz can extract all the undamaged members from the tar.lz archive, skipping over the damaged members, just like the standard (uncompressed) tar. Moreover, the option '--keep-damaged' can be @@ -77,10 +84,11 @@ archive, but it has the following advantages: with standard tar tools. *Note crc32::. Tarlz does not understand other tar formats like 'gnu', 'oldgnu', -'star' or 'v7'. +'star' or 'v7'. 'tarlz -tf archive.tar.lz > /dev/null' can be used to +verify that the format of the archive is compatible with tarlz. -File: tarlz.info, Node: Invoking tarlz, Next: File format, Prev: Introduction, Up: Top +File: tarlz.info, Node: Invoking tarlz, Next: Portable character set, Prev: Introduction, Up: Top 2 Invoking tarlz **************** @@ -94,9 +102,9 @@ FILE is a directory. On archive creation or appending tarlz archives the files specified, but removes from member names any leading and trailing slashes and any -filename prefixes containing a '..' component. On extraction, leading +file name prefixes containing a '..' component. On extraction, leading and trailing slashes are also removed from member names, and archive -members containing a '..' component in the filename are skipped. Tarlz +members containing a '..' component in the file name are skipped. Tarlz detects when the archive being created or enlarged is among the files to be dumped, appended or concatenated, and skips it. @@ -149,30 +157,31 @@ equivalent to '-1 --solid' Change to directory DIR. When creating or appending, the position of each '-C' option in the command line is significant; it will change the current working directory for the following FILES until - a new '-C' option appears in the command line. When extracting, all - the '-C' options are executed in sequence before starting the - extraction. Listing ignores any '-C' options specified. DIR is - relative to the then current working directory, perhaps changed by - a previous '-C' option. + a new '-C' option appears in the command line. When extracting or + comparing, all the '-C' options are executed in sequence before + reading the archive. Listing ignores any '-C' options specified. + DIR is relative to the then current working directory, perhaps + changed by a previous '-C' option. Note that a process can only have one current working directory (CWD). Therefore multi-threading can't be used to create an - archive if a '-C' option appears after a relative filename in the + archive if a '-C' option appears after a relative file name in the command line. '-d' '--diff' - Find differences between archive and file system. For each tar - member in the archive, verify that the corresponding file exists - and is of the same type (regular file, directory, etc). Report on - standard output the differences found in type, mode (permissions), - owner and group IDs, modification time, file size, file contents - (of regular files), target (of symlinks) and device number (of - block/character special files). + Compare and report differences between archive and file system. + For each tar member in the archive, verify that the corresponding + file in the file system exists and is of the same type (regular + file, directory, etc). Report on standard output the differences + found in type, mode (permissions), owner and group IDs, + modification time, file size, file contents (of regular files), + target (of symlinks) and device number (of block/character special + files). As tarlz removes leading slashes from member names, the '-C' - option may be used in combination with '--diff' when absolute - filenames were used on archive creation: 'tarlz -C / -d'. + option may be used in combination with '--diff' when absolute file + names were used on archive creation: 'tarlz -C / -d'. Alternatively, tarlz may be run from the root directory to perform the comparison. @@ -184,15 +193,22 @@ equivalent to '-1 --solid' Delete the specified files and directories from an archive in place. It currently can delete only from uncompressed archives and from archives with individually compressed files ('--no-solid' - archives). To delete a directory without deleting the files under - it, use 'tarlz --delete -f foo --exclude='dir/*' dir'. Deleting in - place may be dangerous. A corrupt archive, a power cut, or an I/O - error may cause data loss. + archives). Note that files of about '--data-size' or larger are + compressed individually even if '--bsolid' is used, and can + therefore be deleted. Tarlz takes care to not delete a tar member + unless it is possible to do so. For example it won't try to delete + a tar member that is not individually compressed. To delete a + directory without deleting the files under it, use + 'tarlz --delete -f foo --exclude='dir/*' dir'. Deleting in place + may be dangerous. A corrupt archive, a power cut, or an I/O error + may cause data loss. '--exclude=PATTERN' Exclude files matching a shell pattern like '*.o'. A file is - considered to match if any component of the filename matches. For - example, '*.o' matches 'foo.o', 'foo.o/bar' and 'foo/bar.o'. + considered to match if any component of the file name matches. For + example, '*.o' matches 'foo.o', 'foo.o/bar' and 'foo/bar.o'. If + PATTERN contains a '/', it matches a corresponding '/' in the file + name. For example, 'foo/*.o' matches 'foo/bar.o'. '-f ARCHIVE' '--file=ARCHIVE' @@ -234,13 +250,15 @@ equivalent to '-1 --solid' Compressed members can't be appended to an uncompressed archive, nor vice versa. If the archive is compressed, it must be a multimember lzip file with the two end-of-file blocks plus any - zero padding contained in the last lzip member of the archive. - Appending works as follows; first the end-of-file blocks are - removed, then the new members are appended, and finally two new - end-of-file blocks are appended to the archive. If the archive is - uncompressed, tarlz parses and skips tar headers until it finds - the end-of-file blocks. Exit with status 0 without modifying the - archive if no FILES have been specified. + zero padding contained in the last lzip member of the archive. It + is possible to append files to an archive with a different + compression granularity. Appending works as follows; first the + end-of-file blocks are removed, then the new members are appended, + and finally two new end-of-file blocks are appended to the + archive. If the archive is uncompressed, tarlz parses and skips + tar headers until it finds the end-of-file blocks. Exit with + status 0 without modifying the archive if no FILES have been + specified. '-t' '--list' @@ -351,7 +369,7 @@ equivalent to '-1 --solid' that a corrupt 'GNU.crc32' keyword, for example 'GNU.crc33', is reported as a missing CRC instead of as a corrupt record. This misleading 'Missing CRC' message is the consequence of a flaw in - the posix pax format; i.e., the lack of a mandatory check sequence + the POSIX pax format; i.e., the lack of a mandatory check sequence in the extended records. *Note crc32::. '--out-slots=N' @@ -369,9 +387,24 @@ invalid input file, 3 for an internal consistency error (eg, bug) which caused tarlz to panic. -File: tarlz.info, Node: File format, Next: Amendments to pax format, Prev: Invoking tarlz, Up: Top +File: tarlz.info, Node: Portable character set, Next: File format, Prev: Invoking tarlz, Up: Top + +3 POSIX portable filename character set +*************************************** + +The set of characters from which portable file names are constructed. + + A B C D E F G H I J K L M N O P Q R S T U V W X Y Z + a b c d e f g h i j k l m n o p q r s t u v w x y z + 0 1 2 3 4 5 6 7 8 9 . _ - + + The last three characters are the period, underscore, and +hyphen-minus characters, respectively. + + +File: tarlz.info, Node: File format, Next: Amendments to pax format, Prev: Portable character set, Up: Top -3 File format +4 File format ************* In the diagram below, a box like this: @@ -393,7 +426,7 @@ sets). The members simply appear one after another in the file, with no additional information before, between, or after them. Each lzip member contains one or more tar members in a simplified -posix pax interchange format. The only pax typeflag value supported by +POSIX pax interchange format. The only pax typeflag value supported by tarlz (in addition to the typeflag values defined by the ustar format) is 'x'. The pax format is an extension on top of the ustar format that removes the size limitations of the ustar format. @@ -438,7 +471,7 @@ tar.lz +===============+=================================================+========+ -3.1 Pax header block +4.1 Pax header block ==================== The pax header block is identical to the ustar header block described @@ -492,7 +525,7 @@ conversion to UTF-8 nor any other transformation. swapping of two bytes. -3.2 Ustar header block +4.2 Ustar header block ====================== The ustar header block has a length of 512 bytes and is structured as @@ -519,11 +552,10 @@ prefix 345 155 All characters in the header block are coded using the ISO/IEC 646:1991 (ASCII) standard, except in fields storing names for files, users, and groups. For maximum portability between implementations, -names should only contain characters from the portable filename -character set. But if an implementation supports the use of characters -outside of '/' and the portable filename character set in names for -files, users, and groups, tarlz will use the byte values in these names -unmodified. +names should only contain characters from the portable character set. +But if an implementation supports the use of characters outside of '/' +and the portable character set in names for files, users, and groups, +tarlz will use the byte values in these names unmodified. The fields name, linkname, and prefix are null-terminated character strings except when all characters in the array contain non-null @@ -618,38 +650,45 @@ character. File: tarlz.info, Node: Amendments to pax format, Next: Multi-threaded tar, Prev: File format, Up: Top -4 The reasons for the differences with pax +5 The reasons for the differences with pax ****************************************** -Tarlz is meant to reliably detect invalid or corrupt metadata during -decoding, and to create safe archives where corrupt metadata can be -reliably detected. In order to achieve these goals, tarlz makes some -changes to the variant of the pax format that it uses. This chapter -describes these changes and the concrete reasons to implement them. +Tarlz creates safe archives that allow the reliable detection of +invalid or corrupt metadata during decoding even when the integrity +checking of lzip can't be used because the lzip members are only +decompressed partially, as it happens in parallel '--list' and +'--extract'. In order to achieve this goal, tarlz makes some changes to +the variant of the pax format that it uses. This chapter describes +these changes and the concrete reasons to implement them. -4.1 Add a CRC of the extended records +5.1 Add a CRC of the extended records ===================================== -The posix pax format has a serious flaw. The metadata stored in pax +The POSIX pax format has a serious flaw. The metadata stored in pax extended records are not protected by any kind of check sequence. -Corruption in a long filename may cause the extraction of the file in +Corruption in a long file name may cause the extraction of the file in the wrong place without warning. Corruption in a large file size may cause the truncation of the file or the appending of garbage to the file, both followed by a spurious warning about a corrupt header far from the place of the undetected corruption. - Metadata like filename and file size must be always protected in an + Metadata like file name and file size must be always protected in an archive format because of the adverse effects of undetected corruption in them, potentially much worse that undetected corruption in the data. Even more so in the case of pax because the amount of metadata it stores is potentially large, making undetected corruption more probable. + Headers and metadata must be protected separately from data because +the integrity checking of lzip may not be able to detect the corruption +before the metadata has been used, for example, to create a new file in +the wrong place. + Because of the above, tarlz protects the extended records with a CRC in a way compatible with standard tar tools. *Note key_crc32::. -4.2 Remove flawed backward compatibility +5.2 Remove flawed backward compatibility ======================================== In order to allow the extraction of pax archives by a tar utility @@ -657,12 +696,12 @@ conforming to the POSIX-2:1993 standard, POSIX.1-2008 recommends selecting extended header field values that allow such tar to create a regular file containing the extended header records as data. This approach is broken because if the extended header is needed because of -a long filename, the name and prefix fields will be unable to contain +a long file name, the name and prefix fields will be unable to contain the full pathname of the file. Therefore the files corresponding to both the extended header and the overridden ustar header will be -extracted using truncated filenames, perhaps overwriting existing files -or directories. It may be a security risk to extract a file with a -truncated filename. +extracted using truncated file names, perhaps overwriting existing +files or directories. It may be a security risk to extract a file with +a truncated file name. To avoid this problem, tarlz writes extended headers with all fields zeroed except size, chksum, typeflag, magic and version. This prevents @@ -672,28 +711,29 @@ overridden by extended records. If an extended header is required for any reason (for example a file size larger than 8 GiB or a link name longer than 100 bytes), tarlz -moves the filename also to the extended header to prevent an ustar tool -from trying to extract the file or link. This also makes easier during -parallel decoding the detection of a tar member split between two lzip -members at the boundary between the extended header and the ustar -header. +moves the file name also to the extended header to prevent an ustar +tool from trying to extract the file or link. This also makes easier +during parallel decoding the detection of a tar member split between +two lzip members at the boundary between the extended header and the +ustar header. -4.3 As simple as possible (but not simpler) +5.3 As simple as possible (but not simpler) =========================================== The tarlz format is mainly ustar. Extended pax headers are used only -when needed because the length of a filename or link name, or the size +when needed because the length of a file name or link name, or the size of a file exceed the limits of the ustar format. Adding extended headers to each member just to record subsecond timestamps seems -wasteful for a backup format. +wasteful for a backup format. Moreover, minimizing the overhead may +help recovering the archive with lziprecover in case of corruption. Global pax headers are tolerated, but not supported; they are parsed and ignored. Some operations may not behave as expected if the archive contains global headers. -4.4 Avoid misconversions to/from UTF-8 +5.4 Avoid misconversions to/from UTF-8 ====================================== There is no portable way to tell what charset a text string is coded @@ -705,7 +745,7 @@ this behavior will be adjusted with a command line option in the future. File: tarlz.info, Node: Multi-threaded tar, Next: Minimum archive sizes, Prev: Amendments to pax format, Up: Top -5 Limitations of parallel tar decoding +6 Limitations of parallel tar decoding ************************************** Safely decoding an arbitrary tar archive in parallel is impossible. For @@ -753,7 +793,7 @@ example listing the Silesia corpus on a dual core machine: File: tarlz.info, Node: Minimum archive sizes, Next: Examples, Prev: Multi-threaded tar, Up: Top -6 Minimum archive sizes required for multi-threaded block compression +7 Minimum archive sizes required for multi-threaded block compression ********************************************************************* When creating or appending to a compressed archive using multi-threaded @@ -791,7 +831,7 @@ Level File: tarlz.info, Node: Examples, Next: Problems, Prev: Minimum archive sizes, Up: Top -7 A small tutorial with examples +8 A small tutorial with examples ******************************** Example 1: Create a multimember compressed archive 'archive.tar.lz' @@ -850,7 +890,7 @@ Example 8: Copy the contents of directory 'sourcedir' to the directory File: tarlz.info, Node: Problems, Next: Concept index, Prev: Examples, Up: Top -8 Reporting bugs +9 Reporting bugs **************** There are probably bugs in tarlz. There are certainly errors and @@ -881,6 +921,9 @@ Concept index * invoking: Invoking tarlz. (line 6) * minimum archive sizes: Minimum archive sizes. (line 6) * options: Invoking tarlz. (line 6) +* parallel tar decoding: Multi-threaded tar. (line 6) +* portable character set: Portable character set. + (line 6) * usage: Invoking tarlz. (line 6) * version: Invoking tarlz. (line 6) @@ -888,20 +931,21 @@ Concept index Tag Table: Node: Top223 -Node: Introduction1086 -Node: Invoking tarlz3337 -Ref: --data-size5489 -Ref: --bsolid12172 -Node: File format15802 -Ref: key_crc3220622 -Node: Amendments to pax format26039 -Ref: crc3226580 -Ref: flawed-compat27605 -Node: Multi-threaded tar30128 -Node: Minimum archive sizes32667 -Node: Examples34800 -Node: Problems36517 -Node: Concept index37043 +Node: Introduction1155 +Node: Invoking tarlz3841 +Ref: --data-size6006 +Ref: --bsolid13287 +Node: Portable character set16917 +Node: File format17420 +Ref: key_crc3222248 +Node: Amendments to pax format27647 +Ref: crc3228304 +Ref: flawed-compat29564 +Node: Multi-threaded tar32198 +Node: Minimum archive sizes34737 +Node: Examples36870 +Node: Problems38587 +Node: Concept index39113 End Tag Table diff --git a/doc/tarlz.texi b/doc/tarlz.texi index 3d2f668..25acaf8 100644 --- a/doc/tarlz.texi +++ b/doc/tarlz.texi @@ -6,8 +6,8 @@ @finalout @c %**end of header -@set UPDATED 11 April 2019 -@set VERSION 0.15 +@set UPDATED 8 October 2019 +@set VERSION 0.16 @dircategory Data Compression @direntry @@ -37,6 +37,7 @@ This manual is for Tarlz (version @value{VERSION}, @value{UPDATED}). @menu * Introduction:: Purpose and features of tarlz * Invoking tarlz:: Command line interface +* Portable character set:: POSIX portable filename character set * File format:: Detailed format of the compressed archive * Amendments to pax format:: The reasons for the differences with pax * Multi-threaded tar:: Limitations of parallel tar decoding @@ -60,13 +61,19 @@ to copy, distribute and modify it. @uref{http://www.nongnu.org/lzip/tarlz.html,,Tarlz} is a massively parallel (multi-threaded) combined implementation of the tar archiver and the @uref{http://www.nongnu.org/lzip/lzip.html,,lzip} compressor. Tarlz creates, -lists and extracts archives in a simplified posix pax format compressed with -lzip, keeping the alignment between tar members and lzip members. This -method adds an indexed lzip layer on top of the tar archive, making it -possible to decode the archive safely in parallel. The resulting multimember -tar.lz archive is fully backward compatible with standard tar tools like GNU -tar, which treat it like any other tar.lz archive. Tarlz can append files to -the end of such compressed archives. +lists and extracts archives in a simplified and safer variant of the POSIX +pax format compressed with lzip, keeping the alignment between tar members +and lzip members. The resulting multimember tar.lz archive is fully backward +compatible with standard tar tools like GNU tar, which treat it like any +other tar.lz archive. Tarlz can append files to the end of such compressed +archives. + +Keeping the alignment between tar members and lzip members has two +advantages. It adds an indexed lzip layer on top of the tar archive, making +it possible to decode the archive safely in parallel. It also minimizes the +amount of data lost in case of corruption. Compressing a tar archive with +plzip may even double the amount of files lost for each lzip member damaged +because it does not keep the members aligned. Tarlz can create tar archives with five levels of compression granularity; per file (---no-solid), per block (---bsolid, default), per directory @@ -88,7 +95,7 @@ member), and unwanted members can be deleted from the archive. Just like an uncompressed tar archive. @item -It is a safe posix-style backup format. In case of corruption, +It is a safe POSIX-style backup format. In case of corruption, tarlz can extract all the undamaged members from the tar.lz archive, skipping over the damaged members, just like the standard (uncompressed) tar. Moreover, the option @samp{--keep-damaged} can be @@ -105,7 +112,9 @@ Tarlz protects the extended records with a CRC in a way compatible with standard tar tools. @xref{crc32}. Tarlz does not understand other tar formats like @samp{gnu}, @samp{oldgnu}, -@samp{star} or @samp{v7}. +@samp{star} or @samp{v7}. @w{@samp{tarlz -tf archive.tar.lz > /dev/null}} +can be used to verify that the format of the archive is compatible with +tarlz. @node Invoking tarlz @@ -126,10 +135,10 @@ All operations except @samp{--concatenate} operate on whole trees if any @var{file} is a directory. On archive creation or appending tarlz archives the files specified, but -removes from member names any leading and trailing slashes and any filename +removes from member names any leading and trailing slashes and any file name prefixes containing a @samp{..} component. On extraction, leading and trailing slashes are also removed from member names, and archive members -containing a @samp{..} component in the filename are skipped. Tarlz detects +containing a @samp{..} component in the file name are skipped. Tarlz detects when the archive being created or enlarged is among the files to be dumped, appended or concatenated, and skips it. @@ -179,30 +188,30 @@ Create a new archive from @var{files}. @item -C @var{dir} @itemx --directory=@var{dir} -Change to directory @var{dir}. When creating or appending, the position -of each @samp{-C} option in the command line is significant; it will -change the current working directory for the following @var{files} until -a new @samp{-C} option appears in the command line. When extracting, all -the @samp{-C} options are executed in sequence before starting the -extraction. Listing ignores any @samp{-C} options specified. @var{dir} -is relative to the then current working directory, perhaps changed by a +Change to directory @var{dir}. When creating or appending, the position of +each @samp{-C} option in the command line is significant; it will change the +current working directory for the following @var{files} until a new +@samp{-C} option appears in the command line. When extracting or comparing, +all the @samp{-C} options are executed in sequence before reading the +archive. Listing ignores any @samp{-C} options specified. @var{dir} is +relative to the then current working directory, perhaps changed by a previous @samp{-C} option. Note that a process can only have one current working directory (CWD). Therefore multi-threading can't be used to create an archive if a @samp{-C} -option appears after a relative filename in the command line. +option appears after a relative file name in the command line. @item -d @itemx --diff -Find differences between archive and file system. For each tar member in the -archive, verify that the corresponding file exists and is of the same type -(regular file, directory, etc). Report on standard output the differences -found in type, mode (permissions), owner and group IDs, modification time, -file size, file contents (of regular files), target (of symlinks) and device -number (of block/character special files). +Compare and report differences between archive and file system. For each tar +member in the archive, verify that the corresponding file in the file system +exists and is of the same type (regular file, directory, etc). Report on +standard output the differences found in type, mode (permissions), owner and +group IDs, modification time, file size, file contents (of regular files), +target (of symlinks) and device number (of block/character special files). As tarlz removes leading slashes from member names, the @samp{-C} option may -be used in combination with @samp{--diff} when absolute filenames were used +be used in combination with @samp{--diff} when absolute file names were used on archive creation: @w{@samp{tarlz -C / -d}}. Alternatively, tarlz may be run from the root directory to perform the comparison. @@ -213,16 +222,22 @@ useful when comparing an @samp{--anonymous} archive. @item --delete Delete the specified files and directories from an archive in place. It currently can delete only from uncompressed archives and from archives with -individually compressed files (@samp{--no-solid} archives). To delete a +individually compressed files (@samp{--no-solid} archives). Note that files +of about @samp{--data-size} or larger are compressed individually even if +@samp{--bsolid} is used, and can therefore be deleted. Tarlz takes care to +not delete a tar member unless it is possible to do so. For example it won't +try to delete a tar member that is not individually compressed. To delete a directory without deleting the files under it, use -@w{@code{tarlz --delete -f foo --exclude='dir/*' dir}}. Deleting in place +@w{@samp{tarlz --delete -f foo --exclude='dir/*' dir}}. Deleting in place may be dangerous. A corrupt archive, a power cut, or an I/O error may cause data loss. @item --exclude=@var{pattern} Exclude files matching a shell pattern like @samp{*.o}. A file is considered -to match if any component of the filename matches. For example, @samp{*.o} -matches @samp{foo.o}, @samp{foo.o/bar} and @samp{foo/bar.o}. +to match if any component of the file name matches. For example, @samp{*.o} +matches @samp{foo.o}, @samp{foo.o/bar} and @samp{foo/bar.o}. If +@var{pattern} contains a @samp{/}, it matches a corresponding @samp{/} in +the file name. For example, @samp{foo/*.o} matches @samp{foo/bar.o}. @item -f @var{archive} @itemx --file=@var{archive} @@ -261,12 +276,13 @@ Append files to the end of an archive. The archive must be a regular be appended to an uncompressed archive, nor vice versa. If the archive is compressed, it must be a multimember lzip file with the two end-of-file blocks plus any zero padding contained in the last lzip member of the -archive. Appending works as follows; first the end-of-file blocks are -removed, then the new members are appended, and finally two new end-of-file -blocks are appended to the archive. If the archive is uncompressed, tarlz -parses and skips tar headers until it finds the end-of-file blocks. Exit -with status 0 without modifying the archive if no @var{files} have been -specified. +archive. It is possible to append files to an archive with a different +compression granularity. Appending works as follows; first the end-of-file +blocks are removed, then the new members are appended, and finally two new +end-of-file blocks are appended to the archive. If the archive is +uncompressed, tarlz parses and skips tar headers until it finds the +end-of-file blocks. Exit with status 0 without modifying the archive if no +@var{files} have been specified. @item -t @itemx --list @@ -282,7 +298,7 @@ Verbosely list files processed. Extract files from an archive. If @var{files} are given, extract only the @var{files} given. Else extract all the files in the archive. To extract a directory without extracting the files under it, use -@w{@code{tarlz -xf foo --exclude='dir/*' dir}}. +@w{@samp{tarlz -xf foo --exclude='dir/*' dir}}. @item -0 .. -9 Set the compression level for @samp{--create} and @samp{--append}. The @@ -326,7 +342,7 @@ compressed data block must contain an integer number of tar members. Block compression is the default because it improves compression ratio for archives with many files smaller than the block size. This option allows tarlz revert to default behavior if, for example, it is invoked through an -alias like @code{tar='tarlz --solid'}. @xref{--data-size}, to set the target +alias like @samp{tar='tarlz --solid'}. @xref{--data-size}, to set the target block size. @item --dsolid @@ -374,7 +390,7 @@ When this option is used, tarlz detects any corruption in the extended records (only limited by CRC collisions). But note that a corrupt @samp{GNU.crc32} keyword, for example @samp{GNU.crc33}, is reported as a missing CRC instead of as a corrupt record. This misleading -@samp{Missing CRC} message is the consequence of a flaw in the posix pax +@samp{Missing CRC} message is the consequence of a flaw in the POSIX pax format; i.e., the lack of a mandatory check sequence in the extended records. @xref{crc32}. @@ -400,6 +416,22 @@ invalid input file, 3 for an internal consistency error (eg, bug) which caused tarlz to panic. +@node Portable character set +@chapter POSIX portable filename character set +@cindex portable character set + +The set of characters from which portable file names are constructed. + +@example +A B C D E F G H I J K L M N O P Q R S T U V W X Y Z +a b c d e f g h i j k l m n o p q r s t u v w x y z +0 1 2 3 4 5 6 7 8 9 . _ - +@end example + +The last three characters are the period, underscore, and hyphen-minus +characters, respectively. + + @node File format @chapter File format @cindex file format @@ -426,7 +458,7 @@ A tar.lz file consists of a series of lzip members (compressed data sets). The members simply appear one after another in the file, with no additional information before, between, or after them. -Each lzip member contains one or more tar members in a simplified posix +Each lzip member contains one or more tar members in a simplified POSIX pax interchange format. The only pax typeflag value supported by tarlz (in addition to the typeflag values defined by the ustar format) is @samp{x}. The pax format is an extension on top of the ustar format that @@ -506,7 +538,7 @@ extraction. @xref{flawed-compat}. The pax extended header data consists of one or more records, each of them constructed as follows:@* -@code{"%d %s=%s\n", <length>, <keyword>, <value>} +@samp{"%d %s=%s\n", <length>, <keyword>, <value>} The <length>, <blank>, <keyword>, <equals-sign>, and <newline> in the record must be limited to the portable character set. The <length> field @@ -577,11 +609,11 @@ shown in the following table. All lengths and offsets are in decimal. All characters in the header block are coded using the ISO/IEC 646:1991 (ASCII) standard, except in fields storing names for files, users, and -groups. For maximum portability between implementations, names should -only contain characters from the portable filename character set. But if -an implementation supports the use of characters outside of @samp{/} and -the portable filename character set in names for files, users, and -groups, tarlz will use the byte values in these names unmodified. +groups. For maximum portability between implementations, names should only +contain characters from the portable character set. But if an implementation +supports the use of characters outside of @samp{/} and the portable +character set in names for files, users, and groups, tarlz will use the byte +values in these names unmodified. The fields name, linkname, and prefix are null-terminated character strings except when all characters in the array contain non-null @@ -679,32 +711,39 @@ ustar by not requiring a terminating null character. @chapter The reasons for the differences with pax @cindex Amendments to pax format -Tarlz is meant to reliably detect invalid or corrupt metadata during -decoding, and to create safe archives where corrupt metadata can be reliably -detected. In order to achieve these goals, tarlz makes some changes to the -variant of the pax format that it uses. This chapter describes these changes -and the concrete reasons to implement them. +Tarlz creates safe archives that allow the reliable detection of invalid or +corrupt metadata during decoding even when the integrity checking of lzip +can't be used because the lzip members are only decompressed partially, as +it happens in parallel @samp{--list} and @samp{--extract}. In order to +achieve this goal, tarlz makes some changes to the variant of the pax format +that it uses. This chapter describes these changes and the concrete reasons +to implement them. @sp 1 @anchor{crc32} @section Add a CRC of the extended records -The posix pax format has a serious flaw. The metadata stored in pax extended +The POSIX pax format has a serious flaw. The metadata stored in pax extended records are not protected by any kind of check sequence. Corruption in a -long filename may cause the extraction of the file in the wrong place +long file name may cause the extraction of the file in the wrong place without warning. Corruption in a large file size may cause the truncation of the file or the appending of garbage to the file, both followed by a spurious warning about a corrupt header far from the place of the undetected corruption. -Metadata like filename and file size must be always protected in an archive +Metadata like file name and file size must be always protected in an archive format because of the adverse effects of undetected corruption in them, potentially much worse that undetected corruption in the data. Even more so in the case of pax because the amount of metadata it stores is potentially large, making undetected corruption more probable. -Because of the above, tarlz protects the extended records with a CRC in -a way compatible with standard tar tools. @xref{key_crc32}. +Headers and metadata must be protected separately from data because the +integrity checking of lzip may not be able to detect the corruption before +the metadata has been used, for example, to create a new file in the wrong +place. + +Because of the above, tarlz protects the extended records with a CRC in a +way compatible with standard tar tools. @xref{key_crc32}. @sp 1 @anchor{flawed-compat} @@ -714,12 +753,12 @@ In order to allow the extraction of pax archives by a tar utility conforming to the POSIX-2:1993 standard, POSIX.1-2008 recommends selecting extended header field values that allow such tar to create a regular file containing the extended header records as data. This approach is broken because if the -extended header is needed because of a long filename, the name and prefix +extended header is needed because of a long file name, the name and prefix fields will be unable to contain the full pathname of the file. Therefore the files corresponding to both the extended header and the overridden ustar -header will be extracted using truncated filenames, perhaps overwriting +header will be extracted using truncated file names, perhaps overwriting existing files or directories. It may be a security risk to extract a file -with a truncated filename. +with a truncated file name. To avoid this problem, tarlz writes extended headers with all fields zeroed except size, chksum, typeflag, magic and version. This prevents old tar @@ -729,8 +768,8 @@ extended records. If an extended header is required for any reason (for example a file size larger than @w{8 GiB} or a link name longer than 100 bytes), tarlz moves the -filename also to the extended header to prevent an ustar tool from trying to -extract the file or link. This also makes easier during parallel decoding +file name also to the extended header to prevent an ustar tool from trying +to extract the file or link. This also makes easier during parallel decoding the detection of a tar member split between two lzip members at the boundary between the extended header and the ustar header. @@ -738,10 +777,11 @@ between the extended header and the ustar header. @section As simple as possible (but not simpler) The tarlz format is mainly ustar. Extended pax headers are used only when -needed because the length of a filename or link name, or the size of a file +needed because the length of a file name or link name, or the size of a file exceed the limits of the ustar format. Adding extended headers to each member just to record subsecond timestamps seems wasteful for a backup -format. +format. Moreover, minimizing the overhead may help recovering the archive +with lziprecover in case of corruption. Global pax headers are tolerated, but not supported; they are parsed and ignored. Some operations may not behave as expected if the archive contains @@ -759,6 +799,7 @@ be adjusted with a command line option in the future. @node Multi-threaded tar @chapter Limitations of parallel tar decoding +@cindex parallel tar decoding Safely decoding an arbitrary tar archive in parallel is impossible. For example, if a tar archive containing another tar archive is decoded starting @@ -47,6 +47,7 @@ bool Exclude::excluded( const char * const filename ) while( *p ) { for( unsigned i = 0; i < patterns.size(); ++i ) + // ignore a trailing sequence starting with '/' in filename #ifdef FNM_LEADING_DIR if( fnmatch( patterns[i].c_str(), p, FNM_LEADING_DIR ) == 0 ) return true; #else @@ -32,9 +32,7 @@ #include <utime.h> #include <sys/stat.h> #include <sys/types.h> -#if defined(__GNU_LIBRARY__) #include <sys/sysmacros.h> // for major, minor, makedev -#endif #include <lzlib.h> #include "arg_parser.h" @@ -181,131 +179,6 @@ int archive_read( const char * const archive_namep, const int infd, } -enum { mode_string_size = 10, - group_string_size = 1 + uname_l + 1 + gname_l + 1 }; // 67 - -void format_mode_string( const Tar_header header, char buf[mode_string_size] ) - { - const Typeflag typeflag = (Typeflag)header[typeflag_o]; - - std::memcpy( buf, "----------", mode_string_size ); - switch( typeflag ) - { - case tf_regular: break; - case tf_link: buf[0] = 'h'; break; - case tf_symlink: buf[0] = 'l'; break; - case tf_chardev: buf[0] = 'c'; break; - case tf_blockdev: buf[0] = 'b'; break; - case tf_directory: buf[0] = 'd'; break; - case tf_fifo: buf[0] = 'p'; break; - case tf_hiperf: buf[0] = 'C'; break; - default: buf[0] = '?'; - } - const mode_t mode = parse_octal( header + mode_o, mode_l ); // 12 bits - const bool setuid = mode & S_ISUID; - const bool setgid = mode & S_ISGID; - const bool sticky = mode & S_ISVTX; - if( mode & S_IRUSR ) buf[1] = 'r'; - if( mode & S_IWUSR ) buf[2] = 'w'; - if( mode & S_IXUSR ) buf[3] = setuid ? 's' : 'x'; - else if( setuid ) buf[3] = 'S'; - if( mode & S_IRGRP ) buf[4] = 'r'; - if( mode & S_IWGRP ) buf[5] = 'w'; - if( mode & S_IXGRP ) buf[6] = setgid ? 's' : 'x'; - else if( setgid ) buf[6] = 'S'; - if( mode & S_IROTH ) buf[7] = 'r'; - if( mode & S_IWOTH ) buf[8] = 'w'; - if( mode & S_IXOTH ) buf[9] = sticky ? 't' : 'x'; - else if( sticky ) buf[9] = 'T'; - } - - -int format_user_group_string( const Tar_header header, - char buf[group_string_size] ) - { - int len; - if( header[uname_o] && header[gname_o] ) - len = snprintf( buf, group_string_size, - " %.32s/%.32s", header + uname_o, header + gname_o ); - else - { - const unsigned uid = parse_octal( header + uid_o, uid_l ); - const unsigned gid = parse_octal( header + gid_o, gid_l ); - len = snprintf( buf, group_string_size, " %u/%u", uid, gid ); - } - return len; - } - -} // end namespace - -bool block_is_zero( const uint8_t * const buf, const int size ) - { - for( int i = 0; i < size; ++i ) if( buf[i] != 0 ) return false; - return true; - } - - -bool format_member_name( const Extended & extended, const Tar_header header, - Resizable_buffer & rbuf, const bool long_format ) - { - if( long_format ) - { - format_mode_string( header, rbuf() ); - const int group_string_len = - format_user_group_string( header, rbuf() + mode_string_size ); - int offset = mode_string_size + group_string_len; - const time_t mtime = parse_octal( header + mtime_o, mtime_l ); // 33 bits - struct tm tms; - const struct tm * tm = localtime_r( &mtime, &tms ); - if( !tm ) - { time_t z = 0; tm = localtime_r( &z, &tms ); if( !tm ) tm = &tms; } - const Typeflag typeflag = (Typeflag)header[typeflag_o]; - const bool islink = ( typeflag == tf_link || typeflag == tf_symlink ); - const char * const link_string = !islink ? "" : - ( ( typeflag == tf_link ) ? " link to " : " -> " ); - if( typeflag == tf_chardev || typeflag == tf_blockdev ) - offset += snprintf( rbuf() + offset, rbuf.size() - offset, " %5u,%u", - (unsigned)parse_octal( header + devmajor_o, devmajor_l ), - (unsigned)parse_octal( header + devminor_o, devminor_l ) ); - else - offset += snprintf( rbuf() + offset, rbuf.size() - offset, " %9llu", - extended.file_size() ); - for( int i = 0; i < 2; ++i ) - { - const int len = snprintf( rbuf() + offset, rbuf.size() - offset, - " %4d-%02u-%02u %02u:%02u %s%s%s\n", - 1900 + tm->tm_year, 1 + tm->tm_mon, tm->tm_mday, - tm->tm_hour, tm->tm_min, extended.path().c_str(), - link_string, islink ? extended.linkpath().c_str() : "" ); - if( (int)rbuf.size() > len + offset ) break; - if( !rbuf.resize( len + offset + 1 ) ) return false; - } - } - else - { - if( rbuf.size() < extended.path().size() + 2 && - !rbuf.resize( extended.path().size() + 2 ) ) return false; - snprintf( rbuf(), rbuf.size(), "%s\n", extended.path().c_str() ); - } - return true; - } - - -bool show_member_name( const Extended & extended, const Tar_header header, - const int vlevel, Resizable_buffer & rbuf ) - { - if( verbosity >= vlevel ) - { - if( !format_member_name( extended, header, rbuf, verbosity > vlevel ) ) - { show_error( mem_msg ); return false; } - std::fputs( rbuf(), stdout ); - std::fflush( stdout ); - } - return true; - } - -namespace { - int skip_member( const char * const archive_namep, const int infd, const Extended & extended ) { @@ -498,7 +371,6 @@ int extract_member( const char * const archive_namep, const int infd, case tf_hiperf: outfd = open_outstream( filename ); if( outfd < 0 ) return 2; - chmod( filename, mode ); // ignore errors break; case tf_link: case tf_symlink: @@ -559,6 +431,9 @@ int extract_member( const char * const archive_namep, const int infd, return 2; } + if( typeflag == tf_regular || typeflag == tf_hiperf ) + fchmod( outfd, mode ); // ignore errors + const int bufsize = 32 * header_size; uint8_t buf[bufsize]; long long rest = extended.file_size(); @@ -597,30 +472,6 @@ int extract_member( const char * const archive_namep, const int infd, return 0; } -} // end namespace - - -// return true if dir is a parent directory of name -bool compare_prefix_dir( const char * const dir, const char * const name ) - { - int len = 0; - while( dir[len] && dir[len] == name[len] ) ++len; - return ( !dir[len] && len > 0 && ( dir[len-1] == '/' || name[len] == '/' ) ); - } - - -// compare two file names ignoring trailing slashes -bool compare_tslash( const char * const name1, const char * const name2 ) - { - const char * p = name1; - const char * q = name2; - while( *p && *p == *q ) { ++p; ++q; } - while( *p == '/' ) ++p; - while( *q == '/' ) ++q; - return ( !*p && !*q ); - } - -namespace { bool parse_records( const char * const archive_namep, const int infd, Extended & extended, const Tar_header header, @@ -638,54 +489,6 @@ bool parse_records( const char * const archive_namep, const int infd, } // end namespace -/* Returns the number of bytes really read. - If (returned value < size) and (errno == 0), means EOF was reached. -*/ -int readblock( const int fd, uint8_t * const buf, const int size ) - { - int sz = 0; - errno = 0; - while( sz < size ) - { - const int n = read( fd, buf + sz, size - sz ); - if( n > 0 ) sz += n; - else if( n == 0 ) break; // EOF - else if( errno != EINTR ) break; - errno = 0; - } - return sz; - } - - -/* Returns the number of bytes really written. - If (returned value < size), it is always an error. -*/ -int writeblock( const int fd, const uint8_t * const buf, const int size ) - { - int sz = 0; - errno = 0; - while( sz < size ) - { - const int n = write( fd, buf + sz, size - sz ); - if( n > 0 ) sz += n; - else if( n < 0 && errno != EINTR ) break; - errno = 0; - } - return sz; - } - - -unsigned long long parse_octal( const uint8_t * const ptr, const int size ) - { - unsigned long long result = 0; - int i = 0; - while( i < size && std::isspace( ptr[i] ) ) ++i; - for( ; i < size && ptr[i] >= '0' && ptr[i] <= '7'; ++i ) - { result <<= 3; result += ptr[i] - '0'; } - return result; - } - - int decode( const std::string & archive_name, const Arg_parser & parser, const int filenames, const int num_workers, const int debug_level, const Program_mode program_mode, const bool ignore_ids, @@ -36,6 +36,8 @@ #include "tarlz.h" +namespace { + // Returns the number of bytes really read. // If (returned value < size) and (errno == 0), means EOF was reached. // @@ -55,7 +57,7 @@ int preadblock( const int fd, uint8_t * const buf, const int size, return sz; } - +/* // Returns the number of bytes really written. // If (returned value < size), it is always an error. // @@ -73,183 +75,7 @@ int pwriteblock( const int fd, const uint8_t * const buf, const int size, } return sz; } - - -void xinit_mutex( pthread_mutex_t * const mutex ) - { - const int errcode = pthread_mutex_init( mutex, 0 ); - if( errcode ) - { show_error( "pthread_mutex_init", errcode ); cleanup_and_fail(); } - } - -void xinit_cond( pthread_cond_t * const cond ) - { - const int errcode = pthread_cond_init( cond, 0 ); - if( errcode ) - { show_error( "pthread_cond_init", errcode ); cleanup_and_fail(); } - } - - -void xdestroy_mutex( pthread_mutex_t * const mutex ) - { - const int errcode = pthread_mutex_destroy( mutex ); - if( errcode ) - { show_error( "pthread_mutex_destroy", errcode ); cleanup_and_fail(); } - } - -void xdestroy_cond( pthread_cond_t * const cond ) - { - const int errcode = pthread_cond_destroy( cond ); - if( errcode ) - { show_error( "pthread_cond_destroy", errcode ); cleanup_and_fail(); } - } - - -void xlock( pthread_mutex_t * const mutex ) - { - const int errcode = pthread_mutex_lock( mutex ); - if( errcode ) - { show_error( "pthread_mutex_lock", errcode ); cleanup_and_fail(); } - } - - -void xunlock( pthread_mutex_t * const mutex ) - { - const int errcode = pthread_mutex_unlock( mutex ); - if( errcode ) - { show_error( "pthread_mutex_unlock", errcode ); cleanup_and_fail(); } - } - - -void xwait( pthread_cond_t * const cond, pthread_mutex_t * const mutex ) - { - const int errcode = pthread_cond_wait( cond, mutex ); - if( errcode ) - { show_error( "pthread_cond_wait", errcode ); cleanup_and_fail(); } - } - - -void xsignal( pthread_cond_t * const cond ) - { - const int errcode = pthread_cond_signal( cond ); - if( errcode ) - { show_error( "pthread_cond_signal", errcode ); cleanup_and_fail(); } - } - - -void xbroadcast( pthread_cond_t * const cond ) - { - const int errcode = pthread_cond_broadcast( cond ); - if( errcode ) - { show_error( "pthread_cond_broadcast", errcode ); cleanup_and_fail(); } - } - - -bool check_skip_filename( const Arg_parser & parser, - std::vector< char > & name_pending, - const char * const filename, const int filenames ) - { - if( Exclude::excluded( filename ) ) return true; // skip excluded files - bool skip = filenames > 0; - if( skip ) - for( int i = 0; i < parser.arguments(); ++i ) - if( !parser.code( i ) && parser.argument( i ).size() ) - { - const char * const name = - remove_leading_dotslash( parser.argument( i ).c_str() ); - if( compare_prefix_dir( name, filename ) || - compare_tslash( name, filename ) ) - { skip = false; name_pending[i] = false; break; } - } - return skip; - } - - -/* Return value: 0 = OK, 1 = damaged member, 2 = fatal error. */ -int archive_read_lz( LZ_Decoder * const decoder, const int infd, - long long & file_pos, const long long member_end, - const long long cdata_size, uint8_t * const buf, - const int size, const char ** msg ) - { - int sz = 0; - - while( sz < size ) - { - const int rd = LZ_decompress_read( decoder, buf + sz, size - sz ); - if( rd < 0 ) - { *msg = LZ_strerror( LZ_decompress_errno( decoder ) ); return 1; } - if( rd == 0 && LZ_decompress_finished( decoder ) == 1 ) - { *msg = end_msg; return 2; } - sz += rd; - if( sz < size && LZ_decompress_write_size( decoder ) > 0 ) - { - const long long ibuf_size = 16384; // try 65536 - uint8_t ibuf[ibuf_size]; - const long long rest = ( file_pos < member_end ) ? - member_end - file_pos : cdata_size - file_pos; - const int rsize = std::min( LZ_decompress_write_size( decoder ), - (int)std::min( ibuf_size, rest ) ); - if( rsize <= 0 ) LZ_decompress_finish( decoder ); - else - { - const int rd = preadblock( infd, ibuf, rsize, file_pos ); - if( LZ_decompress_write( decoder, ibuf, rd ) != rd ) - internal_error( "library error (LZ_decompress_write)." ); - file_pos += rd; - if( rd < rsize ) - { - LZ_decompress_finish( decoder ); - if( errno ) { *msg = "Error reading archive"; return 2; } - } - } - } - } - return 0; - } - - -int parse_records_lz( LZ_Decoder * const decoder, const int infd, - long long & file_pos, const long long member_end, - const long long cdata_size, long long & data_pos, - Extended & extended, const Tar_header header, - Resizable_buffer & rbuf, const char ** msg, - const bool permissive ) - { - const long long edsize = parse_octal( header + size_o, size_l ); - const long long bufsize = round_up( edsize ); - if( edsize <= 0 || edsize >= 1LL << 33 || bufsize >= INT_MAX ) - return 1; // overflow or no extended data - if( !rbuf.resize( bufsize ) ) return 1; // extended records buffer - int retval = archive_read_lz( decoder, infd, file_pos, member_end, - cdata_size, (uint8_t *)rbuf(), bufsize, msg ); - if( retval == 0 ) - { if( extended.parse( rbuf(), edsize, permissive ) ) data_pos += bufsize; - else retval = 1; } - return retval; - } - - -int skip_member_lz( LZ_Decoder * const decoder, const int infd, - long long & file_pos, const long long member_end, - const long long cdata_size, long long & data_pos, - long long rest, const char ** msg ) - { - const int bufsize = 32 * header_size; - uint8_t buf[bufsize]; - while( rest > 0 ) // skip tar member - { - const int rsize = ( rest >= bufsize ) ? bufsize : rest; - const int ret = archive_read_lz( decoder, infd, file_pos, member_end, - cdata_size, buf, rsize, msg ); - if( ret != 0 ) return ret; - data_pos += rsize; - rest -= rsize; - } - return 0; - } - - -namespace { +*/ struct Packet // member name and metadata or error message { @@ -606,6 +432,91 @@ void muxer( const char * const archive_namep, Packet_courier & courier ) } // end namespace +/* Read 'size' decompressed bytes from the archive. + Return value: 0 = OK, 1 = damaged member, 2 = fatal error. */ +int archive_read_lz( LZ_Decoder * const decoder, const int infd, + long long & file_pos, const long long member_end, + const long long cdata_size, uint8_t * const buf, + const int size, const char ** msg ) + { + int sz = 0; + + while( sz < size ) + { + const int rd = LZ_decompress_read( decoder, buf + sz, size - sz ); + if( rd < 0 ) + { *msg = LZ_strerror( LZ_decompress_errno( decoder ) ); return 1; } + if( rd == 0 && LZ_decompress_finished( decoder ) == 1 ) + { *msg = end_msg; return 2; } + sz += rd; + if( sz < size && LZ_decompress_write_size( decoder ) > 0 ) + { + const long long ibuf_size = 16384; // try 65536 + uint8_t ibuf[ibuf_size]; + const long long rest = ( file_pos < member_end ) ? + member_end - file_pos : cdata_size - file_pos; + const int rsize = std::min( LZ_decompress_write_size( decoder ), + (int)std::min( ibuf_size, rest ) ); + if( rsize <= 0 ) LZ_decompress_finish( decoder ); + else + { + const int rd = preadblock( infd, ibuf, rsize, file_pos ); + if( LZ_decompress_write( decoder, ibuf, rd ) != rd ) + internal_error( "library error (LZ_decompress_write)." ); + file_pos += rd; + if( rd < rsize ) + { + LZ_decompress_finish( decoder ); + if( errno ) { *msg = "Error reading archive"; return 2; } + } + } + } + } + return 0; + } + + +int parse_records_lz( LZ_Decoder * const decoder, const int infd, + long long & file_pos, const long long member_end, + const long long cdata_size, long long & data_pos, + Extended & extended, const Tar_header header, + Resizable_buffer & rbuf, const char ** msg, + const bool permissive ) + { + const long long edsize = parse_octal( header + size_o, size_l ); + const long long bufsize = round_up( edsize ); + if( edsize <= 0 || edsize >= 1LL << 33 || bufsize >= INT_MAX ) + return 1; // overflow or no extended data + if( !rbuf.resize( bufsize ) ) return 1; // extended records buffer + int retval = archive_read_lz( decoder, infd, file_pos, member_end, + cdata_size, (uint8_t *)rbuf(), bufsize, msg ); + if( retval == 0 ) + { if( extended.parse( rbuf(), edsize, permissive ) ) data_pos += bufsize; + else retval = 1; } + return retval; + } + + +int skip_member_lz( LZ_Decoder * const decoder, const int infd, + long long & file_pos, const long long member_end, + const long long cdata_size, long long & data_pos, + long long rest, const char ** msg ) + { + const int bufsize = 32 * header_size; + uint8_t buf[bufsize]; + while( rest > 0 ) // skip tar member + { + const int rsize = ( rest >= bufsize ) ? bufsize : rest; + const int ret = archive_read_lz( decoder, infd, file_pos, member_end, + cdata_size, buf, rsize, msg ); + if( ret != 0 ) return ret; + data_pos += rsize; + rest -= rsize; + } + return 0; + } + + // init the courier, then start the workers and call the muxer. int list_lz( const char * const archive_namep, const Arg_parser & parser, std::vector< char > & name_pending, const Lzip_index & lzip_index, @@ -60,7 +60,7 @@ namespace { const char * const program_name = "tarlz"; const char * const program_year = "2019"; -const char * invocation_name = 0; +const char * invocation_name = program_name; // default value bool dereference = false; @@ -68,14 +68,17 @@ void show_help( const long num_online ) { std::printf( "Tarlz is a massively parallel (multi-threaded) combined implementation of\n" "the tar archiver and the lzip compressor. Tarlz creates, lists and extracts\n" - "archives in a simplified posix pax format compressed with lzip, keeping the\n" - "alignment between tar members and lzip members. This method adds an indexed\n" - "lzip layer on top of the tar archive, making it possible to decode the\n" - "archive safely in parallel. The resulting multimember tar.lz archive is\n" - "fully backward compatible with standard tar tools like GNU tar, which treat\n" - "it like any other tar.lz archive. Tarlz can append files to the end of such\n" - "compressed archives.\n" - "\nThe tarlz file format is a safe posix-style backup format. In case of\n" + "archives in a simplified and safer variant of the POSIX pax format\n" + "compressed with lzip, keeping the alignment between tar members and lzip\n" + "members. The resulting multimember tar.lz archive is fully backward\n" + "compatible with standard tar tools like GNU tar, which treat it like any\n" + "other tar.lz archive. Tarlz can append files to the end of such compressed\n" + "archives.\n" + "\nKeeping the alignment between tar members and lzip members has two\n" + "advantages. It adds an indexed lzip layer on top of the tar archive, making\n" + "it possible to decode the archive safely in parallel. It also minimizes the\n" + "amount of data lost in case of corruption.\n" + "\nThe tarlz file format is a safe POSIX-style backup format. In case of\n" "corruption, tarlz can extract all the undamaged members from the tar.lz\n" "archive, skipping over the damaged members, just like the standard\n" "(uncompressed) tar. Moreover, the option '--keep-damaged' can be used to\n" @@ -305,7 +308,7 @@ int main( const int argc, const char * const argv[] ) bool keep_damaged = false; bool missing_crc = false; bool permissive = false; - invocation_name = argv[0]; + if( argc > 0 ) invocation_name = argv[0]; if( LZ_version()[0] < '1' ) { show_error( "Bad library version. At least lzlib 1.0 is required." ); @@ -317,6 +317,31 @@ const char * const fv_msg3 = "Format violation: consecutive extended headers fou const char * const posix_msg = "This does not look like a POSIX tar archive."; const char * const posix_lz_msg = "This does not look like a POSIX tar.lz archive."; +// defined in common.cc +void xinit_mutex( pthread_mutex_t * const mutex ); +void xinit_cond( pthread_cond_t * const cond ); +void xdestroy_mutex( pthread_mutex_t * const mutex ); +void xdestroy_cond( pthread_cond_t * const cond ); +void xlock( pthread_mutex_t * const mutex ); +void xunlock( pthread_mutex_t * const mutex ); +void xwait( pthread_cond_t * const cond, pthread_mutex_t * const mutex ); +void xsignal( pthread_cond_t * const cond ); +void xbroadcast( pthread_cond_t * const cond ); +unsigned long long parse_octal( const uint8_t * const ptr, const int size ); +int readblock( const int fd, uint8_t * const buf, const int size ); +int writeblock( const int fd, const uint8_t * const buf, const int size ); + +// defined in common_decode.cc +class Arg_parser; +bool block_is_zero( const uint8_t * const buf, const int size ); +bool format_member_name( const Extended & extended, const Tar_header header, + Resizable_buffer & rbuf, const bool long_format ); +bool show_member_name( const Extended & extended, const Tar_header header, + const int vlevel, Resizable_buffer & rbuf ); +bool check_skip_filename( const Arg_parser & parser, + std::vector< char > & name_pending, + const char * const filename, const int filenames ); + // defined in create.cc enum Solidity { no_solid, bsolid, dsolid, asolid, solid }; extern int cl_owner; @@ -339,7 +364,6 @@ int final_exit_status( int retval, const bool show_msg = true ); unsigned ustar_chksum( const uint8_t * const header ); bool verify_ustar_chksum( const uint8_t * const header ); bool has_lz_ext( const std::string & name ); -class Arg_parser; int concatenate( const std::string & archive_name, const Arg_parser & parser, const int filenames ); int encode( const std::string & archive_name, const Arg_parser & parser, @@ -381,16 +405,6 @@ bool excluded( const char * const filename ); // defined in extract.cc enum Program_mode { m_none, m_append, m_concatenate, m_create, m_delete, m_diff, m_extract, m_list }; -bool block_is_zero( const uint8_t * const buf, const int size ); -bool format_member_name( const Extended & extended, const Tar_header header, - Resizable_buffer & rbuf, const bool long_format ); -bool show_member_name( const Extended & extended, const Tar_header header, - const int vlevel, Resizable_buffer & rbuf ); -bool compare_prefix_dir( const char * const dir, const char * const name ); -bool compare_tslash( const char * const name1, const char * const name2 ); -int readblock( const int fd, uint8_t * const buf, const int size ); -int writeblock( const int fd, const uint8_t * const buf, const int size ); -unsigned long long parse_octal( const uint8_t * const ptr, const int size ); int decode( const std::string & archive_name, const Arg_parser & parser, const int filenames, const int num_workers, const int debug_level, const Program_mode program_mode, const bool ignore_ids, @@ -398,18 +412,6 @@ int decode( const std::string & archive_name, const Arg_parser & parser, const bool permissive ); // defined in list_lz.cc -void xinit_mutex( pthread_mutex_t * const mutex ); -void xinit_cond( pthread_cond_t * const cond ); -void xdestroy_mutex( pthread_mutex_t * const mutex ); -void xdestroy_cond( pthread_cond_t * const cond ); -void xlock( pthread_mutex_t * const mutex ); -void xunlock( pthread_mutex_t * const mutex ); -void xwait( pthread_cond_t * const cond, pthread_mutex_t * const mutex ); -void xsignal( pthread_cond_t * const cond ); -void xbroadcast( pthread_cond_t * const cond ); -bool check_skip_filename( const Arg_parser & parser, - std::vector< char > & name_pending, - const char * const filename, const int filenames ); struct LZ_Decoder; int archive_read_lz( LZ_Decoder * const decoder, const int infd, long long & file_pos, const long long member_end, diff --git a/testsuite/check.sh b/testsuite/check.sh index 00a56cd..1b6f3f7 100755 --- a/testsuite/check.sh +++ b/testsuite/check.sh @@ -182,7 +182,7 @@ rm -f test.txt || framework_failure cmp "${in}" test.txt || test_failed $LINENO rm -f test.txt || framework_failure -# test3 reference files for diff +# test3 reference files for -t and -tv (list3, vlist3) "${TARLZ}" -tf "${test3}" > list3 || test_failed $LINENO "${TARLZ}" -tvf "${test3}" > vlist3 || test_failed $LINENO "${TARLZ}" -tf "${test3_lz}" > out || test_failed $LINENO @@ -195,6 +195,8 @@ rm -f out || framework_failure cat "${testdir}"/rfoo > cfoo || framework_failure cat "${testdir}"/rbar > cbar || framework_failure cat "${testdir}"/rbaz > cbaz || framework_failure + +# test --list and --extract test3 rm -f foo bar baz || framework_failure "${TARLZ}" -xf "${test3_lz}" --missing-crc || test_failed $LINENO cmp cfoo foo || test_failed $LINENO @@ -253,7 +255,7 @@ for i in "${test3dir}" "${test3dir_lz}" ; do rm -rf dir || framework_failure done -# --exclude +# test --extract --exclude "${TARLZ}" -xf "${test3}" --exclude='f*o' --exclude=baz || test_failed $LINENO [ ! -e foo ] || test_failed $LINENO cmp cbar bar || test_failed $LINENO @@ -288,7 +290,7 @@ rm -rf dir || framework_failure [ ! -e dir ] || test_failed $LINENO rm -rf dir || framework_failure -# eof +# test --list and --extract eof "${TARLZ}" -tvf "${testdir}"/test3_eof1.tar > out 2> /dev/null [ $? = 2 ] || test_failed $LINENO diff -u vlist3 out || test_failed $LINENO @@ -456,8 +458,8 @@ cmp out.tar.lz aout.tar.lz || test_failed $LINENO cmp "${in_tar_lz}" aout.tar.lz || test_failed $LINENO "${TARLZ}" -A "${in_tar_lz}" "${test3_lz}" > aout.tar.lz || test_failed $LINENO cmp out.tar.lz aout.tar.lz || test_failed $LINENO -cat "${eof_lz}" > aout.tar.lz || framework_failure # concatenate to empty archive -"${TARLZ}" -Aqf aout.tar.lz "${in_tar}" +cat "${eof_lz}" > aout.tar.lz || framework_failure +"${TARLZ}" -Aqf aout.tar.lz "${in_tar}" # concatenate to empty archive [ $? = 2 ] || test_failed $LINENO "${TARLZ}" -Af aout.tar.lz "${in_tar_lz}" "${test3_lz}" || test_failed $LINENO cmp out.tar.lz aout.tar.lz || test_failed $LINENO @@ -609,7 +611,7 @@ cmp cbaz dir1/baz || test_failed $LINENO rm -rf dir1 || framework_failure rm -f out.tar.lz aout.tar.lz || framework_failure -# --exclude +# test --create --exclude cat cfoo > foo || framework_failure cat cbar > bar || framework_failure cat cbaz > baz || framework_failure @@ -631,6 +633,31 @@ cmp cfoo foo || test_failed $LINENO [ ! -e baz ] || test_failed $LINENO rm -f out.tar foo bar baz || framework_failure +# test --diff +"${TARLZ}" -xf "${test3_lz}" || test_failed $LINENO +"${TARLZ}" --uncompressed -cf out.tar foo || test_failed $LINENO +"${TARLZ}" --uncompressed -cf aout.tar foo --anonymous || test_failed $LINENO +if cmp out.tar aout.tar > /dev/null ; then + printf "\nwarning: '--diff' test can't be run as root." +else + "${TARLZ}" -df "${test3_lz}" > /dev/null + [ $? = 1 ] || test_failed $LINENO + "${TARLZ}" -df "${test3_lz}" --ignore-ids || test_failed $LINENO + "${TARLZ}" -df "${test3_lz}" --exclude '*' || test_failed $LINENO + "${TARLZ}" -df "${in_tar_lz}" --exclude '*' || test_failed $LINENO + rm -f bar || framework_failure + "${TARLZ}" -df "${test3_lz}" foo baz --ignore-ids || test_failed $LINENO + "${TARLZ}" -df "${test3_lz}" --exclude bar --ignore-ids || + test_failed $LINENO + rm -f foo baz || framework_failure + "${TARLZ}" -q -xf "${test3dir_lz}" || test_failed $LINENO + "${TARLZ}" -q -df "${test3dir_lz}" --ignore-ids || test_failed $LINENO + "${TARLZ}" -q -df "${test3dir_lz}" dir --ignore-ids || test_failed $LINENO + "${TARLZ}" -df "${test3_lz}" --ignore-ids -C dir || test_failed $LINENO + rm -rf dir || framework_failure +fi +rm -f out.tar aout.tar foo bar baz || framework_failure + # test --delete for e in "" .lz ; do "${TARLZ}" -A "${in_tar}"$e "${test3}"$e > out.tar$e || test_failed $LINENO $e @@ -694,6 +721,10 @@ cat "${in}" > test.txt || framework_failure "${TARLZ}" -0 -cf out.tar.lz foo bar baz --asolid || test_failed $LINENO "${TARLZ}" -0 -rf out.tar.lz test.txt || test_failed $LINENO rm -f foo bar baz test.txt || framework_failure +for i in foo bar baz ; do + "${TARLZ}" -qf out.tar.lz --delete $i + [ $? = 2 ] || test_failed $LINENO +done "${TARLZ}" -f out.tar.lz --delete test.txt || test_failed $LINENO "${TARLZ}" -xf out.tar.lz || test_failed $LINENO cmp cfoo foo || test_failed $LINENO @@ -732,7 +763,7 @@ else printf "\nwarning: skipping link test: 'ln' does not work on your system." fi rm -f dummy_slink dummy_link dummy_file || framework_failure - +# if [ "${ln_works}" = yes ] ; then mkdir dir || framework_failure cat cfoo > dir/foo || framework_failure @@ -762,7 +793,7 @@ if [ "${ln_works}" = yes ] ; then done fi -# test --append +# test --append compressed cat cfoo > foo || framework_failure cat cbar > bar || framework_failure cat cbaz > baz || framework_failure @@ -801,7 +832,7 @@ cmp out.tar.lz aout.tar.lz || test_failed $LINENO cmp out.tar.lz aout.tar.lz || test_failed $LINENO rm -f out.tar.lz aout.tar.lz || framework_failure -# --uncompressed +# test --append --uncompressed "${TARLZ}" --un -cf out.tar foo bar baz || test_failed $LINENO "${TARLZ}" --un -cf aout.tar foo || test_failed $LINENO "${TARLZ}" --un -rf aout.tar foo bar baz --exclude foo || test_failed $LINENO @@ -837,7 +868,7 @@ cmp out.tar aout.tar || test_failed $LINENO cmp out.tar aout.tar || test_failed $LINENO rm -f out.tar aout.tar || framework_failure -# append to solid archive +# test --append to solid archive "${TARLZ}" --solid -q -0 -cf out.tar.lz "${in}" foo bar || test_failed $LINENO "${TARLZ}" -q -tf out.tar.lz || test_failed $LINENO # compressed seekable cat out.tar.lz > aout.tar.lz || framework_failure @@ -863,31 +894,7 @@ for i in --asolid --bsolid --dsolid -0 ; do done rm -f foo bar baz || framework_failure -# test --diff -"${TARLZ}" -xf "${test3_lz}" || test_failed $LINENO -"${TARLZ}" --uncompressed -cf out.tar foo || test_failed $LINENO -"${TARLZ}" --uncompressed -cf aout.tar foo --anonymous || test_failed $LINENO -if cmp out.tar aout.tar > /dev/null ; then - printf "\nwarning: --diff test can't be run as root." -else - "${TARLZ}" -df "${test3_lz}" > /dev/null - [ $? = 1 ] || test_failed $LINENO - "${TARLZ}" -df "${test3_lz}" --ignore-ids || test_failed $LINENO - "${TARLZ}" -df "${test3_lz}" --exclude '*' || test_failed $LINENO - "${TARLZ}" -df "${in_tar_lz}" --exclude '*' || test_failed $LINENO - rm -f bar || framework_failure - "${TARLZ}" -df "${test3_lz}" foo baz --ignore-ids || test_failed $LINENO - "${TARLZ}" -df "${test3_lz}" --exclude bar --ignore-ids || - test_failed $LINENO - rm -f foo baz || framework_failure - "${TARLZ}" -q -xf "${test3dir_lz}" || test_failed $LINENO - "${TARLZ}" -q -df "${test3dir_lz}" --ignore-ids || test_failed $LINENO - "${TARLZ}" -q -df "${test3dir_lz}" dir --ignore-ids || test_failed $LINENO - rm -rf dir || framework_failure -fi -rm -f out.tar aout.tar foo bar baz || framework_failure - -# test directories and links +# test -c -d -x on directories and links mkdir dir1 || framework_failure "${TARLZ}" -0 -cf out.tar.lz dir1 || test_failed $LINENO rmdir dir1 || framework_failure @@ -984,6 +991,7 @@ rm -f foo || framework_failure printf "\ntesting bad input..." +# test --extract ".." mkdir dir1 || framework_failure cd dir1 || framework_failure "${TARLZ}" -q -xf "${testdir}"/dotdot1.tar.lz || test_failed $LINENO @@ -999,6 +1007,7 @@ cd dir1 || framework_failure cd .. || framework_failure rm -rf dir1 || framework_failure +# test --list and --extract truncated tar dd if="${in_tar}" of=truncated.tar bs=1000 count=1 2> /dev/null "${TARLZ}" -q -tf truncated.tar > /dev/null [ $? = 2 ] || test_failed $LINENO @@ -1024,7 +1033,7 @@ for i in 1 2 3 4 ; do rm -f out.tar.lz foo bar baz || framework_failure done -# test format violations +# test --list and --extract format violations if [ "${ln_works}" = yes ] ; then mkdir dir1 || framework_failure "${TARLZ}" -C dir1 -xf "${t155}" || test_failed $LINENO @@ -1059,7 +1068,7 @@ if [ "${ln_works}" = yes ] ; then rm -rf dir1 || framework_failure fi -# test compressed and --keep-damaged +# test --extract and --keep-damaged compressed rm -f test.txt || framework_failure for i in "${inbad1}" "${inbad2}" ; do "${TARLZ}" -q -xf "${i}.tar.lz" @@ -1128,7 +1137,7 @@ cmp cfoo foo || test_failed $LINENO cmp cbar bar || test_failed $LINENO cmp cbaz baz || test_failed $LINENO -# test uncompressed and --keep-damaged +# test --extract and --keep-damaged uncompressed rm -f test.txt || framework_failure "${TARLZ}" -q -xf "${inbad1}.tar" [ $? = 2 ] || test_failed $LINENO |