diff options
Diffstat (limited to 'unzcrash.cc')
| -rw-r--r-- | unzcrash.cc | 117 |
1 files changed, 95 insertions, 22 deletions
diff --git a/unzcrash.cc b/unzcrash.cc index 3970638..9a32b82 100644 --- a/unzcrash.cc +++ b/unzcrash.cc @@ -1,6 +1,6 @@ /* Unzcrash - Tests robustness of decompressors to corrupted data. Inspired by unzcrash.c from Julian Seward's bzip2. - Copyright (C) 2008-2016 Antonio Diaz Diaz. + Copyright (C) 2008-2017 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 @@ -40,12 +40,16 @@ #error "Environments where CHAR_BIT != 8 are not supported." #endif +#ifndef INT64_MAX +#define INT64_MAX 0x7FFFFFFFFFFFFFFFLL +#endif + namespace { const char * const Program_name = "Unzcrash"; const char * const program_name = "unzcrash"; -const char * const program_year = "2016"; +const char * const program_year = "2017"; const char * invocation_name = 0; int verbosity = 0; @@ -55,14 +59,27 @@ void show_help() { std::printf( "%s - Tests robustness of decompressors to corrupted data.\n", Program_name ); std::printf( "\nUsage: %s [options] \"lzip -tv\" filename.lz\n", invocation_name ); - std::printf( "\nThis program reads the specified file and then repeatedly decompresses\n" - "it, increasing 256 times each byte of the compressed data, so as to test\n" - "all possible one-byte errors. This should not cause any invalid memory\n" - "accesses. If it does, please, report it as a bug.\n" + std::printf( "\nBy default, unzcrash reads the specified file and then repeatedly\n" + "decompresses it, increasing 256 times each byte of the compressed data,\n" + "so as to test all possible one-byte errors.\n" + "\nIf the '--block' option is given, unzcrash reads the specified file\n" + "and then repeatedly decompresses it, setting all bytes in each\n" + "successive block to the value given, so as to test all possible full\n" + "sector errors.\n" + "\nIf the '--truncate' option is given, unzcrash reads the specified\n" + "file and then repeatedly decompresses it, truncating the file to\n" + "increasing lengths, so as to test all possible truncation points.\n" + "\nNone of the three test modes described above should cause any invalid\n" + "memory accesses. If any of them does, please, report it as a bug to the\n" + "maintainers of the decompressor being tested.\n" "\nIf the decompressor returns with zero status, unzcrash compares the\n" "output of the decompressor for the original and corrupt files. If the\n" - "outputs differ, it means that the decompressor failed to recognize the\n" - "corruption and produced garbage output. Please, report it as a bug.\n" + "outputs differ, it means that the decompressor returned a false\n" + "negative; it failed to recognize the corruption and produced garbage\n" + "output. The only exception is when a multimember file is truncated just\n" + "after the last byte of a member, producing a shorter but valid\n" + "compressed file. Except in this latter case, please, report any false\n" + "negative as a bug.\n" "\nIn order to compare the outputs, unzcrash needs a zcmp program able to\n" "understand the format being tested. For example the one provided by zutils.\n" "Use '--zcmp=false' to disable comparisons.\n" @@ -72,6 +89,7 @@ void show_help() " -b, --bits=<range> test N-bit errors instead of full byte\n" " -B, --block[=<size>][,<val>] test blocks of given size [512,0]\n" " -d, --delta=<n> test one of every n bytes/blocks/truncations\n" + " -e, --set-byte=<pos>,<val> set byte at position <pos> to value <val>\n" " -p, --position=<bytes> first byte position to test [default 0]\n" " -q, --quiet suppress all messages\n" " -s, --size=<bytes> number of byte positions to test [all]\n" @@ -124,12 +142,13 @@ void internal_error( const char * const msg ) } -long getnum( const char * const ptr, const long llimit, const long ulimit, - const bool comma = false ) +long long getnum( const char * const ptr, const long long llimit = -LLONG_MAX, + const long long ulimit = LLONG_MAX, + const char ** const tailp = 0 ) { char * tail; errno = 0; - long result = strtol( ptr, &tail, 0 ); + long long result = strtoll( ptr, &tail, 0 ); if( tail == ptr ) { show_error( "Bad or missing numerical argument.", 0, true ); @@ -138,11 +157,14 @@ long getnum( const char * const ptr, const long llimit, const long ulimit, if( !errno && tail[0] ) { - const int factor = ( tail[1] == 'i' ) ? 1024 : 1000; + char * const p = tail++; + int factor; + bool bsuf; // 'B' suffix is present + if( tail[0] == 'i' ) { ++tail; factor = 1024; } else factor = 1000; + if( tail[0] == 'B' ) { ++tail; bsuf = true; } else bsuf = false; int exponent = -1; // -1 = bad multiplier - switch( tail[0] ) + switch( *p ) { - case ',': if( comma ) exponent = 0; break; case 'Y': exponent = 8; break; case 'Z': exponent = 7; break; case 'E': exponent = 6; break; @@ -152,6 +174,8 @@ long getnum( const char * const ptr, const long llimit, const long ulimit, case 'M': exponent = 2; break; case 'K': if( factor == 1024 ) exponent = 1; break; case 'k': if( factor == 1000 ) exponent = 1; break; + case 'B': if( factor == 1000 && !bsuf ) exponent = 0; break; + default : if( tailp ) { tail = p; exponent = 0; } break; } if( exponent < 0 ) { @@ -160,7 +184,7 @@ long getnum( const char * const ptr, const long llimit, const long ulimit, } for( int i = 0; i < exponent; ++i ) { - if( LONG_MAX / factor >= std::labs( result ) ) result *= factor; + if( LLONG_MAX / factor >= std::labs( result ) ) result *= factor; else { errno = ERANGE; break; } } } @@ -170,23 +194,64 @@ long getnum( const char * const ptr, const long llimit, const long ulimit, show_error( "Numerical argument out of limits." ); std::exit( 1 ); } + if( tailp ) *tailp = tail; return result; } void parse_block( const char * const ptr, long & size, uint8_t & value ) { - const char * const ptr2 = std::strchr( ptr, ',' ); + const char * tail = ptr; + + if( tail[0] != ',' ) + size = getnum( ptr, 1, INT_MAX, &tail ); + if( tail[0] == ',' ) + value = getnum( tail + 1, 0, 255 ); + else if( tail[0] ) + { + show_error( "Bad separator in argument of '--block'", 0, true ); + std::exit( 1 ); + } + } + + +struct Bad_byte + { + enum Mode { literal, delta, flip }; + long long pos; + Mode mode; + uint8_t value; - if( !ptr2 || ptr2 != ptr ) - size = getnum( ptr, 1, INT_MAX, true ); - if( ptr2 ) - value = getnum( ptr2 + 1, 0, 255 ); + Bad_byte() : pos( -1 ), mode( literal ), value( 0 ) {} + uint8_t operator()( const uint8_t old_value ) const + { + if( mode == delta ) return old_value + value; + if( mode == flip ) return old_value ^ value; + return value; + } + }; + + +// Recognized formats: <pos>,<value> <pos>,+<value> <pos>,f<value> +// +void parse_pos_value( const char * const ptr, Bad_byte & bad_byte ) + { + const char * tail; + bad_byte.pos = getnum( ptr, 0, INT64_MAX, &tail ); + if( tail[0] != ',' ) + { + show_error( "Bad separator between <pos> and <val>.", 0, true ); + std::exit( 1 ); + } + if( tail[1] == '+' ) { ++tail; bad_byte.mode = Bad_byte::delta; } + else if( tail[1] == 'f' ) { ++tail; bad_byte.mode = Bad_byte::flip; } + else bad_byte.mode = Bad_byte::literal; + bad_byte.value = getnum( tail + 1, 0, 255 ); } /* Returns the address of a malloc'd buffer containing the file data and - its size in '*size'. + the file size in '*size'. In case of error, returns 0 and does not modify '*size'. */ uint8_t * read_file( const char * const name, long * const size ) @@ -309,6 +374,7 @@ int main( const int argc, const char * const argv[] ) enum Mode { m_block, m_byte, m_truncate }; const char * mode_str[3] = { "block", "byte", "size" }; Bitset8 bits; // if Bitset8::parse not called test full byte + Bad_byte bad_byte; const char * zcmp_program = "zcmp"; long pos = 0; long max_size = LONG_MAX; @@ -324,6 +390,7 @@ int main( const int argc, const char * const argv[] ) { 'b', "bits", Arg_parser::yes }, { 'B', "block", Arg_parser::maybe }, { 'd', "delta", Arg_parser::yes }, + { 'e', "set-byte", Arg_parser::yes }, { 'p', "position", Arg_parser::yes }, { 'q', "quiet", Arg_parser::no }, { 's', "size", Arg_parser::yes }, @@ -331,7 +398,7 @@ int main( const int argc, const char * const argv[] ) { 'v', "verbose", Arg_parser::no }, { 'V', "version", Arg_parser::no }, { 'z', "zcmp", Arg_parser::yes }, - { 0 , 0, Arg_parser::no } }; + { 0 , 0, Arg_parser::no } }; const Arg_parser parser( argc, argv, options ); if( parser.error().size() ) // bad option @@ -350,6 +417,7 @@ int main( const int argc, const char * const argv[] ) case 'B': if( arg[0] ) parse_block( arg, block_size, block_value ); program_mode = m_block; break; case 'd': delta = getnum( arg, 1, INT_MAX ); break; + case 'e': parse_pos_value( arg, bad_byte ); break; case 'p': pos = getnum( arg, -LONG_MAX, LONG_MAX ); break; case 'q': verbosity = -1; break; case 's': max_size = getnum( arg, -LONG_MAX, LONG_MAX ); break; @@ -414,6 +482,11 @@ int main( const int argc, const char * const argv[] ) { show_error( "Nothing to do; domain is empty." ); return 0; } if( max_size < 0 ) max_size += file_size - pos; const long end = ( ( max_size < file_size - pos ) ? pos + max_size : file_size ); + if( bad_byte.pos >= file_size ) + { show_error( "Position of '--set-byte' is beyond end of file." ); + return 1; } + if( bad_byte.pos >= 0 ) + buffer[bad_byte.pos] = bad_byte( buffer[bad_byte.pos] ); long positions = 0, decompressions = 0, successes = 0, failed_comparisons = 0; if( program_mode == m_truncate ) for( long i = pos; i < end; i += std::min( delta, end - i ) ) |