Gurchetan Singh | 876877a | 2023-08-07 15:15:58 -0700 | [diff] [blame] | 1 | #!/usr/bin/env python3 |
| 2 | |
| 3 | import argparse |
| 4 | import os |
| 5 | import platform |
| 6 | import subprocess |
| 7 | |
| 8 | # This list contains symbols that _might_ be exported for some platforms |
| 9 | PLATFORM_SYMBOLS = [ |
| 10 | '__bss_end__', |
| 11 | '__bss_start__', |
| 12 | '__bss_start', |
| 13 | '__cxa_guard_abort', |
| 14 | '__cxa_guard_acquire', |
| 15 | '__cxa_guard_release', |
| 16 | '__cxa_allocate_dependent_exception', |
| 17 | '__cxa_allocate_exception', |
| 18 | '__cxa_begin_catch', |
| 19 | '__cxa_call_unexpected', |
| 20 | '__cxa_current_exception_type', |
| 21 | '__cxa_current_primary_exception', |
| 22 | '__cxa_decrement_exception_refcount', |
| 23 | '__cxa_deleted_virtual', |
| 24 | '__cxa_demangle', |
| 25 | '__cxa_end_catch', |
| 26 | '__cxa_free_dependent_exception', |
| 27 | '__cxa_free_exception', |
| 28 | '__cxa_get_exception_ptr', |
| 29 | '__cxa_get_globals', |
| 30 | '__cxa_get_globals_fast', |
| 31 | '__cxa_increment_exception_refcount', |
| 32 | '__cxa_new_handler', |
| 33 | '__cxa_pure_virtual', |
| 34 | '__cxa_rethrow', |
| 35 | '__cxa_rethrow_primary_exception', |
| 36 | '__cxa_terminate_handler', |
| 37 | '__cxa_throw', |
| 38 | '__cxa_uncaught_exception', |
| 39 | '__cxa_uncaught_exceptions', |
| 40 | '__cxa_unexpected_handler', |
| 41 | '__dynamic_cast', |
| 42 | '__emutls_get_address', |
| 43 | '__gxx_personality_v0', |
| 44 | '__end__', |
| 45 | '__odr_asan._glapi_Context', |
| 46 | '__odr_asan._glapi_Dispatch', |
| 47 | '_bss_end__', |
| 48 | '_edata', |
| 49 | '_end', |
| 50 | '_fini', |
| 51 | '_init', |
| 52 | '_fbss', |
| 53 | '_fdata', |
| 54 | '_ftext', |
| 55 | ] |
| 56 | |
| 57 | def get_symbols_nm(nm, lib): |
| 58 | ''' |
| 59 | List all the (non platform-specific) symbols exported by the library |
| 60 | using `nm` |
| 61 | ''' |
| 62 | symbols = [] |
| 63 | platform_name = platform.system() |
| 64 | output = subprocess.check_output([nm, '-gP', lib], |
| 65 | stderr=open(os.devnull, 'w')).decode("ascii") |
| 66 | for line in output.splitlines(): |
| 67 | fields = line.split() |
| 68 | if len(fields) == 2 or fields[1] == 'U': |
| 69 | continue |
| 70 | symbol_name = fields[0] |
| 71 | if platform_name == 'Linux' or platform_name == 'GNU' or platform_name.startswith('GNU/'): |
| 72 | if symbol_name in PLATFORM_SYMBOLS: |
| 73 | continue |
| 74 | elif platform_name == 'Darwin': |
| 75 | assert symbol_name[0] == '_' |
| 76 | symbol_name = symbol_name[1:] |
| 77 | symbols.append(symbol_name) |
| 78 | return symbols |
| 79 | |
| 80 | |
| 81 | def get_symbols_dumpbin(dumpbin, lib): |
| 82 | ''' |
| 83 | List all the (non platform-specific) symbols exported by the library |
| 84 | using `dumpbin` |
| 85 | ''' |
| 86 | symbols = [] |
| 87 | output = subprocess.check_output([dumpbin, '/exports', lib], |
| 88 | stderr=open(os.devnull, 'w')).decode("ascii") |
| 89 | for line in output.splitlines(): |
| 90 | fields = line.split() |
| 91 | # The lines with the symbols are made of at least 4 columns; see details below |
| 92 | if len(fields) < 4: |
| 93 | continue |
| 94 | try: |
| 95 | # Making sure the first 3 columns are a dec counter, a hex counter |
| 96 | # and a hex address |
| 97 | _ = int(fields[0], 10) |
| 98 | _ = int(fields[1], 16) |
| 99 | _ = int(fields[2], 16) |
| 100 | except ValueError: |
| 101 | continue |
| 102 | symbol_name = fields[3] |
| 103 | # De-mangle symbols |
| 104 | if symbol_name[0] == '_' and '@' in symbol_name: |
| 105 | symbol_name = symbol_name[1:].split('@')[0] |
| 106 | symbols.append(symbol_name) |
| 107 | return symbols |
| 108 | |
| 109 | |
| 110 | def main(): |
| 111 | parser = argparse.ArgumentParser() |
| 112 | parser.add_argument('--symbols-file', |
| 113 | action='store', |
| 114 | required=True, |
| 115 | help='path to file containing symbols') |
| 116 | parser.add_argument('--lib', |
| 117 | action='store', |
| 118 | required=True, |
| 119 | help='path to library') |
| 120 | parser.add_argument('--nm', |
| 121 | action='store', |
| 122 | help='path to binary (or name in $PATH)') |
| 123 | parser.add_argument('--dumpbin', |
| 124 | action='store', |
| 125 | help='path to binary (or name in $PATH)') |
| 126 | parser.add_argument('--ignore-symbol', |
| 127 | action='append', |
| 128 | help='do not process this symbol') |
| 129 | args = parser.parse_args() |
| 130 | |
| 131 | try: |
| 132 | if platform.system() == 'Windows': |
| 133 | if not args.dumpbin: |
| 134 | parser.error('--dumpbin is mandatory') |
| 135 | lib_symbols = get_symbols_dumpbin(args.dumpbin, args.lib) |
| 136 | else: |
| 137 | if not args.nm: |
| 138 | parser.error('--nm is mandatory') |
| 139 | lib_symbols = get_symbols_nm(args.nm, args.lib) |
| 140 | except: |
| 141 | # We can't run this test, but we haven't technically failed it either |
| 142 | # Return the GNU "skip" error code |
| 143 | exit(77) |
| 144 | mandatory_symbols = [] |
| 145 | optional_symbols = [] |
| 146 | with open(args.symbols_file) as symbols_file: |
| 147 | qualifier_optional = '(optional)' |
| 148 | for line in symbols_file.readlines(): |
| 149 | |
| 150 | # Strip comments |
| 151 | line = line.split('#')[0] |
| 152 | line = line.strip() |
| 153 | if not line: |
| 154 | continue |
| 155 | |
| 156 | # Line format: |
| 157 | # [qualifier] symbol |
| 158 | qualifier = None |
| 159 | symbol = None |
| 160 | |
| 161 | fields = line.split() |
| 162 | if len(fields) == 1: |
| 163 | symbol = fields[0] |
| 164 | elif len(fields) == 2: |
| 165 | qualifier = fields[0] |
| 166 | symbol = fields[1] |
| 167 | else: |
| 168 | print(args.symbols_file + ': invalid format: ' + line) |
| 169 | exit(1) |
| 170 | |
| 171 | # The only supported qualifier is 'optional', which means the |
| 172 | # symbol doesn't have to be exported by the library |
| 173 | if qualifier and not qualifier == qualifier_optional: |
| 174 | print(args.symbols_file + ': invalid qualifier: ' + qualifier) |
| 175 | exit(1) |
| 176 | |
| 177 | if qualifier == qualifier_optional: |
| 178 | optional_symbols.append(symbol) |
| 179 | else: |
| 180 | mandatory_symbols.append(symbol) |
| 181 | |
| 182 | unknown_symbols = [] |
| 183 | for symbol in lib_symbols: |
| 184 | if symbol in mandatory_symbols: |
| 185 | continue |
| 186 | if symbol in optional_symbols: |
| 187 | continue |
| 188 | if args.ignore_symbol and symbol in args.ignore_symbol: |
| 189 | continue |
| 190 | if symbol[:2] == '_Z': |
| 191 | # As ajax found out, the compiler intentionally exports symbols |
| 192 | # that we explicitly asked it not to export, and we can't do |
| 193 | # anything about it: |
| 194 | # https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36022#c4 |
| 195 | continue |
| 196 | unknown_symbols.append(symbol) |
| 197 | |
| 198 | missing_symbols = [ |
| 199 | sym for sym in mandatory_symbols if sym not in lib_symbols |
| 200 | ] |
| 201 | |
| 202 | for symbol in unknown_symbols: |
| 203 | print(args.lib + ': unknown symbol exported: ' + symbol) |
| 204 | |
| 205 | for symbol in missing_symbols: |
| 206 | print(args.lib + ': missing symbol: ' + symbol) |
| 207 | |
| 208 | if unknown_symbols or missing_symbols: |
| 209 | exit(1) |
| 210 | exit(0) |
| 211 | |
| 212 | |
| 213 | if __name__ == '__main__': |
| 214 | main() |