NEWS revision 1.1
1 This file contains information about GCC releases which has been generated 2 automatically from the online release notes. It covers releases of GCC 3 (and the former EGCS project) since EGCS 1.0, on the line of development 4 that led to GCC 3. For information on GCC 2.8.1 and older releases of GCC 2, 5 see ONEWS. 6 7 ====================================================================== 8 http://gcc.gnu.org/gcc-4.5/index.html 9 10 GCC 4.5 Release Series 11 12 Apr 28, 2011 13 14 The [1]GNU project and the GCC developers are pleased to announce the 15 release of GCC 4.5.3. 16 17 This release is a bug-fix release, containing fixes for regressions in 18 GCC 4.5.2 relative to previous releases of GCC. 19 20 Release History 21 22 GCC 4.5.3 23 Apr 28, 2011 ([2]changes) 24 25 GCC 4.5.2 26 Dec 16, 2010 ([3]changes) 27 28 GCC 4.5.1 29 Jul 31, 2010 ([4]changes) 30 31 GCC 4.5.0 32 April 14, 2010 ([5]changes) 33 34 References and Acknowledgements 35 36 GCC used to stand for the GNU C Compiler, but since the compiler 37 supports several other languages aside from C, it now stands for the 38 GNU Compiler Collection. 39 40 A list of [6]successful builds is updated as new information becomes 41 available. 42 43 The GCC developers would like to thank the numerous people that have 44 contributed new features, improvements, bug fixes, and other changes as 45 well as test results to GCC. This [7]amazing group of volunteers is 46 what makes GCC successful. 47 48 For additional information about GCC please refer to the [8]GCC project 49 web site or contact the [9]GCC development mailing list. 50 51 To obtain GCC please use [10]our mirror sites or [11]our SVN server. 52 53 54 For questions related to the use of GCC, please consult these web 55 pages and the [12]GCC manuals. If that fails, the 56 [13]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 57 web pages and the development of GCC are welcome on our developer 58 list at [14]gcc (a] gcc.gnu.org. All of [15]our lists have public 59 archives. 60 61 Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and 62 distribution of this entire article is permitted in any medium, 63 provided this notice is preserved. 64 65 These pages are [17]maintained by the GCC team. Last modified 66 2011-04-28[18]. 67 68 References 69 70 1. http://www.gnu.org/ 71 2. http://gcc.gnu.org/gcc-4.5/changes.html 72 3. http://gcc.gnu.org/gcc-4.5/changes.html 73 4. http://gcc.gnu.org/gcc-4.5/changes.html 74 5. http://gcc.gnu.org/gcc-4.5/changes.html 75 6. http://gcc.gnu.org/gcc-4.5/buildstat.html 76 7. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 77 8. http://gcc.gnu.org/index.html 78 9. mailto:gcc (a] gcc.gnu.org 79 10. http://gcc.gnu.org/mirrors.html 80 11. http://gcc.gnu.org/svn.html 81 12. http://gcc.gnu.org/onlinedocs/ 82 13. mailto:gcc-help (a] gcc.gnu.org 83 14. mailto:gcc (a] gcc.gnu.org 84 15. http://gcc.gnu.org/lists.html 85 16. http://www.fsf.org/ 86 17. http://gcc.gnu.org/about.html 87 18. http://validator.w3.org/check/referer 88 ====================================================================== 89 http://gcc.gnu.org/gcc-4.5/changes.html 90 91 GCC 4.5 Release Series 92 Changes, New Features, and Fixes 93 94 Caveats 95 96 * GCC now requires the [1]MPC library in order to build. See the 97 [2]prerequisites page for version requirements. 98 * Support for a number of older systems and recently unmaintained or 99 untested target ports of GCC has been declared obsolete in GCC 4.5. 100 Unless there is activity to revive them, the next release of GCC 101 will have their sources permanently removed. 102 The following ports for individual systems on particular 103 architectures have been obsoleted: 104 + IRIX releases before 6.5 (mips-sgi-irix5*, 105 mips-sgi-irix6.[0-4]) 106 + Solaris 7 (*-*-solaris2.7) 107 + Tru64 UNIX releases before V5.1 (alpha*-dec-osf4*, 108 alpha-dec-osf5.0*) 109 + Details for the IRIX, Solaris 7, and Tru64 UNIX obsoletions 110 can be found in the [3]announcement. 111 Support for the classic POWER architecture implemented in the 112 original RIOS and RIOS2 processors of the old IBM RS/6000 product 113 line has been obsoleted in the rs6000 port. This does not affect 114 the new generation Power and PowerPC architectures. 115 * Support has been removed for all the [4]configurations obsoleted in 116 GCC 4.4. 117 * Support has been removed for the protoize and unprotoize utilities, 118 obsoleted in GCC 4.4. 119 * Support has been removed for tuning for Itanium1 (Merced) variants. 120 Note that code tuned for Itanium2 should also run correctly on 121 Itanium1. 122 * GCC now generates unwind info also for epilogues. DWARF debuginfo 123 generated by GCC now uses more features of DWARF3 than before, and 124 also some DWARF4 features. GDB older than 7.0 is not able to handle 125 either of these, so to debug GCC 4.5 generated binaries or 126 libraries GDB 7.0 or later is needed. You can disable use of DWARF4 127 features with the -gdwarf-3 -gstrict-dwarf options, or use 128 -gdwarf-2 -gstrict-dwarf to restrict GCC to just DWARF2, but 129 epilogue unwind info is emitted unconditionally whenever unwind 130 info is emitted. 131 * On x86 targets, code containing floating-point calculations may run 132 significantly slower when compiled with GCC 4.5 in strict C99 133 conformance mode than they did with earlier GCC versions. This is 134 due to stricter standard conformance of the compiler and can be 135 avoided by using the option -fexcess-precision=fast; also see 136 [5]below. 137 * The function attribute noinline no longer prevents GCC from cloning 138 the function. A new attribute noclone has been introduced for this 139 purpose. Cloning a function means that it is duplicated and the new 140 copy is specialized for certain contexts (for example when a 141 parameter is a known constant). 142 143 General Optimizer Improvements 144 145 * The -save-temps now takes an optional argument. The -save-temps and 146 -save-temps=cwd switches write the temporary files in the current 147 working directory based on the original source file. The 148 -save-temps=obj switch will write files into the directory 149 specified with the -o option, and the intermediate filenames are 150 based on the output file. This will allow the user to get the 151 compiler intermediate files when doing parallel builds without two 152 builds of the same filename located in different directories from 153 interfering with each other. 154 * Debugging dumps are now created in the same directory as the object 155 file rather than in the current working directory. This allows the 156 user to get debugging dumps when doing parallel builds without two 157 builds of the same filename interfering with each other. 158 * GCC has been integrated with the [6]MPC library. This allows GCC to 159 evaluate complex arithmetic at compile time [7]more accurately. It 160 also allows GCC to evaluate calls to complex built-in math 161 functions having constant arguments and replace them at compile 162 time with their mathematically equivalent results. In doing so, GCC 163 can generate correct results regardless of the math library 164 implementation or floating point precision of the host platform. 165 This also allows GCC to generate identical results regardless of 166 whether one compiles in native or cross-compile configurations to a 167 particular target. The following built-in functions take advantage 168 of this new capability: cacos, cacosh, casin, casinh, catan, 169 catanh, ccos, ccosh, cexp, clog, cpow, csin, csinh, csqrt, ctan, 170 and ctanh. The float and long double variants of these functions 171 (e.g. csinf and csinl) are also handled. 172 * A new link-time optimizer has been added ([8]-flto). When this 173 option is used, GCC generates a bytecode representation of each 174 input file and writes it to special ELF sections in each object 175 file. When the object files are linked together, all the function 176 bodies are read from these ELF sections and instantiated as if they 177 had been part of the same translation unit. This enables 178 interprocedural optimizations to work across different files (and 179 even different languages), potentially improving the performance of 180 the generated code. To use the link-timer optimizer, -flto needs to 181 be specified at compile time and during the final link. If the 182 program does not require any symbols to be exported, it is possible 183 to combine -flto and the experimental [9]-fwhopr with 184 [10]-fwhole-program to allow the interprocedural optimizers to use 185 more aggressive assumptions. 186 * The automatic parallelization pass was enhanced to support 187 parallelization of outer loops. 188 * Automatic parallelization can be enabled as part of Graphite. In 189 addition to -ftree-parallelize-loops=, specify 190 -floop-parallelize-all to enable the Graphite-based optimization. 191 * The infrastructure for optimizing based on [11]restrict qualified 192 pointers has been rewritten and should result in code generation 193 improvements. Optimizations based on restrict qualified pointers 194 are now also available when using -fno-strict-aliasing. 195 * There is a new optimization pass that attempts to change prototype 196 of functions to avoid unused parameters, pass only relevant parts 197 of structures and turn arguments passed by reference to arguments 198 passed by value when possible. It is enabled by -O2 and above as 199 well as -Os and can be manually invoked using the new command-line 200 switch -fipa-sra. 201 * GCC now optimize exception handling code. In particular cleanup 202 regions that are proved to not have any effect are optimized out. 203 204 New Languages and Language specific improvements 205 206 All languages 207 208 * The -fshow-column option is now on by default. This means error 209 messages now have a column associated with them. 210 211 Ada 212 213 * Compilation of programs heavily using discriminated record types 214 with variant parts has been sped up and generates more compact 215 code. 216 * Stack checking now works reasonably well on most plaforms. In some 217 specific cases, stack overflows may still fail to be detected, but 218 a compile-time warning will be issued for these cases. 219 220 C family 221 222 * If a header named in a #include directive is not found, the 223 compiler exits immediately. This avoids a cascade of errors arising 224 from declarations expected to be found in that header being 225 missing. 226 * A new built-in function __builtin_unreachable() has been added that 227 tells the compiler that control will never reach that point. It may 228 be used after asm statements that terminate by transferring control 229 elsewhere, and in other places that are known to be unreachable. 230 * The -Wlogical-op option now warns for logical expressions such as 231 (c == 1 && c == 2) and (c != 1 || c != 2), which are likely to be 232 mistakes. This option is disabled by default. 233 * An asm goto feature has been added to allow asm statements that 234 jump to C labels. 235 * C++0x raw strings are supported for C++ and for C with -std=gnu99. 236 * The deprecated attribute now takes an optional string argument, for 237 example, __attribute__((deprecated("text string"))), that will be 238 printed together with the deprecation warning. 239 240 C 241 242 * The -Wenum-compare option, which warns when comparing values of 243 different enum types, now works for C. It formerly only worked for 244 C++. This warning is enabled by -Wall. It may be avoided by using a 245 type cast. 246 * The -Wcast-qual option now warns about casts which are unsafe in 247 that they permit const-correctness to be violated without further 248 warnings. Specifically, it warns about cases where a qualifier is 249 added when all the lower types are not const. For example, it warns 250 about a cast from char ** to const char **. 251 * The -Wc++-compat option is significantly improved. It issues new 252 warnings for: 253 + Using C++ reserved operator names as identifiers. 254 + Conversions to enum types without explicit casts. 255 + Using va_arg with an enum type. 256 + Using different enum types in the two branches of ?:. 257 + Using ++ or -- on a variable of enum type. 258 + Using the same name as both a struct, union or enum tag and a 259 typedef, unless the typedef refers to the tagged type itself. 260 + Using a struct, union, or enum which is defined within another 261 struct or union. 262 + A struct field defined using a typedef if there is a field in 263 the struct, or an enclosing struct, whose name is the typedef 264 name. 265 + Duplicate definitions at file scope. 266 + Uninitialized const variables. 267 + A global variable with an anonymous struct, union, or enum 268 type. 269 + Using a string constant to initialize a char array whose size 270 is the length of the string. 271 * The new -Wjump-misses-init option warns about cases where a goto or 272 switch skips the initialization of a variable. This sort of branch 273 is an error in C++ but not in C. This warning is enabled by 274 -Wc++-compat. 275 * GCC now ensures that a C99-conforming <stdint.h> is present on most 276 targets, and uses information about the types in this header to 277 implement the Fortran bindings to those types. GCC does not ensure 278 the presence of such a header, and does not implement the Fortran 279 bindings, on the following targets: NetBSD, VxWorks, VMS, 280 SymbianOS, WinCE, LynxOS, Netware, QNX, Interix, TPF. 281 * GCC now implements C90- and C99-conforming rules for constant 282 expressions. This may cause warnings or errors for some code using 283 expressions that can be folded to a constant but are not constant 284 expressions as defined by ISO C. 285 * All known target-independent C90 and C90 Amendment 1 conformance 286 bugs, and all known target-independent C99 conformance bugs not 287 related to floating point or extended identifiers, have been fixed. 288 * The C decimal floating point support now includes support for the 289 FLOAT_CONST_DECIMAL64 pragma. 290 * The named address space feature from ISO/IEC TR 18037 is now 291 supported. This is currently only implemented for the SPU 292 processor. 293 294 C++ 295 296 * Improved [12]experimental support for the upcoming C++0x ISO C++ 297 standard, including support for raw strings, lambda expressions and 298 explicit type conversion operators. 299 * When printing the name of a class template specialization, G++ will 300 now omit any template arguments which come from default template 301 arguments. This behavior (and the pretty-printing of function 302 template specializations as template signature and arguments) can 303 be disabled with the -fno-pretty-templates option. 304 * Access control is now applied to typedef names used in a template, 305 which may cause G++ to reject some ill-formed code that was 306 accepted by earlier releases. The -fno-access-control option can be 307 used as a temporary workaround until the code is corrected. 308 * Compilation time for code that uses templates should now scale 309 linearly with the number of instantiations rather than 310 quadratically, as template instantiations are now looked up using 311 hash tables. 312 * Declarations of functions that look like builtin declarations of 313 library functions are only considered to be redeclarations if they 314 are declared with extern "C". This may cause problems with code 315 that omits extern "C" on hand-written declarations of C library 316 functions such as abort or memcpy. Such code is ill-formed, but was 317 accepted by earlier releases. 318 * Diagnostics that used to complain about passing non-POD types to 319 ... or jumping past the declaration of a non-POD variable now check 320 for triviality rather than PODness, as per C++0x. 321 * In C++0x mode local and anonymous classes are now allowed as 322 template arguments, and in declarations of variables and functions 323 with linkage, so long as any such declaration that is used is also 324 defined ([13]DR 757). 325 * Labels may now have attributes, as has been permitted for a while 326 in C. This is only permitted when the label definition and the 327 attribute specifier is followed by a semicolon--i.e., the label 328 applies to an empty statement. The only useful attribute for a 329 label is unused. 330 * G++ now implements [14]DR 176. Previously G++ did not support using 331 the injected-class-name of a template base class as a type name, 332 and lookup of the name found the declaration of the template in the 333 enclosing scope. Now lookup of the name finds the 334 injected-class-name, which can be used either as a type or as a 335 template, depending on whether or not the name is followed by a 336 template argument list. As a result of this change, some code that 337 was previously accepted may be ill-formed because 338 1. The injected-class-name is not accessible because it's from a 339 private base, or 340 2. The injected-class-name cannot be used as an argument for a 341 template template parameter. 342 In either of these cases, the code can be fixed by adding a 343 nested-name-specifier to explicitly name the template. The first 344 can be worked around with -fno-access-control; the second is only 345 rejected with -pedantic. 346 * A new standard mangling for SIMD vector types has been added, to 347 avoid name clashes on systems with vectors of varying length. By 348 default the compiler still uses the old mangling, but emits aliases 349 with the new mangling on targets that support strong aliases. Users 350 can switch over entirely to the new mangling with -fabi-version=4 351 or -fabi-version=0. -Wabi will now warn about code that uses the 352 old mangling. 353 * The command-line option -ftemplate-depth-N is now written as 354 -ftemplate-depth=N and the old form is deprecated. 355 * Conversions between NULL and non-pointer types are now warned by 356 default. The new option -Wno-conversion-null disables these 357 warnings. Previously these warnings were only available when using 358 -Wconversion explicitly. 359 360 Runtime Library (libstdc++) 361 362 * [15]Improved experimental support for the upcoming ISO C++ 363 standard, C++0x, including: 364 + Support for <future>, <functional>, and <random>. 365 + Existing facilities now exploit explicit operators and the 366 newly implemented core C++0x features. 367 * An experimental [16]profile mode has been added. This is an 368 implementation of many C++ standard library constructs with an 369 additional analysis layer that gives performance improvement advice 370 based on recognition of suboptimal usage patterns. For example, 371 #include <vector> 372 int main() 373 { 374 std::vector<int> v; 375 for (int k = 0; k < 1024; ++k) 376 v.insert(v.begin(), k); 377 } 378 379 When instrumented via the profile mode, can return suggestions 380 about the initial size and choice of the container used as follows: 381 vector-to-list: improvement = 5: call stack = 0x804842c ... 382 : advice = change std::vector to std::list 383 vector-size: improvement = 3: call stack = 0x804842c ... 384 : advice = change initial container size from 0 to 1024 385 386 These constructs can be substituted for the normal libstdc++ 387 constructs on a piecemeal basis, or all existing components can be 388 transformed via the -D_GLIBCXX_PROFILE macro. 389 * [17]Support for decimal floating-point arithmetic (aka ISO C++ TR 390 24733) has been added. This support is in header file 391 <decimal/decimal>, uses namespace std::decimal, and includes 392 classes decimal32, decimal64, and decimal128. 393 * Sources have been audited for application of function attributes 394 nothrow, const, pure, and noreturn. 395 * Python pretty-printers have been added for many standard library 396 components that simplify the internal representation and present a 397 more intuitive view of components when used with 398 appropriately-advanced versions of GDB. For more information, 399 please consult the more [18]detailed description. 400 * The default behavior for comparing typeinfo names has changed, so 401 in <typeinfo>, __GXX_MERGED_TYPEINFO_NAMES now defaults to zero. 402 * The new -static-libstdc++ option directs g++ to link the C++ 403 library statically, even if the default would normally be to link 404 it dynamically. 405 406 Fortran 407 408 * The COMMON default padding has been changed - instead of adding the 409 padding before a variable it is now added afterwards, which 410 increases the compatibility with other vendors and helps to obtain 411 the correct output in some cases. Cf. also the -falign-commons 412 option ([19]added in 4.4). 413 * The -finit-real= option now also supports the value snan for 414 signalling not-a-number; to be effective, one additionally needs to 415 enable trapping (e.g. via -ffpe-trap=). Note: Compile-time 416 optimizations can turn a signalling NaN into a quiet one. 417 * The new option -fcheck= has been added with the options bounds, 418 array-temps, do, pointer, and recursive. The bounds and array-temps 419 options are equivalent to -fbounds-check and 420 -fcheck-array-temporaries. The do option checks for invalid 421 modification of loop iteration variables, and the recursive option 422 tests for recursive calls to subroutines/functions which are not 423 marked as recursive. With pointer pointer association checks in 424 calls are performed; however, neither undefined pointers nor 425 pointers in expressions are handled. Using -fcheck=all enables all 426 these run-time checks. 427 * The run-time checking -fcheck=bounds now warns about invalid string 428 lengths of character dummy arguments. Additionally, more 429 compile-time checks have been added. 430 * The new option [20]-fno-protect-parens has been added; if set, the 431 compiler may reorder REAL and COMPLEX expressions without regard to 432 parentheses. 433 * GNU Fortran no longer links against libgfortranbegin. As before, 434 MAIN__ (assembler symbol name) is the actual Fortran main program, 435 which is invoked by the main function. However, main is now 436 generated and put in the same object file as MAIN__. For the time 437 being, libgfortranbegin still exists for backward compatibility. 438 For details see the new [21]Mixed-Language Programming chapter in 439 the manual. 440 * The I/O library was restructured for performance and cleaner code. 441 * Array assignments and WHERE are now run in parallel when OpenMP's 442 WORKSHARE is used. 443 * The experimental option -fwhole-file was added. The option allows 444 whole-file checking of procedure arguments and allows for better 445 optimizations. It can also be used with -fwhole-program, which is 446 now also supported in gfortran. 447 * More Fortran 2003 and Fortran 2008 mathematical functions can now 448 be used as initialization expressions. 449 * Some extended attributes such as STDCALL are now supported via the 450 [22]GCC$ compiler directive. 451 * For Fortran 77 compatibility: If -fno-sign-zero is used, the SIGN 452 intrinsic behaves now as if zero were always positive. 453 * For legacy compatibiliy: On Cygwin and MinGW, the special files 454 CONOUT$ and CONIN$ (and CONERR$ which maps to CONOUT$) are now 455 supported. 456 * Fortran 2003 support has been extended: 457 + Procedure-pointer function results and procedure-pointer 458 components (including PASS), 459 + allocatable scalars (experimental), 460 + DEFERRED type-bound procedures, 461 + the ERRMSG= argument of the ALLOCATE and DEALLOCATE statements 462 have been implemented. 463 + The ALLOCATE statement supports type-specs and the SOURCE= 464 argument. 465 + OPERATOR(*) and ASSIGNMENT(=) are now allowed as GENERIC 466 type-bound procedure (i.e. as type-bound operators). 467 + Rounding (ROUND=, RZ, ...) for output is now supported. 468 + The INT_FAST{8,16,32,64,128}_T kind type parameters of the 469 intrinsic module ISO_C_BINDING are now supported, except for 470 the targets listed above as ones where GCC does not have 471 <stdint.h> type information. 472 + Extensible derived types with type-bound procedure or 473 procedure pointer with PASS attribute now have to use CLASS in 474 line with the Fortran 2003 standard; the workaround to use 475 TYPE is no longer supported. 476 + [23]Experimental, incomplete support for polymorphism, 477 including CLASS, SELECT TYPE and dynamic dispatch of 478 type-bound procedure calls. Some features do not work yet such 479 as unlimited polymorphism (CLASS(*)). 480 * Fortran 2008 support has been extended: 481 + The OPEN statement now supports the NEWUNIT= option, which 482 returns a unique file unit, thus preventing inadvertent use of 483 the same unit in different parts of the program. 484 + Support for unlimited format items has been added. 485 + The INT{8,16,32} and REAL{32,64,128} kind type parameters of 486 the intrinsic module ISO_FORTRAN_ENV are now supported. 487 + Using complex arguments with TAN, SINH, COSH, TANH, ASIN, 488 ACOS, and ATAN is now possible; the functions ASINH, ACOSH, 489 and ATANH have been added (for real and complex arguments) and 490 ATAN(Y,X) is now an alias for ATAN2(Y,X). 491 + The BLOCK construct has been implemented. 492 493 Java (GCJ) 494 495 New Targets and Target Specific Improvements 496 497 AIX 498 499 * Full cross-toolchain support now available with GNU Binutils 500 501 ARM 502 503 * GCC now supports the Cortex-M0 and Cortex-A5 processors. 504 * GCC now supports the ARM v7E-M architecture. 505 * GCC now supports VFPv4-based FPUs and FPUs with 506 single-precision-only VFP. 507 * GCC has many improvements to optimization for other ARM processors, 508 including scheduling support for the integer pipeline on Cortex-A9. 509 * GCC now supports the IEEE 754-2008 half-precision floating-point 510 type, and a variant ARM-specific half-precision type. This type is 511 specified using __fp16, with the layout determined by 512 -mfp16-format. With appropriate -mfpu options, the Cortex-A9 and 513 VFPv4 half-precision instructions will be used. 514 * GCC now supports the variant of AAPCS that uses VFP registers for 515 parameter passing and return values. 516 517 AVR 518 519 * The -mno-tablejump option has been removed because it has the same 520 effect as the -fno-jump-tables option. 521 * Added support for these new AVR devices: 522 + ATmega8U2 523 + ATmega16U2 524 + ATmega32U2 525 526 IA-32/x86-64 527 528 * GCC now will set the default for -march= based on the configure 529 target. 530 * GCC now supports handling floating-point excess precision arising 531 from use of the x87 floating-point unit in a way that conforms to 532 ISO C99. This is enabled with -fexcess-precision=standard and with 533 standards conformance options such as -std=c99, and may be disabled 534 using -fexcess-precision=fast. 535 * Support for the Intel Atom processor is now available through the 536 -march=atom and -mtune=atom options. 537 * A new -mcrc32 option is now available to enable crc32 intrinsics. 538 * A new -mmovbe option is now available to enable GCC to use the 539 movbe instruction to implement __builtin_bswap32 and 540 __builtin_bswap64. 541 * SSE math now can be enabled by default at configure time with the 542 new --with-fpmath=sse option. 543 * There is a new intrinsic header file, <x86intrin.h>. It should be 544 included before using any IA-32/x86-64 intrinsics. 545 * Support for the XOP, FMA4, and LWP instruction sets for the AMD 546 Orochi processors are now available with the -mxop, -mfma4, and 547 -mlwp options. 548 * The -mabm option enables GCC to use the popcnt and lzcnt 549 instructions on AMD processors. 550 * The -mpopcnt option enables GCC to use the popcnt instructions on 551 both AMD and Intel processors. 552 553 M68K/ColdFire 554 555 * GCC now supports ColdFire 51xx, 5221x, 5225x, 52274, 52277, 5301x 556 and 5441x devices. 557 * GCC now supports thread-local storage (TLS) on M68K and ColdFire 558 processors. 559 560 MeP 561 562 Support has been added for the Toshiba Media embedded Processor (MeP, 563 or mep-elf) embedded target. 564 565 MIPS 566 567 * GCC now supports MIPS 1004K processors. 568 * GCC can now be configured with options --with-arch-32, 569 --with-arch-64, --with-tune-32 and --with-tune-64 to control the 570 default optimization separately for 32-bit and 64-bit modes. 571 * MIPS targets now support an alternative _mcount interface, in which 572 register $12 points to the function's save slot for register $31. 573 This interface is selected by the -mcount-ra-address option; see 574 the documentation for more details. 575 * GNU/Linux targets can now generate read-only .eh_frame sections. 576 This optimization requires GNU binutils 2.20 or above, and is only 577 available if GCC is configured with a suitable version of binutils. 578 * GNU/Linux targets can now attach special relocations to indirect 579 calls, so that the linker can turn them into direct jumps or 580 branches. This optimization requires GNU binutils 2.20 or later, 581 and is automatically selected if GCC is configured with an 582 appropriate version of binutils. It can be explicitly enabled or 583 disabled using the -mrelax-pic-calls command-line option. 584 * GCC now generates more heavily-optimized atomic operations on 585 Octeon processors. 586 * MIPS targets now support the -fstack-protector option. 587 * GCC now supports an -msynci option, which specifies that synci is 588 enough to flush the instruction cache, without help from the 589 operating system. GCC uses this information to optimize 590 automatically-generated cache flush operations, such as those used 591 for nested functions in C. There is also a --with-synci 592 configure-time option, which makes -msynci the default. 593 * GCC supports four new function attributes for interrupt handlers: 594 interrupt, use_shadow_register_set, keep_interrupts_masked and 595 use_debug_exception_return. See the documentation for more details 596 about these attributes. 597 598 picochip 599 600 RS/6000 (POWER/PowerPC) 601 602 * GCC now supports the Power ISA 2.06, which includes the VSX 603 instructions that add vector 64-bit floating point support, new 604 population count instructions, and conversions between floating 605 point and unsigned types. 606 * Support for the power7 processor is now available through the 607 -mcpu=power7 and -mtune=power7. 608 * GCC will now vectorize loops that contain simple math functions 609 like copysign when generating code for altivec or VSX targets. 610 * Support for the A2 processor is now available through the -mcpu=a2 611 and -mtune=a2 options. 612 * Support for the 476 processor is now available through the 613 -mcpu={476,476fp} and -mtune={476,476fp} options. 614 * Support for the e500mc64 processor is now available through the 615 -mcpu=e500mc64 and -mtune=e500mc64 options. 616 * GCC can now be configured with options --with-cpu-32, 617 --with-cpu-64, --with-tune-32 and --with-tune-64 to control the 618 default optimization separately for 32-bit and 64-bit modes. 619 620 RX 621 622 Support has been added for the Renesas RX Processor (rx-elf) target. 623 624 Operating Systems 625 626 Windows (Cygwin and MinGW) 627 628 * GCC now installs all the major language runtime libraries as DLLs 629 when configured with the --enable-shared option. 630 * GCC now makes use of the new support for aligned common variables 631 in versions of binutils >= 2.20 to fix bugs in the support for SSE 632 data types. 633 * Improvements to the libffi support library increase the reliability 634 of code generated by GCJ on all Windows platforms. Libgcj is 635 enabled by default for the first time. 636 * Libtool improvements simplify installation by placing the generated 637 DLLs in the correct binaries directory. 638 * Numerous other minor bugfixes and improvements, and substantial 639 enhancements to the Fortran language support library. 640 641 Documentation improvements 642 643 Other significant improvements 644 645 Plugins 646 647 * It is now possible to extend the compiler without having to modify 648 its source code. A new option -fplugin=file.so tells GCC to load 649 the shared object file.so and execute it as part of the compiler. 650 The internal documentation describes the details on how plugins can 651 interact with the compiler. 652 653 Installation changes 654 655 * The move to newer autotools changed default installation 656 directories and switches to control them: The --with-datarootdir, 657 --with-docdir, --with-pdfdir, and --with-htmldir switches are not 658 used any more. Instead, you can now use --datarootdir, --docdir, 659 --htmldir, and --pdfdir. The default installation directories have 660 changed as follows according to the GNU Coding Standards: 661 662 datarootdir read-only architecture-independent data root [PREFIX/share] 663 localedir locale-specific message catalogs [DATAROOTDIR/locale] 664 docdir documentation root [DATAROOTDIR/doc/PACKAGE] 665 htmldir html documentation [DOCDIR] 666 dvidir dvi documentation [DOCDIR] 667 pdfdir pdf documentation [DOCDIR] 668 psdir ps documentation [DOCDIR] 669 The following variables have new default values: 670 671 datadir read-only architecture-independent data [DATAROOTDIR] 672 infodir info documentation [DATAROOTDIR/info] 673 mandir man documentation [DATAROOTDIR/man] 674 675 GCC 4.5.1 676 677 This is the [24]list of problem reports (PRs) from GCC's bug tracking 678 system that are known to be fixed in the 4.5.1 release. This list might 679 not be complete (that is, it is possible that some PRs that have been 680 fixed are not listed here). 681 682 All languages 683 684 * GCC's new link-time optimizer ([25]-flto) now also works on a few 685 non-ELF targets: 686 + Cygwin (*-cygwin*) 687 + MinGW (*-mingw*) 688 + Darwin on x86-64 (x86_64-apple-darwin*) 689 LTO is not enabled by default for these targets. To enable LTO, you 690 should configure with the --enable-lto option. 691 692 GCC 4.5.2 693 694 This is the [26]list of problem reports (PRs) from GCC's bug tracking 695 system that are known to be fixed in the 4.5.2 release. This list might 696 not be complete (that is, it is possible that some PRs that have been 697 fixed are not listed here). 698 699 GCC 4.5.3 700 701 This is the [27]list of problem reports (PRs) from GCC's bug tracking 702 system that are known to be fixed in the 4.5.2 release. This list might 703 not be complete (that is, it is possible that some PRs that have been 704 fixed are not listed here). 705 706 On the PowerPC compiler, the altivec builtin functions vec_ld and 707 vec_st have been modified to generate the Altivec memory instructions 708 LVX and STVX, even if the -mvsx option is used. In the initial GCC 4.5 709 release, these builtin functions were changed to generate VSX memory 710 reference instructions instead of Altivec memory instructions, but 711 there are differences between the two instructions. If the VSX 712 instruction set is available, you can now use the new builtin functions 713 vec_vsx_ld and vec_vsx_st which always generates the VSX memory 714 instructions. 715 716 717 For questions related to the use of GCC, please consult these web 718 pages and the [28]GCC manuals. If that fails, the 719 [29]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 720 web pages and the development of GCC are welcome on our developer 721 list at [30]gcc (a] gcc.gnu.org. All of [31]our lists have public 722 archives. 723 724 Copyright (C) [32]Free Software Foundation, Inc. Verbatim copying and 725 distribution of this entire article is permitted in any medium, 726 provided this notice is preserved. 727 728 These pages are [33]maintained by the GCC team. Last modified 729 2011-04-25[34]. 730 731 References 732 733 1. http://www.multiprecision.org/ 734 2. http://gcc.gnu.org/install/prerequisites.html 735 3. http://gcc.gnu.org/ml/gcc/2010-01/msg00510.html 736 4. http://gcc.gnu.org/gcc-4.4/changes.html#obsoleted 737 5. http://gcc.gnu.org/gcc-4.5/changes.html#x86 738 6. http://www.multiprecision.org/ 739 7. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30789 740 8. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801 741 9. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhopr-802 742 10. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhole-program-800 743 11. http://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html 744 12. http://gcc.gnu.org/gcc-4.5/cxx0x_status.html 745 13. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#757 746 14. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#176 747 15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x 748 16. http://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html 749 17. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.tr24733 750 18. http://sourceware.org/gdb/wiki/STLSupport 751 19. http://gcc.gnu.org/gcc-4.4/changes.html 752 20. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html 753 21. http://gcc.gnu.org/onlinedocs/gfortran/Mixed-Language-Programming.html 754 22. http://gcc.gnu.org/onlinedocs/gfortran/GNU-Fortran-Compiler-Directives.html 755 23. http://gcc.gnu.org/wiki/OOP 756 24. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.1 757 25. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801 758 26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.2 759 27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.2 760 28. http://gcc.gnu.org/onlinedocs/ 761 29. mailto:gcc-help (a] gcc.gnu.org 762 30. mailto:gcc (a] gcc.gnu.org 763 31. http://gcc.gnu.org/lists.html 764 32. http://www.fsf.org/ 765 33. http://gcc.gnu.org/about.html 766 34. http://validator.w3.org/check/referer 767 ====================================================================== 768 http://gcc.gnu.org/gcc-4.4/index.html 769 770 GCC 4.4 Release Series 771 772 April 16, 2011 773 774 The [1]GNU project and the GCC developers are pleased to announce the 775 release of GCC 4.4.6. 776 777 This release is a bug-fix release, containing fixes for regressions in 778 GCC 4.4.5 relative to previous releases of GCC. 779 780 Release History 781 782 GCC 4.4.6 783 April 16, 2011 ([2]changes) 784 785 GCC 4.4.5 786 October 1, 2010 ([3]changes) 787 788 GCC 4.4.4 789 April 29, 2010 ([4]changes) 790 791 GCC 4.4.3 792 January 21, 2010 ([5]changes) 793 794 GCC 4.4.2 795 October 15, 2009 ([6]changes) 796 797 GCC 4.4.1 798 July 22, 2009 ([7]changes) 799 800 GCC 4.4.0 801 April 21, 2009 ([8]changes) 802 803 References and Acknowledgements 804 805 GCC used to stand for the GNU C Compiler, but since the compiler 806 supports several other languages aside from C, it now stands for the 807 GNU Compiler Collection. 808 809 A list of [9]successful builds is updated as new information becomes 810 available. 811 812 The GCC developers would like to thank the numerous people that have 813 contributed new features, improvements, bug fixes, and other changes as 814 well as test results to GCC. This [10]amazing group of volunteers is 815 what makes GCC successful. 816 817 For additional information about GCC please refer to the [11]GCC 818 project web site or contact the [12]GCC development mailing list. 819 820 To obtain GCC please use [13]our mirror sites or [14]our SVN server. 821 822 823 For questions related to the use of GCC, please consult these web 824 pages and the [15]GCC manuals. If that fails, the 825 [16]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 826 web pages and the development of GCC are welcome on our developer 827 list at [17]gcc (a] gcc.gnu.org. All of [18]our lists have public 828 archives. 829 830 Copyright (C) [19]Free Software Foundation, Inc. Verbatim copying and 831 distribution of this entire article is permitted in any medium, 832 provided this notice is preserved. 833 834 These pages are [20]maintained by the GCC team. Last modified 835 2011-04-25[21]. 836 837 References 838 839 1. http://www.gnu.org/ 840 2. http://gcc.gnu.org/gcc-4.4/changes.html 841 3. http://gcc.gnu.org/gcc-4.4/changes.html 842 4. http://gcc.gnu.org/gcc-4.4/changes.html 843 5. http://gcc.gnu.org/gcc-4.4/changes.html 844 6. http://gcc.gnu.org/gcc-4.4/changes.html 845 7. http://gcc.gnu.org/gcc-4.4/changes.html 846 8. http://gcc.gnu.org/gcc-4.4/changes.html 847 9. http://gcc.gnu.org/gcc-4.4/buildstat.html 848 10. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 849 11. http://gcc.gnu.org/index.html 850 12. mailto:gcc (a] gcc.gnu.org 851 13. http://gcc.gnu.org/mirrors.html 852 14. http://gcc.gnu.org/svn.html 853 15. http://gcc.gnu.org/onlinedocs/ 854 16. mailto:gcc-help (a] gcc.gnu.org 855 17. mailto:gcc (a] gcc.gnu.org 856 18. http://gcc.gnu.org/lists.html 857 19. http://www.fsf.org/ 858 20. http://gcc.gnu.org/about.html 859 21. http://validator.w3.org/check/referer 860 ====================================================================== 861 http://gcc.gnu.org/gcc-4.4/changes.html 862 863 GCC 4.4 Release Series 864 Changes, New Features, and Fixes 865 866 The latest release in the 4.4 release series is [1]GCC 4.4.6. 867 868 Caveats 869 870 * __builtin_stdarg_start has been completely removed from GCC. 871 Support for <varargs.h> had been deprecated since GCC 4.0. Use 872 __builtin_va_start as a replacement. 873 * Some of the errors issued by the C++ front end that could be 874 downgraded to warnings in previous releases by using -fpermissive 875 are now warnings by default. They can be converted into errors by 876 using -pedantic-errors. 877 * Use of the cpp assertion extension will now emit a warning when 878 -Wdeprecated or -pedantic is used. This extension has been 879 deprecated for many years, but never warned about. 880 * Packed bit-fields of type char were not properly bit-packed on many 881 targets prior to GCC 4.4. On these targets, the fix in GCC 4.4 882 causes an ABI change. For example there is no longer a 4-bit 883 padding between field a and b in this structure: 884 struct foo 885 { 886 char a:4; 887 char b:8; 888 } __attribute__ ((packed)); 889 There is a new warning to help identify fields that are affected: 890 foo.c:5: note: Offset of packed bit-field 'b' has changed in GCC 4.4 891 The warning can be disabled with -Wno-packed-bitfield-compat. 892 * On ARM EABI targets, the C++ mangling of the va_list type has been 893 changed to conform to the current revision of the EABI. This does 894 not affect the libstdc++ library included with GCC. 895 * The SCOUNT and POS bits of the MIPS DSP control register are now 896 treated as global. Previous versions of GCC treated these fields as 897 call-clobbered instead. 898 * The MIPS port no longer recognizes the h asm constraint. It was 899 necessary to remove this constraint in order to avoid generating 900 unpredictable code sequences. 901 One of the main uses of the h constraint was to extract the high 902 part of a multiplication on 64-bit targets. For example: 903 asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y)); 904 You can now achieve the same effect using 128-bit types: 905 typedef unsigned int uint128_t __attribute__((mode(TI))); 906 result = ((uint128_t) x * y) >> 64; 907 The second sequence is better in many ways. For example, if x and y 908 are constants, the compiler can perform the multiplication at 909 compile time. If x and y are not constants, the compiler can 910 schedule the runtime multiplication better than it can schedule an 911 asm statement. 912 * Support for a number of older systems and recently unmaintained or 913 untested target ports of GCC has been declared obsolete in GCC 4.4. 914 Unless there is activity to revive them, the next release of GCC 915 will have their sources permanently removed. 916 The following ports for individual systems on particular 917 architectures have been obsoleted: 918 + Generic a.out on IA32 and m68k (i[34567]86-*-aout*, 919 m68k-*-aout*) 920 + Generic COFF on ARM, H8300, IA32, m68k and SH (arm-*-coff*, 921 armel-*-coff*, h8300-*-*, i[34567]86-*-coff*, m68k-*-coff*, 922 sh-*-*). This does not affect other more specific targets 923 using the COFF object format on those architectures, or the 924 more specific H8300 and SH targets (h8300-*-rtems*, 925 h8300-*-elf*, sh-*-elf*, sh-*-symbianelf*, sh-*-linux*, 926 sh-*-netbsdelf*, sh-*-rtems*, sh-wrs-vxworks). 927 + 2BSD on PDP-11 (pdp11-*-bsd) 928 + AIX 4.1 and 4.2 on PowerPC (rs6000-ibm-aix4.[12]*, 929 powerpc-ibm-aix4.[12]*) 930 + Tuning support for Itanium1 (Merced) variants. Note that code 931 tuned for Itanium2 should also run correctly on Itanium1. 932 * The protoize and unprotoize utilities have been obsoleted and will 933 be removed in GCC 4.5. These utilities have not been installed by 934 default since GCC 3.0. 935 * Support has been removed for all the [2]configurations obsoleted in 936 GCC 4.3. 937 * Unknown -Wno-* options are now silently ignored by GCC if no other 938 diagnostics are issued. If other diagnostics are issued, then GCC 939 warns about the unknown options. 940 * More information on porting to GCC 4.4 from previous versions of 941 GCC can be found in the [3]porting guide for this release. 942 943 General Optimizer Improvements 944 945 * A new command-line switch -findirect-inlining has been added. When 946 turned on it allows the inliner to also inline indirect calls that 947 are discovered to have known targets at compile time thanks to 948 previous inlining. 949 * A new command-line switch -ftree-switch-conversion has been added. 950 This new pass turns simple initializations of scalar variables in 951 switch statements into initializations from a static array, given 952 that all the values are known at compile time and the ratio between 953 the new array size and the original switch branches does not exceed 954 the parameter --param switch-conversion-max-branch-ratio (default 955 is eight). 956 * A new command-line switch -ftree-builtin-call-dce has been added. 957 This optimization eliminates unnecessary calls to certain builtin 958 functions when the return value is not used, in cases where the 959 calls can not be eliminated entirely because the function may set 960 errno. This optimization is on by default at -O2 and above. 961 * A new command-line switch -fconserve-stack directs the compiler to 962 minimize stack usage even if it makes the generated code slower. 963 This affects inlining decisions. 964 * When the assembler supports it, the compiler will now emit unwind 965 information using assembler .cfi directives. This makes it possible 966 to use such directives in inline assembler code. The new option 967 -fno-dwarf2-cfi-asm directs the compiler to not use .cfi 968 directives. 969 * The [4]Graphite branch has been merged. This merge has brought in a 970 new framework for loop optimizations based on a polyhedral 971 intermediate representation. These optimizations apply to all the 972 languages supported by GCC. The following new code transformations 973 are available in GCC 4.4: 974 + -floop-interchange performs loop interchange transformations 975 on loops. Interchanging two nested loops switches the inner 976 and outer loops. For example, given a loop like: 977 DO J = 1, M 978 DO I = 1, N 979 A(J, I) = A(J, I) * C 980 ENDDO 981 ENDDO 982 983 loop interchange will transform the loop as if the user had 984 written: 985 DO I = 1, N 986 DO J = 1, M 987 A(J, I) = A(J, I) * C 988 ENDDO 989 ENDDO 990 991 which can be beneficial when N is larger than the caches, 992 because in Fortran, the elements of an array are stored in 993 memory contiguously by column, and the original loop iterates 994 over rows, potentially creating at each access a cache miss. 995 + -floop-strip-mine performs loop strip mining transformations 996 on loops. Strip mining splits a loop into two nested loops. 997 The outer loop has strides equal to the strip size and the 998 inner loop has strides of the original loop within a strip. 999 For example, given a loop like: 1000 DO I = 1, N 1001 A(I) = A(I) + C 1002 ENDDO 1003 1004 loop strip mining will transform the loop as if the user had 1005 written: 1006 DO II = 1, N, 4 1007 DO I = II, min (II + 3, N) 1008 A(I) = A(I) + C 1009 ENDDO 1010 ENDDO 1011 1012 + -floop-block performs loop blocking transformations on loops. 1013 Blocking strip mines each loop in the loop nest such that the 1014 memory accesses of the element loops fit inside caches. For 1015 example, given a loop like: 1016 DO I = 1, N 1017 DO J = 1, M 1018 A(J, I) = B(I) + C(J) 1019 ENDDO 1020 ENDDO 1021 1022 loop blocking will transform the loop as if the user had 1023 written: 1024 DO II = 1, N, 64 1025 DO JJ = 1, M, 64 1026 DO I = II, min (II + 63, N) 1027 DO J = JJ, min (JJ + 63, M) 1028 A(J, I) = B(I) + C(J) 1029 ENDDO 1030 ENDDO 1031 ENDDO 1032 ENDDO 1033 1034 which can be beneficial when M is larger than the caches, 1035 because the innermost loop will iterate over a smaller amount 1036 of data that can be kept in the caches. 1037 * A new register allocator has replaced the old one. It is called 1038 integrated register allocator (IRA) because coalescing, register 1039 live range splitting, and hard register preferencing are done 1040 on-the-fly during coloring. It also has better integration with the 1041 reload pass. IRA is a regional register allocator which uses modern 1042 Chaitin-Briggs coloring instead of Chow's priority coloring used in 1043 the old register allocator. More info about IRA internals and 1044 options can be found in the GCC manuals. 1045 * A new instruction scheduler and software pipeliner, based on the 1046 selective scheduling approach, has been added. The new pass 1047 performs instruction unification, register renaming, substitution 1048 through register copies, and speculation during scheduling. The 1049 software pipeliner is able to pipeline non-countable loops. The new 1050 pass is targeted at scheduling-eager in-order platforms. In GCC 4.4 1051 it is available for the Intel Itanium platform working by default 1052 as the second scheduling pass (after register allocation) at the 1053 -O3 optimization level. 1054 * When using -fprofile-generate with a multi-threaded program, the 1055 profile counts may be slightly wrong due to race conditions. The 1056 new -fprofile-correction option directs the compiler to apply 1057 heuristics to smooth out the inconsistencies. By default the 1058 compiler will give an error message when it finds an inconsistent 1059 profile. 1060 * The new -fprofile-dir=PATH option permits setting the directory 1061 where profile data files are stored when using -fprofile-generate 1062 and friends, and the directory used when reading profile data files 1063 using -fprofile-use and friends. 1064 1065 New warning options 1066 1067 * The new -Wframe-larger-than=NUMBER option directs GCC to emit a 1068 warning if any stack frame is larger than NUMBER bytes. This may be 1069 used to help ensure that code fits within a limited amount of stack 1070 space. 1071 * The command-line option -Wlarger-than-N is now written as 1072 -Wlarger-than=N and the old form is deprecated. 1073 * The new -Wno-mudflap option disables warnings about constructs 1074 which can not be instrumented when using -fmudflap. 1075 1076 New Languages and Language specific improvements 1077 1078 * Version 3.0 of the [5]OpenMP specification is now supported for the 1079 C, C++, and Fortran compilers. 1080 * New character data types, per [6]TR 19769: New character types in 1081 C, are now supported for the C compiler in -std=gnu99 mode, as 1082 __CHAR16_TYPE__ and __CHAR32_TYPE__, and for the C++ compiler in 1083 -std=c++0x and -std=gnu++0x modes, as char16_t and char32_t too. 1084 1085 C family 1086 1087 * A new optimize attribute was added to allow programmers to change 1088 the optimization level and particular optimization options for an 1089 individual function. You can also change the optimization options 1090 via the GCC optimize pragma for functions defined after the pragma. 1091 The GCC push_options pragma and the GCC pop_options pragma allow 1092 you temporarily save and restore the options used. The GCC 1093 reset_options pragma restores the options to what was specified on 1094 the command line. 1095 * Uninitialized warnings do not require enabling optimization 1096 anymore, that is, -Wuninitialized can be used together with -O0. 1097 Nonetheless, the warnings given by -Wuninitialized will probably be 1098 more accurate if optimization is enabled. 1099 * -Wparentheses now warns about expressions such as (!x | y) and (!x 1100 & y). Using explicit parentheses, such as in ((!x) | y), silences 1101 this warning. 1102 * -Wsequence-point now warns within if, while,do while and for 1103 conditions, and within for begin/end expressions. 1104 * A new option -dU is available to dump definitions of preprocessor 1105 macros that are tested or expanded. 1106 1107 C++ 1108 1109 * [7]Improved experimental support for the upcoming ISO C++ standard, 1110 C++0x. Including support for auto, inline namespaces, generalized 1111 initializer lists, defaulted and deleted functions, new character 1112 types, and scoped enums. 1113 * Those errors that may be downgraded to warnings to build legacy 1114 code now mention -fpermissive when -fdiagnostics-show-option is 1115 enabled. 1116 * -Wconversion now warns if the result of a static_cast to enumeral 1117 type is unspecified because the value is outside the range of the 1118 enumeral type. 1119 * -Wuninitialized now warns if a non-static reference or non-static 1120 const member appears in a class without constructors. 1121 * G++ now properly implements value-initialization, so objects with 1122 an initializer of () and an implicitly defined default constructor 1123 will be zero-initialized before the default constructor is called. 1124 1125 Runtime Library (libstdc++) 1126 1127 * [8]Improved experimental support for the upcoming ISO C++ standard, 1128 C++0x, including: 1129 + Support for <chrono>, <condition_variable>, <cstdatomic>, 1130 <forward_list>, <initializer_list>, <mutex>, <ratio>, 1131 <system_error>, and <thread>. 1132 + unique_ptr, <algorithm> additions, exception propagation, and 1133 support for the new character types in <string> and <limits>. 1134 + Existing facilities now exploit initializer lists, defaulted 1135 and deleted functions, and the newly implemented core C++0x 1136 features. 1137 + Some standard containers are more efficient together with 1138 stateful allocators, i.e., no allocator is constructed on the 1139 fly at element construction time. 1140 * Experimental support for non-standard pointer types in containers. 1141 * The long standing libstdc++/30928 has been fixed for targets 1142 running glibc 2.10 or later. 1143 * As usual, many small and larger bug fixes, in particular quite a 1144 few corner cases in <locale>. 1145 1146 Fortran 1147 1148 * GNU Fortran now employs libcpp directly instead of using cc1 as an 1149 external preprocessor. The [9]-cpp option was added to allow manual 1150 invocation of the preprocessor without relying on filename 1151 extensions. 1152 * The [10]-Warray-temporaries option warns about array temporaries 1153 generated by the compiler, as an aid to optimization. 1154 * The [11]-fcheck-array-temporaries option has been added, printing a 1155 notification at run time, when an array temporary had to be created 1156 for an function argument. Contrary to -Warray-temporaries the 1157 warning is only printed if the array is noncontiguous. 1158 * Improved generation of DWARF debugging symbols 1159 * If using an intrinsic not part of the selected standard (via -std= 1160 and -fall-intrinsics) gfortran will now treat it as if this 1161 procedure were declared EXTERNAL and try to link to a user-supplied 1162 procedure. -Wintrinsics-std will warn whenever this happens. The 1163 now-useless option -Wnonstd-intrinsic was removed. 1164 * The flag -falign-commons has been added to control the alignment of 1165 variables in COMMON blocks, which is enabled by default in line 1166 with previous GCC version. Using -fno-align-commons one can force 1167 commons to be contiguous in memory as required by the Fortran 1168 standard, however, this slows down the memory access. The option 1169 -Walign-commons, which is enabled by default, warns when padding 1170 bytes were added for alignment. The proper solution is to sort the 1171 common objects by decreasing storage size, which avoids the 1172 alignment problems. 1173 * Fortran 2003 support has been extended: 1174 + Wide characters (ISO 10646, UCS-4, kind=4) and UTF-8 I/O is 1175 now supported (except internal reads from/writes to wide 1176 strings). [12]-fbackslash now supports also \unnnn and 1177 \Unnnnnnnn to enter Unicode characters. 1178 + Asynchronous I/O (implemented as synchronous I/O) and the 1179 decimal=, size=, sign=, pad=, blank=, and delim= specifiers 1180 are now supported in I/O statements. 1181 + Support for Fortran 2003 structure constructors and for array 1182 constructor with typespec has been added. 1183 + Procedure Pointers (but not yet as component in derived types 1184 and as function results) are now supported. 1185 + Abstract types, type extension, and type-bound procedures 1186 (both PROCEDURE and GENERIC but not as operators). Note: As 1187 CLASS/polymorphyic types are not implemented, type-bound 1188 procedures with PASS accept as non-standard extension TYPE 1189 arguments. 1190 * Fortran 2008 support has been added: 1191 + The -std=f2008 option and support for the file extensions 1192 .f2008 and .F2008 has been added. 1193 + The g0 format descriptor is now supported. 1194 + The Fortran 2008 mathematical intrinsics ASINH, ACOSH, ATANH, 1195 ERF, ERFC, GAMMA, LOG_GAMMA, BESSEL_*, HYPOT, and ERFC_SCALED 1196 are now available (some of them existed as GNU extension 1197 before). Note: The hyperbolic functions are not yet supporting 1198 complex arguments and the three- argument version of BESSEL_*N 1199 is not available. 1200 + The bit intrinsics LEADZ and TRAILZ have been added. 1201 1202 Java (GCJ) 1203 1204 Ada 1205 1206 * The Ada runtime now supports multilibs on many platforms including 1207 x86_64, SPARC and PowerPC. Their build is enabled by default. 1208 1209 New Targets and Target Specific Improvements 1210 1211 ARM 1212 1213 * GCC now supports optimizing for the Cortex-A9, Cortex-R4 and 1214 Cortex-R4F processors and has many other improvements to 1215 optimization for ARM processors. 1216 * GCC now supports the VFPv3 variant with 16 double-precision 1217 registers with -mfpu=vfpv3-d16. The option -mfpu=vfp3 has been 1218 renamed to -mfpu=vfpv3. 1219 * GCC now supports the -mfix-cortex-m3-ldrd option to work around an 1220 erratum on Cortex-M3 processors. 1221 * GCC now supports the __sync_* atomic operations for ARM EABI 1222 GNU/Linux. 1223 * The section anchors optimization is now enabled by default when 1224 optimizing for ARM. 1225 * GCC now uses a new EABI-compatible profiling interface for EABI 1226 targets. This requires a function __gnu_mcount_nc, which is 1227 provided by GNU libc versions 2.8 and later. 1228 1229 AVR 1230 1231 * The -mno-tablejump option has been deprecated because it has the 1232 same effect as the -fno-jump-tables option. 1233 * Added support for these new AVR devices: 1234 + ATA6289 1235 + ATtiny13A 1236 + ATtiny87 1237 + ATtiny167 1238 + ATtiny327 1239 + ATmega8C1 1240 + ATmega16C1 1241 + ATmega32C1 1242 + ATmega8M1 1243 + ATmega16M1 1244 + ATmega32M1 1245 + ATmega32U4 1246 + ATmega16HVB 1247 + ATmega4HVD 1248 + ATmega8HVD 1249 + ATmega64C1 1250 + ATmega64M1 1251 + ATmega16U4 1252 + ATmega32U6 1253 + ATmega128RFA1 1254 + AT90PWM81 1255 + AT90SCR100 1256 + M3000F 1257 + M3000S 1258 + M3001B 1259 1260 IA-32/x86-64 1261 1262 * Support for Intel AES built-in functions and code generation is 1263 available via -maes. 1264 * Support for Intel PCLMUL built-in function and code generation is 1265 available via -mpclmul. 1266 * Support for Intel AVX built-in functions and code generation is 1267 available via -mavx. 1268 * Automatically align the stack for local variables with alignment 1269 requirement. 1270 * GCC can now utilize the SVML library for vectorizing calls to a set 1271 of C99 functions if -mveclibabi=svml is specified and you link to 1272 an SVML ABI compatible library. 1273 * On x86-64, the ABI has been changed in the following cases to 1274 conform to the x86-64 ABI: 1275 + Passing/returning structures with flexible array member: 1276 struct foo 1277 { 1278 int i; 1279 int flex[]; 1280 }; 1281 + Passing/returning structures with complex float member: 1282 struct foo 1283 { 1284 int i; 1285 __complex__ float f; 1286 }; 1287 + Passing/returning unions with long double member: 1288 union foo 1289 { 1290 int x; 1291 long double ld; 1292 }; 1293 Code built with previous versions of GCC that uses any of these is 1294 not compatible with code built with GCC 4.4.0 or later. 1295 * A new target attribute was added to allow programmers to change the 1296 target options like -msse2 or -march=k8 for an individual function. 1297 You can also change the target options via the GCC target pragma 1298 for functions defined after the pragma. 1299 * GCC can now be configured with options --with-arch-32, 1300 --with-arch-64, --with-cpu-32, --with-cpu-64, --with-tune-32 and 1301 --with-tune-64 to control the default optimization separately for 1302 32-bit and 64-bit modes. 1303 1304 IA-32/IA64 1305 1306 * Support for __float128 (TFmode) IEEE quad type and corresponding 1307 TCmode IEEE complex quad type is available via the soft-fp library 1308 on IA-32/IA64 targets. This includes basic arithmetic operations 1309 (addition, subtraction, negation, multiplication and division) on 1310 __float128 real and TCmode complex values, the full set of IEEE 1311 comparisons between __float128 values, conversions to and from 1312 float, double and long double floating point types, as well as 1313 conversions to and from signed or unsigned integer, signed or 1314 unsigned long integer and signed or unsigned quad (TImode, IA64 1315 only) integer types. Additionally, all operations generate the full 1316 set of IEEE exceptions and support the full set of IEEE rounding 1317 modes. 1318 1319 M68K/ColdFire 1320 1321 * GCC now supports instruction scheduling for ColdFire V1, V3 and V4 1322 processors. (Scheduling support for ColdFire V2 processors was 1323 added in GCC 4.3.) 1324 * GCC now supports the -mxgot option to support programs requiring 1325 many GOT entries on ColdFire. 1326 * The m68k-*-linux-gnu target now builds multilibs by default. 1327 1328 MIPS 1329 1330 * MIPS Technologies have extended the original MIPS SVR4 ABI to 1331 include support for procedure linkage tables (PLTs) and copy 1332 relocations. These extensions allow GNU/Linux executables to use a 1333 significantly more efficient code model than the one defined by the 1334 original ABI. 1335 GCC support for this code model is available via a new command-line 1336 option, -mplt. There is also a new configure-time option, 1337 --with-mips-plt, to make -mplt the default. 1338 The new code model requires support from the assembler, the linker, 1339 and the runtime C library. This support is available in binutils 1340 2.19 and GLIBC 2.9. 1341 * GCC can now generate MIPS16 code for 32-bit GNU/Linux executables 1342 and 32-bit GNU/Linux shared libraries. This feature requires GNU 1343 binutils 2.19 or above. 1344 * Support for RMI's XLR processor is now available through the 1345 -march=xlr and -mtune=xlr options. 1346 * 64-bit targets can now perform 128-bit multiplications inline, 1347 instead of relying on a libgcc function. 1348 * Native GNU/Linux toolchains now support -march=native and 1349 -mtune=native, which select the host processor. 1350 * GCC now supports the R10K, R12K, R14K and R16K processors. The 1351 canonical -march= and -mtune= names for these processors are 1352 r10000, r12000, r14000 and r16000 respectively. 1353 * GCC can now work around the side effects of speculative execution 1354 on R10K processors. Please see the documentation of the 1355 -mr10k-cache-barrier option for details. 1356 * Support for the MIPS64 Release 2 instruction set has been added. 1357 The option -march=mips64r2 enables generation of these 1358 instructions. 1359 * GCC now supports Cavium Networks' Octeon processor. This support is 1360 available through the -march=octeon and -mtune=octeon options. 1361 * GCC now supports STMicroelectronics' Loongson 2E/2F processors. The 1362 canonical -march= and -mtune= names for these processors are 1363 loongson2e and loongson2f. 1364 1365 picochip 1366 1367 Picochip is a 16-bit processor. A typical picoChip contains over 250 1368 small cores, each with small amounts of memory. There are three 1369 processor variants (STAN, MEM and CTRL) with different instruction sets 1370 and memory configurations and they can be chosen using the -mae option. 1371 1372 This port is intended to be a "C" only port. 1373 1374 Power Architecture and PowerPC 1375 1376 * GCC now supports the e300c2, e300c3 and e500mc processors. 1377 * GCC now supports Xilinx processors with a single-precision FPU. 1378 * Decimal floating point is now supported for e500 processors. 1379 1380 S/390, zSeries and System z9/z10 1381 1382 * Support for the IBM System z10 EC/BC processor has been added. When 1383 using the -march=z10 option, the compiler will generate code making 1384 use of instructions provided by the General-Instruction-Extension 1385 Facility and the Execute-Extension Facility. 1386 1387 VxWorks 1388 1389 * GCC now supports the thread-local storage mechanism used on 1390 VxWorks. 1391 1392 Xtensa 1393 1394 * GCC now supports thread-local storage (TLS) for Xtensa processor 1395 configurations that include the Thread Pointer option. TLS also 1396 requires support from the assembler and linker; this support is 1397 provided in the GNU binutils beginning with version 2.19. 1398 1399 Documentation improvements 1400 1401 Other significant improvements 1402 1403 GCC 4.4.1 1404 1405 This is the [13]list of problem reports (PRs) from GCC's bug tracking 1406 system that are known to be fixed in the 4.4.1 release. This list might 1407 not be complete (that is, it is possible that some PRs that have been 1408 fixed are not listed here). 1409 1410 GCC 4.4.2 1411 1412 This is the [14]list of problem reports (PRs) from GCC's bug tracking 1413 system that are known to be fixed in the 4.4.2 release. This list might 1414 not be complete (that is, it is possible that some PRs that have been 1415 fixed are not listed here). 1416 1417 GCC 4.4.3 1418 1419 This is the [15]list of problem reports (PRs) from GCC's bug tracking 1420 system that are known to be fixed in the 4.4.3 release. This list might 1421 not be complete (that is, it is possible that some PRs that have been 1422 fixed are not listed here). 1423 1424 GCC 4.4.4 1425 1426 This is the [16]list of problem reports (PRs) from GCC's bug tracking 1427 system that are known to be fixed in the 4.4.4 release. This list might 1428 not be complete (that is, it is possible that some PRs that have been 1429 fixed are not listed here). 1430 1431 GCC 4.4.5 1432 1433 This is the [17]list of problem reports (PRs) from GCC's bug tracking 1434 system that are known to be fixed in the 4.4.5 release. This list might 1435 not be complete (that is, it is possible that some PRs that have been 1436 fixed are not listed here). 1437 1438 GCC 4.4.6 1439 1440 This is the [18]list of problem reports (PRs) from GCC's bug tracking 1441 system that are known to be fixed in the 4.4.6 release. This list might 1442 not be complete (that is, it is possible that some PRs that have been 1443 fixed are not listed here). 1444 1445 1446 For questions related to the use of GCC, please consult these web 1447 pages and the [19]GCC manuals. If that fails, the 1448 [20]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 1449 web pages and the development of GCC are welcome on our developer 1450 list at [21]gcc (a] gcc.gnu.org. All of [22]our lists have public 1451 archives. 1452 1453 Copyright (C) [23]Free Software Foundation, Inc. Verbatim copying and 1454 distribution of this entire article is permitted in any medium, 1455 provided this notice is preserved. 1456 1457 These pages are [24]maintained by the GCC team. Last modified 1458 2011-04-25[25]. 1459 1460 References 1461 1462 1. http://gcc.gnu.org/gcc-4.4/changes.html#4.4.6 1463 2. http://gcc.gnu.org/gcc-4.3/changes.html#obsoleted 1464 3. http://gcc.gnu.org/gcc-4.4/porting_to.html 1465 4. http://gcc.gnu.org/wiki/Graphite 1466 5. http://openmp.org/wp/openmp-specifications/ 1467 6. http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1040.pdf 1468 7. http://gcc.gnu.org/gcc-4.4/cxx0x_status.html 1469 8. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#id476343 1470 9. http://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html 1471 10. http://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWarray-temporaries_007d-125 1472 11. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcheck-array-temporaries_007d-221 1473 12. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html#index-g_t_0040code_007bbackslash_007d-34 1474 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.1 1475 14. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.2 1476 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.3 1477 16. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.4 1478 17. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.5 1479 18. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.6 1480 19. http://gcc.gnu.org/onlinedocs/ 1481 20. mailto:gcc-help (a] gcc.gnu.org 1482 21. mailto:gcc (a] gcc.gnu.org 1483 22. http://gcc.gnu.org/lists.html 1484 23. http://www.fsf.org/ 1485 24. http://gcc.gnu.org/about.html 1486 25. http://validator.w3.org/check/referer 1487 ====================================================================== 1488 http://gcc.gnu.org/gcc-4.3/index.html 1489 1490 GCC 4.3 Release Series 1491 1492 May 22, 2010 1493 1494 The [1]GNU project and the GCC developers are pleased to announce the 1495 release of GCC 4.3.5. 1496 1497 This release is a bug-fix release, containing fixes for regressions in 1498 GCC 4.3.4 relative to previous releases of GCC. 1499 1500 Release History 1501 1502 GCC 4.3.5 1503 May 22, 2010 ([2]changes) 1504 1505 GCC 4.3.4 1506 August 4, 2009 ([3]changes) 1507 1508 GCC 4.3.3 1509 January 24, 2009 ([4]changes) 1510 1511 GCC 4.3.2 1512 August 27, 2008 ([5]changes) 1513 1514 GCC 4.3.1 1515 June 6, 2008 ([6]changes) 1516 1517 GCC 4.3.0 1518 March 5, 2008 ([7]changes) 1519 1520 References and Acknowledgements 1521 1522 GCC used to stand for the GNU C Compiler, but since the compiler 1523 supports several other languages aside from C, it now stands for the 1524 GNU Compiler Collection. 1525 1526 A list of [8]successful builds is updated as new information becomes 1527 available. 1528 1529 The GCC developers would like to thank the numerous people that have 1530 contributed new features, improvements, bug fixes, and other changes as 1531 well as test results to GCC. This [9]amazing group of volunteers is 1532 what makes GCC successful. 1533 1534 For additional information about GCC please refer to the [10]GCC 1535 project web site or contact the [11]GCC development mailing list. 1536 1537 To obtain GCC please use [12]our mirror sites or [13]our SVN server. 1538 1539 1540 For questions related to the use of GCC, please consult these web 1541 pages and the [14]GCC manuals. If that fails, the 1542 [15]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 1543 web pages and the development of GCC are welcome on our developer 1544 list at [16]gcc (a] gcc.gnu.org. All of [17]our lists have public 1545 archives. 1546 1547 Copyright (C) [18]Free Software Foundation, Inc. Verbatim copying and 1548 distribution of this entire article is permitted in any medium, 1549 provided this notice is preserved. 1550 1551 These pages are [19]maintained by the GCC team. Last modified 1552 2011-04-25[20]. 1553 1554 References 1555 1556 1. http://www.gnu.org/ 1557 2. http://gcc.gnu.org/gcc-4.3/changes.html 1558 3. http://gcc.gnu.org/gcc-4.3/changes.html 1559 4. http://gcc.gnu.org/gcc-4.3/changes.html 1560 5. http://gcc.gnu.org/gcc-4.3/changes.html 1561 6. http://gcc.gnu.org/gcc-4.3/changes.html 1562 7. http://gcc.gnu.org/gcc-4.3/changes.html 1563 8. http://gcc.gnu.org/gcc-4.3/buildstat.html 1564 9. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 1565 10. http://gcc.gnu.org/index.html 1566 11. mailto:gcc (a] gcc.gnu.org 1567 12. http://gcc.gnu.org/mirrors.html 1568 13. http://gcc.gnu.org/svn.html 1569 14. http://gcc.gnu.org/onlinedocs/ 1570 15. mailto:gcc-help (a] gcc.gnu.org 1571 16. mailto:gcc (a] gcc.gnu.org 1572 17. http://gcc.gnu.org/lists.html 1573 18. http://www.fsf.org/ 1574 19. http://gcc.gnu.org/about.html 1575 20. http://validator.w3.org/check/referer 1576 ====================================================================== 1577 http://gcc.gnu.org/gcc-4.3/changes.html 1578 1579 GCC 4.3 Release Series 1580 Changes, New Features, and Fixes 1581 1582 The latest release in the 4.3 release series is [1]GCC 4.3.5. 1583 1584 Caveats 1585 1586 * GCC requires the [2]GMP and [3]MPFR libraries for building all the 1587 various front-end languages it supports. See the [4]prerequisites 1588 page for version requirements. 1589 * ColdFire targets now treat long double as having the same format as 1590 double. In earlier versions of GCC, they used the 68881 long double 1591 format instead. 1592 * The m68k-uclinux target now uses the same calling conventions as 1593 m68k-linux-gnu. You can select the original calling conventions by 1594 configuring for m68k-uclinuxoldabi instead. Note that 1595 m68k-uclinuxoldabi also retains the original 80-bit long double on 1596 ColdFire targets. 1597 * The -fforce-mem option has been removed because it has had no 1598 effect in the last few GCC releases. 1599 * The i386 -msvr3-shlib option has been removed since it is no longer 1600 used. 1601 * Fastcall for i386 has been changed not to pass aggregate arguments 1602 in registers, following Microsoft compilers. 1603 * Support for the AOF assembler has been removed from the ARM back 1604 end; this affects only the targets arm-semi-aof and armel-semi-aof, 1605 which are no longer recognized. We removed these targets without a 1606 deprecation period because we discovered that they have been 1607 unusable since GCC 4.0.0. 1608 * Support for the TMS320C3x/C4x processor (targets c4x-* and tic4x-*) 1609 has been removed. This support had been deprecated since GCC 4.0.0. 1610 * Support for a number of older systems and recently unmaintained or 1611 untested target ports of GCC has been declared obsolete in GCC 4.3. 1612 Unless there is activity to revive them, the next release of GCC 1613 will have their sources permanently removed. 1614 All GCC ports for the following processor architectures have been 1615 declared obsolete: 1616 + Morpho MT (mt-*) 1617 The following aliases for processor architectures have been 1618 declared obsolete. Users should use the indicated generic target 1619 names instead, with compile-time options such as -mcpu or 1620 configure-time options such as --with-cpu to control the 1621 configuration more precisely. 1622 + strongarm*-*-*, ep9312*-*-*, xscale*-*-* (use arm*-*-* 1623 instead). 1624 + parisc*-*-* (use hppa*-*-* instead). 1625 + m680[012]0-*-* (use m68k-*-* instead). 1626 All GCC ports for the following operating systems have been 1627 declared obsolete: 1628 + BeOS (*-*-beos*) 1629 + kaOS (*-*-kaos*) 1630 + GNU/Linux using the a.out object format (*-*-linux*aout*) 1631 + GNU/Linux using version 1 of the GNU C Library 1632 (*-*-linux*libc1*) 1633 + Solaris versions before Solaris 7 (*-*-solaris2.[0-6], 1634 *-*-solaris2.[0-6].*) 1635 + Miscellaneous System V (*-*-sysv*) 1636 + WindISS (*-*-windiss*) 1637 Also, those for some individual systems on particular architectures 1638 have been obsoleted: 1639 + UNICOS/mk on DEC Alpha (alpha*-*-unicosmk*) 1640 + CRIS with a.out object format (cris-*-aout) 1641 + BSD 4.3 on PA-RISC (hppa1.1-*-bsd*) 1642 + OSF/1 on PA-RISC (hppa1.1-*-osf*) 1643 + PRO on PA-RISC (hppa1.1-*-pro*) 1644 + Sequent PTX on IA32 (i[34567]86-sequent-ptx4*, 1645 i[34567]86-sequent-sysv4*) 1646 + SCO Open Server 5 on IA32 (i[34567]86-*-sco3.2v5*) 1647 + UWIN on IA32 (i[34567]86-*-uwin*) (support for UWIN as a host 1648 was previously [5]removed in 2001, leaving only the support 1649 for UWIN as a target now being deprecated) 1650 + ChorusOS on PowerPC (powerpc-*-chorusos*) 1651 + All VAX configurations apart from NetBSD and OpenBSD 1652 (vax-*-bsd*, vax-*-sysv*, vax-*-ultrix*) 1653 * The [6]-Wconversion option has been modified. Its purpose now is to 1654 warn for implicit conversions that may alter a value. This new 1655 behavior is available for both C and C++. Warnings about 1656 conversions between signed and unsigned integers can be disabled by 1657 using -Wno-sign-conversion. In C++, they are disabled by default 1658 unless -Wsign-conversion is explicitly requested. The old behavior 1659 of -Wconversion, that is, warn for prototypes causing a type 1660 conversion that is different from what would happen to the same 1661 argument in the absence of a prototype, has been moved to a new 1662 option -Wtraditional-conversion, which is only available for C. 1663 * The -m386, -m486, -mpentium and -mpentiumpro tuning options have 1664 been removed because they were deprecated for more than 3 GCC major 1665 releases. Use -mtune=i386, -mtune=i486, -mtune=pentium or 1666 -mtune=pentiumpro as a replacement. 1667 * The -funsafe-math-optimizations option now automatically turns on 1668 -fno-trapping-math in addition to -fno-signed-zeros, as it enables 1669 reassociation and thus may introduce or remove traps. 1670 * The -ftree-vectorize option is now on by default under -O3. In 1671 order to generate code for a SIMD extension, it has to be enabled 1672 as well: use -maltivec for PowerPC platforms and -msse/-msse2 for 1673 i?86 and x86_64. 1674 * More information on porting to GCC 4.3 from previous versions of 1675 GCC can be found in the [7]porting guide for this release. 1676 1677 General Optimizer Improvements 1678 1679 * The GCC middle-end has been integrated with the [8]MPFR library. 1680 This allows GCC to evaluate and replace at compile-time calls to 1681 built-in math functions having constant arguments with their 1682 mathematically equivalent results. In making use of [9]MPFR, GCC 1683 can generate correct results regardless of the math library 1684 implementation or floating point precision of the host platform. 1685 This also allows GCC to generate identical results regardless of 1686 whether one compiles in native or cross-compile configurations to a 1687 particular target. The following built-in functions take advantage 1688 of this new capability: acos, acosh, asin, asinh, atan2, atan, 1689 atanh, cbrt, cos, cosh, drem, erf, erfc, exp10, exp2, exp, expm1, 1690 fdim, fma, fmax, fmin, gamma_r, hypot, j0, j1, jn, lgamma_r, log10, 1691 log1p, log2, log, pow10, pow, remainder, remquo, sin, sincos, sinh, 1692 tan, tanh, tgamma, y0, y1 and yn. The float and long double 1693 variants of these functions (e.g. sinf and sinl) are also handled. 1694 The sqrt and cabs functions with constant arguments were already 1695 optimized in prior GCC releases. Now they also use [10]MPFR. 1696 * A new forward propagation pass on RTL was added. The new pass 1697 replaces several slower transformations, resulting in compile-time 1698 improvements as well as better code generation in some cases. 1699 * A new command-line switch -frecord-gcc-switches has been added to 1700 GCC, although it is only enabled for some targets. The switch 1701 causes the command line that was used to invoke the compiler to be 1702 recorded into the object file that is being created. The exact 1703 format of this recording is target and binary file format 1704 dependent, but it usually takes the form of a note section 1705 containing ASCII text. The switch is related to the -fverbose-asm 1706 switch, but that one only records the information in the assembler 1707 output file as comments, so the information never reaches the 1708 object file. 1709 * The inliner heuristic is now aware of stack frame consumption. New 1710 command-line parameters --param large-stack-frame and --param 1711 large-stack-frame-growth can be used to limit stack frame size 1712 growth caused by inlining. 1713 * During feedback directed optimizations, the expected block size the 1714 memcpy, memset and bzero functions operate on is discovered and for 1715 cases of commonly used small sizes, specialized inline code is 1716 generated. 1717 * __builtin_expect no longer requires its argument to be a compile 1718 time constant. 1719 * Interprocedural optimization was reorganized to work on functions 1720 in SSA form. This enables more precise and cheaper dataflow 1721 analysis and makes writing interprocedural optimizations easier. 1722 The following improvements have been implemented on top of this 1723 framework: 1724 + Pre-inline optimization: Selected local optimization passes 1725 are run before the inliner (and other interprocedural passes) 1726 are executed. This significantly improves the accuracy of code 1727 growth estimates used by the inliner and reduces the overall 1728 memory footprint for large compilation units. 1729 + Early inlining (a simple bottom-up inliner pass inlining only 1730 functions whose body is smaller than the expected call 1731 overhead) is now executed with the early optimization passes, 1732 thus inlining already optimized function bodies into an 1733 unoptimized function that is subsequently optimized by early 1734 optimizers. This enables the compiler to quickly eliminate 1735 abstraction penalty in C++ programs. 1736 + Interprocedural constant propagation now operate on SSA form 1737 increasing accuracy of the analysis. 1738 * A new internal representation for GIMPLE statements has been 1739 contributed, resulting in compile-time memory savings. 1740 * The vectorizer was enhanced to support vectorization of outer 1741 loops, intra-iteration parallelism (loop-aware SLP), vectorization 1742 of strided accesses and loops with multiple data-types. Run-time 1743 dependency testing using loop versioning was added. The cost model, 1744 turned on by -fvect-cost-model, was developed. 1745 1746 New Languages and Language specific improvements 1747 1748 * We have added new command-line options 1749 -finstrument-functions-exclude-function-list and 1750 -finstrument-functions-exclude-file-list. They provide more control 1751 over which functions are annotated by the -finstrument-functions 1752 option. 1753 1754 C family 1755 1756 * Implicit conversions between generic vector types are now only 1757 permitted when the two vectors in question have the same number of 1758 elements and compatible element types. (Note that the restriction 1759 involves compatible element types, not implicitly-convertible 1760 element types: thus, a vector type with element type int may not be 1761 implicitly converted to a vector type with element type unsigned 1762 int.) This restriction, which is in line with specifications for 1763 SIMD architectures such as AltiVec, may be relaxed using the flag 1764 -flax-vector-conversions. This flag is intended only as a 1765 compatibility measure and should not be used for new code. 1766 * -Warray-bounds has been added and is now enabled by default for 1767 -Wall . It produces warnings for array subscripts that can be 1768 determined at compile time to be always out of bounds. 1769 -Wno-array-bounds will disable the warning. 1770 * The constructor and destructor function attributes now accept 1771 optional priority arguments which control the order in which the 1772 constructor and destructor functions are run. 1773 * New [11]command-line options -Wtype-limits, 1774 -Wold-style-declaration, -Wmissing-parameter-type, -Wempty-body, 1775 -Wclobbered and -Wignored-qualifiers have been added for finer 1776 control of the diverse warnings enabled by -Wextra. 1777 * A new function attribute alloc_size has been added to mark up 1778 malloc style functions. For constant sized allocations this can be 1779 used to find out the size of the returned pointer using the 1780 __builtin_object_size() function for buffer overflow checking and 1781 similar. This supplements the already built-in malloc and calloc 1782 constant size handling. 1783 * Integer constants written in binary are now supported as a GCC 1784 extension. They consist of a prefix 0b or 0B, followed by a 1785 sequence of 0 and 1 digits. 1786 * A new predefined macro __COUNTER__ has been added. It expands to 1787 sequential integral values starting from 0. In conjunction with the 1788 ## operator, this provides a convenient means to generate unique 1789 identifiers. 1790 * A new command-line option -fdirectives-only has been added. It 1791 enables a special preprocessing mode which improves the performance 1792 of applications like distcc and ccache. 1793 * Fixed-point data types and operators have been added. They are 1794 based on Chapter 4 of the Embedded-C specification (n1169.pdf). 1795 Currently, only MIPS targets are supported. 1796 * Decimal floating-point arithmetic based on draft ISO/IEC TR 24732, 1797 N1241, is now supported as a GCC extension to C for targets 1798 i[34567]86-*-linux-gnu, powerpc*-*-linux-gnu, s390*-ibm-linux-gnu, 1799 and x86_64-*-linux-gnu. The feature introduces new data types 1800 _Decimal32, _Decimal64, and _Decimal128 with constant suffixes DF, 1801 DD, and DL. 1802 1803 C++ 1804 1805 * [12]Experimental support for the upcoming ISO C++ standard, C++0x. 1806 * -Wc++0x-compat has been added and is now enabled by default for 1807 -Wall. It produces warnings for constructs whose meaning differs 1808 between ISO C++ 1998 and C++0x. 1809 * The -Wparentheses option now works for C++ as it does for C. It 1810 warns if parentheses are omitted when operators with confusing 1811 precedence are nested. It also warns about ambiguous else 1812 statements. Since -Wparentheses is enabled by -Wall, this may cause 1813 additional warnings with existing C++ code which uses -Wall. These 1814 new warnings may be disabled by using -Wall -Wno-parentheses. 1815 * The -Wmissing-declarations now works for C++ as it does for C. 1816 * The -fvisibility-ms-compat flag was added, to make it easier to 1817 port larger projects using shared libraries from Microsoft's Visual 1818 Studio to ELF and Mach-O systems. 1819 * C++ attribute handling has been overhauled for template arguments 1820 (ie dependent types). In particular, __attribute__((aligned(T))); 1821 works for C++ types. 1822 1823 Runtime Library (libstdc++) 1824 1825 * [13]Experimental support for the upcoming ISO C++ standard, C++0x. 1826 * Support for TR1 mathematical special functions and regular 1827 expressions. ([14]Implementation status of TR1) 1828 * Default what implementations give more elaborate exception strings 1829 for bad_cast, bad_typeid, bad_exception, and bad_alloc. 1830 * Header dependencies have been streamlined, reducing unnecessary 1831 includes and pre-processed bloat. 1832 * Variadic template implementations of items in <tuple> and 1833 <functional>. 1834 * An experimental [15]parallel mode has been added. This is a 1835 parallel implementation of many C++ Standard library algorithms, 1836 like std::accumulate, std::for_each, std::transform, or std::sort, 1837 to give but four examples. These algorithms can be substituted for 1838 the normal (sequential) libstdc++ algorithms on a piecemeal basis, 1839 or all existing algorithms can be transformed via the 1840 -D_GLIBCXX_PARALLEL macro. 1841 * Debug mode versions of classes in <unordered_set> and 1842 <unordered_map>. 1843 * Formal deprecation of <ext/hash_set> and <ext/hash_map>, which are 1844 now <backward/hash_set> and <backward/hash_map>. This code: 1845 #include <ext/hash_set> 1846 __gnu_cxx::hash_set<int> s; 1847 1848 Can be transformed (in order of preference) to: 1849 #include <tr1/unordered_set> 1850 std::tr1::unordered_set<int> s; 1851 1852 or 1853 #include <backward/hash_set> 1854 __gnu_cxx::hash_set<int> s; 1855 1856 Similar transformations apply to __gnu_cxx::hash_map, 1857 __gnu_cxx::hash_multimap, __gnu_cxx::hash_set, 1858 __gnu_cxx::hash_multiset. 1859 1860 Fortran 1861 1862 * Due to the fact that the [16]GMP and [17]MPFR libraries are 1863 required for all languages, Fortran is no longer special in this 1864 regard and is available by default. 1865 * The [18]-fexternal-blas option has been added, which generates 1866 calls to BLAS routines for intrinsic matrix operations such as 1867 matmul rather than using the built-in algorithms. 1868 * Support to give a backtrace (compiler flag -fbacktrace or 1869 environment variable GFORTRAN_ERROR_BACKTRACE; on glibc systems 1870 only) or a core dump (-fdump-core, GFORTRAN_ERROR_DUMPCORE) when a 1871 run-time error occured. 1872 * GNU Fortran now defines __GFORTRAN__ when it runs the C 1873 preprocessor (CPP). 1874 * The [19]-finit-local-zero, -finit-real, -finit-integer, 1875 -finit-character, and -finit-logical options have been added, which 1876 can be used to initialize local variables. 1877 * The intrinsic procedures [20]GAMMA and [21]LGAMMA have been added, 1878 which calculate the Gamma function and its logarithm. Use EXTERNAL 1879 gamma if you want to use your own gamma function. 1880 * GNU Fortran now regards the backslash character as literal (as 1881 required by the Fortran 2003 standard); using [22]-fbackslash GNU 1882 Fortran interprets backslashes as C-style escape characters. 1883 * The [23]interpretation of binary, octal and hexadecimal (BOZ) 1884 literal constants has been changed. Before they were always 1885 interpreted as integer; now they are bit-wise transferred as 1886 argument of INT, REAL, DBLE and CMPLX as required by the Fortran 1887 2003 standard, and for real and complex variables in DATA 1888 statements or when directly assigned to real and complex variables. 1889 Everywhere else and especially in expressions they are still 1890 regarded as integer constants. 1891 * Fortran 2003 support has been extended: 1892 + Intrinsic statements IMPORT, PROTECTED, VALUE and VOLATILE 1893 + Pointer intent 1894 + Intrinsic module ISO_ENV_FORTRAN 1895 + Interoperability with C (ISO C Bindings) 1896 + ABSTRACT INTERFACES and PROCEDURE statements (without POINTER 1897 attribute) 1898 + Fortran 2003 BOZ 1899 1900 Java (GCJ) 1901 1902 * GCJ now uses the Eclipse Java compiler for its Java parsing needs. 1903 This enables the use of all 1.5 language features, and fixes most 1904 existing front end bugs. 1905 * libgcj now supports all 1.5 language features which require runtime 1906 support: foreach, enum, annotations, generics, and auto-boxing. 1907 * We've made many changes to the tools shipped with gcj. 1908 + The old jv-scan tool has been removed. This tool never really 1909 worked properly. There is no replacement. 1910 + gcjh has been rewritten. Some of its more obscure options no 1911 longer work, but are still recognized in an attempt at 1912 compatibility. gjavah is a new program with similar 1913 functionality but different command-line options. 1914 + grmic and grmiregistry have been rewritten. grmid has been 1915 added. 1916 + gjar replaces the old fastjar. 1917 + gjarsigner (used for signing jars), gkeytool (used for key 1918 management), gorbd (for CORBA), gserialver (computes 1919 serialization UIDs), and gtnameserv (also for CORBA) are now 1920 installed. 1921 * The ability to dump the contents of the java run time heap to a 1922 file for off-line analysis has been added. The heap dumps may be 1923 analyzed with the new gc-analyze tool. They may be generated on 1924 out-of-memory conditions or on demand and are controlled by the new 1925 run time class gnu.gcj.util.GCInfo. 1926 * java.util.TimeZone can now read files from /usr/share/zoneinfo to 1927 provide correct, updated, timezone information. This means that 1928 packagers no longer have to update libgcj when a time zone change 1929 is published. 1930 1931 New Targets and Target Specific Improvements 1932 1933 IA-32/x86-64 1934 1935 * Tuning for Intel Core 2 processors is available via -mtune=core2 1936 and -march=core2. 1937 * Tuning for AMD Geode processors is available via -mtune=geode and 1938 -march=geode. 1939 * Code generation of block move (memcpy) and block set (memset) was 1940 rewritten. GCC can now pick the best algorithm (loop, unrolled 1941 loop, instruction with rep prefix or a library call) based on the 1942 size of the block being copied and the CPU being optimized for. A 1943 new option -minline-stringops-dynamically has been added. With this 1944 option string operations of unknown size are expanded such that 1945 small blocks are copied by in-line code, while for large blocks a 1946 library call is used. This results in faster code than 1947 -minline-all-stringops when the library implementation is capable 1948 of using cache hierarchy hints. The heuristic choosing the 1949 particular algorithm can be overwritten via -mstringop-strategy. 1950 Newly also memset of values different from 0 is inlined. 1951 * GCC no longer places the cld instruction before string operations. 1952 Both i386 and x86-64 ABI documents mandate the direction flag to be 1953 clear at the entry of a function. It is now invalid to set the flag 1954 in asm statement without reseting it afterward. 1955 * Support for SSSE3 built-in functions and code generation are 1956 available via -mssse3. 1957 * Support for SSE4.1 built-in functions and code generation are 1958 available via -msse4.1. 1959 * Support for SSE4.2 built-in functions and code generation are 1960 available via -msse4.2. 1961 * Both SSE4.1 and SSE4.2 support can be enabled via -msse4. 1962 * A new set of options -mpc32, -mpc64 and -mpc80 have been added to 1963 allow explicit control of x87 floating point precision. 1964 * Support for __float128 (TFmode) IEEE quad type and corresponding 1965 TCmode IEEE complex quad type is available via the soft-fp library 1966 on x86_64 targets. This includes basic arithmetic operations 1967 (addition, subtraction, negation, multiplication and division) on 1968 __float128 real and TCmode complex values, the full set of IEEE 1969 comparisons between __float128 values, conversions to and from 1970 float, double and long double floating point types, as well as 1971 conversions to and from signed or unsigned integer, signed or 1972 unsigned long integer and signed or unsigned quad (TImode) integer 1973 types. Additionally, all operations generate the full set of IEEE 1974 exceptions and support the full set of IEEE rounding modes. 1975 * GCC can now utilize the ACML library for vectorizing calls to a set 1976 of C99 functions on x86_64 if -mveclibabi=acml is specified and you 1977 link to an ACML ABI compatible library. 1978 1979 ARM 1980 1981 * Compiler and Library support for Thumb-2 and the ARMv7 architecture 1982 has been added. 1983 1984 CRIS 1985 1986 New features 1987 1988 * Compiler and Library support for the CRIS v32 architecture, as 1989 found in Axis Communications ETRAX FS and ARTPEC-3 chips, has been 1990 added. 1991 1992 Configuration changes 1993 1994 * The cris-*-elf target now includes support for CRIS v32, including 1995 libraries, through the -march=v32 option. 1996 * A new crisv32-*-elf target defaults to generate code for CRIS v32. 1997 * A new crisv32-*-linux* target defaults to generate code for CRIS 1998 v32. 1999 * The cris-*-aout target has been obsoleted. 2000 2001 Improved support for built-in functions 2002 2003 * GCC can now use the lz and swapwbr instructions to implement the 2004 __builtin_clz, __builtin_ctz and __builtin_ffs family of functions. 2005 * __builtin_bswap32 is now implemented using the swapwb instruction, 2006 when available. 2007 2008 m68k and ColdFire 2009 2010 New features 2011 2012 * Support for several new ColdFire processors has been added. You can 2013 generate code for them using the new -mcpu option. 2014 * All targets now support ColdFire processors. 2015 * m68k-uclinux targets have improved support for C++ constructors and 2016 destructors, and for shared libraries. 2017 * It is now possible to set breakpoints on the first or last line of 2018 a function, even if there are no statements on that line. 2019 2020 Optimizations 2021 2022 * Support for sibling calls has been added. 2023 * More use is now made of the ColdFire mov3q instruction. 2024 * __builtin_clz is now implemented using the ff1 ColdFire 2025 instruction, when available. 2026 * GCC now honors the -m68010 option. 68010 code now uses clr rather 2027 than move to zero volatile memory. 2028 * 68020 targets and above can now use symbol(index.size*scale) 2029 addresses for indexed array accesses. Earlier compilers would 2030 always load the symbol into a base register first. 2031 2032 Configuration changes 2033 2034 * All m68k and ColdFire targets now allow the default processor to be 2035 set at configure time using --with-cpu. 2036 * A --with-arch configuration option has been added. This option 2037 allows you to restrict a target to ColdFire or non-ColdFire 2038 processors. 2039 2040 Preprocessor macros 2041 2042 * An __mcfv*__ macro is now defined for all ColdFire targets. 2043 (Earlier versions of GCC only defined __mcfv4e__.) 2044 * __mcf_cpu_*, __mcf_family_* and __mcffpu__ macros have been added. 2045 * All targets now define __mc68010 and __mc68010__ when generating 2046 68010 code. 2047 2048 Command-line changes 2049 2050 * New command-line options -march, -mcpu, -mtune and -mhard-float 2051 have been added. These options apply to both m68k and ColdFire 2052 targets. 2053 * -mno-short, -mno-bitfield and -mno-rtd are now accepted as negative 2054 versions of -mshort, etc. 2055 * -fforce-addr has been removed. It is now ignored by the compiler. 2056 2057 Other improvements 2058 2059 * ColdFire targets now try to maintain a 4-byte-aligned stack where 2060 possible. 2061 * m68k-uclinux targets now try to avoid situations that lead to the 2062 load-time error: BINFMT_FLAT: reloc outside program. 2063 2064 MIPS 2065 2066 Changes to existing configurations 2067 2068 * libffi and libjava now support all three GNU/Linux ABIs: o32, n32 2069 and n64. Every GNU/Linux configuration now builds these libraries 2070 by default. 2071 * GNU/Linux configurations now generate -mno-shared code unless 2072 overridden by -fpic, -fPIC, -fpie or -fPIE. 2073 * mipsisa32*-linux-gnu configurations now generate hard-float code by 2074 default, just like other mipsisa32* and mips*-linux-gnu 2075 configurations. You can build a soft-float version of any 2076 mips*-linux-gnu configuration by passing --with-float=soft to 2077 configure. 2078 * mips-wrs-vxworks now supports run-time processes (RTPs). 2079 2080 Changes to existing command-line options 2081 2082 * The -march and -mtune options no longer accept 24k as a processor 2083 name. Please use 24kc, 24kf2_1 or 24kf1_1 instead. 2084 * The -march and -mtune options now accept 24kf2_1, 24kef2_1 and 2085 34kf2_1 as synonyms for 24kf, 24kef and 34kf respectively. The 2086 options also accept 24kf1_1, 24kef1_1 and 34kf1_1 as synonyms for 2087 24kx, 24kex and 34kx. 2088 2089 New configurations 2090 2091 GCC now supports the following configurations: 2092 * mipsisa32r2*-linux-gnu*, which generates MIPS32 revision 2 code by 2093 default. Earlier releases also recognized this configuration, but 2094 they treated it in the same way as mipsisa32*-linux-gnu*. Note that 2095 you can customize any mips*-linux-gnu* configuration to a 2096 particular ISA or processor by passing an appropriate --with-arch 2097 option to configure. 2098 * mipsisa*-sde-elf*, which provides compatibility with MIPS 2099 Technologies' SDE toolchains. The configuration uses the SDE 2100 libraries by default, but you can use it like other newlib-based 2101 ELF configurations by passing --with-newlib to configure. It is the 2102 only configuration besides mips64vr*-elf* to build MIPS16 as well 2103 as non-MIPS16 libraries. 2104 * mipsisa*-elfoabi*, which is similar to the general mipsisa*-elf* 2105 configuration, but uses the o32 and o64 ABIs instead of the 32-bit 2106 and 64-bit forms of the EABI. 2107 2108 New processors and application-specific extensions 2109 2110 * Support for the SmartMIPS ASE is available through the new 2111 -msmartmips option. 2112 * Support for revision 2 of the DSP ASE is available through the new 2113 -mdspr2 option. A new preprocessor macro called __mips_dsp_rev 2114 indicates the revision of the ASE in use. 2115 * Support for the 4KS and 74K families of processors is available 2116 through the -march and -mtune options. 2117 2118 Improved support for built-in functions 2119 2120 * GCC can now use load-linked, store-conditional and sync 2121 instructions to implement atomic built-in functions such as 2122 __sync_fetch_and_add. The memory reference must be 4 bytes wide for 2123 32-bit targets and either 4 or 8 bytes wide for 64-bit targets. 2124 * GCC can now use the clz and dclz instructions to implement the 2125 __builtin_ctz and __builtin_ffs families of functions. 2126 * There is a new __builtin___clear_cache function for flushing the 2127 instruction cache. GCC expands this function inline on MIPS32 2128 revision 2 targets, otherwise it calls the function specified by 2129 -mcache-flush-func. 2130 2131 MIPS16 improvements 2132 2133 * GCC can now compile objects that contain a mixture of MIPS16 and 2134 non-MIPS16 code. There are two new attributes, mips16 and nomips16, 2135 for specifying which mode a function should use. 2136 * A new option called -minterlink-mips16 makes non-MIPS16 code 2137 link-compatible with MIPS16 code. 2138 * After many bug fixes, the long-standing MIPS16 -mhard-float support 2139 should now work fairly reliably. 2140 * GCC can now use the MIPS16e save and restore instructions. 2141 * -fsection-anchors now works in MIPS16 mode. MIPS16 code compiled 2142 with -G0 -fsection-anchors is often smaller than code compiled with 2143 -G8. However, please note that you must usually compile all objects 2144 in your application with the same -G option; see the documentation 2145 of -G for details. 2146 * A new option called-mcode-readable specifies which instructions are 2147 allowed to load from the code segment. -mcode-readable=yes is the 2148 default and says that any instruction may load from the code 2149 segment. The other alternatives are -mcode-readable=pcrel, which 2150 says that only PC-relative MIPS16 instructions may load from the 2151 code segment, and -mcode-readable=no, which says that no 2152 instruction may do so. Please see the documentation for more 2153 details, including example uses. 2154 2155 Small-data improvements 2156 2157 There are three new options for controlling small data: 2158 * -mno-extern-sdata, which disables small-data accesses for 2159 externally-defined variables. Code compiled with -Gn 2160 -mno-extern-sdata will be link-compatible with any -G setting 2161 between -G0 and -Gn inclusive. 2162 * -mno-local-sdata, which disables the use of small-data sections for 2163 data that is not externally visible. This option can be a useful 2164 way of reducing small-data usage in less performance-critical parts 2165 of an application. 2166 * -mno-gpopt, which disables the use of the $gp register while still 2167 honoring the -G limit when placing externally-visible data. This 2168 option implies -mno-extern-sdata and -mno-local-sdata and it can be 2169 useful in situations where $gp does not necessarily hold the 2170 expected value. 2171 2172 Miscellaneous improvements 2173 2174 * There is a new option called -mbranch-cost for tweaking the 2175 perceived cost of branches. 2176 * If GCC is configured to use a version of GAS that supports the 2177 .gnu_attribute directive, it will use that directive to record 2178 certain properties of the output code. .gnu_attribute is new to GAS 2179 2.18. 2180 * There are two new function attributes, near and far, for overriding 2181 the command-line setting of -mlong-calls on a function-by-function 2182 basis. 2183 * -mfp64, which previously required a 64-bit target, now works with 2184 MIPS32 revision 2 targets as well. The mipsisa*-elfoabi* and 2185 mipsisa*-sde-elf* configurations provide suitable library support. 2186 * GCC now recognizes the -mdmx and -mmt options and passes them down 2187 to the assembler. It does nothing else with the options at present. 2188 2189 SPU (Synergistic Processor Unit) of the Cell Broadband Engine Architecture 2190 (BEA) 2191 2192 * Support has been added for this new architecture. 2193 2194 RS6000 (POWER/PowerPC) 2195 2196 * Support for the PowerPC 750CL paired-single instructions has been 2197 added with a new powerpc-*-linux*paired* target configuration. It 2198 is enabled by an associated -mpaired option and can be accessed 2199 using new built-in functions. 2200 * Support for auto-detecting architecture and system configuration to 2201 auto-select processor optimization tuning. 2202 * Support for VMX on AIX 5.3 has been added. 2203 * Support for AIX Version 6.1 has been added. 2204 2205 S/390, zSeries and System z9 2206 2207 * Support for the IBM System z9 EC/BC processor (z9 GA3) has been 2208 added. When using the -march=z9-ec option, the compiler will 2209 generate code making use of instructions provided by the decimal 2210 floating point facility and the floating point conversion facility 2211 (pfpo). Besides the instructions used to implement decimal floating 2212 point operations these facilities also contain instructions to move 2213 between general purpose and floating point registers and to modify 2214 and copy the sign-bit of floating point values. 2215 * When the -march=z9-ec option is used the new 2216 -mhard-dfp/-mno-hard-dfp options can be used to specify whether the 2217 decimal floating point hardware instructions will be used or not. 2218 If none of them is given the hardware support is enabled by 2219 default. 2220 * The -mstack-guard option can now be omitted when using stack 2221 checking via -mstack-size in order to let GCC choose a sensible 2222 stack guard value according to the frame size of each function. 2223 * Various changes to improve performance of generated code have been 2224 implemented, including: 2225 + The condition code set by an add logical with carry 2226 instruction is now available for overflow checks like: a + b + 2227 carry < b. 2228 + The test data class instruction is now used to implement 2229 sign-bit and infinity checks of binary and decimal floating 2230 point numbers. 2231 2232 Xtensa 2233 2234 * Stack unwinding for exception handling now uses by default a 2235 specialized version of DWARF unwinding. This is not 2236 binary-compatible with the setjmp/longjmp (sjlj) unwinding used for 2237 Xtensa with previous versions of GCC. 2238 * For Xtensa processors that include the Conditional Store option, 2239 the built-in functions for atomic memory access are now implemented 2240 using S32C1I instructions. 2241 * If the Xtensa NSA option is available, GCC will use it to implement 2242 the __builtin_ctz and __builtin_clz functions. 2243 2244 Documentation improvements 2245 2246 * Existing libstdc++ documentation has been edited and restructured 2247 into a single DocBook XML manual. The results can be viewed online 2248 [24]here. 2249 2250 Other significant improvements 2251 2252 * The compiler's --help command-line option has been extended so that 2253 it now takes an optional set of arguments. These arguments restrict 2254 the information displayed to specific classes of command-line 2255 options, and possibly only a subset of those options. It is also 2256 now possible to replace the descriptive text associated with each 2257 displayed option with an indication of its current value, or for 2258 binary options, whether it has been enabled or disabled. 2259 Here are some examples. The following will display all the options 2260 controlling warning messages: 2261 --help=warnings 2262 2263 Whereas this will display all the undocumented, target specific 2264 options: 2265 --help=target,undocumented 2266 2267 This sequence of commands will display the binary optimizations 2268 that are enabled by -O3: 2269 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts 2270 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts 2271 diff /tmp/O2-opts /tmp/O3-opts | grep enabled 2272 2273 * The configure options --with-pkgversion and --with-bugurl have been 2274 added. These allow distributors of GCC to include a 2275 distributor-specific string in manuals and --version output and to 2276 specify the URL for reporting bugs in their versions of GCC. 2277 2278 GCC 4.3.1 2279 2280 This is the [25]list of problem reports (PRs) from GCC's bug tracking 2281 system that are known to be fixed in the 4.3.1 release. This list might 2282 not be complete (that is, it is possible that some PRs that have been 2283 fixed are not listed here). 2284 2285 Target Specific Changes 2286 2287 IA-32/x86-64 2288 2289 ABI changes 2290 2291 * Starting with GCC 4.3.1, decimal floating point variables are 2292 aligned to their natural boundaries when they are passed on the 2293 stack for i386. 2294 2295 Command-line changes 2296 2297 * Starting with GCC 4.3.1, the -mcld option has been added to 2298 automatically generate a cld instruction in the prologue of 2299 functions that use string instructions. This option is used for 2300 backward compatibility on some operating systems and can be enabled 2301 by default for 32-bit x86 targets by configuring GCC with the 2302 --enable-cld configure option. 2303 2304 GCC 4.3.2 2305 2306 This is the [26]list of problem reports (PRs) from GCC's bug tracking 2307 system that are known to be fixed in the 4.3.2 release. This list might 2308 not be complete (that is, it is possible that some PRs that have been 2309 fixed are not listed here). 2310 2311 GCC 4.3.3 2312 2313 This is the [27]list of problem reports (PRs) from GCC's bug tracking 2314 system that are known to be fixed in the 4.3.3 release. This list might 2315 not be complete (that is, it is possible that some PRs that have been 2316 fixed are not listed here). 2317 2318 GCC 4.3.4 2319 2320 This is the [28]list of problem reports (PRs) from GCC's bug tracking 2321 system that are known to be fixed in the 4.3.4 release. This list might 2322 not be complete (that is, it is possible that some PRs that have been 2323 fixed are not listed here). 2324 2325 GCC 4.3.5 2326 2327 This is the [29]list of problem reports (PRs) from GCC's bug tracking 2328 system that are known to be fixed in the 4.3.5 release. This list might 2329 not be complete (that is, it is possible that some PRs that have been 2330 fixed are not listed here). 2331 2332 2333 For questions related to the use of GCC, please consult these web 2334 pages and the [30]GCC manuals. If that fails, the 2335 [31]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 2336 web pages and the development of GCC are welcome on our developer 2337 list at [32]gcc (a] gcc.gnu.org. All of [33]our lists have public 2338 archives. 2339 2340 Copyright (C) [34]Free Software Foundation, Inc. Verbatim copying and 2341 distribution of this entire article is permitted in any medium, 2342 provided this notice is preserved. 2343 2344 These pages are [35]maintained by the GCC team. Last modified 2345 2011-04-25[36]. 2346 2347 References 2348 2349 1. http://gcc.gnu.org/gcc-4.3/changes.html#4.3.5 2350 2. http://gmplib.org/ 2351 3. http://www.mpfr.org/ 2352 4. http://gcc.gnu.org/install/prerequisites.html 2353 5. http://gcc.gnu.org/ml/gcc-announce/2001/msg00000.html 2354 6. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options 2355 7. http://gcc.gnu.org/gcc-4.3/porting_to.html 2356 8. http://www.mpfr.org/ 2357 9. http://www.mpfr.org/ 2358 10. http://www.mpfr.org/ 2359 11. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html 2360 12. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html 2361 13. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html 2362 14. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#m anual.intro.status.standard.tr1 2363 15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/parallel_mode.html 2364 16. http://gmplib.org/ 2365 17. http://www.mpfr.org/ 2366 18. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#Code-Gen-Options 2367 19. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfinit-local-zero_007d-167 2368 20. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/GAMMA.html 2369 21. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/LGAMMA.html 2370 22. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html 2371 23. http://gcc.gnu.org/onlinedocs/gfortran/BOZ-literal-constants.html 2372 24. http://gcc.gnu.org/onlinedocs/libstdc++/ 2373 25. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.1 2374 26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.2 2375 27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.3 2376 28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.4 2377 29. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.5 2378 30. http://gcc.gnu.org/onlinedocs/ 2379 31. mailto:gcc-help (a] gcc.gnu.org 2380 32. mailto:gcc (a] gcc.gnu.org 2381 33. http://gcc.gnu.org/lists.html 2382 34. http://www.fsf.org/ 2383 35. http://gcc.gnu.org/about.html 2384 36. http://validator.w3.org/check/referer 2385 ====================================================================== 2386 http://gcc.gnu.org/gcc-4.2/index.html 2387 2388 GCC 4.2 Release Series 2389 2390 May 19, 2008 2391 2392 The [1]GNU project and the GCC developers are pleased to announce the 2393 release of GCC 4.2.4. 2394 2395 This release is a bug-fix release, containing fixes for regressions in 2396 GCC 4.2.3 relative to previous releases of GCC. 2397 2398 Release History 2399 2400 GCC 4.2.4 2401 May 19, 2008 ([2]changes) 2402 2403 GCC 4.2.3 2404 February 1, 2008 ([3]changes) 2405 2406 GCC 4.2.2 2407 October 7, 2007 ([4]changes) 2408 2409 GCC 4.2.1 2410 July 18, 2007 ([5]changes) 2411 2412 GCC 4.2.0 2413 May 13, 2007 ([6]changes) 2414 2415 References and Acknowledgements 2416 2417 GCC used to stand for the GNU C Compiler, but since the compiler 2418 supports several other languages aside from C, it now stands for the 2419 GNU Compiler Collection. 2420 2421 A list of [7]successful builds is updated as new information becomes 2422 available. 2423 2424 The GCC developers would like to thank the numerous people that have 2425 contributed new features, improvements, bug fixes, and other changes as 2426 well as test results to GCC. This [8]amazing group of volunteers is 2427 what makes GCC successful. 2428 2429 For additional information about GCC please refer to the [9]GCC project 2430 web site or contact the [10]GCC development mailing list. 2431 2432 To obtain GCC please use [11]our mirror sites or [12]our SVN server. 2433 2434 2435 For questions related to the use of GCC, please consult these web 2436 pages and the [13]GCC manuals. If that fails, the 2437 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 2438 web pages and the development of GCC are welcome on our developer 2439 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 2440 archives. 2441 2442 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 2443 distribution of this entire article is permitted in any medium, 2444 provided this notice is preserved. 2445 2446 These pages are [18]maintained by the GCC team. Last modified 2447 2011-04-25[19]. 2448 2449 References 2450 2451 1. http://www.gnu.org/ 2452 2. http://gcc.gnu.org/gcc-4.2/changes.html 2453 3. http://gcc.gnu.org/gcc-4.2/changes.html 2454 4. http://gcc.gnu.org/gcc-4.2/changes.html 2455 5. http://gcc.gnu.org/gcc-4.2/changes.html 2456 6. http://gcc.gnu.org/gcc-4.2/changes.html 2457 7. http://gcc.gnu.org/gcc-4.2/buildstat.html 2458 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 2459 9. http://gcc.gnu.org/index.html 2460 10. mailto:gcc (a] gcc.gnu.org 2461 11. http://gcc.gnu.org/mirrors.html 2462 12. http://gcc.gnu.org/svn.html 2463 13. http://gcc.gnu.org/onlinedocs/ 2464 14. mailto:gcc-help (a] gcc.gnu.org 2465 15. mailto:gcc (a] gcc.gnu.org 2466 16. http://gcc.gnu.org/lists.html 2467 17. http://www.fsf.org/ 2468 18. http://gcc.gnu.org/about.html 2469 19. http://validator.w3.org/check/referer 2470 ====================================================================== 2471 http://gcc.gnu.org/gcc-4.2/changes.html 2472 2473 GCC 4.2 Release Series 2474 Changes, New Features, and Fixes 2475 2476 Caveats 2477 2478 * GCC no longer accepts the -fshared-data option. This option has had 2479 no effect in any GCC 4 release; the targets to which the option 2480 used to apply had been removed before GCC 4.0. 2481 2482 General Optimizer Improvements 2483 2484 * New command-line options specify the possible relationships among 2485 parameters and between parameters and global data. For example, 2486 -fargument-noalias-anything specifies that arguments do not alias 2487 any other storage. 2488 Each language will automatically use whatever option is required by 2489 the language standard. You should not need to use these options 2490 yourself. 2491 2492 New Languages and Language specific improvements 2493 2494 * [1]OpenMP is now supported for the C, C++ and Fortran compilers. 2495 * New command-line options -fstrict-overflow and -Wstrict-overflow 2496 have been added. -fstrict-overflow tells the compiler that it may 2497 assume that the program follows the strict signed overflow 2498 semantics permitted for the language: for C and C++ this means that 2499 the compiler may assume that signed overflow does not occur. For 2500 example, a loop like 2501 for (i = 1; i > 0; i *= 2) 2502 2503 is presumably intended to continue looping until i overflows. With 2504 -fstrict-overflow, the compiler may assume that signed overflow 2505 will not occur, and transform this into an infinite loop. 2506 -fstrict-overflow is turned on by default at -O2, and may be 2507 disabled via -fno-strict-overflow. The -Wstrict-overflow option may 2508 be used to warn about cases where the compiler assumes that signed 2509 overflow will not occur. It takes five different levels: 2510 -Wstrict-overflow=1 to 5. See the [2]documentation for details. 2511 -Wstrict-overflow=1 is enabled by -Wall. 2512 * The new command-line option -fno-toplevel-reorder directs GCC to 2513 emit top-level functions, variables, and asm statements in the same 2514 order that they appear in the input file. This is intended to 2515 support existing code which relies on a particular ordering (for 2516 example, code which uses top-level asm statements to switch 2517 sections). For new code, it is generally better to use function and 2518 variable attributes. The -fno-toplevel-reorder option may be used 2519 for most cases which currently use -fno-unit-at-a-time. The 2520 -fno-unit-at-a-time option will be removed in some future version 2521 of GCC. If you know of a case which requires -fno-unit-at-a-time 2522 which is not fixed by -fno-toplevel-reorder, please open a bug 2523 report. 2524 2525 C family 2526 2527 * The pragma redefine_extname will now macro expand its tokens for 2528 compatibility with SunPRO. 2529 * In the next release of GCC, 4.3, -std=c99 or -std=gnu99 will direct 2530 GCC to handle inline functions as specified in the C99 standard. In 2531 preparation for this, GCC 4.2 will warn about any use of non-static 2532 inline functions in gnu99 or c99 mode. This new warning may be 2533 disabled with the new gnu_inline function attribute or the new 2534 -fgnu89-inline command-line option. Also, GCC 4.2 and later will 2535 define one of the preprocessor macros __GNUC_GNU_INLINE__ or 2536 __GNUC_STDC_INLINE__ to indicate the semantics of inline functions 2537 in the current compilation. 2538 * A new command-line option -Waddress has been added to warn about 2539 suspicious uses of memory addresses as, for example, using the 2540 address of a function in a conditional expression, and comparisons 2541 against the memory address of a string literal. This warning is 2542 enabled by -Wall. 2543 2544 C++ 2545 2546 * C++ visibility handling has been overhauled. 2547 Restricted visiblity is propagated from classes to members, from 2548 functions to local statics, and from templates and template 2549 arguments to instantiations, unless the latter has explicitly 2550 declared visibility. 2551 The visibility attribute for a class must come between the 2552 class-key and the name, not after the closing brace. 2553 Attributes are now allowed for enums and elaborated-type-specifiers 2554 that only declare a type. 2555 Members of the anonymous namespace are now local to a particular 2556 translation unit, along with any other declarations which use them, 2557 though they are still treated as having external linkage for 2558 language semantics. 2559 * The (undocumented) extension which permitted templates with default 2560 arguments to be bound to template template parameters with fewer 2561 parameters has been removed. For example: 2562 template <template <typename> class C> 2563 void f(C<double>) {} 2564 2565 template <typename T, typename U = int> 2566 struct S {}; 2567 2568 template void f(S<double>); 2569 2570 is no longer accepted by G++. The reason this code is not accepted 2571 is that S is a template with two parameters; therefore, it cannot 2572 be bound to C which has only one parameter. 2573 * The <?, >?, <?=, and >?= operators, deprecated in previous GCC 2574 releases, have been removed. 2575 * The command-line option -fconst-strings, deprecated in previous GCC 2576 releases, has been removed. 2577 * The configure variable enable-__cxa_atexit is now enabled by 2578 default for more targets. Enabling this variable is necessary in 2579 order for static destructors to be executed in the correct order, 2580 but it depends upon the presence of a non-standard C library in the 2581 target library in order to work. The variable is now enabled for 2582 more targets which are known to have suitable C libraries. 2583 * -Wextra will produce warnings for if statements with a semicolon as 2584 the only body, to catch code like: 2585 if (a); 2586 return 1; 2587 return 0; 2588 2589 To suppress the warning in valid cases, use { } instead. 2590 * The C++ frontend now also produces strict aliasing warnings when 2591 -fstrict-aliasing -Wstrict-aliasing is in effect. 2592 2593 Runtime Library (libstdc++) 2594 2595 * Added support for TR1 <random>, <complex>, and C compatibility 2596 headers. In addition, a lock-free version of shared_ptr was 2597 contributed as part of Phillip Jordan's Google Summer of Code 2598 project on lock-free containers. ([3]Implementation status of TR1) 2599 * In association with the Summer of Code work on lock-free 2600 containers, the interface for atomic builtins was adjusted, 2601 creating simpler alternatives for non-threaded code paths. Also, 2602 usage was consolidated and all elements were moved from namespace 2603 std to namespace__gnu_cxx. Affected interfaces are the functions 2604 __exchange_and_add, __atomic_add, and the objects __mutex, 2605 __recursive_mutex, and __scoped_lock. 2606 * Support for versioning weak symbol names via namespace association 2607 was added. However, as this changes the names of exported symbols, 2608 this is turned off by default in the current ABI. Intrepid users 2609 can enable this feature by using 2610 --enable-symvers=gnu-versioned-namespace during configuration. 2611 * Revised, simplified, and expanded policy-based associative 2612 containers, including data types for tree and trie forms 2613 (basic_tree, tree, trie), lists (list_update), and both 2614 collision-chaining and probing hash-based containers 2615 (basic_hash_table, cc_hash_table, gp_hash_table). More details per 2616 the [4]documentation. 2617 * The implementation of the debug mode was modified, whereby the 2618 debug namespaces were nested inside of namespace std and namespace 2619 __gnu_cxx in order to resolve some long standing corner cases 2620 involving name lookup. Debug functionality from the policy-based 2621 data structures was consolidated and enabled with the single macro, 2622 _GLIBCXX_DEBUG. See PR 26142 for more information. 2623 * Added extensions for type traits: __conditional_type, 2624 __numeric_traits, __add_unsigned, __removed_unsigned, __enable_if. 2625 * Added a typelist implementation for compile-time meta-programming. 2626 Elements for typelist construction and operation can be found 2627 within namespace __gnu_cxx::typelist. 2628 * Added a new allocator, __gnu_cxx::throw_allocator, for testing 2629 exception-safety. 2630 * Enabled library-wide visibility control, allowing -fvisibility to 2631 be used. 2632 * Consolidated all nested namespaces and the conversion of 2633 __gnu_internal implementation-private details to anonymous 2634 namespaces whenever possible. 2635 * Implemented LWG resolutions DR 431 and DR 538. 2636 2637 Fortran 2638 2639 * Support for allocatable components has been added (TR 15581 and 2640 Fortran 2003). 2641 * Support for the Fortran 2003 streaming IO extension has been added. 2642 * The GNU Fortran compiler now uses 4-byte record markers by default 2643 for unformatted files to be compatible with g77 and most other 2644 compilers. The implementation allows for records greater than 2 GB 2645 and is compatible with several other compilers. Older versions of 2646 gfortran used 8-byte record markers by default (on most systems). 2647 In order to change the length of the record markers, e.g. to read 2648 unformatted files created by older gfortran versions, the 2649 [5]-frecord-marker=8 option can be used. 2650 2651 Java (GCJ) 2652 2653 * A new command-line option -static-libgcj has been added for targets 2654 that use a linker compatible with GNU Binutils. As its name 2655 implies, this causes libgcj to be linked statically. In some cases 2656 this causes the resulting executable to start faster and use less 2657 memory than if the shared version of libgcj were used. However 2658 caution should be used as it can also cause essential parts of the 2659 library to be omitted. Some of these issues are discussed in: 2660 [6]http://gcc.gnu.org/wiki/Statically_linking_libgcj 2661 * fastjar is no longer bundled with GCC. To build libgcj, you will 2662 need either InfoZIP (both zip and unzip) or an external jar 2663 program. In the former case, the GCC build will install a jar shell 2664 script that is based on InfoZIP and provides the same functionality 2665 as fastjar. 2666 2667 New Targets and Target Specific Improvements 2668 2669 IA-32/x86-64 2670 2671 * -mtune=generic can now be used to generate code running well on 2672 common x86 chips. This includes AMD Athlon, AMD Opteron, Intel 2673 Pentium-M, Intel Pentium 4 and Intel Core 2. 2674 * -mtune=native and -march=native will produce code optimized for the 2675 host architecture as detected using the cpuid instruction. 2676 * Added a new command-line option -fstackrealign and and 2677 __attribute__ ((force_align_arg_pointer)) to realign the stack at 2678 runtime. This allows functions compiled with a vector-aligned stack 2679 to be invoked from legacy objects that keep only word-alignment. 2680 2681 SPARC 2682 2683 * The default CPU setting has been changed from V7 to V9 in 32-bit 2684 mode on Solaris 7 and above. This is already the case in 64-bit 2685 mode. It can be overridden by specifying --with-cpu at configure 2686 time. 2687 * Back-end support of built-in functions for atomic memory access has 2688 been implemented. 2689 * Support for the Sun UltraSPARC T1 (Niagara) processor has been 2690 added. 2691 2692 M32C 2693 2694 * Various bug fixes have made some functions (notably, functions 2695 returning structures) incompatible with previous releases. 2696 Recompiling all libraries is recommended. Note that code quality 2697 has considerably improved since 4.1, making a recompile even more 2698 beneficial. 2699 2700 MIPS 2701 2702 * Added support for the Broadcom SB-1A core. 2703 2704 IA-64 2705 2706 * Added support for IA-64 data and control speculation. By default 2707 speculation is enabled only during second scheduler pass. A number 2708 of machine flags was introduced to control the usage of speculation 2709 for both scheduler passes. 2710 2711 HPPA 2712 2713 * Added Java language support (libffi and libjava) for 32-bit HP-UX 2714 11 target. 2715 2716 Obsolete Systems 2717 2718 Documentation improvements 2719 2720 PDF Documentation 2721 2722 * A make pdf target has been added to the top-level makefile, 2723 enabling automated production of PDF documentation files. 2724 (Front-ends external to GCC should modify their Make-lang.in file 2725 to add a lang.pdf: target.) 2726 2727 Other significant improvements 2728 2729 Build system improvements 2730 2731 * All the components of the compiler are now bootstrapped by default. 2732 This improves the resilience to bugs in the system compiler or 2733 binary compatibility problems, as well as providing better testing 2734 of GCC 4.2 itself. In addition, if you build the compiler from a 2735 combined tree, the assembler, linker, etc. will also be 2736 bootstrapped (i.e. built with themselves). 2737 You can disable this behavior, and go back to the pre-GCC 4.2 set 2738 up, by configuring GCC with --disable-bootstrap. 2739 * The rules that configure follows to find target tools resemble more 2740 closely the locations that the built compiler will search. In 2741 addition, you can use the new configure option --with-target-tools 2742 to specify where to find the target tools used during the build, 2743 without affecting what the built compiler will use. 2744 This can be especially useful when building packages of GCC. For 2745 example, you may want to build GCC with GNU as or ld, even if the 2746 resulting compiler to work with the native assembler and linker. To 2747 do so, you can use --with-target-tools to point to the native 2748 tools. 2749 2750 Incompatible changes to the build system 2751 2752 * Front-ends external to GCC should modify their Make-lang.in file to 2753 replace double-colon rules (e.g. dvi::) with normal rules (like 2754 lang.dvi:). Front-end makefile hooks do not use double-colon rules 2755 anymore. 2756 * Up to GCC 4.1, a popular way to specify the target tools used 2757 during the build was to create directories named gas, binutils, 2758 etc. in the build tree, and create links to the tools from there. 2759 This does not work any more when the compiler is bootstrapped. The 2760 new configure option --with-target-tools provides a better way to 2761 achieve the same effect, and works for all native and cross 2762 settings. 2763 2764 2765 For questions related to the use of GCC, please consult these web 2766 pages and the [7]GCC manuals. If that fails, the 2767 [8]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 2768 web pages and the development of GCC are welcome on our developer 2769 list at [9]gcc (a] gcc.gnu.org. All of [10]our lists have public 2770 archives. 2771 2772 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and 2773 distribution of this entire article is permitted in any medium, 2774 provided this notice is preserved. 2775 2776 These pages are [12]maintained by the GCC team. Last modified 2777 2011-04-25[13]. 2778 2779 References 2780 2781 1. http://gcc.gnu.org/projects/gomp/ 2782 2. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html 2783 3. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.tr1 2784 4. http://gcc.gnu.org/onlinedocs/libstdc++/ext/pb_ds/index.html 2785 5. http://gcc.gnu.org/onlinedocs/gfortran/Runtime-Options.html 2786 6. http://gcc.gnu.org/wiki/Statically_linking_libgcj 2787 7. http://gcc.gnu.org/onlinedocs/ 2788 8. mailto:gcc-help (a] gcc.gnu.org 2789 9. mailto:gcc (a] gcc.gnu.org 2790 10. http://gcc.gnu.org/lists.html 2791 11. http://www.fsf.org/ 2792 12. http://gcc.gnu.org/about.html 2793 13. http://validator.w3.org/check/referer 2794 ====================================================================== 2795 http://gcc.gnu.org/gcc-4.1/index.html 2796 2797 GCC 4.1 Release Series 2798 2799 February 13, 2007 2800 2801 The [1]GNU project and the GCC developers are pleased to announce the 2802 release of GCC 4.1.2. 2803 2804 This release is a bug-fix release, containing fixes for regressions in 2805 GCC 4.1.1 relative to previous releases of GCC. 2806 2807 Release History 2808 2809 GCC 4.1.2 2810 February 13, 2007 ([2]changes) 2811 2812 GCC 4.1.1 2813 May 24, 2006 ([3]changes) 2814 2815 GCC 4.1.0 2816 February 28, 2006 ([4]changes) 2817 2818 References and Acknowledgements 2819 2820 GCC used to stand for the GNU C Compiler, but since the compiler 2821 supports several other languages aside from C, it now stands for the 2822 GNU Compiler Collection. 2823 2824 A list of [5]successful builds is updated as new information becomes 2825 available. 2826 2827 The GCC developers would like to thank the numerous people that have 2828 contributed new features, improvements, bug fixes, and other changes as 2829 well as test results to GCC. This [6]amazing group of volunteers is 2830 what makes GCC successful. 2831 2832 For additional information about GCC please refer to the [7]GCC project 2833 web site or contact the [8]GCC development mailing list. 2834 2835 To obtain GCC please use [9]our mirror sites or [10]our SVN server. 2836 2837 2838 For questions related to the use of GCC, please consult these web 2839 pages and the [11]GCC manuals. If that fails, the 2840 [12]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 2841 web pages and the development of GCC are welcome on our developer 2842 list at [13]gcc (a] gcc.gnu.org. All of [14]our lists have public 2843 archives. 2844 2845 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and 2846 distribution of this entire article is permitted in any medium, 2847 provided this notice is preserved. 2848 2849 These pages are [16]maintained by the GCC team. Last modified 2850 2011-04-25[17]. 2851 2852 References 2853 2854 1. http://www.gnu.org/ 2855 2. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2 2856 3. http://gcc.gnu.org/gcc-4.1/changes.html 2857 4. http://gcc.gnu.org/gcc-4.1/changes.html 2858 5. http://gcc.gnu.org/gcc-4.1/buildstat.html 2859 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 2860 7. http://gcc.gnu.org/index.html 2861 8. mailto:gcc (a] gcc.gnu.org 2862 9. http://gcc.gnu.org/mirrors.html 2863 10. http://gcc.gnu.org/svn.html 2864 11. http://gcc.gnu.org/onlinedocs/ 2865 12. mailto:gcc-help (a] gcc.gnu.org 2866 13. mailto:gcc (a] gcc.gnu.org 2867 14. http://gcc.gnu.org/lists.html 2868 15. http://www.fsf.org/ 2869 16. http://gcc.gnu.org/about.html 2870 17. http://validator.w3.org/check/referer 2871 ====================================================================== 2872 http://gcc.gnu.org/gcc-4.1/changes.html 2873 2874 GCC 4.1 Release Series 2875 Changes, New Features, and Fixes 2876 2877 The latest release in the 4.1 release series is [1]GCC 4.1.2. 2878 2879 Caveats 2880 2881 General Optimizer Improvements 2882 2883 * GCC now has infrastructure for inter-procedural optimizations and 2884 the following inter-procedural optimizations are implemented: 2885 + Profile guided inlining. When doing profile feedback guided 2886 optimization, GCC can now use the profile to make better 2887 informed decisions on whether inlining of a function is 2888 profitable or not. This means that GCC will no longer inline 2889 functions at call sites that are not executed very often, and 2890 that functions at hot call sites are more likely to be 2891 inlined. 2892 A new parameter min-inline-recursive-probability is also now 2893 available to throttle recursive inlining of functions with 2894 small average recursive depths. 2895 + Discovery of pure and const functions, a form of side-effects 2896 analysis. While older GCC releases could also discover such 2897 special functions, the new IPA-based pass runs earlier so that 2898 the results are available to more optimizers. The pass is also 2899 simply more powerful than the old one. 2900 + Analysis of references to static variables and type escape 2901 analysis, also forms of side-effects analysis. The results of 2902 these passes allow the compiler to be less conservative about 2903 call-clobbered variables and references. This results in more 2904 redundant loads being eliminated and in making static 2905 variables candidates for register promotion. 2906 + Improvement of RTL-based alias analysis. The results of type 2907 escape analysis are fed to the RTL type-based alias analyzer, 2908 allowing it to disambiguate more memory references. 2909 + Interprocedural constant propagation and function versioning. 2910 This pass looks for functions that are always called with the 2911 same constant value for one or more of the function arguments, 2912 and propagates those constants into those functions. 2913 + GCC will now eliminate static variables whose usage was 2914 optimized out. 2915 + -fwhole-program --combine can now be used to make all 2916 functions in program static allowing whole program 2917 optimization. As an exception, the main function and all 2918 functions marked with the new externally_visible attribute are 2919 kept global so that programs can link with runtime libraries. 2920 * GCC can now do a form of partial dead code elimination (PDCE) that 2921 allows code motion of expressions to the paths where the result of 2922 the expression is actually needed. This is not always a win, so the 2923 pass has been limited to only consider profitable cases. Here is an 2924 example: 2925 int foo (int *, int *); 2926 int 2927 bar (int d) 2928 { 2929 int a, b, c; 2930 b = d + 1; 2931 c = d + 2; 2932 a = b + c; 2933 if (d) 2934 { 2935 foo (&b, &c); 2936 a = b + c; 2937 } 2938 printf ("%d\n", a); 2939 } 2940 2941 The a = b + c can be sunk to right before the printf. Normal code 2942 sinking will not do this, it will sink the first one above into the 2943 else-branch of the conditional jump, which still gives you two 2944 copies of the code. 2945 * GCC now has a value range propagation pass. This allows the 2946 compiler to eliminate bounds checks and branches. The results of 2947 the pass can also be used to accurately compute branch 2948 probabilities. 2949 * The pass to convert PHI nodes to straight-line code (a form of 2950 if-conversion for GIMPLE) has been improved significantly. The two 2951 most significant improvements are an improved algorithm to 2952 determine the order in which the PHI nodes are considered, and an 2953 improvement that allow the pass to consider if-conversions of basic 2954 blocks with more than two predecessors. 2955 * Alias analysis improvements. GCC can now differentiate between 2956 different fields of structures in Tree-SSA's virtual operands form. 2957 This lets stores/loads from non-overlapping structure fields not 2958 conflict. A new algorithm to compute points-to sets was contributed 2959 that can allows GCC to see now that p->a and p->b, where p is a 2960 pointer to a structure, can never point to the same field. 2961 * Various enhancements to auto-vectorization: 2962 + Incrementally preserve SSA form when vectorizing. 2963 + Incrementally preserve loop-closed form when vectorizing. 2964 + Improvements to peeling for alignment: generate better code 2965 when the misalignment of an access is known at compile time, 2966 or when different accesses are known to have the same 2967 misalignment, even if the misalignment amount itself is 2968 unknown. 2969 + Consider dependence distance in the vectorizer. 2970 + Externalize generic parts of data reference analysis to make 2971 this analysis available to other passes. 2972 + Vectorization of conditional code. 2973 + Reduction support. 2974 * GCC can now partition functions in sections of hot and cold code. 2975 This can significantly improve performance due to better 2976 instruction cache locality. This feature works best together with 2977 profile feedback driven optimization. 2978 * A new pass to avoid saving of unneeded arguments to the stack in 2979 vararg functions if the compiler can prove that they will not be 2980 needed. 2981 * Transition of basic block profiling to tree level implementation 2982 has been completed. The new implementation should be considerably 2983 more reliable (hopefully avoiding profile mismatch errors when 2984 using -fprofile-use or -fbranch-probabilities) and can be used to 2985 drive higher level optimizations, such as inlining. 2986 The -ftree-based-profiling command-line option was removed and 2987 -fprofile-use now implies disabling old RTL level loop optimizer 2988 (-fno-loop-optimize). Speculative prefetching optimization 2989 (originally enabled by -fspeculative-prefetching) was removed. 2990 2991 New Languages and Language specific improvements 2992 2993 C and Objective-C 2994 2995 * The old Bison-based C and Objective-C parser has been replaced by a 2996 new, faster hand-written recursive-descent parser. 2997 2998 Ada 2999 3000 * The build infrastructure for the Ada runtime library and tools has 3001 been changed to be better integrated with the rest of the build 3002 infrastructure of GCC. This should make doing cross builds of Ada a 3003 bit easier. 3004 3005 C++ 3006 3007 * ARM-style name-injection of friend declarations is no longer the 3008 default. For example: 3009 struct S { 3010 friend void f(); 3011 }; 3012 3013 void g() { f(); } 3014 will not be accepted; instead a declaration of f will need to be 3015 present outside of the scope of S. The new -ffriend-injection 3016 option will enable the old behavior. 3017 * The (undocumented) extension which permitted templates with default 3018 arguments to be bound to template template parameters with fewer 3019 parameters has been deprecated, and will be removed in the next 3020 major release of G++. For example: 3021 template <template <typename> class C> 3022 void f(C<double>) {} 3023 3024 template <typename T, typename U = int> 3025 struct S {}; 3026 3027 template void f(S<double>); 3028 3029 makes use of the deprecated extension. The reason this code is not 3030 valid ISO C++ is that S is a template with two parameters; 3031 therefore, it cannot be bound to C which has only one parameter. 3032 3033 Runtime Library (libstdc++) 3034 3035 * Optimization work: 3036 + A new implementation of std::search_n is provided, better 3037 performing in case of random access iterators. 3038 + Added further efficient specializations of istream functions, 3039 i.e., character array and string extractors. 3040 + Other smaller improvements throughout. 3041 * Policy-based associative containers, designed for high-performance, 3042 flexibility and semantic safety are delivered in ext/pb_assoc. 3043 * A versatile string class, __gnu_cxx::__versa_string, providing 3044 facilities conforming to the standard requirements for 3045 basic_string, is delivered in <ext/vstring.h>. In particular: 3046 + Two base classes are provided: the default one avoids 3047 reference counting and is optimized for short strings; the 3048 alternate one, still uses it while improving in a few low 3049 level areas (e.g., alignment). See vstring_fwd.h for some 3050 useful typedefs. 3051 + Various algorithms have been rewritten (e.g., replace), the 3052 code streamlined and simple optimizations added. 3053 + Option 3 of DR 431 is implemented for both available bases, 3054 thus improving the support for stateful allocators. 3055 * As usual, many bugs have been fixed (e.g., libstdc++/13583, 3056 libstdc++/23953) and LWG resolutions put into effect for the first 3057 time (e.g., DR 280, DR 464, N1780 recommendations for DR 233, TR1 3058 Issue 6.19). The implementation status of TR1 is now tracked in the 3059 docs in tr1.html. 3060 3061 Objective-C++ 3062 3063 * A new language front end for Objective-C++ has been added. This 3064 language allows users to mix the object oriented features of 3065 Objective-C with those of C++. 3066 3067 Java (GCJ) 3068 3069 * Core library (libgcj) updates based on GNU Classpath 0.15 - 0.19 3070 features (plus some 0.20 bug-fixes) 3071 + Networking 3072 o The java.net.HttpURLConnection implementation no longer 3073 buffers the entire response body in memory. This means 3074 that response bodies larger than available memory can now 3075 be handled. 3076 + (N)IO 3077 o NIO FileChannel.map implementation, fast bulk put 3078 implementation for DirectByteBuffer (speeds up this 3079 method 10x). 3080 o FileChannel.lock() and FileChannel.force() implemented. 3081 + XML 3082 o gnu.xml fix for nodes created outside a namespace 3083 context. 3084 o Add support for output indenting and 3085 cdata-section-elements output instruction in 3086 xml.transform. 3087 o xml.xpath corrections for cases where elements/attributes 3088 might have been created in non-namespace-aware mode. 3089 Corrections to handling of XSL variables and minor 3090 conformance updates. 3091 + AWT 3092 o GNU JAWT implementation, the AWT Native Interface, which 3093 allows direct access to native screen resources from 3094 within a Canvas's paint method. GNU Classpath Examples 3095 comes with a Demo, see libjava/classpath/examples/README. 3096 o awt.datatransfer updated to 1.5 with support for 3097 FlavorEvents. The gtk+ awt peers now allow copy/paste of 3098 text, images, URIs/files and serialized objects with 3099 other applications and tracking clipboard change events 3100 with gtk+ 2.6 (for gtk+ 2.4 only text and serialized 3101 objects are supported). A GNU Classpath Examples 3102 datatransfer Demo was added to show the new 3103 functionality. 3104 o Split gtk+ awt peers event handling in two threads and 3105 improve gdk lock handling (solves several awt lock ups). 3106 o Speed up awt Image loading. 3107 o Better gtk+ scrollbar peer implementation when using gtk+ 3108 >= 2.6. 3109 o Handle image loading errors correctly for gdkpixbuf and 3110 MediaTracker. 3111 o Better handle GDK lock. Properly prefix gtkpeer native 3112 functions (cp_gtk). 3113 o GdkGraphics2D has been updated to use Cairo 0.5.x or 3114 higher. 3115 o BufferedImage and GtkImage rewrites. All image drawing 3116 operations should now work correctly (flipping requires 3117 gtk+ >= 2.6) 3118 o Future Graphics2D, image and text work is documented at: 3119 [2]http://developer.classpath.org/mediation/ClasspathGrap 3120 hicsImagesText 3121 o When gtk+ 2.6 or higher is installed the default log 3122 handler will produce stack traces whenever a WARNING, 3123 CRITICAL or ERROR message is produced. 3124 + Free Swing 3125 o The RepaintManager has been reworked for more efficient 3126 painting, especially for large GUIs. 3127 o The layout manager OverlayLayout has been implemented, 3128 the BoxLayout has been rewritten to make use of the 3129 SizeRequirements utility class and caching for more 3130 efficient layout. 3131 o Improved accessibility support. 3132 o Significant progress has been made in the implementation 3133 of the javax.swing.plaf.metal package, with most UI 3134 delegates in a working state now. Please test this with 3135 your own applications and provide feedback that will help 3136 us to improve this package. 3137 o The GUI demo (gnu.classpath.examples.swing.Demo) has been 3138 extended to highlight various features in our Free Swing 3139 implementation. And it includes a look and feel switcher 3140 for Metal (default), Ocean and GNU themes. 3141 o The javax.swing.plaf.multi package is now implemented. 3142 o Editing and several key actions for JTree and JTable were 3143 implemented. 3144 o Lots of icons and look and feel improvements for Free 3145 Swing basic and metal themes were added. Try running the 3146 GNU Classpath Swing Demo in examples 3147 (gnu.classpath.examples.swing.Demo) with: 3148 -Dswing.defaultlaf=javax.swing.plaf.basic.BasicLookAndFee 3149 l or 3150 -Dswing.defaultlaf=javax.swing.plaf.metal.MetalLookAndFee 3151 l 3152 o Start of styled text capabilites for java.swing.text. 3153 o DefaultMutableTreeNode pre-order, post-order, depth-first 3154 and breadth-first traversal enumerations implemented. 3155 o JInternalFrame colors and titlebar draw properly. 3156 o JTree is working up to par (icons, selection and keyboard 3157 traversal). 3158 o JMenus were made more compatible in visual and 3159 programmatic behavior. 3160 o JTable changeSelection and multiple selections 3161 implemented. 3162 o JButton and JToggleButton change states work properly 3163 now. 3164 o JFileChooser fixes. 3165 o revalidate() and repaint() fixes which make Free Swing 3166 much more responsive. 3167 o MetalIconFactory implemented. 3168 o Free Swing Top-Level Compatibility. JFrame, JDialog, 3169 JApplet, JInternalFrame, and JWindow are now 1.5 3170 compatible in the sense that you can call add() and 3171 setLayout() directly on them, which will have the same 3172 effect as calling getContentPane().add() and 3173 getContentPane().setLayout(). 3174 o The JTree interface has been completed. JTrees now 3175 recognizes mouse clicks and selections work. 3176 o BoxLayout works properly now. 3177 o Fixed GrayFilter to actually work. 3178 o Metal SplitPane implemented. 3179 o Lots of Free Swing text and editor stuff work now. 3180 + Free RMI and Corba 3181 o Andrew Watson, Vice President and Technical Director of 3182 the Object Management Group, has officially assigned us 3183 20 bit Vendor Minor Code Id: 0x47430 ("GC") that will 3184 mark remote classpath-specific system exceptions. 3185 Obtaining the VMCID means that GNU Classpath now is a 3186 recogniseable type of node in a highly interoperable 3187 CORBA world. 3188 o GNU Classpath now includes the first working draft to 3189 support the RMI over IIOP protocol. The current 3190 implementation is capable of remote invocations, 3191 transferring various Serializables and Externalizables 3192 via RMI-IIOP protocol. It can flatten graphs and, at 3193 least for the simple cases, is interoperable with 1.5 3194 JDKs. 3195 o org.omg.PortableInterceptor and related functionality in 3196 other packages is now implemented: 3197 # The sever and client interceptors work as required 3198 since 1.4. 3199 # The IOR interceptor works as needed for 1.5. 3200 o The org.omg.DynamicAny package is completed and passes 3201 the prepared tests. 3202 o The Portable Object Adapter should now support the output 3203 of the recent IDL to java compilers. These compilers now 3204 generate servants and not CORBA objects as before, making 3205 the output depend on the existing POA implementation. 3206 Completing POA means that such code can already be tried 3207 to run on Classpath. Our POA is tested for the following 3208 usager scenarios: 3209 # POA converts servant to the CORBA object. 3210 # Servant provides to the CORBA object. 3211 # POA activates new CORBA object with the given Object 3212 Id (byte array) that is later accessible for the 3213 servant. 3214 # During the first call, the ServantActivator provides 3215 servant for this and all subsequent calls on the 3216 current object. 3217 # During each call, the ServantLocator provides 3218 servant for this call only. 3219 # ServantLocator or ServantActivator forwards call to 3220 another server. 3221 # POA has a single servant, responsible for all 3222 objects. 3223 # POA has a default servant, but some objects are 3224 explicitly connected to they specific servants. 3225 The POA is verified using tests from the former 3226 cost.omg.org. 3227 o The CORBA implementation is now a working prototype that 3228 should support features up to 1.3 inclusive. We invite 3229 groups writing CORBA dependent applications to try 3230 Classpath implementation, reporting any possible bugs. 3231 The CORBA prototype is interoperable with Sun's 3232 implementation v 1.4, transferring object references, 3233 primitive types, narrow and wide strings, arrays, 3234 structures, trees, abstract interfaces and value types 3235 (feature of CORBA 2.3) between these two platforms. 3236 Remote exceptions are transferred and handled correctly. 3237 The stringified object references (IORs) from various 3238 sources are parsed as required. The transient (for 3239 current session) and permanent (till jre restart) 3240 redirections work. Both Little and Big Endian encoded 3241 messages are accepted. The implementation is verified 3242 using tests from the former cost.omg.org. The current 3243 release includes working examples (see the examples 3244 directory), demonstrating the client-server 3245 communication, using either CORBA Request or IDL-based 3246 stub (usually generated by a IDL to java compiler). These 3247 examples also show how to use the Classpath CORBA naming 3248 service. The IDL to java compiler is not yet written, but 3249 as our library must be compatible, it naturally accepts 3250 the output of other idlj implementations. 3251 + Misc 3252 o Updated TimeZone data against Olson tzdata2005l. 3253 o Make zip and jar packages UTF-8 clean. 3254 o "native" code builds and compiles (warning free) on 3255 Darwin and Solaris. 3256 o java.util.logging.FileHandler now rotates files. 3257 o Start of a generic JDWP framework in gnu/classpath/jdwp. 3258 This is unfinished, but feedback (at classpath (a] gnu.org) 3259 from runtime hackers is greatly appreciated. Although 3260 most of the work is currently being done around gcj/gij 3261 we want this framework to be as VM neutral as possible. 3262 Early design is described in: 3263 [3]http://gcc.gnu.org/ml/java/2005-05/msg00260.html 3264 o QT4 AWT peers, enable by giving configure 3265 --enable-qt-peer. Included, but not ready for production 3266 yet. They are explicitly disabled and not supported. But 3267 if you want to help with the development of these new 3268 features we are interested in feedback. You will have to 3269 explicitly enable them to try them out (and they will 3270 most likely contain bugs). 3271 o Documentation fixes all over the place. See 3272 [4]http://developer.classpath.org/doc/ 3273 3274 New Targets and Target Specific Improvements 3275 3276 IA-32/x86-64 3277 3278 * The x86-64 medium model (that allows building applications whose 3279 data segment exceeds 4GB) was redesigned to match latest ABI draft. 3280 New implementation split large datastructures into separate segment 3281 improving performance of accesses to small datastructures and also 3282 allows linking of small model libraries into medium model programs 3283 as long as the libraries are not accessing the large datastructures 3284 directly. Medium model is also supported in position independent 3285 code now. 3286 The ABI change results in partial incompatibility among medium 3287 model objects. Linking medium model libraries (or objects) compiled 3288 with new compiler into medium model program compiled with older 3289 will likely result in exceeding ranges of relocations. 3290 Binutils 2.16.91 or newer are required for compiling medium model 3291 now. 3292 3293 RS6000 (POWER/PowerPC) 3294 3295 * The AltiVec vector primitives in <altivec.h> are now implemented in 3296 a way that puts a smaller burden on the preprocessor, instead 3297 processing the "overloading" in the front ends. This should benefit 3298 compilation speed on AltiVec vector code. 3299 * AltiVec initializers now are generated more efficiently. 3300 * The popcountb instruction available on POWER5 now is generated. 3301 * The floating point round to integer instructions available on 3302 POWER5+ now is generated. 3303 * Floating point divides can be synthesized using the floating point 3304 reciprocal estimate instructions. 3305 * Double precision floating point constants are initialized as single 3306 precision values if they can be represented exactly. 3307 3308 S/390, zSeries and System z9 3309 3310 * Support for the IBM System z9 109 processor has been added. When 3311 using the -march=z9-109 option, the compiler will generate code 3312 making use of instructions provided by the extended immediate 3313 facility. 3314 * Support for 128-bit IEEE floating point has been added. When using 3315 the -mlong-double-128 option, the compiler will map the long double 3316 data type to 128-bit IEEE floating point. Using this option 3317 constitutes an ABI change, and requires glibc support. 3318 * Various changes to improve performance of generated code have been 3319 implemented, including: 3320 + In functions that do not require a literal pool, register %r13 3321 (which is traditionally reserved as literal pool pointer), can 3322 now be freely used for other purposes by the compiler. 3323 + More precise tracking of register use allows the compiler to 3324 generate more efficient function prolog and epilog code in 3325 certain cases. 3326 + The SEARCH STRING, COMPARE LOGICAL STRING, and MOVE STRING 3327 instructions are now used to implement C string functions. 3328 + The MOVE CHARACTER instruction with single byte overlap is now 3329 used to implement the memset function with non-zero fill byte. 3330 + The LOAD ZERO instructions are now used where appropriate. 3331 + The INSERT CHARACTERS UNDER MASK, STORE CHARACTERS UNDER MASK, 3332 and INSERT IMMEDIATE instructions are now used more frequently 3333 to optimize bitfield operations. 3334 + The BRANCH ON COUNT instruction is now used more frequently. 3335 In particular, the fact that a loop contains a subroutine call 3336 no longer prevents the compiler from using this instruction. 3337 + The compiler is now aware that all shift and rotate 3338 instructions implicitly truncate the shift count to six bits. 3339 * Back-end support for the following generic features has been 3340 implemented: 3341 + The full set of [5]built-in functions for atomic memory 3342 access. 3343 + The -fstack-protector feature. 3344 + The optimization pass avoiding unnecessary stores of incoming 3345 argument registers in functions with variable argument list. 3346 3347 SPARC 3348 3349 * The default code model in 64-bit mode has been changed from 3350 Medium/Anywhere to Medium/Middle on Solaris. 3351 * TLS support is disabled by default on Solaris prior to release 10. 3352 It can be enabled on TLS-capable Solaris 9 versions (4/04 release 3353 and later) by specifying --enable-tls at configure time. 3354 3355 MorphoSys 3356 3357 * Support has been added for this new architecture. 3358 3359 Obsolete Systems 3360 3361 Documentation improvements 3362 3363 Other significant improvements 3364 3365 * GCC can now emit code for protecting applications from 3366 stack-smashing attacks. The protection is realized by buffer 3367 overflow detection and reordering of stack variables to avoid 3368 pointer corruption. 3369 * Some built-in functions have been fortified to protect them against 3370 various buffer overflow (and format string) vulnerabilities. 3371 Compared to the mudflap bounds checking feature, the safe builtins 3372 have far smaller overhead. This means that programs built using 3373 safe builtins should not experience any measurable slowdown. 3374 3375 GCC 4.1.2 3376 3377 This is the [6]list of problem reports (PRs) from GCC's bug tracking 3378 system that are known to be fixed in the 4.1.2 release. This list might 3379 not be complete (that is, it is possible that some PRs that have been 3380 fixed are not listed here). 3381 3382 When generating code for a shared library, GCC now recognizes that 3383 global functions may be replaced when the program runs. Therefore, it 3384 is now more conservative in deducing information from the bodies of 3385 functions. For example, in this example: 3386 void f() {} 3387 void g() { 3388 try { f(); } 3389 catch (...) { 3390 cout << "Exception"; 3391 } 3392 } 3393 3394 G++ would previously have optimized away the catch clause, since it 3395 would have concluded that f cannot throw exceptions. Because users may 3396 replace f with another function in the main body of the program, this 3397 optimization is unsafe, and is no longer performed. If you wish G++ to 3398 continue to optimize as before, you must add a throw() clause to the 3399 declaration of f to make clear that it does not throw exceptions. 3400 3401 3402 For questions related to the use of GCC, please consult these web 3403 pages and the [7]GCC manuals. If that fails, the 3404 [8]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 3405 web pages and the development of GCC are welcome on our developer 3406 list at [9]gcc (a] gcc.gnu.org. All of [10]our lists have public 3407 archives. 3408 3409 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and 3410 distribution of this entire article is permitted in any medium, 3411 provided this notice is preserved. 3412 3413 These pages are [12]maintained by the GCC team. Last modified 3414 2011-04-25[13]. 3415 3416 References 3417 3418 1. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2 3419 2. http://developer.classpath.org/mediation/ClasspathGraphicsImagesText 3420 3. http://gcc.gnu.org/ml/java/2005-05/msg00260.html 3421 4. http://developer.classpath.org/doc/ 3422 5. http://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html 3423 6. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.1.2 3424 7. http://gcc.gnu.org/onlinedocs/ 3425 8. mailto:gcc-help (a] gcc.gnu.org 3426 9. mailto:gcc (a] gcc.gnu.org 3427 10. http://gcc.gnu.org/lists.html 3428 11. http://www.fsf.org/ 3429 12. http://gcc.gnu.org/about.html 3430 13. http://validator.w3.org/check/referer 3431 ====================================================================== 3432 http://gcc.gnu.org/gcc-4.0/index.html 3433 3434 GCC 4.0 Release Series 3435 3436 January 31, 2007 3437 3438 The [1]GNU project and the GCC developers are pleased to announce the 3439 release of GCC 4.0.4. 3440 3441 This release is a bug-fix release, containing fixes for regressions in 3442 GCC 4.0.3 relative to previous releases of GCC. 3443 3444 Release History 3445 3446 GCC 4.0.4 3447 January 31, 2007 ([2]changes) 3448 3449 GCC 4.0.3 3450 March 10, 2006 ([3]changes) 3451 3452 GCC 4.0.2 3453 September 28, 2005 ([4]changes) 3454 3455 GCC 4.0.1 3456 July 7, 2005 ([5]changes) 3457 3458 GCC 4.0.0 3459 April 20, 2005 ([6]changes) 3460 3461 References and Acknowledgements 3462 3463 GCC used to stand for the GNU C Compiler, but since the compiler 3464 supports several other languages aside from C, it now stands for the 3465 GNU Compiler Collection. 3466 3467 A list of [7]successful builds is updated as new information becomes 3468 available. 3469 3470 The GCC developers would like to thank the numerous people that have 3471 contributed new features, improvements, bug fixes, and other changes as 3472 well as test results to GCC. This [8]amazing group of volunteers is 3473 what makes GCC successful. 3474 3475 For additional information about GCC please refer to the [9]GCC project 3476 web site or contact the [10]GCC development mailing list. 3477 3478 To obtain GCC please use [11]our mirror sites, or [12]our SVN server. 3479 3480 3481 For questions related to the use of GCC, please consult these web 3482 pages and the [13]GCC manuals. If that fails, the 3483 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 3484 web pages and the development of GCC are welcome on our developer 3485 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 3486 archives. 3487 3488 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 3489 distribution of this entire article is permitted in any medium, 3490 provided this notice is preserved. 3491 3492 These pages are [18]maintained by the GCC team. Last modified 3493 2011-04-25[19]. 3494 3495 References 3496 3497 1. http://www.gnu.org/ 3498 2. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4 3499 3. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.3 3500 4. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.2 3501 5. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.1 3502 6. http://gcc.gnu.org/gcc-4.0/changes.html 3503 7. http://gcc.gnu.org/gcc-4.0/buildstat.html 3504 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 3505 9. http://gcc.gnu.org/index.html 3506 10. mailto:gcc (a] gcc.gnu.org 3507 11. http://gcc.gnu.org/mirrors.html 3508 12. http://gcc.gnu.org/svn.html 3509 13. http://gcc.gnu.org/onlinedocs/ 3510 14. mailto:gcc-help (a] gcc.gnu.org 3511 15. mailto:gcc (a] gcc.gnu.org 3512 16. http://gcc.gnu.org/lists.html 3513 17. http://www.fsf.org/ 3514 18. http://gcc.gnu.org/about.html 3515 19. http://validator.w3.org/check/referer 3516 ====================================================================== 3517 http://gcc.gnu.org/gcc-4.0/changes.html 3518 3519 GCC 4.0 Release Series 3520 Changes, New Features, and Fixes 3521 3522 The latest release in the 4.0 release series is [1]GCC 4.0.4. 3523 3524 Caveats 3525 3526 * GCC now generates location lists by default when compiling with 3527 debug info and optimization. 3528 + GDB 6.0 and older crashes when it sees location lists. GDB 6.1 3529 or later is needed to debug binaries containing location 3530 lists. 3531 + When you are trying to view a value of a variable in a part of 3532 a function where it has no location (for example when the 3533 variable is no longer used and thus its location was used for 3534 something else) GDB will say that it is not available. 3535 You can disable generating location lists by -fno-var-tracking. 3536 * GCC no longer accepts the -fwritable-strings option. Use named 3537 character arrays when you need a writable string. 3538 * The options -freduce-all-givs and -fmove-all-movables have been 3539 discontinued. They were used to circumvent a shortcoming in the 3540 heuristics of the old loop optimization code with respect to common 3541 Fortran constructs. The new (tree) loop optimizer works differently 3542 and doesn't need those work-arounds. 3543 * The graph-coloring register allocator, formerly enabled by the 3544 option -fnew-ra, has been discontinued. 3545 * -I- has been deprecated. -iquote is meant to replace the need for 3546 this option. 3547 * The MIPS -membedded-pic and -mrnames options have been removed. 3548 * All MIPS targets now require the GNU assembler. In particular, IRIX 3549 configurations can no longer use the MIPSpro assemblers, although 3550 they do still support the MIPSpro linkers. 3551 * The SPARC option -mflat has been removed. 3552 * English-language diagnostic messages will now use Unicode quotation 3553 marks in UTF-8 locales. (Non-English messages already used the 3554 quotes appropriate for the language in previous releases.) If your 3555 terminal does not support UTF-8 but you are using a UTF-8 locale 3556 (such locales are the default on many GNU/Linux systems) then you 3557 should set LC_CTYPE=C in the environment to disable that locale. 3558 Programs that parse diagnostics and expect plain ASCII 3559 English-language messages should set LC_ALL=C. See [2]Markus Kuhn's 3560 explanation of Unicode quotation marks for more information. 3561 * The specs file is no longer installed on most platforms. Most users 3562 will be totally unaffected. However, if you are accustomed to 3563 editing the specs file yourself, you will now have to use the 3564 -dumpspecs option to generate the specs file, and then edit the 3565 resulting file. 3566 3567 General Optimizer Improvements 3568 3569 * The [3]tree ssa branch has been merged. This merge has brought in a 3570 completely new optimization framework based on a higher level 3571 intermediate representation than the existing RTL representation. 3572 Numerous new code transformations based on the new framework are 3573 available in GCC 4.0, including: 3574 + Scalar replacement of aggregates 3575 + Constant propagation 3576 + Value range propagation 3577 + Partial redundancy elimination 3578 + Load and store motion 3579 + Strength reduction 3580 + Dead store elimination 3581 + Dead and unreachable code elimination 3582 + [4]Autovectorization 3583 + Loop interchange 3584 + Tail recursion by accumulation 3585 Many of these passes outperform their counterparts from previous 3586 GCC releases. 3587 * [5]Swing Modulo Scheduling (SMS). An RTL level instruction 3588 scheduling optimization intended for loops that perform heavy 3589 computations. 3590 3591 New Languages and Language specific improvements 3592 3593 C family 3594 3595 * The sentinel attribute has been added to GCC. This function 3596 attribute allows GCC to warn when variadic functions such as execl 3597 are not NULL terminated. See the GCC manual for a complete 3598 description of its behavior. 3599 * Given __attribute__((alias("target"))) it is now an error if target 3600 is not a symbol, defined in the same translation unit. This also 3601 applies to aliases created by #pragma weak alias=target. This is 3602 because it's meaningless to define an alias to an undefined symbol. 3603 On Solaris, the native assembler would have caught this error, but 3604 GNU as does not. 3605 3606 C and Objective-C 3607 3608 * The -Wstrict-aliasing=2 option has been added. This warning catches 3609 all unsafe cases, but it may also give a warning for some cases 3610 that are safe. 3611 * The cast-as-lvalue, conditional-expression-as-lvalue and 3612 compound-expression-as-lvalue extensions, which were deprecated in 3613 3.3.4 and 3.4, have been removed. 3614 * The -fwritable-strings option, which was deprecated in 3.4, has 3615 been removed. 3616 * #pragma pack() semantics have been brought closer to those used by 3617 other compilers. This also applies to C++. 3618 * Taking the address of a variable with register storage is invalid 3619 in C. GCC now issues an error instead of a warning. 3620 * Arrays of incomplete element type are invalid in C. GCC now issues 3621 an error for such arrays. Declarations such as extern struct s x[]; 3622 (where struct s has not been defined) can be moved after the 3623 definition of struct s. Function parameters declared as arrays of 3624 incomplete type can instead be declared as pointers. 3625 3626 C++ 3627 3628 * When compiling without optimizations (-O0), the C++ frontend is 3629 much faster than in any previous versions of GCC. Independent 3630 testers have measured speed-ups up to 25% in real-world production 3631 code, compared to the 3.4 family (which was already the fastest 3632 version to date). Upgrading from older versions might show even 3633 bigger improvements. 3634 * ELF visibility attributes can now be applied to a class type, so 3635 that it affects every member function of a class at once, without 3636 having to specify each individually: 3637 class __attribute__ ((visibility("hidden"))) Foo 3638 { 3639 int foo1(); 3640 void foo2(); 3641 }; 3642 The syntax is deliberately similar to the __declspec() system used 3643 by Microsoft Windows based compilers, allowing cross-platform 3644 projects to easily reuse their existing macro system for denoting 3645 exports and imports. By explicitly marking internal classes never 3646 used outside a binary as hidden, one can completely avoid PLT 3647 indirection overheads during their usage by the compiler. You can 3648 find out more about the advantages of this at 3649 [6]http://www.akkadia.org/drepper/dsohowto.pdf 3650 * The -fvisibility-inlines-hidden option has been added which marks 3651 all inlineable functions as having hidden ELF visibility, thus 3652 removing their symbol and typeinfo from the exported symbol table 3653 of the output ELF binary. Using this option can reduce the exported 3654 symbol count of template-heavy code by up to 40% with no code 3655 change at all, thus notably improving link and load times for the 3656 binary as well as a reduction in size of up to 10%. Also, check the 3657 new [7]-fvisibility option. 3658 * The compiler now uses the library interface specified by the [8]C++ 3659 ABI for thread-safe initialization of function-scope static 3660 variables. Most users should leave this alone, but embedded 3661 programmers may want to disable this by specifying 3662 -fno-threadsafe-statics for a small savings in code size. 3663 * Taking the address of an explicit register variable is no longer 3664 supported. Note that C++ allows taking the address of variables 3665 with register storage so this will continue to compile with a 3666 warning. For example, assuming that r0 is a machine register: 3667 register int foo asm ("r0"); 3668 register int bar; 3669 &foo; // error, no longer accepted 3670 &bar; // OK, with a warning 3671 * G++ has an undocumented extension to virtual function covariancy 3672 rules that allowed the overrider to return a type that was 3673 implicitly convertable to the overridden function's return type. 3674 For instance a function returning void * could be overridden by a 3675 function returning T *. This is now deprecated and will be removed 3676 in a future release. 3677 * The G++ minimum and maximum operators (<? and >?) and their 3678 compound forms (<?=) and >?=) have been deprecated and will be 3679 removed in a future version. Code using these operators should be 3680 modified to use std::min and std::max instead. 3681 * Declaration of nested classes of class templates as friends are 3682 supported: 3683 template <typename T> struct A { 3684 class B {}; 3685 }; 3686 class C { 3687 template <typename T> friend class A<T>::B; 3688 }; 3689 This complements the feature member functions of class templates as 3690 friends introduced in GCC 3.4.0. 3691 * When declaring a friend class using an unqualified name, classes 3692 outside the innermost non-class scope are not searched: 3693 class A; 3694 namespace N { 3695 class B { 3696 friend class A; // Refer to N::A which has not been declared yet 3697 // because name outside namespace N are not searched 3698 friend class ::A; // Refer to ::A 3699 }; 3700 } 3701 Hiding the friend name until declaration is still not implemented. 3702 * Friends of classes defined outside their namespace are correctly 3703 handled: 3704 namespace N { 3705 class A; 3706 } 3707 class N::A { 3708 friend class B; // Refer to N::B in GCC 4.0.0 3709 // but ::B in earlier versions of GCC 3710 }; 3711 3712 Runtime Library (libstdc++) 3713 3714 * Optimization work: 3715 + Added efficient specializations of istream functions for char 3716 and wchar_t. 3717 + Further performance tuning of strings, in particular wrt 3718 single-char append and getline. 3719 + iter_swap - and therefore most of the mutating algorithms - 3720 now makes an unqualified call to swap when the value_type of 3721 the two iterators is the same. 3722 * A large subset of the features in Technical Report 1 (TR1 for 3723 short) is experimentally delivered (i.e., no guarantees about the 3724 implementation are provided. In particular it is not promised that 3725 the library will remain link-compatible when code using TR1 is 3726 used): 3727 + General utilities such as reference_wrapper and shared_ptr. 3728 + Function objects, i.e., result_of, mem_fn, bind, function. 3729 + Support for metaprogramming. 3730 + New containers such as tuple, array, unordered_set, 3731 unordered_map, unordered_multiset, unordered_multimap. 3732 * As usual, many bugs have been fixed and LWG resolutions implemented 3733 for the first time (e.g., DR 409). 3734 3735 Java 3736 3737 * In order to prevent naming conflicts with other implementations of 3738 these tools, some GCJ binaries have been renamed: 3739 + rmic is now grmic, 3740 + rmiregistry is now grmiregistry, and 3741 + jar is now fastjar. 3742 In particular, these names were problematic for the jpackage.org 3743 packaging conventions which install symlinks in /usr/bin that point 3744 to the preferred versions of these tools. 3745 * The -findirect-dispatch argument to the compiler now works and 3746 generates code following a new "binary compatibility" ABI. Code 3747 compiled this way follows the binary compatibility rules of the 3748 Java Language Specification. 3749 * libgcj now has support for using GCJ as a JIT, using the 3750 gnu.gcj.jit family of system properties. 3751 * libgcj can now find a shared library corresponding to the bytecode 3752 representation of a class. See the documentation for the new 3753 gcj-dbtool program, and the new gnu.gcj.precompiled.db.path system 3754 property. 3755 * There have been many improvements to the class library. Here are 3756 some highlights: 3757 + Much more of AWT and Swing exist. 3758 + Many new packages and classes were added, including 3759 java.util.regex, java.net.URI, javax.crypto, 3760 javax.crypto.interfaces, javax.crypto.spec, javax.net, 3761 javax.net.ssl, javax.security.auth, 3762 javax.security.auth.callback, javax.security.auth.login, 3763 javax.security.auth.x500, javax.security.sasl, org.ietf.jgss, 3764 javax.imageio, javax.imageio.event, javax.imageio.spi, 3765 javax.print, javax.print.attribute, 3766 javax.print.attribute.standard, javax.print.event, and 3767 javax.xml 3768 + Updated SAX and DOM, and imported GNU JAXP 3769 3770 Fortran 3771 3772 * A new [9]Fortran front end has replaced the aging GNU Fortran 77 3773 front end. The new front end supports Fortran 90 and Fortran 95. It 3774 may not yet be as stable as the old Fortran front end. 3775 3776 Ada 3777 3778 * Ada (with tasking and Zero Cost Exceptions) is now available on 3779 many more targets, including but not limited to: alpha-linux, 3780 hppa-hpux, hppa-linux, powerpc-darwin, powerpc-linux, s390-linux, 3781 s390x-linux, sparc-linux. 3782 * Some of the new Ada 2005 features are now implemented like 3783 Wide_Wide_Character and Ada.Containers. 3784 * Many bugs have been fixed, tools and documentation improved. 3785 * To compile Ada from the sources, install an older working Ada 3786 compiler and then use --enable-languages=ada at configuration time, 3787 since the Ada frontend is not currently activated by default. See 3788 the [10]Installing GCC for details. 3789 3790 New Targets and Target Specific Improvements 3791 3792 H8/300 3793 3794 * The frame layout has changed. In the new layout, the prologue of a 3795 function first saves registers and then allocate space for locals, 3796 resulting in an 1% improvement on code size. 3797 3798 IA-32/x86-64 (AMD64) 3799 3800 * The acos, asin, drem, exp10, exp2, expm1, fmod, ilogb, log10, 3801 log1p, log2, logb and tan mathematical builtins (and their float 3802 and long double variants) are now implemented as inline x87 3803 intrinsics when using -ffast-math. 3804 * The ceil, floor, nearbyint, rint and trunc mathematical builtins 3805 (and their float and long double variants) are now implemented as 3806 inline x87 intrinsics when using -ffast-math. 3807 * The x87's fsincos instruction is now used automatically with 3808 -ffast-math when calculating both the sin and cos of the same 3809 argument. 3810 * Instruction selection for multiplication and division by constants 3811 has been improved. 3812 3813 IA-64 3814 3815 * Floating point division, integer division and sqrt are now inlined, 3816 resulting in significant performance improvements on some codes. 3817 3818 MIPS 3819 3820 * Division by zero checks now use conditional traps if the target 3821 processor supports them. This decreases code size by one word per 3822 division operation. The old behavior (branch and break) can be 3823 obtained either at configure time by passing --with-divide=breaks 3824 to configure or at runtime by passing -mdivide-breaks to GCC. 3825 * Support for MIPS64 paired-single instructions has been added. It is 3826 enabled by -mpaired-single and can be accessed using both the 3827 target-independent vector extensions and new MIPS-specific built-in 3828 functions. 3829 * Support for the MIPS-3D ASE has been added. It is enabled by 3830 -mips3d and provides new MIPS-3D-specific built-in functions. 3831 * The -mexplicit-relocs option now supports static n64 code (as is 3832 used, for example, in 64-bit linux kernels). -mexplicit-relocs 3833 should now be feature-complete and is enabled by default when GCC 3834 is configured to use a compatible assembler. 3835 * Support for the NEC VR4130 series has been added. This support 3836 includes the use of VR-specific instructions and a new VR4130 3837 scheduler. Full VR4130 support can be selected with -march=vr4130 3838 while code for any ISA can be tuned for the VR4130 using 3839 -mtune=vr4130. There is also a new -mvr4130-align option that 3840 produces better schedules at the cost of increased code size. 3841 * Support for the Broadcom SB-1 has been extended. There is now an 3842 SB-1 scheduler as well as support for the SB-1-specific 3843 paired-single instructions. Full SB-1 support can be selected with 3844 -march=sb1 while code for any ISA can be optimized for the SB-1 3845 using -mtune=sb1. 3846 * The compiler can now work around errata in R4000, R4400, VR4120 and 3847 VR4130 processors. These workarounds are enabled by -mfix-r4000, 3848 -mfix-r4400, -mfix-vr4120 and -mfix-vr4130 respectively. The VR4120 3849 and VR4130 workarounds need binutils 2.16 or above. 3850 * IRIX shared libraries are now installed into the standard library 3851 directories: o32 libraries go into lib/, n32 libraries go into 3852 lib32/ and n64 libraries go into lib64/. 3853 * The compiler supports a new -msym32 option. It can be used to 3854 optimize n64 code in which all symbols are known to have 32-bit 3855 values. 3856 3857 S/390 and zSeries 3858 3859 * New command-line options help to generate code intended to run in 3860 an environment where stack space is restricted, e.g. Linux kernel 3861 code: 3862 + -mwarn-framesize and -mwarn-dynamicstack trigger compile-time 3863 warnings for single functions that require large or dynamic 3864 stack frames. 3865 + -mstack-size and -mstack-guard generate code that checks for 3866 stack overflow at run time. 3867 + -mpacked-stack generates code that reduces the stack frame 3868 size of many functions by reusing unneeded parts of the stack 3869 bias area. 3870 * The -msoft-float option now ensures that generated code never 3871 accesses floating point registers. 3872 * The s390x-ibm-tpf target now fully supports C++, including 3873 exceptions and threads. 3874 * Various changes to improve performance of the generated code have 3875 been implemented, including: 3876 + GCC now uses sibling calls where possible. 3877 + Condition code handling has been optimized, allowing GCC to 3878 omit redundant comparisons in certain cases. 3879 + The cost function guiding many optimizations has been refined 3880 to more accurately represent the z900 and z990 processors. 3881 + The ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL WITH BORROW 3882 instructions are now used to avoid conditional branches in 3883 certain cases. 3884 + The back end now uses the LEGITIMIZE_RELOAD_ADDRESS feature to 3885 optimize address arithmetic required to access large stack 3886 frames. 3887 + GCC now makes more efficient use of memory-to-memory type 3888 instructions (MVC, CLC, ...). 3889 + More precise tracking of special register use allows better 3890 instruction scheduling, in particular of the function prologue 3891 and epilogue sequences. 3892 + The Java front end now generates inline code to implement 3893 integer division, instead of calling library routines. 3894 3895 SPARC 3896 3897 * The options -mv8, -msparclite, -mcypress, -msupersparc, -mf930 and 3898 -mf934 have been removed. They have been replaced with -mcpu=xxx. 3899 * The internal model used to estimate the relative cost of each 3900 instruction has been updated. It is expected to give better results 3901 on recent UltraSPARC processors. 3902 * Code generation for function prologues and epilogues has been 3903 improved, resulting in better scheduling and allowing multiple exit 3904 points in functions. 3905 * Support for Sun's Visual Instruction Set (VIS) has been enhanced. 3906 It is enabled by -mvis and provides new built-in functions for VIS 3907 instructions on UltraSPARC processors. 3908 * The option -mapp-regs has been turned on by default on Solaris too. 3909 3910 NetWare 3911 3912 * Novell NetWare (on ix86, no other hardware platform was ever really 3913 supported by this OS) has been re-enabled and the ABI supported by 3914 GCC has been brought into sync with that of MetroWerks CodeWarrior 3915 (the ABI previously supported was that of some Unix systems, which 3916 NetWare never tried to support). 3917 3918 Obsolete Systems 3919 3920 Support for a number of older systems has been declared obsolete in GCC 3921 4.0. Unless there is activity to revive them, the next release of GCC 3922 will have their sources permanently removed. 3923 3924 All GCC ports for the following processor architectures have been 3925 declared obsolete: 3926 * Intel i860 3927 * Ubicom IP2022 3928 * National Semiconductor NS32K 3929 * Texas Instruments TMS320C[34]x 3930 3931 Also, those for some individual systems have been obsoleted: 3932 * SPARC family 3933 + SPARClite-based systems (sparclite-*-coff, sparclite-*-elf, 3934 sparc86x-*-elf) 3935 + OpenBSD 32-bit (sparc-*-openbsd*) 3936 3937 Documentation improvements 3938 3939 Other significant improvements 3940 3941 * Location lists are now generated by default when compiling with 3942 debug info and optimization. Location lists provide more accurate 3943 debug info about locations of variables and they allow debugging 3944 code compiled with -fomit-frame-pointer. 3945 * The -fvisibility option has been added which allows the default ELF 3946 visibility of all symbols to be set per compilation and the new 3947 #pragma GCC visibility preprocessor command allows the setting of 3948 default ELF visibility for a region of code. Using 3949 -fvisibility=hidden especially in combination with the new 3950 -fvisibility-inlines-hidden can yield substantial improvements in 3951 output binary quality including avoiding PLT indirection overheads, 3952 reduction of the exported symbol count by up to 60% (with resultant 3953 improvements to link and load times), better scope for the 3954 optimizer to improve code and up to a 20% reduction in binary size. 3955 Using these options correctly yields a binary with a similar symbol 3956 count to a Windows DLL. 3957 Perhaps more importantly, this new feature finally allows (with 3958 careful planning) complete avoidance of symbol clashes when 3959 manually loading shared objects with RTLD_GLOBAL, thus finally 3960 solving problems many projects such as python were forced to use 3961 RTLD_LOCAL for (with its resulting issues for C++ correctness). You 3962 can find more information about using these options at 3963 [11]http://gcc.gnu.org/wiki/Visibility. 3964 __________________________________________________________________ 3965 3966 GCC 4.0.1 3967 3968 This is the [12]list of problem reports (PRs) from GCC's bug tracking 3969 system that are known to be fixed in the 4.0.1 release. This list might 3970 not be complete (that is, it is possible that some PRs that have been 3971 fixed are not listed here). 3972 3973 GCC 4.0.2 3974 3975 This is the [13]list of problem reports (PRs) from GCC's bug tracking 3976 system that are known to be fixed in the 4.0.2 release. This list might 3977 not be complete (that is, it is possible that some PRs that have been 3978 fixed are not listed here). 3979 3980 Unfortunately, due to a release engineering failure, this release has a 3981 regression on Solaris that will affect some C++ programs. We suggest 3982 that Solaris users apply a [14]patch that corrects the problem. Users 3983 who do not wish to apply the patch should explicitly link C++ programs 3984 with the -pthreads option, even if they do not use threads. This 3985 problem has been corrected in the current 4.0 branch sources and will 3986 not be present in GCC 4.0.3. 3987 3988 GCC 4.0.3 3989 3990 Starting with this release, the function getcontext is recognized by 3991 the compiler as having the same semantics as the setjmp function. In 3992 particular, the compiler will ensure that all registers are dead before 3993 calling such a function and will emit a warning about the variables 3994 that may be clobbered after the second return from the function. 3995 3996 GCC 4.0.4 3997 3998 This is the [15]list of problem reports (PRs) from GCC's bug tracking 3999 system that are known to be fixed in the 4.0.4 release. This list might 4000 not be complete (that is, it is possible that some PRs that have been 4001 fixed are not listed here). 4002 4003 The 4.0.4 release is provided for those that require a high degree of 4004 binary compatibility with previous 4.0.x releases. For most users, the 4005 GCC team recommends that version 4.1.1 or later be used instead." 4006 4007 4008 For questions related to the use of GCC, please consult these web 4009 pages and the [16]GCC manuals. If that fails, the 4010 [17]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 4011 web pages and the development of GCC are welcome on our developer 4012 list at [18]gcc (a] gcc.gnu.org. All of [19]our lists have public 4013 archives. 4014 4015 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and 4016 distribution of this entire article is permitted in any medium, 4017 provided this notice is preserved. 4018 4019 These pages are [21]maintained by the GCC team. Last modified 4020 2011-04-25[22]. 4021 4022 References 4023 4024 1. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4 4025 2. http://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html 4026 3. http://gcc.gnu.org/projects/tree-ssa/ 4027 4. http://gcc.gnu.org/projects/tree-ssa/vectorization.html 4028 5. http://gcc.gnu.org/news/sms.html 4029 6. http://www.akkadia.org/drepper/dsohowto.pdf 4030 7. http://gcc.gnu.org/gcc-4.0/changes.html#visibility 4031 8. http://www.codesourcery.com/public/cxx-abi/ 4032 9. http://gcc.gnu.org/fortran/ 4033 10. http://gcc.gnu.org/install/ 4034 11. http://gcc.gnu.org/wiki/Visibility 4035 12. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.1 4036 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.2 4037 14. http://gcc.gnu.org/ml/gcc-cvs/2005-09/msg00984.html 4038 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.4 4039 16. http://gcc.gnu.org/onlinedocs/ 4040 17. mailto:gcc-help (a] gcc.gnu.org 4041 18. mailto:gcc (a] gcc.gnu.org 4042 19. http://gcc.gnu.org/lists.html 4043 20. http://www.fsf.org/ 4044 21. http://gcc.gnu.org/about.html 4045 22. http://validator.w3.org/check/referer 4046 ====================================================================== 4047 http://gcc.gnu.org/gcc-3.4/index.html 4048 4049 GCC 3.4 Release Series 4050 4051 May 26, 2006 4052 4053 The [1]GNU project and the GCC developers are pleased to announce the 4054 release of GCC 3.4.6. 4055 4056 This release is a bug-fix release, containing fixes for regressions in 4057 GCC 3.4.4 relative to previous releases of GCC. This is the last of the 4058 3.4.x series. 4059 4060 The GCC 3.4 release series includes numerous [2]new features, 4061 improvements, bug fixes, and other changes, thanks to an [3]amazing 4062 group of volunteers. 4063 4064 Release History 4065 4066 GCC 3.4.6 4067 March 6, 2006 ([4]changes) 4068 4069 GCC 3.4.5 4070 November 30, 2005 ([5]changes) 4071 4072 GCC 3.4.4 4073 May 18, 2005 ([6]changes) 4074 4075 GCC 3.4.3 4076 November 4, 2004 ([7]changes) 4077 4078 GCC 3.4.2 4079 September 6, 2004 ([8]changes) 4080 4081 GCC 3.4.1 4082 July 1, 2004 ([9]changes) 4083 4084 GCC 3.4.0 4085 April 18, 2004 ([10]changes) 4086 4087 References and Acknowledgements 4088 4089 GCC used to stand for the GNU C Compiler, but since the compiler 4090 supports several other languages aside from C, it now stands for the 4091 GNU Compiler Collection. 4092 4093 A list of [11]successful builds is updated as new information becomes 4094 available. 4095 4096 The GCC developers would like to thank the numerous people that have 4097 contributed new features, improvements, bug fixes, and other changes as 4098 well as test results to GCC. This [12]amazing group of volunteers is 4099 what makes GCC successful. 4100 4101 For additional information about GCC please refer to the [13]GCC 4102 project web site or contact the [14]GCC development mailing list. 4103 4104 To obtain GCC please use [15]our mirror sites, or [16]our SVN server. 4105 4106 4107 For questions related to the use of GCC, please consult these web 4108 pages and the [17]GCC manuals. If that fails, the 4109 [18]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 4110 web pages and the development of GCC are welcome on our developer 4111 list at [19]gcc (a] gcc.gnu.org. All of [20]our lists have public 4112 archives. 4113 4114 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and 4115 distribution of this entire article is permitted in any medium, 4116 provided this notice is preserved. 4117 4118 These pages are [22]maintained by the GCC team. Last modified 4119 2011-04-25[23]. 4120 4121 References 4122 4123 1. http://www.gnu.org/ 4124 2. http://gcc.gnu.org/gcc-3.4/changes.html 4125 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 4126 4. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6 4127 5. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.5 4128 6. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.4 4129 7. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.3 4130 8. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.2 4131 9. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.1 4132 10. http://gcc.gnu.org/gcc-3.4/changes.html 4133 11. http://gcc.gnu.org/gcc-3.4/buildstat.html 4134 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 4135 13. http://gcc.gnu.org/index.html 4136 14. mailto:gcc (a] gcc.gnu.org 4137 15. http://gcc.gnu.org/mirrors.html 4138 16. http://gcc.gnu.org/svn.html 4139 17. http://gcc.gnu.org/onlinedocs/ 4140 18. mailto:gcc-help (a] gcc.gnu.org 4141 19. mailto:gcc (a] gcc.gnu.org 4142 20. http://gcc.gnu.org/lists.html 4143 21. http://www.fsf.org/ 4144 22. http://gcc.gnu.org/about.html 4145 23. http://validator.w3.org/check/referer 4146 ====================================================================== 4147 http://gcc.gnu.org/gcc-3.4/changes.html 4148 4149 GCC 3.4 Release Series 4150 Changes, New Features, and Fixes 4151 4152 The final release in the 3.4 release series is [1]GCC 3.4.6. The series 4153 is now closed. 4154 4155 GCC 3.4 has [2]many improvements in the C++ frontend. Before reporting 4156 a bug, please make sure it's really GCC, and not your code, that is 4157 broken. 4158 4159 Caveats 4160 4161 * GNU Make is now required to build GCC. 4162 * With -nostdinc the preprocessor used to ignore both standard 4163 include paths and include paths contained in environment variables. 4164 It was neither documented nor intended that environment variable 4165 paths be ignored, so this has been corrected. 4166 * GCC no longer accepts the options -fvolatile, -fvolatile-global and 4167 -fvolatile-static. It is unlikely that they worked correctly in any 4168 3.x release. 4169 * GCC no longer ships <varargs.h>. Use <stdarg.h> instead. 4170 * Support for all the systems [3]obsoleted in GCC 3.3 has been 4171 removed from GCC 3.4. See below for a [4]list of systems which are 4172 obsoleted in this release. 4173 * GCC now requires an ISO C90 (ANSI C89) C compiler to build. K&R C 4174 compilers will not work. 4175 * The implementation of the [5]MIPS ABIs has changed. As a result, 4176 the code generated for certain MIPS targets will not be binary 4177 compatible with earlier releases. 4178 * In previous releases, the MIPS port had a fake "hilo" register with 4179 the user-visible name accum. This register has been removed. 4180 * The implementation of the [6]SPARC ABIs has changed. As a result, 4181 the code generated will not be binary compatible with earlier 4182 releases in certain cases. 4183 * The configure option --enable-threads=pthreads has been removed; 4184 use --enable-threads=posix instead, which should have the same 4185 effect. 4186 * Code size estimates used by inlining heuristics for C, Objective-C, 4187 C++ and Java have been redesigned significantly. As a result the 4188 parameters of -finline-insns, --param max-inline-insns-single and 4189 --param max-inline-insns-auto need to be reconsidered. 4190 * --param max-inline-slope and --param min-inline-insns have been 4191 removed; they are not needed for the new bottom-up inlining 4192 heuristics. 4193 * The new unit-at-a-time compilation scheme has several compatibility 4194 issues: 4195 + The order in which functions, variables, and top-level asm 4196 statements are emitted may have changed. Code relying on some 4197 particular ordering needs to be updated. The majority of such 4198 top-level asm statements can be replaced by section 4199 attributes. 4200 + Unreferenced static variables and functions are removed. This 4201 may result in undefined references when an asm statement 4202 refers to the variable/function directly. In that case either 4203 the variable/function shall be listed in asm statement operand 4204 or in the case of top-level asm statements the attribute used 4205 shall be used to force function/variable to be always output 4206 and considered as a possibly used by unknown code. 4207 For variables the attribute is accepted only by GCC 3.4 and 4208 newer, while for earlier versions it is sufficient to use 4209 unused to silence warnings about the variables not being 4210 referenced. To keep code portable across different GCC 4211 versions, you can use appropriate preprocessor conditionals. 4212 + Static functions now can use non-standard passing conventions 4213 that may break asm statements calling functions directly. 4214 Again the attribute used shall be used to prevent this 4215 behavior. 4216 As a temporary workaround, -fno-unit-at-a-time can be used, but 4217 this scheme may not be supported by future releases of GCC. 4218 * GCC 3.4 automatically places zero-initialized variables in the .bss 4219 section on some operating systems. Versions of GNU Emacs up to (and 4220 including) 21.3 will not work correctly when using this 4221 optimization; you can use -fno-zero-initialized-in-bss to disable 4222 it. 4223 * If GCC 3.4 is configured with --enable-threads=posix (the default 4224 on most targets that support pthreads) then _REENTRANT will be 4225 defined unconditionally by some libstdc++ headers. C++ code which 4226 relies on that macro to detect whether multi-threaded code is being 4227 compiled might change in meaning, possibly resulting in linker 4228 errors for single-threaded programs. Affected users of [7]Boost 4229 should compile single-threaded code with -DBOOST_DISABLE_THREADS. 4230 See Bugzilla for [8]more information. 4231 4232 General Optimizer Improvements 4233 4234 * Usability of the profile feedback and coverage testing has been 4235 improved. 4236 + Performance of profiled programs has been improved by faster 4237 profile merging code. 4238 + Better use of the profile feedback for optimization (loop 4239 unrolling and loop peeling). 4240 + File locking support allowing fork() calls and parallel runs 4241 of profiled programs. 4242 + Coverage file format has been redesigned. 4243 + gcov coverage tool has been improved. 4244 + make profiledbootstrap available to build a faster compiler. 4245 Experiments made on i386 hardware showed an 11% speedup on -O0 4246 and a 7.5% speedup on -O2 compilation of a [9]large C++ 4247 testcase. 4248 + New value profiling pass enabled via -fprofile-values 4249 + New value profile transformations pass enabled via -fvpt aims 4250 to optimize some code sequences by exploiting knowledge about 4251 value ranges or other properties of the operands. At the 4252 moment a conversion of expensive divisions into cheaper 4253 operations has been implemented. 4254 + New -fprofile-generate and -fprofile-use command-line options 4255 to simplify the use of profile feedback. 4256 * A new unit-at-a-time compilation scheme for C, Objective-C, C++ and 4257 Java which is enabled via -funit-at-a-time (and implied by -O2). In 4258 this scheme a whole file is parsed first and optimized later. The 4259 following basic inter-procedural optimizations are implemented: 4260 + Removal of unreachable functions and variables 4261 + Discovery of local functions (functions with static linkage 4262 whose address is never taken) 4263 + On i386, these local functions use register parameter passing 4264 conventions. 4265 + Reordering of functions in topological order of the call graph 4266 to enable better propagation of optimizing hints (such as the 4267 stack alignments needed by functions) in the back end. 4268 + Call graph based out-of-order inlining heuristics which allows 4269 to limit overall compilation unit growth (--param 4270 inline-unit-growth). 4271 Overall, the unit-at-a-time scheme produces a 1.3% improvement for 4272 the SPECint2000 benchmark on the i386 architecture (AMD Athlon 4273 CPU). 4274 * More realistic code size estimates used by inlining for C, 4275 Objective-C, C++ and Java. The growth of large functions can now be 4276 limited via --param large-function-insns and --param 4277 large-function-growth. 4278 * A new cfg-level loop optimizer pass replaces the old loop unrolling 4279 pass and adds two other loop transformations -- loop peeling and 4280 loop unswitching -- and also uses the profile feedback to limit 4281 code growth. (The three optimizations are enabled by 4282 -funroll-loops, -fpeel-loops and -funswitch-loops flags, 4283 respectively). 4284 The old loop unroller still can be enabled by -fold-unroll-loops 4285 and may produce better code in some cases, especially when the 4286 webizer optimization pass is not run. 4287 * A new web construction pass enabled via -fweb (and implied by -O3) 4288 improves the quality of register allocation, CSE, first scheduling 4289 pass and some other optimization passes by avoiding re-use of 4290 pseudo registers with non-overlapping live ranges. The pass almost 4291 always improves code quality but does make debugging difficult and 4292 thus is not enabled by default by -O2 4293 The pass is especially effective as cleanup after code duplication 4294 passes, such as the loop unroller or the tracer. 4295 * Experimental implementations of superblock or trace scheduling in 4296 the second scheduling pass can be enabled via 4297 -fsched2-use-superblocks and -fsched2-use-traces, respectively. 4298 4299 New Languages and Language specific improvements 4300 4301 Ada 4302 4303 * The Ada front end has been updated to include numerous bug fixes 4304 and enhancements. These include: 4305 + Improved project file support 4306 + Additional set of warnings about potential wrong code 4307 + Improved error messages 4308 + Improved code generation 4309 + Improved cross reference information 4310 + Improved inlining 4311 + Better run-time check elimination 4312 + Better error recovery 4313 + More efficient implementation of unbounded strings 4314 + Added features in GNAT.Sockets, GNAT.OS_Lib, GNAT.Debug_Pools, 4315 ... 4316 + New GNAT.xxxx packages (e.g. GNAT.Strings, 4317 GNAT.Exception_Action) 4318 + New pragmas 4319 + New -gnatS switch replacing gnatpsta 4320 + Implementation of new Ada features (in particular limited 4321 with, limited aggregates) 4322 4323 C/Objective-C/C++ 4324 4325 * Precompiled headers are now supported. Precompiled headers can 4326 dramatically speed up compilation of some projects. There are some 4327 known defects in the current precompiled header implementation that 4328 will result in compiler crashes in relatively rare situations. 4329 Therefore, precompiled headers should be considered a "technology 4330 preview" in this release. Read the manual for details about how to 4331 use precompiled headers. 4332 * File handling in the preprocessor has been rewritten. GCC no longer 4333 gets confused by symlinks and hardlinks, and now has a correct 4334 implementation of #import and #pragma once. These two directives 4335 have therefore been un-deprecated. 4336 * The undocumented extension that allowed C programs to have a label 4337 at the end of a compound statement, which has been deprecated since 4338 GCC 3.0, has been removed. 4339 * The cast-as-lvalue extension has been removed for C++ and 4340 deprecated for C and Objective-C. In particular, code like this: 4341 int i; 4342 (char) i = 5; 4343 4344 or this: 4345 char *p; 4346 ((int *) p)++; 4347 4348 is no longer accepted for C++ and will not be accepted for C and 4349 Objective-C in a future version. 4350 * The conditional-expression-as-lvalue extension has been deprecated 4351 for C and Objective-C. In particular, code like this: 4352 int a, b, c; 4353 (a ? b : c) = 2; 4354 4355 will not be accepted for C and Objective-C in a future version. 4356 * The compound-expression-as-lvalue extension has been deprecated for 4357 C and Objective-C. In particular, code like this: 4358 int a, b; 4359 (a, b) = 2; 4360 4361 will not be accepted for C and Objective-C in a future version. A 4362 possible non-intrusive workaround is the following: 4363 (*(a, &b)) = 2; 4364 4365 * Several [10]built-in functions such as __builtin_popcount for 4366 counting bits, finding the highest and lowest bit in a word, and 4367 parity have been added. 4368 * The -fwritable-strings option has been deprecated and will be 4369 removed. 4370 * Many C math library functions are now recognized as built-ins and 4371 optimized. 4372 * The C, C++, and Objective-C compilers can now handle source files 4373 written in any character encoding supported by the host C library. 4374 The default input character set is taken from the current locale, 4375 and may be overridden with the -finput-charset command line option. 4376 In the future we will add support for inline encoding markers. 4377 4378 C++ 4379 4380 * G++ is now much closer to full conformance to the ISO/ANSI C++ 4381 standard. This means, among other things, that a lot of invalid 4382 constructs which used to be accepted in previous versions will now 4383 be rejected. It is very likely that existing C++ code will need to 4384 be fixed. This document lists some of the most common issues. 4385 * A hand-written recursive-descent C++ parser has replaced the 4386 YACC-derived C++ parser from previous GCC releases. The new parser 4387 contains much improved infrastructure needed for better parsing of 4388 C++ source codes, handling of extensions, and clean separation 4389 (where possible) between proper semantics analysis and parsing. The 4390 new parser fixes many bugs that were found in the old parser. 4391 * You must now use the typename and template keywords to disambiguate 4392 dependent names, as required by the C++ standard. 4393 struct K { 4394 typedef int mytype_t; 4395 }; 4396 4397 template <class T1> struct A { 4398 template <class T2> struct B { 4399 void callme(void); 4400 }; 4401 4402 template <int N> void bar(void) 4403 { 4404 // Use 'typename' to tell the parser that T1::mytype_t names 4405 // a type. This is needed because the name is dependent (in 4406 // this case, on template parameter T1). 4407 typename T1::mytype_t x; 4408 x = 0; 4409 } 4410 }; 4411 4412 template <class T> void template_func(void) 4413 { 4414 // Use 'template' to prefix member templates within 4415 // dependent types (a has type A<T>, which depends on 4416 // the template parameter T). 4417 A<T> a; 4418 a.template bar<0>(); 4419 4420 // Use 'template' to tell the parser that B is a nested 4421 // template class (dependent on template parameter T), and 4422 // 'typename' because the whole A<T>::B<int> is 4423 // the name of a type (again, dependent). 4424 typename A<T>::template B<int> b; 4425 b.callme(); 4426 } 4427 4428 void non_template_func(void) 4429 { 4430 // Outside of any template class or function, no names can be 4431 // dependent, so the use of the keyword 'typename' and 'template' 4432 // is not needed (and actually forbidden). 4433 A<K> a; 4434 a.bar<0>(); 4435 A<K>::B<float> b; 4436 b.callme(); 4437 } 4438 * In a template definition, unqualified names will no longer find 4439 members of a dependent base (as specified by [temp.dep]/3 in the 4440 C++ standard). For example, 4441 template <typename T> struct B { 4442 int m; 4443 int n; 4444 int f (); 4445 int g (); 4446 }; 4447 int n; 4448 int g (); 4449 template <typename T> struct C : B<T> { 4450 void h () 4451 { 4452 m = 0; // error 4453 f (); // error 4454 n = 0; // ::n is modified 4455 g (); // ::g is called 4456 } 4457 }; 4458 You must make the names dependent, e.g. by prefixing them with 4459 this->. Here is the corrected definition of C<T>::h, 4460 template <typename T> void C<T>::h () 4461 { 4462 this->m = 0; 4463 this->f (); 4464 this->n = 0 4465 this->g (); 4466 } 4467 As an alternative solution (unfortunately not backwards compatible 4468 with GCC 3.3), you may use using declarations instead of this->: 4469 template <typename T> struct C : B<T> { 4470 using B<T>::m; 4471 using B<T>::f; 4472 using B<T>::n; 4473 using B<T>::g; 4474 void h () 4475 { 4476 m = 0; 4477 f (); 4478 n = 0; 4479 g (); 4480 } 4481 }; 4482 * In templates, all non-dependent names are now looked up and bound 4483 at definition time (while parsing the code), instead of later when 4484 the template is instantiated. For instance: 4485 void foo(int); 4486 4487 template <int> struct A { 4488 static void bar(void){ 4489 foo('a'); 4490 } 4491 }; 4492 4493 void foo(char); 4494 4495 int main() 4496 { 4497 A<0>::bar(); // Calls foo(int), used to call foo(char). 4498 } 4499 4500 * In an explicit instantiation of a class template, you must use 4501 class or struct before the template-id: 4502 template <int N> 4503 class A {}; 4504 4505 template A<0>; // error, not accepted anymore 4506 template class A<0>; // OK 4507 * The "named return value" and "implicit typename" extensions have 4508 been removed. 4509 * Default arguments in function types have been deprecated and will 4510 be removed. 4511 * ARM-style name-injection of friend declarations has been deprecated 4512 and will be removed. For example: struct S { friend void f(); }; 4513 void g() { f(); } will not be accepted by future versions of G++; 4514 instead a declaration of "f" will need to be present outside of the 4515 scope of "S". 4516 * Covariant returns are implemented for all but varadic functions 4517 that require an adjustment. 4518 * When -pedantic is used, G++ now issues errors about spurious 4519 semicolons. For example, 4520 namespace N {}; // Invalid semicolon. 4521 void f() {}; // Invalid semicolon. 4522 * G++ no longer accepts attributes for a declarator after the 4523 initializer associated with that declarator. For example, 4524 X x(1) __attribute__((...)); 4525 is no longer accepted. Instead, use: 4526 X x __attribute__((...)) (1); 4527 * Inside the scope of a template class, the name of the class itself 4528 can be treated as either a class or a template. So GCC used to 4529 accept the class name as argument of type template, and template 4530 template parameter. However this is not C++ standard compliant. Now 4531 the name is not treated as a valid template template argument 4532 unless you qualify the name by its scope. For example, the code 4533 below no longer compiles. 4534 template <template <class> class TT> class X {}; 4535 template <class T> class Y { 4536 X<Y> x; // Invalid, Y is always a type template parameter. 4537 }; 4538 The valid code for the above example is 4539 X< ::Y> x; // Valid. 4540 (Notice the space between < and : to prevent GCC to interpret this 4541 as a digraph for [.) 4542 * Friend declarations that refer to template specializations are 4543 rejected if the template has not already been declared. For 4544 example, 4545 template <typename T> 4546 class C { 4547 friend void f<> (C&); 4548 }; 4549 is rejected. You must first declare f as a template, 4550 template <typename T> 4551 void f(T); 4552 * In case of friend declarations, every name used in the friend 4553 declaration must be accessible at the point of that declaration. 4554 Previous versions of G++ used to be less strict about this and 4555 allowed friend declarations for private class members, for example. 4556 See the ISO C++ Standard Committee's [11]defect report #209 for 4557 details. 4558 * Declaration of member functions of class templates as friends are 4559 supported. For example, 4560 template <typename T> struct A { 4561 void f(); 4562 }; 4563 class C { 4564 template <typename T> friend void A<T>::f(); 4565 }; 4566 * You must use template <> to introduce template specializations, as 4567 required by the standard. For example, 4568 template <typename T> 4569 struct S; 4570 4571 struct S<int> { }; 4572 is rejected. You must write, 4573 template <> struct S<int> {}; 4574 * G++ used to accept code like this, 4575 struct S { 4576 int h(); 4577 void f(int i = g()); 4578 int g(int i = h()); 4579 }; 4580 This behavior is not mandated by the standard. Now G++ issues an 4581 error about this code. To avoid the error, you must move the 4582 declaration of g before the declaration of f. The default arguments 4583 for g must be visible at the point where it is called. 4584 * The C++ ABI Section 3.3.3 specifications for the array construction 4585 routines __cxa_vec_new2 and __cxa_vec_new3 were changed to return 4586 NULL when the allocator argument returns NULL. These changes are 4587 incorporated into the libstdc++ runtime library. 4588 * Using a name introduced by a typedef in a friend declaration or in 4589 an explicit instantiation is now rejected, as specified by the ISO 4590 C++ standard. 4591 class A; 4592 typedef A B; 4593 class C { 4594 friend class B; // error, no typedef name here 4595 friend B; // error, friend always needs class/struct/enum 4596 friend class A; // OK 4597 }; 4598 4599 template <int> class Q {}; 4600 typedef Q<0> R; 4601 template class R; // error, no typedef name here 4602 template class Q<0>; // OK 4603 * When allocating an array with a new expression, GCC used to allow 4604 parentheses around the type name. This is actually ill-formed and 4605 it is now rejected: 4606 int* a = new (int)[10]; // error, not accepted anymore 4607 int* a = new int[10]; // OK 4608 * When binding an rvalue of class type to a reference, the copy 4609 constructor of the class must be accessible. For instance, consider 4610 the following code: 4611 class A 4612 { 4613 public: 4614 A(); 4615 4616 private: 4617 A(const A&); // private copy ctor 4618 }; 4619 4620 A makeA(void); 4621 void foo(const A&); 4622 4623 void bar(void) 4624 { 4625 foo(A()); // error, copy ctor is not accessible 4626 foo(makeA()); // error, copy ctor is not accessible 4627 4628 A a1; 4629 foo(a1); // OK, a1 is a lvalue 4630 } 4631 This might be surprising at first sight, especially since most 4632 popular compilers do not correctly implement this rule ([12]further 4633 details). 4634 * When forming a pointer to member or a pointer to member function, 4635 access checks for class visibility (public, protected, private) are 4636 now performed using the qualifying scope of the name itself. This 4637 is better explained with an example: 4638 class A 4639 { 4640 public: 4641 void pub_func(); 4642 protected: 4643 void prot_func(); 4644 private: 4645 void priv_func(); 4646 }; 4647 4648 class B : public A 4649 { 4650 public: 4651 void foo() 4652 { 4653 &A::pub_func; // OK, pub_func is accessible through A 4654 &A::prot_func; // error, cannot access prot_func through A 4655 &A::priv_func; // error, cannot access priv_func through A 4656 4657 &B::pub_func; // OK, pub_func is accessible through B 4658 &B::prot_func; // OK, can access prot_func through B (within B) 4659 &B::priv_func; // error, cannot access priv_func through B 4660 } 4661 }; 4662 4663 Runtime Library (libstdc++) 4664 4665 * Optimization work: 4666 + Streamlined streambuf, filebuf, separate synched with C 4667 Standard I/O streambuf. 4668 + All formatted I/O now uses cached locale information. 4669 + STL optimizations (memory/speed for list, red-black trees as 4670 used by sets and maps). 4671 + More use of GCC builtins. 4672 + String optimizations (avoid contention on 4673 increment/decrement-and-test of the reference count in the 4674 empty-string object, constructor from input_iterators 4675 speedup). 4676 * Static linkage size reductions. 4677 * Large File Support (files larger than 2 GB on 32-bit systems). 4678 * Wide character and variable encoding filebuf work (UTF-8, Unicode). 4679 * Generic character traits. 4680 * Also support wchar_t specializations on Mac OS 10.3.x, FreeBSD 5.x, 4681 Solaris 2.7 and above, AIX 5.x, Irix 6.5. 4682 * The allocator class is now standard-conformant, and two additional 4683 extension allocators have been added, mt_alloc and 4684 bitmap_allocator. 4685 * PCH support: -include bits/stdc++.h (2x compile speedup). 4686 * Rewrote __cxa_demangle with support for C++ style allocators. 4687 * New debug modes for STL containers and iterators. 4688 * Testsuite rewrite: five times as many tests, plus increasingly 4689 sophisticated tests, including I/O, MT, multi-locale, wide and 4690 narrow characters. 4691 * Use current versions of GNU "autotools" for build/configuration. 4692 4693 Objective-C 4694 4695 * The Objective-C front end has been updated to include the numerous 4696 bug fixes and enhancements previously available only in Apple's 4697 version of GCC. These include: 4698 + Structured exception (@try... @catch... @finally, @throw) and 4699 synchronization (@synchronized) support. These are accessible 4700 via the -fobjc-exceptions switch; as of this writing, they may 4701 only be used in conjunction with -fnext-runtime on Mac OS X 4702 10.3 and later. See [13]Options Controlling Objective-C 4703 Dialect for more information. 4704 + An overhaul of @encode logic. The C99 _Bool and C++ bool type 4705 may now be encoded as 'B'. In addition, the back-end/codegen 4706 dependencies have been removed. 4707 + An overhaul of message dispatch construction, ensuring that 4708 the various receiver types (and casts thereof) are handled 4709 properly, and that correct diagnostics are issued. 4710 + Support for "Zero-Link" (-fzero-link) and "Fix-and-Continue" 4711 (-freplace-objc-classes) debugging modes, currently available 4712 on Mac OS X 10.3 and later. See [14]Options Controlling 4713 Objective-C Dialect for more information. 4714 + Access to optimized runtime entry points (-fno-nil-receivers ) 4715 on the assumption that message receivers are never nil. This 4716 is currently available on Mac OS X 10.3 and later. See 4717 [15]Options Controlling Objective-C Dialect for more 4718 information. 4719 4720 Java 4721 4722 * Compiling a .jar file will now cause non-.class entries to be 4723 automatically compiled as resources. 4724 * libgcj has been ported to Darwin. 4725 * Jeff Sturm has adapted Jan Hubicka's call graph optimization code 4726 to gcj. 4727 * libgcj has a new gcjlib URL type; this lets URLClassLoader load 4728 code from shared libraries. 4729 * libgcj has been much more completely merged with [16]GNU Classpath. 4730 * Class loading is now much more correct; in particular the caller's 4731 class loader is now used when that is required. 4732 * [17]Eclipse 2.x will run out of the box using gij. 4733 * Parts of java.nio have been implemented. Direct and indirect 4734 buffers work, as do fundamental file and socket operations. 4735 * java.awt has been improved, though it is still not ready for 4736 general use. 4737 * The HTTP protocol handler now uses HTTP/1.1 and can handle the POST 4738 method. 4739 * The MinGW port has matured. Enhancements include socket timeout 4740 support, thread interruption, improved Runtime.exec() handling and 4741 support for accented characters in filenames. 4742 4743 Fortran 4744 4745 * Fortran improvements are listed in the [18]Fortran documentation. 4746 4747 New Targets and Target Specific Improvements 4748 4749 Alpha 4750 4751 * Several [19]built-in functions have been added such as 4752 __builtin_alpha_zap to allow utilizing the more obscure 4753 instructions of the CPU. 4754 * Parameter passing of complex arguments has changed to match the 4755 [20]ABI. This change is incompatible with previous GCC versions, 4756 but does fix compatibility with the Tru64 compiler and several 4757 corner cases where GCC was incompatible with itself. 4758 4759 ARM 4760 4761 * Nicolas Pitre has contributed his hand-coded floating-point support 4762 code for ARM. It is both significantly smaller and faster than the 4763 existing C-based implementation, even when building applications 4764 for Thumb. The arm-elf configuration has been converted to use the 4765 new code. 4766 * Support for the Intel's iWMMXt architecture, a second generation 4767 XScale processor, has been added. Enabled at run time with the 4768 -mcpu=iwmmxt command line switch. 4769 * A new ARM target has been added: arm-wince-pe. This is similar to 4770 the arm-pe target, but it defaults to using the APCS32 ABI. 4771 * The existing ARM pipeline description has been converted to the use 4772 the [21]DFA processor pipeline model. There is not much change in 4773 code performance, but the description is now [22]easier to 4774 understand. 4775 * Support for the Cirrus EP9312 Maverick floating point co-processor 4776 added. Enabled at run time with the -mcpu=ep9312 command line 4777 switch. Note however that the multilibs to support this chip are 4778 currently disabled in gcc/config/arm/t-arm-elf, so if you want to 4779 enable their production you will have to uncomment the entries in 4780 that file. 4781 4782 H8/300 4783 4784 * Support for long long has been added. 4785 * Support for saveall attribute has been added. 4786 * Pavel Pisa contributed hand-written 32-bit-by-32-bit division code 4787 for H8/300H and H8S, which is much faster than the previous 4788 implementation. 4789 * A lot of small performance improvements. 4790 4791 IA-32/AMD64 (x86-64) 4792 4793 * Tuning for K8 (AMD Opteron/Athlon64) core is available via 4794 -march=k8 and -mcpu=k8. 4795 * Scalar SSE code generation carefully avoids reformatting penalties, 4796 hidden dependencies and minimizes the number of uops generated on 4797 both Intel and AMD CPUs. 4798 * Vector MMX and SSE operands are now passed in registers to improve 4799 performance and match the argument passing convention used by the 4800 Intel C++ Compiler. As a result it is not possible to call 4801 functions accepting vector arguments compiled by older GCC version. 4802 * Conditional jump elimination is now more aggressive on modern CPUs. 4803 * The Athlon ports has been converted to use the DFA processor 4804 pipeline description. 4805 * Optimization of indirect tail calls is now possible in a similar 4806 fashion as direct sibcall optimization. 4807 * Further small performance improvements. 4808 * -m128bit-long-double is now less buggy. 4809 * __float128 support in 64-bit compilation. 4810 * Support for data structures exceeding 2GB in 64-bit mode. 4811 * -mcpu has been renamed to -mtune. 4812 4813 IA-64 4814 4815 * Tuning code for the Itanium 2 processor has been added. The 4816 generation of code tuned for Itanium 2 (option -mtune=itanium2) is 4817 enabled by default now. To generate code tuned for Itanium 1 the 4818 option -mtune=itanium1 should be used. 4819 * [23]DFA processor pipeline descriptions for the IA-64 processors 4820 have been added. This resulted in about 3% improvement on the 4821 SPECInt2000 benchmark for Itanium 2. 4822 * Instruction bundling for the IA-64 processors has been rewritten 4823 using the DFA pipeline hazard recognizer. It resulted in about 60% 4824 compiler speedup on the SPECInt2000 C programs. 4825 4826 M32R 4827 4828 * Support for the M32R/2 processor has been added by Renesas. 4829 * Support for an M32R Linux target and PIC code generation has been 4830 added by Renesas. 4831 4832 M68000 4833 4834 * Bernardo Innocenti (Develer S.r.l.) has contributed the 4835 m68k-uclinux target, based on former work done by Paul Dale 4836 (SnapGear Inc.). Code generation for the ColdFire processors family 4837 has been enhanced and extended to support the MCF 53xx and MCF 54xx 4838 cores, integrating former work done by Peter Barada (Motorola). 4839 4840 MIPS 4841 4842 Processor-specific changes 4843 4844 * Support for the RM7000 and RM9000 processors has been added. It can 4845 be selected using the -march compiler option and should work with 4846 any MIPS I (mips-*) or MIPS III (mips64-*) configuration. 4847 * Support for revision 2 of the MIPS32 ISA has been added. It can be 4848 selected with the command-line option -march=mips32r2. 4849 * There is a new option, -mfix-sb1, to work around certain SB-1 4850 errata. 4851 4852 Configuration 4853 4854 * It is possible to customize GCC using the following configure-time 4855 options: 4856 + --with-arch, which specifies the default value of the -march 4857 option. 4858 + --with-tune, which specifies the default value of the -mtune 4859 option. 4860 + --with-abi, which specifies the default ABI. 4861 + --with-float=soft, which tells GCC to use software floating 4862 point by default. 4863 + --with-float=hard, which tells GCC to use hardware floating 4864 point by default. 4865 * A 64-bit GNU/Linux port has been added. The associated 4866 configurations are mips64-linux-gnu and mips64el-linux-gnu. 4867 * The 32-bit GNU/Linux port now supports Java. 4868 * The IRIX 6 configuration now supports the o32 ABI and will build 4869 o32 multilibs by default. This support is compatible with both 4870 binutils and the SGI tools, but note that several features, 4871 including debugging information and DWARF2 exception handling, are 4872 only available when using the GNU assembler. Use of the GNU 4873 assembler and linker (version 2.15 or above) is strongly 4874 recommended. 4875 * The IRIX 6 configuration now supports 128-bit long doubles. 4876 * There are two new RTEMS-specific configurations, mips-rtems and 4877 mipsel-rtems. 4878 * There are two new *-elf configurations, mipsisa32r2-elf and 4879 mipsisa32r2el-elf. 4880 4881 General 4882 4883 * Several [24]ABI bugs have been fixed. Unfortunately, these changes 4884 will break binary compatibility with earlier releases. 4885 * GCC can now use explicit relocation operators when generating 4886 -mabicalls code. This behavior is controlled by -mexplicit-relocs 4887 and can have several performance benefits. For example: 4888 + It allows for more optimization of GOT accesses, including 4889 better scheduling and redundancy elimination. 4890 + It allows sibling calls to be implemented as jumps. 4891 + n32 and n64 leaf functions can use a call-clobbered global 4892 pointer instead of $28. 4893 + The code to set up $gp can be removed from functions that 4894 don't need it. 4895 * A new option, -mxgot, allows the GOT to be bigger than 64k. This 4896 option is equivalent to the assembler's -xgot option and should be 4897 used instead of -Wa,-xgot. 4898 * Frame pointer elimination is now supported when generating 64-bit 4899 MIPS16 code. 4900 * Inline block moves have been optimized to take more account of 4901 alignment information. 4902 * Many internal changes have been made to the MIPS port, mostly aimed 4903 at reducing the reliance on assembler macros. 4904 4905 PowerPC 4906 4907 * GCC 3.4 releases have a number of fixes for PowerPC and PowerPC64 4908 [25]ABI incompatibilities regarding the way parameters are passed 4909 during functions calls. These changes may result in incompatibility 4910 between code compiled with GCC 3.3 and GCC 3.4. 4911 4912 PowerPC Darwin 4913 4914 * Support for shared/dylib gcc libraries has been added. It is 4915 enabled by default on powerpc-apple-darwin7.0.0 and up. 4916 * Libgcj is enabled by default. On systems older than 4917 powerpc-apple-darwin7.0.0 you need to install dlcompat. 4918 * 128-bit IBM extended precision format support added for long 4919 double. 4920 4921 PowerPC64 GNU/Linux 4922 4923 * By default, PowerPC64 GNU/Linux now uses natural alignment of 4924 structure elements. The old four byte alignment for double, with 4925 special rules for a struct starting with a double, can be chosen 4926 with -malign-power. This change may result in incompatibility 4927 between code compiled with GCC 3.3 and GCC 3.4. 4928 * -mabi=altivec is now the default rather than -mabi=no-altivec. 4929 * 128-bit IBM extended precision format support added for long 4930 double. 4931 4932 S/390 and zSeries 4933 4934 * New command-line options allow to specify the intended execution 4935 environment for generated code: 4936 + -mesa/-mzarch allows to specify whether to generate code 4937 running in ESA/390 mode or in z/Architecture mode (this is 4938 applicable to 31-bit code only). 4939 + -march allows to specify a minimum processor architecture 4940 level (g5, g6, z900, or z990). 4941 + -mtune allows to specify which processor to tune for. 4942 * It is possible to customize GCC using the following configure-time 4943 options: 4944 + --with-mode, which specifies whether to default to assuming 4945 ESA/390 or z/Architecture mode. 4946 + --with-arch, which specifies the default value of the -march 4947 option. 4948 + --with-tune, which specifies the default value of the -mtune 4949 option. 4950 * Support for the z990 processor has been added, and can be selected 4951 using -march=z990 or -mtune=z990. This includes instruction 4952 scheduling tuned for the superscalar instruction pipeline of the 4953 z990 processor as well as support for all new instructions provided 4954 by the long-displacement facility. 4955 * Support to generate 31-bit code optimized for zSeries processors 4956 (running in ESA/390 or in z/Architecture mode) has been added. This 4957 can be selected using -march=z900 and -mzarch respectively. 4958 * Instruction scheduling for the z900 and z990 processors now uses 4959 the DFA pipeline hazard recognizer. 4960 * GCC no longer generates code to maintain a stack backchain, 4961 previously used to generate stack backtraces for debugging 4962 purposes. As replacement that does not incur runtime overhead, 4963 DWARF-2 call frame information is provided by GCC; this is 4964 supported by GDB 6.1. The old behavior can be restored using the 4965 -mbackchain option. 4966 * The stack frame size of functions may now exceed 2 GB in 64-bit 4967 code. 4968 * A port for the 64-bit IBM TPF operating system has been added; the 4969 configuration is s390x-ibm-tpf. This configuration is supported as 4970 cross-compilation target only. 4971 * Various changes to improve the generated code have been 4972 implemented, including: 4973 + GCC now uses the MULTIPLY AND ADD and MULTIPLY AND SUBTRACT 4974 instructions to significantly speed up many floating-point 4975 applications. 4976 + GCC now uses the ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL 4977 WITH BORROW instructions to speed up long long arithmetic. 4978 + GCC now uses the SEARCH STRING instruction to implement 4979 strlen(). 4980 + In many cases, function call overhead for 31-bit code has been 4981 reduced by placing the literal pool after the function code 4982 instead of after the function prolog. 4983 + Register 14 is no longer reserved in 64-bit code. 4984 + Handling of global register variables has been improved. 4985 4986 SPARC 4987 4988 * The option -mflat is deprecated. 4989 * Support for large (> 2GB) frames has been added to the 64-bit port. 4990 * Several [26]ABI bugs have been fixed. Unfortunately, these changes 4991 will break binary compatibility with earlier releases. 4992 * The default debugging format has been switched from STABS to 4993 DWARF-2 for 32-bit code on Solaris 7 and later. DWARF-2 is already 4994 the default debugging format for 64-bit code on Solaris. 4995 4996 SuperH 4997 4998 * Support for the SH2E processor has been added. Enabled at run time 4999 with the -m2e command line switch, or at configure time by 5000 specifying sh2e as the machine part of the target triple. 5001 5002 V850 5003 5004 * Support for the Mitsubishi V850E1 processor has been added. This is 5005 a variant of the V850E processor with some additional debugging 5006 instructions. 5007 5008 Xtensa 5009 5010 * Several ABI bugs have been fixed. Unfortunately, these changes 5011 break binary compatibility with earlier releases. 5012 + For big-endian processors, the padding of aggregate return 5013 values larger than a word has changed. If the size of an 5014 aggregate return value is not a multiple of 32 bits, previous 5015 versions of GCC inserted padding in the most-significant bytes 5016 of the first return value register. Aggregates larger than a 5017 word are now padded in the least-significant bytes of the last 5018 return value register used. Aggregates smaller than a word are 5019 still padded in the most-significant bytes. The return value 5020 padding has not changed for little-endian processors. 5021 + Function arguments with 16-byte alignment are now properly 5022 aligned. 5023 + The implementation of the va_list type has changed. A va_list 5024 value created by va_start from a previous release cannot be 5025 used with va_arg from this release, or vice versa. 5026 * More processor configuration options for Xtensa processors are 5027 supported: 5028 + the ABS instruction is now optional; 5029 + the ADDX* and SUBX* instructions are now optional; 5030 + an experimental CONST16 instruction can be used to synthesize 5031 constants instead of loading them from constant pools. 5032 These and other Xtensa processor configuration options can no 5033 longer be enabled or disabled by command-line options; the 5034 processor configuration must be specified by the xtensa-config.h 5035 header file when building GCC. Additionally, the 5036 -mno-serialize-volatile option is no longer supported. 5037 5038 Obsolete Systems 5039 5040 Support for a number of older systems has been declared obsolete in GCC 5041 3.4. Unless there is activity to revive them, the next release of GCC 5042 will have their sources permanently removed. 5043 5044 All configurations of the following processor architectures have been 5045 declared obsolete: 5046 * Mitsubishi D30V, d30v-* 5047 * AT&T DSP1600 and DSP1610, dsp16xx-* 5048 * Intel 80960, i960 5049 5050 Also, some individual systems have been obsoleted: 5051 * ARM Family 5052 + Support for generating code for operation in APCS/26 mode 5053 (-mapcs-26). 5054 * IBM ESA/390 5055 + "Bigfoot" port, i370-*. (The other port, s390-*, is actively 5056 maintained and supported.) 5057 * Intel 386 family 5058 + MOSS, i?86-moss-msdos and i?86-*-moss* 5059 + NCR 3000 running System V r.4, i?86-ncr-sysv4* 5060 + FreeBSD with a.out object format, i?86-*-freebsd*aout* and 5061 i?86-*-freebsd2* 5062 + GNU/Linux with a.out object format, i?86-linux*aout* 5063 + GNU/Linux with libc5, a.k.a. glibc1, i?86-linux*libc1* 5064 + Interix versions before Interix 3, i?86-*-interix 5065 + Mach microkernel, i?86-mach* 5066 + SCO UnixWare with UDK, i?86-*-udk* 5067 + Generic System V releases 1, 2, and 3, i?86-*-sysv[123]* 5068 + VSTa microkernel, i386-*-vsta 5069 * Motorola M68000 family 5070 + HPUX, m68k-hp-hpux* and m68000-hp-hpux* 5071 + NetBSD with a.out object format (before NetBSD 1.4), 5072 m68k-*-*-netbsd* except m68k-*-*-netbsdelf* 5073 + Generic System V r.4, m68k-*-sysv4* 5074 * VAX 5075 + Generic VAX, vax-*-* (This is generic VAX only; we have not 5076 obsoleted any VAX triples for specific operating systems.) 5077 5078 Documentation improvements 5079 5080 Other significant improvements 5081 5082 * The build system has undergone several significant cleanups. 5083 Subdirectories will only be configured if they are being built, and 5084 all subdirectory configures are run from the make command. The top 5085 level has been autoconfiscated. 5086 * Building GCC no longer writes to its source directory. This should 5087 help those wishing to share a read-only source directory over NFS 5088 or build from a CD. The exceptions to this feature are if you 5089 configure with either --enable-maintainer-mode or 5090 --enable-generated-files-in-srcdir. 5091 * The -W warning option has been renamed to -Wextra, which is more 5092 easily understood. The older spelling will be retained for 5093 backwards compatibility. 5094 * Substantial improvements in compile time have been made, 5095 particularly for non-optimizing compilations. 5096 __________________________________________________________________ 5097 5098 GCC 3.4.0 5099 5100 Bug Fixes 5101 5102 A vast number of bugs have been fixed in 3.4.0, too many to publish a 5103 complete list here. [27]Follow this link to query the Bugzilla database 5104 for the list of over 900 bugs fixed in 3.4.0. This is the list of all 5105 bugs marked as resolved and fixed in 3.4.0 that are not flagged as 3.4 5106 regressions. 5107 __________________________________________________________________ 5108 5109 GCC 3.4.1 5110 5111 Bug Fixes 5112 5113 This section lists the problem reports (PRs) from GCC's bug tracking 5114 system that are known to be fixed in the 3.4.1 release. This list might 5115 not be complete (that is, it is possible that some PRs that have been 5116 fixed are not listed here). 5117 5118 Bootstrap failures 5119 5120 * [28]10129 Ada bootstrap fails on PPC-Darwin - invalid assembler 5121 emitted - PIC related 5122 * [29]14576 [ARM] ICE in libiberty when building gcc-3.4 for arm-elf 5123 * [30]14760 A bug in configure.in prevents using both 5124 --program-suffix and --program-prefix 5125 * [31]14671 [hppa64] bootstrap fails: ICE in 5126 save_call_clobbered_regs, in caller_save.c 5127 * [32]15093 [alpha][Java] make bootstrap fails to configure libffi on 5128 Alpha 5129 * [33]15178 Solaris 9/x86 fails linking after stage 3 5130 5131 Multi-platform internal compiler errors (ICEs) 5132 5133 * [34]12753 (preprocessor) Memory corruption in preprocessor on bad 5134 input 5135 * [35]13985 ICE in gcc.c-torture/compile/930621-1.c 5136 * [36]14810 (c++) tree check failures with invalid code involving 5137 templates 5138 * [37]14883 (c++) ICE on invalid code, in cp_parser_lookup_name, in 5139 cp/parser.c 5140 * [38]15044 (c++) ICE on syntax error, template header 5141 * [39]15057 (c++) Compiling of conditional value throw constructs 5142 cause a segmentation violation 5143 * [40]15064 (c++) typeid of template parameter gives ICE 5144 * [41]15142 (c++) ICE when passing a string where a char* is expected 5145 in a throw statement 5146 * [42]15159 ICE in rtl_verify_flow_info_1 5147 * [43]15165 (c++) ICE in instantiate_template 5148 * [44]15193 Unary minus using pointer to V4SF vector causes 5149 -fforce-mem to exhaust all memory 5150 * [45]15209 (c++) Runs out of memory with packed structs 5151 * [46]15227 (c++) Trouble with invalid function definition 5152 * [47]15285 (c++) instantiate_type ICE when forming pointer to 5153 template function 5154 * [48]15299 (c++) ICE in resolve_overloaded_unification 5155 * [49]15329 (c++) ICE on constructor of member template 5156 * [50]15550 ICE in extract_insn, in recog.c 5157 * [51]15554 (c++) ICE in tsubst_copy, in cp/pt.c 5158 * [52]15640 (c++) ICE on invalid code in arg_assoc, in 5159 cp/name-lookup.c 5160 * [53]15666 [unit-at-a-time] Gcc abort on valid code 5161 * [54]15696 (c++) ICE with bad pointer-to-member code 5162 * [55]15701 (c++) ICE with friends and template template parameter 5163 * [56]15761 ICE in do_SUBST, in combine.c 5164 * [57]15829 (c++) ICE on Botan-1.3.13 due to -funroll-loops 5165 5166 Ada 5167 5168 * [58]14538 All RTEMS targets broken for gnat 5169 5170 C front end 5171 5172 * [59]12391 missing warning about assigning to an incomplete type 5173 * [60]14649 atan(1.0) should not be a constant expression 5174 * [61]15004 [unit-at-a-time] no warning for unused paramater in 5175 static function 5176 * [62]15749 --pedantic-errors behaves differently from --pedantic 5177 with C-compiler on GNU/Linux 5178 5179 C++ compiler and library 5180 5181 * [63]10646 non-const reference is incorrectly matched in a "const T" 5182 partial specialization 5183 * [64]12077 wcin.rdbuf()->in_avail() return value too high 5184 * [65]13598 enc_filebuf doesn't work 5185 * [66]14211 const_cast returns lvalue but should be rvalue 5186 * [67]14220 num_put::do_put() undesired float/double behavior 5187 * [68]14245 problem with user-defined allocators in std::basic_string 5188 * [69]14340 libstdc++ Debug mode: failure to convert iterator to 5189 const_iterator 5190 * [70]14600 __gnu_cxx::stdio_sync_filebuf should expose internal 5191 FILE* 5192 * [71]14668 no warning anymore for reevaluation of declaration 5193 * [72]14775 LFS (large file support) tests missing 5194 * [73]14821 Duplicate namespace alias declaration should not conflict 5195 * [74]14930 Friend declaration ignored 5196 * [75]14932 cannot use offsetof to get offsets of array elements in 5197 g++ 3.4.0 5198 * [76]14950 [non unit-at-a-time] always_inline does not mix with 5199 templates and -O0 5200 * [77]14962 g++ ignores #pragma redefine_extname 5201 * [78]14975 Segfault on low-level write error during imbue 5202 * [79]15002 Linewise stream input is unusably slow (std::string slow) 5203 * [80]15025 compiler accepts redeclaration of template as 5204 non-template 5205 * [81]15046 [arm] Math functions misdetected by cross configuration 5206 * [82]15069 a bit test on a variable of enum type is miscompiled 5207 * [83]15074 g++ -lsupc++ still links against libstdc++ 5208 * [84]15083 spurious "statement has no effect" warning 5209 * [85]15096 parse error with templates and pointer to const member 5210 * [86]15287 combination of operator[] and operator .* fails in 5211 templates 5212 * [87]15317 __attribute__ unused in first parameter of constructor 5213 gives error 5214 * [88]15337 sizeof on incomplete type diagnostic 5215 * [89]15361 bitset<>::_Find_next fails 5216 * [90]15412 _GLIBCXX_ symbols symbols defined and used in different 5217 namespaces 5218 * [91]15427 valid code results in incomplete type error 5219 * [92]15471 Incorrect member pointer offsets in anonymous 5220 structs/unions 5221 * [93]15503 nested template problem 5222 * [94]15507 compiler hangs while laying out union 5223 * [95]15542 operator & and template definitions 5224 * [96]15565 SLES9: leading + sign for unsigned int with showpos 5225 * [97]15625 friend defined inside a template fails to find static 5226 function 5227 * [98]15629 Function templates, overloads, and friend name injection 5228 * [99]15742 'noreturn' attribute ignored in method of template 5229 functions. 5230 * [100]15775 Allocator::pointer consistently ignored 5231 * [101]15821 Duplicate namespace alias within namespace rejected 5232 * [102]15862 'enum yn' fails (confict with undeclared builtin) 5233 * [103]15875 rejects pointer to member in template 5234 * [104]15877 valid code using templates and anonymous enums is 5235 rejected 5236 * [105]15947 Puzzling error message for wrong destructor declaration 5237 in template class 5238 * [106]16020 cannot copy __gnu_debug::bitset 5239 * [107]16154 input iterator concept too restrictive 5240 * [108]16174 deducing top-level consts 5241 5242 Java 5243 5244 * [109]14315 Java compiler is not parallel make safe 5245 5246 Fortran 5247 5248 * [110]15151 [g77] incorrect logical i/o in 64-bit mode 5249 5250 Objective-C 5251 5252 * [111]7993 private variables cannot be shadowed in subclasses 5253 5254 Optimization bugs 5255 5256 * [112]15228 useless copies of floating point operands 5257 * [113]15345 [non-unit-at-a-time] unreferenced nested inline 5258 functions not optimized away 5259 * [114]15945 Incorrect floating point optimization 5260 * [115]15526 ftrapv aborts on 0 * (-1) 5261 * [116]14690 Miscompiled POOMA tests 5262 * [117]15112 GCC generates code to write to unchanging memory 5263 5264 Preprocessor 5265 5266 * [118]15067 Minor glitch in the source of cpp 5267 5268 Main driver program bugs 5269 5270 * [119]1963 collect2 interprets -oldstyle_liblookup as -o 5271 ldstyle_liblookup 5272 5273 x86-specific (Intel/AMD) 5274 5275 * [120]15717 Error: can't resolve `L0' {*ABS* section} - `xx' {*UND* 5276 section} 5277 5278 HPPA-specific 5279 5280 * [121]14782 GCC produces an unaligned data access at -O2 5281 * [122]14828 FAIL: gcc.c-torture/execute/20030408-1.c execution, -O2 5282 * [123]15202 ICE in reload_cse_simplify_operands, in postreload.c 5283 5284 IA64-specific 5285 5286 * [124]14610 __float80 constants incorrectly emitted 5287 * [125]14813 init_array sections are initialized in the wrong order 5288 * [126]14857 GCC segfault on duplicated asm statement 5289 * [127]15598 Gcc 3.4 ICE on valid code 5290 * [128]15653 Gcc 3.4 ICE on valid code 5291 5292 MIPS-specific 5293 5294 * [129]15189 wrong filling of delay slot with -march=mips1 -G0 5295 -mno-split-addresses -mno-explicit-relocs 5296 * [130]15331 Assembler error building gnatlib on IRIX 6.5 with GNU as 5297 2.14.91 5298 * [131]16144 Bogus reference to __divdf3 when -O1 5299 * [132]16176 Miscompilation of unaligned data in MIPS backend 5300 5301 PowerPC-specific 5302 5303 * [133]11591 ICE in gcc.dg/altivec-5.c 5304 * [134]12028 powerpc-eabispe produces bad sCOND operation 5305 * [135]14478 rs6000 geu/ltu patterns generate incorrect code 5306 * [136]14567 long double and va_arg complex args 5307 * [137]14715 Altivec stack layout may overlap gpr save with stack 5308 temps 5309 * [138]14902 (libstdc++) Stream checking functions fail when -pthread 5310 option is used. 5311 * [139]14924 Compiler ICE on valid code 5312 * [140]14960 -maltivec affects vector return with -mabi=no-altivec 5313 * [141]15106 vector varargs failure passing from altivec to 5314 non-altivec code for -m32 5315 * [142]16026 ICE in function.c:4804, assign_parms, when -mpowerpc64 & 5316 half-word operation 5317 * [143]15191 -maltivec -mabi=no-altivec results in mis-aligned lvx 5318 and stvx 5319 * [144]15662 Segmentation fault when an exception is thrown - even if 5320 try and catch are specified 5321 5322 s390-specific 5323 5324 * [145]15054 Bad code due to overlapping stack temporaries 5325 5326 SPARC-specific 5327 5328 * [146]15783 ICE with union assignment in 64-bit mode 5329 * [147]15626 GCC 3.4 emits "ld: warning: relocation error: 5330 R_SPARC_UA32" 5331 5332 x86-64-specific 5333 5334 * [148]14326 boehm-gc hardcodes to 3DNow! prefetch for x86_64 5335 * [149]14723 Backported -march=nocona from mainline 5336 * [150]15290 __float128 failed to pass to function properly 5337 5338 Cygwin/Mingw32-specific 5339 5340 * [151]15250 Option -mms-bitfields support on GCC 3.4 is not 5341 conformant to MS layout 5342 * [152]15551 -mtune=pentium4 -O2 with sjlj EH breaks stack probe 5343 worker on windows32 targets 5344 5345 Bugs specific to embedded processors 5346 5347 * [153]8309 [m68k] -m5200 produces erroneous SImode set of short 5348 varaible on stack 5349 * [154]13250 [SH] Gcc code for rotation clobbers the register, but 5350 gcc continues to use the register as if it was not clobbered 5351 * [155]13803 [coldfire] movqi operand constraints too restrictivefor 5352 TARGET_COLDFIRE 5353 * [156]14093 [SH] ICE for code when using -mhitachi option in SH 5354 * [157]14457 [m6811hc] ICE with simple c++ source 5355 * [158]14542 [m6811hc] ICE on simple source 5356 * [159]15100 [SH] cc1plus got hang-up on 5357 libstdc++-v3/testsuite/abi_check.cc 5358 * [160]15296 [CRIS] Delayed branch scheduling causing invalid code on 5359 cris-* 5360 * [161]15396 [SH] ICE with -O2 -fPIC 5361 * [162]15782 [coldfire] m68k_output_mi_thunk emits wrong code for 5362 ColdFire 5363 5364 Testsuite problems (compiler not affected) 5365 5366 * [163]11610 libstdc++ testcases 27_io/* don't work properly remotely 5367 * [164]15488 (libstdc++) possibly insufficient file permissions for 5368 executing test suite 5369 * [165]15489 (libstdc++) testsuite_files determined incorrectly 5370 5371 Documentation bugs 5372 5373 * [166]13928 (libstdc++) no whatis info in some man pages generated 5374 by doxygen 5375 * [167]14150 Ada documentation out of date 5376 * [168]14949 (c++) Need to document method visibility changes 5377 * [169]15123 libstdc++-doc: Allocators.3 manpage is empty 5378 __________________________________________________________________ 5379 5380 GCC 3.4.2 5381 5382 Bug Fixes 5383 5384 This section lists the problem reports (PRs) from GCC's bug tracking 5385 system that are known to be fixed in the 3.4.2 release. This list might 5386 not be complete (that is, it is possible that some PRs that have been 5387 fixed are not listed here). 5388 5389 Bootstrap failures and issues 5390 5391 * [170]16469 [mips-sgi-irix5.3] bootstrap fails in 5392 libstdc++-v3/testsuite 5393 * [171]16344 [hppa-linux-gnu] libstdc++'s PCH built by 5394 profiledbootstrap does not work with the built compiler 5395 * [172]16842 [Solaris/x86] mkheaders can not find mkheaders.conf 5396 5397 Multi-platform internal compiler errors (ICEs) 5398 5399 * [173]12608 (c++) ICE: expected class 't', have 'x' (error_mark) in 5400 cp_parser_class_specifier, in cp/parser.c 5401 * [174]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c 5402 * [175]15461 (c++) ICE due to NRV and inlining 5403 * [176]15890 (c++) ICE in c_expand_expr, in c-common.c 5404 * [177]16180 ICE: segmentation fault in RTL optimization 5405 * [178]16224 (c++) ICE in write_unscoped_name (template/namespace) 5406 * [179]16408 ICE: in delete_insn, in cfgrtl.c 5407 * [180]16529 (c++) ICE for: namespace-alias shall not be declared as 5408 the name of any other entity 5409 * [181]16698 (c++) ICE with exceptions and declaration of __cxa_throw 5410 * [182]16706 (c++) ICE in finish_member_declaration, in 5411 cp/semantics.c 5412 * [183]16810 (c++) Legal C++ program with cast gives ICE in 5413 build_ptrmemfunc 5414 * [184]16851 (c++) ICE when throwing a comma expression 5415 * [185]16870 (c++) Boost.Spirit causes ICE in tsubst, in cp/pt.c 5416 * [186]16904 (c++) ICE in finish_class_member_access_expr, in 5417 cp/typeck.c 5418 * [187]16905 (c++) ICE (segfault) with exceptions 5419 * [188]16964 (c++) ICE in cp_parser_class_specifier due to 5420 redefinition 5421 * [189]17068 (c++) ICE: tree check: expected class 'd', have 'x' 5422 (identifier_node) in dependent_template_p, in cp/pt.c 5423 5424 Preprocessor bugs 5425 5426 * [190]16366 Preprocessor option -remap causes memory corruption 5427 5428 Optimization 5429 5430 * [191]15345 unreferenced nested inline functions not optimized away 5431 * [192]16590 Incorrect execution when compiling with -O2 5432 * [193]16693 Bitwise AND is lost when used within a cast to an enum 5433 of the same precision 5434 * [194]17078 Jump into if(0) substatement fails 5435 5436 Problems in generated debug information 5437 5438 * [195]13956 incorrect stabs for nested local variables 5439 5440 C front end bugs 5441 5442 * [196]16684 GCC should not warn about redundant redeclarations of 5443 built-ins 5444 5445 C++ compiler and library 5446 5447 * [197]12658 Thread safety problems in locale::global() and 5448 locale::locale() 5449 * [198]13092 g++ accepts invalid pointer-to-member conversion 5450 * [199]15320 Excessive memory consumption 5451 * [200]16246 Incorrect template argument deduction 5452 * [201]16273 Memory exhausted when using nested classes and virtual 5453 functions 5454 * [202]16401 ostringstream in gcc 3.4.x very slow for big data 5455 * [203]16411 undefined reference to 5456 __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char> 5457 >::file() 5458 * [204]16489 G++ incorrectly rejects use of a null constant integral 5459 expression as a null constant pointer 5460 * [205]16618 offsetof fails with constant member 5461 * [206]16637 syntax error reported for valid input code 5462 * [207]16717 __attribute__((constructor)) broken in C++ 5463 * [208]16813 compiler error in DEBUG version of range insertion 5464 std::map::insert 5465 * [209]16853 pointer-to-member initialization from incompatible one 5466 accepted 5467 * [210]16889 ambiguity is not detected 5468 * [211]16959 Segmentation fault in ios_base::sync_with_stdio 5469 5470 Java compiler and library 5471 5472 * [212]7587 direct threaded interpreter not thread-safe 5473 * [213]16473 ServerSocket accept() leaks file descriptors 5474 * [214]16478 Hash synchronization deadlock with finalizers 5475 5476 Alpha-specific 5477 5478 * [215]10695 ICE in dwarf2out_frame_debug_expr, in dwarf2out.c 5479 * [216]16974 could not split insn (ice in final_scan_insn, in 5480 final.c) 5481 5482 x86-specific 5483 5484 * [217]16298 ICE in output_operand 5485 * [218]17113 ICE with SSE2 intrinsics 5486 5487 x86-64 specific 5488 5489 * [219]14697 libstdc++ couldn't find 32bit libgcc_s 5490 5491 MIPS-specific 5492 5493 * [220]15869 [mips64] No NOP after LW (with -mips1 -O0) 5494 * [221]16325 [mips64] value profiling clobbers gp on mips 5495 * [222]16357 [mipsisa64-elf] ICE copying 7 bytes between extern 5496 char[]s 5497 * [223]16380 [mips64] Use of uninitialised register after dbra 5498 conversion 5499 * [224]16407 [mips64] Unaligned access to local variables 5500 * [225]16643 [mips64] verify_local_live_at_start ICE after 5501 crossjumping & cfgcleanup 5502 5503 ARM-specific 5504 5505 * [226]15927 THUMB -O2: strength-reduced iteration variable ends up 5506 off by 1 5507 * [227]15948 THUMB: ICE with non-commutative cbranch 5508 * [228]17019 THUMB: bad switch statement in md code for 5509 addsi3_cbranch_scratch 5510 5511 IA64-specific 5512 5513 * [229]16130 ICE on valid code: in bundling, in config/ia64/ia64.c 5514 (-mtune=merced) 5515 * [230]16142 ICE on valid code: in bundling, in config/ia64/ia64.c 5516 (-mtune=itanium) 5517 * [231]16278 Gcc failed to build Linux kernel with -mtune=merced 5518 * [232]16414 ICE on valid code: typo in comparison of asm_noperands 5519 result 5520 * [233]16445 ICE on valid code: don't count ignored insns 5521 * [234]16490 ICE (segfault) while compiling with -fprofile-use 5522 * [235]16683 ia64 does not honor SUBTARGET_EXTRA_SPECS 5523 5524 PowerPC-specific 5525 5526 * [236]16195 (ppc64): Miscompilation of GCC 3.3.x by 3.4.x 5527 * [237]16239 ICE on ppc64 (mozilla 1.7 compile, -O1 -fno-exceptions 5528 issue) 5529 5530 SPARC-specific 5531 5532 * [238]16199 ICE while compiling apache 2.0.49 5533 * [239]16416 -m64 doesn't imply -mcpu=v9 anymore 5534 * [240]16430 ICE when returning non-C aggregates larger than 16 bytes 5535 5536 Bugs specific to embedded processors 5537 5538 * [241]16379 [m32r] can't output large model function call of memcpy 5539 * [242]17093 [m32r] ICE with -msdata=use -O0 5540 * [243]17119 [m32r] ICE at switch case 0x8000 5541 5542 DJGPP-specific 5543 5544 * [244]15928 libstdc++ in 3.4.x doesn't cross-compile for djgpp 5545 5546 Alpha Tru64-specific 5547 5548 * [245]16210 libstdc++ gratuitously omits "long long" I/O 5549 5550 Testsuite, documentation issues (compiler is not affected): 5551 5552 * [246]15488 (libstdc++) possibly insufficient file permissions for 5553 executing test suite 5554 * [247]16250 ada/doctools runs makeinfo even in release tarball 5555 __________________________________________________________________ 5556 5557 GCC 3.4.3 5558 5559 This is the [248]list of problem reports (PRs) from GCC's bug tracking 5560 system that are known to be fixed in the 3.4.3 release. This list might 5561 not be complete (that is, it is possible that some PRs that have been 5562 fixed are not listed here). 5563 5564 Bootstrap failures 5565 5566 * [249]17369 [ia64] Bootstrap failure with binutils-2.15.90.0.1.1 5567 * [250]17850 [arm-elf] bootstrap failure - libstdc++ uses strtold 5568 when undeclared 5569 5570 Internal compiler errors (ICEs) affecting multiple platforms 5571 5572 * [251]13948 (java) GCJ segmentation fault while compiling GL4Java 5573 .class files 5574 * [252]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c 5575 * [253]16301 (c++) ICE when "strong" attribute is attached to a using 5576 directive 5577 * [254]16566 ICE with flexible arrays 5578 * [255]17023 ICE with nested functions in parameter declaration 5579 * [256]17027 ICE with noreturn function in loop at -O2 5580 * [257]17524 ICE in grokdeclarator, in cp/decl.c 5581 * [258]17826 (c++) ICE in cp_tree_equal 5582 5583 C and optimization bugs 5584 5585 * [259]15526 -ftrapv aborts on 0 * (-1) 5586 * [260]16999 #ident stopped working 5587 * [261]17503 quadratic behaviour in invalid_mode_change_p 5588 * [262]17581 Long long arithmetic fails inside a switch/case 5589 statement when compiled with -O2 5590 * [263]18129 -fwritable-strings doesn't work 5591 5592 C++ compiler and library bugs 5593 5594 * [264]10975 incorrect initial ostringstream::tellp() 5595 * [265]11722 Unbuffered filebuf::sgetn is slow 5596 * [266]14534 Unrecognizing static function as a template parameter 5597 when its return value is also templated 5598 * [267]15172 Copy constructor optimization in aggregate 5599 initialization 5600 * [268]15786 Bad error message for frequently occuring error. 5601 * [269]16162 Rejects valid member-template-definition 5602 * [270]16612 empty basic_strings can't live in shared memory 5603 * [271]16715 std::basic_iostream is instantiated when used, even 5604 though instantiations are already contained in libstdc++ 5605 * [272]16848 code in /ext/demangle.h appears broken 5606 * [273]17132 GCC fails to eliminate function template specialization 5607 when argument deduction fails 5608 * [274]17259 One more _S_leaf incorrectly qualified with _RopeRep:: 5609 in ropeimpl.h 5610 * [275]17327 use of `enumeral_type' in template type unification 5611 * [276]17393 "unused variable '._0'" warning with -Wall 5612 * [277]17501 Confusion with member templates 5613 * [278]17537 g++ not passing -lstdc++ to linker when all command line 5614 arguments are libraries 5615 * [279]17585 usage of unqualified name of static member from within 5616 class not allowed 5617 * [280]17821 Poor diagnostic for using "." instead of "->" 5618 * [281]17829 wrong error: call of overloaded function is ambiguous 5619 * [282]17851 Misleading diagnostic for invalid function declarations 5620 with undeclared types 5621 * [283]17976 Destructor is called twice 5622 * [284]18020 rejects valid definition of enum value in template 5623 * [285]18093 bogus conflict in namespace aliasing 5624 * [286]18140 C++ parser bug when using >> in templates 5625 5626 Fortran 5627 5628 * [287]17541 data statements with double precision constants fail 5629 5630 x86-specific 5631 5632 * [288]17853 -O2 ICE for MMX testcase 5633 5634 SPARC-specific 5635 5636 * [289]17245 ICE compiling gsl-1.5 statistics/lag1.c 5637 5638 Darwin-specific 5639 5640 * [290]17167 FATAL:Symbol L_foo$stub already defined. 5641 5642 AIX-specific 5643 5644 * [291]17277 could not catch an exception when specified -maix64 5645 5646 Solaris-specific 5647 5648 * [292]17505 <cmath> calls acosf(), ceilf(), and other functions 5649 missing from system libraries 5650 5651 HP/UX specific: 5652 5653 * [293]17684 /usr/ccs/bin/ld: Can't create libgcc_s.sl 5654 5655 ARM-specific 5656 5657 * [294]17384 ICE with mode attribute on structures 5658 5659 MIPS-specific 5660 5661 * [295]17770 No NOP after LWL with -mips1 5662 5663 Other embedded target specific 5664 5665 * [296]11476 [arc-elf] gcc ICE on newlib's vfprintf.c 5666 * [297]14064 [avr-elf] -fdata-sections triggers ICE 5667 * [298]14678 [m68hc11-elf] gcc ICE 5668 * [299]15583 [powerpc-rtems] powerpc-rtems lacks __USE_INIT_FINI__ 5669 * [300]15790 [i686-coff] Alignment error building gcc with i686-coff 5670 target 5671 * [301]15886 [SH] Miscompilation with -O2 -fPIC 5672 * [302]16884 [avr-elf] [fweb related] bug while initializing 5673 variables 5674 5675 Bugs relating to debugger support 5676 5677 * [303]13841 missing debug info for _Complex function arguments 5678 * [304]15860 [big-endian targets] No DW_AT_location debug info is 5679 emitted for formal arguments to a function that uses "register" 5680 qualifiers 5681 5682 Testsuite issues (compiler not affected) 5683 5684 * [305]17465 Testsuite in libffi overrides LD_LIBRARY_PATH 5685 * [306]17469 Testsuite in libstdc++ overrides LD_LIBRARY_PATH 5686 * [307]18138 [mips-sgi-irix6.5] libgcc_s.so.1 not found by 64-bit 5687 testsuite 5688 5689 Documentation 5690 5691 * [308]15498 typo in gcc manual: non-existing locale example en_UK, 5692 should be en_GB 5693 * [309]15747 [mips-sgi-irix5.3] /bin/sh hangs during bootstrap: 5694 document broken shell 5695 * [310]16406 USE_LD_AS_NEEDED undocumented 5696 __________________________________________________________________ 5697 5698 GCC 3.4.4 5699 5700 This is the [311]list of problem reports (PRs) from GCC's bug tracking 5701 system that are known to be fixed in the 3.4.4 release. This list might 5702 not be complete (that is, it is possible that some PRs that have been 5703 fixed are not listed here). 5704 __________________________________________________________________ 5705 5706 GCC 3.4.5 5707 5708 This is the [312]list of problem reports (PRs) from GCC's bug tracking 5709 system that are known to be fixed in the 3.4.5 release. This list might 5710 not be complete (that is, it is possible that some PRs that have been 5711 fixed are not listed here). 5712 5713 Bootstrap issues 5714 5715 * [313]24688 sco_math fixincl breaks math.h 5716 5717 C compiler bugs 5718 5719 * [314]17188 struct Foo { } redefinition 5720 * [315]20187 wrong code for ((unsigned char)(unsigned long 5721 long)((a?a:1)&(a*b)))?0:1) 5722 * [316]21873 infinite warning loop on bad array initializer 5723 * [317]21899 enum definition accepts values to be overriden 5724 * [318]22061 ICE in find_function_data, in function.c 5725 * [319]22308 Failure to diagnose violation of constraint 6.516p2 5726 * [320]22458 ICE on missing brace 5727 * [321]22589 ICE casting to long long 5728 * [322]24101 Segfault with preprocessed source 5729 5730 C++ compiler and library bugs 5731 5732 * [323]10611 operations on vector mode not recognized in C++ 5733 * [324]13377 unexpected behavior of namespace usage directive 5734 * [325]16002 Strange error message with new parser 5735 * [326]17413 local classes as template argument 5736 * [327]17609 spurious error message after using keyword 5737 * [328]17618 ICE in cp_convert_to_pointer, in cp/cvt.c 5738 * [329]18124 ICE with invalid template template parameter 5739 * [330]18155 typedef in template declaration not rejected 5740 * [331]18177 ICE with const_cast for undeclared variable 5741 * [332]18368 C++ error message regression 5742 * [333]16378 ICE when returning a copy of a packed member 5743 * [334]18466 int ::i; accepted 5744 * [335]18512 ICE on invalid usage of template base class 5745 * [336]18454 ICE when returning undefined type 5746 * [337]18738 typename not allowed with non-dependent qualified name 5747 * [338]18803 rejects access to operator() in template 5748 * [339]19004 ICE in uses_template_parms, in cp/pt.c 5749 * [340]19208 Spurious error about variably modified type 5750 * [341]18253 bad error message / ICE for invalid template parameter 5751 * [342]19608 ICE after friend function definition in local class 5752 * [343]19884 ICE on explicit instantiation of a non-template 5753 constructor 5754 * [344]20153 ICE when C++ template function contains anonymous union 5755 * [345]20563 Infinite loop in diagnostic (and ice after error 5756 message) 5757 * [346]20789 ICE with incomplete type in template 5758 * [347]21336 Internal compiler error when using custom new operators 5759 * [348]21768 ICE in error message due to violation of coding 5760 conventions 5761 * [349]21853 constness of pointer to data member ignored 5762 * [350]21903 Default argument of template function causes a 5763 compile-time error 5764 * [351]21983 multiple diagnostics 5765 * [352]21987 New testsuite failure 5766 g++.dg/warn/conversion-function-1.C 5767 * [353]22153 ICE on invalid template specialization 5768 * [354]22172 Internal compiler error, seg fault. 5769 * [355]21286 filebuf::xsgetn vs pipes 5770 * [356]22233 ICE with wrong number of template parameters 5771 * [357]22508 ICE after invalid operator new 5772 * [358]22545 ICE with pointer to class member & user defined 5773 conversion operator 5774 * [359]23528 Wrong default allocator in ext/hash_map 5775 * [360]23550 char_traits requirements/1.cc test bad math 5776 * [361]23586 Bad diagnostic for invalid namespace-name 5777 * [362]23624 ICE in invert_truthvalue, in fold-const.c 5778 * [363]23639 Bad error message: not a member of '<declaration error>' 5779 * [364]23797 ICE on typename outside template 5780 * [365]23965 Bogus error message: no matching function for call to 5781 'foo(<type error>)' 5782 * [366]24052 &#`label_decl' not supported by dump_expr#<expression 5783 error> 5784 * [367]24580 virtual base class cause exception not to be caught 5785 5786 Problems in generated debug information 5787 5788 * [368]24267 Bad DWARF for altivec vectors 5789 5790 Optimizations issues 5791 5792 * [369]17810 ICE in verify_local_live_at_start 5793 * [370]17860 Wrong generated code for loop with varying bound 5794 * [371]21709 ICE on compile-time complex NaN 5795 * [372]21964 broken tail call at -O2 or more 5796 * [373]22167 Strange optimization bug when using -Os 5797 * [374]22619 Compilation failure for real_const_1.f and 5798 real_const_2.f90 5799 * [375]23241 Invalid code generated for comparison of uchar to 255 5800 * [376]23478 Miscompilation due to reloading of a var that is also 5801 used in EH pad 5802 * [377]24470 segmentation fault in cc1plus when compiling with -O 5803 * [378]24950 ICE in operand_subword_force 5804 5805 Precompiled headers problems 5806 5807 * [379]14400 Cannot compile qt-x11-free-3.3.0 5808 * [380]14940 PCH largefile test fails on various platforms 5809 5810 Preprocessor bugs 5811 5812 * [381]20239 ICE on empty preprocessed input 5813 * [382]15220 "gcc -E -MM -MG" reports missing system headers in 5814 source directory 5815 5816 Testsuite issues 5817 5818 * [383]19275 gcc.dg/20020919-1.c fails with -fpic/-fPIC on 5819 i686-pc-linux-gnu 5820 5821 Alpha specific 5822 5823 * [384]21888 bootstrap failure with linker relaxation enabled 5824 5825 ARM specific 5826 5827 * [385]15342 [arm-linux]: ICE in verify_local_live_at_start 5828 * [386]23985 Memory aliasing information incorrect in inlined memcpy 5829 5830 ColdFile specific 5831 5832 * [387]16719 Illegal move of byte into address register causes 5833 compiler to ICE 5834 5835 HPPA specific 5836 5837 * [388]21723 ICE while building libgfortran 5838 * [389]21841 -mhp-ld/-mgnu-ld documentation 5839 5840 IA-64 specific 5841 5842 * [390]23644 IA-64 hardware models and configuration options 5843 documentation error 5844 * [391]24718 Shared libgcc not used for linking by default 5845 5846 M68000 specific 5847 5848 * [392]18421 ICE in reload_cse_simplify_operands, in postreload.c 5849 5850 MIPS specific 5851 5852 * [393]20621 ICE in change_address_1, in emit-rtl.c 5853 5854 PowerPC and PowerPC64 specific 5855 5856 * [394]18583 error on valid code: const 5857 __attribute__((altivec(vector__))) doesn't work in arrays 5858 * [395]20191 ICE in reload_cse_simplify_operands 5859 * [396]22083 AIX: TARGET_C99_FUNCTIONS is wrongly defined 5860 * [397]23070 CALL_V4_CLEAR_FP_ARGS flag not properly set 5861 * [398]23404 gij trashes args of functions with more than 8 fp args 5862 * [399]23539 C & C++ compiler generating misaligned references 5863 regardless of compiler flags 5864 * [400]24102 floatdisf2_internal2 broken 5865 * [401]24465 -mminimal-toc miscompilation of __thread vars 5866 5867 Solaris specific 5868 5869 * [402]19933 Problem with define of HUGE_VAL in math_c99 5870 * [403]21889 Native Solaris assembler cannot grok DTP-relative debug 5871 symbols 5872 5873 SPARC specific 5874 5875 * [404]19300 PCH failures on sparc-linux 5876 * [405]20301 Assembler labels have a leading "-" 5877 * [406]20673 C PCH testsuite assembly comparison failure 5878 5879 x86 and x86_64 specific 5880 5881 * [407]18582 ICE with arrays of type V2DF 5882 * [408]19340 Compilation SEGFAULTs with -O1 -fschedule-insns2 5883 -fsched2-use-traces 5884 * [409]21716 ICE in reg-stack.c's swap_rtx_condition 5885 * [410]24315 amd64 fails -fpeephole2 5886 __________________________________________________________________ 5887 5888 GCC 3.4.6 5889 5890 This is the [411]list of problem reports (PRs) from GCC's bug tracking 5891 system that are known to be fixed in the 3.4.6 release. This list might 5892 not be complete (that is, it is possible that some PRs that have been 5893 fixed are not listed here). 5894 5895 5896 For questions related to the use of GCC, please consult these web 5897 pages and the [412]GCC manuals. If that fails, the 5898 [413]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 5899 web pages and the development of GCC are welcome on our developer 5900 list at [414]gcc (a] gcc.gnu.org. All of [415]our lists have public 5901 archives. 5902 5903 Copyright (C) [416]Free Software Foundation, Inc. Verbatim copying and 5904 distribution of this entire article is permitted in any medium, 5905 provided this notice is preserved. 5906 5907 These pages are [417]maintained by the GCC team. Last modified 5908 2011-04-25[418]. 5909 5910 References 5911 5912 1. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6 5913 2. http://gcc.gnu.org/gcc-3.4/changes.html#cplusplus 5914 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems 5915 4. http://gcc.gnu.org/gcc-3.4/changes.html#obsolete_systems 5916 5. http://gcc.gnu.org/gcc-3.4/mips-abi.html 5917 6. http://gcc.gnu.org/gcc-3.4/sparc-abi.html 5918 7. http://www.boost.org/ 5919 8. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11953 5920 9. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8361 5921 10. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Other-Builtins.html#Other%20Builtins 5922 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#209 5923 12. http://gcc.gnu.org/bugs/#cxx_rvalbind 5924 13. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html 5925 14. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html 5926 15. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html 5927 16. http://www.gnu.org/software/classpath/ 5928 17. http://www.eclipse.org/ 5929 18. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/g77/News.html 5930 19. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Alpha-Built-in-Functions.html 5931 20. http://h30097.www3.hp.com/docs/base_doc/DOCUMENTATION/V51A_HTML/ARH9MBTE/DTMNPLTN.HTM#normal-argument-list-structure 5932 21. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html 5933 22. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Comparison-of-the-two-descriptions.html 5934 23. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html 5935 24. http://gcc.gnu.org/gcc-3.4/mips-abi.html 5936 25. http://gcc.gnu.org/gcc-3.4/powerpc-abi.html 5937 26. http://gcc.gnu.org/gcc-3.4/sparc-abi.html 5938 27. http://gcc.gnu.org/bugzilla/buglist.cgi?short_desc_type=notregexp&short_desc=%5C%5B3%5C.4.*%5BRr%5Degression&target_milestone=3.4.0&bug_status=RESOLVED&resolution=FIXED 5939 28. http://gcc.gnu.org/PR10129 5940 29. http://gcc.gnu.org/PR14576 5941 30. http://gcc.gnu.org/PR14760 5942 31. http://gcc.gnu.org/PR14671 5943 32. http://gcc.gnu.org/PR15093 5944 33. http://gcc.gnu.org/PR15178 5945 34. http://gcc.gnu.org/PR12753 5946 35. http://gcc.gnu.org/PR13985 5947 36. http://gcc.gnu.org/PR14810 5948 37. http://gcc.gnu.org/PR14883 5949 38. http://gcc.gnu.org/PR15044 5950 39. http://gcc.gnu.org/PR15057 5951 40. http://gcc.gnu.org/PR15064 5952 41. http://gcc.gnu.org/PR15142 5953 42. http://gcc.gnu.org/PR15159 5954 43. http://gcc.gnu.org/PR15165 5955 44. http://gcc.gnu.org/PR15193 5956 45. http://gcc.gnu.org/PR15209 5957 46. http://gcc.gnu.org/PR15227 5958 47. http://gcc.gnu.org/PR15285 5959 48. http://gcc.gnu.org/PR15299 5960 49. http://gcc.gnu.org/PR15329 5961 50. http://gcc.gnu.org/PR15550 5962 51. http://gcc.gnu.org/PR15554 5963 52. http://gcc.gnu.org/PR15640 5964 53. http://gcc.gnu.org/PR15666 5965 54. http://gcc.gnu.org/PR15696 5966 55. http://gcc.gnu.org/PR15701 5967 56. http://gcc.gnu.org/PR15761 5968 57. http://gcc.gnu.org/PR15829 5969 58. http://gcc.gnu.org/PR14538 5970 59. http://gcc.gnu.org/PR12391 5971 60. http://gcc.gnu.org/PR14649 5972 61. http://gcc.gnu.org/PR15004 5973 62. http://gcc.gnu.org/PR15749 5974 63. http://gcc.gnu.org/PR10646 5975 64. http://gcc.gnu.org/PR12077 5976 65. http://gcc.gnu.org/PR13598 5977 66. http://gcc.gnu.org/PR14211 5978 67. http://gcc.gnu.org/PR14220 5979 68. http://gcc.gnu.org/PR14245 5980 69. http://gcc.gnu.org/PR14340 5981 70. http://gcc.gnu.org/PR14600 5982 71. http://gcc.gnu.org/PR14668 5983 72. http://gcc.gnu.org/PR14775 5984 73. http://gcc.gnu.org/PR14821 5985 74. http://gcc.gnu.org/PR14930 5986 75. http://gcc.gnu.org/PR14932 5987 76. http://gcc.gnu.org/PR14950 5988 77. http://gcc.gnu.org/PR14962 5989 78. http://gcc.gnu.org/PR14975 5990 79. http://gcc.gnu.org/PR15002 5991 80. http://gcc.gnu.org/PR15025 5992 81. http://gcc.gnu.org/PR15046 5993 82. http://gcc.gnu.org/PR15069 5994 83. http://gcc.gnu.org/PR15074 5995 84. http://gcc.gnu.org/PR15083 5996 85. http://gcc.gnu.org/PR15096 5997 86. http://gcc.gnu.org/PR15287 5998 87. http://gcc.gnu.org/PR15317 5999 88. http://gcc.gnu.org/PR15337 6000 89. http://gcc.gnu.org/PR15361 6001 90. http://gcc.gnu.org/PR15412 6002 91. http://gcc.gnu.org/PR15427 6003 92. http://gcc.gnu.org/PR15471 6004 93. http://gcc.gnu.org/PR15503 6005 94. http://gcc.gnu.org/PR15507 6006 95. http://gcc.gnu.org/PR15542 6007 96. http://gcc.gnu.org/PR15565 6008 97. http://gcc.gnu.org/PR15625 6009 98. http://gcc.gnu.org/PR15629 6010 99. http://gcc.gnu.org/PR15742 6011 100. http://gcc.gnu.org/PR15775 6012 101. http://gcc.gnu.org/PR15821 6013 102. http://gcc.gnu.org/PR15862 6014 103. http://gcc.gnu.org/PR15875 6015 104. http://gcc.gnu.org/PR15877 6016 105. http://gcc.gnu.org/PR15947 6017 106. http://gcc.gnu.org/PR16020 6018 107. http://gcc.gnu.org/PR16154 6019 108. http://gcc.gnu.org/PR16174 6020 109. http://gcc.gnu.org/PR14315 6021 110. http://gcc.gnu.org/PR15151 6022 111. http://gcc.gnu.org/PR7993 6023 112. http://gcc.gnu.org/PR15228 6024 113. http://gcc.gnu.org/PR15345 6025 114. http://gcc.gnu.org/PR15945 6026 115. http://gcc.gnu.org/PR15526 6027 116. http://gcc.gnu.org/PR14690 6028 117. http://gcc.gnu.org/PR15112 6029 118. http://gcc.gnu.org/PR15067 6030 119. http://gcc.gnu.org/PR1963 6031 120. http://gcc.gnu.org/PR15717 6032 121. http://gcc.gnu.org/PR14782 6033 122. http://gcc.gnu.org/PR14828 6034 123. http://gcc.gnu.org/PR15202 6035 124. http://gcc.gnu.org/PR14610 6036 125. http://gcc.gnu.org/PR14813 6037 126. http://gcc.gnu.org/PR14857 6038 127. http://gcc.gnu.org/PR15598 6039 128. http://gcc.gnu.org/PR15653 6040 129. http://gcc.gnu.org/PR15189 6041 130. http://gcc.gnu.org/PR15331 6042 131. http://gcc.gnu.org/PR16144 6043 132. http://gcc.gnu.org/PR16176 6044 133. http://gcc.gnu.org/PR11591 6045 134. http://gcc.gnu.org/PR12028 6046 135. http://gcc.gnu.org/PR14478 6047 136. http://gcc.gnu.org/PR14567 6048 137. http://gcc.gnu.org/PR14715 6049 138. http://gcc.gnu.org/PR14902 6050 139. http://gcc.gnu.org/PR14924 6051 140. http://gcc.gnu.org/PR14960 6052 141. http://gcc.gnu.org/PR15106 6053 142. http://gcc.gnu.org/PR16026 6054 143. http://gcc.gnu.org/PR15191 6055 144. http://gcc.gnu.org/PR15662 6056 145. http://gcc.gnu.org/PR15054 6057 146. http://gcc.gnu.org/PR15783 6058 147. http://gcc.gnu.org/PR15626 6059 148. http://gcc.gnu.org/PR14326 6060 149. http://gcc.gnu.org/PR14723 6061 150. http://gcc.gnu.org/PR15290 6062 151. http://gcc.gnu.org/PR15250 6063 152. http://gcc.gnu.org/PR15551 6064 153. http://gcc.gnu.org/PR8309 6065 154. http://gcc.gnu.org/PR13250 6066 155. http://gcc.gnu.org/PR13803 6067 156. http://gcc.gnu.org/PR14093 6068 157. http://gcc.gnu.org/PR14457 6069 158. http://gcc.gnu.org/PR14542 6070 159. http://gcc.gnu.org/PR15100 6071 160. http://gcc.gnu.org/PR15296 6072 161. http://gcc.gnu.org/PR15396 6073 162. http://gcc.gnu.org/PR15782 6074 163. http://gcc.gnu.org/PR11610 6075 164. http://gcc.gnu.org/PR15488 6076 165. http://gcc.gnu.org/PR15489 6077 166. http://gcc.gnu.org/PR13928 6078 167. http://gcc.gnu.org/PR14150 6079 168. http://gcc.gnu.org/PR14949 6080 169. http://gcc.gnu.org/PR15123 6081 170. http://gcc.gnu.org/PR16469 6082 171. http://gcc.gnu.org/PR16344 6083 172. http://gcc.gnu.org/PR16842 6084 173. http://gcc.gnu.org/PR12608 6085 174. http://gcc.gnu.org/PR14492 6086 175. http://gcc.gnu.org/PR15461 6087 176. http://gcc.gnu.org/PR15890 6088 177. http://gcc.gnu.org/PR16180 6089 178. http://gcc.gnu.org/PR16224 6090 179. http://gcc.gnu.org/PR16408 6091 180. http://gcc.gnu.org/PR16529 6092 181. http://gcc.gnu.org/PR16698 6093 182. http://gcc.gnu.org/PR16706 6094 183. http://gcc.gnu.org/PR16810 6095 184. http://gcc.gnu.org/PR16851 6096 185. http://gcc.gnu.org/PR16870 6097 186. http://gcc.gnu.org/PR16904 6098 187. http://gcc.gnu.org/PR16905 6099 188. http://gcc.gnu.org/PR16964 6100 189. http://gcc.gnu.org/PR17068 6101 190. http://gcc.gnu.org/PR16366 6102 191. http://gcc.gnu.org/PR15345 6103 192. http://gcc.gnu.org/PR16590 6104 193. http://gcc.gnu.org/PR16693 6105 194. http://gcc.gnu.org/PR17078 6106 195. http://gcc.gnu.org/PR13956 6107 196. http://gcc.gnu.org/PR16684 6108 197. http://gcc.gnu.org/PR12658 6109 198. http://gcc.gnu.org/PR13092 6110 199. http://gcc.gnu.org/PR15320 6111 200. http://gcc.gnu.org/PR16246 6112 201. http://gcc.gnu.org/PR16273 6113 202. http://gcc.gnu.org/PR16401 6114 203. http://gcc.gnu.org/PR16411 6115 204. http://gcc.gnu.org/PR16489 6116 205. http://gcc.gnu.org/PR16618 6117 206. http://gcc.gnu.org/PR16637 6118 207. http://gcc.gnu.org/PR16717 6119 208. http://gcc.gnu.org/PR16813 6120 209. http://gcc.gnu.org/PR16853 6121 210. http://gcc.gnu.org/PR16889 6122 211. http://gcc.gnu.org/PR16959 6123 212. http://gcc.gnu.org/PR7587 6124 213. http://gcc.gnu.org/PR16473 6125 214. http://gcc.gnu.org/PR16478 6126 215. http://gcc.gnu.org/PR10695 6127 216. http://gcc.gnu.org/PR16974 6128 217. http://gcc.gnu.org/PR16298 6129 218. http://gcc.gnu.org/PR17113 6130 219. http://gcc.gnu.org/PR14697 6131 220. http://gcc.gnu.org/PR15869 6132 221. http://gcc.gnu.org/PR16325 6133 222. http://gcc.gnu.org/PR16357 6134 223. http://gcc.gnu.org/PR16380 6135 224. http://gcc.gnu.org/PR16407 6136 225. http://gcc.gnu.org/PR16643 6137 226. http://gcc.gnu.org/PR15927 6138 227. http://gcc.gnu.org/PR15948 6139 228. http://gcc.gnu.org/PR17019 6140 229. http://gcc.gnu.org/PR16130 6141 230. http://gcc.gnu.org/PR16142 6142 231. http://gcc.gnu.org/PR16278 6143 232. http://gcc.gnu.org/PR16414 6144 233. http://gcc.gnu.org/PR16445 6145 234. http://gcc.gnu.org/PR16490 6146 235. http://gcc.gnu.org/PR16683 6147 236. http://gcc.gnu.org/PR16195 6148 237. http://gcc.gnu.org/PR16239 6149 238. http://gcc.gnu.org/PR16199 6150 239. http://gcc.gnu.org/PR16416 6151 240. http://gcc.gnu.org/PR16430 6152 241. http://gcc.gnu.org/PR16379 6153 242. http://gcc.gnu.org/PR17093 6154 243. http://gcc.gnu.org/PR17119 6155 244. http://gcc.gnu.org/PR15928 6156 245. http://gcc.gnu.org/PR16210 6157 246. http://gcc.gnu.org/PR15488 6158 247. http://gcc.gnu.org/PR16250 6159 248. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.3 6160 249. http://gcc.gnu.org/PR17369 6161 250. http://gcc.gnu.org/PR17850 6162 251. http://gcc.gnu.org/PR13948 6163 252. http://gcc.gnu.org/PR14492 6164 253. http://gcc.gnu.org/PR16301 6165 254. http://gcc.gnu.org/PR16566 6166 255. http://gcc.gnu.org/PR17023 6167 256. http://gcc.gnu.org/PR17027 6168 257. http://gcc.gnu.org/PR17524 6169 258. http://gcc.gnu.org/PR17826 6170 259. http://gcc.gnu.org/PR15526 6171 260. http://gcc.gnu.org/PR16999 6172 261. http://gcc.gnu.org/PR17503 6173 262. http://gcc.gnu.org/PR17581 6174 263. http://gcc.gnu.org/PR18129 6175 264. http://gcc.gnu.org/PR10975 6176 265. http://gcc.gnu.org/PR11722 6177 266. http://gcc.gnu.org/PR14534 6178 267. http://gcc.gnu.org/PR15172 6179 268. http://gcc.gnu.org/PR15786 6180 269. http://gcc.gnu.org/PR16162 6181 270. http://gcc.gnu.org/PR16612 6182 271. http://gcc.gnu.org/PR16715 6183 272. http://gcc.gnu.org/PR16848 6184 273. http://gcc.gnu.org/PR17132 6185 274. http://gcc.gnu.org/PR17259 6186 275. http://gcc.gnu.org/PR17327 6187 276. http://gcc.gnu.org/PR17393 6188 277. http://gcc.gnu.org/PR17501 6189 278. http://gcc.gnu.org/PR17537 6190 279. http://gcc.gnu.org/PR17585 6191 280. http://gcc.gnu.org/PR17821 6192 281. http://gcc.gnu.org/PR17829 6193 282. http://gcc.gnu.org/PR17851 6194 283. http://gcc.gnu.org/PR17976 6195 284. http://gcc.gnu.org/PR18020 6196 285. http://gcc.gnu.org/PR18093 6197 286. http://gcc.gnu.org/PR18140 6198 287. http://gcc.gnu.org/PR17541 6199 288. http://gcc.gnu.org/PR17853 6200 289. http://gcc.gnu.org/PR17245 6201 290. http://gcc.gnu.org/PR17167 6202 291. http://gcc.gnu.org/PR17277 6203 292. http://gcc.gnu.org/PR17505 6204 293. http://gcc.gnu.org/PR17684 6205 294. http://gcc.gnu.org/PR17384 6206 295. http://gcc.gnu.org/PR17770 6207 296. http://gcc.gnu.org/PR11476 6208 297. http://gcc.gnu.org/PR14064 6209 298. http://gcc.gnu.org/PR14678 6210 299. http://gcc.gnu.org/PR15583 6211 300. http://gcc.gnu.org/PR15790 6212 301. http://gcc.gnu.org/PR15886 6213 302. http://gcc.gnu.org/PR16884 6214 303. http://gcc.gnu.org/PR13841 6215 304. http://gcc.gnu.org/PR15860 6216 305. http://gcc.gnu.org/PR17465 6217 306. http://gcc.gnu.org/PR17469 6218 307. http://gcc.gnu.org/PR18138 6219 308. http://gcc.gnu.org/PR15498 6220 309. http://gcc.gnu.org/PR15747 6221 310. http://gcc.gnu.org/PR16406 6222 311. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.4 6223 312. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.5 6224 313. http://gcc.gnu.org/PR24688 6225 314. http://gcc.gnu.org/PR17188 6226 315. http://gcc.gnu.org/PR20187 6227 316. http://gcc.gnu.org/PR21873 6228 317. http://gcc.gnu.org/PR21899 6229 318. http://gcc.gnu.org/PR22061 6230 319. http://gcc.gnu.org/PR22208 6231 320. http://gcc.gnu.org/PR22458 6232 321. http://gcc.gnu.org/PR22589 6233 322. http://gcc.gnu.org/PR24101 6234 323. http://gcc.gnu.org/PR10611 6235 324. http://gcc.gnu.org/PR13377 6236 325. http://gcc.gnu.org/PR16002 6237 326. http://gcc.gnu.org/PR17413 6238 327. http://gcc.gnu.org/PR17609 6239 328. http://gcc.gnu.org/PR17618 6240 329. http://gcc.gnu.org/PR18124 6241 330. http://gcc.gnu.org/PR18155 6242 331. http://gcc.gnu.org/PR18177 6243 332. http://gcc.gnu.org/PR18368 6244 333. http://gcc.gnu.org/PR18378 6245 334. http://gcc.gnu.org/PR18466 6246 335. http://gcc.gnu.org/PR18512 6247 336. http://gcc.gnu.org/PR18545 6248 337. http://gcc.gnu.org/PR18738 6249 338. http://gcc.gnu.org/PR18803 6250 339. http://gcc.gnu.org/PR19004 6251 340. http://gcc.gnu.org/PR19208 6252 341. http://gcc.gnu.org/PR19253 6253 342. http://gcc.gnu.org/PR19608 6254 343. http://gcc.gnu.org/PR19884 6255 344. http://gcc.gnu.org/PR20153 6256 345. http://gcc.gnu.org/PR20563 6257 346. http://gcc.gnu.org/PR20789 6258 347. http://gcc.gnu.org/PR21336 6259 348. http://gcc.gnu.org/PR21768 6260 349. http://gcc.gnu.org/PR21853 6261 350. http://gcc.gnu.org/PR21903 6262 351. http://gcc.gnu.org/PR21983 6263 352. http://gcc.gnu.org/PR21987 6264 353. http://gcc.gnu.org/PR22153 6265 354. http://gcc.gnu.org/PR22172 6266 355. http://gcc.gnu.org/PR21286 6267 356. http://gcc.gnu.org/PR22233 6268 357. http://gcc.gnu.org/PR22508 6269 358. http://gcc.gnu.org/PR22545 6270 359. http://gcc.gnu.org/PR23528 6271 360. http://gcc.gnu.org/PR23550 6272 361. http://gcc.gnu.org/PR23586 6273 362. http://gcc.gnu.org/PR23624 6274 363. http://gcc.gnu.org/PR23639 6275 364. http://gcc.gnu.org/PR23797 6276 365. http://gcc.gnu.org/PR23965 6277 366. http://gcc.gnu.org/PR24052 6278 367. http://gcc.gnu.org/PR24580 6279 368. http://gcc.gnu.org/PR24267 6280 369. http://gcc.gnu.org/PR17810 6281 370. http://gcc.gnu.org/PR17860 6282 371. http://gcc/gnu.org/PR21709 6283 372. http://gcc.gnu.org/PR21964 6284 373. http://gcc.gnu.org/PR22167 6285 374. http://gcc.gnu.org/PR22619 6286 375. http://gcc.gnu.org/PR23241 6287 376. http://gcc.gnu.org/PR23478 6288 377. http://gcc.gnu.org/PR24470 6289 378. http://gcc.gnu.org/PR24950 6290 379. http://gcc.gnu.org/PR14400 6291 380. http://gcc.gnu.org/PR14940 6292 381. http://gcc.gnu.org/PR20239 6293 382. http://gcc.gnu.org/PR15220 6294 383. http://gcc.gnu.org/PR19275 6295 384. http://gcc.gnu.org/PR21888 6296 385. http://gcc.gnu.org/PR15342 6297 386. http://gcc.gnu.org/PR23985 6298 387. http://gcc.gnu.org/PR16719 6299 388. http://gcc.gnu.org/PR21723 6300 389. http://gcc.gnu.org/PR21841 6301 390. http://gcc.gnu.org/PR23644 6302 391. http://gcc.gnu.org/PR24718 6303 392. http://gcc.gnu.org/PR18421 6304 393. http://gcc.gnu.org/PR20621 6305 394. http://gcc.gnu.org/PR18583 6306 395. http://gcc.gnu.org/PR20191 6307 396. http://gcc.gnu.org/PR22083 6308 397. http://gcc.gnu.org/PR23070 6309 398. http://gcc.gnu.org/PR23404 6310 399. http://gcc.gnu.org/PR23539 6311 400. http://gcc.gnu.org/PR24102 6312 401. http://gcc.gnu.org/PR24465 6313 402. http://gcc.gnu.org/PR19933 6314 403. http://gcc.gnu.org/PR21889 6315 404. http://gcc.gnu.org/PR19300 6316 405. http://gcc.gnu.org/PR20301 6317 406. http://gcc.gnu.org/PR20673 6318 407. http://gcc.gnu.org/PR18582 6319 408. http://gcc.gnu.org/PR19340 6320 409. http://gcc.gnu.org/PR21716 6321 410. http://gcc.gnu.org/PR24315 6322 411. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.6 6323 412. http://gcc.gnu.org/onlinedocs/ 6324 413. mailto:gcc-help (a] gcc.gnu.org 6325 414. mailto:gcc (a] gcc.gnu.org 6326 415. http://gcc.gnu.org/lists.html 6327 416. http://www.fsf.org/ 6328 417. http://gcc.gnu.org/about.html 6329 418. http://validator.w3.org/check/referer 6330 ====================================================================== 6331 http://gcc.gnu.org/gcc-3.3/index.html 6332 6333 GCC 3.3 Release Series 6334 6335 May 03, 2005 6336 6337 The [1]GNU project and the GCC developers are pleased to announce the 6338 release of GCC 3.3.6. 6339 6340 This release is a bug-fix release, containing fixes for regressions in 6341 GCC 3.3.5 relative to previous releases of GCC. 6342 6343 This release is the last of the series 3.3.x. 6344 6345 The GCC 3.3 release series includes numerous [2]new features, 6346 improvements, bug fixes, and other changes, thanks to an [3]amazing 6347 group of volunteers. 6348 6349 Release History 6350 6351 GCC 3.3.6 6352 May 3, 2005 ([4]changes) 6353 6354 GCC 3.3.5 6355 September 30, 2004 ([5]changes) 6356 6357 GCC 3.3.4 6358 May 31, 2004 ([6]changes) 6359 6360 GCC 3.3.3 6361 February 14, 2004 ([7]changes) 6362 6363 GCC 3.3.2 6364 October 16, 2003 ([8]changes) 6365 6366 GCC 3.3.1 6367 August 8, 2003 ([9]changes) 6368 6369 GCC 3.3 6370 May 14, 2003 ([10]changes) 6371 6372 References and Acknowledgements 6373 6374 GCC used to stand for the GNU C Compiler, but since the compiler 6375 supports several other languages aside from C, it now stands for the 6376 GNU Compiler Collection. 6377 6378 A list of [11]successful builds is updated as new information becomes 6379 available. 6380 6381 The GCC developers would like to thank the numerous people that have 6382 contributed new features, improvements, bug fixes, and other changes as 6383 well as test results to GCC. This [12]amazing group of volunteers is 6384 what makes GCC successful. 6385 6386 For additional information about GCC please refer to the [13]GCC 6387 project web site or contact the [14]GCC development mailing list. 6388 6389 To obtain GCC please use [15]our mirror sites, or our CVS server. 6390 6391 6392 For questions related to the use of GCC, please consult these web 6393 pages and the [16]GCC manuals. If that fails, the 6394 [17]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 6395 web pages and the development of GCC are welcome on our developer 6396 list at [18]gcc (a] gcc.gnu.org. All of [19]our lists have public 6397 archives. 6398 6399 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and 6400 distribution of this entire article is permitted in any medium, 6401 provided this notice is preserved. 6402 6403 These pages are [21]maintained by the GCC team. Last modified 6404 2011-04-25[22]. 6405 6406 References 6407 6408 1. http://www.gnu.org/ 6409 2. http://gcc.gnu.org/gcc-3.3/changes.html 6410 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 6411 4. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6 6412 5. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.5 6413 6. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.4 6414 7. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.3 6415 8. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.2 6416 9. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.1 6417 10. http://gcc.gnu.org/gcc-3.3/changes.html 6418 11. http://gcc.gnu.org/gcc-3.3/buildstat.html 6419 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 6420 13. http://gcc.gnu.org/index.html 6421 14. mailto:gcc (a] gcc.gnu.org 6422 15. http://gcc.gnu.org/mirrors.html 6423 16. http://gcc.gnu.org/onlinedocs/ 6424 17. mailto:gcc-help (a] gcc.gnu.org 6425 18. mailto:gcc (a] gcc.gnu.org 6426 19. http://gcc.gnu.org/lists.html 6427 20. http://www.fsf.org/ 6428 21. http://gcc.gnu.org/about.html 6429 22. http://validator.w3.org/check/referer 6430 ====================================================================== 6431 http://gcc.gnu.org/gcc-3.3/changes.html 6432 6433 GCC 3.3 Release Series 6434 Changes, New Features, and Fixes 6435 6436 The latest release in the 3.3 release series is [1]GCC 3.3.6. 6437 6438 Caveats 6439 6440 * The preprocessor no longer accepts multi-line string literals. They 6441 were deprecated in 3.0, 3.1, and 3.2. 6442 * The preprocessor no longer supports the -A- switch when appearing 6443 alone. -A- followed by an assertion is still supported. 6444 * Support for all the systems [2]obsoleted in GCC 3.1 has been 6445 removed from GCC 3.3. See below for a [3]list of systems which are 6446 obsoleted in this release. 6447 * Checking for null format arguments has been decoupled from the rest 6448 of the format checking mechanism. Programs which use the format 6449 attribute may regain this functionality by using the new [4]nonnull 6450 function attribute. Note that all functions for which GCC has a 6451 built-in format attribute, an appropriate built-in nonnull 6452 attribute is also applied. 6453 * The DWARF (version 1) debugging format has been deprecated and will 6454 be removed in a future version of GCC. Version 2 of the DWARF 6455 debugging format will continue to be supported for the foreseeable 6456 future. 6457 * The C and Objective-C compilers no longer accept the "Naming Types" 6458 extension (typedef foo = bar); it was already unavailable in C++. 6459 Code which uses it will need to be changed to use the "typeof" 6460 extension instead: typedef typeof(bar) foo. (We have removed this 6461 extension without a period of deprecation because it has caused the 6462 compiler to crash since version 3.0 and no one noticed until very 6463 recently. Thus we conclude it is not in widespread use.) 6464 * The -traditional C compiler option has been removed. It was 6465 deprecated in 3.1 and 3.2. (Traditional preprocessing remains 6466 available.) The <varargs.h> header, used for writing variadic 6467 functions in traditional C, still exists but will produce an error 6468 message if used. 6469 * GCC 3.3.1 automatically places zero-initialized variables in the 6470 .bss section on some operating systems. Versions of GNU Emacs up to 6471 (and including) 21.3 will not work correctly when using this 6472 optimization; you can use -fno-zero-initialized-in-bss to disable 6473 it. 6474 6475 General Optimizer Improvements 6476 6477 * A new scheme for accurately describing processor pipelines, the 6478 [5]DFA scheduler, has been added. 6479 * Pavel Nejedly, Charles University Prague, has contributed new file 6480 format used by the edge coverage profiler (-fprofile-arcs). 6481 The new format is robust and diagnoses common mistakes where 6482 profiles from different versions (or compilations) of the program 6483 are combined resulting in nonsensical profiles and slow code to 6484 produced with profile feedback. Additionally this format allows 6485 extra data to be gathered. Currently, overall statistics are 6486 produced helping optimizers to identify hot spots of a program 6487 globally replacing the old intra-procedural scheme and resulting in 6488 better code. Note that the gcov tool from older GCC versions will 6489 not be able to parse the profiles generated by GCC 3.3 and vice 6490 versa. 6491 * Jan Hubicka, SuSE Labs, has contributed a new superblock formation 6492 pass enabled using -ftracer. This pass simplifies the control flow 6493 of functions allowing other optimizations to do better job. 6494 He also contributed the function reordering pass 6495 (-freorder-functions) to optimize function placement using profile 6496 feedback. 6497 6498 New Languages and Language specific improvements 6499 6500 C/ObjC/C++ 6501 6502 * The preprocessor now accepts directives within macro arguments. It 6503 processes them just as if they had not been within macro arguments. 6504 * The separate ISO and traditional preprocessors have been completely 6505 removed. The front end handles either type of preprocessed output 6506 if necessary. 6507 * In C99 mode preprocessor arithmetic is done in the precision of the 6508 target's intmax_t, as required by that standard. 6509 * The preprocessor can now copy comments inside macros to the output 6510 file when the macro is expanded. This feature, enabled using the 6511 -CC option, is intended for use by applications which place 6512 metadata or directives inside comments, such as lint. 6513 * The method of constructing the list of directories to be searched 6514 for header files has been revised. If a directory named by a -I 6515 option is a standard system include directory, the option is 6516 ignored to ensure that the default search order for system 6517 directories and the special treatment of system header files are 6518 not defeated. 6519 * A few more [6]ISO C99 features now work correctly. 6520 * A new function attribute, nonnull, has been added which allows 6521 pointer arguments to functions to be specified as requiring a 6522 non-null value. The compiler currently uses this information to 6523 issue a warning when it detects a null value passed in such an 6524 argument slot. 6525 * A new type attribute, may_alias, has been added. Accesses to 6526 objects with types with this attribute are not subjected to 6527 type-based alias analysis, but are instead assumed to be able to 6528 alias any other type of objects, just like the char type. 6529 6530 C++ 6531 6532 * Type based alias analysis has been implemented for C++ aggregate 6533 types. 6534 6535 Objective-C 6536 6537 * Generate an error if Objective-C objects are passed by value in 6538 function and method calls. 6539 * When -Wselector is used, check the whole list of selectors at the 6540 end of compilation, and emit a warning if a @selector() is not 6541 known. 6542 * Define __NEXT_RUNTIME__ when compiling for the NeXT runtime. 6543 * No longer need to include objc/objc-class.h to compile self calls 6544 in class methods (NeXT runtime only). 6545 * New -Wundeclared-selector option. 6546 * Removed selector bloating which was causing object files to be 10% 6547 bigger on average (GNU runtime only). 6548 * Using at run time @protocol() objects has been fixed in certain 6549 situations (GNU runtime only). 6550 * Type checking has been fixed and improved in many situations 6551 involving protocols. 6552 6553 Java 6554 6555 * The java.sql and javax.sql packages now implement the JDBC 3.0 (JDK 6556 1.4) API. 6557 * The JDK 1.4 assert facility has been implemented. 6558 * The bytecode interpreter is now direct threaded and thus faster. 6559 6560 Fortran 6561 6562 * Fortran improvements are listed in [7]the Fortran documentation. 6563 6564 Ada 6565 6566 * Ada tasking now works with glibc 2.3.x threading libraries. 6567 6568 New Targets and Target Specific Improvements 6569 6570 * The following changes have been made to the HP-PA port: 6571 + The port now defaults to scheduling for the PA8000 series of 6572 processors. 6573 + Scheduling support for the PA7300 processor has been added. 6574 + The 32-bit port now supports weak symbols under HP-UX 11. 6575 + The handling of initializers and finalizers has been improved 6576 under HP-UX 11. The 64-bit port no longer uses collect2. 6577 + Dwarf2 EH support has been added to the 32-bit linux port. 6578 + ABI fixes to correct the passing of small structures by value. 6579 * The SPARC, HP-PA, SH4, and x86/pentium ports have been converted to 6580 use the DFA processor pipeline description. 6581 * The following NetBSD configurations for the SuperH processor family 6582 have been added: 6583 + SH3, big-endian, sh-*-netbsdelf* 6584 + SH3, little-endian, shle-*-netbsdelf* 6585 + SH5, SHmedia, big-endian, 32-bit default, sh5-*-netbsd* 6586 + SH5, SHmedia, little-endian, 32-bit default, sh5le-*-netbsd* 6587 + SH5, SHmedia, big-endian, 64-bit default, sh64-*-netbsd* 6588 + SH5, SHmedia, little-endian, 64-bit default, sh64le-*-netbsd* 6589 * The following changes have been made to the IA-32/x86-64 port: 6590 + SSE2 and 3dNOW! intrinsics are now supported. 6591 + Support for thread local storage has been added to the IA-32 6592 and x86-64 ports. 6593 + The x86-64 port has been significantly improved. 6594 * The following changes have been made to the MIPS port: 6595 + All configurations now accept the -mabi switch. Note that you 6596 will need appropriate multilibs for this option to work 6597 properly. 6598 + ELF configurations will always pass an ABI flag to the 6599 assembler, except when the MIPS EABI is selected. 6600 + -mabi=64 no longer selects MIPS IV code. 6601 + The -mcpu option, which was deprecated in 3.1 and 3.2, has 6602 been removed from this release. 6603 + -march now changes the core ISA level. In previous releases, 6604 it would change the use of processor-specific extensions, but 6605 would leave the core ISA unchanged. For example, mips64-elf 6606 -march=r8000 will now generate MIPS IV code. 6607 + Under most configurations, -mipsN now acts as a synonym for 6608 -march. 6609 + There are some new preprocessor macros to describe the -march 6610 and -mtune settings. See the documentation of those options 6611 for details. 6612 + Support for the NEC VR-Series processors has been added. This 6613 includes the 54xx, 5500, and 41xx series. 6614 + Support for the Sandcraft sr71k processor has been added. 6615 * The following changes have been made to the S/390 port: 6616 + Support to build the Java runtime libraries has been added. 6617 Java is now enabled by default on s390-*-linux* and 6618 s390x-*-linux* targets. 6619 + Multilib support for the s390x-*-linux* target has been added; 6620 this allows to build 31-bit binaries using the -m31 option. 6621 + Support for thread local storage has been added. 6622 + Inline assembler code may now use the 'Q' constraint to 6623 specify memory operands without index register. 6624 + Various platform-specific performance improvements have been 6625 implemented; in particular, the compiler now uses the BRANCH 6626 ON COUNT family of instructions and makes more frequent use of 6627 the TEST UNDER MASK family of instructions. 6628 * The following changes have been made to the PowerPC port: 6629 + Support for IBM Power4 processor added. 6630 + Support for Motorola e500 SPE added. 6631 + Support for AIX 5.2 added. 6632 + Function and Data sections now supported on AIX. 6633 + Sibcall optimizations added. 6634 * The support for H8 Tiny is added to the H8/300 port with -mn. 6635 6636 Obsolete Systems 6637 6638 Support for a number of older systems has been declared obsolete in GCC 6639 3.3. Unless there is activity to revive them, the next release of GCC 6640 will have their sources permanently removed. 6641 6642 All configurations of the following processor architectures have been 6643 declared obsolete: 6644 * Matsushita MN10200, mn10200-*-* 6645 * Motorola 88000, m88k-*-* 6646 * IBM ROMP, romp-*-* 6647 6648 Also, some individual systems have been obsoleted: 6649 * Alpha 6650 + Interix, alpha*-*-interix* 6651 + Linux libc1, alpha*-*-linux*libc1* 6652 + Linux ECOFF, alpha*-*-linux*ecoff* 6653 * ARM 6654 + Generic a.out, arm*-*-aout* 6655 + Conix, arm*-*-conix* 6656 + "Old ABI," arm*-*-oabi 6657 + StrongARM/COFF, strongarm-*-coff* 6658 * HPPA (PA-RISC) 6659 + Generic OSF, hppa1.0-*-osf* 6660 + Generic BSD, hppa1.0-*-bsd* 6661 + HP/UX versions 7, 8, and 9, hppa1.[01]-*-hpux[789]* 6662 + HiUX, hppa*-*-hiux* 6663 + Mach Lites, hppa*-*-lites* 6664 * Intel 386 family 6665 + Windows NT 3.x, i?86-*-win32 6666 * MC68000 family 6667 + HP systems, m68000-hp-bsd* and m68k-hp-bsd* 6668 + Sun systems, m68000-sun-sunos*, m68k-sun-sunos*, and 6669 m68k-sun-mach* 6670 + AT&T systems, m68000-att-sysv* 6671 + Atari systems, m68k-atari-sysv* 6672 + Motorola systems, m68k-motorola-sysv* 6673 + NCR systems, m68k-ncr-sysv* 6674 + Plexus systems, m68k-plexus-sysv* 6675 + Commodore systems, m68k-cbm-sysv* 6676 + Citicorp TTI, m68k-tti-* 6677 + Unos, m68k-crds-unos* 6678 + Concurrent RTU, m68k-ccur-rtu* 6679 + Linux a.out, m68k-*-linux*aout* 6680 + Linux libc1, m68k-*-linux*libc1* 6681 + pSOS, m68k-*-psos* 6682 * MIPS 6683 + Generic ECOFF, mips*-*-ecoff* 6684 + SINIX, mips-sni-sysv4 6685 + Orion RTEMS, mips64orion-*-rtems* 6686 * National Semiconductor 32000 6687 + OpenBSD, ns32k-*-openbsd* 6688 * POWER (aka RS/6000) and PowerPC 6689 + AIX versions 1, 2, and 3, rs6000-ibm-aix[123]* 6690 + Bull BOSX, rs6000-bull-bosx 6691 + Generic Mach, rs6000-*-mach* 6692 + Generic SysV, powerpc*-*-sysv* 6693 + Linux libc1, powerpc*-*-linux*libc1* 6694 * Sun SPARC 6695 + Generic a.out, sparc-*-aout*, sparclet-*-aout*, 6696 sparclite-*-aout*, and sparc86x-*-aout* 6697 + NetBSD a.out, sparc-*-netbsd*aout* 6698 + Generic BSD, sparc-*-bsd* 6699 + ChorusOS, sparc-*-chorusos* 6700 + Linux a.out, sparc-*-linux*aout* 6701 + Linux libc1, sparc-*-linux*libc1* 6702 + LynxOS, sparc-*-lynxos* 6703 + Solaris on HAL hardware, sparc-hal-solaris2* 6704 + SunOS versions 3 and 4, sparc-*-sunos[34]* 6705 * NEC V850 6706 + RTEMS, v850-*-rtems* 6707 * VAX 6708 + VMS, vax-*-vms* 6709 6710 Documentation improvements 6711 6712 Other significant improvements 6713 6714 * Almost all front-end dependencies in the compiler have been 6715 separated out into a set of language hooks. This should make adding 6716 a new front end clearer and easier. 6717 * One effect of removing the separate preprocessor is a small 6718 increase in the robustness of the compiler in general, and the 6719 maintainability of target descriptions. Previously target-specific 6720 built-in macros and others, such as __FAST_MATH__, had to be 6721 handled with so-called specs that were hard to maintain. Often they 6722 would fail to behave properly when conflicting options were 6723 supplied on the command line, and define macros in the user's 6724 namespace even when strict ISO compliance was requested. 6725 Integrating the preprocessor has cleanly solved these issues. 6726 * The Makefile suite now supports redirection of make install by 6727 means of the variable DESTDIR. 6728 __________________________________________________________________ 6729 6730 GCC 3.3 6731 6732 Detailed release notes for the GCC 3.3 release follow. 6733 6734 Bug Fixes 6735 6736 bootstrap failures 6737 6738 * [8]10140 cross compiler build failures: missing __mempcpy (DUP: 6739 [9]10198,[10]10338) 6740 6741 Internal compiler errors (multi-platform) 6742 6743 * [11]3581 large string causes segmentation fault in cc1 6744 * [12]4382 __builtin_{set,long}jmp with -O3 can crash the compiler 6745 * [13]5533 (c++) ICE when processing std::accumulate(begin, end, 6746 init, invalid_op) 6747 * [14]6387 -fpic -gdwarf-2 -g1 combination gives ICE in dwarf2out 6748 * [15]6412 (c++) ICE in retrieve_specialization 6749 * [16]6620 (c++) partial template specialization causes an ICE 6750 (segmentation fault) 6751 * [17]6663 (c++) ICE with attribute aligned 6752 * [18]7068 ICE with incomplete types 6753 * [19]7083 (c++) ICE using -gstabs with dodgy class derivation 6754 * [20]7647 (c++) ICE when data member has the name of the enclosing 6755 class 6756 * [21]7675 ICE in fixup_var_refs_1 6757 * [22]7718 'complex' template instantiation causes ICE 6758 * [23]8116 (c++) ICE in member template function 6759 * [24]8358 (ada) Ada compiler accesses freed memory, crashes 6760 * [25]8511 (c++) ICE: (hopefully) reproducible cc1plus segmentation 6761 fault 6762 * [26]8564 (c++) ICE in find_function_data, in function.c 6763 * [27]8660 (c++) template overloading ICE in tsubst_expr, in cp/pt.c 6764 * [28]8766 (c++) ICE after failed initialization of static template 6765 variable 6766 * [29]8803 ICE in instantiate_virtual_regs_1, in function.c 6767 * [30]8846 (c++) ICE after diagnostic if fr_FR@euro locale is set 6768 * [31]8906 (c++) ICE (Segmentation fault) when parsing nested-class 6769 definition 6770 * [32]9216 (c++) ICE on missing template parameter 6771 * [33]9261 (c++) ICE in arg_assoc, in cp/decl2.c 6772 * [34]9263 (fortran) ICE caused by invalid PARAMETER in implied DO 6773 loop 6774 * [35]9429 (c++) ICE in template instantiation with a pointered new 6775 operator 6776 * [36]9516 Internal error when using a big array 6777 * [37]9600 (c++) ICE with typedefs in template class 6778 * [38]9629 (c++) virtual inheritance segfault 6779 * [39]9672 (c++) ICE: Error reporting routines re-entered 6780 * [40]9749 (c++) ICE in write_expression on invalid function 6781 prototype 6782 * [41]9794 (fortran) ICE: floating point exception during constant 6783 folding 6784 * [42]9829 (c++) Missing colon in nested namespace usage causes ICE 6785 * [43]9916 (c++) ICE with noreturn function in ?: statement 6786 * [44]9936 ICE with local function and variable-length 2d array 6787 * [45]10262 (c++) cc1plus crashes with large generated code 6788 * [46]10278 (c++) ICE in parser for invalid code 6789 * [47]10446 (c++) ICE on definition of nonexistent member function of 6790 nested class in a class template 6791 * [48]10451 (c++) ICE in grokdeclarator on spurious mutable 6792 declaration 6793 * [49]10506 (c++) ICE in build_new at cp/init.c with 6794 -fkeep-inline-functions and multiple inheritance 6795 * [50]10549 (c++) ICE in store_bit_field on bitfields that exceed the 6796 precision of the declared type 6797 6798 Optimization bugs 6799 6800 * [51]2001 Inordinately long compile times in reload CSE regs 6801 * [52]2391 Exponential compilation time explosion in combine 6802 * [53]2960 Duplicate loop conditions even with -Os 6803 * [54]4046 redundant conditional branch 6804 * [55]6405 Loop-unrolling related performance regressions 6805 * [56]6798 very long compile time with large case-statement 6806 * [57]6871 const objects shouldn't be moved to .bss 6807 * [58]6909 problem w/ -Os on modified loop-2c.c test case 6808 * [59]7189 gcc -O2 -Wall does not print ``control reaches end of 6809 non-void function'' warning 6810 * [60]7642 optimization problem with signbit() 6811 * [61]8634 incorrect code for inlining of memcpy under -O2 6812 * [62]8750 Cygwin prolog generation erroneously emitting __alloca as 6813 regular function call 6814 6815 C front end 6816 6817 * [63]2161 long if-else cascade overflows parser stack 6818 * [64]4319 short accepted on typedef'd char 6819 * [65]8602 incorrect line numbers in warning messages when using 6820 inline functions 6821 * [66]9177 -fdump-translation-unit: C front end deletes function_decl 6822 AST nodes and breaks debugging dumps 6823 * [67]9853 miscompilation of non-constant structure initializer 6824 6825 c++ compiler and library 6826 6827 * [68]45 legal template specialization code is rejected (DUP: 6828 [69]3784) 6829 * [70]764 lookup failure: friend operator and dereferencing a pointer 6830 and templates (DUP: [71]5116) 6831 * [72]2862 gcc accepts invalid explicit instantiation syntax (DUP: 6832 2863) 6833 * [73]3663 G++ doesn't check access control during template 6834 instantiation 6835 * [74]3797 gcc fails to emit explicit specialization of a template 6836 member 6837 * [75]3948 Two destructors are called when no copy destructor is 6838 defined (ABI change) 6839 * [76]4137 Conversion operator within template is not accepted 6840 * [77]4361 bogus ambiguity taking the address of a member template 6841 * [78]4802 g++ accepts illegal template code (access to private 6842 member; DUP: [79]5837) 6843 * [80]4803 inline function is used but never defined, and g++ does 6844 not object 6845 * [81]5094 Partial specialization cannot be friend? 6846 * [82]5730 complex<double>::norm() -- huge slowdown from egcs-2.91.66 6847 * [83]6713 Regression wrt 3.0.4: g++ -O2 leads to seg fault at run 6848 time 6849 * [84]7015 certain __asm__ constructs rejected 6850 * [85]7086 compile time regression (quadratic behavior in 6851 fixup_var_refs) 6852 * [86]7099 G++ doesn't set the noreturn attribute on std::exit and 6853 std::abort 6854 * [87]7247 copy constructor missing when inlining enabled (invalid 6855 optimization?) 6856 * [88]7441 string array initialization compilation time regression 6857 from seconds to minutes 6858 * [89]7768 __PRETTY_FUNCTION__ for template destructor is wrong 6859 * [90]7804 bad printing of floating point constant in warning message 6860 * [91]8099 Friend classes and template specializations 6861 * [92]8117 member function pointers and multiple inheritance 6862 * [93]8205 using declaration and multiple inheritance 6863 * [94]8645 unnecessary non-zero checks in stl_tree.h 6864 * [95]8724 explicit destructor call for incomplete class allowed 6865 * [96]8805 compile time regression with many member variables 6866 * [97]8691 -O3 and -fno-implicit-templates are incompatible 6867 * [98]8700 unhelpful error message for binding temp to reference 6868 * [99]8724 explicit destructor call for incomplete class allowed 6869 * [100]8949 numeric_limits<>::denorm_min() and is_iec559 problems 6870 * [101]9016 Failure to consistently constant fold "constant" C++ 6871 objects 6872 * [102]9053 g++ confused about ambiguity of overloaded function 6873 templates 6874 * [103]9152 undefined virtual thunks 6875 * [104]9182 basic_filebuf<> does not report errors in codecvt<>::out 6876 * [105]9297 data corruption due to codegen bug (when copying.) 6877 * [106]9318 i/ostream::operator>>/<<(streambuf*) broken 6878 * [107]9320 Incorrect usage of traits_type::int_type in stdio_filebuf 6879 * [108]9400 bogus -Wshadow warning: shadowed declaration of this in 6880 local classes 6881 * [109]9424 i/ostream::operator>>/<<(streambuf*) drops characters 6882 * [110]9425 filebuf::pbackfail broken (DUP: [111]9439) 6883 * [112]9474 GCC freezes in compiling a weird code mixing <iostream> 6884 and <iostream.h> 6885 * [113]9548 Incorrect results from setf(ios::fixed) and precision(-1) 6886 [114][DR 231] 6887 * [115]9555 ostream inserters fail to set badbit on exception 6888 * [116]9561 ostream inserters rethrow exception of wrong type 6889 * [117]9563 ostream::sentry returns true after a failed preparation 6890 * [118]9582 one-definition rule violation in std::allocator 6891 * [119]9622 __PRETTY_FUNCTION__ incorrect in template destructors 6892 * [120]9683 bug in initialization chains for static const variables 6893 from template classes 6894 * [121]9791 -Woverloaded-virtual reports hiding of destructor 6895 * [122]9817 collate::compare doesn't handle nul characters 6896 * [123]9825 filebuf::sputbackc breaks sbumpc 6897 * [124]9826 operator>>(basic_istream, basic_string) fails to compile 6898 with custom traits 6899 * [125]9924 Multiple using statements for builtin functions not 6900 allowed 6901 * [126]9946 destructor is not called for temporary object 6902 * [127]9964 filebuf::close() sometimes fails to close file 6903 * [128]9988 filebuf::overflow writes EOF to file 6904 * [129]10033 optimization breaks polymorphic references w/ typeid 6905 operator 6906 * [130]10097 filebuf::underflow drops characters 6907 * [131]10132 filebuf destructor can throw exceptions 6908 * [132]10180 gcc fails to warn about non-inlined function 6909 * [133]10199 method parametrized by template does not work everywhere 6910 * [134]10300 use of array-new (nothrow) in segfaults on NULL return 6911 * [135]10427 Stack corruption with variable-length automatic arrays 6912 and virtual destructors 6913 * [136]10503 Compilation never stops in fixed_type_or_null 6914 6915 Objective-C 6916 6917 * [137]5956 selectors aren't matched properly when added to the 6918 selector table 6919 6920 Fortran compiler and library 6921 6922 * [138]1832 list directed i/o overflow hangs, -fbounds-check doesn't 6923 detect 6924 * [139]3924 g77 generates code that is rejected by GAS if COFF debug 6925 info requested 6926 * [140]5634 doc: explain that configure --prefix=~/... does not work 6927 * [141]6367 multiple repeat counts confuse namelist read into array 6928 * [142]6491 Logical operations error on logicals when using 6929 -fugly-logint 6930 * [143]6742 Generation of C++ Prototype for FORTRAN and extern "C" 6931 * [144]7113 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os 6932 on irix6.5 6933 * [145]7236 OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should 6934 assume a direct access file 6935 * [146]7278 g77 "bug"; the executable misbehaves (with -O2 6936 -fno-automatic) 6937 * [147]7384 DATE_AND_TIME milliseconds field inactive on Windows 6938 * [148]7388 Incorrect output with 0-based array of characters 6939 * [149]8587 Double complex zero ** double precision number -> NaN 6940 instead of zero 6941 * [150]9038 -ffixed-line-length-none -x f77-cpp-input gives: Warning: 6942 unknown register name line-length-none 6943 * [151]10197 Direct access files not unformatted by default 6944 6945 Java compiler and library 6946 6947 * [152]6005 gcj fails to build rhug on alpha 6948 * [153]6389 System.getProperty("") should always throw an 6949 IllegalArgumentException 6950 * [154]6576 java.util.ResourceBundle.getResource ignores locale 6951 * [155]6652 new java.io.File("").getCanonicalFile() throws exception 6952 * [156]7060 getMethod() doesn't search super interface 6953 * [157]7073 bytecode interpreter gives wrong answer for interface 6954 getSuperclass() 6955 * [158]7180 possible bug in 6956 javax.naming.spi.NamingManager.getPlusPath() 6957 * [159]7416 java.security startup refs "GNU libgcj.security" 6958 * [160]7570 Runtime.exec with null envp: child doesn't inherit parent 6959 env (DUP: [161]7578) 6960 * [162]7611 Internal error while compiling libjava with -O 6961 * [163]7709 NullPointerException in _Jv_ResolvePoolEntry 6962 * [164]7766 ZipInputStream.available returns 0 immediately after 6963 construction 6964 * [165]7785 Calendar.getTimeInMillis/setTimeInMillis should be public 6965 * [166]7786 TimeZone.getDSTSavings() from JDK1.4 not implemented 6966 * [167]8142 '$' in class names vs. dlopen 'dynamic string tokens' 6967 * [168]8234 ZipInputStream chokes when InputStream.read() returns 6968 small chunks 6969 * [169]8415 reflection bug: exception info for Method 6970 * [170]8481 java.Random.nextInt(int) may return negative 6971 * [171]8593 Error reading GZIPped files with BufferedReader 6972 * [172]8759 java.beans.Introspector has no flushCaches() or 6973 flushFromCaches() methods 6974 * [173]8997 spin() calls Thread.sleep 6975 * [174]9253 on win32, java.io.File.listFiles("C:\\") returns pwd 6976 instead of the root content of C: 6977 * [175]9254 java::lang::Object::wait(), threads-win32.cc returns 6978 wrong return codes 6979 * [176]9271 Severe bias in java.security.SecureRandom 6980 6981 Ada compiler and library 6982 6983 * [177]6767 make gnatlib-shared fails on -laddr2line 6984 * [178]9911 gnatmake fails to link when GCC configured with 6985 --with-sjlj-exceptions=yes 6986 * [179]10020 Can't bootstrap gcc on AIX with Ada enabled 6987 * [180]10546 Ada tasking not working on Red Hat 9 6988 6989 preprocessor 6990 6991 * [181]7029 preprocessor should ignore #warning with -M 6992 6993 ARM-specific 6994 6995 * [182]2903 [arm] Optimization bug with long long arithmetic 6996 * [183]7873 arm-linux-gcc fails when assigning address to a bit field 6997 6998 FreeBSD-specific 6999 7000 * [184]7680 float functions undefined in math.h/cmath with #define 7001 _XOPEN_SOURCE 7002 7003 HP-UX or HP-PA-specific 7004 7005 * [185]8705 [HP-PA] ICE in emit_move_insn_1, in expr.c 7006 * [186]9986 [HP-UX] Incorrect transformation of fputs_unlocked to 7007 fputc_unlocked 7008 * [187]10056 [HP-PA] ICE at -O2 when building c++ code from doxygen 7009 7010 m68hc11-specific 7011 7012 * [188]6744 Bad assembler code generated: reference to pseudo 7013 register z 7014 * [189]7361 Internal compiler error in reload_cse_simplify_operands, 7015 in reload1.c 7016 7017 MIPS-specific 7018 7019 * [190]9496 [mips-linux] bug in optimizer? 7020 7021 PowerPC-specific 7022 7023 * [191]7067 -Os with -mcpu=powerpc optimizes for speed (?) instead of 7024 space 7025 * [192]8480 reload ICEs for LAPACK code on powerpc64-linux 7026 * [193]8784 [AIX] Internal compiler error in simplify_gen_subreg 7027 * [194]10315 [powerpc] ICE: in extract_insn, in recog.c 7028 7029 SPARC-specific 7030 7031 * [195]10267 (documentation) Wrong build instructions for 7032 *-*-solaris2* 7033 7034 x86-specific (Intel/AMD) 7035 7036 * [196]7916 ICE in instantiate_virtual_register_1 7037 * [197]7926 (c++) i486 instructions in header files make c++ programs 7038 crash on i386 7039 * [198]8555 ICE in gen_split_1231 7040 * [199]8994 ICE with -O -march=pentium4 7041 * [200]9426 ICE with -fssa -funroll-loops -fprofile-arcs 7042 * [201]9806 ICE in inline assembly with -fPIC flag 7043 * [202]10077 gcc -msse2 generates movd to move dwords between xmm 7044 regs 7045 * [203]10233 64-bit comparison only comparing bottom 32-bits 7046 * [204]10286 type-punning doesn't work with __m64 and -O 7047 * [205]10308 [x86] ICE with -O -fgcse or -O2 7048 __________________________________________________________________ 7049 7050 GCC 3.3.1 7051 7052 Bug Fixes 7053 7054 This section lists the problem reports (PRs) from GCC's bug tracking 7055 system that are known to be fixed in the 3.3.1 release. This list might 7056 not be complete (that is, it is possible that some PRs that have been 7057 fixed are not listed here). 7058 7059 Bootstrap failures 7060 7061 * [206]11272 [Solaris] make bootstrap fails while building libstdc++ 7062 7063 Internal compiler errors (multi-platform) 7064 7065 * [207]5754 ICE on invalid nested template class 7066 * [208]6597 ICE in set_mem_alias_set compiling Qt with -O2 on ia64 7067 and --enable-checking 7068 * [209]6949 (c++) ICE in tsubst_decl, in cp/pt.c 7069 * [210]7053 (c++) ICE when declaring a function already defined as a 7070 friend method of a template class 7071 * [211]8164 (c++) ICE when using different const expressions as 7072 template parameter 7073 * [212]8384 (c++) ICE in is_base_type, in dwarf2out.c 7074 * [213]9559 (c++) ICE with invalid initialization of a static const 7075 * [214]9649 (c++) ICE in finish_member_declaration, in cp/semantics.c 7076 when redeclaring a static member variable 7077 * [215]9864 (fortran) ICE in add_abstract_origin_attribute, in 7078 dwarfout.c with -g -O -finline-functions 7079 * [216]10432 (c++) ICE in poplevel, in cp/decl.c 7080 * [217]10475 ICE in subreg_highpart_offset for code with long long 7081 * [218]10635 (c++) ICE when dereferencing an incomplete type casted 7082 from a void pointer 7083 * [219]10661 (c++) ICE in instantiate_decl, in cp/pt.c while 7084 instantiating static member variables 7085 * [220]10700 ICE in copy_to_mode_reg on 64-bit targets 7086 * [221]10712 (c++) ICE in constructor_name_full, in cp/decl2.c 7087 * [222]10796 (c++) ICE when defining an enum with two values: -1 and 7088 MAX_INT_64BIT 7089 * [223]10890 ICE in merge_assigned_reloads building Linux 2.4.2x 7090 sched.c 7091 * [224]10939 (c++) ICE with template code 7092 * [225]10956 (c++) ICE when specializing a template member function 7093 of a template class, in tsubst, in cp/pt.c 7094 * [226]11041 (c++) ICE: const myclass &x = *x; (when operator*() 7095 defined) 7096 * [227]11059 (c++) ICE with empty union 7097 * [228]11083 (c++) ICE in commit_one_edge_insertion, in cfgrtl.c with 7098 -O2 -fnon-call-exceptions 7099 * [229]11105 (c++) ICE in mangle_conv_op_name_for_type 7100 * [230]11149 (c++) ICE on error when instantiation with call function 7101 of a base type 7102 * [231]11228 (c++) ICE on new-expression using array operator new and 7103 default-initialization 7104 * [232]11282 (c++) Infinite memory usage after syntax error 7105 * [233]11301 (fortran) ICE with -fno-globals 7106 * [234]11308 (c++) ICE when using an enum type name as if it were a 7107 class or namespace 7108 * [235]11473 (c++) ICE with -gstabs when empty struct inherits from 7109 an empty struct 7110 * [236]11503 (c++) ICE when instantiating template with ADDR_EXPR 7111 * [237]11513 (c++) ICE in push_template_decl_real, in cp/pt.c: 7112 template member functions 7113 7114 Optimization bugs 7115 7116 * [238]11198 -O2 -frename-registers generates wrong code (aliasing 7117 problem) 7118 * [239]11304 Wrong code production with -fomit-frame-pointer 7119 * [240]11381 volatile memory access optimized away 7120 * [241]11536 [strength-reduce] -O2 optimization produces wrong code 7121 * [242]11557 constant folding bug generates wrong code 7122 7123 C front end 7124 7125 * [243]5897 No warning for statement after return 7126 * [244]11279 DWARF-2 output mishandles large enums 7127 7128 Preprocessor bugs 7129 7130 * [245]11022 no warning for non-compatible macro redefinition 7131 7132 C++ compiler and library 7133 7134 * [246]2330 static_cast<>() to a private base is allowed 7135 * [247]5388 Incorrect message "operands to ?: have different types" 7136 * [248]5390 Libiberty fails to demangle multi-digit template 7137 parameters 7138 * [249]7877 Incorrect parameter passing to specializations of member 7139 function templates 7140 * [250]9393 Anonymous namespaces and compiling the same file twice 7141 * [251]10032 -pedantic converts some errors to warnings 7142 * [252]10468 const typeof(x) is non-const, but only in templates 7143 * [253]10527 confused error message with "new int()" parameter 7144 initializer 7145 * [254]10679 parameter MIN_INLINE_INSNS is not honored 7146 * [255]10682 gcc chokes on a typedef for an enum inside a class 7147 template 7148 * [256]10689 pow(std::complex(0),1/3) returns (nan, nan) instead of 7149 0. 7150 * [257]10845 template member function (with nested template as 7151 parameter) cannot be called anymore if another unrelated template 7152 member function is defined 7153 * [258]10849 Cannot define an out-of-class specialization of a 7154 private nested template class 7155 * [259]10888 Suppress -Winline warnings for system headers 7156 * [260]10929 -Winline warns about functions for which no definition 7157 is visible 7158 * [261]10931 valid conversion static_cast<const unsigned 7159 int&>(lvalue-of-type-int) is rejected 7160 * [262]10940 Bad code with explicit specialization 7161 * [263]10968 If member function implicitly instantiated, explicit 7162 instantiation of class fails to instantiate it 7163 * [264]10990 Cannot convert with dynamic_cast<> to a private base 7164 class from within a member function 7165 * [265]11039 Bad interaction between implicit typename deprecation 7166 and friendship 7167 * [266]11062 (libstdc++) avoid __attribute__ ((unused)); say 7168 "__unused__" instead 7169 * [267]11095 C++ iostream manipulator causes segfault when called 7170 with negative argument 7171 * [268]11098 g++ doesn't emit complete debugging information for 7172 local variables in destructors 7173 * [269]11137 Linux shared library constructors not called unless 7174 there's one global object 7175 * [270]11154 spurious ambiguity report for template class 7176 specialization 7177 * [271]11329 Compiler cannot find user defined implicit typecast 7178 * [272]11332 Spurious error with casts in ?: expression 7179 * [273]11431 static_cast behavior with subclasses when default 7180 constructor available 7181 * [274]11528 money_get facet does not accept "$.00" as valid 7182 * [275]11546 Type lookup problems in out-of-line definition of a 7183 class doubly nested from a template class 7184 * [276]11567 C++ code containing templated member function with same 7185 name as pure virtual member function results in linking failure 7186 * [277]11645 Failure to deal with using and private inheritance 7187 7188 Java compiler and library 7189 7190 * [278]5179 Qualified static field access doesn't initialize its 7191 class 7192 * [279]8204 gcj -O2 to native reorders certain instructions 7193 improperly 7194 * [280]10838 java.io.ObjectInputStream syntax error 7195 * [281]10886 The RMI registry that comes with GCJ does not work 7196 correctly 7197 * [282]11349 JNDI URL context factories not located correctly 7198 7199 x86-specific (Intel/AMD) 7200 7201 * [283]4823 ICE on inline assembly code 7202 * [284]8878 miscompilation with -O and SSE 7203 * [285]9815 (c++ library) atomicity.h - fails to compile with -O3 7204 -masm=intel 7205 * [286]10402 (inline assembly) [x86] ICE in merge_assigned_reloads, 7206 in reload1.c 7207 * [287]10504 ICE with SSE2 code and -O3 -mcpu=pentium4 -msse2 7208 * [288]10673 ICE for x86-64 on freebsd libc vfprintf.c source 7209 * [289]11044 [x86] out of range loop instructions for FP code on K6 7210 * [290]11089 ICE: instantiate_virtual_regs_lossage while using SSE 7211 built-ins 7212 * [291]11420 [x86_64] gcc generates invalid asm code when "-O -fPIC" 7213 is used 7214 7215 SPARC- or Solaris- specific 7216 7217 * [292]9362 solaris 'as' dies when fed .s and "-gstabs" 7218 * [293]10142 [SPARC64] gcc produces wrong code when passing 7219 structures by value 7220 * [294]10663 New configure check aborts with Sun tools. 7221 * [295]10835 combinatorial explosion in scheduler on HyperSPARC 7222 * [296]10876 ICE in calculate_giv_inc when building KDE 7223 * [297]10955 wrong code at -O3 for structure argument in context of 7224 structure return 7225 * [298]11018 -mcpu=ultrasparc busts tar-1.13.25 7226 * [299]11556 [sparc64] ICE in gen_reg_rtx() while compiling 2.6.x 7227 Linux kernel 7228 7229 ia64 specific 7230 7231 * [300]10907 gcc violates the ia64 ABI (GP must be preserved) 7232 * [301]11320 scheduler bug (in machine depended reorganization pass) 7233 * [302]11599 bug with conditional and __builtin_prefetch 7234 7235 PowerPC specific 7236 7237 * [303]9745 [powerpc] gcc mis-compiles libmcrypt (alias problem 7238 during loop) 7239 * [304]10871 error in rs6000_stack_info save_size computation 7240 * [305]11440 gcc mis-compiles c++ code (libkhtml) with -O2, -fno-gcse 7241 cures it 7242 7243 m68k-specific 7244 7245 * [306]7594 [m68k] ICE on legal code associated with simplify-rtx 7246 * [307]10557 [m68k] ICE in subreg_offset_representable_p 7247 * [308]11054 [m68k] ICE in reg_overlap_mentioned_p 7248 7249 ARM-specific 7250 7251 * [309]10834 [arm] GCC 3.3 still generates incorrect instructions for 7252 functions with __attribute__ ((interrupt ("IRQ"))) 7253 * [310]10842 [arm] Clobbered link register is copied to pc under 7254 certain circumstances 7255 * [311]11052 [arm] noce_process_if_block() can lose REG_INC notes 7256 * [312]11183 [arm] ICE in change_address_1 (3.3) / subreg_hard_regno 7257 (3.4) 7258 7259 MIPS-specific 7260 7261 * [313]11084 ICE in propagate_one_insn, in flow.c 7262 7263 SH-specific 7264 7265 * [314]10331 can't compile c++ part of gcc cross compiler for sh-elf 7266 * [315]10413 [SH] ICE in reload_cse_simplify_operands, in reload1.c 7267 * [316]11096 i686-linux to sh-linux cross compiler fails to compile 7268 C++ files 7269 7270 GNU/Linux (or Hurd?) specific 7271 7272 * [317]2873 Bogus fixinclude of stdio.h from glibc 2.2.3 7273 7274 UnixWare specific 7275 7276 * [318]3163 configure bug: gcc/aclocal.m4 mmap test fails on UnixWare 7277 7.1.1 7278 7279 Cygwin (or mingw) specific 7280 7281 * [319]5287 ICE with dllimport attribute 7282 * [320]10148 [MingW/CygWin] Compiler dumps core 7283 7284 DJGPP specific 7285 7286 * [321]8787 GCC fails to emit .intel_syntax when invoked with 7287 -masm=intel on DJGPP 7288 7289 Darwin (and MacOS X) specific 7290 7291 * [322]10900 trampolines crash 7292 7293 Documentation 7294 7295 * [323]1607 (c++) Format attributes on methods undocumented 7296 * [324]4252 Invalid option `-fdump-translation-unit' 7297 * [325]4490 Clarify restrictions on -m96bit-long-double, 7298 -m128bit-long-double 7299 * [326]10355 document an issue with regparm attribute on some systems 7300 (e.g. Solaris) 7301 * [327]10726 (fortran) Documentation for function "IDate Intrinsic 7302 (Unix)" is wrong 7303 * [328]10805 document bug in old version of Sun assembler 7304 * [329]10815 warn against GNU binutils on AIX 7305 * [330]10877 document need for newer binutils on i?86-*-linux-gnu 7306 * [331]11280 Manual incorrect with respect to -freorder-blocks 7307 * [332]11466 Document -mlittle-endian and its restrictions for the 7308 sparc64 port 7309 7310 Testsuite bugs (compiler itself is not affected) 7311 7312 * [333]10737 newer bison causes g++.dg/parse/crash2.C to incorrectly 7313 report failure 7314 * [334]10810 gcc-3.3 fails make check: buffer overrun in 7315 test_demangle.c 7316 __________________________________________________________________ 7317 7318 GCC 3.3.2 7319 7320 Bug Fixes 7321 7322 This section lists the problem reports (PRs) from [335]GCC's bug 7323 tracking system that are known to be fixed in the 3.3.2 release. This 7324 list might not be complete (that is, it is possible that some PRs that 7325 have been fixed are not listed here). 7326 7327 Bootstrap failures and problems 7328 7329 * [336]8336 [SCO5] bootstrap config still tries to use COFF options 7330 * [337]9330 [alpha-osf] Bootstrap failure on Compaq Tru64 with 7331 --enable-threads=posix 7332 * [338]9631 [hppa64-linux] gcc-3.3 fails to bootstrap 7333 * [339]9877 fixincludes makes a bad sys/byteorder.h on svr5 (UnixWare 7334 7.1.1) 7335 * [340]11687 xstormy16-elf build fails in libf2c 7336 * [341]12263 [SGI IRIX] bootstrap fails during compile of 7337 libf2c/libI77/backspace.c 7338 * [342]12490 buffer overflow in scan-decls.c (during Solaris 9 7339 fix-header processing) 7340 7341 Internal compiler errors (multi-platform) 7342 7343 * [343]7277 Casting integers to vector types causes ICE 7344 * [344]7939 (c++) ICE on invalid function template specialization 7345 * [345]11063 (c++) ICE on parsing initialization list of const array 7346 member 7347 * [346]11207 ICE with negative index in array element designator 7348 * [347]11522 (fortran) g77 dwarf-2 ICE in 7349 add_abstract_origin_attribute 7350 * [348]11595 (c++) ICE on duplicate label definition 7351 * [349]11646 (c++) ICE in commit_one_edge_insertion with 7352 -fnon-call-exceptions -fgcse -O 7353 * [350]11665 ICE in struct initializer when taking address 7354 * [351]11852 (c++) ICE with bad struct initializer. 7355 * [352]11878 (c++) ICE in cp_expr_size 7356 * [353]11883 ICE with any -O on mercury-generated C code 7357 * [354]11991 (c++) ICE in cxx_incomplete_type_diagnostic, in 7358 cp/typeck2.c when applying typeid operator to template template 7359 parameter 7360 * [355]12146 ICE in lookup_template_function, in cp/pt.c 7361 * [356]12215 ICE in make_label_edge with -fnon-call-exceptions 7362 -fno-gcse -O2 7363 * [357]12369 (c++) ICE with templates and friends 7364 * [358]12446 ICE in emit_move_insn on complicated array reference 7365 * [359]12510 ICE in final_scan_insn 7366 * [360]12544 ICE with large parameters used in nested functions 7367 7368 C and optimization bugs 7369 7370 * [361]9862 spurious warnings with -W -finline-functions 7371 * [362]10962 lookup_field is a linear search on a linked list (can be 7372 slow if large struct) 7373 * [363]11370 -Wunreachable-code gives false complaints 7374 * [364]11637 invalid assembly with -fnon-call-exceptions 7375 * [365]11885 Problem with bitfields in packed structs 7376 * [366]12082 Inappropriate unreachable code warnings 7377 * [367]12180 Inline optimization fails for variadic function 7378 * [368]12340 loop unroller + gcse produces wrong code 7379 7380 C++ compiler and library 7381 7382 * [369]3907 nested template parameter collides with member name 7383 * [370]5293 confusing message when binding a temporary to a reference 7384 * [371]5296 [DR115] Pointers to functions and to template functions 7385 behave differently in deduction 7386 * [372]7939 ICE on function template specialization 7387 * [373]8656 Unable to assign function with __attribute__ and pointer 7388 return type to an appropriate variable 7389 * [374]10147 Confusing error message for invalid template function 7390 argument 7391 * [375]11400 std::search_n() makes assumptions about Size parameter 7392 * [376]11409 issues with using declarations, overloading, and 7393 built-in functions 7394 * [377]11740 ctype<wchar_t>::do_is(mask, wchar_t) doesn't handle 7395 multiple bits in mask 7396 * [378]11786 operator() call on variable in other namespace not 7397 recognized 7398 * [379]11867 static_cast ignores ambiguity 7399 * [380]11928 bug with conversion operators that are typedefs 7400 * [381]12114 Uninitialized memory accessed in dtor 7401 * [382]12163 static_cast + explicit constructor regression 7402 * [383]12181 Wrong code with comma operator and c++ 7403 * [384]12236 regparm and fastcall messes up parameters 7404 * [385]12266 incorrect instantiation of unneeded template during 7405 overload resolution 7406 * [386]12296 istream::peek() doesn't set eofbit 7407 * [387]12298 [sjlj exceptions] Stack unwind destroys 7408 not-yet-constructed object 7409 * [388]12369 ICE with templates and friends 7410 * [389]12337 apparently infinite loop in g++ 7411 * [390]12344 stdcall attribute ignored if function returns a pointer 7412 * [391]12451 missing(late) class forward declaration in cxxabi.h 7413 * [392]12486 g++ accepts invalid use of a qualified name 7414 7415 x86 specific (Intel/AMD) 7416 7417 * [393]8869 [x86 MMX] ICE with const variable optimization and MMX 7418 builtins 7419 * [394]9786 ICE in fixup_abnormal_edges with -fnon-call-exceptions 7420 -O2 7421 * [395]11689 g++3.3 emits un-assembleable code for k6 architecture 7422 * [396]12116 [k6] Invalid assembly output values with X-MAME code 7423 * [397]12070 ICE converting between double and long double with 7424 -msoft-float 7425 7426 ia64-specific 7427 7428 * [398]11184 [ia64 hpux] ICE on __builtin_apply building libobjc 7429 * [399]11535 __builtin_return_address may not work on ia64 7430 * [400]11693 [ia64] ICE in gen_nop_type 7431 * [401]12224 [ia64] Thread-local storage doesn't work 7432 7433 PowerPC-specific 7434 7435 * [402]11087 [powerpc64-linux] GCC miscompiles raid1.c from linux 7436 kernel 7437 * [403]11319 loop miscompiled on ppc32 7438 * [404]11949 ICE Compiler segfault with ffmpeg -maltivec code 7439 7440 SPARC-specific 7441 7442 * [405]11662 wrong code for expr. with cast to long long and 7443 exclusive or 7444 * [406]11965 invalid assembler code for a shift < 32 operation 7445 * [407]12301 (c++) stack corruption when a returned expression throws 7446 an exception 7447 7448 Alpha-specific 7449 7450 * [408]11717 [alpha-linux] unrecognizable insn compiling for.c of 7451 kernel 2.4.22-pre8 7452 7453 HPUX-specific 7454 7455 * [409]11313 problem with #pragma weak and static inline functions 7456 * [410]11712 __STDC_EXT__ not defined for C++ by default anymore? 7457 7458 Solaris specific 7459 7460 * [411]12166 Profiled programs crash if PROFDIR is set 7461 7462 Solaris-x86 specific 7463 7464 * [412]12101 i386 Solaris no longer works with GNU as? 7465 7466 Miscellaneous embedded target-specific bugs 7467 7468 * [413]10988 [m32r-elf] wrong blockmove code with -O3 7469 * [414]11805 [h8300-unknown-coff] [H8300] ICE for simple code with 7470 -O2 7471 * [415]11902 [sh4] spec file improperly inserts rpath even when none 7472 needed 7473 * [416]11903 [sh4] -pthread fails to link due to error in spec file 7474 on sh4 7475 __________________________________________________________________ 7476 7477 GCC 3.3.3 7478 7479 Minor features 7480 7481 In addition to the bug fixes documented below, this release contains 7482 few minor features such as: 7483 * Support for --with-sysroot 7484 * Support for automatic detection of executable stacks 7485 * Support for SSE3 instructions 7486 * Support for thread local storage debugging under GDB on S390 7487 7488 Bug Fixes 7489 7490 This section lists the problem reports (PRs) from [417]GCC's bug 7491 tracking system that are known to be fixed in the 3.3.3 release. This 7492 list might not be complete (that is, it is possible that some PRs that 7493 have been fixed are not listed here). 7494 7495 Bootstrap failures and issues 7496 7497 * [418]11890 Building cross gcc-3.3.1 for sparc-sun-solaris2.6 fails 7498 * [419]12399 boehm-gc fails (when building a cross compiler): libtool 7499 unable to infer tagged configuration 7500 * [420]13068 mklibgcc.in doesn't handle multi-level multilib 7501 subdirectories properly 7502 7503 Internal compiler errors (multi-platform) 7504 7505 * [421]10060 ICE (stack overflow) on huge file (300k lines) due to 7506 recursive behaviour of copy_rtx_if_shared, in emit_rtl.c 7507 * [422]10555 (c++) ICE on undefined template argument 7508 * [423]10706 (c++) ICE in mangle_class_name_for_template 7509 * [424]11496 (fortran) error in flow_loops_find when -funroll-loops 7510 active 7511 * [425]11741 ICE in pre_insert_copy_insn, in gcse.c 7512 * [426]12440 GCC crashes during compilation of quicktime4linux 2.0.0 7513 * [427]12632 (fortran) -fbounds-check ICE 7514 * [428]12712 (c++) ICE on short legit C++ code fragment with gcc 7515 3.3.2 7516 * [429]12726 (c++) ICE (segfault) on trivial code 7517 * [430]12890 (c++) ICE on compilation of class with throwing method 7518 * [431]12900 (c++) ICE in rtl_verify_flow_info_1 7519 * [432]13060 (fortran) ICE in fixup_var_refs_1, in function.c on 7520 correct code with -O2 -fno-force-mem 7521 * [433]13289 (c++) ICE in regenerate_decl_from_template on recursive 7522 template 7523 * [434]13318 ICE: floating point exception in the loop optimizer 7524 * [435]13392 (c++) ICE in convert_from_eh_region_ranges_1, in 7525 except.c 7526 * [436]13574 (c++) invalid array default initializer in class lets 7527 gcc consume all memory and die 7528 * [437]13475 ICE on SIMD variables with partial value initialization 7529 * [438]13797 (c++) ICE on invalid template parameter 7530 * [439]13824 (java) gcj SEGV with simple .java program 7531 7532 C and optimization bugs 7533 7534 * [440]8776 loop invariants are not removed (most likely) 7535 * [441]10339 [sparc,ppc,ppc64] Invalid optimization: replacing 7536 strncmp by memcmp 7537 * [442]11350 undefined labels with -Os -fPIC 7538 * [443]12826 Optimizer removes reference through volatile pointer 7539 * [444]12500 stabs debug info: void no longer a predefined / builtin 7540 type 7541 * [445]12941 builtin-bitops-1.c miscompilation (latent bug) 7542 * [446]12953 tree inliner bug (in inline_forbidden_p) and fix 7543 * [447]13041 linux-2.6/sound/core/oss/rate.c miscompiled 7544 * [448]13507 spurious printf format warning 7545 * [449]13382 Type information for const pointer disappears during 7546 optimization. 7547 * [450]13394 noreturn attribute ignored on recursive invokation 7548 * [451]13400 Compiled code crashes storing to read-only location 7549 * [452]13521 Endless loop in calculate_global_regs_live 7550 7551 C++ compiler and library 7552 7553 Some of the bug fixes in this list were made to implement decisions 7554 that the ISO C++ standards committee has made concerning several defect 7555 reports (DRs). Links in the list below point to detailed discussion of 7556 the relevant defect report. 7557 * [453]2094 unimplemented: use of `ptrmem_cst' in template type 7558 unification 7559 * [454]2294 using declaration confusion 7560 * [455]5050 template instantiation depth exceeds limit: recursion 7561 problem? 7562 * [456]9371 Bad exception handling in 7563 i/ostream::operator>>/<<(streambuf*) 7564 * [457]9546 bad exception handling in ostream members 7565 * [458]10081 basic_ios::_M_cache_locale leaves NULL members in the 7566 face of unknown locales 7567 * [459]10093 [460][DR 61] Setting failbit in exceptions doesn't work 7568 * [461]10095 istream::operator>>(int&) sets ios::badbit when 7569 ios::failbit is set. 7570 * [462]11554 Warning about reordering of initializers doesn't mention 7571 location of constructor 7572 * [463]12297 istream::sentry::sentry() handles eof() incorrectly. 7573 * [464]12352 Exception safety problems in src/localename.cc 7574 * [465]12438 Memory leak in locale::combine() 7575 * [466]12540 Memory leak in locale::locale(const char*) 7576 * [467]12594 DRs [468]60 [TC] and [469]63 [TC] not implemented 7577 * [470]12657 Resolution of [471]DR 292 (WP) still unimplemented 7578 * [472]12696 memory eating infinite loop in diagnostics (error 7579 recovery problem) 7580 * [473]12815 Code compiled with optimization behaves unexpectedly 7581 * [474]12862 Conflicts between typedefs/enums and namespace member 7582 declarations 7583 * [475]12926 Wrong value after assignment in initialize list using 7584 bit-fields 7585 * [476]12967 Resolution of [477]DR 300 [WP] still unimplemented 7586 * [478]12971 Resolution of [479]DR 328 [WP] still unimplemented 7587 * [480]13007 basic_streambuf::pubimbue, imbue wrong 7588 * [481]13009 Implicitly-defined assignment operator writes to wrong 7589 memory 7590 * [482]13057 regparm attribute not applied to destructor 7591 * [483]13070 -Wformat option ignored in g++ 7592 * [484]13081 forward template declarations in <complex> let inlining 7593 fail 7594 * [485]13239 Assertion does not seem to work correctly anymore 7595 * [486]13262 "xxx is private within this context" when initializing a 7596 self-contained template class 7597 * [487]13290 simple typo in concept checking for std::generate_n 7598 * [488]13323 Template code does not compile in presence of typedef 7599 * [489]13369 __verify_grouping (and __add_grouping?) not correct 7600 * [490]13371 infinite loop with packed struct and inlining 7601 * [491]13445 Template argument replacement "dereferences" a typedef 7602 * [492]13461 Fails to access protected-ctor from public constant 7603 * [493]13462 Non-standard-conforming type set::pointer 7604 * [494]13478 gcc uses wrong constructor to initialize a const 7605 reference 7606 * [495]13544 "conflicting types" for enums in different scopes 7607 * [496]13650 string::compare should not (always) use 7608 traits_type::length() 7609 * [497]13683 bogus warning about passing non-PODs through ellipsis 7610 * [498]13688 Derived class is denied access to protected base class 7611 member class 7612 * [499]13774 Member variable cleared in virtual multiple inheritance 7613 class 7614 * [500]13884 Protect sstream.tcc from extern template use 7615 7616 Java compiler and library 7617 7618 * [501]10746 [win32] garbage collection crash in GCJ 7619 7620 Objective-C compiler and library 7621 7622 * [502]11433 Crash due to dereferencing null pointer when querying 7623 protocol 7624 7625 Fortran compiler and library 7626 7627 * [503]12633 logical expression gives incorrect result with 7628 -fugly-logint option 7629 * [504]13037 [gcse-lm] g77 generates incorrect code 7630 * [505]13213 Hex constant problem when compiling with -fugly-logint 7631 and -ftypeless-boz 7632 7633 x86-specific (Intel/AMD) 7634 7635 * [506]4490 ICE with -m128bit-long-double 7636 * [507]12292 [x86_64] ICE: RTL check: expected code `const_int', have 7637 `reg' in make_field_assignment, in combine.c 7638 * [508]12441 ICE: can't find a register to spill 7639 * [509]12943 array static-init failure under -fpic, -fPIC 7640 * [510]13608 Incorrect code with -O3 -ffast-math 7641 7642 PowerPC-specific 7643 7644 * [511]11598 testcase gcc.dg/20020118-1.c fails runtime check of 7645 __attribute__((aligned(16))) 7646 * [512]11793 ICE in extract_insn, in recog.c (const_vector's) 7647 * [513]12467 vmsumubm emitted when vmsummbm appropriate (typo in 7648 altivec.md) 7649 * [514]12537 g++ generates writeable text sections 7650 7651 SPARC-specific 7652 7653 * [515]12496 wrong result for __atomic_add(&value, -1) when using -O0 7654 -m64 7655 * [516]12865 mprotect call to make trampoline executable may fail 7656 * [517]13354 ICE in sparc_emit_set_const32 7657 7658 ARM-specific 7659 7660 * [518]10467 [arm] ICE in pre_insert_copy_insn, 7661 7662 ia64-specific 7663 7664 * [519]11226 ICE passing struct arg with two floats 7665 * [520]11227 ICE for _Complex float, _Complex long double args 7666 * [521]12644 GCC 3.3.2 fails to compile glibc on ia64 7667 * [522]13149 build gcc-3.3.2 1305 error:unrecognizable insn 7668 * Various fixes for libunwind 7669 7670 Alpha-specific 7671 7672 * [523]12654 Incorrect comparison code generated for Alpha 7673 * [524]12965 SEGV+ICE in cc1plus on alpha-linux with -O2 7674 * [525]13031 ICE (unrecognizable insn) when building gnome-libs-1.4.2 7675 7676 HPPA-specific 7677 7678 * [526]11634 [hppa] ICE in verify_local_live_at_start, in flow.c 7679 * [527]12158 [hppa] compilation does not terminate at -O1 7680 7681 S390-specific 7682 7683 * [528]11992 Wrong built-in code for memcmp with length 1<<24: only 7684 (1<<24)-1 possible for CLCL-Instruction 7685 7686 SH-specific 7687 7688 * [529]9365 segfault in gen_far_branch (config/sh/sh.c) 7689 * [530]10392 optimizer generates faulty array indexing 7690 * [531]11322 SH profiler outputs multiple definitions of symbol 7691 * [532]13069 gcc/config/sh/rtems.h broken 7692 * [533]13302 Putting a va_list in a struct causes seg fault 7693 * [534]13585 Incorrect optimization of call to sfunc 7694 * Fix inappropriately exported libgcc functions from the shared 7695 library 7696 7697 Other embedded target specific 7698 7699 * [535]8916 [mcore] unsigned char assign gets hosed. 7700 * [536]11576 [h8300] ICE in change_address_1, in emit-rtl.c 7701 * [537]13122 [h8300] local variable gets corrupted by function call 7702 when -fomit-frame-pointer is given 7703 * [538]13256 [cris] strict_low_part mistreated in delay slots 7704 * [539]13373 [mcore] optimization with -frerun-cse-after-loop 7705 -fexpensive-optimizations produces wrong code on mcore 7706 7707 GNU HURD-specific 7708 7709 * [540]12561 gcc/config/t-gnu needs updating to work with 7710 --with-sysroot 7711 7712 Tru64 Unix specific 7713 7714 * [541]6243 testsuite fails almost all tests due to no libintl in 7715 LD_LIBRARY_PATH during test. 7716 * [542]11397 weak aliases broken on Tru64 UNIX 7717 7718 AIX-specific 7719 7720 * [543]12505 build failure due to defines of uchar in cpphash.h and 7721 sys/types.h 7722 * [544]13150 WEAK symbols not exported by collect2 7723 7724 IRIX-specific 7725 7726 * [545]12666 fixincludes problem on IRIX 6.5.19m 7727 7728 Solaris-specific 7729 7730 * [546]12969 Including sys/byteorder.h breaks configure checks 7731 7732 Testsuite problems (compiler is not affected) 7733 7734 * [547]10819 testsuite creates CR+LF on compiler version lines in 7735 test summary files 7736 * [548]11612 abi_check not finding correct libgcc_s.so.1 7737 7738 Miscellaneous 7739 7740 * [549]13211 using -###, incorrect warnings about unused linker file 7741 are produced 7742 __________________________________________________________________ 7743 7744 GCC 3.3.4 7745 7746 This is the [550]list of problem reports (PRs) from GCC's bug tracking 7747 system that are known to be fixed in the 3.3.4 release. This list might 7748 not be complete (that is, it is possible that some PRs that have been 7749 fixed are not listed here). 7750 __________________________________________________________________ 7751 7752 GCC 3.3.5 7753 7754 This is the [551]list of problem reports (PRs) from GCC's bug tracking 7755 system that are known to be fixed in the 3.3.5 release. This list might 7756 not be complete (that is, it is possible that some PRs that have been 7757 fixed are not listed here). 7758 __________________________________________________________________ 7759 7760 GCC 3.3.6 7761 7762 This is the [552]list of problem reports (PRs) from GCC's bug tracking 7763 system that are known to be fixed in the 3.3.6 release. This list might 7764 not be complete (that is, it is possible that some PRs that have been 7765 fixed are not listed here). 7766 7767 7768 For questions related to the use of GCC, please consult these web 7769 pages and the [553]GCC manuals. If that fails, the 7770 [554]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 7771 web pages and the development of GCC are welcome on our developer 7772 list at [555]gcc (a] gcc.gnu.org. All of [556]our lists have public 7773 archives. 7774 7775 Copyright (C) [557]Free Software Foundation, Inc. Verbatim copying and 7776 distribution of this entire article is permitted in any medium, 7777 provided this notice is preserved. 7778 7779 These pages are [558]maintained by the GCC team. Last modified 7780 2011-04-25[559]. 7781 7782 References 7783 7784 1. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6 7785 2. http://gcc.gnu.org/gcc-3.1/changes.html#obsolete_systems 7786 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems 7787 4. http://gcc.gnu.org/gcc-3.3/changes.html#nonnull_attribute 7788 5. http://gcc.gnu.org/news/dfa.html 7789 6. http://gcc.gnu.org/gcc-3.3/c99status.html 7790 7. http://gcc.gnu.org/onlinedocs/gcc-3.3.6/g77/News.html 7791 8. http://gcc.gnu.org/PR10140 7792 9. http://gcc.gnu.org/PR10198 7793 10. http://gcc.gnu.org/PR10338 7794 11. http://gcc.gnu.org/PR3581 7795 12. http://gcc.gnu.org/PR4382 7796 13. http://gcc.gnu.org/PR5533 7797 14. http://gcc.gnu.org/PR6387 7798 15. http://gcc.gnu.org/PR6412 7799 16. http://gcc.gnu.org/PR6620 7800 17. http://gcc.gnu.org/PR6663 7801 18. http://gcc.gnu.org/PR7068 7802 19. http://gcc.gnu.org/PR7083 7803 20. http://gcc.gnu.org/PR7647 7804 21. http://gcc.gnu.org/PR7675 7805 22. http://gcc.gnu.org/PR7718 7806 23. http://gcc.gnu.org/PR8116 7807 24. http://gcc.gnu.org/PR8358 7808 25. http://gcc.gnu.org/PR8511 7809 26. http://gcc.gnu.org/PR8564 7810 27. http://gcc.gnu.org/PR8660 7811 28. http://gcc.gnu.org/PR8766 7812 29. http://gcc.gnu.org/PR8803 7813 30. http://gcc.gnu.org/PR8846 7814 31. http://gcc.gnu.org/PR8906 7815 32. http://gcc.gnu.org/PR9216 7816 33. http://gcc.gnu.org/PR9261 7817 34. http://gcc.gnu.org/PR9263 7818 35. http://gcc.gnu.org/PR9429 7819 36. http://gcc.gnu.org/PR9516 7820 37. http://gcc.gnu.org/PR9600 7821 38. http://gcc.gnu.org/PR9629 7822 39. http://gcc.gnu.org/PR9672 7823 40. http://gcc.gnu.org/PR9749 7824 41. http://gcc.gnu.org/PR9794 7825 42. http://gcc.gnu.org/PR9829 7826 43. http://gcc.gnu.org/PR9916 7827 44. http://gcc.gnu.org/PR9936 7828 45. http://gcc.gnu.org/PR10262 7829 46. http://gcc.gnu.org/PR10278 7830 47. http://gcc.gnu.org/PR10446 7831 48. http://gcc.gnu.org/PR10451 7832 49. http://gcc.gnu.org/PR10506 7833 50. http://gcc.gnu.org/PR10549 7834 51. http://gcc.gnu.org/PR2001 7835 52. http://gcc.gnu.org/PR2391 7836 53. http://gcc.gnu.org/PR2960 7837 54. http://gcc.gnu.org/PR4046 7838 55. http://gcc.gnu.org/PR6405 7839 56. http://gcc.gnu.org/PR6798 7840 57. http://gcc.gnu.org/PR6871 7841 58. http://gcc.gnu.org/PR6909 7842 59. http://gcc.gnu.org/PR7189 7843 60. http://gcc.gnu.org/PR7642 7844 61. http://gcc.gnu.org/PR8634 7845 62. http://gcc.gnu.org/PR8750 7846 63. http://gcc.gnu.org/PR2161 7847 64. http://gcc.gnu.org/PR4319 7848 65. http://gcc.gnu.org/PR8602 7849 66. http://gcc.gnu.org/PR9177 7850 67. http://gcc.gnu.org/PR9853 7851 68. http://gcc.gnu.org/PR45 7852 69. http://gcc.gnu.org/PR3784 7853 70. http://gcc.gnu.org/PR764 7854 71. http://gcc.gnu.org/PR5116 7855 72. http://gcc.gnu.org/PR2862 7856 73. http://gcc.gnu.org/PR3663 7857 74. http://gcc.gnu.org/PR3797 7858 75. http://gcc.gnu.org/PR3948 7859 76. http://gcc.gnu.org/PR4137 7860 77. http://gcc.gnu.org/PR4361 7861 78. http://gcc.gnu.org/PR4802 7862 79. http://gcc.gnu.org/PR5837 7863 80. http://gcc.gnu.org/PR4803 7864 81. http://gcc.gnu.org/PR5094 7865 82. http://gcc.gnu.org/PR5730 7866 83. http://gcc.gnu.org/PR6713 7867 84. http://gcc.gnu.org/PR7015 7868 85. http://gcc.gnu.org/PR7086 7869 86. http://gcc.gnu.org/PR7099 7870 87. http://gcc.gnu.org/PR7247 7871 88. http://gcc.gnu.org/PR7441 7872 89. http://gcc.gnu.org/PR7768 7873 90. http://gcc.gnu.org/PR7804 7874 91. http://gcc.gnu.org/PR8099 7875 92. http://gcc.gnu.org/PR8117 7876 93. http://gcc.gnu.org/PR8205 7877 94. http://gcc.gnu.org/PR8645 7878 95. http://gcc.gnu.org/PR8724 7879 96. http://gcc.gnu.org/PR8805 7880 97. http://gcc.gnu.org/PR8691 7881 98. http://gcc.gnu.org/PR8700 7882 99. http://gcc.gnu.org/PR8724 7883 100. http://gcc.gnu.org/PR8949 7884 101. http://gcc.gnu.org/PR9016 7885 102. http://gcc.gnu.org/PR9053 7886 103. http://gcc.gnu.org/PR9152 7887 104. http://gcc.gnu.org/PR9182 7888 105. http://gcc.gnu.org/PR9297 7889 106. http://gcc.gnu.org/PR9318 7890 107. http://gcc.gnu.org/PR9320 7891 108. http://gcc.gnu.org/PR9400 7892 109. http://gcc.gnu.org/PR9424 7893 110. http://gcc.gnu.org/PR9425 7894 111. http://gcc.gnu.org/PR9439 7895 112. http://gcc.gnu.org/PR9474 7896 113. http://gcc.gnu.org/PR9548 7897 114. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#231 7898 115. http://gcc.gnu.org/PR9555 7899 116. http://gcc.gnu.org/PR9561 7900 117. http://gcc.gnu.org/PR9563 7901 118. http://gcc.gnu.org/PR9582 7902 119. http://gcc.gnu.org/PR9622 7903 120. http://gcc.gnu.org/PR9683 7904 121. http://gcc.gnu.org/PR9791 7905 122. http://gcc.gnu.org/PR9817 7906 123. http://gcc.gnu.org/PR9825 7907 124. http://gcc.gnu.org/PR9826 7908 125. http://gcc.gnu.org/PR9924 7909 126. http://gcc.gnu.org/PR9946 7910 127. http://gcc.gnu.org/PR9964 7911 128. http://gcc.gnu.org/PR9988 7912 129. http://gcc.gnu.org/PR10033 7913 130. http://gcc.gnu.org/PR10097 7914 131. http://gcc.gnu.org/PR10132 7915 132. http://gcc.gnu.org/PR10180 7916 133. http://gcc.gnu.org/PR10199 7917 134. http://gcc.gnu.org/PR10300 7918 135. http://gcc.gnu.org/PR10427 7919 136. http://gcc.gnu.org/PR10503 7920 137. http://gcc.gnu.org/PR5956 7921 138. http://gcc.gnu.org/PR1832 7922 139. http://gcc.gnu.org/PR3924 7923 140. http://gcc.gnu.org/PR5634 7924 141. http://gcc.gnu.org/PR6367 7925 142. http://gcc.gnu.org/PR6491 7926 143. http://gcc.gnu.org/PR6742 7927 144. http://gcc.gnu.org/PR7113 7928 145. http://gcc.gnu.org/PR7236 7929 146. http://gcc.gnu.org/PR7278 7930 147. http://gcc.gnu.org/PR7384 7931 148. http://gcc.gnu.org/PR7388 7932 149. http://gcc.gnu.org/PR8587 7933 150. http://gcc.gnu.org/PR9038 7934 151. http://gcc.gnu.org/PR10197 7935 152. http://gcc.gnu.org/PR6005 7936 153. http://gcc.gnu.org/PR6389 7937 154. http://gcc.gnu.org/PR6576 7938 155. http://gcc.gnu.org/PR6652 7939 156. http://gcc.gnu.org/PR7060 7940 157. http://gcc.gnu.org/PR7073 7941 158. http://gcc.gnu.org/PR7180 7942 159. http://gcc.gnu.org/PR7416 7943 160. http://gcc.gnu.org/PR7570 7944 161. http://gcc.gnu.org/PR7578 7945 162. http://gcc.gnu.org/PR7611 7946 163. http://gcc.gnu.org/PR7709 7947 164. http://gcc.gnu.org/PR7766 7948 165. http://gcc.gnu.org/PR7785 7949 166. http://gcc.gnu.org/PR7786 7950 167. http://gcc.gnu.org/PR8142 7951 168. http://gcc.gnu.org/PR8234 7952 169. http://gcc.gnu.org/PR8415 7953 170. http://gcc.gnu.org/PR8481 7954 171. http://gcc.gnu.org/PR8593 7955 172. http://gcc.gnu.org/PR8759 7956 173. http://gcc.gnu.org/PR8997 7957 174. http://gcc.gnu.org/PR9253 7958 175. http://gcc.gnu.org/PR9254 7959 176. http://gcc.gnu.org/PR9271 7960 177. http://gcc.gnu.org/PR6767 7961 178. http://gcc.gnu.org/PR9911 7962 179. http://gcc.gnu.org/PR10020 7963 180. http://gcc.gnu.org/PR10546 7964 181. http://gcc.gnu.org/PR7029 7965 182. http://gcc.gnu.org/PR2903 7966 183. http://gcc.gnu.org/PR7873 7967 184. http://gcc.gnu.org/PR7680 7968 185. http://gcc.gnu.org/PR8705 7969 186. http://gcc.gnu.org/PR9986 7970 187. http://gcc.gnu.org/PR10056 7971 188. http://gcc.gnu.org/PR6744 7972 189. http://gcc.gnu.org/PR7361 7973 190. http://gcc.gnu.org/PR9496 7974 191. http://gcc.gnu.org/PR7067 7975 192. http://gcc.gnu.org/PR8480 7976 193. http://gcc.gnu.org/PR8784 7977 194. http://gcc.gnu.org/PR10315 7978 195. http://gcc.gnu.org/PR10267 7979 196. http://gcc.gnu.org/PR7916 7980 197. http://gcc.gnu.org/PR7926 7981 198. http://gcc.gnu.org/PR8555 7982 199. http://gcc.gnu.org/PR8994 7983 200. http://gcc.gnu.org/PR9426 7984 201. http://gcc.gnu.org/PR9806 7985 202. http://gcc.gnu.org/PR10077 7986 203. http://gcc.gnu.org/PR10233 7987 204. http://gcc.gnu.org/PR10286 7988 205. http://gcc.gnu.org/PR10308 7989 206. http://gcc.gnu.org/PR11272 7990 207. http://gcc.gnu.org/PR5754 7991 208. http://gcc.gnu.org/PR6597 7992 209. http://gcc.gnu.org/PR6949 7993 210. http://gcc.gnu.org/PR7053 7994 211. http://gcc.gnu.org/PR8164 7995 212. http://gcc.gnu.org/PR8384 7996 213. http://gcc.gnu.org/PR9559 7997 214. http://gcc.gnu.org/PR9649 7998 215. http://gcc.gnu.org/PR9864 7999 216. http://gcc.gnu.org/PR10432 8000 217. http://gcc.gnu.org/PR10475 8001 218. http://gcc.gnu.org/PR10635 8002 219. http://gcc.gnu.org/PR10661 8003 220. http://gcc.gnu.org/PR10700 8004 221. http://gcc.gnu.org/PR10712 8005 222. http://gcc.gnu.org/PR10796 8006 223. http://gcc.gnu.org/PR10890 8007 224. http://gcc.gnu.org/PR10939 8008 225. http://gcc.gnu.org/PR10956 8009 226. http://gcc.gnu.org/PR11041 8010 227. http://gcc.gnu.org/PR11059 8011 228. http://gcc.gnu.org/PR11083 8012 229. http://gcc.gnu.org/PR11105 8013 230. http://gcc.gnu.org/PR11149 8014 231. http://gcc.gnu.org/PR11228 8015 232. http://gcc.gnu.org/PR11282 8016 233. http://gcc.gnu.org/PR11301 8017 234. http://gcc.gnu.org/PR11308 8018 235. http://gcc.gnu.org/PR11473 8019 236. http://gcc.gnu.org/PR11503 8020 237. http://gcc.gnu.org/PR11513 8021 238. http://gcc.gnu.org/PR11198 8022 239. http://gcc.gnu.org/PR11304 8023 240. http://gcc.gnu.org/PR11381 8024 241. http://gcc.gnu.org/PR11536 8025 242. http://gcc.gnu.org/PR11557 8026 243. http://gcc.gnu.org/PR5897 8027 244. http://gcc.gnu.org/PR11279 8028 245. http://gcc.gnu.org/PR11022 8029 246. http://gcc.gnu.org/PR2330 8030 247. http://gcc.gnu.org/PR5388 8031 248. http://gcc.gnu.org/PR5390 8032 249. http://gcc.gnu.org/PR7877 8033 250. http://gcc.gnu.org/PR9393 8034 251. http://gcc.gnu.org/PR10032 8035 252. http://gcc.gnu.org/PR10468 8036 253. http://gcc.gnu.org/PR10527 8037 254. http://gcc.gnu.org/PR10679 8038 255. http://gcc.gnu.org/PR10682 8039 256. http://gcc.gnu.org/PR10689 8040 257. http://gcc.gnu.org/PR10845 8041 258. http://gcc.gnu.org/PR10849 8042 259. http://gcc.gnu.org/PR10888 8043 260. http://gcc.gnu.org/PR10929 8044 261. http://gcc.gnu.org/PR10931 8045 262. http://gcc.gnu.org/PR10940 8046 263. http://gcc.gnu.org/PR10968 8047 264. http://gcc.gnu.org/PR10990 8048 265. http://gcc.gnu.org/PR11039 8049 266. http://gcc.gnu.org/PR11062 8050 267. http://gcc.gnu.org/PR11095 8051 268. http://gcc.gnu.org/PR11098 8052 269. http://gcc.gnu.org/PR11137 8053 270. http://gcc.gnu.org/PR11154 8054 271. http://gcc.gnu.org/PR11329 8055 272. http://gcc.gnu.org/PR11332 8056 273. http://gcc.gnu.org/PR11431 8057 274. http://gcc.gnu.org/PR11528 8058 275. http://gcc.gnu.org/PR11546 8059 276. http://gcc.gnu.org/PR11567 8060 277. http://gcc.gnu.org/PR11645 8061 278. http://gcc.gnu.org/PR5179 8062 279. http://gcc.gnu.org/PR8204 8063 280. http://gcc.gnu.org/PR10838 8064 281. http://gcc.gnu.org/PR10886 8065 282. http://gcc.gnu.org/PR11349 8066 283. http://gcc.gnu.org/PR4823 8067 284. http://gcc.gnu.org/PR8878 8068 285. http://gcc.gnu.org/PR9815 8069 286. http://gcc.gnu.org/PR10402 8070 287. http://gcc.gnu.org/PR10504 8071 288. http://gcc.gnu.org/PR10673 8072 289. http://gcc.gnu.org/PR11044 8073 290. http://gcc.gnu.org/PR11089 8074 291. http://gcc.gnu.org/PR11420 8075 292. http://gcc.gnu.org/PR9362 8076 293. http://gcc.gnu.org/PR10142 8077 294. http://gcc.gnu.org/PR10663 8078 295. http://gcc.gnu.org/PR10835 8079 296. http://gcc.gnu.org/PR10876 8080 297. http://gcc.gnu.org/PR10955 8081 298. http://gcc.gnu.org/PR11018 8082 299. http://gcc.gnu.org/PR11556 8083 300. http://gcc.gnu.org/PR10907 8084 301. http://gcc.gnu.org/PR11320 8085 302. http://gcc.gnu.org/PR11599 8086 303. http://gcc.gnu.org/PR9745 8087 304. http://gcc.gnu.org/PR10871 8088 305. http://gcc.gnu.org/PR11440 8089 306. http://gcc.gnu.org/PR7594 8090 307. http://gcc.gnu.org/PR10557 8091 308. http://gcc.gnu.org/PR11054 8092 309. http://gcc.gnu.org/PR10834 8093 310. http://gcc.gnu.org/PR10842 8094 311. http://gcc.gnu.org/PR11052 8095 312. http://gcc.gnu.org/PR11183 8096 313. http://gcc.gnu.org/PR11084 8097 314. http://gcc.gnu.org/PR10331 8098 315. http://gcc.gnu.org/PR10413 8099 316. http://gcc.gnu.org/PR11096 8100 317. http://gcc.gnu.org/PR2873 8101 318. http://gcc.gnu.org/PR3163 8102 319. http://gcc.gnu.org/PR5287 8103 320. http://gcc.gnu.org/PR10148 8104 321. http://gcc.gnu.org/PR8787 8105 322. http://gcc.gnu.org/PR10900 8106 323. http://gcc.gnu.org/PR1607 8107 324. http://gcc.gnu.org/PR4252 8108 325. http://gcc.gnu.org/PR4490 8109 326. http://gcc.gnu.org/PR10355 8110 327. http://gcc.gnu.org/PR10726 8111 328. http://gcc.gnu.org/PR10805 8112 329. http://gcc.gnu.org/PR10815 8113 330. http://gcc.gnu.org/PR10877 8114 331. http://gcc.gnu.org/PR11280 8115 332. http://gcc.gnu.org/PR11466 8116 333. http://gcc.gnu.org/PR10737 8117 334. http://gcc.gnu.org/PR10810 8118 335. http://gcc.gnu.org/bugzilla/ 8119 336. http://gcc.gnu.org/PR8336 8120 337. http://gcc.gnu.org/PR9330 8121 338. http://gcc.gnu.org/PR9631 8122 339. http://gcc.gnu.org/PR9877 8123 340. http://gcc.gnu.org/PR11687 8124 341. http://gcc.gnu.org/PR12263 8125 342. http://gcc.gnu.org/PR12490 8126 343. http://gcc.gnu.org/PR7277 8127 344. http://gcc.gnu.org/PR7939 8128 345. http://gcc.gnu.org/PR11063 8129 346. http://gcc.gnu.org/PR11207 8130 347. http://gcc.gnu.org/PR11522 8131 348. http://gcc.gnu.org/PR11595 8132 349. http://gcc.gnu.org/PR11646 8133 350. http://gcc.gnu.org/PR11665 8134 351. http://gcc.gnu.org/PR11852 8135 352. http://gcc.gnu.org/PR11878 8136 353. http://gcc.gnu.org/PR11883 8137 354. http://gcc.gnu.org/PR11991 8138 355. http://gcc.gnu.org/PR12146 8139 356. http://gcc.gnu.org/PR12215 8140 357. http://gcc.gnu.org/PR12369 8141 358. http://gcc.gnu.org/PR12446 8142 359. http://gcc.gnu.org/PR12510 8143 360. http://gcc.gnu.org/PR12544 8144 361. http://gcc.gnu.org/PR9862 8145 362. http://gcc.gnu.org/PR10962 8146 363. http://gcc.gnu.org/PR11370 8147 364. http://gcc.gnu.org/PR11637 8148 365. http://gcc.gnu.org/PR11885 8149 366. http://gcc.gnu.org/PR12082 8150 367. http://gcc.gnu.org/PR12180 8151 368. http://gcc.gnu.org/PR12340 8152 369. http://gcc.gnu.org/PR3907 8153 370. http://gcc.gnu.org/PR5293 8154 371. http://gcc.gnu.org/PR5296 8155 372. http://gcc.gnu.org/PR7939 8156 373. http://gcc.gnu.org/PR8656 8157 374. http://gcc.gnu.org/PR10147 8158 375. http://gcc.gnu.org/PR11400 8159 376. http://gcc.gnu.org/PR11409 8160 377. http://gcc.gnu.org/PR11740 8161 378. http://gcc.gnu.org/PR11786 8162 379. http://gcc.gnu.org/PR11867 8163 380. http://gcc.gnu.org/PR11928 8164 381. http://gcc.gnu.org/PR12114 8165 382. http://gcc.gnu.org/PR12163 8166 383. http://gcc.gnu.org/PR12181 8167 384. http://gcc.gnu.org/PR12236 8168 385. http://gcc.gnu.org/PR12266 8169 386. http://gcc.gnu.org/PR12296 8170 387. http://gcc.gnu.org/PR12298 8171 388. http://gcc.gnu.org/PR12369 8172 389. http://gcc.gnu.org/PR12337 8173 390. http://gcc.gnu.org/PR12344 8174 391. http://gcc.gnu.org/PR12451 8175 392. http://gcc.gnu.org/PR12486 8176 393. http://gcc.gnu.org/PR8869 8177 394. http://gcc.gnu.org/PR9786 8178 395. http://gcc.gnu.org/PR11689 8179 396. http://gcc.gnu.org/PR12116 8180 397. http://gcc.gnu.org/PR12070 8181 398. http://gcc.gnu.org/PR11184 8182 399. http://gcc.gnu.org/PR11535 8183 400. http://gcc.gnu.org/PR11693 8184 401. http://gcc.gnu.org/PR12224 8185 402. http://gcc.gnu.org/PR11087 8186 403. http://gcc.gnu.org/PR11319 8187 404. http://gcc.gnu.org/PR11949 8188 405. http://gcc.gnu.org/PR11662 8189 406. http://gcc.gnu.org/PR11965 8190 407. http://gcc.gnu.org/PR12301 8191 408. http://gcc.gnu.org/PR11717 8192 409. http://gcc.gnu.org/PR11313 8193 410. http://gcc.gnu.org/PR11712 8194 411. http://gcc.gnu.org/PR12166 8195 412. http://gcc.gnu.org/PR12101 8196 413. http://gcc.gnu.org/PR10988 8197 414. http://gcc.gnu.org/PR11805 8198 415. http://gcc.gnu.org/PR11902 8199 416. http://gcc.gnu.org/PR11903 8200 417. http://gcc.gnu.org/bugzilla/ 8201 418. http://gcc.gnu.org/PR11890 8202 419. http://gcc.gnu.org/PR12399 8203 420. http://gcc.gnu.org/PR13068 8204 421. http://gcc.gnu.org/PR10060 8205 422. http://gcc.gnu.org/PR10555 8206 423. http://gcc.gnu.org/PR10706 8207 424. http://gcc.gnu.org/PR11496 8208 425. http://gcc.gnu.org/PR11741 8209 426. http://gcc.gnu.org/PR12440 8210 427. http://gcc.gnu.org/PR12632 8211 428. http://gcc.gnu.org/PR12712 8212 429. http://gcc.gnu.org/PR12726 8213 430. http://gcc.gnu.org/PR12890 8214 431. http://gcc.gnu.org/PR12900 8215 432. http://gcc.gnu.org/PR13060 8216 433. http://gcc.gnu.org/PR13289 8217 434. http://gcc.gnu.org/PR13318 8218 435. http://gcc.gnu.org/PR13392 8219 436. http://gcc.gnu.org/PR13574 8220 437. http://gcc.gnu.org/PR13475 8221 438. http://gcc.gnu.org/PR13797 8222 439. http://gcc.gnu.org/PR13824 8223 440. http://gcc.gnu.org/PR8776 8224 441. http://gcc.gnu.org/PR10339 8225 442. http://gcc.gnu.org/PR11350 8226 443. http://gcc.gnu.org/PR12826 8227 444. http://gcc.gnu.org/PR12500 8228 445. http://gcc.gnu.org/PR12941 8229 446. http://gcc.gnu.org/PR12953 8230 447. http://gcc.gnu.org/PR13041 8231 448. http://gcc.gnu.org/PR13507 8232 449. http://gcc.gnu.org/PR13382 8233 450. http://gcc.gnu.org/PR13394 8234 451. http://gcc.gnu.org/PR13400 8235 452. http://gcc.gnu.org/PR13521 8236 453. http://gcc.gnu.org/PR2094 8237 454. http://gcc.gnu.org/PR2294 8238 455. http://gcc.gnu.org/PR5050 8239 456. http://gcc.gnu.org/PR9371 8240 457. http://gcc.gnu.org/PR9546 8241 458. http://gcc.gnu.org/PR10081 8242 459. http://gcc.gnu.org/PR10093 8243 460. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#61 8244 461. http://gcc.gnu.org/PR10095 8245 462. http://gcc.gnu.org/PR11554 8246 463. http://gcc.gnu.org/PR12297 8247 464. http://gcc.gnu.org/PR12352 8248 465. http://gcc.gnu.org/PR12438 8249 466. http://gcc.gnu.org/PR12540 8250 467. http://gcc.gnu.org/PR12594 8251 468. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#60 8252 469. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#63 8253 470. http://gcc.gnu.org/PR12657 8254 471. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#292 8255 472. http://gcc.gnu.org/PR12696 8256 473. http://gcc.gnu.org/PR12815 8257 474. http://gcc.gnu.org/PR12862 8258 475. http://gcc.gnu.org/PR12926 8259 476. http://gcc.gnu.org/PR12967 8260 477. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html 8261 478. http://gcc.gnu.org/PR12971 8262 479. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#328 8263 480. http://gcc.gnu.org/PR13007 8264 481. http://gcc.gnu.org/PR13009 8265 482. http://gcc.gnu.org/PR13057 8266 483. http://gcc.gnu.org/PR13070 8267 484. http://gcc.gnu.org/PR13081 8268 485. http://gcc.gnu.org/PR13239 8269 486. http://gcc.gnu.org/PR13262 8270 487. http://gcc.gnu.org/PR13290 8271 488. http://gcc.gnu.org/PR13323 8272 489. http://gcc.gnu.org/PR13369 8273 490. http://gcc.gnu.org/PR13371 8274 491. http://gcc.gnu.org/PR13445 8275 492. http://gcc.gnu.org/PR13461 8276 493. http://gcc.gnu.org/PR13462 8277 494. http://gcc.gnu.org/PR13478 8278 495. http://gcc.gnu.org/PR13544 8279 496. http://gcc.gnu.org/PR13650 8280 497. http://gcc.gnu.org/PR13683 8281 498. http://gcc.gnu.org/PR13688 8282 499. http://gcc.gnu.org/PR13774 8283 500. http://gcc.gnu.org/PR13884 8284 501. http://gcc.gnu.org/PR10746 8285 502. http://gcc.gnu.org/PR11433 8286 503. http://gcc.gnu.org/PR12633 8287 504. http://gcc.gnu.org/PR13037 8288 505. http://gcc.gnu.org/PR13213 8289 506. http://gcc.gnu.org/PR4490 8290 507. http://gcc.gnu.org/PR12292 8291 508. http://gcc.gnu.org/PR12441 8292 509. http://gcc.gnu.org/PR12943 8293 510. http://gcc.gnu.org/PR13608 8294 511. http://gcc.gnu.org/PR11598 8295 512. http://gcc.gnu.org/PR11793 8296 513. http://gcc.gnu.org/PR12467 8297 514. http://gcc.gnu.org/PR12537 8298 515. http://gcc.gnu.org/PR12496 8299 516. http://gcc.gnu.org/PR12865 8300 517. http://gcc.gnu.org/PR13354 8301 518. http://gcc.gnu.org/PR10467 8302 519. http://gcc.gnu.org/PR11226 8303 520. http://gcc.gnu.org/PR11227 8304 521. http://gcc.gnu.org/PR12644 8305 522. http://gcc.gnu.org/PR13149 8306 523. http://gcc.gnu.org/PR12654 8307 524. http://gcc.gnu.org/PR12965 8308 525. http://gcc.gnu.org/PR13031 8309 526. http://gcc.gnu.org/PR11634 8310 527. http://gcc.gnu.org/PR12158 8311 528. http://gcc.gnu.org/PR11992 8312 529. http://gcc.gnu.org/PR9365 8313 530. http://gcc.gnu.org/PR10392 8314 531. http://gcc.gnu.org/PR11322 8315 532. http://gcc.gnu.org/PR13069 8316 533. http://gcc.gnu.org/PR13302 8317 534. http://gcc.gnu.org/PR13585 8318 535. http://gcc.gnu.org/PR8916 8319 536. http://gcc.gnu.org/PR11576 8320 537. http://gcc.gnu.org/PR13122 8321 538. http://gcc.gnu.org/PR13256 8322 539. http://gcc.gnu.org/PR13373 8323 540. http://gcc.gnu.org/PR12561 8324 541. http://gcc.gnu.org/PR6243 8325 542. http://gcc.gnu.org/PR11397 8326 543. http://gcc.gnu.org/PR12505 8327 544. http://gcc.gnu.org/PR13150 8328 545. http://gcc.gnu.org/PR12666 8329 546. http://gcc.gnu.org/PR12969 8330 547. http://gcc.gnu.org/PR10819 8331 548. http://gcc.gnu.org/PR11612 8332 549. http://gcc.gnu.org/PR13211 8333 550. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.4 8334 551. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.5 8335 552. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.6 8336 553. http://gcc.gnu.org/onlinedocs/ 8337 554. mailto:gcc-help (a] gcc.gnu.org 8338 555. mailto:gcc (a] gcc.gnu.org 8339 556. http://gcc.gnu.org/lists.html 8340 557. http://www.fsf.org/ 8341 558. http://gcc.gnu.org/about.html 8342 559. http://validator.w3.org/check/referer 8343 ====================================================================== 8344 http://gcc.gnu.org/gcc-3.2/index.html 8345 8346 GCC 3.2 Release Series 8347 8348 April 25, 2003 8349 8350 The [1]GNU project and the GCC developers are pleased to announce the 8351 release of GCC 3.2.3. 8352 8353 The purpose of the GCC 3.2 release series is to provide a stable 8354 platform for OS distributors to use building their next releases. A 8355 primary objective was to stabilize the C++ ABI; we believe that the 8356 interface to the compiler and the C++ standard library are now 8357 relatively stable. 8358 8359 Be aware that C++ code compiled by GCC 3.2.x will (in general) not 8360 interoperate with code compiled by GCC 3.1.1 or earlier. 8361 8362 Please refer to our [2]detailed list of news, caveats, and bug-fixes 8363 for further information. 8364 8365 Release History 8366 8367 GCC 3.2.3 8368 April 25, 2003 ([3]changes) 8369 8370 GCC 3.2.2 8371 February 5, 2003 ([4]changes) 8372 8373 GCC 3.2.1 8374 November 19, 2002 ([5]changes) 8375 8376 GCC 3.2 8377 August 14, 2002 ([6]changes) 8378 8379 References and Acknowledgements 8380 8381 GCC used to stand for the GNU C Compiler, but since the compiler 8382 supports several other languages aside from C, it now stands for the 8383 GNU Compiler Collection. 8384 8385 A list of [7]successful builds is updated as new information becomes 8386 available. 8387 8388 The GCC developers would like to thank the numerous people that have 8389 contributed new features, improvements, bug fixes, and other changes as 8390 well as test results to GCC. This [8]amazing group of volunteers is 8391 what makes GCC successful. 8392 8393 For additional information about GCC please refer to the [9]GCC project 8394 web site or contact the [10]GCC development mailing list. 8395 8396 To obtain GCC please use [11]our mirror sites, or our CVS server. 8397 8398 8399 For questions related to the use of GCC, please consult these web 8400 pages and the [12]GCC manuals. If that fails, the 8401 [13]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 8402 web pages and the development of GCC are welcome on our developer 8403 list at [14]gcc (a] gcc.gnu.org. All of [15]our lists have public 8404 archives. 8405 8406 Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and 8407 distribution of this entire article is permitted in any medium, 8408 provided this notice is preserved. 8409 8410 These pages are [17]maintained by the GCC team. Last modified 8411 2011-04-25[18]. 8412 8413 References 8414 8415 1. http://www.gnu.org/ 8416 2. http://gcc.gnu.org/gcc-3.2/changes.html 8417 3. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3 8418 4. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.2 8419 5. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.1 8420 6. http://gcc.gnu.org/gcc-3.2/changes.html#3.2 8421 7. http://gcc.gnu.org/gcc-3.2/buildstat.html 8422 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 8423 9. http://gcc.gnu.org/index.html 8424 10. mailto:gcc (a] gcc.gnu.org 8425 11. http://gcc.gnu.org/mirrors.html 8426 12. http://gcc.gnu.org/onlinedocs/ 8427 13. mailto:gcc-help (a] gcc.gnu.org 8428 14. mailto:gcc (a] gcc.gnu.org 8429 15. http://gcc.gnu.org/lists.html 8430 16. http://www.fsf.org/ 8431 17. http://gcc.gnu.org/about.html 8432 18. http://validator.w3.org/check/referer 8433 ====================================================================== 8434 http://gcc.gnu.org/gcc-3.2/changes.html 8435 8436 GCC 3.2 Release Series 8437 Changes, New Features, and Fixes 8438 8439 The latest release in the 3.2 release series is [1]GCC 3.2.3. 8440 8441 Caveats and New Features 8442 8443 Caveats 8444 8445 * The C++ compiler does not correctly zero-initialize 8446 pointers-to-data members. You must explicitly initialize them. For 8447 example: int S::*m(0); will work, but depending on 8448 default-initialization to zero will not work. This bug cannot be 8449 fixed in GCC 3.2 without inducing unacceptable risks. It will be 8450 fixed in GCC 3.3. 8451 * This GCC release is based on the GCC 3.1 sourcebase, and thus has 8452 all the [2]changes in the GCC 3.1 series. In addition, GCC 3.2 has 8453 a number of C++ ABI fixes which make its C++ compiler generate 8454 binary code which is incompatible with the C++ compilers found in 8455 earlier GCC releases, including GCC 3.1 and GCC 3.1.1. 8456 8457 Frontend Enhancements 8458 8459 C/C++/Objective-C 8460 8461 * The method of constructing the list of directories to be searched 8462 for header files has been revised. If a directory named by a -I 8463 option is a standard system include directory, the option is 8464 ignored to ensure that the default search order for system 8465 directories and the special treatment of system header files are 8466 not defeated. 8467 * The C and Objective-C compilers no longer accept the "Naming Types" 8468 extension (typedef foo = bar); it was already unavailable in C++. 8469 Code which uses it will need to be changed to use the "typeof" 8470 extension instead: typedef typeof(bar) foo. (We have removed this 8471 extension without a period of deprecation because it has caused the 8472 compiler to crash since version 3.0 and no one noticed until very 8473 recently. Thus we conclude it is not in widespread use.) 8474 8475 C++ 8476 8477 * GCC 3.2 fixed serveral differences between the C++ ABI implemented 8478 in GCC and the multi-vendor standard, but more have been found 8479 since the release. 3.2.1 adds a new warning, -Wabi, to warn about 8480 code which is affected by these bugs. We will fix these bugs in 8481 some future release, once we are confident that all have been 8482 found; until then, it is our intention to make changes to the ABI 8483 only if they are necessary for correct compilation of C++, as 8484 opposed to conformance to the ABI documents. 8485 * For details on how to build an ABI compliant compiler for GNU/Linux 8486 systems, check the [3]common C++ ABI page. 8487 8488 New Targets and Target Specific Improvements 8489 8490 IA-32 8491 8492 * Fixed a number of bugs in SSE and MMX intrinsics. 8493 * Fixed common compiler crashes with SSE instruction set enabled 8494 (implied by -march=pentium3, pentium4, athlon-xp) 8495 * __m128 and __m128i is not 128bit aligned when used in structures. 8496 8497 x86-64 8498 8499 * A bug whereby the compiler could generate bad code for bzero has 8500 been fixed. 8501 * ABI fixes (implying ABI incompatibilities with previous version in 8502 some corner cases) 8503 * Fixed prefetch code generation 8504 __________________________________________________________________ 8505 8506 GCC 3.2.3 8507 8508 3.2.3 is a bug fix release only; there are no new features that were 8509 not present in GCC 3.2.2. 8510 8511 Bug Fixes 8512 8513 This section lists the problem reports (PRs) from GCC's bug tracking 8514 system that are known to be fixed in the 3.2.3 release. This list might 8515 not be complete (that is, it is possible that some PRs that have been 8516 fixed are not listed here), and some of the titles have been changed to 8517 make them more clear. 8518 8519 Internal Compiler Errors (multi-platform) 8520 8521 * [4]3782: (c++) -quiet -fstats produces a segmentation fault in 8522 cc1plus 8523 * [5]6440: (c++) template specializations cause ICE 8524 * [6]7050: (c++) ICE on: (i ? get_string() : throw) 8525 * [7]7741: ICE on conflicting types (make_decl_rtl in varasm.c) 8526 * [8]7982: (c++) ICE due to infinite recursion (using STL set) 8527 * [9]8068: exceedingly high (infinite) memory usage 8528 * [10]8178: ICE with __builtin_ffs 8529 * [11]8396: ICE in copy_to_mode_reg, in explow.c 8530 * [12]8674: (c++) ICE in cp_expr_size, in cp/cp-lang.c 8531 * [13]9768: ICE when optimizing inline code at -O2 8532 * [14]9798: (c++) Infinite recursion (segfault) in 8533 cp/decl.c:push_using_directive with recursive using directives 8534 * [15]9799: mismatching structure initializer with nested flexible 8535 array member: ICE 8536 * [16]9928: ICE on duplicate enum declaration 8537 * [17]10114: ICE in mem_loc_descriptor, in dwarf2out.c (affects 8538 sparc, alpha) 8539 * [18]10352: ICE in find_reloads_toplev 8540 * [19]10336: ICE with -Wunreachable-code 8541 8542 C/optimizer bugs: 8543 8544 * [20]8224: Incorrect joining of signed and unsigned division 8545 * [21]8613: -O2 produces wrong code with builtin strlen and 8546 postincrements 8547 * [22]8828: gcc reports some code is unreachable when it is not 8548 * [23]9226: GCSE breaking argument passing 8549 * [24]9853: miscompilation of non-constant structure initializer 8550 * [25]9797: C99-style struct initializers are miscompiled 8551 * [26]9967: Some standard C function calls should not be replaced 8552 when optimizing for size 8553 * [27]10116: ce2: invalid merge of join_bb in the context of switch 8554 statements 8555 * [28]10171: wrong code for inlined function 8556 * [29]10175: -Wunreachable-code doesn't work for single lines 8557 8558 C++ compiler and library: 8559 8560 * [30]8316: Confusing diagnostic for code that misuses conversion 8561 operators 8562 * [31]9169: filebuf output fails if codecvt<>::out returns noconv 8563 * [32]9420: incomplete type incorrectly reported 8564 * [33]9459: typeof in return type specification of template not 8565 supported 8566 * [34]9507: filebuf::open handles ios_base::ate incorrectly 8567 * [35]9538: Out-of-bounds memory access in streambuf::sputbackc 8568 * [36]9602: Total confusion about template/friend/virtual/abstract 8569 * [37]9993: destructor not called for local object created within and 8570 returned from infinite loop 8571 * [38]10167: ieee_1003.1-2001 locale specialisations on a glibc-2.3.2 8572 system 8573 8574 Java compiler and library: 8575 8576 * [39]9652: libgcj build fails on irix6.5.1[78] 8577 * [40]10144: gas on solaris complains about bad .stabs lines for 8578 java, native as unaffected 8579 8580 x86-specific (Intel/AMD): 8581 8582 * [41]8746: gcc miscompiles Linux kernel ppa driver on x86 8583 * [42]9888: -mcpu=k6 -Os produces out of range loop instructions 8584 * [43]9638: Cross-build for target i386-elf and i586-pc-linux-gnu 8585 failed 8586 * [44]9954: Cross-build for target i586-pc-linux-gnu (--with-newlib) 8587 failed 8588 8589 SPARC-specific: 8590 8591 * [45]7784: [Sparc] ICE in extract_insn, in recog.c 8592 * [46]7796: sparc extra failure with -m64 on execute/930921-1.c in 8593 unroll.c 8594 * [47]8281: ICE when compiling with -O2 -fPIC for Ultrasparc 8595 * [48]8366: [Sparc] C testsuite failure with -m64 -fpic -O in 8596 execute/loop-2d.c 8597 * [49]8726: gcc -O2 miscompiles Samba 2.2.7 on 32-bit sparc 8598 * [50]9414: Scheduling bug on Ultrasparc 8599 * [51]10067: GCC-3.2.2 outputs invalid asm on sparc64 8600 8601 m68k-specific: 8602 8603 * [52]7248: broken "inclusive or" code 8604 * [53]8343: m68k-elf/rtems ICE at instantiate_virtual_regs_1 8605 8606 PowerPC-specific: 8607 8608 * [54]9732: Wrong code with -O2 -fPIC 8609 * [55]10073: ICE: powerpc cannot split insn 8610 8611 Alpha-specific: 8612 8613 * [56]7702: optimization problem on a DEC alpha under OSF1 8614 * [57]9671: gcc.3.2.2 does not build on a HP Tru64 Unix v5.1B system 8615 8616 HP-specific: 8617 8618 * [58]8694: <string> breaks <ctype.h> on HP-UX 10.20 (DUP: 9275) 8619 * [59]9953: (ada) gcc 3.2.x can't build 3.3-branch ada on HP-UX 10 8620 (missing symbol) 8621 * [60]10271: Floating point args don't get reloaded across function 8622 calls with -O2 8623 8624 MIPS specific: 8625 8626 * [61]6362: mips-irix6 gcc-3.1 C testsuite failure with -mips4 in 8627 compile/920501-4.c 8628 8629 CRIS specific: 8630 8631 * [62]10377: gcc-3.2.2 creates bad assembler code for cris 8632 8633 Miscellaneous and minor bugs: 8634 8635 * [63]6955: collect2 says "core dumped" when there is no core 8636 __________________________________________________________________ 8637 8638 GCC 3.2.2 8639 8640 Beginning with 3.2.2, GCC's Makefile suite supports redirection of make 8641 install by means of the DESTDIR variable. Parts of the GCC tree have 8642 featured that support long before, but now it is available even from 8643 the top level. 8644 8645 Other than that, GCC 3.2.2 is a bug fix release only; there are no new 8646 features that were not present in GCC 3.2.1. 8647 8648 Bug Fixes 8649 8650 On the following i386-based systems GCC 3.2.1 broke the C ABI wrt. 8651 functions returning structures: Cygwin, FreeBSD (GCC 3.2.1 as shipped 8652 with FreeBSD 5.0 does not have this problem), Interix, a.out-based 8653 GNU/Linux and NetBSD, OpenBSD, and Darwin. GCC 3.2.2 reverts this ABI 8654 change, and thus restores ABI-compatibility with previous releases 8655 (except GCC 3.2.1) on these platforms. 8656 8657 This section lists the problem reports (PRs) from GCC's bug tracking 8658 system that are known to be fixed in the 3.2.2 release. This list might 8659 not be complete (that is, it is possible that some PRs that have been 8660 fixed are not listed here) and some of the titles have been changed to 8661 make them more clear. 8662 8663 Internal Compiler Errors (multi-platform) 8664 8665 * [64]5919: (c++) ICE when passing variable array to template 8666 function 8667 * [65]7129: (c++) ICE with min/max assignment operators (<?= and >?=) 8668 * [66]7507: ICE with -O2 when address of called function is a 8669 complicated expression 8670 * [67]7622: ICE with nested inline functions if function's address is 8671 taken 8672 * [68]7681: (fortran) ICE in compensate_edge, in reg-stack.c (also PR 8673 [69]9258) 8674 * [70]8031: (c++) ICE in code comparing typeids and casting from 8675 virtual base 8676 * [71]8275: ICE in simplify_subreg 8677 * [72]8332: (c++) builtin strlen/template interaction causes ICE 8678 * [73]8372: (c++) ICE on explicit call of destructor 8679 * [74]8439: (c, not c++) empty struct causes ICE 8680 * [75]8442: (c++) ICE with nested template classes 8681 * [76]8518: ICE when compiling mplayer ("extern inline" issue) 8682 * [77]8615: (c++) ICE with out-of-range character constant template 8683 argument 8684 * [78]8663: (c++) ICE in cp_expr_size, at cp-lang.c:307 8685 * [79]8799: (c++) ICE: error reporting routines re-entered 8686 * [80]9328: (c++) ICE with typeof(X) for overloaded X 8687 * [81]9465: (preprocessor) cpp -traditional ICE on null bytes 8688 8689 C++ (compiler and library) bugs 8690 8691 * [82]47: scoping in nested classes is broken 8692 * [83]6745: problems with iostream rdbuf() member function 8693 * [84]8214: conversion from const char* const to char* sometimes 8694 accepted illegally 8695 * [85]8493: builtin strlen and overload resolution (same bug as 8696 [86]8332) 8697 * [87]8503: strange behaviour of function types 8698 * [88]8727: compiler confused by inheritance from an anonymous struct 8699 * [89]7445: poor performance of std::locale::classic() in 8700 multi-threaded applications 8701 * [90]8230: mishandling of overflow in vector<T>::resize 8702 * [91]8399: sync_with_stdio(false) breaks unformatted input 8703 * [92]8662: illegal access of private member of unnamed class is 8704 accepted 8705 * [93]8707: "make distclean" fails in libstdc++-v3 directory 8706 * [94]8708: __USE_MALLOC doesn't work 8707 * [95]8790: Use of non-thread-safe strtok in src/localename.cc 8708 * [96]8887: Bug in date formats with --enable-clocale=generic 8709 * [97]9076: Call Frame Instructions are not handled correctly during 8710 unwind operation 8711 * [98]9151: std::setprecision limited to 16 digits when outputting a 8712 double to a stream 8713 * [99]9168: codecvt<char, char, mbstate_t> overwrites output buffers 8714 * [100]9269: libstdc++ headers: explicit specialization of function 8715 must precede its first use 8716 * [101]9322: return value of basic_streambuf<>::getloc affected by 8717 locale::global 8718 * [102]9433: segfault in runtime support for dynamic_cast 8719 8720 C and optimizer bugs 8721 8722 * [103]8032: GCC incorrectly initializes static structs that have 8723 flexible arrays 8724 * [104]8639: simple arithmetic expression broken 8725 * [105]8794: optimization improperly eliminates certain expressions 8726 * [106]8832: traditional "asm volatile" code is illegally optimized 8727 * [107]8988: loop optimizer bug: with -O2, code is generated that 8728 segfaults (found on i386, bug present for all platforms) 8729 * [108]9492: structure copy clobbers subsequent stores to structure 8730 8731 Objective-C bugs 8732 8733 * [109]9267: Objective-C parser won't build with newer bison versions 8734 (e.g. 1.875) 8735 8736 Ada bugs 8737 8738 * [110]8344: Ada build problem due to conflict between gcc/final.o, 8739 gcc/ada/final.o 8740 8741 Preprocessor bugs 8742 8743 * [111]8524: _Pragma within macros is improperly expanded 8744 * [112]8880: __WCHAR_TYPE__ macro incorrectly set to "long int" with 8745 -fshort-wchar 8746 8747 ARM-specific 8748 8749 * [113]9090: arm ICE with >= -O2; regression from gcc-2.95 8750 8751 x86-specific (Intel/AMD) 8752 8753 * [114]8588: ICE in extract_insn, at recog.c:NNNN (shift instruction) 8754 * [115]8599: loop unroll bug with -march=k6-3 8755 * [116]9506: ABI breakage in structure return (affects BSD and 8756 Cygwin, but not GNU/Linux) 8757 8758 FreeBSD 5.0 specific 8759 8760 * [117]9484: GCC 3.2.1 Bootstrap failure on FreeBSD 5.0 8761 8762 RTEMS-specific 8763 8764 * [118]9292: hppa1.1-rtems configurery problems 8765 * [119]9293: [m68k-elf/rtems] config/m68k/t-crtstuff bug 8766 * [120]9295: [mips-rtems] config/mips/rtems.h init/fini issue 8767 * [121]9296: gthr-rtems regression 8768 * [122]9316: powerpc-rtems: extending multilibs 8769 8770 HP-PA specific 8771 8772 * [123]9493: ICE with -O2 when building a simple function 8773 8774 Documentation 8775 8776 * [124]7341: hyperlink to gcov in GCC documentation doesn't work 8777 * [125]8947: Please add a warning about "-malign-double" in docs 8778 * [126]7448, [127]8882: typo cleanups 8779 __________________________________________________________________ 8780 8781 GCC 3.2.1 8782 8783 3.2.1 adds a new warning, -Wabi. This option warns when GNU C++ 8784 generates code that is known not to be binary-compatible with the 8785 vendor-neutral ia32/ia64 ABI. Please consult the GCC manual, included 8786 in the distribution, for details. 8787 8788 This release also removes an old GCC extension, "naming types", and the 8789 documentation now directs users to use a different GCC extension, 8790 __typeof__, instead. The feature had evidently been broken for a while. 8791 8792 Otherwise, 3.2.1 is a bug fix release only; other than bug fixes and 8793 the new warning there are no new features that were not present in GCC 8794 3.2. 8795 8796 In addition, the previous fix for [128]PR 7445 (poor performance of 8797 std::locale::classic() in multi-threaded applications) was reverted 8798 ("unfixed"), because the "fix" was not thread-safe. 8799 8800 Bug Fixes 8801 8802 This section lists the problem reports (PRs) from GCC's bug tracking 8803 system that are known to be fixed in the 3.2.1 release. This list might 8804 not be complete (that is, it is possible that some PRs that have been 8805 fixed are not listed here). As you can see, the number of bug fixes is 8806 quite large, so it is strongly recommended that users of earlier GCC 8807 3.x releases upgrade to GCC 3.2.1. 8808 8809 Internal Compiler Errors (multi-platform) 8810 8811 * [129]2521: (c++) ICE in build_ptrmemfunc, in cp/typeck.c 8812 * [130]5661: (c++) ICE instantiating template on array of unknown 8813 size (bad code) 8814 * [131]6419: (c++) ICE in make_decl_rtl for "longest" attribute on 8815 64-bit platforms 8816 * [132]6994: (c++) ICE in find_function_data 8817 * [133]7150: preprocessor: GCC -dM -E gives an ICE 8818 * [134]7160: ICE when optimizing branches without a return value 8819 * [135]7228: (c++) ICE when using member template and template 8820 function 8821 * [136]7266: (c++) ICE with -pedantic on missing typename 8822 * [137]7353: ICE from use of "Naming Types" extension, see above 8823 * [138]7411: ICE in instantiate_virtual_regs_1, in function.c 8824 * [139]7478: (c++) ICE on static_cast inside template 8825 * [140]7526: preprocessor core dump when _Pragma implies #pragma 8826 dependency 8827 * [141]7721: (c++) ICE on simple (but incorrect) template ([142]7803 8828 is a duplicate) 8829 * [143]7754: (c++) ICE on union with template parameter 8830 * [144]7788: (c++) redeclaring a definition as an incomplete class 8831 causes ICE 8832 * [145]8031: (c++) ICE in comptypes, in cp/typeck.c 8833 * [146]8055: preprocessor dies with SIG11 when building FreeBSD 8834 kernel 8835 * [147]8067: (c++) ICE due to mishandling of __FUNCTION__ and related 8836 variables 8837 * [148]8134: (c++) ICE in force_store_init_value on legal code 8838 * [149]8149: (c++) ICE on incomplete type 8839 * [150]8160: (c++) ICE in build_modify_expr, in cp/typeck.c: array 8840 initialization 8841 8842 C++ (compiler and library) bugs 8843 8844 * [151]5607: No pointer adjustment in covariant return types 8845 * [152]6579: Infinite loop with statement expressions in member 8846 initialization 8847 * [153]6803: Default copy constructor bug in GCC 3.1 8848 * [154]7176: g++ confused by friend and static member with same name 8849 * [155]7188: Segfault with template class and recursive (incorrect) 8850 initializer list 8851 * [156]7306: Regression: GCC 3.x fails to compile code with virtual 8852 inheritance if a method has a variable number of arguments 8853 * [157]7461: ctype<char>::classic_table() returns offset array on 8854 Cygwin 8855 * [158]7524: f(const float arg[3]) fails 8856 * [159]7584: Erroneous ambiguous base error on using declaration 8857 * [160]7676: Member template overloading problem 8858 * [161]7679: infinite loop when a right parenthesis is missing 8859 * [162]7811: default locale not taken from environment 8860 * [163]7961: compare( char *) implemented incorrectly in 8861 basic_string<> 8862 * [164]8071: basic_ostream::operator<<(streambuf*) loops forever if 8863 streambuf::underflow() leaves gptr() NULL (dups: [165]8127, 8864 [166]6745) 8865 * [167]8096: deque::at() throws std::range_error instead of 8866 std::out_of_range 8867 * [168]8127: cout << cin.rdbuf() infinite loop 8868 * [169]8218: Excessively large memory consumed for classes with large 8869 array members 8870 * [170]8287: GCC 3.2: Destructor called for non-constructed local 8871 object 8872 * [171]8347: empty vector range used in string construction causes 8873 core dump 8874 * [172]8348: fail() flag is set in istringstream when eof() flag is 8875 set 8876 * [173]8391: regression: infinite loop in cp/decl2.c(finish_file) 8877 8878 C and optimizer bugs 8879 8880 * [174]6627: -fno-align-functions doesn't seem to disable function 8881 alignment 8882 * [175]6631: life_analysis misoptimizes code to initialize fields of 8883 a structure 8884 * [176]7102: unsigned char division results in floating exception 8885 * [177]7120: Run once loop should *always* be unrolled 8886 (pessimization) 8887 * [178]7209: Bug involving array referencing and ?: operator 8888 * [179]7515: invalid inlining of global function with -O3 8889 * [180]7814: incorrect scheduling for glibc-2.2.92 strcpy test 8890 * [181]8467: bug in sibling call optimization 8891 8892 Preprocessor bugs 8893 8894 * [182]4890: incorrect line markers from the traditional preprocessor 8895 * [183]7357: -M option omits system headers files (making it the same 8896 as -MM) 8897 * [184]7358: Changes to Sun's make Dependencies 8898 * [185]7602: C++ header files found in CPLUS_INCLUDE_PATH treated as 8899 C headers 8900 * [186]7862: Interrupting GCC -MD removes .d file but not .o 8901 * [187]8190: Failed compilation deletes -MD dependency file 8902 * [188]8524: _Pragma within macro is improperly expanded 8903 8904 x86 specific (Intel/AMD) 8905 8906 * [189]5351: (i686-only) function pass-by-value structure copy 8907 corrupts stack ([190]7591 is a duplicate) 8908 * [191]6845, [192]7034, [193]7124, [194]7174: ICE's with 8909 -march=pentium3/pentium2/athlon (these are all the same underlying 8910 bug, in MMX register use) 8911 * [195]7134, [196]7375, [197]7390: ICE with -march=athlon (maybe same 8912 as above?) 8913 * [198]6890: xmmintrin.h, _MM_TRANSPOSE4_PS is broken 8914 * [199]6981: wrong code in 64-bit manipulation on x86 8915 * [200]7242: GCC -mcpu=pentium[23] doesn't define __tune_pentiumpro__ 8916 macro 8917 * [201]7396: ix86: cmpgt_ss, cmpge_ss, cmpngt_ss, and cmpnge_ss SSE 8918 intrinsics are broken 8919 * [202]7630: GCC 3.2 breaks on Mozilla 1.0's JS sources with 8920 -march=pentium4 8921 * [203]7693: Typo in i386 mmintrin.h header 8922 * [204]7723: ICE - Pentium3 sse - GCC 3.2 8923 * [205]7951: ICE on -march=pentium4 -O2 -mfpmath=sse 8924 * [206]8146: (i686 only) gcc 3.2 miscompiles gcc 2.95.3 8925 8926 PowerPC specific 8927 8928 * [207]5967: GCC bug when profiling nested functions on powerpc 8929 * [208]6984: wrong code generated with -O2, -O3, -Os for do-while 8930 loop on PowerPC 8931 * [209]7114: PowerPC: ICE building strcoll.op from glibc-2.2.5 8932 * [210]7130: miscompiled code for GCC-3.1 in powerpc linux with 8933 -funroll-all-loops 8934 * [211]7133: PowerPC ICE: unrecognizable insn 8935 * [212]7380: ICE in extract_insn, at recog.c:2148 8936 * [213]8252: ICE on Altivec code with optimization turned on 8937 * [214]8451: Altivec ICE in GCC 3.2 8938 8939 HP/PA specific 8940 8941 * [215]7250: __ashrdi3 returns wrong value on 32 bit hppa 8942 8943 SPARC specific 8944 8945 * [216]6668: when using --disable-multilib, libgcc_s.so is installed 8946 in the wrong place on sparc-solaris 8947 * [217]7151: ICE when compiling for UltraSPARC 8948 * [218]7335: SPARC: ICE in verify_wide_reg (flow.c:557) with long 8949 double and -O1 8950 * [219]7842: [REGRESSION] SPARC code gen bug 8951 8952 ARM specific 8953 8954 * [220]7856: [arm] invalid offset in constant pool reference 8955 * [221]7967: optimization produces wrong code (ARM) 8956 8957 Alpha specific 8958 8959 * [222]7374: __builtin_fabsl broken on alpha 8960 8961 IBM s390 specific 8962 8963 * [223]7370: ICE in fixup_var_refs_1 on s390x 8964 * [224]7409: loop optimization bug on s390x-linux-gnu 8965 * [225]8232: s390x: ICE when using bcmp with int length argument 8966 8967 SCO specific 8968 8969 * [226]7623: SCO OpenServer build fails with machmode.def: undefined 8970 symbol: BITS_PER_UNIT 8971 8972 m68k/Coldfire specific 8973 8974 * [227]8314: crtbegin, crtend need to be multilib'ed for this 8975 platform 8976 8977 Documentation 8978 8979 * [228]761: Document some undocumented options 8980 * [229]5610: Fix documentation about invoking SSE instructions 8981 (-mfpmath=sse) 8982 * [230]7484: List -Wmissing-declarations as C-only option 8983 * [231]7531: -mcmodel not documented for x86-64 8984 * [232]8120: Update documentation of bad use of ## 8985 __________________________________________________________________ 8986 8987 GCC 3.2 8988 8989 3.2 is a small bug fix release, but there is a change to the 8990 application binary interface (ABI), hence the change to the second part 8991 of the version number. 8992 8993 The main purpose of the 3.2 release is to correct a couple of problems 8994 in the C++ ABI, with the intention of providing a stable interface 8995 going forward. Accordingly, 3.2 is only a small change to 3.1.1. 8996 8997 Bug Fixes 8998 8999 C++ 9000 9001 * [233]7320: g++ 3.2 relocation problem 9002 * [234]7470: vtable: virtual function pointers not in declaration 9003 order 9004 9005 libstdc++ 9006 9007 * [235]6410: Trouble with non-ASCII monetary symbols and wchar_t 9008 * [236]6503, [237]6642, [238]7186: Problems with comparing or 9009 subtracting various types of const and non-const iterators 9010 * [239]7216: ambiguity with basic_iostream::traits_type 9011 * [240]7220: problem with basic_istream::ignore(0,delimiter) 9012 * [241]7222: locale::operator==() doesn't work on std::locale("") 9013 * [242]7286: placement operator delete issue 9014 * [243]7442: cxxabi.h does not match the C++ ABI 9015 * [244]7445: poor performance of std::locale::classic() in 9016 multi-threaded applications 9017 9018 x86-64 specific 9019 9020 * [245]7291: off-by-one in generated inline bzero code for x86-64 9021 9022 9023 For questions related to the use of GCC, please consult these web 9024 pages and the [246]GCC manuals. If that fails, the 9025 [247]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 9026 web pages and the development of GCC are welcome on our developer 9027 list at [248]gcc (a] gcc.gnu.org. All of [249]our lists have public 9028 archives. 9029 9030 Copyright (C) [250]Free Software Foundation, Inc. Verbatim copying and 9031 distribution of this entire article is permitted in any medium, 9032 provided this notice is preserved. 9033 9034 These pages are [251]maintained by the GCC team. Last modified 9035 2011-04-25[252]. 9036 9037 References 9038 9039 1. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3 9040 2. http://gcc.gnu.org/gcc-3.1/changes.html 9041 3. http://gcc.gnu.org/gcc-3.2/c++-abi.html 9042 4. http://gcc.gnu.org/PR3782 9043 5. http://gcc.gnu.org/PR6440 9044 6. http://gcc.gnu.org/PR7050 9045 7. http://gcc.gnu.org/PR7741 9046 8. http://gcc.gnu.org/PR7982 9047 9. http://gcc.gnu.org/PR8068 9048 10. http://gcc.gnu.org/PR8178 9049 11. http://gcc.gnu.org/PR8396 9050 12. http://gcc.gnu.org/PR8674 9051 13. http://gcc.gnu.org/PR9768 9052 14. http://gcc.gnu.org/PR9798 9053 15. http://gcc.gnu.org/PR9799 9054 16. http://gcc.gnu.org/PR9928 9055 17. http://gcc.gnu.org/PR10114 9056 18. http://gcc.gnu.org/PR10352 9057 19. http://gcc.gnu.org/PR10336 9058 20. http://gcc.gnu.org/PR8224 9059 21. http://gcc.gnu.org/PR8613 9060 22. http://gcc.gnu.org/PR8828 9061 23. http://gcc.gnu.org/PR9226 9062 24. http://gcc.gnu.org/PR9853 9063 25. http://gcc.gnu.org/PR9797 9064 26. http://gcc.gnu.org/PR9967 9065 27. http://gcc.gnu.org/PR10116 9066 28. http://gcc.gnu.org/PR10171 9067 29. http://gcc.gnu.org/PR10175 9068 30. http://gcc.gnu.org/PR8316 9069 31. http://gcc.gnu.org/PR9169 9070 32. http://gcc.gnu.org/PR9420 9071 33. http://gcc.gnu.org/PR9459 9072 34. http://gcc.gnu.org/PR9507 9073 35. http://gcc.gnu.org/PR9538 9074 36. http://gcc.gnu.org/PR9602 9075 37. http://gcc.gnu.org/PR9993 9076 38. http://gcc.gnu.org/PR10167 9077 39. http://gcc.gnu.org/PR9652 9078 40. http://gcc.gnu.org/PR10144 9079 41. http://gcc.gnu.org/PR8746 9080 42. http://gcc.gnu.org/PR9888 9081 43. http://gcc.gnu.org/PR9638 9082 44. http://gcc.gnu.org/PR9954 9083 45. http://gcc.gnu.org/PR7784 9084 46. http://gcc.gnu.org/PR7796 9085 47. http://gcc.gnu.org/PR8281 9086 48. http://gcc.gnu.org/PR8366 9087 49. http://gcc.gnu.org/PR8726 9088 50. http://gcc.gnu.org/PR9414 9089 51. http://gcc.gnu.org/PR10067 9090 52. http://gcc.gnu.org/PR7248 9091 53. http://gcc.gnu.org/PR8343 9092 54. http://gcc.gnu.org/PR9732 9093 55. http://gcc.gnu.org/PR10073 9094 56. http://gcc.gnu.org/PR7702 9095 57. http://gcc.gnu.org/PR9671 9096 58. http://gcc.gnu.org/PR8694 9097 59. http://gcc.gnu.org/PR9953 9098 60. http://gcc.gnu.org/PR10271 9099 61. http://gcc.gnu.org/PR6362 9100 62. http://gcc.gnu.org/PR10377 9101 63. http://gcc.gnu.org/PR6955 9102 64. http://gcc.gnu.org/PR5919 9103 65. http://gcc.gnu.org/PR7129 9104 66. http://gcc.gnu.org/PR7507 9105 67. http://gcc.gnu.org/PR7622 9106 68. http://gcc.gnu.org/PR7681 9107 69. http://gcc.gnu.org/PR9528 9108 70. http://gcc.gnu.org/PR8031 9109 71. http://gcc.gnu.org/PR8275 9110 72. http://gcc.gnu.org/PR8332 9111 73. http://gcc.gnu.org/PR8372 9112 74. http://gcc.gnu.org/PR8439 9113 75. http://gcc.gnu.org/PR8442 9114 76. http://gcc.gnu.org/PR8518 9115 77. http://gcc.gnu.org/PR8615 9116 78. http://gcc.gnu.org/PR8663 9117 79. http://gcc.gnu.org/PR8799 9118 80. http://gcc.gnu.org/PR9328 9119 81. http://gcc.gnu.org/PR9465 9120 82. http://gcc.gnu.org/PR47 9121 83. http://gcc.gnu.org/PR6745 9122 84. http://gcc.gnu.org/PR8214 9123 85. http://gcc.gnu.org/PR8493 9124 86. http://gcc.gnu.org/PR8332 9125 87. http://gcc.gnu.org/PR8503 9126 88. http://gcc.gnu.org/PR8727 9127 89. http://gcc.gnu.org/PR7445 9128 90. http://gcc.gnu.org/PR8230 9129 91. http://gcc.gnu.org/PR8399 9130 92. http://gcc.gnu.org/PR8662 9131 93. http://gcc.gnu.org/PR8707 9132 94. http://gcc.gnu.org/PR8708 9133 95. http://gcc.gnu.org/PR8790 9134 96. http://gcc.gnu.org/PR8887 9135 97. http://gcc.gnu.org/PR9076 9136 98. http://gcc.gnu.org/PR9151 9137 99. http://gcc.gnu.org/PR9168 9138 100. http://gcc.gnu.org/PR9269 9139 101. http://gcc.gnu.org/PR9322 9140 102. http://gcc.gnu.org/PR9433 9141 103. http://gcc.gnu.org/PR8032 9142 104. http://gcc.gnu.org/PR8639 9143 105. http://gcc.gnu.org/PR8794 9144 106. http://gcc.gnu.org/PR8832 9145 107. http://gcc.gnu.org/PR8988 9146 108. http://gcc.gnu.org/PR9492 9147 109. http://gcc.gnu.org/PR9267 9148 110. http://gcc.gnu.org/PR8344 9149 111. http://gcc.gnu.org/PR8524 9150 112. http://gcc.gnu.org/PR8880 9151 113. http://gcc.gnu.org/PR9090 9152 114. http://gcc.gnu.org/PR8588 9153 115. http://gcc.gnu.org/PR8599 9154 116. http://gcc.gnu.org/PR9506 9155 117. http://gcc.gnu.org/PR9484 9156 118. http://gcc.gnu.org/PR9292 9157 119. http://gcc.gnu.org/PR9293 9158 120. http://gcc.gnu.org/PR9295 9159 121. http://gcc.gnu.org/PR9296 9160 122. http://gcc.gnu.org/PR9316 9161 123. http://gcc.gnu.org/PR9493 9162 124. http://gcc.gnu.org/PR7341 9163 125. http://gcc.gnu.org/PR8947 9164 126. http://gcc.gnu.org/PR7448 9165 127. http://gcc.gnu.org/PR8882 9166 128. http://gcc.gnu.org/PR7445 9167 129. http://gcc.gnu.org/PR2521 9168 130. http://gcc.gnu.org/PR5661 9169 131. http://gcc.gnu.org/PR6419 9170 132. http://gcc.gnu.org/PR6994 9171 133. http://gcc.gnu.org/PR7150 9172 134. http://gcc.gnu.org/PR7160 9173 135. http://gcc.gnu.org/PR7228 9174 136. http://gcc.gnu.org/PR7266 9175 137. http://gcc.gnu.org/PR7353 9176 138. http://gcc.gnu.org/PR7411 9177 139. http://gcc.gnu.org/PR7478 9178 140. http://gcc.gnu.org/PR7526 9179 141. http://gcc.gnu.org/PR7721 9180 142. http://gcc.gnu.org/PR7803 9181 143. http://gcc.gnu.org/PR7754 9182 144. http://gcc.gnu.org/PR7788 9183 145. http://gcc.gnu.org/PR8031 9184 146. http://gcc.gnu.org/PR8055 9185 147. http://gcc.gnu.org/PR8067 9186 148. http://gcc.gnu.org/PR8134 9187 149. http://gcc.gnu.org/PR8149 9188 150. http://gcc.gnu.org/PR8160 9189 151. http://gcc.gnu.org/PR5607 9190 152. http://gcc.gnu.org/PR6579 9191 153. http://gcc.gnu.org/PR6803 9192 154. http://gcc.gnu.org/PR7176 9193 155. http://gcc.gnu.org/PR7188 9194 156. http://gcc.gnu.org/PR7306 9195 157. http://gcc.gnu.org/PR7461 9196 158. http://gcc.gnu.org/PR7524 9197 159. http://gcc.gnu.org/PR7584 9198 160. http://gcc.gnu.org/PR7676 9199 161. http://gcc.gnu.org/PR7679 9200 162. http://gcc.gnu.org/PR7811 9201 163. http://gcc.gnu.org/PR7961 9202 164. http://gcc.gnu.org/PR8071 9203 165. http://gcc.gnu.org/PR8127 9204 166. http://gcc.gnu.org/PR6745 9205 167. http://gcc.gnu.org/PR8096 9206 168. http://gcc.gnu.org/PR8127 9207 169. http://gcc.gnu.org/PR8218 9208 170. http://gcc.gnu.org/PR8287 9209 171. http://gcc.gnu.org/PR8347 9210 172. http://gcc.gnu.org/PR8348 9211 173. http://gcc.gnu.org/PR8391 9212 174. http://gcc.gnu.org/PR6627 9213 175. http://gcc.gnu.org/PR6631 9214 176. http://gcc.gnu.org/PR7102 9215 177. http://gcc.gnu.org/PR7120 9216 178. http://gcc.gnu.org/PR7209 9217 179. http://gcc.gnu.org/PR7515 9218 180. http://gcc.gnu.org/PR7814 9219 181. http://gcc.gnu.org/PR8467 9220 182. http://gcc.gnu.org/PR4890 9221 183. http://gcc.gnu.org/PR7357 9222 184. http://gcc.gnu.org/PR7358 9223 185. http://gcc.gnu.org/PR7602 9224 186. http://gcc.gnu.org/PR7862 9225 187. http://gcc.gnu.org/PR8190 9226 188. http://gcc.gnu.org/PR8524 9227 189. http://gcc.gnu.org/PR5351 9228 190. http://gcc.gnu.org/PR7591 9229 191. http://gcc.gnu.org/PR6845 9230 192. http://gcc.gnu.org/PR7034 9231 193. http://gcc.gnu.org/PR7124 9232 194. http://gcc.gnu.org/PR7174 9233 195. http://gcc.gnu.org/PR7134 9234 196. http://gcc.gnu.org/PR7375 9235 197. http://gcc.gnu.org/PR7390 9236 198. http://gcc.gnu.org/PR6890 9237 199. http://gcc.gnu.org/PR6981 9238 200. http://gcc.gnu.org/PR7242 9239 201. http://gcc.gnu.org/PR7396 9240 202. http://gcc.gnu.org/PR7630 9241 203. http://gcc.gnu.org/PR7693 9242 204. http://gcc.gnu.org/PR7723 9243 205. http://gcc.gnu.org/PR7951 9244 206. http://gcc.gnu.org/PR8146 9245 207. http://gcc.gnu.org/PR5967 9246 208. http://gcc.gnu.org/PR6984 9247 209. http://gcc.gnu.org/PR7114 9248 210. http://gcc.gnu.org/PR7130 9249 211. http://gcc.gnu.org/PR7133 9250 212. http://gcc.gnu.org/PR7380 9251 213. http://gcc.gnu.org/PR8252 9252 214. http://gcc.gnu.org/PR8451 9253 215. http://gcc.gnu.org/PR7250 9254 216. http://gcc.gnu.org/PR6668 9255 217. http://gcc.gnu.org/PR7151 9256 218. http://gcc.gnu.org/PR7335 9257 219. http://gcc.gnu.org/PR7842 9258 220. http://gcc.gnu.org/PR7856 9259 221. http://gcc.gnu.org/PR7967 9260 222. http://gcc.gnu.org/PR7374 9261 223. http://gcc.gnu.org/PR7370 9262 224. http://gcc.gnu.org/PR7409 9263 225. http://gcc.gnu.org/PR8232 9264 226. http://gcc.gnu.org/PR7623 9265 227. http://gcc.gnu.org/PR8314 9266 228. http://gcc.gnu.org/PR761 9267 229. http://gcc.gnu.org/PR5610 9268 230. http://gcc.gnu.org/PR7484 9269 231. http://gcc.gnu.org/PR7531 9270 232. http://gcc.gnu.org/PR8120 9271 233. http://gcc.gnu.org/PR7320 9272 234. http://gcc.gnu.org/PR7470 9273 235. http://gcc.gnu.org/PR6410 9274 236. http://gcc.gnu.org/PR6503 9275 237. http://gcc.gnu.org/PR6642 9276 238. http://gcc.gnu.org/PR7186 9277 239. http://gcc.gnu.org/PR7216 9278 240. http://gcc.gnu.org/PR7220 9279 241. http://gcc.gnu.org/PR7222 9280 242. http://gcc.gnu.org/PR7286 9281 243. http://gcc.gnu.org/PR7442 9282 244. http://gcc.gnu.org/PR7445 9283 245. http://gcc.gnu.org/PR7291 9284 246. http://gcc.gnu.org/onlinedocs/ 9285 247. mailto:gcc-help (a] gcc.gnu.org 9286 248. mailto:gcc (a] gcc.gnu.org 9287 249. http://gcc.gnu.org/lists.html 9288 250. http://www.fsf.org/ 9289 251. http://gcc.gnu.org/about.html 9290 252. http://validator.w3.org/check/referer 9291 ====================================================================== 9292 http://gcc.gnu.org/gcc-3.1/index.html 9293 9294 GCC 3.1 9295 9296 July 27, 2002 9297 9298 The [1]GNU project and the GCC developers are pleased to announce the 9299 release of GCC 3.1.1. 9300 9301 The links below still apply to GCC 3.1.1. 9302 9303 May 15, 2002 9304 9305 The [2]GNU project and the GCC developers are pleased to announce the 9306 release of GCC 3.1. 9307 9308 GCC used to stand for the GNU C Compiler, but since the compiler 9309 supports several other languages aside from C, it now stands for the 9310 GNU Compiler Collection. 9311 9312 A list of [3]successful builds is updated as new information becomes 9313 available. 9314 9315 The GCC developers would like to thank the numerous people that have 9316 contributed [4]new features, improvements, bug fixes, and other changes 9317 as well as test results to GCC. This [5]amazing group of volunteers is 9318 what makes GCC successful. 9319 9320 For additional information about GCC please refer to the [6]GCC project 9321 web site or contact the [7]GCC development mailing list. 9322 9323 To obtain GCC please use [8]our mirror sites, or our CVS server. 9324 __________________________________________________________________ 9325 9326 9327 For questions related to the use of GCC, please consult these web 9328 pages and the [9]GCC manuals. If that fails, the 9329 [10]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 9330 web pages and the development of GCC are welcome on our developer 9331 list at [11]gcc (a] gcc.gnu.org. All of [12]our lists have public 9332 archives. 9333 9334 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and 9335 distribution of this entire article is permitted in any medium, 9336 provided this notice is preserved. 9337 9338 These pages are [14]maintained by the GCC team. Last modified 9339 2011-04-25[15]. 9340 9341 References 9342 9343 1. http://www.gnu.org/ 9344 2. http://www.gnu.org/ 9345 3. http://gcc.gnu.org/gcc-3.1/buildstat.html 9346 4. http://gcc.gnu.org/gcc-3.1/changes.html 9347 5. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 9348 6. http://gcc.gnu.org/index.html 9349 7. mailto:gcc (a] gcc.gnu.org 9350 8. http://gcc.gnu.org/mirrors.html 9351 9. http://gcc.gnu.org/onlinedocs/ 9352 10. mailto:gcc-help (a] gcc.gnu.org 9353 11. mailto:gcc (a] gcc.gnu.org 9354 12. http://gcc.gnu.org/lists.html 9355 13. http://www.fsf.org/ 9356 14. http://gcc.gnu.org/about.html 9357 15. http://validator.w3.org/check/referer 9358 ====================================================================== 9359 http://gcc.gnu.org/gcc-3.1/changes.html 9360 9361 GCC 3.1 Release Series 9362 Changes, New Features, and Fixes 9363 9364 Additional changes in GCC 3.1.1 9365 9366 * A bug related to how structures and unions are returned has been 9367 fixed for powerpc-*-netbsd*. 9368 * An important bug in the implementation of -fprefetch-loop-arrays 9369 has been fixed. Previously the optimization prefetched random 9370 blocks of memory for most targets except for i386. 9371 * The Java compiler now compiles Java programs much faster and also 9372 works with parallel make. 9373 * Nested functions have been fixed for mips*-*-netbsd*. 9374 * Some missing floating point support routines have beed added for 9375 mips*-*-netbsd*. 9376 * This [1]message gives additional information about the bugs fixed 9377 in this release. 9378 9379 Caveats 9380 9381 * The -traditional C compiler option has been deprecated and will be 9382 removed in GCC 3.3. (It remains possible to preprocess non-C code 9383 with the traditional preprocessor.) 9384 * The default debugging format for most ELF platforms (including 9385 GNU/Linux and FreeBSD; notable exception is Solaris) has changed 9386 from stabs to DWARF2. This requires GDB 5.1.1 or later. 9387 9388 General Optimizer Improvements 9389 9390 * Jan Hubicka, SuSE Labs, together with Richard Henderson, Red Hat, 9391 and Andreas Jaeger, SuSE Labs, has contributed [2]infrastructure 9392 for profile driven optimizations. 9393 Options -fprofile-arcs and -fbranch-probabilities can now be used 9394 to improve speed of the generated code by profiling the actual 9395 program behaviour on typical runs. In the absence of profile info 9396 the compiler attempts to guess the profile statically. 9397 * [3]SPEC2000 and SPEC95 benchmark suites are now used daily to 9398 monitor performance of the generated code. 9399 According to the SPECInt2000 results on an AMD Athlon CPU, the code 9400 generated by GCC 3.1 is 6% faster on the average (8.2% faster with 9401 profile feedback) compared to GCC 3.0. The code produced by GCC 3.0 9402 is about 2.1% faster compared to 2.95.3. Tests were done using the 9403 -O2 -march=athlon command-line options. 9404 * Alexandre Oliva, of Red Hat, has generalized the tree inlining 9405 infrastructure developed by CodeSourcery, LLC for the C++ front 9406 end, so that it is now used in the C front end too. Inlining 9407 functions as trees exposes them earlier to the compiler, giving it 9408 more opportunities for optimization. 9409 * Support for data prefetching instructions has been added to the GCC 9410 back end and several targets. A new __builtin_prefetch intrinsic is 9411 available to explicitly insert prefetch instructions and 9412 experimental support for loop array prefetching has been added (see 9413 -fprefetch-loop-array documentation). 9414 * Support for emitting debugging information for macros has been 9415 added for DWARF2. It is activated using -g3. 9416 9417 New Languages and Language specific improvements 9418 9419 C/C++ 9420 9421 * A few more [4]ISO C99 features. 9422 * The preprocessor is 10-50% faster than the preprocessor in GCC 3.0. 9423 * The preprocessor's symbol table has been merged with the symbol 9424 table of the C, C++ and Objective-C front ends. 9425 * The preprocessor consumes less memory than the preprocessor in GCC 9426 3.0, often significantly so. On normal input files, it typically 9427 consumes less memory than pre-3.0 cccp-based GCC, too. 9428 9429 C++ 9430 9431 * -fhonor-std and -fno-honor-std have been removed. -fno-honor-std 9432 was a workaround to allow std compliant code to work with the 9433 non-std compliant libstdc++-v2. libstdc++-v3 is std compliant. 9434 * The C++ ABI has been fixed so that void (A::*)() const is mangled 9435 as "M1AKFvvE", rather than "MK1AFvvE" as before. This change only 9436 affects pointer to cv-qualified member function types. 9437 * The C++ ABI has been changed to correctly handle this code: 9438 struct A { 9439 void operator delete[] (void *, size_t); 9440 }; 9441 9442 struct B : public A { 9443 }; 9444 9445 new B[10]; 9446 9447 The amount of storage allocated for the array will be greater than 9448 it was in 3.0, in order to store the number of elements in the 9449 array, so that the correct size can be passed to operator delete[] 9450 when the array is deleted. Previously, the value passed to operator 9451 delete[] was unpredictable. 9452 This change will only affect code that declares a two-argument 9453 operator delete[] with a second parameter of type size_t in a base 9454 class, and does not override that definition in a derived class. 9455 * The C++ ABI has been changed so that: 9456 struct A { 9457 void operator delete[] (void *, size_t); 9458 void operator delete[] (void *); 9459 }; 9460 9461 does not cause unnecessary storage to be allocated when an array of 9462 A objects is allocated. 9463 This change will only affect code that declares both of these forms 9464 of operator delete[], and declared the two-argument form before the 9465 one-argument form. 9466 * The C++ ABI has been changed so that when a parameter is passed by 9467 value, any cleanup for that parameter is performed in the caller, 9468 as specified by the ia64 C++ ABI, rather than the called function 9469 as before. As a result, classes with a non-trivial destructor but a 9470 trivial copy constructor will be passed and returned by invisible 9471 reference, rather than by bitwise copy as before. 9472 * G++ now supports the "named return value optimization": for code 9473 like 9474 A f () { 9475 A a; 9476 ... 9477 return a; 9478 } 9479 9480 G++ will allocate a in the return value slot, so that the return 9481 becomes a no-op. For this to work, all return statements in the 9482 function must return the same variable. 9483 * Improvements to the C++ library are listed in [5]the libstdc++-v3 9484 FAQ. 9485 9486 Objective-C 9487 9488 * Annoying linker warnings (due to incorrect code being generated) 9489 have been fixed. 9490 * If a class method cannot be found, the compiler no longer issues a 9491 warning if a corresponding instance method exists in the root 9492 class. 9493 * Forward @protocol declarations have been fixed. 9494 * Loading of categories has been fixed in certain situations (GNU run 9495 time only). 9496 * The class lookup in the run-time library has been rewritten so that 9497 class method dispatch is more than twice as fast as it used to be 9498 (GNU run time only). 9499 9500 Java 9501 9502 * libgcj now includes RMI, java.lang.ref.*, javax.naming, and 9503 javax.transaction. 9504 * Property files and other system resources can be compiled into 9505 executables which use libgcj using the new gcj --resource feature. 9506 * libgcj has been ported to more platforms. In particular there is 9507 now a mostly-functional mingw32 (Windows) target port. 9508 * JNI and CNI invocation interfaces were implemented, so gcj-compiled 9509 Java code can now be called from a C/C++ application. 9510 * gcj can now use builtin functions for certain known methods, for 9511 instance Math.cos. 9512 * gcj can now automatically remove redundant array-store checks in 9513 some common cases. 9514 * The --no-store-checks optimization option was added. This can be 9515 used to omit runtime store checks for code which is known not to 9516 throw ArrayStoreException 9517 * The following third party interface standards were added to libgcj: 9518 org.w3c.dom and org.xml.sax. 9519 * java.security has been merged with GNU Classpath. The new package 9520 is now JDK 1.2 compliant, and much more complete. 9521 * A bytecode verifier was added to the libgcj interpreter. 9522 * java.lang.Character was rewritten to comply with the Unicode 3.0 9523 standard, and improve performance. 9524 * Partial support for many more locales was added to libgcj. 9525 * Socket timeouts have been implemented. 9526 * libgcj has been merged into a single shared library. There are no 9527 longer separate shared libraries for the garbage collector and 9528 zlib. 9529 * Several performance improvements were made to gcj and libgcj: 9530 + Hash synchronization (thin locks) 9531 + A special allocation path for finalizer-free objects 9532 + Thread-local allocation 9533 + Parallel GC, and other GC tweaks 9534 9535 Fortran 9536 9537 Fortran improvements are listed in [6]the Fortran documentation. 9538 9539 Ada 9540 9541 [7]Ada Core Technologies, Inc, has contributed its GNAT Ada 95 front 9542 end and associated tools. The GNAT compiler fully implements the Ada 9543 language as defined by the ISO/IEC 8652 standard. 9544 9545 Please note that the integration of the Ada front end is still work in 9546 progress. 9547 9548 New Targets and Target Specific Improvements 9549 9550 * Hans-Peter Nilsson has contributed a port to [8]MMIX, the CPU 9551 architecture used in new editions of Donald E. Knuth's The Art of 9552 Computer Programming. 9553 * [9]Axis Communications has contributed its port to the CRIS CPU 9554 architecture, used in the ETRAX system-on-a-chip series. See 9555 [10]Axis' developer site for technical information. 9556 * Alexandre Oliva, of Red Hat, has contributed a port to the 9557 [11]SuperH SH5 64-bit RISC microprocessor architecture, extending 9558 the existing SH port. 9559 * UltraSPARC is fully supported in 64-bit mode. The option -m64 9560 enables it. 9561 * For compatibility with the Sun compiler #pragma redefine_extname 9562 has been implemented on Solaris. 9563 * The x86 back end has had some noticeable work done to it. 9564 + SuSE Labs developers Jan Hubicka, Bo Thorsen and Andreas 9565 Jaeger have contributed a port to the AMD x86-64 architecture. 9566 For more information on x86-64 see [12]http://www.x86-64.org. 9567 + The compiler now supports MMX, 3DNow!, SSE, and SSE2 9568 instructions. Options -mmmx, -m3dnow, -msse, and -msse2 will 9569 enable the respective instruction sets. Intel C++ compatible 9570 MMX/3DNow!/SSE intrinsics are implemented. SSE2 intrinsics 9571 will be added in next major release. 9572 + Following those improvements, targets for Pentium MMX, K6-2, 9573 K6-3, Pentium III, Pentium 4, and Athlon 4 Mobile/XP/MP were 9574 added. Refer to the documentation on -march= and -mcpu= 9575 options for details. 9576 + For those targets that support it, -mfpmath=sse will cause the 9577 compiler to generate SSE/SSE2 instructions for floating point 9578 math instead of x87 instructions. Usually, this will lead to 9579 quicker code -- especially on the Pentium 4. Note that only 9580 scalar floating point instructions are used and GCC does not 9581 exploit SIMD features yet. 9582 + Prefetch support has been added to the Pentium III, Pentium 4, 9583 K6-2, K6-3, and Athlon series. 9584 + Code generated for floating point to integer conversions has 9585 been improved leading to better performance of many 3D 9586 applications. 9587 * The PowerPC back end has added 64-bit PowerPC GNU/Linux support. 9588 * C++ support for AIX has been improved. 9589 * Aldy Hernandez, of Red Hat, Inc, has contributed extensions to the 9590 PowerPC port supporting the AltiVec programming model (SIMD). The 9591 support, though presently useful, is experimental and is expected 9592 to stabilize for 3.2. The support is written to conform to 9593 Motorola's AltiVec specs. See -maltivec. 9594 9595 Obsolete Systems 9596 9597 Support for a number of older systems has been declared obsolete in GCC 9598 3.1. Unless there is activity to revive them, the next release of GCC 9599 will have their sources permanently removed. 9600 9601 All configurations of the following processor architectures have been 9602 declared obsolete: 9603 * MIL-STD-1750A, 1750a-*-* 9604 * AMD A29k, a29k-*-* 9605 * Convex, c*-convex-* 9606 * Clipper, clipper-*-* 9607 * Elxsi, elxsi-*-* 9608 * Intel i860, i860-*-* 9609 * Sun picoJava, pj-*-* and pjl-*-* 9610 * Western Electric 32000, we32k-*-* 9611 9612 Most configurations of the following processor architectures have been 9613 declared obsolete, but we are preserving a few systems which may have 9614 active developers. It is unlikely that the remaining systems will 9615 survive much longer unless we see definite signs of port activity. 9616 * Motorola 88000 except 9617 + Generic a.out, m88k-*-aout* 9618 + Generic SVR4, m88k-*-sysv4 9619 + OpenBSD, m88k-*-openbsd* 9620 * NS32k except 9621 + NetBSD, ns32k-*-netbsd* 9622 + OpenBSD, ns32k-*-openbsd*. 9623 * ROMP except 9624 + OpenBSD, romp-*-openbsd*. 9625 9626 Finally, only some configurations of these processor architectures are 9627 being obsoleted. 9628 * Alpha: 9629 + OSF/1, alpha*-*-osf[123]*. (Digital Unix and Tru64 Unix, aka 9630 alpha*-*-osf[45], are still supported.) 9631 * ARM: 9632 + RISCiX, arm-*-riscix*. 9633 * i386: 9634 + 386BSD, i?86-*-bsd* 9635 + Chorus, i?86-*-chorusos* 9636 + DG/UX, i?86-*-dgux* 9637 + FreeBSD 1.x, i?86-*-freebsd1.* 9638 + IBM AIX, i?86-*-aix* 9639 + ISC UNIX, i?86-*-isc* 9640 + GNU/Linux with pre-BFD linker, i?86-*-linux*oldld* 9641 + NEXTstep, i?86-next-* 9642 + OSF UNIX, i?86-*-osf1* and i?86-*-osfrose* 9643 + RTEMS/coff, i?86-*-rtemscoff* 9644 + RTEMS/go32, i?86-go32-rtems* 9645 + Sequent/BSD, i?86-sequent-bsd* 9646 + Sequent/ptx before version 3, i?86-sequent-ptx[12]* and 9647 i?86-sequent-sysv3* 9648 + SunOS, i?86-*-sunos* 9649 * Motorola 68000: 9650 + Altos, m68[k0]*-altos-* 9651 + Apollo, m68[k0]*-apollo-* 9652 + Apple A/UX, m68[k0]*-apple-* 9653 + Bull, m68[k0]*-bull-* 9654 + Convergent, m68[k0]*-convergent-* 9655 + Generic SVR3, m68[k0]*-*-sysv3* 9656 + ISI, m68[k0]*-isi-* 9657 + LynxOS, m68[k0]*-*-lynxos* 9658 + NEXT, m68[k0]*-next-* 9659 + RTEMS/coff, m68[k0]*-*-rtemscoff* 9660 + Sony, m68[k0]*-sony-* 9661 * MIPS: 9662 + DEC Ultrix, mips-*-ultrix* and mips-dec-* 9663 + Generic BSD, mips-*-bsd* 9664 + Generic System V, mips-*-sysv* 9665 + IRIX before version 5, mips-sgi-irix[1234]* 9666 + RiscOS, mips-*-riscos* 9667 + Sony, mips-sony-* 9668 + Tandem, mips-tandem-* 9669 * SPARC: 9670 + RTEMS/a.out, sparc-*-rtemsaout*. 9671 9672 Documentation improvements 9673 9674 * The old manual ("Using and Porting the GNU Compiler Collection") 9675 has been replaced by a users manual ("Using the GNU Compiler 9676 Collection") and a separate internals reference manual ("GNU 9677 Compiler Collection Internals"). 9678 * More complete and much improved documentation about GCC's internal 9679 representation used by the C and C++ front ends. 9680 * Many cleanups and improvements in general. 9681 9682 9683 For questions related to the use of GCC, please consult these web 9684 pages and the [13]GCC manuals. If that fails, the 9685 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 9686 web pages and the development of GCC are welcome on our developer 9687 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 9688 archives. 9689 9690 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 9691 distribution of this entire article is permitted in any medium, 9692 provided this notice is preserved. 9693 9694 These pages are [18]maintained by the GCC team. Last modified 9695 2011-04-25[19]. 9696 9697 References 9698 9699 1. http://gcc.gnu.org/ml/gcc/2002-07/msg01208.html 9700 2. http://gcc.gnu.org/news/profiledriven.html 9701 3. http://gcc.gnu.org/benchmarks/ 9702 4. http://gcc.gnu.org/gcc-3.1/c99status.html 9703 5. http://gcc.gnu.org/onlinedocs/libstdc++/faq.html 9704 6. http://gcc.gnu.org/onlinedocs/gcc-3.1.1/g77/News.html 9705 7. http://www.adacore.com/home/ 9706 8. http://www-cs-faculty.stanford.edu/~knuth/mmix.html 9707 9. http://www.axis.com/ 9708 10. http://developer.axis.com/ 9709 11. http://www.superh.com/ 9710 12. http://www.x86-64.org/ 9711 13. http://gcc.gnu.org/onlinedocs/ 9712 14. mailto:gcc-help (a] gcc.gnu.org 9713 15. mailto:gcc (a] gcc.gnu.org 9714 16. http://gcc.gnu.org/lists.html 9715 17. http://www.fsf.org/ 9716 18. http://gcc.gnu.org/about.html 9717 19. http://validator.w3.org/check/referer 9718 ====================================================================== 9719 http://gcc.gnu.org/gcc-3.0/gcc-3.0.html 9720 9721 GCC 3.0.4 9722 9723 February 20, 2002 9724 9725 The [1]GNU project and the GCC developers are pleased to announce the 9726 release of GCC 3.0.4, which is a bug-fix release for the GCC 3.0 9727 series. 9728 9729 GCC used to stand for the GNU C Compiler, but since the compiler 9730 supports several other languages aside from C, it now stands for the 9731 GNU Compiler Collection. 9732 9733 GCC 3.0.x has several new optimizations, new targets, new languages and 9734 many other new features, relative to GCC 2.95.x. See the [2]new 9735 features page for a more complete list. 9736 9737 A list of [3]successful builds is updated as new information becomes 9738 available. 9739 9740 The GCC developers would like to thank the numerous people that have 9741 contributed new features, test results, bug fixes, etc to GCC. This 9742 [4]amazing group of volunteers is what makes GCC successful. 9743 9744 And finally, we can't in good conscience fail to mention some 9745 [5]caveats to using GCC 3.0.x. 9746 9747 For additional information about GCC please refer to the [6]GCC project 9748 web site or contact the [7]GCC development mailing list. 9749 9750 To obtain GCC please use [8]our mirror sites, or our CVS server. 9751 __________________________________________________________________ 9752 9753 Previous 3.0.x Releases 9754 9755 December 20, 2001: GCC 3.0.3 has been released. 9756 October 25, 2001: GCC 3.0.2 has been released. 9757 August 20, 2001: GCC 3.0.1 has been released. 9758 June 18, 2001: GCC 3.0 has been released. 9759 9760 9761 For questions related to the use of GCC, please consult these web 9762 pages and the [9]GCC manuals. If that fails, the 9763 [10]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 9764 web pages and the development of GCC are welcome on our developer 9765 list at [11]gcc (a] gcc.gnu.org. All of [12]our lists have public 9766 archives. 9767 9768 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and 9769 distribution of this entire article is permitted in any medium, 9770 provided this notice is preserved. 9771 9772 These pages are [14]maintained by the GCC team. Last modified 9773 2011-04-25[15]. 9774 9775 References 9776 9777 1. http://www.gnu.org/ 9778 2. http://gcc.gnu.org/gcc-3.0/features.html 9779 3. http://gcc.gnu.org/gcc-3.0/buildstat.html 9780 4. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 9781 5. http://gcc.gnu.org/gcc-3.0/caveats.html 9782 6. http://gcc.gnu.org/index.html 9783 7. mailto:gcc (a] gcc.gnu.org 9784 8. http://gcc.gnu.org/mirrors.html 9785 9. http://gcc.gnu.org/onlinedocs/ 9786 10. mailto:gcc-help (a] gcc.gnu.org 9787 11. mailto:gcc (a] gcc.gnu.org 9788 12. http://gcc.gnu.org/lists.html 9789 13. http://www.fsf.org/ 9790 14. http://gcc.gnu.org/about.html 9791 15. http://validator.w3.org/check/referer 9792 ====================================================================== 9793 http://gcc.gnu.org/gcc-3.0/features.html 9794 9795 GCC 3.0 New Features 9796 9797 Additional changes in GCC 3.0.4 9798 9799 * GCC 3.0 now supports newer versions of the [1]NetBSD operating 9800 system, which use the ELF object file format, on x86 processors. 9801 * Correct debugging information is generated from functions that have 9802 lines from multiple files (e.g. yacc output). 9803 * A fix for whitespace handling in the -traditional preprocessor, 9804 which can affect Fortran. 9805 * Fixes to the exception handling runtime. 9806 * More fixes for bad code generation in C++. 9807 * A fix for shared library generation under AIX 4.3. 9808 * Documentation updates. 9809 * Port of GCC to Tensilica's Xtensa processor contributed. 9810 * A fix for compiling the PPC Linux kernel (FAT fs wouldn't link). 9811 9812 Additional changes in GCC 3.0.3 9813 9814 * A fix to correct an accidental change to the PowerPC ABI. 9815 * Fixes for bad code generation on a variety of architectures. 9816 * Improvements to the debugging information generated for C++ 9817 classes. 9818 * Fixes for bad code generation in C++. 9819 * A fix to avoid crashes in the C++ demangler. 9820 * A fix to the C++ standard library to avoid buffer overflows. 9821 * Miscellaneous improvements for a variety of architectures. 9822 9823 Additional changes in GCC 3.0.2 9824 9825 * Fixes for bad code generation during loop unrolling. 9826 * Fixes for bad code generation by the sibling call optimization. 9827 * Minor improvements to x86 code generation. 9828 * Implementation of function descriptors in C++ vtables for IA64. 9829 * Numerous minor bug-fixes. 9830 9831 Additional changes in GCC 3.0.1 9832 9833 * C++ fixes for incorrect code-generation. 9834 * Improved cross-compiling support for the C++ standard library. 9835 * Fixes for some embedded targets that worked in GCC 2.95.3, but not 9836 in GCC 3.0. 9837 * Fixes for various exception-handling bugs. 9838 * A port to the S/390 architecture. 9839 9840 General Optimizer Improvements 9841 9842 * [2]Basic block reordering pass. 9843 * New if-conversion pass with support for conditional (predicated) 9844 execution. 9845 * New tail call and sibling call elimination optimizations. 9846 * New register renaming pass. 9847 * New (experimental) [3]static single assignment (SSA) representation 9848 support. 9849 * New dead-code elimination pass implemented using the SSA 9850 representation. 9851 * [4]Global null pointer test elimination. 9852 * [5]Global code hoisting/unification. 9853 * More builtins and optimizations for stdio.h, string.h and old BSD 9854 functions, as well as for ISO C99 functions. 9855 * New builtin __builtin_expect for giving hints to the branch 9856 predictor. 9857 9858 New Languages and Language specific improvements 9859 9860 * The GNU Compiler for the Java(TM) language (GCJ) is now integrated 9861 and supported, including the run-time library containing most 9862 common non-GUI Java classes, a bytecode interpreter, and the Boehm 9863 conservative garbage collector. Many bugs have been fixed. GCJ can 9864 compile Java source or Java bytecodes to either native code or Java 9865 class files, and supports native methods written in either the 9866 standard JNI or the more efficient and convenient CNI. 9867 * Here is a [6]partial list of C++ improvements, both new features 9868 and those no longer supported. 9869 * New C++ ABI. On the IA-64 platform GCC is capable of 9870 inter-operating with other IA-64 compilers. 9871 * The new ABI also significantly reduces the size of symbol and debug 9872 information. 9873 * New [7]C++ support library and many C++ bug fixes, vastly improving 9874 our conformance to the ISO C++ standard. 9875 * New [8]inliner for C++. 9876 * Rewritten C preprocessor, integrated into the C, C++ and Objective 9877 C compilers, with very many improvements including ISO C99 support 9878 and [9]improvements to dependency generation. 9879 * Support for more [10]ISO C99 features. 9880 * Many improvements to support for checking calls to format functions 9881 such as printf and scanf, including support for ISO C99 format 9882 features, extensions from the Single Unix Specification and GNU 9883 libc 2.2, checking of strfmon formats and features to assist in 9884 auditing for format string security bugs. 9885 * New warnings for C code that may have undefined semantics because 9886 of violations of sequence point rules in the C standard (such as a 9887 = a++;, a[n] = b[n++]; and a[i++] = i;), included in -Wall. 9888 * Additional warning option -Wfloat-equal. 9889 * Improvements to -Wtraditional. 9890 * Fortran improvements are listed in [11]the Fortran documentation. 9891 9892 New Targets and Target Specific Improvements 9893 9894 * New x86 back-end, generating much improved code. 9895 * Support for a generic i386-elf target contributed. 9896 * New option to emit x86 assembly code using Intel style syntax 9897 (-mintel-syntax). 9898 * HPUX 11 support contributed. 9899 * Improved PowerPC code generation, including scheduled prologue and 9900 epilogue. 9901 * Port of GCC to Intel's IA-64 processor contributed. 9902 * Port of GCC to Motorola's MCore 210 and 340 contributed. 9903 * New unified back-end for Arm, Thumb and StrongArm contributed. 9904 * Port of GCC to Intel's XScale processor contributed. 9905 * Port of GCC to Atmel's AVR microcontrollers contributed. 9906 * Port of GCC to Mitsubishi's D30V processor contributed. 9907 * Port of GCC to Matsushita's AM33 processor (a member of the MN10300 9908 processor family) contributed. 9909 * Port of GCC to Fujitsu's FR30 processor contributed. 9910 * Port of GCC to Motorola's 68HC11 and 68HC12 processors contributed. 9911 * Port of GCC to Sun's picoJava processor core contributed. 9912 9913 Documentation improvements 9914 9915 * Substantially rewritten and improved C preprocessor manual. 9916 * Many improvements to other documentation. 9917 * Manpages for gcc, cpp and gcov are now generated automatically from 9918 the master Texinfo manual, eliminating the problem of manpages 9919 being out of date. (The generated manpages are only extracts from 9920 the full manual, which is provided in Texinfo form, from which 9921 info, HTML, other formats and a printed manual can be generated.) 9922 * Generated info files are included in the release tarballs alongside 9923 their Texinfo sources, avoiding problems on some platforms with 9924 building makeinfo as part of the GCC distribution. 9925 9926 Other significant improvements 9927 9928 * Garbage collection used internally by the compiler for most memory 9929 allocation instead of obstacks. 9930 * Lengauer and Tarjan algorithm used for computing dominators in the 9931 CFG. This algorithm can be significantly faster and more space 9932 efficient than our older algorithm. 9933 * gccbug script provided to assist in submitting bug reports to our 9934 bug tracking system. (Bug reports previously submitted directly to 9935 our mailing lists, for which you received no bug tracking number, 9936 should be submitted again using gccbug if you can reproduce the 9937 problem with GCC 3.0.) 9938 * The internal libgcc library is [12]built as a shared library on 9939 systems that support it. 9940 * Extensive testsuite included with GCC, with many new tests. In 9941 addition to tests for GCC bugs that have been fixed, many tests 9942 have been added for language features, compiler warnings and 9943 builtin functions. 9944 * Additional language-independent warning options -Wpacked, -Wpadded, 9945 -Wunreachable-code and -Wdisabled-optimization. 9946 * Target-independent options -falign-functions, -falign-loops and 9947 -falign-jumps. 9948 9949 Plus a great many bug fixes and almost all the [13]features found in 9950 GCC 2.95. 9951 9952 9953 For questions related to the use of GCC, please consult these web 9954 pages and the [14]GCC manuals. If that fails, the 9955 [15]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 9956 web pages and the development of GCC are welcome on our developer 9957 list at [16]gcc (a] gcc.gnu.org. All of [17]our lists have public 9958 archives. 9959 9960 Copyright (C) [18]Free Software Foundation, Inc. Verbatim copying and 9961 distribution of this entire article is permitted in any medium, 9962 provided this notice is preserved. 9963 9964 These pages are [19]maintained by the GCC team. Last modified 9965 2011-04-25[20]. 9966 9967 References 9968 9969 1. http://www.netbsd.org/ 9970 2. http://gcc.gnu.org/news/reorder.html 9971 3. http://gcc.gnu.org/news/ssa.html 9972 4. http://gcc.gnu.org/news/null.html 9973 5. http://gcc.gnu.org/news/unify.html 9974 6. http://gcc.gnu.org/gcc-3.0/c++features.html 9975 7. http://gcc.gnu.org/libstdc++/ 9976 8. http://gcc.gnu.org/news/inlining.html 9977 9. http://gcc.gnu.org/news/dependencies.html 9978 10. http://gcc.gnu.org/gcc-3.0/c99status.html 9979 11. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html 9980 12. http://gcc.gnu.org/gcc-3.0/libgcc.html 9981 13. http://gcc.gnu.org/gcc-2.95/features.html 9982 14. http://gcc.gnu.org/onlinedocs/ 9983 15. mailto:gcc-help (a] gcc.gnu.org 9984 16. mailto:gcc (a] gcc.gnu.org 9985 17. http://gcc.gnu.org/lists.html 9986 18. http://www.fsf.org/ 9987 19. http://gcc.gnu.org/about.html 9988 20. http://validator.w3.org/check/referer 9989 ====================================================================== 9990 http://gcc.gnu.org/gcc-3.0/caveats.html 9991 9992 GCC 3.0 Caveats 9993 9994 * -fstrict-aliasing is now part of -O2 and higher optimization 9995 levels. This allows the compiler to assume the strictest aliasing 9996 rules applicable to the language being compiled. For C and C++, 9997 this activates optimizations based on the type of expressions. This 9998 optimization may thus break old, non-compliant code. 9999 * Enumerations are now properly promoted to int in function 10000 parameters and function returns. Normally this change is not 10001 visible, but when using -fshort-enums this is an ABI change. 10002 * The undocumented extension that allowed C programs to have a label 10003 at the end of a compound statement has been deprecated and may be 10004 removed in a future version. Programs that now generate a warning 10005 about this may be fixed by adding a null statement (a single 10006 semicolon) after the label. 10007 * The poorly documented extension that allowed string constants in C, 10008 C++ and Objective C to contain unescaped newlines has been 10009 deprecated and may be removed in a future version. Programs using 10010 this extension may be fixed in several ways: the bare newline may 10011 be replaced by \n, or preceded by \n\, or string concatenation may 10012 be used with the bare newline preceded by \n" and " placed at the 10013 start of the next line. 10014 * The Chill compiler is not included in GCC 3.0, because of the lack 10015 of a volunteer to convert it to use garbage collection. 10016 * Certain non-standard iostream methods from earlier versions of 10017 libstdc++ are not included in libstdc++ v3, i.e. filebuf::attach, 10018 ostream::form, and istream::gets. 10019 * The new C++ ABI is not yet fully supported by current (as of 10020 2001-07-01) releases and development versions of GDB, or any 10021 earlier versions. There is a problem setting breakpoints by line 10022 number, and other related issues that have been fixed in GCC 3.0 10023 but not yet handled in GDB: 10024 [1]http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html 10025 10026 10027 For questions related to the use of GCC, please consult these web 10028 pages and the [2]GCC manuals. If that fails, the 10029 [3]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10030 web pages and the development of GCC are welcome on our developer 10031 list at [4]gcc (a] gcc.gnu.org. All of [5]our lists have public archives. 10032 10033 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and 10034 distribution of this entire article is permitted in any medium, 10035 provided this notice is preserved. 10036 10037 These pages are [7]maintained by the GCC team. Last modified 10038 2011-04-25[8]. 10039 10040 References 10041 10042 1. http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html 10043 2. http://gcc.gnu.org/onlinedocs/ 10044 3. mailto:gcc-help (a] gcc.gnu.org 10045 4. mailto:gcc (a] gcc.gnu.org 10046 5. http://gcc.gnu.org/lists.html 10047 6. http://www.fsf.org/ 10048 7. http://gcc.gnu.org/about.html 10049 8. http://validator.w3.org/check/referer 10050 ====================================================================== 10051 http://gcc.gnu.org/gcc-2.95/index.html 10052 10053 GCC 2.95 10054 10055 March 16, 2001: The GNU project and the GCC developers are pleased to 10056 announce the release of GCC version 2.95.3. 10057 10058 Release History 10059 10060 GCC 2.95.3 10061 March 16, 2001 10062 10063 GCC 2.95.2 10064 October 27, 1999 10065 10066 GCC 2.95.1 10067 August 19, 1999 10068 10069 GCC 2.95 10070 July 31, 1999. This is the first release of GCC since the April 10071 1999 GCC/EGCS reunification and includes nearly a year's worth 10072 of new development and bugfixes. 10073 10074 References and Acknowledgements 10075 10076 GCC used to stand for the GNU C Compiler, but since the compiler 10077 supports several other languages aside from C, it now stands for the 10078 GNU Compiler Collection. 10079 10080 The whole suite has been extensively [1]regression tested and 10081 [2]package tested. It should be reliable and suitable for widespread 10082 use. 10083 10084 The compiler has several new optimizations, new targets, new languages 10085 and other new features. See the [3]new features page for a more 10086 complete list of new features found in the GCC 2.95 releases. 10087 10088 The sources include installation instructions in both HTML and 10089 plaintext forms in the install directory in the distribution. However, 10090 the most up to date [4]installation instructions and [5]build/test 10091 status are on the web pages. We will update those pages as new 10092 information becomes available. 10093 10094 The GCC developers would like to thank the numerous people that have 10095 contributed new features, test results, bugfixes, etc to GCC. This 10096 [6]amazing group of volunteers is what makes GCC successful. 10097 10098 And finally, we can't in good conscience fail to mention some 10099 [7]caveats to using GCC 2.95. 10100 10101 Download GCC 2.95 from one of our many [8]mirror sites. 10102 10103 For additional information about GCC please see the [9]GCC project web 10104 server or contact the [10]GCC development mailing list. 10105 10106 10107 For questions related to the use of GCC, please consult these web 10108 pages and the [11]GCC manuals. If that fails, the 10109 [12]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10110 web pages and the development of GCC are welcome on our developer 10111 list at [13]gcc (a] gcc.gnu.org. All of [14]our lists have public 10112 archives. 10113 10114 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and 10115 distribution of this entire article is permitted in any medium, 10116 provided this notice is preserved. 10117 10118 These pages are [16]maintained by the GCC team. Last modified 10119 2011-04-25[17]. 10120 10121 References 10122 10123 1. http://gcc.gnu.org/gcc-2.95/regress.html 10124 2. http://gcc.gnu.org/gcc-2.95/othertest.html 10125 3. http://gcc.gnu.org/gcc-2.95/features.html 10126 4. http://gcc.gnu.org/install/ 10127 5. http://gcc.gnu.org/gcc-2.95/buildstat.html 10128 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 10129 7. http://gcc.gnu.org/gcc-2.95/caveats.html 10130 8. http://gcc.gnu.org/mirrors.html 10131 9. http://gcc.gnu.org/index.html 10132 10. mailto:gcc (a] gcc.gnu.org 10133 11. http://gcc.gnu.org/onlinedocs/ 10134 12. mailto:gcc-help (a] gcc.gnu.org 10135 13. mailto:gcc (a] gcc.gnu.org 10136 14. http://gcc.gnu.org/lists.html 10137 15. http://www.fsf.org/ 10138 16. http://gcc.gnu.org/about.html 10139 17. http://validator.w3.org/check/referer 10140 ====================================================================== 10141 http://gcc.gnu.org/gcc-2.95/features.html 10142 10143 GCC 2.95 New Features 10144 10145 * General Optimizer Improvements: 10146 + [1]Localized register spilling to improve speed and code 10147 density especially on small register class machines. 10148 + [2]Global CSE using lazy code motion algorithms. 10149 + [3]Improved global constant/copy propagation. 10150 + [4]Improved control flow graph analysis and manipulation. 10151 + [5]Local dead store elimination. 10152 + [6]Memory Load hoisting/store sinking in loops. 10153 + [7]Type based alias analysis is enabled by default. Note this 10154 feature will expose bugs in the Linux kernel. Please refer to 10155 the FAQ (as shipped with GCC 2.95) for additional information 10156 on this issue. 10157 + Major revamp of GIV detection, combination and simplification 10158 to improve loop performance. 10159 + Major improvements to register allocation and reloading. 10160 * New Languages and Language specific improvements 10161 + [8]Many C++ improvements. 10162 + [9]Many Fortran improvements. 10163 + [10]Java front-end has been integrated. [11]runtime library is 10164 available separately. 10165 + [12]ISO C99 support 10166 + [13]Chill front-end and runtime has been integrated. 10167 + Boehm garbage collector support in libobjc. 10168 + More support for various pragmas which appear in vendor 10169 include files 10170 * New Targets and Target Specific Improvements 10171 + [14]SPARC backend rewrite. 10172 + -mschedule=8000 will optimize code for PA8000 class 10173 processors; -mpa-risc-2-0 will generate code for PA2.0 10174 processors 10175 + Various micro-optimizations for the ia32 port. K6 10176 optimizations 10177 + Compiler will attempt to align doubles in the stack on the 10178 ia32 port 10179 + Alpha EV6 support 10180 + PowerPC 750 10181 + RS6000/PowerPC: -mcpu=401 was added as an alias for -mcpu=403. 10182 -mcpu=e603e was added to do -mcpu=603e and -msoft-float. 10183 + c3x, c4x 10184 + HyperSPARC 10185 + SparcLite86x 10186 + sh4 10187 + Support for new systems (OpenBSD, FreeBSD, UWIN, Interix, 10188 arm-linux) 10189 + vxWorks targets include support for vxWorks threads 10190 + StrongARM 110 and ARM9 support added. ARM Scheduling 10191 parameters rewritten. 10192 + Various changes to the MIPS port to avoid assembler macros, 10193 which in turn improves performance 10194 + Various performance improvements to the i960 port. 10195 + Major rewrite of ns32k port 10196 * Other significant improvements 10197 + [15]Ability to dump cfg information and display it using vcg. 10198 + The new faster scheme for fixing vendor header files is 10199 enabled by default. 10200 + Experimental internationalization support. 10201 + multibyte character support 10202 + Some compile-time speedups for pathological problems 10203 + Better support for complex types 10204 * Plus the usual mountain of bugfixes 10205 * Core compiler is based on the gcc2 development tree from Sept 30, 10206 1998, so we have all of the [16]features found in GCC 2.8. 10207 10208 Additional Changes in GCC 2.95.1 10209 10210 * Generic bugfixes and improvements 10211 + Various documentation fixes related to the GCC/EGCS merger. 10212 + Fix memory management bug which could lead to spurious aborts, 10213 core dumps or random parsing errors in the compiler. 10214 + Fix a couple bugs in the dwarf1 and dwarf2 debug record 10215 support. 10216 + Fix infinite loop in the CSE optimizer. 10217 + Avoid undefined behavior in compiler FP emulation code 10218 + Fix install problem when prefix is overridden on the make 10219 install command. 10220 + Fix problem with unwanted installation of assert.h on some 10221 systems. 10222 + Fix problem with finding the wrong assembler in a single tree 10223 build. 10224 + Avoid increasing the known alignment of a register that is 10225 already known to be a pointer. 10226 * Platform specific bugfixes and improvements 10227 + Codegen bugfix for prologue/epilogue for cpu32 target. 10228 + Fix long long code generation bug for the Coldfire target. 10229 + Fix various aborts in the SH compiler. 10230 + Fix bugs in libgcc support library for the SH. 10231 + Fix alpha ev6 code generation bug. 10232 + Fix problems with EXIT_SUCCESS/EXIT_FAILURE redefinitions on 10233 AIX platforms. 10234 + Fix -fpic code generation bug for rs6000/ppc svr4 targets. 10235 + Fix varargs/stdarg code generation bug for rs6000/ppc svr4 10236 targets. 10237 + Fix weak symbol handling for rs6000/ppc svr4 targets. 10238 + Fix various problems with 64bit code generation for the 10239 rs6000/ppc port. 10240 + Fix codegen bug which caused tetex to be mis-compiled on the 10241 x86. 10242 + Fix compiler abort in new cfg code exposed by x86 port. 10243 + Fix out of range array reference in code convert flat 10244 registers to the x87 stacked FP register file. 10245 + Fix minor vxworks configuration bug. 10246 + Fix return type of bsearch for SunOS 4.x. 10247 * Language & Runtime specific fixes. 10248 + The G++ signature extension has been deprecated. It will be 10249 removed in the next major release of G++. Use of signatures 10250 will result in a warning from the compiler. 10251 + Several bugs relating to templates and namespaces were fixed. 10252 + A bug that caused crashes when combining templates with -g on 10253 DWARF1 platforms was fixed. 10254 + Pointers-to-members, virtual functions, and multiple 10255 inheritance should now work together correctly. 10256 + Some code-generation bugs relating to function try blocks were 10257 fixed. 10258 + G++ is a little bit more lenient with certain archaic 10259 constructs than in GCC 2.95. 10260 + Fix to prevent shared library version #s from bring truncated 10261 to 1 digit 10262 + Fix missing std:: in the libstdc++ library. 10263 + Fix stream locking problems in libio. 10264 + Fix problem in java compiler driver. 10265 10266 Additional Changes in GCC 2.95.2 10267 10268 The -fstrict-aliasing is not enabled by default for GCC 2.95.2. While 10269 the optimizations performed by -fstrict-aliasing are valid according to 10270 the C and C++ standards, the optimization have caused some problems, 10271 particularly with old non-conforming code. 10272 10273 The GCC developers are experimenting with ways to warn users about code 10274 which violates the C/C++ standards, but those warnings are not ready 10275 for widespread use at this time. Rather than wait for those warnings 10276 the GCC developers have chosen to disable -fstrict-aliasing by default 10277 for the GCC 2.95.2 release. 10278 10279 We strongly encourage developers to find and fix code which violates 10280 the C/C++ standards as -fstrict-aliasing may be enabled by default in 10281 future releases. Use the option -fstrict-aliasing to re-enable these 10282 optimizations. 10283 * Generic bugfixes and improvements 10284 + Fix incorrectly optimized memory reference in global common 10285 subexpression elimination (GCSE) optimization pass. 10286 + Fix code generation bug in regmove.c in which it could 10287 incorrectly change a "const" value. 10288 + Fix bug in optimization of conditionals involving volatile 10289 memory references. 10290 + Avoid over-allocation of stack space for some procedures. 10291 + Fixed bug in the compiler which caused incorrect optimization 10292 of an obscure series of bit manipulations, shifts and 10293 arithmetic. 10294 + Fixed register allocator bug which caused teTeX to be 10295 mis-compiled on SPARC targets. 10296 + Avoid incorrect optimization of degenerate case statements for 10297 certain targets such as the ARM. 10298 + Fix out of range memory reference in the jump optimizer. 10299 + Avoid dereferencing null pointer in fix-header. 10300 + Fix test for GCC specific features so that it is possible to 10301 bootstrap with gcc-2.6.2 and older versions of GCC. 10302 + Fix typo in scheduler which could potentially cause out of 10303 range memory accesses. 10304 + Avoid incorrect loop reversal which caused incorrect code for 10305 certain loops on PowerPC targets. 10306 + Avoid incorrect optimization of switch statements on certain 10307 targets (for example the ARM). 10308 * Platform specific bugfixes and improvements 10309 + Work around bug in Sun V5.0 compilers which caused bootstrap 10310 comparison failures on SPARC targets. 10311 + Fix SPARC backend bug which caused aborts in final.c. 10312 + Fix sparc-hal-solaris2* configuration fragments. 10313 + Fix bug in sparc block profiling. 10314 + Fix obscure code generation bug for the PARISC targets. 10315 + Define __STDC_EXT__ for HPUX configurations. 10316 + Various POWERPC64 code generation bugfixes. 10317 + Fix abort for PPC targets using ELF (ex GNU/Linux). 10318 + Fix collect2 problems for AIX targets. 10319 + Correct handling of .file directive for PPC targets. 10320 + Fix bug in fix_trunc x86 patterns. 10321 + Fix x86 port to correctly pop the FP stack for functions that 10322 return structures in memory. 10323 + Fix minor bug in strlen x86 pattern. 10324 + Use stabs debugging instead of dwarf1 for x86-solaris targets. 10325 + Fix template repository code to handle leading underscore in 10326 mangled names. 10327 + Fix weak/weak alias support for OpenBSD. 10328 + GNU/Linux for the ARM has C++ compatible include files. 10329 * Language & Runtime specific fixes. 10330 + Fix handling of constructor attribute in the C front-end which 10331 caused problems building the Chill runtime library on some 10332 targets. 10333 + Fix minor problem merging type qualifiers in the C front-end. 10334 + Fix aliasing bug for pointers and references (C/C++). 10335 + Fix incorrect "non-constant initializer bug" when -traditional 10336 or -fwritable-strings is enabled. 10337 + Fix build error for Chill front-end on SunOS. 10338 + Do not complain about duplicate instantiations when using 10339 -frepo (C++). 10340 + Fix array bounds handling in C++ front-end which caused 10341 problems with dwarf debugging information in some 10342 circumstances. 10343 + Fix minor namespace problem. 10344 + Fix problem linking java programs. 10345 10346 Additional Changes in GCC 2.95.3 10347 10348 * Generic bugfixes and improvements 10349 + Fix numerous problems that caused incorrect optimization in 10350 the register reloading code. 10351 + Fix numerous problems that caused incorrect optimization in 10352 the loop optimizer. 10353 + Fix aborts in the functions build_insn_chain and scan_loops 10354 under some circumstances. 10355 + Fix an alias analysis bug. 10356 + Fix an infinite compilation bug in the combiner. 10357 + A few problems with complex number support have been fixed. 10358 + It is no longer possible for gcc to act as a fork bomb when 10359 installed incorrectly. 10360 + The -fpack-struct option should be recognized now. 10361 + Fixed a bug that caused incorrect code to be generated due to 10362 a lost stack adjustment. 10363 * Platform specific bugfixes and improvements 10364 + Support building ARM toolchains hosted on Windows. 10365 + Fix attribute calculations in ARM toolchains. 10366 + arm-linux support has been improved. 10367 + Fix a PIC failure on sparc targets. 10368 + On ix86 targets, the regparm attribute should now work 10369 reliably. 10370 + Several updates for the h8300 port. 10371 + Fix problem building libio with glibc 2.2. 10372 10373 10374 For questions related to the use of GCC, please consult these web 10375 pages and the [17]GCC manuals. If that fails, the 10376 [18]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10377 web pages and the development of GCC are welcome on our developer 10378 list at [19]gcc (a] gcc.gnu.org. All of [20]our lists have public 10379 archives. 10380 10381 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and 10382 distribution of this entire article is permitted in any medium, 10383 provided this notice is preserved. 10384 10385 These pages are [22]maintained by the GCC team. Last modified 10386 2011-04-25[23]. 10387 10388 References 10389 10390 1. http://gcc.gnu.org/news/spill.html 10391 2. http://gcc.gnu.org/news/lcm.html 10392 3. http://gcc.gnu.org/news/cprop.html 10393 4. http://gcc.gnu.org/news/cfg.html 10394 5. http://gcc.gnu.org/news/dse.html 10395 6. http://gcc.gnu.org/news/hoist.html 10396 7. http://gcc.gnu.org/news/alias.html 10397 8. http://gcc.gnu.org/gcc-2.95/c++features.html 10398 9. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html 10399 10. http://gcc.gnu.org/java/gcj-announce.txt 10400 11. http://gcc.gnu.org/news/javaannounce.html 10401 12. http://gcc.gnu.org/c99status.html 10402 13. http://gcc.gnu.org/news/chill.html 10403 14. http://gcc.gnu.org/news/sparc.html 10404 15. http://gcc.gnu.org/news/egcs-vcg.html 10405 16. http://gcc.gnu.org/egcs-1.0/features-2.8.html 10406 17. http://gcc.gnu.org/onlinedocs/ 10407 18. mailto:gcc-help (a] gcc.gnu.org 10408 19. mailto:gcc (a] gcc.gnu.org 10409 20. http://gcc.gnu.org/lists.html 10410 21. http://www.fsf.org/ 10411 22. http://gcc.gnu.org/about.html 10412 23. http://validator.w3.org/check/referer 10413 ====================================================================== 10414 http://gcc.gnu.org/gcc-2.95/caveats.html 10415 10416 GCC 2.95 Caveats 10417 10418 * GCC 2.95 will issue an error for invalid asm statements that had 10419 been silently accepted by earlier versions of the compiler. This is 10420 particularly noticeable when compiling older versions of the Linux 10421 kernel (2.0.xx). Please refer to the FAQ (as shipped with GCC 2.95) 10422 for more information on this issue. 10423 * GCC 2.95 implements type based alias analysis to disambiguate 10424 memory references. Some programs, particularly the Linux kernel 10425 violate ANSI/ISO aliasing rules and therefore may not operate 10426 correctly when compiled with GCC 2.95. Please refer to the FAQ (as 10427 shipped with GCC 2.95) for more information on this issue. 10428 * GCC 2.95 has a known bug in its handling of complex variables for 10429 64bit targets. Instead of silently generating incorrect code, GCC 10430 2.95 will issue a fatal error for situations it can not handle. 10431 This primarily affects the Fortran community as Fortran makes more 10432 use of complex variables than C or C++. 10433 * GCC 2.95 has an integrated libstdc++, but does not have an 10434 integrated libg++. Furthermore old libg++ releases will not work 10435 with GCC 2.95. You can retrieve a recent copy of libg++ from the 10436 [1]GCC ftp server. 10437 Note most C++ programs only need libstdc++. 10438 * Exception handling may not work with shared libraries, particularly 10439 on alphas, hppas, rs6000/powerpc and mips based platforms. 10440 Exception handling is known to work on x86 GNU/Linux platforms with 10441 shared libraries. 10442 * In general, GCC 2.95 is more rigorous about rejecting invalid C++ 10443 code or deprecated C++ constructs than G++ 2.7, G++ 2.8, EGCS 1.0, 10444 or EGCS 1.1. As a result it may be necessary to fix C++ code before 10445 it will compile with GCC 2.95. 10446 * G++ is also converting toward the ISO C++ standard; as a result 10447 code which was previously valid (and thus accepted by other 10448 compilers and older versions of g++) may no longer be accepted. The 10449 flag -fpermissive may allow some non-conforming code to compile 10450 with GCC 2.95. 10451 * GCC 2.95 compiled C++ code is not binary compatible with EGCS 10452 1.1.x, EGCS 1.0.x or GCC 2.8.x. 10453 * GCC 2.95 does not have changes from the GCC 2.8 tree that were made 10454 between Sept 30, 1998 and April 30, 1999 (the official end of the 10455 GCC 2.8 project). Future GCC releases will include all the changes 10456 from the defunct GCC 2.8 sources. 10457 10458 10459 For questions related to the use of GCC, please consult these web 10460 pages and the [2]GCC manuals. If that fails, the 10461 [3]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10462 web pages and the development of GCC are welcome on our developer 10463 list at [4]gcc (a] gcc.gnu.org. All of [5]our lists have public archives. 10464 10465 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and 10466 distribution of this entire article is permitted in any medium, 10467 provided this notice is preserved. 10468 10469 These pages are [7]maintained by the GCC team. Last modified 10470 2011-04-25[8]. 10471 10472 References 10473 10474 1. ftp://gcc.gnu.org/pub/gcc/infrastructure/libg++-2.8.1.3.tar.gz 10475 2. http://gcc.gnu.org/onlinedocs/ 10476 3. mailto:gcc-help (a] gcc.gnu.org 10477 4. mailto:gcc (a] gcc.gnu.org 10478 5. http://gcc.gnu.org/lists.html 10479 6. http://www.fsf.org/ 10480 7. http://gcc.gnu.org/about.html 10481 8. http://validator.w3.org/check/referer 10482 ====================================================================== 10483 http://gcc.gnu.org/egcs-1.1/index.html 10484 10485 EGCS 1.1 10486 10487 September 3, 1998: We are pleased to announce the release of EGCS 1.1. 10488 December 1, 1998: We are pleased to announce the release of EGCS 1.1.1. 10489 March 15, 1999: We are pleased to announce the release of EGCS 1.1.2. 10490 10491 EGCS is a free software project to further the development of the GNU 10492 compilers using an open development environment. 10493 10494 EGCS 1.1 is a major new release of the EGCS compiler system. It has 10495 been [1]extensively tested and is believed to be stable and suitable 10496 for widespread use. 10497 10498 EGCS 1.1 is based on an June 6, 1998 snapshot of the GCC 2.8 10499 development sources; it contains all of the new features found in GCC 10500 2.8.1 as well as all new development from GCC up to June 6, 1998. 10501 10502 EGCS 1.1 also contains many improvements and features not found in GCC 10503 or in older versions of EGCS: 10504 * Global common subexpression elimination and global constant/copy 10505 propagation (aka [2]gcse) 10506 * Ongoing improvements to the [3]alias analysis support to allow for 10507 better optimizations throughout the compiler. 10508 * Vastly improved [4]C++ compiler and integrated C++ runtime 10509 libraries. 10510 * Fixes for the /tmp symlink race security problems. 10511 * New targets including mips16, arm-thumb and 64 bit PowerPC. 10512 * Improvements to GNU Fortran (g77) compiler and runtime library made 10513 since g77 version 0.5.23. 10514 10515 See the [5]new features page for a more complete list of new features 10516 found in EGCS 1.1 releases. 10517 10518 EGCS 1.1.1 is a minor update to fix several serious problems in EGCS 10519 1.1: 10520 * General improvements and fixes 10521 + Avoid some stack overflows when compiling large functions. 10522 + Avoid incorrect loop invariant code motions. 10523 + Fix some core dumps on Linux kernel code. 10524 + Bring back the imake -Di386 and friends fix from EGCS 1.0.2. 10525 + Fix code generation problem in gcse. 10526 + Various documentation related fixes. 10527 * g++/libstdc++ improvements and fixes 10528 + MT safe EH fix for setjmp/longjmp based exception handling. 10529 + Fix a few bad interactions between optimization and exception 10530 handling. 10531 + Fixes for demangling of template names starting with "__". 10532 + Fix a bug that would fail to run destructors in some cases 10533 with -O2. 10534 + Fix 'new' of classes with virtual bases. 10535 + Fix crash building Qt on the Alpha. 10536 + Fix failure compiling WIFEXITED macro on GNU/Linux. 10537 + Fix some -frepo failures. 10538 * g77 and libf2c improvements and fixes 10539 + Various documentation fixes. 10540 + Avoid compiler crash on RAND intrinsic. 10541 + Fix minor bugs in makefiles exposed by BSD make programs. 10542 + Define _XOPEN_SOURCE for libI77 build to avoid potential 10543 problems on some 64-bit systems. 10544 + Fix problem with implicit endfile on rewind. 10545 + Fix spurious recursive I/O errors. 10546 * platform specific improvements and fixes 10547 + Match all versions of UnixWare7. 10548 + Do not assume x86 SVR4 or UnixWare targets can handle stabs. 10549 + Fix PPC/RS6000 LEGITIMIZE_ADDRESS macro and bug in conversion 10550 from unsigned ints to double precision floats. 10551 + Fix ARM ABI issue with NetBSD. 10552 + Fix a few arm code generation bugs. 10553 + Fixincludes will fix additional broken SCO OpenServer header 10554 files. 10555 + Fix a m68k backend bug which caused invalid offsets in reg+d 10556 addresses. 10557 + Fix problems with 64bit AIX 4.3 support. 10558 + Fix handling of long longs for varargs/stdarg functions on the 10559 ppc. 10560 + Minor fixes to CPP predefines for Windows. 10561 + Fix code generation problems with gpr<->fpr copies for 64bit 10562 ppc. 10563 + Fix a few coldfire code generation bugs. 10564 + Fix some more header file problems on SunOS 4.x. 10565 + Fix assert.h handling for RTEMS. 10566 + Fix Windows handling of TREE_SYMBOL_REFERENCED. 10567 + Fix x86 compiler abort in reg-stack pass. 10568 + Fix cygwin/windows problem with section attributes. 10569 + Fix Alpha code generation problem exposed by SMP Linux 10570 kernels. 10571 + Fix typo in m68k 32->64bit integer conversion. 10572 + Make sure target libraries build with -fPIC for PPC & Alpha 10573 targets. 10574 10575 EGCS 1.1.2 is a minor update to fix several serious problems in EGCS 10576 1.1.1: 10577 * General improvements and fixes 10578 + Fix bug in loop optimizer which caused the SPARC (and 10579 potentially other) ports to segfault. 10580 + Fix infinite recursion in alias analysis and combiner code. 10581 + Fix bug in regclass preferencing. 10582 + Fix incorrect loop reversal which caused incorrect code to be 10583 generated for several targets. 10584 + Fix return value for builtin memcpy. 10585 + Reduce compile time for certain loops which exposed quadratic 10586 behavior in the loop optimizer. 10587 + Fix bug which caused volatile memory to be written multiple 10588 times when only one write was needed/desired. 10589 + Fix compiler abort in caller-save.c 10590 + Fix combiner bug which caused incorrect code generation for 10591 certain division by constant operations. 10592 + Fix incorrect code generation due to a bug in range check 10593 optimizations. 10594 + Fix incorrect code generation due to mis-handling of clobbered 10595 values in CSE. 10596 + Fix compiler abort/segfault due to incorrect register 10597 splitting when unrolling loops. 10598 + Fix code generation involving autoincremented addresses with 10599 ternary operators. 10600 + Work around bug in the scheduler which caused qt to be 10601 mis-compiled on some platforms. 10602 + Fix code generation problems with -fshort-enums. 10603 + Tighten security for temporary files. 10604 + Improve compile time for codes which make heavy use of 10605 overloaded functions. 10606 + Fix multiply defined constructor/destructor symbol problems. 10607 + Avoid setting bogus RPATH environment variable during 10608 bootstrap. 10609 + Avoid GNU-make dependencies in the texinfo subdir. 10610 + Install CPP wrapper script in $(prefix)/bin if --enable-cpp. 10611 --enable-cpp=<dirname> can be used to specify an additional 10612 install directory for the cpp wrapper script. 10613 + Fix CSE bug which caused incorrect label-label refs to appear 10614 on some platforms. 10615 + Avoid linking in EH routines from libgcc if they are not 10616 needed. 10617 + Avoid obscure bug in aliasing code. 10618 + Fix bug in weak symbol handling. 10619 * Platform-specific improvements and fixes 10620 + Fix detection of PPro/PII on Unixware 7. 10621 + Fix compiler segfault when building spec99 and other programs 10622 for SPARC targets. 10623 + Fix code-generation bugs for integer and floating point 10624 conditional move instructions on the PPro/PII. 10625 + Use fixincludes to fix byteorder problems on i?86-*-sysv. 10626 + Fix build failure for the arc port. 10627 + Fix floating point format configuration for i?86-gnu port. 10628 + Fix problems with hppa1.0-hp-hpux10.20 configuration when 10629 threads are enabled. 10630 + Fix coldfire code generation bugs. 10631 + Fix "unrecognized insn" problems for Alpha and PPC ports. 10632 + Fix h8/300 code generation problem with floating point values 10633 in memory. 10634 + Fix unrecognized insn problems for the m68k port. 10635 + Fix namespace-pollution problem for the x86 port. 10636 + Fix problems with old assembler on x86 NeXT systems. 10637 + Fix PIC code-generation problems for the SPARC port. 10638 + Fix minor bug with LONG_CALLS in PowerPC SVR4 support. 10639 + Fix minor ISO namespace violation in Alpha varargs/stdarg 10640 support. 10641 + Fix incorrect "braf" instruction usage for the SH port. 10642 + Fix minor bug in va-sh which prevented its use with -ansi. 10643 + Fix problems recognizing and supporting FreeBSD. 10644 + Handle OpenBSD systems correctly. 10645 + Minor fixincludes fix for Digital UNIX 4.0B. 10646 + Fix problems with ctors/dtors in SCO shared libraries. 10647 + Abort instead of generating incorrect code for PPro/PII 10648 floating point conditional moves. 10649 + Avoid multiply defined symbols on Linux/GNU systems using 10650 libc-5.4.xx. 10651 + Fix abort in alpha compiler. 10652 * Fortran-specific fixes 10653 + Fix the IDate intrinsic (VXT) (in libg2c) so the returned year 10654 is in the documented, non-Y2K-compliant range of 0-99, instead 10655 of being returned as 100 in the year 2000. 10656 + Fix the `Date_and_Time' intrinsic (in libg2c) to return the 10657 milliseconds value properly in Values(8). 10658 + Fix the `LStat' intrinsic (in libg2c) to return device-ID 10659 information properly in SArray(7). 10660 10661 Each release includes installation instructions in both HTML and 10662 plaintext forms (see the INSTALL directory in the toplevel directory of 10663 the distribution). However, we also keep the most up to date 10664 [6]installation instructions and [7]build/test status on our web page. 10665 We will update those pages as new information becomes available. 10666 10667 The EGCS project would like to thank the numerous people that have 10668 contributed new features, test results, bugfixes, etc. This [8]amazing 10669 group of volunteers is what makes EGCS successful. 10670 10671 And finally, we can't in good conscience fail to mention some 10672 [9]caveats to using EGCS 1.1. 10673 10674 Download EGCS from egcs.cygnus.com (USA California). 10675 10676 The EGCS 1.1 release is also available on many mirror sites. 10677 [10]Goto mirror list to find a closer site. 10678 10679 10680 For questions related to the use of GCC, please consult these web 10681 pages and the [11]GCC manuals. If that fails, the 10682 [12]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10683 web pages and the development of GCC are welcome on our developer 10684 list at [13]gcc (a] gcc.gnu.org. All of [14]our lists have public 10685 archives. 10686 10687 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and 10688 distribution of this entire article is permitted in any medium, 10689 provided this notice is preserved. 10690 10691 These pages are [16]maintained by the GCC team. Last modified 10692 2011-04-25[17]. 10693 10694 References 10695 10696 1. http://gcc.gnu.org/egcs-1.1/egcs-1.1-test.html 10697 2. http://gcc.gnu.org/news/gcse.html 10698 3. http://gcc.gnu.org/news/alias.html 10699 4. http://gcc.gnu.org/egcs-1.1/c++features.html 10700 5. http://gcc.gnu.org/egcs-1.1/features.html 10701 6. http://gcc.gnu.org/install/ 10702 7. http://gcc.gnu.org/egcs-1.1/buildstat.html 10703 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 10704 9. http://gcc.gnu.org/egcs-1.1/caveats.html 10705 10. http://gcc.gnu.org/mirrors.html 10706 11. http://gcc.gnu.org/onlinedocs/ 10707 12. mailto:gcc-help (a] gcc.gnu.org 10708 13. mailto:gcc (a] gcc.gnu.org 10709 14. http://gcc.gnu.org/lists.html 10710 15. http://www.fsf.org/ 10711 16. http://gcc.gnu.org/about.html 10712 17. http://validator.w3.org/check/referer 10713 ====================================================================== 10714 http://gcc.gnu.org/egcs-1.1/features.html 10715 10716 EGCS 1.1 new features 10717 10718 * Integrated GNU Fortran (g77) compiler and runtime library with 10719 improvements, based on g77 version 0.5.23. 10720 * Vast improvements in the C++ compiler; so many they have [1]page of 10721 their own! 10722 * Compiler implements [2]global common subexpression elimination and 10723 global copy/constant propagation. 10724 * More major improvements in the [3]alias analysis code. 10725 * More major improvements in the exception handling code to improve 10726 performance, lower static overhead and provide the infrastructure 10727 for future improvements. 10728 * The infamous /tmp symlink race security problems have been fixed. 10729 * The regmove optimization pass has been nearly completely rewritten 10730 to improve performance of generated code. 10731 * The compiler now recomputes register usage information before local 10732 register allocation. By providing more accurate information to the 10733 priority based allocator, we get better register allocation. 10734 * The register reloading phase of the compiler optimizes spill code 10735 much better than in previous releases. 10736 * Some bad interactions between the register allocator and 10737 instruction scheduler have been fixed, resulting in much better 10738 code for certain programs. Additionally, we have tuned the 10739 scheduler in various ways to improve performance of generated code 10740 for some architectures. 10741 * The compiler's branch shortening algorithms have been significantly 10742 improved to work better on targets which align jump targets. 10743 * The compiler now supports -Os to prefer optimizing for code space 10744 over optimizing for code speed. 10745 * The compiler will now totally eliminate library calls which compute 10746 constant values. This primarily helps targets with no integer 10747 div/mul support and targets without floating point support. 10748 * The compiler now supports an extensive "--help" option. 10749 * cpplib has been greatly improved and may be suitable for limited 10750 use. 10751 * Memory footprint for the compiler has been significantly reduced 10752 for some pathological cases. 10753 * The time to build EGCS has been improved for certain targets 10754 (particularly the alpha and mips platforms). 10755 * Many infrastructure improvements throughout the compiler, plus the 10756 usual mountain of bugfixes and minor improvements. 10757 * Target dependent improvements: 10758 + SPARC port now includes V8 plus and V9 support as well as 10759 performance tuning for Ultra class machines. The SPARC port 10760 now uses the Haifa scheduler. 10761 + Alpha port has been tuned for the EV6 processor and has an 10762 optimized expansion of memcpy/bzero. The Alpha port now uses 10763 the Haifa scheduler. 10764 + RS6000/PowerPC: support for the Power64 architecture and AIX 10765 4.3. The RS6000/PowerPC port now uses the Haifa scheduler. 10766 + x86: Alignment of static store data and jump targets is per 10767 Intel recommendations now. Various improvements throughout the 10768 x86 port to improve performance on Pentium processors 10769 (including improved epilogue sequences for Pentium chips and 10770 backend improvements which should help register allocation on 10771 all x86 variants. Conditional move support has been fixed and 10772 enabled for PPro processors. The x86 port also better supports 10773 64bit operations now. Unixware 7, a System V Release 5 target, 10774 is now supported and SCO OpenServer targets can support GAS. 10775 + MIPS has improved multiply/multiply-add support and now 10776 includes mips16 ISA support. 10777 + M68k has many micro-optimizations and Coldfire fixes. 10778 * Core compiler is based on the GCC development tree from June 9, 10779 1998, so we have all of the [4]features found in GCC 2.8. 10780 10781 10782 For questions related to the use of GCC, please consult these web 10783 pages and the [5]GCC manuals. If that fails, the 10784 [6]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10785 web pages and the development of GCC are welcome on our developer 10786 list at [7]gcc (a] gcc.gnu.org. All of [8]our lists have public archives. 10787 10788 Copyright (C) [9]Free Software Foundation, Inc. Verbatim copying and 10789 distribution of this entire article is permitted in any medium, 10790 provided this notice is preserved. 10791 10792 These pages are [10]maintained by the GCC team. Last modified 10793 2011-04-25[11]. 10794 10795 References 10796 10797 1. http://gcc.gnu.org/egcs-1.1/c++features.html 10798 2. http://gcc.gnu.org/news/gcse.html 10799 3. http://gcc.gnu.org/news/alias.html 10800 4. http://gcc.gnu.org/egcs-1.0/features-2.8.html 10801 5. http://gcc.gnu.org/onlinedocs/ 10802 6. mailto:gcc-help (a] gcc.gnu.org 10803 7. mailto:gcc (a] gcc.gnu.org 10804 8. http://gcc.gnu.org/lists.html 10805 9. http://www.fsf.org/ 10806 10. http://gcc.gnu.org/about.html 10807 11. http://validator.w3.org/check/referer 10808 ====================================================================== 10809 http://gcc.gnu.org/egcs-1.1/caveats.html 10810 10811 EGCS 1.1 Caveats 10812 10813 * EGCS has an integrated libstdc++, but does not have an integrated 10814 libg++. Furthermore old libg++ releases will not work with EGCS; HJ 10815 Lu has made a libg++-2.8.1.2 snapshot available which may work with 10816 EGCS. 10817 Note most C++ programs only need libstdc++. 10818 * Exception handling may not work with shared libraries, particularly 10819 on alphas, hppas, rs6000/powerpc and mips based platforms. 10820 Exception handling is known to work on x86-linux platforms with 10821 shared libraries. 10822 * Some versions of the Linux kernel have bugs which prevent them from 10823 being compiled or from running when compiled by EGCS. See the FAQ 10824 (as shipped with EGCS 1.1) for additional information. 10825 * In general, EGCS is more rigorous about rejecting invalid C++ code 10826 or deprecated C++ constructs than g++-2.7, g++-2.8 or EGCS 1.0. As 10827 a result it may be necessary to fix C++ code before it will compile 10828 with EGCS. 10829 * G++ is also converting toward the ISO C++ standard; as a result 10830 code which was previously valid (and thus accepted by other 10831 compilers and older versions of g++) may no longer be accepted. 10832 * EGCS 1.1 compiled C++ code is not binary compatible with EGCS 1.0.x 10833 or GCC 2.8.x due to changes necessary to support thread safe 10834 exception handling. 10835 10836 10837 For questions related to the use of GCC, please consult these web 10838 pages and the [1]GCC manuals. If that fails, the 10839 [2]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10840 web pages and the development of GCC are welcome on our developer 10841 list at [3]gcc (a] gcc.gnu.org. All of [4]our lists have public archives. 10842 10843 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and 10844 distribution of this entire article is permitted in any medium, 10845 provided this notice is preserved. 10846 10847 These pages are [6]maintained by the GCC team. Last modified 10848 2011-04-25[7]. 10849 10850 References 10851 10852 1. http://gcc.gnu.org/onlinedocs/ 10853 2. mailto:gcc-help (a] gcc.gnu.org 10854 3. mailto:gcc (a] gcc.gnu.org 10855 4. http://gcc.gnu.org/lists.html 10856 5. http://www.fsf.org/ 10857 6. http://gcc.gnu.org/about.html 10858 7. http://validator.w3.org/check/referer 10859 ====================================================================== 10860 http://gcc.gnu.org/egcs-1.0/index.html 10861 10862 EGCS 1.0 10863 10864 December 3, 1997: We are pleased to announce the release of EGCS 1.0. 10865 January 6, 1998: We are pleased to announce the release of EGCS 1.0.1. 10866 March 16, 1998: We are pleased to announce the release of EGCS 1.0.2. 10867 May 15, 1998 We are pleased to announce the release of EGCS 1.0.3. 10868 10869 EGCS is a collaborative effort involving several groups of hackers 10870 using an open development model to accelerate development and testing 10871 of GNU compilers and runtime libraries. 10872 10873 An important goal of EGCS is to allow wide scale testing of 10874 experimental features and optimizations; therefore, EGCS contains some 10875 features and optimizations which are still under development. However, 10876 EGCS has been carefully tested and should be comparable in quality to 10877 most GCC releases. 10878 10879 EGCS 1.0 is based on an August 2, 1997 snapshot of the GCC 2.8 10880 development sources; it contains nearly all of the new features found 10881 in GCC 2.8. 10882 10883 EGCS 1.0 also contains many improvements and features not found in GCC 10884 2.7 and even the GCC 2.8 series (which was released after the original 10885 EGCS 1.0 release). 10886 * Integrated C++ runtime libraries, including support for most major 10887 GNU/Linux systems! 10888 * The integrated libstdc++ library includes a verbatim copy of SGI's 10889 STL release. 10890 * Integrated GNU Fortran compiler. 10891 * New instruction scheduler. 10892 * New alias analysis code. 10893 10894 See the [1]new features page for a more complete list of new features. 10895 10896 EGCS 1.0.1 is a minor update to the EGCS 1.0 compiler to fix a few 10897 critical bugs and add support for Red Hat 5.0 Linux. Changes since the 10898 EGCS 1.0 release: 10899 * Add support for Red Hat 5.0 Linux and better support for Linux 10900 systems using glibc2. 10901 Many programs failed to link when compiled with EGCS 1.0 on Red Hat 10902 5.0 or on systems with newer versions of glibc2. EGCS 1.0.1 should 10903 fix these problems. 10904 * Compatibility with both EGCS 1.0 and GCC 2.8 libgcc exception 10905 handling interfaces. 10906 To avoid future compatibility problems, we strongly urge anyone who 10907 is planning on distributing shared libraries that contain C++ code 10908 to upgrade to EGCS 1.0.1 first. 10909 Soon after EGCS 1.0 was released, the GCC developers made some 10910 incompatible changes in libgcc's exception handling interfaces. 10911 These changes were needed to solve problems on some platforms. This 10912 means that GCC 2.8.0, when released, will not be seamlessly 10913 compatible with shared libraries built by EGCS 1.0. The reason is 10914 that the libgcc.a in GCC 2.8.0 will not contain a function needed 10915 by the old interface. 10916 The result of this is that there may be compatibility problems with 10917 shared libraries built by EGCS 1.0 when used with GCC 2.8.0. 10918 With EGCS 1.0.1, generated code uses the new (GCC 2.8.0) interface, 10919 and libgcc.a has the support routines for both the old and the new 10920 interfaces (so EGCS 1.0.1 and EGCS 1.0 code can be freely mixed, 10921 and EGCS 1.0.1 and GCC 2.8.0 code can be freely mixed). 10922 The maintainers of GCC 2.x have decided against including seamless 10923 support for the old interface in 2.8.0, since it was never 10924 "official", so to avoid future compatibility problems we recommend 10925 against distributing any shared libraries built by EGCS 1.0 that 10926 contain C++ code (upgrade to 1.0.1 and use that). 10927 * Various bugfixes in the x86, hppa, mips, and rs6000/ppc backends. 10928 The x86 changes fix code generation errors exposed when building 10929 glibc2 and the Linux dynamic linker (ld.so). 10930 The hppa change fixes a compiler abort when configured for use with 10931 RTEMS. 10932 The MIPS changes fix problems with the definition of LONG_MAX on 10933 newer systems, allow for command line selection of the target ABI, 10934 and fix one code generation problem. 10935 The rs6000/ppc change fixes some problems with passing structures 10936 to varargs/stdarg functions. 10937 * A few machine independent bugfixes, mostly to fix code generation 10938 errors when building Linux kernels or glibc. 10939 * Fix a few critical exception handling and template bugs in the C++ 10940 compiler. 10941 * Fix Fortran namelist bug on alphas. 10942 * Fix build problems on x86-solaris systems. 10943 10944 EGCS 1.0.2 is a minor update to the EGCS 1.0.1 compiler to fix several 10945 serious problems in EGCS 1.0.1. 10946 * General improvements and fixes 10947 + Memory consumption significantly reduced, especially for 10948 templates and inline functions. 10949 + Fix various problems with glibc2.1. 10950 + Fix loop optimization bug exposed by rs6000/ppc port. 10951 + Fix to avoid potential code generation problems in jump.c. 10952 + Fix some undefined symbol problems in dwarf1 debug support. 10953 * g++/libstdc++ improvements and fixes 10954 + libstdc++ in the EGCS release has been updated and should be 10955 link compatible with libstdc++-2.8. 10956 + Various fixes in libio/libstdc++ to work better on Linux 10957 systems. 10958 + Fix problems with duplicate symbols on systems that do not 10959 support weak symbols. 10960 + Memory corruption bug and undefined symbols in bastring have 10961 been fixed. 10962 + Various exception handling fixes. 10963 + Fix compiler abort for very long thunk names. 10964 * g77 improvements and fixes 10965 + Fix compiler crash for omitted bound in Fortran CASE 10966 statement. 10967 + Add missing entries to g77 lang-options. 10968 + Fix problem with -fpedantic in the g77 compiler. 10969 + Fix "backspace" problem with g77 on alphas. 10970 + Fix x86 backend problem with Fortran literals and -fpic. 10971 + Fix some of the problems with negative subscripts for g77 on 10972 alphas. 10973 + Fixes for Fortran builds on cygwin32/mingw32. 10974 * platform specific improvements and fixes 10975 + Fix long double problems on x86 (exposed by glibc). 10976 + x86 ports define i386 again to keep imake happy. 10977 + Fix exception handling support on NetBSD ports. 10978 + Several changes to collect2 to fix many problems with AIX. 10979 + Define __ELF__ for rs6000/linux. 10980 + Fix -mcall-linux problem on rs6000/linux. 10981 + Fix stdarg/vararg problem for rs6000/linux. 10982 + Allow autoconf to select a proper install problem on AIX 3.1. 10983 + m68k port support includes -mcpu32 option as well as cpu32 10984 multilibs. 10985 + Fix stdarg bug for irix6. 10986 + Allow EGCS to build on irix5 without the gnu assembler. 10987 + Fix problem with static linking on sco5. 10988 + Fix bootstrap on sco5 with native compiler. 10989 + Fix for abort building newlib on H8 target. 10990 + Fix fixincludes handling of math.h on SunOS. 10991 + Minor fix for Motorola 3300 m68k systems. 10992 10993 EGCS 1.0.3 is a minor update to the EGCS 1.0.2 compiler to fix a few 10994 problems reported by Red Hat for builds of Red Hat 5.1. 10995 * Generic bugfixes: 10996 + Fix a typo in the libio library which resulted in incorrect 10997 behavior of istream::get. 10998 + Fix the Fortran negative array index problem. 10999 + Fix a major problem with the ObjC runtime thread support 11000 exposed by glibc2. 11001 + Reduce memory consumption of the Haifa scheduler. 11002 * Target specific bugfixes: 11003 + Fix one x86 floating point code generation bug exposed by 11004 glibc2 builds. 11005 + Fix one x86 internal compiler error exposed by glibc2 builds. 11006 + Fix profiling bugs on the Alpha. 11007 + Fix ImageMagick & emacs 20.2 build problems on the Alpha. 11008 + Fix rs6000/ppc bug when converting values from integer types 11009 to floating point types. 11010 11011 The EGCS 1.0 releases include installation instructions in both HTML 11012 and plaintext forms (see the INSTALL directory in the toplevel 11013 directory of the distribution). However, we also keep the most up to 11014 date [2]installation instructions and [3]build/test status on our web 11015 page. We will update those pages as new information becomes available. 11016 11017 And, we can't in good conscience fail to mention some [4]caveats to 11018 using EGCS. 11019 11020 Update: Big thanks to Stanford for providing a high speed link for 11021 downloading EGCS (go.cygnus.com)! 11022 11023 Download EGCS from ftp.cygnus.com (USA California) or go.cygnus.com 11024 (USA California -- High speed link provided by Stanford). 11025 11026 The EGCS 1.0 release is also available many mirror sites. 11027 [5]Goto mirror list to find a closer site 11028 11029 We'd like to thank the numerous people that have contributed new 11030 features, test results, bugfixes, etc. Unfortunately, they're far too 11031 numerous to mention by name. 11032 11033 11034 For questions related to the use of GCC, please consult these web 11035 pages and the [6]GCC manuals. If that fails, the 11036 [7]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 11037 web pages and the development of GCC are welcome on our developer 11038 list at [8]gcc (a] gcc.gnu.org. All of [9]our lists have public archives. 11039 11040 Copyright (C) [10]Free Software Foundation, Inc. Verbatim copying and 11041 distribution of this entire article is permitted in any medium, 11042 provided this notice is preserved. 11043 11044 These pages are [11]maintained by the GCC team. Last modified 11045 2011-04-25[12]. 11046 11047 References 11048 11049 1. http://gcc.gnu.org/egcs-1.0/features.html 11050 2. http://gcc.gnu.org/install/ 11051 3. http://gcc.gnu.org/egcs-1.0/buildstat.html 11052 4. http://gcc.gnu.org/egcs-1.0/caveats.html 11053 5. http://gcc.gnu.org/mirrors.html 11054 6. http://gcc.gnu.org/onlinedocs/ 11055 7. mailto:gcc-help (a] gcc.gnu.org 11056 8. mailto:gcc (a] gcc.gnu.org 11057 9. http://gcc.gnu.org/lists.html 11058 10. http://www.fsf.org/ 11059 11. http://gcc.gnu.org/about.html 11060 12. http://validator.w3.org/check/referer 11061 ====================================================================== 11062 http://gcc.gnu.org/egcs-1.0/features.html 11063 11064 EGCS 1.0 features 11065 11066 * Core compiler is based on the gcc2 development tree from Aug 2, 11067 1997, so we have most of the [1]features found in GCC 2.8. 11068 * Integrated GNU Fortran compiler based on g77-0.5.22-19970929. 11069 * Vast improvements in the C++ compiler; so many they have [2]page of 11070 their own! 11071 * Integrated C++ runtime libraries, including support for most major 11072 linux systems! 11073 * New instruction scheduler from IBM Haifa which includes support for 11074 function wide instruction scheduling as well as superscalar 11075 scheduling. 11076 * Significantly improved alias analysis code. 11077 * Improved register allocation for two address machines. 11078 * Significant code generation improvements for Fortran code on 11079 Alphas. 11080 * Various optimizations from the g77 project as well as improved loop 11081 optimizations. 11082 * Dwarf2 debug format support for some targets. 11083 * egcs libstdc++ includes the SGI STL implementation without changes. 11084 * As a result of these and other changes, egcs libstc++ is not binary 11085 compatible with previous releases of libstdc++. 11086 * Various new ports -- UltraSPARC, Irix6.2 & Irix6.3 support, The SCO 11087 Openserver 5 family (5.0.{0,2,4} and Internet FastStart 1.0 and 11088 1.1), Support for RTEMS on several embedded targets, Support for 11089 arm-linux, Mitsubishi M32R, Hitachi H8/S, Matsushita MN102 and 11090 MN103, NEC V850, Sparclet, Solaris & Linux on PowerPCs, etc. 11091 * Integrated testsuites for gcc, g++, g77, libstdc++ and libio. 11092 * RS6000/PowerPC ports generate code which can run on all 11093 RS6000/PowerPC variants by default. 11094 * -mcpu= and -march= switches for the x86 port to allow better 11095 control over how the x86 port generates code. 11096 * Includes the template repository patch (aka repo patch); note the 11097 new template code makes repo obsolete for ELF systems using gnu-ld 11098 such as Linux. 11099 * Plus the usual assortment of bugfixes and improvements. 11100 11101 11102 For questions related to the use of GCC, please consult these web 11103 pages and the [3]GCC manuals. If that fails, the 11104 [4]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 11105 web pages and the development of GCC are welcome on our developer 11106 list at [5]gcc (a] gcc.gnu.org. All of [6]our lists have public archives. 11107 11108 Copyright (C) [7]Free Software Foundation, Inc. Verbatim copying and 11109 distribution of this entire article is permitted in any medium, 11110 provided this notice is preserved. 11111 11112 These pages are [8]maintained by the GCC team. Last modified 11113 2011-04-25[9]. 11114 11115 References 11116 11117 1. http://gcc.gnu.org/egcs-1.0/features-2.8.html 11118 2. http://gcc.gnu.org/egcs-1.0/c++features.html 11119 3. http://gcc.gnu.org/onlinedocs/ 11120 4. mailto:gcc-help (a] gcc.gnu.org 11121 5. mailto:gcc (a] gcc.gnu.org 11122 6. http://gcc.gnu.org/lists.html 11123 7. http://www.fsf.org/ 11124 8. http://gcc.gnu.org/about.html 11125 9. http://validator.w3.org/check/referer 11126 ====================================================================== 11127 http://gcc.gnu.org/egcs-1.0/caveats.html 11128 11129 EGCS 1.0 Caveats 11130 11131 * EGCS has an integrated libstdc++, but does not have an integrated 11132 libg++. Furthermore old libg++ releases will not work with egc; HJ 11133 Lu has made a libg++-2.8.1.2 available which may work with EGCS. 11134 Note most C++ programs only need libstdc++. 11135 * Note that using -pedantic or -Wreturn-type can cause an explosion 11136 in the amount of memory needed for template-heavy C++ code, such as 11137 code that uses STL. Also note that -Wall includes -Wreturn-type, so 11138 if you use -Wall you will need to specify -Wno-return-type to turn 11139 it off. 11140 * Exception handling may not work with shared libraries, particularly 11141 on alphas, hppas, and mips based platforms. Exception handling is 11142 known to work on x86-linux platforms with shared libraries. 11143 * Some versions of the Linux kernel have bugs which prevent them from 11144 being compiled or from running when compiled by EGCS. See the FAQ 11145 (as shipped with EGCS 1.0) for additional information. 11146 * In general, EGCS is more rigorous about rejecting invalid C++ code 11147 or deprecated C++ constructs than G++ 2.7. As a result it may be 11148 necessary to fix C++ code before it will compile with EGCS. 11149 * G++ is also aggressively tracking the C++ standard; as a result 11150 code which was previously valid (and thus accepted by other 11151 compilers and older versions of G++) may no longer be accepted. 11152 * EGCS 1.0 may not work with Red Hat Linux 5.0 on all targets. EGCS 11153 1.0.x and later releases should work with Red Hat Linux 5.0. 11154 11155 11156 For questions related to the use of GCC, please consult these web 11157 pages and the [1]GCC manuals. If that fails, the 11158 [2]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 11159 web pages and the development of GCC are welcome on our developer 11160 list at [3]gcc (a] gcc.gnu.org. All of [4]our lists have public archives. 11161 11162 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and 11163 distribution of this entire article is permitted in any medium, 11164 provided this notice is preserved. 11165 11166 These pages are [6]maintained by the GCC team. Last modified 11167 2011-04-25[7]. 11168 11169 References 11170 11171 1. http://gcc.gnu.org/onlinedocs/ 11172 2. mailto:gcc-help (a] gcc.gnu.org 11173 3. mailto:gcc (a] gcc.gnu.org 11174 4. http://gcc.gnu.org/lists.html 11175 5. http://www.fsf.org/ 11176 6. http://gcc.gnu.org/about.html 11177 7. http://validator.w3.org/check/referer 11178 ====================================================================== 11179
Indexes created Tue Oct 27 00:25:24 UTC 2026