NEWS revision 1.7
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-5/index.html 9 GCC 5 Release Series 10 11 October 10, 2017 12 13 The [1]GNU project and the GCC developers are pleased to announce the 14 release of GCC 5.5. 15 16 This release is a bug-fix release, containing fixes for regressions in 17 GCC 5.4 relative to previous releases of GCC. 18 19 Release History 20 21 GCC 5.5 22 October 10, 2017 ([2]changes, [3]documentation) 23 24 GCC 5.4 25 June 3, 2016 ([4]changes, [5]documentation) 26 27 GCC 5.3 28 December 4, 2015 ([6]changes, [7]documentation) 29 30 GCC 5.2 31 July 16, 2015 ([8]changes, [9]documentation) 32 33 GCC 5.1 34 April 22, 2015 ([10]changes, [11]documentation) 35 36 References and Acknowledgements 37 38 GCC used to stand for the GNU C Compiler, but since the compiler 39 supports several other languages aside from C, it now stands for the 40 GNU Compiler Collection. 41 42 A list of [12]successful builds is updated as new information becomes 43 available. 44 45 The GCC developers would like to thank the numerous people that have 46 contributed new features, improvements, bug fixes, and other changes as 47 well as test results to GCC. This [13]amazing group of volunteers is 48 what makes GCC successful. 49 50 For additional information about GCC please refer to the [14]GCC 51 project web site or contact the [15]GCC development mailing list. 52 53 To obtain GCC please use [16]our mirror sites or [17]our SVN server. 54 55 56 For questions related to the use of GCC, please consult these web 57 pages and the [18]GCC manuals. If that fails, the 58 [19]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 59 web pages and the development of GCC are welcome on our developer 60 list at [20]gcc (a] gcc.gnu.org. All of [21]our lists have public 61 archives. 62 63 Copyright (C) [22]Free Software Foundation, Inc. Verbatim copying and 64 distribution of this entire article is permitted in any medium, 65 provided this notice is preserved. 66 67 These pages are [23]maintained by the GCC team. Last modified 68 2017-10-10[24]. 69 70 References 71 72 1. http://www.gnu.org/ 73 2. http://gcc.gnu.org/gcc-5/changes.html 74 3. http://gcc.gnu.org/onlinedocs/5.5.0/ 75 4. http://gcc.gnu.org/gcc-5/changes.html 76 5. http://gcc.gnu.org/onlinedocs/5.4.0/ 77 6. http://gcc.gnu.org/gcc-5/changes.html 78 7. http://gcc.gnu.org/onlinedocs/5.3.0/ 79 8. http://gcc.gnu.org/gcc-5/changes.html 80 9. http://gcc.gnu.org/onlinedocs/5.2.0/ 81 10. http://gcc.gnu.org/gcc-5/changes.html 82 11. http://gcc.gnu.org/onlinedocs/5.1.0/ 83 12. http://gcc.gnu.org/gcc-5/buildstat.html 84 13. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html 85 14. http://gcc.gnu.org/index.html 86 15. mailto:gcc (a] gcc.gnu.org 87 16. http://gcc.gnu.org/mirrors.html 88 17. http://gcc.gnu.org/svn.html 89 18. https://gcc.gnu.org/onlinedocs/ 90 19. mailto:gcc-help (a] gcc.gnu.org 91 20. mailto:gcc (a] gcc.gnu.org 92 21. https://gcc.gnu.org/lists.html 93 22. http://www.fsf.org/ 94 23. https://gcc.gnu.org/about.html 95 24. http://validator.w3.org/check/referer 96 ====================================================================== 97 http://gcc.gnu.org/gcc-5/changes.html 98 GCC 5 Release Series 99 Changes, New Features, and Fixes 100 101 Caveats 102 103 * The default mode for C is now -std=gnu11 instead of -std=gnu89. 104 * The C++ runtime library (libstdc++) uses a new ABI by default (see 105 [1]below). 106 * The Graphite framework for loop optimizations no longer requires 107 the CLooG library, only ISL version 0.14 (recommended) or 0.12.2. 108 The installation manual contains more information about 109 requirements to build GCC. 110 * The non-standard C++0x type traits has_trivial_default_constructor, 111 has_trivial_copy_constructor and has_trivial_copy_assign have been 112 deprecated and will be removed in a future version. The standard 113 C++11 traits is_trivially_default_constructible, 114 is_trivially_copy_constructible and is_trivially_copy_assignable 115 should be used instead. 116 * On AVR, support has been added for the devices 117 ATtiny4/5/9/10/20/40. This requires Binutils 2.25 or newer. 118 * The AVR port uses a new scheme to describe supported devices: For 119 each supported device the compiler provides a device-specific 120 [2]spec file. If the compiler is used together with AVR-LibC, this 121 requires at least GCC 5.2 and a version of AVR-LibC which 122 implements [3]feature #44574. 123 124 General Optimizer Improvements 125 126 * Inter-procedural optimization improvements: 127 + An Identical Code Folding (ICF) pass (controlled via 128 -fipa-icf) has been added. Compared to the identical code 129 folding performed by the Gold linker this pass does not 130 require function sections. It also performs merging before 131 inlining, so inter-procedural optimizations are aware of the 132 code re-use. On the other hand not all unifications performed 133 by a linker are doable by GCC which must honor aliasing 134 information. During link-time optimization of Firefox, this 135 pass unifies about 31000 functions, that is 14% overall. 136 + The devirtualization pass was significantly improved by adding 137 better support for speculative devirtualization and dynamic 138 type detection. About 50% of virtual calls in Firefox are now 139 speculatively devirtualized during link-time optimization. 140 + A new comdat localization pass allows the linker to eliminate 141 more dead code in presence of C++ inline functions. 142 + Virtual tables are now optimized. Local aliases are used to 143 reduce dynamic linking time of C++ virtual tables on ELF 144 targets and data alignment has been reduced to limit data 145 segment bloat. 146 + A new -fno-semantic-interposition option can be used to 147 improve code quality of shared libraries where interposition 148 of exported symbols is not allowed. 149 + Write-only variables are now detected and optimized out. 150 + With profile feedback the function inliner can now bypass 151 --param inline-insns-auto and --param inline-insns-single 152 limits for hot calls. 153 + The IPA reference pass was significantly sped up making it 154 feasible to enable -fipa-reference with -fprofile-generate. 155 This also solves a bottleneck seen when building Chromium with 156 link-time optimization. 157 + The symbol table and call-graph API was reworked to C++ and 158 simplified. 159 + The interprocedural propagation of constants now also 160 propagates alignments of pointer parameters. This for example 161 means that the vectorizer often does not need to generate loop 162 prologues and epilogues to make up for potential 163 misalignments. 164 * Link-time optimization improvements: 165 + One Definition Rule based merging of C++ types has been 166 implemented. Type merging enables better devirtualization and 167 alias analysis. Streaming extra information needed to merge 168 types adds about 2-6% of memory size and object size increase. 169 This can be controlled by -flto-odr-type-merging. 170 + Command-line optimization and target options are now streamed 171 on a per-function basis and honored by the link-time 172 optimizer. This change makes link-time optimization a more 173 transparent replacement of per-file optimizations. It is now 174 possible to build projects that require different optimization 175 settings for different translation units (such as -ffast-math, 176 -mavx, or -finline). Contrary to earlier GCC releases, the 177 optimization and target options passed on the link command 178 line are ignored. 179 Note that this applies only to those command-line options that 180 can be passed to optimize and target attributes. Command-line 181 options affecting global code generation (such as -fpic), 182 warnings (such as -Wodr), optimizations affecting the way 183 static variables are optimized (such as -fcommon), debug 184 output (such as -g), and --param parameters can be applied 185 only to the whole link-time optimization unit. In these cases, 186 it is recommended to consistently use the same options at both 187 compile time and link time. 188 + GCC bootstrap now uses slim LTO object files. 189 + Memory usage and link times were improved. Tree merging was 190 sped up, memory usage of GIMPLE declarations and types was 191 reduced, and, support for on-demand streaming of variable 192 constructors was added. 193 * Feedback directed optimization improvements: 194 + A new auto-FDO mode uses profiles collected by low overhead 195 profiling tools (perf) instead of more expensive program 196 instrumentation (via -fprofile-generate). SPEC2006 benchmarks 197 on x86-64 improve by 4.7% with auto-FDO and by 7.3% with 198 traditional feedback directed optimization. 199 + Profile precision was improved in presence of C++ inline and 200 extern inline functions. 201 + The new gcov-tool utility allows manipulating profiles. 202 + Profiles are now more tolerant to source file changes (this 203 can be controlled by --param profile-func-internal-id). 204 * Register allocation improvements: 205 + A new local register allocator (LRA) sub-pass, controlled by 206 -flra-remat, implements control-flow sensitive global register 207 rematerialization. Instead of spilling and restoring a 208 register value, it is recalculated if it is profitable. The 209 sub-pass improved SPEC2000 generated code by 1% and 0.5% 210 correspondingly on ARM and x86-64. 211 + Reuse of the PIC hard register, instead of using a fixed 212 register, was implemented on x86/x86-64 targets. This improves 213 generated PIC code performance as more hard registers can be 214 used. Shared libraries can significantly benefit from this 215 optimization. Currently it is switched on only for x86/x86-64 216 targets. As RA infrastructure is already implemented for PIC 217 register reuse, other targets might follow this in the future. 218 + A simple form of inter-procedural RA was implemented. When it 219 is known that a called function does not use caller-saved 220 registers, save/restore code is not generated around the call 221 for such registers. This optimization can be controlled by 222 -fipa-ra 223 + LRA is now much more effective at generating spills of general 224 registers into vector registers instead of memory on 225 architectures (e.g., modern Intel processors) where this is 226 profitable. 227 * UndefinedBehaviorSanitizer gained a few new sanitization options: 228 + -fsanitize=float-divide-by-zero: detect floating-point 229 division by zero; 230 + -fsanitize=float-cast-overflow: check that the result of 231 floating-point type to integer conversions do not overflow; 232 + -fsanitize=bounds: enable instrumentation of array bounds and 233 detect out-of-bounds accesses; 234 + -fsanitize=alignment: enable alignment checking, detect 235 various misaligned objects; 236 + -fsanitize=object-size: enable object size checking, detect 237 various out-of-bounds accesses. 238 + -fsanitize=vptr: enable checking of C++ member function calls, 239 member accesses and some conversions between pointers to base 240 and derived classes, detect if the referenced object does not 241 have the correct dynamic type. 242 * Pointer Bounds Checker, a bounds violation detector, has been added 243 and can be enabled via -fcheck-pointer-bounds. Memory accesses are 244 instrumented with run-time checks of used pointers against their 245 bounds to detect pointer bounds violations (overflows). The Pointer 246 Bounds Checker is available on x86/x86-64 GNU/Linux targets with a 247 new ISA extension Intel MPX support. See the Pointer Bounds Checker 248 [4]Wiki page for more details. 249 250 New Languages and Language specific improvements 251 252 * [5]OpenMP 4.0 specification offloading features are now supported 253 by the C, C++, and Fortran compilers. Generic changes: 254 + Infrastructure (suitable for any vendor). 255 + Testsuite which covers offloading from the [6]OpenMP 4.0 256 Examples document. 257 Specific for upcoming Intel Xeon Phi products: 258 + Run-time library. 259 + Card emulator. 260 * GCC 5 includes a preliminary implementation of the OpenACC 2.0a 261 specification. OpenACC is intended for programming accelerator 262 devices such as GPUs. See [7]the OpenACC wiki page for more 263 information. 264 265 C family 266 267 * The default setting of the -fdiagnostics-color= command-line option 268 is now [8]configurable when building GCC using configuration option 269 --with-diagnostics-color=. The possible values are: never, always, 270 auto and auto-if-env. The new default auto uses color only when the 271 standard error is a terminal. The default in GCC 4.9 was 272 auto-if-env, which is equivalent to auto if there is a non-empty 273 GCC_COLORS environment variable, and never otherwise. As in GCC 274 4.9, an empty GCC_COLORS variable in the environment will always 275 disable colors, no matter what the default is or what command-line 276 options are used. 277 * A new command-line option -Wswitch-bool has been added for the C 278 and C++ compilers, which warns whenever a switch statement has an 279 index of boolean type. 280 * A new command-line option -Wlogical-not-parentheses has been added 281 for the C and C++ compilers, which warns about "logical not" used 282 on the left hand side operand of a comparison. 283 * A new command-line option -Wsizeof-array-argument has been added 284 for the C and C++ compilers, which warns when the sizeof operator 285 is applied to a parameter that has been declared as an array in a 286 function definition. 287 * A new command-line option -Wbool-compare has been added for the C 288 and C++ compilers, which warns about boolean expressions compared 289 with an integer value different from true/false. 290 * Full support for [9]Cilk Plus has been added to the GCC compiler. 291 Cilk Plus is an extension to the C and C++ languages to support 292 data and task parallelism. 293 * A new attribute no_reorder prevents reordering of selected symbols 294 against other such symbols or inline assembler. This enables to 295 link-time optimize the Linux kernel without having to resort to 296 -fno-toplevel-reorder that disables several optimizations. 297 * New preprocessor constructs, __has_include and __has_include_next, 298 to test the availability of headers have been added. 299 This demonstrates a way to include the header <optional> only if it 300 is available: 301 302 #ifdef __has_include 303 # if __has_include(<optional>) 304 # include <optional> 305 # define have_optional 1 306 # elif __has_include(<experimental/optional>) 307 # include <experimental/optional> 308 # define have_optional 1 309 # define experimental_optional 310 # else 311 # define have_optional 0 312 # endif 313 #endif 314 315 The header search paths for __has_include and __has_include_next 316 are equivalent to those of the standard directive #include and the 317 extension #include_next respectively. 318 * A new built-in function-like macro to determine the existence of an 319 attribute, __has_attribute, has been added. The equivalent built-in 320 macro __has_cpp_attribute was added to C++ to support 321 [10]Feature-testing recommendations for C++. The macro 322 __has_attribute is added to all C-like languages as an extension: 323 324 int 325 #ifdef __has_attribute 326 # if __has_attribute(__noinline__) 327 __attribute__((__noinline__)) 328 # endif 329 #endif 330 foo(int x); 331 332 If an attribute exists, a nonzero constant integer is returned. For 333 standardized C++ attributes a date is returned, otherwise the 334 constant returned is 1. Both __has_attribute and 335 __has_cpp_attribute will add underscores to an attribute name if 336 necessary to resolve the name. For C++11 and onwards the attribute 337 may be scoped. 338 * A new set of built-in functions for arithmetics with overflow 339 checking has been added: __builtin_add_overflow, 340 __builtin_sub_overflow and __builtin_mul_overflow and for 341 compatibility with clang also other variants. These builtins have 342 two integral arguments (which don't need to have the same type), 343 the arguments are extended to infinite precision signed type, +, - 344 or * is performed on those, and the result is stored in an integer 345 variable pointed to by the last argument. If the stored value is 346 equal to the infinite precision result, the built-in functions 347 return false, otherwise true. The type of the integer variable that 348 will hold the result can be different from the types of the first 349 two arguments. The following snippet demonstrates how this can be 350 used in computing the size for the calloc function: 351 352 void * 353 calloc (size_t x, size_t y) 354 { 355 size_t sz; 356 if (__builtin_mul_overflow (x, y, &sz)) 357 return NULL; 358 void *ret = malloc (sz); 359 if (ret) memset (res, 0, sz); 360 return ret; 361 } 362 363 On e.g. i?86 or x86-64 the above will result in a mul instruction 364 followed by a jump on overflow. 365 * The option -fextended-identifiers is now enabled by default for 366 C++, and for C99 and later C versions. Various bugs in the 367 implementation of extended identifiers have been fixed. 368 369 C 370 371 * The default mode has been changed to -std=gnu11. 372 * A new command-line option -Wc90-c99-compat has been added to warn 373 about features not present in ISO C90, but present in ISO C99. 374 * A new command-line option -Wc99-c11-compat has been added to warn 375 about features not present in ISO C99, but present in ISO C11. 376 * It is possible to disable warnings about conversions between 377 pointers that have incompatible types via a new warning option 378 -Wno-incompatible-pointer-types; warnings about implicit 379 incompatible integer to pointer and pointer to integer conversions 380 via a new warning option -Wno-int-conversion; and warnings about 381 qualifiers on pointers being discarded via a new warning option 382 -Wno-discarded-qualifiers. 383 * To allow proper use of const qualifiers with multidimensional 384 arrays, GCC will not warn about incompatible pointer types anymore 385 for conversions between pointers to arrays with and without const 386 qualifier (except when using -pedantic). Instead, a new warning is 387 emitted only if the const qualifier is lost. This can be controlled 388 with a new warning option -Wno-discarded-array-qualifiers. 389 * The C front end now generates more precise caret diagnostics. 390 * The -pg command-line option now only affects the current file in an 391 LTO build. 392 393 C++ 394 395 * G++ now supports [11]C++14 variable templates. 396 * -Wnon-virtual-dtor doesn't warn anymore for final classes. 397 * Excessive template instantiation depth is now a fatal error. This 398 prevents excessive diagnostics that usually do not help to identify 399 the problem. 400 * G++ and libstdc++ now implement the feature-testing macros from 401 [12]Feature-testing recommendations for C++. 402 * G++ now allows typename in a template template parameter. 403 404 template<template<typename> typename X> struct D; // OK 405 406 * G++ now supports [13]C++14 aggregates with non-static data member 407 initializers. 408 409 struct A { int i, j = i; }; 410 A a = { 42 }; // a.j is also 42 411 412 * G++ now supports [14]C++14 extended constexpr. 413 414 constexpr int f (int i) 415 { 416 int j = 0; 417 for (; i > 0; --i) 418 ++j; 419 return j; 420 } 421 422 constexpr int i = f(42); // i is 42 423 424 * G++ now supports the [15]C++14 sized deallocation functions. 425 426 void operator delete (void *, std::size_t) noexcept; 427 void operator delete[] (void *, std::size_t) noexcept; 428 429 * A new One Definition Rule violation warning (controlled by -Wodr) 430 detects mismatches in type definitions and virtual table contents 431 during link-time optimization. 432 * New warnings -Wsuggest-final-types and -Wsuggest-final-methods help 433 developers to annotate programs with final specifiers (or anonymous 434 namespaces) to improve code generation. These warnings can be used 435 at compile time, but they are more useful in combination with 436 link-time optimization. 437 * G++ no longer supports [16]N3639 variable length arrays, as they 438 were removed from the C++14 working paper prior to ratification. 439 GNU VLAs are still supported, so VLA support is now the same in 440 C++14 mode as in C++98 and C++11 modes. 441 * G++ now allows passing a non-trivially-copyable class via C 442 varargs, which is conditionally-supported with 443 implementation-defined semantics in the standard. This uses the 444 same calling convention as a normal value parameter. 445 * G++ now defaults to -fabi-version=9 and -fabi-compat-version=2. So 446 various mangling bugs are fixed, but G++ will still emit aliases 447 with the old, wrong mangling where feasible. -Wabi=2 will warn 448 about differences between ABI version 2 and the current setting. 449 * G++ 5.2 fixes the alignment of std::nullptr_t. Most code is likely 450 to be unaffected, but -Wabi=8 will warn about a non-static data 451 member with type std::nullptr_t which changes position due to this 452 change. 453 454 Runtime Library (libstdc++) 455 456 * A [17]Dual ABI is provided by the library. A new ABI is enabled by 457 default. The old ABI is still supported and can be used by defining 458 the macro _GLIBCXX_USE_CXX11_ABI to 0 before including any C++ 459 standard library headers. 460 * A new implementation of std::string is enabled by default, using 461 the small string optimization instead of copy-on-write reference 462 counting. 463 * A new implementation of std::list is enabled by default, with an 464 O(1) size() function; 465 * [18]Full support for C++11, including the following new features: 466 + std::deque and std::vector<bool> meet the allocator-aware 467 container requirements; 468 + movable and swappable iostream classes; 469 + support for std::align and std::aligned_union; 470 + type traits std::is_trivially_copyable, 471 std::is_trivially_constructible, std::is_trivially_assignable 472 etc.; 473 + I/O manipulators std::put_time, std::get_time, std::hexfloat 474 and std::defaultfloat; 475 + generic locale-aware std::isblank; 476 + locale facets for Unicode conversion; 477 + atomic operations for std::shared_ptr; 478 + std::notify_all_at_thread_exit() and functions for making 479 futures ready at thread exit. 480 * Support for the C++11 hexfloat manipulator changes how the num_put 481 facet formats floating point types when 482 ios_base::fixed|ios_base::scientific is set in a stream's fmtflags. 483 This change affects all language modes, even though the C++98 484 standard gave no special meaning to that combination of flags. To 485 prevent the use of hexadecimal notation for floating point types 486 use str.unsetf(std::ios_base::floatfield) to clear the relevant 487 bits in str.flags(). 488 * [19]Full experimental support for C++14, including the following 489 new features: 490 + std::is_final type trait; 491 + heterogeneous comparison lookup in associative containers. 492 + global functions cbegin, cend, rbegin, rend, crbegin, and 493 crend for range access to containers, arrays and initializer 494 lists. 495 * [20]Improved experimental support for the Library Fundamentals TS, 496 including: 497 + class std::experimental::any; 498 + function template std::experimental::apply; 499 + function template std::experimental::sample; 500 + function template std::experimental::search and related 501 searcher types; 502 + variable templates for type traits; 503 + function template std::experimental::not_fn. 504 * New random number distributions logistic_distribution and 505 uniform_on_sphere_distribution as extensions. 506 * [21]GDB Xmethods for containers and std::unique_ptr. 507 508 Fortran 509 510 * Compatibility notice: 511 + The version of the module files (.mod) has been incremented. 512 + For free-form source files [22]-Werror=line-truncation is now 513 enabled by default. Note that comments exceeding the line 514 length are not diagnosed. (For fixed-form source code, the 515 same warning is available but turned off by default, such that 516 excess characters are ignored. -ffree-line-length-n and 517 -ffixed-line-length-n can be used to modify the default line 518 lengths of 132 and 72 columns, respectively.) 519 + The -Wtabs option is now more sensible: with -Wtabs the 520 compiler warns if it encounters tabs and with -Wno-tabs this 521 warning is turned off. Before, -Wno-tabs warned and -Wtabs 522 disabled the warning. As before, this warning is also enabled 523 by -Wall, -pedantic and the f95, f2003, f2008 and f2008ts 524 options of -std=. 525 * Incomplete support for colorizing diagnostics emitted by gfortran 526 has been added. The option [23]-fdiagnostics-color controls when 527 color is used in diagnostics. The default value of this option can 528 be [24]configured when building GCC. The GCC_COLORS environment 529 variable can be used to customize the colors or disable coloring 530 completely. Sample diagnostics output: 531 $ gfortran -fdiagnostics-color=always -Wuse-without-only test.f90 532 test.f90:6:1: 533 534 0 continue 535 1 536 Error: Zero is not a valid statement label at (1) 537 test.f90:9:6: 538 539 USE foo 540 1 541 Warning: USE statement at (1) has no ONLY qualifier [-Wuse-without-only] 542 543 * The -Wuse-without-only option has been added to warn when a USE 544 statement has no ONLY qualifier and thus implicitly imports all 545 public entities of the used module. 546 * Formatted READ and WRITE statements now work correctly in 547 locale-aware programs. For more information and potential caveats, 548 see [25]Section 5.3 Thread-safety of the runtime library in the 549 manual. 550 * [26]Fortran 2003: 551 + The intrinsic IEEE modules (IEEE_FEATURES, IEEE_EXCEPTIONS and 552 IEEE_ARITHMETIC) are now supported. 553 * [27]Fortran 2008: 554 + [28]Coarrays: Full experimental support of Fortran 2008's 555 coarrays with -fcoarray=lib except for allocatable/pointer 556 components of derived-type coarrays. GCC currently only ships 557 with a single-image library (libcaf_single), but multi-image 558 support based on MPI and GASNet is provided by the libraries 559 of the [29]OpenCoarrays project. 560 * TS18508 Additional Parallel Features in Fortran: 561 + Support for the collective intrinsic subroutines CO_MAX, 562 CO_MIN, CO_SUM, CO_BROADCAST and CO_REDUCE has been added, 563 including -fcoarray=lib support. 564 + Support for the new atomic intrinsics has been added, 565 including -fcoarray=lib support. 566 * Fortran 2015: 567 + Support for IMPLICIT NONE (external, type). 568 + ERROR STOP is now permitted in pure procedures. 569 570 Go 571 572 * GCC 5 provides a complete implementation of the Go 1.4.2 release. 573 * Building GCC 5 with Go enabled will install two new programs: 574 [30]go and [31]gofmt. 575 576 libgccjit 577 578 New in GCC 5 is the ability to build GCC as a shared library for 579 embedding in other processes (such as interpreters), suitable for 580 Just-In-Time compilation to machine code. 581 582 The shared library has a [32]C API and a [33]C++ wrapper API providing 583 some "syntactic sugar". There are also bindings available from 3rd 584 parties for [34]Python and for [35]D. 585 586 For example, this library can be used by interpreters for [36]compiling 587 functions from bytecode to machine code. 588 589 The library can also be used for ahead-of-time compilation, enabling 590 GCC to be plugged into a pre-existing frontend. An example of using 591 this to build a compiler for an esoteric language we'll refer to as 592 "brainf" can be seen [37]here. 593 594 libgccjit is licensed under the GPLv3 (or at your option, any later 595 version) 596 597 It should be regarded as experimental at this time. 598 599 New Targets and Target Specific Improvements 600 601 Reporting stack usage 602 603 * The BFIN, FT32, H8300, IQ2000 and M32C targets now support the 604 -fstack-usage option. 605 606 AArch64 607 608 * Code generation for the ARM Cortex-A57 processor has been improved. 609 A more accurate instruction scheduling model for the processor is 610 now used, and a number of compiler tuning parameters have been set 611 to offer increased performance when compiling with -mcpu=cortex-a57 612 or -mtune=cortex-a57. 613 * A workaround for the ARM Cortex-A53 erratum 835769 has been added 614 and can be enabled by giving the -mfix-cortex-a53-835769 option. 615 Alternatively it can be enabled by default by configuring GCC with 616 the --enable-fix-cortex-a53-835769 option. 617 * The optional cryptographic extensions to the ARMv8-A architecture 618 are no longer enabled by default when specifying the 619 -mcpu=cortex-a53, -mcpu=cortex-a57 or -mcpu=cortex-a57.cortex-a53 620 options. To enable these extensions add +crypto to the value of 621 -mcpu or -march e.g. -mcpu=cortex-a53+crypto. 622 * Support has been added for the following processors (GCC 623 identifiers in parentheses): ARM Cortex-A72 (cortex-a72) and 624 initial support for its big.LITTLE combination with the ARM 625 Cortex-A53 (cortex-a72.cortex-a53), Cavium ThunderX (thunderx), 626 Applied Micro X-Gene 1 (xgene1), and Samsung Exynos M1 (exynos-m1). 627 The GCC identifiers can be used as arguments to the -mcpu or -mtune 628 options, for example: -mcpu=xgene1 or -mtune=cortex-a72.cortex-a53. 629 Using -mcpu=cortex-a72 requires a version of GNU binutils that has 630 support for the Cortex-A72. 631 * The transitional options -mlra and -mno-lra have been removed. The 632 AArch64 backend now uses the local register allocator (LRA) only. 633 634 ARM 635 636 * Thumb-1 assembly code is now generated in unified syntax. The new 637 option -masm-syntax-unified specifies whether inline assembly code 638 is using unified syntax. By default the option is off which means 639 non-unified syntax is used. However this is subject to change in 640 future releases. Eventually the non-unified syntax will be 641 deprecated. 642 * It is now a configure-time error to use the --with-cpu configure 643 option with either of --with-tune or --with-arch. 644 * Code generation for the ARM Cortex-A57 processor has been improved. 645 A more accurate instruction scheduling model for the processor is 646 now used, and a number of compiler tuning parameters have been set 647 to offer increased performance when compiling with -mcpu=cortex-a57 648 or -mtune=cortex-a57. 649 * Support has been added for the following processors (GCC 650 identifiers in parentheses): ARM Cortex-A17 (cortex-a17) and 651 initial support for its big.LITTLE combination with the ARM 652 Cortex-A7 (cortex-a17.cortex-a7), ARM Cortex-A72 (cortex-a72) and 653 initial support for its big.LITTLE combination with the ARM 654 Cortex-A53 (cortex-a72.cortex-a53), ARM Cortex-M7 (cortex-m7), 655 Applied Micro X-Gene 1 (xgene1), and Samsung Exynos M1 (exynos-m1). 656 The GCC identifiers can be used as arguments to the -mcpu or -mtune 657 options, for example: -mcpu=xgene1 or -mtune=cortex-a72.cortex-a53. 658 Using -mcpu=cortex-a72 requires a version of GNU binutils that has 659 support for the Cortex-A72. 660 * The deprecated option -mwords-little-endian has been removed. 661 * The options -mapcs, -mapcs-frame, -mtpcs-frame and 662 -mtpcs-leaf-frame which are only applicable to the old ABI have 663 been deprecated. 664 * The transitional options -mlra and -mno-lra have been removed. The 665 ARM backend now uses the local register allocator (LRA) only. 666 667 AVR 668 669 * The compiler no more supports individual devices like ATmega8. 670 Specifying, say, -mmcu=atmega8 triggers the usage of the 671 device-specific [38]spec file specs-atmega8 which is part of the 672 installation and describes options for the sub-processes like 673 compiler proper, assembler and linker. You can add support for a 674 new device -mmcu=mydevice as follows: 675 1. In an empty directory /someplace, create a new directory 676 device-specs. 677 2. Copy a device spec file from the installed device-specs 678 folder, follow the comments in that file and then save it as 679 /someplace/device-specs/specs-mydevice. 680 3. Add -B /someplace -mmcu=mydevice to the compiler's 681 command-line options. Notice that /someplace must specify an 682 absolute path and that mydevice must not start with "avr". 683 4. Provided you have a device-specific library libmydevice.a 684 available, you can put it at /someplace, dito for a 685 device-specific startup file crtmydevice.o. 686 The contents of the device spec files depend on the compiler's 687 configuration, in particular on --with-avrlibc=no and whether or 688 not it is configured for RTEMS. 689 * A new command-line option -nodevicelib has been added. It prevents 690 the compiler from linking against AVR-LibC's device-specific 691 library libdevice.a. 692 * The following three command-line options have been added: 693 694 -mrmw 695 Set if the device supports the read-modify-write 696 instructions LAC, LAS, LAT and XCH. 697 698 -mn-flash=size 699 Specify the flash size of the device in units of 64 KiB, 700 rounded up to the next integer as needed. This option 701 affects the availability of the [39]AVR address-spaces. 702 703 -mskip-bug 704 Set if the device is affected by the respective silicon 705 bug. 706 707 In general, you don't need to set these options by hand. The new 708 device-specific spec file will set them as needed. 709 710 IA-32/x86-64 711 712 * New [40]ISA extensions support AVX-512{BW,DQ,VL,IFMA,VBMI} of 713 Intel's CPU codenamed Skylake Server was added to GCC. That 714 includes inline assembly support, new intrinsics, and basic 715 autovectorization. These new AVX-512 extensions are available via 716 the following GCC switches: AVX-512 Vector Length EVEX feature: 717 -mavx512vl, AVX-512 Byte and Word instructions: -mavx512bw, AVX-512 718 Dword and Qword instructions: -mavx512dq, AVX-512 FMA-52 719 instructions: -mavx512ifma and for AVX-512 Vector Bit Manipulation 720 Instructions: -mavx512vbmi. 721 * New ISA extensions support Intel MPX was added to GCC. This new 722 extension is available via the -mmpx compiler switch. Intel MPX is 723 a set of processor features which, with compiler, run-time library 724 and OS support, brings increased robustness to software by run-time 725 checking pointer references against their bounds. In GCC Intel MPX 726 is supported by Pointer Bounds Checker and libmpx run-time 727 libraries. 728 * The new -mrecord-mcount option for -pg generates a Linux kernel 729 style table of pointers to mcount or __fentry__ calls at the 730 beginning of functions. The new -mnop-mcount option in addition 731 also generates nops in place of the __fentry__ or mcount call, so 732 that a call per function can be later patched in. This can be used 733 for low overhead tracing or hot code patching. 734 * The new -malign-data option controls how GCC aligns variables. 735 -malign-data=compat uses increased alignment compatible with GCC 736 4.8 and earlier, -malign-data=abi uses alignment as specified by 737 the psABI, and -malign-data=cacheline uses increased alignment to 738 match the cache line size. -malign-data=compat is the default. 739 * The new -mskip-rax-setup option skips setting up the RAX register 740 when SSE is disabled and there are no variable arguments passed in 741 vector registers. This can be used to optimize the Linux kernel. 742 743 MIPS 744 745 * MIPS Releases 3 and 5 are now directly supported. Use the 746 command-line options -mips32r3, -mips64r3, -mips32r5 and -mips64r5 747 to enable code-generation for these processors. 748 * The Imagination P5600 processor is now supported using the 749 -march=p5600 command-line option. 750 * The Cavium Octeon3 processor is now supported using the 751 -march=octeon3 command-line option. 752 * MIPS Release 6 is now supported using the -mips32r6 and -mips64r6 753 command-line options. 754 * The o32 ABI has been modified and extended. The o32 64-bit 755 floating-point register support is now obsolete and has been 756 removed. It has been replaced by three ABI extensions FPXX, FP64A, 757 and FP64. The meaning of the -mfp64 command-line option has 758 changed. It is now used to enable the FP64A and FP64 ABI 759 extensions. 760 + The FPXX extension requires that code generated to access 761 double-precision values use even-numbered registers. Code that 762 adheres to this extension is link-compatible with all other 763 o32 double-precision ABI variants and will execute correctly 764 in all hardware FPU modes. The command-line options -mabi=32 765 -mfpxx can be used to enable this extension. MIPS II is the 766 minimum processor required. 767 + The o32 FP64A extension requires that floating-point registers 768 be 64-bit and odd-numbered single-precision registers are not 769 allowed. Code that adheres to the o32 FP64A variant is 770 link-compatible with all other o32 double-precision ABI 771 variants. The command-line options -mabi=32 -mfp64 772 -mno-odd-spreg can be used to enable this extension. MIPS32R2 773 is the minimum processor required. 774 + The o32 FP64 extension also requires that floating-point 775 registers be 64-bit, but permits the use of single-precision 776 registers. Code that adheres to the o32 FP64 variant is 777 link-compatible with o32 FPXX and o32 FP64A variants only, 778 i.e. it is not compatible with the original o32 779 double-precision ABI. The command-line options -mabi=32 -mfp64 780 -modd-spreg can be used to enable this extension. MIPS32R2 is 781 the minimum processor required. 782 The new ABI variants can be enabled by default using the configure 783 time options --with-fp-32=[32|xx|64] and --with(out)-odd-sp-reg-32. 784 It is strongly recommended that all vendors begin to set o32 FPXX 785 as the default ABI. This will be required to run the generated code 786 on MIPSR5 cores in conjunction with future MIPS SIMD (MSA) code and 787 MIPSR6 cores. 788 * GCC will now pass all floating-point options to the assembler if 789 GNU binutils 2.25 is used. As a result, any inline assembly code 790 that uses hard-float instructions should be amended to include a 791 .set directive to override the global assembler options when 792 compiling for soft-float targets. 793 794 NDS32 795 796 * The variadic function ABI implementation is now compatible with 797 past Andes toolchains where the caller uses registers to pass 798 arguments and the callee is in charge of pushing them on stack. 799 * The options -mforce-fp-as-gp, -mforbid-fp-as-gp, and -mex9 have 800 been removed since they are not yet available in the nds32 port of 801 GNU binutils. 802 * A new option -mcmodel=[small|medium|large] supports varied code 803 models on code generation. The -mgp-direct option became 804 meaningless and can be discarded. 805 806 RX 807 808 * A new command line option -mno-allow-string-insns can be used to 809 disable the generation of the SCMPU, SMOVU, SMOVB, SMOVF, SUNTIL, 810 SWHILE and RMPA instructions. An erratum released by Renesas shows 811 that it is unsafe to use these instructions on addresses within the 812 I/O space of the processor. The new option can be used when the 813 programmer is concerned that the I/O space might be accessed. The 814 default is still to enable these instructions. 815 816 SH 817 818 * The compiler will now pass the appropriate --isa= option to the 819 assembler. 820 * The default handling for the GBR has been changed from call 821 clobbered to call preserved. The old behavior can be reinstated by 822 specifying the option -fcall-used-gbr. 823 * Support for the SH4A fpchg instruction has been added which will be 824 utilized when switching between single and double precision FPU 825 modes. 826 * The compiler no longer uses the __fpscr_values array for switching 827 between single and double FPU precision modes on non-SH4A targets. 828 Instead mode switching will now be performed by storing, modifying 829 and reloading the FPSCR, so that other FPSCR bits are preserved 830 across mode switches. The __fpscr_values array that is defined in 831 libgcc is still present for backwards compatibility, but it will 832 not be referenced by compiler generated code anymore. 833 * New builtin functions __builtin_sh_get_fpscr and 834 __builtin_sh_set_fpscr have been added. The __builtin_sh_set_fpscr 835 function will mask the specified bits in such a way that the SZ, PR 836 and FR mode bits will be preserved, while changing the other bits. 837 These new functions do not reference the __fpscr_values array. The 838 old functions __set_fpscr and __get_fpscr in libgcc which access 839 the __fpscr_values array are still present for backwards 840 compatibility, but their usage is highly discouraged. 841 * Some improvements to code generated for __atomic built-in 842 functions. 843 * When compiling for SH2E the compiler will no longer force the usage 844 of delay slots for conditional branch instructions bt and bf. The 845 old behavior can be reinstated (e.g. to work around a hardware bug 846 in the original SH7055) by specifying the new option 847 -mcbranch-force-delay-slot. 848 849 Operating Systems 850 851 AIX 852 853 * GCC now supports stabs debugging continuation lines to allow long 854 stabs debug information without overflow that generates AIX linker 855 errors. 856 857 DragonFly BSD 858 859 * GCC now supports the DragonFly BSD operating system. 860 861 FreeBSD 862 863 * GCC now supports the FreeBSD operating system for the arm port 864 through the arm*-*-freebsd* target triplets. 865 866 VxWorks MILS 867 868 * GCC now supports the MILS (Multiple Independent Levels of Security) 869 variant of WindRiver's VxWorks operating system for PowerPC 870 targets. 871 872 Other significant improvements 873 874 * The gcc-ar, gcc-nm, gcc-ranlib wrappers now understand a -B option 875 to set the compiler to use. 876 877 * When the new command-line option -freport-bug is used, GCC 878 automatically generates a developer-friendly reproducer whenever an 879 internal compiler error is encountered. 880 881 GCC 5.2 882 883 This is the [41]list of problem reports (PRs) from GCC's bug tracking 884 system that are known to be fixed in the 5.2 release. This list might 885 not be complete (that is, it is possible that some PRs that have been 886 fixed are not listed here). 887 888 Target Specific Changes 889 890 IA-32/x86-64 891 892 * Support for new AMD instructions monitorx and mwaitx has been 893 added. This includes new intrinsic and built-in support. It is 894 enabled through option -mmwaitx. The instructions monitorx and 895 mwaitx implement the same functionality as the old monitor and 896 mwait instructions. In addition mwaitx adds a configurable timer. 897 The timer value is received as third argument and stored in 898 register %ebx. 899 900 S/390, System z, IBM z Systems 901 902 * Support for the IBM z13 processor has been added. When using the 903 -march=z13 option, the compiler will generate code making use of 904 the new instructions and registers introduced with the vector 905 extension facility. The -mtune=z13 option enables z13 specific 906 instruction scheduling without making use of new instructions. 907 Compiling code with -march=z13 reduces the default alignment of 908 vector types bigger than 8 bytes to 8. This is an ABI change and 909 care must be taken when linking modules compiled with different 910 arch levels which interchange variables containing vector type 911 values. For newly compiled code the GNU linker will emit a warning. 912 * The -mzvector option enables a C/C++ language extension. This 913 extension provides a new keyword vector which can be used to define 914 vector type variables. (Note: This is not available when enforcing 915 strict standard compliance e.g. with -std=c99. Either enable GNU 916 extensions with e.g. -std=gnu99 or use __vector instead of vector.) 917 Additionally a set of overloaded builtins is provided which is 918 partially compatible to the PowerPC Altivec builtins. In order to 919 make use of these builtins the vecintrin.h header file needs to be 920 included. 921 922 GCC 5.3 923 924 This is the [42]list of problem reports (PRs) from GCC's bug tracking 925 system that are known to be fixed in the 5.3 release. This list might 926 not be complete (that is, it is possible that some PRs that have been 927 fixed are not listed here). 928 929 Target Specific Changes 930 931 IA-32/x86-64 932 933 * GCC now supports the Intel CPU named Skylake with AVX-512 934 extensions through -march=skylake-avx512. The switch enables the 935 following ISA extensions: AVX-512F, AVX512VL, AVX-512CD, AVX-512BW, 936 AVX-512DQ. 937 938 S/390, System z, IBM z Systems 939 940 * With this version of GCC IBM z Systems support has been added to 941 the GO runtime environment. GCC 5.3 has proven to be able to 942 compile larger GO applications on IBM z Systems. 943 944 GCC 5.4 945 946 This is the [43]list of problem reports (PRs) from GCC's bug tracking 947 system that are known to be fixed in the 5.4 release. This list might 948 not be complete (that is, it is possible that some PRs that have been 949 fixed are not listed here). 950 951 GCC 5.5 952 953 This is the [44]list of problem reports (PRs) from GCC's bug tracking 954 system that are known to be fixed in the 5.5 release. This list might 955 not be complete (that is, it is possible that some PRs that have been 956 fixed are not listed here). 957 958 Target Specific Changes 959 960 IA-32/x86-64 961 962 * Support for the [45]deprecated pcommit instruction has been 963 removed. 964 965 966 For questions related to the use of GCC, please consult these web 967 pages and the [46]GCC manuals. If that fails, the 968 [47]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 969 web pages and the development of GCC are welcome on our developer 970 list at [48]gcc (a] gcc.gnu.org. All of [49]our lists have public 971 archives. 972 973 Copyright (C) [50]Free Software Foundation, Inc. Verbatim copying and 974 distribution of this entire article is permitted in any medium, 975 provided this notice is preserved. 976 977 These pages are [51]maintained by the GCC team. Last modified 978 2017-10-10[52]. 979 980 References 981 982 1. http://gcc.gnu.org/gcc-5/changes.html#libstdcxx 983 2. https://gcc.gnu.org/onlinedocs/gcc/Spec-Files.html 984 3. https://savannah.nongnu.org/bugs/?44574 985 4. https://gcc.gnu.org/wiki/Intel MPX support in the GCC compiler 986 5. http://www.openmp.org/wp-content/uploads/OpenMP4.0.0.pdf 987 6. http://www.openmp.org/wp-content/uploads/OpenMP4.0.0.Examples.pdf 988 7. https://gcc.gnu.org/wiki/OpenACC 989 8. https://gcc.gnu.org/install/configure.html 990 9. https://www.cilkplus.org/ 991 10. https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations 992 11. http://gcc.gnu.org/projects/cxx1y.html 993 12. https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations 994 13. http://gcc.gnu.org/projects/cxx1y.html 995 14. http://gcc.gnu.org/projects/cxx1y.html 996 15. http://gcc.gnu.org/projects/cxx1y.html 997 16. http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2013/n3639.html 998 17. http://gcc.gnu.org/onlinedocs/libstdc++/manual/using_dual_abi.html 999 18. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/libstdc++/manual/manual/status.html#status.iso.2011 1000 19. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/libstdc++/manual/manual/status.html#status.iso.2014 1001 20. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/libstdc++/manual/manual/status.html#status.iso.2014 1002 21. https://sourceware.org/gdb/current/onlinedocs/gdb/Xmethods-In-Python.html 1003 22. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/gfortran/Error-and-Warning-Options.html 1004 23. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/gcc/Language-Independent-Options.html 1005 24. https://gcc.gnu.org/install/configure.html 1006 25. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/gfortran/Thread-safety-of-the-runtime-library.html 1007 26. https://gcc.gnu.org/wiki/Fortran2003Status 1008 27. https://gcc.gnu.org/wiki/Fortran2008Status 1009 28. https://gcc.gnu.org/wiki/Coarray 1010 29. http://www.opencoarrays.org/ 1011 30. https://golang.org/cmd/go/ 1012 31. https://golang.org/cmd/gofmt/ 1013 32. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/jit/intro/index.html 1014 33. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/jit/cp/index.html 1015 34. https://github.com/davidmalcolm/pygccjit 1016 35. https://github.com/ibuclaw/gccjitd 1017 36. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/jit/intro/tutorial04.html 1018 37. https://gcc.gnu.org/onlinedocs/gcc-5.1.0/jit/intro/tutorial05.html 1019 38. https://gcc.gnu.org/onlinedocs/gcc/Spec-Files.html 1020 39. https://gcc.gnu.org/onlinedocs/gcc/Named-Address-Spaces.html 1021 40. https://software.intel.com/sites/default/files/managed/39/c5/325462-sdm-vol-1-2abcd-3abcd.pdf 1022 41. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=5.2 1023 42. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=5.3 1024 43. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=5.4 1025 44. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=5.5 1026 45. https://software.intel.com/en-us/blogs/2016/09/12/deprecate-pcommit-instruction 1027 46. https://gcc.gnu.org/onlinedocs/ 1028 47. mailto:gcc-help (a] gcc.gnu.org 1029 48. mailto:gcc (a] gcc.gnu.org 1030 49. https://gcc.gnu.org/lists.html 1031 50. http://www.fsf.org/ 1032 51. https://gcc.gnu.org/about.html 1033 52. http://validator.w3.org/check/referer 1034 ====================================================================== 1035 http://gcc.gnu.org/gcc-4.9/index.html 1036 GCC 4.9 Release Series 1037 1038 Aug 3, 2016 1039 1040 The [1]GNU project and the GCC developers are pleased to announce the 1041 release of GCC 4.9.4. 1042 1043 This release is a bug-fix release, containing fixes for regressions in 1044 GCC 4.9.3 relative to previous releases of GCC. 1045 1046 This release series is no longer maintained. 1047 1048 Release History 1049 1050 GCC 4.9.4 1051 Aug 3, 2016 ([2]changes, [3]documentation) 1052 1053 GCC 4.9.3 1054 June 26, 2015 ([4]changes, [5]documentation) 1055 1056 GCC 4.9.2 1057 October 30, 2014 ([6]changes, [7]documentation) 1058 1059 GCC 4.9.1 1060 July 16, 2014 ([8]changes, [9]documentation) 1061 1062 GCC 4.9.0 1063 April 22, 2014 ([10]changes, [11]documentation) 1064 1065 References and Acknowledgements 1066 1067 GCC used to stand for the GNU C Compiler, but since the compiler 1068 supports several other languages aside from C, it now stands for the 1069 GNU Compiler Collection. 1070 1071 A list of [12]successful builds is updated as new information becomes 1072 available. 1073 1074 The GCC developers would like to thank the numerous people that have 1075 contributed new features, improvements, bug fixes, and other changes as 1076 well as test results to GCC. This [13]amazing group of volunteers is 1077 what makes GCC successful. 1078 1079 For additional information about GCC please refer to the [14]GCC 1080 project web site or contact the [15]GCC development mailing list. 1081 1082 To obtain GCC please use [16]our mirror sites or [17]our SVN server. 1083 1084 1085 For questions related to the use of GCC, please consult these web 1086 pages and the [18]GCC manuals. If that fails, the 1087 [19]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 1088 web pages and the development of GCC are welcome on our developer 1089 list at [20]gcc (a] gcc.gnu.org. All of [21]our lists have public 1090 archives. 1091 1092 Copyright (C) [22]Free Software Foundation, Inc. Verbatim copying and 1093 distribution of this entire article is permitted in any medium, 1094 provided this notice is preserved. 1095 1096 These pages are [23]maintained by the GCC team. Last modified 1097 2016-09-30[24]. 1098 1099 References 1100 1101 1. http://www.gnu.org/ 1102 2. http://gcc.gnu.org/gcc-4.9/changes.html 1103 3. http://gcc.gnu.org/onlinedocs/4.9.4/ 1104 4. http://gcc.gnu.org/gcc-4.9/changes.html 1105 5. http://gcc.gnu.org/onlinedocs/4.9.3/ 1106 6. http://gcc.gnu.org/gcc-4.9/changes.html 1107 7. http://gcc.gnu.org/onlinedocs/4.9.2/ 1108 8. http://gcc.gnu.org/gcc-4.9/changes.html 1109 9. http://gcc.gnu.org/onlinedocs/4.9.1/ 1110 10. http://gcc.gnu.org/gcc-4.9/changes.html 1111 11. https://gcc.gnu.org/onlinedocs/4.9.0/ 1112 12. http://gcc.gnu.org/gcc-4.9/buildstat.html 1113 13. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 1114 14. http://gcc.gnu.org/index.html 1115 15. mailto:gcc (a] gcc.gnu.org 1116 16. http://gcc.gnu.org/mirrors.html 1117 17. http://gcc.gnu.org/svn.html 1118 18. https://gcc.gnu.org/onlinedocs/ 1119 19. mailto:gcc-help (a] gcc.gnu.org 1120 20. mailto:gcc (a] gcc.gnu.org 1121 21. https://gcc.gnu.org/lists.html 1122 22. http://www.fsf.org/ 1123 23. https://gcc.gnu.org/about.html 1124 24. http://validator.w3.org/check/referer 1125 ====================================================================== 1126 http://gcc.gnu.org/gcc-4.9/changes.html 1127 GCC 4.9 Release Series 1128 Changes, New Features, and Fixes 1129 1130 Caveats 1131 1132 * The mudflap run time checker has been removed. The mudflap options 1133 remain, but do nothing. 1134 * Support for a number of older systems and recently unmaintained or 1135 untested target ports of GCC has been declared obsolete in GCC 4.9. 1136 Unless there is activity to revive them, the next release of GCC 1137 will have their sources permanently removed. 1138 The following ports for individual systems on particular 1139 architectures have been obsoleted: 1140 + Solaris 9 (*-*-solaris2.9). Details can be found in the 1141 [1]announcement. 1142 * On AArch64, the singleton vector types int64x1_t, uint64x1_t and 1143 float64x1_t exported by arm_neon.h are defined to be the same as 1144 their base types. This results in incorrect application of 1145 parameter passing rules to arguments of types int64x1_t and 1146 uint64x1_t, with respect to the AAPCS64 ABI specification. In 1147 addition, names of C++ functions with parameters of these types 1148 (including float64x1_t) are not mangled correctly. The current 1149 typedef declarations also unintentionally allow implicit casting 1150 between singleton vector types and their base types. These issues 1151 will be resolved in a near future release. See [2]PR60825 for more 1152 information. 1153 1154 More information on porting to GCC 4.9 from previous versions of GCC 1155 can be found in the [3]porting guide for this release. 1156 1157 General Optimizer Improvements 1158 1159 * AddressSanitizer, a fast memory error detector, is now available on 1160 ARM. 1161 * UndefinedBehaviorSanitizer (ubsan), a fast undefined behavior 1162 detector, has been added and can be enabled via 1163 -fsanitize=undefined. Various computations will be instrumented to 1164 detect undefined behavior at runtime. UndefinedBehaviorSanitizer is 1165 currently available for the C and C++ languages. 1166 * Link-time optimization (LTO) improvements: 1167 + Type merging was rewritten. The new implementation is 1168 significantly faster and uses less memory. 1169 + Better partitioning algorithm resulting in less streaming 1170 during link time. 1171 + Early removal of virtual methods reduces the size of object 1172 files and improves link-time memory usage and compile time. 1173 + Function bodies are now loaded on-demand and released early 1174 improving overall memory usage at link time. 1175 + C++ hidden keyed methods can now be optimized out. 1176 + When using a linker plugin, compiling with the -flto option 1177 now generates slim object files (.o) which only contain 1178 intermediate language representation for LTO. Use 1179 -ffat-lto-objects to create files which contain additionally 1180 the object code. To generate static libraries suitable for LTO 1181 processing, use gcc-ar and gcc-ranlib; to list symbols from a 1182 slim object file use gcc-nm. (This requires that ar, ranlib 1183 and nm have been compiled with plugin support.) 1184 Memory usage building Firefox with debug enabled was reduced from 1185 15GB to 3.5GB; link time from 1700 seconds to 350 seconds. 1186 * Inter-procedural optimization improvements: 1187 + New type inheritance analysis module improving 1188 devirtualization. Devirtualization now takes into account 1189 anonymous name-spaces and the C++11 final keyword. 1190 + New speculative devirtualization pass (controlled by 1191 -fdevirtualize-speculatively. 1192 + Calls that were speculatively made direct are turned back to 1193 indirect where direct call is not cheaper. 1194 + Local aliases are introduced for symbols that are known to be 1195 semantically equivalent across shared libraries improving 1196 dynamic linking times. 1197 * Feedback directed optimization improvements: 1198 + Profiling of programs using C++ inline functions is now more 1199 reliable. 1200 + New time profiling determines typical order in which functions 1201 are executed. 1202 + A new function reordering pass (controlled by 1203 -freorder-functions) significantly reduces startup time of 1204 large applications. Until binutils support is completed, it is 1205 effective only with link-time optimization. 1206 + Feedback driven indirect call removal and devirtualization now 1207 handle cross-module calls when link-time optimization is 1208 enabled. 1209 1210 New Languages and Language specific improvements 1211 1212 * Version 4.0 of the [4]OpenMP specification is now supported in the 1213 C and C++ compilers and starting with the 4.9.1 release also in the 1214 Fortran compiler. The new -fopenmp-simd option can be used to 1215 enable OpenMP's SIMD directives while ignoring other OpenMP 1216 directives. The new [5]-fsimd-cost-model= option permits to tune 1217 the vectorization cost model for loops annotated with OpenMP and 1218 Cilk Plus simd directives. -Wopenmp-simd warns when the current 1219 cost model overrides simd directives set by the user. 1220 * The -Wdate-time option has been added for the C, C++ and Fortran 1221 compilers, which warns when the __DATE__, __TIME__ or __TIMESTAMP__ 1222 macros are used. Those macros might prevent bit-wise-identical 1223 reproducible compilations. 1224 1225 Ada 1226 1227 * GNAT switched to Ada 2012 instead of Ada 2005 by default. 1228 1229 C family 1230 1231 * Support for colorizing diagnostics emitted by GCC has been added. 1232 The [6]-fdiagnostics-color=auto will enable it when outputting to 1233 terminals, -fdiagnostics-color=always unconditionally. The 1234 GCC_COLORS environment variable can be used to customize the colors 1235 or disable coloring. If GCC_COLORS variable is present in the 1236 environment, the default is -fdiagnostics-color=auto, otherwise 1237 -fdiagnostics-color=never. 1238 Sample diagnostics output: 1239 $ g++ -fdiagnostics-color=always -S -Wall test.C 1240 test.C: In function int foo(): 1241 test.C:1:14: warning: no return statement in function returning non-void [-W 1242 return-type] 1243 int foo () { } 1244 ^ 1245 test.C:2:46: error: template instantiation depth exceeds maximum of 900 (use 1246 -ftemplate-depth= to increase the maximum) instantiating struct X<100> 1247 template <int N> struct X { static const int value = X<N-1>::value; }; temp 1248 late struct X<1000>; 1249 ^ 1250 test.C:2:46: recursively required from const int X<999>::value 1251 test.C:2:46: required from const int X<1000>::value 1252 test.C:2:88: required from here 1253 1254 test.C:2:46: error: incomplete type X<100> used in nested name specifier 1255 1256 * With the new [7]#pragma GCC ivdep, the user can assert that there 1257 are no loop-carried dependencies which would prevent concurrent 1258 execution of consecutive iterations using SIMD (single instruction 1259 multiple data) instructions. 1260 * Support for [8]Cilk Plus has been added and can be enabled with the 1261 -fcilkplus option. Cilk Plus is an extension to the C and C++ 1262 languages to support data and task parallelism. The present 1263 implementation follows ABI version 1.2; all features but _Cilk_for 1264 have been implemented. 1265 1266 C 1267 1268 * ISO C11 atomics (the _Atomic type specifier and qualifier and the 1269 <stdatomic.h> header) are now supported. 1270 * ISO C11 generic selections (_Generic keyword) are now supported. 1271 * ISO C11 thread-local storage (_Thread_local, similar to GNU C 1272 __thread) is now supported. 1273 * ISO C11 support is now at a similar level of completeness to ISO 1274 C99 support: substantially complete modulo bugs, extended 1275 identifiers (supported except for corner cases when 1276 -fextended-identifiers is used), floating-point issues (mainly but 1277 not entirely relating to optional C99 features from Annexes F and 1278 G) and the optional Annexes K (Bounds-checking interfaces) and L 1279 (Analyzability). 1280 * A new C extension __auto_type provides a subset of the 1281 functionality of C++11 auto in GNU C. 1282 1283 C++ 1284 1285 * The G++ implementation of [9]C++1y return type deduction for normal 1286 functions has been updated to conform to [10]N3638, the proposal 1287 accepted into the working paper. Most notably, it adds 1288 decltype(auto) for getting decltype semantics rather than the 1289 template argument deduction semantics of plain auto: 1290 1291 int& f(); 1292 auto i1 = f(); // int 1293 decltype(auto) i2 = f(); // int& 1294 1295 * G++ supports [11]C++1y lambda capture initializers: 1296 1297 [x = 42]{ ... }; 1298 1299 Actually, they have been accepted since GCC 4.5, but now the 1300 compiler doesn't warn about them with -std=c++1y, and supports 1301 parenthesized and brace-enclosed initializers as well. 1302 * G++ supports [12]C++1y variable length arrays. G++ has supported 1303 GNU/C99-style VLAs for a long time, but now additionally supports 1304 initializers and lambda capture by reference. In C++1y mode G++ 1305 will complain about VLA uses that are not permitted by the draft 1306 standard, such as forming a pointer to VLA type or applying sizeof 1307 to a VLA variable. Note that it now appears that VLAs will not be 1308 part of C++14, but will be part of a separate document and then 1309 perhaps C++17. 1310 1311 void f(int n) { 1312 int a[n] = { 1, 2, 3 }; // throws std::bad_array_length if n < 3 1313 [&a]{ for (int i : a) { cout << i << endl; } }(); 1314 &a; // error, taking address of VLA 1315 } 1316 1317 * G++ supports the [13]C++1y [[deprecated]] attribute modulo bugs in 1318 the underlying [[gnu::deprecated]] attribute. Classes and functions 1319 can be marked deprecated and a diagnostic message added: 1320 1321 class A; 1322 int bar(int n); 1323 #if __cplusplus > 201103 1324 class [[deprecated("A is deprecated in C++14; Use B instead")]] A; 1325 [[deprecated("bar is unsafe; use foo() instead")]] 1326 int bar(int n); 1327 1328 int foo(int n); 1329 class B; 1330 #endif 1331 A aa; // warning: 'A' is deprecated : A is deprecated in C++14; Use B instead 1332 int j = bar(2); // warning: 'int bar(int)' is deprecated : bar is unsafe; use fo 1333 o() instead 1334 1335 * G++ supports [14]C++1y digit separators. Long numeric literals can 1336 be subdivided with a single quote ' to enhance readability: 1337 1338 int i = 1048576; 1339 int j = 1'048'576; 1340 int k = 0x10'0000; 1341 int m = 0'004'000'000; 1342 int n = 0b0001'0000'0000'0000'0000'0000; 1343 1344 double x = 1.602'176'565e-19; 1345 double y = 1.602'176'565e-1'9; 1346 1347 * G++ supports [15]C++1y generic (polymorphic) lambdas. 1348 1349 // a functional object that will increment any type 1350 auto incr = [](auto x) { return x++; }; 1351 1352 * As a GNU extension, G++ supports explicit template parameter syntax 1353 for generic lambdas. This can be combined in the expected way with 1354 the standard auto syntax. 1355 1356 // a functional object that will add two like-type objects 1357 auto add = [] <typename T> (T a, T b) { return a + b; }; 1358 1359 * G++ supports unconstrained generic functions as specified by 4.1.2 1360 and 5.1.1 of [16]N3889: Concepts Lite Specification. Briefly, auto 1361 may be used as a type-specifier in a parameter declaration of any 1362 function declarator in order to introduce an implicit function 1363 template parameter, akin to generic lambdas. 1364 1365 // the following two function declarations are equivalent 1366 auto incr(auto x) { return x++; } 1367 template <typename T> 1368 auto incr(T x) { return x++; } 1369 1370 Runtime Library (libstdc++) 1371 1372 * [17]Improved support for C++11, including: 1373 + support for <regex>; 1374 + The associative containers in <map> and <set> and the 1375 unordered associative containers in <unordered_map> and 1376 <unordered_set> meet the allocator-aware container 1377 requirements; 1378 * [18]Improved experimental support for the upcoming ISO C++ 1379 standard, C++14, including: 1380 + fixing constexpr member functions without const; 1381 + implementation of the std::exchange() utility function; 1382 + addressing tuples by type; 1383 + implemention of std::make_unique; 1384 + implemention of std::shared_lock; 1385 + making std::result_of SFINAE-friendly; 1386 + adding operator() to std::integral_constant; 1387 + adding user-defined literals for standard library types 1388 std::basic_string, std::chrono::duration, and std::complex; 1389 + adding two range overloads to non-modifying sequence oprations 1390 std::equal and std::mismatch; 1391 + adding IO manipulators for quoted strings; 1392 + adding constexpr members to <utility>, <complex>, <chrono>, 1393 and some containers; 1394 + adding compile-time std::integer_sequence; 1395 + adding cleaner transformation traits; 1396 + making <functional>s operator functors easier to use and more 1397 generic; 1398 * An implementation of std::experimental::optional. 1399 * An implementation of std::experimental::string_view. 1400 * The non-standard function std::copy_exception has been deprecated 1401 and will be removed in a future version. std::make_exception_ptr 1402 should be used instead. 1403 1404 Fortran 1405 1406 * Compatibility notice: 1407 + Module files: The version of the module files (.mod) has been 1408 incremented; additionally, module files are now compressed. 1409 Fortran MODULEs compiled by earlier GCC versions have to be 1410 recompiled, when they are USEd by files compiled with GCC 4.9. 1411 GCC 4.9 is not able to read .mod files of earlier GCC 1412 versions; attempting to do so gives an error message. Note: 1413 The ABI of the produced assembler data itself has not changed: 1414 object files and libraries are fully compatible with older 1415 versions (except as stated below). 1416 + ABI changes: 1417 o The [19]argument passing ABI has changed for scalar dummy 1418 arguments of type INTEGER, REAL, COMPLEX and LOGICAL, 1419 which have both the VALUE and the OPTIONAL attributes. 1420 o To support finalization the virtual table associated with 1421 polymorphic variables has changed. Code containing CLASS 1422 should be recompiled, including all files which define 1423 derived types involved in the type definition used by 1424 polymorphic variables. (Note: Due to the incremented 1425 module version, trying to mix old code with new code will 1426 usually give an error message.) 1427 + GNU Fortran no longer deallocates allocatable variables or 1428 allocatable components of variables declared in the main 1429 program. Since Fortran 2008, the standard explicitly states 1430 that variables declared in the Fortran main program 1431 automatically have the SAVE attribute. 1432 + When opening files, the close-on-exec flag is set if the 1433 system supports such a feature. This is generally considered 1434 good practice these days, but if there is a need to pass file 1435 descriptors to child processes the parent process must now 1436 remember to clear the close-on-exec flag by calling fcntl(), 1437 e.g. via ISO_C_BINDING, before executing the child process. 1438 * The deprecated command-line option -fno-whole-file has been 1439 removed. (-fwhole-file is the default since GCC 4.6.) 1440 -fwhole-file/-fno-whole-file continue to be accepted but do not 1441 influence the code generation. 1442 * The compiler no longer unconditionally warns about DO loops with 1443 zero iterations. This warning is now controlled by the -Wzerotrip 1444 option, which is implied by -Wall. 1445 * The new NO_ARG_CHECK attribute of the [20]!GCC$ directive can be 1446 used to disable the type-kind-rank (TKR) argument check for a dummy 1447 argument. The feature is similar to ISO/IEC TS 29133:2012's 1448 TYPE(*), except that it additionally also disables the rank check. 1449 Variables with NO_ARG_CHECK have to be dummy arguments and may only 1450 be used as argument to ISO_C_BINDING's C_LOC and as actual argument 1451 to another NO_ARG_CHECK dummy argument; also the other constraints 1452 of TYPE(*) apply. The dummy arguments should be declared as scalar 1453 or assumed-size variable of type type(*) (recommended) or of type 1454 integer, real, complex or logical. With NO_ARG_CHECK, a pointer to 1455 the data without further type or shape information is passed, 1456 similar to C's void*. Note that also TS 29113's 1457 type(*),dimension(..) accepts arguments of any type and rank; 1458 contrary to NO_ARG_CHECK assumed-rank arguments pass an array 1459 descriptor which contains the array shape and stride of the 1460 argument. 1461 * [21]Fortran 2003: 1462 + Finalization is now supported. It is currently only done for a 1463 subset of those situations in which it should occur. 1464 + Experimental support for scalar character components with 1465 deferred length (i.e. allocatable string length) in derived 1466 types has been added. (Deferred-length character variables are 1467 supported since GCC 4.6.) 1468 * [22]Fortran 2008: 1469 + When STOP or ERROR STOP are used to terminate the execution 1470 and any exception (but inexact) is signaling, a warning is 1471 printed to ERROR_UNIT, indicating which exceptions are 1472 signaling. The [23]-ffpe-summary= command-line option can be 1473 used to fine-tune for which exceptions the warning should be 1474 shown. 1475 + Rounding on input (READ) is now handled on systems where 1476 strtod honours the rounding mode. (For output, rounding is 1477 supported since GCC 4.5.) Note that for input, the compatible 1478 rounding mode is handled as nearest (i.e., rounding to an even 1479 least significant [cf. IEC 60559:1989] for a tie, while 1480 compatible rounds away from zero in that case). 1481 1482 Go 1483 1484 * GCC 4.9 provides a complete implementation of the Go 1.2.1 release. 1485 1486 New Targets and Target Specific Improvements 1487 1488 AArch64 1489 1490 * The ARMv8-A crypto and CRC instructions are now supported through 1491 intrinsics. These are enabled when the architecture supports these 1492 and are available through the -march=armv8-a+crc and 1493 -march=armv8-a+crypto options. 1494 * Initial support for ILP32 has now been added to the compiler. This 1495 is now available through the command-line option -mabi=ilp32. 1496 Support for ILP32 is considered experimental as the ABI 1497 specification is still beta. 1498 * Coverage of more of the ISA including the SIMD extensions has been 1499 added. The Advanced SIMD intrinsics have also been improved. 1500 * The new local register allocator (LRA) is now on by default for the 1501 AArch64 backend. 1502 * The REE (Redundant extension elimination) pass has now been enabled 1503 by default for the AArch64 backend. 1504 * Tuning for the Cortex-A53 and Cortex-A57 has been improved. 1505 * Initial big.LITTLE tuning support for the combination of Cortex-A57 1506 and Cortex-A53 was added through the -mcpu=cortex-a57.cortex-a53 1507 option. 1508 * A number of structural changes have been made to both the ARM and 1509 AArch64 backends to facilitate improved code-generation. 1510 * As of GCC 4.9.2 a workaround for the ARM Cortex-A53 erratum 835769 1511 has been added and can be enabled by giving the 1512 -mfix-cortex-a53-835769 option. Alternatively it can be enabled by 1513 default by configuring GCC with the --enable-fix-cortex-a53-835769 1514 option. 1515 1516 ARC 1517 1518 * A port for Synopsys Designware ARC has been contributed by Embecosm 1519 and Synopsys Inc. 1520 1521 ARM 1522 1523 * Use of Advanced SIMD (Neon) for 64-bit scalar computations has been 1524 disabled by default. This was found to generate better code in only 1525 a small number of cases. It can be turned back on with the 1526 -mneon-for-64bits option. 1527 * Further support for the ARMv8-A architecture, notably implementing 1528 the restriction around IT blocks in the Thumb32 instruction set has 1529 been added. The -mrestrict-it option can be used with 1530 -march=armv7-a or the -march=armv7ve options to make code 1531 generation fully compatible with the deprecated instructions in 1532 ARMv8-A. 1533 * Support has now been added for the ARMv7ve variant of the 1534 architecture. This can be used by the -march=armv7ve option. 1535 * The ARMv8-A crypto and CRC instructions are now supported through 1536 intrinsics and are available through the -march=armv8-a+crc and 1537 mfpu=crypto-neon-fp-armv8 options. 1538 * LRA is now on by default for the ARM target. This can be turned off 1539 using the -mno-lra option. This option is a purely transitionary 1540 command-line option and will be removed in a future release. We are 1541 interested in any bug reports regarding functional and performance 1542 regressions with LRA. 1543 * A new option -mslow-flash-data to improve performance of programs 1544 fetching data on slow flash memory has now been introduced for the 1545 ARMv7-M profile cores. 1546 * A new option -mpic-data-is-text-relative for targets that allows 1547 data segments to be relative to text segments has been added. This 1548 is on by default for all targets except VxWorks RTP. 1549 * A number of infrastructural changes have been made to both the ARM 1550 and AArch64 backends to facilitate improved code-generation. 1551 * GCC now supports Cortex-A12 and the Cortex-R7 through the 1552 -mcpu=cortex-a12 and -mcpu=cortex-r7 options. 1553 * GCC now has tuning for the Cortex-A57 and Cortex-A53 through the 1554 -mcpu=cortex-a57 and -mcpu=cortex-a53 options. 1555 * Initial big.LITTLE tuning support for the combination of Cortex-A57 1556 and Cortex-A53 was added through the -mcpu=cortex-a57.cortex-a53 1557 option. Similar support was added for the combination of Cortex-A15 1558 and Cortex-A7 through the -mcpu=cortex-a15.cortex-a7 option. 1559 * Further performance optimizations for the Cortex-A15 and the 1560 Cortex-M4 have been added. 1561 * A number of code generation improvements for Thumb2 to reduce code 1562 size when compiling for the M-profile processors. 1563 1564 AVR 1565 1566 * A new command-line option -mfract-convert-truncate has been added. 1567 It allows compiler to use truncation instead of rounding towards 1568 zero for fractional fixed-point types. 1569 1570 IA-32/x86-64 1571 1572 * -mfpmath=sse is now implied by -ffast-math on all targets where 1573 SSE2 is supported. 1574 * Intel AVX-512 support was added to GCC. That includes inline 1575 assembly support, new registers and extending existing ones, new 1576 intrinsics (covered by corresponding testsuite), and basic 1577 autovectorization. AVX-512 instructions are available via the 1578 following GCC switches: AVX-512 foundation instructions: -mavx512f, 1579 AVX-512 prefetch instructions: -mavx512pf, AVX-512 exponential and 1580 reciprocal instructions: -mavx512er, AVX-512 conflict detection 1581 instructions: -mavx512cd. 1582 * It is now possible to call x86 intrinsics from select functions in 1583 a file that are tagged with the corresponding target attribute 1584 without having to compile the entire file with the -mxxx option. 1585 This improves the usability of x86 intrinsics and is particularly 1586 useful when doing [24]Function Multiversioning. 1587 * GCC now supports the new Intel microarchitecture named Silvermont 1588 through -march=silvermont. 1589 * GCC now supports the new Intel microarchitecture named Broadwell 1590 through -march=broadwell. 1591 * Optimizing for other Intel microarchitectures have been renamed to 1592 -march=nehalem, westmere, sandybridge, ivybridge, haswell, bonnell. 1593 * -march=generic has been retuned for better support of Intel core 1594 and AMD Bulldozer architectures. Performance of AMD K7, K8, Intel 1595 Pentium-M, and Pentium4 based CPUs is no longer considered 1596 important for generic. 1597 * -mtune=intel can now be used to generate code running well on the 1598 most current Intel processors, which are Haswell and Silvermont for 1599 GCC 4.9. 1600 * Support to encode 32-bit assembly instructions in 16-bit format is 1601 now available through the -m16 command-line option. 1602 * Better inlining of memcpy and memset that is aware of value ranges 1603 and produces shorter alignment prologues. 1604 * -mno-accumulate-outgoing-args is now honored when unwind 1605 information is output. Argument accumulation is also now turned off 1606 for portions of programs optimized for size. 1607 * Support for new AMD family 15h processors (Excavator core) is now 1608 available through the -march=bdver4 and -mtune=bdver4 options. 1609 1610 MSP430 1611 1612 * A new command-line option -mcpu= has been added to the MSP430 1613 backend. This option is used to specify the ISA to be used. 1614 Accepted values are msp430 (the default), msp430x and msp430xv2. 1615 The ISA is no longer deduced from the -mmcu= option as there are 1616 far too many different MCU names. The -mmcu= option is still 1617 supported, and this is still used to select linker scripts and 1618 generate a C preprocessor symbol that will be recognised by the 1619 msp430.h header file. 1620 1621 NDS32 1622 1623 * A new nds32 port supports the 32-bit architecture from Andes 1624 Technology Corporation. 1625 * The port provides initial support for the V2, V3, V3m instruction 1626 set architectures. 1627 1628 Nios II 1629 1630 * A port for the Altera Nios II has been contributed by Mentor 1631 Graphics. 1632 1633 PowerPC / PowerPC64 / RS6000 1634 1635 * GCC now supports Power ISA 2.07, which includes support for 1636 Hardware Transactional Memory (HTM), Quadword atomics and several 1637 VMX and VSX additions, including Crypto, 64-bit integer, 128-bit 1638 integer and decimal integer operations. 1639 * Support for the POWER8 processor is now available through the 1640 -mcpu=power8 and -mtune=power8 options. 1641 * The libitm library has been modified to add a HTM fastpath that 1642 automatically uses POWER's HTM hardware instructions when it is 1643 executing on a HTM enabled processor. 1644 * Support for the new powerpc64le-linux platform has been added. It 1645 defaults to generating code that conforms to the ELFV2 ABI. 1646 1647 S/390, System z 1648 1649 * Support for the Transactional Execution Facility included with the 1650 IBM zEnterprise zEC12 processor has been added. A set of GCC style 1651 builtins as well as XLC style builtins are provided. The builtins 1652 are enabled by default when using the -march=zEC12 option but can 1653 explicitly be disabled with -mno-htm. Using the GCC builtins also 1654 libitm supports hardware transactions on S/390. 1655 * The hotpatch features allows to prepare functions for hotpatching. 1656 A certain amount of bytes is reserved before the function entry 1657 label plus a NOP is inserted at its very beginning to implement a 1658 backward jump when applying a patch. The feature can either be 1659 enabled per compilation unit via the command-line option -mhotpatch 1660 or per function using the hotpatch attribute. 1661 * The shrink wrap optimization is now supported on S/390 and enabled 1662 by default. 1663 * A major rework of the routines to determine which registers need to 1664 be saved and restored in function prologue/epilogue now allow to 1665 use floating point registers as save slots. This will happen for 1666 certain leaf function with -march=z10 or higher. 1667 * The LRA rtl pass replaces reload by default on S/390. 1668 1669 RX 1670 1671 * The port now allows to specify the RX100, RX200, and RX600 1672 processors with the command-line options -mcpu=rx100, -mcpu=rx200 1673 and -mcpu=rx600. 1674 1675 SH 1676 1677 * Minor improvements to code generated for integer arithmetic and 1678 code that involves the T bit. 1679 * Added support for the SH2A clips and clipu instructions. The 1680 compiler will now try to utilize them for min/max expressions such 1681 as max (-128, min (127, x)). 1682 * Added support for the cmp/str instruction through built-in 1683 functions such as __builtin_strlen. When not optimizing for size, 1684 the compiler will now expand calls to e.g. strlen as an inlined 1685 sequences which utilize the cmp/str instruction. 1686 * Improved code generated around volatile memory loads and stores. 1687 * The option -mcbranchdi has been deprecated. Specifying it will 1688 result in a warning and will not influence code generation. 1689 * The option -mcmpeqdi has been deprecated. Specifying it will result 1690 in a warning and will not influence code generation. 1691 1692 GCC 4.9.1 1693 1694 This is the [25]list of problem reports (PRs) from GCC's bug tracking 1695 system that are known to be fixed in the 4.9.1 release. This list might 1696 not be complete (that is, it is possible that some PRs that have been 1697 fixed are not listed here). 1698 1699 Version 4.0 of the OpenMP specification is supported even in Fortran, 1700 not just C and C++. 1701 1702 GCC 4.9.2 1703 1704 This is the [26]list of problem reports (PRs) from GCC's bug tracking 1705 system that are known to be fixed in the 4.9.2 release. This list might 1706 not be complete (that is, it is possible that some PRs that have been 1707 fixed are not listed here). 1708 1709 GCC 4.9.3 1710 1711 This is the [27]list of problem reports (PRs) from GCC's bug tracking 1712 system that are known to be fixed in the 4.9.3 release. This list might 1713 not be complete (that is, it is possible that some PRs that have been 1714 fixed are not listed here). 1715 1716 GCC 4.9.4 1717 1718 This is the [28]list of problem reports (PRs) from GCC's bug tracking 1719 system that are known to be fixed in the 4.9.4 release. This list might 1720 not be complete (that is, it is possible that some PRs that have been 1721 fixed are not listed here). 1722 1723 1724 For questions related to the use of GCC, please consult these web 1725 pages and the [29]GCC manuals. If that fails, the 1726 [30]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 1727 web pages and the development of GCC are welcome on our developer 1728 list at [31]gcc (a] gcc.gnu.org. All of [32]our lists have public 1729 archives. 1730 1731 Copyright (C) [33]Free Software Foundation, Inc. Verbatim copying and 1732 distribution of this entire article is permitted in any medium, 1733 provided this notice is preserved. 1734 1735 These pages are [34]maintained by the GCC team. Last modified 1736 2016-12-28[35]. 1737 1738 References 1739 1740 1. https://gcc.gnu.org/ml/gcc-patches/2013-05/msg00728.html 1741 2. https://gcc.gnu.org/PR60825 1742 3. https://gcc.gnu.org/gcc-4.9/porting_to.html 1743 4. http://www.openmp.org/specifications/ 1744 5. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gcc/Optimize-Options.html#index-fsimd-cost-model-908 1745 6. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gcc/Language-Independent-Options.html#index-fdiagnostics-color-252 1746 7. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gcc/Loop-Specific-Pragmas.html 1747 8. https://www.cilkplus.org/ 1748 9. http://gcc.gnu.org/projects/cxx1y.html 1749 10. http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2013/n3638.html 1750 11. http://gcc.gnu.org/projects/cxx1y.html 1751 12. http://gcc.gnu.org/projects/cxx1y.html 1752 13. http://gcc.gnu.org/projects/cxx1y.html 1753 14. http://gcc.gnu.org/projects/cxx1y.html 1754 15. http://gcc.gnu.org/projects/cxx1y.html 1755 16. http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2014/n3889.pdf 1756 17. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/libstdc++/manual/manual/status.html#status.iso.2011 1757 18. https://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.2014 1758 19. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gfortran/Argument-passing-conventions.html 1759 20. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gfortran/GNU-Fortran-Compiler-Directives.html 1760 21. https://gcc.gnu.org/wiki/Fortran2003Status 1761 22. https://gcc.gnu.org/wiki/Fortran2008Status 1762 23. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gfortran/Debugging-Options.html 1763 24. https://gcc.gnu.org/onlinedocs/gcc-4.9.2/gcc/Function-Multiversioning.html 1764 25. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.9.1 1765 26. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.9.2 1766 27. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.9.3 1767 28. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.9.4 1768 29. https://gcc.gnu.org/onlinedocs/ 1769 30. mailto:gcc-help (a] gcc.gnu.org 1770 31. mailto:gcc (a] gcc.gnu.org 1771 32. https://gcc.gnu.org/lists.html 1772 33. http://www.fsf.org/ 1773 34. https://gcc.gnu.org/about.html 1774 35. http://validator.w3.org/check/referer 1775 ====================================================================== 1776 http://gcc.gnu.org/gcc-4.8/index.html 1777 GCC 4.8 Release Series 1778 1779 June 23, 2015 1780 1781 The [1]GNU project and the GCC developers are pleased to announce the 1782 release of GCC 4.8.5. 1783 1784 This release is a bug-fix release, containing fixes for regressions in 1785 GCC 4.8.4 relative to previous releases of GCC. 1786 1787 This release series is no longer maintained. 1788 1789 Release History 1790 1791 GCC 4.8.5 1792 June 23, 2015 ([2]changes, [3]documentation) 1793 1794 GCC 4.8.4 1795 December 19, 2014 ([4]changes, [5]documentation) 1796 1797 GCC 4.8.3 1798 May 22, 2014 ([6]changes, [7]documentation) 1799 1800 GCC 4.8.2 1801 October 16, 2013 ([8]changes, [9]documentation) 1802 1803 GCC 4.8.1 1804 May 31, 2013 ([10]changes, [11]documentation) 1805 1806 GCC 4.8.0 1807 March 22, 2013 ([12]changes, [13]documentation) 1808 1809 References and Acknowledgements 1810 1811 GCC used to stand for the GNU C Compiler, but since the compiler 1812 supports several other languages aside from C, it now stands for the 1813 GNU Compiler Collection. 1814 1815 A list of [14]successful builds is updated as new information becomes 1816 available. 1817 1818 The GCC developers would like to thank the numerous people that have 1819 contributed new features, improvements, bug fixes, and other changes as 1820 well as test results to GCC. This [15]amazing group of volunteers is 1821 what makes GCC successful. 1822 1823 For additional information about GCC please refer to the [16]GCC 1824 project web site or contact the [17]GCC development mailing list. 1825 1826 To obtain GCC please use [18]our mirror sites or [19]our SVN server. 1827 1828 1829 For questions related to the use of GCC, please consult these web 1830 pages and the [20]GCC manuals. If that fails, the 1831 [21]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 1832 web pages and the development of GCC are welcome on our developer 1833 list at [22]gcc (a] gcc.gnu.org. All of [23]our lists have public 1834 archives. 1835 1836 Copyright (C) [24]Free Software Foundation, Inc. Verbatim copying and 1837 distribution of this entire article is permitted in any medium, 1838 provided this notice is preserved. 1839 1840 These pages are [25]maintained by the GCC team. Last modified 1841 2016-09-30[26]. 1842 1843 References 1844 1845 1. http://www.gnu.org/ 1846 2. http://gcc.gnu.org/gcc-4.8/changes.html 1847 3. https://gcc.gnu.org/onlinedocs/4.8.5/ 1848 4. http://gcc.gnu.org/gcc-4.8/changes.html 1849 5. https://gcc.gnu.org/onlinedocs/4.8.4/ 1850 6. http://gcc.gnu.org/gcc-4.8/changes.html 1851 7. https://gcc.gnu.org/onlinedocs/4.8.3/ 1852 8. http://gcc.gnu.org/gcc-4.8/changes.html 1853 9. https://gcc.gnu.org/onlinedocs/4.8.2/ 1854 10. http://gcc.gnu.org/gcc-4.8/changes.html 1855 11. https://gcc.gnu.org/onlinedocs/4.8.1/ 1856 12. http://gcc.gnu.org/gcc-4.8/changes.html 1857 13. https://gcc.gnu.org/onlinedocs/4.8.0/ 1858 14. http://gcc.gnu.org/gcc-4.8/buildstat.html 1859 15. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 1860 16. http://gcc.gnu.org/index.html 1861 17. mailto:gcc (a] gcc.gnu.org 1862 18. http://gcc.gnu.org/mirrors.html 1863 19. http://gcc.gnu.org/svn.html 1864 20. https://gcc.gnu.org/onlinedocs/ 1865 21. mailto:gcc-help (a] gcc.gnu.org 1866 22. mailto:gcc (a] gcc.gnu.org 1867 23. https://gcc.gnu.org/lists.html 1868 24. http://www.fsf.org/ 1869 25. https://gcc.gnu.org/about.html 1870 26. http://validator.w3.org/check/referer 1871 ====================================================================== 1872 http://gcc.gnu.org/gcc-4.8/changes.html 1873 GCC 4.8 Release Series 1874 Changes, New Features, and Fixes 1875 1876 Caveats 1877 1878 GCC now uses C++ as its implementation language. This means that to 1879 build GCC from sources, you will need a C++ compiler that understands 1880 C++ 2003. For more details on the rationale and specific changes, 1881 please refer to the [1]C++ conversion page. 1882 1883 To enable the Graphite framework for loop optimizations you now need 1884 CLooG version 0.18.0 and ISL version 0.11.1. Both can be obtained from 1885 the [2]GCC infrastructure directory. The installation manual contains 1886 more information about requirements to build GCC. 1887 1888 GCC now uses a more aggressive analysis to derive an upper bound for 1889 the number of iterations of loops using constraints imposed by language 1890 standards. This may cause non-conforming programs to no longer work as 1891 expected, such as SPEC CPU 2006 464.h264ref and 416.gamess. A new 1892 option, -fno-aggressive-loop-optimizations, was added to disable this 1893 aggressive analysis. In some loops that have known constant number of 1894 iterations, but undefined behavior is known to occur in the loop before 1895 reaching or during the last iteration, GCC will warn about the 1896 undefined behavior in the loop instead of deriving lower upper bound of 1897 the number of iterations for the loop. The warning can be disabled with 1898 -Wno-aggressive-loop-optimizations. 1899 1900 On ARM, a bug has been fixed in GCC's implementation of the AAPCS rules 1901 for the layout of vectors that could lead to wrong code being 1902 generated. Vectors larger than 8 bytes in size are now by default 1903 aligned to an 8-byte boundary. This is an ABI change: code that makes 1904 explicit use of vector types may be incompatible with binary objects 1905 built with older versions of GCC. Auto-vectorized code is not affected 1906 by this change. 1907 1908 On AVR, support has been removed for the command-line option 1909 -mshort-calls deprecated in GCC 4.7. 1910 1911 On AVR, the configure option --with-avrlibc supported since GCC 4.7.2 1912 is turned on per default for all non-RTEMS configurations. This option 1913 arranges for a better integration of [3]AVR Libc with avr-gcc. For 1914 technical details, see [4]PR54461. To turn off the option in non-RTEMS 1915 configurations, use --with-avrlibc=no. If the compiler is configured 1916 for RTEMS, the option is always turned off. 1917 1918 More information on porting to GCC 4.8 from previous versions of GCC 1919 can be found in the [5]porting guide for this release. 1920 1921 General Optimizer Improvements (and Changes) 1922 1923 * DWARF4 is now the default when generating DWARF debug information. 1924 When -g is used on a platform that uses DWARF debugging 1925 information, GCC will now default to -gdwarf-4 1926 -fno-debug-types-section. 1927 GDB 7.5, Valgrind 3.8.0 and elfutils 0.154 debug information 1928 consumers support DWARF4 by default. Before GCC 4.8 the default 1929 version used was DWARF2. To make GCC 4.8 generate an older DWARF 1930 version use -g together with -gdwarf-2 or -gdwarf-3. The default 1931 for Darwin and VxWorks is still -gdwarf-2 -gstrict-dwarf. 1932 * A new general optimization level, -Og, has been introduced. It 1933 addresses the need for fast compilation and a superior debugging 1934 experience while providing a reasonable level of run-time 1935 performance. Overall experience for development should be better 1936 than the default optimization level -O0. 1937 * A new option -ftree-partial-pre was added to control the partial 1938 redundancy elimination (PRE) optimization. This option is enabled 1939 by default at the -O3 optimization level, and it makes PRE more 1940 aggressive. 1941 * The option -fconserve-space has been removed; it was no longer 1942 useful on most targets since GCC supports putting variables into 1943 BSS without making them common. 1944 * The struct reorg and matrix reorg optimizations (command-line 1945 options -fipa-struct-reorg and -fipa-matrix-reorg) have been 1946 removed. They did not always work correctly, nor did they work with 1947 link-time optimization (LTO), hence were only applicable to 1948 programs consisting of a single translation unit. 1949 * Several scalability bottle-necks have been removed from GCC's 1950 optimization passes. Compilation of extremely large functions, e.g. 1951 due to the use of the flatten attribute in the "Eigen" C++ linear 1952 algebra templates library, is significantly faster than previous 1953 releases of GCC. 1954 * Link-time optimization (LTO) improvements: 1955 + LTO partitioning has been rewritten for better reliability and 1956 maintanibility. Several important bugs leading to link 1957 failures have been fixed. 1958 * Interprocedural optimization improvements: 1959 + A new symbol table has been implemented. It builds on existing 1960 callgraph and varpool modules and provide a new API. Unusual 1961 symbol visibilities and aliases are handled more consistently 1962 leading to, for example, more aggressive unreachable code 1963 removal with LTO. 1964 + The inline heuristic can now bypass limits on the size of of 1965 inlined functions when the inlining is particularly 1966 profitable. This happens, for example, when loop bounds or 1967 array strides get propagated. 1968 + Values passed through aggregates (either by value or 1969 reference) are now propagated at the inter-procedural level 1970 leading to better inlining decisions (for example in the case 1971 of Fortran array descriptors) and devirtualization. 1972 * [6]AddressSanitizer , a fast memory error detector, has been added 1973 and can be enabled via -fsanitize=address. Memory access 1974 instructions will be instrumented to detect heap-, stack-, and 1975 global-buffer overflow as well as use-after-free bugs. To get nicer 1976 stacktraces, use -fno-omit-frame-pointer. The AddressSanitizer is 1977 available on IA-32/x86-64/x32/PowerPC/PowerPC64 GNU/Linux and on 1978 x86-64 Darwin. 1979 * [7]ThreadSanitizer has been added and can be enabled via 1980 -fsanitize=thread. Instructions will be instrumented to detect data 1981 races. The ThreadSanitizer is available on x86-64 GNU/Linux. 1982 * A new local register allocator (LRA) has been implemented, which 1983 replaces the 26 year old reload pass and improves generated code 1984 quality. For now it is active on the IA-32 and x86-64 targets. 1985 * Support for transactional memory has been implemented on the 1986 following architectures: IA-32/x86-64, ARM, PowerPC, SH, SPARC, and 1987 Alpha. 1988 1989 New Languages and Language specific improvements 1990 1991 C family 1992 1993 * Each diagnostic emitted now includes the original source line and a 1994 caret '^' indicating the column. The option 1995 -fno-diagnostics-show-caret suppresses this information. 1996 * The option -ftrack-macro-expansion=2 is now enabled by default. 1997 This allows the compiler to display the macro expansion stack in 1998 diagnostics. Combined with the caret information, an example 1999 diagnostic showing these two features is: 2000 2001 t.c:1:94: error: invalid operands to binary < (have struct mystruct and float 2002 ) 2003 #define MYMAX(A,B) __extension__ ({ __typeof__(A) __a = (A); __typeof__(B) _ 2004 _b = (B); __a < __b ? __b : __a; }) 2005 2006 ^ 2007 t.c:7:7: note: in expansion of macro 'MYMAX' 2008 X = MYMAX(P, F); 2009 ^ 2010 2011 * A new -Wsizeof-pointer-memaccess warning has been added (also 2012 enabled by -Wall) to warn about suspicious length parameters to 2013 certain string and memory built-in functions if the argument uses 2014 sizeof. This warning warns e.g. about memset (ptr, 0, sizeof 2015 (ptr)); if ptr is not an array, but a pointer, and suggests a 2016 possible fix, or about memcpy (&foo, ptr, sizeof (&foo));. 2017 * The new option -Wpedantic is an alias for -pedantic, which is now 2018 deprecated. The forms -Wno-pedantic, -Werror=pedantic, and 2019 -Wno-error=pedantic work in the same way as for any other -W 2020 option. One caveat is that -Werror=pedantic is not equivalent to 2021 -pedantic-errors, since the latter makes into errors some warnings 2022 that are not controlled by -Wpedantic, and the former only affects 2023 diagnostics that are disabled when using -Wno-pedantic. 2024 * The option -Wshadow no longer warns if a declaration shadows a 2025 function declaration, unless the former declares a function or 2026 pointer to function, because this is [8]a common and valid case in 2027 real-world code. 2028 2029 C++ 2030 2031 * G++ now implements the [9]C++11 thread_local keyword; this differs 2032 from the GNU __thread keyword primarily in that it allows dynamic 2033 initialization and destruction semantics. Unfortunately, this 2034 support requires a run-time penalty for references to 2035 non-function-local thread_local variables defined in a different 2036 translation unit even if they don't need dynamic initialization, so 2037 users may want to continue to use __thread for TLS variables with 2038 static initialization semantics. 2039 If the programmer can be sure that no use of the variable in a 2040 non-defining TU needs to trigger dynamic initialization (either 2041 because the variable is statically initialized, or a use of the 2042 variable in the defining TU will be executed before any uses in 2043 another TU), they can avoid this overhead with the 2044 -fno-extern-tls-init option. 2045 OpenMP threadprivate variables now also support dynamic 2046 initialization and destruction by the same mechanism. 2047 * G++ now implements the [10]C++11 attribute syntax, e.g. 2048 2049 [[noreturn]] void f(); 2050 2051 and also the alignment specifier, e.g. 2052 2053 alignas(double) int i; 2054 2055 * G++ now implements [11]C++11 inheriting constructors, e.g. 2056 2057 struct A { A(int); }; 2058 struct B: A { using A::A; }; // defines B::B(int) 2059 B b(42); // OK 2060 2061 * As of GCC 4.8.1, G++ implements the change to decltype semantics 2062 from [12]N3276. 2063 2064 struct A f(); 2065 decltype(f()) g(); // OK, return type of f() is not required to be complete. 2066 2067 * As of GCC 4.8.1, G++ implements [13]C++11 ref-qualifiers, e.g. 2068 2069 struct A { int f() &; }; 2070 int i = A().f(); // error, f() requires an lvalue object 2071 2072 * G++ now supports a -std=c++1y option for experimentation with 2073 features proposed for the next revision of the standard, expected 2074 around 2014. Currently the only difference from -std=c++11 is 2075 support for return type deduction in normal functions, as proposed 2076 in [14]N3386. Status of C++1y features in GCC 4.8 can be found 2077 [15]here. 2078 * The G++ namespace association extension, __attribute ((strong)), 2079 has been deprecated. Inline namespaces should be used instead. 2080 * G++ now supports a -fext-numeric-literal option to control whether 2081 GNU numeric literal suffixes are accepted as extensions or 2082 processed as C++11 user-defined numeric literal suffixes. The flag 2083 is on (use suffixes for GNU literals) by default for -std=gnu++*, 2084 and -std=c++98. The flag is off (use suffixes for user-defined 2085 literals) by default for -std=c++11 and later. 2086 2087 Runtime Library (libstdc++) 2088 2089 * [16]Improved experimental support for the new ISO C++ standard, 2090 C++11, including: 2091 + forward_list meets the allocator-aware container requirements; 2092 + this_thread::sleep_for(), this_thread::sleep_until() and 2093 this_thread::yield() are defined without requiring the 2094 configure option --enable-libstdcxx-time; 2095 * Improvements to <random>: 2096 + SSE optimized normal_distribution. 2097 + Use of hardware RNG instruction for random_device on new x86 2098 processors (requires the assembler to support the 2099 instruction.) 2100 and <ext/random>: 2101 + New random number engine simd_fast_mersenne_twister_engine 2102 with an optimized SSE implementation. 2103 + New random number distributions beta_distribution, 2104 normal_mv_distribution, rice_distribution, 2105 nakagami_distribution, pareto_distribution, k_distribution, 2106 arcsine_distribution, hoyt_distribution. 2107 * Added --disable-libstdcxx-verbose configure option to disable 2108 diagnostic messages issued when a process terminates abnormally. 2109 This may be useful for embedded systems to reduce the size of 2110 executables that link statically to the library. 2111 2112 Fortran 2113 2114 * Compatibility notice: 2115 + Module files: The version of module files (.mod) has been 2116 incremented. Fortran MODULEs compiled by earlier GCC versions 2117 have to be recompiled, when they are USEd by files compiled 2118 with GCC 4.8. GCC 4.8 is not able to read .mod files created 2119 by earlier versions; attempting to do so gives an error 2120 message. 2121 Note: The ABI of the produced assembler data itself has not 2122 changed; object files and libraries are fully compatible with 2123 older versions except as noted below. 2124 + ABI: Some internal names (used in the assembler/object file) 2125 have changed for symbols declared in the specification part of 2126 a module. If an affected module or a file using it via use 2127 association is recompiled, the module and all files which 2128 directly use such symbols have to be recompiled as well. This 2129 change only affects the following kind of module symbols: 2130 o Procedure pointers. Note: C-interoperable function 2131 pointers (type(c_funptr)) are not affected nor are 2132 procedure-pointer components. 2133 o Deferred-length character strings. 2134 * The [17]BACKTRACE intrinsic subroutine has been added. It shows a 2135 backtrace at an arbitrary place in user code; program execution 2136 continues normally afterwards. 2137 * The [18]-Wc-binding-type warning option has been added (disabled by 2138 default). It warns if the a variable might not be C interoperable; 2139 in particular, if the variable has been declared using an intrinsic 2140 type with default kind instead of using a kind parameter defined 2141 for C interoperability in the intrinsic ISO_C_Binding module. 2142 Before, this warning was always printed. The -Wc-binding-type 2143 option is enabled by -Wall. 2144 * The [19]-Wrealloc-lhs and -Wrealloc-lhs-all warning command-line 2145 options have been added, which diagnose when code is inserted for 2146 automatic (re)allocation of a variable during assignment. This 2147 option can be used to decide whether it is safe to use 2148 [20]-fno-realloc-lhs. Additionally, it can be used to find 2149 automatic (re)allocation in hot loops. (For arrays, replacing 2150 "var=" by "var(:)=" disables the automatic reallocation.) 2151 * The [21]-Wcompare-reals command-line option has been added. When 2152 this is set, warnings are issued when comparing REAL or COMPLEX 2153 types for equality and inequality; consider replacing a == b by 2154 abs(ab) < eps with a suitable eps. -Wcompare-reals is enabled by 2155 -Wextra. 2156 * The [22]-Wtarget-lifetime command-line option has been added 2157 (enabled with -Wall), which warns if the pointer in a pointer 2158 assignment might outlive its target. 2159 * Reading floating point numbers which use "q" for the exponential 2160 (such as 4.0q0) is now supported as vendor extension for better 2161 compatibility with old data files. It is strongly recommended to 2162 use for I/O the equivalent but standard conforming "e" (such as 2163 4.0e0). 2164 (For Fortran source code, consider replacing the "q" in 2165 floating-point literals by a kind parameter (e.g. 4.0e0_qp with a 2166 suitable qp). Note that in Fortran source code replacing "q" by 2167 a simple "e" is not equivalent.) 2168 * The GFORTRAN_TMPDIR environment variable for specifying a 2169 non-default directory for files opened with STATUS="SCRATCH", is 2170 not used anymore. Instead gfortran checks the POSIX/GNU standard 2171 TMPDIR environment variable. If TMPDIR is not defined, gfortran 2172 falls back to other methods to determine the directory for 2173 temporary files as documented in the [23]user manual. 2174 * [24]Fortran 2003: 2175 + Support for unlimited polymorphic variables (CLASS(*)) has 2176 been added. Nonconstant character lengths are not yet 2177 supported. 2178 * [25]TS 29113: 2179 + Assumed types (TYPE(*)) are now supported. 2180 + Experimental support for assumed-rank arrays (dimension(..)) 2181 has been added. Note that currently gfortran's own array 2182 descriptor is used, which is different from the one defined in 2183 TS29113, see [26]gfortran's header file or use the [27]Chasm 2184 Language Interoperability Tools. 2185 2186 Go 2187 2188 * GCC 4.8.2 provides a complete implementation of the Go 1.1.2 2189 release. 2190 * GCC 4.8.0 and 4.8.1 implement a preliminary version of the Go 1.1 2191 release. The library support is not quite complete. 2192 * Go has been tested on GNU/Linux and Solaris platforms for various 2193 processors including x86, x86_64, PowerPC, SPARC, and Alpha. It may 2194 work on other platforms as well. 2195 2196 New Targets and Target Specific Improvements 2197 2198 AArch64 2199 2200 * A new port has been added to support AArch64, the new 64-bit 2201 architecture from ARM. Note that this is a separate port from the 2202 existing 32-bit ARM port. 2203 * The port provides initial support for the Cortex-A53 and the 2204 Cortex-A57 processors with the command line options 2205 -mcpu=cortex-a53 and -mcpu=cortex-a57. 2206 * As of GCC 4.8.4 a workaround for the ARM Cortex-A53 erratum 835769 2207 has been added and can be enabled by giving the 2208 -mfix-cortex-a53-835769 option. Alternatively it can be enabled by 2209 default by configuring GCC with the --enable-fix-cortex-a53-835769 2210 option. 2211 2212 ARM 2213 2214 * Initial support has been added for the AArch32 extensions defined 2215 in the ARMv8 architecture. 2216 * Code generation improvements for the Cortex-A7 and Cortex-A15 CPUs. 2217 * A new option, -mcpu=marvell-pj4, has been added to generate code 2218 for the Marvell PJ4 processor. 2219 * The compiler can now automatically generate the VFMA, VFMS, REVSH 2220 and REV16 instructions. 2221 * A new vectorizer cost model for Advanced SIMD configurations to 2222 improve the auto-vectorization strategies used. 2223 * The scheduler now takes into account the number of live registers 2224 to reduce the amount of spilling that can occur. This should 2225 improve code performance in large functions. The limit can be 2226 removed by using the option -fno-sched-pressure. 2227 * Improvements have been made to the Marvell iWMMX code generation 2228 and support for the iWMMX2 SIMD unit has been added. The option 2229 -mcpu=iwmmxt2 can be used to enable code generation for the latter. 2230 * A number of code generation improvements for Thumb2 to reduce code 2231 size when compiling for the M-profile processors. 2232 * The RTEMS (arm-rtems) port has been updated to use the EABI. 2233 * Code generation support for the old FPA and Maverick floating-point 2234 architectures has been removed. Ports that previously relied on 2235 these features have also been removed. This includes the targets: 2236 + arm*-*-linux-gnu (use arm*-*-linux-gnueabi) 2237 + arm*-*-elf (use arm*-*-eabi) 2238 + arm*-*-uclinux* (use arm*-*-uclinux*eabi) 2239 + arm*-*-ecos-elf (no alternative) 2240 + arm*-*-freebsd (no alternative) 2241 + arm*-wince-pe* (no alternative). 2242 2243 AVR 2244 2245 * Support for the "Embedded C" fixed-point has been added. For 2246 details, see the [28]GCC wiki and the [29]user manual. The support 2247 is not complete. 2248 * A new print modifier %r for register operands in inline assembler 2249 is supported. It will print the raw register number without the 2250 register prefix 'r': 2251 /* Return the most significant byte of 'val', a 64-bit value. */ 2252 2253 unsigned char msb (long long val) 2254 { 2255 unsigned char c; 2256 __asm__ ("mov %0, %r1+7" : "=r" (c) : "r" (val)); 2257 return c; 2258 } 2259 The inline assembler in this example will generate code like 2260 mov r24, 8+7 2261 provided c is allocated to R24 and val is allocated to R8R15. This 2262 works because the GNU assembler accepts plain register numbers 2263 without register prefix. 2264 * Static initializers with 3-byte symbols are supported now: 2265 extern const __memx char foo; 2266 const __memx void *pfoo = &foo; 2267 This requires at least Binutils 2.23. 2268 2269 IA-32/x86-64 2270 2271 * Allow -mpreferred-stack-boundary=3 for the x86-64 architecture with 2272 SSE extensions disabled. Since the x86-64 ABI requires 16 byte 2273 stack alignment, this is ABI incompatible and intended to be used 2274 in controlled environments where stack space is an important 2275 limitation. This option will lead to wrong code when functions 2276 compiled with 16 byte stack alignment (such as functions from a 2277 standard library) are called with misaligned stack. In this case, 2278 SSE instructions may lead to misaligned memory access traps. In 2279 addition, variable arguments will be handled incorrectly for 16 2280 byte aligned objects (including x87 long double and __int128), 2281 leading to wrong results. You must build all modules with 2282 -mpreferred-stack-boundary=3, including any libraries. This 2283 includes the system libraries and startup modules. 2284 * Support for the new Intel processor codename Broadwell with RDSEED, 2285 ADCX, ADOX, PREFETCHW is available through -madx, -mprfchw, 2286 -mrdseed command-line options. 2287 * Support for the Intel RTM and HLE intrinsics, built-in functions 2288 and code generation is available via -mrtm and -mhle. 2289 * Support for the Intel FXSR, XSAVE and XSAVEOPT instruction sets. 2290 Intrinsics and built-in functions are available via -mfxsr, -mxsave 2291 and -mxsaveopt respectively. 2292 * New -maddress-mode=[short|long] options for x32. 2293 -maddress-mode=short overrides default 64-bit addresses to 32-bit 2294 by emitting the 0x67 address-size override prefix. This is the 2295 default address mode for x32. 2296 * New built-in functions to detect run-time CPU type and ISA: 2297 + A built-in function __builtin_cpu_is has been added to detect 2298 if the run-time CPU is of a particular type. It returns a 2299 positive integer on a match and zero otherwise. It accepts one 2300 string literal argument, the CPU name. For example, 2301 __builtin_cpu_is("westmere") returns a positive integer if the 2302 run-time CPU is an Intel Core i7 Westmere processor. Please 2303 refer to the [30]user manual for the list of valid CPU names 2304 recognized. 2305 + A built-in function __builtin_cpu_supports has been added to 2306 detect if the run-time CPU supports a particular ISA feature. 2307 It returns a positive integer on a match and zero otherwise. 2308 It accepts one string literal argument, the ISA feature. For 2309 example, __builtin_cpu_supports("ssse3") returns a positive 2310 integer if the run-time CPU supports SSSE3 instructions. 2311 Please refer to the [31]user manual for the list of valid ISA 2312 names recognized. 2313 Caveat: If these built-in functions are called before any static 2314 constructors are invoked, like during IFUNC initialization, then 2315 the CPU detection initialization must be explicitly run using this 2316 newly provided built-in function, __builtin_cpu_init. The 2317 initialization needs to be done only once. For example, this is how 2318 the invocation would look like inside an IFUNC initializer: 2319 static void (*some_ifunc_resolver(void))(void) 2320 { 2321 __builtin_cpu_init(); 2322 if (__builtin_cpu_is("amdfam10h") ... 2323 if (__builtin_cpu_supports("popcnt") ... 2324 } 2325 2326 * Function Multiversioning Support with G++: 2327 It is now possible to create multiple function versions each 2328 targeting a specific processor and/or ISA. Function versions have 2329 the same signature but different target attributes. For example, 2330 here is a program with function versions: 2331 __attribute__ ((target ("default"))) 2332 int foo(void) 2333 { 2334 return 1; 2335 } 2336 2337 __attribute__ ((target ("sse4.2"))) 2338 int foo(void) 2339 { 2340 return 2; 2341 } 2342 2343 int main (void) 2344 { 2345 int (*p) = &foo; 2346 assert ((*p)() == foo()); 2347 return 0; 2348 } 2349 2350 Please refer to this [32]wiki for more information. 2351 * The x86 back end has been improved to allow option -fschedule-insns 2352 to work reliably. This option can be used to schedule instructions 2353 better and leads to improved performace in certain cases. 2354 * Windows MinGW-w64 targets (*-w64-mingw*) require at least r5437 2355 from the Mingw-w64 trunk. 2356 * Support for new AMD family 15h processors (Steamroller core) is now 2357 available through the -march=bdver3 and -mtune=bdver3 options. 2358 * Support for new AMD family 16h processors (Jaguar core) is now 2359 available through the -march=btver2 and -mtune=btver2 options. 2360 2361 FRV 2362 2363 * This target now supports the -fstack-usage command-line option. 2364 2365 MIPS 2366 2367 * GCC can now generate code specifically for the R4700, Broadcom XLP 2368 and MIPS 34kn processors. The associated -march options are 2369 -march=r4700, -march=xlp and -march=34kn respectively. 2370 * GCC now generates better DSP code for MIPS 74k cores thanks to 2371 further scheduling optimizations. 2372 * The MIPS port now supports the -fstack-check option. 2373 * GCC now passes the -mmcu and -mno-mcu options to the assembler. 2374 * Previous versions of GCC would silently accept -fpic and -fPIC for 2375 -mno-abicalls targets like mips*-elf. This combination was not 2376 intended or supported, and did not generate position-independent 2377 code. GCC 4.8 now reports an error when this combination is used. 2378 2379 PowerPC / PowerPC64 / RS6000 2380 2381 * SVR4 configurations (GNU/Linux, FreeBSD, NetBSD) no longer save, 2382 restore or update the VRSAVE register by default. The respective 2383 operating systems manage the VRSAVE register directly. 2384 * Large TOC support has been added for AIX through the command line 2385 option -mcmodel=large. 2386 * Native Thread-Local Storage support has been added for AIX. 2387 * VMX (Altivec) and VSX instruction sets now are enabled implicitly 2388 when targetting processors that support those hardware features on 2389 AIX 6.1 and above. 2390 2391 RX 2392 2393 * This target will now issue a warning message whenever multiple fast 2394 interrupt handlers are found in the same compilation unit. This 2395 feature can be turned off by the new 2396 -mno-warn-multiple-fast-interrupts command-line option. 2397 2398 S/390, System z 2399 2400 * Support for the IBM zEnterprise zEC12 processor has been added. 2401 When using the -march=zEC12 option, the compiler will generate code 2402 making use of the following new instructions: 2403 + load and trap instructions 2404 + 2 new compare and trap instructions 2405 + rotate and insert selected bits - without CC clobber 2406 The -mtune=zEC12 option enables zEC12 specific instruction 2407 scheduling without making use of new instructions. 2408 * Register pressure sensitive instruction scheduling is enabled by 2409 default. 2410 * The ifunc function attribute is enabled by default. 2411 * memcpy and memcmp invokations on big memory chunks or with run time 2412 lengths are not generated inline anymore when tuning for z10 or 2413 higher. The purpose is to make use of the IFUNC optimized versions 2414 in Glibc. 2415 2416 SH 2417 2418 * The default alignment settings have been reduced to be less 2419 aggressive. This results in more compact code for optimization 2420 levels other than -Os. 2421 * Improved support for the __atomic built-in functions: 2422 + A new option -matomic-model=model selects the model for the 2423 generated atomic sequences. The following models are 2424 supported: 2425 2426 soft-gusa 2427 Software gUSA sequences (SH3* and SH4* only). On 2428 SH4A targets this will now also partially utilize 2429 the movco.l and movli.l instructions. This is the 2430 default when the target is sh3*-*-linux* or 2431 sh4*-*-linux*. 2432 2433 hard-llcs 2434 Hardware movco.l / movli.l sequences (SH4A only). 2435 2436 soft-tcb 2437 Software thread control block sequences. 2438 2439 soft-imask 2440 Software interrupt flipping sequences (privileged 2441 mode only). This is the default when the target is 2442 sh1*-*-linux* or sh2*-*-linux*. 2443 2444 none 2445 Generates function calls to the respective __atomic 2446 built-in functions. This is the default for SH64 2447 targets or when the target is not sh*-*-linux*. 2448 2449 + The option -msoft-atomic has been deprecated. It is now an 2450 alias for -matomic-model=soft-gusa. 2451 + A new option -mtas makes the compiler generate the tas.b 2452 instruction for the __atomic_test_and_set built-in function 2453 regardless of the selected atomic model. 2454 + The __sync functions in libgcc now reflect the selected atomic 2455 model when building the toolchain. 2456 * Added support for the mov.b and mov.w instructions with 2457 displacement addressing. 2458 * Added support for the SH2A instructions movu.b and movu.w. 2459 * Various improvements to code generated for integer arithmetic. 2460 * Improvements to conditional branches and code that involves the T 2461 bit. A new option -mzdcbranch tells the compiler to favor 2462 zero-displacement branches. This is enabled by default for SH4* 2463 targets. 2464 * The pref instruction will now be emitted by the __builtin_prefetch 2465 built-in function for SH3* targets. 2466 * The fmac instruction will now be emitted by the fmaf standard 2467 function and the __builtin_fmaf built-in function. 2468 * The -mfused-madd option has been deprecated in favor of the 2469 machine-independent -ffp-contract option. Notice that the fmac 2470 instruction will now be generated by default for expressions like a 2471 * b + c. This is due to the compiler default setting 2472 -ffp-contract=fast. 2473 * Added new options -mfsrra and -mfsca to allow the compiler using 2474 the fsrra and fsca instructions on targets other than SH4A (where 2475 they are already enabled by default). 2476 * Added support for the __builtin_bswap32 built-in function. It is 2477 now expanded as a sequence of swap.b and swap.w instructions 2478 instead of a library function call. 2479 * The behavior of the -mieee option has been fixed and the negative 2480 form -mno-ieee has been added to control the IEEE conformance of 2481 floating point comparisons. By default -mieee is now enabled and 2482 the option -ffinite-math-only implicitly sets -mno-ieee. 2483 * Added support for the built-in functions __builtin_thread_pointer 2484 and __builtin_set_thread_pointer. This assumes that GBR is used to 2485 hold the thread pointer of the current thread. Memory loads and 2486 stores relative to the address returned by __builtin_thread_pointer 2487 will now also utilize GBR based displacement address modes. 2488 * The -mdiv= option for targets other than SHmedia has been fixed and 2489 documented. 2490 2491 SPARC 2492 2493 * Added optimized instruction scheduling for Niagara4. 2494 2495 TILE-Gx 2496 2497 * Added support for the -mcmodel=MODEL command-line option. The 2498 models supported are small and large. 2499 2500 V850 2501 2502 * This target now supports the E3V5 architecture via the use of the 2503 new -mv850e3v5 command-line option. It also has experimental 2504 support for the e3v5 LOOP instruction which can be enabled via the 2505 new -mloop command-line option. 2506 2507 XStormy16 2508 2509 * This target now supports the -fstack-usage command-line option. 2510 2511 Operating Systems 2512 2513 OpenBSD 2514 2515 * Support for OpenBSD/amd64 (x86_64-*-openbsd*) has been added and 2516 support for OpenBSD/i386 (i386-*-openbsd*) has been rejuvenated. 2517 2518 Windows (Cygwin) 2519 2520 * Executables are now linked against shared libgcc by default. The 2521 previous default was to link statically, which can still be done by 2522 explicitly specifying -static or static-libgcc on the command line. 2523 However it is strongly advised against, as it will cause problems 2524 for any application that makes use of DLLs compiled by GCC. It 2525 should be alright for a monolithic stand-alone application that 2526 only links against the Windows DLLs, but offers little or no 2527 benefit. 2528 2529 GCC 4.8.1 2530 2531 This is the [33]list of problem reports (PRs) from GCC's bug tracking 2532 system that are known to be fixed in the 4.8.1 release. This list might 2533 not be complete (that is, it is possible that some PRs that have been 2534 fixed are not listed here). 2535 2536 The C++11 <chrono> std::chrono::system_clock and 2537 std::chrono::steady_clock classes have changed ABI in GCC 4.8.1, they 2538 both are now separate (never typedefs of each other), both use 2539 std::chrono::nanoseconds resolution, on most GNU/Linux configurations 2540 std::chrono::steady_clock is now finally monotonic, and both classes 2541 are mangled differently than in the previous GCC releases. 2542 std::chrono::system_clock::now() with std::chrono::microseconds resp. 2543 std::chrono::seconds resolution is still exported for backwards 2544 compatibility with default configured libstdc++. Note that libstdc++ 2545 configured with --enable-libstdcxx-time= used to be ABI incompatible 2546 with default configured libstdc++ for those two classes and no ABI 2547 compatibility can be offered for those configurations, so any C++11 2548 code that uses those classes and has been compiled and linked against 2549 libstdc++ configured with the non-default --enable-libstdcxx-time= 2550 configuration option needs to be recompiled. 2551 2552 GCC 4.8.2 2553 2554 This is the [34]list of problem reports (PRs) from GCC's bug tracking 2555 system that are known to be fixed in the 4.8.2 release. This list might 2556 not be complete (that is, it is possible that some PRs that have been 2557 fixed are not listed here). 2558 2559 GCC 4.8.3 2560 2561 This is the [35]list of problem reports (PRs) from GCC's bug tracking 2562 system that are known to be fixed in the 4.8.3 release. This list might 2563 not be complete (that is, it is possible that some PRs that have been 2564 fixed are not listed here). 2565 2566 Support for the new powerpc64le-linux platform has been added. It 2567 defaults to generating code that conforms to the ELFV2 ABI. 2568 2569 GCC 4.8.4 2570 2571 This is the [36]list of problem reports (PRs) from GCC's bug tracking 2572 system that are known to be fixed in the 4.8.4 release. This list might 2573 not be complete (that is, it is possible that some PRs that have been 2574 fixed are not listed here). 2575 2576 GCC 4.8.5 2577 2578 This is the [37]list of problem reports (PRs) from GCC's bug tracking 2579 system that are known to be fixed in the 4.8.5 release. This list might 2580 not be complete (that is, it is possible that some PRs that have been 2581 fixed are not listed here). 2582 2583 2584 For questions related to the use of GCC, please consult these web 2585 pages and the [38]GCC manuals. If that fails, the 2586 [39]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 2587 web pages and the development of GCC are welcome on our developer 2588 list at [40]gcc (a] gcc.gnu.org. All of [41]our lists have public 2589 archives. 2590 2591 Copyright (C) [42]Free Software Foundation, Inc. Verbatim copying and 2592 distribution of this entire article is permitted in any medium, 2593 provided this notice is preserved. 2594 2595 These pages are [43]maintained by the GCC team. Last modified 2596 2017-02-27[44]. 2597 2598 References 2599 2600 1. https://gcc.gnu.org/wiki/cxx-conversion 2601 2. ftp://gcc.gnu.org/pub/gcc/infrastructure/ 2602 3. http://www.nongnu.org/avr-libc/ 2603 4. https://gcc.gnu.org/PR54461 2604 5. https://gcc.gnu.org/gcc-4.8/porting_to.html 2605 6. https://github.com/google/sanitizers 2606 7. https://code.google.com/archive/p/data-race-test/wikis/ThreadSanitizer.wiki 2607 8. https://lkml.org/lkml/2006/11/28/239 2608 9. http://gcc.gnu.org/gcc-4.8/cxx0x_status.html 2609 10. http://gcc.gnu.org/gcc-4.8/cxx0x_status.html 2610 11. http://gcc.gnu.org/gcc-4.8/cxx0x_status.html 2611 12. http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3276.pdf 2612 13. http://gcc.gnu.org/gcc-4.8/cxx0x_status.html 2613 14. http://www.open-std.org/JTC1/SC22/WG21/docs/papers/2012/n3386.html 2614 15. http://gcc.gnu.org/projects/cxx1y.html 2615 16. https://gcc.gnu.org/onlinedocs/gcc-4.8.4/libstdc++/manual/manual/status.html#status.iso.2011 2616 17. https://gcc.gnu.org/onlinedocs/gfortran/BACKTRACE.html 2617 18. https://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html 2618 19. https://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html 2619 20. https://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html 2620 21. https://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html 2621 22. https://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html 2622 23. https://gcc.gnu.org/onlinedocs/gfortran/TMPDIR.html 2623 24. https://gcc.gnu.org/wiki/Fortran2003Status 2624 25. https://gcc.gnu.org/wiki/TS29113Status 2625 26. https://gcc.gnu.org/viewcvs/trunk/libgfortran/libgfortran.h?content-type=text/plain&view=co 2626 27. http://chasm-interop.sourceforge.net/ 2627 28. https://gcc.gnu.org/wiki/avr-gcc#Fixed-Point_Support 2628 29. https://gcc.gnu.org/onlinedocs/gcc/Fixed-Point.html 2629 30. https://gcc.gnu.org/onlinedocs/gcc/x86-Built-in-Functions.html 2630 31. https://gcc.gnu.org/onlinedocs/gcc/x86-Built-in-Functions.html 2631 32. https://gcc.gnu.org/wiki/FunctionMultiVersioning 2632 33. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.8.1 2633 34. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.8.2 2634 35. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.8.3 2635 36. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.8.4 2636 37. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.8.5 2637 38. https://gcc.gnu.org/onlinedocs/ 2638 39. mailto:gcc-help (a] gcc.gnu.org 2639 40. mailto:gcc (a] gcc.gnu.org 2640 41. https://gcc.gnu.org/lists.html 2641 42. http://www.fsf.org/ 2642 43. https://gcc.gnu.org/about.html 2643 44. http://validator.w3.org/check/referer 2644 ====================================================================== 2645 http://gcc.gnu.org/gcc-4.7/index.html 2646 GCC 4.7 Release Series 2647 2648 June 12, 2014 2649 2650 The [1]GNU project and the GCC developers are pleased to announce the 2651 release of GCC 4.7.4. 2652 2653 This release is a bug-fix release, containing fixes for regressions in 2654 GCC 4.7.3 relative to previous releases of GCC. 2655 2656 This release series is no longer maintained. 2657 2658 Release History 2659 2660 GCC 4.7.4 2661 June 12, 2014 ([2]changes, [3]documentation) 2662 2663 GCC 4.7.3 2664 April 11, 2013 ([4]changes, [5]documentation) 2665 2666 GCC 4.7.2 2667 September 20, 2012 ([6]changes, [7]documentation) 2668 2669 GCC 4.7.1 2670 June 14, 2012 ([8]changes, [9]documentation) 2671 2672 GCC 4.7.0 2673 March 22, 2012 ([10]changes, [11]documentation) 2674 2675 References and Acknowledgements 2676 2677 GCC used to stand for the GNU C Compiler, but since the compiler 2678 supports several other languages aside from C, it now stands for the 2679 GNU Compiler Collection. 2680 2681 A list of [12]successful builds is updated as new information becomes 2682 available. 2683 2684 The GCC developers would like to thank the numerous people that have 2685 contributed new features, improvements, bug fixes, and other changes as 2686 well as test results to GCC. This [13]amazing group of volunteers is 2687 what makes GCC successful. 2688 2689 For additional information about GCC please refer to the [14]GCC 2690 project web site or contact the [15]GCC development mailing list. 2691 2692 To obtain GCC please use [16]our mirror sites or [17]our SVN server. 2693 2694 2695 For questions related to the use of GCC, please consult these web 2696 pages and the [18]GCC manuals. If that fails, the 2697 [19]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 2698 web pages and the development of GCC are welcome on our developer 2699 list at [20]gcc (a] gcc.gnu.org. All of [21]our lists have public 2700 archives. 2701 2702 Copyright (C) [22]Free Software Foundation, Inc. Verbatim copying and 2703 distribution of this entire article is permitted in any medium, 2704 provided this notice is preserved. 2705 2706 These pages are [23]maintained by the GCC team. Last modified 2707 2016-09-30[24]. 2708 2709 References 2710 2711 1. http://www.gnu.org/ 2712 2. http://gcc.gnu.org/gcc-4.7/changes.html 2713 3. https://gcc.gnu.org/onlinedocs/4.7.4/ 2714 4. http://gcc.gnu.org/gcc-4.7/changes.html 2715 5. https://gcc.gnu.org/onlinedocs/4.7.3/ 2716 6. http://gcc.gnu.org/gcc-4.7/changes.html 2717 7. https://gcc.gnu.org/onlinedocs/4.7.2/ 2718 8. http://gcc.gnu.org/gcc-4.7/changes.html 2719 9. https://gcc.gnu.org/onlinedocs/4.7.1/ 2720 10. http://gcc.gnu.org/gcc-4.7/changes.html 2721 11. https://gcc.gnu.org/onlinedocs/4.7.0/ 2722 12. http://gcc.gnu.org/gcc-4.7/buildstat.html 2723 13. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 2724 14. http://gcc.gnu.org/index.html 2725 15. mailto:gcc (a] gcc.gnu.org 2726 16. http://gcc.gnu.org/mirrors.html 2727 17. http://gcc.gnu.org/svn.html 2728 18. https://gcc.gnu.org/onlinedocs/ 2729 19. mailto:gcc-help (a] gcc.gnu.org 2730 20. mailto:gcc (a] gcc.gnu.org 2731 21. https://gcc.gnu.org/lists.html 2732 22. http://www.fsf.org/ 2733 23. https://gcc.gnu.org/about.html 2734 24. http://validator.w3.org/check/referer 2735 ====================================================================== 2736 http://gcc.gnu.org/gcc-4.7/changes.html 2737 GCC 4.7 Release Series 2738 Changes, New Features, and Fixes 2739 2740 Caveats 2741 2742 * The -fconserve-space flag has been deprecated. The flag had no 2743 effect for most targets: only targets without a global .bss section 2744 and without support for switchable sections. Furthermore, the flag 2745 only had an effect for G++, where it could result in wrong 2746 semantics (please refer to the GCC manual for further details). The 2747 flag will be removed in GCC 4.8 2748 * Support for a number of older systems and recently unmaintained or 2749 untested target ports of GCC has been declared obsolete in GCC 4.7. 2750 Unless there is activity to revive them, the next release of GCC 2751 will have their sources permanently removed. 2752 All GCC ports for the following processor architectures have been 2753 declared obsolete: 2754 + picoChip (picochip-*) 2755 The following ports for individual systems on particular 2756 architectures have been obsoleted: 2757 + IRIX 6.5 (mips-sgi-irix6.5) 2758 + MIPS OpenBSD (mips*-*-openbsd*) 2759 + Solaris 8 (*-*-solaris2.8). Details can be found in the 2760 [1]announcement. 2761 + Tru64 UNIX V5.1 (alpha*-dec-osf5.1*) 2762 * On ARM, when compiling for ARMv6 (but not ARMv6-M), ARMv7-A, 2763 ARMv7-R, or ARMv7-M, the new option -munaligned-access is active by 2764 default, which for some sources generates code that accesses memory 2765 on unaligned addresses. This requires the kernel of those systems 2766 to enable such accesses (controlled by CP15 register c1, refer to 2767 ARM documentation). Alternatively, or for compatibility with 2768 kernels where unaligned accesses are not supported, all code has to 2769 be compiled with -mno-unaligned-access. Upstream Linux kernel 2770 releases have automatically and unconditionally supported unaligned 2771 accesses as emitted by GCC due to this option being active since 2772 version 2.6.28. 2773 * Support on ARM for the legacy floating-point accelerator (FPA) and 2774 the mixed-endian floating-point format that it used has been 2775 obsoleted. The ports that still use this format have been obsoleted 2776 as well. Many legacy ARM ports already provide an alternative that 2777 uses the VFP floating-point format. The obsolete ports will be 2778 deleted in the next release. 2779 The obsolete ports with alternatives are: 2780 + arm*-*-rtems (use arm*-*-rtemseabi) 2781 + arm*-*-linux-gnu (use arm*-*-linux-gnueabi) 2782 + arm*-*-elf (use arm*-*-eabi) 2783 + arm*-*-uclinux* (use arm*-*-uclinux*eabi) 2784 Note, however, that these alternatives are not binary compatible 2785 with their legacy counterparts (although some can support running 2786 legacy applications). 2787 The obsolete ports that currently lack a modern alternative are: 2788 + arm*-*-ecos-elf 2789 + arm*-*-freebsd 2790 + arm*-wince-pe* 2791 New ports that support more recent versions of the architecture are 2792 welcome. 2793 * Support for the Maverick co-processor on ARM has been obsoleted. 2794 Code to support it will be deleted in the next release. 2795 * Support has been removed for Unix International threads on Solaris 2796 2, so the --enable-threads=solaris configure option and the 2797 -threads compiler option don't work any longer. 2798 * Support has been removed for the Solaris BSD Compatibility Package, 2799 which lives in /usr/ucbinclude and /usr/ucblib. It has been removed 2800 from Solaris 11, and was only intended as a migration aid from 2801 SunOS 4 to SunOS 5. The -compat-bsd compiler option is not 2802 recognized any longer. 2803 * The AVR port's libgcc has been improved and its multilib structure 2804 has been enhanced. As a result, all objects contributing to an 2805 application must either be compiled with GCC versions up to 4.6.x 2806 or with GCC versions 4.7.1 or later. If the compiler is used with 2807 AVR Libc, you need a version that supports the new layout, i.e. 2808 implements [2]#35407. 2809 * The AVR port's -mshort-calls command-line option has been 2810 deprecated. It will be removed in the GCC 4.8 release. See -mrelax 2811 for a replacement. 2812 * The AVR port only references startup code that clears .bss and the 2813 common section resp. initializes the .data and .rodata section 2814 provided respective sections (or subsections thereof) are not 2815 empty, see [3]PR18145. Applications that put all static storage 2816 objects into non-standard sections and / or define all static 2817 storage objects in assembler modules, must reference __do_clear_bss 2818 resp. __do_copy_data by hand or undefine the symbol(s) by means of 2819 -Wl,-u,__do_clear_bss resp. -Wl,-u,__do_copy_data. 2820 * The ARM port's -mwords-little-endian option has been deprecated. It 2821 will be removed in a future release. 2822 * Support has been removed for the NetWare x86 configuration 2823 obsoleted in GCC 4.6. 2824 * It is no longer possible to use the "l" constraint in MIPS16 asm 2825 statements. 2826 * GCC versions 4.7.0 and 4.7.1 had changes to the C++ standard 2827 library which affected the ABI in C++11 mode: a data member was 2828 added to std::list changing its size and altering the definitions 2829 of some member functions, and std::pair's move constructor was 2830 non-trivial which altered the calling convention for functions with 2831 std::pair arguments or return types. The ABI incompatibilities have 2832 been fixed for GCC version 4.7.2 but as a result C++11 code 2833 compiled with GCC 4.7.0 or 4.7.1 may be incompatible with C++11 2834 code compiled with different GCC versions and with C++98/C++03 code 2835 compiled with any version. 2836 * On ARM, a bug has been fixed in GCC's implementation of the AAPCS 2837 rules for the layout of vectors that could lead to wrong code being 2838 generated. Vectors larger than 8 bytes in size are now by default 2839 aligned to an 8-byte boundary. This is an ABI change: code that 2840 makes explicit use of vector types may be incompatible with binary 2841 objects built with older versions of GCC. Auto-vectorized code is 2842 not affected by this change. (This change affects GCC versions 2843 4.7.2 and later.) 2844 * More information on porting to GCC 4.7 from previous versions of 2845 GCC can be found in the [4]porting guide for this release. 2846 2847 General Optimizer Improvements 2848 2849 * Support for a new parameter --param case-values-threshold=n was 2850 added to allow users to control the cutoff between doing switch 2851 statements as a series of if statements and using a jump table. 2852 * Link-time optimization (LTO) improvements: 2853 + Improved scalability and reduced memory usage. Link time 2854 optimization of Firefox now requires 3GB of RAM on a 64-bit 2855 system, while over 8GB was needed previously. Linking time has 2856 been improved, too. The serial stage of linking Firefox has 2857 been sped up by about a factor of 10. 2858 + Reduced size of object files and temporary storage used during 2859 linking. 2860 + Streaming performance (both outbound and inbound) has been 2861 improved. 2862 + ld -r is now supported with LTO. 2863 + Several bug fixes, especially in symbol table handling and 2864 merging. 2865 * Interprocedural optimization improvements: 2866 + Heuristics now take into account that after inlining code will 2867 be optimized out because of known values (or properties) of 2868 function parameters. For example: 2869 void foo(int a) 2870 { 2871 if (a > 10) 2872 ... huge code ... 2873 } 2874 void bar (void) 2875 { 2876 foo (0); 2877 } 2878 2879 The call of foo will be inlined into bar even when optimizing 2880 for code size. Constructs based on __builtin_constant_p are 2881 now understood by the inliner and code size estimates are 2882 evaluated a lot more realistically. 2883 + The representation of C++ virtual thunks and aliases (both 2884 implicit and defined via the alias attribute) has been 2885 re-engineered. Aliases no longer pose optimization barriers 2886 and calls to an alias can be inlined and otherwise optimized. 2887 + The inter-procedural constant propagation pass has been 2888 rewritten. It now performs generic function specialization. 2889 For example when compiling the following: 2890 void foo(bool flag) 2891 { 2892 if (flag) 2893 ... do something ... 2894 else 2895 ... do something else ... 2896 } 2897 void bar (void) 2898 { 2899 foo (false); 2900 foo (true); 2901 foo (false); 2902 foo (true); 2903 foo (false); 2904 foo (true); 2905 } 2906 2907 GCC will now produce two copies of foo. One with flag being 2908 true, while other with flag being false. This leads to 2909 performance improvements previously possible only by inlining 2910 all calls. Cloning causes a lot less code size growth. 2911 * A string length optimization pass has been added. It attempts to 2912 track string lengths and optimize various standard C string 2913 functions like strlen, strchr, strcpy, strcat, stpcpy and their 2914 _FORTIFY_SOURCE counterparts into faster alternatives. This pass is 2915 enabled by default at -O2 or above, unless optimizing for size, and 2916 can be disabled by the -fno-optimize-strlen option. The pass can 2917 e.g. optimize 2918 char *bar (const char *a) 2919 { 2920 size_t l = strlen (a) + 2; 2921 char *p = malloc (l); if (p == NULL) return p; 2922 strcpy (p, a); strcat (p, "/"); return p; 2923 } 2924 2925 into: 2926 char *bar (const char *a) 2927 { 2928 size_t tmp = strlen (a); 2929 char *p = malloc (tmp + 2); if (p == NULL) return p; 2930 memcpy (p, a, tmp); memcpy (p + tmp, "/", 2); return p; 2931 } 2932 2933 or for hosted compilations where stpcpy is available in the runtime 2934 and headers provide its prototype, e.g. 2935 void foo (char *a, const char *b, const char *c, const char *d) 2936 { 2937 strcpy (a, b); strcat (a, c); strcat (a, d); 2938 } 2939 2940 can be optimized into: 2941 void foo (char *a, const char *b, const char *c, const char *d) 2942 { 2943 strcpy (stpcpy (stpcpy (a, b), c), d); 2944 } 2945 2946 New Languages and Language specific improvements 2947 2948 * Version 3.1 of the [5]OpenMP specification is now supported for the 2949 C, C++, and Fortran compilers. 2950 2951 Ada 2952 2953 * The command-line option -feliminate-unused-debug-types has been 2954 re-enabled by default, as it is for the other languages, leading to 2955 a reduction in debug info size of 12.5% and more for relevant 2956 cases, as well as to a small compilation speedup. 2957 2958 C family 2959 2960 * A new built-in, __builtin_assume_aligned, has been added, through 2961 which the compiler can be hinted about pointer alignment and can 2962 use it to improve generated code. 2963 * A new warning option -Wunused-local-typedefs was added for C, C++, 2964 Objective-C and Objective-C++. This warning diagnoses typedefs 2965 locally defined in a function, and otherwise not used. 2966 * A new experimental command-line option -ftrack-macro-expansion was 2967 added for C, C++, Objective-C, Objective-C++ and Fortran. It allows 2968 the compiler to emit diagnostic about the current macro expansion 2969 stack when a compilation error occurs in a macro expansion. 2970 * Experimental support for transactional memory has been added. It 2971 includes support in the compiler, as well as a supporting runtime 2972 library called libitm. To compile code with transactional memory 2973 constructs, use the -fgnu-tm option. 2974 Support is currently available for Alpha, ARM, PowerPC, SH, SPARC, 2975 and 32-bit/64-bit x86 platforms. 2976 For more details on transactional memory see [6]the GCC WiKi. 2977 * Support for atomic operations specifying the C++11/C11 memory model 2978 has been added. These new __atomic routines replace the existing 2979 __sync built-in routines. 2980 Atomic support is also available for memory blocks. Lock-free 2981 instructions will be used if a memory block is the same size and 2982 alignment as a supported integer type. Atomic operations which do 2983 not have lock-free support are left as function calls. A set of 2984 library functions is available on the GCC atomic wiki in the 2985 "External Atomics Library" section. 2986 For more details on the memory models and features, see the 2987 [7]atomic wiki. 2988 * When a binary operation is performed on vector types and one of the 2989 operands is a uniform vector, it is possible to replace the vector 2990 with the generating element. For example: 2991 typedef int v4si __attribute__ ((vector_size (16))); 2992 v4si res, a = {1,2,3,4}; 2993 int x; 2994 2995 res = 2 + a; /* means {2,2,2,2} + a */ 2996 res = a - x; /* means a - {x,x,x,x} */ 2997 2998 C 2999 3000 * There is support for some more features from the C11 revision of 3001 the ISO C standard. GCC now accepts the options -std=c11 and 3002 -std=gnu11, in addition to the previous -std=c1x and -std=gnu1x. 3003 + Unicode strings (previously supported only with options such 3004 as -std=gnu11, now supported with -std=c11), and the 3005 predefined macros __STDC_UTF_16__ and __STDC_UTF_32__. 3006 + Nonreturning functions (_Noreturn and <stdnoreturn.h>). 3007 + Alignment support (_Alignas, _Alignof, max_align_t, 3008 <stdalign.h>). 3009 + A built-in function __builtin_complex is provided to support C 3010 library implementation of the CMPLX family of macros. 3011 3012 C++ 3013 3014 * G++ now accepts the -std=c++11, -std=gnu++11, and -Wc++11-compat 3015 options, which are equivalent to -std=c++0x, -std=gnu++0x, and 3016 -Wc++0x-compat, respectively. 3017 * G++ now implements [8]C++11 extended friend syntax: 3018 3019 template<class W> 3020 class Q 3021 { 3022 static const int I = 2; 3023 public: 3024 friend W; 3025 }; 3026 3027 struct B 3028 { 3029 int ar[Q<B>::I]; 3030 }; 3031 3032 * Thanks to Ville Voutilainen, G++ now implements [9]C++11 explicit 3033 override control. 3034 3035 struct B { 3036 virtual void f() const final; 3037 virtual void f(int); 3038 }; 3039 3040 struct D : B { 3041 void f() const; // error: D::f attempts to override final B::f 3042 void f(long) override; // error: doesn't override anything 3043 void f(int) override; // ok 3044 }; 3045 3046 struct E final { }; 3047 struct F: E { }; // error: deriving from final class 3048 3049 * G++ now implements [10]C++11 non-static data member initializers. 3050 3051 struct A { 3052 int i = 42; 3053 } a; // initializes a.i to 42 3054 3055 * Thanks to Ed Smith-Rowland, G++ now implements [11]C++11 3056 user-defined literals. 3057 3058 // Not actually a good approximation. :) 3059 constexpr long double operator"" _degrees (long double d) { return d * 0.0175; } 3060 long double pi = 180.0_degrees; 3061 3062 * G++ now implements [12]C++11 alias-declarations. 3063 3064 template <class T> using Ptr = T*; 3065 Ptr<int> ip; // decltype(ip) is int* 3066 3067 * Thanks to Ville Voutilainen and Pedro Lamaro, G++ now implements 3068 [13]C++11 delegating constructors. 3069 3070 struct A { 3071 A(int); 3072 A(): A(42) { } // delegate to the A(int) constructor 3073 }; 3074 3075 * G++ now fully implements C++11 atomic classes rather than just 3076 integer derived classes. 3077 3078 class POD { 3079 int a; 3080 int b; 3081 }; 3082 std::atomic<POD> my_atomic_POD; 3083 3084 * G++ now sets the predefined macro __cplusplus to the correct value, 3085 199711L for C++98/03, and 201103L for C++11. 3086 * G++ now correctly implements the two-phase lookup rules such that 3087 an unqualified name used in a template must have an appropriate 3088 declaration found either in scope at the point of definition of the 3089 template or by argument-dependent lookup at the point of 3090 instantiation. As a result, code that relies on a second 3091 unqualified lookup at the point of instantiation to find functions 3092 declared after the template or in dependent bases will be rejected. 3093 The compiler will suggest ways to fix affected code, and using the 3094 -fpermissive compiler flag will allow the code to compile with a 3095 warning. 3096 3097 template <class T> 3098 void f() { g(T()); } // error, g(int) not found by argument-dependent lookup 3099 void g(int) { } // fix by moving this declaration before the declaration of f 3100 3101 template <class T> 3102 struct A: T { 3103 // error, B::g(B) not found by argument-dependent lookup 3104 void f() { g(T()); } // fix by using this->g or A::g 3105 }; 3106 3107 struct B { void g(B); }; 3108 3109 int main() 3110 { 3111 f<int>(); 3112 A<B>().f(); 3113 } 3114 3115 * G++ now properly re-uses stack space allocated for temporary 3116 objects when their lifetime ends, which can significantly lower 3117 stack consumption for some C++ functions. As a result of this, some 3118 code with undefined behavior will now break: 3119 3120 const int &f(const int &i) { return i; } 3121 .... 3122 const int &x = f(1); 3123 const int &y = f(2); 3124 3125 Here, x refers to the temporary allocated to hold the 1 argument, 3126 which only lives until the end of the initialization; it 3127 immediately becomes a dangling reference. So the next statement 3128 re-uses the stack slot to hold the 2 argument, and users of x get 3129 that value instead. 3130 Note that this should not cause any change of behavior for 3131 temporaries of types with non-trivial destructors, as they are 3132 already destroyed at end of full-expression; the change is that now 3133 the storage is released as well. 3134 * A new command-line option -Wdelete-non-virtual-dtor has been added 3135 to warn when delete is used to destroy an instance of a class which 3136 has virtual functions and non-virtual destructor. It is unsafe to 3137 delete an instance of a derived class through a pointer to a base 3138 class if the base class does not have a virtual destructor. This 3139 warning is enabled by -Wall. 3140 * A new command-line option -Wzero-as-null-pointer-constant has been 3141 added to warn when a literal '0' is used as null pointer constant. 3142 It can be useful to facilitate the conversion to nullptr in C++11. 3143 * As per C++98, access-declarations are now deprecated by G++. 3144 Using-declarations are to be used instead. Furthermore, some 3145 efforts have been made to improve the support of class scope 3146 using-declarations. In particular, using-declarations referring to 3147 a dependent type now work as expected ([14]bug c++/14258). 3148 * The ELF symbol visibility of a template instantiation is now 3149 properly constrained by the visibility of its template arguments 3150 ([15]bug c++/35688). 3151 3152 Runtime Library (libstdc++) 3153 3154 * [16]Improved experimental support for the new ISO C++ standard, 3155 C++11, including: 3156 + using noexcept in most of the library; 3157 + implementations of pointer_traits, allocator_traits and 3158 scoped_allocator_adaptor; 3159 + uses-allocator construction for tuple; 3160 + vector meets the allocator-aware container requirements; 3161 + replacing monotonic_clock with steady_clock; 3162 + enabling the thread support library on most POSIX targets; 3163 + many small improvements to conform to the FDIS. 3164 * Added --enable-clocale=newlib configure option. 3165 * Debug Mode iterators for unordered associative containers. 3166 * Avoid polluting the global namespace and do not include <unistd.h>. 3167 3168 Fortran 3169 3170 * The compile flag [17]-fstack-arrays has been added, which causes 3171 all local arrays to be put on stack memory. For some programs this 3172 will improve the performance significantly. If your program uses 3173 very large local arrays, it is possible that you will have to 3174 extend your runtime limits for stack memory. 3175 * The [18]-Ofast flag now also implies [19]-fno-protect-parens and 3176 [20]-fstack-arrays. 3177 * Front-end optimizations can now be selected by the 3178 [21]-ffrontend-optimize option and deselected by the 3179 -fno-frontend-optimize option. 3180 * When front-end optimization removes a function call, 3181 [22]-Wfunction-elimination warns about that. 3182 * When performing front-end-optimization, the 3183 [23]-faggressive-function-elimination option allows the removal of 3184 duplicate function calls even for impure functions. 3185 * The flag [24]-Wreal-q-constant has been added, which warns if 3186 floating-point literals have been specified using q (such as 3187 1.0q0); the q marker is now supported as a vendor extension to 3188 denote quad precision (REAL(16) or, if not available, REAL(10)). 3189 Consider using a kind parameter (such as in 1.0_qp) instead, which 3190 can be obtained via [25]SELECTED_REAL_KIND. 3191 * The GFORTRAN_USE_STDERR environment variable has been removed. GNU 3192 Fortran now always prints error messages to standard error. If you 3193 wish to redirect standard error, please consult the manual for your 3194 OS, shell, batch environment etc. as appropriate. 3195 * The -fdump-core option and GFORTRAN_ERROR_DUMPCORE environment 3196 variable have been removed. When encountering a serious error, 3197 gfortran will now always abort the program. Whether a core dump is 3198 generated depends on the user environment settings; see the ulimit 3199 -c setting for POSIX shells, limit coredumpsize for C shells, and 3200 the [26]WER user-mode dumps settings on Windows. 3201 * The [27]-fbacktrace option is now enabled by default. When 3202 encountering a fatal error, gfortran will attempt to print a 3203 backtrace to standard error before aborting. It can be disabled 3204 with -fno-backtrace. Note: On POSIX targets with the addr2line 3205 utility from GNU binutils, GNU Fortran can print a backtrace with 3206 function name, file name, line number information in addition to 3207 the addresses; otherwise only the addresses are printed. 3208 * [28]Fortran 2003: 3209 + Generic interface names which have the same name as derived 3210 types are now supported, which allows to write constructor 3211 functions. Note that Fortran does not support static 3212 constructor functions; only default initialization or an 3213 explicit structure-constructor initialization are available. 3214 + [29]Polymorphic (class) arrays are now supported. 3215 * [30]Fortran 2008: 3216 + Support for the DO CONCURRENT construct has been added, which 3217 allows the user to specify that individual loop iterations 3218 have no interdependencies. 3219 + [31]Coarrays: Full single-image support except for polymorphic 3220 coarrays. Additionally, preliminary support for multiple 3221 images via an MPI-based [32]coarray communication library has 3222 been added. Note: The library version is not yet usable as 3223 remote coarray access is not yet possible. 3224 * [33]TS 29113: 3225 + New flag [34]-std=f2008ts permits programs that are expected 3226 to conform to the Fortran 2008 standard and the draft 3227 Technical Specification (TS) 29113 on Further Interoperability 3228 of Fortran with C. 3229 + The OPTIONAL attribute is now allowed for dummy arguments of 3230 BIND(C) procedures. 3231 + The RANK intrinsic has been added. 3232 + The implementation of the ASYNCHRONOUS attribute in GCC is 3233 compatible with the candidate draft of TS 29113 (since GCC 3234 4.6). 3235 3236 Go 3237 3238 * GCC 4.7 implements the [35]Go 1 language standard. The library 3239 support in 4.7.0 is not quite complete, due to release timing. 3240 Release 4.7.1 includes complete support for Go 1. The Go library is 3241 from the Go 1.0.1 release. 3242 * Go has been tested on GNU/Linux and Solaris platforms. It may work 3243 on other platforms as well. 3244 3245 New Targets and Target Specific Improvements 3246 3247 ARM 3248 3249 * GCC now supports the Cortex-A7 processor implementing the v7-a 3250 version of the architecture using the option -mcpu=cortex-a7. 3251 * The default vector size in auto-vectorization for NEON is now 128 3252 bits. If vectorization fails thusly, the vectorizer tries again 3253 with 64-bit vectors. 3254 * A new option -mvectorize-with-neon-double was added to allow users 3255 to change the vector size to 64 bits. 3256 3257 AVR 3258 3259 * GCC now supports the XMEGA architecture. This requires GNU binutils 3260 2.22 or later. 3261 * Support for the [36]named address spaces __flash, __flash1, , 3262 __flash5 and __memx has been added. These address spaces locate 3263 read-only data in flash memory and allow reading from flash memory 3264 by means of ordinary C code, i.e. without the need of (inline) 3265 assembler code: 3266 3267 const __flash int values[] = { 42, 31 }; 3268 3269 int add_values (const __flash int *p, int i) 3270 { 3271 return values[i] + *p; 3272 } 3273 3274 * Support has been added for the AVR-specific configure option 3275 --with-avrlibc=yes in order to arrange for better integration of 3276 [37]AVR-Libc. This configure option is supported in avr-gcc 4.7.2 3277 and newer and will only take effect in non-RTEMS configurations. If 3278 avr-gcc is configured for RTEMS, the option will be ignored which 3279 is the same as specifying --with-avrlibc=no. See [38]PR54461 for 3280 more technical details. 3281 * Support for AVR-specific [39]built-in functions has been added. 3282 * Support has been added for the signed and unsigned 24-bit scalar 3283 integer types __int24 and __uint24. 3284 * New command-line options -maccumulate-args, -mbranch-cost=cost and 3285 -mstrict-X were added to allow better fine-tuning of code 3286 optimization. 3287 * The command-line option -fdata-sections now also takes affect on 3288 the section names of variables with the progmem attribute. 3289 * A new inline assembler print modifier %i to print a RAM address as 3290 I/O address has been added: 3291 3292 #include <avr/io.h> /* Port Definitions from AVR-LibC */ 3293 3294 void set_portb (uint8_t value) 3295 { 3296 asm volatile ("out %i0, %1" :: "n" (&PORTB), "r" (value) : "memory"); 3297 } 3298 3299 The offset between an I/O address and the RAM address for that I/O 3300 location is device-specific. This offset is taken into account when 3301 printing a RAM address with the %i modifier so that the address is 3302 suitable to be used as operand in an I/O command. The address must 3303 be a constant integer known at compile time. 3304 * The inline assembler constraint "R" to represent integers in the 3305 range 6 5 has been removed without replacement. 3306 * Many optimizations to: 3307 + 64-bit integer arithmetic 3308 + Widening multiplication 3309 + Integer division by a constant 3310 + Avoid constant reloading in multi-byte instructions. 3311 + Micro-optimizations for special instruction sequences. 3312 + Generic built-in functions like __builtin_ffs*, 3313 __builtin_clz*, etc. 3314 + If-else decision trees generated by switch instructions 3315 + Merging of data located in flash memory 3316 + New libgcc variants for devices with 8-bit wide stack pointer 3317 + 3318 * Better documentation: 3319 + Handling of EIND and indirect jumps on devices with more than 3320 128 KiB of program memory. 3321 + Handling of the RAMPD, RAMPX, RAMPY and RAMPZ special function 3322 registers. 3323 + Function attributes OS_main and OS_task. 3324 + AVR-specific built-in macros. 3325 3326 C6X 3327 3328 * Support has been added for the Texas Instruments C6X family of 3329 processors. 3330 3331 CR16 3332 3333 * Support has been added for National Semiconductor's CR16 3334 architecture. 3335 3336 Epiphany 3337 3338 * Support has been added for Adapteva's Epiphany architecture. 3339 3340 IA-32/x86-64 3341 3342 * Support for Intel AVX2 intrinsics, built-in functions and code 3343 generation is available via -mavx2. 3344 * Support for Intel BMI2 intrinsics, built-in functions and code 3345 generation is available via -mbmi2. 3346 * Implementation and automatic generation of __builtin_clz* using the 3347 lzcnt instruction is available via -mlzcnt. 3348 * Support for Intel FMA3 intrinsics and code generation is available 3349 via -mfma. 3350 * A new -mfsgsbase command-line option is available that makes GCC 3351 generate new segment register read/write instructions through 3352 dedicated built-ins. 3353 * Support for the new Intel rdrnd instruction is available via 3354 -mrdrnd. 3355 * Two additional AVX vector conversion instructions are available via 3356 -mf16c. 3357 * Support for new Intel processor codename IvyBridge with RDRND, 3358 FSGSBASE and F16C is available through -march=core-avx-i. 3359 * Support for the new Intel processor codename Haswell with AVX2, 3360 FMA, BMI, BMI2, LZCNT is available through -march=core-avx2. 3361 * Support for new AMD family 15h processors (Piledriver core) is now 3362 available through -march=bdver2 and -mtune=bdver2 options. 3363 * Support for [40]the x32 psABI is now available through the -mx32 3364 option. 3365 * Windows mingw targets are using the -mms-bitfields option by 3366 default. 3367 * Windows x86 targets are using the __thiscall calling convention for 3368 C++ class-member functions. 3369 * Support for the configure option --with-threads=posix for Windows 3370 mingw targets. 3371 3372 MIPS 3373 3374 * GCC now supports thread-local storage (TLS) for MIPS16. This 3375 requires GNU binutils 2.22 or later. 3376 * GCC can now generate code specifically for the Cavium Octeon+ and 3377 Octeon2 processors. The associated command-line options are 3378 -march=octeon+ and -march=octeon2 respectively. Both options 3379 require GNU binutils 2.22 or later. 3380 * GCC can now work around certain 24k errata, under the control of 3381 the command-line option -mfix-24k. These workarounds require GNU 3382 binutils 2.20 or later. 3383 * 32-bit MIPS GNU/Linux targets such as mips-linux-gnu can now build 3384 n32 and n64 multilibs. The result is effectively a 64-bit GNU/Linux 3385 toolchain that generates 32-bit code by default. Use the 3386 configure-time option --enable-targets=all to select these extra 3387 multilibs. 3388 * Passing -fno-delayed-branch now also stops the assembler from 3389 automatically filling delay slots. 3390 3391 PowerPC/PowerPC64 3392 3393 * Vectors of type vector long long or vector long are passed and 3394 returned using the same method as other vectors with the VSX 3395 instruction set. Previously GCC did not adhere to the ABI for 3396 128-bit vectors with 64-bit integer base types (PR 48857). This 3397 will also be fixed in the GCC 4.6.1 and 4.5.4 releases. 3398 * A new option -mno-pointers-to-nested-functions was added to allow 3399 AIX 32-bit/64-bit and GNU/Linux 64-bit PowerPC users to specify 3400 that the compiler should not load up the chain register (r11) 3401 before calling a function through a pointer. If you use this 3402 option, you cannot call nested functions through a pointer, or call 3403 other languages that might use the static chain. 3404 * A new option msave-toc-indirect was added to allow AIX 3405 32-bit/64-bit and GNU/Linux 64-bit PowerPC users control whether we 3406 save the TOC in the prologue for indirect calls or generate the 3407 save inline. This can speed up some programs that call through a 3408 function pointer a lot, but it can slow down other functions that 3409 only call through a function pointer in exceptional cases. 3410 * The PowerPC port will now enable machine-specific built-in 3411 functions when the user switches the target machine using the 3412 #pragma GCC target or __attribute__ ((__target__ ("target"))) code 3413 sequences. In addition, the target macros are updated. However, due 3414 to the way the -save-temps switch is implemented, you won't see the 3415 effect of these additional macros being defined in preprocessor 3416 output. 3417 3418 SH 3419 3420 * A new option -msoft-atomic has been added. When it is specified, 3421 GCC will generate GNU/Linux-compatible gUSA atomic sequences for 3422 the new __atomic routines. 3423 * Since it is neither supported by GAS nor officially documented, 3424 code generation for little endian SH2A has been disabled. 3425 Specifying -ml with -m2a* will now result in a compiler error. 3426 * The defunct -mbranch-cost option has been fixed. 3427 * Some improvements to the generated code of: 3428 + Utilization of the tst #imm,R0 instruction. 3429 + Dynamic shift instructions on SH2A. 3430 + Integer absolute value calculations. 3431 * The -mdiv= option for targets other than SHmedia has been fixed and 3432 documented. 3433 3434 SPARC 3435 3436 * The option -mflat has been reinstated. When it is specified, the 3437 compiler will generate code for a single register window model. 3438 This is essentially a new implementation and the corresponding 3439 debugger support has been added to GDB 7.4. 3440 * Support for the options -mtune=native and -mcpu=native has been 3441 added on selected native platforms (GNU/Linux and Solaris). 3442 * Support for the SPARC T3 (Niagara 3) processor has been added. 3443 * VIS: 3444 + An intrinsics header visintrin.h has been added. 3445 + Builtin intrinsics for the VIS 1.0 edge handling and pixel 3446 compare instructions have been added. 3447 + The little-endian version of alignaddr is now supported. 3448 + When possible, VIS builtins are marked const, which should 3449 increase the compiler's ability to optimize VIS operations. 3450 + The compiler now properly tracks the %gsr register and how it 3451 behaves as an input for various VIS instructions. 3452 + Akin to fzero, the compiler can now generate fone instructions 3453 in order to set all of the bits of a floating-point register 3454 to 1. 3455 + The documentation for the VIS intrinsics in the GCC manual has 3456 been brought up to date and many inaccuracies were fixed. 3457 + Intrinsics for the VIS 2.0 bmask, bshuffle, and 3458 non-condition-code setting edge instructions have been added. 3459 Their availability is controlled by the new -mvis2 and 3460 -mno-vis2 options. They are enabled by default on 3461 UltraSPARC-III and later CPUs. 3462 * Support for UltraSPARC Fused Multiply-Add floating-point extensions 3463 has been added. These instructions are enabled by default on SPARC 3464 T3 (Niagara 3) and later CPUs. 3465 3466 TILE-Gx/TILEPro 3467 3468 * Support has been added for the Tilera TILE-Gx and TILEPro families 3469 of processors. 3470 3471 Other significant improvements 3472 3473 * A new option (-grecord-gcc-switches) was added that appends 3474 compiler command-line options that might affect code generation to 3475 the DW_AT_producer attribute string in the DWARF debugging 3476 information. 3477 * GCC now supports various new GNU extensions to the DWARF debugging 3478 information format, like [41]entry value and [42]call site 3479 information, [43]typed DWARF stack or [44]a more compact macro 3480 representation. Support for these extensions has been added to GDB 3481 7.4. They can be disabled through the -gstrict-dwarf command-line 3482 option. 3483 3484 GCC 4.7.1 3485 3486 This is the [45]list of problem reports (PRs) from GCC's bug tracking 3487 system that are known to be fixed in the 4.7.1 release. This list might 3488 not be complete (that is, it is possible that some PRs that have been 3489 fixed are not listed here). 3490 3491 The Go frontend in the 4.7.1 release fully supports the [46]Go 1 3492 language standard. 3493 3494 GCC 4.7.2 3495 3496 This is the [47]list of problem reports (PRs) from GCC's bug tracking 3497 system that are known to be fixed in the 4.7.2 release. This list might 3498 not be complete (that is, it is possible that some PRs that have been 3499 fixed are not listed here). 3500 3501 GCC 4.7.3 3502 3503 This is the [48]list of problem reports (PRs) from GCC's bug tracking 3504 system that are known to be fixed in the 4.7.3 release. This list might 3505 not be complete (that is, it is possible that some PRs that have been 3506 fixed are not listed here). 3507 3508 GCC 4.7.4 3509 3510 This is the [49]list of problem reports (PRs) from GCC's bug tracking 3511 system that are known to be fixed in the 4.7.4 release. This list might 3512 not be complete (that is, it is possible that some PRs that have been 3513 fixed are not listed here). 3514 3515 3516 For questions related to the use of GCC, please consult these web 3517 pages and the [50]GCC manuals. If that fails, the 3518 [51]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 3519 web pages and the development of GCC are welcome on our developer 3520 list at [52]gcc (a] gcc.gnu.org. All of [53]our lists have public 3521 archives. 3522 3523 Copyright (C) [54]Free Software Foundation, Inc. Verbatim copying and 3524 distribution of this entire article is permitted in any medium, 3525 provided this notice is preserved. 3526 3527 These pages are [55]maintained by the GCC team. Last modified 3528 2017-03-03[56]. 3529 3530 References 3531 3532 1. https://gcc.gnu.org/ml/gcc-patches/2011-03/msg01263.html 3533 2. http://savannah.nongnu.org/bugs/?35407 3534 3. https://gcc.gnu.org/PR18145 3535 4. https://gcc.gnu.org/gcc-4.7/porting_to.html 3536 5. http://www.openmp.org/specifications/ 3537 6. https://gcc.gnu.org/wiki/TransactionalMemory 3538 7. https://gcc.gnu.org/wiki/Atomic/GCCMM 3539 8. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html 3540 9. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html 3541 10. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html 3542 11. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html 3543 12. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html 3544 13. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html 3545 14. https://gcc.gnu.org/PR14258 3546 15. https://gcc.gnu.org/PR35688 3547 16. https://gcc.gnu.org/onlinedocs/gcc-4.7.4/libstdc++/manual/manual/status.html#status.iso.2011 3548 17. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfstack-arrays_007d-254 3549 18. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Optimize-Options.html#index-Ofast-689 3550 19. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfno-protect-parens_007d-270 3551 20. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfstack-arrays_007d-254 3552 21. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfrontend-optimize_007d-275 3553 22. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWfunction-elimination_007d-170 3554 23. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfaggressive-function-elimination_007d-270 3555 24. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWreal-q-constant_007d-149 3556 25. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/SELECTED_005fREAL_005fKIND.html 3557 26. https://msdn.microsoft.com/en-us/library/bb787181(v=vs.85).aspx 3558 27. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Debugging-Options.html#index-g_t_0040code_007bfno-backtrace_007d-183 3559 28. https://gcc.gnu.org/wiki/Fortran2003Status 3560 29. https://gcc.gnu.org/wiki/OOP 3561 30. https://gcc.gnu.org/wiki/Fortran2008Status 3562 31. https://gcc.gnu.org/wiki/Coarray 3563 32. https://gcc.gnu.org/wiki/CoarrayLib 3564 33. https://gcc.gnu.org/wiki/TS29113Status 3565 34. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Fortran-Dialect-Options.html#index-g_t_0040code_007bstd_003d_007d_0040var_007bstd_007d-option-53 3566 35. https://golang.org/doc/go1 3567 36. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Named-Address-Spaces.html 3568 37. http://nongnu.org/avr-libc/ 3569 38. https://gcc.gnu.org/PR54461 3570 39. https://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/AVR-Built_002din-Functions.html 3571 40. https://sites.google.com/site/x32abi/ 3572 41. http://www.dwarfstd.org/ShowIssue.php?issue=100909.1 3573 42. http://www.dwarfstd.org/ShowIssue.php?issue=100909.2 3574 43. http://www.dwarfstd.org/ShowIssue.php?issue=140425.1 3575 44. http://www.dwarfstd.org/ShowIssue.php?issue=110722.1 3576 45. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.1 3577 46. https://golang.org/doc/go1 3578 47. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.2 3579 48. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.3 3580 49. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.4 3581 50. https://gcc.gnu.org/onlinedocs/ 3582 51. mailto:gcc-help (a] gcc.gnu.org 3583 52. mailto:gcc (a] gcc.gnu.org 3584 53. https://gcc.gnu.org/lists.html 3585 54. http://www.fsf.org/ 3586 55. https://gcc.gnu.org/about.html 3587 56. http://validator.w3.org/check/referer 3588 ====================================================================== 3589 http://gcc.gnu.org/gcc-4.6/index.html 3590 GCC 4.6 Release Series 3591 3592 April 12, 2013 3593 3594 The [1]GNU project and the GCC developers are pleased to announce the 3595 release of GCC 4.6.4. 3596 3597 This release is a bug-fix release, containing fixes for regressions in 3598 GCC 4.6.3 relative to previous releases of GCC. 3599 3600 This release series is no longer maintained. 3601 3602 Release History 3603 3604 GCC 4.6.4 3605 April 12, 2013 ([2]changes, [3]documentation) 3606 3607 GCC 4.6.3 3608 March 1, 2012 ([4]changes, [5]documentation) 3609 3610 GCC 4.6.2 3611 October 26, 2011 ([6]changes, [7]documentation) 3612 3613 GCC 4.6.1 3614 June 27, 2011 ([8]changes, [9]documentation) 3615 3616 GCC 4.6.0 3617 March 25, 2011 ([10]changes, [11]documentation) 3618 3619 References and Acknowledgements 3620 3621 GCC used to stand for the GNU C Compiler, but since the compiler 3622 supports several other languages aside from C, it now stands for the 3623 GNU Compiler Collection. 3624 3625 A list of [12]successful builds is updated as new information becomes 3626 available. 3627 3628 The GCC developers would like to thank the numerous people that have 3629 contributed new features, improvements, bug fixes, and other changes as 3630 well as test results to GCC. This [13]amazing group of volunteers is 3631 what makes GCC successful. 3632 3633 For additional information about GCC please refer to the [14]GCC 3634 project web site or contact the [15]GCC development mailing list. 3635 3636 To obtain GCC please use [16]our mirror sites or [17]our SVN server. 3637 3638 3639 For questions related to the use of GCC, please consult these web 3640 pages and the [18]GCC manuals. If that fails, the 3641 [19]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 3642 web pages and the development of GCC are welcome on our developer 3643 list at [20]gcc (a] gcc.gnu.org. All of [21]our lists have public 3644 archives. 3645 3646 Copyright (C) [22]Free Software Foundation, Inc. Verbatim copying and 3647 distribution of this entire article is permitted in any medium, 3648 provided this notice is preserved. 3649 3650 These pages are [23]maintained by the GCC team. Last modified 3651 2016-09-30[24]. 3652 3653 References 3654 3655 1. http://www.gnu.org/ 3656 2. http://gcc.gnu.org/gcc-4.6/changes.html 3657 3. https://gcc.gnu.org/onlinedocs/4.6.4/ 3658 4. http://gcc.gnu.org/gcc-4.6/changes.html 3659 5. https://gcc.gnu.org/onlinedocs/4.6.3/ 3660 6. http://gcc.gnu.org/gcc-4.6/changes.html 3661 7. https://gcc.gnu.org/onlinedocs/4.6.2/ 3662 8. http://gcc.gnu.org/gcc-4.6/changes.html 3663 9. https://gcc.gnu.org/onlinedocs/4.6.1/ 3664 10. http://gcc.gnu.org/gcc-4.6/changes.html 3665 11. https://gcc.gnu.org/onlinedocs/4.6.0/ 3666 12. http://gcc.gnu.org/gcc-4.6/buildstat.html 3667 13. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 3668 14. http://gcc.gnu.org/index.html 3669 15. mailto:gcc (a] gcc.gnu.org 3670 16. http://gcc.gnu.org/mirrors.html 3671 17. http://gcc.gnu.org/svn.html 3672 18. https://gcc.gnu.org/onlinedocs/ 3673 19. mailto:gcc-help (a] gcc.gnu.org 3674 20. mailto:gcc (a] gcc.gnu.org 3675 21. https://gcc.gnu.org/lists.html 3676 22. http://www.fsf.org/ 3677 23. https://gcc.gnu.org/about.html 3678 24. http://validator.w3.org/check/referer 3679 ====================================================================== 3680 http://gcc.gnu.org/gcc-4.6/changes.html 3681 GCC 4.6 Release Series 3682 Changes, New Features, and Fixes 3683 3684 Caveats 3685 3686 * The options -b <machine> and -V <version> have been removed because 3687 they were unreliable. Instead, users should directly run 3688 <machine>-gcc when cross-compiling, or <machine>-gcc-<version> to 3689 run a different version of gcc. 3690 * GCC now has stricter checks for invalid command-line options. In 3691 particular, when gcc was called to link object files rather than 3692 compile source code, it would previously accept and ignore all 3693 options starting with --, including linker options such as 3694 --as-needed and --export-dynamic, although such options would 3695 result in errors if any source code was compiled. Such options, if 3696 unknown to the compiler, are now rejected in all cases; if the 3697 intent was to pass them to the linker, options such as 3698 -Wl,--as-needed should be used. 3699 * Versions of the GNU C library up to and including 2.11.1 included 3700 an [1]incorrect implementation of the cproj function. GCC optimizes 3701 its builtin cproj according to the behavior specified and allowed 3702 by the ISO C99 standard. If you want to avoid discrepancies between 3703 the C library and GCC's builtin transformations when using cproj in 3704 your code, use GLIBC 2.12 or later. If you are using an older GLIBC 3705 and actually rely on the incorrect behavior of cproj, then you can 3706 disable GCC's transformations using -fno-builtin-cproj. 3707 * The C-only intermodule optimization framework (IMA, enabled by 3708 -combine) has been removed in favor of the new generic link-time 3709 optimization framework (LTO) introduced in [2]GCC 4.5.0. 3710 * GCC now ships with the LGPL-licensed libquadmath library, which 3711 provides quad-precision mathematical functions for targets with a 3712 __float128 datatype. __float128 is available for targets on 32-bit 3713 x86, x86-64 and Itanium architectures. The libquadmath library is 3714 automatically built on such targets when building the Fortran 3715 compiler. 3716 * New -Wunused-but-set-variable and -Wunused-but-set-parameter 3717 warnings were added for C, C++, Objective-C and Objective-C++. 3718 These warnings diagnose variables respective parameters which are 3719 only set in the code and never otherwise used. Usually such 3720 variables are useless and often even the value assigned to them is 3721 computed needlessly, sometimes expensively. The 3722 -Wunused-but-set-variable warning is enabled by default by -Wall 3723 flag and -Wunused-but-set-parameter by -Wall -Wextra flags. 3724 * On ARM, a bug has been fixed in GCC's implementation of the AAPCS 3725 rules for the layout of vectors that could lead to wrong code being 3726 generated. Vectors larger than 8 bytes in size are now by default 3727 aligned to an 8-byte boundary. This is an ABI change: code that 3728 makes explicit use of vector types may be incompatible with binary 3729 objects built with older versions of GCC. Auto-vectorized code is 3730 not affected by this change. (This change affects GCC versions 3731 4.6.4 and later, with the exception of versions 4.7.0 and 4.7.1.) 3732 * On AVR, variables with the progmem attribute to locate data in 3733 flash memory must be qualified as const. 3734 * Support for a number of older systems and recently unmaintained or 3735 untested target ports of GCC has been declared obsolete in GCC 4.6. 3736 Unless there is activity to revive them, the next release of GCC 3737 will have their sources permanently removed. 3738 All GCC ports for the following processor architectures have been 3739 declared obsolete: 3740 + Argonaut ARC (arc-*) 3741 + National Semiconductor CRX (crx-*) 3742 + Motorola 68HC11 and 68HC12 (m68hc11-*-*, m6811-*-*, 3743 m68hc12-*-*, m6812-*-*) 3744 + Sunplus S+core (score-*) 3745 The following ports for individual systems on particular 3746 architectures have been obsoleted: 3747 + Interix (i[34567]86-*-interix3*) 3748 + NetWare x86 (i[3456x]86-*-netware*) 3749 + Generic ARM PE (arm-*-pe* other than arm*-wince-pe*) 3750 + MCore PE (mcore-*-pe*) 3751 + SH SymbianOS (sh*-*-symbianelf*) 3752 + GNU Hurd on Alpha and PowerPC (alpha*-*-gnu*, powerpc*-*-gnu*) 3753 + M68K uClinux old ABI (m68k-*-uclinuxoldabi*) 3754 + a.out NetBSD (arm*-*-netbsd*, i[34567]86-*-netbsd*, 3755 vax-*-netbsd*, but not *-*-netbsdelf*) 3756 The i[34567]86-*-pe alias for Cygwin targets has also been 3757 obsoleted; users should configure for i[34567]86-*-cygwin* instead. 3758 Certain configure options to control the set of libraries built 3759 with GCC on some targets have been obsoleted. On ARM targets, the 3760 options --disable-fpu, --disable-26bit, --disable-underscore, 3761 --disable-interwork, --disable-biendian and --disable-nofmult have 3762 been obsoleted. On MIPS targets, the options 3763 --disable-single-float, --disable-biendian and --disable-softfloat 3764 have been obsoleted. 3765 * Support has been removed for all the [3]configurations obsoleted in 3766 GCC 4.5. 3767 * More information on porting to GCC 4.6 from previous versions of 3768 GCC can be found in the [4]porting guide for this release. 3769 3770 General Optimizer Improvements 3771 3772 * A new general optimization level, -Ofast, has been introduced. It 3773 combines the existing optimization level -O3 with options that can 3774 affect standards compliance but result in better optimized code. 3775 For example, -Ofast enables -ffast-math. 3776 * Link-time optimization improvements: 3777 + The [5]Scalable Whole Program Optimizer (WHOPR) project has 3778 stabilized to the point of being usable. It has become the 3779 default mode when using the LTO optimization model. Link time 3780 optimization can now split itself into multiple parallel 3781 compilations. Parallelism is controlled with -flto=n (where n 3782 specifies the number of compilations to execute in parallel). 3783 GCC can also cooperate with a GNU make job server by 3784 specifying the -flto=jobserver option and adding + to the 3785 beginning of the Makefile rule executing the linker. 3786 Classical LTO mode can be enforced by -flto-partition=none. 3787 This may result in small code quality improvements. 3788 + A large number of bugs were fixed. GCC itself, Mozilla Firefox 3789 and other large applications can be built with LTO enabled. 3790 + The linker plugin support improvements 3791 o Linker plugin is now enabled by default when the linker 3792 is detected to have plugin support. This is the case for 3793 GNU ld 2.21.51 or newer (on ELF and Cygwin targets) and 3794 the Gold linker on ELF targets. Plugin support of the 3795 Apple linker on Darwin is not compatible with GCC. The 3796 linker plugin can also be controlled by the 3797 -fuse-linker-plugin command-line option. 3798 o Resolution information from the linker plugin is used to 3799 drive whole program assumptions. Use of the linker plugin 3800 results in more aggressive optimization on binaries and 3801 on shared libraries that use the hidden visibility 3802 attribute. Consequently the use of -fwhole-program is not 3803 necessary in addition to LTO. 3804 + Hidden symbols used from non-LTO objects now have to be 3805 explicitly annotated with externally_visible when the linker 3806 plugin is not used. 3807 + C++ inline functions and virtual tables are now privatized 3808 more aggressively, leading to better inter-procedural 3809 optimization and faster dynamic linking. 3810 + Memory usage and intermediate language streaming performance 3811 have been improved. 3812 + Static constructors and destructors from individual units are 3813 inlined into a single function. This can significantly improve 3814 startup times of large C++ applications where static 3815 constructors are very common. For example, static constructors 3816 are used when including the iostream header. 3817 + Support for the Ada language has been added. 3818 * Interprocedural optimization improvements 3819 + The interprocedural framework was re-tuned for link time 3820 optimization. Several scalability issues were resolved. 3821 + Improved auto-detection of const and pure functions. Newly, 3822 noreturn functions are auto-detected. 3823 The [6]-Wsuggest-attribute=[const|pure|noreturn] flag is 3824 available that informs users when adding attributes to headers 3825 might improve code generation. 3826 + A number of inlining heuristic improvements. In particular: 3827 o Partial inlining is now supported and enabled by default 3828 at -O2 and greater. The feature can be controlled via 3829 -fpartial-inlining. 3830 Partial inlining splits functions with short hot path to 3831 return. This allows more aggressive inlining of the hot 3832 path leading to better performance and often to code size 3833 reductions (because cold parts of functions are not 3834 duplicated). 3835 o Scalability for large compilation units was improved 3836 significantly. 3837 o Inlining of callbacks is now more aggressive. 3838 o Virtual methods are considered for inlining when the 3839 caller is inlined and devirtualization is then possible. 3840 o Inlining when optimizing for size (either in cold regions 3841 of a program or when compiling with -Os) was improved to 3842 better handle C++ programs with larger abstraction 3843 penalty, leading to smaller and faster code. 3844 + The IPA reference optimization pass detecting global variables 3845 used or modified by functions was strengthened and sped up. 3846 + Functions whose address was taken are now optimized out when 3847 all references to them are dead. 3848 + A new inter-procedural static profile estimation pass detects 3849 functions that are executed once or unlikely to be executed. 3850 Unlikely executed functions are optimized for size. Functions 3851 executed once are optimized for size except for the inner 3852 loops. 3853 + On most targets with named section support, functions used 3854 only at startup (static constructors and main), functions used 3855 only at exit and functions detected to be cold are placed into 3856 separate text segment subsections. This extends the 3857 -freorder-functions feature and is controlled by the same 3858 switch. The goal is to improve the startup time of large C++ 3859 programs. 3860 Proper function placement requires linker support. GNU ld 3861 2.21.51 on ELF targets was updated to place those functions 3862 together within the text section leading to better code 3863 locality and faster startup times of large C++ programs. The 3864 feature is also supported in the Apple linker. Support in the 3865 gold linker is planned. 3866 * A new switch -fstack-usage has been added. It makes the compiler 3867 output stack usage information for the program, on a per-function 3868 basis, in an auxiliary file. 3869 * A new switch -fcombine-stack-adjustments has been added. It can be 3870 used to enable or disable the compiler's stack-slot combining pass 3871 which before was enabled automatically at -O1 and above, but could 3872 not be controlled on its own. 3873 * A new switch -fstrict-volatile-bitfields has been added. Using it 3874 indicates that accesses to volatile bitfields should use a single 3875 access of the width of the field's type. This option can be useful 3876 for precisely defining and accessing memory-mapped peripheral 3877 registers from C or C++. 3878 3879 Compile time and memory usage improvements 3880 3881 * Datastructures used by the dataflow framework in GCC were 3882 reorganized for better memory usage and more cache locality. 3883 Compile time is improved especially on units with large functions 3884 (possibly resulting from a lot of inlining) not fitting into the 3885 processor cache. The compile time of the GCC C compiler binary with 3886 link-time optimization went down by over 10% (benchmarked on x86-64 3887 target). 3888 3889 New Languages and Language specific improvements 3890 3891 Ada 3892 3893 * Stack checking has been improved on selected architectures (Alpha, 3894 IA-32/x86-64, RS/6000 and SPARC): it now will detect stack 3895 overflows in all cases on these architectures. 3896 * Initial support for Ada 2012 has been added. 3897 3898 C family 3899 3900 * A new warning, enabled by -Wdouble-promotion, has been added that 3901 warns about cases where a value of type float is implicitly 3902 promoted to double. This is especially helpful for CPUs that handle 3903 the former in hardware, but emulate the latter in software. 3904 * A new function attribute leaf was introduced. This attribute allows 3905 better inter-procedural optimization across calls to functions that 3906 return to the current unit only via returning or exception 3907 handling. This is the case for most library functions that have no 3908 callbacks. 3909 * Support for a new data type __int128 for targets having wide enough 3910 machine-mode support. 3911 * The new function attribute callee_pop_aggregate allows to specify 3912 if the caller or callee is responsible for popping the aggregate 3913 return pointer value from the stack. 3914 * Support for selectively enabling and disabling warnings via #pragma 3915 GCC diagnostic has been added. For instance: 3916 #pragma GCC diagnostic error "-Wuninitialized" 3917 foo(a); /* error is given for this one */ 3918 #pragma GCC diagnostic push 3919 #pragma GCC diagnostic ignored "-Wuninitialized" 3920 foo(b); /* no diagnostic for this one */ 3921 #pragma GCC diagnostic pop 3922 foo(c); /* error is given for this one */ 3923 #pragma GCC diagnostic pop 3924 foo(d); /* depends on command-line options */ 3925 3926 * The -fmax-errors=N option is now supported. Using this option 3927 causes the compiler to exit after N errors have been issued. 3928 3929 C 3930 3931 * There is now experimental support for some features from the 3932 upcoming C1X revision of the ISO C standard. This support may be 3933 selected with -std=c1x, or -std=gnu1x for C1X with GNU extensions. 3934 Note that this support is experimental and may change incompatibly 3935 in future releases for consistency with changes to the C1X standard 3936 draft. The following features are newly supported as described in 3937 the N1539 draft of C1X (with changes agreed at the March 2011 WG14 3938 meeting); some other features were already supported with no 3939 compiler changes being needed, or have some support but not in full 3940 accord with N1539 (as amended). 3941 + Static assertions (_Static_assert keyword) 3942 + Typedef redefinition 3943 + New macros in <float.h> 3944 + Anonymous structures and unions 3945 * The new -fplan9-extensions option directs the compiler to support 3946 some extensions for anonymous struct fields which are implemented 3947 by the Plan 9 compiler. A pointer to a struct may be automatically 3948 converted to a pointer to an anonymous field when calling a 3949 function, in order to make the types match. An anonymous struct 3950 field whose type is a typedef name may be referred to using the 3951 typedef name. 3952 3953 C++ 3954 3955 * Improved [7]experimental support for the upcoming C++0x ISO C++ 3956 standard, including support for constexpr (thanks to Gabriel Dos 3957 Reis and Jason Merrill), nullptr (thanks to Magnus Fromreide), 3958 noexcept, unrestricted unions, range-based for loops (thanks to 3959 Rodrigo Rivas Costa), opaque enum declarations (thanks also to 3960 Rodrigo), implicitly deleted functions and implicit move 3961 constructors. 3962 * When an extern declaration within a function does not match a 3963 declaration in the enclosing context, G++ now properly declares the 3964 name within the namespace of the function rather than the namespace 3965 which was open just before the function definition ([8]c++/43145). 3966 * GCC now warns by default when casting integers to larger pointer 3967 types. These warnings can be disabled with the option 3968 -Wno-int-to-pointer-cast, which is now also available in C++. 3969 * G++ no longer optimizes using the assumption that a value of 3970 enumeration type will fall within the range specified by the 3971 standard, since that assumption is easily violated with a 3972 conversion from integer type ([9]c++/43680). The old behavior can 3973 be restored with -fstrict-enums. 3974 * The new -fnothrow-opt flag changes the semantics of a throw() 3975 exception specification to match the proposed semantics of the 3976 noexcept specification: just call terminate if an exception tries 3977 to propagate out of a function with such an exception 3978 specification. This dramatically reduces or eliminates the code 3979 size overhead from adding the exception specification. 3980 * The new -Wnoexcept flag will suggest adding a noexcept qualifier to 3981 a function that the compiler can tell doesn't throw if it would 3982 change the value of a noexcept expression. 3983 * The -Wshadow option now warns if a local variable or type 3984 declaration shadows another type in C++. Note that the compiler 3985 will not warn if a local variable shadows a struct/class/enum, but 3986 will warn if it shadows an explicit typedef. 3987 * When an identifier is not found in the current scope, G++ now 3988 offers suggestions about which identifier might have been intended. 3989 * G++ now issues clearer diagnostics for missing semicolons after 3990 class, struct, and union definitions. 3991 * G++ now issues clearer diagnostics for missing semicolons after 3992 class member declarations. 3993 * G++ now issues clearer diagnostics when a colon is used in a place 3994 where a double-colon was intended. 3995 * G++ no longer accepts mutable on reference members ([10]c++/33558). 3996 Use -fpermissive to allow the old, non-conforming behaviour. 3997 * A few mangling fixes have been made, to attribute const/volatile on 3998 function pointer types, decltype of a plain decl, and use of a 3999 function parameter in the declaration of another parameter. By 4000 default the compiler still uses the old mangling, but emits aliases 4001 with the new mangling on targets that support strong aliases. Users 4002 can switch over entirely to the new mangling with -fabi-version=5 4003 or -fabi-version=0. -Wabi will now warn about code that uses the 4004 old mangling. 4005 * In 4.6.0 and 4.6.1 G++ no longer allows objects of const-qualified 4006 type to be default initialized unless the type has a user-declared 4007 default constructor. In 4.6.2 G++ implements the proposed 4008 resolution of [11]DR 253, so default initialization is allowed if 4009 it initializes all subobjects. Code that fails to compile can be 4010 fixed by providing an initializer e.g. 4011 struct A { A(); }; 4012 struct B : A { int i; }; 4013 const B b = B(); 4014 Use -fpermissive to allow the old, non-conforming behaviour. 4015 4016 Runtime Library (libstdc++) 4017 4018 * [12]Improved experimental support for the upcoming ISO C++ 4019 standard, C++0x, including using constexpr and nullptr. 4020 * Performance improvements to the [13]Debug Mode, thanks to Franois 4021 Dumont. 4022 * Atomic operations used for reference-counting are annotated so that 4023 they can be understood by race detectors such as Helgrind, see 4024 [14]Data Race Hunting. 4025 * Most libstdc++ standard headers have been changed to no longer 4026 include the cstddef header as an implementation detail. Code that 4027 relied on that header being included as side-effect of including 4028 other standard headers will need to include cstddef explicitly. 4029 4030 Fortran 4031 4032 * On systems supporting the libquadmath library, GNU Fortran now also 4033 supports a quad-precision, kind=16 floating-point data type 4034 (REAL(16), COMPLEX(16)). As the data type is not fully supported in 4035 hardware, calculations might be one to two orders of magnitude 4036 slower than with the 4, 8 or 10 bytes floating-point data types. 4037 This change does not affect systems which support REAL(16) in 4038 hardware nor those which do not support libquadmath. 4039 * Much improved compile time for large array constructors. 4040 * In order to reduce execution time and memory consumption, use of 4041 temporary arrays in assignment expressions is avoided for many 4042 cases. The compiler now reverses loops in order to avoid generating 4043 a temporary array where possible. 4044 * Improved diagnostics, especially with -fwhole-file. 4045 * The -fwhole-file flag is now enabled by default. This improves code 4046 generation and diagnostics. It can be disabled using the deprecated 4047 -fno-whole-file flag. 4048 * Support the generation of Makefile dependencies via the [15]-M... 4049 flags of GCC; you may need to specify the -cpp option in addition. 4050 The dependencies take modules, Fortran's include, and CPP's 4051 #include into account. Note: Using -M for the module path is no 4052 longer supported, use -J instead. 4053 * The flag -Wconversion has been modified to only issue warnings 4054 where a conversion leads to information loss. This drastically 4055 reduces the number of warnings; -Wconversion is thus now enabled 4056 with -Wall. The flag -Wconversion-extra has been added and also 4057 warns about other conversions; -Wconversion-extra typically issues 4058 a huge number of warnings, most of which can be ignored. 4059 * A new command-line option -Wunused-dummy-argument warns about 4060 unused dummy arguments and is included in -Wall. Before, 4061 -Wunused-variable also warned about unused dummy arguments. 4062 * Fortran 2003 support has been extended: 4063 + Improved support for polymorphism between libraries and 4064 programs and for complicated inheritance patterns (cf. 4065 [16]object-oriented programming). 4066 + Experimental support of the ASSOCIATE construct. 4067 + In pointer assignments it is now possible to specify the lower 4068 bounds of the pointer and, for a rank-1 or a simply contiguous 4069 data-target, to remap the bounds. 4070 + Automatic (re)allocation: In intrinsic assignments to 4071 allocatable variables the left-hand side will be automatically 4072 allocated (if unallocated) or reallocated (if the shape or 4073 type parameter is different). To avoid the small performance 4074 penalty, you can use a(:) = ... instead of a = ... for arrays 4075 and character strings or disable the feature using -std=f95 4076 or -fno-realloc-lhs. 4077 + Deferred type parameter: For scalar allocatable and pointer 4078 variables the character length can be deferred. 4079 + Namelist variables with allocatable and pointer attribute and 4080 nonconstant length type parameter are supported. 4081 * Fortran 2008 support has been extended: 4082 + Experimental [17]coarray support (for one image only, i.e. 4083 num_images() == 1); use the [18]-fcoarray=single flag to 4084 enable it. 4085 + The STOP and the new ERROR STOP statements now support all 4086 constant expressions. 4087 + Support for the CONTIGUOUS attribute. 4088 + Support for ALLOCATE with MOLD. 4089 + Support for the STORAGE_SIZE intrinsic inquiry function. 4090 + Support of the NORM2 and PARITY intrinsic functions. 4091 + The following bit intrinsics were added: POPCNT and POPPAR for 4092 counting the number of 1 bits and returning the parity; BGE, 4093 BGT, BLE, and BLT for bitwise comparisons; DSHIFTL and DSHIFTR 4094 for combined left and right shifts, MASKL and MASKR for simple 4095 left and right justified masks, MERGE_BITS for a bitwise merge 4096 using a mask, SHIFTA, SHIFTL and SHIFTR for shift operations, 4097 and the transformational bit intrinsics IALL, IANY and 4098 IPARITY. 4099 + Support of the EXECUTE_COMMAND_LINE intrinsic subroutine. 4100 + Support for the IMPURE attribute for procedures, which allows 4101 for ELEMENTAL procedures without the restrictions of PURE. 4102 + Null pointers (including NULL()) and not allocated variables 4103 can be used as actual argument to optional non-pointer, 4104 non-allocatable dummy arguments, denoting an absent argument. 4105 + Non-pointer variables with TARGET attribute can be used as 4106 actual argument to POINTER dummies with INTENT(IN) 4107 + Pointers including procedure pointers and those in a derived 4108 type (pointer components) can now be initialized by a target 4109 instead of only by NULL. 4110 + The EXIT statement (with construct-name) can now be used to 4111 leave not only the DO but also the ASSOCIATE, BLOCK, IF, 4112 SELECT CASE and SELECT TYPE constructs. 4113 + Internal procedures can now be used as actual argument. 4114 + The named constants INTEGER_KINDS, LOGICAL_KINDS, REAL_KINDS 4115 and CHARACTER_KINDS of the intrinsic module ISO_FORTRAN_ENV 4116 have been added; these arrays contain the supported kind 4117 values for the respective types. 4118 + The module procedures C_SIZEOF of the intrinsic module 4119 ISO_C_BINDINGS and COMPILER_VERSION and COMPILER_OPTIONS of 4120 ISO_FORTRAN_ENV have been implemented. 4121 + Minor changes: obsolescence diagnostics for ENTRY was added 4122 for -std=f2008; a line may start with a semicolon; for 4123 internal and module procedures END can be used instead of END 4124 SUBROUTINE and END FUNCTION; SELECTED_REAL_KIND now also takes 4125 a RADIX argument; intrinsic types are supported for 4126 TYPE(intrinsic-type-spec); multiple type-bound procedures can 4127 be declared in a single PROCEDURE statement; implied-shape 4128 arrays are supported for named constants (PARAMETER). The 4129 transformational, three argument versions of BESSEL_JN and 4130 BESSEL_YN were added the elemental, two-argument version had 4131 been added in GCC 4.4; note that the transformational 4132 functions use a recurrence algorithm. 4133 4134 Go 4135 4136 Support for the [19]Go programming language has been added to GCC. It 4137 is not enabled by default when you build GCC; use the 4138 --enable-languages configure option to build it. The driver program for 4139 compiling Go code is gccgo. 4140 4141 Go is currently known to work on GNU/Linux and RTEMS. Solaris support 4142 is in progress. It may or may not work on other platforms. 4143 4144 Objective-C and Objective-C++ 4145 4146 * The -fobjc-exceptions flag is now required to enable Objective-C 4147 exception and synchronization syntax (introduced by the keywords 4148 @try, @catch, @finally and @synchronized). 4149 * A number of Objective-C 2.0 features and extensions are now 4150 supported by GCC. These features are enabled by default; you can 4151 disable them by using the new -fobjc-std=objc1 command-line option. 4152 * The Objective-C 2.0 dot-syntax is now supported. It is an 4153 alternative syntax for using getters and setters; object.count is 4154 automatically converted into [object count] or [object setCount: 4155 ...] depending on context; for example if (object.count > 0) is 4156 automatically compiled into the equivalent of if ([object count] > 4157 0) while object.count = 0; is automatically compiled into the 4158 equivalent ot [object setCount: 0];. The dot-syntax can be used 4159 with instance and class objects and with any setters or getters, no 4160 matter if they are part of a declared property or not. 4161 * Objective-C 2.0 declared properties are now supported. They are 4162 declared using the new @property keyword, and are most commonly 4163 used in conjunction with the new Objective-C 2.0 dot-syntax. The 4164 nonatomic, readonly, readwrite, assign, retain, copy, setter and 4165 getter attributes are all supported. Marking declared properties 4166 with __attribute__ ((deprecated)) is supported too. 4167 * The Objective-C 2.0 @synthesize and @dynamic keywords are 4168 supported. @synthesize causes the compiler to automatically 4169 synthesize a declared property, while @dynamic is used to disable 4170 all warnings for a declared property for which no implementation is 4171 provided at compile time. Synthesizing declared properties requires 4172 runtime support in most useful cases; to be able to use it with the 4173 GNU runtime, appropriate helper functions have been added to the 4174 GNU Objective-C runtime ABI, and are implemented by the GNU 4175 Objective-C runtime library shipped with GCC. 4176 * The Objective-C 2.0 fast enumeration syntax is supported in 4177 Objective-C. This is currently not yet available in Objective-C++. 4178 Fast enumeration requires support in the runtime, and such support 4179 has been added to the GNU Objective-C runtime library (shipped with 4180 GCC). 4181 * The Objective-C 2.0 @optional keyword is supported. It allows you 4182 to mark methods or properties in a protocol as optional as opposed 4183 to required. 4184 * The Objective-C 2.0 @package keyword is supported. It has currently 4185 the same effect as the @public keyword. 4186 * Objective-C 2.0 method attributes are supported. Currently the 4187 supported attributes are deprecated, sentinel, noreturn and format. 4188 * Objective-C 2.0 method argument attributes are supported. The most 4189 widely used attribute is unused, to mark an argument as unused in 4190 the implementation. 4191 * Objective-C 2.0 class and protocol attributes are supported. 4192 Currently the only supported attribute is deprecated. 4193 * Objective-C 2.0 class extensions are supported. A class extension 4194 has the same syntax as a category declaration with no category 4195 name, and the methods and properties declared in it are added 4196 directly to the main class. It is mostly used as an alternative to 4197 a category to add methods to a class without advertising them in 4198 the public headers, with the advantage that for class extensions 4199 the compiler checks that all the privately declared methods are 4200 actually implemented. 4201 * As a result of these enhancements, GCC can now be used to build 4202 Objective-C and Objective-C++ software that uses Foundation and 4203 other important system frameworks with the NeXT runtime on Darwin 9 4204 and Darwin 10 (OSX 10.5 and 10.6). 4205 * Many bugs in the compiler have been fixed in this release; in 4206 particular, LTO can now be used when compiling Objective-C and 4207 Objective-C++ and the parser is much more robust in dealing with 4208 invalid code. 4209 4210 Runtime Library (libobjc) 4211 4212 * The GNU Objective-C runtime library now defines the macro 4213 __GNU_LIBOBJC__ (with a value that is increased at every release 4214 where there is any change to the API) in objc/objc.h, making it 4215 easy to determine if the GNU Objective-C runtime library is being 4216 used, and if so, which version. Previous versions of the GNU 4217 Objective-C runtime library (and other Objective-C runtime 4218 libraries such as the Apple one) do not define this macro. 4219 * A new Objective-C 2.0 API, almost identical to the one implemented 4220 by the Apple Objective-C runtime, has been implemented in the GNU 4221 Objective-C runtime library. The new API hides the internals of 4222 most runtime structures but provides a more extensive set of 4223 functions to operate on them. It is much easier, for example, to 4224 create or modify classes at runtime. The new API also makes it 4225 easier to port software from Apple to GNU as almost no changes 4226 should be required. The old API is still supported for backwards 4227 compatibility; including the old objc/objc-api.h header file 4228 automatically selects the old API, while including the new 4229 objc/runtime.h header file automatically selects the new API. 4230 Support for the old API is being phased out and upgrading the 4231 software to use the new API is strongly recommended. To check for 4232 the availability of the new API, the __GNU_LIBOBJC__ macro can be 4233 used as older versions of the GNU Objective-C runtime library, 4234 which do not support the new API, do not define such a macro. 4235 * Runtime support for @synchronized has been added. 4236 * Runtime support for Objective-C 2.0 synthesized property accessors 4237 has been added. 4238 * Runtime support for Objective-C 2.0 fast enumeration has been 4239 added. 4240 4241 New Targets and Target Specific Improvements 4242 4243 ARM 4244 4245 * GCC now supports the Cortex-M4 processor implementing the v7-em 4246 version of the architecture using the option -mcpu=cortex-m4. 4247 * Scheduling descriptions for the Cortex-M4, the Neon and the 4248 floating point units of the Cortex-A9 and a pipeline description 4249 for the Cortex-A5 have been added. 4250 * Synchronization primitives such as __sync_fetch_and_add and friends 4251 are now inlined for supported architectures rather than calling 4252 into a kernel helper function. 4253 * SSA loop prefetching is enabled by default for the Cortex-A9 at 4254 -O3. 4255 * Several improvements were committed to improve code generation for 4256 the ARM architecture including a rewritten implementation for load 4257 and store multiples. 4258 * Several enhancements were committed to improve SIMD code generation 4259 for NEON by adding support for widening instructions, misaligned 4260 loads and stores, vector conditionals and support for 64 bit 4261 arithmetic. 4262 * Support was added for the Faraday cores fa526, fa606te, fa626te, 4263 fmp626te, fmp626 and fa726te and can be used with the respective 4264 names as parameters to the -mcpu= option. 4265 * Basic support was added for Cortex-A15 and is available through 4266 -mcpu=cortex-a15. 4267 * GCC for AAPCS configurations now more closely adheres to the AAPCS 4268 specification by enabling -fstrict-volatile-bitfields by default. 4269 4270 IA-32/x86-64 4271 4272 * The new -fsplit-stack option permits programs to use a 4273 discontiguous stack. This is useful for threaded programs, in that 4274 it is no longer necessary to specify the maximum stack size when 4275 creating a thread. This feature is currently only implemented for 4276 32-bit and 64-bit x86 GNU/Linux targets. 4277 * Support for emitting profiler counter calls before function 4278 prologues. This is enabled via a new command-line option -mfentry. 4279 * Optimization for the Intel Core 2 processors is now available 4280 through the -march=core2 and -mtune=core2 options. 4281 * Support for Intel Core i3/i5/i7 processors is now available through 4282 the -march=corei7 and -mtune=corei7 options. 4283 * Support for Intel Core i3/i5/i7 processors with AVX is now 4284 available through the -march=corei7-avx and -mtune=corei7-avx 4285 options. 4286 * Support for AMD Bobcat (family 14) processors is now available 4287 through the -march=btver1 and -mtune=btver1 options. 4288 * Support for AMD Bulldozer (family 15) processors is now available 4289 through the -march=bdver1 and -mtune=bdver1 options. 4290 * The default setting (when not optimizing for size) for 32-bit 4291 GNU/Linux and Darwin x86 targets has been changed to 4292 -fomit-frame-pointer. The default can be reverted to 4293 -fno-omit-frame-pointer by configuring GCC with the 4294 --enable-frame-pointer configure option. 4295 * Darwin, FreeBSD, Solaris 2, MinGW and Cygwin now all support 4296 __float128 on 32-bit and 64-bit x86 targets. 4297 * AVX floating-point arithmetic can now be enabled by default at 4298 configure time with the new --with-fpmath=avx option. 4299 * The SSA loop prefetching pass is enabled when using -O3 when 4300 optimizing for CPUs where prefetching is beneficial (AMD CPUs newer 4301 than K6). 4302 * Support for TBM (Trailing Bit Manipulation) built-in functions and 4303 code generation is available via -mtbm. 4304 * Support for AMD's BMI (Bit Manipulation) built-in functions and 4305 code generation is available via -mbmi. 4306 4307 MicroBlaze 4308 4309 * Support has been added for the Xilinx MicroBlaze softcore processor 4310 (microblaze-elf) embedded target. This configurable processor is 4311 supported on several Xilinx Spartan and Virtex FPGAs. 4312 4313 MIPS 4314 4315 * GCC now supports the Loongson 3A processor. Its canonical -march= 4316 and -mtune= name is loongson3a. 4317 4318 MN10300 / AM33 4319 4320 * The inline assembly register constraint "A" has been renamed "c". 4321 This constraint is used to select a floating-point register that 4322 can be used as the destination of a multiply-accumulate 4323 instruction. 4324 * New inline assembly register constraints "A" and "D" have been 4325 added. These constraint letters resolve to all general registers 4326 when compiling for AM33, and resolve to address registers only or 4327 data registers only when compiling for MN10300. 4328 * The MDR register is represented in the compiler. One can access the 4329 register via the "z" constraint in inline assembly. It can be 4330 marked as clobbered or used as a local register variable via the 4331 "mdr" name. The compiler uses the RETF instruction if the function 4332 does not modify the MDR register, so it is important that inline 4333 assembly properly annotate any usage of the register. 4334 4335 PowerPC/PowerPC64 4336 4337 * GCC now supports the Applied Micro Titan processor with 4338 -mcpu=titan. 4339 * The -mrecip option has been added, which indicates whether the 4340 reciprocal and reciprocal square root instructions should be used. 4341 * The -mveclibabi=mass option can be used to enable the compiler to 4342 autovectorize mathematical functions using the Mathematical 4343 Acceleration Subsystem library. 4344 * The -msingle-pic-base option has been added, which instructs the 4345 compiler to avoid loading the PIC base register in function 4346 prologues. The PIC base register must be initialized by the runtime 4347 system. 4348 * The -mblock-move-inline-limit option has been added, which enables 4349 the user to control the maximum size of inlined memcpy calls and 4350 similar. 4351 * PowerPC64 GNU/Linux support for applications requiring a large TOC 4352 section has been improved. A new command-line option, 4353 -mcmodel=MODEL, controls this feature; valid values for MODEL are 4354 small, medium, or large. 4355 * The Altivec builtin functions vec_ld and vec_st have been modified 4356 to generate the Altivec memory instructions LVX and STVX, even if 4357 the -mvsx option is used. In the initial GCC 4.5 release, these 4358 builtin functions were changed to generate VSX memory reference 4359 instructions instead of Altivec memory instructions, but there are 4360 differences between the two instructions. If the VSX instruction 4361 set is available, you can now use the new builtin functions 4362 vec_vsx_ld and vec_vsx_st which always generates the VSX memory 4363 instructions. 4364 * The GCC compiler on AIX now defaults to a process layout with a 4365 larger data space allowing larger programs to be compiled. 4366 * The GCC long double type on AIX 6.1 and above has reverted to 64 4367 bit double precision, matching the AIX XL compiler default, because 4368 of missing C99 symbols required by the GCC runtime. 4369 * The default processor scheduling model and tuning for PowerPC64 4370 GNU/Linux and for AIX 6.1 and above now is POWER7. 4371 * Starting with GCC 4.6.1, vectors of type vector long long or vector 4372 long are passed and returned in the same method as other vectors 4373 with the VSX instruction set. Previously the GCC compiler did not 4374 adhere to the ABI for 128-bit vectors with 64-bit integer base 4375 types (PR 48857). This is also fixed in the GCC 4.5.4 release. 4376 4377 S/390, zSeries and System z9/z10, IBM zEnterprise z196 4378 4379 * Support for the zEnterprise z196 processor has been added. When 4380 using the -march=z196 option, the compiler will generate code 4381 making use of the following instruction facilities: 4382 + Conditional load/store 4383 + Distinct-operands 4384 + Floating-point-extension 4385 + Interlocked-access 4386 + Population-count 4387 The -mtune=z196 option avoids the compare and branch instructions 4388 as well as the load address instruction with an index register as 4389 much as possible and performs instruction scheduling appropriate 4390 for the new out-of-order pipeline architecture. 4391 * When using the -m31 -mzarch options the generated code still 4392 conforms to the 32-bit ABI but uses the general purpose registers 4393 as 64-bit registers internally. This requires a Linux kernel saving 4394 the whole 64-bit registers when doing a context switch. Kernels 4395 providing that feature indicate that by the 'highgprs' string in 4396 /proc/cpuinfo. 4397 * The SSA loop prefetching pass is enabled when using -O3. 4398 4399 SPARC 4400 4401 * GCC now supports the LEON series of SPARC V8 processors. The code 4402 generated by the compiler can either be tuned to it by means of the 4403 --with-tune=leon configure option and -mtune=leon compilation 4404 option, or the compiler can be built for the sparc-leon-{elf,linux} 4405 and sparc-leon3-{elf,linux} targets directly. 4406 * GCC has stopped sign/zero-extending parameter registers in the 4407 callee for functions taking parameters with sub-word size in 32-bit 4408 mode, since this is redundant with the specification of the ABI. 4409 GCC has never done so in 64-bit mode since this is also redundant. 4410 * The command-line option -mfix-at697f has been added to enable the 4411 documented workaround for the single erratum of the Atmel AT697F 4412 processor. 4413 4414 Operating Systems 4415 4416 Android 4417 4418 * GCC now supports the Bionic C library and provides a convenient way 4419 of building native libraries and applications for the Android 4420 platform. Refer to the documentation of the -mandroid and -mbionic 4421 options for details on building native code. At the moment, Android 4422 support is enabled only for ARM. 4423 4424 Darwin/Mac OS X 4425 4426 * General 4427 + Initial support for CFString types has been added. 4428 This allows GCC to build projects including the system Core 4429 Foundation frameworks. The GCC Objective-C family supports 4430 CFString "toll-free bridged" as per the Mac OS X system tools. 4431 CFString is also recognized in the context of format 4432 attributes and arguments (see the documentation for format 4433 attributes for limitations). At present, 8-bit character types 4434 are supported. 4435 + Object file size reduction. 4436 The Darwin zeroed memory allocators have been re-written to 4437 make more use of .zerofill sections. For non-debug code, this 4438 can reduce object file size significantly. 4439 + Objective-C family 64-bit support (NeXT ABI 2). 4440 Initial support has been added to support 64-bit Objective-C 4441 code using the Darwin/OS X native (NeXT) runtime. ABI version 4442 2 will be selected automatically when 64-bit code is built. 4443 + Objective-C family 32-bit ABI 1. 4444 For 32-bit code ABI 1 is also now also allowed. At present it 4445 must be selected manually using -fobjc-abi-version=1 where 4446 applicable - i.e. on Darwin 9/10 (OS X 10.5/10.6). 4447 * x86 Architecture 4448 + The -mdynamic-no-pic option has been enabled. 4449 Code supporting -mdynamic-no-pic optimization has been added 4450 and is applicable to -m32 builds. The compiler bootstrap uses 4451 the option where appropriate. 4452 + The default value for -mtune= has been changed. 4453 Since Darwin systems are primarily Xeon, Core-2 or similar the 4454 default tuning has been changed to -mtune=core2. 4455 + Enable 128-bit long double (__float128) support on Darwin. 4456 * PPC Architecture 4457 + Darwin64 ABI. 4458 Several significant bugs have been fixed, such that GCC now 4459 produces code compatible with the Darwin64 PowerPC ABI. 4460 + libffi and boehm-gc. 4461 The Darwin ports of the libffi and boehm-gc libraries have 4462 been upgraded to include a Darwin64 implementation. This means 4463 that powerpc*-*-darwin9 platforms may now, for example, build 4464 Java applications with -m64 enabled. 4465 + Plug-in support has been enabled. 4466 + The -fsection-anchors option is now available although, 4467 presently, not heavily tested. 4468 4469 Solaris 2 4470 4471 New Features 4472 4473 * Support symbol versioning with the Sun linker. 4474 * Allow libstdc++ to leverage full ISO C99 support on Solaris 10+. 4475 * Support thread-local storage (TLS) with the Sun assembler on 4476 Solaris 2/x86. 4477 * Support TLS on Solaris 8/9 if prerequisites are met. 4478 * Support COMDAT group with the GNU assembler and recent Sun linker. 4479 * Support the Sun assembler visibility syntax. 4480 * Default Solaris 2/x86 to -march=pentium4 (Solaris 10+) resp. 4481 -march=pentiumpro (Solaris 8/9). 4482 * Don't use SSE on Solaris 8/9 x86 by default. 4483 * Enable 128-bit long double (__float128) support on Solaris 2/x86. 4484 4485 ABI Change 4486 4487 * Change the ABI for returning 8-byte vectors like __m64 in MMX 4488 registers on Solaris 10+/x86 to match the Sun Studio 12.1+ 4489 compilers. This is an incompatible change. If you use such types, 4490 you must either recompile all your code with the new compiler or 4491 use the new -mvect8-ret-in-mem option to remain compatible with 4492 previous versions of GCC and Sun Studio. 4493 4494 Windows x86/x86_64 4495 4496 * Initial support for decimal floating point. 4497 * Support for the __thiscall calling-convention. 4498 * Support for hot-patchable function prologues via the 4499 ms_hook_prologue attribute for x86_64 in addition to 32-bit x86. 4500 * Improvements of stack-probing and stack-allocation mechanisms. 4501 * Support of push/pop-macro pragma as preprocessor command. 4502 With #pragma push_macro("macro-name") the current definition of 4503 macro-name is saved and can be restored with #pragma 4504 pop_macro("macro-name") to its saved definition. 4505 * Enable 128-bit long double (__float128) support on MinGW and 4506 Cygwin. 4507 4508 Other significant improvements 4509 4510 Installation changes 4511 4512 * An install-strip make target is provided that installs stripped 4513 executables, and may install libraries with unneeded or debugging 4514 sections stripped. 4515 * On Power7 systems, there is a potential problem if you build the 4516 GCC compiler with a host compiler using options that enable the VSX 4517 instruction set generation. If the host compiler has been patched 4518 so that the vec_ld and vec_st builtin functions generate Altivec 4519 memory instructions instead of VSX memory instructions, then you 4520 should be able to build the compiler with VSX instruction 4521 generation. 4522 4523 Changes for GCC Developers 4524 4525 Note: these changes concern developers that develop GCC itself or 4526 software that integrates with GCC, such as plugins, and not the general 4527 GCC users. 4528 * The gengtype utility, which previously was internal to the GCC 4529 build process, has been enchanced to provide GC root information 4530 for plugins as necessary. 4531 * The old GC allocation interface of ggc_alloc and friends was 4532 replaced with a type-safe alternative. 4533 4534 GCC 4.6.1 4535 4536 This is the [20]list of problem reports (PRs) from GCC's bug tracking 4537 system that are known to be fixed in the 4.6.1 release. This list might 4538 not be complete (that is, it is possible that some PRs that have been 4539 fixed are not listed here). 4540 4541 GCC 4.6.2 4542 4543 This is the [21]list of problem reports (PRs) from GCC's bug tracking 4544 system that are known to be fixed in the 4.6.2 release. This list might 4545 not be complete (that is, it is possible that some PRs that have been 4546 fixed are not listed here). 4547 4548 GCC 4.6.3 4549 4550 This is the [22]list of problem reports (PRs) from GCC's bug tracking 4551 system that are known to be fixed in the 4.6.3 release. This list might 4552 not be complete (that is, it is possible that some PRs that have been 4553 fixed are not listed here). 4554 4555 GCC 4.6.4 4556 4557 This is the [23]list of problem reports (PRs) from GCC's bug tracking 4558 system that are known to be fixed in the 4.6.4 release. This list might 4559 not be complete (that is, it is possible that some PRs that have been 4560 fixed are not listed here). 4561 4562 4563 For questions related to the use of GCC, please consult these web 4564 pages and the [24]GCC manuals. If that fails, the 4565 [25]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 4566 web pages and the development of GCC are welcome on our developer 4567 list at [26]gcc (a] gcc.gnu.org. All of [27]our lists have public 4568 archives. 4569 4570 Copyright (C) [28]Free Software Foundation, Inc. Verbatim copying and 4571 distribution of this entire article is permitted in any medium, 4572 provided this notice is preserved. 4573 4574 These pages are [29]maintained by the GCC team. Last modified 4575 2016-11-27[30]. 4576 4577 References 4578 4579 1. https://sourceware.org/bugzilla/show_bug.cgi?id=10401 4580 2. http://gcc.gnu.org/gcc-4.5/changes.html 4581 3. http://gcc.gnu.org/gcc-4.5/changes.html#obsoleted 4582 4. http://gcc.gnu.org/gcc-4.6/porting_to.html 4583 5. http://gcc.gnu.org/projects/lto/whopr.pdf 4584 6. https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options 4585 7. http://gcc.gnu.org/gcc-4.6/cxx0x_status.html 4586 8. https://gcc.gnu.org/PR43145 4587 9. https://gcc.gnu.org/PR43680 4588 10. https://gcc.gnu.org/PR33558 4589 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_active.html#253 4590 12. https://gcc.gnu.org/onlinedocs/gcc-4.6.4/libstdc++/manual/manual/status.html#status.iso.200x 4591 13. https://gcc.gnu.org/onlinedocs/libstdc++/manual/debug_mode.html 4592 14. https://gcc.gnu.org/onlinedocs/libstdc++/manual/debug.html#debug.races 4593 15. https://gcc.gnu.org/onlinedocs/gcc/Preprocessor-Options.html 4594 16. https://gcc.gnu.org/wiki/OOP 4595 17. https://gcc.gnu.org/wiki/Coarray 4596 18. https://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcoarray_007d-233 4597 19. https://golang.org/ 4598 20. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.1 4599 21. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.2 4600 22. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.3 4601 23. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.4 4602 24. https://gcc.gnu.org/onlinedocs/ 4603 25. mailto:gcc-help (a] gcc.gnu.org 4604 26. mailto:gcc (a] gcc.gnu.org 4605 27. https://gcc.gnu.org/lists.html 4606 28. http://www.fsf.org/ 4607 29. https://gcc.gnu.org/about.html 4608 30. http://validator.w3.org/check/referer 4609 ====================================================================== 4610 http://gcc.gnu.org/gcc-4.5/index.html 4611 GCC 4.5 Release Series 4612 4613 Jul 2, 2012 4614 4615 The [1]GNU project and the GCC developers are pleased to announce the 4616 release of GCC 4.5.4. 4617 4618 This release is a bug-fix release, containing fixes for regressions in 4619 GCC 4.5.3 relative to previous releases of GCC. 4620 4621 This release series is no longer maintained. 4622 4623 Release History 4624 4625 GCC 4.5.4 4626 Jul 2, 2012 ([2]changes) 4627 4628 GCC 4.5.3 4629 Apr 28, 2011 ([3]changes) 4630 4631 GCC 4.5.2 4632 Dec 16, 2010 ([4]changes) 4633 4634 GCC 4.5.1 4635 Jul 31, 2010 ([5]changes) 4636 4637 GCC 4.5.0 4638 April 14, 2010 ([6]changes) 4639 4640 References and Acknowledgements 4641 4642 GCC used to stand for the GNU C Compiler, but since the compiler 4643 supports several other languages aside from C, it now stands for the 4644 GNU Compiler Collection. 4645 4646 A list of [7]successful builds is updated as new information becomes 4647 available. 4648 4649 The GCC developers would like to thank the numerous people that have 4650 contributed new features, improvements, bug fixes, and other changes as 4651 well as test results to GCC. This [8]amazing group of volunteers is 4652 what makes GCC successful. 4653 4654 For additional information about GCC please refer to the [9]GCC project 4655 web site or contact the [10]GCC development mailing list. 4656 4657 To obtain GCC please use [11]our mirror sites or [12]our SVN server. 4658 4659 4660 For questions related to the use of GCC, please consult these web 4661 pages and the [13]GCC manuals. If that fails, the 4662 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 4663 web pages and the development of GCC are welcome on our developer 4664 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 4665 archives. 4666 4667 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 4668 distribution of this entire article is permitted in any medium, 4669 provided this notice is preserved. 4670 4671 These pages are [18]maintained by the GCC team. Last modified 4672 2016-09-30[19]. 4673 4674 References 4675 4676 1. http://www.gnu.org/ 4677 2. http://gcc.gnu.org/gcc-4.5/changes.html 4678 3. http://gcc.gnu.org/gcc-4.5/changes.html 4679 4. http://gcc.gnu.org/gcc-4.5/changes.html 4680 5. http://gcc.gnu.org/gcc-4.5/changes.html 4681 6. http://gcc.gnu.org/gcc-4.5/changes.html 4682 7. http://gcc.gnu.org/gcc-4.5/buildstat.html 4683 8. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 4684 9. http://gcc.gnu.org/index.html 4685 10. mailto:gcc (a] gcc.gnu.org 4686 11. http://gcc.gnu.org/mirrors.html 4687 12. http://gcc.gnu.org/svn.html 4688 13. https://gcc.gnu.org/onlinedocs/ 4689 14. mailto:gcc-help (a] gcc.gnu.org 4690 15. mailto:gcc (a] gcc.gnu.org 4691 16. https://gcc.gnu.org/lists.html 4692 17. http://www.fsf.org/ 4693 18. https://gcc.gnu.org/about.html 4694 19. http://validator.w3.org/check/referer 4695 ====================================================================== 4696 http://gcc.gnu.org/gcc-4.5/changes.html 4697 GCC 4.5 Release Series 4698 Changes, New Features, and Fixes 4699 4700 Caveats 4701 4702 * GCC now requires the [1]MPC library in order to build. See the 4703 [2]prerequisites page for version requirements. 4704 * Support for a number of older systems and recently unmaintained or 4705 untested target ports of GCC has been declared obsolete in GCC 4.5. 4706 Unless there is activity to revive them, the next release of GCC 4707 will have their sources permanently removed. 4708 The following ports for individual systems on particular 4709 architectures have been obsoleted: 4710 + IRIX releases before 6.5 (mips-sgi-irix5*, 4711 mips-sgi-irix6.[0-4]) 4712 + Solaris 7 (*-*-solaris2.7) 4713 + Tru64 UNIX releases before V5.1 (alpha*-dec-osf4*, 4714 alpha-dec-osf5.0*) 4715 + Details for the IRIX, Solaris 7, and Tru64 UNIX obsoletions 4716 can be found in the [3]announcement. 4717 Support for the classic POWER architecture implemented in the 4718 original RIOS and RIOS2 processors of the old IBM RS/6000 product 4719 line has been obsoleted in the rs6000 port. This does not affect 4720 the new generation Power and PowerPC architectures. 4721 * Support has been removed for all the [4]configurations obsoleted in 4722 GCC 4.4. 4723 * Support has been removed for the protoize and unprotoize utilities, 4724 obsoleted in GCC 4.4. 4725 * Support has been removed for tuning for Itanium1 (Merced) variants. 4726 Note that code tuned for Itanium2 should also run correctly on 4727 Itanium1. 4728 * GCC now generates unwind info also for epilogues. DWARF debuginfo 4729 generated by GCC now uses more features of DWARF3 than before, and 4730 also some DWARF4 features. GDB older than 7.0 is not able to handle 4731 either of these, so to debug GCC 4.5 generated binaries or 4732 libraries GDB 7.0 or later is needed. You can disable use of DWARF4 4733 features with the -gdwarf-3 -gstrict-dwarf options, or use 4734 -gdwarf-2 -gstrict-dwarf to restrict GCC to just DWARF2, but 4735 epilogue unwind info is emitted unconditionally whenever unwind 4736 info is emitted. 4737 * On x86 targets, code containing floating-point calculations may run 4738 significantly more slowly when compiled with GCC 4.5 in strict C99 4739 conformance mode than they did with earlier GCC versions. This is 4740 due to stricter standard conformance of the compiler and can be 4741 avoided by using the option -fexcess-precision=fast; also see 4742 [5]below. 4743 * The function attribute noinline no longer prevents GCC from cloning 4744 the function. A new attribute noclone has been introduced for this 4745 purpose. Cloning a function means that it is duplicated and the new 4746 copy is specialized for certain contexts (for example when a 4747 parameter is a known constant). 4748 4749 General Optimizer Improvements 4750 4751 * The -save-temps now takes an optional argument. The -save-temps and 4752 -save-temps=cwd switches write the temporary files in the current 4753 working directory based on the original source file. The 4754 -save-temps=obj switch will write files into the directory 4755 specified with the -o option, and the intermediate filenames are 4756 based on the output file. This will allow the user to get the 4757 compiler intermediate files when doing parallel builds without two 4758 builds of the same filename located in different directories from 4759 interfering with each other. 4760 * Debugging dumps are now created in the same directory as the object 4761 file rather than in the current working directory. This allows the 4762 user to get debugging dumps when doing parallel builds without two 4763 builds of the same filename interfering with each other. 4764 * GCC has been integrated with the [6]MPC library. This allows GCC to 4765 evaluate complex arithmetic at compile time [7]more accurately. It 4766 also allows GCC to evaluate calls to complex built-in math 4767 functions having constant arguments and replace them at compile 4768 time with their mathematically equivalent results. In doing so, GCC 4769 can generate correct results regardless of the math library 4770 implementation or floating point precision of the host platform. 4771 This also allows GCC to generate identical results regardless of 4772 whether one compiles in native or cross-compile configurations to a 4773 particular target. The following built-in functions take advantage 4774 of this new capability: cacos, cacosh, casin, casinh, catan, 4775 catanh, ccos, ccosh, cexp, clog, cpow, csin, csinh, csqrt, ctan, 4776 and ctanh. The float and long double variants of these functions 4777 (e.g. csinf and csinl) are also handled. 4778 * A new link-time optimizer has been added ([8]-flto). When this 4779 option is used, GCC generates a bytecode representation of each 4780 input file and writes it to specially-named sections in each object 4781 file. When the object files are linked together, all the function 4782 bodies are read from these named sections and instantiated as if 4783 they had been part of the same translation unit. This enables 4784 interprocedural optimizations to work across different files (and 4785 even different languages), potentially improving the performance of 4786 the generated code. To use the link-timer optimizer, -flto needs to 4787 be specified at compile time and during the final link. If the 4788 program does not require any symbols to be exported, it is possible 4789 to combine -flto and the experimental [9]-fwhopr with 4790 [10]-fwhole-program to allow the interprocedural optimizers to use 4791 more aggressive assumptions. 4792 * The automatic parallelization pass was enhanced to support 4793 parallelization of outer loops. 4794 * Automatic parallelization can be enabled as part of Graphite. In 4795 addition to -ftree-parallelize-loops=, specify 4796 -floop-parallelize-all to enable the Graphite-based optimization. 4797 * The infrastructure for optimizing based on [11]restrict qualified 4798 pointers has been rewritten and should result in code generation 4799 improvements. Optimizations based on restrict qualified pointers 4800 are now also available when using -fno-strict-aliasing. 4801 * There is a new optimization pass that attempts to change prototype 4802 of functions to avoid unused parameters, pass only relevant parts 4803 of structures and turn arguments passed by reference to arguments 4804 passed by value when possible. It is enabled by -O2 and above as 4805 well as -Os and can be manually invoked using the new command-line 4806 switch -fipa-sra. 4807 * GCC now optimize exception handling code. In particular cleanup 4808 regions that are proved to not have any effect are optimized out. 4809 4810 New Languages and Language specific improvements 4811 4812 All languages 4813 4814 * The -fshow-column option is now on by default. This means error 4815 messages now have a column associated with them. 4816 4817 Ada 4818 4819 * Compilation of programs heavily using discriminated record types 4820 with variant parts has been sped up and generates more compact 4821 code. 4822 * Stack checking now works reasonably well on most plaforms. In some 4823 specific cases, stack overflows may still fail to be detected, but 4824 a compile-time warning will be issued for these cases. 4825 4826 C family 4827 4828 * If a header named in a #include directive is not found, the 4829 compiler exits immediately. This avoids a cascade of errors arising 4830 from declarations expected to be found in that header being 4831 missing. 4832 * A new built-in function __builtin_unreachable() has been added that 4833 tells the compiler that control will never reach that point. It may 4834 be used after asm statements that terminate by transferring control 4835 elsewhere, and in other places that are known to be unreachable. 4836 * The -Wlogical-op option now warns for logical expressions such as 4837 (c == 1 && c == 2) and (c != 1 || c != 2), which are likely to be 4838 mistakes. This option is disabled by default. 4839 * An asm goto feature has been added to allow asm statements that 4840 jump to C labels. 4841 * C++0x raw strings are supported for C++ and for C with -std=gnu99. 4842 * The deprecated attribute now takes an optional string argument, for 4843 example, __attribute__((deprecated("text string"))), that will be 4844 printed together with the deprecation warning. 4845 4846 C 4847 4848 * The -Wenum-compare option, which warns when comparing values of 4849 different enum types, now works for C. It formerly only worked for 4850 C++. This warning is enabled by -Wall. It may be avoided by using a 4851 type cast. 4852 * The -Wcast-qual option now warns about casts which are unsafe in 4853 that they permit const-correctness to be violated without further 4854 warnings. Specifically, it warns about cases where a qualifier is 4855 added when all the lower types are not const. For example, it warns 4856 about a cast from char ** to const char **. 4857 * The -Wc++-compat option is significantly improved. It issues new 4858 warnings for: 4859 + Using C++ reserved operator names as identifiers. 4860 + Conversions to enum types without explicit casts. 4861 + Using va_arg with an enum type. 4862 + Using different enum types in the two branches of ?:. 4863 + Using ++ or -- on a variable of enum type. 4864 + Using the same name as both a struct, union or enum tag and a 4865 typedef, unless the typedef refers to the tagged type itself. 4866 + Using a struct, union, or enum which is defined within another 4867 struct or union. 4868 + A struct field defined using a typedef if there is a field in 4869 the struct, or an enclosing struct, whose name is the typedef 4870 name. 4871 + Duplicate definitions at file scope. 4872 + Uninitialized const variables. 4873 + A global variable with an anonymous struct, union, or enum 4874 type. 4875 + Using a string constant to initialize a char array whose size 4876 is the length of the string. 4877 * The new -Wjump-misses-init option warns about cases where a goto or 4878 switch skips the initialization of a variable. This sort of branch 4879 is an error in C++ but not in C. This warning is enabled by 4880 -Wc++-compat. 4881 * GCC now ensures that a C99-conforming <stdint.h> is present on most 4882 targets, and uses information about the types in this header to 4883 implement the Fortran bindings to those types. GCC does not ensure 4884 the presence of such a header, and does not implement the Fortran 4885 bindings, on the following targets: NetBSD, VxWorks, VMS, 4886 SymbianOS, WinCE, LynxOS, Netware, QNX, Interix, TPF. 4887 * GCC now implements C90- and C99-conforming rules for constant 4888 expressions. This may cause warnings or errors for some code using 4889 expressions that can be folded to a constant but are not constant 4890 expressions as defined by ISO C. 4891 * All known target-independent C90 and C90 Amendment 1 conformance 4892 bugs, and all known target-independent C99 conformance bugs not 4893 related to floating point or extended identifiers, have been fixed. 4894 * The C decimal floating point support now includes support for the 4895 FLOAT_CONST_DECIMAL64 pragma. 4896 * The named address space feature from ISO/IEC TR 18037 is now 4897 supported. This is currently only implemented for the SPU 4898 processor. 4899 4900 C++ 4901 4902 * Improved [12]experimental support for the upcoming C++0x ISO C++ 4903 standard, including support for raw strings, lambda expressions and 4904 explicit type conversion operators. 4905 * When printing the name of a class template specialization, G++ will 4906 now omit any template arguments which come from default template 4907 arguments. This behavior (and the pretty-printing of function 4908 template specializations as template signature and arguments) can 4909 be disabled with the -fno-pretty-templates option. 4910 * Access control is now applied to typedef names used in a template, 4911 which may cause G++ to reject some ill-formed code that was 4912 accepted by earlier releases. The -fno-access-control option can be 4913 used as a temporary workaround until the code is corrected. 4914 * Compilation time for code that uses templates should now scale 4915 linearly with the number of instantiations rather than 4916 quadratically, as template instantiations are now looked up using 4917 hash tables. 4918 * Declarations of functions that look like builtin declarations of 4919 library functions are only considered to be redeclarations if they 4920 are declared with extern "C". This may cause problems with code 4921 that omits extern "C" on hand-written declarations of C library 4922 functions such as abort or memcpy. Such code is ill-formed, but was 4923 accepted by earlier releases. 4924 * Diagnostics that used to complain about passing non-POD types to 4925 ... or jumping past the declaration of a non-POD variable now check 4926 for triviality rather than PODness, as per C++0x. 4927 * In C++0x mode local and anonymous classes are now allowed as 4928 template arguments, and in declarations of variables and functions 4929 with linkage, so long as any such declaration that is used is also 4930 defined ([13]DR 757). 4931 * Labels may now have attributes, as has been permitted for a while 4932 in C. This is only permitted when the label definition and the 4933 attribute specifier is followed by a semicoloni.e., the label 4934 applies to an empty statement. The only useful attribute for a 4935 label is unused. 4936 * G++ now implements [14]DR 176. Previously G++ did not support using 4937 the injected-class-name of a template base class as a type name, 4938 and lookup of the name found the declaration of the template in the 4939 enclosing scope. Now lookup of the name finds the 4940 injected-class-name, which can be used either as a type or as a 4941 template, depending on whether or not the name is followed by a 4942 template argument list. As a result of this change, some code that 4943 was previously accepted may be ill-formed because 4944 1. The injected-class-name is not accessible because it's from a 4945 private base, or 4946 2. The injected-class-name cannot be used as an argument for a 4947 template template parameter. 4948 In either of these cases, the code can be fixed by adding a 4949 nested-name-specifier to explicitly name the template. The first 4950 can be worked around with -fno-access-control; the second is only 4951 rejected with -pedantic. 4952 * A new standard mangling for SIMD vector types has been added, to 4953 avoid name clashes on systems with vectors of varying length. By 4954 default the compiler still uses the old mangling, but emits aliases 4955 with the new mangling on targets that support strong aliases. Users 4956 can switch over entirely to the new mangling with -fabi-version=4 4957 or -fabi-version=0. -Wabi will now warn about code that uses the 4958 old mangling. 4959 * The command-line option -ftemplate-depth-N is now written as 4960 -ftemplate-depth=N and the old form is deprecated. 4961 * Conversions between NULL and non-pointer types are now warned by 4962 default. The new option -Wno-conversion-null disables these 4963 warnings. Previously these warnings were only available when using 4964 -Wconversion explicitly. 4965 4966 Runtime Library (libstdc++) 4967 4968 * Improved experimental support for the upcoming ISO C++ standard, 4969 C++0x, including: 4970 + Support for <future>, <functional>, and <random>. 4971 + Existing facilities now exploit explicit operators and the 4972 newly implemented core C++0x features. 4973 + The header <cstdatomic> has been renamed to <atomic>. 4974 * An experimental [15]profile mode has been added. This is an 4975 implementation of many C++ standard library constructs with an 4976 additional analysis layer that gives performance improvement advice 4977 based on recognition of suboptimal usage patterns. For example, 4978 #include <vector> 4979 int main() 4980 { 4981 std::vector<int> v; 4982 for (int k = 0; k < 1024; ++k) 4983 v.insert(v.begin(), k); 4984 } 4985 4986 When instrumented via the profile mode, can return suggestions 4987 about the initial size and choice of the container used as follows: 4988 vector-to-list: improvement = 5: call stack = 0x804842c ... 4989 : advice = change std::vector to std::list 4990 vector-size: improvement = 3: call stack = 0x804842c ... 4991 : advice = change initial container size from 0 to 1024 4992 4993 These constructs can be substituted for the normal libstdc++ 4994 constructs on a piecemeal basis, or all existing components can be 4995 transformed via the -D_GLIBCXX_PROFILE macro. 4996 * [16]Support for decimal floating-point arithmetic (aka ISO C++ TR 4997 24733) has been added. This support is in header file 4998 <decimal/decimal>, uses namespace std::decimal, and includes 4999 classes decimal32, decimal64, and decimal128. 5000 * Sources have been audited for application of function attributes 5001 nothrow, const, pure, and noreturn. 5002 * Python pretty-printers have been added for many standard library 5003 components that simplify the internal representation and present a 5004 more intuitive view of components when used with 5005 appropriately-advanced versions of GDB. For more information, 5006 please consult the more [17]detailed description. 5007 * The default behavior for comparing typeinfo names has changed, so 5008 in <typeinfo>, __GXX_MERGED_TYPEINFO_NAMES now defaults to zero. 5009 * The new -static-libstdc++ option directs g++ to link the C++ 5010 library statically, even if the default would normally be to link 5011 it dynamically. 5012 5013 Fortran 5014 5015 * The COMMON default padding has been changed instead of adding the 5016 padding before a variable it is now added afterwards, which 5017 increases the compatibility with other vendors and helps to obtain 5018 the correct output in some cases. Cf. also the -falign-commons 5019 option ([18]added in 4.4). 5020 * The -finit-real= option now also supports the value snan for 5021 signaling not-a-number; to be effective, one additionally needs to 5022 enable trapping (e.g. via -ffpe-trap=). Note: Compile-time 5023 optimizations can turn a signaling NaN into a quiet one. 5024 * The new option -fcheck= has been added with the options bounds, 5025 array-temps, do, pointer, and recursive. The bounds and array-temps 5026 options are equivalent to -fbounds-check and 5027 -fcheck-array-temporaries. The do option checks for invalid 5028 modification of loop iteration variables, and the recursive option 5029 tests for recursive calls to subroutines/functions which are not 5030 marked as recursive. With pointer pointer association checks in 5031 calls are performed; however, neither undefined pointers nor 5032 pointers in expressions are handled. Using -fcheck=all enables all 5033 these run-time checks. 5034 * The run-time checking -fcheck=bounds now warns about invalid string 5035 lengths of character dummy arguments. Additionally, more 5036 compile-time checks have been added. 5037 * The new option [19]-fno-protect-parens has been added; if set, the 5038 compiler may reorder REAL and COMPLEX expressions without regard to 5039 parentheses. 5040 * GNU Fortran no longer links against libgfortranbegin. As before, 5041 MAIN__ (assembler symbol name) is the actual Fortran main program, 5042 which is invoked by the main function. However, main is now 5043 generated and put in the same object file as MAIN__. For the time 5044 being, libgfortranbegin still exists for backward compatibility. 5045 For details see the new [20]Mixed-Language Programming chapter in 5046 the manual. 5047 * The I/O library was restructured for performance and cleaner code. 5048 * Array assignments and WHERE are now run in parallel when OpenMP's 5049 WORKSHARE is used. 5050 * The experimental option -fwhole-file was added. The option allows 5051 whole-file checking of procedure arguments and allows for better 5052 optimizations. It can also be used with -fwhole-program, which is 5053 now also supported in gfortran. 5054 * More Fortran 2003 and Fortran 2008 mathematical functions can now 5055 be used as initialization expressions. 5056 * Some extended attributes such as STDCALL are now supported via the 5057 [21]GCC$ compiler directive. 5058 * For Fortran 77 compatibility: If -fno-sign-zero is used, the SIGN 5059 intrinsic behaves now as if zero were always positive. 5060 * For legacy compatibiliy: On Cygwin and MinGW, the special files 5061 CONOUT$ and CONIN$ (and CONERR$ which maps to CONOUT$) are now 5062 supported. 5063 * Fortran 2003 support has been extended: 5064 + Procedure-pointer function results and procedure-pointer 5065 components (including PASS), 5066 + allocatable scalars (experimental), 5067 + DEFERRED type-bound procedures, 5068 + the ERRMSG= argument of the ALLOCATE and DEALLOCATE statements 5069 have been implemented. 5070 + The ALLOCATE statement supports type-specs and the SOURCE= 5071 argument. 5072 + OPERATOR(*) and ASSIGNMENT(=) are now allowed as GENERIC 5073 type-bound procedure (i.e. as type-bound operators). 5074 + Rounding (ROUND=, RZ, ...) for output is now supported. 5075 + The INT_FAST{8,16,32,64,128}_T kind type parameters of the 5076 intrinsic module ISO_C_BINDING are now supported, except for 5077 the targets listed above as ones where GCC does not have 5078 <stdint.h> type information. 5079 + Extensible derived types with type-bound procedure or 5080 procedure pointer with PASS attribute now have to use CLASS in 5081 line with the Fortran 2003 standard; the workaround to use 5082 TYPE is no longer supported. 5083 + [22]Experimental, incomplete support for polymorphism, 5084 including CLASS, SELECT TYPE and dynamic dispatch of 5085 type-bound procedure calls. Some features do not work yet such 5086 as unlimited polymorphism (CLASS(*)). 5087 * Fortran 2008 support has been extended: 5088 + The OPEN statement now supports the NEWUNIT= option, which 5089 returns a unique file unit, thus preventing inadvertent use of 5090 the same unit in different parts of the program. 5091 + Support for unlimited format items has been added. 5092 + The INT{8,16,32} and REAL{32,64,128} kind type parameters of 5093 the intrinsic module ISO_FORTRAN_ENV are now supported. 5094 + Using complex arguments with TAN, SINH, COSH, TANH, ASIN, 5095 ACOS, and ATAN is now possible; the functions ASINH, ACOSH, 5096 and ATANH have been added (for real and complex arguments) and 5097 ATAN(Y,X) is now an alias for ATAN2(Y,X). 5098 + The BLOCK construct has been implemented. 5099 5100 New Targets and Target Specific Improvements 5101 5102 AIX 5103 5104 * Full cross-toolchain support now available with GNU Binutils 5105 5106 ARM 5107 5108 * GCC now supports the Cortex-M0 and Cortex-A5 processors. 5109 * GCC now supports the ARM v7E-M architecture. 5110 * GCC now supports VFPv4-based FPUs and FPUs with 5111 single-precision-only VFP. 5112 * GCC has many improvements to optimization for other ARM processors, 5113 including scheduling support for the integer pipeline on Cortex-A9. 5114 * GCC now supports the IEEE 754-2008 half-precision floating-point 5115 type, and a variant ARM-specific half-precision type. This type is 5116 specified using __fp16, with the layout determined by 5117 -mfp16-format. With appropriate -mfpu options, the Cortex-A9 and 5118 VFPv4 half-precision instructions will be used. 5119 * GCC now supports the variant of AAPCS that uses VFP registers for 5120 parameter passing and return values. 5121 5122 AVR 5123 5124 * The -mno-tablejump option has been removed because it has the same 5125 effect as the -fno-jump-tables option. 5126 * Added support for these new AVR devices: 5127 + ATmega8U2 5128 + ATmega16U2 5129 + ATmega32U2 5130 5131 IA-32/x86-64 5132 5133 * GCC now will set the default for -march= based on the configure 5134 target. 5135 * GCC now supports handling floating-point excess precision arising 5136 from use of the x87 floating-point unit in a way that conforms to 5137 ISO C99. This is enabled with -fexcess-precision=standard and with 5138 standards conformance options such as -std=c99, and may be disabled 5139 using -fexcess-precision=fast. 5140 * Support for the Intel Atom processor is now available through the 5141 -march=atom and -mtune=atom options. 5142 * A new -mcrc32 option is now available to enable crc32 intrinsics. 5143 * A new -mmovbe option is now available to enable GCC to use the 5144 movbe instruction to implement __builtin_bswap32 and 5145 __builtin_bswap64. 5146 * SSE math now can be enabled by default at configure time with the 5147 new --with-fpmath=sse option. 5148 * There is a new intrinsic header file, <x86intrin.h>. It should be 5149 included before using any IA-32/x86-64 intrinsics. 5150 * Support for the XOP, FMA4, and LWP instruction sets for the AMD 5151 Orochi processors are now available with the -mxop, -mfma4, and 5152 -mlwp options. 5153 * The -mabm option enables GCC to use the popcnt and lzcnt 5154 instructions on AMD processors. 5155 * The -mpopcnt option enables GCC to use the popcnt instructions on 5156 both AMD and Intel processors. 5157 5158 M68K/ColdFire 5159 5160 * GCC now supports ColdFire 51xx, 5221x, 5225x, 52274, 52277, 5301x 5161 and 5441x devices. 5162 * GCC now supports thread-local storage (TLS) on M68K and ColdFire 5163 processors. 5164 5165 MeP 5166 5167 Support has been added for the Toshiba Media embedded Processor (MeP, 5168 or mep-elf) embedded target. 5169 5170 MIPS 5171 5172 * GCC now supports MIPS 1004K processors. 5173 * GCC can now be configured with options --with-arch-32, 5174 --with-arch-64, --with-tune-32 and --with-tune-64 to control the 5175 default optimization separately for 32-bit and 64-bit modes. 5176 * MIPS targets now support an alternative _mcount interface, in which 5177 register $12 points to the function's save slot for register $31. 5178 This interface is selected by the -mcount-ra-address option; see 5179 the documentation for more details. 5180 * GNU/Linux targets can now generate read-only .eh_frame sections. 5181 This optimization requires GNU binutils 2.20 or above, and is only 5182 available if GCC is configured with a suitable version of binutils. 5183 * GNU/Linux targets can now attach special relocations to indirect 5184 calls, so that the linker can turn them into direct jumps or 5185 branches. This optimization requires GNU binutils 2.20 or later, 5186 and is automatically selected if GCC is configured with an 5187 appropriate version of binutils. It can be explicitly enabled or 5188 disabled using the -mrelax-pic-calls command-line option. 5189 * GCC now generates more heavily-optimized atomic operations on 5190 Octeon processors. 5191 * MIPS targets now support the -fstack-protector option. 5192 * GCC now supports an -msynci option, which specifies that synci is 5193 enough to flush the instruction cache, without help from the 5194 operating system. GCC uses this information to optimize 5195 automatically-generated cache flush operations, such as those used 5196 for nested functions in C. There is also a --with-synci 5197 configure-time option, which makes -msynci the default. 5198 * GCC supports four new function attributes for interrupt handlers: 5199 interrupt, use_shadow_register_set, keep_interrupts_masked and 5200 use_debug_exception_return. See the documentation for more details 5201 about these attributes. 5202 5203 RS/6000 (POWER/PowerPC) 5204 5205 * GCC now supports the Power ISA 2.06, which includes the VSX 5206 instructions that add vector 64-bit floating point support, new 5207 population count instructions, and conversions between floating 5208 point and unsigned types. 5209 * Support for the power7 processor is now available through the 5210 -mcpu=power7 and -mtune=power7. 5211 * GCC will now vectorize loops that contain simple math functions 5212 like copysign when generating code for altivec or VSX targets. 5213 * Support for the A2 processor is now available through the -mcpu=a2 5214 and -mtune=a2 options. 5215 * Support for the 476 processor is now available through the 5216 -mcpu={476,476fp} and -mtune={476,476fp} options. 5217 * Support for the e500mc64 processor is now available through the 5218 -mcpu=e500mc64 and -mtune=e500mc64 options. 5219 * GCC can now be configured with options --with-cpu-32, 5220 --with-cpu-64, --with-tune-32 and --with-tune-64 to control the 5221 default optimization separately for 32-bit and 64-bit modes. 5222 * Starting with GCC 4.5.4, vectors of type vector long long or vector 5223 long are passed and returned in the same method as other vectors 5224 with the VSX instruction set. Previously the GCC compiler did not 5225 adhere to the ABI for 128-bit vectors with 64-bit integer base 5226 types (PR 48857). This is also fixed in the GCC 4.6.1 release. 5227 5228 RX 5229 5230 Support has been added for the Renesas RX Processor (rx-elf) target. 5231 5232 Operating Systems 5233 5234 Windows (Cygwin and MinGW) 5235 5236 * GCC now installs all the major language runtime libraries as DLLs 5237 when configured with the --enable-shared option. 5238 * GCC now makes use of the new support for aligned common variables 5239 in versions of binutils >= 2.20 to fix bugs in the support for SSE 5240 data types. 5241 * Improvements to the libffi support library increase the reliability 5242 of code generated by GCJ on all Windows platforms. Libgcj is 5243 enabled by default for the first time. 5244 * Libtool improvements simplify installation by placing the generated 5245 DLLs in the correct binaries directory. 5246 * Numerous other minor bugfixes and improvements, and substantial 5247 enhancements to the Fortran language support library. 5248 5249 > 5250 5251 Other significant improvements 5252 5253 Plugins 5254 5255 * It is now possible to extend the compiler without having to modify 5256 its source code. A new option -fplugin=file.so tells GCC to load 5257 the shared object file.so and execute it as part of the compiler. 5258 The internal documentation describes the details on how plugins can 5259 interact with the compiler. 5260 5261 Installation changes 5262 5263 * The move to newer autotools changed default installation 5264 directories and switches to control them: The --with-datarootdir, 5265 --with-docdir, --with-pdfdir, and --with-htmldir switches are not 5266 used any more. Instead, you can now use --datarootdir, --docdir, 5267 --htmldir, and --pdfdir. The default installation directories have 5268 changed as follows according to the GNU Coding Standards: 5269 5270 datarootdir read-only architecture-independent data root [PREFIX/share] 5271 localedir locale-specific message catalogs [DATAROOTDIR/locale] 5272 docdir documentation root [DATAROOTDIR/doc/PACKAGE] 5273 htmldir html documentation [DOCDIR] 5274 dvidir dvi documentation [DOCDIR] 5275 pdfdir pdf documentation [DOCDIR] 5276 psdir ps documentation [DOCDIR] 5277 The following variables have new default values: 5278 5279 datadir read-only architecture-independent data [DATAROOTDIR] 5280 infodir info documentation [DATAROOTDIR/info] 5281 mandir man documentation [DATAROOTDIR/man] 5282 5283 GCC 4.5.1 5284 5285 This is the [23]list of problem reports (PRs) from GCC's bug tracking 5286 system that are known to be fixed in the 4.5.1 release. This list might 5287 not be complete (that is, it is possible that some PRs that have been 5288 fixed are not listed here). 5289 5290 All languages 5291 5292 * GCC's new link-time optimizer ([24]-flto) now also works on a few 5293 non-ELF targets: 5294 + Cygwin (*-cygwin*) 5295 + MinGW (*-mingw*) 5296 + Darwin on x86-64 (x86_64-apple-darwin*) 5297 LTO is not enabled by default for these targets. To enable LTO, you 5298 should configure with the --enable-lto option. 5299 5300 GCC 4.5.2 5301 5302 This is the [25]list of problem reports (PRs) from GCC's bug tracking 5303 system that are known to be fixed in the 4.5.2 release. This list might 5304 not be complete (that is, it is possible that some PRs that have been 5305 fixed are not listed here). 5306 5307 GCC 4.5.3 5308 5309 This is the [26]list of problem reports (PRs) from GCC's bug tracking 5310 system that are known to be fixed in the 4.5.3 release. This list might 5311 not be complete (that is, it is possible that some PRs that have been 5312 fixed are not listed here). 5313 5314 On the PowerPC compiler, the Altivec builtin functions vec_ld and 5315 vec_st have been modified to generate the Altivec memory instructions 5316 LVX and STVX, even if the -mvsx option is used. In the initial GCC 4.5 5317 release, these builtin functions were changed to generate VSX memory 5318 reference instructions instead of Altivec memory instructions, but 5319 there are differences between the two instructions. If the VSX 5320 instruction set is available, you can now use the new builtin functions 5321 vec_vsx_ld and vec_vsx_st which always generates the VSX memory 5322 instructions. 5323 5324 GCC 4.5.4 5325 5326 This is the [27]list of problem reports (PRs) from GCC's bug tracking 5327 system that are known to be fixed in the 4.5.4 release. This list might 5328 not be complete (that is, it is possible that some PRs that have been 5329 fixed are not listed here). 5330 5331 5332 For questions related to the use of GCC, please consult these web 5333 pages and the [28]GCC manuals. If that fails, the 5334 [29]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 5335 web pages and the development of GCC are welcome on our developer 5336 list at [30]gcc (a] gcc.gnu.org. All of [31]our lists have public 5337 archives. 5338 5339 Copyright (C) [32]Free Software Foundation, Inc. Verbatim copying and 5340 distribution of this entire article is permitted in any medium, 5341 provided this notice is preserved. 5342 5343 These pages are [33]maintained by the GCC team. Last modified 5344 2017-02-28[34]. 5345 5346 References 5347 5348 1. http://www.multiprecision.org/ 5349 2. https://gcc.gnu.org/install/prerequisites.html 5350 3. https://gcc.gnu.org/ml/gcc/2010-01/msg00510.html 5351 4. http://gcc.gnu.org/gcc-4.4/changes.html#obsoleted 5352 5. http://gcc.gnu.org/gcc-4.5/changes.html#x86 5353 6. http://www.multiprecision.org/ 5354 7. https://gcc.gnu.org/PR30789 5355 8. https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801 5356 9. https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhopr-802 5357 10. https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhole-program-800 5358 11. https://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html 5359 12. http://gcc.gnu.org/gcc-4.5/cxx0x_status.html 5360 13. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#757 5361 14. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#176 5362 15. https://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html 5363 16. https://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.tr24733 5364 17. https://sourceware.org/gdb/wiki/STLSupport 5365 18. http://gcc.gnu.org/gcc-4.4/changes.html 5366 19. https://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html 5367 20. https://gcc.gnu.org/onlinedocs/gfortran/Mixed-Language-Programming.html 5368 21. https://gcc.gnu.org/onlinedocs/gfortran/GNU-Fortran-Compiler-Directives.html 5369 22. https://gcc.gnu.org/wiki/OOP 5370 23. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.1 5371 24. https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801 5372 25. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.2 5373 26. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.3 5374 27. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.4 5375 28. https://gcc.gnu.org/onlinedocs/ 5376 29. mailto:gcc-help (a] gcc.gnu.org 5377 30. mailto:gcc (a] gcc.gnu.org 5378 31. https://gcc.gnu.org/lists.html 5379 32. http://www.fsf.org/ 5380 33. https://gcc.gnu.org/about.html 5381 34. http://validator.w3.org/check/referer 5382 ====================================================================== 5383 http://gcc.gnu.org/gcc-4.4/index.html 5384 GCC 4.4 Release Series 5385 5386 March 13, 2012 5387 5388 The [1]GNU project and the GCC developers are pleased to announce the 5389 release of GCC 4.4.7. 5390 5391 This release is a bug-fix release, containing fixes for regressions in 5392 GCC 4.4.6 relative to previous releases of GCC. 5393 5394 This release series is no longer maintained. 5395 5396 Release History 5397 5398 GCC 4.4.7 5399 March 13, 2012 ([2]changes) 5400 5401 GCC 4.4.6 5402 April 16, 2011 ([3]changes) 5403 5404 GCC 4.4.5 5405 October 1, 2010 ([4]changes) 5406 5407 GCC 4.4.4 5408 April 29, 2010 ([5]changes) 5409 5410 GCC 4.4.3 5411 January 21, 2010 ([6]changes) 5412 5413 GCC 4.4.2 5414 October 15, 2009 ([7]changes) 5415 5416 GCC 4.4.1 5417 July 22, 2009 ([8]changes) 5418 5419 GCC 4.4.0 5420 April 21, 2009 ([9]changes) 5421 5422 References and Acknowledgements 5423 5424 GCC used to stand for the GNU C Compiler, but since the compiler 5425 supports several other languages aside from C, it now stands for the 5426 GNU Compiler Collection. 5427 5428 A list of [10]successful builds is updated as new information becomes 5429 available. 5430 5431 The GCC developers would like to thank the numerous people that have 5432 contributed new features, improvements, bug fixes, and other changes as 5433 well as test results to GCC. This [11]amazing group of volunteers is 5434 what makes GCC successful. 5435 5436 For additional information about GCC please refer to the [12]GCC 5437 project web site or contact the [13]GCC development mailing list. 5438 5439 To obtain GCC please use [14]our mirror sites or [15]our SVN server. 5440 5441 5442 For questions related to the use of GCC, please consult these web 5443 pages and the [16]GCC manuals. If that fails, the 5444 [17]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 5445 web pages and the development of GCC are welcome on our developer 5446 list at [18]gcc (a] gcc.gnu.org. All of [19]our lists have public 5447 archives. 5448 5449 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and 5450 distribution of this entire article is permitted in any medium, 5451 provided this notice is preserved. 5452 5453 These pages are [21]maintained by the GCC team. Last modified 5454 2016-09-30[22]. 5455 5456 References 5457 5458 1. http://www.gnu.org/ 5459 2. http://gcc.gnu.org/gcc-4.4/changes.html 5460 3. http://gcc.gnu.org/gcc-4.4/changes.html 5461 4. http://gcc.gnu.org/gcc-4.4/changes.html 5462 5. http://gcc.gnu.org/gcc-4.4/changes.html 5463 6. http://gcc.gnu.org/gcc-4.4/changes.html 5464 7. http://gcc.gnu.org/gcc-4.4/changes.html 5465 8. http://gcc.gnu.org/gcc-4.4/changes.html 5466 9. http://gcc.gnu.org/gcc-4.4/changes.html 5467 10. http://gcc.gnu.org/gcc-4.4/buildstat.html 5468 11. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 5469 12. http://gcc.gnu.org/index.html 5470 13. mailto:gcc (a] gcc.gnu.org 5471 14. http://gcc.gnu.org/mirrors.html 5472 15. http://gcc.gnu.org/svn.html 5473 16. https://gcc.gnu.org/onlinedocs/ 5474 17. mailto:gcc-help (a] gcc.gnu.org 5475 18. mailto:gcc (a] gcc.gnu.org 5476 19. https://gcc.gnu.org/lists.html 5477 20. http://www.fsf.org/ 5478 21. https://gcc.gnu.org/about.html 5479 22. http://validator.w3.org/check/referer 5480 ====================================================================== 5481 http://gcc.gnu.org/gcc-4.4/changes.html 5482 GCC 4.4 Release Series 5483 Changes, New Features, and Fixes 5484 5485 The latest release in the 4.4 release series is [1]GCC 4.4.7. 5486 5487 Caveats 5488 5489 * __builtin_stdarg_start has been completely removed from GCC. 5490 Support for <varargs.h> had been deprecated since GCC 4.0. Use 5491 __builtin_va_start as a replacement. 5492 * Some of the errors issued by the C++ front end that could be 5493 downgraded to warnings in previous releases by using -fpermissive 5494 are now warnings by default. They can be converted into errors by 5495 using -pedantic-errors. 5496 * Use of the cpp assertion extension will now emit a warning when 5497 -Wdeprecated or -pedantic is used. This extension has been 5498 deprecated for many years, but never warned about. 5499 * Packed bit-fields of type char were not properly bit-packed on many 5500 targets prior to GCC 4.4. On these targets, the fix in GCC 4.4 5501 causes an ABI change. For example there is no longer a 4-bit 5502 padding between field a and b in this structure: 5503 struct foo 5504 { 5505 char a:4; 5506 char b:8; 5507 } __attribute__ ((packed)); 5508 There is a new warning to help identify fields that are affected: 5509 foo.c:5: note: Offset of packed bit-field 'b' has changed in GCC 4.4 5510 The warning can be disabled with -Wno-packed-bitfield-compat. 5511 * On ARM EABI targets, the C++ mangling of the va_list type has been 5512 changed to conform to the current revision of the EABI. This does 5513 not affect the libstdc++ library included with GCC. 5514 * The SCOUNT and POS bits of the MIPS DSP control register are now 5515 treated as global. Previous versions of GCC treated these fields as 5516 call-clobbered instead. 5517 * The MIPS port no longer recognizes the h asm constraint. It was 5518 necessary to remove this constraint in order to avoid generating 5519 unpredictable code sequences. 5520 One of the main uses of the h constraint was to extract the high 5521 part of a multiplication on 64-bit targets. For example: 5522 asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y)); 5523 You can now achieve the same effect using 128-bit types: 5524 typedef unsigned int uint128_t __attribute__((mode(TI))); 5525 result = ((uint128_t) x * y) >> 64; 5526 The second sequence is better in many ways. For example, if x and y 5527 are constants, the compiler can perform the multiplication at 5528 compile time. If x and y are not constants, the compiler can 5529 schedule the runtime multiplication better than it can schedule an 5530 asm statement. 5531 * Support for a number of older systems and recently unmaintained or 5532 untested target ports of GCC has been declared obsolete in GCC 4.4. 5533 Unless there is activity to revive them, the next release of GCC 5534 will have their sources permanently removed. 5535 The following ports for individual systems on particular 5536 architectures have been obsoleted: 5537 + Generic a.out on IA32 and m68k (i[34567]86-*-aout*, 5538 m68k-*-aout*) 5539 + Generic COFF on ARM, H8300, IA32, m68k and SH (arm-*-coff*, 5540 armel-*-coff*, h8300-*-*, i[34567]86-*-coff*, m68k-*-coff*, 5541 sh-*-*). This does not affect other more specific targets 5542 using the COFF object format on those architectures, or the 5543 more specific H8300 and SH targets (h8300-*-rtems*, 5544 h8300-*-elf*, sh-*-elf*, sh-*-symbianelf*, sh-*-linux*, 5545 sh-*-netbsdelf*, sh-*-rtems*, sh-wrs-vxworks). 5546 + 2BSD on PDP-11 (pdp11-*-bsd) 5547 + AIX 4.1 and 4.2 on PowerPC (rs6000-ibm-aix4.[12]*, 5548 powerpc-ibm-aix4.[12]*) 5549 + Tuning support for Itanium1 (Merced) variants. Note that code 5550 tuned for Itanium2 should also run correctly on Itanium1. 5551 * The protoize and unprotoize utilities have been obsoleted and will 5552 be removed in GCC 4.5. These utilities have not been installed by 5553 default since GCC 3.0. 5554 * Support has been removed for all the [2]configurations obsoleted in 5555 GCC 4.3. 5556 * Unknown -Wno-* options are now silently ignored by GCC if no other 5557 diagnostics are issued. If other diagnostics are issued, then GCC 5558 warns about the unknown options. 5559 * More information on porting to GCC 4.4 from previous versions of 5560 GCC can be found in the [3]porting guide for this release. 5561 5562 General Optimizer Improvements 5563 5564 * A new command-line switch -findirect-inlining has been added. When 5565 turned on it allows the inliner to also inline indirect calls that 5566 are discovered to have known targets at compile time thanks to 5567 previous inlining. 5568 * A new command-line switch -ftree-switch-conversion has been added. 5569 This new pass turns simple initializations of scalar variables in 5570 switch statements into initializations from a static array, given 5571 that all the values are known at compile time and the ratio between 5572 the new array size and the original switch branches does not exceed 5573 the parameter --param switch-conversion-max-branch-ratio (default 5574 is eight). 5575 * A new command-line switch -ftree-builtin-call-dce has been added. 5576 This optimization eliminates unnecessary calls to certain builtin 5577 functions when the return value is not used, in cases where the 5578 calls can not be eliminated entirely because the function may set 5579 errno. This optimization is on by default at -O2 and above. 5580 * A new command-line switch -fconserve-stack directs the compiler to 5581 minimize stack usage even if it makes the generated code slower. 5582 This affects inlining decisions. 5583 * When the assembler supports it, the compiler will now emit unwind 5584 information using assembler .cfi directives. This makes it possible 5585 to use such directives in inline assembler code. The new option 5586 -fno-dwarf2-cfi-asm directs the compiler to not use .cfi 5587 directives. 5588 * The [4]Graphite branch has been merged. This merge has brought in a 5589 new framework for loop optimizations based on a polyhedral 5590 intermediate representation. These optimizations apply to all the 5591 languages supported by GCC. The following new code transformations 5592 are available in GCC 4.4: 5593 + -floop-interchange performs loop interchange transformations 5594 on loops. Interchanging two nested loops switches the inner 5595 and outer loops. For example, given a loop like: 5596 DO J = 1, M 5597 DO I = 1, N 5598 A(J, I) = A(J, I) * C 5599 ENDDO 5600 ENDDO 5601 5602 loop interchange will transform the loop as if the user had 5603 written: 5604 DO I = 1, N 5605 DO J = 1, M 5606 A(J, I) = A(J, I) * C 5607 ENDDO 5608 ENDDO 5609 5610 which can be beneficial when N is larger than the caches, 5611 because in Fortran, the elements of an array are stored in 5612 memory contiguously by column, and the original loop iterates 5613 over rows, potentially creating at each access a cache miss. 5614 + -floop-strip-mine performs loop strip mining transformations 5615 on loops. Strip mining splits a loop into two nested loops. 5616 The outer loop has strides equal to the strip size and the 5617 inner loop has strides of the original loop within a strip. 5618 For example, given a loop like: 5619 DO I = 1, N 5620 A(I) = A(I) + C 5621 ENDDO 5622 5623 loop strip mining will transform the loop as if the user had 5624 written: 5625 DO II = 1, N, 4 5626 DO I = II, min (II + 3, N) 5627 A(I) = A(I) + C 5628 ENDDO 5629 ENDDO 5630 5631 + -floop-block performs loop blocking transformations on loops. 5632 Blocking strip mines each loop in the loop nest such that the 5633 memory accesses of the element loops fit inside caches. For 5634 example, given a loop like: 5635 DO I = 1, N 5636 DO J = 1, M 5637 A(J, I) = B(I) + C(J) 5638 ENDDO 5639 ENDDO 5640 5641 loop blocking will transform the loop as if the user had 5642 written: 5643 DO II = 1, N, 64 5644 DO JJ = 1, M, 64 5645 DO I = II, min (II + 63, N) 5646 DO J = JJ, min (JJ + 63, M) 5647 A(J, I) = B(I) + C(J) 5648 ENDDO 5649 ENDDO 5650 ENDDO 5651 ENDDO 5652 5653 which can be beneficial when M is larger than the caches, 5654 because the innermost loop will iterate over a smaller amount 5655 of data that can be kept in the caches. 5656 * A new register allocator has replaced the old one. It is called 5657 integrated register allocator (IRA) because coalescing, register 5658 live range splitting, and hard register preferencing are done 5659 on-the-fly during coloring. It also has better integration with the 5660 reload pass. IRA is a regional register allocator which uses modern 5661 Chaitin-Briggs coloring instead of Chow's priority coloring used in 5662 the old register allocator. More info about IRA internals and 5663 options can be found in the GCC manuals. 5664 * A new instruction scheduler and software pipeliner, based on the 5665 selective scheduling approach, has been added. The new pass 5666 performs instruction unification, register renaming, substitution 5667 through register copies, and speculation during scheduling. The 5668 software pipeliner is able to pipeline non-countable loops. The new 5669 pass is targeted at scheduling-eager in-order platforms. In GCC 4.4 5670 it is available for the Intel Itanium platform working by default 5671 as the second scheduling pass (after register allocation) at the 5672 -O3 optimization level. 5673 * When using -fprofile-generate with a multi-threaded program, the 5674 profile counts may be slightly wrong due to race conditions. The 5675 new -fprofile-correction option directs the compiler to apply 5676 heuristics to smooth out the inconsistencies. By default the 5677 compiler will give an error message when it finds an inconsistent 5678 profile. 5679 * The new -fprofile-dir=PATH option permits setting the directory 5680 where profile data files are stored when using -fprofile-generate 5681 and friends, and the directory used when reading profile data files 5682 using -fprofile-use and friends. 5683 5684 New warning options 5685 5686 * The new -Wframe-larger-than=NUMBER option directs GCC to emit a 5687 warning if any stack frame is larger than NUMBER bytes. This may be 5688 used to help ensure that code fits within a limited amount of stack 5689 space. 5690 * The command-line option -Wlarger-than-N is now written as 5691 -Wlarger-than=N and the old form is deprecated. 5692 * The new -Wno-mudflap option disables warnings about constructs 5693 which can not be instrumented when using -fmudflap. 5694 5695 New Languages and Language specific improvements 5696 5697 * Version 3.0 of the OpenMP specification is now supported for the C, 5698 C++, and Fortran compilers. 5699 * New character data types, per [5]TR 19769: New character types in 5700 C, are now supported for the C compiler in -std=gnu99 mode, as 5701 __CHAR16_TYPE__ and __CHAR32_TYPE__, and for the C++ compiler in 5702 -std=c++0x and -std=gnu++0x modes, as char16_t and char32_t too. 5703 5704 C family 5705 5706 * A new optimize attribute was added to allow programmers to change 5707 the optimization level and particular optimization options for an 5708 individual function. You can also change the optimization options 5709 via the GCC optimize pragma for functions defined after the pragma. 5710 The GCC push_options pragma and the GCC pop_options pragma allow 5711 you temporarily save and restore the options used. The GCC 5712 reset_options pragma restores the options to what was specified on 5713 the command line. 5714 * Uninitialized warnings do not require enabling optimization 5715 anymore, that is, -Wuninitialized can be used together with -O0. 5716 Nonetheless, the warnings given by -Wuninitialized will probably be 5717 more accurate if optimization is enabled. 5718 * -Wparentheses now warns about expressions such as (!x | y) and (!x 5719 & y). Using explicit parentheses, such as in ((!x) | y), silences 5720 this warning. 5721 * -Wsequence-point now warns within if, while,do while and for 5722 conditions, and within for begin/end expressions. 5723 * A new option -dU is available to dump definitions of preprocessor 5724 macros that are tested or expanded. 5725 5726 C++ 5727 5728 * [6]Improved experimental support for the upcoming ISO C++ standard, 5729 C++0x. Including support for auto, inline namespaces, generalized 5730 initializer lists, defaulted and deleted functions, new character 5731 types, and scoped enums. 5732 * Those errors that may be downgraded to warnings to build legacy 5733 code now mention -fpermissive when -fdiagnostics-show-option is 5734 enabled. 5735 * -Wconversion now warns if the result of a static_cast to enumeral 5736 type is unspecified because the value is outside the range of the 5737 enumeral type. 5738 * -Wuninitialized now warns if a non-static reference or non-static 5739 const member appears in a class without constructors. 5740 * G++ now properly implements value-initialization, so objects with 5741 an initializer of () and an implicitly defined default constructor 5742 will be zero-initialized before the default constructor is called. 5743 5744 Runtime Library (libstdc++) 5745 5746 * Improved experimental support for the upcoming ISO C++ standard, 5747 C++0x, including: 5748 + Support for <chrono>, <condition_variable>, <cstdatomic>, 5749 <forward_list>, <initializer_list>, <mutex>, <ratio>, 5750 <system_error>, and <thread>. 5751 + unique_ptr, <algorithm> additions, exception propagation, and 5752 support for the new character types in <string> and <limits>. 5753 + Existing facilities now exploit initializer lists, defaulted 5754 and deleted functions, and the newly implemented core C++0x 5755 features. 5756 + Some standard containers are more efficient together with 5757 stateful allocators, i.e., no allocator is constructed on the 5758 fly at element construction time. 5759 * Experimental support for non-standard pointer types in containers. 5760 * The long standing libstdc++/30928 has been fixed for targets 5761 running glibc 2.10 or later. 5762 * As usual, many small and larger bug fixes, in particular quite a 5763 few corner cases in <locale>. 5764 5765 Fortran 5766 5767 * GNU Fortran now employs libcpp directly instead of using cc1 as an 5768 external preprocessor. The [7]-cpp option was added to allow manual 5769 invocation of the preprocessor without relying on filename 5770 extensions. 5771 * The [8]-Warray-temporaries option warns about array temporaries 5772 generated by the compiler, as an aid to optimization. 5773 * The [9]-fcheck-array-temporaries option has been added, printing a 5774 notification at run time, when an array temporary had to be created 5775 for an function argument. Contrary to -Warray-temporaries the 5776 warning is only printed if the array is noncontiguous. 5777 * Improved generation of DWARF debugging symbols 5778 * If using an intrinsic not part of the selected standard (via -std= 5779 and -fall-intrinsics) gfortran will now treat it as if this 5780 procedure were declared EXTERNAL and try to link to a user-supplied 5781 procedure. -Wintrinsics-std will warn whenever this happens. The 5782 now-useless option -Wnonstd-intrinsic was removed. 5783 * The flag -falign-commons has been added to control the alignment of 5784 variables in COMMON blocks, which is enabled by default in line 5785 with previous GCC version. Using -fno-align-commons one can force 5786 commons to be contiguous in memory as required by the Fortran 5787 standard, however, this slows down the memory access. The option 5788 -Walign-commons, which is enabled by default, warns when padding 5789 bytes were added for alignment. The proper solution is to sort the 5790 common objects by decreasing storage size, which avoids the 5791 alignment problems. 5792 * Fortran 2003 support has been extended: 5793 + Wide characters (ISO 10646, UCS-4, kind=4) and UTF-8 I/O is 5794 now supported (except internal reads from/writes to wide 5795 strings). [10]-fbackslash now supports also \unnnn and 5796 \Unnnnnnnn to enter Unicode characters. 5797 + Asynchronous I/O (implemented as synchronous I/O) and the 5798 decimal=, size=, sign=, pad=, blank=, and delim= specifiers 5799 are now supported in I/O statements. 5800 + Support for Fortran 2003 structure constructors and for array 5801 constructor with typespec has been added. 5802 + Procedure Pointers (but not yet as component in derived types 5803 and as function results) are now supported. 5804 + Abstract types, type extension, and type-bound procedures 5805 (both PROCEDURE and GENERIC but not as operators). Note: As 5806 CLASS/polymorphyic types are not implemented, type-bound 5807 procedures with PASS accept as non-standard extension TYPE 5808 arguments. 5809 * Fortran 2008 support has been added: 5810 + The -std=f2008 option and support for the file extensions 5811 .f2008 and .F2008 has been added. 5812 + The g0 format descriptor is now supported. 5813 + The Fortran 2008 mathematical intrinsics ASINH, ACOSH, ATANH, 5814 ERF, ERFC, GAMMA, LOG_GAMMA, BESSEL_*, HYPOT, and ERFC_SCALED 5815 are now available (some of them existed as GNU extension 5816 before). Note: The hyperbolic functions are not yet supporting 5817 complex arguments and the three- argument version of BESSEL_*N 5818 is not available. 5819 + The bit intrinsics LEADZ and TRAILZ have been added. 5820 5821 Java (GCJ) 5822 5823 Ada 5824 5825 * The Ada runtime now supports multilibs on many platforms including 5826 x86_64, SPARC and PowerPC. Their build is enabled by default. 5827 5828 New Targets and Target Specific Improvements 5829 5830 ARM 5831 5832 * GCC now supports optimizing for the Cortex-A9, Cortex-R4 and 5833 Cortex-R4F processors and has many other improvements to 5834 optimization for ARM processors. 5835 * GCC now supports the VFPv3 variant with 16 double-precision 5836 registers with -mfpu=vfpv3-d16. The option -mfpu=vfp3 has been 5837 renamed to -mfpu=vfpv3. 5838 * GCC now supports the -mfix-cortex-m3-ldrd option to work around an 5839 erratum on Cortex-M3 processors. 5840 * GCC now supports the __sync_* atomic operations for ARM EABI 5841 GNU/Linux. 5842 * The section anchors optimization is now enabled by default when 5843 optimizing for ARM. 5844 * GCC now uses a new EABI-compatible profiling interface for EABI 5845 targets. This requires a function __gnu_mcount_nc, which is 5846 provided by GNU libc versions 2.8 and later. 5847 5848 AVR 5849 5850 * The -mno-tablejump option has been deprecated because it has the 5851 same effect as the -fno-jump-tables option. 5852 * Added support for these new AVR devices: 5853 + ATA6289 5854 + ATtiny13A 5855 + ATtiny87 5856 + ATtiny167 5857 + ATtiny327 5858 + ATmega8C1 5859 + ATmega16C1 5860 + ATmega32C1 5861 + ATmega8M1 5862 + ATmega16M1 5863 + ATmega32M1 5864 + ATmega32U4 5865 + ATmega16HVB 5866 + ATmega4HVD 5867 + ATmega8HVD 5868 + ATmega64C1 5869 + ATmega64M1 5870 + ATmega16U4 5871 + ATmega32U6 5872 + ATmega128RFA1 5873 + AT90PWM81 5874 + AT90SCR100 5875 + M3000F 5876 + M3000S 5877 + M3001B 5878 5879 IA-32/x86-64 5880 5881 * Support for Intel AES built-in functions and code generation is 5882 available via -maes. 5883 * Support for Intel PCLMUL built-in function and code generation is 5884 available via -mpclmul. 5885 * Support for Intel AVX built-in functions and code generation is 5886 available via -mavx. 5887 * Automatically align the stack for local variables with alignment 5888 requirement. 5889 * GCC can now utilize the SVML library for vectorizing calls to a set 5890 of C99 functions if -mveclibabi=svml is specified and you link to 5891 an SVML ABI compatible library. 5892 * On x86-64, the ABI has been changed in the following cases to 5893 conform to the x86-64 ABI: 5894 + Passing/returning structures with flexible array member: 5895 struct foo 5896 { 5897 int i; 5898 int flex[]; 5899 }; 5900 + Passing/returning structures with complex float member: 5901 struct foo 5902 { 5903 int i; 5904 __complex__ float f; 5905 }; 5906 + Passing/returning unions with long double member: 5907 union foo 5908 { 5909 int x; 5910 long double ld; 5911 }; 5912 Code built with previous versions of GCC that uses any of these is 5913 not compatible with code built with GCC 4.4.0 or later. 5914 * A new target attribute was added to allow programmers to change the 5915 target options like -msse2 or -march=k8 for an individual function. 5916 You can also change the target options via the GCC target pragma 5917 for functions defined after the pragma. 5918 * GCC can now be configured with options --with-arch-32, 5919 --with-arch-64, --with-cpu-32, --with-cpu-64, --with-tune-32 and 5920 --with-tune-64 to control the default optimization separately for 5921 32-bit and 64-bit modes. 5922 5923 IA-32/IA64 5924 5925 * Support for __float128 (TFmode) IEEE quad type and corresponding 5926 TCmode IEEE complex quad type is available via the soft-fp library 5927 on IA-32/IA64 targets. This includes basic arithmetic operations 5928 (addition, subtraction, negation, multiplication and division) on 5929 __float128 real and TCmode complex values, the full set of IEEE 5930 comparisons between __float128 values, conversions to and from 5931 float, double and long double floating point types, as well as 5932 conversions to and from signed or unsigned integer, signed or 5933 unsigned long integer and signed or unsigned quad (TImode, IA64 5934 only) integer types. Additionally, all operations generate the full 5935 set of IEEE exceptions and support the full set of IEEE rounding 5936 modes. 5937 5938 M68K/ColdFire 5939 5940 * GCC now supports instruction scheduling for ColdFire V1, V3 and V4 5941 processors. (Scheduling support for ColdFire V2 processors was 5942 added in GCC 4.3.) 5943 * GCC now supports the -mxgot option to support programs requiring 5944 many GOT entries on ColdFire. 5945 * The m68k-*-linux-gnu target now builds multilibs by default. 5946 5947 MIPS 5948 5949 * MIPS Technologies have extended the original MIPS SVR4 ABI to 5950 include support for procedure linkage tables (PLTs) and copy 5951 relocations. These extensions allow GNU/Linux executables to use a 5952 significantly more efficient code model than the one defined by the 5953 original ABI. 5954 GCC support for this code model is available via a new command-line 5955 option, -mplt. There is also a new configure-time option, 5956 --with-mips-plt, to make -mplt the default. 5957 The new code model requires support from the assembler, the linker, 5958 and the runtime C library. This support is available in binutils 5959 2.19 and GLIBC 2.9. 5960 * GCC can now generate MIPS16 code for 32-bit GNU/Linux executables 5961 and 32-bit GNU/Linux shared libraries. This feature requires GNU 5962 binutils 2.19 or above. 5963 * Support for RMI's XLR processor is now available through the 5964 -march=xlr and -mtune=xlr options. 5965 * 64-bit targets can now perform 128-bit multiplications inline, 5966 instead of relying on a libgcc function. 5967 * Native GNU/Linux toolchains now support -march=native and 5968 -mtune=native, which select the host processor. 5969 * GCC now supports the R10K, R12K, R14K and R16K processors. The 5970 canonical -march= and -mtune= names for these processors are 5971 r10000, r12000, r14000 and r16000 respectively. 5972 * GCC can now work around the side effects of speculative execution 5973 on R10K processors. Please see the documentation of the 5974 -mr10k-cache-barrier option for details. 5975 * Support for the MIPS64 Release 2 instruction set has been added. 5976 The option -march=mips64r2 enables generation of these 5977 instructions. 5978 * GCC now supports Cavium Networks' Octeon processor. This support is 5979 available through the -march=octeon and -mtune=octeon options. 5980 * GCC now supports STMicroelectronics' Loongson 2E/2F processors. The 5981 canonical -march= and -mtune= names for these processors are 5982 loongson2e and loongson2f. 5983 5984 picochip 5985 5986 Picochip is a 16-bit processor. A typical picoChip contains over 250 5987 small cores, each with small amounts of memory. There are three 5988 processor variants (STAN, MEM and CTRL) with different instruction sets 5989 and memory configurations and they can be chosen using the -mae option. 5990 5991 This port is intended to be a "C" only port. 5992 5993 Power Architecture and PowerPC 5994 5995 * GCC now supports the e300c2, e300c3 and e500mc processors. 5996 * GCC now supports Xilinx processors with a single-precision FPU. 5997 * Decimal floating point is now supported for e500 processors. 5998 5999 S/390, zSeries and System z9/z10 6000 6001 * Support for the IBM System z10 EC/BC processor has been added. When 6002 using the -march=z10 option, the compiler will generate code making 6003 use of instructions provided by the General-Instruction-Extension 6004 Facility and the Execute-Extension Facility. 6005 6006 VxWorks 6007 6008 * GCC now supports the thread-local storage mechanism used on 6009 VxWorks. 6010 6011 Xtensa 6012 6013 * GCC now supports thread-local storage (TLS) for Xtensa processor 6014 configurations that include the Thread Pointer option. TLS also 6015 requires support from the assembler and linker; this support is 6016 provided in the GNU binutils beginning with version 2.19. 6017 6018 Documentation improvements 6019 6020 Other significant improvements 6021 6022 GCC 4.4.1 6023 6024 This is the [11]list of problem reports (PRs) from GCC's bug tracking 6025 system that are known to be fixed in the 4.4.1 release. This list might 6026 not be complete (that is, it is possible that some PRs that have been 6027 fixed are not listed here). 6028 6029 GCC 4.4.2 6030 6031 This is the [12]list of problem reports (PRs) from GCC's bug tracking 6032 system that are known to be fixed in the 4.4.2 release. This list might 6033 not be complete (that is, it is possible that some PRs that have been 6034 fixed are not listed here). 6035 6036 GCC 4.4.3 6037 6038 This is the [13]list of problem reports (PRs) from GCC's bug tracking 6039 system that are known to be fixed in the 4.4.3 release. This list might 6040 not be complete (that is, it is possible that some PRs that have been 6041 fixed are not listed here). 6042 6043 GCC 4.4.4 6044 6045 This is the [14]list of problem reports (PRs) from GCC's bug tracking 6046 system that are known to be fixed in the 4.4.4 release. This list might 6047 not be complete (that is, it is possible that some PRs that have been 6048 fixed are not listed here). 6049 6050 GCC 4.4.5 6051 6052 This is the [15]list of problem reports (PRs) from GCC's bug tracking 6053 system that are known to be fixed in the 4.4.5 release. This list might 6054 not be complete (that is, it is possible that some PRs that have been 6055 fixed are not listed here). 6056 6057 GCC 4.4.6 6058 6059 This is the [16]list of problem reports (PRs) from GCC's bug tracking 6060 system that are known to be fixed in the 4.4.6 release. This list might 6061 not be complete (that is, it is possible that some PRs that have been 6062 fixed are not listed here). 6063 6064 GCC 4.4.7 6065 6066 This is the [17]list of problem reports (PRs) from GCC's bug tracking 6067 system that are known to be fixed in the 4.4.7 release. This list might 6068 not be complete (that is, it is possible that some PRs that have been 6069 fixed are not listed here). 6070 6071 6072 For questions related to the use of GCC, please consult these web 6073 pages and the [18]GCC manuals. If that fails, the 6074 [19]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 6075 web pages and the development of GCC are welcome on our developer 6076 list at [20]gcc (a] gcc.gnu.org. All of [21]our lists have public 6077 archives. 6078 6079 Copyright (C) [22]Free Software Foundation, Inc. Verbatim copying and 6080 distribution of this entire article is permitted in any medium, 6081 provided this notice is preserved. 6082 6083 These pages are [23]maintained by the GCC team. Last modified 6084 2016-12-28[24]. 6085 6086 References 6087 6088 1. http://gcc.gnu.org/gcc-4.4/changes.html#4.4.7 6089 2. http://gcc.gnu.org/gcc-4.3/changes.html#obsoleted 6090 3. http://gcc.gnu.org/gcc-4.4/porting_to.html 6091 4. https://gcc.gnu.org/wiki/Graphite 6092 5. http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1040.pdf 6093 6. http://gcc.gnu.org/gcc-4.4/cxx0x_status.html 6094 7. https://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html 6095 8. https://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWarray-temporaries_007d-125 6096 9. https://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcheck-array-temporaries_007d-221 6097 10. https://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html#index-g_t_0040code_007bbackslash_007d-34 6098 11. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.1 6099 12. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.2 6100 13. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.3 6101 14. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.4 6102 15. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.5 6103 16. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.6 6104 17. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.7 6105 18. https://gcc.gnu.org/onlinedocs/ 6106 19. mailto:gcc-help (a] gcc.gnu.org 6107 20. mailto:gcc (a] gcc.gnu.org 6108 21. https://gcc.gnu.org/lists.html 6109 22. http://www.fsf.org/ 6110 23. https://gcc.gnu.org/about.html 6111 24. http://validator.w3.org/check/referer 6112 ====================================================================== 6113 http://gcc.gnu.org/gcc-4.3/index.html 6114 GCC 4.3 Release Series 6115 6116 Jun 27, 2011 6117 6118 The [1]GNU project and the GCC developers are pleased to announce the 6119 release of GCC 4.3.6. 6120 6121 This release is a bug-fix release, containing fixes for regressions in 6122 GCC 4.3.5 relative to previous releases of GCC. 6123 6124 This release series is no longer maintained. 6125 6126 Release History 6127 6128 GCC 4.3.6 6129 Jun 27, 2011 ([2]changes) 6130 6131 GCC 4.3.5 6132 May 22, 2010 ([3]changes) 6133 6134 GCC 4.3.4 6135 August 4, 2009 ([4]changes) 6136 6137 GCC 4.3.3 6138 January 24, 2009 ([5]changes) 6139 6140 GCC 4.3.2 6141 August 27, 2008 ([6]changes) 6142 6143 GCC 4.3.1 6144 June 6, 2008 ([7]changes) 6145 6146 GCC 4.3.0 6147 March 5, 2008 ([8]changes) 6148 6149 References and Acknowledgements 6150 6151 GCC used to stand for the GNU C Compiler, but since the compiler 6152 supports several other languages aside from C, it now stands for the 6153 GNU Compiler Collection. 6154 6155 A list of [9]successful builds is updated as new information becomes 6156 available. 6157 6158 The GCC developers would like to thank the numerous people that have 6159 contributed new features, improvements, bug fixes, and other changes as 6160 well as test results to GCC. This [10]amazing group of volunteers is 6161 what makes GCC successful. 6162 6163 For additional information about GCC please refer to the [11]GCC 6164 project web site or contact the [12]GCC development mailing list. 6165 6166 To obtain GCC please use [13]our mirror sites or [14]our SVN server. 6167 6168 6169 For questions related to the use of GCC, please consult these web 6170 pages and the [15]GCC manuals. If that fails, the 6171 [16]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 6172 web pages and the development of GCC are welcome on our developer 6173 list at [17]gcc (a] gcc.gnu.org. All of [18]our lists have public 6174 archives. 6175 6176 Copyright (C) [19]Free Software Foundation, Inc. Verbatim copying and 6177 distribution of this entire article is permitted in any medium, 6178 provided this notice is preserved. 6179 6180 These pages are [20]maintained by the GCC team. Last modified 6181 2016-09-30[21]. 6182 6183 References 6184 6185 1. http://www.gnu.org/ 6186 2. http://gcc.gnu.org/gcc-4.3/changes.html 6187 3. http://gcc.gnu.org/gcc-4.3/changes.html 6188 4. http://gcc.gnu.org/gcc-4.3/changes.html 6189 5. http://gcc.gnu.org/gcc-4.3/changes.html 6190 6. http://gcc.gnu.org/gcc-4.3/changes.html 6191 7. http://gcc.gnu.org/gcc-4.3/changes.html 6192 8. http://gcc.gnu.org/gcc-4.3/changes.html 6193 9. http://gcc.gnu.org/gcc-4.3/buildstat.html 6194 10. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 6195 11. http://gcc.gnu.org/index.html 6196 12. mailto:gcc (a] gcc.gnu.org 6197 13. http://gcc.gnu.org/mirrors.html 6198 14. http://gcc.gnu.org/svn.html 6199 15. https://gcc.gnu.org/onlinedocs/ 6200 16. mailto:gcc-help (a] gcc.gnu.org 6201 17. mailto:gcc (a] gcc.gnu.org 6202 18. https://gcc.gnu.org/lists.html 6203 19. http://www.fsf.org/ 6204 20. https://gcc.gnu.org/about.html 6205 21. http://validator.w3.org/check/referer 6206 ====================================================================== 6207 http://gcc.gnu.org/gcc-4.3/changes.html 6208 GCC 4.3 Release Series 6209 Changes, New Features, and Fixes 6210 6211 The latest release in the 4.3 release series is [1]GCC 4.3.5. 6212 6213 Caveats 6214 6215 * GCC requires the [2]GMP and [3]MPFR libraries for building all the 6216 various front-end languages it supports. See the [4]prerequisites 6217 page for version requirements. 6218 * ColdFire targets now treat long double as having the same format as 6219 double. In earlier versions of GCC, they used the 68881 long double 6220 format instead. 6221 * The m68k-uclinux target now uses the same calling conventions as 6222 m68k-linux-gnu. You can select the original calling conventions by 6223 configuring for m68k-uclinuxoldabi instead. Note that 6224 m68k-uclinuxoldabi also retains the original 80-bit long double on 6225 ColdFire targets. 6226 * The -fforce-mem option has been removed because it has had no 6227 effect in the last few GCC releases. 6228 * The i386 -msvr3-shlib option has been removed since it is no longer 6229 used. 6230 * Fastcall for i386 has been changed not to pass aggregate arguments 6231 in registers, following Microsoft compilers. 6232 * Support for the AOF assembler has been removed from the ARM back 6233 end; this affects only the targets arm-semi-aof and armel-semi-aof, 6234 which are no longer recognized. We removed these targets without a 6235 deprecation period because we discovered that they have been 6236 unusable since GCC 4.0.0. 6237 * Support for the TMS320C3x/C4x processor (targets c4x-* and tic4x-*) 6238 has been removed. This support had been deprecated since GCC 4.0.0. 6239 * Support for a number of older systems and recently unmaintained or 6240 untested target ports of GCC has been declared obsolete in GCC 4.3. 6241 Unless there is activity to revive them, the next release of GCC 6242 will have their sources permanently removed. 6243 All GCC ports for the following processor architectures have been 6244 declared obsolete: 6245 + Morpho MT (mt-*) 6246 The following aliases for processor architectures have been 6247 declared obsolete. Users should use the indicated generic target 6248 names instead, with compile-time options such as -mcpu or 6249 configure-time options such as --with-cpu to control the 6250 configuration more precisely. 6251 + strongarm*-*-*, ep9312*-*-*, xscale*-*-* (use arm*-*-* 6252 instead). 6253 + parisc*-*-* (use hppa*-*-* instead). 6254 + m680[012]0-*-* (use m68k-*-* instead). 6255 All GCC ports for the following operating systems have been 6256 declared obsolete: 6257 + BeOS (*-*-beos*) 6258 + kaOS (*-*-kaos*) 6259 + GNU/Linux using the a.out object format (*-*-linux*aout*) 6260 + GNU/Linux using version 1 of the GNU C Library 6261 (*-*-linux*libc1*) 6262 + Solaris versions before Solaris 7 (*-*-solaris2.[0-6], 6263 *-*-solaris2.[0-6].*) 6264 + Miscellaneous System V (*-*-sysv*) 6265 + WindISS (*-*-windiss*) 6266 Also, those for some individual systems on particular architectures 6267 have been obsoleted: 6268 + UNICOS/mk on DEC Alpha (alpha*-*-unicosmk*) 6269 + CRIS with a.out object format (cris-*-aout) 6270 + BSD 4.3 on PA-RISC (hppa1.1-*-bsd*) 6271 + OSF/1 on PA-RISC (hppa1.1-*-osf*) 6272 + PRO on PA-RISC (hppa1.1-*-pro*) 6273 + Sequent PTX on IA32 (i[34567]86-sequent-ptx4*, 6274 i[34567]86-sequent-sysv4*) 6275 + SCO Open Server 5 on IA32 (i[34567]86-*-sco3.2v5*) 6276 + UWIN on IA32 (i[34567]86-*-uwin*) (support for UWIN as a host 6277 was previously [5]removed in 2001, leaving only the support 6278 for UWIN as a target now being deprecated) 6279 + ChorusOS on PowerPC (powerpc-*-chorusos*) 6280 + All VAX configurations apart from NetBSD and OpenBSD 6281 (vax-*-bsd*, vax-*-sysv*, vax-*-ultrix*) 6282 * The [6]-Wconversion option has been modified. Its purpose now is to 6283 warn for implicit conversions that may alter a value. This new 6284 behavior is available for both C and C++. Warnings about 6285 conversions between signed and unsigned integers can be disabled by 6286 using -Wno-sign-conversion. In C++, they are disabled by default 6287 unless -Wsign-conversion is explicitly requested. The old behavior 6288 of -Wconversion, that is, warn for prototypes causing a type 6289 conversion that is different from what would happen to the same 6290 argument in the absence of a prototype, has been moved to a new 6291 option -Wtraditional-conversion, which is only available for C. 6292 * The -m386, -m486, -mpentium and -mpentiumpro tuning options have 6293 been removed because they were deprecated for more than 3 GCC major 6294 releases. Use -mtune=i386, -mtune=i486, -mtune=pentium or 6295 -mtune=pentiumpro as a replacement. 6296 * The -funsafe-math-optimizations option now automatically turns on 6297 -fno-trapping-math in addition to -fno-signed-zeros, as it enables 6298 reassociation and thus may introduce or remove traps. 6299 * The -ftree-vectorize option is now on by default under -O3. In 6300 order to generate code for a SIMD extension, it has to be enabled 6301 as well: use -maltivec for PowerPC platforms and -msse/-msse2 for 6302 i?86 and x86_64. 6303 * More information on porting to GCC 4.3 from previous versions of 6304 GCC can be found in the [7]porting guide for this release. 6305 6306 General Optimizer Improvements 6307 6308 * The GCC middle-end has been integrated with the [8]MPFR library. 6309 This allows GCC to evaluate and replace at compile-time calls to 6310 built-in math functions having constant arguments with their 6311 mathematically equivalent results. In making use of [9]MPFR, GCC 6312 can generate correct results regardless of the math library 6313 implementation or floating point precision of the host platform. 6314 This also allows GCC to generate identical results regardless of 6315 whether one compiles in native or cross-compile configurations to a 6316 particular target. The following built-in functions take advantage 6317 of this new capability: acos, acosh, asin, asinh, atan2, atan, 6318 atanh, cbrt, cos, cosh, drem, erf, erfc, exp10, exp2, exp, expm1, 6319 fdim, fma, fmax, fmin, gamma_r, hypot, j0, j1, jn, lgamma_r, log10, 6320 log1p, log2, log, pow10, pow, remainder, remquo, sin, sincos, sinh, 6321 tan, tanh, tgamma, y0, y1 and yn. The float and long double 6322 variants of these functions (e.g. sinf and sinl) are also handled. 6323 The sqrt and cabs functions with constant arguments were already 6324 optimized in prior GCC releases. Now they also use [10]MPFR. 6325 * A new forward propagation pass on RTL was added. The new pass 6326 replaces several slower transformations, resulting in compile-time 6327 improvements as well as better code generation in some cases. 6328 * A new command-line switch -frecord-gcc-switches has been added to 6329 GCC, although it is only enabled for some targets. The switch 6330 causes the command line that was used to invoke the compiler to be 6331 recorded into the object file that is being created. The exact 6332 format of this recording is target and binary file format 6333 dependent, but it usually takes the form of a note section 6334 containing ASCII text. The switch is related to the -fverbose-asm 6335 switch, but that one only records the information in the assembler 6336 output file as comments, so the information never reaches the 6337 object file. 6338 * The inliner heuristic is now aware of stack frame consumption. New 6339 command-line parameters --param large-stack-frame and --param 6340 large-stack-frame-growth can be used to limit stack frame size 6341 growth caused by inlining. 6342 * During feedback directed optimizations, the expected block size the 6343 memcpy, memset and bzero functions operate on is discovered and for 6344 cases of commonly used small sizes, specialized inline code is 6345 generated. 6346 * __builtin_expect no longer requires its argument to be a compile 6347 time constant. 6348 * Interprocedural optimization was reorganized to work on functions 6349 in SSA form. This enables more precise and cheaper dataflow 6350 analysis and makes writing interprocedural optimizations easier. 6351 The following improvements have been implemented on top of this 6352 framework: 6353 + Pre-inline optimization: Selected local optimization passes 6354 are run before the inliner (and other interprocedural passes) 6355 are executed. This significantly improves the accuracy of code 6356 growth estimates used by the inliner and reduces the overall 6357 memory footprint for large compilation units. 6358 + Early inlining (a simple bottom-up inliner pass inlining only 6359 functions whose body is smaller than the expected call 6360 overhead) is now executed with the early optimization passes, 6361 thus inlining already optimized function bodies into an 6362 unoptimized function that is subsequently optimized by early 6363 optimizers. This enables the compiler to quickly eliminate 6364 abstraction penalty in C++ programs. 6365 + Interprocedural constant propagation now operate on SSA form 6366 increasing accuracy of the analysis. 6367 * A new internal representation for GIMPLE statements has been 6368 contributed, resulting in compile-time memory savings. 6369 * The vectorizer was enhanced to support vectorization of outer 6370 loops, intra-iteration parallelism (loop-aware SLP), vectorization 6371 of strided accesses and loops with multiple data-types. Run-time 6372 dependency testing using loop versioning was added. The cost model, 6373 turned on by -fvect-cost-model, was developed. 6374 6375 New Languages and Language specific improvements 6376 6377 * We have added new command-line options 6378 -finstrument-functions-exclude-function-list and 6379 -finstrument-functions-exclude-file-list. They provide more control 6380 over which functions are annotated by the -finstrument-functions 6381 option. 6382 6383 C family 6384 6385 * Implicit conversions between generic vector types are now only 6386 permitted when the two vectors in question have the same number of 6387 elements and compatible element types. (Note that the restriction 6388 involves compatible element types, not implicitly-convertible 6389 element types: thus, a vector type with element type int may not be 6390 implicitly converted to a vector type with element type unsigned 6391 int.) This restriction, which is in line with specifications for 6392 SIMD architectures such as AltiVec, may be relaxed using the flag 6393 -flax-vector-conversions. This flag is intended only as a 6394 compatibility measure and should not be used for new code. 6395 * -Warray-bounds has been added and is now enabled by default for 6396 -Wall . It produces warnings for array subscripts that can be 6397 determined at compile time to be always out of bounds. 6398 -Wno-array-bounds will disable the warning. 6399 * The constructor and destructor function attributes now accept 6400 optional priority arguments which control the order in which the 6401 constructor and destructor functions are run. 6402 * New [11]command-line options -Wtype-limits, 6403 -Wold-style-declaration, -Wmissing-parameter-type, -Wempty-body, 6404 -Wclobbered and -Wignored-qualifiers have been added for finer 6405 control of the diverse warnings enabled by -Wextra. 6406 * A new function attribute alloc_size has been added to mark up 6407 malloc style functions. For constant sized allocations this can be 6408 used to find out the size of the returned pointer using the 6409 __builtin_object_size() function for buffer overflow checking and 6410 similar. This supplements the already built-in malloc and calloc 6411 constant size handling. 6412 * Integer constants written in binary are now supported as a GCC 6413 extension. They consist of a prefix 0b or 0B, followed by a 6414 sequence of 0 and 1 digits. 6415 * A new predefined macro __COUNTER__ has been added. It expands to 6416 sequential integral values starting from 0. In conjunction with the 6417 ## operator, this provides a convenient means to generate unique 6418 identifiers. 6419 * A new command-line option -fdirectives-only has been added. It 6420 enables a special preprocessing mode which improves the performance 6421 of applications like distcc and ccache. 6422 * Fixed-point data types and operators have been added. They are 6423 based on Chapter 4 of the Embedded-C specification (n1169.pdf). 6424 Currently, only MIPS targets are supported. 6425 * Decimal floating-point arithmetic based on draft ISO/IEC TR 24732, 6426 N1241, is now supported as a GCC extension to C for targets 6427 i[34567]86-*-linux-gnu, powerpc*-*-linux-gnu, s390*-ibm-linux-gnu, 6428 and x86_64-*-linux-gnu. The feature introduces new data types 6429 _Decimal32, _Decimal64, and _Decimal128 with constant suffixes DF, 6430 DD, and DL. 6431 6432 C++ 6433 6434 * [12]Experimental support for the upcoming ISO C++ standard, C++0x. 6435 * -Wc++0x-compat has been added and is now enabled by default for 6436 -Wall. It produces warnings for constructs whose meaning differs 6437 between ISO C++ 1998 and C++0x. 6438 * The -Wparentheses option now works for C++ as it does for C. It 6439 warns if parentheses are omitted when operators with confusing 6440 precedence are nested. It also warns about ambiguous else 6441 statements. Since -Wparentheses is enabled by -Wall, this may cause 6442 additional warnings with existing C++ code which uses -Wall. These 6443 new warnings may be disabled by using -Wall -Wno-parentheses. 6444 * The -Wmissing-declarations now works for C++ as it does for C. 6445 * The -fvisibility-ms-compat flag was added, to make it easier to 6446 port larger projects using shared libraries from Microsoft's Visual 6447 Studio to ELF and Mach-O systems. 6448 * C++ attribute handling has been overhauled for template arguments 6449 (ie dependent types). In particular, __attribute__((aligned(T))); 6450 works for C++ types. 6451 6452 Runtime Library (libstdc++) 6453 6454 * [13]Experimental support for the upcoming ISO C++ standard, C++0x. 6455 * Support for TR1 mathematical special functions and regular 6456 expressions. 6457 * Default what implementations give more elaborate exception strings 6458 for bad_cast, bad_typeid, bad_exception, and bad_alloc. 6459 * Header dependencies have been streamlined, reducing unnecessary 6460 includes and pre-processed bloat. 6461 * Variadic template implementations of items in <tuple> and 6462 <functional>. 6463 * An experimental [14]parallel mode has been added. This is a 6464 parallel implementation of many C++ Standard library algorithms, 6465 like std::accumulate, std::for_each, std::transform, or std::sort, 6466 to give but four examples. These algorithms can be substituted for 6467 the normal (sequential) libstdc++ algorithms on a piecemeal basis, 6468 or all existing algorithms can be transformed via the 6469 -D_GLIBCXX_PARALLEL macro. 6470 * Debug mode versions of classes in <unordered_set> and 6471 <unordered_map>. 6472 * Formal deprecation of <ext/hash_set> and <ext/hash_map>, which are 6473 now <backward/hash_set> and <backward/hash_map>. This code: 6474 #include <ext/hash_set> 6475 __gnu_cxx::hash_set<int> s; 6476 6477 Can be transformed (in order of preference) to: 6478 #include <tr1/unordered_set> 6479 std::tr1::unordered_set<int> s; 6480 6481 or 6482 #include <backward/hash_set> 6483 __gnu_cxx::hash_set<int> s; 6484 6485 Similar transformations apply to __gnu_cxx::hash_map, 6486 __gnu_cxx::hash_multimap, __gnu_cxx::hash_set, 6487 __gnu_cxx::hash_multiset. 6488 6489 Fortran 6490 6491 * Due to the fact that the GMP and MPFR libraries are required for 6492 all languages, Fortran is no longer special in this regard and is 6493 available by default. 6494 * The [15]-fexternal-blas option has been added, which generates 6495 calls to BLAS routines for intrinsic matrix operations such as 6496 matmul rather than using the built-in algorithms. 6497 * Support to give a backtrace (compiler flag -fbacktrace or 6498 environment variable GFORTRAN_ERROR_BACKTRACE; on glibc systems 6499 only) or a core dump (-fdump-core, GFORTRAN_ERROR_DUMPCORE) when a 6500 run-time error occured. 6501 * GNU Fortran now defines __GFORTRAN__ when it runs the C 6502 preprocessor (CPP). 6503 * The [16]-finit-local-zero, -finit-real, -finit-integer, 6504 -finit-character, and -finit-logical options have been added, which 6505 can be used to initialize local variables. 6506 * The intrinsic procedures [17]GAMMA and [18]LGAMMA have been added, 6507 which calculate the Gamma function and its logarithm. Use EXTERNAL 6508 gamma if you want to use your own gamma function. 6509 * GNU Fortran now regards the backslash character as literal (as 6510 required by the Fortran 2003 standard); using [19]-fbackslash GNU 6511 Fortran interprets backslashes as C-style escape characters. 6512 * The [20]interpretation of binary, octal and hexadecimal (BOZ) 6513 literal constants has been changed. Before they were always 6514 interpreted as integer; now they are bit-wise transferred as 6515 argument of INT, REAL, DBLE and CMPLX as required by the Fortran 6516 2003 standard, and for real and complex variables in DATA 6517 statements or when directly assigned to real and complex variables. 6518 Everywhere else and especially in expressions they are still 6519 regarded as integer constants. 6520 * Fortran 2003 support has been extended: 6521 + Intrinsic statements IMPORT, PROTECTED, VALUE and VOLATILE 6522 + Pointer intent 6523 + Intrinsic module ISO_ENV_FORTRAN 6524 + Interoperability with C (ISO C Bindings) 6525 + ABSTRACT INTERFACES and PROCEDURE statements (without POINTER 6526 attribute) 6527 + Fortran 2003 BOZ 6528 6529 Java (GCJ) 6530 6531 * GCJ now uses the Eclipse Java compiler for its Java parsing needs. 6532 This enables the use of all 1.5 language features, and fixes most 6533 existing front end bugs. 6534 * libgcj now supports all 1.5 language features which require runtime 6535 support: foreach, enum, annotations, generics, and auto-boxing. 6536 * We've made many changes to the tools shipped with gcj. 6537 + The old jv-scan tool has been removed. This tool never really 6538 worked properly. There is no replacement. 6539 + gcjh has been rewritten. Some of its more obscure options no 6540 longer work, but are still recognized in an attempt at 6541 compatibility. gjavah is a new program with similar 6542 functionality but different command-line options. 6543 + grmic and grmiregistry have been rewritten. grmid has been 6544 added. 6545 + gjar replaces the old fastjar. 6546 + gjarsigner (used for signing jars), gkeytool (used for key 6547 management), gorbd (for CORBA), gserialver (computes 6548 serialization UIDs), and gtnameserv (also for CORBA) are now 6549 installed. 6550 * The ability to dump the contents of the java run time heap to a 6551 file for off-line analysis has been added. The heap dumps may be 6552 analyzed with the new gc-analyze tool. They may be generated on 6553 out-of-memory conditions or on demand and are controlled by the new 6554 run time class gnu.gcj.util.GCInfo. 6555 * java.util.TimeZone can now read files from /usr/share/zoneinfo to 6556 provide correct, updated, timezone information. This means that 6557 packagers no longer have to update libgcj when a time zone change 6558 is published. 6559 6560 New Targets and Target Specific Improvements 6561 6562 IA-32/x86-64 6563 6564 * Tuning for Intel Core 2 processors is available via -mtune=core2 6565 and -march=core2. 6566 * Tuning for AMD Geode processors is available via -mtune=geode and 6567 -march=geode. 6568 * Code generation of block move (memcpy) and block set (memset) was 6569 rewritten. GCC can now pick the best algorithm (loop, unrolled 6570 loop, instruction with rep prefix or a library call) based on the 6571 size of the block being copied and the CPU being optimized for. A 6572 new option -minline-stringops-dynamically has been added. With this 6573 option string operations of unknown size are expanded such that 6574 small blocks are copied by in-line code, while for large blocks a 6575 library call is used. This results in faster code than 6576 -minline-all-stringops when the library implementation is capable 6577 of using cache hierarchy hints. The heuristic choosing the 6578 particular algorithm can be overwritten via -mstringop-strategy. 6579 Newly also memset of values different from 0 is inlined. 6580 * GCC no longer places the cld instruction before string operations. 6581 Both i386 and x86-64 ABI documents mandate the direction flag to be 6582 clear at the entry of a function. It is now invalid to set the flag 6583 in asm statement without reseting it afterward. 6584 * Support for SSSE3 built-in functions and code generation are 6585 available via -mssse3. 6586 * Support for SSE4.1 built-in functions and code generation are 6587 available via -msse4.1. 6588 * Support for SSE4.2 built-in functions and code generation are 6589 available via -msse4.2. 6590 * Both SSE4.1 and SSE4.2 support can be enabled via -msse4. 6591 * A new set of options -mpc32, -mpc64 and -mpc80 have been added to 6592 allow explicit control of x87 floating point precision. 6593 * Support for __float128 (TFmode) IEEE quad type and corresponding 6594 TCmode IEEE complex quad type is available via the soft-fp library 6595 on x86_64 targets. This includes basic arithmetic operations 6596 (addition, subtraction, negation, multiplication and division) on 6597 __float128 real and TCmode complex values, the full set of IEEE 6598 comparisons between __float128 values, conversions to and from 6599 float, double and long double floating point types, as well as 6600 conversions to and from signed or unsigned integer, signed or 6601 unsigned long integer and signed or unsigned quad (TImode) integer 6602 types. Additionally, all operations generate the full set of IEEE 6603 exceptions and support the full set of IEEE rounding modes. 6604 * GCC can now utilize the ACML library for vectorizing calls to a set 6605 of C99 functions on x86_64 if -mveclibabi=acml is specified and you 6606 link to an ACML ABI compatible library. 6607 6608 ARM 6609 6610 * Compiler and Library support for Thumb-2 and the ARMv7 architecture 6611 has been added. 6612 6613 CRIS 6614 6615 New features 6616 6617 * Compiler and Library support for the CRIS v32 architecture, as 6618 found in Axis Communications ETRAX FS and ARTPEC-3 chips, has been 6619 added. 6620 6621 Configuration changes 6622 6623 * The cris-*-elf target now includes support for CRIS v32, including 6624 libraries, through the -march=v32 option. 6625 * A new crisv32-*-elf target defaults to generate code for CRIS v32. 6626 * A new crisv32-*-linux* target defaults to generate code for CRIS 6627 v32. 6628 * The cris-*-aout target has been obsoleted. 6629 6630 Improved support for built-in functions 6631 6632 * GCC can now use the lz and swapwbr instructions to implement the 6633 __builtin_clz, __builtin_ctz and __builtin_ffs family of functions. 6634 * __builtin_bswap32 is now implemented using the swapwb instruction, 6635 when available. 6636 6637 m68k and ColdFire 6638 6639 New features 6640 6641 * Support for several new ColdFire processors has been added. You can 6642 generate code for them using the new -mcpu option. 6643 * All targets now support ColdFire processors. 6644 * m68k-uclinux targets have improved support for C++ constructors and 6645 destructors, and for shared libraries. 6646 * It is now possible to set breakpoints on the first or last line of 6647 a function, even if there are no statements on that line. 6648 6649 Optimizations 6650 6651 * Support for sibling calls has been added. 6652 * More use is now made of the ColdFire mov3q instruction. 6653 * __builtin_clz is now implemented using the ff1 ColdFire 6654 instruction, when available. 6655 * GCC now honors the -m68010 option. 68010 code now uses clr rather 6656 than move to zero volatile memory. 6657 * 68020 targets and above can now use symbol(index.size*scale) 6658 addresses for indexed array accesses. Earlier compilers would 6659 always load the symbol into a base register first. 6660 6661 Configuration changes 6662 6663 * All m68k and ColdFire targets now allow the default processor to be 6664 set at configure time using --with-cpu. 6665 * A --with-arch configuration option has been added. This option 6666 allows you to restrict a target to ColdFire or non-ColdFire 6667 processors. 6668 6669 Preprocessor macros 6670 6671 * An __mcfv*__ macro is now defined for all ColdFire targets. 6672 (Earlier versions of GCC only defined __mcfv4e__.) 6673 * __mcf_cpu_*, __mcf_family_* and __mcffpu__ macros have been added. 6674 * All targets now define __mc68010 and __mc68010__ when generating 6675 68010 code. 6676 6677 Command-line changes 6678 6679 * New command-line options -march, -mcpu, -mtune and -mhard-float 6680 have been added. These options apply to both m68k and ColdFire 6681 targets. 6682 * -mno-short, -mno-bitfield and -mno-rtd are now accepted as negative 6683 versions of -mshort, etc. 6684 * -fforce-addr has been removed. It is now ignored by the compiler. 6685 6686 Other improvements 6687 6688 * ColdFire targets now try to maintain a 4-byte-aligned stack where 6689 possible. 6690 * m68k-uclinux targets now try to avoid situations that lead to the 6691 load-time error: BINFMT_FLAT: reloc outside program. 6692 6693 MIPS 6694 6695 Changes to existing configurations 6696 6697 * libffi and libjava now support all three GNU/Linux ABIs: o32, n32 6698 and n64. Every GNU/Linux configuration now builds these libraries 6699 by default. 6700 * GNU/Linux configurations now generate -mno-shared code unless 6701 overridden by -fpic, -fPIC, -fpie or -fPIE. 6702 * mipsisa32*-linux-gnu configurations now generate hard-float code by 6703 default, just like other mipsisa32* and mips*-linux-gnu 6704 configurations. You can build a soft-float version of any 6705 mips*-linux-gnu configuration by passing --with-float=soft to 6706 configure. 6707 * mips-wrs-vxworks now supports run-time processes (RTPs). 6708 6709 Changes to existing command-line options 6710 6711 * The -march and -mtune options no longer accept 24k as a processor 6712 name. Please use 24kc, 24kf2_1 or 24kf1_1 instead. 6713 * The -march and -mtune options now accept 24kf2_1, 24kef2_1 and 6714 34kf2_1 as synonyms for 24kf, 24kef and 34kf respectively. The 6715 options also accept 24kf1_1, 24kef1_1 and 34kf1_1 as synonyms for 6716 24kx, 24kex and 34kx. 6717 6718 New configurations 6719 6720 GCC now supports the following configurations: 6721 * mipsisa32r2*-linux-gnu*, which generates MIPS32 revision 2 code by 6722 default. Earlier releases also recognized this configuration, but 6723 they treated it in the same way as mipsisa32*-linux-gnu*. Note that 6724 you can customize any mips*-linux-gnu* configuration to a 6725 particular ISA or processor by passing an appropriate --with-arch 6726 option to configure. 6727 * mipsisa*-sde-elf*, which provides compatibility with MIPS 6728 Technologies' SDE toolchains. The configuration uses the SDE 6729 libraries by default, but you can use it like other newlib-based 6730 ELF configurations by passing --with-newlib to configure. It is the 6731 only configuration besides mips64vr*-elf* to build MIPS16 as well 6732 as non-MIPS16 libraries. 6733 * mipsisa*-elfoabi*, which is similar to the general mipsisa*-elf* 6734 configuration, but uses the o32 and o64 ABIs instead of the 32-bit 6735 and 64-bit forms of the EABI. 6736 6737 New processors and application-specific extensions 6738 6739 * Support for the SmartMIPS ASE is available through the new 6740 -msmartmips option. 6741 * Support for revision 2 of the DSP ASE is available through the new 6742 -mdspr2 option. A new preprocessor macro called __mips_dsp_rev 6743 indicates the revision of the ASE in use. 6744 * Support for the 4KS and 74K families of processors is available 6745 through the -march and -mtune options. 6746 6747 Improved support for built-in functions 6748 6749 * GCC can now use load-linked, store-conditional and sync 6750 instructions to implement atomic built-in functions such as 6751 __sync_fetch_and_add. The memory reference must be 4 bytes wide for 6752 32-bit targets and either 4 or 8 bytes wide for 64-bit targets. 6753 * GCC can now use the clz and dclz instructions to implement the 6754 __builtin_ctz and __builtin_ffs families of functions. 6755 * There is a new __builtin___clear_cache function for flushing the 6756 instruction cache. GCC expands this function inline on MIPS32 6757 revision 2 targets, otherwise it calls the function specified by 6758 -mcache-flush-func. 6759 6760 MIPS16 improvements 6761 6762 * GCC can now compile objects that contain a mixture of MIPS16 and 6763 non-MIPS16 code. There are two new attributes, mips16 and nomips16, 6764 for specifying which mode a function should use. 6765 * A new option called -minterlink-mips16 makes non-MIPS16 code 6766 link-compatible with MIPS16 code. 6767 * After many bug fixes, the long-standing MIPS16 -mhard-float support 6768 should now work fairly reliably. 6769 * GCC can now use the MIPS16e save and restore instructions. 6770 * -fsection-anchors now works in MIPS16 mode. MIPS16 code compiled 6771 with -G0 -fsection-anchors is often smaller than code compiled with 6772 -G8. However, please note that you must usually compile all objects 6773 in your application with the same -G option; see the documentation 6774 of -G for details. 6775 * A new option called-mcode-readable specifies which instructions are 6776 allowed to load from the code segment. -mcode-readable=yes is the 6777 default and says that any instruction may load from the code 6778 segment. The other alternatives are -mcode-readable=pcrel, which 6779 says that only PC-relative MIPS16 instructions may load from the 6780 code segment, and -mcode-readable=no, which says that no 6781 instruction may do so. Please see the documentation for more 6782 details, including example uses. 6783 6784 Small-data improvements 6785 6786 There are three new options for controlling small data: 6787 * -mno-extern-sdata, which disables small-data accesses for 6788 externally-defined variables. Code compiled with -Gn 6789 -mno-extern-sdata will be link-compatible with any -G setting 6790 between -G0 and -Gn inclusive. 6791 * -mno-local-sdata, which disables the use of small-data sections for 6792 data that is not externally visible. This option can be a useful 6793 way of reducing small-data usage in less performance-critical parts 6794 of an application. 6795 * -mno-gpopt, which disables the use of the $gp register while still 6796 honoring the -G limit when placing externally-visible data. This 6797 option implies -mno-extern-sdata and -mno-local-sdata and it can be 6798 useful in situations where $gp does not necessarily hold the 6799 expected value. 6800 6801 Miscellaneous improvements 6802 6803 * There is a new option called -mbranch-cost for tweaking the 6804 perceived cost of branches. 6805 * If GCC is configured to use a version of GAS that supports the 6806 .gnu_attribute directive, it will use that directive to record 6807 certain properties of the output code. .gnu_attribute is new to GAS 6808 2.18. 6809 * There are two new function attributes, near and far, for overriding 6810 the command-line setting of -mlong-calls on a function-by-function 6811 basis. 6812 * -mfp64, which previously required a 64-bit target, now works with 6813 MIPS32 revision 2 targets as well. The mipsisa*-elfoabi* and 6814 mipsisa*-sde-elf* configurations provide suitable library support. 6815 * GCC now recognizes the -mdmx and -mmt options and passes them down 6816 to the assembler. It does nothing else with the options at present. 6817 6818 SPU (Synergistic Processor Unit) of the Cell Broadband Engine Architecture 6819 (BEA) 6820 6821 * Support has been added for this new architecture. 6822 6823 RS6000 (POWER/PowerPC) 6824 6825 * Support for the PowerPC 750CL paired-single instructions has been 6826 added with a new powerpc-*-linux*paired* target configuration. It 6827 is enabled by an associated -mpaired option and can be accessed 6828 using new built-in functions. 6829 * Support for auto-detecting architecture and system configuration to 6830 auto-select processor optimization tuning. 6831 * Support for VMX on AIX 5.3 has been added. 6832 * Support for AIX Version 6.1 has been added. 6833 6834 S/390, zSeries and System z9 6835 6836 * Support for the IBM System z9 EC/BC processor (z9 GA3) has been 6837 added. When using the -march=z9-ec option, the compiler will 6838 generate code making use of instructions provided by the decimal 6839 floating point facility and the floating point conversion facility 6840 (pfpo). Besides the instructions used to implement decimal floating 6841 point operations these facilities also contain instructions to move 6842 between general purpose and floating point registers and to modify 6843 and copy the sign-bit of floating point values. 6844 * When the -march=z9-ec option is used the new 6845 -mhard-dfp/-mno-hard-dfp options can be used to specify whether the 6846 decimal floating point hardware instructions will be used or not. 6847 If none of them is given the hardware support is enabled by 6848 default. 6849 * The -mstack-guard option can now be omitted when using stack 6850 checking via -mstack-size in order to let GCC choose a sensible 6851 stack guard value according to the frame size of each function. 6852 * Various changes to improve performance of generated code have been 6853 implemented, including: 6854 + The condition code set by an add logical with carry 6855 instruction is now available for overflow checks like: a + b + 6856 carry < b. 6857 + The test data class instruction is now used to implement 6858 sign-bit and infinity checks of binary and decimal floating 6859 point numbers. 6860 6861 SPARC 6862 6863 * Support for the Sun UltraSPARC T2 (Niagara 2) processor has been 6864 added. 6865 6866 Xtensa 6867 6868 * Stack unwinding for exception handling now uses by default a 6869 specialized version of DWARF unwinding. This is not 6870 binary-compatible with the setjmp/longjmp (sjlj) unwinding used for 6871 Xtensa with previous versions of GCC. 6872 * For Xtensa processors that include the Conditional Store option, 6873 the built-in functions for atomic memory access are now implemented 6874 using S32C1I instructions. 6875 * If the Xtensa NSA option is available, GCC will use it to implement 6876 the __builtin_ctz and __builtin_clz functions. 6877 6878 Documentation improvements 6879 6880 * Existing libstdc++ documentation has been edited and restructured 6881 into a single DocBook XML manual. The results can be viewed online 6882 [21]here. 6883 6884 Other significant improvements 6885 6886 * The compiler's --help command-line option has been extended so that 6887 it now takes an optional set of arguments. These arguments restrict 6888 the information displayed to specific classes of command-line 6889 options, and possibly only a subset of those options. It is also 6890 now possible to replace the descriptive text associated with each 6891 displayed option with an indication of its current value, or for 6892 binary options, whether it has been enabled or disabled. 6893 Here are some examples. The following will display all the options 6894 controlling warning messages: 6895 --help=warnings 6896 6897 Whereas this will display all the undocumented, target specific 6898 options: 6899 --help=target,undocumented 6900 6901 This sequence of commands will display the binary optimizations 6902 that are enabled by -O3: 6903 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts 6904 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts 6905 diff /tmp/O2-opts /tmp/O3-opts | grep enabled 6906 6907 * The configure options --with-pkgversion and --with-bugurl have been 6908 added. These allow distributors of GCC to include a 6909 distributor-specific string in manuals and --version output and to 6910 specify the URL for reporting bugs in their versions of GCC. 6911 6912 GCC 4.3.1 6913 6914 This is the [22]list of problem reports (PRs) from GCC's bug tracking 6915 system that are known to be fixed in the 4.3.1 release. This list might 6916 not be complete (that is, it is possible that some PRs that have been 6917 fixed are not listed here). 6918 6919 Target Specific Changes 6920 6921 IA-32/x86-64 6922 6923 ABI changes 6924 6925 * Starting with GCC 4.3.1, decimal floating point variables are 6926 aligned to their natural boundaries when they are passed on the 6927 stack for i386. 6928 6929 Command-line changes 6930 6931 * Starting with GCC 4.3.1, the -mcld option has been added to 6932 automatically generate a cld instruction in the prologue of 6933 functions that use string instructions. This option is used for 6934 backward compatibility on some operating systems and can be enabled 6935 by default for 32-bit x86 targets by configuring GCC with the 6936 --enable-cld configure option. 6937 6938 GCC 4.3.2 6939 6940 This is the [23]list of problem reports (PRs) from GCC's bug tracking 6941 system that are known to be fixed in the 4.3.2 release. This list might 6942 not be complete (that is, it is possible that some PRs that have been 6943 fixed are not listed here). 6944 6945 GCC 4.3.3 6946 6947 This is the [24]list of problem reports (PRs) from GCC's bug tracking 6948 system that are known to be fixed in the 4.3.3 release. This list might 6949 not be complete (that is, it is possible that some PRs that have been 6950 fixed are not listed here). 6951 6952 GCC 4.3.4 6953 6954 This is the [25]list of problem reports (PRs) from GCC's bug tracking 6955 system that are known to be fixed in the 4.3.4 release. This list might 6956 not be complete (that is, it is possible that some PRs that have been 6957 fixed are not listed here). 6958 6959 GCC 4.3.5 6960 6961 This is the [26]list of problem reports (PRs) from GCC's bug tracking 6962 system that are known to be fixed in the 4.3.5 release. This list might 6963 not be complete (that is, it is possible that some PRs that have been 6964 fixed are not listed here). 6965 6966 GCC 4.3.6 6967 6968 This is the [27]list of problem reports (PRs) from GCC's bug tracking 6969 system that are known to be fixed in the 4.3.6 release. This list might 6970 not be complete (that is, it is possible that some PRs that have been 6971 fixed are not listed here). 6972 6973 6974 For questions related to the use of GCC, please consult these web 6975 pages and the [28]GCC manuals. If that fails, the 6976 [29]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 6977 web pages and the development of GCC are welcome on our developer 6978 list at [30]gcc (a] gcc.gnu.org. All of [31]our lists have public 6979 archives. 6980 6981 Copyright (C) [32]Free Software Foundation, Inc. Verbatim copying and 6982 distribution of this entire article is permitted in any medium, 6983 provided this notice is preserved. 6984 6985 These pages are [33]maintained by the GCC team. Last modified 6986 2017-02-07[34]. 6987 6988 References 6989 6990 1. http://gcc.gnu.org/gcc-4.3/changes.html#4.3.5 6991 2. https://gmplib.org/ 6992 3. http://www.mpfr.org/ 6993 4. https://gcc.gnu.org/install/prerequisites.html 6994 5. https://gcc.gnu.org/ml/gcc-announce/2001/msg00000.html 6995 6. https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options 6996 7. http://gcc.gnu.org/gcc-4.3/porting_to.html 6997 8. http://www.mpfr.org/ 6998 9. http://www.mpfr.org/ 6999 10. http://www.mpfr.org/ 7000 11. https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html 7001 12. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html 7002 13. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html 7003 14. https://gcc.gnu.org/onlinedocs/libstdc++/manual/parallel_mode.html 7004 15. https://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#Code-Gen-Options 7005 16. https://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfinit-local-zero_007d-167 7006 17. https://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/GAMMA.html 7007 18. https://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/LGAMMA.html 7008 19. https://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html 7009 20. https://gcc.gnu.org/onlinedocs/gfortran/BOZ-literal-constants.html 7010 21. https://gcc.gnu.org/onlinedocs/libstdc++/ 7011 22. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.1 7012 23. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.2 7013 24. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.3 7014 25. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.4 7015 26. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.5 7016 27. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.6 7017 28. https://gcc.gnu.org/onlinedocs/ 7018 29. mailto:gcc-help (a] gcc.gnu.org 7019 30. mailto:gcc (a] gcc.gnu.org 7020 31. https://gcc.gnu.org/lists.html 7021 32. http://www.fsf.org/ 7022 33. https://gcc.gnu.org/about.html 7023 34. http://validator.w3.org/check/referer 7024 ====================================================================== 7025 http://gcc.gnu.org/gcc-4.2/index.html 7026 GCC 4.2 Release Series 7027 7028 May 19, 2008 7029 7030 The [1]GNU project and the GCC developers are pleased to announce the 7031 release of GCC 4.2.4. 7032 7033 This release is a bug-fix release, containing fixes for regressions in 7034 GCC 4.2.3 relative to previous releases of GCC. 7035 7036 This release series is no longer maintained. 7037 7038 Release History 7039 7040 GCC 4.2.4 7041 May 19, 2008 ([2]changes) 7042 7043 GCC 4.2.3 7044 February 1, 2008 ([3]changes) 7045 7046 GCC 4.2.2 7047 October 7, 2007 ([4]changes) 7048 7049 GCC 4.2.1 7050 July 18, 2007 ([5]changes) 7051 7052 GCC 4.2.0 7053 May 13, 2007 ([6]changes) 7054 7055 References and Acknowledgements 7056 7057 GCC used to stand for the GNU C Compiler, but since the compiler 7058 supports several other languages aside from C, it now stands for the 7059 GNU Compiler Collection. 7060 7061 A list of [7]successful builds is updated as new information becomes 7062 available. 7063 7064 The GCC developers would like to thank the numerous people that have 7065 contributed new features, improvements, bug fixes, and other changes as 7066 well as test results to GCC. This [8]amazing group of volunteers is 7067 what makes GCC successful. 7068 7069 For additional information about GCC please refer to the [9]GCC project 7070 web site or contact the [10]GCC development mailing list. 7071 7072 To obtain GCC please use [11]our mirror sites or [12]our SVN server. 7073 7074 7075 For questions related to the use of GCC, please consult these web 7076 pages and the [13]GCC manuals. If that fails, the 7077 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 7078 web pages and the development of GCC are welcome on our developer 7079 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 7080 archives. 7081 7082 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 7083 distribution of this entire article is permitted in any medium, 7084 provided this notice is preserved. 7085 7086 These pages are [18]maintained by the GCC team. Last modified 7087 2016-09-30[19]. 7088 7089 References 7090 7091 1. http://www.gnu.org/ 7092 2. http://gcc.gnu.org/gcc-4.2/changes.html 7093 3. http://gcc.gnu.org/gcc-4.2/changes.html 7094 4. http://gcc.gnu.org/gcc-4.2/changes.html 7095 5. http://gcc.gnu.org/gcc-4.2/changes.html 7096 6. http://gcc.gnu.org/gcc-4.2/changes.html 7097 7. http://gcc.gnu.org/gcc-4.2/buildstat.html 7098 8. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 7099 9. http://gcc.gnu.org/index.html 7100 10. mailto:gcc (a] gcc.gnu.org 7101 11. http://gcc.gnu.org/mirrors.html 7102 12. http://gcc.gnu.org/svn.html 7103 13. https://gcc.gnu.org/onlinedocs/ 7104 14. mailto:gcc-help (a] gcc.gnu.org 7105 15. mailto:gcc (a] gcc.gnu.org 7106 16. https://gcc.gnu.org/lists.html 7107 17. http://www.fsf.org/ 7108 18. https://gcc.gnu.org/about.html 7109 19. http://validator.w3.org/check/referer 7110 ====================================================================== 7111 http://gcc.gnu.org/gcc-4.2/changes.html 7112 GCC 4.2 Release Series 7113 Changes, New Features, and Fixes 7114 7115 Caveats 7116 7117 * GCC no longer accepts the -fshared-data option. This option has had 7118 no effect in any GCC 4 release; the targets to which the option 7119 used to apply had been removed before GCC 4.0. 7120 7121 General Optimizer Improvements 7122 7123 * New command-line options specify the possible relationships among 7124 parameters and between parameters and global data. For example, 7125 -fargument-noalias-anything specifies that arguments do not alias 7126 any other storage. 7127 Each language will automatically use whatever option is required by 7128 the language standard. You should not need to use these options 7129 yourself. 7130 7131 New Languages and Language specific improvements 7132 7133 * [1]OpenMP is now supported for the C, C++ and Fortran compilers. 7134 * New command-line options -fstrict-overflow and -Wstrict-overflow 7135 have been added. -fstrict-overflow tells the compiler that it may 7136 assume that the program follows the strict signed overflow 7137 semantics permitted for the language: for C and C++ this means that 7138 the compiler may assume that signed overflow does not occur. For 7139 example, a loop like 7140 for (i = 1; i > 0; i *= 2) 7141 7142 is presumably intended to continue looping until i overflows. With 7143 -fstrict-overflow, the compiler may assume that signed overflow 7144 will not occur, and transform this into an infinite loop. 7145 -fstrict-overflow is turned on by default at -O2, and may be 7146 disabled via -fno-strict-overflow. The -Wstrict-overflow option may 7147 be used to warn about cases where the compiler assumes that signed 7148 overflow will not occur. It takes five different levels: 7149 -Wstrict-overflow=1 to 5. See the [2]documentation for details. 7150 -Wstrict-overflow=1 is enabled by -Wall. 7151 * The new command-line option -fno-toplevel-reorder directs GCC to 7152 emit top-level functions, variables, and asm statements in the same 7153 order that they appear in the input file. This is intended to 7154 support existing code which relies on a particular ordering (for 7155 example, code which uses top-level asm statements to switch 7156 sections). For new code, it is generally better to use function and 7157 variable attributes. The -fno-toplevel-reorder option may be used 7158 for most cases which currently use -fno-unit-at-a-time. The 7159 -fno-unit-at-a-time option will be removed in some future version 7160 of GCC. If you know of a case which requires -fno-unit-at-a-time 7161 which is not fixed by -fno-toplevel-reorder, please open a bug 7162 report. 7163 7164 C family 7165 7166 * The pragma redefine_extname will now macro expand its tokens for 7167 compatibility with SunPRO. 7168 * In the next release of GCC, 4.3, -std=c99 or -std=gnu99 will direct 7169 GCC to handle inline functions as specified in the C99 standard. In 7170 preparation for this, GCC 4.2 will warn about any use of non-static 7171 inline functions in gnu99 or c99 mode. This new warning may be 7172 disabled with the new gnu_inline function attribute or the new 7173 -fgnu89-inline command-line option. Also, GCC 4.2 and later will 7174 define one of the preprocessor macros __GNUC_GNU_INLINE__ or 7175 __GNUC_STDC_INLINE__ to indicate the semantics of inline functions 7176 in the current compilation. 7177 * A new command-line option -Waddress has been added to warn about 7178 suspicious uses of memory addresses as, for example, using the 7179 address of a function in a conditional expression, and comparisons 7180 against the memory address of a string literal. This warning is 7181 enabled by -Wall. 7182 7183 C++ 7184 7185 * C++ visibility handling has been overhauled. 7186 Restricted visiblity is propagated from classes to members, from 7187 functions to local statics, and from templates and template 7188 arguments to instantiations, unless the latter has explicitly 7189 declared visibility. 7190 The visibility attribute for a class must come between the 7191 class-key and the name, not after the closing brace. 7192 Attributes are now allowed for enums and elaborated-type-specifiers 7193 that only declare a type. 7194 Members of the anonymous namespace are now local to a particular 7195 translation unit, along with any other declarations which use them, 7196 though they are still treated as having external linkage for 7197 language semantics. 7198 * The (undocumented) extension which permitted templates with default 7199 arguments to be bound to template template parameters with fewer 7200 parameters has been removed. For example: 7201 template <template <typename> class C> 7202 void f(C<double>) {} 7203 7204 template <typename T, typename U = int> 7205 struct S {}; 7206 7207 template void f(S<double>); 7208 7209 is no longer accepted by G++. The reason this code is not accepted 7210 is that S is a template with two parameters; therefore, it cannot 7211 be bound to C which has only one parameter. 7212 * The <?, >?, <?=, and >?= operators, deprecated in previous GCC 7213 releases, have been removed. 7214 * The command-line option -fconst-strings, deprecated in previous GCC 7215 releases, has been removed. 7216 * The configure variable enable-__cxa_atexit is now enabled by 7217 default for more targets. Enabling this variable is necessary in 7218 order for static destructors to be executed in the correct order, 7219 but it depends upon the presence of a non-standard C library in the 7220 target library in order to work. The variable is now enabled for 7221 more targets which are known to have suitable C libraries. 7222 * -Wextra will produce warnings for if statements with a semicolon as 7223 the only body, to catch code like: 7224 if (a); 7225 return 1; 7226 return 0; 7227 7228 To suppress the warning in valid cases, use { } instead. 7229 * The C++ frontend now also produces strict aliasing warnings when 7230 -fstrict-aliasing -Wstrict-aliasing is in effect. 7231 7232 Runtime Library (libstdc++) 7233 7234 * Added support for TR1 <random>, <complex>, and C compatibility 7235 headers. In addition, a lock-free version of shared_ptr was 7236 contributed as part of Phillip Jordan's Google Summer of Code 7237 project on lock-free containers. 7238 * In association with the Summer of Code work on lock-free 7239 containers, the interface for atomic builtins was adjusted, 7240 creating simpler alternatives for non-threaded code paths. Also, 7241 usage was consolidated and all elements were moved from namespace 7242 std to namespace__gnu_cxx. Affected interfaces are the functions 7243 __exchange_and_add, __atomic_add, and the objects __mutex, 7244 __recursive_mutex, and __scoped_lock. 7245 * Support for versioning weak symbol names via namespace association 7246 was added. However, as this changes the names of exported symbols, 7247 this is turned off by default in the current ABI. Intrepid users 7248 can enable this feature by using 7249 --enable-symvers=gnu-versioned-namespace during configuration. 7250 * Revised, simplified, and expanded policy-based associative 7251 containers, including data types for tree and trie forms 7252 (basic_tree, tree, trie), lists (list_update), and both 7253 collision-chaining and probing hash-based containers 7254 (basic_hash_table, cc_hash_table, gp_hash_table). More details per 7255 the [3]documentation. 7256 * The implementation of the debug mode was modified, whereby the 7257 debug namespaces were nested inside of namespace std and namespace 7258 __gnu_cxx in order to resolve some long standing corner cases 7259 involving name lookup. Debug functionality from the policy-based 7260 data structures was consolidated and enabled with the single macro, 7261 _GLIBCXX_DEBUG. See PR 26142 for more information. 7262 * Added extensions for type traits: __conditional_type, 7263 __numeric_traits, __add_unsigned, __removed_unsigned, __enable_if. 7264 * Added a typelist implementation for compile-time meta-programming. 7265 Elements for typelist construction and operation can be found 7266 within namespace __gnu_cxx::typelist. 7267 * Added a new allocator, __gnu_cxx::throw_allocator, for testing 7268 exception-safety. 7269 * Enabled library-wide visibility control, allowing -fvisibility to 7270 be used. 7271 * Consolidated all nested namespaces and the conversion of 7272 __gnu_internal implementation-private details to anonymous 7273 namespaces whenever possible. 7274 * Implemented LWG resolutions DR 431 and DR 538. 7275 7276 Fortran 7277 7278 * Support for allocatable components has been added (TR 15581 and 7279 Fortran 2003). 7280 * Support for the Fortran 2003 streaming IO extension has been added. 7281 * The GNU Fortran compiler now uses 4-byte record markers by default 7282 for unformatted files to be compatible with g77 and most other 7283 compilers. The implementation allows for records greater than 2 GB 7284 and is compatible with several other compilers. Older versions of 7285 gfortran used 8-byte record markers by default (on most systems). 7286 In order to change the length of the record markers, e.g. to read 7287 unformatted files created by older gfortran versions, the 7288 [4]-frecord-marker=8 option can be used. 7289 7290 Java (GCJ) 7291 7292 * A new command-line option -static-libgcj has been added for targets 7293 that use a linker compatible with GNU Binutils. As its name 7294 implies, this causes libgcj to be linked statically. In some cases 7295 this causes the resulting executable to start faster and use less 7296 memory than if the shared version of libgcj were used. However 7297 caution should be used as it can also cause essential parts of the 7298 library to be omitted. Some of these issues are discussed in: 7299 [5]https://gcc.gnu.org/wiki/Statically_linking_libgcj 7300 * fastjar is no longer bundled with GCC. To build libgcj, you will 7301 need either InfoZIP (both zip and unzip) or an external jar 7302 program. In the former case, the GCC build will install a jar shell 7303 script that is based on InfoZIP and provides the same functionality 7304 as fastjar. 7305 7306 New Targets and Target Specific Improvements 7307 7308 IA-32/x86-64 7309 7310 * -mtune=generic can now be used to generate code running well on 7311 common x86 chips. This includes AMD Athlon, AMD Opteron, Intel 7312 Pentium-M, Intel Pentium 4 and Intel Core 2. 7313 * -mtune=native and -march=native will produce code optimized for the 7314 host architecture as detected using the cpuid instruction. 7315 * Added a new command-line option -fstackrealign and and 7316 __attribute__ ((force_align_arg_pointer)) to realign the stack at 7317 runtime. This allows functions compiled with a vector-aligned stack 7318 to be invoked from legacy objects that keep only word-alignment. 7319 7320 SPARC 7321 7322 * The default CPU setting has been changed from V7 to V9 in 32-bit 7323 mode on Solaris 7 and above. This is already the case in 64-bit 7324 mode. It can be overridden by specifying --with-cpu at configure 7325 time. 7326 * Back-end support of built-in functions for atomic memory access has 7327 been implemented. 7328 * Support for the Sun UltraSPARC T1 (Niagara) processor has been 7329 added. 7330 7331 M32C 7332 7333 * Various bug fixes have made some functions (notably, functions 7334 returning structures) incompatible with previous releases. 7335 Recompiling all libraries is recommended. Note that code quality 7336 has considerably improved since 4.1, making a recompile even more 7337 beneficial. 7338 7339 MIPS 7340 7341 * Added support for the Broadcom SB-1A core. 7342 7343 IA-64 7344 7345 * Added support for IA-64 data and control speculation. By default 7346 speculation is enabled only during second scheduler pass. A number 7347 of machine flags was introduced to control the usage of speculation 7348 for both scheduler passes. 7349 7350 HPPA 7351 7352 * Added Java language support (libffi and libjava) for 32-bit HP-UX 7353 11 target. 7354 7355 Obsolete Systems 7356 7357 Documentation improvements 7358 7359 PDF Documentation 7360 7361 * A make pdf target has been added to the top-level makefile, 7362 enabling automated production of PDF documentation files. 7363 (Front-ends external to GCC should modify their Make-lang.in file 7364 to add a lang.pdf: target.) 7365 7366 Other significant improvements 7367 7368 Build system improvements 7369 7370 * All the components of the compiler are now bootstrapped by default. 7371 This improves the resilience to bugs in the system compiler or 7372 binary compatibility problems, as well as providing better testing 7373 of GCC 4.2 itself. In addition, if you build the compiler from a 7374 combined tree, the assembler, linker, etc. will also be 7375 bootstrapped (i.e. built with themselves). 7376 You can disable this behavior, and go back to the pre-GCC 4.2 set 7377 up, by configuring GCC with --disable-bootstrap. 7378 * The rules that configure follows to find target tools resemble more 7379 closely the locations that the built compiler will search. In 7380 addition, you can use the new configure option --with-target-tools 7381 to specify where to find the target tools used during the build, 7382 without affecting what the built compiler will use. 7383 This can be especially useful when building packages of GCC. For 7384 example, you may want to build GCC with GNU as or ld, even if the 7385 resulting compiler to work with the native assembler and linker. To 7386 do so, you can use --with-target-tools to point to the native 7387 tools. 7388 7389 Incompatible changes to the build system 7390 7391 * Front-ends external to GCC should modify their Make-lang.in file to 7392 replace double-colon rules (e.g. dvi::) with normal rules (like 7393 lang.dvi:). Front-end makefile hooks do not use double-colon rules 7394 anymore. 7395 * Up to GCC 4.1, a popular way to specify the target tools used 7396 during the build was to create directories named gas, binutils, 7397 etc. in the build tree, and create links to the tools from there. 7398 This does not work any more when the compiler is bootstrapped. The 7399 new configure option --with-target-tools provides a better way to 7400 achieve the same effect, and works for all native and cross 7401 settings. 7402 7403 7404 For questions related to the use of GCC, please consult these web 7405 pages and the [6]GCC manuals. If that fails, the 7406 [7]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 7407 web pages and the development of GCC are welcome on our developer 7408 list at [8]gcc (a] gcc.gnu.org. All of [9]our lists have public archives. 7409 7410 Copyright (C) [10]Free Software Foundation, Inc. Verbatim copying and 7411 distribution of this entire article is permitted in any medium, 7412 provided this notice is preserved. 7413 7414 These pages are [11]maintained by the GCC team. Last modified 7415 2017-02-07[12]. 7416 7417 References 7418 7419 1. http://gcc.gnu.org/projects/gomp/ 7420 2. https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html 7421 3. https://gcc.gnu.org/onlinedocs/libstdc++/ext/pb_ds/index.html 7422 4. https://gcc.gnu.org/onlinedocs/gfortran/Runtime-Options.html 7423 5. https://gcc.gnu.org/wiki/Statically_linking_libgcj 7424 6. https://gcc.gnu.org/onlinedocs/ 7425 7. mailto:gcc-help (a] gcc.gnu.org 7426 8. mailto:gcc (a] gcc.gnu.org 7427 9. https://gcc.gnu.org/lists.html 7428 10. http://www.fsf.org/ 7429 11. https://gcc.gnu.org/about.html 7430 12. http://validator.w3.org/check/referer 7431 ====================================================================== 7432 http://gcc.gnu.org/gcc-4.1/index.html 7433 GCC 4.1 Release Series 7434 7435 February 13, 2007 7436 7437 The [1]GNU project and the GCC developers are pleased to announce the 7438 release of GCC 4.1.2. 7439 7440 This release is a bug-fix release, containing fixes for regressions in 7441 GCC 4.1.1 relative to previous releases of GCC. 7442 7443 This release series is no longer maintained. 7444 7445 Release History 7446 7447 GCC 4.1.2 7448 February 13, 2007 ([2]changes) 7449 7450 GCC 4.1.1 7451 May 24, 2006 ([3]changes) 7452 7453 GCC 4.1.0 7454 February 28, 2006 ([4]changes) 7455 7456 References and Acknowledgements 7457 7458 GCC used to stand for the GNU C Compiler, but since the compiler 7459 supports several other languages aside from C, it now stands for the 7460 GNU Compiler Collection. 7461 7462 A list of [5]successful builds is updated as new information becomes 7463 available. 7464 7465 The GCC developers would like to thank the numerous people that have 7466 contributed new features, improvements, bug fixes, and other changes as 7467 well as test results to GCC. This [6]amazing group of volunteers is 7468 what makes GCC successful. 7469 7470 For additional information about GCC please refer to the [7]GCC project 7471 web site or contact the [8]GCC development mailing list. 7472 7473 To obtain GCC please use [9]our mirror sites or [10]our SVN server. 7474 7475 7476 For questions related to the use of GCC, please consult these web 7477 pages and the [11]GCC manuals. If that fails, the 7478 [12]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 7479 web pages and the development of GCC are welcome on our developer 7480 list at [13]gcc (a] gcc.gnu.org. All of [14]our lists have public 7481 archives. 7482 7483 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and 7484 distribution of this entire article is permitted in any medium, 7485 provided this notice is preserved. 7486 7487 These pages are [16]maintained by the GCC team. Last modified 7488 2016-09-30[17]. 7489 7490 References 7491 7492 1. http://www.gnu.org/ 7493 2. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2 7494 3. http://gcc.gnu.org/gcc-4.1/changes.html 7495 4. http://gcc.gnu.org/gcc-4.1/changes.html 7496 5. http://gcc.gnu.org/gcc-4.1/buildstat.html 7497 6. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 7498 7. http://gcc.gnu.org/index.html 7499 8. mailto:gcc (a] gcc.gnu.org 7500 9. http://gcc.gnu.org/mirrors.html 7501 10. http://gcc.gnu.org/svn.html 7502 11. https://gcc.gnu.org/onlinedocs/ 7503 12. mailto:gcc-help (a] gcc.gnu.org 7504 13. mailto:gcc (a] gcc.gnu.org 7505 14. https://gcc.gnu.org/lists.html 7506 15. http://www.fsf.org/ 7507 16. https://gcc.gnu.org/about.html 7508 17. http://validator.w3.org/check/referer 7509 ====================================================================== 7510 http://gcc.gnu.org/gcc-4.1/changes.html 7511 GCC 4.1 Release Series 7512 Changes, New Features, and Fixes 7513 7514 The latest release in the 4.1 release series is [1]GCC 4.1.2. 7515 7516 Caveats 7517 7518 General Optimizer Improvements 7519 7520 * GCC now has infrastructure for inter-procedural optimizations and 7521 the following inter-procedural optimizations are implemented: 7522 + Profile guided inlining. When doing profile feedback guided 7523 optimization, GCC can now use the profile to make better 7524 informed decisions on whether inlining of a function is 7525 profitable or not. This means that GCC will no longer inline 7526 functions at call sites that are not executed very often, and 7527 that functions at hot call sites are more likely to be 7528 inlined. 7529 A new parameter min-inline-recursive-probability is also now 7530 available to throttle recursive inlining of functions with 7531 small average recursive depths. 7532 + Discovery of pure and const functions, a form of side-effects 7533 analysis. While older GCC releases could also discover such 7534 special functions, the new IPA-based pass runs earlier so that 7535 the results are available to more optimizers. The pass is also 7536 simply more powerful than the old one. 7537 + Analysis of references to static variables and type escape 7538 analysis, also forms of side-effects analysis. The results of 7539 these passes allow the compiler to be less conservative about 7540 call-clobbered variables and references. This results in more 7541 redundant loads being eliminated and in making static 7542 variables candidates for register promotion. 7543 + Improvement of RTL-based alias analysis. The results of type 7544 escape analysis are fed to the RTL type-based alias analyzer, 7545 allowing it to disambiguate more memory references. 7546 + Interprocedural constant propagation and function versioning. 7547 This pass looks for functions that are always called with the 7548 same constant value for one or more of the function arguments, 7549 and propagates those constants into those functions. 7550 + GCC will now eliminate static variables whose usage was 7551 optimized out. 7552 + -fwhole-program --combine can now be used to make all 7553 functions in program static allowing whole program 7554 optimization. As an exception, the main function and all 7555 functions marked with the new externally_visible attribute are 7556 kept global so that programs can link with runtime libraries. 7557 * GCC can now do a form of partial dead code elimination (PDCE) that 7558 allows code motion of expressions to the paths where the result of 7559 the expression is actually needed. This is not always a win, so the 7560 pass has been limited to only consider profitable cases. Here is an 7561 example: 7562 int foo (int *, int *); 7563 int 7564 bar (int d) 7565 { 7566 int a, b, c; 7567 b = d + 1; 7568 c = d + 2; 7569 a = b + c; 7570 if (d) 7571 { 7572 foo (&b, &c); 7573 a = b + c; 7574 } 7575 printf ("%d\n", a); 7576 } 7577 7578 The a = b + c can be sunk to right before the printf. Normal code 7579 sinking will not do this, it will sink the first one above into the 7580 else-branch of the conditional jump, which still gives you two 7581 copies of the code. 7582 * GCC now has a value range propagation pass. This allows the 7583 compiler to eliminate bounds checks and branches. The results of 7584 the pass can also be used to accurately compute branch 7585 probabilities. 7586 * The pass to convert PHI nodes to straight-line code (a form of 7587 if-conversion for GIMPLE) has been improved significantly. The two 7588 most significant improvements are an improved algorithm to 7589 determine the order in which the PHI nodes are considered, and an 7590 improvement that allow the pass to consider if-conversions of basic 7591 blocks with more than two predecessors. 7592 * Alias analysis improvements. GCC can now differentiate between 7593 different fields of structures in Tree-SSA's virtual operands form. 7594 This lets stores/loads from non-overlapping structure fields not 7595 conflict. A new algorithm to compute points-to sets was contributed 7596 that can allows GCC to see now that p->a and p->b, where p is a 7597 pointer to a structure, can never point to the same field. 7598 * Various enhancements to auto-vectorization: 7599 + Incrementally preserve SSA form when vectorizing. 7600 + Incrementally preserve loop-closed form when vectorizing. 7601 + Improvements to peeling for alignment: generate better code 7602 when the misalignment of an access is known at compile time, 7603 or when different accesses are known to have the same 7604 misalignment, even if the misalignment amount itself is 7605 unknown. 7606 + Consider dependence distance in the vectorizer. 7607 + Externalize generic parts of data reference analysis to make 7608 this analysis available to other passes. 7609 + Vectorization of conditional code. 7610 + Reduction support. 7611 * GCC can now partition functions in sections of hot and cold code. 7612 This can significantly improve performance due to better 7613 instruction cache locality. This feature works best together with 7614 profile feedback driven optimization. 7615 * A new pass to avoid saving of unneeded arguments to the stack in 7616 vararg functions if the compiler can prove that they will not be 7617 needed. 7618 * Transition of basic block profiling to tree level implementation 7619 has been completed. The new implementation should be considerably 7620 more reliable (hopefully avoiding profile mismatch errors when 7621 using -fprofile-use or -fbranch-probabilities) and can be used to 7622 drive higher level optimizations, such as inlining. 7623 The -ftree-based-profiling command-line option was removed and 7624 -fprofile-use now implies disabling old RTL level loop optimizer 7625 (-fno-loop-optimize). Speculative prefetching optimization 7626 (originally enabled by -fspeculative-prefetching) was removed. 7627 7628 New Languages and Language specific improvements 7629 7630 C and Objective-C 7631 7632 * The old Bison-based C and Objective-C parser has been replaced by a 7633 new, faster hand-written recursive-descent parser. 7634 7635 Ada 7636 7637 * The build infrastructure for the Ada runtime library and tools has 7638 been changed to be better integrated with the rest of the build 7639 infrastructure of GCC. This should make doing cross builds of Ada a 7640 bit easier. 7641 7642 C++ 7643 7644 * ARM-style name-injection of friend declarations is no longer the 7645 default. For example: 7646 struct S { 7647 friend void f(); 7648 }; 7649 7650 void g() { f(); } 7651 will not be accepted; instead a declaration of f will need to be 7652 present outside of the scope of S. The new -ffriend-injection 7653 option will enable the old behavior. 7654 * The (undocumented) extension which permitted templates with default 7655 arguments to be bound to template template parameters with fewer 7656 parameters has been deprecated, and will be removed in the next 7657 major release of G++. For example: 7658 template <template <typename> class C> 7659 void f(C<double>) {} 7660 7661 template <typename T, typename U = int> 7662 struct S {}; 7663 7664 template void f(S<double>); 7665 7666 makes use of the deprecated extension. The reason this code is not 7667 valid ISO C++ is that S is a template with two parameters; 7668 therefore, it cannot be bound to C which has only one parameter. 7669 7670 Runtime Library (libstdc++) 7671 7672 * Optimization work: 7673 + A new implementation of std::search_n is provided, better 7674 performing in case of random access iterators. 7675 + Added further efficient specializations of istream functions, 7676 i.e., character array and string extractors. 7677 + Other smaller improvements throughout. 7678 * Policy-based associative containers, designed for high-performance, 7679 flexibility and semantic safety are delivered in ext/pb_assoc. 7680 * A versatile string class, __gnu_cxx::__versa_string, providing 7681 facilities conforming to the standard requirements for 7682 basic_string, is delivered in <ext/vstring.h>. In particular: 7683 + Two base classes are provided: the default one avoids 7684 reference counting and is optimized for short strings; the 7685 alternate one, still uses it while improving in a few low 7686 level areas (e.g., alignment). See vstring_fwd.h for some 7687 useful typedefs. 7688 + Various algorithms have been rewritten (e.g., replace), the 7689 code streamlined and simple optimizations added. 7690 + Option 3 of DR 431 is implemented for both available bases, 7691 thus improving the support for stateful allocators. 7692 * As usual, many bugs have been fixed (e.g., libstdc++/13583, 7693 libstdc++/23953) and LWG resolutions put into effect for the first 7694 time (e.g., DR 280, DR 464, N1780 recommendations for DR 233, TR1 7695 Issue 6.19). The implementation status of TR1 is now tracked in the 7696 docs in tr1.html. 7697 7698 Objective-C++ 7699 7700 * A new language front end for Objective-C++ has been added. This 7701 language allows users to mix the object oriented features of 7702 Objective-C with those of C++. 7703 7704 Java (GCJ) 7705 7706 * Core library (libgcj) updates based on GNU Classpath 0.15 - 0.19 7707 features (plus some 0.20 bug-fixes) 7708 + Networking 7709 o The java.net.HttpURLConnection implementation no longer 7710 buffers the entire response body in memory. This means 7711 that response bodies larger than available memory can now 7712 be handled. 7713 + (N)IO 7714 o NIO FileChannel.map implementation, fast bulk put 7715 implementation for DirectByteBuffer (speeds up this 7716 method 10x). 7717 o FileChannel.lock() and FileChannel.force() implemented. 7718 + XML 7719 o gnu.xml fix for nodes created outside a namespace 7720 context. 7721 o Add support for output indenting and 7722 cdata-section-elements output instruction in 7723 xml.transform. 7724 o xml.xpath corrections for cases where elements/attributes 7725 might have been created in non-namespace-aware mode. 7726 Corrections to handling of XSL variables and minor 7727 conformance updates. 7728 + AWT 7729 o GNU JAWT implementation, the AWT Native Interface, which 7730 allows direct access to native screen resources from 7731 within a Canvas's paint method. GNU Classpath Examples 7732 comes with a Demo, see libjava/classpath/examples/README. 7733 o awt.datatransfer updated to 1.5 with support for 7734 FlavorEvents. The gtk+ awt peers now allow copy/paste of 7735 text, images, URIs/files and serialized objects with 7736 other applications and tracking clipboard change events 7737 with gtk+ 2.6 (for gtk+ 2.4 only text and serialized 7738 objects are supported). A GNU Classpath Examples 7739 datatransfer Demo was added to show the new 7740 functionality. 7741 o Split gtk+ awt peers event handling in two threads and 7742 improve gdk lock handling (solves several awt lock ups). 7743 o Speed up awt Image loading. 7744 o Better gtk+ scrollbar peer implementation when using gtk+ 7745 >= 2.6. 7746 o Handle image loading errors correctly for gdkpixbuf and 7747 MediaTracker. 7748 o Better handle GDK lock. Properly prefix gtkpeer native 7749 functions (cp_gtk). 7750 o GdkGraphics2D has been updated to use Cairo 0.5.x or 7751 higher. 7752 o BufferedImage and GtkImage rewrites. All image drawing 7753 operations should now work correctly (flipping requires 7754 gtk+ >= 2.6) 7755 o Future Graphics2D, image and text work is documented at: 7756 [2]http://developer.classpath.org/mediation/ClasspathGrap 7757 hicsImagesText 7758 o When gtk+ 2.6 or higher is installed the default log 7759 handler will produce stack traces whenever a WARNING, 7760 CRITICAL or ERROR message is produced. 7761 + Free Swing 7762 o The RepaintManager has been reworked for more efficient 7763 painting, especially for large GUIs. 7764 o The layout manager OverlayLayout has been implemented, 7765 the BoxLayout has been rewritten to make use of the 7766 SizeRequirements utility class and caching for more 7767 efficient layout. 7768 o Improved accessibility support. 7769 o Significant progress has been made in the implementation 7770 of the javax.swing.plaf.metal package, with most UI 7771 delegates in a working state now. Please test this with 7772 your own applications and provide feedback that will help 7773 us to improve this package. 7774 o The GUI demo (gnu.classpath.examples.swing.Demo) has been 7775 extended to highlight various features in our Free Swing 7776 implementation. And it includes a look and feel switcher 7777 for Metal (default), Ocean and GNU themes. 7778 o The javax.swing.plaf.multi package is now implemented. 7779 o Editing and several key actions for JTree and JTable were 7780 implemented. 7781 o Lots of icons and look and feel improvements for Free 7782 Swing basic and metal themes were added. Try running the 7783 GNU Classpath Swing Demo in examples 7784 (gnu.classpath.examples.swing.Demo) with: 7785 -Dswing.defaultlaf=javax.swing.plaf.basic.BasicLookAndFee 7786 l or 7787 -Dswing.defaultlaf=javax.swing.plaf.metal.MetalLookAndFee 7788 l 7789 o Start of styled text capabilites for java.swing.text. 7790 o DefaultMutableTreeNode pre-order, post-order, depth-first 7791 and breadth-first traversal enumerations implemented. 7792 o JInternalFrame colors and titlebar draw properly. 7793 o JTree is working up to par (icons, selection and keyboard 7794 traversal). 7795 o JMenus were made more compatible in visual and 7796 programmatic behavior. 7797 o JTable changeSelection and multiple selections 7798 implemented. 7799 o JButton and JToggleButton change states work properly 7800 now. 7801 o JFileChooser fixes. 7802 o revalidate() and repaint() fixes which make Free Swing 7803 much more responsive. 7804 o MetalIconFactory implemented. 7805 o Free Swing Top-Level Compatibility. JFrame, JDialog, 7806 JApplet, JInternalFrame, and JWindow are now 1.5 7807 compatible in the sense that you can call add() and 7808 setLayout() directly on them, which will have the same 7809 effect as calling getContentPane().add() and 7810 getContentPane().setLayout(). 7811 o The JTree interface has been completed. JTrees now 7812 recognizes mouse clicks and selections work. 7813 o BoxLayout works properly now. 7814 o Fixed GrayFilter to actually work. 7815 o Metal SplitPane implemented. 7816 o Lots of Free Swing text and editor stuff work now. 7817 + Free RMI and Corba 7818 o Andrew Watson, Vice President and Technical Director of 7819 the Object Management Group, has officially assigned us 7820 20 bit Vendor Minor Code Id: 0x47430 ("GC") that will 7821 mark remote classpath-specific system exceptions. 7822 Obtaining the VMCID means that GNU Classpath now is a 7823 recogniseable type of node in a highly interoperable 7824 CORBA world. 7825 o GNU Classpath now includes the first working draft to 7826 support the RMI over IIOP protocol. The current 7827 implementation is capable of remote invocations, 7828 transferring various Serializables and Externalizables 7829 via RMI-IIOP protocol. It can flatten graphs and, at 7830 least for the simple cases, is interoperable with 1.5 7831 JDKs. 7832 o org.omg.PortableInterceptor and related functionality in 7833 other packages is now implemented: 7834 # The sever and client interceptors work as required 7835 since 1.4. 7836 # The IOR interceptor works as needed for 1.5. 7837 o The org.omg.DynamicAny package is completed and passes 7838 the prepared tests. 7839 o The Portable Object Adapter should now support the output 7840 of the recent IDL to java compilers. These compilers now 7841 generate servants and not CORBA objects as before, making 7842 the output depend on the existing POA implementation. 7843 Completing POA means that such code can already be tried 7844 to run on Classpath. Our POA is tested for the following 7845 usager scenarios: 7846 # POA converts servant to the CORBA object. 7847 # Servant provides to the CORBA object. 7848 # POA activates new CORBA object with the given Object 7849 Id (byte array) that is later accessible for the 7850 servant. 7851 # During the first call, the ServantActivator provides 7852 servant for this and all subsequent calls on the 7853 current object. 7854 # During each call, the ServantLocator provides 7855 servant for this call only. 7856 # ServantLocator or ServantActivator forwards call to 7857 another server. 7858 # POA has a single servant, responsible for all 7859 objects. 7860 # POA has a default servant, but some objects are 7861 explicitly connected to they specific servants. 7862 The POA is verified using tests from the former 7863 cost.omg.org. 7864 o The CORBA implementation is now a working prototype that 7865 should support features up to 1.3 inclusive. We invite 7866 groups writing CORBA dependent applications to try 7867 Classpath implementation, reporting any possible bugs. 7868 The CORBA prototype is interoperable with Sun's 7869 implementation v 1.4, transferring object references, 7870 primitive types, narrow and wide strings, arrays, 7871 structures, trees, abstract interfaces and value types 7872 (feature of CORBA 2.3) between these two platforms. 7873 Remote exceptions are transferred and handled correctly. 7874 The stringified object references (IORs) from various 7875 sources are parsed as required. The transient (for 7876 current session) and permanent (till jre restart) 7877 redirections work. Both Little and Big Endian encoded 7878 messages are accepted. The implementation is verified 7879 using tests from the former cost.omg.org. The current 7880 release includes working examples (see the examples 7881 directory), demonstrating the client-server 7882 communication, using either CORBA Request or IDL-based 7883 stub (usually generated by a IDL to java compiler). These 7884 examples also show how to use the Classpath CORBA naming 7885 service. The IDL to java compiler is not yet written, but 7886 as our library must be compatible, it naturally accepts 7887 the output of other idlj implementations. 7888 + Misc 7889 o Updated TimeZone data against Olson tzdata2005l. 7890 o Make zip and jar packages UTF-8 clean. 7891 o "native" code builds and compiles (warning free) on 7892 Darwin and Solaris. 7893 o java.util.logging.FileHandler now rotates files. 7894 o Start of a generic JDWP framework in gnu/classpath/jdwp. 7895 This is unfinished, but feedback (at classpath (a] gnu.org) 7896 from runtime hackers is greatly appreciated. Although 7897 most of the work is currently being done around gcj/gij 7898 we want this framework to be as VM neutral as possible. 7899 Early design is described in: 7900 [3]https://gcc.gnu.org/ml/java/2005-05/msg00260.html 7901 o QT4 AWT peers, enable by giving configure 7902 --enable-qt-peer. Included, but not ready for production 7903 yet. They are explicitly disabled and not supported. But 7904 if you want to help with the development of these new 7905 features we are interested in feedback. You will have to 7906 explicitly enable them to try them out (and they will 7907 most likely contain bugs). 7908 o Documentation fixes all over the place. See 7909 [4]http://developer.classpath.org/doc/ 7910 7911 New Targets and Target Specific Improvements 7912 7913 IA-32/x86-64 7914 7915 * The x86-64 medium model (that allows building applications whose 7916 data segment exceeds 4GB) was redesigned to match latest ABI draft. 7917 New implementation split large datastructures into separate segment 7918 improving performance of accesses to small datastructures and also 7919 allows linking of small model libraries into medium model programs 7920 as long as the libraries are not accessing the large datastructures 7921 directly. Medium model is also supported in position independent 7922 code now. 7923 The ABI change results in partial incompatibility among medium 7924 model objects. Linking medium model libraries (or objects) compiled 7925 with new compiler into medium model program compiled with older 7926 will likely result in exceeding ranges of relocations. 7927 Binutils 2.16.91 or newer are required for compiling medium model 7928 now. 7929 7930 RS6000 (POWER/PowerPC) 7931 7932 * The AltiVec vector primitives in <altivec.h> are now implemented in 7933 a way that puts a smaller burden on the preprocessor, instead 7934 processing the "overloading" in the front ends. This should benefit 7935 compilation speed on AltiVec vector code. 7936 * AltiVec initializers now are generated more efficiently. 7937 * The popcountb instruction available on POWER5 now is generated. 7938 * The floating point round to integer instructions available on 7939 POWER5+ now is generated. 7940 * Floating point divides can be synthesized using the floating point 7941 reciprocal estimate instructions. 7942 * Double precision floating point constants are initialized as single 7943 precision values if they can be represented exactly. 7944 7945 S/390, zSeries and System z9 7946 7947 * Support for the IBM System z9 109 processor has been added. When 7948 using the -march=z9-109 option, the compiler will generate code 7949 making use of instructions provided by the extended immediate 7950 facility. 7951 * Support for 128-bit IEEE floating point has been added. When using 7952 the -mlong-double-128 option, the compiler will map the long double 7953 data type to 128-bit IEEE floating point. Using this option 7954 constitutes an ABI change, and requires glibc support. 7955 * Various changes to improve performance of generated code have been 7956 implemented, including: 7957 + In functions that do not require a literal pool, register %r13 7958 (which is traditionally reserved as literal pool pointer), can 7959 now be freely used for other purposes by the compiler. 7960 + More precise tracking of register use allows the compiler to 7961 generate more efficient function prolog and epilog code in 7962 certain cases. 7963 + The SEARCH STRING, COMPARE LOGICAL STRING, and MOVE STRING 7964 instructions are now used to implement C string functions. 7965 + The MOVE CHARACTER instruction with single byte overlap is now 7966 used to implement the memset function with non-zero fill byte. 7967 + The LOAD ZERO instructions are now used where appropriate. 7968 + The INSERT CHARACTERS UNDER MASK, STORE CHARACTERS UNDER MASK, 7969 and INSERT IMMEDIATE instructions are now used more frequently 7970 to optimize bitfield operations. 7971 + The BRANCH ON COUNT instruction is now used more frequently. 7972 In particular, the fact that a loop contains a subroutine call 7973 no longer prevents the compiler from using this instruction. 7974 + The compiler is now aware that all shift and rotate 7975 instructions implicitly truncate the shift count to six bits. 7976 * Back-end support for the following generic features has been 7977 implemented: 7978 + The full set of [5]built-in functions for atomic memory 7979 access. 7980 + The -fstack-protector feature. 7981 + The optimization pass avoiding unnecessary stores of incoming 7982 argument registers in functions with variable argument list. 7983 7984 SPARC 7985 7986 * The default code model in 64-bit mode has been changed from 7987 Medium/Anywhere to Medium/Middle on Solaris. 7988 * TLS support is disabled by default on Solaris prior to release 10. 7989 It can be enabled on TLS-capable Solaris 9 versions (4/04 release 7990 and later) by specifying --enable-tls at configure time. 7991 7992 MorphoSys 7993 7994 * Support has been added for this new architecture. 7995 7996 Obsolete Systems 7997 7998 Documentation improvements 7999 8000 Other significant improvements 8001 8002 * GCC can now emit code for protecting applications from 8003 stack-smashing attacks. The protection is realized by buffer 8004 overflow detection and reordering of stack variables to avoid 8005 pointer corruption. 8006 * Some built-in functions have been fortified to protect them against 8007 various buffer overflow (and format string) vulnerabilities. 8008 Compared to the mudflap bounds checking feature, the safe builtins 8009 have far smaller overhead. This means that programs built using 8010 safe builtins should not experience any measurable slowdown. 8011 8012 GCC 4.1.2 8013 8014 This is the [6]list of problem reports (PRs) from GCC's bug tracking 8015 system that are known to be fixed in the 4.1.2 release. This list might 8016 not be complete (that is, it is possible that some PRs that have been 8017 fixed are not listed here). 8018 8019 When generating code for a shared library, GCC now recognizes that 8020 global functions may be replaced when the program runs. Therefore, it 8021 is now more conservative in deducing information from the bodies of 8022 functions. For example, in this example: 8023 void f() {} 8024 void g() { 8025 try { f(); } 8026 catch (...) { 8027 cout << "Exception"; 8028 } 8029 } 8030 8031 G++ would previously have optimized away the catch clause, since it 8032 would have concluded that f cannot throw exceptions. Because users may 8033 replace f with another function in the main body of the program, this 8034 optimization is unsafe, and is no longer performed. If you wish G++ to 8035 continue to optimize as before, you must add a throw() clause to the 8036 declaration of f to make clear that it does not throw exceptions. 8037 8038 8039 For questions related to the use of GCC, please consult these web 8040 pages and the [7]GCC manuals. If that fails, the 8041 [8]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 8042 web pages and the development of GCC are welcome on our developer 8043 list at [9]gcc (a] gcc.gnu.org. All of [10]our lists have public 8044 archives. 8045 8046 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and 8047 distribution of this entire article is permitted in any medium, 8048 provided this notice is preserved. 8049 8050 These pages are [12]maintained by the GCC team. Last modified 8051 2016-01-30[13]. 8052 8053 References 8054 8055 1. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2 8056 2. http://developer.classpath.org/mediation/ClasspathGraphicsImagesText 8057 3. https://gcc.gnu.org/ml/java/2005-05/msg00260.html 8058 4. http://developer.classpath.org/doc/ 8059 5. https://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html 8060 6. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.1.2 8061 7. https://gcc.gnu.org/onlinedocs/ 8062 8. mailto:gcc-help (a] gcc.gnu.org 8063 9. mailto:gcc (a] gcc.gnu.org 8064 10. https://gcc.gnu.org/lists.html 8065 11. http://www.fsf.org/ 8066 12. https://gcc.gnu.org/about.html 8067 13. http://validator.w3.org/check/referer 8068 ====================================================================== 8069 http://gcc.gnu.org/gcc-4.0/index.html 8070 GCC 4.0 Release Series 8071 8072 January 31, 2007 8073 8074 The [1]GNU project and the GCC developers are pleased to announce the 8075 release of GCC 4.0.4. 8076 8077 This release is a bug-fix release, containing fixes for regressions in 8078 GCC 4.0.3 relative to previous releases of GCC. 8079 8080 This release series is no longer maintained. 8081 8082 Release History 8083 8084 GCC 4.0.4 8085 January 31, 2007 ([2]changes) 8086 8087 GCC 4.0.3 8088 March 10, 2006 ([3]changes) 8089 8090 GCC 4.0.2 8091 September 28, 2005 ([4]changes) 8092 8093 GCC 4.0.1 8094 July 7, 2005 ([5]changes) 8095 8096 GCC 4.0.0 8097 April 20, 2005 ([6]changes) 8098 8099 References and Acknowledgements 8100 8101 GCC used to stand for the GNU C Compiler, but since the compiler 8102 supports several other languages aside from C, it now stands for the 8103 GNU Compiler Collection. 8104 8105 A list of [7]successful builds is updated as new information becomes 8106 available. 8107 8108 The GCC developers would like to thank the numerous people that have 8109 contributed new features, improvements, bug fixes, and other changes as 8110 well as test results to GCC. This [8]amazing group of volunteers is 8111 what makes GCC successful. 8112 8113 For additional information about GCC please refer to the [9]GCC project 8114 web site or contact the [10]GCC development mailing list. 8115 8116 To obtain GCC please use [11]our mirror sites, or [12]our SVN server. 8117 8118 8119 For questions related to the use of GCC, please consult these web 8120 pages and the [13]GCC manuals. If that fails, the 8121 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 8122 web pages and the development of GCC are welcome on our developer 8123 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 8124 archives. 8125 8126 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 8127 distribution of this entire article is permitted in any medium, 8128 provided this notice is preserved. 8129 8130 These pages are [18]maintained by the GCC team. Last modified 8131 2016-09-30[19]. 8132 8133 References 8134 8135 1. http://www.gnu.org/ 8136 2. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4 8137 3. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.3 8138 4. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.2 8139 5. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.1 8140 6. http://gcc.gnu.org/gcc-4.0/changes.html 8141 7. http://gcc.gnu.org/gcc-4.0/buildstat.html 8142 8. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 8143 9. http://gcc.gnu.org/index.html 8144 10. mailto:gcc (a] gcc.gnu.org 8145 11. http://gcc.gnu.org/mirrors.html 8146 12. http://gcc.gnu.org/svn.html 8147 13. https://gcc.gnu.org/onlinedocs/ 8148 14. mailto:gcc-help (a] gcc.gnu.org 8149 15. mailto:gcc (a] gcc.gnu.org 8150 16. https://gcc.gnu.org/lists.html 8151 17. http://www.fsf.org/ 8152 18. https://gcc.gnu.org/about.html 8153 19. http://validator.w3.org/check/referer 8154 ====================================================================== 8155 http://gcc.gnu.org/gcc-4.0/changes.html 8156 GCC 4.0 Release Series 8157 Changes, New Features, and Fixes 8158 8159 The latest release in the 4.0 release series is [1]GCC 4.0.4. 8160 8161 Caveats 8162 8163 * GCC now generates location lists by default when compiling with 8164 debug info and optimization. 8165 + GDB 6.0 and older crashes when it sees location lists. GDB 6.1 8166 or later is needed to debug binaries containing location 8167 lists. 8168 + When you are trying to view a value of a variable in a part of 8169 a function where it has no location (for example when the 8170 variable is no longer used and thus its location was used for 8171 something else) GDB will say that it is not available. 8172 You can disable generating location lists by -fno-var-tracking. 8173 * GCC no longer accepts the -fwritable-strings option. Use named 8174 character arrays when you need a writable string. 8175 * The options -freduce-all-givs and -fmove-all-movables have been 8176 discontinued. They were used to circumvent a shortcoming in the 8177 heuristics of the old loop optimization code with respect to common 8178 Fortran constructs. The new (tree) loop optimizer works differently 8179 and doesn't need those work-arounds. 8180 * The graph-coloring register allocator, formerly enabled by the 8181 option -fnew-ra, has been discontinued. 8182 * -I- has been deprecated. -iquote is meant to replace the need for 8183 this option. 8184 * The MIPS -membedded-pic and -mrnames options have been removed. 8185 * All MIPS targets now require the GNU assembler. In particular, IRIX 8186 configurations can no longer use the MIPSpro assemblers, although 8187 they do still support the MIPSpro linkers. 8188 * The SPARC option -mflat has been removed. 8189 * English-language diagnostic messages will now use Unicode quotation 8190 marks in UTF-8 locales. (Non-English messages already used the 8191 quotes appropriate for the language in previous releases.) If your 8192 terminal does not support UTF-8 but you are using a UTF-8 locale 8193 (such locales are the default on many GNU/Linux systems) then you 8194 should set LC_CTYPE=C in the environment to disable that locale. 8195 Programs that parse diagnostics and expect plain ASCII 8196 English-language messages should set LC_ALL=C. See [2]Markus Kuhn's 8197 explanation of Unicode quotation marks for more information. 8198 * The specs file is no longer installed on most platforms. Most users 8199 will be totally unaffected. However, if you are accustomed to 8200 editing the specs file yourself, you will now have to use the 8201 -dumpspecs option to generate the specs file, and then edit the 8202 resulting file. 8203 8204 General Optimizer Improvements 8205 8206 * The [3]tree ssa branch has been merged. This merge has brought in a 8207 completely new optimization framework based on a higher level 8208 intermediate representation than the existing RTL representation. 8209 Numerous new code transformations based on the new framework are 8210 available in GCC 4.0, including: 8211 + Scalar replacement of aggregates 8212 + Constant propagation 8213 + Value range propagation 8214 + Partial redundancy elimination 8215 + Load and store motion 8216 + Strength reduction 8217 + Dead store elimination 8218 + Dead and unreachable code elimination 8219 + [4]Autovectorization 8220 + Loop interchange 8221 + Tail recursion by accumulation 8222 Many of these passes outperform their counterparts from previous 8223 GCC releases. 8224 * [5]Swing Modulo Scheduling (SMS). An RTL level instruction 8225 scheduling optimization intended for loops that perform heavy 8226 computations. 8227 8228 New Languages and Language specific improvements 8229 8230 C family 8231 8232 * The sentinel attribute has been added to GCC. This function 8233 attribute allows GCC to warn when variadic functions such as execl 8234 are not NULL terminated. See the GCC manual for a complete 8235 description of its behavior. 8236 * Given __attribute__((alias("target"))) it is now an error if target 8237 is not a symbol, defined in the same translation unit. This also 8238 applies to aliases created by #pragma weak alias=target. This is 8239 because it's meaningless to define an alias to an undefined symbol. 8240 On Solaris, the native assembler would have caught this error, but 8241 GNU as does not. 8242 8243 C and Objective-C 8244 8245 * The -Wstrict-aliasing=2 option has been added. This warning catches 8246 all unsafe cases, but it may also give a warning for some cases 8247 that are safe. 8248 * The cast-as-lvalue, conditional-expression-as-lvalue and 8249 compound-expression-as-lvalue extensions, which were deprecated in 8250 3.3.4 and 3.4, have been removed. 8251 * The -fwritable-strings option, which was deprecated in 3.4, has 8252 been removed. 8253 * #pragma pack() semantics have been brought closer to those used by 8254 other compilers. This also applies to C++. 8255 * Taking the address of a variable with register storage is invalid 8256 in C. GCC now issues an error instead of a warning. 8257 * Arrays of incomplete element type are invalid in C. GCC now issues 8258 an error for such arrays. Declarations such as extern struct s x[]; 8259 (where struct s has not been defined) can be moved after the 8260 definition of struct s. Function parameters declared as arrays of 8261 incomplete type can instead be declared as pointers. 8262 8263 C++ 8264 8265 * When compiling without optimizations (-O0), the C++ frontend is 8266 much faster than in any previous versions of GCC. Independent 8267 testers have measured speed-ups up to 25% in real-world production 8268 code, compared to the 3.4 family (which was already the fastest 8269 version to date). Upgrading from older versions might show even 8270 bigger improvements. 8271 * ELF visibility attributes can now be applied to a class type, so 8272 that it affects every member function of a class at once, without 8273 having to specify each individually: 8274 class __attribute__ ((visibility("hidden"))) Foo 8275 { 8276 int foo1(); 8277 void foo2(); 8278 }; 8279 The syntax is deliberately similar to the __declspec() system used 8280 by Microsoft Windows based compilers, allowing cross-platform 8281 projects to easily reuse their existing macro system for denoting 8282 exports and imports. By explicitly marking internal classes never 8283 used outside a binary as hidden, one can completely avoid PLT 8284 indirection overheads during their usage by the compiler. You can 8285 find out more about the advantages of this at 8286 [6]https://www.akkadia.org/drepper/dsohowto.pdf 8287 * The -fvisibility-inlines-hidden option has been added which marks 8288 all inlineable functions as having hidden ELF visibility, thus 8289 removing their symbol and typeinfo from the exported symbol table 8290 of the output ELF binary. Using this option can reduce the exported 8291 symbol count of template-heavy code by up to 40% with no code 8292 change at all, thus notably improving link and load times for the 8293 binary as well as a reduction in size of up to 10%. Also, check the 8294 new [7]-fvisibility option. 8295 * The compiler now uses the library interface specified by the [8]C++ 8296 ABI for thread-safe initialization of function-scope static 8297 variables. Most users should leave this alone, but embedded 8298 programmers may want to disable this by specifying 8299 -fno-threadsafe-statics for a small savings in code size. 8300 * Taking the address of an explicit register variable is no longer 8301 supported. Note that C++ allows taking the address of variables 8302 with register storage so this will continue to compile with a 8303 warning. For example, assuming that r0 is a machine register: 8304 register int foo asm ("r0"); 8305 register int bar; 8306 &foo; // error, no longer accepted 8307 &bar; // OK, with a warning 8308 * G++ has an undocumented extension to virtual function covariancy 8309 rules that allowed the overrider to return a type that was 8310 implicitly convertable to the overridden function's return type. 8311 For instance a function returning void * could be overridden by a 8312 function returning T *. This is now deprecated and will be removed 8313 in a future release. 8314 * The G++ minimum and maximum operators (<? and >?) and their 8315 compound forms (<?=) and >?=) have been deprecated and will be 8316 removed in a future version. Code using these operators should be 8317 modified to use std::min and std::max instead. 8318 * Declaration of nested classes of class templates as friends are 8319 supported: 8320 template <typename T> struct A { 8321 class B {}; 8322 }; 8323 class C { 8324 template <typename T> friend class A<T>::B; 8325 }; 8326 This complements the feature member functions of class templates as 8327 friends introduced in GCC 3.4.0. 8328 * When declaring a friend class using an unqualified name, classes 8329 outside the innermost non-class scope are not searched: 8330 class A; 8331 namespace N { 8332 class B { 8333 friend class A; // Refer to N::A which has not been declared yet 8334 // because name outside namespace N are not searched 8335 friend class ::A; // Refer to ::A 8336 }; 8337 } 8338 Hiding the friend name until declaration is still not implemented. 8339 * Friends of classes defined outside their namespace are correctly 8340 handled: 8341 namespace N { 8342 class A; 8343 } 8344 class N::A { 8345 friend class B; // Refer to N::B in GCC 4.0.0 8346 // but ::B in earlier versions of GCC 8347 }; 8348 8349 Runtime Library (libstdc++) 8350 8351 * Optimization work: 8352 + Added efficient specializations of istream functions for char 8353 and wchar_t. 8354 + Further performance tuning of strings, in particular wrt 8355 single-char append and getline. 8356 + iter_swap - and therefore most of the mutating algorithms - 8357 now makes an unqualified call to swap when the value_type of 8358 the two iterators is the same. 8359 * A large subset of the features in Technical Report 1 (TR1 for 8360 short) is experimentally delivered (i.e., no guarantees about the 8361 implementation are provided. In particular it is not promised that 8362 the library will remain link-compatible when code using TR1 is 8363 used): 8364 + General utilities such as reference_wrapper and shared_ptr. 8365 + Function objects, i.e., result_of, mem_fn, bind, function. 8366 + Support for metaprogramming. 8367 + New containers such as tuple, array, unordered_set, 8368 unordered_map, unordered_multiset, unordered_multimap. 8369 * As usual, many bugs have been fixed and LWG resolutions implemented 8370 for the first time (e.g., DR 409). 8371 8372 Java 8373 8374 * In order to prevent naming conflicts with other implementations of 8375 these tools, some GCJ binaries have been renamed: 8376 + rmic is now grmic, 8377 + rmiregistry is now grmiregistry, and 8378 + jar is now fastjar. 8379 In particular, these names were problematic for the jpackage.org 8380 packaging conventions which install symlinks in /usr/bin that point 8381 to the preferred versions of these tools. 8382 * The -findirect-dispatch argument to the compiler now works and 8383 generates code following a new "binary compatibility" ABI. Code 8384 compiled this way follows the binary compatibility rules of the 8385 Java Language Specification. 8386 * libgcj now has support for using GCJ as a JIT, using the 8387 gnu.gcj.jit family of system properties. 8388 * libgcj can now find a shared library corresponding to the bytecode 8389 representation of a class. See the documentation for the new 8390 gcj-dbtool program, and the new gnu.gcj.precompiled.db.path system 8391 property. 8392 * There have been many improvements to the class library. Here are 8393 some highlights: 8394 + Much more of AWT and Swing exist. 8395 + Many new packages and classes were added, including 8396 java.util.regex, java.net.URI, javax.crypto, 8397 javax.crypto.interfaces, javax.crypto.spec, javax.net, 8398 javax.net.ssl, javax.security.auth, 8399 javax.security.auth.callback, javax.security.auth.login, 8400 javax.security.auth.x500, javax.security.sasl, org.ietf.jgss, 8401 javax.imageio, javax.imageio.event, javax.imageio.spi, 8402 javax.print, javax.print.attribute, 8403 javax.print.attribute.standard, javax.print.event, and 8404 javax.xml 8405 + Updated SAX and DOM, and imported GNU JAXP 8406 8407 Fortran 8408 8409 * A new [9]Fortran front end has replaced the aging GNU Fortran 77 8410 front end. The new front end supports Fortran 90 and Fortran 95. It 8411 may not yet be as stable as the old Fortran front end. 8412 8413 Ada 8414 8415 * Ada (with tasking and Zero Cost Exceptions) is now available on 8416 many more targets, including but not limited to: alpha-linux, 8417 hppa-hpux, hppa-linux, powerpc-darwin, powerpc-linux, s390-linux, 8418 s390x-linux, sparc-linux. 8419 * Some of the new Ada 2005 features are now implemented like 8420 Wide_Wide_Character and Ada.Containers. 8421 * Many bugs have been fixed, tools and documentation improved. 8422 * To compile Ada from the sources, install an older working Ada 8423 compiler and then use --enable-languages=ada at configuration time, 8424 since the Ada frontend is not currently activated by default. See 8425 the [10]Installing GCC for details. 8426 8427 New Targets and Target Specific Improvements 8428 8429 H8/300 8430 8431 * The frame layout has changed. In the new layout, the prologue of a 8432 function first saves registers and then allocate space for locals, 8433 resulting in an 1% improvement on code size. 8434 8435 IA-32/x86-64 (AMD64) 8436 8437 * The acos, asin, drem, exp10, exp2, expm1, fmod, ilogb, log10, 8438 log1p, log2, logb and tan mathematical builtins (and their float 8439 and long double variants) are now implemented as inline x87 8440 intrinsics when using -ffast-math. 8441 * The ceil, floor, nearbyint, rint and trunc mathematical builtins 8442 (and their float and long double variants) are now implemented as 8443 inline x87 intrinsics when using -ffast-math. 8444 * The x87's fsincos instruction is now used automatically with 8445 -ffast-math when calculating both the sin and cos of the same 8446 argument. 8447 * Instruction selection for multiplication and division by constants 8448 has been improved. 8449 8450 IA-64 8451 8452 * Floating point division, integer division and sqrt are now inlined, 8453 resulting in significant performance improvements on some codes. 8454 8455 MIPS 8456 8457 * Division by zero checks now use conditional traps if the target 8458 processor supports them. This decreases code size by one word per 8459 division operation. The old behavior (branch and break) can be 8460 obtained either at configure time by passing --with-divide=breaks 8461 to configure or at runtime by passing -mdivide-breaks to GCC. 8462 * Support for MIPS64 paired-single instructions has been added. It is 8463 enabled by -mpaired-single and can be accessed using both the 8464 target-independent vector extensions and new MIPS-specific built-in 8465 functions. 8466 * Support for the MIPS-3D ASE has been added. It is enabled by 8467 -mips3d and provides new MIPS-3D-specific built-in functions. 8468 * The -mexplicit-relocs option now supports static n64 code (as is 8469 used, for example, in 64-bit linux kernels). -mexplicit-relocs 8470 should now be feature-complete and is enabled by default when GCC 8471 is configured to use a compatible assembler. 8472 * Support for the NEC VR4130 series has been added. This support 8473 includes the use of VR-specific instructions and a new VR4130 8474 scheduler. Full VR4130 support can be selected with -march=vr4130 8475 while code for any ISA can be tuned for the VR4130 using 8476 -mtune=vr4130. There is also a new -mvr4130-align option that 8477 produces better schedules at the cost of increased code size. 8478 * Support for the Broadcom SB-1 has been extended. There is now an 8479 SB-1 scheduler as well as support for the SB-1-specific 8480 paired-single instructions. Full SB-1 support can be selected with 8481 -march=sb1 while code for any ISA can be optimized for the SB-1 8482 using -mtune=sb1. 8483 * The compiler can now work around errata in R4000, R4400, VR4120 and 8484 VR4130 processors. These workarounds are enabled by -mfix-r4000, 8485 -mfix-r4400, -mfix-vr4120 and -mfix-vr4130 respectively. The VR4120 8486 and VR4130 workarounds need binutils 2.16 or above. 8487 * IRIX shared libraries are now installed into the standard library 8488 directories: o32 libraries go into lib/, n32 libraries go into 8489 lib32/ and n64 libraries go into lib64/. 8490 * The compiler supports a new -msym32 option. It can be used to 8491 optimize n64 code in which all symbols are known to have 32-bit 8492 values. 8493 8494 S/390 and zSeries 8495 8496 * New command-line options help to generate code intended to run in 8497 an environment where stack space is restricted, e.g. Linux kernel 8498 code: 8499 + -mwarn-framesize and -mwarn-dynamicstack trigger compile-time 8500 warnings for single functions that require large or dynamic 8501 stack frames. 8502 + -mstack-size and -mstack-guard generate code that checks for 8503 stack overflow at run time. 8504 + -mpacked-stack generates code that reduces the stack frame 8505 size of many functions by reusing unneeded parts of the stack 8506 bias area. 8507 * The -msoft-float option now ensures that generated code never 8508 accesses floating point registers. 8509 * The s390x-ibm-tpf target now fully supports C++, including 8510 exceptions and threads. 8511 * Various changes to improve performance of the generated code have 8512 been implemented, including: 8513 + GCC now uses sibling calls where possible. 8514 + Condition code handling has been optimized, allowing GCC to 8515 omit redundant comparisons in certain cases. 8516 + The cost function guiding many optimizations has been refined 8517 to more accurately represent the z900 and z990 processors. 8518 + The ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL WITH BORROW 8519 instructions are now used to avoid conditional branches in 8520 certain cases. 8521 + The back end now uses the LEGITIMIZE_RELOAD_ADDRESS feature to 8522 optimize address arithmetic required to access large stack 8523 frames. 8524 + GCC now makes more efficient use of memory-to-memory type 8525 instructions (MVC, CLC, ...). 8526 + More precise tracking of special register use allows better 8527 instruction scheduling, in particular of the function prologue 8528 and epilogue sequences. 8529 + The Java front end now generates inline code to implement 8530 integer division, instead of calling library routines. 8531 8532 SPARC 8533 8534 * The options -mv8, -msparclite, -mcypress, -msupersparc, -mf930 and 8535 -mf934 have been removed. They have been replaced with -mcpu=xxx. 8536 * The internal model used to estimate the relative cost of each 8537 instruction has been updated. It is expected to give better results 8538 on recent UltraSPARC processors. 8539 * Code generation for function prologues and epilogues has been 8540 improved, resulting in better scheduling and allowing multiple exit 8541 points in functions. 8542 * Support for Sun's Visual Instruction Set (VIS) has been enhanced. 8543 It is enabled by -mvis and provides new built-in functions for VIS 8544 instructions on UltraSPARC processors. 8545 * The option -mapp-regs has been turned on by default on Solaris too. 8546 8547 NetWare 8548 8549 * Novell NetWare (on ix86, no other hardware platform was ever really 8550 supported by this OS) has been re-enabled and the ABI supported by 8551 GCC has been brought into sync with that of MetroWerks CodeWarrior 8552 (the ABI previously supported was that of some Unix systems, which 8553 NetWare never tried to support). 8554 8555 Obsolete Systems 8556 8557 Support for a number of older systems has been declared obsolete in GCC 8558 4.0. Unless there is activity to revive them, the next release of GCC 8559 will have their sources permanently removed. 8560 8561 All GCC ports for the following processor architectures have been 8562 declared obsolete: 8563 * Intel i860 8564 * Ubicom IP2022 8565 * National Semiconductor NS32K (ns32k) 8566 * Texas Instruments TMS320C[34]x 8567 8568 Also, those for some individual systems have been obsoleted: 8569 * SPARC family 8570 + SPARClite-based systems (sparclite-*-coff, sparclite-*-elf, 8571 sparc86x-*-elf) 8572 + OpenBSD 32-bit (sparc-*-openbsd*) 8573 8574 Documentation improvements 8575 8576 Other significant improvements 8577 8578 * Location lists are now generated by default when compiling with 8579 debug info and optimization. Location lists provide more accurate 8580 debug info about locations of variables and they allow debugging 8581 code compiled with -fomit-frame-pointer. 8582 * The -fvisibility option has been added which allows the default ELF 8583 visibility of all symbols to be set per compilation and the new 8584 #pragma GCC visibility preprocessor command allows the setting of 8585 default ELF visibility for a region of code. Using 8586 -fvisibility=hidden especially in combination with the new 8587 -fvisibility-inlines-hidden can yield substantial improvements in 8588 output binary quality including avoiding PLT indirection overheads, 8589 reduction of the exported symbol count by up to 60% (with resultant 8590 improvements to link and load times), better scope for the 8591 optimizer to improve code and up to a 20% reduction in binary size. 8592 Using these options correctly yields a binary with a similar symbol 8593 count to a Windows DLL. 8594 Perhaps more importantly, this new feature finally allows (with 8595 careful planning) complete avoidance of symbol clashes when 8596 manually loading shared objects with RTLD_GLOBAL, thus finally 8597 solving problems many projects such as python were forced to use 8598 RTLD_LOCAL for (with its resulting issues for C++ correctness). You 8599 can find more information about using these options at 8600 [11]https://gcc.gnu.org/wiki/Visibility. 8601 __________________________________________________________________ 8602 8603 GCC 4.0.1 8604 8605 This is the [12]list of problem reports (PRs) from GCC's bug tracking 8606 system that are known to be fixed in the 4.0.1 release. This list might 8607 not be complete (that is, it is possible that some PRs that have been 8608 fixed are not listed here). 8609 8610 GCC 4.0.2 8611 8612 This is the [13]list of problem reports (PRs) from GCC's bug tracking 8613 system that are known to be fixed in the 4.0.2 release. This list might 8614 not be complete (that is, it is possible that some PRs that have been 8615 fixed are not listed here). 8616 8617 Unfortunately, due to a release engineering failure, this release has a 8618 regression on Solaris that will affect some C++ programs. We suggest 8619 that Solaris users apply a [14]patch that corrects the problem. Users 8620 who do not wish to apply the patch should explicitly link C++ programs 8621 with the -pthreads option, even if they do not use threads. This 8622 problem has been corrected in the current 4.0 branch sources and will 8623 not be present in GCC 4.0.3. 8624 8625 GCC 4.0.3 8626 8627 Starting with this release, the function getcontext is recognized by 8628 the compiler as having the same semantics as the setjmp function. In 8629 particular, the compiler will ensure that all registers are dead before 8630 calling such a function and will emit a warning about the variables 8631 that may be clobbered after the second return from the function. 8632 8633 GCC 4.0.4 8634 8635 This is the [15]list of problem reports (PRs) from GCC's bug tracking 8636 system that are known to be fixed in the 4.0.4 release. This list might 8637 not be complete (that is, it is possible that some PRs that have been 8638 fixed are not listed here). 8639 8640 The 4.0.4 release is provided for those that require a high degree of 8641 binary compatibility with previous 4.0.x releases. For most users, the 8642 GCC team recommends that version 4.1.1 or later be used instead." 8643 8644 8645 For questions related to the use of GCC, please consult these web 8646 pages and the [16]GCC manuals. If that fails, the 8647 [17]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 8648 web pages and the development of GCC are welcome on our developer 8649 list at [18]gcc (a] gcc.gnu.org. All of [19]our lists have public 8650 archives. 8651 8652 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and 8653 distribution of this entire article is permitted in any medium, 8654 provided this notice is preserved. 8655 8656 These pages are [21]maintained by the GCC team. Last modified 8657 2017-03-29[22]. 8658 8659 References 8660 8661 1. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4 8662 2. http://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html 8663 3. http://gcc.gnu.org/projects/tree-ssa/ 8664 4. http://gcc.gnu.org/projects/tree-ssa/vectorization.html 8665 5. http://gcc.gnu.org/news/sms.html 8666 6. https://www.akkadia.org/drepper/dsohowto.pdf 8667 7. http://gcc.gnu.org/gcc-4.0/changes.html#visibility 8668 8. https://itanium-cxx-abi.github.io/cxx-abi/ 8669 9. http://gcc.gnu.org/fortran/ 8670 10. https://gcc.gnu.org/install/ 8671 11. https://gcc.gnu.org/wiki/Visibility 8672 12. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.1 8673 13. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.2 8674 14. https://gcc.gnu.org/ml/gcc-cvs/2005-09/msg00984.html 8675 15. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.4 8676 16. https://gcc.gnu.org/onlinedocs/ 8677 17. mailto:gcc-help (a] gcc.gnu.org 8678 18. mailto:gcc (a] gcc.gnu.org 8679 19. https://gcc.gnu.org/lists.html 8680 20. http://www.fsf.org/ 8681 21. https://gcc.gnu.org/about.html 8682 22. http://validator.w3.org/check/referer 8683 ====================================================================== 8684 http://gcc.gnu.org/gcc-3.4/index.html 8685 GCC 3.4 Release Series 8686 8687 May 26, 2006 8688 8689 The [1]GNU project and the GCC developers are pleased to announce the 8690 release of GCC 3.4.6. 8691 8692 This release is a bug-fix release, containing fixes for regressions in 8693 GCC 3.4.4 relative to previous releases of GCC. This is the last of the 8694 3.4.x series. 8695 8696 The GCC 3.4 release series includes numerous [2]new features, 8697 improvements, bug fixes, and other changes, thanks to an [3]amazing 8698 group of volunteers. 8699 8700 This release series is no longer maintained. 8701 8702 Release History 8703 8704 GCC 3.4.6 8705 March 6, 2006 ([4]changes) 8706 8707 GCC 3.4.5 8708 November 30, 2005 ([5]changes) 8709 8710 GCC 3.4.4 8711 May 18, 2005 ([6]changes) 8712 8713 GCC 3.4.3 8714 November 4, 2004 ([7]changes) 8715 8716 GCC 3.4.2 8717 September 6, 2004 ([8]changes) 8718 8719 GCC 3.4.1 8720 July 1, 2004 ([9]changes) 8721 8722 GCC 3.4.0 8723 April 18, 2004 ([10]changes) 8724 8725 References and Acknowledgements 8726 8727 GCC used to stand for the GNU C Compiler, but since the compiler 8728 supports several other languages aside from C, it now stands for the 8729 GNU Compiler Collection. 8730 8731 A list of [11]successful builds is updated as new information becomes 8732 available. 8733 8734 The GCC developers would like to thank the numerous people that have 8735 contributed new features, improvements, bug fixes, and other changes as 8736 well as test results to GCC. This [12]amazing group of volunteers is 8737 what makes GCC successful. 8738 8739 For additional information about GCC please refer to the [13]GCC 8740 project web site or contact the [14]GCC development mailing list. 8741 8742 To obtain GCC please use [15]our mirror sites, or [16]our SVN server. 8743 8744 8745 For questions related to the use of GCC, please consult these web 8746 pages and the [17]GCC manuals. If that fails, the 8747 [18]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 8748 web pages and the development of GCC are welcome on our developer 8749 list at [19]gcc (a] gcc.gnu.org. All of [20]our lists have public 8750 archives. 8751 8752 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and 8753 distribution of this entire article is permitted in any medium, 8754 provided this notice is preserved. 8755 8756 These pages are [22]maintained by the GCC team. Last modified 8757 2016-09-30[23]. 8758 8759 References 8760 8761 1. http://www.gnu.org/ 8762 2. http://gcc.gnu.org/gcc-3.4/changes.html 8763 3. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 8764 4. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6 8765 5. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.5 8766 6. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.4 8767 7. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.3 8768 8. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.2 8769 9. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.1 8770 10. http://gcc.gnu.org/gcc-3.4/changes.html 8771 11. http://gcc.gnu.org/gcc-3.4/buildstat.html 8772 12. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 8773 13. http://gcc.gnu.org/index.html 8774 14. mailto:gcc (a] gcc.gnu.org 8775 15. http://gcc.gnu.org/mirrors.html 8776 16. http://gcc.gnu.org/svn.html 8777 17. https://gcc.gnu.org/onlinedocs/ 8778 18. mailto:gcc-help (a] gcc.gnu.org 8779 19. mailto:gcc (a] gcc.gnu.org 8780 20. https://gcc.gnu.org/lists.html 8781 21. http://www.fsf.org/ 8782 22. https://gcc.gnu.org/about.html 8783 23. http://validator.w3.org/check/referer 8784 ====================================================================== 8785 http://gcc.gnu.org/gcc-3.4/changes.html 8786 GCC 3.4 Release Series 8787 Changes, New Features, and Fixes 8788 8789 The final release in the 3.4 release series is [1]GCC 3.4.6. The series 8790 is now closed. 8791 8792 GCC 3.4 has [2]many improvements in the C++ frontend. Before reporting 8793 a bug, please make sure it's really GCC, and not your code, that is 8794 broken. 8795 8796 Caveats 8797 8798 * GNU Make is now required to build GCC. 8799 * With -nostdinc the preprocessor used to ignore both standard 8800 include paths and include paths contained in environment variables. 8801 It was neither documented nor intended that environment variable 8802 paths be ignored, so this has been corrected. 8803 * GCC no longer accepts the options -fvolatile, -fvolatile-global and 8804 -fvolatile-static. It is unlikely that they worked correctly in any 8805 3.x release. 8806 * GCC no longer ships <varargs.h>. Use <stdarg.h> instead. 8807 * Support for all the systems [3]obsoleted in GCC 3.3 has been 8808 removed from GCC 3.4. See below for a [4]list of systems which are 8809 obsoleted in this release. 8810 * GCC now requires an ISO C90 (ANSI C89) C compiler to build. K&R C 8811 compilers will not work. 8812 * The implementation of the [5]MIPS ABIs has changed. As a result, 8813 the code generated for certain MIPS targets will not be binary 8814 compatible with earlier releases. 8815 * In previous releases, the MIPS port had a fake "hilo" register with 8816 the user-visible name accum. This register has been removed. 8817 * The implementation of the [6]SPARC ABIs has changed. As a result, 8818 the code generated will not be binary compatible with earlier 8819 releases in certain cases. 8820 * The configure option --enable-threads=pthreads has been removed; 8821 use --enable-threads=posix instead, which should have the same 8822 effect. 8823 * Code size estimates used by inlining heuristics for C, Objective-C, 8824 C++ and Java have been redesigned significantly. As a result the 8825 parameters of -finline-insns, --param max-inline-insns-single and 8826 --param max-inline-insns-auto need to be reconsidered. 8827 * --param max-inline-slope and --param min-inline-insns have been 8828 removed; they are not needed for the new bottom-up inlining 8829 heuristics. 8830 * The new unit-at-a-time compilation scheme has several compatibility 8831 issues: 8832 + The order in which functions, variables, and top-level asm 8833 statements are emitted may have changed. Code relying on some 8834 particular ordering needs to be updated. The majority of such 8835 top-level asm statements can be replaced by section 8836 attributes. 8837 + Unreferenced static variables and functions are removed. This 8838 may result in undefined references when an asm statement 8839 refers to the variable/function directly. In that case either 8840 the variable/function shall be listed in asm statement operand 8841 or in the case of top-level asm statements the attribute used 8842 shall be used to force function/variable to be always output 8843 and considered as a possibly used by unknown code. 8844 For variables the attribute is accepted only by GCC 3.4 and 8845 newer, while for earlier versions it is sufficient to use 8846 unused to silence warnings about the variables not being 8847 referenced. To keep code portable across different GCC 8848 versions, you can use appropriate preprocessor conditionals. 8849 + Static functions now can use non-standard passing conventions 8850 that may break asm statements calling functions directly. 8851 Again the attribute used shall be used to prevent this 8852 behavior. 8853 As a temporary workaround, -fno-unit-at-a-time can be used, but 8854 this scheme may not be supported by future releases of GCC. 8855 * GCC 3.4 automatically places zero-initialized variables in the .bss 8856 section on some operating systems. Versions of GNU Emacs up to (and 8857 including) 21.3 will not work correctly when using this 8858 optimization; you can use -fno-zero-initialized-in-bss to disable 8859 it. 8860 * If GCC 3.4 is configured with --enable-threads=posix (the default 8861 on most targets that support pthreads) then _REENTRANT will be 8862 defined unconditionally by some libstdc++ headers. C++ code which 8863 relies on that macro to detect whether multi-threaded code is being 8864 compiled might change in meaning, possibly resulting in linker 8865 errors for single-threaded programs. Affected users of [7]Boost 8866 should compile single-threaded code with -DBOOST_DISABLE_THREADS. 8867 See Bugzilla for [8]more information. 8868 8869 General Optimizer Improvements 8870 8871 * Usability of the profile feedback and coverage testing has been 8872 improved. 8873 + Performance of profiled programs has been improved by faster 8874 profile merging code. 8875 + Better use of the profile feedback for optimization (loop 8876 unrolling and loop peeling). 8877 + File locking support allowing fork() calls and parallel runs 8878 of profiled programs. 8879 + Coverage file format has been redesigned. 8880 + gcov coverage tool has been improved. 8881 + make profiledbootstrap available to build a faster compiler. 8882 Experiments made on i386 hardware showed an 11% speedup on -O0 8883 and a 7.5% speedup on -O2 compilation of a [9]large C++ 8884 testcase. 8885 + New value profiling pass enabled via -fprofile-values 8886 + New value profile transformations pass enabled via -fvpt aims 8887 to optimize some code sequences by exploiting knowledge about 8888 value ranges or other properties of the operands. At the 8889 moment a conversion of expensive divisions into cheaper 8890 operations has been implemented. 8891 + New -fprofile-generate and -fprofile-use command-line options 8892 to simplify the use of profile feedback. 8893 * A new unit-at-a-time compilation scheme for C, Objective-C, C++ and 8894 Java which is enabled via -funit-at-a-time (and implied by -O2). In 8895 this scheme a whole file is parsed first and optimized later. The 8896 following basic inter-procedural optimizations are implemented: 8897 + Removal of unreachable functions and variables 8898 + Discovery of local functions (functions with static linkage 8899 whose address is never taken) 8900 + On i386, these local functions use register parameter passing 8901 conventions. 8902 + Reordering of functions in topological order of the call graph 8903 to enable better propagation of optimizing hints (such as the 8904 stack alignments needed by functions) in the back end. 8905 + Call graph based out-of-order inlining heuristics which allows 8906 to limit overall compilation unit growth (--param 8907 inline-unit-growth). 8908 Overall, the unit-at-a-time scheme produces a 1.3% improvement for 8909 the SPECint2000 benchmark on the i386 architecture (AMD Athlon 8910 CPU). 8911 * More realistic code size estimates used by inlining for C, 8912 Objective-C, C++ and Java. The growth of large functions can now be 8913 limited via --param large-function-insns and --param 8914 large-function-growth. 8915 * A new cfg-level loop optimizer pass replaces the old loop unrolling 8916 pass and adds two other loop transformations -- loop peeling and 8917 loop unswitching -- and also uses the profile feedback to limit 8918 code growth. (The three optimizations are enabled by 8919 -funroll-loops, -fpeel-loops and -funswitch-loops flags, 8920 respectively). 8921 The old loop unroller still can be enabled by -fold-unroll-loops 8922 and may produce better code in some cases, especially when the 8923 webizer optimization pass is not run. 8924 * A new web construction pass enabled via -fweb (and implied by -O3) 8925 improves the quality of register allocation, CSE, first scheduling 8926 pass and some other optimization passes by avoiding re-use of 8927 pseudo registers with non-overlapping live ranges. The pass almost 8928 always improves code quality but does make debugging difficult and 8929 thus is not enabled by default by -O2 8930 The pass is especially effective as cleanup after code duplication 8931 passes, such as the loop unroller or the tracer. 8932 * Experimental implementations of superblock or trace scheduling in 8933 the second scheduling pass can be enabled via 8934 -fsched2-use-superblocks and -fsched2-use-traces, respectively. 8935 8936 New Languages and Language specific improvements 8937 8938 Ada 8939 8940 * The Ada front end has been updated to include numerous bug fixes 8941 and enhancements. These include: 8942 + Improved project file support 8943 + Additional set of warnings about potential wrong code 8944 + Improved error messages 8945 + Improved code generation 8946 + Improved cross reference information 8947 + Improved inlining 8948 + Better run-time check elimination 8949 + Better error recovery 8950 + More efficient implementation of unbounded strings 8951 + Added features in GNAT.Sockets, GNAT.OS_Lib, GNAT.Debug_Pools, 8952 ... 8953 + New GNAT.xxxx packages (e.g. GNAT.Strings, 8954 GNAT.Exception_Action) 8955 + New pragmas 8956 + New -gnatS switch replacing gnatpsta 8957 + Implementation of new Ada features (in particular limited 8958 with, limited aggregates) 8959 8960 C/Objective-C/C++ 8961 8962 * Precompiled headers are now supported. Precompiled headers can 8963 dramatically speed up compilation of some projects. There are some 8964 known defects in the current precompiled header implementation that 8965 will result in compiler crashes in relatively rare situations. 8966 Therefore, precompiled headers should be considered a "technology 8967 preview" in this release. Read the manual for details about how to 8968 use precompiled headers. 8969 * File handling in the preprocessor has been rewritten. GCC no longer 8970 gets confused by symlinks and hardlinks, and now has a correct 8971 implementation of #import and #pragma once. These two directives 8972 have therefore been un-deprecated. 8973 * The undocumented extension that allowed C programs to have a label 8974 at the end of a compound statement, which has been deprecated since 8975 GCC 3.0, has been removed. 8976 * The cast-as-lvalue extension has been removed for C++ and 8977 deprecated for C and Objective-C. In particular, code like this: 8978 int i; 8979 (char) i = 5; 8980 8981 or this: 8982 char *p; 8983 ((int *) p)++; 8984 8985 is no longer accepted for C++ and will not be accepted for C and 8986 Objective-C in a future version. 8987 * The conditional-expression-as-lvalue extension has been deprecated 8988 for C and Objective-C. In particular, code like this: 8989 int a, b, c; 8990 (a ? b : c) = 2; 8991 8992 will not be accepted for C and Objective-C in a future version. 8993 * The compound-expression-as-lvalue extension has been deprecated for 8994 C and Objective-C. In particular, code like this: 8995 int a, b; 8996 (a, b) = 2; 8997 8998 will not be accepted for C and Objective-C in a future version. A 8999 possible non-intrusive workaround is the following: 9000 (*(a, &b)) = 2; 9001 9002 * Several [10]built-in functions such as __builtin_popcount for 9003 counting bits, finding the highest and lowest bit in a word, and 9004 parity have been added. 9005 * The -fwritable-strings option has been deprecated and will be 9006 removed. 9007 * Many C math library functions are now recognized as built-ins and 9008 optimized. 9009 * The C, C++, and Objective-C compilers can now handle source files 9010 written in any character encoding supported by the host C library. 9011 The default input character set is taken from the current locale, 9012 and may be overridden with the -finput-charset command line option. 9013 In the future we will add support for inline encoding markers. 9014 9015 C++ 9016 9017 * G++ is now much closer to full conformance to the ISO/ANSI C++ 9018 standard. This means, among other things, that a lot of invalid 9019 constructs which used to be accepted in previous versions will now 9020 be rejected. It is very likely that existing C++ code will need to 9021 be fixed. This document lists some of the most common issues. 9022 * A hand-written recursive-descent C++ parser has replaced the 9023 YACC-derived C++ parser from previous GCC releases. The new parser 9024 contains much improved infrastructure needed for better parsing of 9025 C++ source codes, handling of extensions, and clean separation 9026 (where possible) between proper semantics analysis and parsing. The 9027 new parser fixes many bugs that were found in the old parser. 9028 * You must now use the typename and template keywords to disambiguate 9029 dependent names, as required by the C++ standard. 9030 struct K { 9031 typedef int mytype_t; 9032 }; 9033 9034 template <class T1> struct A { 9035 template <class T2> struct B { 9036 void callme(void); 9037 }; 9038 9039 template <int N> void bar(void) 9040 { 9041 // Use 'typename' to tell the parser that T1::mytype_t names 9042 // a type. This is needed because the name is dependent (in 9043 // this case, on template parameter T1). 9044 typename T1::mytype_t x; 9045 x = 0; 9046 } 9047 }; 9048 9049 template <class T> void template_func(void) 9050 { 9051 // Use 'template' to prefix member templates within 9052 // dependent types (a has type A<T>, which depends on 9053 // the template parameter T). 9054 A<T> a; 9055 a.template bar<0>(); 9056 9057 // Use 'template' to tell the parser that B is a nested 9058 // template class (dependent on template parameter T), and 9059 // 'typename' because the whole A<T>::B<int> is 9060 // the name of a type (again, dependent). 9061 typename A<T>::template B<int> b; 9062 b.callme(); 9063 } 9064 9065 void non_template_func(void) 9066 { 9067 // Outside of any template class or function, no names can be 9068 // dependent, so the use of the keyword 'typename' and 'template' 9069 // is not needed (and actually forbidden). 9070 A<K> a; 9071 a.bar<0>(); 9072 A<K>::B<float> b; 9073 b.callme(); 9074 } 9075 * In a template definition, unqualified names will no longer find 9076 members of a dependent base (as specified by [temp.dep]/3 in the 9077 C++ standard). For example, 9078 template <typename T> struct B { 9079 int m; 9080 int n; 9081 int f (); 9082 int g (); 9083 }; 9084 int n; 9085 int g (); 9086 template <typename T> struct C : B<T> { 9087 void h () 9088 { 9089 m = 0; // error 9090 f (); // error 9091 n = 0; // ::n is modified 9092 g (); // ::g is called 9093 } 9094 }; 9095 You must make the names dependent, e.g. by prefixing them with 9096 this->. Here is the corrected definition of C<T>::h, 9097 template <typename T> void C<T>::h () 9098 { 9099 this->m = 0; 9100 this->f (); 9101 this->n = 0 9102 this->g (); 9103 } 9104 As an alternative solution (unfortunately not backwards compatible 9105 with GCC 3.3), you may use using declarations instead of this->: 9106 template <typename T> struct C : B<T> { 9107 using B<T>::m; 9108 using B<T>::f; 9109 using B<T>::n; 9110 using B<T>::g; 9111 void h () 9112 { 9113 m = 0; 9114 f (); 9115 n = 0; 9116 g (); 9117 } 9118 }; 9119 * In templates, all non-dependent names are now looked up and bound 9120 at definition time (while parsing the code), instead of later when 9121 the template is instantiated. For instance: 9122 void foo(int); 9123 9124 template <int> struct A { 9125 static void bar(void){ 9126 foo('a'); 9127 } 9128 }; 9129 9130 void foo(char); 9131 9132 int main() 9133 { 9134 A<0>::bar(); // Calls foo(int), used to call foo(char). 9135 } 9136 9137 * In an explicit instantiation of a class template, you must use 9138 class or struct before the template-id: 9139 template <int N> 9140 class A {}; 9141 9142 template A<0>; // error, not accepted anymore 9143 template class A<0>; // OK 9144 * The "named return value" and "implicit typename" extensions have 9145 been removed. 9146 * Default arguments in function types have been deprecated and will 9147 be removed. 9148 * ARM-style name-injection of friend declarations has been deprecated 9149 and will be removed. For example: struct S { friend void f(); }; 9150 void g() { f(); } will not be accepted by future versions of G++; 9151 instead a declaration of "f" will need to be present outside of the 9152 scope of "S". 9153 * Covariant returns are implemented for all but varadic functions 9154 that require an adjustment. 9155 * When -pedantic is used, G++ now issues errors about spurious 9156 semicolons. For example, 9157 namespace N {}; // Invalid semicolon. 9158 void f() {}; // Invalid semicolon. 9159 * G++ no longer accepts attributes for a declarator after the 9160 initializer associated with that declarator. For example, 9161 X x(1) __attribute__((...)); 9162 is no longer accepted. Instead, use: 9163 X x __attribute__((...)) (1); 9164 * Inside the scope of a template class, the name of the class itself 9165 can be treated as either a class or a template. So GCC used to 9166 accept the class name as argument of type template, and template 9167 template parameter. However this is not C++ standard compliant. Now 9168 the name is not treated as a valid template template argument 9169 unless you qualify the name by its scope. For example, the code 9170 below no longer compiles. 9171 template <template <class> class TT> class X {}; 9172 template <class T> class Y { 9173 X<Y> x; // Invalid, Y is always a type template parameter. 9174 }; 9175 The valid code for the above example is 9176 X< ::Y> x; // Valid. 9177 (Notice the space between < and : to prevent GCC to interpret this 9178 as a digraph for [.) 9179 * Friend declarations that refer to template specializations are 9180 rejected if the template has not already been declared. For 9181 example, 9182 template <typename T> 9183 class C { 9184 friend void f<> (C&); 9185 }; 9186 is rejected. You must first declare f as a template, 9187 template <typename T> 9188 void f(T); 9189 * In case of friend declarations, every name used in the friend 9190 declaration must be accessible at the point of that declaration. 9191 Previous versions of G++ used to be less strict about this and 9192 allowed friend declarations for private class members, for example. 9193 See the ISO C++ Standard Committee's [11]defect report #209 for 9194 details. 9195 * Declaration of member functions of class templates as friends are 9196 supported. For example, 9197 template <typename T> struct A { 9198 void f(); 9199 }; 9200 class C { 9201 template <typename T> friend void A<T>::f(); 9202 }; 9203 * You must use template <> to introduce template specializations, as 9204 required by the standard. For example, 9205 template <typename T> 9206 struct S; 9207 9208 struct S<int> { }; 9209 is rejected. You must write, 9210 template <> struct S<int> {}; 9211 * G++ used to accept code like this, 9212 struct S { 9213 int h(); 9214 void f(int i = g()); 9215 int g(int i = h()); 9216 }; 9217 This behavior is not mandated by the standard. Now G++ issues an 9218 error about this code. To avoid the error, you must move the 9219 declaration of g before the declaration of f. The default arguments 9220 for g must be visible at the point where it is called. 9221 * The C++ ABI Section 3.3.3 specifications for the array construction 9222 routines __cxa_vec_new2 and __cxa_vec_new3 were changed to return 9223 NULL when the allocator argument returns NULL. These changes are 9224 incorporated into the libstdc++ runtime library. 9225 * Using a name introduced by a typedef in a friend declaration or in 9226 an explicit instantiation is now rejected, as specified by the ISO 9227 C++ standard. 9228 class A; 9229 typedef A B; 9230 class C { 9231 friend class B; // error, no typedef name here 9232 friend B; // error, friend always needs class/struct/enum 9233 friend class A; // OK 9234 }; 9235 9236 template <int> class Q {}; 9237 typedef Q<0> R; 9238 template class R; // error, no typedef name here 9239 template class Q<0>; // OK 9240 * When allocating an array with a new expression, GCC used to allow 9241 parentheses around the type name. This is actually ill-formed and 9242 it is now rejected: 9243 int* a = new (int)[10]; // error, not accepted anymore 9244 int* a = new int[10]; // OK 9245 * When binding an rvalue of class type to a reference, the copy 9246 constructor of the class must be accessible. For instance, consider 9247 the following code: 9248 class A 9249 { 9250 public: 9251 A(); 9252 9253 private: 9254 A(const A&); // private copy ctor 9255 }; 9256 9257 A makeA(void); 9258 void foo(const A&); 9259 9260 void bar(void) 9261 { 9262 foo(A()); // error, copy ctor is not accessible 9263 foo(makeA()); // error, copy ctor is not accessible 9264 9265 A a1; 9266 foo(a1); // OK, a1 is a lvalue 9267 } 9268 This might be surprising at first sight, especially since most 9269 popular compilers do not correctly implement this rule ([12]further 9270 details). 9271 * When forming a pointer to member or a pointer to member function, 9272 access checks for class visibility (public, protected, private) are 9273 now performed using the qualifying scope of the name itself. This 9274 is better explained with an example: 9275 class A 9276 { 9277 public: 9278 void pub_func(); 9279 protected: 9280 void prot_func(); 9281 private: 9282 void priv_func(); 9283 }; 9284 9285 class B : public A 9286 { 9287 public: 9288 void foo() 9289 { 9290 &A::pub_func; // OK, pub_func is accessible through A 9291 &A::prot_func; // error, cannot access prot_func through A 9292 &A::priv_func; // error, cannot access priv_func through A 9293 9294 &B::pub_func; // OK, pub_func is accessible through B 9295 &B::prot_func; // OK, can access prot_func through B (within B) 9296 &B::priv_func; // error, cannot access priv_func through B 9297 } 9298 }; 9299 9300 Runtime Library (libstdc++) 9301 9302 * Optimization work: 9303 + Streamlined streambuf, filebuf, separate synched with C 9304 Standard I/O streambuf. 9305 + All formatted I/O now uses cached locale information. 9306 + STL optimizations (memory/speed for list, red-black trees as 9307 used by sets and maps). 9308 + More use of GCC builtins. 9309 + String optimizations (avoid contention on 9310 increment/decrement-and-test of the reference count in the 9311 empty-string object, constructor from input_iterators 9312 speedup). 9313 * Static linkage size reductions. 9314 * Large File Support (files larger than 2 GB on 32-bit systems). 9315 * Wide character and variable encoding filebuf work (UTF-8, Unicode). 9316 * Generic character traits. 9317 * Also support wchar_t specializations on Mac OS 10.3.x, FreeBSD 5.x, 9318 Solaris 2.7 and above, AIX 5.x, Irix 6.5. 9319 * The allocator class is now standard-conformant, and two additional 9320 extension allocators have been added, mt_alloc and 9321 bitmap_allocator. 9322 * PCH support: -include bits/stdc++.h (2x compile speedup). 9323 * Rewrote __cxa_demangle with support for C++ style allocators. 9324 * New debug modes for STL containers and iterators. 9325 * Testsuite rewrite: five times as many tests, plus increasingly 9326 sophisticated tests, including I/O, MT, multi-locale, wide and 9327 narrow characters. 9328 * Use current versions of GNU "autotools" for build/configuration. 9329 9330 Objective-C 9331 9332 * The Objective-C front end has been updated to include the numerous 9333 bug fixes and enhancements previously available only in Apple's 9334 version of GCC. These include: 9335 + Structured exception (@try... @catch... @finally, @throw) and 9336 synchronization (@synchronized) support. These are accessible 9337 via the -fobjc-exceptions switch; as of this writing, they may 9338 only be used in conjunction with -fnext-runtime on Mac OS X 9339 10.3 and later. See [13]Options Controlling Objective-C 9340 Dialect for more information. 9341 + An overhaul of @encode logic. The C99 _Bool and C++ bool type 9342 may now be encoded as 'B'. In addition, the back-end/codegen 9343 dependencies have been removed. 9344 + An overhaul of message dispatch construction, ensuring that 9345 the various receiver types (and casts thereof) are handled 9346 properly, and that correct diagnostics are issued. 9347 + Support for "Zero-Link" (-fzero-link) and "Fix-and-Continue" 9348 (-freplace-objc-classes) debugging modes, currently available 9349 on Mac OS X 10.3 and later. See [14]Options Controlling 9350 Objective-C Dialect for more information. 9351 + Access to optimized runtime entry points (-fno-nil-receivers ) 9352 on the assumption that message receivers are never nil. This 9353 is currently available on Mac OS X 10.3 and later. See 9354 [15]Options Controlling Objective-C Dialect for more 9355 information. 9356 9357 Java 9358 9359 * Compiling a .jar file will now cause non-.class entries to be 9360 automatically compiled as resources. 9361 * libgcj has been ported to Darwin. 9362 * Jeff Sturm has adapted Jan Hubicka's call graph optimization code 9363 to gcj. 9364 * libgcj has a new gcjlib URL type; this lets URLClassLoader load 9365 code from shared libraries. 9366 * libgcj has been much more completely merged with [16]GNU Classpath. 9367 * Class loading is now much more correct; in particular the caller's 9368 class loader is now used when that is required. 9369 * [17]Eclipse 2.x will run out of the box using gij. 9370 * Parts of java.nio have been implemented. Direct and indirect 9371 buffers work, as do fundamental file and socket operations. 9372 * java.awt has been improved, though it is still not ready for 9373 general use. 9374 * The HTTP protocol handler now uses HTTP/1.1 and can handle the POST 9375 method. 9376 * The MinGW port has matured. Enhancements include socket timeout 9377 support, thread interruption, improved Runtime.exec() handling and 9378 support for accented characters in filenames. 9379 9380 Fortran 9381 9382 * Fortran improvements are listed in the [18]Fortran documentation. 9383 9384 New Targets and Target Specific Improvements 9385 9386 Alpha 9387 9388 * Several [19]built-in functions have been added such as 9389 __builtin_alpha_zap to allow utilizing the more obscure 9390 instructions of the CPU. 9391 * Parameter passing of complex arguments has changed to match the 9392 ABI. This change is incompatible with previous GCC versions, but 9393 does fix compatibility with the Tru64 compiler and several corner 9394 cases where GCC was incompatible with itself. 9395 9396 ARM 9397 9398 * Nicolas Pitre has contributed his hand-coded floating-point support 9399 code for ARM. It is both significantly smaller and faster than the 9400 existing C-based implementation, even when building applications 9401 for Thumb. The arm-elf configuration has been converted to use the 9402 new code. 9403 * Support for the Intel's iWMMXt architecture, a second generation 9404 XScale processor, has been added. Enabled at run time with the 9405 -mcpu=iwmmxt command line switch. 9406 * A new ARM target has been added: arm-wince-pe. This is similar to 9407 the arm-pe target, but it defaults to using the APCS32 ABI. 9408 * The existing ARM pipeline description has been converted to the use 9409 the [20]DFA processor pipeline model. There is not much change in 9410 code performance, but the description is now [21]easier to 9411 understand. 9412 * Support for the Cirrus EP9312 Maverick floating point co-processor 9413 added. Enabled at run time with the -mcpu=ep9312 command line 9414 switch. Note however that the multilibs to support this chip are 9415 currently disabled in gcc/config/arm/t-arm-elf, so if you want to 9416 enable their production you will have to uncomment the entries in 9417 that file. 9418 9419 H8/300 9420 9421 * Support for long long has been added. 9422 * Support for saveall attribute has been added. 9423 * Pavel Pisa contributed hand-written 32-bit-by-32-bit division code 9424 for H8/300H and H8S, which is much faster than the previous 9425 implementation. 9426 * A lot of small performance improvements. 9427 9428 IA-32/AMD64 (x86-64) 9429 9430 * Tuning for K8 (AMD Opteron/Athlon64) core is available via 9431 -march=k8 and -mcpu=k8. 9432 * Scalar SSE code generation carefully avoids reformatting penalties, 9433 hidden dependencies and minimizes the number of uops generated on 9434 both Intel and AMD CPUs. 9435 * Vector MMX and SSE operands are now passed in registers to improve 9436 performance and match the argument passing convention used by the 9437 Intel C++ Compiler. As a result it is not possible to call 9438 functions accepting vector arguments compiled by older GCC version. 9439 * Conditional jump elimination is now more aggressive on modern CPUs. 9440 * The Athlon ports has been converted to use the DFA processor 9441 pipeline description. 9442 * Optimization of indirect tail calls is now possible in a similar 9443 fashion as direct sibcall optimization. 9444 * Further small performance improvements. 9445 * -m128bit-long-double is now less buggy. 9446 * __float128 support in 64-bit compilation. 9447 * Support for data structures exceeding 2GB in 64-bit mode. 9448 * -mcpu has been renamed to -mtune. 9449 9450 IA-64 9451 9452 * Tuning code for the Itanium 2 processor has been added. The 9453 generation of code tuned for Itanium 2 (option -mtune=itanium2) is 9454 enabled by default now. To generate code tuned for Itanium 1 the 9455 option -mtune=itanium1 should be used. 9456 * [22]DFA processor pipeline descriptions for the IA-64 processors 9457 have been added. This resulted in about 3% improvement on the 9458 SPECInt2000 benchmark for Itanium 2. 9459 * Instruction bundling for the IA-64 processors has been rewritten 9460 using the DFA pipeline hazard recognizer. It resulted in about 60% 9461 compiler speedup on the SPECInt2000 C programs. 9462 9463 M32R 9464 9465 * Support for the M32R/2 processor has been added by Renesas. 9466 * Support for an M32R GNU/Linux target and PIC code generation has 9467 been added by Renesas. 9468 9469 M68000 9470 9471 * Bernardo Innocenti (Develer S.r.l.) has contributed the 9472 m68k-uclinux target, based on former work done by Paul Dale 9473 (SnapGear Inc.). Code generation for the ColdFire processors family 9474 has been enhanced and extended to support the MCF 53xx and MCF 54xx 9475 cores, integrating former work done by Peter Barada (Motorola). 9476 9477 MIPS 9478 9479 Processor-specific changes 9480 9481 * Support for the RM7000 and RM9000 processors has been added. It can 9482 be selected using the -march compiler option and should work with 9483 any MIPS I (mips-*) or MIPS III (mips64-*) configuration. 9484 * Support for revision 2 of the MIPS32 ISA has been added. It can be 9485 selected with the command-line option -march=mips32r2. 9486 * There is a new option, -mfix-sb1, to work around certain SB-1 9487 errata. 9488 9489 Configuration 9490 9491 * It is possible to customize GCC using the following configure-time 9492 options: 9493 + --with-arch, which specifies the default value of the -march 9494 option. 9495 + --with-tune, which specifies the default value of the -mtune 9496 option. 9497 + --with-abi, which specifies the default ABI. 9498 + --with-float=soft, which tells GCC to use software floating 9499 point by default. 9500 + --with-float=hard, which tells GCC to use hardware floating 9501 point by default. 9502 * A 64-bit GNU/Linux port has been added. The associated 9503 configurations are mips64-linux-gnu and mips64el-linux-gnu. 9504 * The 32-bit GNU/Linux port now supports Java. 9505 * The IRIX 6 configuration now supports the o32 ABI and will build 9506 o32 multilibs by default. This support is compatible with both 9507 binutils and the SGI tools, but note that several features, 9508 including debugging information and DWARF2 exception handling, are 9509 only available when using the GNU assembler. Use of the GNU 9510 assembler and linker (version 2.15 or above) is strongly 9511 recommended. 9512 * The IRIX 6 configuration now supports 128-bit long doubles. 9513 * There are two new RTEMS-specific configurations, mips-rtems and 9514 mipsel-rtems. 9515 * There are two new *-elf configurations, mipsisa32r2-elf and 9516 mipsisa32r2el-elf. 9517 9518 General 9519 9520 * Several [23]ABI bugs have been fixed. Unfortunately, these changes 9521 will break binary compatibility with earlier releases. 9522 * GCC can now use explicit relocation operators when generating 9523 -mabicalls code. This behavior is controlled by -mexplicit-relocs 9524 and can have several performance benefits. For example: 9525 + It allows for more optimization of GOT accesses, including 9526 better scheduling and redundancy elimination. 9527 + It allows sibling calls to be implemented as jumps. 9528 + n32 and n64 leaf functions can use a call-clobbered global 9529 pointer instead of $28. 9530 + The code to set up $gp can be removed from functions that 9531 don't need it. 9532 * A new option, -mxgot, allows the GOT to be bigger than 64k. This 9533 option is equivalent to the assembler's -xgot option and should be 9534 used instead of -Wa,-xgot. 9535 * Frame pointer elimination is now supported when generating 64-bit 9536 MIPS16 code. 9537 * Inline block moves have been optimized to take more account of 9538 alignment information. 9539 * Many internal changes have been made to the MIPS port, mostly aimed 9540 at reducing the reliance on assembler macros. 9541 9542 PowerPC 9543 9544 * GCC 3.4 releases have a number of fixes for PowerPC and PowerPC64 9545 [24]ABI incompatibilities regarding the way parameters are passed 9546 during functions calls. These changes may result in incompatibility 9547 between code compiled with GCC 3.3 and GCC 3.4. 9548 9549 PowerPC Darwin 9550 9551 * Support for shared/dylib gcc libraries has been added. It is 9552 enabled by default on powerpc-apple-darwin7.0.0 and up. 9553 * Libgcj is enabled by default. On systems older than 9554 powerpc-apple-darwin7.0.0 you need to install dlcompat. 9555 * 128-bit IBM extended precision format support added for long 9556 double. 9557 9558 PowerPC64 GNU/Linux 9559 9560 * By default, PowerPC64 GNU/Linux now uses natural alignment of 9561 structure elements. The old four byte alignment for double, with 9562 special rules for a struct starting with a double, can be chosen 9563 with -malign-power. This change may result in incompatibility 9564 between code compiled with GCC 3.3 and GCC 3.4. 9565 * -mabi=altivec is now the default rather than -mabi=no-altivec. 9566 * 128-bit IBM extended precision format support added for long 9567 double. 9568 9569 S/390 and zSeries 9570 9571 * New command-line options allow to specify the intended execution 9572 environment for generated code: 9573 + -mesa/-mzarch allows to specify whether to generate code 9574 running in ESA/390 mode or in z/Architecture mode (this is 9575 applicable to 31-bit code only). 9576 + -march allows to specify a minimum processor architecture 9577 level (g5, g6, z900, or z990). 9578 + -mtune allows to specify which processor to tune for. 9579 * It is possible to customize GCC using the following configure-time 9580 options: 9581 + --with-mode, which specifies whether to default to assuming 9582 ESA/390 or z/Architecture mode. 9583 + --with-arch, which specifies the default value of the -march 9584 option. 9585 + --with-tune, which specifies the default value of the -mtune 9586 option. 9587 * Support for the z990 processor has been added, and can be selected 9588 using -march=z990 or -mtune=z990. This includes instruction 9589 scheduling tuned for the superscalar instruction pipeline of the 9590 z990 processor as well as support for all new instructions provided 9591 by the long-displacement facility. 9592 * Support to generate 31-bit code optimized for zSeries processors 9593 (running in ESA/390 or in z/Architecture mode) has been added. This 9594 can be selected using -march=z900 and -mzarch respectively. 9595 * Instruction scheduling for the z900 and z990 processors now uses 9596 the DFA pipeline hazard recognizer. 9597 * GCC no longer generates code to maintain a stack backchain, 9598 previously used to generate stack backtraces for debugging 9599 purposes. As replacement that does not incur runtime overhead, 9600 DWARF-2 call frame information is provided by GCC; this is 9601 supported by GDB 6.1. The old behavior can be restored using the 9602 -mbackchain option. 9603 * The stack frame size of functions may now exceed 2 GB in 64-bit 9604 code. 9605 * A port for the 64-bit IBM TPF operating system has been added; the 9606 configuration is s390x-ibm-tpf. This configuration is supported as 9607 cross-compilation target only. 9608 * Various changes to improve the generated code have been 9609 implemented, including: 9610 + GCC now uses the MULTIPLY AND ADD and MULTIPLY AND SUBTRACT 9611 instructions to significantly speed up many floating-point 9612 applications. 9613 + GCC now uses the ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL 9614 WITH BORROW instructions to speed up long long arithmetic. 9615 + GCC now uses the SEARCH STRING instruction to implement 9616 strlen(). 9617 + In many cases, function call overhead for 31-bit code has been 9618 reduced by placing the literal pool after the function code 9619 instead of after the function prolog. 9620 + Register 14 is no longer reserved in 64-bit code. 9621 + Handling of global register variables has been improved. 9622 9623 SPARC 9624 9625 * The option -mflat is deprecated. 9626 * Support for large (> 2GB) frames has been added to the 64-bit port. 9627 * Several [25]ABI bugs have been fixed. Unfortunately, these changes 9628 will break binary compatibility with earlier releases. 9629 * The default debugging format has been switched from STABS to 9630 DWARF-2 for 32-bit code on Solaris 7 and later. DWARF-2 is already 9631 the default debugging format for 64-bit code on Solaris. 9632 9633 SuperH 9634 9635 * Support for the SH2E processor has been added. Enabled at run time 9636 with the -m2e command line switch, or at configure time by 9637 specifying sh2e as the machine part of the target triple. 9638 9639 V850 9640 9641 * Support for the Mitsubishi V850E1 processor has been added. This is 9642 a variant of the V850E processor with some additional debugging 9643 instructions. 9644 9645 Xtensa 9646 9647 * Several ABI bugs have been fixed. Unfortunately, these changes 9648 break binary compatibility with earlier releases. 9649 + For big-endian processors, the padding of aggregate return 9650 values larger than a word has changed. If the size of an 9651 aggregate return value is not a multiple of 32 bits, previous 9652 versions of GCC inserted padding in the most-significant bytes 9653 of the first return value register. Aggregates larger than a 9654 word are now padded in the least-significant bytes of the last 9655 return value register used. Aggregates smaller than a word are 9656 still padded in the most-significant bytes. The return value 9657 padding has not changed for little-endian processors. 9658 + Function arguments with 16-byte alignment are now properly 9659 aligned. 9660 + The implementation of the va_list type has changed. A va_list 9661 value created by va_start from a previous release cannot be 9662 used with va_arg from this release, or vice versa. 9663 * More processor configuration options for Xtensa processors are 9664 supported: 9665 + the ABS instruction is now optional; 9666 + the ADDX* and SUBX* instructions are now optional; 9667 + an experimental CONST16 instruction can be used to synthesize 9668 constants instead of loading them from constant pools. 9669 These and other Xtensa processor configuration options can no 9670 longer be enabled or disabled by command-line options; the 9671 processor configuration must be specified by the xtensa-config.h 9672 header file when building GCC. Additionally, the 9673 -mno-serialize-volatile option is no longer supported. 9674 9675 Obsolete Systems 9676 9677 Support for a number of older systems has been declared obsolete in GCC 9678 3.4. Unless there is activity to revive them, the next release of GCC 9679 will have their sources permanently removed. 9680 9681 All configurations of the following processor architectures have been 9682 declared obsolete: 9683 * Mitsubishi D30V, d30v-* 9684 * AT&T DSP1600 and DSP1610, dsp16xx-* 9685 * Intel 80960, i960 9686 9687 Also, some individual systems have been obsoleted: 9688 * ARM Family 9689 + Support for generating code for operation in APCS/26 mode 9690 (-mapcs-26). 9691 * IBM ESA/390 9692 + "Bigfoot" port, i370-*. (The other port, s390-*, is actively 9693 maintained and supported.) 9694 * Intel 386 family 9695 + MOSS, i?86-moss-msdos and i?86-*-moss* 9696 + NCR 3000 running System V r.4, i?86-ncr-sysv4* 9697 + FreeBSD with a.out object format, i?86-*-freebsd*aout* and 9698 i?86-*-freebsd2* 9699 + GNU/Linux with a.out object format, i?86-linux*aout* 9700 + GNU/Linux with libc5, a.k.a. glibc1, i?86-linux*libc1* 9701 + Interix versions before Interix 3, i?86-*-interix 9702 + Mach microkernel, i?86-mach* 9703 + SCO UnixWare with UDK, i?86-*-udk* 9704 + Generic System V releases 1, 2, and 3, i?86-*-sysv[123]* 9705 + VSTa microkernel, i386-*-vsta 9706 * Motorola M68000 family 9707 + HPUX, m68k-hp-hpux* and m68000-hp-hpux* 9708 + NetBSD with a.out object format (before NetBSD 1.4), 9709 m68k-*-*-netbsd* except m68k-*-*-netbsdelf* 9710 + Generic System V r.4, m68k-*-sysv4* 9711 * VAX 9712 + Generic VAX, vax-*-* (This is generic VAX only; we have not 9713 obsoleted any VAX triples for specific operating systems.) 9714 9715 Documentation improvements 9716 9717 Other significant improvements 9718 9719 * The build system has undergone several significant cleanups. 9720 Subdirectories will only be configured if they are being built, and 9721 all subdirectory configures are run from the make command. The top 9722 level has been autoconfiscated. 9723 * Building GCC no longer writes to its source directory. This should 9724 help those wishing to share a read-only source directory over NFS 9725 or build from a CD. The exceptions to this feature are if you 9726 configure with either --enable-maintainer-mode or 9727 --enable-generated-files-in-srcdir. 9728 * The -W warning option has been renamed to -Wextra, which is more 9729 easily understood. The older spelling will be retained for 9730 backwards compatibility. 9731 * Substantial improvements in compile time have been made, 9732 particularly for non-optimizing compilations. 9733 __________________________________________________________________ 9734 9735 GCC 3.4.0 9736 9737 Bug Fixes 9738 9739 A vast number of bugs have been fixed in 3.4.0, too many to publish a 9740 complete list here. [26]Follow this link to query the Bugzilla database 9741 for the list of over 900 bugs fixed in 3.4.0. This is the list of all 9742 bugs marked as resolved and fixed in 3.4.0 that are not flagged as 3.4 9743 regressions. 9744 __________________________________________________________________ 9745 9746 GCC 3.4.1 9747 9748 Bug Fixes 9749 9750 This section lists the problem reports (PRs) from GCC's bug tracking 9751 system that are known to be fixed in the 3.4.1 release. This list might 9752 not be complete (that is, it is possible that some PRs that have been 9753 fixed are not listed here). 9754 9755 Bootstrap failures 9756 9757 * [27]10129 Ada bootstrap fails on PPC-Darwin - invalid assembler 9758 emitted - PIC related 9759 * [28]14576 [ARM] ICE in libiberty when building gcc-3.4 for arm-elf 9760 * [29]14760 A bug in configure.in prevents using both 9761 --program-suffix and --program-prefix 9762 * [30]14671 [hppa64] bootstrap fails: ICE in 9763 save_call_clobbered_regs, in caller_save.c 9764 * [31]15093 [alpha][Java] make bootstrap fails to configure libffi on 9765 Alpha 9766 * [32]15178 Solaris 9/x86 fails linking after stage 3 9767 9768 Multi-platform internal compiler errors (ICEs) 9769 9770 * [33]12753 (preprocessor) Memory corruption in preprocessor on bad 9771 input 9772 * [34]13985 ICE in gcc.c-torture/compile/930621-1.c 9773 * [35]14810 (c++) tree check failures with invalid code involving 9774 templates 9775 * [36]14883 (c++) ICE on invalid code, in cp_parser_lookup_name, in 9776 cp/parser.c 9777 * [37]15044 (c++) ICE on syntax error, template header 9778 * [38]15057 (c++) Compiling of conditional value throw constructs 9779 cause a segmentation violation 9780 * [39]15064 (c++) typeid of template parameter gives ICE 9781 * [40]15142 (c++) ICE when passing a string where a char* is expected 9782 in a throw statement 9783 * [41]15159 ICE in rtl_verify_flow_info_1 9784 * [42]15165 (c++) ICE in instantiate_template 9785 * [43]15193 Unary minus using pointer to V4SF vector causes 9786 -fforce-mem to exhaust all memory 9787 * [44]15209 (c++) Runs out of memory with packed structs 9788 * [45]15227 (c++) Trouble with invalid function definition 9789 * [46]15285 (c++) instantiate_type ICE when forming pointer to 9790 template function 9791 * [47]15299 (c++) ICE in resolve_overloaded_unification 9792 * [48]15329 (c++) ICE on constructor of member template 9793 * [49]15550 ICE in extract_insn, in recog.c 9794 * [50]15554 (c++) ICE in tsubst_copy, in cp/pt.c 9795 * [51]15640 (c++) ICE on invalid code in arg_assoc, in 9796 cp/name-lookup.c 9797 * [52]15666 [unit-at-a-time] Gcc abort on valid code 9798 * [53]15696 (c++) ICE with bad pointer-to-member code 9799 * [54]15701 (c++) ICE with friends and template template parameter 9800 * [55]15761 ICE in do_SUBST, in combine.c 9801 * [56]15829 (c++) ICE on Botan-1.3.13 due to -funroll-loops 9802 9803 Ada 9804 9805 * [57]14538 All RTEMS targets broken for gnat 9806 9807 C front end 9808 9809 * [58]12391 missing warning about assigning to an incomplete type 9810 * [59]14649 atan(1.0) should not be a constant expression 9811 * [60]15004 [unit-at-a-time] no warning for unused paramater in 9812 static function 9813 * [61]15749 --pedantic-errors behaves differently from --pedantic 9814 with C-compiler on GNU/Linux 9815 9816 C++ compiler and library 9817 9818 * [62]10646 non-const reference is incorrectly matched in a "const T" 9819 partial specialization 9820 * [63]12077 wcin.rdbuf()->in_avail() return value too high 9821 * [64]13598 enc_filebuf doesn't work 9822 * [65]14211 const_cast returns lvalue but should be rvalue 9823 * [66]14220 num_put::do_put() undesired float/double behavior 9824 * [67]14245 problem with user-defined allocators in std::basic_string 9825 * [68]14340 libstdc++ Debug mode: failure to convert iterator to 9826 const_iterator 9827 * [69]14600 __gnu_cxx::stdio_sync_filebuf should expose internal 9828 FILE* 9829 * [70]14668 no warning anymore for reevaluation of declaration 9830 * [71]14775 LFS (large file support) tests missing 9831 * [72]14821 Duplicate namespace alias declaration should not conflict 9832 * [73]14930 Friend declaration ignored 9833 * [74]14932 cannot use offsetof to get offsets of array elements in 9834 g++ 3.4.0 9835 * [75]14950 [non unit-at-a-time] always_inline does not mix with 9836 templates and -O0 9837 * [76]14962 g++ ignores #pragma redefine_extname 9838 * [77]14975 Segfault on low-level write error during imbue 9839 * [78]15002 Linewise stream input is unusably slow (std::string slow) 9840 * [79]15025 compiler accepts redeclaration of template as 9841 non-template 9842 * [80]15046 [arm] Math functions misdetected by cross configuration 9843 * [81]15069 a bit test on a variable of enum type is miscompiled 9844 * [82]15074 g++ -lsupc++ still links against libstdc++ 9845 * [83]15083 spurious "statement has no effect" warning 9846 * [84]15096 parse error with templates and pointer to const member 9847 * [85]15287 combination of operator[] and operator .* fails in 9848 templates 9849 * [86]15317 __attribute__ unused in first parameter of constructor 9850 gives error 9851 * [87]15337 sizeof on incomplete type diagnostic 9852 * [88]15361 bitset<>::_Find_next fails 9853 * [89]15412 _GLIBCXX_ symbols symbols defined and used in different 9854 namespaces 9855 * [90]15427 valid code results in incomplete type error 9856 * [91]15471 Incorrect member pointer offsets in anonymous 9857 structs/unions 9858 * [92]15503 nested template problem 9859 * [93]15507 compiler hangs while laying out union 9860 * [94]15542 operator & and template definitions 9861 * [95]15565 SLES9: leading + sign for unsigned int with showpos 9862 * [96]15625 friend defined inside a template fails to find static 9863 function 9864 * [97]15629 Function templates, overloads, and friend name injection 9865 * [98]15742 'noreturn' attribute ignored in method of template 9866 functions. 9867 * [99]15775 Allocator::pointer consistently ignored 9868 * [100]15821 Duplicate namespace alias within namespace rejected 9869 * [101]15862 'enum yn' fails (confict with undeclared builtin) 9870 * [102]15875 rejects pointer to member in template 9871 * [103]15877 valid code using templates and anonymous enums is 9872 rejected 9873 * [104]15947 Puzzling error message for wrong destructor declaration 9874 in template class 9875 * [105]16020 cannot copy __gnu_debug::bitset 9876 * [106]16154 input iterator concept too restrictive 9877 * [107]16174 deducing top-level consts 9878 9879 Java 9880 9881 * [108]14315 Java compiler is not parallel make safe 9882 9883 Fortran 9884 9885 * [109]15151 [g77] incorrect logical i/o in 64-bit mode 9886 9887 Objective-C 9888 9889 * [110]7993 private variables cannot be shadowed in subclasses 9890 9891 Optimization bugs 9892 9893 * [111]15228 useless copies of floating point operands 9894 * [112]15345 [non-unit-at-a-time] unreferenced nested inline 9895 functions not optimized away 9896 * [113]15945 Incorrect floating point optimization 9897 * [114]15526 ftrapv aborts on 0 * (-1) 9898 * [115]14690 Miscompiled POOMA tests 9899 * [116]15112 GCC generates code to write to unchanging memory 9900 9901 Preprocessor 9902 9903 * [117]15067 Minor glitch in the source of cpp 9904 9905 Main driver program bugs 9906 9907 * [118]1963 collect2 interprets -oldstyle_liblookup as -o 9908 ldstyle_liblookup 9909 9910 x86-specific (Intel/AMD) 9911 9912 * [119]15717 Error: can't resolve `L0' {*ABS* section} - `xx' {*UND* 9913 section} 9914 9915 HPPA-specific 9916 9917 * [120]14782 GCC produces an unaligned data access at -O2 9918 * [121]14828 FAIL: gcc.c-torture/execute/20030408-1.c execution, -O2 9919 * [122]15202 ICE in reload_cse_simplify_operands, in postreload.c 9920 9921 IA64-specific 9922 9923 * [123]14610 __float80 constants incorrectly emitted 9924 * [124]14813 init_array sections are initialized in the wrong order 9925 * [125]14857 GCC segfault on duplicated asm statement 9926 * [126]15598 Gcc 3.4 ICE on valid code 9927 * [127]15653 Gcc 3.4 ICE on valid code 9928 9929 MIPS-specific 9930 9931 * [128]15189 wrong filling of delay slot with -march=mips1 -G0 9932 -mno-split-addresses -mno-explicit-relocs 9933 * [129]15331 Assembler error building gnatlib on IRIX 6.5 with GNU as 9934 2.14.91 9935 * [130]16144 Bogus reference to __divdf3 when -O1 9936 * [131]16176 Miscompilation of unaligned data in MIPS backend 9937 9938 PowerPC-specific 9939 9940 * [132]11591 ICE in gcc.dg/altivec-5.c 9941 * [133]12028 powerpc-eabispe produces bad sCOND operation 9942 * [134]14478 rs6000 geu/ltu patterns generate incorrect code 9943 * [135]14567 long double and va_arg complex args 9944 * [136]14715 Altivec stack layout may overlap gpr save with stack 9945 temps 9946 * [137]14902 (libstdc++) Stream checking functions fail when -pthread 9947 option is used. 9948 * [138]14924 Compiler ICE on valid code 9949 * [139]14960 -maltivec affects vector return with -mabi=no-altivec 9950 * [140]15106 vector varargs failure passing from altivec to 9951 non-altivec code for -m32 9952 * [141]16026 ICE in function.c:4804, assign_parms, when -mpowerpc64 & 9953 half-word operation 9954 * [142]15191 -maltivec -mabi=no-altivec results in mis-aligned lvx 9955 and stvx 9956 * [143]15662 Segmentation fault when an exception is thrown - even if 9957 try and catch are specified 9958 9959 s390-specific 9960 9961 * [144]15054 Bad code due to overlapping stack temporaries 9962 9963 SPARC-specific 9964 9965 * [145]15783 ICE with union assignment in 64-bit mode 9966 * [146]15626 GCC 3.4 emits "ld: warning: relocation error: 9967 R_SPARC_UA32" 9968 9969 x86-64-specific 9970 9971 * [147]14326 boehm-gc hardcodes to 3DNow! prefetch for x86_64 9972 * [148]14723 Backported -march=nocona from mainline 9973 * [149]15290 __float128 failed to pass to function properly 9974 9975 Cygwin/Mingw32-specific 9976 9977 * [150]15250 Option -mms-bitfields support on GCC 3.4 is not 9978 conformant to MS layout 9979 * [151]15551 -mtune=pentium4 -O2 with sjlj EH breaks stack probe 9980 worker on windows32 targets 9981 9982 Bugs specific to embedded processors 9983 9984 * [152]8309 [m68k] -m5200 produces erroneous SImode set of short 9985 varaible on stack 9986 * [153]13250 [SH] Gcc code for rotation clobbers the register, but 9987 gcc continues to use the register as if it was not clobbered 9988 * [154]13803 [coldfire] movqi operand constraints too restrictivefor 9989 TARGET_COLDFIRE 9990 * [155]14093 [SH] ICE for code when using -mhitachi option in SH 9991 * [156]14457 [m6811hc] ICE with simple c++ source 9992 * [157]14542 [m6811hc] ICE on simple source 9993 * [158]15100 [SH] cc1plus got hang-up on 9994 libstdc++-v3/testsuite/abi_check.cc 9995 * [159]15296 [CRIS] Delayed branch scheduling causing invalid code on 9996 cris-* 9997 * [160]15396 [SH] ICE with -O2 -fPIC 9998 * [161]15782 [coldfire] m68k_output_mi_thunk emits wrong code for 9999 ColdFire 10000 10001 Testsuite problems (compiler not affected) 10002 10003 * [162]11610 libstdc++ testcases 27_io/* don't work properly remotely 10004 * [163]15488 (libstdc++) possibly insufficient file permissions for 10005 executing test suite 10006 * [164]15489 (libstdc++) testsuite_files determined incorrectly 10007 10008 Documentation bugs 10009 10010 * [165]13928 (libstdc++) no whatis info in some man pages generated 10011 by doxygen 10012 * [166]14150 Ada documentation out of date 10013 * [167]14949 (c++) Need to document method visibility changes 10014 * [168]15123 libstdc++-doc: Allocators.3 manpage is empty 10015 __________________________________________________________________ 10016 10017 GCC 3.4.2 10018 10019 Bug Fixes 10020 10021 This section lists the problem reports (PRs) from GCC's bug tracking 10022 system that are known to be fixed in the 3.4.2 release. This list might 10023 not be complete (that is, it is possible that some PRs that have been 10024 fixed are not listed here). 10025 10026 Bootstrap failures and issues 10027 10028 * [169]16469 [mips-sgi-irix5.3] bootstrap fails in 10029 libstdc++-v3/testsuite 10030 * [170]16344 [hppa-linux-gnu] libstdc++'s PCH built by 10031 profiledbootstrap does not work with the built compiler 10032 * [171]16842 [Solaris/x86] mkheaders can not find mkheaders.conf 10033 10034 Multi-platform internal compiler errors (ICEs) 10035 10036 * [172]12608 (c++) ICE: expected class 't', have 'x' (error_mark) in 10037 cp_parser_class_specifier, in cp/parser.c 10038 * [173]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c 10039 * [174]15461 (c++) ICE due to NRV and inlining 10040 * [175]15890 (c++) ICE in c_expand_expr, in c-common.c 10041 * [176]16180 ICE: segmentation fault in RTL optimization 10042 * [177]16224 (c++) ICE in write_unscoped_name (template/namespace) 10043 * [178]16408 ICE: in delete_insn, in cfgrtl.c 10044 * [179]16529 (c++) ICE for: namespace-alias shall not be declared as 10045 the name of any other entity 10046 * [180]16698 (c++) ICE with exceptions and declaration of __cxa_throw 10047 * [181]16706 (c++) ICE in finish_member_declaration, in 10048 cp/semantics.c 10049 * [182]16810 (c++) Legal C++ program with cast gives ICE in 10050 build_ptrmemfunc 10051 * [183]16851 (c++) ICE when throwing a comma expression 10052 * [184]16870 (c++) Boost.Spirit causes ICE in tsubst, in cp/pt.c 10053 * [185]16904 (c++) ICE in finish_class_member_access_expr, in 10054 cp/typeck.c 10055 * [186]16905 (c++) ICE (segfault) with exceptions 10056 * [187]16964 (c++) ICE in cp_parser_class_specifier due to 10057 redefinition 10058 * [188]17068 (c++) ICE: tree check: expected class 'd', have 'x' 10059 (identifier_node) in dependent_template_p, in cp/pt.c 10060 10061 Preprocessor bugs 10062 10063 * [189]16366 Preprocessor option -remap causes memory corruption 10064 10065 Optimization 10066 10067 * [190]15345 unreferenced nested inline functions not optimized away 10068 * [191]16590 Incorrect execution when compiling with -O2 10069 * [192]16693 Bitwise AND is lost when used within a cast to an enum 10070 of the same precision 10071 * [193]17078 Jump into if(0) substatement fails 10072 10073 Problems in generated debug information 10074 10075 * [194]13956 incorrect stabs for nested local variables 10076 10077 C front end bugs 10078 10079 * [195]16684 GCC should not warn about redundant redeclarations of 10080 built-ins 10081 10082 C++ compiler and library 10083 10084 * [196]12658 Thread safety problems in locale::global() and 10085 locale::locale() 10086 * [197]13092 g++ accepts invalid pointer-to-member conversion 10087 * [198]15320 Excessive memory consumption 10088 * [199]16246 Incorrect template argument deduction 10089 * [200]16273 Memory exhausted when using nested classes and virtual 10090 functions 10091 * [201]16401 ostringstream in gcc 3.4.x very slow for big data 10092 * [202]16411 undefined reference to 10093 __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char> 10094 >::file() 10095 * [203]16489 G++ incorrectly rejects use of a null constant integral 10096 expression as a null constant pointer 10097 * [204]16618 offsetof fails with constant member 10098 * [205]16637 syntax error reported for valid input code 10099 * [206]16717 __attribute__((constructor)) broken in C++ 10100 * [207]16813 compiler error in DEBUG version of range insertion 10101 std::map::insert 10102 * [208]16853 pointer-to-member initialization from incompatible one 10103 accepted 10104 * [209]16889 ambiguity is not detected 10105 * [210]16959 Segmentation fault in ios_base::sync_with_stdio 10106 10107 Java compiler and library 10108 10109 * [211]7587 direct threaded interpreter not thread-safe 10110 * [212]16473 ServerSocket accept() leaks file descriptors 10111 * [213]16478 Hash synchronization deadlock with finalizers 10112 10113 Alpha-specific 10114 10115 * [214]10695 ICE in dwarf2out_frame_debug_expr, in dwarf2out.c 10116 * [215]16974 could not split insn (ice in final_scan_insn, in 10117 final.c) 10118 10119 x86-specific 10120 10121 * [216]16298 ICE in output_operand 10122 * [217]17113 ICE with SSE2 intrinsics 10123 10124 x86-64 specific 10125 10126 * [218]14697 libstdc++ couldn't find 32bit libgcc_s 10127 10128 MIPS-specific 10129 10130 * [219]15869 [mips64] No NOP after LW (with -mips1 -O0) 10131 * [220]16325 [mips64] value profiling clobbers gp on mips 10132 * [221]16357 [mipsisa64-elf] ICE copying 7 bytes between extern 10133 char[]s 10134 * [222]16380 [mips64] Use of uninitialised register after dbra 10135 conversion 10136 * [223]16407 [mips64] Unaligned access to local variables 10137 * [224]16643 [mips64] verify_local_live_at_start ICE after 10138 crossjumping & cfgcleanup 10139 10140 ARM-specific 10141 10142 * [225]15927 THUMB -O2: strength-reduced iteration variable ends up 10143 off by 1 10144 * [226]15948 THUMB: ICE with non-commutative cbranch 10145 * [227]17019 THUMB: bad switch statement in md code for 10146 addsi3_cbranch_scratch 10147 10148 IA64-specific 10149 10150 * [228]16130 ICE on valid code: in bundling, in config/ia64/ia64.c 10151 (-mtune=merced) 10152 * [229]16142 ICE on valid code: in bundling, in config/ia64/ia64.c 10153 (-mtune=itanium) 10154 * [230]16278 Gcc failed to build Linux kernel with -mtune=merced 10155 * [231]16414 ICE on valid code: typo in comparison of asm_noperands 10156 result 10157 * [232]16445 ICE on valid code: don't count ignored insns 10158 * [233]16490 ICE (segfault) while compiling with -fprofile-use 10159 * [234]16683 ia64 does not honor SUBTARGET_EXTRA_SPECS 10160 10161 PowerPC-specific 10162 10163 * [235]16195 (ppc64): Miscompilation of GCC 3.3.x by 3.4.x 10164 * [236]16239 ICE on ppc64 (mozilla 1.7 compile, -O1 -fno-exceptions 10165 issue) 10166 10167 SPARC-specific 10168 10169 * [237]16199 ICE while compiling apache 2.0.49 10170 * [238]16416 -m64 doesn't imply -mcpu=v9 anymore 10171 * [239]16430 ICE when returning non-C aggregates larger than 16 bytes 10172 10173 Bugs specific to embedded processors 10174 10175 * [240]16379 [m32r] can't output large model function call of memcpy 10176 * [241]17093 [m32r] ICE with -msdata=use -O0 10177 * [242]17119 [m32r] ICE at switch case 0x8000 10178 10179 DJGPP-specific 10180 10181 * [243]15928 libstdc++ in 3.4.x doesn't cross-compile for djgpp 10182 10183 Alpha Tru64-specific 10184 10185 * [244]16210 libstdc++ gratuitously omits "long long" I/O 10186 10187 Testsuite, documentation issues (compiler is not affected): 10188 10189 * [245]15488 (libstdc++) possibly insufficient file permissions for 10190 executing test suite 10191 * [246]16250 ada/doctools runs makeinfo even in release tarball 10192 __________________________________________________________________ 10193 10194 GCC 3.4.3 10195 10196 This is the [247]list of problem reports (PRs) from GCC's bug tracking 10197 system that are known to be fixed in the 3.4.3 release. This list might 10198 not be complete (that is, it is possible that some PRs that have been 10199 fixed are not listed here). 10200 10201 Bootstrap failures 10202 10203 * [248]17369 [ia64] Bootstrap failure with binutils-2.15.90.0.1.1 10204 * [249]17850 [arm-elf] bootstrap failure - libstdc++ uses strtold 10205 when undeclared 10206 10207 Internal compiler errors (ICEs) affecting multiple platforms 10208 10209 * [250]13948 (java) GCJ segmentation fault while compiling GL4Java 10210 .class files 10211 * [251]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c 10212 * [252]16301 (c++) ICE when "strong" attribute is attached to a using 10213 directive 10214 * [253]16566 ICE with flexible arrays 10215 * [254]17023 ICE with nested functions in parameter declaration 10216 * [255]17027 ICE with noreturn function in loop at -O2 10217 * [256]17524 ICE in grokdeclarator, in cp/decl.c 10218 * [257]17826 (c++) ICE in cp_tree_equal 10219 10220 C and optimization bugs 10221 10222 * [258]15526 -ftrapv aborts on 0 * (-1) 10223 * [259]16999 #ident stopped working 10224 * [260]17503 quadratic behaviour in invalid_mode_change_p 10225 * [261]17581 Long long arithmetic fails inside a switch/case 10226 statement when compiled with -O2 10227 * [262]18129 -fwritable-strings doesn't work 10228 10229 C++ compiler and library bugs 10230 10231 * [263]10975 incorrect initial ostringstream::tellp() 10232 * [264]11722 Unbuffered filebuf::sgetn is slow 10233 * [265]14534 Unrecognizing static function as a template parameter 10234 when its return value is also templated 10235 * [266]15172 Copy constructor optimization in aggregate 10236 initialization 10237 * [267]15786 Bad error message for frequently occuring error. 10238 * [268]16162 Rejects valid member-template-definition 10239 * [269]16612 empty basic_strings can't live in shared memory 10240 * [270]16715 std::basic_iostream is instantiated when used, even 10241 though instantiations are already contained in libstdc++ 10242 * [271]16848 code in /ext/demangle.h appears broken 10243 * [272]17132 GCC fails to eliminate function template specialization 10244 when argument deduction fails 10245 * [273]17259 One more _S_leaf incorrectly qualified with _RopeRep:: 10246 in ropeimpl.h 10247 * [274]17327 use of `enumeral_type' in template type unification 10248 * [275]17393 "unused variable '._0'" warning with -Wall 10249 * [276]17501 Confusion with member templates 10250 * [277]17537 g++ not passing -lstdc++ to linker when all command line 10251 arguments are libraries 10252 * [278]17585 usage of unqualified name of static member from within 10253 class not allowed 10254 * [279]17821 Poor diagnostic for using "." instead of "->" 10255 * [280]17829 wrong error: call of overloaded function is ambiguous 10256 * [281]17851 Misleading diagnostic for invalid function declarations 10257 with undeclared types 10258 * [282]17976 Destructor is called twice 10259 * [283]18020 rejects valid definition of enum value in template 10260 * [284]18093 bogus conflict in namespace aliasing 10261 * [285]18140 C++ parser bug when using >> in templates 10262 10263 Fortran 10264 10265 * [286]17541 data statements with double precision constants fail 10266 10267 x86-specific 10268 10269 * [287]17853 -O2 ICE for MMX testcase 10270 10271 SPARC-specific 10272 10273 * [288]17245 ICE compiling gsl-1.5 statistics/lag1.c 10274 10275 Darwin-specific 10276 10277 * [289]17167 FATAL:Symbol L_foo$stub already defined. 10278 10279 AIX-specific 10280 10281 * [290]17277 could not catch an exception when specified -maix64 10282 10283 Solaris-specific 10284 10285 * [291]17505 <cmath> calls acosf(), ceilf(), and other functions 10286 missing from system libraries 10287 10288 HP/UX specific: 10289 10290 * [292]17684 /usr/ccs/bin/ld: Can't create libgcc_s.sl 10291 10292 ARM-specific 10293 10294 * [293]17384 ICE with mode attribute on structures 10295 10296 MIPS-specific 10297 10298 * [294]17770 No NOP after LWL with -mips1 10299 10300 Other embedded target specific 10301 10302 * [295]11476 [arc-elf] gcc ICE on newlib's vfprintf.c 10303 * [296]14064 [avr-elf] -fdata-sections triggers ICE 10304 * [297]14678 [m68hc11-elf] gcc ICE 10305 * [298]15583 [powerpc-rtems] powerpc-rtems lacks __USE_INIT_FINI__ 10306 * [299]15790 [i686-coff] Alignment error building gcc with i686-coff 10307 target 10308 * [300]15886 [SH] Miscompilation with -O2 -fPIC 10309 * [301]16884 [avr-elf] [fweb related] bug while initializing 10310 variables 10311 10312 Bugs relating to debugger support 10313 10314 * [302]13841 missing debug info for _Complex function arguments 10315 * [303]15860 [big-endian targets] No DW_AT_location debug info is 10316 emitted for formal arguments to a function that uses "register" 10317 qualifiers 10318 10319 Testsuite issues (compiler not affected) 10320 10321 * [304]17465 Testsuite in libffi overrides LD_LIBRARY_PATH 10322 * [305]17469 Testsuite in libstdc++ overrides LD_LIBRARY_PATH 10323 * [306]18138 [mips-sgi-irix6.5] libgcc_s.so.1 not found by 64-bit 10324 testsuite 10325 10326 Documentation 10327 10328 * [307]15498 typo in gcc manual: non-existing locale example en_UK, 10329 should be en_GB 10330 * [308]15747 [mips-sgi-irix5.3] /bin/sh hangs during bootstrap: 10331 document broken shell 10332 * [309]16406 USE_LD_AS_NEEDED undocumented 10333 __________________________________________________________________ 10334 10335 GCC 3.4.4 10336 10337 This is the [310]list of problem reports (PRs) from GCC's bug tracking 10338 system that are known to be fixed in the 3.4.4 release. This list might 10339 not be complete (that is, it is possible that some PRs that have been 10340 fixed are not listed here). 10341 __________________________________________________________________ 10342 10343 GCC 3.4.5 10344 10345 This is the [311]list of problem reports (PRs) from GCC's bug tracking 10346 system that are known to be fixed in the 3.4.5 release. This list might 10347 not be complete (that is, it is possible that some PRs that have been 10348 fixed are not listed here). 10349 10350 Bootstrap issues 10351 10352 * [312]24688 sco_math fixincl breaks math.h 10353 10354 C compiler bugs 10355 10356 * [313]17188 struct Foo { } redefinition 10357 * [314]20187 wrong code for ((unsigned char)(unsigned long 10358 long)((a?a:1)&(a*b)))?0:1) 10359 * [315]21873 infinite warning loop on bad array initializer 10360 * [316]21899 enum definition accepts values to be overriden 10361 * [317]22061 ICE in find_function_data, in function.c 10362 * [318]22308 Failure to diagnose violation of constraint 6.516p2 10363 * [319]22458 ICE on missing brace 10364 * [320]22589 ICE casting to long long 10365 * [321]24101 Segfault with preprocessed source 10366 10367 C++ compiler and library bugs 10368 10369 * [322]10611 operations on vector mode not recognized in C++ 10370 * [323]13377 unexpected behavior of namespace usage directive 10371 * [324]16002 Strange error message with new parser 10372 * [325]17413 local classes as template argument 10373 * [326]17609 spurious error message after using keyword 10374 * [327]17618 ICE in cp_convert_to_pointer, in cp/cvt.c 10375 * [328]18124 ICE with invalid template template parameter 10376 * [329]18155 typedef in template declaration not rejected 10377 * [330]18177 ICE with const_cast for undeclared variable 10378 * [331]18368 C++ error message regression 10379 * [332]16378 ICE when returning a copy of a packed member 10380 * [333]18466 int ::i; accepted 10381 * [334]18512 ICE on invalid usage of template base class 10382 * [335]18454 ICE when returning undefined type 10383 * [336]18738 typename not allowed with non-dependent qualified name 10384 * [337]18803 rejects access to operator() in template 10385 * [338]19004 ICE in uses_template_parms, in cp/pt.c 10386 * [339]19208 Spurious error about variably modified type 10387 * [340]18253 bad error message / ICE for invalid template parameter 10388 * [341]19608 ICE after friend function definition in local class 10389 * [342]19884 ICE on explicit instantiation of a non-template 10390 constructor 10391 * [343]20153 ICE when C++ template function contains anonymous union 10392 * [344]20563 Infinite loop in diagnostic (and ice after error 10393 message) 10394 * [345]20789 ICE with incomplete type in template 10395 * [346]21336 Internal compiler error when using custom new operators 10396 * [347]21768 ICE in error message due to violation of coding 10397 conventions 10398 * [348]21853 constness of pointer to data member ignored 10399 * [349]21903 Default argument of template function causes a 10400 compile-time error 10401 * [350]21983 multiple diagnostics 10402 * [351]21987 New testsuite failure 10403 g++.dg/warn/conversion-function-1.C 10404 * [352]22153 ICE on invalid template specialization 10405 * [353]22172 Internal compiler error, seg fault. 10406 * [354]21286 filebuf::xsgetn vs pipes 10407 * [355]22233 ICE with wrong number of template parameters 10408 * [356]22508 ICE after invalid operator new 10409 * [357]22545 ICE with pointer to class member & user defined 10410 conversion operator 10411 * [358]23528 Wrong default allocator in ext/hash_map 10412 * [359]23550 char_traits requirements/1.cc test bad math 10413 * [360]23586 Bad diagnostic for invalid namespace-name 10414 * [361]23624 ICE in invert_truthvalue, in fold-const.c 10415 * [362]23639 Bad error message: not a member of '<declaration error>' 10416 * [363]23797 ICE on typename outside template 10417 * [364]23965 Bogus error message: no matching function for call to 10418 'foo(<type error>)' 10419 * [365]24052 &#`label_decl' not supported by dump_expr#<expression 10420 error> 10421 * [366]24580 virtual base class cause exception not to be caught 10422 10423 Problems in generated debug information 10424 10425 * [367]24267 Bad DWARF for altivec vectors 10426 10427 Optimizations issues 10428 10429 * [368]17810 ICE in verify_local_live_at_start 10430 * [369]17860 Wrong generated code for loop with varying bound 10431 * [370]21709 ICE on compile-time complex NaN 10432 * [371]21964 broken tail call at -O2 or more 10433 * [372]22167 Strange optimization bug when using -Os 10434 * [373]22619 Compilation failure for real_const_1.f and 10435 real_const_2.f90 10436 * [374]23241 Invalid code generated for comparison of uchar to 255 10437 * [375]23478 Miscompilation due to reloading of a var that is also 10438 used in EH pad 10439 * [376]24470 segmentation fault in cc1plus when compiling with -O 10440 * [377]24950 ICE in operand_subword_force 10441 10442 Precompiled headers problems 10443 10444 * [378]14400 Cannot compile qt-x11-free-3.3.0 10445 * [379]14940 PCH largefile test fails on various platforms 10446 10447 Preprocessor bugs 10448 10449 * [380]20239 ICE on empty preprocessed input 10450 * [381]15220 "gcc -E -MM -MG" reports missing system headers in 10451 source directory 10452 10453 Testsuite issues 10454 10455 * [382]19275 gcc.dg/20020919-1.c fails with -fpic/-fPIC on 10456 i686-pc-linux-gnu 10457 10458 Alpha specific 10459 10460 * [383]21888 bootstrap failure with linker relaxation enabled 10461 10462 ARM specific 10463 10464 * [384]15342 [arm-linux]: ICE in verify_local_live_at_start 10465 * [385]23985 Memory aliasing information incorrect in inlined memcpy 10466 10467 ColdFile specific 10468 10469 * [386]16719 Illegal move of byte into address register causes 10470 compiler to ICE 10471 10472 HPPA specific 10473 10474 * [387]21723 ICE while building libgfortran 10475 * [388]21841 -mhp-ld/-mgnu-ld documentation 10476 10477 IA-64 specific 10478 10479 * [389]23644 IA-64 hardware models and configuration options 10480 documentation error 10481 * [390]24718 Shared libgcc not used for linking by default 10482 10483 M68000 specific 10484 10485 * [391]18421 ICE in reload_cse_simplify_operands, in postreload.c 10486 10487 MIPS specific 10488 10489 * [392]20621 ICE in change_address_1, in emit-rtl.c 10490 10491 PowerPC and PowerPC64 specific 10492 10493 * [393]18583 error on valid code: const 10494 __attribute__((altivec(vector__))) doesn't work in arrays 10495 * [394]20191 ICE in reload_cse_simplify_operands 10496 * [395]22083 AIX: TARGET_C99_FUNCTIONS is wrongly defined 10497 * [396]23070 CALL_V4_CLEAR_FP_ARGS flag not properly set 10498 * [397]23404 gij trashes args of functions with more than 8 fp args 10499 * [398]23539 C & C++ compiler generating misaligned references 10500 regardless of compiler flags 10501 * [399]24102 floatdisf2_internal2 broken 10502 * [400]24465 -mminimal-toc miscompilation of __thread vars 10503 10504 Solaris specific 10505 10506 * [401]19933 Problem with define of HUGE_VAL in math_c99 10507 * [402]21889 Native Solaris assembler cannot grok DTP-relative debug 10508 symbols 10509 10510 SPARC specific 10511 10512 * [403]19300 PCH failures on sparc-linux 10513 * [404]20301 Assembler labels have a leading "-" 10514 * [405]20673 C PCH testsuite assembly comparison failure 10515 10516 x86 and x86_64 specific 10517 10518 * [406]18582 ICE with arrays of type V2DF 10519 * [407]19340 Compilation SEGFAULTs with -O1 -fschedule-insns2 10520 -fsched2-use-traces 10521 * [408]21716 ICE in reg-stack.c's swap_rtx_condition 10522 * [409]24315 amd64 fails -fpeephole2 10523 __________________________________________________________________ 10524 10525 GCC 3.4.6 10526 10527 This is the [410]list of problem reports (PRs) from GCC's bug tracking 10528 system that are known to be fixed in the 3.4.6 release. This list might 10529 not be complete (that is, it is possible that some PRs that have been 10530 fixed are not listed here). 10531 10532 10533 For questions related to the use of GCC, please consult these web 10534 pages and the [411]GCC manuals. If that fails, the 10535 [412]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 10536 web pages and the development of GCC are welcome on our developer 10537 list at [413]gcc (a] gcc.gnu.org. All of [414]our lists have public 10538 archives. 10539 10540 Copyright (C) [415]Free Software Foundation, Inc. Verbatim copying and 10541 distribution of this entire article is permitted in any medium, 10542 provided this notice is preserved. 10543 10544 These pages are [416]maintained by the GCC team. Last modified 10545 2016-12-30[417]. 10546 10547 References 10548 10549 1. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6 10550 2. http://gcc.gnu.org/gcc-3.4/changes.html#cplusplus 10551 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems 10552 4. http://gcc.gnu.org/gcc-3.4/changes.html#obsolete_systems 10553 5. http://gcc.gnu.org/gcc-3.4/mips-abi.html 10554 6. http://gcc.gnu.org/gcc-3.4/sparc-abi.html 10555 7. http://www.boost.org/ 10556 8. https://gcc.gnu.org/PR11953 10557 9. https://gcc.gnu.org/PR8361 10558 10. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Other-Builtins.html#Other Builtins 10559 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#209 10560 12. http://gcc.gnu.org/bugs/#cxx_rvalbind 10561 13. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html 10562 14. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html 10563 15. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html 10564 16. http://www.gnu.org/software/classpath/ 10565 17. http://www.eclipse.org/ 10566 18. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/g77/News.html 10567 19. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Alpha-Built-in-Functions.html 10568 20. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html 10569 21. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Comparison-of-the-two-descriptions.html 10570 22. https://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html 10571 23. http://gcc.gnu.org/gcc-3.4/mips-abi.html 10572 24. http://gcc.gnu.org/gcc-3.4/powerpc-abi.html 10573 25. http://gcc.gnu.org/gcc-3.4/sparc-abi.html 10574 26. https://gcc.gnu.org/bugzilla/buglist.cgi?short_desc_type=notregexp&short_desc=\[3\.4.*[Rr]egression&target_milestone=3.4.0&bug_status=RESOLVED&resolution=FIXED 10575 27. https://gcc.gnu.org/PR10129 10576 28. https://gcc.gnu.org/PR14576 10577 29. https://gcc.gnu.org/PR14760 10578 30. https://gcc.gnu.org/PR14671 10579 31. https://gcc.gnu.org/PR15093 10580 32. https://gcc.gnu.org/PR15178 10581 33. https://gcc.gnu.org/PR12753 10582 34. https://gcc.gnu.org/PR13985 10583 35. https://gcc.gnu.org/PR14810 10584 36. https://gcc.gnu.org/PR14883 10585 37. https://gcc.gnu.org/PR15044 10586 38. https://gcc.gnu.org/PR15057 10587 39. https://gcc.gnu.org/PR15064 10588 40. https://gcc.gnu.org/PR15142 10589 41. https://gcc.gnu.org/PR15159 10590 42. https://gcc.gnu.org/PR15165 10591 43. https://gcc.gnu.org/PR15193 10592 44. https://gcc.gnu.org/PR15209 10593 45. https://gcc.gnu.org/PR15227 10594 46. https://gcc.gnu.org/PR15285 10595 47. https://gcc.gnu.org/PR15299 10596 48. https://gcc.gnu.org/PR15329 10597 49. https://gcc.gnu.org/PR15550 10598 50. https://gcc.gnu.org/PR15554 10599 51. https://gcc.gnu.org/PR15640 10600 52. https://gcc.gnu.org/PR15666 10601 53. https://gcc.gnu.org/PR15696 10602 54. https://gcc.gnu.org/PR15701 10603 55. https://gcc.gnu.org/PR15761 10604 56. https://gcc.gnu.org/PR15829 10605 57. https://gcc.gnu.org/PR14538 10606 58. https://gcc.gnu.org/PR12391 10607 59. https://gcc.gnu.org/PR14649 10608 60. https://gcc.gnu.org/PR15004 10609 61. https://gcc.gnu.org/PR15749 10610 62. https://gcc.gnu.org/PR10646 10611 63. https://gcc.gnu.org/PR12077 10612 64. https://gcc.gnu.org/PR13598 10613 65. https://gcc.gnu.org/PR14211 10614 66. https://gcc.gnu.org/PR14220 10615 67. https://gcc.gnu.org/PR14245 10616 68. https://gcc.gnu.org/PR14340 10617 69. https://gcc.gnu.org/PR14600 10618 70. https://gcc.gnu.org/PR14668 10619 71. https://gcc.gnu.org/PR14775 10620 72. https://gcc.gnu.org/PR14821 10621 73. https://gcc.gnu.org/PR14930 10622 74. https://gcc.gnu.org/PR14932 10623 75. https://gcc.gnu.org/PR14950 10624 76. https://gcc.gnu.org/PR14962 10625 77. https://gcc.gnu.org/PR14975 10626 78. https://gcc.gnu.org/PR15002 10627 79. https://gcc.gnu.org/PR15025 10628 80. https://gcc.gnu.org/PR15046 10629 81. https://gcc.gnu.org/PR15069 10630 82. https://gcc.gnu.org/PR15074 10631 83. https://gcc.gnu.org/PR15083 10632 84. https://gcc.gnu.org/PR15096 10633 85. https://gcc.gnu.org/PR15287 10634 86. https://gcc.gnu.org/PR15317 10635 87. https://gcc.gnu.org/PR15337 10636 88. https://gcc.gnu.org/PR15361 10637 89. https://gcc.gnu.org/PR15412 10638 90. https://gcc.gnu.org/PR15427 10639 91. https://gcc.gnu.org/PR15471 10640 92. https://gcc.gnu.org/PR15503 10641 93. https://gcc.gnu.org/PR15507 10642 94. https://gcc.gnu.org/PR15542 10643 95. https://gcc.gnu.org/PR15565 10644 96. https://gcc.gnu.org/PR15625 10645 97. https://gcc.gnu.org/PR15629 10646 98. https://gcc.gnu.org/PR15742 10647 99. https://gcc.gnu.org/PR15775 10648 100. https://gcc.gnu.org/PR15821 10649 101. https://gcc.gnu.org/PR15862 10650 102. https://gcc.gnu.org/PR15875 10651 103. https://gcc.gnu.org/PR15877 10652 104. https://gcc.gnu.org/PR15947 10653 105. https://gcc.gnu.org/PR16020 10654 106. https://gcc.gnu.org/PR16154 10655 107. https://gcc.gnu.org/PR16174 10656 108. https://gcc.gnu.org/PR14315 10657 109. https://gcc.gnu.org/PR15151 10658 110. https://gcc.gnu.org/PR7993 10659 111. https://gcc.gnu.org/PR15228 10660 112. https://gcc.gnu.org/PR15345 10661 113. https://gcc.gnu.org/PR15945 10662 114. https://gcc.gnu.org/PR15526 10663 115. https://gcc.gnu.org/PR14690 10664 116. https://gcc.gnu.org/PR15112 10665 117. https://gcc.gnu.org/PR15067 10666 118. https://gcc.gnu.org/PR1963 10667 119. https://gcc.gnu.org/PR15717 10668 120. https://gcc.gnu.org/PR14782 10669 121. https://gcc.gnu.org/PR14828 10670 122. https://gcc.gnu.org/PR15202 10671 123. https://gcc.gnu.org/PR14610 10672 124. https://gcc.gnu.org/PR14813 10673 125. https://gcc.gnu.org/PR14857 10674 126. https://gcc.gnu.org/PR15598 10675 127. https://gcc.gnu.org/PR15653 10676 128. https://gcc.gnu.org/PR15189 10677 129. https://gcc.gnu.org/PR15331 10678 130. https://gcc.gnu.org/PR16144 10679 131. https://gcc.gnu.org/PR16176 10680 132. https://gcc.gnu.org/PR11591 10681 133. https://gcc.gnu.org/PR12028 10682 134. https://gcc.gnu.org/PR14478 10683 135. https://gcc.gnu.org/PR14567 10684 136. https://gcc.gnu.org/PR14715 10685 137. https://gcc.gnu.org/PR14902 10686 138. https://gcc.gnu.org/PR14924 10687 139. https://gcc.gnu.org/PR14960 10688 140. https://gcc.gnu.org/PR15106 10689 141. https://gcc.gnu.org/PR16026 10690 142. https://gcc.gnu.org/PR15191 10691 143. https://gcc.gnu.org/PR15662 10692 144. https://gcc.gnu.org/PR15054 10693 145. https://gcc.gnu.org/PR15783 10694 146. https://gcc.gnu.org/PR15626 10695 147. https://gcc.gnu.org/PR14326 10696 148. https://gcc.gnu.org/PR14723 10697 149. https://gcc.gnu.org/PR15290 10698 150. https://gcc.gnu.org/PR15250 10699 151. https://gcc.gnu.org/PR15551 10700 152. https://gcc.gnu.org/PR8309 10701 153. https://gcc.gnu.org/PR13250 10702 154. https://gcc.gnu.org/PR13803 10703 155. https://gcc.gnu.org/PR14093 10704 156. https://gcc.gnu.org/PR14457 10705 157. https://gcc.gnu.org/PR14542 10706 158. https://gcc.gnu.org/PR15100 10707 159. https://gcc.gnu.org/PR15296 10708 160. https://gcc.gnu.org/PR15396 10709 161. https://gcc.gnu.org/PR15782 10710 162. https://gcc.gnu.org/PR11610 10711 163. https://gcc.gnu.org/PR15488 10712 164. https://gcc.gnu.org/PR15489 10713 165. https://gcc.gnu.org/PR13928 10714 166. https://gcc.gnu.org/PR14150 10715 167. https://gcc.gnu.org/PR14949 10716 168. https://gcc.gnu.org/PR15123 10717 169. https://gcc.gnu.org/PR16469 10718 170. https://gcc.gnu.org/PR16344 10719 171. https://gcc.gnu.org/PR16842 10720 172. https://gcc.gnu.org/PR12608 10721 173. https://gcc.gnu.org/PR14492 10722 174. https://gcc.gnu.org/PR15461 10723 175. https://gcc.gnu.org/PR15890 10724 176. https://gcc.gnu.org/PR16180 10725 177. https://gcc.gnu.org/PR16224 10726 178. https://gcc.gnu.org/PR16408 10727 179. https://gcc.gnu.org/PR16529 10728 180. https://gcc.gnu.org/PR16698 10729 181. https://gcc.gnu.org/PR16706 10730 182. https://gcc.gnu.org/PR16810 10731 183. https://gcc.gnu.org/PR16851 10732 184. https://gcc.gnu.org/PR16870 10733 185. https://gcc.gnu.org/PR16904 10734 186. https://gcc.gnu.org/PR16905 10735 187. https://gcc.gnu.org/PR16964 10736 188. https://gcc.gnu.org/PR17068 10737 189. https://gcc.gnu.org/PR16366 10738 190. https://gcc.gnu.org/PR15345 10739 191. https://gcc.gnu.org/PR16590 10740 192. https://gcc.gnu.org/PR16693 10741 193. https://gcc.gnu.org/PR17078 10742 194. https://gcc.gnu.org/PR13956 10743 195. https://gcc.gnu.org/PR16684 10744 196. https://gcc.gnu.org/PR12658 10745 197. https://gcc.gnu.org/PR13092 10746 198. https://gcc.gnu.org/PR15320 10747 199. https://gcc.gnu.org/PR16246 10748 200. https://gcc.gnu.org/PR16273 10749 201. https://gcc.gnu.org/PR16401 10750 202. https://gcc.gnu.org/PR16411 10751 203. https://gcc.gnu.org/PR16489 10752 204. https://gcc.gnu.org/PR16618 10753 205. https://gcc.gnu.org/PR16637 10754 206. https://gcc.gnu.org/PR16717 10755 207. https://gcc.gnu.org/PR16813 10756 208. https://gcc.gnu.org/PR16853 10757 209. https://gcc.gnu.org/PR16889 10758 210. https://gcc.gnu.org/PR16959 10759 211. https://gcc.gnu.org/PR7587 10760 212. https://gcc.gnu.org/PR16473 10761 213. https://gcc.gnu.org/PR16478 10762 214. https://gcc.gnu.org/PR10695 10763 215. https://gcc.gnu.org/PR16974 10764 216. https://gcc.gnu.org/PR16298 10765 217. https://gcc.gnu.org/PR17113 10766 218. https://gcc.gnu.org/PR14697 10767 219. https://gcc.gnu.org/PR15869 10768 220. https://gcc.gnu.org/PR16325 10769 221. https://gcc.gnu.org/PR16357 10770 222. https://gcc.gnu.org/PR16380 10771 223. https://gcc.gnu.org/PR16407 10772 224. https://gcc.gnu.org/PR16643 10773 225. https://gcc.gnu.org/PR15927 10774 226. https://gcc.gnu.org/PR15948 10775 227. https://gcc.gnu.org/PR17019 10776 228. https://gcc.gnu.org/PR16130 10777 229. https://gcc.gnu.org/PR16142 10778 230. https://gcc.gnu.org/PR16278 10779 231. https://gcc.gnu.org/PR16414 10780 232. https://gcc.gnu.org/PR16445 10781 233. https://gcc.gnu.org/PR16490 10782 234. https://gcc.gnu.org/PR16683 10783 235. https://gcc.gnu.org/PR16195 10784 236. https://gcc.gnu.org/PR16239 10785 237. https://gcc.gnu.org/PR16199 10786 238. https://gcc.gnu.org/PR16416 10787 239. https://gcc.gnu.org/PR16430 10788 240. https://gcc.gnu.org/PR16379 10789 241. https://gcc.gnu.org/PR17093 10790 242. https://gcc.gnu.org/PR17119 10791 243. https://gcc.gnu.org/PR15928 10792 244. https://gcc.gnu.org/PR16210 10793 245. https://gcc.gnu.org/PR15488 10794 246. https://gcc.gnu.org/PR16250 10795 247. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.3 10796 248. https://gcc.gnu.org/PR17369 10797 249. https://gcc.gnu.org/PR17850 10798 250. https://gcc.gnu.org/PR13948 10799 251. https://gcc.gnu.org/PR14492 10800 252. https://gcc.gnu.org/PR16301 10801 253. https://gcc.gnu.org/PR16566 10802 254. https://gcc.gnu.org/PR17023 10803 255. https://gcc.gnu.org/PR17027 10804 256. https://gcc.gnu.org/PR17524 10805 257. https://gcc.gnu.org/PR17826 10806 258. https://gcc.gnu.org/PR15526 10807 259. https://gcc.gnu.org/PR16999 10808 260. https://gcc.gnu.org/PR17503 10809 261. https://gcc.gnu.org/PR17581 10810 262. https://gcc.gnu.org/PR18129 10811 263. https://gcc.gnu.org/PR10975 10812 264. https://gcc.gnu.org/PR11722 10813 265. https://gcc.gnu.org/PR14534 10814 266. https://gcc.gnu.org/PR15172 10815 267. https://gcc.gnu.org/PR15786 10816 268. https://gcc.gnu.org/PR16162 10817 269. https://gcc.gnu.org/PR16612 10818 270. https://gcc.gnu.org/PR16715 10819 271. https://gcc.gnu.org/PR16848 10820 272. https://gcc.gnu.org/PR17132 10821 273. https://gcc.gnu.org/PR17259 10822 274. https://gcc.gnu.org/PR17327 10823 275. https://gcc.gnu.org/PR17393 10824 276. https://gcc.gnu.org/PR17501 10825 277. https://gcc.gnu.org/PR17537 10826 278. https://gcc.gnu.org/PR17585 10827 279. https://gcc.gnu.org/PR17821 10828 280. https://gcc.gnu.org/PR17829 10829 281. https://gcc.gnu.org/PR17851 10830 282. https://gcc.gnu.org/PR17976 10831 283. https://gcc.gnu.org/PR18020 10832 284. https://gcc.gnu.org/PR18093 10833 285. https://gcc.gnu.org/PR18140 10834 286. https://gcc.gnu.org/PR17541 10835 287. https://gcc.gnu.org/PR17853 10836 288. https://gcc.gnu.org/PR17245 10837 289. https://gcc.gnu.org/PR17167 10838 290. https://gcc.gnu.org/PR17277 10839 291. https://gcc.gnu.org/PR17505 10840 292. https://gcc.gnu.org/PR17684 10841 293. https://gcc.gnu.org/PR17384 10842 294. https://gcc.gnu.org/PR17770 10843 295. https://gcc.gnu.org/PR11476 10844 296. https://gcc.gnu.org/PR14064 10845 297. https://gcc.gnu.org/PR14678 10846 298. https://gcc.gnu.org/PR15583 10847 299. https://gcc.gnu.org/PR15790 10848 300. https://gcc.gnu.org/PR15886 10849 301. https://gcc.gnu.org/PR16884 10850 302. https://gcc.gnu.org/PR13841 10851 303. https://gcc.gnu.org/PR15860 10852 304. https://gcc.gnu.org/PR17465 10853 305. https://gcc.gnu.org/PR17469 10854 306. https://gcc.gnu.org/PR18138 10855 307. https://gcc.gnu.org/PR15498 10856 308. https://gcc.gnu.org/PR15747 10857 309. https://gcc.gnu.org/PR16406 10858 310. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.4 10859 311. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.5 10860 312. https://gcc.gnu.org/PR24688 10861 313. https://gcc.gnu.org/PR17188 10862 314. https://gcc.gnu.org/PR20187 10863 315. https://gcc.gnu.org/PR21873 10864 316. https://gcc.gnu.org/PR21899 10865 317. https://gcc.gnu.org/PR22061 10866 318. https://gcc.gnu.org/PR22208 10867 319. https://gcc.gnu.org/PR22458 10868 320. https://gcc.gnu.org/PR22589 10869 321. https://gcc.gnu.org/PR24101 10870 322. https://gcc.gnu.org/PR10611 10871 323. https://gcc.gnu.org/PR13377 10872 324. https://gcc.gnu.org/PR16002 10873 325. https://gcc.gnu.org/PR17413 10874 326. https://gcc.gnu.org/PR17609 10875 327. https://gcc.gnu.org/PR17618 10876 328. https://gcc.gnu.org/PR18124 10877 329. https://gcc.gnu.org/PR18155 10878 330. https://gcc.gnu.org/PR18177 10879 331. https://gcc.gnu.org/PR18368 10880 332. https://gcc.gnu.org/PR18378 10881 333. https://gcc.gnu.org/PR18466 10882 334. https://gcc.gnu.org/PR18512 10883 335. https://gcc.gnu.org/PR18545 10884 336. https://gcc.gnu.org/PR18738 10885 337. https://gcc.gnu.org/PR18803 10886 338. https://gcc.gnu.org/PR19004 10887 339. https://gcc.gnu.org/PR19208 10888 340. https://gcc.gnu.org/PR19253 10889 341. https://gcc.gnu.org/PR19608 10890 342. https://gcc.gnu.org/PR19884 10891 343. https://gcc.gnu.org/PR20153 10892 344. https://gcc.gnu.org/PR20563 10893 345. https://gcc.gnu.org/PR20789 10894 346. https://gcc.gnu.org/PR21336 10895 347. https://gcc.gnu.org/PR21768 10896 348. https://gcc.gnu.org/PR21853 10897 349. https://gcc.gnu.org/PR21903 10898 350. https://gcc.gnu.org/PR21983 10899 351. https://gcc.gnu.org/PR21987 10900 352. https://gcc.gnu.org/PR22153 10901 353. https://gcc.gnu.org/PR22172 10902 354. https://gcc.gnu.org/PR21286 10903 355. https://gcc.gnu.org/PR22233 10904 356. https://gcc.gnu.org/PR22508 10905 357. https://gcc.gnu.org/PR22545 10906 358. https://gcc.gnu.org/PR23528 10907 359. https://gcc.gnu.org/PR23550 10908 360. https://gcc.gnu.org/PR23586 10909 361. https://gcc.gnu.org/PR23624 10910 362. https://gcc.gnu.org/PR23639 10911 363. https://gcc.gnu.org/PR23797 10912 364. https://gcc.gnu.org/PR23965 10913 365. https://gcc.gnu.org/PR24052 10914 366. https://gcc.gnu.org/PR24580 10915 367. https://gcc.gnu.org/PR24267 10916 368. https://gcc.gnu.org/PR17810 10917 369. https://gcc.gnu.org/PR17860 10918 370. https://gcc.gnu.org/PR21709 10919 371. https://gcc.gnu.org/PR21964 10920 372. https://gcc.gnu.org/PR22167 10921 373. https://gcc.gnu.org/PR22619 10922 374. https://gcc.gnu.org/PR23241 10923 375. https://gcc.gnu.org/PR23478 10924 376. https://gcc.gnu.org/PR24470 10925 377. https://gcc.gnu.org/PR24950 10926 378. https://gcc.gnu.org/PR14400 10927 379. https://gcc.gnu.org/PR14940 10928 380. https://gcc.gnu.org/PR20239 10929 381. https://gcc.gnu.org/PR15220 10930 382. https://gcc.gnu.org/PR19275 10931 383. https://gcc.gnu.org/PR21888 10932 384. https://gcc.gnu.org/PR15342 10933 385. https://gcc.gnu.org/PR23985 10934 386. https://gcc.gnu.org/PR16719 10935 387. https://gcc.gnu.org/PR21723 10936 388. https://gcc.gnu.org/PR21841 10937 389. https://gcc.gnu.org/PR23644 10938 390. https://gcc.gnu.org/PR24718 10939 391. https://gcc.gnu.org/PR18421 10940 392. https://gcc.gnu.org/PR20621 10941 393. https://gcc.gnu.org/PR18583 10942 394. https://gcc.gnu.org/PR20191 10943 395. https://gcc.gnu.org/PR22083 10944 396. https://gcc.gnu.org/PR23070 10945 397. https://gcc.gnu.org/PR23404 10946 398. https://gcc.gnu.org/PR23539 10947 399. https://gcc.gnu.org/PR24102 10948 400. https://gcc.gnu.org/PR24465 10949 401. https://gcc.gnu.org/PR19933 10950 402. https://gcc.gnu.org/PR21889 10951 403. https://gcc.gnu.org/PR19300 10952 404. https://gcc.gnu.org/PR20301 10953 405. https://gcc.gnu.org/PR20673 10954 406. https://gcc.gnu.org/PR18582 10955 407. https://gcc.gnu.org/PR19340 10956 408. https://gcc.gnu.org/PR21716 10957 409. https://gcc.gnu.org/PR24315 10958 410. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.6 10959 411. https://gcc.gnu.org/onlinedocs/ 10960 412. mailto:gcc-help (a] gcc.gnu.org 10961 413. mailto:gcc (a] gcc.gnu.org 10962 414. https://gcc.gnu.org/lists.html 10963 415. http://www.fsf.org/ 10964 416. https://gcc.gnu.org/about.html 10965 417. http://validator.w3.org/check/referer 10966 ====================================================================== 10967 http://gcc.gnu.org/gcc-3.3/index.html 10968 GCC 3.3 Release Series 10969 10970 May 03, 2005 10971 10972 The [1]GNU project and the GCC developers are pleased to announce the 10973 release of GCC 3.3.6. 10974 10975 This release is a bug-fix release, containing fixes for regressions in 10976 GCC 3.3.5 relative to previous releases of GCC. 10977 10978 This release is the last of the series 3.3.x. 10979 10980 The GCC 3.3 release series includes numerous [2]new features, 10981 improvements, bug fixes, and other changes, thanks to an [3]amazing 10982 group of volunteers. 10983 10984 This release series is no longer maintained. 10985 10986 Release History 10987 10988 GCC 3.3.6 10989 May 3, 2005 ([4]changes) 10990 10991 GCC 3.3.5 10992 September 30, 2004 ([5]changes) 10993 10994 GCC 3.3.4 10995 May 31, 2004 ([6]changes) 10996 10997 GCC 3.3.3 10998 February 14, 2004 ([7]changes) 10999 11000 GCC 3.3.2 11001 October 16, 2003 ([8]changes) 11002 11003 GCC 3.3.1 11004 August 8, 2003 ([9]changes) 11005 11006 GCC 3.3 11007 May 14, 2003 ([10]changes) 11008 11009 References and Acknowledgements 11010 11011 GCC used to stand for the GNU C Compiler, but since the compiler 11012 supports several other languages aside from C, it now stands for the 11013 GNU Compiler Collection. 11014 11015 A list of [11]successful builds is updated as new information becomes 11016 available. 11017 11018 The GCC developers would like to thank the numerous people that have 11019 contributed new features, improvements, bug fixes, and other changes as 11020 well as test results to GCC. This [12]amazing group of volunteers is 11021 what makes GCC successful. 11022 11023 For additional information about GCC please refer to the [13]GCC 11024 project web site or contact the [14]GCC development mailing list. 11025 11026 To obtain GCC please use [15]our mirror sites, or our CVS server. 11027 11028 11029 For questions related to the use of GCC, please consult these web 11030 pages and the [16]GCC manuals. If that fails, the 11031 [17]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 11032 web pages and the development of GCC are welcome on our developer 11033 list at [18]gcc (a] gcc.gnu.org. All of [19]our lists have public 11034 archives. 11035 11036 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and 11037 distribution of this entire article is permitted in any medium, 11038 provided this notice is preserved. 11039 11040 These pages are [21]maintained by the GCC team. Last modified 11041 2016-09-30[22]. 11042 11043 References 11044 11045 1. http://www.gnu.org/ 11046 2. http://gcc.gnu.org/gcc-3.3/changes.html 11047 3. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 11048 4. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6 11049 5. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.5 11050 6. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.4 11051 7. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.3 11052 8. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.2 11053 9. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.1 11054 10. http://gcc.gnu.org/gcc-3.3/changes.html 11055 11. http://gcc.gnu.org/gcc-3.3/buildstat.html 11056 12. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 11057 13. http://gcc.gnu.org/index.html 11058 14. mailto:gcc (a] gcc.gnu.org 11059 15. http://gcc.gnu.org/mirrors.html 11060 16. https://gcc.gnu.org/onlinedocs/ 11061 17. mailto:gcc-help (a] gcc.gnu.org 11062 18. mailto:gcc (a] gcc.gnu.org 11063 19. https://gcc.gnu.org/lists.html 11064 20. http://www.fsf.org/ 11065 21. https://gcc.gnu.org/about.html 11066 22. http://validator.w3.org/check/referer 11067 ====================================================================== 11068 http://gcc.gnu.org/gcc-3.3/changes.html 11069 GCC 3.3 Release Series 11070 Changes, New Features, and Fixes 11071 11072 The latest release in the 3.3 release series is [1]GCC 3.3.6. 11073 11074 Caveats 11075 11076 * The preprocessor no longer accepts multi-line string literals. They 11077 were deprecated in 3.0, 3.1, and 3.2. 11078 * The preprocessor no longer supports the -A- switch when appearing 11079 alone. -A- followed by an assertion is still supported. 11080 * Support for all the systems [2]obsoleted in GCC 3.1 has been 11081 removed from GCC 3.3. See below for a [3]list of systems which are 11082 obsoleted in this release. 11083 * Checking for null format arguments has been decoupled from the rest 11084 of the format checking mechanism. Programs which use the format 11085 attribute may regain this functionality by using the new [4]nonnull 11086 function attribute. Note that all functions for which GCC has a 11087 built-in format attribute, an appropriate built-in nonnull 11088 attribute is also applied. 11089 * The DWARF (version 1) debugging format has been deprecated and will 11090 be removed in a future version of GCC. Version 2 of the DWARF 11091 debugging format will continue to be supported for the foreseeable 11092 future. 11093 * The C and Objective-C compilers no longer accept the "Naming Types" 11094 extension (typedef foo = bar); it was already unavailable in C++. 11095 Code which uses it will need to be changed to use the "typeof" 11096 extension instead: typedef typeof(bar) foo. (We have removed this 11097 extension without a period of deprecation because it has caused the 11098 compiler to crash since version 3.0 and no one noticed until very 11099 recently. Thus we conclude it is not in widespread use.) 11100 * The -traditional C compiler option has been removed. It was 11101 deprecated in 3.1 and 3.2. (Traditional preprocessing remains 11102 available.) The <varargs.h> header, used for writing variadic 11103 functions in traditional C, still exists but will produce an error 11104 message if used. 11105 * GCC 3.3.1 automatically places zero-initialized variables in the 11106 .bss section on some operating systems. Versions of GNU Emacs up to 11107 (and including) 21.3 will not work correctly when using this 11108 optimization; you can use -fno-zero-initialized-in-bss to disable 11109 it. 11110 11111 General Optimizer Improvements 11112 11113 * A new scheme for accurately describing processor pipelines, the 11114 [5]DFA scheduler, has been added. 11115 * Pavel Nejedly, Charles University Prague, has contributed new file 11116 format used by the edge coverage profiler (-fprofile-arcs). 11117 The new format is robust and diagnoses common mistakes where 11118 profiles from different versions (or compilations) of the program 11119 are combined resulting in nonsensical profiles and slow code to 11120 produced with profile feedback. Additionally this format allows 11121 extra data to be gathered. Currently, overall statistics are 11122 produced helping optimizers to identify hot spots of a program 11123 globally replacing the old intra-procedural scheme and resulting in 11124 better code. Note that the gcov tool from older GCC versions will 11125 not be able to parse the profiles generated by GCC 3.3 and vice 11126 versa. 11127 * Jan Hubicka, SuSE Labs, has contributed a new superblock formation 11128 pass enabled using -ftracer. This pass simplifies the control flow 11129 of functions allowing other optimizations to do better job. 11130 He also contributed the function reordering pass 11131 (-freorder-functions) to optimize function placement using profile 11132 feedback. 11133 11134 New Languages and Language specific improvements 11135 11136 C/ObjC/C++ 11137 11138 * The preprocessor now accepts directives within macro arguments. It 11139 processes them just as if they had not been within macro arguments. 11140 * The separate ISO and traditional preprocessors have been completely 11141 removed. The front end handles either type of preprocessed output 11142 if necessary. 11143 * In C99 mode preprocessor arithmetic is done in the precision of the 11144 target's intmax_t, as required by that standard. 11145 * The preprocessor can now copy comments inside macros to the output 11146 file when the macro is expanded. This feature, enabled using the 11147 -CC option, is intended for use by applications which place 11148 metadata or directives inside comments, such as lint. 11149 * The method of constructing the list of directories to be searched 11150 for header files has been revised. If a directory named by a -I 11151 option is a standard system include directory, the option is 11152 ignored to ensure that the default search order for system 11153 directories and the special treatment of system header files are 11154 not defeated. 11155 * A few more [6]ISO C99 features now work correctly. 11156 * A new function attribute, nonnull, has been added which allows 11157 pointer arguments to functions to be specified as requiring a 11158 non-null value. The compiler currently uses this information to 11159 issue a warning when it detects a null value passed in such an 11160 argument slot. 11161 * A new type attribute, may_alias, has been added. Accesses to 11162 objects with types with this attribute are not subjected to 11163 type-based alias analysis, but are instead assumed to be able to 11164 alias any other type of objects, just like the char type. 11165 11166 C++ 11167 11168 * Type based alias analysis has been implemented for C++ aggregate 11169 types. 11170 11171 Objective-C 11172 11173 * Generate an error if Objective-C objects are passed by value in 11174 function and method calls. 11175 * When -Wselector is used, check the whole list of selectors at the 11176 end of compilation, and emit a warning if a @selector() is not 11177 known. 11178 * Define __NEXT_RUNTIME__ when compiling for the NeXT runtime. 11179 * No longer need to include objc/objc-class.h to compile self calls 11180 in class methods (NeXT runtime only). 11181 * New -Wundeclared-selector option. 11182 * Removed selector bloating which was causing object files to be 10% 11183 bigger on average (GNU runtime only). 11184 * Using at run time @protocol() objects has been fixed in certain 11185 situations (GNU runtime only). 11186 * Type checking has been fixed and improved in many situations 11187 involving protocols. 11188 11189 Java 11190 11191 * The java.sql and javax.sql packages now implement the JDBC 3.0 (JDK 11192 1.4) API. 11193 * The JDK 1.4 assert facility has been implemented. 11194 * The bytecode interpreter is now direct threaded and thus faster. 11195 11196 Fortran 11197 11198 * Fortran improvements are listed in [7]the Fortran documentation. 11199 11200 Ada 11201 11202 * Ada tasking now works with glibc 2.3.x threading libraries. 11203 11204 New Targets and Target Specific Improvements 11205 11206 * The following changes have been made to the HP-PA port: 11207 + The port now defaults to scheduling for the PA8000 series of 11208 processors. 11209 + Scheduling support for the PA7300 processor has been added. 11210 + The 32-bit port now supports weak symbols under HP-UX 11. 11211 + The handling of initializers and finalizers has been improved 11212 under HP-UX 11. The 64-bit port no longer uses collect2. 11213 + Dwarf2 EH support has been added to the 32-bit GNU/Linux port. 11214 + ABI fixes to correct the passing of small structures by value. 11215 * The SPARC, HP-PA, SH4, and x86/pentium ports have been converted to 11216 use the DFA processor pipeline description. 11217 * The following NetBSD configurations for the SuperH processor family 11218 have been added: 11219 + SH3, big-endian, sh-*-netbsdelf* 11220 + SH3, little-endian, shle-*-netbsdelf* 11221 + SH5, SHmedia, big-endian, 32-bit default, sh5-*-netbsd* 11222 + SH5, SHmedia, little-endian, 32-bit default, sh5le-*-netbsd* 11223 + SH5, SHmedia, big-endian, 64-bit default, sh64-*-netbsd* 11224 + SH5, SHmedia, little-endian, 64-bit default, sh64le-*-netbsd* 11225 * The following changes have been made to the IA-32/x86-64 port: 11226 + SSE2 and 3dNOW! intrinsics are now supported. 11227 + Support for thread local storage has been added to the IA-32 11228 and x86-64 ports. 11229 + The x86-64 port has been significantly improved. 11230 * The following changes have been made to the MIPS port: 11231 + All configurations now accept the -mabi switch. Note that you 11232 will need appropriate multilibs for this option to work 11233 properly. 11234 + ELF configurations will always pass an ABI flag to the 11235 assembler, except when the MIPS EABI is selected. 11236 + -mabi=64 no longer selects MIPS IV code. 11237 + The -mcpu option, which was deprecated in 3.1 and 3.2, has 11238 been removed from this release. 11239 + -march now changes the core ISA level. In previous releases, 11240 it would change the use of processor-specific extensions, but 11241 would leave the core ISA unchanged. For example, mips64-elf 11242 -march=r8000 will now generate MIPS IV code. 11243 + Under most configurations, -mipsN now acts as a synonym for 11244 -march. 11245 + There are some new preprocessor macros to describe the -march 11246 and -mtune settings. See the documentation of those options 11247 for details. 11248 + Support for the NEC VR-Series processors has been added. This 11249 includes the 54xx, 5500, and 41xx series. 11250 + Support for the Sandcraft sr71k processor has been added. 11251 * The following changes have been made to the S/390 port: 11252 + Support to build the Java runtime libraries has been added. 11253 Java is now enabled by default on s390-*-linux* and 11254 s390x-*-linux* targets. 11255 + Multilib support for the s390x-*-linux* target has been added; 11256 this allows to build 31-bit binaries using the -m31 option. 11257 + Support for thread local storage has been added. 11258 + Inline assembler code may now use the 'Q' constraint to 11259 specify memory operands without index register. 11260 + Various platform-specific performance improvements have been 11261 implemented; in particular, the compiler now uses the BRANCH 11262 ON COUNT family of instructions and makes more frequent use of 11263 the TEST UNDER MASK family of instructions. 11264 * The following changes have been made to the PowerPC port: 11265 + Support for IBM Power4 processor added. 11266 + Support for Motorola e500 SPE added. 11267 + Support for AIX 5.2 added. 11268 + Function and Data sections now supported on AIX. 11269 + Sibcall optimizations added. 11270 * The support for H8 Tiny is added to the H8/300 port with -mn. 11271 11272 Obsolete Systems 11273 11274 Support for a number of older systems has been declared obsolete in GCC 11275 3.3. Unless there is activity to revive them, the next release of GCC 11276 will have their sources permanently removed. 11277 11278 All configurations of the following processor architectures have been 11279 declared obsolete: 11280 * Matsushita MN10200, mn10200-*-* 11281 * Motorola 88000, m88k-*-* 11282 * IBM ROMP, romp-*-* 11283 11284 Also, some individual systems have been obsoleted: 11285 * Alpha 11286 + Interix, alpha*-*-interix* 11287 + Linux libc1, alpha*-*-linux*libc1* 11288 + Linux ECOFF, alpha*-*-linux*ecoff* 11289 * ARM 11290 + Generic a.out, arm*-*-aout* 11291 + Conix, arm*-*-conix* 11292 + "Old ABI," arm*-*-oabi 11293 + StrongARM/COFF, strongarm-*-coff* 11294 * HPPA (PA-RISC) 11295 + Generic OSF, hppa1.0-*-osf* 11296 + Generic BSD, hppa1.0-*-bsd* 11297 + HP/UX versions 7, 8, and 9, hppa1.[01]-*-hpux[789]* 11298 + HiUX, hppa*-*-hiux* 11299 + Mach Lites, hppa*-*-lites* 11300 * Intel 386 family 11301 + Windows NT 3.x, i?86-*-win32 11302 * MC68000 family 11303 + HP systems, m68000-hp-bsd* and m68k-hp-bsd* 11304 + Sun systems, m68000-sun-sunos*, m68k-sun-sunos*, and 11305 m68k-sun-mach* 11306 + AT&T systems, m68000-att-sysv* 11307 + Atari systems, m68k-atari-sysv* 11308 + Motorola systems, m68k-motorola-sysv* 11309 + NCR systems, m68k-ncr-sysv* 11310 + Plexus systems, m68k-plexus-sysv* 11311 + Commodore systems, m68k-cbm-sysv* 11312 + Citicorp TTI, m68k-tti-* 11313 + Unos, m68k-crds-unos* 11314 + Concurrent RTU, m68k-ccur-rtu* 11315 + Linux a.out, m68k-*-linux*aout* 11316 + Linux libc1, m68k-*-linux*libc1* 11317 + pSOS, m68k-*-psos* 11318 * MIPS 11319 + Generic ECOFF, mips*-*-ecoff* 11320 + SINIX, mips-sni-sysv4 11321 + Orion RTEMS, mips64orion-*-rtems* 11322 * National Semiconductor 32000 11323 + OpenBSD, ns32k-*-openbsd* 11324 * POWER (aka RS/6000) and PowerPC 11325 + AIX versions 1, 2, and 3, rs6000-ibm-aix[123]* 11326 + Bull BOSX, rs6000-bull-bosx 11327 + Generic Mach, rs6000-*-mach* 11328 + Generic SysV, powerpc*-*-sysv* 11329 + Linux libc1, powerpc*-*-linux*libc1* 11330 * Sun SPARC 11331 + Generic a.out, sparc-*-aout*, sparclet-*-aout*, 11332 sparclite-*-aout*, and sparc86x-*-aout* 11333 + NetBSD a.out, sparc-*-netbsd*aout* 11334 + Generic BSD, sparc-*-bsd* 11335 + ChorusOS, sparc-*-chorusos* 11336 + Linux a.out, sparc-*-linux*aout* 11337 + Linux libc1, sparc-*-linux*libc1* 11338 + LynxOS, sparc-*-lynxos* 11339 + Solaris on HAL hardware, sparc-hal-solaris2* 11340 + SunOS versions 3 and 4, sparc-*-sunos[34]* 11341 * NEC V850 11342 + RTEMS, v850-*-rtems* 11343 * VAX 11344 + VMS, vax-*-vms* 11345 11346 Documentation improvements 11347 11348 Other significant improvements 11349 11350 * Almost all front-end dependencies in the compiler have been 11351 separated out into a set of language hooks. This should make adding 11352 a new front end clearer and easier. 11353 * One effect of removing the separate preprocessor is a small 11354 increase in the robustness of the compiler in general, and the 11355 maintainability of target descriptions. Previously target-specific 11356 built-in macros and others, such as __FAST_MATH__, had to be 11357 handled with so-called specs that were hard to maintain. Often they 11358 would fail to behave properly when conflicting options were 11359 supplied on the command line, and define macros in the user's 11360 namespace even when strict ISO compliance was requested. 11361 Integrating the preprocessor has cleanly solved these issues. 11362 * The Makefile suite now supports redirection of make install by 11363 means of the variable DESTDIR. 11364 __________________________________________________________________ 11365 11366 GCC 3.3 11367 11368 Detailed release notes for the GCC 3.3 release follow. 11369 11370 Bug Fixes 11371 11372 bootstrap failures 11373 11374 * [8]10140 cross compiler build failures: missing __mempcpy (DUP: 11375 [9]10198,[10]10338) 11376 11377 Internal compiler errors (multi-platform) 11378 11379 * [11]3581 large string causes segmentation fault in cc1 11380 * [12]4382 __builtin_{set,long}jmp with -O3 can crash the compiler 11381 * [13]5533 (c++) ICE when processing std::accumulate(begin, end, 11382 init, invalid_op) 11383 * [14]6387 -fpic -gdwarf-2 -g1 combination gives ICE in dwarf2out 11384 * [15]6412 (c++) ICE in retrieve_specialization 11385 * [16]6620 (c++) partial template specialization causes an ICE 11386 (segmentation fault) 11387 * [17]6663 (c++) ICE with attribute aligned 11388 * [18]7068 ICE with incomplete types 11389 * [19]7083 (c++) ICE using -gstabs with dodgy class derivation 11390 * [20]7647 (c++) ICE when data member has the name of the enclosing 11391 class 11392 * [21]7675 ICE in fixup_var_refs_1 11393 * [22]7718 'complex' template instantiation causes ICE 11394 * [23]8116 (c++) ICE in member template function 11395 * [24]8358 (ada) Ada compiler accesses freed memory, crashes 11396 * [25]8511 (c++) ICE: (hopefully) reproducible cc1plus segmentation 11397 fault 11398 * [26]8564 (c++) ICE in find_function_data, in function.c 11399 * [27]8660 (c++) template overloading ICE in tsubst_expr, in cp/pt.c 11400 * [28]8766 (c++) ICE after failed initialization of static template 11401 variable 11402 * [29]8803 ICE in instantiate_virtual_regs_1, in function.c 11403 * [30]8846 (c++) ICE after diagnostic if fr_FR@euro locale is set 11404 * [31]8906 (c++) ICE (Segmentation fault) when parsing nested-class 11405 definition 11406 * [32]9216 (c++) ICE on missing template parameter 11407 * [33]9261 (c++) ICE in arg_assoc, in cp/decl2.c 11408 * [34]9263 (fortran) ICE caused by invalid PARAMETER in implied DO 11409 loop 11410 * [35]9429 (c++) ICE in template instantiation with a pointered new 11411 operator 11412 * [36]9516 Internal error when using a big array 11413 * [37]9600 (c++) ICE with typedefs in template class 11414 * [38]9629 (c++) virtual inheritance segfault 11415 * [39]9672 (c++) ICE: Error reporting routines re-entered 11416 * [40]9749 (c++) ICE in write_expression on invalid function 11417 prototype 11418 * [41]9794 (fortran) ICE: floating point exception during constant 11419 folding 11420 * [42]9829 (c++) Missing colon in nested namespace usage causes ICE 11421 * [43]9916 (c++) ICE with noreturn function in ?: statement 11422 * [44]9936 ICE with local function and variable-length 2d array 11423 * [45]10262 (c++) cc1plus crashes with large generated code 11424 * [46]10278 (c++) ICE in parser for invalid code 11425 * [47]10446 (c++) ICE on definition of nonexistent member function of 11426 nested class in a class template 11427 * [48]10451 (c++) ICE in grokdeclarator on spurious mutable 11428 declaration 11429 * [49]10506 (c++) ICE in build_new at cp/init.c with 11430 -fkeep-inline-functions and multiple inheritance 11431 * [50]10549 (c++) ICE in store_bit_field on bitfields that exceed the 11432 precision of the declared type 11433 11434 Optimization bugs 11435 11436 * [51]2001 Inordinately long compile times in reload CSE regs 11437 * [52]2391 Exponential compilation time explosion in combine 11438 * [53]2960 Duplicate loop conditions even with -Os 11439 * [54]4046 redundant conditional branch 11440 * [55]6405 Loop-unrolling related performance regressions 11441 * [56]6798 very long compile time with large case-statement 11442 * [57]6871 const objects shouldn't be moved to .bss 11443 * [58]6909 problem w/ -Os on modified loop-2c.c test case 11444 * [59]7189 gcc -O2 -Wall does not print ``control reaches end of 11445 non-void function'' warning 11446 * [60]7642 optimization problem with signbit() 11447 * [61]8634 incorrect code for inlining of memcpy under -O2 11448 * [62]8750 Cygwin prolog generation erroneously emitting __alloca as 11449 regular function call 11450 11451 C front end 11452 11453 * [63]2161 long if-else cascade overflows parser stack 11454 * [64]4319 short accepted on typedef'd char 11455 * [65]8602 incorrect line numbers in warning messages when using 11456 inline functions 11457 * [66]9177 -fdump-translation-unit: C front end deletes function_decl 11458 AST nodes and breaks debugging dumps 11459 * [67]9853 miscompilation of non-constant structure initializer 11460 11461 c++ compiler and library 11462 11463 * [68]45 legal template specialization code is rejected (DUP: 11464 [69]3784) 11465 * [70]764 lookup failure: friend operator and dereferencing a pointer 11466 and templates (DUP: [71]5116) 11467 * [72]2862 gcc accepts invalid explicit instantiation syntax (DUP: 11468 2863) 11469 * [73]3663 G++ doesn't check access control during template 11470 instantiation 11471 * [74]3797 gcc fails to emit explicit specialization of a template 11472 member 11473 * [75]3948 Two destructors are called when no copy destructor is 11474 defined (ABI change) 11475 * [76]4137 Conversion operator within template is not accepted 11476 * [77]4361 bogus ambiguity taking the address of a member template 11477 * [78]4802 g++ accepts illegal template code (access to private 11478 member; DUP: [79]5837) 11479 * [80]4803 inline function is used but never defined, and g++ does 11480 not object 11481 * [81]5094 Partial specialization cannot be friend? 11482 * [82]5730 complex<double>::norm() -- huge slowdown from egcs-2.91.66 11483 * [83]6713 Regression wrt 3.0.4: g++ -O2 leads to seg fault at run 11484 time 11485 * [84]7015 certain __asm__ constructs rejected 11486 * [85]7086 compile time regression (quadratic behavior in 11487 fixup_var_refs) 11488 * [86]7099 G++ doesn't set the noreturn attribute on std::exit and 11489 std::abort 11490 * [87]7247 copy constructor missing when inlining enabled (invalid 11491 optimization?) 11492 * [88]7441 string array initialization compilation time regression 11493 from seconds to minutes 11494 * [89]7768 __PRETTY_FUNCTION__ for template destructor is wrong 11495 * [90]7804 bad printing of floating point constant in warning message 11496 * [91]8099 Friend classes and template specializations 11497 * [92]8117 member function pointers and multiple inheritance 11498 * [93]8205 using declaration and multiple inheritance 11499 * [94]8645 unnecessary non-zero checks in stl_tree.h 11500 * [95]8724 explicit destructor call for incomplete class allowed 11501 * [96]8805 compile time regression with many member variables 11502 * [97]8691 -O3 and -fno-implicit-templates are incompatible 11503 * [98]8700 unhelpful error message for binding temp to reference 11504 * [99]8724 explicit destructor call for incomplete class allowed 11505 * [100]8949 numeric_limits<>::denorm_min() and is_iec559 problems 11506 * [101]9016 Failure to consistently constant fold "constant" C++ 11507 objects 11508 * [102]9053 g++ confused about ambiguity of overloaded function 11509 templates 11510 * [103]9152 undefined virtual thunks 11511 * [104]9182 basic_filebuf<> does not report errors in codecvt<>::out 11512 * [105]9297 data corruption due to codegen bug (when copying.) 11513 * [106]9318 i/ostream::operator>>/<<(streambuf*) broken 11514 * [107]9320 Incorrect usage of traits_type::int_type in stdio_filebuf 11515 * [108]9400 bogus -Wshadow warning: shadowed declaration of this in 11516 local classes 11517 * [109]9424 i/ostream::operator>>/<<(streambuf*) drops characters 11518 * [110]9425 filebuf::pbackfail broken (DUP: [111]9439) 11519 * [112]9474 GCC freezes in compiling a weird code mixing <iostream> 11520 and <iostream.h> 11521 * [113]9548 Incorrect results from setf(ios::fixed) and precision(-1) 11522 [114][DR 231] 11523 * [115]9555 ostream inserters fail to set badbit on exception 11524 * [116]9561 ostream inserters rethrow exception of wrong type 11525 * [117]9563 ostream::sentry returns true after a failed preparation 11526 * [118]9582 one-definition rule violation in std::allocator 11527 * [119]9622 __PRETTY_FUNCTION__ incorrect in template destructors 11528 * [120]9683 bug in initialization chains for static const variables 11529 from template classes 11530 * [121]9791 -Woverloaded-virtual reports hiding of destructor 11531 * [122]9817 collate::compare doesn't handle nul characters 11532 * [123]9825 filebuf::sputbackc breaks sbumpc 11533 * [124]9826 operator>>(basic_istream, basic_string) fails to compile 11534 with custom traits 11535 * [125]9924 Multiple using statements for builtin functions not 11536 allowed 11537 * [126]9946 destructor is not called for temporary object 11538 * [127]9964 filebuf::close() sometimes fails to close file 11539 * [128]9988 filebuf::overflow writes EOF to file 11540 * [129]10033 optimization breaks polymorphic references w/ typeid 11541 operator 11542 * [130]10097 filebuf::underflow drops characters 11543 * [131]10132 filebuf destructor can throw exceptions 11544 * [132]10180 gcc fails to warn about non-inlined function 11545 * [133]10199 method parametrized by template does not work everywhere 11546 * [134]10300 use of array-new (nothrow) in segfaults on NULL return 11547 * [135]10427 Stack corruption with variable-length automatic arrays 11548 and virtual destructors 11549 * [136]10503 Compilation never stops in fixed_type_or_null 11550 11551 Objective-C 11552 11553 * [137]5956 selectors aren't matched properly when added to the 11554 selector table 11555 11556 Fortran compiler and library 11557 11558 * [138]1832 list directed i/o overflow hangs, -fbounds-check doesn't 11559 detect 11560 * [139]3924 g77 generates code that is rejected by GAS if COFF debug 11561 info requested 11562 * [140]5634 doc: explain that configure --prefix=~/... does not work 11563 * [141]6367 multiple repeat counts confuse namelist read into array 11564 * [142]6491 Logical operations error on logicals when using 11565 -fugly-logint 11566 * [143]6742 Generation of C++ Prototype for FORTRAN and extern "C" 11567 * [144]7113 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os 11568 on irix6.5 11569 * [145]7236 OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should 11570 assume a direct access file 11571 * [146]7278 g77 "bug"; the executable misbehaves (with -O2 11572 -fno-automatic) 11573 * [147]7384 DATE_AND_TIME milliseconds field inactive on Windows 11574 * [148]7388 Incorrect output with 0-based array of characters 11575 * [149]8587 Double complex zero ** double precision number -> NaN 11576 instead of zero 11577 * [150]9038 -ffixed-line-length-none -x f77-cpp-input gives: Warning: 11578 unknown register name line-length-none 11579 * [151]10197 Direct access files not unformatted by default 11580 11581 Java compiler and library 11582 11583 * [152]6005 gcj fails to build rhug on alpha 11584 * [153]6389 System.getProperty("") should always throw an 11585 IllegalArgumentException 11586 * [154]6576 java.util.ResourceBundle.getResource ignores locale 11587 * [155]6652 new java.io.File("").getCanonicalFile() throws exception 11588 * [156]7060 getMethod() doesn't search super interface 11589 * [157]7073 bytecode interpreter gives wrong answer for interface 11590 getSuperclass() 11591 * [158]7180 possible bug in 11592 javax.naming.spi.NamingManager.getPlusPath() 11593 * [159]7416 java.security startup refs "GNU libgcj.security" 11594 * [160]7570 Runtime.exec with null envp: child doesn't inherit parent 11595 env (DUP: [161]7578) 11596 * [162]7611 Internal error while compiling libjava with -O 11597 * [163]7709 NullPointerException in _Jv_ResolvePoolEntry 11598 * [164]7766 ZipInputStream.available returns 0 immediately after 11599 construction 11600 * [165]7785 Calendar.getTimeInMillis/setTimeInMillis should be public 11601 * [166]7786 TimeZone.getDSTSavings() from JDK1.4 not implemented 11602 * [167]8142 '$' in class names vs. dlopen 'dynamic string tokens' 11603 * [168]8234 ZipInputStream chokes when InputStream.read() returns 11604 small chunks 11605 * [169]8415 reflection bug: exception info for Method 11606 * [170]8481 java.Random.nextInt(int) may return negative 11607 * [171]8593 Error reading GZIPped files with BufferedReader 11608 * [172]8759 java.beans.Introspector has no flushCaches() or 11609 flushFromCaches() methods 11610 * [173]8997 spin() calls Thread.sleep 11611 * [174]9253 on win32, java.io.File.listFiles("C:\\") returns pwd 11612 instead of the root content of C: 11613 * [175]9254 java::lang::Object::wait(), threads-win32.cc returns 11614 wrong return codes 11615 * [176]9271 Severe bias in java.security.SecureRandom 11616 11617 Ada compiler and library 11618 11619 * [177]6767 make gnatlib-shared fails on -laddr2line 11620 * [178]9911 gnatmake fails to link when GCC configured with 11621 --with-sjlj-exceptions=yes 11622 * [179]10020 Can't bootstrap gcc on AIX with Ada enabled 11623 * [180]10546 Ada tasking not working on Red Hat 9 11624 11625 preprocessor 11626 11627 * [181]7029 preprocessor should ignore #warning with -M 11628 11629 ARM-specific 11630 11631 * [182]2903 [arm] Optimization bug with long long arithmetic 11632 * [183]7873 arm-linux-gcc fails when assigning address to a bit field 11633 11634 FreeBSD-specific 11635 11636 * [184]7680 float functions undefined in math.h/cmath with #define 11637 _XOPEN_SOURCE 11638 11639 HP-UX or HP-PA-specific 11640 11641 * [185]8705 [HP-PA] ICE in emit_move_insn_1, in expr.c 11642 * [186]9986 [HP-UX] Incorrect transformation of fputs_unlocked to 11643 fputc_unlocked 11644 * [187]10056 [HP-PA] ICE at -O2 when building c++ code from doxygen 11645 11646 m68hc11-specific 11647 11648 * [188]6744 Bad assembler code generated: reference to pseudo 11649 register z 11650 * [189]7361 Internal compiler error in reload_cse_simplify_operands, 11651 in reload1.c 11652 11653 MIPS-specific 11654 11655 * [190]9496 [mips-linux] bug in optimizer? 11656 11657 PowerPC-specific 11658 11659 * [191]7067 -Os with -mcpu=powerpc optimizes for speed (?) instead of 11660 space 11661 * [192]8480 reload ICEs for LAPACK code on powerpc64-linux 11662 * [193]8784 [AIX] Internal compiler error in simplify_gen_subreg 11663 * [194]10315 [powerpc] ICE: in extract_insn, in recog.c 11664 11665 SPARC-specific 11666 11667 * [195]10267 (documentation) Wrong build instructions for 11668 *-*-solaris2* 11669 11670 x86-specific (Intel/AMD) 11671 11672 * [196]7916 ICE in instantiate_virtual_register_1 11673 * [197]7926 (c++) i486 instructions in header files make c++ programs 11674 crash on i386 11675 * [198]8555 ICE in gen_split_1231 11676 * [199]8994 ICE with -O -march=pentium4 11677 * [200]9426 ICE with -fssa -funroll-loops -fprofile-arcs 11678 * [201]9806 ICE in inline assembly with -fPIC flag 11679 * [202]10077 gcc -msse2 generates movd to move dwords between xmm 11680 regs 11681 * [203]10233 64-bit comparison only comparing bottom 32-bits 11682 * [204]10286 type-punning doesn't work with __m64 and -O 11683 * [205]10308 [x86] ICE with -O -fgcse or -O2 11684 __________________________________________________________________ 11685 11686 GCC 3.3.1 11687 11688 Bug Fixes 11689 11690 This section lists the problem reports (PRs) from GCC's bug tracking 11691 system that are known to be fixed in the 3.3.1 release. This list might 11692 not be complete (that is, it is possible that some PRs that have been 11693 fixed are not listed here). 11694 11695 Bootstrap failures 11696 11697 * [206]11272 [Solaris] make bootstrap fails while building libstdc++ 11698 11699 Internal compiler errors (multi-platform) 11700 11701 * [207]5754 ICE on invalid nested template class 11702 * [208]6597 ICE in set_mem_alias_set compiling Qt with -O2 on ia64 11703 and --enable-checking 11704 * [209]6949 (c++) ICE in tsubst_decl, in cp/pt.c 11705 * [210]7053 (c++) ICE when declaring a function already defined as a 11706 friend method of a template class 11707 * [211]8164 (c++) ICE when using different const expressions as 11708 template parameter 11709 * [212]8384 (c++) ICE in is_base_type, in dwarf2out.c 11710 * [213]9559 (c++) ICE with invalid initialization of a static const 11711 * [214]9649 (c++) ICE in finish_member_declaration, in cp/semantics.c 11712 when redeclaring a static member variable 11713 * [215]9864 (fortran) ICE in add_abstract_origin_attribute, in 11714 dwarfout.c with -g -O -finline-functions 11715 * [216]10432 (c++) ICE in poplevel, in cp/decl.c 11716 * [217]10475 ICE in subreg_highpart_offset for code with long long 11717 * [218]10635 (c++) ICE when dereferencing an incomplete type casted 11718 from a void pointer 11719 * [219]10661 (c++) ICE in instantiate_decl, in cp/pt.c while 11720 instantiating static member variables 11721 * [220]10700 ICE in copy_to_mode_reg on 64-bit targets 11722 * [221]10712 (c++) ICE in constructor_name_full, in cp/decl2.c 11723 * [222]10796 (c++) ICE when defining an enum with two values: -1 and 11724 MAX_INT_64BIT 11725 * [223]10890 ICE in merge_assigned_reloads building Linux 2.4.2x 11726 sched.c 11727 * [224]10939 (c++) ICE with template code 11728 * [225]10956 (c++) ICE when specializing a template member function 11729 of a template class, in tsubst, in cp/pt.c 11730 * [226]11041 (c++) ICE: const myclass &x = *x; (when operator*() 11731 defined) 11732 * [227]11059 (c++) ICE with empty union 11733 * [228]11083 (c++) ICE in commit_one_edge_insertion, in cfgrtl.c with 11734 -O2 -fnon-call-exceptions 11735 * [229]11105 (c++) ICE in mangle_conv_op_name_for_type 11736 * [230]11149 (c++) ICE on error when instantiation with call function 11737 of a base type 11738 * [231]11228 (c++) ICE on new-expression using array operator new and 11739 default-initialization 11740 * [232]11282 (c++) Infinite memory usage after syntax error 11741 * [233]11301 (fortran) ICE with -fno-globals 11742 * [234]11308 (c++) ICE when using an enum type name as if it were a 11743 class or namespace 11744 * [235]11473 (c++) ICE with -gstabs when empty struct inherits from 11745 an empty struct 11746 * [236]11503 (c++) ICE when instantiating template with ADDR_EXPR 11747 * [237]11513 (c++) ICE in push_template_decl_real, in cp/pt.c: 11748 template member functions 11749 11750 Optimization bugs 11751 11752 * [238]11198 -O2 -frename-registers generates wrong code (aliasing 11753 problem) 11754 * [239]11304 Wrong code production with -fomit-frame-pointer 11755 * [240]11381 volatile memory access optimized away 11756 * [241]11536 [strength-reduce] -O2 optimization produces wrong code 11757 * [242]11557 constant folding bug generates wrong code 11758 11759 C front end 11760 11761 * [243]5897 No warning for statement after return 11762 * [244]11279 DWARF-2 output mishandles large enums 11763 11764 Preprocessor bugs 11765 11766 * [245]11022 no warning for non-compatible macro redefinition 11767 11768 C++ compiler and library 11769 11770 * [246]2330 static_cast<>() to a private base is allowed 11771 * [247]5388 Incorrect message "operands to ?: have different types" 11772 * [248]5390 Libiberty fails to demangle multi-digit template 11773 parameters 11774 * [249]7877 Incorrect parameter passing to specializations of member 11775 function templates 11776 * [250]9393 Anonymous namespaces and compiling the same file twice 11777 * [251]10032 -pedantic converts some errors to warnings 11778 * [252]10468 const typeof(x) is non-const, but only in templates 11779 * [253]10527 confused error message with "new int()" parameter 11780 initializer 11781 * [254]10679 parameter MIN_INLINE_INSNS is not honored 11782 * [255]10682 gcc chokes on a typedef for an enum inside a class 11783 template 11784 * [256]10689 pow(std::complex(0),1/3) returns (nan, nan) instead of 11785 0. 11786 * [257]10845 template member function (with nested template as 11787 parameter) cannot be called anymore if another unrelated template 11788 member function is defined 11789 * [258]10849 Cannot define an out-of-class specialization of a 11790 private nested template class 11791 * [259]10888 Suppress -Winline warnings for system headers 11792 * [260]10929 -Winline warns about functions for which no definition 11793 is visible 11794 * [261]10931 valid conversion static_cast<const unsigned 11795 int&>(lvalue-of-type-int) is rejected 11796 * [262]10940 Bad code with explicit specialization 11797 * [263]10968 If member function implicitly instantiated, explicit 11798 instantiation of class fails to instantiate it 11799 * [264]10990 Cannot convert with dynamic_cast<> to a private base 11800 class from within a member function 11801 * [265]11039 Bad interaction between implicit typename deprecation 11802 and friendship 11803 * [266]11062 (libstdc++) avoid __attribute__ ((unused)); say 11804 "__unused__" instead 11805 * [267]11095 C++ iostream manipulator causes segfault when called 11806 with negative argument 11807 * [268]11098 g++ doesn't emit complete debugging information for 11808 local variables in destructors 11809 * [269]11137 GNU/Linux shared library constructors not called unless 11810 there's one global object 11811 * [270]11154 spurious ambiguity report for template class 11812 specialization 11813 * [271]11329 Compiler cannot find user defined implicit typecast 11814 * [272]11332 Spurious error with casts in ?: expression 11815 * [273]11431 static_cast behavior with subclasses when default 11816 constructor available 11817 * [274]11528 money_get facet does not accept "$.00" as valid 11818 * [275]11546 Type lookup problems in out-of-line definition of a 11819 class doubly nested from a template class 11820 * [276]11567 C++ code containing templated member function with same 11821 name as pure virtual member function results in linking failure 11822 * [277]11645 Failure to deal with using and private inheritance 11823 11824 Java compiler and library 11825 11826 * [278]5179 Qualified static field access doesn't initialize its 11827 class 11828 * [279]8204 gcj -O2 to native reorders certain instructions 11829 improperly 11830 * [280]10838 java.io.ObjectInputStream syntax error 11831 * [281]10886 The RMI registry that comes with GCJ does not work 11832 correctly 11833 * [282]11349 JNDI URL context factories not located correctly 11834 11835 x86-specific (Intel/AMD) 11836 11837 * [283]4823 ICE on inline assembly code 11838 * [284]8878 miscompilation with -O and SSE 11839 * [285]9815 (c++ library) atomicity.h - fails to compile with -O3 11840 -masm=intel 11841 * [286]10402 (inline assembly) [x86] ICE in merge_assigned_reloads, 11842 in reload1.c 11843 * [287]10504 ICE with SSE2 code and -O3 -mcpu=pentium4 -msse2 11844 * [288]10673 ICE for x86-64 on freebsd libc vfprintf.c source 11845 * [289]11044 [x86] out of range loop instructions for FP code on K6 11846 * [290]11089 ICE: instantiate_virtual_regs_lossage while using SSE 11847 built-ins 11848 * [291]11420 [x86_64] gcc generates invalid asm code when "-O -fPIC" 11849 is used 11850 11851 SPARC- or Solaris- specific 11852 11853 * [292]9362 solaris 'as' dies when fed .s and "-gstabs" 11854 * [293]10142 [SPARC64] gcc produces wrong code when passing 11855 structures by value 11856 * [294]10663 New configure check aborts with Sun tools. 11857 * [295]10835 combinatorial explosion in scheduler on HyperSPARC 11858 * [296]10876 ICE in calculate_giv_inc when building KDE 11859 * [297]10955 wrong code at -O3 for structure argument in context of 11860 structure return 11861 * [298]11018 -mcpu=ultrasparc busts tar-1.13.25 11862 * [299]11556 [sparc64] ICE in gen_reg_rtx() while compiling 2.6.x 11863 Linux kernel 11864 11865 ia64 specific 11866 11867 * [300]10907 gcc violates the ia64 ABI (GP must be preserved) 11868 * [301]11320 scheduler bug (in machine depended reorganization pass) 11869 * [302]11599 bug with conditional and __builtin_prefetch 11870 11871 PowerPC specific 11872 11873 * [303]9745 [powerpc] gcc mis-compiles libmcrypt (alias problem 11874 during loop) 11875 * [304]10871 error in rs6000_stack_info save_size computation 11876 * [305]11440 gcc mis-compiles c++ code (libkhtml) with -O2, -fno-gcse 11877 cures it 11878 11879 m68k-specific 11880 11881 * [306]7594 [m68k] ICE on legal code associated with simplify-rtx 11882 * [307]10557 [m68k] ICE in subreg_offset_representable_p 11883 * [308]11054 [m68k] ICE in reg_overlap_mentioned_p 11884 11885 ARM-specific 11886 11887 * [309]10834 [arm] GCC 3.3 still generates incorrect instructions for 11888 functions with __attribute__ ((interrupt ("IRQ"))) 11889 * [310]10842 [arm] Clobbered link register is copied to pc under 11890 certain circumstances 11891 * [311]11052 [arm] noce_process_if_block() can lose REG_INC notes 11892 * [312]11183 [arm] ICE in change_address_1 (3.3) / subreg_hard_regno 11893 (3.4) 11894 11895 MIPS-specific 11896 11897 * [313]11084 ICE in propagate_one_insn, in flow.c 11898 11899 SH-specific 11900 11901 * [314]10331 can't compile c++ part of gcc cross compiler for sh-elf 11902 * [315]10413 [SH] ICE in reload_cse_simplify_operands, in reload1.c 11903 * [316]11096 i686-linux to sh-linux cross compiler fails to compile 11904 C++ files 11905 11906 GNU/Linux (or Hurd?) specific 11907 11908 * [317]2873 Bogus fixinclude of stdio.h from glibc 2.2.3 11909 11910 UnixWare specific 11911 11912 * [318]3163 configure bug: gcc/aclocal.m4 mmap test fails on UnixWare 11913 7.1.1 11914 11915 Cygwin (or mingw) specific 11916 11917 * [319]5287 ICE with dllimport attribute 11918 * [320]10148 [MingW/CygWin] Compiler dumps core 11919 11920 DJGPP specific 11921 11922 * [321]8787 GCC fails to emit .intel_syntax when invoked with 11923 -masm=intel on DJGPP 11924 11925 Darwin (and MacOS X) specific 11926 11927 * [322]10900 trampolines crash 11928 11929 Documentation 11930 11931 * [323]1607 (c++) Format attributes on methods undocumented 11932 * [324]4252 Invalid option `-fdump-translation-unit' 11933 * [325]4490 Clarify restrictions on -m96bit-long-double, 11934 -m128bit-long-double 11935 * [326]10355 document an issue with regparm attribute on some systems 11936 (e.g. Solaris) 11937 * [327]10726 (fortran) Documentation for function "IDate Intrinsic 11938 (Unix)" is wrong 11939 * [328]10805 document bug in old version of Sun assembler 11940 * [329]10815 warn against GNU binutils on AIX 11941 * [330]10877 document need for newer binutils on i?86-*-linux-gnu 11942 * [331]11280 Manual incorrect with respect to -freorder-blocks 11943 * [332]11466 Document -mlittle-endian and its restrictions for the 11944 sparc64 port 11945 11946 Testsuite bugs (compiler itself is not affected) 11947 11948 * [333]10737 newer bison causes g++.dg/parse/crash2.C to incorrectly 11949 report failure 11950 * [334]10810 gcc-3.3 fails make check: buffer overrun in 11951 test_demangle.c 11952 __________________________________________________________________ 11953 11954 GCC 3.3.2 11955 11956 Bug Fixes 11957 11958 This section lists the problem reports (PRs) from GCC's bug tracker 11959 that are known to be fixed in the 3.3.2 release. This list might not be 11960 complete (that is, it is possible that some PRs that have been fixed 11961 are not listed here). 11962 11963 Bootstrap failures and problems 11964 11965 * [335]8336 [SCO5] bootstrap config still tries to use COFF options 11966 * [336]9330 [alpha-osf] Bootstrap failure on Compaq Tru64 with 11967 --enable-threads=posix 11968 * [337]9631 [hppa64-linux] gcc-3.3 fails to bootstrap 11969 * [338]9877 fixincludes makes a bad sys/byteorder.h on svr5 (UnixWare 11970 7.1.1) 11971 * [339]11687 xstormy16-elf build fails in libf2c 11972 * [340]12263 [SGI IRIX] bootstrap fails during compile of 11973 libf2c/libI77/backspace.c 11974 * [341]12490 buffer overflow in scan-decls.c (during Solaris 9 11975 fix-header processing) 11976 11977 Internal compiler errors (multi-platform) 11978 11979 * [342]7277 Casting integers to vector types causes ICE 11980 * [343]7939 (c++) ICE on invalid function template specialization 11981 * [344]11063 (c++) ICE on parsing initialization list of const array 11982 member 11983 * [345]11207 ICE with negative index in array element designator 11984 * [346]11522 (fortran) g77 dwarf-2 ICE in 11985 add_abstract_origin_attribute 11986 * [347]11595 (c++) ICE on duplicate label definition 11987 * [348]11646 (c++) ICE in commit_one_edge_insertion with 11988 -fnon-call-exceptions -fgcse -O 11989 * [349]11665 ICE in struct initializer when taking address 11990 * [350]11852 (c++) ICE with bad struct initializer. 11991 * [351]11878 (c++) ICE in cp_expr_size 11992 * [352]11883 ICE with any -O on mercury-generated C code 11993 * [353]11991 (c++) ICE in cxx_incomplete_type_diagnostic, in 11994 cp/typeck2.c when applying typeid operator to template template 11995 parameter 11996 * [354]12146 ICE in lookup_template_function, in cp/pt.c 11997 * [355]12215 ICE in make_label_edge with -fnon-call-exceptions 11998 -fno-gcse -O2 11999 * [356]12369 (c++) ICE with templates and friends 12000 * [357]12446 ICE in emit_move_insn on complicated array reference 12001 * [358]12510 ICE in final_scan_insn 12002 * [359]12544 ICE with large parameters used in nested functions 12003 12004 C and optimization bugs 12005 12006 * [360]9862 spurious warnings with -W -finline-functions 12007 * [361]10962 lookup_field is a linear search on a linked list (can be 12008 slow if large struct) 12009 * [362]11370 -Wunreachable-code gives false complaints 12010 * [363]11637 invalid assembly with -fnon-call-exceptions 12011 * [364]11885 Problem with bitfields in packed structs 12012 * [365]12082 Inappropriate unreachable code warnings 12013 * [366]12180 Inline optimization fails for variadic function 12014 * [367]12340 loop unroller + gcse produces wrong code 12015 12016 C++ compiler and library 12017 12018 * [368]3907 nested template parameter collides with member name 12019 * [369]5293 confusing message when binding a temporary to a reference 12020 * [370]5296 [DR115] Pointers to functions and to template functions 12021 behave differently in deduction 12022 * [371]7939 ICE on function template specialization 12023 * [372]8656 Unable to assign function with __attribute__ and pointer 12024 return type to an appropriate variable 12025 * [373]10147 Confusing error message for invalid template function 12026 argument 12027 * [374]11400 std::search_n() makes assumptions about Size parameter 12028 * [375]11409 issues with using declarations, overloading, and 12029 built-in functions 12030 * [376]11740 ctype<wchar_t>::do_is(mask, wchar_t) doesn't handle 12031 multiple bits in mask 12032 * [377]11786 operator() call on variable in other namespace not 12033 recognized 12034 * [378]11867 static_cast ignores ambiguity 12035 * [379]11928 bug with conversion operators that are typedefs 12036 * [380]12114 Uninitialized memory accessed in dtor 12037 * [381]12163 static_cast + explicit constructor regression 12038 * [382]12181 Wrong code with comma operator and c++ 12039 * [383]12236 regparm and fastcall messes up parameters 12040 * [384]12266 incorrect instantiation of unneeded template during 12041 overload resolution 12042 * [385]12296 istream::peek() doesn't set eofbit 12043 * [386]12298 [sjlj exceptions] Stack unwind destroys 12044 not-yet-constructed object 12045 * [387]12369 ICE with templates and friends 12046 * [388]12337 apparently infinite loop in g++ 12047 * [389]12344 stdcall attribute ignored if function returns a pointer 12048 * [390]12451 missing(late) class forward declaration in cxxabi.h 12049 * [391]12486 g++ accepts invalid use of a qualified name 12050 12051 x86 specific (Intel/AMD) 12052 12053 * [392]8869 [x86 MMX] ICE with const variable optimization and MMX 12054 builtins 12055 * [393]9786 ICE in fixup_abnormal_edges with -fnon-call-exceptions 12056 -O2 12057 * [394]11689 g++3.3 emits un-assembleable code for k6 architecture 12058 * [395]12116 [k6] Invalid assembly output values with X-MAME code 12059 * [396]12070 ICE converting between double and long double with 12060 -msoft-float 12061 12062 ia64-specific 12063 12064 * [397]11184 [ia64 hpux] ICE on __builtin_apply building libobjc 12065 * [398]11535 __builtin_return_address may not work on ia64 12066 * [399]11693 [ia64] ICE in gen_nop_type 12067 * [400]12224 [ia64] Thread-local storage doesn't work 12068 12069 PowerPC-specific 12070 12071 * [401]11087 [powerpc64-linux] GCC miscompiles raid1.c from linux 12072 kernel 12073 * [402]11319 loop miscompiled on ppc32 12074 * [403]11949 ICE Compiler segfault with ffmpeg -maltivec code 12075 12076 SPARC-specific 12077 12078 * [404]11662 wrong code for expr. with cast to long long and 12079 exclusive or 12080 * [405]11965 invalid assembler code for a shift < 32 operation 12081 * [406]12301 (c++) stack corruption when a returned expression throws 12082 an exception 12083 12084 Alpha-specific 12085 12086 * [407]11717 [alpha-linux] unrecognizable insn compiling for.c of 12087 kernel 2.4.22-pre8 12088 12089 HPUX-specific 12090 12091 * [408]11313 problem with #pragma weak and static inline functions 12092 * [409]11712 __STDC_EXT__ not defined for C++ by default anymore? 12093 12094 Solaris specific 12095 12096 * [410]12166 Profiled programs crash if PROFDIR is set 12097 12098 Solaris-x86 specific 12099 12100 * [411]12101 i386 Solaris no longer works with GNU as? 12101 12102 Miscellaneous embedded target-specific bugs 12103 12104 * [412]10988 [m32r-elf] wrong blockmove code with -O3 12105 * [413]11805 [h8300-unknown-coff] [H8300] ICE for simple code with 12106 -O2 12107 * [414]11902 [sh4] spec file improperly inserts rpath even when none 12108 needed 12109 * [415]11903 [sh4] -pthread fails to link due to error in spec file 12110 on sh4 12111 __________________________________________________________________ 12112 12113 GCC 3.3.3 12114 12115 Minor features 12116 12117 In addition to the bug fixes documented below, this release contains 12118 few minor features such as: 12119 * Support for --with-sysroot 12120 * Support for automatic detection of executable stacks 12121 * Support for SSE3 instructions 12122 * Support for thread local storage debugging under GDB on S390 12123 12124 Bug Fixes 12125 12126 This section lists the problem reports (PRs) from GCC's bug tracker 12127 that are known to be fixed in the 3.3.3 release. This list might not be 12128 complete (that is, it is possible that some PRs that have been fixed 12129 are not listed here). 12130 12131 Bootstrap failures and issues 12132 12133 * [416]11890 Building cross gcc-3.3.1 for sparc-sun-solaris2.6 fails 12134 * [417]12399 boehm-gc fails (when building a cross compiler): libtool 12135 unable to infer tagged configuration 12136 * [418]13068 mklibgcc.in doesn't handle multi-level multilib 12137 subdirectories properly 12138 12139 Internal compiler errors (multi-platform) 12140 12141 * [419]10060 ICE (stack overflow) on huge file (300k lines) due to 12142 recursive behaviour of copy_rtx_if_shared, in emit_rtl.c 12143 * [420]10555 (c++) ICE on undefined template argument 12144 * [421]10706 (c++) ICE in mangle_class_name_for_template 12145 * [422]11496 (fortran) error in flow_loops_find when -funroll-loops 12146 active 12147 * [423]11741 ICE in pre_insert_copy_insn, in gcse.c 12148 * [424]12440 GCC crashes during compilation of quicktime4linux 2.0.0 12149 * [425]12632 (fortran) -fbounds-check ICE 12150 * [426]12712 (c++) ICE on short legit C++ code fragment with gcc 12151 3.3.2 12152 * [427]12726 (c++) ICE (segfault) on trivial code 12153 * [428]12890 (c++) ICE on compilation of class with throwing method 12154 * [429]12900 (c++) ICE in rtl_verify_flow_info_1 12155 * [430]13060 (fortran) ICE in fixup_var_refs_1, in function.c on 12156 correct code with -O2 -fno-force-mem 12157 * [431]13289 (c++) ICE in regenerate_decl_from_template on recursive 12158 template 12159 * [432]13318 ICE: floating point exception in the loop optimizer 12160 * [433]13392 (c++) ICE in convert_from_eh_region_ranges_1, in 12161 except.c 12162 * [434]13574 (c++) invalid array default initializer in class lets 12163 gcc consume all memory and die 12164 * [435]13475 ICE on SIMD variables with partial value initialization 12165 * [436]13797 (c++) ICE on invalid template parameter 12166 * [437]13824 (java) gcj SEGV with simple .java program 12167 12168 C and optimization bugs 12169 12170 * [438]8776 loop invariants are not removed (most likely) 12171 * [439]10339 [sparc,ppc,ppc64] Invalid optimization: replacing 12172 strncmp by memcmp 12173 * [440]11350 undefined labels with -Os -fPIC 12174 * [441]12826 Optimizer removes reference through volatile pointer 12175 * [442]12500 stabs debug info: void no longer a predefined / builtin 12176 type 12177 * [443]12941 builtin-bitops-1.c miscompilation (latent bug) 12178 * [444]12953 tree inliner bug (in inline_forbidden_p) and fix 12179 * [445]13041 linux-2.6/sound/core/oss/rate.c miscompiled 12180 * [446]13507 spurious printf format warning 12181 * [447]13382 Type information for const pointer disappears during 12182 optimization. 12183 * [448]13394 noreturn attribute ignored on recursive invokation 12184 * [449]13400 Compiled code crashes storing to read-only location 12185 * [450]13521 Endless loop in calculate_global_regs_live 12186 12187 C++ compiler and library 12188 12189 Some of the bug fixes in this list were made to implement decisions 12190 that the ISO C++ standards committee has made concerning several defect 12191 reports (DRs). Links in the list below point to detailed discussion of 12192 the relevant defect report. 12193 * [451]2094 unimplemented: use of `ptrmem_cst' in template type 12194 unification 12195 * [452]2294 using declaration confusion 12196 * [453]5050 template instantiation depth exceeds limit: recursion 12197 problem? 12198 * [454]9371 Bad exception handling in 12199 i/ostream::operator>>/<<(streambuf*) 12200 * [455]9546 bad exception handling in ostream members 12201 * [456]10081 basic_ios::_M_cache_locale leaves NULL members in the 12202 face of unknown locales 12203 * [457]10093 [458][DR 61] Setting failbit in exceptions doesn't work 12204 * [459]10095 istream::operator>>(int&) sets ios::badbit when 12205 ios::failbit is set. 12206 * [460]11554 Warning about reordering of initializers doesn't mention 12207 location of constructor 12208 * [461]12297 istream::sentry::sentry() handles eof() incorrectly. 12209 * [462]12352 Exception safety problems in src/localename.cc 12210 * [463]12438 Memory leak in locale::combine() 12211 * [464]12540 Memory leak in locale::locale(const char*) 12212 * [465]12594 DRs [466]60 [TC] and [467]63 [TC] not implemented 12213 * [468]12657 Resolution of [469]DR 292 (WP) still unimplemented 12214 * [470]12696 memory eating infinite loop in diagnostics (error 12215 recovery problem) 12216 * [471]12815 Code compiled with optimization behaves unexpectedly 12217 * [472]12862 Conflicts between typedefs/enums and namespace member 12218 declarations 12219 * [473]12926 Wrong value after assignment in initialize list using 12220 bit-fields 12221 * [474]12967 Resolution of [475]DR 300 [WP] still unimplemented 12222 * [476]12971 Resolution of [477]DR 328 [WP] still unimplemented 12223 * [478]13007 basic_streambuf::pubimbue, imbue wrong 12224 * [479]13009 Implicitly-defined assignment operator writes to wrong 12225 memory 12226 * [480]13057 regparm attribute not applied to destructor 12227 * [481]13070 -Wformat option ignored in g++ 12228 * [482]13081 forward template declarations in <complex> let inlining 12229 fail 12230 * [483]13239 Assertion does not seem to work correctly anymore 12231 * [484]13262 "xxx is private within this context" when initializing a 12232 self-contained template class 12233 * [485]13290 simple typo in concept checking for std::generate_n 12234 * [486]13323 Template code does not compile in presence of typedef 12235 * [487]13369 __verify_grouping (and __add_grouping?) not correct 12236 * [488]13371 infinite loop with packed struct and inlining 12237 * [489]13445 Template argument replacement "dereferences" a typedef 12238 * [490]13461 Fails to access protected-ctor from public constant 12239 * [491]13462 Non-standard-conforming type set::pointer 12240 * [492]13478 gcc uses wrong constructor to initialize a const 12241 reference 12242 * [493]13544 "conflicting types" for enums in different scopes 12243 * [494]13650 string::compare should not (always) use 12244 traits_type::length() 12245 * [495]13683 bogus warning about passing non-PODs through ellipsis 12246 * [496]13688 Derived class is denied access to protected base class 12247 member class 12248 * [497]13774 Member variable cleared in virtual multiple inheritance 12249 class 12250 * [498]13884 Protect sstream.tcc from extern template use 12251 12252 Java compiler and library 12253 12254 * [499]10746 [win32] garbage collection crash in GCJ 12255 12256 Objective-C compiler and library 12257 12258 * [500]11433 Crash due to dereferencing null pointer when querying 12259 protocol 12260 12261 Fortran compiler and library 12262 12263 * [501]12633 logical expression gives incorrect result with 12264 -fugly-logint option 12265 * [502]13037 [gcse-lm] g77 generates incorrect code 12266 * [503]13213 Hex constant problem when compiling with -fugly-logint 12267 and -ftypeless-boz 12268 12269 x86-specific (Intel/AMD) 12270 12271 * [504]4490 ICE with -m128bit-long-double 12272 * [505]12292 [x86_64] ICE: RTL check: expected code `const_int', have 12273 `reg' in make_field_assignment, in combine.c 12274 * [506]12441 ICE: can't find a register to spill 12275 * [507]12943 array static-init failure under -fpic, -fPIC 12276 * [508]13608 Incorrect code with -O3 -ffast-math 12277 12278 PowerPC-specific 12279 12280 * [509]11598 testcase gcc.dg/20020118-1.c fails runtime check of 12281 __attribute__((aligned(16))) 12282 * [510]11793 ICE in extract_insn, in recog.c (const_vector's) 12283 * [511]12467 vmsumubm emitted when vmsummbm appropriate (typo in 12284 altivec.md) 12285 * [512]12537 g++ generates writeable text sections 12286 12287 SPARC-specific 12288 12289 * [513]12496 wrong result for __atomic_add(&value, -1) when using -O0 12290 -m64 12291 * [514]12865 mprotect call to make trampoline executable may fail 12292 * [515]13354 ICE in sparc_emit_set_const32 12293 12294 ARM-specific 12295 12296 * [516]10467 [arm] ICE in pre_insert_copy_insn, 12297 12298 ia64-specific 12299 12300 * [517]11226 ICE passing struct arg with two floats 12301 * [518]11227 ICE for _Complex float, _Complex long double args 12302 * [519]12644 GCC 3.3.2 fails to compile glibc on ia64 12303 * [520]13149 build gcc-3.3.2 1305 error:unrecognizable insn 12304 * Various fixes for libunwind 12305 12306 Alpha-specific 12307 12308 * [521]12654 Incorrect comparison code generated for Alpha 12309 * [522]12965 SEGV+ICE in cc1plus on alpha-linux with -O2 12310 * [523]13031 ICE (unrecognizable insn) when building gnome-libs-1.4.2 12311 12312 HPPA-specific 12313 12314 * [524]11634 [hppa] ICE in verify_local_live_at_start, in flow.c 12315 * [525]12158 [hppa] compilation does not terminate at -O1 12316 12317 S390-specific 12318 12319 * [526]11992 Wrong built-in code for memcmp with length 1<<24: only 12320 (1<<24)-1 possible for CLCL-Instruction 12321 12322 SH-specific 12323 12324 * [527]9365 segfault in gen_far_branch (config/sh/sh.c) 12325 * [528]10392 optimizer generates faulty array indexing 12326 * [529]11322 SH profiler outputs multiple definitions of symbol 12327 * [530]13069 gcc/config/sh/rtems.h broken 12328 * [531]13302 Putting a va_list in a struct causes seg fault 12329 * [532]13585 Incorrect optimization of call to sfunc 12330 * Fix inappropriately exported libgcc functions from the shared 12331 library 12332 12333 Other embedded target specific 12334 12335 * [533]8916 [mcore] unsigned char assign gets hosed. 12336 * [534]11576 [h8300] ICE in change_address_1, in emit-rtl.c 12337 * [535]13122 [h8300] local variable gets corrupted by function call 12338 when -fomit-frame-pointer is given 12339 * [536]13256 [cris] strict_low_part mistreated in delay slots 12340 * [537]13373 [mcore] optimization with -frerun-cse-after-loop 12341 -fexpensive-optimizations produces wrong code on mcore 12342 12343 GNU HURD-specific 12344 12345 * [538]12561 gcc/config/t-gnu needs updating to work with 12346 --with-sysroot 12347 12348 Tru64 Unix specific 12349 12350 * [539]6243 testsuite fails almost all tests due to no libintl in 12351 LD_LIBRARY_PATH during test. 12352 * [540]11397 weak aliases broken on Tru64 UNIX 12353 12354 AIX-specific 12355 12356 * [541]12505 build failure due to defines of uchar in cpphash.h and 12357 sys/types.h 12358 * [542]13150 WEAK symbols not exported by collect2 12359 12360 IRIX-specific 12361 12362 * [543]12666 fixincludes problem on IRIX 6.5.19m 12363 12364 Solaris-specific 12365 12366 * [544]12969 Including sys/byteorder.h breaks configure checks 12367 12368 Testsuite problems (compiler is not affected) 12369 12370 * [545]10819 testsuite creates CR+LF on compiler version lines in 12371 test summary files 12372 * [546]11612 abi_check not finding correct libgcc_s.so.1 12373 12374 Miscellaneous 12375 12376 * [547]13211 using -###, incorrect warnings about unused linker file 12377 are produced 12378 __________________________________________________________________ 12379 12380 GCC 3.3.4 12381 12382 This is the [548]list of problem reports (PRs) from GCC's bug tracking 12383 system that are known to be fixed in the 3.3.4 release. This list might 12384 not be complete (that is, it is possible that some PRs that have been 12385 fixed are not listed here). 12386 __________________________________________________________________ 12387 12388 GCC 3.3.5 12389 12390 This is the [549]list of problem reports (PRs) from GCC's bug tracking 12391 system that are known to be fixed in the 3.3.5 release. This list might 12392 not be complete (that is, it is possible that some PRs that have been 12393 fixed are not listed here). 12394 __________________________________________________________________ 12395 12396 GCC 3.3.6 12397 12398 This is the [550]list of problem reports (PRs) from GCC's bug tracking 12399 system that are known to be fixed in the 3.3.6 release. This list might 12400 not be complete (that is, it is possible that some PRs that have been 12401 fixed are not listed here). 12402 12403 12404 For questions related to the use of GCC, please consult these web 12405 pages and the [551]GCC manuals. If that fails, the 12406 [552]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 12407 web pages and the development of GCC are welcome on our developer 12408 list at [553]gcc (a] gcc.gnu.org. All of [554]our lists have public 12409 archives. 12410 12411 Copyright (C) [555]Free Software Foundation, Inc. Verbatim copying and 12412 distribution of this entire article is permitted in any medium, 12413 provided this notice is preserved. 12414 12415 These pages are [556]maintained by the GCC team. Last modified 12416 2016-01-30[557]. 12417 12418 References 12419 12420 1. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6 12421 2. http://gcc.gnu.org/gcc-3.1/changes.html#obsolete_systems 12422 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems 12423 4. http://gcc.gnu.org/gcc-3.3/changes.html#nonnull_attribute 12424 5. http://gcc.gnu.org/news/dfa.html 12425 6. http://gcc.gnu.org/c99status.html 12426 7. https://gcc.gnu.org/onlinedocs/gcc-3.3.6/g77/News.html 12427 8. https://gcc.gnu.org/PR10140 12428 9. https://gcc.gnu.org/PR10198 12429 10. https://gcc.gnu.org/PR10338 12430 11. https://gcc.gnu.org/PR3581 12431 12. https://gcc.gnu.org/PR4382 12432 13. https://gcc.gnu.org/PR5533 12433 14. https://gcc.gnu.org/PR6387 12434 15. https://gcc.gnu.org/PR6412 12435 16. https://gcc.gnu.org/PR6620 12436 17. https://gcc.gnu.org/PR6663 12437 18. https://gcc.gnu.org/PR7068 12438 19. https://gcc.gnu.org/PR7083 12439 20. https://gcc.gnu.org/PR7647 12440 21. https://gcc.gnu.org/PR7675 12441 22. https://gcc.gnu.org/PR7718 12442 23. https://gcc.gnu.org/PR8116 12443 24. https://gcc.gnu.org/PR8358 12444 25. https://gcc.gnu.org/PR8511 12445 26. https://gcc.gnu.org/PR8564 12446 27. https://gcc.gnu.org/PR8660 12447 28. https://gcc.gnu.org/PR8766 12448 29. https://gcc.gnu.org/PR8803 12449 30. https://gcc.gnu.org/PR8846 12450 31. https://gcc.gnu.org/PR8906 12451 32. https://gcc.gnu.org/PR9216 12452 33. https://gcc.gnu.org/PR9261 12453 34. https://gcc.gnu.org/PR9263 12454 35. https://gcc.gnu.org/PR9429 12455 36. https://gcc.gnu.org/PR9516 12456 37. https://gcc.gnu.org/PR9600 12457 38. https://gcc.gnu.org/PR9629 12458 39. https://gcc.gnu.org/PR9672 12459 40. https://gcc.gnu.org/PR9749 12460 41. https://gcc.gnu.org/PR9794 12461 42. https://gcc.gnu.org/PR9829 12462 43. https://gcc.gnu.org/PR9916 12463 44. https://gcc.gnu.org/PR9936 12464 45. https://gcc.gnu.org/PR10262 12465 46. https://gcc.gnu.org/PR10278 12466 47. https://gcc.gnu.org/PR10446 12467 48. https://gcc.gnu.org/PR10451 12468 49. https://gcc.gnu.org/PR10506 12469 50. https://gcc.gnu.org/PR10549 12470 51. https://gcc.gnu.org/PR2001 12471 52. https://gcc.gnu.org/PR2391 12472 53. https://gcc.gnu.org/PR2960 12473 54. https://gcc.gnu.org/PR4046 12474 55. https://gcc.gnu.org/PR6405 12475 56. https://gcc.gnu.org/PR6798 12476 57. https://gcc.gnu.org/PR6871 12477 58. https://gcc.gnu.org/PR6909 12478 59. https://gcc.gnu.org/PR7189 12479 60. https://gcc.gnu.org/PR7642 12480 61. https://gcc.gnu.org/PR8634 12481 62. https://gcc.gnu.org/PR8750 12482 63. https://gcc.gnu.org/PR2161 12483 64. https://gcc.gnu.org/PR4319 12484 65. https://gcc.gnu.org/PR8602 12485 66. https://gcc.gnu.org/PR9177 12486 67. https://gcc.gnu.org/PR9853 12487 68. https://gcc.gnu.org/PR45 12488 69. https://gcc.gnu.org/PR3784 12489 70. https://gcc.gnu.org/PR764 12490 71. https://gcc.gnu.org/PR5116 12491 72. https://gcc.gnu.org/PR2862 12492 73. https://gcc.gnu.org/PR3663 12493 74. https://gcc.gnu.org/PR3797 12494 75. https://gcc.gnu.org/PR3948 12495 76. https://gcc.gnu.org/PR4137 12496 77. https://gcc.gnu.org/PR4361 12497 78. https://gcc.gnu.org/PR4802 12498 79. https://gcc.gnu.org/PR5837 12499 80. https://gcc.gnu.org/PR4803 12500 81. https://gcc.gnu.org/PR5094 12501 82. https://gcc.gnu.org/PR5730 12502 83. https://gcc.gnu.org/PR6713 12503 84. https://gcc.gnu.org/PR7015 12504 85. https://gcc.gnu.org/PR7086 12505 86. https://gcc.gnu.org/PR7099 12506 87. https://gcc.gnu.org/PR7247 12507 88. https://gcc.gnu.org/PR7441 12508 89. https://gcc.gnu.org/PR7768 12509 90. https://gcc.gnu.org/PR7804 12510 91. https://gcc.gnu.org/PR8099 12511 92. https://gcc.gnu.org/PR8117 12512 93. https://gcc.gnu.org/PR8205 12513 94. https://gcc.gnu.org/PR8645 12514 95. https://gcc.gnu.org/PR8724 12515 96. https://gcc.gnu.org/PR8805 12516 97. https://gcc.gnu.org/PR8691 12517 98. https://gcc.gnu.org/PR8700 12518 99. https://gcc.gnu.org/PR8724 12519 100. https://gcc.gnu.org/PR8949 12520 101. https://gcc.gnu.org/PR9016 12521 102. https://gcc.gnu.org/PR9053 12522 103. https://gcc.gnu.org/PR9152 12523 104. https://gcc.gnu.org/PR9182 12524 105. https://gcc.gnu.org/PR9297 12525 106. https://gcc.gnu.org/PR9318 12526 107. https://gcc.gnu.org/PR9320 12527 108. https://gcc.gnu.org/PR9400 12528 109. https://gcc.gnu.org/PR9424 12529 110. https://gcc.gnu.org/PR9425 12530 111. https://gcc.gnu.org/PR9439 12531 112. https://gcc.gnu.org/PR9474 12532 113. https://gcc.gnu.org/PR9548 12533 114. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#231 12534 115. https://gcc.gnu.org/PR9555 12535 116. https://gcc.gnu.org/PR9561 12536 117. https://gcc.gnu.org/PR9563 12537 118. https://gcc.gnu.org/PR9582 12538 119. https://gcc.gnu.org/PR9622 12539 120. https://gcc.gnu.org/PR9683 12540 121. https://gcc.gnu.org/PR9791 12541 122. https://gcc.gnu.org/PR9817 12542 123. https://gcc.gnu.org/PR9825 12543 124. https://gcc.gnu.org/PR9826 12544 125. https://gcc.gnu.org/PR9924 12545 126. https://gcc.gnu.org/PR9946 12546 127. https://gcc.gnu.org/PR9964 12547 128. https://gcc.gnu.org/PR9988 12548 129. https://gcc.gnu.org/PR10033 12549 130. https://gcc.gnu.org/PR10097 12550 131. https://gcc.gnu.org/PR10132 12551 132. https://gcc.gnu.org/PR10180 12552 133. https://gcc.gnu.org/PR10199 12553 134. https://gcc.gnu.org/PR10300 12554 135. https://gcc.gnu.org/PR10427 12555 136. https://gcc.gnu.org/PR10503 12556 137. https://gcc.gnu.org/PR5956 12557 138. https://gcc.gnu.org/PR1832 12558 139. https://gcc.gnu.org/PR3924 12559 140. https://gcc.gnu.org/PR5634 12560 141. https://gcc.gnu.org/PR6367 12561 142. https://gcc.gnu.org/PR6491 12562 143. https://gcc.gnu.org/PR6742 12563 144. https://gcc.gnu.org/PR7113 12564 145. https://gcc.gnu.org/PR7236 12565 146. https://gcc.gnu.org/PR7278 12566 147. https://gcc.gnu.org/PR7384 12567 148. https://gcc.gnu.org/PR7388 12568 149. https://gcc.gnu.org/PR8587 12569 150. https://gcc.gnu.org/PR9038 12570 151. https://gcc.gnu.org/PR10197 12571 152. https://gcc.gnu.org/PR6005 12572 153. https://gcc.gnu.org/PR6389 12573 154. https://gcc.gnu.org/PR6576 12574 155. https://gcc.gnu.org/PR6652 12575 156. https://gcc.gnu.org/PR7060 12576 157. https://gcc.gnu.org/PR7073 12577 158. https://gcc.gnu.org/PR7180 12578 159. https://gcc.gnu.org/PR7416 12579 160. https://gcc.gnu.org/PR7570 12580 161. https://gcc.gnu.org/PR7578 12581 162. https://gcc.gnu.org/PR7611 12582 163. https://gcc.gnu.org/PR7709 12583 164. https://gcc.gnu.org/PR7766 12584 165. https://gcc.gnu.org/PR7785 12585 166. https://gcc.gnu.org/PR7786 12586 167. https://gcc.gnu.org/PR8142 12587 168. https://gcc.gnu.org/PR8234 12588 169. https://gcc.gnu.org/PR8415 12589 170. https://gcc.gnu.org/PR8481 12590 171. https://gcc.gnu.org/PR8593 12591 172. https://gcc.gnu.org/PR8759 12592 173. https://gcc.gnu.org/PR8997 12593 174. https://gcc.gnu.org/PR9253 12594 175. https://gcc.gnu.org/PR9254 12595 176. https://gcc.gnu.org/PR9271 12596 177. https://gcc.gnu.org/PR6767 12597 178. https://gcc.gnu.org/PR9911 12598 179. https://gcc.gnu.org/PR10020 12599 180. https://gcc.gnu.org/PR10546 12600 181. https://gcc.gnu.org/PR7029 12601 182. https://gcc.gnu.org/PR2903 12602 183. https://gcc.gnu.org/PR7873 12603 184. https://gcc.gnu.org/PR7680 12604 185. https://gcc.gnu.org/PR8705 12605 186. https://gcc.gnu.org/PR9986 12606 187. https://gcc.gnu.org/PR10056 12607 188. https://gcc.gnu.org/PR6744 12608 189. https://gcc.gnu.org/PR7361 12609 190. https://gcc.gnu.org/PR9496 12610 191. https://gcc.gnu.org/PR7067 12611 192. https://gcc.gnu.org/PR8480 12612 193. https://gcc.gnu.org/PR8784 12613 194. https://gcc.gnu.org/PR10315 12614 195. https://gcc.gnu.org/PR10267 12615 196. https://gcc.gnu.org/PR7916 12616 197. https://gcc.gnu.org/PR7926 12617 198. https://gcc.gnu.org/PR8555 12618 199. https://gcc.gnu.org/PR8994 12619 200. https://gcc.gnu.org/PR9426 12620 201. https://gcc.gnu.org/PR9806 12621 202. https://gcc.gnu.org/PR10077 12622 203. https://gcc.gnu.org/PR10233 12623 204. https://gcc.gnu.org/PR10286 12624 205. https://gcc.gnu.org/PR10308 12625 206. https://gcc.gnu.org/PR11272 12626 207. https://gcc.gnu.org/PR5754 12627 208. https://gcc.gnu.org/PR6597 12628 209. https://gcc.gnu.org/PR6949 12629 210. https://gcc.gnu.org/PR7053 12630 211. https://gcc.gnu.org/PR8164 12631 212. https://gcc.gnu.org/PR8384 12632 213. https://gcc.gnu.org/PR9559 12633 214. https://gcc.gnu.org/PR9649 12634 215. https://gcc.gnu.org/PR9864 12635 216. https://gcc.gnu.org/PR10432 12636 217. https://gcc.gnu.org/PR10475 12637 218. https://gcc.gnu.org/PR10635 12638 219. https://gcc.gnu.org/PR10661 12639 220. https://gcc.gnu.org/PR10700 12640 221. https://gcc.gnu.org/PR10712 12641 222. https://gcc.gnu.org/PR10796 12642 223. https://gcc.gnu.org/PR10890 12643 224. https://gcc.gnu.org/PR10939 12644 225. https://gcc.gnu.org/PR10956 12645 226. https://gcc.gnu.org/PR11041 12646 227. https://gcc.gnu.org/PR11059 12647 228. https://gcc.gnu.org/PR11083 12648 229. https://gcc.gnu.org/PR11105 12649 230. https://gcc.gnu.org/PR11149 12650 231. https://gcc.gnu.org/PR11228 12651 232. https://gcc.gnu.org/PR11282 12652 233. https://gcc.gnu.org/PR11301 12653 234. https://gcc.gnu.org/PR11308 12654 235. https://gcc.gnu.org/PR11473 12655 236. https://gcc.gnu.org/PR11503 12656 237. https://gcc.gnu.org/PR11513 12657 238. https://gcc.gnu.org/PR11198 12658 239. https://gcc.gnu.org/PR11304 12659 240. https://gcc.gnu.org/PR11381 12660 241. https://gcc.gnu.org/PR11536 12661 242. https://gcc.gnu.org/PR11557 12662 243. https://gcc.gnu.org/PR5897 12663 244. https://gcc.gnu.org/PR11279 12664 245. https://gcc.gnu.org/PR11022 12665 246. https://gcc.gnu.org/PR2330 12666 247. https://gcc.gnu.org/PR5388 12667 248. https://gcc.gnu.org/PR5390 12668 249. https://gcc.gnu.org/PR7877 12669 250. https://gcc.gnu.org/PR9393 12670 251. https://gcc.gnu.org/PR10032 12671 252. https://gcc.gnu.org/PR10468 12672 253. https://gcc.gnu.org/PR10527 12673 254. https://gcc.gnu.org/PR10679 12674 255. https://gcc.gnu.org/PR10682 12675 256. https://gcc.gnu.org/PR10689 12676 257. https://gcc.gnu.org/PR10845 12677 258. https://gcc.gnu.org/PR10849 12678 259. https://gcc.gnu.org/PR10888 12679 260. https://gcc.gnu.org/PR10929 12680 261. https://gcc.gnu.org/PR10931 12681 262. https://gcc.gnu.org/PR10940 12682 263. https://gcc.gnu.org/PR10968 12683 264. https://gcc.gnu.org/PR10990 12684 265. https://gcc.gnu.org/PR11039 12685 266. https://gcc.gnu.org/PR11062 12686 267. https://gcc.gnu.org/PR11095 12687 268. https://gcc.gnu.org/PR11098 12688 269. https://gcc.gnu.org/PR11137 12689 270. https://gcc.gnu.org/PR11154 12690 271. https://gcc.gnu.org/PR11329 12691 272. https://gcc.gnu.org/PR11332 12692 273. https://gcc.gnu.org/PR11431 12693 274. https://gcc.gnu.org/PR11528 12694 275. https://gcc.gnu.org/PR11546 12695 276. https://gcc.gnu.org/PR11567 12696 277. https://gcc.gnu.org/PR11645 12697 278. https://gcc.gnu.org/PR5179 12698 279. https://gcc.gnu.org/PR8204 12699 280. https://gcc.gnu.org/PR10838 12700 281. https://gcc.gnu.org/PR10886 12701 282. https://gcc.gnu.org/PR11349 12702 283. https://gcc.gnu.org/PR4823 12703 284. https://gcc.gnu.org/PR8878 12704 285. https://gcc.gnu.org/PR9815 12705 286. https://gcc.gnu.org/PR10402 12706 287. https://gcc.gnu.org/PR10504 12707 288. https://gcc.gnu.org/PR10673 12708 289. https://gcc.gnu.org/PR11044 12709 290. https://gcc.gnu.org/PR11089 12710 291. https://gcc.gnu.org/PR11420 12711 292. https://gcc.gnu.org/PR9362 12712 293. https://gcc.gnu.org/PR10142 12713 294. https://gcc.gnu.org/PR10663 12714 295. https://gcc.gnu.org/PR10835 12715 296. https://gcc.gnu.org/PR10876 12716 297. https://gcc.gnu.org/PR10955 12717 298. https://gcc.gnu.org/PR11018 12718 299. https://gcc.gnu.org/PR11556 12719 300. https://gcc.gnu.org/PR10907 12720 301. https://gcc.gnu.org/PR11320 12721 302. https://gcc.gnu.org/PR11599 12722 303. https://gcc.gnu.org/PR9745 12723 304. https://gcc.gnu.org/PR10871 12724 305. https://gcc.gnu.org/PR11440 12725 306. https://gcc.gnu.org/PR7594 12726 307. https://gcc.gnu.org/PR10557 12727 308. https://gcc.gnu.org/PR11054 12728 309. https://gcc.gnu.org/PR10834 12729 310. https://gcc.gnu.org/PR10842 12730 311. https://gcc.gnu.org/PR11052 12731 312. https://gcc.gnu.org/PR11183 12732 313. https://gcc.gnu.org/PR11084 12733 314. https://gcc.gnu.org/PR10331 12734 315. https://gcc.gnu.org/PR10413 12735 316. https://gcc.gnu.org/PR11096 12736 317. https://gcc.gnu.org/PR2873 12737 318. https://gcc.gnu.org/PR3163 12738 319. https://gcc.gnu.org/PR5287 12739 320. https://gcc.gnu.org/PR10148 12740 321. https://gcc.gnu.org/PR8787 12741 322. https://gcc.gnu.org/PR10900 12742 323. https://gcc.gnu.org/PR1607 12743 324. https://gcc.gnu.org/PR4252 12744 325. https://gcc.gnu.org/PR4490 12745 326. https://gcc.gnu.org/PR10355 12746 327. https://gcc.gnu.org/PR10726 12747 328. https://gcc.gnu.org/PR10805 12748 329. https://gcc.gnu.org/PR10815 12749 330. https://gcc.gnu.org/PR10877 12750 331. https://gcc.gnu.org/PR11280 12751 332. https://gcc.gnu.org/PR11466 12752 333. https://gcc.gnu.org/PR10737 12753 334. https://gcc.gnu.org/PR10810 12754 335. https://gcc.gnu.org/PR8336 12755 336. https://gcc.gnu.org/PR9330 12756 337. https://gcc.gnu.org/PR9631 12757 338. https://gcc.gnu.org/PR9877 12758 339. https://gcc.gnu.org/PR11687 12759 340. https://gcc.gnu.org/PR12263 12760 341. https://gcc.gnu.org/PR12490 12761 342. https://gcc.gnu.org/PR7277 12762 343. https://gcc.gnu.org/PR7939 12763 344. https://gcc.gnu.org/PR11063 12764 345. https://gcc.gnu.org/PR11207 12765 346. https://gcc.gnu.org/PR11522 12766 347. https://gcc.gnu.org/PR11595 12767 348. https://gcc.gnu.org/PR11646 12768 349. https://gcc.gnu.org/PR11665 12769 350. https://gcc.gnu.org/PR11852 12770 351. https://gcc.gnu.org/PR11878 12771 352. https://gcc.gnu.org/PR11883 12772 353. https://gcc.gnu.org/PR11991 12773 354. https://gcc.gnu.org/PR12146 12774 355. https://gcc.gnu.org/PR12215 12775 356. https://gcc.gnu.org/PR12369 12776 357. https://gcc.gnu.org/PR12446 12777 358. https://gcc.gnu.org/PR12510 12778 359. https://gcc.gnu.org/PR12544 12779 360. https://gcc.gnu.org/PR9862 12780 361. https://gcc.gnu.org/PR10962 12781 362. https://gcc.gnu.org/PR11370 12782 363. https://gcc.gnu.org/PR11637 12783 364. https://gcc.gnu.org/PR11885 12784 365. https://gcc.gnu.org/PR12082 12785 366. https://gcc.gnu.org/PR12180 12786 367. https://gcc.gnu.org/PR12340 12787 368. https://gcc.gnu.org/PR3907 12788 369. https://gcc.gnu.org/PR5293 12789 370. https://gcc.gnu.org/PR5296 12790 371. https://gcc.gnu.org/PR7939 12791 372. https://gcc.gnu.org/PR8656 12792 373. https://gcc.gnu.org/PR10147 12793 374. https://gcc.gnu.org/PR11400 12794 375. https://gcc.gnu.org/PR11409 12795 376. https://gcc.gnu.org/PR11740 12796 377. https://gcc.gnu.org/PR11786 12797 378. https://gcc.gnu.org/PR11867 12798 379. https://gcc.gnu.org/PR11928 12799 380. https://gcc.gnu.org/PR12114 12800 381. https://gcc.gnu.org/PR12163 12801 382. https://gcc.gnu.org/PR12181 12802 383. https://gcc.gnu.org/PR12236 12803 384. https://gcc.gnu.org/PR12266 12804 385. https://gcc.gnu.org/PR12296 12805 386. https://gcc.gnu.org/PR12298 12806 387. https://gcc.gnu.org/PR12369 12807 388. https://gcc.gnu.org/PR12337 12808 389. https://gcc.gnu.org/PR12344 12809 390. https://gcc.gnu.org/PR12451 12810 391. https://gcc.gnu.org/PR12486 12811 392. https://gcc.gnu.org/PR8869 12812 393. https://gcc.gnu.org/PR9786 12813 394. https://gcc.gnu.org/PR11689 12814 395. https://gcc.gnu.org/PR12116 12815 396. https://gcc.gnu.org/PR12070 12816 397. https://gcc.gnu.org/PR11184 12817 398. https://gcc.gnu.org/PR11535 12818 399. https://gcc.gnu.org/PR11693 12819 400. https://gcc.gnu.org/PR12224 12820 401. https://gcc.gnu.org/PR11087 12821 402. https://gcc.gnu.org/PR11319 12822 403. https://gcc.gnu.org/PR11949 12823 404. https://gcc.gnu.org/PR11662 12824 405. https://gcc.gnu.org/PR11965 12825 406. https://gcc.gnu.org/PR12301 12826 407. https://gcc.gnu.org/PR11717 12827 408. https://gcc.gnu.org/PR11313 12828 409. https://gcc.gnu.org/PR11712 12829 410. https://gcc.gnu.org/PR12166 12830 411. https://gcc.gnu.org/PR12101 12831 412. https://gcc.gnu.org/PR10988 12832 413. https://gcc.gnu.org/PR11805 12833 414. https://gcc.gnu.org/PR11902 12834 415. https://gcc.gnu.org/PR11903 12835 416. https://gcc.gnu.org/PR11890 12836 417. https://gcc.gnu.org/PR12399 12837 418. https://gcc.gnu.org/PR13068 12838 419. https://gcc.gnu.org/PR10060 12839 420. https://gcc.gnu.org/PR10555 12840 421. https://gcc.gnu.org/PR10706 12841 422. https://gcc.gnu.org/PR11496 12842 423. https://gcc.gnu.org/PR11741 12843 424. https://gcc.gnu.org/PR12440 12844 425. https://gcc.gnu.org/PR12632 12845 426. https://gcc.gnu.org/PR12712 12846 427. https://gcc.gnu.org/PR12726 12847 428. https://gcc.gnu.org/PR12890 12848 429. https://gcc.gnu.org/PR12900 12849 430. https://gcc.gnu.org/PR13060 12850 431. https://gcc.gnu.org/PR13289 12851 432. https://gcc.gnu.org/PR13318 12852 433. https://gcc.gnu.org/PR13392 12853 434. https://gcc.gnu.org/PR13574 12854 435. https://gcc.gnu.org/PR13475 12855 436. https://gcc.gnu.org/PR13797 12856 437. https://gcc.gnu.org/PR13824 12857 438. https://gcc.gnu.org/PR8776 12858 439. https://gcc.gnu.org/PR10339 12859 440. https://gcc.gnu.org/PR11350 12860 441. https://gcc.gnu.org/PR12826 12861 442. https://gcc.gnu.org/PR12500 12862 443. https://gcc.gnu.org/PR12941 12863 444. https://gcc.gnu.org/PR12953 12864 445. https://gcc.gnu.org/PR13041 12865 446. https://gcc.gnu.org/PR13507 12866 447. https://gcc.gnu.org/PR13382 12867 448. https://gcc.gnu.org/PR13394 12868 449. https://gcc.gnu.org/PR13400 12869 450. https://gcc.gnu.org/PR13521 12870 451. https://gcc.gnu.org/PR2094 12871 452. https://gcc.gnu.org/PR2294 12872 453. https://gcc.gnu.org/PR5050 12873 454. https://gcc.gnu.org/PR9371 12874 455. https://gcc.gnu.org/PR9546 12875 456. https://gcc.gnu.org/PR10081 12876 457. https://gcc.gnu.org/PR10093 12877 458. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#61 12878 459. https://gcc.gnu.org/PR10095 12879 460. https://gcc.gnu.org/PR11554 12880 461. https://gcc.gnu.org/PR12297 12881 462. https://gcc.gnu.org/PR12352 12882 463. https://gcc.gnu.org/PR12438 12883 464. https://gcc.gnu.org/PR12540 12884 465. https://gcc.gnu.org/PR12594 12885 466. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#60 12886 467. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#63 12887 468. https://gcc.gnu.org/PR12657 12888 469. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#292 12889 470. https://gcc.gnu.org/PR12696 12890 471. https://gcc.gnu.org/PR12815 12891 472. https://gcc.gnu.org/PR12862 12892 473. https://gcc.gnu.org/PR12926 12893 474. https://gcc.gnu.org/PR12967 12894 475. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html 12895 476. https://gcc.gnu.org/PR12971 12896 477. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#328 12897 478. https://gcc.gnu.org/PR13007 12898 479. https://gcc.gnu.org/PR13009 12899 480. https://gcc.gnu.org/PR13057 12900 481. https://gcc.gnu.org/PR13070 12901 482. https://gcc.gnu.org/PR13081 12902 483. https://gcc.gnu.org/PR13239 12903 484. https://gcc.gnu.org/PR13262 12904 485. https://gcc.gnu.org/PR13290 12905 486. https://gcc.gnu.org/PR13323 12906 487. https://gcc.gnu.org/PR13369 12907 488. https://gcc.gnu.org/PR13371 12908 489. https://gcc.gnu.org/PR13445 12909 490. https://gcc.gnu.org/PR13461 12910 491. https://gcc.gnu.org/PR13462 12911 492. https://gcc.gnu.org/PR13478 12912 493. https://gcc.gnu.org/PR13544 12913 494. https://gcc.gnu.org/PR13650 12914 495. https://gcc.gnu.org/PR13683 12915 496. https://gcc.gnu.org/PR13688 12916 497. https://gcc.gnu.org/PR13774 12917 498. https://gcc.gnu.org/PR13884 12918 499. https://gcc.gnu.org/PR10746 12919 500. https://gcc.gnu.org/PR11433 12920 501. https://gcc.gnu.org/PR12633 12921 502. https://gcc.gnu.org/PR13037 12922 503. https://gcc.gnu.org/PR13213 12923 504. https://gcc.gnu.org/PR4490 12924 505. https://gcc.gnu.org/PR12292 12925 506. https://gcc.gnu.org/PR12441 12926 507. https://gcc.gnu.org/PR12943 12927 508. https://gcc.gnu.org/PR13608 12928 509. https://gcc.gnu.org/PR11598 12929 510. https://gcc.gnu.org/PR11793 12930 511. https://gcc.gnu.org/PR12467 12931 512. https://gcc.gnu.org/PR12537 12932 513. https://gcc.gnu.org/PR12496 12933 514. https://gcc.gnu.org/PR12865 12934 515. https://gcc.gnu.org/PR13354 12935 516. https://gcc.gnu.org/PR10467 12936 517. https://gcc.gnu.org/PR11226 12937 518. https://gcc.gnu.org/PR11227 12938 519. https://gcc.gnu.org/PR12644 12939 520. https://gcc.gnu.org/PR13149 12940 521. https://gcc.gnu.org/PR12654 12941 522. https://gcc.gnu.org/PR12965 12942 523. https://gcc.gnu.org/PR13031 12943 524. https://gcc.gnu.org/PR11634 12944 525. https://gcc.gnu.org/PR12158 12945 526. https://gcc.gnu.org/PR11992 12946 527. https://gcc.gnu.org/PR9365 12947 528. https://gcc.gnu.org/PR10392 12948 529. https://gcc.gnu.org/PR11322 12949 530. https://gcc.gnu.org/PR13069 12950 531. https://gcc.gnu.org/PR13302 12951 532. https://gcc.gnu.org/PR13585 12952 533. https://gcc.gnu.org/PR8916 12953 534. https://gcc.gnu.org/PR11576 12954 535. https://gcc.gnu.org/PR13122 12955 536. https://gcc.gnu.org/PR13256 12956 537. https://gcc.gnu.org/PR13373 12957 538. https://gcc.gnu.org/PR12561 12958 539. https://gcc.gnu.org/PR6243 12959 540. https://gcc.gnu.org/PR11397 12960 541. https://gcc.gnu.org/PR12505 12961 542. https://gcc.gnu.org/PR13150 12962 543. https://gcc.gnu.org/PR12666 12963 544. https://gcc.gnu.org/PR12969 12964 545. https://gcc.gnu.org/PR10819 12965 546. https://gcc.gnu.org/PR11612 12966 547. https://gcc.gnu.org/PR13211 12967 548. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.4 12968 549. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.5 12969 550. https://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.6 12970 551. https://gcc.gnu.org/onlinedocs/ 12971 552. mailto:gcc-help (a] gcc.gnu.org 12972 553. mailto:gcc (a] gcc.gnu.org 12973 554. https://gcc.gnu.org/lists.html 12974 555. http://www.fsf.org/ 12975 556. https://gcc.gnu.org/about.html 12976 557. http://validator.w3.org/check/referer 12977 ====================================================================== 12978 http://gcc.gnu.org/gcc-3.2/index.html 12979 GCC 3.2 Release Series 12980 12981 April 25, 2003 12982 12983 The [1]GNU project and the GCC developers are pleased to announce the 12984 release of GCC 3.2.3. 12985 12986 The purpose of the GCC 3.2 release series is to provide a stable 12987 platform for OS distributors to use building their next releases. A 12988 primary objective was to stabilize the C++ ABI; we believe that the 12989 interface to the compiler and the C++ standard library are now 12990 relatively stable. 12991 12992 Be aware that C++ code compiled by GCC 3.2.x will (in general) not 12993 interoperate with code compiled by GCC 3.1.1 or earlier. 12994 12995 Please refer to our [2]detailed list of news, caveats, and bug-fixes 12996 for further information. 12997 12998 This release series is no longer maintained. 12999 13000 Release History 13001 13002 GCC 3.2.3 13003 April 25, 2003 ([3]changes) 13004 13005 GCC 3.2.2 13006 February 5, 2003 ([4]changes) 13007 13008 GCC 3.2.1 13009 November 19, 2002 ([5]changes) 13010 13011 GCC 3.2 13012 August 14, 2002 ([6]changes) 13013 13014 References and Acknowledgements 13015 13016 GCC used to stand for the GNU C Compiler, but since the compiler 13017 supports several other languages aside from C, it now stands for the 13018 GNU Compiler Collection. 13019 13020 A list of [7]successful builds is updated as new information becomes 13021 available. 13022 13023 The GCC developers would like to thank the numerous people that have 13024 contributed new features, improvements, bug fixes, and other changes as 13025 well as test results to GCC. This [8]amazing group of volunteers is 13026 what makes GCC successful. 13027 13028 For additional information about GCC please refer to the [9]GCC project 13029 web site or contact the [10]GCC development mailing list. 13030 13031 To obtain GCC please use [11]our mirror sites, or our CVS server. 13032 13033 13034 For questions related to the use of GCC, please consult these web 13035 pages and the [12]GCC manuals. If that fails, the 13036 [13]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 13037 web pages and the development of GCC are welcome on our developer 13038 list at [14]gcc (a] gcc.gnu.org. All of [15]our lists have public 13039 archives. 13040 13041 Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and 13042 distribution of this entire article is permitted in any medium, 13043 provided this notice is preserved. 13044 13045 These pages are [17]maintained by the GCC team. Last modified 13046 2016-09-30[18]. 13047 13048 References 13049 13050 1. http://www.gnu.org/ 13051 2. http://gcc.gnu.org/gcc-3.2/changes.html 13052 3. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3 13053 4. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.2 13054 5. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.1 13055 6. http://gcc.gnu.org/gcc-3.2/changes.html#3.2 13056 7. http://gcc.gnu.org/gcc-3.2/buildstat.html 13057 8. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 13058 9. http://gcc.gnu.org/index.html 13059 10. mailto:gcc (a] gcc.gnu.org 13060 11. http://gcc.gnu.org/mirrors.html 13061 12. https://gcc.gnu.org/onlinedocs/ 13062 13. mailto:gcc-help (a] gcc.gnu.org 13063 14. mailto:gcc (a] gcc.gnu.org 13064 15. https://gcc.gnu.org/lists.html 13065 16. http://www.fsf.org/ 13066 17. https://gcc.gnu.org/about.html 13067 18. http://validator.w3.org/check/referer 13068 ====================================================================== 13069 http://gcc.gnu.org/gcc-3.2/changes.html 13070 GCC 3.2 Release Series 13071 Changes, New Features, and Fixes 13072 13073 The latest release in the 3.2 release series is [1]GCC 3.2.3. 13074 13075 Caveats and New Features 13076 13077 Caveats 13078 13079 * The C++ compiler does not correctly zero-initialize 13080 pointers-to-data members. You must explicitly initialize them. For 13081 example: int S::*m(0); will work, but depending on 13082 default-initialization to zero will not work. This bug cannot be 13083 fixed in GCC 3.2 without inducing unacceptable risks. It will be 13084 fixed in GCC 3.3. 13085 * This GCC release is based on the GCC 3.1 sourcebase, and thus has 13086 all the [2]changes in the GCC 3.1 series. In addition, GCC 3.2 has 13087 a number of C++ ABI fixes which make its C++ compiler generate 13088 binary code which is incompatible with the C++ compilers found in 13089 earlier GCC releases, including GCC 3.1 and GCC 3.1.1. 13090 13091 Frontend Enhancements 13092 13093 C/C++/Objective-C 13094 13095 * The method of constructing the list of directories to be searched 13096 for header files has been revised. If a directory named by a -I 13097 option is a standard system include directory, the option is 13098 ignored to ensure that the default search order for system 13099 directories and the special treatment of system header files are 13100 not defeated. 13101 * The C and Objective-C compilers no longer accept the "Naming Types" 13102 extension (typedef foo = bar); it was already unavailable in C++. 13103 Code which uses it will need to be changed to use the "typeof" 13104 extension instead: typedef typeof(bar) foo. (We have removed this 13105 extension without a period of deprecation because it has caused the 13106 compiler to crash since version 3.0 and no one noticed until very 13107 recently. Thus we conclude it is not in widespread use.) 13108 13109 C++ 13110 13111 * GCC 3.2 fixed serveral differences between the C++ ABI implemented 13112 in GCC and the multi-vendor standard, but more have been found 13113 since the release. 3.2.1 adds a new warning, -Wabi, to warn about 13114 code which is affected by these bugs. We will fix these bugs in 13115 some future release, once we are confident that all have been 13116 found; until then, it is our intention to make changes to the ABI 13117 only if they are necessary for correct compilation of C++, as 13118 opposed to conformance to the ABI documents. 13119 * For details on how to build an ABI compliant compiler for GNU/Linux 13120 systems, check the [3]common C++ ABI page. 13121 13122 New Targets and Target Specific Improvements 13123 13124 IA-32 13125 13126 * Fixed a number of bugs in SSE and MMX intrinsics. 13127 * Fixed common compiler crashes with SSE instruction set enabled 13128 (implied by -march=pentium3, pentium4, athlon-xp) 13129 * __m128 and __m128i is not 128bit aligned when used in structures. 13130 13131 x86-64 13132 13133 * A bug whereby the compiler could generate bad code for bzero has 13134 been fixed. 13135 * ABI fixes (implying ABI incompatibilities with previous version in 13136 some corner cases) 13137 * Fixed prefetch code generation 13138 __________________________________________________________________ 13139 13140 GCC 3.2.3 13141 13142 3.2.3 is a bug fix release only; there are no new features that were 13143 not present in GCC 3.2.2. 13144 13145 Bug Fixes 13146 13147 This section lists the problem reports (PRs) from GCC's bug tracking 13148 system that are known to be fixed in the 3.2.3 release. This list might 13149 not be complete (that is, it is possible that some PRs that have been 13150 fixed are not listed here), and some of the titles have been changed to 13151 make them more clear. 13152 13153 Internal Compiler Errors (multi-platform) 13154 13155 * [4]3782: (c++) -quiet -fstats produces a segmentation fault in 13156 cc1plus 13157 * [5]6440: (c++) template specializations cause ICE 13158 * [6]7050: (c++) ICE on: (i ? get_string() : throw) 13159 * [7]7741: ICE on conflicting types (make_decl_rtl in varasm.c) 13160 * [8]7982: (c++) ICE due to infinite recursion (using STL set) 13161 * [9]8068: exceedingly high (infinite) memory usage 13162 * [10]8178: ICE with __builtin_ffs 13163 * [11]8396: ICE in copy_to_mode_reg, in explow.c 13164 * [12]8674: (c++) ICE in cp_expr_size, in cp/cp-lang.c 13165 * [13]9768: ICE when optimizing inline code at -O2 13166 * [14]9798: (c++) Infinite recursion (segfault) in 13167 cp/decl.c:push_using_directive with recursive using directives 13168 * [15]9799: mismatching structure initializer with nested flexible 13169 array member: ICE 13170 * [16]9928: ICE on duplicate enum declaration 13171 * [17]10114: ICE in mem_loc_descriptor, in dwarf2out.c (affects 13172 sparc, alpha) 13173 * [18]10352: ICE in find_reloads_toplev 13174 * [19]10336: ICE with -Wunreachable-code 13175 13176 C/optimizer bugs: 13177 13178 * [20]8224: Incorrect joining of signed and unsigned division 13179 * [21]8613: -O2 produces wrong code with builtin strlen and 13180 postincrements 13181 * [22]8828: gcc reports some code is unreachable when it is not 13182 * [23]9226: GCSE breaking argument passing 13183 * [24]9853: miscompilation of non-constant structure initializer 13184 * [25]9797: C99-style struct initializers are miscompiled 13185 * [26]9967: Some standard C function calls should not be replaced 13186 when optimizing for size 13187 * [27]10116: ce2: invalid merge of join_bb in the context of switch 13188 statements 13189 * [28]10171: wrong code for inlined function 13190 * [29]10175: -Wunreachable-code doesn't work for single lines 13191 13192 C++ compiler and library: 13193 13194 * [30]8316: Confusing diagnostic for code that misuses conversion 13195 operators 13196 * [31]9169: filebuf output fails if codecvt<>::out returns noconv 13197 * [32]9420: incomplete type incorrectly reported 13198 * [33]9459: typeof in return type specification of template not 13199 supported 13200 * [34]9507: filebuf::open handles ios_base::ate incorrectly 13201 * [35]9538: Out-of-bounds memory access in streambuf::sputbackc 13202 * [36]9602: Total confusion about template/friend/virtual/abstract 13203 * [37]9993: destructor not called for local object created within and 13204 returned from infinite loop 13205 * [38]10167: ieee_1003.1-2001 locale specialisations on a glibc-2.3.2 13206 system 13207 13208 Java compiler and library: 13209 13210 * [39]9652: libgcj build fails on irix6.5.1[78] 13211 * [40]10144: gas on solaris complains about bad .stabs lines for 13212 java, native as unaffected 13213 13214 x86-specific (Intel/AMD): 13215 13216 * [41]8746: gcc miscompiles Linux kernel ppa driver on x86 13217 * [42]9888: -mcpu=k6 -Os produces out of range loop instructions 13218 * [43]9638: Cross-build for target i386-elf and i586-pc-linux-gnu 13219 failed 13220 * [44]9954: Cross-build for target i586-pc-linux-gnu (--with-newlib) 13221 failed 13222 13223 SPARC-specific: 13224 13225 * [45]7784: [Sparc] ICE in extract_insn, in recog.c 13226 * [46]7796: sparc extra failure with -m64 on execute/930921-1.c in 13227 unroll.c 13228 * [47]8281: ICE when compiling with -O2 -fPIC for Ultrasparc 13229 * [48]8366: [Sparc] C testsuite failure with -m64 -fpic -O in 13230 execute/loop-2d.c 13231 * [49]8726: gcc -O2 miscompiles Samba 2.2.7 on 32-bit sparc 13232 * [50]9414: Scheduling bug on Ultrasparc 13233 * [51]10067: GCC-3.2.2 outputs invalid asm on sparc64 13234 13235 m68k-specific: 13236 13237 * [52]7248: broken "inclusive or" code 13238 * [53]8343: m68k-elf/rtems ICE at instantiate_virtual_regs_1 13239 13240 PowerPC-specific: 13241 13242 * [54]9732: Wrong code with -O2 -fPIC 13243 * [55]10073: ICE: powerpc cannot split insn 13244 13245 Alpha-specific: 13246 13247 * [56]7702: optimization problem on a DEC alpha under OSF1 13248 * [57]9671: gcc.3.2.2 does not build on a HP Tru64 Unix v5.1B system 13249 13250 HP-specific: 13251 13252 * [58]8694: <string> breaks <ctype.h> on HP-UX 10.20 (DUP: 9275) 13253 * [59]9953: (ada) gcc 3.2.x can't build 3.3-branch ada on HP-UX 10 13254 (missing symbol) 13255 * [60]10271: Floating point args don't get reloaded across function 13256 calls with -O2 13257 13258 MIPS specific: 13259 13260 * [61]6362: mips-irix6 gcc-3.1 C testsuite failure with -mips4 in 13261 compile/920501-4.c 13262 13263 CRIS specific: 13264 13265 * [62]10377: gcc-3.2.2 creates bad assembler code for cris 13266 13267 Miscellaneous and minor bugs: 13268 13269 * [63]6955: collect2 says "core dumped" when there is no core 13270 __________________________________________________________________ 13271 13272 GCC 3.2.2 13273 13274 Beginning with 3.2.2, GCC's Makefile suite supports redirection of make 13275 install by means of the DESTDIR variable. Parts of the GCC tree have 13276 featured that support long before, but now it is available even from 13277 the top level. 13278 13279 Other than that, GCC 3.2.2 is a bug fix release only; there are no new 13280 features that were not present in GCC 3.2.1. 13281 13282 Bug Fixes 13283 13284 On the following i386-based systems GCC 3.2.1 broke the C ABI wrt. 13285 functions returning structures: Cygwin, FreeBSD (GCC 3.2.1 as shipped 13286 with FreeBSD 5.0 does not have this problem), Interix, a.out-based 13287 GNU/Linux and NetBSD, OpenBSD, and Darwin. GCC 3.2.2 reverts this ABI 13288 change, and thus restores ABI-compatibility with previous releases 13289 (except GCC 3.2.1) on these platforms. 13290 13291 This section lists the problem reports (PRs) from GCC's bug tracking 13292 system that are known to be fixed in the 3.2.2 release. This list might 13293 not be complete (that is, it is possible that some PRs that have been 13294 fixed are not listed here) and some of the titles have been changed to 13295 make them more clear. 13296 13297 Internal Compiler Errors (multi-platform) 13298 13299 * [64]5919: (c++) ICE when passing variable array to template 13300 function 13301 * [65]7129: (c++) ICE with min/max assignment operators (<?= and >?=) 13302 * [66]7507: ICE with -O2 when address of called function is a 13303 complicated expression 13304 * [67]7622: ICE with nested inline functions if function's address is 13305 taken 13306 * [68]7681: (fortran) ICE in compensate_edge, in reg-stack.c (also PR 13307 [69]9258) 13308 * [70]8031: (c++) ICE in code comparing typeids and casting from 13309 virtual base 13310 * [71]8275: ICE in simplify_subreg 13311 * [72]8332: (c++) builtin strlen/template interaction causes ICE 13312 * [73]8372: (c++) ICE on explicit call of destructor 13313 * [74]8439: (c, not c++) empty struct causes ICE 13314 * [75]8442: (c++) ICE with nested template classes 13315 * [76]8518: ICE when compiling mplayer ("extern inline" issue) 13316 * [77]8615: (c++) ICE with out-of-range character constant template 13317 argument 13318 * [78]8663: (c++) ICE in cp_expr_size, at cp-lang.c:307 13319 * [79]8799: (c++) ICE: error reporting routines re-entered 13320 * [80]9328: (c++) ICE with typeof(X) for overloaded X 13321 * [81]9465: (preprocessor) cpp -traditional ICE on null bytes 13322 13323 C++ (compiler and library) bugs 13324 13325 * [82]47: scoping in nested classes is broken 13326 * [83]6745: problems with iostream rdbuf() member function 13327 * [84]8214: conversion from const char* const to char* sometimes 13328 accepted illegally 13329 * [85]8493: builtin strlen and overload resolution (same bug as 13330 [86]8332) 13331 * [87]8503: strange behaviour of function types 13332 * [88]8727: compiler confused by inheritance from an anonymous struct 13333 * [89]7445: poor performance of std::locale::classic() in 13334 multi-threaded applications 13335 * [90]8230: mishandling of overflow in vector<T>::resize 13336 * [91]8399: sync_with_stdio(false) breaks unformatted input 13337 * [92]8662: illegal access of private member of unnamed class is 13338 accepted 13339 * [93]8707: "make distclean" fails in libstdc++-v3 directory 13340 * [94]8708: __USE_MALLOC doesn't work 13341 * [95]8790: Use of non-thread-safe strtok in src/localename.cc 13342 * [96]8887: Bug in date formats with --enable-clocale=generic 13343 * [97]9076: Call Frame Instructions are not handled correctly during 13344 unwind operation 13345 * [98]9151: std::setprecision limited to 16 digits when outputting a 13346 double to a stream 13347 * [99]9168: codecvt<char, char, mbstate_t> overwrites output buffers 13348 * [100]9269: libstdc++ headers: explicit specialization of function 13349 must precede its first use 13350 * [101]9322: return value of basic_streambuf<>::getloc affected by 13351 locale::global 13352 * [102]9433: segfault in runtime support for dynamic_cast 13353 13354 C and optimizer bugs 13355 13356 * [103]8032: GCC incorrectly initializes static structs that have 13357 flexible arrays 13358 * [104]8639: simple arithmetic expression broken 13359 * [105]8794: optimization improperly eliminates certain expressions 13360 * [106]8832: traditional "asm volatile" code is illegally optimized 13361 * [107]8988: loop optimizer bug: with -O2, code is generated that 13362 segfaults (found on i386, bug present for all platforms) 13363 * [108]9492: structure copy clobbers subsequent stores to structure 13364 13365 Objective-C bugs 13366 13367 * [109]9267: Objective-C parser won't build with newer bison versions 13368 (e.g. 1.875) 13369 13370 Ada bugs 13371 13372 * [110]8344: Ada build problem due to conflict between gcc/final.o, 13373 gcc/ada/final.o 13374 13375 Preprocessor bugs 13376 13377 * [111]8524: _Pragma within macros is improperly expanded 13378 * [112]8880: __WCHAR_TYPE__ macro incorrectly set to "long int" with 13379 -fshort-wchar 13380 13381 ARM-specific 13382 13383 * [113]9090: arm ICE with >= -O2; regression from gcc-2.95 13384 13385 x86-specific (Intel/AMD) 13386 13387 * [114]8588: ICE in extract_insn, at recog.c:NNNN (shift instruction) 13388 * [115]8599: loop unroll bug with -march=k6-3 13389 * [116]9506: ABI breakage in structure return (affects BSD and 13390 Cygwin, but not GNU/Linux) 13391 13392 FreeBSD 5.0 specific 13393 13394 * [117]9484: GCC 3.2.1 Bootstrap failure on FreeBSD 5.0 13395 13396 RTEMS-specific 13397 13398 * [118]9292: hppa1.1-rtems configurery problems 13399 * [119]9293: [m68k-elf/rtems] config/m68k/t-crtstuff bug 13400 * [120]9295: [mips-rtems] config/mips/rtems.h init/fini issue 13401 * [121]9296: gthr-rtems regression 13402 * [122]9316: powerpc-rtems: extending multilibs 13403 13404 HP-PA specific 13405 13406 * [123]9493: ICE with -O2 when building a simple function 13407 13408 Documentation 13409 13410 * [124]7341: hyperlink to gcov in GCC documentation doesn't work 13411 * [125]8947: Please add a warning about "-malign-double" in docs 13412 * [126]7448, [127]8882: typo cleanups 13413 __________________________________________________________________ 13414 13415 GCC 3.2.1 13416 13417 3.2.1 adds a new warning, -Wabi. This option warns when GNU C++ 13418 generates code that is known not to be binary-compatible with the 13419 vendor-neutral ia32/ia64 ABI. Please consult the GCC manual, included 13420 in the distribution, for details. 13421 13422 This release also removes an old GCC extension, "naming types", and the 13423 documentation now directs users to use a different GCC extension, 13424 __typeof__, instead. The feature had evidently been broken for a while. 13425 13426 Otherwise, 3.2.1 is a bug fix release only; other than bug fixes and 13427 the new warning there are no new features that were not present in GCC 13428 3.2. 13429 13430 In addition, the previous fix for [128]PR 7445 (poor performance of 13431 std::locale::classic() in multi-threaded applications) was reverted 13432 ("unfixed"), because the "fix" was not thread-safe. 13433 13434 Bug Fixes 13435 13436 This section lists the problem reports (PRs) from GCC's bug tracking 13437 system that are known to be fixed in the 3.2.1 release. This list might 13438 not be complete (that is, it is possible that some PRs that have been 13439 fixed are not listed here). As you can see, the number of bug fixes is 13440 quite large, so it is strongly recommended that users of earlier GCC 13441 3.x releases upgrade to GCC 3.2.1. 13442 13443 Internal Compiler Errors (multi-platform) 13444 13445 * [129]2521: (c++) ICE in build_ptrmemfunc, in cp/typeck.c 13446 * [130]5661: (c++) ICE instantiating template on array of unknown 13447 size (bad code) 13448 * [131]6419: (c++) ICE in make_decl_rtl for "longest" attribute on 13449 64-bit platforms 13450 * [132]6994: (c++) ICE in find_function_data 13451 * [133]7150: preprocessor: GCC -dM -E gives an ICE 13452 * [134]7160: ICE when optimizing branches without a return value 13453 * [135]7228: (c++) ICE when using member template and template 13454 function 13455 * [136]7266: (c++) ICE with -pedantic on missing typename 13456 * [137]7353: ICE from use of "Naming Types" extension, see above 13457 * [138]7411: ICE in instantiate_virtual_regs_1, in function.c 13458 * [139]7478: (c++) ICE on static_cast inside template 13459 * [140]7526: preprocessor core dump when _Pragma implies #pragma 13460 dependency 13461 * [141]7721: (c++) ICE on simple (but incorrect) template ([142]7803 13462 is a duplicate) 13463 * [143]7754: (c++) ICE on union with template parameter 13464 * [144]7788: (c++) redeclaring a definition as an incomplete class 13465 causes ICE 13466 * [145]8031: (c++) ICE in comptypes, in cp/typeck.c 13467 * [146]8055: preprocessor dies with SIG11 when building FreeBSD 13468 kernel 13469 * [147]8067: (c++) ICE due to mishandling of __FUNCTION__ and related 13470 variables 13471 * [148]8134: (c++) ICE in force_store_init_value on legal code 13472 * [149]8149: (c++) ICE on incomplete type 13473 * [150]8160: (c++) ICE in build_modify_expr, in cp/typeck.c: array 13474 initialization 13475 13476 C++ (compiler and library) bugs 13477 13478 * [151]5607: No pointer adjustment in covariant return types 13479 * [152]6579: Infinite loop with statement expressions in member 13480 initialization 13481 * [153]6803: Default copy constructor bug in GCC 3.1 13482 * [154]7176: g++ confused by friend and static member with same name 13483 * [155]7188: Segfault with template class and recursive (incorrect) 13484 initializer list 13485 * [156]7306: Regression: GCC 3.x fails to compile code with virtual 13486 inheritance if a method has a variable number of arguments 13487 * [157]7461: ctype<char>::classic_table() returns offset array on 13488 Cygwin 13489 * [158]7524: f(const float arg[3]) fails 13490 * [159]7584: Erroneous ambiguous base error on using declaration 13491 * [160]7676: Member template overloading problem 13492 * [161]7679: infinite loop when a right parenthesis is missing 13493 * [162]7811: default locale not taken from environment 13494 * [163]7961: compare( char *) implemented incorrectly in 13495 basic_string<> 13496 * [164]8071: basic_ostream::operator<<(streambuf*) loops forever if 13497 streambuf::underflow() leaves gptr() NULL (dups: [165]8127, 13498 [166]6745) 13499 * [167]8096: deque::at() throws std::range_error instead of 13500 std::out_of_range 13501 * [168]8127: cout << cin.rdbuf() infinite loop 13502 * [169]8218: Excessively large memory consumed for classes with large 13503 array members 13504 * [170]8287: GCC 3.2: Destructor called for non-constructed local 13505 object 13506 * [171]8347: empty vector range used in string construction causes 13507 core dump 13508 * [172]8348: fail() flag is set in istringstream when eof() flag is 13509 set 13510 * [173]8391: regression: infinite loop in cp/decl2.c(finish_file) 13511 13512 C and optimizer bugs 13513 13514 * [174]6627: -fno-align-functions doesn't seem to disable function 13515 alignment 13516 * [175]6631: life_analysis misoptimizes code to initialize fields of 13517 a structure 13518 * [176]7102: unsigned char division results in floating exception 13519 * [177]7120: Run once loop should *always* be unrolled 13520 (pessimization) 13521 * [178]7209: Bug involving array referencing and ?: operator 13522 * [179]7515: invalid inlining of global function with -O3 13523 * [180]7814: incorrect scheduling for glibc-2.2.92 strcpy test 13524 * [181]8467: bug in sibling call optimization 13525 13526 Preprocessor bugs 13527 13528 * [182]4890: incorrect line markers from the traditional preprocessor 13529 * [183]7357: -M option omits system headers files (making it the same 13530 as -MM) 13531 * [184]7358: Changes to Sun's make Dependencies 13532 * [185]7602: C++ header files found in CPLUS_INCLUDE_PATH treated as 13533 C headers 13534 * [186]7862: Interrupting GCC -MD removes .d file but not .o 13535 * [187]8190: Failed compilation deletes -MD dependency file 13536 * [188]8524: _Pragma within macro is improperly expanded 13537 13538 x86 specific (Intel/AMD) 13539 13540 * [189]5351: (i686-only) function pass-by-value structure copy 13541 corrupts stack ([190]7591 is a duplicate) 13542 * [191]6845, [192]7034, [193]7124, [194]7174: ICE's with 13543 -march=pentium3/pentium2/athlon (these are all the same underlying 13544 bug, in MMX register use) 13545 * [195]7134, [196]7375, [197]7390: ICE with -march=athlon (maybe same 13546 as above?) 13547 * [198]6890: xmmintrin.h, _MM_TRANSPOSE4_PS is broken 13548 * [199]6981: wrong code in 64-bit manipulation on x86 13549 * [200]7242: GCC -mcpu=pentium[23] doesn't define __tune_pentiumpro__ 13550 macro 13551 * [201]7396: ix86: cmpgt_ss, cmpge_ss, cmpngt_ss, and cmpnge_ss SSE 13552 intrinsics are broken 13553 * [202]7630: GCC 3.2 breaks on Mozilla 1.0's JS sources with 13554 -march=pentium4 13555 * [203]7693: Typo in i386 mmintrin.h header 13556 * [204]7723: ICE - Pentium3 sse - GCC 3.2 13557 * [205]7951: ICE on -march=pentium4 -O2 -mfpmath=sse 13558 * [206]8146: (i686 only) gcc 3.2 miscompiles gcc 2.95.3 13559 13560 PowerPC specific 13561 13562 * [207]5967: GCC bug when profiling nested functions on powerpc 13563 * [208]6984: wrong code generated with -O2, -O3, -Os for do-while 13564 loop on PowerPC 13565 * [209]7114: PowerPC: ICE building strcoll.op from glibc-2.2.5 13566 * [210]7130: miscompiled code for GCC-3.1 on 13567 powerpc-unknown-linux-gnu with -funroll-all-loops 13568 * [211]7133: PowerPC ICE: unrecognizable insn 13569 * [212]7380: ICE in extract_insn, at recog.c:2148 13570 * [213]8252: ICE on Altivec code with optimization turned on 13571 * [214]8451: Altivec ICE in GCC 3.2 13572 13573 HP/PA specific 13574 13575 * [215]7250: __ashrdi3 returns wrong value on 32 bit hppa 13576 13577 SPARC specific 13578 13579 * [216]6668: when using --disable-multilib, libgcc_s.so is installed 13580 in the wrong place on sparc-solaris 13581 * [217]7151: ICE when compiling for UltraSPARC 13582 * [218]7335: SPARC: ICE in verify_wide_reg (flow.c:557) with long 13583 double and -O1 13584 * [219]7842: [REGRESSION] SPARC code gen bug 13585 13586 ARM specific 13587 13588 * [220]7856: [arm] invalid offset in constant pool reference 13589 * [221]7967: optimization produces wrong code (ARM) 13590 13591 Alpha specific 13592 13593 * [222]7374: __builtin_fabsl broken on alpha 13594 13595 IBM s390 specific 13596 13597 * [223]7370: ICE in fixup_var_refs_1 on s390x 13598 * [224]7409: loop optimization bug on s390x-linux-gnu 13599 * [225]8232: s390x: ICE when using bcmp with int length argument 13600 13601 SCO specific 13602 13603 * [226]7623: SCO OpenServer build fails with machmode.def: undefined 13604 symbol: BITS_PER_UNIT 13605 13606 m68k/Coldfire specific 13607 13608 * [227]8314: crtbegin, crtend need to be multilib'ed for this 13609 platform 13610 13611 Documentation 13612 13613 * [228]761: Document some undocumented options 13614 * [229]5610: Fix documentation about invoking SSE instructions 13615 (-mfpmath=sse) 13616 * [230]7484: List -Wmissing-declarations as C-only option 13617 * [231]7531: -mcmodel not documented for x86-64 13618 * [232]8120: Update documentation of bad use of ## 13619 __________________________________________________________________ 13620 13621 GCC 3.2 13622 13623 3.2 is a small bug fix release, but there is a change to the 13624 application binary interface (ABI), hence the change to the second part 13625 of the version number. 13626 13627 The main purpose of the 3.2 release is to correct a couple of problems 13628 in the C++ ABI, with the intention of providing a stable interface 13629 going forward. Accordingly, 3.2 is only a small change to 3.1.1. 13630 13631 Bug Fixes 13632 13633 C++ 13634 13635 * [233]7320: g++ 3.2 relocation problem 13636 * [234]7470: vtable: virtual function pointers not in declaration 13637 order 13638 13639 libstdc++ 13640 13641 * [235]6410: Trouble with non-ASCII monetary symbols and wchar_t 13642 * [236]6503, [237]6642, [238]7186: Problems with comparing or 13643 subtracting various types of const and non-const iterators 13644 * [239]7216: ambiguity with basic_iostream::traits_type 13645 * [240]7220: problem with basic_istream::ignore(0,delimiter) 13646 * [241]7222: locale::operator==() doesn't work on std::locale("") 13647 * [242]7286: placement operator delete issue 13648 * [243]7442: cxxabi.h does not match the C++ ABI 13649 * [244]7445: poor performance of std::locale::classic() in 13650 multi-threaded applications 13651 13652 x86-64 specific 13653 13654 * [245]7291: off-by-one in generated inline bzero code for x86-64 13655 13656 13657 For questions related to the use of GCC, please consult these web 13658 pages and the [246]GCC manuals. If that fails, the 13659 [247]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 13660 web pages and the development of GCC are welcome on our developer 13661 list at [248]gcc (a] gcc.gnu.org. All of [249]our lists have public 13662 archives. 13663 13664 Copyright (C) [250]Free Software Foundation, Inc. Verbatim copying and 13665 distribution of this entire article is permitted in any medium, 13666 provided this notice is preserved. 13667 13668 These pages are [251]maintained by the GCC team. Last modified 13669 2016-01-30[252]. 13670 13671 References 13672 13673 1. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3 13674 2. http://gcc.gnu.org/gcc-3.1/changes.html 13675 3. http://gcc.gnu.org/gcc-3.2/c++-abi.html 13676 4. https://gcc.gnu.org/PR3782 13677 5. https://gcc.gnu.org/PR6440 13678 6. https://gcc.gnu.org/PR7050 13679 7. https://gcc.gnu.org/PR7741 13680 8. https://gcc.gnu.org/PR7982 13681 9. https://gcc.gnu.org/PR8068 13682 10. https://gcc.gnu.org/PR8178 13683 11. https://gcc.gnu.org/PR8396 13684 12. https://gcc.gnu.org/PR8674 13685 13. https://gcc.gnu.org/PR9768 13686 14. https://gcc.gnu.org/PR9798 13687 15. https://gcc.gnu.org/PR9799 13688 16. https://gcc.gnu.org/PR9928 13689 17. https://gcc.gnu.org/PR10114 13690 18. https://gcc.gnu.org/PR10352 13691 19. https://gcc.gnu.org/PR10336 13692 20. https://gcc.gnu.org/PR8224 13693 21. https://gcc.gnu.org/PR8613 13694 22. https://gcc.gnu.org/PR8828 13695 23. https://gcc.gnu.org/PR9226 13696 24. https://gcc.gnu.org/PR9853 13697 25. https://gcc.gnu.org/PR9797 13698 26. https://gcc.gnu.org/PR9967 13699 27. https://gcc.gnu.org/PR10116 13700 28. https://gcc.gnu.org/PR10171 13701 29. https://gcc.gnu.org/PR10175 13702 30. https://gcc.gnu.org/PR8316 13703 31. https://gcc.gnu.org/PR9169 13704 32. https://gcc.gnu.org/PR9420 13705 33. https://gcc.gnu.org/PR9459 13706 34. https://gcc.gnu.org/PR9507 13707 35. https://gcc.gnu.org/PR9538 13708 36. https://gcc.gnu.org/PR9602 13709 37. https://gcc.gnu.org/PR9993 13710 38. https://gcc.gnu.org/PR10167 13711 39. https://gcc.gnu.org/PR9652 13712 40. https://gcc.gnu.org/PR10144 13713 41. https://gcc.gnu.org/PR8746 13714 42. https://gcc.gnu.org/PR9888 13715 43. https://gcc.gnu.org/PR9638 13716 44. https://gcc.gnu.org/PR9954 13717 45. https://gcc.gnu.org/PR7784 13718 46. https://gcc.gnu.org/PR7796 13719 47. https://gcc.gnu.org/PR8281 13720 48. https://gcc.gnu.org/PR8366 13721 49. https://gcc.gnu.org/PR8726 13722 50. https://gcc.gnu.org/PR9414 13723 51. https://gcc.gnu.org/PR10067 13724 52. https://gcc.gnu.org/PR7248 13725 53. https://gcc.gnu.org/PR8343 13726 54. https://gcc.gnu.org/PR9732 13727 55. https://gcc.gnu.org/PR10073 13728 56. https://gcc.gnu.org/PR7702 13729 57. https://gcc.gnu.org/PR9671 13730 58. https://gcc.gnu.org/PR8694 13731 59. https://gcc.gnu.org/PR9953 13732 60. https://gcc.gnu.org/PR10271 13733 61. https://gcc.gnu.org/PR6362 13734 62. https://gcc.gnu.org/PR10377 13735 63. https://gcc.gnu.org/PR6955 13736 64. https://gcc.gnu.org/PR5919 13737 65. https://gcc.gnu.org/PR7129 13738 66. https://gcc.gnu.org/PR7507 13739 67. https://gcc.gnu.org/PR7622 13740 68. https://gcc.gnu.org/PR7681 13741 69. https://gcc.gnu.org/PR9528 13742 70. https://gcc.gnu.org/PR8031 13743 71. https://gcc.gnu.org/PR8275 13744 72. https://gcc.gnu.org/PR8332 13745 73. https://gcc.gnu.org/PR8372 13746 74. https://gcc.gnu.org/PR8439 13747 75. https://gcc.gnu.org/PR8442 13748 76. https://gcc.gnu.org/PR8518 13749 77. https://gcc.gnu.org/PR8615 13750 78. https://gcc.gnu.org/PR8663 13751 79. https://gcc.gnu.org/PR8799 13752 80. https://gcc.gnu.org/PR9328 13753 81. https://gcc.gnu.org/PR9465 13754 82. https://gcc.gnu.org/PR47 13755 83. https://gcc.gnu.org/PR6745 13756 84. https://gcc.gnu.org/PR8214 13757 85. https://gcc.gnu.org/PR8493 13758 86. https://gcc.gnu.org/PR8332 13759 87. https://gcc.gnu.org/PR8503 13760 88. https://gcc.gnu.org/PR8727 13761 89. https://gcc.gnu.org/PR7445 13762 90. https://gcc.gnu.org/PR8230 13763 91. https://gcc.gnu.org/PR8399 13764 92. https://gcc.gnu.org/PR8662 13765 93. https://gcc.gnu.org/PR8707 13766 94. https://gcc.gnu.org/PR8708 13767 95. https://gcc.gnu.org/PR8790 13768 96. https://gcc.gnu.org/PR8887 13769 97. https://gcc.gnu.org/PR9076 13770 98. https://gcc.gnu.org/PR9151 13771 99. https://gcc.gnu.org/PR9168 13772 100. https://gcc.gnu.org/PR9269 13773 101. https://gcc.gnu.org/PR9322 13774 102. https://gcc.gnu.org/PR9433 13775 103. https://gcc.gnu.org/PR8032 13776 104. https://gcc.gnu.org/PR8639 13777 105. https://gcc.gnu.org/PR8794 13778 106. https://gcc.gnu.org/PR8832 13779 107. https://gcc.gnu.org/PR8988 13780 108. https://gcc.gnu.org/PR9492 13781 109. https://gcc.gnu.org/PR9267 13782 110. https://gcc.gnu.org/PR8344 13783 111. https://gcc.gnu.org/PR8524 13784 112. https://gcc.gnu.org/PR8880 13785 113. https://gcc.gnu.org/PR9090 13786 114. https://gcc.gnu.org/PR8588 13787 115. https://gcc.gnu.org/PR8599 13788 116. https://gcc.gnu.org/PR9506 13789 117. https://gcc.gnu.org/PR9484 13790 118. https://gcc.gnu.org/PR9292 13791 119. https://gcc.gnu.org/PR9293 13792 120. https://gcc.gnu.org/PR9295 13793 121. https://gcc.gnu.org/PR9296 13794 122. https://gcc.gnu.org/PR9316 13795 123. https://gcc.gnu.org/PR9493 13796 124. https://gcc.gnu.org/PR7341 13797 125. https://gcc.gnu.org/PR8947 13798 126. https://gcc.gnu.org/PR7448 13799 127. https://gcc.gnu.org/PR8882 13800 128. https://gcc.gnu.org/PR7445 13801 129. https://gcc.gnu.org/PR2521 13802 130. https://gcc.gnu.org/PR5661 13803 131. https://gcc.gnu.org/PR6419 13804 132. https://gcc.gnu.org/PR6994 13805 133. https://gcc.gnu.org/PR7150 13806 134. https://gcc.gnu.org/PR7160 13807 135. https://gcc.gnu.org/PR7228 13808 136. https://gcc.gnu.org/PR7266 13809 137. https://gcc.gnu.org/PR7353 13810 138. https://gcc.gnu.org/PR7411 13811 139. https://gcc.gnu.org/PR7478 13812 140. https://gcc.gnu.org/PR7526 13813 141. https://gcc.gnu.org/PR7721 13814 142. https://gcc.gnu.org/PR7803 13815 143. https://gcc.gnu.org/PR7754 13816 144. https://gcc.gnu.org/PR7788 13817 145. https://gcc.gnu.org/PR8031 13818 146. https://gcc.gnu.org/PR8055 13819 147. https://gcc.gnu.org/PR8067 13820 148. https://gcc.gnu.org/PR8134 13821 149. https://gcc.gnu.org/PR8149 13822 150. https://gcc.gnu.org/PR8160 13823 151. https://gcc.gnu.org/PR5607 13824 152. https://gcc.gnu.org/PR6579 13825 153. https://gcc.gnu.org/PR6803 13826 154. https://gcc.gnu.org/PR7176 13827 155. https://gcc.gnu.org/PR7188 13828 156. https://gcc.gnu.org/PR7306 13829 157. https://gcc.gnu.org/PR7461 13830 158. https://gcc.gnu.org/PR7524 13831 159. https://gcc.gnu.org/PR7584 13832 160. https://gcc.gnu.org/PR7676 13833 161. https://gcc.gnu.org/PR7679 13834 162. https://gcc.gnu.org/PR7811 13835 163. https://gcc.gnu.org/PR7961 13836 164. https://gcc.gnu.org/PR8071 13837 165. https://gcc.gnu.org/PR8127 13838 166. https://gcc.gnu.org/PR6745 13839 167. https://gcc.gnu.org/PR8096 13840 168. https://gcc.gnu.org/PR8127 13841 169. https://gcc.gnu.org/PR8218 13842 170. https://gcc.gnu.org/PR8287 13843 171. https://gcc.gnu.org/PR8347 13844 172. https://gcc.gnu.org/PR8348 13845 173. https://gcc.gnu.org/PR8391 13846 174. https://gcc.gnu.org/PR6627 13847 175. https://gcc.gnu.org/PR6631 13848 176. https://gcc.gnu.org/PR7102 13849 177. https://gcc.gnu.org/PR7120 13850 178. https://gcc.gnu.org/PR7209 13851 179. https://gcc.gnu.org/PR7515 13852 180. https://gcc.gnu.org/PR7814 13853 181. https://gcc.gnu.org/PR8467 13854 182. https://gcc.gnu.org/PR4890 13855 183. https://gcc.gnu.org/PR7357 13856 184. https://gcc.gnu.org/PR7358 13857 185. https://gcc.gnu.org/PR7602 13858 186. https://gcc.gnu.org/PR7862 13859 187. https://gcc.gnu.org/PR8190 13860 188. https://gcc.gnu.org/PR8524 13861 189. https://gcc.gnu.org/PR5351 13862 190. https://gcc.gnu.org/PR7591 13863 191. https://gcc.gnu.org/PR6845 13864 192. https://gcc.gnu.org/PR7034 13865 193. https://gcc.gnu.org/PR7124 13866 194. https://gcc.gnu.org/PR7174 13867 195. https://gcc.gnu.org/PR7134 13868 196. https://gcc.gnu.org/PR7375 13869 197. https://gcc.gnu.org/PR7390 13870 198. https://gcc.gnu.org/PR6890 13871 199. https://gcc.gnu.org/PR6981 13872 200. https://gcc.gnu.org/PR7242 13873 201. https://gcc.gnu.org/PR7396 13874 202. https://gcc.gnu.org/PR7630 13875 203. https://gcc.gnu.org/PR7693 13876 204. https://gcc.gnu.org/PR7723 13877 205. https://gcc.gnu.org/PR7951 13878 206. https://gcc.gnu.org/PR8146 13879 207. https://gcc.gnu.org/PR5967 13880 208. https://gcc.gnu.org/PR6984 13881 209. https://gcc.gnu.org/PR7114 13882 210. https://gcc.gnu.org/PR7130 13883 211. https://gcc.gnu.org/PR7133 13884 212. https://gcc.gnu.org/PR7380 13885 213. https://gcc.gnu.org/PR8252 13886 214. https://gcc.gnu.org/PR8451 13887 215. https://gcc.gnu.org/PR7250 13888 216. https://gcc.gnu.org/PR6668 13889 217. https://gcc.gnu.org/PR7151 13890 218. https://gcc.gnu.org/PR7335 13891 219. https://gcc.gnu.org/PR7842 13892 220. https://gcc.gnu.org/PR7856 13893 221. https://gcc.gnu.org/PR7967 13894 222. https://gcc.gnu.org/PR7374 13895 223. https://gcc.gnu.org/PR7370 13896 224. https://gcc.gnu.org/PR7409 13897 225. https://gcc.gnu.org/PR8232 13898 226. https://gcc.gnu.org/PR7623 13899 227. https://gcc.gnu.org/PR8314 13900 228. https://gcc.gnu.org/PR761 13901 229. https://gcc.gnu.org/PR5610 13902 230. https://gcc.gnu.org/PR7484 13903 231. https://gcc.gnu.org/PR7531 13904 232. https://gcc.gnu.org/PR8120 13905 233. https://gcc.gnu.org/PR7320 13906 234. https://gcc.gnu.org/PR7470 13907 235. https://gcc.gnu.org/PR6410 13908 236. https://gcc.gnu.org/PR6503 13909 237. https://gcc.gnu.org/PR6642 13910 238. https://gcc.gnu.org/PR7186 13911 239. https://gcc.gnu.org/PR7216 13912 240. https://gcc.gnu.org/PR7220 13913 241. https://gcc.gnu.org/PR7222 13914 242. https://gcc.gnu.org/PR7286 13915 243. https://gcc.gnu.org/PR7442 13916 244. https://gcc.gnu.org/PR7445 13917 245. https://gcc.gnu.org/PR7291 13918 246. https://gcc.gnu.org/onlinedocs/ 13919 247. mailto:gcc-help (a] gcc.gnu.org 13920 248. mailto:gcc (a] gcc.gnu.org 13921 249. https://gcc.gnu.org/lists.html 13922 250. http://www.fsf.org/ 13923 251. https://gcc.gnu.org/about.html 13924 252. http://validator.w3.org/check/referer 13925 ====================================================================== 13926 http://gcc.gnu.org/gcc-3.1/index.html 13927 GCC 3.1 13928 13929 July 27, 2002 13930 13931 The [1]GNU project and the GCC developers are pleased to announce the 13932 release of GCC 3.1.1. 13933 13934 The links below still apply to GCC 3.1.1. 13935 13936 This release series is no longer maintained. 13937 13938 May 15, 2002 13939 13940 The [2]GNU project and the GCC developers are pleased to announce the 13941 release of GCC 3.1. 13942 13943 GCC used to stand for the GNU C Compiler, but since the compiler 13944 supports several other languages aside from C, it now stands for the 13945 GNU Compiler Collection. 13946 13947 A list of [3]successful builds is updated as new information becomes 13948 available. 13949 13950 The GCC developers would like to thank the numerous people that have 13951 contributed [4]new features, improvements, bug fixes, and other changes 13952 as well as test results to GCC. This [5]amazing group of volunteers is 13953 what makes GCC successful. 13954 13955 For additional information about GCC please refer to the [6]GCC project 13956 web site or contact the [7]GCC development mailing list. 13957 13958 To obtain GCC please use [8]our mirror sites, or our CVS server. 13959 __________________________________________________________________ 13960 13961 13962 For questions related to the use of GCC, please consult these web 13963 pages and the [9]GCC manuals. If that fails, the 13964 [10]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 13965 web pages and the development of GCC are welcome on our developer 13966 list at [11]gcc (a] gcc.gnu.org. All of [12]our lists have public 13967 archives. 13968 13969 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and 13970 distribution of this entire article is permitted in any medium, 13971 provided this notice is preserved. 13972 13973 These pages are [14]maintained by the GCC team. Last modified 13974 2016-09-30[15]. 13975 13976 References 13977 13978 1. http://www.gnu.org/ 13979 2. http://www.gnu.org/ 13980 3. http://gcc.gnu.org/gcc-3.1/buildstat.html 13981 4. http://gcc.gnu.org/gcc-3.1/changes.html 13982 5. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 13983 6. http://gcc.gnu.org/index.html 13984 7. mailto:gcc (a] gcc.gnu.org 13985 8. http://gcc.gnu.org/mirrors.html 13986 9. https://gcc.gnu.org/onlinedocs/ 13987 10. mailto:gcc-help (a] gcc.gnu.org 13988 11. mailto:gcc (a] gcc.gnu.org 13989 12. https://gcc.gnu.org/lists.html 13990 13. http://www.fsf.org/ 13991 14. https://gcc.gnu.org/about.html 13992 15. http://validator.w3.org/check/referer 13993 ====================================================================== 13994 http://gcc.gnu.org/gcc-3.1/changes.html 13995 GCC 3.1 Release Series 13996 Changes, New Features, and Fixes 13997 13998 Additional changes in GCC 3.1.1 13999 14000 * A bug related to how structures and unions are returned has been 14001 fixed for powerpc-*-netbsd*. 14002 * An important bug in the implementation of -fprefetch-loop-arrays 14003 has been fixed. Previously the optimization prefetched random 14004 blocks of memory for most targets except for i386. 14005 * The Java compiler now compiles Java programs much faster and also 14006 works with parallel make. 14007 * Nested functions have been fixed for mips*-*-netbsd*. 14008 * Some missing floating point support routines have beed added for 14009 mips*-*-netbsd*. 14010 * This [1]message gives additional information about the bugs fixed 14011 in this release. 14012 14013 Caveats 14014 14015 * The -traditional C compiler option has been deprecated and will be 14016 removed in GCC 3.3. (It remains possible to preprocess non-C code 14017 with the traditional preprocessor.) 14018 * The default debugging format for most ELF platforms (including 14019 GNU/Linux and FreeBSD; notable exception is Solaris) has changed 14020 from stabs to DWARF2. This requires GDB 5.1.1 or later. 14021 14022 General Optimizer Improvements 14023 14024 * Jan Hubicka, SuSE Labs, together with Richard Henderson, Red Hat, 14025 and Andreas Jaeger, SuSE Labs, has contributed [2]infrastructure 14026 for profile driven optimizations. 14027 Options -fprofile-arcs and -fbranch-probabilities can now be used 14028 to improve speed of the generated code by profiling the actual 14029 program behaviour on typical runs. In the absence of profile info 14030 the compiler attempts to guess the profile statically. 14031 * [3]SPEC2000 and SPEC95 benchmark suites are now used daily to 14032 monitor performance of the generated code. 14033 According to the SPECInt2000 results on an AMD Athlon CPU, the code 14034 generated by GCC 3.1 is 6% faster on the average (8.2% faster with 14035 profile feedback) compared to GCC 3.0. The code produced by GCC 3.0 14036 is about 2.1% faster compared to 2.95.3. Tests were done using the 14037 -O2 -march=athlon command-line options. 14038 * Alexandre Oliva, of Red Hat, has generalized the tree inlining 14039 infrastructure developed by CodeSourcery, LLC for the C++ front 14040 end, so that it is now used in the C front end too. Inlining 14041 functions as trees exposes them earlier to the compiler, giving it 14042 more opportunities for optimization. 14043 * Support for data prefetching instructions has been added to the GCC 14044 back end and several targets. A new __builtin_prefetch intrinsic is 14045 available to explicitly insert prefetch instructions and 14046 experimental support for loop array prefetching has been added (see 14047 -fprefetch-loop-array documentation). 14048 * Support for emitting debugging information for macros has been 14049 added for DWARF2. It is activated using -g3. 14050 14051 New Languages and Language specific improvements 14052 14053 C/C++ 14054 14055 * A few more [4]ISO C99 features. 14056 * The preprocessor is 10-50% faster than the preprocessor in GCC 3.0. 14057 * The preprocessor's symbol table has been merged with the symbol 14058 table of the C, C++ and Objective-C front ends. 14059 * The preprocessor consumes less memory than the preprocessor in GCC 14060 3.0, often significantly so. On normal input files, it typically 14061 consumes less memory than pre-3.0 cccp-based GCC, too. 14062 14063 C++ 14064 14065 * -fhonor-std and -fno-honor-std have been removed. -fno-honor-std 14066 was a workaround to allow std compliant code to work with the 14067 non-std compliant libstdc++-v2. libstdc++-v3 is std compliant. 14068 * The C++ ABI has been fixed so that void (A::*)() const is mangled 14069 as "M1AKFvvE", rather than "MK1AFvvE" as before. This change only 14070 affects pointer to cv-qualified member function types. 14071 * The C++ ABI has been changed to correctly handle this code: 14072 struct A { 14073 void operator delete[] (void *, size_t); 14074 }; 14075 14076 struct B : public A { 14077 }; 14078 14079 new B[10]; 14080 14081 The amount of storage allocated for the array will be greater than 14082 it was in 3.0, in order to store the number of elements in the 14083 array, so that the correct size can be passed to operator delete[] 14084 when the array is deleted. Previously, the value passed to operator 14085 delete[] was unpredictable. 14086 This change will only affect code that declares a two-argument 14087 operator delete[] with a second parameter of type size_t in a base 14088 class, and does not override that definition in a derived class. 14089 * The C++ ABI has been changed so that: 14090 struct A { 14091 void operator delete[] (void *, size_t); 14092 void operator delete[] (void *); 14093 }; 14094 14095 does not cause unnecessary storage to be allocated when an array of 14096 A objects is allocated. 14097 This change will only affect code that declares both of these forms 14098 of operator delete[], and declared the two-argument form before the 14099 one-argument form. 14100 * The C++ ABI has been changed so that when a parameter is passed by 14101 value, any cleanup for that parameter is performed in the caller, 14102 as specified by the ia64 C++ ABI, rather than the called function 14103 as before. As a result, classes with a non-trivial destructor but a 14104 trivial copy constructor will be passed and returned by invisible 14105 reference, rather than by bitwise copy as before. 14106 * G++ now supports the "named return value optimization": for code 14107 like 14108 A f () { 14109 A a; 14110 ... 14111 return a; 14112 } 14113 14114 G++ will allocate a in the return value slot, so that the return 14115 becomes a no-op. For this to work, all return statements in the 14116 function must return the same variable. 14117 * Improvements to the C++ library are listed in [5]the libstdc++-v3 14118 FAQ. 14119 14120 Objective-C 14121 14122 * Annoying linker warnings (due to incorrect code being generated) 14123 have been fixed. 14124 * If a class method cannot be found, the compiler no longer issues a 14125 warning if a corresponding instance method exists in the root 14126 class. 14127 * Forward @protocol declarations have been fixed. 14128 * Loading of categories has been fixed in certain situations (GNU run 14129 time only). 14130 * The class lookup in the run-time library has been rewritten so that 14131 class method dispatch is more than twice as fast as it used to be 14132 (GNU run time only). 14133 14134 Java 14135 14136 * libgcj now includes RMI, java.lang.ref.*, javax.naming, and 14137 javax.transaction. 14138 * Property files and other system resources can be compiled into 14139 executables which use libgcj using the new gcj --resource feature. 14140 * libgcj has been ported to more platforms. In particular there is 14141 now a mostly-functional mingw32 (Windows) target port. 14142 * JNI and CNI invocation interfaces were implemented, so gcj-compiled 14143 Java code can now be called from a C/C++ application. 14144 * gcj can now use builtin functions for certain known methods, for 14145 instance Math.cos. 14146 * gcj can now automatically remove redundant array-store checks in 14147 some common cases. 14148 * The --no-store-checks optimization option was added. This can be 14149 used to omit runtime store checks for code which is known not to 14150 throw ArrayStoreException 14151 * The following third party interface standards were added to libgcj: 14152 org.w3c.dom and org.xml.sax. 14153 * java.security has been merged with GNU Classpath. The new package 14154 is now JDK 1.2 compliant, and much more complete. 14155 * A bytecode verifier was added to the libgcj interpreter. 14156 * java.lang.Character was rewritten to comply with the Unicode 3.0 14157 standard, and improve performance. 14158 * Partial support for many more locales was added to libgcj. 14159 * Socket timeouts have been implemented. 14160 * libgcj has been merged into a single shared library. There are no 14161 longer separate shared libraries for the garbage collector and 14162 zlib. 14163 * Several performance improvements were made to gcj and libgcj: 14164 + Hash synchronization (thin locks) 14165 + A special allocation path for finalizer-free objects 14166 + Thread-local allocation 14167 + Parallel GC, and other GC tweaks 14168 14169 Fortran 14170 14171 Fortran improvements are listed in [6]the Fortran documentation. 14172 14173 Ada 14174 14175 [7]Ada Core Technologies, Inc, has contributed its GNAT Ada 95 front 14176 end and associated tools. The GNAT compiler fully implements the Ada 14177 language as defined by the ISO/IEC 8652 standard. 14178 14179 Please note that the integration of the Ada front end is still work in 14180 progress. 14181 14182 New Targets and Target Specific Improvements 14183 14184 * Hans-Peter Nilsson has contributed a port to MMIX, the CPU 14185 architecture used in new editions of Donald E. Knuth's The Art of 14186 Computer Programming. 14187 * Axis Communications has contributed its port to the CRIS CPU 14188 architecture, used in the ETRAX system-on-a-chip series. 14189 * Alexandre Oliva, of Red Hat, has contributed a port to the SuperH 14190 SH5 64-bit RISC microprocessor architecture, extending the existing 14191 SH port. 14192 * UltraSPARC is fully supported in 64-bit mode. The option -m64 14193 enables it. 14194 * For compatibility with the Sun compiler #pragma redefine_extname 14195 has been implemented on Solaris. 14196 * The x86 back end has had some noticeable work done to it. 14197 + SuSE Labs developers Jan Hubicka, Bo Thorsen and Andreas 14198 Jaeger have contributed a port to the AMD x86-64 architecture. 14199 For more information on x86-64 see http://www.x86-64.org. 14200 + The compiler now supports MMX, 3DNow!, SSE, and SSE2 14201 instructions. Options -mmmx, -m3dnow, -msse, and -msse2 will 14202 enable the respective instruction sets. Intel C++ compatible 14203 MMX/3DNow!/SSE intrinsics are implemented. SSE2 intrinsics 14204 will be added in next major release. 14205 + Following those improvements, targets for Pentium MMX, K6-2, 14206 K6-3, Pentium III, Pentium 4, and Athlon 4 Mobile/XP/MP were 14207 added. Refer to the documentation on -march= and -mcpu= 14208 options for details. 14209 + For those targets that support it, -mfpmath=sse will cause the 14210 compiler to generate SSE/SSE2 instructions for floating point 14211 math instead of x87 instructions. Usually, this will lead to 14212 quicker code especially on the Pentium 4. Note that only 14213 scalar floating point instructions are used and GCC does not 14214 exploit SIMD features yet. 14215 + Prefetch support has been added to the Pentium III, Pentium 4, 14216 K6-2, K6-3, and Athlon series. 14217 + Code generated for floating point to integer conversions has 14218 been improved leading to better performance of many 3D 14219 applications. 14220 * The PowerPC back end has added 64-bit PowerPC GNU/Linux support. 14221 * C++ support for AIX has been improved. 14222 * Aldy Hernandez, of Red Hat, Inc, has contributed extensions to the 14223 PowerPC port supporting the AltiVec programming model (SIMD). The 14224 support, though presently useful, is experimental and is expected 14225 to stabilize for 3.2. The support is written to conform to 14226 Motorola's AltiVec specs. See -maltivec. 14227 14228 Obsolete Systems 14229 14230 Support for a number of older systems has been declared obsolete in GCC 14231 3.1. Unless there is activity to revive them, the next release of GCC 14232 will have their sources permanently removed. 14233 14234 All configurations of the following processor architectures have been 14235 declared obsolete: 14236 * MIL-STD-1750A, 1750a-*-* 14237 * AMD A29k, a29k-*-* 14238 * Convex, c*-convex-* 14239 * Clipper, clipper-*-* 14240 * Elxsi, elxsi-*-* 14241 * Intel i860, i860-*-* 14242 * Sun picoJava, pj-*-* and pjl-*-* 14243 * Western Electric 32000, we32k-*-* 14244 14245 Most configurations of the following processor architectures have been 14246 declared obsolete, but we are preserving a few systems which may have 14247 active developers. It is unlikely that the remaining systems will 14248 survive much longer unless we see definite signs of port activity. 14249 * Motorola 88000 except 14250 + Generic a.out, m88k-*-aout* 14251 + Generic SVR4, m88k-*-sysv4 14252 + OpenBSD, m88k-*-openbsd* 14253 * NS32k except 14254 + NetBSD, ns32k-*-netbsd* 14255 + OpenBSD, ns32k-*-openbsd*. 14256 * ROMP except 14257 + OpenBSD, romp-*-openbsd*. 14258 14259 Finally, only some configurations of these processor architectures are 14260 being obsoleted. 14261 * Alpha: 14262 + OSF/1, alpha*-*-osf[123]*. (Digital Unix and Tru64 Unix, aka 14263 alpha*-*-osf[45], are still supported.) 14264 * ARM: 14265 + RISCiX, arm-*-riscix*. 14266 * i386: 14267 + 386BSD, i?86-*-bsd* 14268 + Chorus, i?86-*-chorusos* 14269 + DG/UX, i?86-*-dgux* 14270 + FreeBSD 1.x, i?86-*-freebsd1.* 14271 + IBM AIX, i?86-*-aix* 14272 + ISC UNIX, i?86-*-isc* 14273 + GNU/Linux with pre-BFD linker, i?86-*-linux*oldld* 14274 + NEXTstep, i?86-next-* 14275 + OSF UNIX, i?86-*-osf1* and i?86-*-osfrose* 14276 + RTEMS/coff, i?86-*-rtemscoff* 14277 + RTEMS/go32, i?86-go32-rtems* 14278 + Sequent/BSD, i?86-sequent-bsd* 14279 + Sequent/ptx before version 3, i?86-sequent-ptx[12]* and 14280 i?86-sequent-sysv3* 14281 + SunOS, i?86-*-sunos* 14282 * Motorola 68000: 14283 + Altos, m68[k0]*-altos-* 14284 + Apollo, m68[k0]*-apollo-* 14285 + Apple A/UX, m68[k0]*-apple-* 14286 + Bull, m68[k0]*-bull-* 14287 + Convergent, m68[k0]*-convergent-* 14288 + Generic SVR3, m68[k0]*-*-sysv3* 14289 + ISI, m68[k0]*-isi-* 14290 + LynxOS, m68[k0]*-*-lynxos* 14291 + NEXT, m68[k0]*-next-* 14292 + RTEMS/coff, m68[k0]*-*-rtemscoff* 14293 + Sony, m68[k0]*-sony-* 14294 * MIPS: 14295 + DEC Ultrix, mips-*-ultrix* and mips-dec-* 14296 + Generic BSD, mips-*-bsd* 14297 + Generic System V, mips-*-sysv* 14298 + IRIX before version 5, mips-sgi-irix[1234]* 14299 + RiscOS, mips-*-riscos* 14300 + Sony, mips-sony-* 14301 + Tandem, mips-tandem-* 14302 * SPARC: 14303 + RTEMS/a.out, sparc-*-rtemsaout*. 14304 14305 Documentation improvements 14306 14307 * The old manual ("Using and Porting the GNU Compiler Collection") 14308 has been replaced by a users manual ("Using the GNU Compiler 14309 Collection") and a separate internals reference manual ("GNU 14310 Compiler Collection Internals"). 14311 * More complete and much improved documentation about GCC's internal 14312 representation used by the C and C++ front ends. 14313 * Many cleanups and improvements in general. 14314 14315 14316 For questions related to the use of GCC, please consult these web 14317 pages and the [8]GCC manuals. If that fails, the 14318 [9]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 14319 web pages and the development of GCC are welcome on our developer 14320 list at [10]gcc (a] gcc.gnu.org. All of [11]our lists have public 14321 archives. 14322 14323 Copyright (C) [12]Free Software Foundation, Inc. Verbatim copying and 14324 distribution of this entire article is permitted in any medium, 14325 provided this notice is preserved. 14326 14327 These pages are [13]maintained by the GCC team. Last modified 14328 2017-09-09[14]. 14329 14330 References 14331 14332 1. https://gcc.gnu.org/ml/gcc/2002-07/msg01208.html 14333 2. http://gcc.gnu.org/news/profiledriven.html 14334 3. http://gcc.gnu.org/benchmarks/ 14335 4. http://gcc.gnu.org/c99status.html 14336 5. https://gcc.gnu.org/onlinedocs/libstdc++/faq.html 14337 6. https://gcc.gnu.org/onlinedocs/gcc-3.1.1/g77/News.html 14338 7. http://www.adacore.com/ 14339 8. https://gcc.gnu.org/onlinedocs/ 14340 9. mailto:gcc-help (a] gcc.gnu.org 14341 10. mailto:gcc (a] gcc.gnu.org 14342 11. https://gcc.gnu.org/lists.html 14343 12. http://www.fsf.org/ 14344 13. https://gcc.gnu.org/about.html 14345 14. http://validator.w3.org/check/referer 14346 ====================================================================== 14347 http://gcc.gnu.org/gcc-3.0/index.html 14348 GCC 3.0.4 14349 14350 February 20, 2002 14351 14352 The [1]GNU project and the GCC developers are pleased to announce the 14353 release of GCC 3.0.4, which is a bug-fix release for the GCC 3.0 14354 series. 14355 14356 This release series is no longer maintained. 14357 14358 GCC used to stand for the GNU C Compiler, but since the compiler 14359 supports several other languages aside from C, it now stands for the 14360 GNU Compiler Collection. 14361 14362 GCC 3.0.x has several new optimizations, new targets, new languages and 14363 many other new features, relative to GCC 2.95.x. See the [2]new 14364 features page for a more complete list. 14365 14366 A list of [3]successful builds is updated as new information becomes 14367 available. 14368 14369 The GCC developers would like to thank the numerous people that have 14370 contributed new features, test results, bug fixes, etc to GCC. This 14371 [4]amazing group of volunteers is what makes GCC successful. 14372 14373 And finally, we can't in good conscience fail to mention some 14374 [5]caveats to using GCC 3.0.x. 14375 14376 For additional information about GCC please refer to the [6]GCC project 14377 web site or contact the [7]GCC development mailing list. 14378 14379 To obtain GCC please use [8]our mirror sites, or our CVS server. 14380 __________________________________________________________________ 14381 14382 Previous 3.0.x Releases 14383 14384 December 20, 2001: GCC 3.0.3 has been released. 14385 October 25, 2001: GCC 3.0.2 has been released. 14386 August 20, 2001: GCC 3.0.1 has been released. 14387 June 18, 2001: GCC 3.0 has been released. 14388 14389 14390 For questions related to the use of GCC, please consult these web 14391 pages and the [9]GCC manuals. If that fails, the 14392 [10]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 14393 web pages and the development of GCC are welcome on our developer 14394 list at [11]gcc (a] gcc.gnu.org. All of [12]our lists have public 14395 archives. 14396 14397 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and 14398 distribution of this entire article is permitted in any medium, 14399 provided this notice is preserved. 14400 14401 These pages are [14]maintained by the GCC team. Last modified 14402 2016-09-30[15]. 14403 14404 References 14405 14406 1. http://www.gnu.org/ 14407 2. http://gcc.gnu.org/gcc-3.0/features.html 14408 3. http://gcc.gnu.org/gcc-3.0/buildstat.html 14409 4. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 14410 5. http://gcc.gnu.org/gcc-3.0/caveats.html 14411 6. http://gcc.gnu.org/index.html 14412 7. mailto:gcc (a] gcc.gnu.org 14413 8. http://gcc.gnu.org/mirrors.html 14414 9. https://gcc.gnu.org/onlinedocs/ 14415 10. mailto:gcc-help (a] gcc.gnu.org 14416 11. mailto:gcc (a] gcc.gnu.org 14417 12. https://gcc.gnu.org/lists.html 14418 13. http://www.fsf.org/ 14419 14. https://gcc.gnu.org/about.html 14420 15. http://validator.w3.org/check/referer 14421 ====================================================================== 14422 http://gcc.gnu.org/gcc-3.0/features.html 14423 GCC 3.0 New Features 14424 14425 Additional changes in GCC 3.0.4 14426 14427 * GCC 3.0 now supports newer versions of the [1]NetBSD operating 14428 system, which use the ELF object file format, on x86 processors. 14429 * Correct debugging information is generated from functions that have 14430 lines from multiple files (e.g. yacc output). 14431 * A fix for whitespace handling in the -traditional preprocessor, 14432 which can affect Fortran. 14433 * Fixes to the exception handling runtime. 14434 * More fixes for bad code generation in C++. 14435 * A fix for shared library generation under AIX 4.3. 14436 * Documentation updates. 14437 * Port of GCC to Tensilica's Xtensa processor contributed. 14438 * A fix for compiling the PPC Linux kernel (FAT fs wouldn't link). 14439 14440 Additional changes in GCC 3.0.3 14441 14442 * A fix to correct an accidental change to the PowerPC ABI. 14443 * Fixes for bad code generation on a variety of architectures. 14444 * Improvements to the debugging information generated for C++ 14445 classes. 14446 * Fixes for bad code generation in C++. 14447 * A fix to avoid crashes in the C++ demangler. 14448 * A fix to the C++ standard library to avoid buffer overflows. 14449 * Miscellaneous improvements for a variety of architectures. 14450 14451 Additional changes in GCC 3.0.2 14452 14453 * Fixes for bad code generation during loop unrolling. 14454 * Fixes for bad code generation by the sibling call optimization. 14455 * Minor improvements to x86 code generation. 14456 * Implementation of function descriptors in C++ vtables for IA64. 14457 * Numerous minor bug-fixes. 14458 14459 Additional changes in GCC 3.0.1 14460 14461 * C++ fixes for incorrect code-generation. 14462 * Improved cross-compiling support for the C++ standard library. 14463 * Fixes for some embedded targets that worked in GCC 2.95.3, but not 14464 in GCC 3.0. 14465 * Fixes for various exception-handling bugs. 14466 * A port to the S/390 architecture. 14467 14468 General Optimizer Improvements 14469 14470 * [2]Basic block reordering pass. 14471 * New if-conversion pass with support for conditional (predicated) 14472 execution. 14473 * New tail call and sibling call elimination optimizations. 14474 * New register renaming pass. 14475 * New (experimental) [3]static single assignment (SSA) representation 14476 support. 14477 * New dead-code elimination pass implemented using the SSA 14478 representation. 14479 * [4]Global null pointer test elimination. 14480 * [5]Global code hoisting/unification. 14481 * More builtins and optimizations for stdio.h, string.h and old BSD 14482 functions, as well as for ISO C99 functions. 14483 * New builtin __builtin_expect for giving hints to the branch 14484 predictor. 14485 14486 New Languages and Language specific improvements 14487 14488 * The GNU Compiler for the Java(TM) language (GCJ) is now integrated 14489 and supported, including the run-time library containing most 14490 common non-GUI Java classes, a bytecode interpreter, and the Boehm 14491 conservative garbage collector. Many bugs have been fixed. GCJ can 14492 compile Java source or Java bytecodes to either native code or Java 14493 class files, and supports native methods written in either the 14494 standard JNI or the more efficient and convenient CNI. 14495 * Here is a [6]partial list of C++ improvements, both new features 14496 and those no longer supported. 14497 * New C++ ABI. On the IA-64 platform GCC is capable of 14498 inter-operating with other IA-64 compilers. 14499 * The new ABI also significantly reduces the size of symbol and debug 14500 information. 14501 * New C++ support library and many C++ bug fixes, vastly improving 14502 our conformance to the ISO C++ standard. 14503 * New [7]inliner for C++. 14504 * Rewritten C preprocessor, integrated into the C, C++ and Objective 14505 C compilers, with very many improvements including ISO C99 support 14506 and [8]improvements to dependency generation. 14507 * Support for more [9]ISO C99 features. 14508 * Many improvements to support for checking calls to format functions 14509 such as printf and scanf, including support for ISO C99 format 14510 features, extensions from the Single Unix Specification and GNU 14511 libc 2.2, checking of strfmon formats and features to assist in 14512 auditing for format string security bugs. 14513 * New warnings for C code that may have undefined semantics because 14514 of violations of sequence point rules in the C standard (such as a 14515 = a++;, a[n] = b[n++]; and a[i++] = i;), included in -Wall. 14516 * Additional warning option -Wfloat-equal. 14517 * Improvements to -Wtraditional. 14518 * Fortran improvements are listed in [10]the Fortran documentation. 14519 14520 New Targets and Target Specific Improvements 14521 14522 * New x86 back end, generating much improved code. 14523 * Support for a generic i386-elf target contributed. 14524 * New option to emit x86 assembly code using Intel style syntax 14525 (-mintel-syntax). 14526 * HPUX 11 support contributed. 14527 * Improved PowerPC code generation, including scheduled prologue and 14528 epilogue. 14529 * Port of GCC to Intel's IA-64 processor contributed. 14530 * Port of GCC to Motorola's MCore 210 and 340 contributed. 14531 * New unified back-end for Arm, Thumb and StrongArm contributed. 14532 * Port of GCC to Intel's XScale processor contributed. 14533 * Port of GCC to Atmel's AVR microcontrollers contributed. 14534 * Port of GCC to Mitsubishi's D30V processor contributed. 14535 * Port of GCC to Matsushita's AM33 processor (a member of the MN10300 14536 processor family) contributed. 14537 * Port of GCC to Fujitsu's FR30 processor contributed. 14538 * Port of GCC to Motorola's 68HC11 and 68HC12 processors contributed. 14539 * Port of GCC to Sun's picoJava processor core contributed. 14540 14541 Documentation improvements 14542 14543 * Substantially rewritten and improved C preprocessor manual. 14544 * Many improvements to other documentation. 14545 * Manpages for gcc, cpp and gcov are now generated automatically from 14546 the master Texinfo manual, eliminating the problem of manpages 14547 being out of date. (The generated manpages are only extracts from 14548 the full manual, which is provided in Texinfo form, from which 14549 info, HTML, other formats and a printed manual can be generated.) 14550 * Generated info files are included in the release tarballs alongside 14551 their Texinfo sources, avoiding problems on some platforms with 14552 building makeinfo as part of the GCC distribution. 14553 14554 Other significant improvements 14555 14556 * Garbage collection used internally by the compiler for most memory 14557 allocation instead of obstacks. 14558 * Lengauer and Tarjan algorithm used for computing dominators in the 14559 CFG. This algorithm can be significantly faster and more space 14560 efficient than our older algorithm. 14561 * gccbug script provided to assist in submitting bug reports to our 14562 bug tracking system. (Bug reports previously submitted directly to 14563 our mailing lists, for which you received no bug tracking number, 14564 should be submitted again using gccbug if you can reproduce the 14565 problem with GCC 3.0.) 14566 * The internal libgcc library is [11]built as a shared library on 14567 systems that support it. 14568 * Extensive testsuite included with GCC, with many new tests. In 14569 addition to tests for GCC bugs that have been fixed, many tests 14570 have been added for language features, compiler warnings and 14571 builtin functions. 14572 * Additional language-independent warning options -Wpacked, -Wpadded, 14573 -Wunreachable-code and -Wdisabled-optimization. 14574 * Target-independent options -falign-functions, -falign-loops and 14575 -falign-jumps. 14576 14577 Plus a great many bug fixes and almost all the [12]features found in 14578 GCC 2.95. 14579 14580 14581 For questions related to the use of GCC, please consult these web 14582 pages and the [13]GCC manuals. If that fails, the 14583 [14]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 14584 web pages and the development of GCC are welcome on our developer 14585 list at [15]gcc (a] gcc.gnu.org. All of [16]our lists have public 14586 archives. 14587 14588 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and 14589 distribution of this entire article is permitted in any medium, 14590 provided this notice is preserved. 14591 14592 These pages are [18]maintained by the GCC team. Last modified 14593 2016-08-23[19]. 14594 14595 References 14596 14597 1. http://www.netbsd.org/ 14598 2. http://gcc.gnu.org/news/reorder.html 14599 3. http://gcc.gnu.org/news/ssa.html 14600 4. http://gcc.gnu.org/news/null.html 14601 5. http://gcc.gnu.org/news/unify.html 14602 6. http://gcc.gnu.org/gcc-3.0/c++features.html 14603 7. http://gcc.gnu.org/news/inlining.html 14604 8. http://gcc.gnu.org/news/dependencies.html 14605 9. http://gcc.gnu.org/c99status.html 14606 10. https://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html 14607 11. http://gcc.gnu.org/gcc-3.0/libgcc.html 14608 12. http://gcc.gnu.org/gcc-2.95/features.html 14609 13. https://gcc.gnu.org/onlinedocs/ 14610 14. mailto:gcc-help (a] gcc.gnu.org 14611 15. mailto:gcc (a] gcc.gnu.org 14612 16. https://gcc.gnu.org/lists.html 14613 17. http://www.fsf.org/ 14614 18. https://gcc.gnu.org/about.html 14615 19. http://validator.w3.org/check/referer 14616 ====================================================================== 14617 http://gcc.gnu.org/gcc-3.0/caveats.html 14618 GCC 3.0 Caveats 14619 14620 * -fstrict-aliasing is now part of -O2 and higher optimization 14621 levels. This allows the compiler to assume the strictest aliasing 14622 rules applicable to the language being compiled. For C and C++, 14623 this activates optimizations based on the type of expressions. This 14624 optimization may thus break old, non-compliant code. 14625 * Enumerations are now properly promoted to int in function 14626 parameters and function returns. Normally this change is not 14627 visible, but when using -fshort-enums this is an ABI change. 14628 * The undocumented extension that allowed C programs to have a label 14629 at the end of a compound statement has been deprecated and may be 14630 removed in a future version. Programs that now generate a warning 14631 about this may be fixed by adding a null statement (a single 14632 semicolon) after the label. 14633 * The poorly documented extension that allowed string constants in C, 14634 C++ and Objective C to contain unescaped newlines has been 14635 deprecated and may be removed in a future version. Programs using 14636 this extension may be fixed in several ways: the bare newline may 14637 be replaced by \n, or preceded by \n\, or string concatenation may 14638 be used with the bare newline preceded by \n" and " placed at the 14639 start of the next line. 14640 * The Chill compiler is not included in GCC 3.0, because of the lack 14641 of a volunteer to convert it to use garbage collection. 14642 * Certain non-standard iostream methods from earlier versions of 14643 libstdc++ are not included in libstdc++ v3, i.e. filebuf::attach, 14644 ostream::form, and istream::gets. 14645 * The new C++ ABI is not yet fully supported by current (as of 14646 2001-07-01) releases and development versions of GDB, or any 14647 earlier versions. There is a problem setting breakpoints by line 14648 number, and other related issues that have been fixed in GCC 3.0 14649 but not yet handled in GDB: 14650 [1]https://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html 14651 14652 14653 For questions related to the use of GCC, please consult these web 14654 pages and the [2]GCC manuals. If that fails, the 14655 [3]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 14656 web pages and the development of GCC are welcome on our developer 14657 list at [4]gcc (a] gcc.gnu.org. All of [5]our lists have public archives. 14658 14659 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and 14660 distribution of this entire article is permitted in any medium, 14661 provided this notice is preserved. 14662 14663 These pages are [7]maintained by the GCC team. Last modified 14664 2016-01-30[8]. 14665 14666 References 14667 14668 1. https://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html 14669 2. https://gcc.gnu.org/onlinedocs/ 14670 3. mailto:gcc-help (a] gcc.gnu.org 14671 4. mailto:gcc (a] gcc.gnu.org 14672 5. https://gcc.gnu.org/lists.html 14673 6. http://www.fsf.org/ 14674 7. https://gcc.gnu.org/about.html 14675 8. http://validator.w3.org/check/referer 14676 ====================================================================== 14677 http://gcc.gnu.org/gcc-2.95/index.html 14678 GCC 2.95 14679 14680 March 16, 2001: The GNU project and the GCC developers are pleased to 14681 announce the release of GCC version 2.95.3. 14682 14683 This release is no longer maintained. 14684 14685 Release History 14686 14687 GCC 2.95.3 14688 March 16, 2001 14689 14690 GCC 2.95.2 14691 October 27, 1999 14692 14693 GCC 2.95.1 14694 August 19, 1999 14695 14696 GCC 2.95 14697 July 31, 1999. This is the first release of GCC since the April 14698 1999 GCC/EGCS reunification and includes nearly a year's worth 14699 of new development and bugfixes. 14700 14701 References and Acknowledgements 14702 14703 GCC used to stand for the GNU C Compiler, but since the compiler 14704 supports several other languages aside from C, it now stands for the 14705 GNU Compiler Collection. 14706 14707 The whole suite has been extensively [1]regression tested and 14708 [2]package tested. It should be reliable and suitable for widespread 14709 use. 14710 14711 The compiler has several new optimizations, new targets, new languages 14712 and other new features. See the [3]new features page for a more 14713 complete list of new features found in the GCC 2.95 releases. 14714 14715 The sources include installation instructions in both HTML and 14716 plaintext forms in the install directory in the distribution. However, 14717 the most up to date installation instructions and [4]build/test status 14718 are on the web pages. We will update those pages as new information 14719 becomes available. 14720 14721 The GCC developers would like to thank the numerous people that have 14722 contributed new features, test results, bugfixes, etc to GCC. This 14723 [5]amazing group of volunteers is what makes GCC successful. 14724 14725 And finally, we can't in good conscience fail to mention some 14726 [6]caveats to using GCC 2.95. 14727 14728 Download GCC 2.95 from one of our many [7]mirror sites. 14729 14730 For additional information about GCC please see the [8]GCC project web 14731 server or contact the [9]GCC development mailing list. 14732 14733 14734 For questions related to the use of GCC, please consult these web 14735 pages and the [10]GCC manuals. If that fails, the 14736 [11]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 14737 web pages and the development of GCC are welcome on our developer 14738 list at [12]gcc (a] gcc.gnu.org. All of [13]our lists have public 14739 archives. 14740 14741 Copyright (C) [14]Free Software Foundation, Inc. Verbatim copying and 14742 distribution of this entire article is permitted in any medium, 14743 provided this notice is preserved. 14744 14745 These pages are [15]maintained by the GCC team. Last modified 14746 2016-09-30[16]. 14747 14748 References 14749 14750 1. http://gcc.gnu.org/gcc-2.95/regress.html 14751 2. http://gcc.gnu.org/gcc-2.95/othertest.html 14752 3. http://gcc.gnu.org/gcc-2.95/features.html 14753 4. http://gcc.gnu.org/gcc-2.95/buildstat.html 14754 5. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 14755 6. http://gcc.gnu.org/gcc-2.95/caveats.html 14756 7. http://gcc.gnu.org/mirrors.html 14757 8. http://gcc.gnu.org/index.html 14758 9. mailto:gcc (a] gcc.gnu.org 14759 10. https://gcc.gnu.org/onlinedocs/ 14760 11. mailto:gcc-help (a] gcc.gnu.org 14761 12. mailto:gcc (a] gcc.gnu.org 14762 13. https://gcc.gnu.org/lists.html 14763 14. http://www.fsf.org/ 14764 15. https://gcc.gnu.org/about.html 14765 16. http://validator.w3.org/check/referer 14766 ====================================================================== 14767 http://gcc.gnu.org/gcc-2.95/features.html 14768 GCC 2.95 New Features 14769 14770 * General Optimizer Improvements: 14771 + [1]Localized register spilling to improve speed and code 14772 density especially on small register class machines. 14773 + [2]Global CSE using lazy code motion algorithms. 14774 + [3]Improved global constant/copy propagation. 14775 + [4]Improved control flow graph analysis and manipulation. 14776 + [5]Local dead store elimination. 14777 + [6]Memory Load hoisting/store sinking in loops. 14778 + [7]Type based alias analysis is enabled by default. Note this 14779 feature will expose bugs in the Linux kernel. Please refer to 14780 the FAQ (as shipped with GCC 2.95) for additional information 14781 on this issue. 14782 + Major revamp of GIV detection, combination and simplification 14783 to improve loop performance. 14784 + Major improvements to register allocation and reloading. 14785 * New Languages and Language specific improvements 14786 + [8]Many C++ improvements. 14787 + [9]Many Fortran improvements. 14788 + [10]Java front-end has been integrated. A [11]runtime library 14789 is available separately. 14790 + [12]ISO C99 support 14791 + [13]Chill front-end and runtime has been integrated. 14792 + Boehm garbage collector support in libobjc. 14793 + More support for various pragmas which appear in vendor 14794 include files 14795 * New Targets and Target Specific Improvements 14796 + [14]SPARC backend rewrite. 14797 + -mschedule=8000 will optimize code for PA8000 class 14798 processors; -mpa-risc-2-0 will generate code for PA2.0 14799 processors 14800 + Various micro-optimizations for the ia32 port. K6 14801 optimizations 14802 + Compiler will attempt to align doubles in the stack on the 14803 ia32 port 14804 + Alpha EV6 support 14805 + PowerPC 750 14806 + RS6000/PowerPC: -mcpu=401 was added as an alias for -mcpu=403. 14807 -mcpu=e603e was added to do -mcpu=603e and -msoft-float. 14808 + c3x, c4x 14809 + HyperSPARC 14810 + SparcLite86x 14811 + sh4 14812 + Support for new systems (OpenBSD, FreeBSD, UWIN, Interix, 14813 arm-linux) 14814 + vxWorks targets include support for vxWorks threads 14815 + StrongARM 110 and ARM9 support added. ARM Scheduling 14816 parameters rewritten. 14817 + Various changes to the MIPS port to avoid assembler macros, 14818 which in turn improves performance 14819 + Various performance improvements to the i960 port. 14820 + Major rewrite of ns32k port 14821 * Other significant improvements 14822 + [15]Ability to dump cfg information and display it using vcg. 14823 + The new faster scheme for fixing vendor header files is 14824 enabled by default. 14825 + Experimental internationalization support. 14826 + multibyte character support 14827 + Some compile-time speedups for pathological problems 14828 + Better support for complex types 14829 * Plus the usual mountain of bugfixes 14830 * Core compiler is based on the gcc2 development tree from Sept 30, 14831 1998, so we have all of the [16]features found in GCC 2.8. 14832 14833 Additional Changes in GCC 2.95.1 14834 14835 * Generic bugfixes and improvements 14836 + Various documentation fixes related to the GCC/EGCS merger. 14837 + Fix memory management bug which could lead to spurious aborts, 14838 core dumps or random parsing errors in the compiler. 14839 + Fix a couple bugs in the dwarf1 and dwarf2 debug record 14840 support. 14841 + Fix infinite loop in the CSE optimizer. 14842 + Avoid undefined behavior in compiler FP emulation code 14843 + Fix install problem when prefix is overridden on the make 14844 install command. 14845 + Fix problem with unwanted installation of assert.h on some 14846 systems. 14847 + Fix problem with finding the wrong assembler in a single tree 14848 build. 14849 + Avoid increasing the known alignment of a register that is 14850 already known to be a pointer. 14851 * Platform specific bugfixes and improvements 14852 + Codegen bugfix for prologue/epilogue for cpu32 target. 14853 + Fix long long code generation bug for the Coldfire target. 14854 + Fix various aborts in the SH compiler. 14855 + Fix bugs in libgcc support library for the SH. 14856 + Fix alpha ev6 code generation bug. 14857 + Fix problems with EXIT_SUCCESS/EXIT_FAILURE redefinitions on 14858 AIX platforms. 14859 + Fix -fpic code generation bug for rs6000/ppc svr4 targets. 14860 + Fix varargs/stdarg code generation bug for rs6000/ppc svr4 14861 targets. 14862 + Fix weak symbol handling for rs6000/ppc svr4 targets. 14863 + Fix various problems with 64bit code generation for the 14864 rs6000/ppc port. 14865 + Fix codegen bug which caused tetex to be mis-compiled on the 14866 x86. 14867 + Fix compiler abort in new cfg code exposed by x86 port. 14868 + Fix out of range array reference in code convert flat 14869 registers to the x87 stacked FP register file. 14870 + Fix minor vxworks configuration bug. 14871 + Fix return type of bsearch for SunOS 4.x. 14872 * Language & Runtime specific fixes. 14873 + The G++ signature extension has been deprecated. It will be 14874 removed in the next major release of G++. Use of signatures 14875 will result in a warning from the compiler. 14876 + Several bugs relating to templates and namespaces were fixed. 14877 + A bug that caused crashes when combining templates with -g on 14878 DWARF1 platforms was fixed. 14879 + Pointers-to-members, virtual functions, and multiple 14880 inheritance should now work together correctly. 14881 + Some code-generation bugs relating to function try blocks were 14882 fixed. 14883 + G++ is a little bit more lenient with certain archaic 14884 constructs than in GCC 2.95. 14885 + Fix to prevent shared library version #s from bring truncated 14886 to 1 digit 14887 + Fix missing std:: in the libstdc++ library. 14888 + Fix stream locking problems in libio. 14889 + Fix problem in java compiler driver. 14890 14891 Additional Changes in GCC 2.95.2 14892 14893 The -fstrict-aliasing is not enabled by default for GCC 2.95.2. While 14894 the optimizations performed by -fstrict-aliasing are valid according to 14895 the C and C++ standards, the optimization have caused some problems, 14896 particularly with old non-conforming code. 14897 14898 The GCC developers are experimenting with ways to warn users about code 14899 which violates the C/C++ standards, but those warnings are not ready 14900 for widespread use at this time. Rather than wait for those warnings 14901 the GCC developers have chosen to disable -fstrict-aliasing by default 14902 for the GCC 2.95.2 release. 14903 14904 We strongly encourage developers to find and fix code which violates 14905 the C/C++ standards as -fstrict-aliasing may be enabled by default in 14906 future releases. Use the option -fstrict-aliasing to re-enable these 14907 optimizations. 14908 * Generic bugfixes and improvements 14909 + Fix incorrectly optimized memory reference in global common 14910 subexpression elimination (GCSE) optimization pass. 14911 + Fix code generation bug in regmove.c in which it could 14912 incorrectly change a "const" value. 14913 + Fix bug in optimization of conditionals involving volatile 14914 memory references. 14915 + Avoid over-allocation of stack space for some procedures. 14916 + Fixed bug in the compiler which caused incorrect optimization 14917 of an obscure series of bit manipulations, shifts and 14918 arithmetic. 14919 + Fixed register allocator bug which caused teTeX to be 14920 mis-compiled on SPARC targets. 14921 + Avoid incorrect optimization of degenerate case statements for 14922 certain targets such as the ARM. 14923 + Fix out of range memory reference in the jump optimizer. 14924 + Avoid dereferencing null pointer in fix-header. 14925 + Fix test for GCC specific features so that it is possible to 14926 bootstrap with gcc-2.6.2 and older versions of GCC. 14927 + Fix typo in scheduler which could potentially cause out of 14928 range memory accesses. 14929 + Avoid incorrect loop reversal which caused incorrect code for 14930 certain loops on PowerPC targets. 14931 + Avoid incorrect optimization of switch statements on certain 14932 targets (for example the ARM). 14933 * Platform specific bugfixes and improvements 14934 + Work around bug in Sun V5.0 compilers which caused bootstrap 14935 comparison failures on SPARC targets. 14936 + Fix SPARC backend bug which caused aborts in final.c. 14937 + Fix sparc-hal-solaris2* configuration fragments. 14938 + Fix bug in sparc block profiling. 14939 + Fix obscure code generation bug for the PARISC targets. 14940 + Define __STDC_EXT__ for HPUX configurations. 14941 + Various POWERPC64 code generation bugfixes. 14942 + Fix abort for PPC targets using ELF (ex GNU/Linux). 14943 + Fix collect2 problems for AIX targets. 14944 + Correct handling of .file directive for PPC targets. 14945 + Fix bug in fix_trunc x86 patterns. 14946 + Fix x86 port to correctly pop the FP stack for functions that 14947 return structures in memory. 14948 + Fix minor bug in strlen x86 pattern. 14949 + Use stabs debugging instead of dwarf1 for x86-solaris targets. 14950 + Fix template repository code to handle leading underscore in 14951 mangled names. 14952 + Fix weak/weak alias support for OpenBSD. 14953 + GNU/Linux for the ARM has C++ compatible include files. 14954 * Language & Runtime specific fixes. 14955 + Fix handling of constructor attribute in the C front-end which 14956 caused problems building the Chill runtime library on some 14957 targets. 14958 + Fix minor problem merging type qualifiers in the C front-end. 14959 + Fix aliasing bug for pointers and references (C/C++). 14960 + Fix incorrect "non-constant initializer bug" when -traditional 14961 or -fwritable-strings is enabled. 14962 + Fix build error for Chill front-end on SunOS. 14963 + Do not complain about duplicate instantiations when using 14964 -frepo (C++). 14965 + Fix array bounds handling in C++ front-end which caused 14966 problems with dwarf debugging information in some 14967 circumstances. 14968 + Fix minor namespace problem. 14969 + Fix problem linking java programs. 14970 14971 Additional Changes in GCC 2.95.3 14972 14973 * Generic bugfixes and improvements 14974 + Fix numerous problems that caused incorrect optimization in 14975 the register reloading code. 14976 + Fix numerous problems that caused incorrect optimization in 14977 the loop optimizer. 14978 + Fix aborts in the functions build_insn_chain and scan_loops 14979 under some circumstances. 14980 + Fix an alias analysis bug. 14981 + Fix an infinite compilation bug in the combiner. 14982 + A few problems with complex number support have been fixed. 14983 + It is no longer possible for gcc to act as a fork bomb when 14984 installed incorrectly. 14985 + The -fpack-struct option should be recognized now. 14986 + Fixed a bug that caused incorrect code to be generated due to 14987 a lost stack adjustment. 14988 * Platform specific bugfixes and improvements 14989 + Support building ARM toolchains hosted on Windows. 14990 + Fix attribute calculations in ARM toolchains. 14991 + arm-linux support has been improved. 14992 + Fix a PIC failure on sparc targets. 14993 + On ix86 targets, the regparm attribute should now work 14994 reliably. 14995 + Several updates for the h8300 port. 14996 + Fix problem building libio with glibc 2.2. 14997 14998 14999 For questions related to the use of GCC, please consult these web 15000 pages and the [17]GCC manuals. If that fails, the 15001 [18]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15002 web pages and the development of GCC are welcome on our developer 15003 list at [19]gcc (a] gcc.gnu.org. All of [20]our lists have public 15004 archives. 15005 15006 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and 15007 distribution of this entire article is permitted in any medium, 15008 provided this notice is preserved. 15009 15010 These pages are [22]maintained by the GCC team. Last modified 15011 2016-12-29[23]. 15012 15013 References 15014 15015 1. http://gcc.gnu.org/news/spill.html 15016 2. http://gcc.gnu.org/news/lcm.html 15017 3. http://gcc.gnu.org/news/cprop.html 15018 4. http://gcc.gnu.org/news/cfg.html 15019 5. http://gcc.gnu.org/news/dse.html 15020 6. http://gcc.gnu.org/news/hoist.html 15021 7. http://gcc.gnu.org/news/alias.html 15022 8. http://gcc.gnu.org/gcc-2.95/c++features.html 15023 9. https://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html 15024 10. http://gcc.gnu.org/news/gcj-announce.txt 15025 11. http://gcc.gnu.org/news/javaannounce.html 15026 12. http://gcc.gnu.org/c99status.html 15027 13. http://gcc.gnu.org/news/chill.html 15028 14. http://gcc.gnu.org/news/sparc.html 15029 15. http://gcc.gnu.org/news/egcs-vcg.html 15030 16. http://gcc.gnu.org/egcs-1.0/features-2.8.html 15031 17. https://gcc.gnu.org/onlinedocs/ 15032 18. mailto:gcc-help (a] gcc.gnu.org 15033 19. mailto:gcc (a] gcc.gnu.org 15034 20. https://gcc.gnu.org/lists.html 15035 21. http://www.fsf.org/ 15036 22. https://gcc.gnu.org/about.html 15037 23. http://validator.w3.org/check/referer 15038 ====================================================================== 15039 http://gcc.gnu.org/gcc-2.95/caveats.html 15040 GCC 2.95 Caveats 15041 15042 * GCC 2.95 will issue an error for invalid asm statements that had 15043 been silently accepted by earlier versions of the compiler. This is 15044 particularly noticeable when compiling older versions of the Linux 15045 kernel (2.0.xx). Please refer to the FAQ (as shipped with GCC 2.95) 15046 for more information on this issue. 15047 * GCC 2.95 implements type based alias analysis to disambiguate 15048 memory references. Some programs, particularly the Linux kernel 15049 violate ANSI/ISO aliasing rules and therefore may not operate 15050 correctly when compiled with GCC 2.95. Please refer to the FAQ (as 15051 shipped with GCC 2.95) for more information on this issue. 15052 * GCC 2.95 has a known bug in its handling of complex variables for 15053 64bit targets. Instead of silently generating incorrect code, GCC 15054 2.95 will issue a fatal error for situations it can not handle. 15055 This primarily affects the Fortran community as Fortran makes more 15056 use of complex variables than C or C++. 15057 * GCC 2.95 has an integrated libstdc++, but does not have an 15058 integrated libg++. Furthermore old libg++ releases will not work 15059 with GCC 2.95. You can retrieve a recent copy of libg++ from the 15060 [1]GCC ftp server. 15061 Note most C++ programs only need libstdc++. 15062 * Exception handling may not work with shared libraries, particularly 15063 on alphas, hppas, rs6000/powerpc and mips based platforms. 15064 Exception handling is known to work on x86 GNU/Linux platforms with 15065 shared libraries. 15066 * In general, GCC 2.95 is more rigorous about rejecting invalid C++ 15067 code or deprecated C++ constructs than G++ 2.7, G++ 2.8, EGCS 1.0, 15068 or EGCS 1.1. As a result it may be necessary to fix C++ code before 15069 it will compile with GCC 2.95. 15070 * G++ is also converting toward the ISO C++ standard; as a result 15071 code which was previously valid (and thus accepted by other 15072 compilers and older versions of g++) may no longer be accepted. The 15073 flag -fpermissive may allow some non-conforming code to compile 15074 with GCC 2.95. 15075 * GCC 2.95 compiled C++ code is not binary compatible with EGCS 15076 1.1.x, EGCS 1.0.x or GCC 2.8.x. 15077 * GCC 2.95 does not have changes from the GCC 2.8 tree that were made 15078 between Sept 30, 1998 and April 30, 1999 (the official end of the 15079 GCC 2.8 project). Future GCC releases will include all the changes 15080 from the defunct GCC 2.8 sources. 15081 15082 15083 For questions related to the use of GCC, please consult these web 15084 pages and the [2]GCC manuals. If that fails, the 15085 [3]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15086 web pages and the development of GCC are welcome on our developer 15087 list at [4]gcc (a] gcc.gnu.org. All of [5]our lists have public archives. 15088 15089 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and 15090 distribution of this entire article is permitted in any medium, 15091 provided this notice is preserved. 15092 15093 These pages are [7]maintained by the GCC team. Last modified 15094 2016-01-30[8]. 15095 15096 References 15097 15098 1. ftp://gcc.gnu.org/pub/gcc/infrastructure/libg++-2.8.1.3.tar.gz 15099 2. https://gcc.gnu.org/onlinedocs/ 15100 3. mailto:gcc-help (a] gcc.gnu.org 15101 4. mailto:gcc (a] gcc.gnu.org 15102 5. https://gcc.gnu.org/lists.html 15103 6. http://www.fsf.org/ 15104 7. https://gcc.gnu.org/about.html 15105 8. http://validator.w3.org/check/referer 15106 ====================================================================== 15107 http://gcc.gnu.org/egcs-1.1/index.html 15108 EGCS 1.1 15109 15110 September 3, 1998: We are pleased to announce the release of EGCS 1.1. 15111 December 1, 1998: We are pleased to announce the release of EGCS 1.1.1. 15112 March 15, 1999: We are pleased to announce the release of EGCS 1.1.2. 15113 15114 EGCS is a free software project to further the development of the GNU 15115 compilers using an open development environment. 15116 15117 EGCS 1.1 is a major new release of the EGCS compiler system. It has 15118 been [1]extensively tested and is believed to be stable and suitable 15119 for widespread use. 15120 15121 EGCS 1.1 is based on an June 6, 1998 snapshot of the GCC 2.8 15122 development sources; it contains all of the new features found in GCC 15123 2.8.1 as well as all new development from GCC up to June 6, 1998. 15124 15125 EGCS 1.1 also contains many improvements and features not found in GCC 15126 or in older versions of EGCS: 15127 * Global common subexpression elimination and global constant/copy 15128 propagation (aka [2]gcse) 15129 * Ongoing improvements to the [3]alias analysis support to allow for 15130 better optimizations throughout the compiler. 15131 * Vastly improved [4]C++ compiler and integrated C++ runtime 15132 libraries. 15133 * Fixes for the /tmp symlink race security problems. 15134 * New targets including mips16, arm-thumb and 64 bit PowerPC. 15135 * Improvements to GNU Fortran (g77) compiler and runtime library made 15136 since g77 version 0.5.23. 15137 15138 See the [5]new features page for a more complete list of new features 15139 found in EGCS 1.1 releases. 15140 15141 EGCS 1.1.1 is a minor update to fix several serious problems in EGCS 15142 1.1: 15143 * General improvements and fixes 15144 + Avoid some stack overflows when compiling large functions. 15145 + Avoid incorrect loop invariant code motions. 15146 + Fix some core dumps on Linux kernel code. 15147 + Bring back the imake -Di386 and friends fix from EGCS 1.0.2. 15148 + Fix code generation problem in gcse. 15149 + Various documentation related fixes. 15150 * g++/libstdc++ improvements and fixes 15151 + MT safe EH fix for setjmp/longjmp based exception handling. 15152 + Fix a few bad interactions between optimization and exception 15153 handling. 15154 + Fixes for demangling of template names starting with "__". 15155 + Fix a bug that would fail to run destructors in some cases 15156 with -O2. 15157 + Fix 'new' of classes with virtual bases. 15158 + Fix crash building Qt on the Alpha. 15159 + Fix failure compiling WIFEXITED macro on GNU/Linux. 15160 + Fix some -frepo failures. 15161 * g77 and libf2c improvements and fixes 15162 + Various documentation fixes. 15163 + Avoid compiler crash on RAND intrinsic. 15164 + Fix minor bugs in makefiles exposed by BSD make programs. 15165 + Define _XOPEN_SOURCE for libI77 build to avoid potential 15166 problems on some 64-bit systems. 15167 + Fix problem with implicit endfile on rewind. 15168 + Fix spurious recursive I/O errors. 15169 * platform specific improvements and fixes 15170 + Match all versions of UnixWare7. 15171 + Do not assume x86 SVR4 or UnixWare targets can handle stabs. 15172 + Fix PPC/RS6000 LEGITIMIZE_ADDRESS macro and bug in conversion 15173 from unsigned ints to double precision floats. 15174 + Fix ARM ABI issue with NetBSD. 15175 + Fix a few arm code generation bugs. 15176 + Fixincludes will fix additional broken SCO OpenServer header 15177 files. 15178 + Fix a m68k backend bug which caused invalid offsets in reg+d 15179 addresses. 15180 + Fix problems with 64bit AIX 4.3 support. 15181 + Fix handling of long longs for varargs/stdarg functions on the 15182 ppc. 15183 + Minor fixes to CPP predefines for Windows. 15184 + Fix code generation problems with gpr<->fpr copies for 64bit 15185 ppc. 15186 + Fix a few coldfire code generation bugs. 15187 + Fix some more header file problems on SunOS 4.x. 15188 + Fix assert.h handling for RTEMS. 15189 + Fix Windows handling of TREE_SYMBOL_REFERENCED. 15190 + Fix x86 compiler abort in reg-stack pass. 15191 + Fix cygwin/windows problem with section attributes. 15192 + Fix Alpha code generation problem exposed by SMP Linux 15193 kernels. 15194 + Fix typo in m68k 32->64bit integer conversion. 15195 + Make sure target libraries build with -fPIC for PPC & Alpha 15196 targets. 15197 15198 EGCS 1.1.2 is a minor update to fix several serious problems in EGCS 15199 1.1.1: 15200 * General improvements and fixes 15201 + Fix bug in loop optimizer which caused the SPARC (and 15202 potentially other) ports to segfault. 15203 + Fix infinite recursion in alias analysis and combiner code. 15204 + Fix bug in regclass preferencing. 15205 + Fix incorrect loop reversal which caused incorrect code to be 15206 generated for several targets. 15207 + Fix return value for builtin memcpy. 15208 + Reduce compile time for certain loops which exposed quadratic 15209 behavior in the loop optimizer. 15210 + Fix bug which caused volatile memory to be written multiple 15211 times when only one write was needed/desired. 15212 + Fix compiler abort in caller-save.c 15213 + Fix combiner bug which caused incorrect code generation for 15214 certain division by constant operations. 15215 + Fix incorrect code generation due to a bug in range check 15216 optimizations. 15217 + Fix incorrect code generation due to mis-handling of clobbered 15218 values in CSE. 15219 + Fix compiler abort/segfault due to incorrect register 15220 splitting when unrolling loops. 15221 + Fix code generation involving autoincremented addresses with 15222 ternary operators. 15223 + Work around bug in the scheduler which caused qt to be 15224 mis-compiled on some platforms. 15225 + Fix code generation problems with -fshort-enums. 15226 + Tighten security for temporary files. 15227 + Improve compile time for codes which make heavy use of 15228 overloaded functions. 15229 + Fix multiply defined constructor/destructor symbol problems. 15230 + Avoid setting bogus RPATH environment variable during 15231 bootstrap. 15232 + Avoid GNU-make dependencies in the texinfo subdir. 15233 + Install CPP wrapper script in $(prefix)/bin if --enable-cpp. 15234 --enable-cpp=<dirname> can be used to specify an additional 15235 install directory for the cpp wrapper script. 15236 + Fix CSE bug which caused incorrect label-label refs to appear 15237 on some platforms. 15238 + Avoid linking in EH routines from libgcc if they are not 15239 needed. 15240 + Avoid obscure bug in aliasing code. 15241 + Fix bug in weak symbol handling. 15242 * Platform-specific improvements and fixes 15243 + Fix detection of PPro/PII on Unixware 7. 15244 + Fix compiler segfault when building spec99 and other programs 15245 for SPARC targets. 15246 + Fix code-generation bugs for integer and floating point 15247 conditional move instructions on the PPro/PII. 15248 + Use fixincludes to fix byteorder problems on i?86-*-sysv. 15249 + Fix build failure for the arc port. 15250 + Fix floating point format configuration for i?86-gnu port. 15251 + Fix problems with hppa1.0-hp-hpux10.20 configuration when 15252 threads are enabled. 15253 + Fix coldfire code generation bugs. 15254 + Fix "unrecognized insn" problems for Alpha and PPC ports. 15255 + Fix h8/300 code generation problem with floating point values 15256 in memory. 15257 + Fix unrecognized insn problems for the m68k port. 15258 + Fix namespace-pollution problem for the x86 port. 15259 + Fix problems with old assembler on x86 NeXT systems. 15260 + Fix PIC code-generation problems for the SPARC port. 15261 + Fix minor bug with LONG_CALLS in PowerPC SVR4 support. 15262 + Fix minor ISO namespace violation in Alpha varargs/stdarg 15263 support. 15264 + Fix incorrect "braf" instruction usage for the SH port. 15265 + Fix minor bug in va-sh which prevented its use with -ansi. 15266 + Fix problems recognizing and supporting FreeBSD. 15267 + Handle OpenBSD systems correctly. 15268 + Minor fixincludes fix for Digital UNIX 4.0B. 15269 + Fix problems with ctors/dtors in SCO shared libraries. 15270 + Abort instead of generating incorrect code for PPro/PII 15271 floating point conditional moves. 15272 + Avoid multiply defined symbols on GNU/Linux systems using 15273 libc-5.4.xx. 15274 + Fix abort in alpha compiler. 15275 * Fortran-specific fixes 15276 + Fix the IDate intrinsic (VXT) (in libg2c) so the returned year 15277 is in the documented, non-Y2K-compliant range of 0-99, instead 15278 of being returned as 100 in the year 2000. 15279 + Fix the `Date_and_Time' intrinsic (in libg2c) to return the 15280 milliseconds value properly in Values(8). 15281 + Fix the `LStat' intrinsic (in libg2c) to return device-ID 15282 information properly in SArray(7). 15283 15284 Each release includes installation instructions in both HTML and 15285 plaintext forms (see the INSTALL directory in the toplevel directory of 15286 the distribution). However, we also keep the most up to date 15287 installation instructions and [6]build/test status on our web page. We 15288 will update those pages as new information becomes available. 15289 15290 The EGCS project would like to thank the numerous people that have 15291 contributed new features, test results, bugfixes, etc. This [7]amazing 15292 group of volunteers is what makes EGCS successful. 15293 15294 And finally, we can't in good conscience fail to mention some 15295 [8]caveats to using EGCS 1.1. 15296 15297 Download EGCS from egcs.cygnus.com (USA California). 15298 15299 The EGCS 1.1 release is also available on many mirror sites. 15300 [9]Goto mirror list to find a closer site. 15301 15302 15303 For questions related to the use of GCC, please consult these web 15304 pages and the [10]GCC manuals. If that fails, the 15305 [11]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15306 web pages and the development of GCC are welcome on our developer 15307 list at [12]gcc (a] gcc.gnu.org. All of [13]our lists have public 15308 archives. 15309 15310 Copyright (C) [14]Free Software Foundation, Inc. Verbatim copying and 15311 distribution of this entire article is permitted in any medium, 15312 provided this notice is preserved. 15313 15314 These pages are [15]maintained by the GCC team. Last modified 15315 2016-01-30[16]. 15316 15317 References 15318 15319 1. http://gcc.gnu.org/egcs-1.1/egcs-1.1-test.html 15320 2. http://gcc.gnu.org/news/gcse.html 15321 3. http://gcc.gnu.org/news/alias.html 15322 4. http://gcc.gnu.org/egcs-1.1/c++features.html 15323 5. http://gcc.gnu.org/egcs-1.1/features.html 15324 6. http://gcc.gnu.org/egcs-1.1/buildstat.html 15325 7. https://gcc.gnu.org/onlinedocs/gcc/Contributors.html 15326 8. http://gcc.gnu.org/egcs-1.1/caveats.html 15327 9. http://gcc.gnu.org/mirrors.html 15328 10. https://gcc.gnu.org/onlinedocs/ 15329 11. mailto:gcc-help (a] gcc.gnu.org 15330 12. mailto:gcc (a] gcc.gnu.org 15331 13. https://gcc.gnu.org/lists.html 15332 14. http://www.fsf.org/ 15333 15. https://gcc.gnu.org/about.html 15334 16. http://validator.w3.org/check/referer 15335 ====================================================================== 15336 http://gcc.gnu.org/egcs-1.1/features.html 15337 EGCS 1.1 new features 15338 15339 * Integrated GNU Fortran (g77) compiler and runtime library with 15340 improvements, based on g77 version 0.5.23. 15341 * Vast improvements in the C++ compiler; so many they have [1]page of 15342 their own! 15343 * Compiler implements [2]global common subexpression elimination and 15344 global copy/constant propagation. 15345 * More major improvements in the [3]alias analysis code. 15346 * More major improvements in the exception handling code to improve 15347 performance, lower static overhead and provide the infrastructure 15348 for future improvements. 15349 * The infamous /tmp symlink race security problems have been fixed. 15350 * The regmove optimization pass has been nearly completely rewritten 15351 to improve performance of generated code. 15352 * The compiler now recomputes register usage information before local 15353 register allocation. By providing more accurate information to the 15354 priority based allocator, we get better register allocation. 15355 * The register reloading phase of the compiler optimizes spill code 15356 much better than in previous releases. 15357 * Some bad interactions between the register allocator and 15358 instruction scheduler have been fixed, resulting in much better 15359 code for certain programs. Additionally, we have tuned the 15360 scheduler in various ways to improve performance of generated code 15361 for some architectures. 15362 * The compiler's branch shortening algorithms have been significantly 15363 improved to work better on targets which align jump targets. 15364 * The compiler now supports -Os to prefer optimizing for code space 15365 over optimizing for code speed. 15366 * The compiler will now totally eliminate library calls which compute 15367 constant values. This primarily helps targets with no integer 15368 div/mul support and targets without floating point support. 15369 * The compiler now supports an extensive "--help" option. 15370 * cpplib has been greatly improved and may be suitable for limited 15371 use. 15372 * Memory footprint for the compiler has been significantly reduced 15373 for some pathological cases. 15374 * The time to build EGCS has been improved for certain targets 15375 (particularly the alpha and mips platforms). 15376 * Many infrastructure improvements throughout the compiler, plus the 15377 usual mountain of bugfixes and minor improvements. 15378 * Target dependent improvements: 15379 + SPARC port now includes V8 plus and V9 support as well as 15380 performance tuning for Ultra class machines. The SPARC port 15381 now uses the Haifa scheduler. 15382 + Alpha port has been tuned for the EV6 processor and has an 15383 optimized expansion of memcpy/bzero. The Alpha port now uses 15384 the Haifa scheduler. 15385 + RS6000/PowerPC: support for the Power64 architecture and AIX 15386 4.3. The RS6000/PowerPC port now uses the Haifa scheduler. 15387 + x86: Alignment of static store data and jump targets is per 15388 Intel recommendations now. Various improvements throughout the 15389 x86 port to improve performance on Pentium processors 15390 (including improved epilogue sequences for Pentium chips and 15391 backend improvements which should help register allocation on 15392 all x86 variants. Conditional move support has been fixed and 15393 enabled for PPro processors. The x86 port also better supports 15394 64bit operations now. Unixware 7, a System V Release 5 target, 15395 is now supported and SCO OpenServer targets can support GAS. 15396 + MIPS has improved multiply/multiply-add support and now 15397 includes mips16 ISA support. 15398 + M68k has many micro-optimizations and Coldfire fixes. 15399 * Core compiler is based on the GCC development tree from June 9, 15400 1998, so we have all of the [4]features found in GCC 2.8. 15401 15402 15403 For questions related to the use of GCC, please consult these web 15404 pages and the [5]GCC manuals. If that fails, the 15405 [6]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15406 web pages and the development of GCC are welcome on our developer 15407 list at [7]gcc (a] gcc.gnu.org. All of [8]our lists have public archives. 15408 15409 Copyright (C) [9]Free Software Foundation, Inc. Verbatim copying and 15410 distribution of this entire article is permitted in any medium, 15411 provided this notice is preserved. 15412 15413 These pages are [10]maintained by the GCC team. Last modified 15414 2016-01-30[11]. 15415 15416 References 15417 15418 1. http://gcc.gnu.org/egcs-1.1/c++features.html 15419 2. http://gcc.gnu.org/news/gcse.html 15420 3. http://gcc.gnu.org/news/alias.html 15421 4. http://gcc.gnu.org/egcs-1.0/features-2.8.html 15422 5. https://gcc.gnu.org/onlinedocs/ 15423 6. mailto:gcc-help (a] gcc.gnu.org 15424 7. mailto:gcc (a] gcc.gnu.org 15425 8. https://gcc.gnu.org/lists.html 15426 9. http://www.fsf.org/ 15427 10. https://gcc.gnu.org/about.html 15428 11. http://validator.w3.org/check/referer 15429 ====================================================================== 15430 http://gcc.gnu.org/egcs-1.1/caveats.html 15431 EGCS 1.1 Caveats 15432 15433 * EGCS has an integrated libstdc++, but does not have an integrated 15434 libg++. Furthermore old libg++ releases will not work with EGCS; HJ 15435 Lu has made a libg++-2.8.1.2 snapshot available which may work with 15436 EGCS. 15437 Note most C++ programs only need libstdc++. 15438 * Exception handling may not work with shared libraries, particularly 15439 on alphas, hppas, rs6000/powerpc and mips based platforms. 15440 Exception handling is known to work on x86-linux platforms with 15441 shared libraries. 15442 * Some versions of the Linux kernel have bugs which prevent them from 15443 being compiled or from running when compiled by EGCS. See the FAQ 15444 (as shipped with EGCS 1.1) for additional information. 15445 * In general, EGCS is more rigorous about rejecting invalid C++ code 15446 or deprecated C++ constructs than g++-2.7, g++-2.8 or EGCS 1.0. As 15447 a result it may be necessary to fix C++ code before it will compile 15448 with EGCS. 15449 * G++ is also converting toward the ISO C++ standard; as a result 15450 code which was previously valid (and thus accepted by other 15451 compilers and older versions of g++) may no longer be accepted. 15452 * EGCS 1.1 compiled C++ code is not binary compatible with EGCS 1.0.x 15453 or GCC 2.8.x due to changes necessary to support thread safe 15454 exception handling. 15455 15456 15457 For questions related to the use of GCC, please consult these web 15458 pages and the [1]GCC manuals. If that fails, the 15459 [2]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15460 web pages and the development of GCC are welcome on our developer 15461 list at [3]gcc (a] gcc.gnu.org. All of [4]our lists have public archives. 15462 15463 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and 15464 distribution of this entire article is permitted in any medium, 15465 provided this notice is preserved. 15466 15467 These pages are [6]maintained by the GCC team. Last modified 15468 2016-01-30[7]. 15469 15470 References 15471 15472 1. https://gcc.gnu.org/onlinedocs/ 15473 2. mailto:gcc-help (a] gcc.gnu.org 15474 3. mailto:gcc (a] gcc.gnu.org 15475 4. https://gcc.gnu.org/lists.html 15476 5. http://www.fsf.org/ 15477 6. https://gcc.gnu.org/about.html 15478 7. http://validator.w3.org/check/referer 15479 ====================================================================== 15480 http://gcc.gnu.org/egcs-1.0/index.html 15481 EGCS 1.0 15482 15483 December 3, 1997: We are pleased to announce the release of EGCS 1.0. 15484 January 6, 1998: We are pleased to announce the release of EGCS 1.0.1. 15485 March 16, 1998: We are pleased to announce the release of EGCS 1.0.2. 15486 May 15, 1998 We are pleased to announce the release of EGCS 1.0.3. 15487 15488 EGCS is a collaborative effort involving several groups of hackers 15489 using an open development model to accelerate development and testing 15490 of GNU compilers and runtime libraries. 15491 15492 An important goal of EGCS is to allow wide scale testing of 15493 experimental features and optimizations; therefore, EGCS contains some 15494 features and optimizations which are still under development. However, 15495 EGCS has been carefully tested and should be comparable in quality to 15496 most GCC releases. 15497 15498 EGCS 1.0 is based on an August 2, 1997 snapshot of the GCC 2.8 15499 development sources; it contains nearly all of the new features found 15500 in GCC 2.8. 15501 15502 EGCS 1.0 also contains many improvements and features not found in GCC 15503 2.7 and even the GCC 2.8 series (which was released after the original 15504 EGCS 1.0 release). 15505 * Integrated C++ runtime libraries, including support for most major 15506 GNU/Linux systems! 15507 * The integrated libstdc++ library includes a verbatim copy of SGI's 15508 STL release. 15509 * Integrated GNU Fortran compiler. 15510 * New instruction scheduler. 15511 * New alias analysis code. 15512 15513 See the [1]new features page for a more complete list of new features. 15514 15515 EGCS 1.0.1 is a minor update to the EGCS 1.0 compiler to fix a few 15516 critical bugs and add support for Red Hat 5.0 Linux. Changes since the 15517 EGCS 1.0 release: 15518 * Add support for Red Hat 5.0 Linux and better support for Linux 15519 systems using glibc2. 15520 Many programs failed to link when compiled with EGCS 1.0 on Red Hat 15521 5.0 or on systems with newer versions of glibc2. EGCS 1.0.1 should 15522 fix these problems. 15523 * Compatibility with both EGCS 1.0 and GCC 2.8 libgcc exception 15524 handling interfaces. 15525 To avoid future compatibility problems, we strongly urge anyone who 15526 is planning on distributing shared libraries that contain C++ code 15527 to upgrade to EGCS 1.0.1 first. 15528 Soon after EGCS 1.0 was released, the GCC developers made some 15529 incompatible changes in libgcc's exception handling interfaces. 15530 These changes were needed to solve problems on some platforms. This 15531 means that GCC 2.8.0, when released, will not be seamlessly 15532 compatible with shared libraries built by EGCS 1.0. The reason is 15533 that the libgcc.a in GCC 2.8.0 will not contain a function needed 15534 by the old interface. 15535 The result of this is that there may be compatibility problems with 15536 shared libraries built by EGCS 1.0 when used with GCC 2.8.0. 15537 With EGCS 1.0.1, generated code uses the new (GCC 2.8.0) interface, 15538 and libgcc.a has the support routines for both the old and the new 15539 interfaces (so EGCS 1.0.1 and EGCS 1.0 code can be freely mixed, 15540 and EGCS 1.0.1 and GCC 2.8.0 code can be freely mixed). 15541 The maintainers of GCC 2.x have decided against including seamless 15542 support for the old interface in 2.8.0, since it was never 15543 "official", so to avoid future compatibility problems we recommend 15544 against distributing any shared libraries built by EGCS 1.0 that 15545 contain C++ code (upgrade to 1.0.1 and use that). 15546 * Various bugfixes in the x86, hppa, mips, and rs6000/ppc back ends. 15547 The x86 changes fix code generation errors exposed when building 15548 glibc2 and the usual GNU/Linux dynamic linker (ld.so). 15549 The hppa change fixes a compiler abort when configured for use with 15550 RTEMS. 15551 The MIPS changes fix problems with the definition of LONG_MAX on 15552 newer systems, allow for command line selection of the target ABI, 15553 and fix one code generation problem. 15554 The rs6000/ppc change fixes some problems with passing structures 15555 to varargs/stdarg functions. 15556 * A few machine independent bugfixes, mostly to fix code generation 15557 errors when building Linux kernels or glibc. 15558 * Fix a few critical exception handling and template bugs in the C++ 15559 compiler. 15560 * Fix Fortran namelist bug on alphas. 15561 * Fix build problems on x86-solaris systems. 15562 15563 EGCS 1.0.2 is a minor update to the EGCS 1.0.1 compiler to fix several 15564 serious problems in EGCS 1.0.1. 15565 * General improvements and fixes 15566 + Memory consumption significantly reduced, especially for 15567 templates and inline functions. 15568 + Fix various problems with glibc2.1. 15569 + Fix loop optimization bug exposed by rs6000/ppc port. 15570 + Fix to avoid potential code generation problems in jump.c. 15571 + Fix some undefined symbol problems in dwarf1 debug support. 15572 * g++/libstdc++ improvements and fixes 15573 + libstdc++ in the EGCS release has been updated and should be 15574 link compatible with libstdc++-2.8. 15575 + Various fixes in libio/libstdc++ to work better on GNU/Linux 15576 systems. 15577 + Fix problems with duplicate symbols on systems that do not 15578 support weak symbols. 15579 + Memory corruption bug and undefined symbols in bastring have 15580 been fixed. 15581 + Various exception handling fixes. 15582 + Fix compiler abort for very long thunk names. 15583 * g77 improvements and fixes 15584 + Fix compiler crash for omitted bound in Fortran CASE 15585 statement. 15586 + Add missing entries to g77 lang-options. 15587 + Fix problem with -fpedantic in the g77 compiler. 15588 + Fix "backspace" problem with g77 on alphas. 15589 + Fix x86 backend problem with Fortran literals and -fpic. 15590 + Fix some of the problems with negative subscripts for g77 on 15591 alphas. 15592 + Fixes for Fortran builds on cygwin32/mingw32. 15593 * platform specific improvements and fixes 15594 + Fix long double problems on x86 (exposed by glibc). 15595 + x86 ports define i386 again to keep imake happy. 15596 + Fix exception handling support on NetBSD ports. 15597 + Several changes to collect2 to fix many problems with AIX. 15598 + Define __ELF__ for GNU/Linux on rs6000. 15599 + Fix -mcall-linux problem on GNU/Linux on rs6000. 15600 + Fix stdarg/vararg problem for GNU/Linux on rs6000. 15601 + Allow autoconf to select a proper install problem on AIX 3.1. 15602 + m68k port support includes -mcpu32 option as well as cpu32 15603 multilibs. 15604 + Fix stdarg bug for irix6. 15605 + Allow EGCS to build on irix5 without the gnu assembler. 15606 + Fix problem with static linking on sco5. 15607 + Fix bootstrap on sco5 with native compiler. 15608 + Fix for abort building newlib on H8 target. 15609 + Fix fixincludes handling of math.h on SunOS. 15610 + Minor fix for Motorola 3300 m68k systems. 15611 15612 EGCS 1.0.3 is a minor update to the EGCS 1.0.2 compiler to fix a few 15613 problems reported by Red Hat for builds of Red Hat 5.1. 15614 * Generic bugfixes: 15615 + Fix a typo in the libio library which resulted in incorrect 15616 behavior of istream::get. 15617 + Fix the Fortran negative array index problem. 15618 + Fix a major problem with the ObjC runtime thread support 15619 exposed by glibc2. 15620 + Reduce memory consumption of the Haifa scheduler. 15621 * Target specific bugfixes: 15622 + Fix one x86 floating point code generation bug exposed by 15623 glibc2 builds. 15624 + Fix one x86 internal compiler error exposed by glibc2 builds. 15625 + Fix profiling bugs on the Alpha. 15626 + Fix ImageMagick & emacs 20.2 build problems on the Alpha. 15627 + Fix rs6000/ppc bug when converting values from integer types 15628 to floating point types. 15629 15630 The EGCS 1.0 releases include installation instructions in both HTML 15631 and plaintext forms (see the INSTALL directory in the toplevel 15632 directory of the distribution). However, we also keep the most up to 15633 date installation instructions and [2]build/test status on our web 15634 page. We will update those pages as new information becomes available. 15635 15636 And, we can't in good conscience fail to mention some [3]caveats to 15637 using EGCS. 15638 15639 Update: Big thanks to Stanford for providing a high speed link for 15640 downloading EGCS (go.cygnus.com)! 15641 15642 Download EGCS from ftp.cygnus.com (USA California) or go.cygnus.com 15643 (USA California -- High speed link provided by Stanford). 15644 15645 The EGCS 1.0 release is also available many mirror sites. 15646 [4]Goto mirror list to find a closer site 15647 15648 We'd like to thank the numerous people that have contributed new 15649 features, test results, bugfixes, etc. Unfortunately, they're far too 15650 numerous to mention by name. 15651 15652 15653 For questions related to the use of GCC, please consult these web 15654 pages and the [5]GCC manuals. If that fails, the 15655 [6]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15656 web pages and the development of GCC are welcome on our developer 15657 list at [7]gcc (a] gcc.gnu.org. All of [8]our lists have public archives. 15658 15659 Copyright (C) [9]Free Software Foundation, Inc. Verbatim copying and 15660 distribution of this entire article is permitted in any medium, 15661 provided this notice is preserved. 15662 15663 These pages are [10]maintained by the GCC team. Last modified 15664 2016-01-30[11]. 15665 15666 References 15667 15668 1. http://gcc.gnu.org/egcs-1.0/features.html 15669 2. http://gcc.gnu.org/egcs-1.0/buildstat.html 15670 3. http://gcc.gnu.org/egcs-1.0/caveats.html 15671 4. http://gcc.gnu.org/mirrors.html 15672 5. https://gcc.gnu.org/onlinedocs/ 15673 6. mailto:gcc-help (a] gcc.gnu.org 15674 7. mailto:gcc (a] gcc.gnu.org 15675 8. https://gcc.gnu.org/lists.html 15676 9. http://www.fsf.org/ 15677 10. https://gcc.gnu.org/about.html 15678 11. http://validator.w3.org/check/referer 15679 ====================================================================== 15680 http://gcc.gnu.org/egcs-1.0/features.html 15681 EGCS 1.0 features 15682 15683 * Core compiler is based on the gcc2 development tree from Aug 2, 15684 1997, so we have most of the [1]features found in GCC 2.8. 15685 * Integrated GNU Fortran compiler based on g77-0.5.22-19970929. 15686 * Vast improvements in the C++ compiler; so many they have [2]page of 15687 their own! 15688 * Integrated C++ runtime libraries, including support for most major 15689 GNU/Linux systems! 15690 * New instruction scheduler from IBM Haifa which includes support for 15691 function wide instruction scheduling as well as superscalar 15692 scheduling. 15693 * Significantly improved alias analysis code. 15694 * Improved register allocation for two address machines. 15695 * Significant code generation improvements for Fortran code on 15696 Alphas. 15697 * Various optimizations from the g77 project as well as improved loop 15698 optimizations. 15699 * Dwarf2 debug format support for some targets. 15700 * egcs libstdc++ includes the SGI STL implementation without changes. 15701 * As a result of these and other changes, egcs libstc++ is not binary 15702 compatible with previous releases of libstdc++. 15703 * Various new ports -- UltraSPARC, Irix6.2 & Irix6.3 support, The SCO 15704 Openserver 5 family (5.0.{0,2,4} and Internet FastStart 1.0 and 15705 1.1), Support for RTEMS on several embedded targets, Support for 15706 arm-linux, Mitsubishi M32R, Hitachi H8/S, Matsushita MN102 and 15707 MN103, NEC V850, Sparclet, Solaris & GNU/Linux on PowerPCs, etc. 15708 * Integrated testsuites for gcc, g++, g77, libstdc++ and libio. 15709 * RS6000/PowerPC ports generate code which can run on all 15710 RS6000/PowerPC variants by default. 15711 * -mcpu= and -march= switches for the x86 port to allow better 15712 control over how the x86 port generates code. 15713 * Includes the template repository patch (aka repo patch); note the 15714 new template code makes repo obsolete for ELF systems using gnu-ld 15715 such as GNU/Linux. 15716 * Plus the usual assortment of bugfixes and improvements. 15717 15718 15719 For questions related to the use of GCC, please consult these web 15720 pages and the [3]GCC manuals. If that fails, the 15721 [4]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15722 web pages and the development of GCC are welcome on our developer 15723 list at [5]gcc (a] gcc.gnu.org. All of [6]our lists have public archives. 15724 15725 Copyright (C) [7]Free Software Foundation, Inc. Verbatim copying and 15726 distribution of this entire article is permitted in any medium, 15727 provided this notice is preserved. 15728 15729 These pages are [8]maintained by the GCC team. Last modified 15730 2016-01-30[9]. 15731 15732 References 15733 15734 1. http://gcc.gnu.org/egcs-1.0/features-2.8.html 15735 2. http://gcc.gnu.org/egcs-1.0/c++features.html 15736 3. https://gcc.gnu.org/onlinedocs/ 15737 4. mailto:gcc-help (a] gcc.gnu.org 15738 5. mailto:gcc (a] gcc.gnu.org 15739 6. https://gcc.gnu.org/lists.html 15740 7. http://www.fsf.org/ 15741 8. https://gcc.gnu.org/about.html 15742 9. http://validator.w3.org/check/referer 15743 ====================================================================== 15744 http://gcc.gnu.org/egcs-1.0/caveats.html 15745 EGCS 1.0 Caveats 15746 15747 * EGCS has an integrated libstdc++, but does not have an integrated 15748 libg++. Furthermore old libg++ releases will not work with egc; HJ 15749 Lu has made a libg++-2.8.1.2 available which may work with EGCS. 15750 Note most C++ programs only need libstdc++. 15751 * Note that using -pedantic or -Wreturn-type can cause an explosion 15752 in the amount of memory needed for template-heavy C++ code, such as 15753 code that uses STL. Also note that -Wall includes -Wreturn-type, so 15754 if you use -Wall you will need to specify -Wno-return-type to turn 15755 it off. 15756 * Exception handling may not work with shared libraries, particularly 15757 on alphas, hppas, and mips based platforms. Exception handling is 15758 known to work on x86-linux platforms with shared libraries. 15759 * Some versions of the Linux kernel have bugs which prevent them from 15760 being compiled or from running when compiled by EGCS. See the FAQ 15761 (as shipped with EGCS 1.0) for additional information. 15762 * In general, EGCS is more rigorous about rejecting invalid C++ code 15763 or deprecated C++ constructs than G++ 2.7. As a result it may be 15764 necessary to fix C++ code before it will compile with EGCS. 15765 * G++ is also aggressively tracking the C++ standard; as a result 15766 code which was previously valid (and thus accepted by other 15767 compilers and older versions of G++) may no longer be accepted. 15768 * EGCS 1.0 may not work with Red Hat Linux 5.0 on all targets. EGCS 15769 1.0.x and later releases should work with Red Hat Linux 5.0. 15770 15771 15772 For questions related to the use of GCC, please consult these web 15773 pages and the [1]GCC manuals. If that fails, the 15774 [2]gcc-help (a] gcc.gnu.org mailing list might help. Comments on these 15775 web pages and the development of GCC are welcome on our developer 15776 list at [3]gcc (a] gcc.gnu.org. All of [4]our lists have public archives. 15777 15778 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and 15779 distribution of this entire article is permitted in any medium, 15780 provided this notice is preserved. 15781 15782 These pages are [6]maintained by the GCC team. Last modified 15783 2016-01-30[7]. 15784 15785 References 15786 15787 1. https://gcc.gnu.org/onlinedocs/ 15788 2. mailto:gcc-help (a] gcc.gnu.org 15789 3. mailto:gcc (a] gcc.gnu.org 15790 4. https://gcc.gnu.org/lists.html 15791 5. http://www.fsf.org/ 15792 6. https://gcc.gnu.org/about.html 15793 7. http://validator.w3.org/check/referer 15794 ====================================================================== 15795
Indexes created Tue Oct 27 00:25:24 UTC 2026